paperID
stringlengths 36
36
| pwc_id
stringlengths 8
47
| arxiv_id
stringlengths 6
16
⌀ | nips_id
float64 | url_abs
stringlengths 18
329
| url_pdf
stringlengths 18
742
| title
stringlengths 8
325
| abstract
stringlengths 1
7.27k
⌀ | authors
stringlengths 2
7.06k
| published
stringlengths 10
10
⌀ | conference
stringlengths 12
47
⌀ | conference_url_abs
stringlengths 16
198
⌀ | conference_url_pdf
stringlengths 27
199
⌀ | proceeding
stringlengths 6
47
⌀ | taskID
stringlengths 7
1.44k
| areaID
stringclasses 688
values | embedding
stringlengths 9.26k
12.5k
| umap_embedding
stringlengths 29
44
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
f2d1a241-15dc-476d-9c87-4033771974fe
|
rest-retrieve-self-train-for-generative
|
2209.15000
| null |
https://arxiv.org/abs/2209.15000v1
|
https://arxiv.org/pdf/2209.15000v1.pdf
|
REST: REtrieve & Self-Train for generative action recognition
|
This work is on training a generative action/video recognition model whose output is a free-form action-specific caption describing the video (rather than an action class label). A generative approach has practical advantages like producing more fine-grained and human-readable output, and being naturally open-world. To this end, we propose to adapt a pre-trained generative Vision & Language (V&L) Foundation Model for video/action recognition. While recently there have been a few attempts to adapt V&L models trained with contrastive learning (e.g. CLIP) for video/action, to the best of our knowledge, we propose the very first method that sets outs to accomplish this goal for a generative model. We firstly show that direct fine-tuning of a generative model to produce action classes suffers from severe overfitting. To alleviate this, we introduce REST, a training framework consisting of two key components: an unsupervised method for adapting the generative model to action/video by means of pseudo-caption generation and Self-training, i.e. without using any action-specific labels; (b) a Retrieval approach based on CLIP for discovering a diverse set of pseudo-captions for each video to train the model. Importantly, we show that both components are necessary to obtain high accuracy. We evaluate REST on the problem of zero-shot action recognition where we show that our approach is very competitive when compared to contrastive learning-based methods. Code will be made available.
|
['Georgios Tzimiropoulos', 'Brais Martinez', 'Enrique Sanchez', 'Adrian Bulat']
|
2022-09-29
| null | null | null | null |
['zero-shot-action-recognition', 'video-recognition']
|
['computer-vision', 'computer-vision']
|
[ 7.05423176e-01 1.65324256e-01 -3.70762721e-02 -3.44363511e-01
-9.98378694e-01 -6.73666954e-01 9.48385358e-01 -5.88854671e-01
-4.02495414e-02 6.58121765e-01 3.56193125e-01 -1.71634257e-02
-9.28891897e-02 -6.20579243e-01 -9.72633541e-01 -7.02976465e-01
1.07079387e-01 6.87475562e-01 3.65687519e-01 -1.26492754e-01
1.88699573e-01 4.48815405e-01 -1.87480831e+00 6.59076035e-01
4.92758632e-01 8.18628311e-01 1.80517435e-01 9.59003210e-01
-3.18050273e-02 1.10817242e+00 -5.69362998e-01 -3.45926851e-01
3.21691096e-01 -1.10422146e+00 -1.03734112e+00 7.00666785e-01
3.68467867e-01 -3.75404030e-01 -2.63104647e-01 6.57367527e-01
2.46723369e-01 2.63251603e-01 9.69913661e-01 -1.33540225e+00
-8.04426134e-01 4.17696804e-01 6.77513853e-02 7.88964033e-02
6.97155774e-01 3.38541389e-01 6.88718498e-01 -8.05451512e-01
9.02080119e-01 1.09465611e+00 4.37342405e-01 9.97837782e-01
-1.20779264e+00 -1.66993514e-01 -8.90115350e-02 2.69355685e-01
-1.33883607e+00 -7.11215854e-01 7.92437077e-01 -5.53667009e-01
9.59105670e-01 3.00281614e-01 7.69119263e-01 1.61490297e+00
-1.20737232e-01 9.20142174e-01 9.66750562e-01 -6.73402905e-01
3.93980294e-01 2.21822873e-01 -3.35926354e-01 5.84235072e-01
-1.86785445e-01 1.73003401e-03 -4.37123686e-01 7.06909373e-02
9.30828929e-01 -3.27278525e-02 -2.02444285e-01 -6.75812364e-01
-9.82286274e-01 7.39389777e-01 9.53502655e-02 4.85931247e-01
-4.27287221e-01 2.70846516e-01 3.67686421e-01 3.77741843e-01
3.78244221e-01 3.44073623e-01 -2.29967266e-01 -4.08156872e-01
-1.08697832e+00 2.35515103e-01 9.20143604e-01 1.00425220e+00
6.86770976e-01 5.54725341e-02 -3.59493345e-01 5.90221167e-01
3.58290553e-01 3.26404274e-01 7.48292923e-01 -1.10230350e+00
1.69270352e-01 4.01772290e-01 1.05859227e-01 -6.81975424e-01
3.37248929e-02 -5.42930234e-03 -4.15854543e-01 2.66764700e-01
2.37893179e-01 6.76142722e-02 -1.14492548e+00 1.70183504e+00
4.67279814e-02 2.13587046e-01 3.71716917e-01 6.98436797e-01
7.85148561e-01 7.62382925e-01 1.97645172e-01 -3.38648856e-01
9.20341492e-01 -1.11491728e+00 -5.78479409e-01 -1.76985398e-01
5.06300032e-01 -6.67576849e-01 9.01360571e-01 2.21855134e-01
-1.14655101e+00 -7.73153186e-01 -7.95055926e-01 1.86700866e-01
-4.06710207e-01 1.10402294e-01 5.96832275e-01 6.47136569e-01
-1.30967319e+00 5.54976046e-01 -8.97615254e-01 -8.30576420e-01
3.19878191e-01 1.62226990e-01 -4.72636551e-01 -1.44499302e-01
-9.04159784e-01 7.55257905e-01 6.19677365e-01 -2.94504642e-01
-1.31697893e+00 -1.62264839e-01 -8.40775013e-01 -5.04804626e-02
5.57904184e-01 -8.98284018e-01 1.30560684e+00 -1.70135152e+00
-1.76283395e+00 9.42596912e-01 1.07901758e-02 -4.80565518e-01
4.72161561e-01 -8.39272738e-02 -3.12815338e-01 4.65731472e-01
2.79653650e-02 9.68241334e-01 1.26076484e+00 -1.27544403e+00
-4.39261675e-01 -4.14846614e-02 2.96769530e-01 1.43715993e-01
-2.85083592e-01 9.76784527e-02 -5.81082284e-01 -6.67923689e-01
-2.25846991e-01 -9.92948055e-01 -1.17979795e-02 -3.06573659e-01
-1.45498171e-01 -2.20397353e-01 7.12403536e-01 -4.22381788e-01
1.14486527e+00 -2.01515222e+00 2.40812361e-01 -7.06954896e-02
-2.27073550e-01 6.91677690e-01 -4.22461241e-01 8.29581618e-01
-2.65168726e-01 1.04281582e-01 -3.12792569e-01 -3.06974381e-01
-6.05114183e-05 5.42415440e-01 -3.23929787e-01 1.96769282e-01
4.87999886e-01 1.05447483e+00 -9.71093237e-01 -5.59786916e-01
3.58481497e-01 6.46906137e-01 -6.34026885e-01 4.68707830e-01
-4.23908114e-01 4.96249735e-01 -5.19758403e-01 5.88685811e-01
1.28857151e-01 -2.20691562e-01 1.32411048e-01 -2.50088191e-03
4.46056165e-02 2.51151845e-02 -1.19232261e+00 1.73858202e+00
-2.36289054e-01 4.48610991e-01 -4.09251928e-01 -1.17375684e+00
8.71209025e-01 5.88654578e-01 6.23549283e-01 -3.91799867e-01
1.15990259e-01 1.48550659e-01 -2.89581358e-01 -8.84680748e-01
2.54301667e-01 -2.41262719e-01 -9.06744525e-02 4.92597371e-01
6.41126394e-01 -7.22395182e-02 5.46766222e-01 1.33087337e-01
1.24900568e+00 9.43330288e-01 3.34485978e-01 1.83521062e-01
6.10059977e-01 1.01381145e-01 3.05098534e-01 7.94422507e-01
-1.11448886e-02 1.00846183e+00 3.04885954e-01 -4.55562800e-01
-1.05663991e+00 -9.37112749e-01 2.46666804e-01 1.15259206e+00
-3.47308069e-01 -5.42718768e-01 -1.06787634e+00 -7.47578382e-01
-4.71817255e-01 6.84108496e-01 -7.04831362e-01 -1.65361449e-01
-5.93430102e-01 -4.52293575e-01 3.48856956e-01 6.57680631e-01
3.47313762e-01 -1.50415277e+00 -6.77182078e-01 2.70325005e-01
-1.11265786e-01 -1.03207672e+00 -3.22386801e-01 1.70606658e-01
-7.83300877e-01 -1.15623295e+00 -8.11916351e-01 -7.41682887e-01
7.37263680e-01 8.97871628e-02 1.17683291e+00 -8.04246366e-02
-1.61201179e-01 9.25510466e-01 -8.99028420e-01 -1.74172670e-01
-9.29398239e-01 -1.46077856e-01 -1.88304484e-01 2.73400187e-01
4.13415998e-01 -6.93776190e-01 -3.39717329e-01 2.44105205e-01
-1.41640031e+00 2.03829587e-01 8.40419292e-01 5.77558100e-01
6.54013038e-01 -2.47849807e-01 3.15364540e-01 -8.52852046e-01
3.57064128e-01 -3.57810497e-01 -3.00462514e-01 3.39243889e-01
-4.12028641e-01 1.03303313e-01 6.16135836e-01 -5.56019068e-01
-9.61163819e-01 6.34284675e-01 -2.46658802e-01 -8.00725937e-01
-5.39159358e-01 2.03287959e-01 -1.70734271e-01 -5.44348918e-02
8.54172587e-01 5.57101667e-01 -4.95695509e-02 -5.05342841e-01
5.95698237e-01 6.98239982e-01 7.19665825e-01 -2.69136310e-01
8.57482493e-01 3.06435317e-01 -8.79187882e-02 -6.69285417e-01
-7.72211611e-01 -5.09824932e-01 -9.26109433e-01 -4.34571505e-01
9.86850739e-01 -7.62105286e-01 -9.67368037e-02 3.64349961e-01
-9.29316282e-01 -5.00936747e-01 -6.33617461e-01 3.62455338e-01
-1.37332964e+00 3.89342487e-01 -4.57225263e-01 -7.84373224e-01
-7.44935274e-02 -9.63340282e-01 1.21632695e+00 1.73814461e-01
-1.55002192e-01 -9.49041069e-01 4.04540390e-01 3.87794524e-01
3.70531082e-01 4.11668509e-01 3.28930557e-01 -7.90554464e-01
-6.95225894e-01 -1.51010990e-01 1.19500279e-01 6.17074311e-01
1.14787541e-01 1.31630570e-01 -9.57508147e-01 -1.30387872e-01
5.54015227e-02 -5.87623954e-01 8.26300979e-01 2.55754977e-01
9.29991961e-01 -4.42206919e-01 -2.80442387e-01 5.48113585e-01
1.40815926e+00 3.11296821e-01 1.03868449e+00 2.78355986e-01
5.61011076e-01 2.88157433e-01 6.65678740e-01 4.14431661e-01
6.50794953e-02 9.50758755e-01 2.43561998e-01 1.18256338e-01
-4.77467030e-01 -4.88590389e-01 6.38480783e-01 5.00513077e-01
-4.53011692e-01 -4.17296052e-01 -6.94545984e-01 6.02057815e-01
-2.00553179e+00 -1.39520550e+00 6.03564829e-02 2.13689923e+00
7.39684463e-01 2.93766405e-03 3.76625359e-01 2.39090938e-02
5.37322700e-01 4.71465513e-02 -2.38281444e-01 -4.34192181e-01
1.05263904e-01 3.24330211e-01 1.43745929e-01 2.49220878e-01
-1.34723806e+00 1.07485998e+00 6.51269054e+00 7.70353496e-01
-9.25917506e-01 1.61648110e-01 3.15103799e-01 6.98067322e-02
-1.43111438e-01 1.15209684e-01 -7.17186570e-01 3.93947661e-01
1.18017769e+00 -7.10306019e-02 3.99317354e-01 9.81824338e-01
1.00608513e-01 1.00464962e-01 -1.29594445e+00 1.01673520e+00
7.54985988e-01 -1.23324978e+00 3.64038557e-01 -1.06459605e-02
7.74101257e-01 -1.92839250e-01 -2.71155149e-01 4.64709967e-01
2.14958698e-01 -8.07605982e-01 8.00297201e-01 8.53451610e-01
7.45958447e-01 -4.13510919e-01 5.07764399e-01 3.95635933e-01
-9.88772750e-01 -9.56699625e-03 -2.23327786e-01 1.62529387e-02
2.90144235e-01 5.86669855e-02 -7.21444070e-01 5.44794798e-01
4.59936470e-01 7.57468283e-01 -6.29935622e-01 1.13774216e+00
-4.16673392e-01 6.07508123e-01 -2.81330012e-03 1.79765716e-01
3.09781820e-01 1.40129933e-02 5.69747150e-01 1.36863899e+00
3.70109379e-01 1.93747506e-01 3.55179787e-01 7.27024376e-01
1.47607476e-01 9.10240635e-02 -8.03764284e-01 -4.16587114e-01
-1.91862255e-01 1.15282285e+00 -7.26631463e-01 -5.14645934e-01
-4.03714627e-01 1.25919652e+00 1.27702609e-01 3.05379331e-01
-8.83450150e-01 -2.74262968e-02 1.89519197e-01 2.00731277e-01
6.65153325e-01 -2.33560111e-02 4.54405844e-01 -1.28551531e+00
-1.02100313e-01 -8.85470331e-01 3.00345242e-01 -1.01818752e+00
-1.12290704e+00 7.85294533e-01 2.75751323e-01 -1.51484394e+00
-1.13267434e+00 -5.22381008e-01 -3.75179946e-01 5.11584520e-01
-1.19364250e+00 -1.33299136e+00 -2.63008714e-01 6.79086328e-01
8.91045451e-01 -2.71324873e-01 1.02495778e+00 1.31941080e-01
-2.06968993e-01 2.27479503e-01 -1.08606897e-01 1.27695948e-01
6.45972610e-01 -1.12551129e+00 1.91014603e-01 9.51983571e-01
5.87193429e-01 2.98962682e-01 7.99046993e-01 -5.02874434e-01
-1.33708906e+00 -1.22638786e+00 9.67313766e-01 -7.59657383e-01
4.34349447e-01 -2.94704854e-01 -7.74539411e-01 9.57918882e-01
1.66243300e-01 1.67591088e-02 7.74976015e-01 -3.03725451e-01
-1.54213041e-01 4.91941795e-02 -9.05813754e-01 3.73348325e-01
1.25358164e+00 -4.89606917e-01 -9.46135879e-01 6.33084536e-01
3.19571823e-01 -2.30279118e-01 -7.23322392e-01 3.28895539e-01
3.54896158e-01 -1.11502314e+00 9.28140223e-01 -6.79191768e-01
5.72741270e-01 -3.59182775e-01 -1.40972018e-01 -1.02109647e+00
-4.00119156e-01 -7.04383433e-01 -2.43130594e-01 1.33864582e+00
2.14009792e-01 -2.16483712e-01 6.89699471e-01 4.82871771e-01
-3.86143774e-01 -5.76464236e-01 -7.06033647e-01 -9.57703054e-01
-4.05633658e-01 -3.93668860e-01 1.17576472e-01 6.60314143e-01
-1.76477402e-01 4.63563561e-01 -5.94244361e-01 -2.83766955e-01
2.01360270e-01 1.82163894e-01 1.08596027e+00 -9.13867772e-01
-7.03770816e-01 -2.40031183e-01 -7.01022267e-01 -1.02654052e+00
1.12300538e-01 -6.85691297e-01 1.68977112e-01 -1.57936037e+00
4.58900720e-01 -1.88131109e-02 -7.06718937e-02 7.66618848e-01
1.35716602e-01 6.61712229e-01 2.07255274e-01 4.95695382e-01
-9.97640967e-01 4.91341412e-01 1.10947227e+00 -7.24901503e-04
-2.56488889e-01 8.44053999e-02 -5.54254889e-01 6.29885972e-01
5.33648610e-01 -4.84480500e-01 -5.93567967e-01 -1.40143618e-01
-2.62929630e-02 1.40157208e-01 5.42542934e-01 -1.27372670e+00
-9.25171897e-02 -2.01378822e-01 2.85606682e-01 -1.92351028e-01
4.09721971e-01 -5.67145884e-01 5.71985841e-01 2.31179059e-01
-3.81857812e-01 -3.85393649e-01 -7.06621930e-02 4.99304742e-01
-3.50324392e-01 -5.57950795e-01 6.28524482e-01 -4.36348200e-01
-1.04599249e+00 1.89774811e-01 -5.10028481e-01 -2.29297325e-01
1.18746042e+00 -5.08482575e-01 -5.65993339e-02 -6.03199482e-01
-9.89629447e-01 -3.02155316e-01 7.16212213e-01 5.73739409e-01
4.52542067e-01 -1.26730251e+00 -7.60280192e-01 1.51938021e-01
2.97904551e-01 -4.79463309e-01 2.22950369e-01 6.04830146e-01
-4.06314045e-01 5.67613184e-01 -4.71690804e-01 -4.64708477e-01
-1.19873631e+00 9.13222909e-01 1.62993714e-01 -2.53016979e-01
-6.15430772e-01 6.32820129e-01 1.77883506e-01 5.35954721e-02
3.93289849e-02 2.32384168e-02 -3.28516811e-01 -9.29186270e-02
6.18447304e-01 2.84585450e-02 -1.43051475e-01 -1.02093518e+00
-2.26249322e-01 4.81702864e-01 2.18312547e-01 -1.97052971e-01
1.39805198e+00 -3.94991376e-02 2.35369623e-01 4.30856913e-01
1.08898485e+00 -2.66775548e-01 -1.48753536e+00 6.33494109e-02
-3.09539974e-01 -4.67649609e-01 -2.48511896e-01 -7.33340621e-01
-8.73508334e-01 5.24406135e-01 6.12637877e-01 4.22142118e-01
1.40752518e+00 3.58879626e-01 5.48057199e-01 3.78627151e-01
3.51587862e-01 -1.02941155e+00 3.48693311e-01 3.11428487e-01
9.64063883e-01 -1.07103717e+00 -1.95442200e-01 -3.21910650e-01
-6.87206984e-01 1.19933236e+00 3.36057574e-01 -2.15874374e-01
2.81186968e-01 1.96854919e-02 3.33824456e-02 -1.95077226e-01
-7.74344742e-01 -6.66327834e-01 4.18092340e-01 8.50001812e-01
3.29063058e-01 -1.69880927e-01 -4.48368013e-01 1.95943460e-01
9.22624022e-02 5.47823966e-01 5.00734329e-01 1.12912452e+00
-4.52922046e-01 -1.52601063e+00 -1.85799927e-01 1.76848650e-01
-3.50735694e-01 7.29394183e-02 -5.92812836e-01 7.90856421e-01
2.18170717e-01 8.00651550e-01 -4.04324718e-02 -3.42631817e-01
2.28810549e-01 5.08627653e-01 7.00097740e-01 -8.47789824e-01
-3.38772893e-01 1.11966133e-01 2.39322875e-02 -7.06277728e-01
-1.04619205e+00 -7.90138304e-01 -7.52635896e-01 2.63276100e-01
6.06059358e-02 6.66007102e-02 4.69680548e-01 1.08332574e+00
3.67704511e-01 3.34410042e-01 5.10366321e-01 -1.07247972e+00
-3.58398288e-01 -8.93457651e-01 -4.81900632e-01 8.30772042e-01
-1.55096799e-01 -6.03567362e-01 -3.35750282e-01 8.46936345e-01]
|
[8.751005172729492, 0.7382466197013855]
|
a3e93e8c-bad8-4c69-9d34-5817093e017a
|
autoencoder-based-time-series-clustering-with
|
2002.03624
| null |
https://arxiv.org/abs/2002.03624v1
|
https://arxiv.org/pdf/2002.03624v1.pdf
|
Autoencoder-based time series clustering with energy applications
|
Time series clustering is a challenging task due to the specific nature of the data. Classical approaches do not perform well and need to be adapted either through a new distance measure or a data transformation. In this paper we investigate the combination of a convolutional autoencoder and a k-medoids algorithm to perfom time series clustering. The convolutional autoencoder allows to extract meaningful features and reduce the dimension of the data, leading to an improvement of the subsequent clustering. Using simulation and energy related data to validate the approach, experimental results show that the clustering is robust to outliers thus leading to finer clusters than with standard methods.
|
['Georges Hébrail', 'Benoît Grossin', 'Anne de Moliner', 'Guillaume Richard', 'Guillaume Germaine']
|
2020-02-10
| null | null | null | null |
['time-series-clustering']
|
['time-series']
|
[-3.65751863e-01 -4.21252191e-01 5.14922619e-01 -3.40325147e-01
-6.87947720e-02 -4.10206497e-01 5.82779765e-01 7.09189355e-01
-6.32945895e-01 3.72063965e-01 2.99344957e-02 9.02372971e-02
-7.18707442e-01 -7.99433470e-01 -4.51193601e-01 -8.27299297e-01
-5.06298304e-01 5.44068933e-01 1.18479632e-01 -2.09577024e-01
1.07437909e-01 7.71029770e-01 -1.84915841e+00 9.02404189e-02
6.32442117e-01 9.24856007e-01 -1.11935087e-01 3.97900701e-01
-6.66823611e-02 2.92187810e-01 -7.97630191e-01 4.90175158e-01
4.39033270e-01 -3.81174773e-01 -4.20690596e-01 1.65314794e-01
-2.85352021e-01 1.03655301e-01 -2.47352887e-02 8.26286018e-01
4.90491986e-01 8.03539515e-01 7.18382061e-01 -1.24794924e+00
-1.00785628e-01 6.03295922e-01 1.78941116e-02 1.56581417e-01
7.82107934e-02 -2.85254955e-01 4.16016161e-01 -4.58867490e-01
3.77007425e-01 8.50029647e-01 7.94557333e-01 3.24351639e-02
-1.38443506e+00 -2.45415777e-01 -3.93610716e-01 5.36095381e-01
-1.51729858e+00 2.58368775e-02 1.07301760e+00 -5.23119152e-01
9.68070984e-01 1.24610782e-01 8.86448205e-01 7.67504513e-01
-2.11961623e-02 1.15443230e-01 1.11492455e+00 -5.27766824e-01
9.05844510e-01 -1.29460152e-02 6.33806586e-02 5.02031706e-02
1.66850075e-01 6.86213449e-02 1.69674844e-01 1.51904076e-02
2.46063873e-01 3.28722805e-01 -8.77919346e-02 -3.60474348e-01
-9.34521437e-01 9.53781545e-01 4.95141536e-01 1.22345102e+00
-7.74446070e-01 1.56466551e-02 6.58980668e-01 3.63222361e-01
2.77492374e-01 5.21914840e-01 -3.12456936e-01 -6.84534907e-02
-1.29184973e+00 1.54549494e-01 6.85645819e-01 1.76841184e-01
6.69405162e-01 3.69299918e-01 2.31648907e-01 3.77964973e-01
3.04968059e-02 9.22750235e-02 9.59328711e-01 -6.92298830e-01
-2.38248810e-01 9.02807534e-01 -2.50658661e-01 -1.16695559e+00
-6.92167342e-01 -2.10533664e-01 -9.81015980e-01 5.38651228e-01
4.22913015e-01 -2.63363957e-01 -9.37074840e-01 1.31852686e+00
3.58700454e-01 1.67324558e-01 1.23680435e-01 6.82382703e-01
3.24461013e-01 9.28139210e-01 -2.68302113e-01 -3.57740968e-01
8.45089614e-01 -2.46091306e-01 -8.75544190e-01 5.12544334e-01
5.37289441e-01 -5.86384058e-01 5.85441530e-01 7.32581854e-01
-5.88795543e-01 -8.01409185e-01 -1.02975667e+00 5.53849638e-01
-8.86543214e-01 -6.28859699e-02 1.58009037e-01 5.75560570e-01
-1.08519816e+00 1.10827613e+00 -9.89501536e-01 -5.39207041e-01
-4.74113561e-02 5.82529068e-01 -4.42650139e-01 3.91566336e-01
-1.04965186e+00 6.90574229e-01 1.25048900e+00 -1.07056074e-01
-2.01208889e-01 -4.58867699e-01 -6.39644027e-01 2.96504557e-01
-3.16137820e-02 -7.91233927e-02 5.68929672e-01 -1.17241204e+00
-1.39224374e+00 1.24654099e-01 3.55666608e-01 -8.62025499e-01
4.35432136e-01 5.06608635e-02 -6.72469020e-01 2.29463249e-01
-2.95072287e-01 3.80142689e-01 9.27452207e-01 -9.81493413e-01
-3.25414926e-01 -2.28043064e-01 -4.11841154e-01 -3.16849977e-01
-8.10689092e-01 -1.79709926e-01 -3.09012039e-03 -7.80297279e-01
-7.85991624e-02 -9.52025652e-01 -3.89537126e-01 -7.39113986e-01
-3.33235897e-02 -3.64230663e-01 1.09522486e+00 -5.75852215e-01
1.20430493e+00 -2.37437034e+00 2.31621206e-01 7.54590571e-01
4.06176150e-02 1.92328304e-01 3.67466003e-01 6.42249227e-01
-6.68978393e-01 -1.88634351e-01 -3.23263317e-01 -3.03261392e-02
4.73907404e-02 3.20052266e-01 3.74984145e-02 6.48463011e-01
4.41584438e-02 1.63923740e-01 -5.02801776e-01 -2.87409365e-01
8.13485205e-01 8.71833384e-01 -3.44470978e-01 7.54454136e-02
-6.72513247e-02 3.75527918e-01 -1.25048891e-01 -8.25469419e-02
4.96159732e-01 3.87928821e-02 -1.18438313e-02 -3.75755847e-01
-3.08553100e-01 -3.84440154e-01 -1.52557230e+00 1.26684570e+00
-1.64852872e-01 6.29198015e-01 -4.70464528e-01 -1.42682076e+00
1.19227970e+00 3.93236011e-01 1.19376576e+00 -5.52941501e-01
6.46280825e-01 1.46642849e-01 2.25987762e-01 -3.10674697e-01
2.80942082e-01 -4.01157513e-02 5.76163195e-02 4.62843746e-01
2.70498157e-01 -2.31874269e-03 3.60794306e-01 -2.72928566e-01
8.88650060e-01 -2.49858484e-01 2.33088017e-01 -5.25995851e-01
6.67304337e-01 6.04782104e-02 2.93991894e-01 1.15638383e-01
9.80036780e-02 4.09815490e-01 1.05194561e-01 -8.13572884e-01
-1.19384265e+00 -6.91013873e-01 -1.10170059e-01 4.85962719e-01
-4.21280086e-01 -3.46367866e-01 -1.12918270e+00 -3.50192815e-01
-1.42392904e-01 6.46116376e-01 -7.75751233e-01 -3.88372928e-01
-6.72182262e-01 -8.63489032e-01 2.94034660e-01 4.73660916e-01
1.28358394e-01 -1.11414456e+00 -9.94424045e-01 3.64983529e-01
7.95119181e-02 -8.82115722e-01 5.34822680e-02 6.50232017e-01
-1.11674368e+00 -8.22423935e-01 -3.91511947e-01 -4.37699705e-01
4.98094112e-01 -3.09777111e-01 6.94302261e-01 2.75842138e-02
-3.13523233e-01 1.94025591e-01 -7.46483088e-01 -4.33043033e-01
-6.44967020e-01 -4.17946316e-02 3.06455195e-01 2.74523228e-01
5.61367333e-01 -8.85581255e-01 -4.47064310e-01 3.57575491e-02
-1.25873411e+00 -7.66730428e-01 3.32750797e-01 4.60192114e-01
5.06927550e-01 9.44598377e-01 3.53954762e-01 -1.82480097e-01
8.41258347e-01 -4.42018002e-01 -8.60408187e-01 -1.88453063e-01
-8.19245577e-01 1.95088476e-01 1.30627978e+00 -4.58270311e-01
-4.08616871e-01 4.51139599e-01 -1.51882857e-01 -6.69258535e-01
-6.70498550e-01 5.31546593e-01 9.83736888e-02 -7.38540441e-02
8.20714116e-01 1.59158155e-01 1.95818599e-02 -4.79006529e-01
1.33873641e-01 4.28175181e-01 3.25518787e-01 -1.84300572e-01
7.09812462e-01 5.57169437e-01 6.92308843e-02 -9.95457292e-01
1.30544171e-01 -4.40482289e-01 -9.16560054e-01 -4.62538481e-01
1.08973122e+00 -4.69839215e-01 -7.30342269e-01 2.05233485e-01
-8.45915556e-01 -1.89098254e-01 -6.52520418e-01 8.24924350e-01
-5.25940418e-01 4.07201529e-01 -3.49209547e-01 -5.89314580e-01
-2.02789247e-01 -9.95773971e-01 5.20368934e-01 1.75393134e-01
-1.97116643e-01 -1.09732115e+00 2.86436409e-01 -2.71460682e-01
4.48008180e-01 7.73214698e-01 8.10236216e-01 -1.06401455e+00
8.41014087e-02 -5.71287513e-01 3.79973024e-01 6.04236543e-01
6.36677220e-02 2.85738289e-01 -7.46035516e-01 -3.19911122e-01
3.86085451e-01 2.53456086e-01 5.96574247e-01 5.61674893e-01
1.24732888e+00 -3.83918956e-02 -2.33472176e-02 3.99302900e-01
1.57816637e+00 5.78357041e-01 5.66718221e-01 4.14842695e-01
4.22380239e-01 5.76125026e-01 3.03324968e-01 7.89105773e-01
7.14776898e-03 5.29609025e-01 4.24058586e-01 1.26434509e-02
3.40817094e-01 4.26690608e-01 1.74233407e-01 1.25196099e+00
-3.87995899e-01 -2.29953602e-02 -1.16423881e+00 6.27317846e-01
-1.83492148e+00 -1.07340741e+00 -4.78155971e-01 2.04456592e+00
3.08323681e-01 1.24972701e-01 4.40262765e-01 1.06267452e+00
6.96696281e-01 -1.85478181e-01 -1.37731254e-01 -6.86671376e-01
2.43424997e-02 4.84458625e-01 3.11231792e-01 2.46953905e-01
-1.18715405e+00 3.28996450e-01 6.25818443e+00 5.98219991e-01
-1.39142454e+00 -3.06266025e-02 8.89699087e-02 1.09380111e-01
2.04813153e-01 -2.96114028e-01 -6.04977794e-02 7.74339914e-01
1.38469136e+00 -3.55849355e-01 5.32105565e-01 7.38136470e-01
2.38492414e-01 8.19489136e-02 -8.80430996e-01 1.07559550e+00
2.12304592e-01 -8.76036823e-01 -2.33908966e-01 9.02484544e-03
5.71519315e-01 -3.16890553e-02 -2.61767775e-01 1.10131949e-01
2.06408694e-01 -9.89293098e-01 4.46839362e-01 7.45205998e-01
-1.20613866e-01 -1.11864436e+00 8.56791854e-01 1.35466680e-01
-1.12857890e+00 -2.85668761e-01 -1.74941316e-01 -1.17418855e-01
1.03751272e-01 7.50833035e-01 -8.14266086e-01 6.79552317e-01
1.09856915e+00 5.06193161e-01 -5.09264946e-01 1.26646245e+00
3.52962524e-01 5.17728031e-01 -8.04073691e-01 -8.38229656e-02
3.13071668e-01 -3.30027223e-01 4.77771431e-01 1.13713813e+00
6.69888794e-01 -1.82587221e-01 1.67792231e-01 6.72976673e-01
4.29385066e-01 3.57503653e-01 -4.02957261e-01 -9.87337008e-02
2.50514179e-01 1.23969519e+00 -1.14613163e+00 -3.09951782e-01
-1.51843384e-01 8.83219361e-01 -1.63753122e-01 2.60733277e-01
-7.89076626e-01 -5.46688497e-01 2.90050924e-01 7.92881660e-03
5.80914021e-01 -4.35269475e-01 1.02948964e-01 -8.55273724e-01
-8.47434402e-02 -6.73007071e-01 5.38554609e-01 -5.64298332e-01
-1.12219429e+00 9.11104023e-01 1.27721980e-01 -1.56656742e+00
-6.84443414e-01 -5.49149752e-01 -5.07275403e-01 4.94018644e-01
-9.01597083e-01 -5.29884756e-01 -4.79749531e-01 1.01873457e+00
2.04666898e-01 -2.82130301e-01 9.34462011e-01 5.01463652e-01
-3.49677861e-01 2.55544353e-02 5.33933818e-01 1.22027270e-01
5.20437837e-01 -1.42414963e+00 -8.13064501e-02 9.52247798e-01
7.39201829e-02 2.91770488e-01 1.06253350e+00 -3.57075185e-01
-1.00212026e+00 -1.03668141e+00 4.90416259e-01 -1.82447750e-02
7.04562843e-01 -2.13576838e-01 -1.06863427e+00 3.26848149e-01
5.12583971e-01 1.06352344e-01 9.83203709e-01 -2.20373869e-01
1.41168581e-02 -3.81021976e-01 -1.16405642e+00 1.34760022e-01
1.04826033e-01 -3.80589753e-01 -8.99115443e-01 2.27285046e-02
3.61158282e-01 2.17363343e-01 -1.39808452e+00 3.01851153e-01
1.79926544e-01 -1.25857711e+00 8.16773772e-01 -3.47627491e-01
7.94338435e-03 -6.03242338e-01 -3.63212824e-02 -1.54525137e+00
-2.99022466e-01 -2.37148523e-01 -8.06204379e-02 1.07796776e+00
-4.68305834e-02 -2.92219430e-01 5.31848788e-01 2.13576689e-01
3.92821617e-02 -1.83710545e-01 -8.83809209e-01 -9.40570951e-01
2.37252116e-02 -4.95289981e-01 7.44253635e-01 1.37553179e+00
6.81438297e-02 -9.52577144e-02 1.06121581e-02 1.50155768e-01
5.69421351e-01 1.84210986e-01 6.01025760e-01 -1.73988032e+00
-1.56754315e-01 -6.58150673e-01 -8.15815151e-01 2.15117499e-01
4.85561714e-02 -6.40416265e-01 -9.14326012e-02 -1.21763682e+00
-4.69343275e-01 -8.83838683e-02 -6.24991059e-01 2.54474133e-01
2.78641164e-01 1.95326120e-01 9.30925608e-02 -7.61908665e-02
-4.41810489e-01 5.36193371e-01 3.34491700e-01 7.62248561e-02
-4.82975036e-01 -1.66418552e-01 6.82718977e-02 6.82832658e-01
1.06510293e+00 -5.06036699e-01 -3.92800301e-01 -5.47972322e-02
1.34019718e-01 -1.47748336e-01 2.90050477e-01 -1.65238237e+00
3.74110609e-01 2.78106540e-01 7.43833363e-01 -6.88009024e-01
8.14551562e-02 -1.57016826e+00 5.79094708e-01 5.40059090e-01
-4.77377102e-02 3.80216241e-01 4.29636359e-01 3.99967402e-01
-5.21797240e-01 -1.15693681e-01 7.32239246e-01 9.22602713e-02
-7.09719539e-01 -4.65109237e-02 -7.06190825e-01 -5.20271361e-01
1.29175997e+00 -4.25293952e-01 3.53016347e-01 -3.48784328e-01
-1.11650980e+00 -1.11999944e-01 3.91566664e-01 3.65030825e-01
3.66631478e-01 -1.44805455e+00 -3.93489301e-01 2.11165994e-01
-1.70451716e-01 -2.37402484e-01 2.78461188e-01 9.39036787e-01
-5.67114115e-01 1.31274819e-01 -7.37934589e-01 -7.01370955e-01
-1.13212061e+00 1.06441915e+00 4.90231425e-01 6.87541366e-02
-6.92313731e-01 -4.86793481e-02 -6.09714985e-01 -3.66763502e-01
1.81759864e-01 -6.17818475e-01 -6.51069105e-01 4.85410571e-01
3.60800713e-01 6.05608523e-01 4.30001706e-01 -6.89034581e-01
-3.95794600e-01 5.57305753e-01 3.94276381e-01 1.15045086e-01
1.87500536e+00 3.69496010e-02 -2.10023224e-01 6.50098145e-01
1.22316718e+00 -2.14881629e-01 -9.32856977e-01 2.36322701e-01
3.46025854e-01 -1.72395602e-01 2.04103053e-01 -3.37785512e-01
-1.08538520e+00 8.01876247e-01 1.04095340e+00 9.55918074e-01
1.53380334e+00 -4.01136696e-01 5.35181046e-01 3.39860409e-01
8.24058950e-02 -1.54029882e+00 -8.53187963e-02 3.62392694e-01
6.46455169e-01 -9.59902465e-01 -1.29781112e-01 1.71419784e-01
-2.18690947e-01 1.58942282e+00 9.89099294e-02 -5.83676696e-01
1.01240396e+00 5.33017337e-01 6.29077330e-02 -1.17428638e-01
-2.31382146e-01 -4.45524722e-01 3.66972655e-01 4.57359046e-01
4.10254538e-01 8.59824847e-03 -4.21248734e-01 5.95030904e-01
-2.70536870e-01 9.84402895e-02 4.59587634e-01 6.54388189e-01
-2.80880541e-01 -1.09219182e+00 -6.79588377e-01 2.70899534e-01
-3.86562943e-01 5.09585619e-01 -4.32261765e-01 9.09758985e-01
3.33050847e-01 9.46216702e-01 2.23664090e-01 -6.48308218e-01
4.59191412e-01 3.51721138e-01 1.75764859e-01 -1.48525136e-02
-8.55026245e-01 4.12225932e-01 -6.13867402e-01 -5.32988489e-01
-7.80883014e-01 -7.86189020e-01 -1.51702857e+00 -3.42406958e-01
-2.83325613e-01 3.17182809e-01 9.72169816e-01 9.17561412e-01
3.13923866e-01 8.28476548e-01 7.44238257e-01 -8.72517526e-01
-1.13472506e-01 -8.48556757e-01 -6.05195761e-01 8.07626605e-01
2.69754767e-01 -5.03766537e-01 -3.94566774e-01 1.76074773e-01]
|
[7.247866630554199, 3.2769339084625244]
|
18a911da-c52d-4a2a-a959-f0fe42908e9b
|
one-size-fits-all-hypernetwork-for-tunable
|
2206.05970
| null |
https://arxiv.org/abs/2206.05970v3
|
https://arxiv.org/pdf/2206.05970v3.pdf
|
Hypernetwork-Based Adaptive Image Restoration
|
Adaptive image restoration models can restore images with different degradation levels at inference time without the need to retrain the model. We present an approach that is highly accurate and allows a significant reduction in the number of parameters. In contrast to existing methods, our approach can restore images using a single fixed-size model, regardless of the number of degradation levels. On popular datasets, our approach yields state-of-the-art results in terms of size and accuracy for a variety of image restoration tasks, including denoising, deJPEG, and super-resolution.
|
['Gil Ben-Artzi', 'Shai Aharon']
|
2022-06-13
| null | null | null | null |
['color-image-denoising', 'jpeg-artifact-correction', 'jpeg-artifact-removal']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 5.25588453e-01 -5.06068170e-01 3.86919707e-01 -2.32001230e-01
-1.02090251e+00 -2.12438941e-01 4.07466471e-01 -2.85693705e-01
-2.03525782e-01 6.50000870e-01 4.68992233e-01 -7.26372376e-02
-2.03386098e-01 -5.30070186e-01 -6.73626900e-01 -9.29563463e-01
1.63004503e-01 -6.37572408e-02 3.15301478e-01 -4.10089344e-01
4.76238728e-01 4.62418854e-01 -1.64946079e+00 4.33847368e-01
1.04172921e+00 8.65813792e-01 4.86079425e-01 9.12169337e-01
4.25516546e-01 7.52315462e-01 -4.72575516e-01 -9.35767740e-02
3.97081167e-01 -4.76595372e-01 -6.64754450e-01 5.80679417e-01
9.21074688e-01 -6.19971275e-01 -6.86712265e-01 1.14958036e+00
8.71405303e-01 3.14129204e-01 2.34200120e-01 -5.47984540e-01
-1.01020539e+00 1.68091744e-01 -6.48876309e-01 6.46395385e-01
3.41826677e-01 1.59836665e-01 3.88025343e-01 -9.29232955e-01
6.67149127e-01 1.44363976e+00 8.45539927e-01 3.64216596e-01
-1.78922105e+00 -4.11657184e-01 7.39916787e-02 6.82524383e-01
-1.15355384e+00 -8.86806488e-01 4.23646122e-01 -1.09077990e-01
9.84496593e-01 1.94446251e-01 3.77031773e-01 8.04976404e-01
4.47832674e-01 2.20725134e-01 1.54658604e+00 -4.59785521e-01
2.50074655e-01 -5.94142497e-01 5.73850535e-02 1.51880577e-01
-8.99593234e-02 2.01619297e-01 -7.51783907e-01 -6.30090982e-02
9.91997600e-01 -1.64606184e-01 -5.21049082e-01 1.69510692e-01
-1.03577960e+00 3.95366877e-01 3.21312159e-01 1.94212347e-01
-6.81823552e-01 1.88705072e-01 2.33315870e-01 4.93253142e-01
9.15630221e-01 1.24452397e-01 -4.19570714e-01 1.93964317e-01
-1.10575747e+00 1.92392319e-01 2.59483010e-01 4.78448391e-01
6.42290413e-01 2.67007142e-01 -3.43316346e-01 1.10798180e+00
-8.20645764e-02 4.10342783e-01 2.62379795e-01 -1.69348657e+00
6.81062788e-02 -8.20602700e-02 2.10470274e-01 -8.44483256e-01
-1.98115215e-01 -4.84572381e-01 -1.18874824e+00 7.09097683e-01
2.03127298e-03 3.00377458e-01 -1.31663835e+00 1.52031541e+00
2.32255593e-01 6.05839193e-01 -1.89208854e-02 8.92076313e-01
6.92307532e-01 7.17699587e-01 -1.40624166e-01 -4.77019250e-01
1.24981511e+00 -9.91277754e-01 -1.01555145e+00 -5.37559628e-01
-2.76583731e-01 -1.09228933e+00 1.09937835e+00 6.80359483e-01
-1.39251196e+00 -7.17104375e-01 -8.61979544e-01 -5.03540635e-01
2.26818115e-01 -1.85865030e-01 4.91581380e-01 4.31308538e-01
-1.63531578e+00 9.13549125e-01 -6.84393048e-01 -2.89757401e-01
5.07983744e-01 1.16585426e-01 -4.63064134e-01 -6.70757234e-01
-6.26751959e-01 1.15404046e+00 9.91078094e-02 2.57545799e-01
-1.10426426e+00 -9.24282193e-01 -5.86536586e-01 3.96092348e-02
1.39311654e-02 -8.94952774e-01 8.97871971e-01 -7.02672124e-01
-1.36824644e+00 7.28266060e-01 -5.28010905e-01 -5.61422765e-01
5.11114240e-01 -3.34023863e-01 -4.73969370e-01 2.61099815e-01
-3.74825895e-02 5.16355574e-01 1.18085372e+00 -1.57708418e+00
-3.82673055e-01 -3.43341261e-01 -8.05465952e-02 2.96900749e-01
-9.00094137e-02 2.54130900e-01 -5.41304171e-01 -7.52690434e-01
4.26477760e-01 -7.28692055e-01 -4.03534085e-01 2.41698712e-01
-2.33330086e-01 5.09796083e-01 8.06662440e-01 -1.23753715e+00
9.12283003e-01 -2.25110006e+00 4.05775279e-01 -3.06073636e-01
1.80636242e-01 1.52332813e-01 -3.82479399e-01 2.18293056e-01
-1.90711230e-01 8.42227563e-02 -6.07654512e-01 -6.79875791e-01
-3.95875990e-01 4.55827236e-01 -2.70140827e-01 6.36308372e-01
-2.82223880e-01 4.81573522e-01 -5.38719177e-01 -3.30155760e-01
6.22700274e-01 9.20063496e-01 -4.68647480e-01 1.24486409e-01
2.79645294e-01 6.97731972e-01 2.73879737e-01 5.40607393e-01
1.07478178e+00 -2.15705290e-01 1.34093529e-02 -4.49430376e-01
3.92553583e-02 1.11866184e-01 -1.31433463e+00 1.64610565e+00
-4.97346669e-01 7.52994478e-01 4.05677348e-01 -7.20173001e-01
5.50443590e-01 3.48472118e-01 3.30942899e-01 -8.56946945e-01
-3.79757553e-01 9.12113264e-02 -3.80597740e-01 -3.91155094e-01
4.84661728e-01 -1.82554349e-01 5.35904765e-01 1.57980278e-01
-4.46414314e-02 -1.37906328e-01 2.99808294e-01 -3.03684901e-02
1.21549416e+00 -1.12899773e-01 2.24918604e-01 -3.48962098e-01
4.02847707e-01 -2.03795671e-01 5.20558536e-01 8.28736186e-01
-4.44306396e-02 9.53866601e-01 -1.23405354e-02 -4.27588284e-01
-1.21269858e+00 -1.24064815e+00 -2.59273350e-01 9.02489781e-01
2.27585748e-01 -8.83083865e-02 -7.53829479e-01 2.22364932e-01
-2.53783792e-01 6.54029310e-01 -4.36870575e-01 2.07709540e-02
-6.70747757e-01 -1.19308126e+00 1.42978743e-01 1.79808468e-01
7.80111730e-01 -9.97322619e-01 -3.09167653e-01 1.13489591e-01
-6.63053095e-01 -1.40883279e+00 -5.32716036e-01 -2.35588863e-01
-1.43804264e+00 -8.88944447e-01 -5.80599487e-01 -7.20647693e-01
8.06702793e-01 7.20417321e-01 1.29882681e+00 3.88030112e-01
-3.01414520e-01 3.14583927e-01 -1.14158615e-01 1.80716962e-01
-4.25343126e-01 -4.43127543e-01 -1.28003374e-01 -8.89544338e-02
-4.19059128e-01 -1.00576520e+00 -8.89501929e-01 2.04258993e-01
-1.25844753e+00 5.79293678e-03 3.44991833e-01 8.33592832e-01
9.73686218e-01 4.40058231e-01 3.32548559e-01 -5.81531107e-01
8.64673436e-01 -1.16019137e-01 -3.49029034e-01 1.05383731e-01
-9.37456131e-01 -1.05626322e-01 5.58646858e-01 -5.66584945e-01
-1.31254184e+00 -1.49776265e-01 -9.59966257e-02 -2.35795543e-01
-1.39873818e-01 2.68267930e-01 1.26642585e-01 -4.78508204e-01
7.99699843e-01 4.65398967e-01 -1.26169607e-01 -1.03144562e+00
4.54554796e-01 3.06790024e-01 1.14959514e+00 -1.13049336e-01
7.50234425e-01 8.22313666e-01 1.54743806e-01 -8.30095768e-01
-6.69994235e-01 -1.68901533e-01 -6.40617371e-01 -1.75077543e-01
5.49794137e-01 -1.10658324e+00 -3.54904234e-01 9.86424804e-01
-1.01839721e+00 -4.48051393e-01 -3.20247382e-01 2.04350308e-01
-4.58181381e-01 7.57310033e-01 -1.07152581e+00 -3.86478961e-01
-5.30316651e-01 -1.08204818e+00 9.33086574e-01 2.06873506e-01
3.64912421e-01 -8.02621841e-01 -2.00105265e-01 5.19611359e-01
8.51987243e-01 1.20293960e-01 8.46112728e-01 4.48642910e-01
-7.28620231e-01 6.75261067e-03 -4.37257022e-01 6.26645505e-01
2.17763260e-01 -3.74958545e-01 -8.02381933e-01 -6.78129852e-01
3.44188482e-01 7.08574131e-02 1.14952278e+00 9.49316919e-01
1.23897111e+00 -3.56719285e-01 1.25885278e-01 9.65404928e-01
1.67282033e+00 -1.67680368e-01 1.43440270e+00 6.15255952e-01
2.99767077e-01 8.66387710e-02 2.57994980e-01 3.92137825e-01
2.81629533e-01 6.38475776e-01 6.14110291e-01 -3.70906919e-01
-8.28877866e-01 1.70962647e-01 3.79296303e-01 5.85245013e-01
-3.74313682e-01 -7.16289431e-02 -3.92607957e-01 7.22449183e-01
-1.60774672e+00 -1.06571782e+00 -1.83058783e-01 2.15464640e+00
1.06443882e+00 -3.24016720e-01 -4.40127581e-01 1.42767519e-01
7.31068969e-01 4.21847016e-01 -7.09252298e-01 -1.37571707e-01
-6.05763435e-01 2.97641069e-01 5.92832446e-01 9.01222706e-01
-8.76449287e-01 8.56285214e-01 8.57653141e+00 7.75219798e-01
-7.64238477e-01 2.81681448e-01 7.56159246e-01 -2.14664817e-01
-8.61533880e-02 8.40196982e-02 -4.39444363e-01 4.26566809e-01
9.79954064e-01 -5.83244562e-02 1.21097541e+00 2.73108751e-01
5.18745601e-01 -3.94238472e-01 -6.60933793e-01 9.43284214e-01
2.66115099e-01 -1.47572863e+00 1.73646323e-02 -1.22487359e-01
8.89471710e-01 8.99501517e-02 3.03125475e-02 -2.94713587e-01
2.45646700e-01 -9.75275457e-01 5.65774262e-01 8.41569424e-01
6.82776868e-01 -4.14824903e-01 5.38707674e-01 6.88930452e-02
-7.72135913e-01 -1.14188857e-01 -6.05043948e-01 -3.53602767e-02
6.56620920e-01 8.58532488e-01 4.04256843e-02 4.67362821e-01
1.36397004e+00 6.32381737e-01 -6.91720366e-01 1.22973955e+00
-3.45078290e-01 6.72276676e-01 -2.90347636e-01 1.14846039e+00
-5.10005474e-01 -2.90552795e-01 6.53906703e-01 1.05320501e+00
4.75190610e-01 2.98207551e-01 2.52245646e-02 4.49098080e-01
-1.63923148e-02 -3.48392814e-01 4.28704992e-02 6.75380230e-01
4.31723624e-01 9.04800415e-01 -4.27187026e-01 -3.71850938e-01
-1.12490319e-01 1.35972953e+00 4.80457908e-03 5.94866276e-01
-5.10265410e-01 1.71229213e-01 7.80148089e-01 1.83001831e-01
4.44858968e-01 -4.53030556e-01 -5.14236271e-01 -9.60800290e-01
1.58606768e-01 -1.15784931e+00 3.08685362e-01 -1.26745677e+00
-1.36300302e+00 6.87104821e-01 4.22351882e-02 -1.07907796e+00
-8.50799009e-02 -2.64953375e-01 -2.49010250e-01 1.06484091e+00
-1.91230524e+00 -9.68661606e-01 -6.35719895e-01 7.78456628e-01
7.03514576e-01 2.13365793e-01 7.17239976e-01 5.17549098e-01
-4.10249382e-01 9.72300619e-02 4.02780026e-01 -4.47654665e-01
8.92779231e-01 -9.51972067e-01 5.81712842e-01 1.45249391e+00
-1.88647881e-01 3.39953899e-01 1.28302252e+00 -5.52675664e-01
-1.22205138e+00 -6.68056190e-01 6.73563123e-01 -2.07514763e-02
3.48995745e-01 1.83640569e-01 -1.09224594e+00 5.67407072e-01
4.87455547e-01 3.69277060e-01 2.63591945e-01 -1.03207313e-01
-3.73621523e-01 -3.20550591e-01 -1.41795707e+00 5.73084712e-01
1.12686074e+00 -5.28805017e-01 -1.47196874e-01 5.50899386e-01
6.13582909e-01 -6.78584754e-01 -1.05287111e+00 2.87966043e-01
3.77893448e-01 -1.19933140e+00 1.66850126e+00 -2.51005858e-01
5.55706441e-01 -4.69348282e-01 -4.08704340e-01 -1.36786079e+00
-8.36619198e-01 -4.79703963e-01 -3.06494057e-01 8.98015082e-01
-4.83215824e-02 -7.55967140e-01 1.57350510e-01 5.67502677e-01
-2.46360987e-01 -2.87726074e-01 -1.23505652e+00 -7.90627420e-01
-2.82633275e-01 -9.78620574e-02 4.64151233e-01 5.55245697e-01
-8.65654945e-01 -2.84693502e-02 -8.44408929e-01 5.06086171e-01
1.16548765e+00 3.37720290e-02 4.60241824e-01 -1.05677867e+00
-1.68611199e-01 -1.44767791e-01 -1.77049279e-01 -1.04440916e+00
-1.01997338e-01 -4.32115048e-01 1.42730713e-01 -2.06284547e+00
3.62655401e-01 -2.50622779e-02 -3.78767461e-01 6.36110961e-01
-3.70174885e-01 7.95519292e-01 2.03807965e-01 5.99718750e-01
-2.91576594e-01 4.28250819e-01 1.27206147e+00 -6.35662302e-02
-1.25948131e-01 -3.45698327e-01 -9.53875542e-01 5.12929738e-01
6.43409431e-01 -5.68825424e-01 -3.27522069e-01 -9.47707355e-01
-1.38730139e-01 6.69499189e-02 7.66493618e-01 -9.75770533e-01
1.55905887e-01 -2.56556511e-01 5.58200300e-01 -3.83811265e-01
6.76009476e-01 -5.34639239e-01 5.60481012e-01 2.06179738e-01
-1.39136299e-01 1.80115491e-01 3.95733267e-01 5.94724000e-01
-1.32868543e-01 -1.21713810e-01 1.21979606e+00 -2.09657982e-01
-8.07477653e-01 1.03577673e-01 -4.27217066e-01 -3.34524155e-01
4.36542213e-01 -3.64401281e-01 -7.46467471e-01 -5.05869806e-01
-8.23096156e-01 -1.24610677e-01 7.63027310e-01 3.59878868e-01
7.42666245e-01 -1.06308484e+00 -1.18763375e+00 -7.05363080e-02
-4.82662529e-01 -2.55815566e-01 9.98168290e-01 7.63543546e-01
-5.31490386e-01 -3.30439359e-01 -3.45937282e-01 -5.55899203e-01
-1.49793792e+00 5.75544238e-01 5.52056193e-01 -3.56546968e-01
-1.21363294e+00 5.39803922e-01 -1.15670137e-01 -9.50530916e-02
-1.28810406e-01 1.66516036e-01 -6.39169142e-02 -1.96008444e-01
9.85942960e-01 6.86441958e-01 2.42945060e-01 -7.11230338e-01
-1.68836787e-01 6.17766678e-01 -5.40618710e-02 -7.44014606e-02
1.67908752e+00 -7.58842587e-01 -5.59253335e-01 -1.75146312e-02
6.35898769e-01 -2.51568496e-01 -1.66761887e+00 -3.16495836e-01
-4.69478786e-01 -1.04022443e+00 6.66775405e-01 -1.18335187e+00
-1.26735973e+00 4.33205754e-01 1.15738201e+00 -7.76600242e-02
1.94101024e+00 -2.16877535e-01 8.04224849e-01 6.80908486e-02
3.46995503e-01 -9.53463376e-01 -7.02438205e-02 2.23585829e-01
1.29099941e+00 -1.09184921e+00 6.05460167e-01 -5.71555138e-01
-3.67833823e-01 8.33793640e-01 1.49124339e-01 -2.71164119e-01
5.98131776e-01 4.15237665e-01 2.27183923e-01 1.97886396e-02
-6.93367124e-01 2.42989305e-02 1.95909888e-01 9.42197978e-01
1.62025928e-01 -1.33565560e-01 -3.11923653e-01 -3.87787744e-02
4.75051589e-02 1.26671791e-02 7.86408961e-01 7.23593056e-01
-5.77793539e-01 -1.06233692e+00 -7.35387444e-01 2.22897053e-01
-5.19716144e-01 -5.20180047e-01 2.13038713e-01 2.17517719e-01
-1.43203780e-01 1.39356887e+00 -1.84427872e-01 -4.09906171e-02
4.50456351e-01 -3.21850985e-01 7.61317253e-01 -1.77679673e-01
-5.08721530e-01 1.67820558e-01 -9.77498814e-02 -7.57257581e-01
-6.79539621e-01 -6.60882473e-01 -7.52498806e-01 -6.52134597e-01
-1.66070864e-01 -4.55288023e-01 4.41868871e-01 8.75260651e-01
6.05327606e-01 6.59928739e-01 5.51504493e-01 -1.32556331e+00
-4.28100824e-01 -1.15434861e+00 -6.03226006e-01 5.98626614e-01
5.76426327e-01 -3.75653952e-01 -4.88627225e-01 4.79948997e-01]
|
[11.156813621520996, -2.2280805110931396]
|
21c1efba-7804-48d0-b4dc-51ccc8ff7265
|
patient-contrastive-learning-a-performant
|
2104.04569
| null |
https://arxiv.org/abs/2104.04569v1
|
https://arxiv.org/pdf/2104.04569v1.pdf
|
Patient Contrastive Learning: a Performant, Expressive, and Practical Approach to ECG Modeling
|
Supervised machine learning applications in health care are often limited due to a scarcity of labeled training data. To mitigate this effect of small sample size, we introduce a pre-training approach, Patient Contrastive Learning of Representations (PCLR), which creates latent representations of ECGs from a large number of unlabeled examples. The resulting representations are expressive, performant, and practical across a wide spectrum of clinical tasks. We develop PCLR using a large health care system with over 3.2 million 12-lead ECGs, and demonstrate substantial improvements across multiple new tasks when there are fewer than 5,000 labels. We release our model to extract ECG representations at https://github.com/broadinstitute/ml4h/tree/master/model_zoo/PCLR.
|
['Puneet Batra', 'Collin Stultz', 'Aaron Aguirre', 'Steven Song', 'Erik Reinertsen', 'Nathaniel Diamant']
|
2021-04-09
| null | null | null | null |
['electrocardiography-ecg', 'unsupervised-pre-training']
|
['methodology', 'methodology']
|
[ 4.81579214e-01 2.02459097e-01 -5.32436728e-01 -5.58401465e-01
-1.52862239e+00 -4.18151855e-01 -7.08299596e-03 4.46285546e-01
-1.02310263e-01 9.91175354e-01 4.86880153e-01 -5.40392280e-01
-1.18972279e-01 -3.68012309e-01 -4.06552553e-01 -3.05612624e-01
-1.90025315e-01 5.85265279e-01 -5.87432325e-01 4.38942134e-01
-2.48531371e-01 2.37052485e-01 -6.20855331e-01 5.73575079e-01
5.50246716e-01 6.03567719e-01 -2.98961908e-01 8.35310578e-01
6.35644913e-01 9.75766957e-01 -5.43371260e-01 -2.05071680e-02
1.07742943e-01 -6.19210899e-01 -8.90229225e-01 1.00373529e-01
5.75207360e-02 -4.06285614e-01 -3.61334950e-01 6.11876190e-01
1.02666724e+00 -2.26262286e-01 7.97869086e-01 -9.74892139e-01
-6.63343549e-01 6.09231353e-01 -4.86678660e-01 5.33172607e-01
1.40328214e-01 1.93181351e-01 9.03170466e-01 -8.38241577e-01
6.48839653e-01 8.14757705e-01 8.89046490e-01 8.79795730e-01
-1.52993631e+00 -8.75968099e-01 -2.35199615e-01 -3.45453799e-01
-1.38518178e+00 -6.42155766e-01 4.10078049e-01 -5.87874770e-01
7.46103764e-01 2.72965491e-01 4.64483112e-01 1.25138378e+00
7.47861564e-02 5.79013169e-01 1.08095324e+00 -2.29224682e-01
1.09651111e-01 -1.53970674e-01 4.68119204e-01 5.65838695e-01
2.58541584e-01 -1.22048475e-01 -3.05739731e-01 -9.22274411e-01
1.00391388e+00 5.18439829e-01 -4.90820885e-01 2.97807157e-02
-1.27721918e+00 8.86395335e-01 1.42746717e-01 4.02760804e-02
-4.74361479e-01 2.18672439e-01 5.51608741e-01 2.88938135e-01
3.70477498e-01 6.93388939e-01 -6.57905579e-01 -8.98597687e-02
-8.38808835e-01 -7.70939738e-02 4.85020310e-01 1.04794776e+00
2.80796975e-01 7.44211599e-02 -3.55465412e-01 1.03932214e+00
1.94465995e-01 1.90913111e-01 6.60102725e-01 -1.15267849e+00
1.74298570e-01 3.80705059e-01 -4.74429615e-02 -5.66283107e-01
-4.99614537e-01 -6.70449913e-01 -1.12311816e+00 -2.38973141e-01
1.37610003e-01 -4.84860986e-01 -1.04425609e+00 1.69044566e+00
-5.81133850e-02 2.00155824e-01 2.77353525e-01 6.34360492e-01
1.05924797e+00 3.47947687e-01 4.19721663e-01 -3.14807028e-01
1.36547244e+00 -7.37292647e-01 -5.30693233e-01 -2.92929173e-01
9.59009349e-01 -4.34996217e-01 8.48482907e-01 5.14930844e-01
-1.07634163e+00 -3.62334967e-01 -7.87845433e-01 -2.56727543e-02
1.81495100e-01 2.29601026e-01 6.46677852e-01 3.55971545e-01
-9.59582150e-01 7.12556720e-01 -9.72402394e-01 -1.89678773e-01
9.85376477e-01 3.34613800e-01 -4.19526488e-01 -3.12692761e-01
-1.00515330e+00 4.67415273e-01 2.02062249e-01 -2.26267427e-01
-1.08218896e+00 -8.33619475e-01 -8.04237843e-01 3.58705234e-04
1.15630463e-01 -8.92945886e-01 1.24945486e+00 -6.35660529e-01
-7.84061670e-01 1.03811109e+00 -3.75598781e-02 -4.45286334e-01
3.72404784e-01 -3.24779838e-01 -4.42357361e-01 3.93815547e-01
2.05227524e-01 6.79413438e-01 6.16706192e-01 -1.05469882e+00
-1.68801978e-01 -2.21818119e-01 -4.00262833e-01 7.72830620e-02
-1.29817411e-01 -4.51506861e-02 -1.85334042e-01 -8.86938870e-01
8.91128033e-02 -9.29904163e-01 -7.83934951e-01 -7.18148500e-02
-4.44724321e-01 5.01087829e-02 1.09288253e-01 -6.71010256e-01
1.10015726e+00 -2.35105824e+00 -2.65112400e-01 4.52683158e-02
7.62592316e-01 5.71504645e-02 -2.08292291e-01 3.53686154e-01
-7.10527301e-01 3.76440734e-01 -3.17686230e-01 -1.65950522e-01
-4.85929489e-01 1.51712194e-01 -2.05462456e-01 3.98281336e-01
3.65828961e-01 1.06824625e+00 -1.05537736e+00 -5.91284335e-01
-2.13793442e-02 4.33726311e-01 -5.54413080e-01 2.96602309e-01
1.47588104e-01 8.53862584e-01 -7.23753572e-01 7.73944855e-01
1.11737683e-01 -1.03785467e+00 5.23566008e-01 1.32399142e-01
4.95551050e-01 4.55494314e-01 -7.69616604e-01 1.92140329e+00
-1.60457753e-02 1.48942947e-01 -4.11453485e-01 -9.37069714e-01
7.80719101e-01 7.75769770e-01 7.77457416e-01 -7.28206038e-02
1.38153076e-01 2.16144875e-01 8.95448178e-02 -3.19732934e-01
-2.34187454e-01 -3.45910400e-01 -1.72573447e-01 5.24731815e-01
4.49841209e-02 1.37584433e-01 -1.08295478e-01 3.64589989e-01
1.60981297e+00 -1.10381678e-01 7.74659216e-01 -5.35119586e-02
-2.55606771e-01 3.63174751e-02 1.06029606e+00 8.49607587e-01
-3.65481973e-01 1.14314497e+00 4.67988908e-01 -5.10523200e-01
-7.03553081e-01 -1.09614873e+00 -6.06841862e-01 7.34727323e-01
-5.79301834e-01 -6.28434837e-01 -2.33704537e-01 -7.65478015e-01
1.17520832e-01 4.63377059e-01 -6.21648669e-01 -2.34267190e-01
-4.12122905e-01 -1.15270185e+00 9.19069648e-01 9.05282378e-01
-2.35582680e-01 -1.25671530e+00 -7.23688543e-01 4.40612793e-01
-2.05626875e-01 -8.39121163e-01 -4.04922098e-01 3.60798955e-01
-1.31140208e+00 -1.30751431e+00 -9.25099969e-01 -6.40615940e-01
8.53349864e-01 -3.14467758e-01 1.41788578e+00 2.98510492e-01
-8.71953368e-01 3.77155006e-01 -3.54956686e-01 -6.17025375e-01
-5.89211047e-01 -5.00581274e-03 4.66720350e-02 -2.94738919e-01
2.79675961e-01 -5.49419701e-01 -8.68707061e-01 -8.28023627e-02
-5.23872733e-01 2.77687758e-02 6.35854900e-01 1.19195759e+00
8.82283032e-01 -6.68537974e-01 1.28340745e+00 -1.64072835e+00
6.24564290e-01 -8.47193837e-01 -1.28802896e-01 8.25214610e-02
-8.40925336e-01 -2.12964341e-01 3.92633736e-01 -4.44878429e-01
-4.12156761e-01 1.12029932e-01 -2.12672800e-01 -6.04323745e-01
-4.39027697e-01 5.80101550e-01 4.65601057e-01 5.59781849e-01
9.69421268e-01 3.45726907e-02 3.21707651e-02 -4.54691708e-01
1.13039039e-01 9.43544745e-01 4.01547909e-01 -4.83040333e-01
3.80803466e-01 2.15970248e-01 -1.23510718e-01 -4.28723425e-01
-8.73862088e-01 -6.53862357e-01 -3.21349442e-01 1.93370059e-01
6.72896922e-01 -1.17866814e+00 -3.99506241e-01 -4.59306836e-02
-7.24481940e-01 -6.39735579e-01 -7.95477152e-01 7.64690757e-01
-5.45218885e-01 4.63837534e-01 -1.06458092e+00 -5.12516141e-01
-9.75071907e-01 -1.02380705e+00 9.39026713e-01 -2.38446280e-01
-8.49647522e-01 -8.14457893e-01 2.79483378e-01 2.60841757e-01
9.59869996e-02 4.67974484e-01 1.31434155e+00 -1.04751050e+00
-1.75390631e-01 -1.93817899e-01 -1.20072260e-01 4.66707230e-01
5.84533811e-01 -2.94492096e-01 -9.64569092e-01 -5.30776441e-01
-6.28717691e-02 -7.07155645e-01 7.65765727e-01 5.27575612e-01
1.63127875e+00 -1.36967123e-01 -3.63392472e-01 7.96223581e-01
1.09299946e+00 2.36934498e-01 5.80689549e-01 -2.30315730e-01
5.70727289e-01 1.97205991e-01 5.17939925e-01 7.78252482e-01
1.09743319e-01 1.13811374e-01 -1.45287424e-01 -6.41782761e-01
-4.32893559e-02 1.58908572e-02 -1.47338033e-01 9.62788105e-01
-1.46696895e-01 6.12029806e-02 -1.19921207e+00 5.90363503e-01
-1.71371460e+00 -5.18759012e-01 1.47602946e-01 2.07038450e+00
1.24658704e+00 9.11936313e-02 3.35409269e-02 7.37929791e-02
5.29898107e-01 -1.44772545e-01 -8.48449349e-01 -4.93772067e-02
2.81363726e-01 5.83469391e-01 3.48974556e-01 1.12707660e-01
-1.08512449e+00 5.07176399e-01 7.00698757e+00 5.34214020e-01
-9.66976285e-01 3.15754026e-01 1.31492686e+00 -1.55609772e-01
-8.29187781e-02 -3.14510494e-01 -3.33265007e-01 2.72866040e-01
1.15106916e+00 -1.73973411e-01 1.03169069e-01 9.14987743e-01
1.84771389e-01 4.23519671e-01 -1.32115996e+00 1.22685742e+00
-8.63929391e-02 -1.53549981e+00 -1.06286839e-01 9.21800286e-02
5.91054559e-01 2.51573026e-01 8.40830617e-03 2.20546633e-01
2.93159336e-01 -1.39491010e+00 5.19608334e-03 4.54034984e-01
1.68608701e+00 -3.22033197e-01 6.57835662e-01 1.17960259e-01
-7.92474926e-01 -2.04443792e-03 -2.42986351e-01 1.40921935e-01
2.70380471e-02 5.23069024e-01 -1.10171533e+00 4.38641787e-01
6.70426428e-01 1.03033483e+00 -5.12977481e-01 7.91809559e-01
-1.18120596e-01 1.04844344e+00 1.06284767e-02 6.22967422e-01
-2.78815120e-01 1.54319286e-01 1.83658719e-01 1.26310217e+00
8.85325149e-02 4.17569250e-01 3.05040061e-01 6.88249290e-01
-5.12576222e-01 2.81840473e-01 -7.32258737e-01 -2.37301111e-01
4.98894215e-01 1.50259960e+00 -6.86872661e-01 -6.73825502e-01
-3.62636983e-01 6.25895023e-01 2.87800878e-01 4.24138248e-01
-6.33879125e-01 -1.21208265e-01 1.44577220e-01 2.66383141e-01
-3.54780018e-01 4.15031999e-01 -2.92360812e-01 -1.14064479e+00
-3.48024905e-01 -1.27643394e+00 8.82264912e-01 -7.75498986e-01
-1.71317255e+00 6.91019773e-01 -8.66205171e-02 -1.48036063e+00
-4.82043415e-01 -3.64124656e-01 -2.85138547e-01 9.35084164e-01
-1.36890769e+00 -7.98945546e-01 -1.77740380e-01 4.99767780e-01
5.10639071e-01 -3.47019106e-01 1.64804900e+00 5.49880803e-01
-4.48690712e-01 7.12787986e-01 -2.18833014e-01 5.26668429e-01
9.95227695e-01 -1.32391465e+00 4.30823088e-01 2.18029812e-01
9.00404435e-03 1.00082171e+00 2.00147182e-02 -6.73234820e-01
-9.68383193e-01 -1.31873643e+00 7.37414479e-01 -6.20906353e-01
2.38058954e-01 -1.99106202e-01 -9.75122809e-01 1.21790719e+00
-9.90931466e-02 5.11019111e-01 1.47781539e+00 3.73422086e-01
-3.32965821e-01 1.22518979e-01 -1.02804279e+00 4.11819071e-01
9.35982525e-01 -5.10472953e-01 -6.63544774e-01 5.77783048e-01
4.16740030e-01 -3.59740436e-01 -1.51869905e+00 7.41127491e-01
3.79752338e-01 -2.61170894e-01 9.62912917e-01 -9.73186135e-01
3.16606224e-01 1.35468602e-01 2.88860463e-02 -1.00605607e+00
-5.34128547e-01 -6.67177796e-01 -9.05060396e-02 8.30231965e-01
6.82549417e-01 -9.38599229e-01 7.28978455e-01 5.93812227e-01
-2.15458676e-01 -1.19797587e+00 -5.76019645e-01 -2.81272769e-01
9.86528993e-02 -2.19927862e-01 2.64201164e-01 1.34575677e+00
3.15189779e-01 6.14519835e-01 -4.14419740e-01 -6.70870841e-02
5.03024697e-01 1.68195561e-01 3.90378952e-01 -1.25764465e+00
-7.95995891e-01 1.41980737e-01 -1.62504151e-01 -5.72020173e-01
-2.69414306e-01 -1.20484698e+00 -1.73892424e-01 -1.64058089e+00
6.47759259e-01 -8.32760990e-01 -9.32199180e-01 1.10983682e+00
-4.15498555e-01 5.08109093e-01 -9.27276835e-02 5.68925440e-01
-6.78354323e-01 6.80252090e-02 1.04209971e+00 1.84237752e-02
-1.39908448e-01 1.17227435e-01 -1.02381516e+00 7.39307463e-01
1.01660657e+00 -9.66609478e-01 -4.17640924e-01 -2.32720405e-01
-1.62015110e-01 5.21495700e-01 1.79648921e-01 -7.82182813e-01
-3.81506801e-01 1.80557027e-01 8.00926864e-01 -1.85174301e-01
3.58878791e-01 -3.28116566e-01 2.27782294e-01 7.38999426e-01
-8.48961830e-01 2.40771651e-01 1.41852498e-01 3.79939377e-01
-1.34875979e-02 -1.92177482e-02 7.94340014e-01 -4.32429314e-01
-1.23000242e-01 5.07035375e-01 -3.98440748e-01 5.59466481e-01
6.87841058e-01 1.43468425e-01 -3.57778251e-01 -2.29433730e-01
-1.24857903e+00 2.03140453e-01 1.69950217e-01 1.21358119e-01
7.60298908e-01 -1.16041398e+00 -1.04786789e+00 1.73941895e-01
3.86028290e-01 5.47123775e-02 3.71505886e-01 8.88914764e-01
-5.96866429e-01 1.15877889e-01 -4.04870696e-02 -4.74975169e-01
-1.23150635e+00 6.26759052e-01 3.15410852e-01 -4.24594104e-01
-1.19470847e+00 5.22204697e-01 3.63155752e-01 -2.97494382e-01
2.68097579e-01 -2.35611200e-01 -1.06581286e-01 -2.61805981e-01
4.93119419e-01 2.22671196e-01 -1.04646280e-01 -3.00422728e-01
-3.95764738e-01 1.60403147e-01 -2.87984580e-01 1.56227291e-01
1.44226289e+00 3.83228600e-01 1.75911024e-01 6.02384388e-01
1.08615661e+00 -2.78823376e-01 -1.07862186e+00 -4.55252826e-02
-1.29985094e-01 -1.34294450e-01 -2.65480995e-01 -7.57068992e-01
-9.24083710e-01 9.34308648e-01 9.02431011e-01 -1.81000620e-01
1.08679676e+00 3.44775207e-02 5.63473225e-01 3.39593738e-01
1.77020967e-01 -6.55767381e-01 2.57121861e-01 -9.49761271e-02
6.34420156e-01 -1.17287266e+00 2.52746582e-01 -3.66288185e-01
-9.73734677e-01 7.58796692e-01 1.96220964e-01 -1.98261723e-01
6.79544091e-01 3.20753962e-01 5.74524760e-01 -3.90897870e-01
-9.78755176e-01 1.17990986e-01 6.30829409e-02 5.30743420e-01
8.62301648e-01 3.44882011e-01 -1.28357857e-01 1.00179851e+00
2.52347887e-01 3.93028140e-01 5.32543540e-01 1.09232640e+00
1.26068145e-01 -1.27415717e+00 6.74313307e-02 1.09470820e+00
-1.17746639e+00 -4.68831837e-01 -9.05349106e-03 4.27497387e-01
-1.88790828e-01 8.63141716e-01 -2.23566204e-01 -1.15683332e-01
2.49663264e-01 3.23766053e-01 3.10982466e-01 -1.35896683e+00
-5.04245639e-01 5.97723186e-01 1.43139213e-01 -2.92613089e-01
-2.92128772e-01 -6.40662313e-01 -1.52408302e+00 4.46117789e-01
7.88656697e-02 1.91911638e-01 -6.65558353e-02 4.54780877e-01
7.99908459e-01 8.87713432e-01 5.37388444e-01 -3.06965649e-01
-7.64236987e-01 -1.09800982e+00 -7.35375226e-01 6.94862008e-01
2.73565233e-01 -2.56681055e-01 -8.50972906e-02 4.15870100e-01]
|
[8.024843215942383, 6.51099157333374]
|
be4e516c-ed5f-49c4-bb1d-118a3032c8db
|
laxity-aware-scalable-reinforcement-learning
|
2306.16619
| null |
https://arxiv.org/abs/2306.16619v1
|
https://arxiv.org/pdf/2306.16619v1.pdf
|
Laxity-Aware Scalable Reinforcement Learning for HVAC Control
|
Demand flexibility plays a vital role in maintaining grid balance, reducing peak demand, and saving customers' energy bills. Given their highly shiftable load and significant contribution to a building's energy consumption, Heating, Ventilation, and Air Conditioning (HVAC) systems can provide valuable demand flexibility to the power systems by adjusting their energy consumption in response to electricity price and power system needs. To exploit this flexibility in both operation time and power, it is imperative to accurately model and aggregate the load flexibility of a large population of HVAC systems as well as designing effective control algorithms. In this paper, we tackle the curse of dimensionality issue in modeling and control by utilizing the concept of laxity to quantify the emergency level of each HVAC operation request. We further propose a two-level approach to address energy optimization for a large population of HVAC systems. The lower level involves an aggregator to aggregate HVAC load laxity information and use least-laxity-first (LLF) rule to allocate real-time power for individual HVAC systems based on the controller's total power. Due to the complex and uncertain nature of HVAC systems, we leverage a reinforcement learning (RL)-based controller to schedule the total power based on the aggregated laxity information and electricity price. We evaluate the temperature control and energy cost saving performance of a large-scale group of HVAC systems in both single-zone and multi-zone scenarios, under varying climate and electricity market conditions. The experiment results indicate that proposed approach outperforms the centralized methods in the majority of test scenarios, and performs comparably to model-based method in some scenarios.
|
['Yize Chen', 'Yuxin Pan', 'Ruohong Liu']
|
2023-06-29
| null | null | null | null |
['reinforcement-learning-1']
|
['methodology']
|
[-3.44645113e-01 -7.00545162e-02 -1.09071657e-01 8.76877010e-02
-3.70920926e-01 -1.15849543e+00 3.43879133e-01 3.12479973e-01
3.38526279e-01 1.08364928e+00 2.16565922e-01 -3.16044539e-01
-7.79721320e-01 -1.02915072e+00 -7.40064457e-02 -1.12876284e+00
-1.48373097e-01 7.62939215e-01 -3.51078063e-01 -1.69760048e-01
-1.81607325e-02 4.93932843e-01 -1.42261684e+00 -3.68518978e-01
1.06687295e+00 1.14359415e+00 2.77234435e-01 5.37631094e-01
5.66909492e-01 2.51440525e-01 -5.40764511e-01 7.55887389e-01
4.77037013e-01 -2.94381857e-01 -3.91686797e-01 3.72540727e-02
-6.93859696e-01 -4.29672599e-01 2.78467864e-01 6.58236980e-01
7.55657077e-01 6.86897039e-01 6.41526699e-01 -1.66503334e+00
-2.22463533e-02 4.92442667e-01 -4.14765060e-01 4.14998293e-01
1.46707281e-01 5.08597076e-01 8.57352495e-01 1.12681836e-01
-2.23308414e-01 7.86313772e-01 9.98219326e-02 1.88841283e-01
-1.23393321e+00 -4.83239502e-01 1.91050664e-01 2.89122458e-03
-1.11983407e+00 6.23600557e-02 7.31662810e-01 -3.45775396e-01
1.44145751e+00 6.86342239e-01 1.27643585e+00 -8.08251575e-02
3.12295437e-01 3.70513052e-01 1.37454283e+00 -1.99408993e-01
6.20551348e-01 -2.20470786e-01 -3.30301821e-01 1.12725785e-02
1.00026511e-01 2.27349158e-02 3.67648959e-01 -3.20584089e-01
2.40134954e-01 1.15811611e-02 -3.67903829e-01 -1.85709864e-01
-7.49739110e-01 6.73611641e-01 3.22571695e-01 2.73534298e-01
-6.67037487e-01 -6.89480156e-02 3.28043520e-01 7.67503399e-03
-4.66000661e-02 7.99976945e-01 -7.33200610e-01 1.49516696e-02
-9.13056433e-01 2.69696295e-01 8.70932877e-01 6.93114400e-01
1.93471804e-01 4.88970488e-01 -3.94541085e-01 5.39289415e-01
1.65940568e-01 6.56690419e-01 6.27779245e-01 -1.00067806e+00
1.74961254e-01 5.11198461e-01 5.76367140e-01 -3.11221123e-01
-5.40000677e-01 -3.25792253e-01 -9.43863630e-01 1.64443895e-01
-2.91868329e-01 -6.34687424e-01 -7.29046047e-01 1.57155895e+00
5.43488860e-01 -3.47514540e-01 -7.83000290e-02 8.05558801e-01
-3.70247245e-01 1.05333519e+00 -6.49798254e-04 -1.07061660e+00
1.45787191e+00 -5.78734100e-01 -8.45565081e-01 4.51581419e-01
2.21500039e-01 -7.02427268e-01 4.73700076e-01 6.91487044e-02
-1.43359900e+00 -1.36508763e-01 -1.08812511e+00 6.56632423e-01
-5.21924257e-01 -6.45480901e-02 5.58874719e-02 4.52869922e-01
-8.63296926e-01 3.70145530e-01 -7.15050817e-01 -2.45868906e-01
-9.12141204e-02 4.36968595e-01 4.68404979e-01 5.15461087e-01
-1.33103085e+00 1.08239973e+00 7.46926367e-01 2.64919668e-01
-6.22199357e-01 -9.45492268e-01 -5.83449006e-01 4.03457582e-01
5.18044472e-01 -7.33858407e-01 1.46020615e+00 3.34919170e-02
-1.58727634e+00 -4.32650864e-01 4.56336915e-01 -4.38320071e-01
3.61515552e-01 2.25829497e-01 -3.74027729e-01 -1.42437348e-03
-4.17495966e-01 3.25986631e-02 2.58173645e-01 -1.18979037e+00
-1.04242992e+00 -2.09538832e-01 -2.18312368e-01 7.47321248e-01
-1.81842610e-01 -6.08604193e-01 7.48875141e-01 -2.25386322e-01
-3.58314365e-01 -1.06439936e+00 -2.83786714e-01 -9.87513423e-01
-3.28980118e-01 -7.54963160e-01 1.33775353e+00 -6.74014688e-01
1.56846619e+00 -1.63943934e+00 3.02533448e-01 7.40385473e-01
-5.10701120e-01 1.70646325e-01 6.41195595e-01 1.17334819e+00
-1.83163807e-01 1.49125814e-01 -3.11528713e-01 2.21466348e-01
4.80495691e-01 6.74032211e-01 -3.19673568e-01 1.16735488e-01
-2.40375131e-01 3.70050699e-01 -7.85009503e-01 -2.82084458e-02
8.00844669e-01 3.73298936e-02 -3.18644643e-01 4.69523519e-01
-4.17144626e-01 2.75510520e-01 -6.37276411e-01 5.08447111e-01
5.24807811e-01 -1.32530381e-03 4.34560120e-01 -4.44424331e-01
-3.73726785e-01 -3.42235267e-01 -1.55209172e+00 1.02530348e+00
-8.30294609e-01 -6.04925305e-02 4.52505141e-01 -1.20027030e+00
5.20760834e-01 4.77546304e-01 1.17792773e+00 -7.19419837e-01
-6.82069287e-02 9.21153054e-02 9.19924397e-03 -3.82830203e-01
3.45363200e-01 -2.05675021e-01 -6.07790500e-02 7.29592741e-01
-3.36943388e-01 -7.16807008e-01 4.24761564e-01 -5.75898923e-02
8.04785728e-01 -3.21127206e-01 4.69873220e-01 -8.28362763e-01
8.35473359e-01 -3.10819566e-01 6.40280604e-01 -1.79698497e-01
-2.62654603e-01 -4.57460910e-01 7.70965517e-02 -2.50635952e-01
-1.15582108e+00 -9.51639593e-01 -1.33506447e-01 6.55203521e-01
6.28426746e-02 2.38371521e-01 -5.85084260e-01 6.11592503e-03
2.44695082e-01 1.42786431e+00 -1.51939139e-01 -5.41112013e-02
-6.40831947e-01 -1.06155384e+00 -4.65319097e-01 4.20333505e-01
5.27077734e-01 -5.80608010e-01 -1.25942159e+00 3.20017517e-01
1.15841575e-01 -8.79663169e-01 -6.99876428e-01 3.96926820e-01
-4.08372700e-01 -9.68666136e-01 -2.43937179e-01 -3.75764966e-01
7.21496940e-01 -1.64815560e-01 9.49899137e-01 7.21252486e-02
-4.39245909e-01 5.17446280e-01 1.94549352e-01 -4.47076291e-01
-1.48415787e-03 1.35626733e-01 3.22164863e-01 -4.09067959e-01
-3.19291174e-01 -5.66789687e-01 -1.27384591e+00 5.09372413e-01
-1.00766253e+00 1.71967037e-02 2.73509622e-01 7.82725394e-01
5.99532604e-01 1.24502742e+00 1.03519952e+00 -3.76796186e-01
9.99895632e-01 -5.16157806e-01 -9.80393410e-01 5.13913393e-01
-1.07320392e+00 9.50935185e-02 1.02277637e+00 4.10947688e-02
-1.05304694e+00 1.43932328e-01 3.75757366e-01 6.53049443e-03
2.09041864e-01 1.55139163e-01 -1.42023891e-01 3.51945430e-01
-3.81014258e-01 3.19975704e-01 -3.02684933e-01 -1.64794326e-01
3.47918868e-02 4.27655309e-01 3.95670235e-01 -8.45616698e-01
1.15734220e+00 -2.50910670e-01 4.60514754e-01 -1.48109674e-01
-1.80101112e-01 -3.82583886e-01 -2.01669589e-01 -2.53530174e-01
8.00271094e-01 -8.34043205e-01 -1.58388638e+00 2.61791915e-01
-3.55281562e-01 -4.33168739e-01 -2.34117523e-01 1.84370875e-01
-7.46594012e-01 -2.49311160e-02 -2.69212157e-01 -1.38241804e+00
-7.17612207e-01 -1.16741586e+00 5.61250806e-01 6.79337263e-01
7.01823756e-02 -9.97286797e-01 2.75906563e-01 2.62934238e-01
7.25984871e-01 8.00833762e-01 1.25476611e+00 -1.66669816e-01
-5.63392401e-01 6.79242611e-02 3.40297282e-01 3.95840436e-01
4.05912638e-01 -5.73963746e-02 -5.62851071e-01 -1.06733346e+00
-3.66219804e-02 -2.23282829e-01 -1.44415587e-01 3.46198559e-01
1.32394505e+00 -9.17502940e-01 -3.34172755e-01 5.94900586e-02
2.16305399e+00 8.44348848e-01 1.64955124e-01 1.16269283e-01
1.88594237e-01 4.72041547e-01 6.04918301e-01 1.14964557e+00
5.53690255e-01 6.24005556e-01 6.78916991e-01 -1.89095512e-02
7.32334197e-01 -3.69019173e-02 -4.40496989e-02 8.07325304e-01
-3.96240130e-02 -3.88382226e-01 -6.78051472e-01 5.49736023e-01
-1.69952118e+00 -1.03302431e+00 4.45856333e-01 2.37293291e+00
8.34235311e-01 -1.49796039e-01 2.02745736e-01 3.89028162e-01
4.99219716e-01 2.18257736e-02 -6.97306633e-01 -9.65734243e-01
2.06395835e-01 -1.51576847e-01 7.78679550e-01 3.52551758e-01
-4.98594522e-01 -1.71059608e-01 5.66296148e+00 6.96713686e-01
-9.32696581e-01 -2.59711057e-01 1.02942634e+00 -3.68785411e-01
-2.60728121e-01 -1.02696449e-01 -4.59296793e-01 1.06128240e+00
1.35945380e+00 -8.36244881e-01 1.19840395e+00 5.63876927e-01
9.61336851e-01 -5.50836682e-01 -1.16667747e+00 4.83822703e-01
-6.53944790e-01 -9.45885956e-01 -4.73758966e-01 2.40723625e-01
1.13759017e+00 -1.39358386e-01 -2.90489584e-01 1.96007401e-01
4.67299640e-01 -7.20958114e-01 2.84987897e-01 5.19558787e-01
1.30826265e-01 -1.38847327e+00 8.64504337e-01 5.48811316e-01
-1.62447512e+00 -8.21478844e-01 1.95683911e-01 2.09797937e-02
6.16010785e-01 4.00899678e-01 -7.53287137e-01 7.50681102e-01
7.13460326e-01 -1.84961751e-01 7.38027245e-02 6.70492709e-01
3.24788392e-01 2.56885618e-01 -7.05738068e-01 -5.03408499e-02
1.89881653e-01 -5.08636773e-01 4.93810587e-02 5.03733635e-01
3.93823922e-01 7.46471941e-01 6.17689490e-01 5.81978381e-01
2.12691560e-01 -2.13190675e-01 -3.17846447e-01 3.27540468e-03
9.93377507e-01 1.68152320e+00 -6.53795719e-01 -3.91932845e-01
8.52125362e-02 3.48093122e-01 -5.84195614e-01 4.52736616e-01
-7.93960690e-01 -2.55607873e-01 6.24194205e-01 7.50139728e-02
6.63338080e-02 -2.24187791e-01 -2.28854790e-01 -3.32230598e-01
-1.42981410e-01 -4.78651762e-01 6.33548737e-01 -7.18305230e-01
-1.21386886e+00 2.45903675e-02 4.66294825e-01 -9.38209057e-01
-8.01156580e-01 -1.87507495e-01 -9.88420367e-01 1.11531353e+00
-1.84908617e+00 -7.81362772e-01 -8.78150687e-02 5.57138383e-01
8.03547382e-01 8.24871138e-02 6.70639932e-01 3.00631970e-01
-9.73922849e-01 6.27482608e-02 7.01651096e-01 -5.73465824e-01
-9.54334065e-02 -1.61256182e+00 -4.43439811e-01 6.82288349e-01
-1.10432911e+00 1.31971136e-01 9.04740155e-01 -2.55859792e-01
-2.02944446e+00 -7.93836534e-01 1.72955066e-01 1.50178075e-01
6.35467470e-01 -7.12970048e-02 -6.07567251e-01 2.71446496e-01
1.10144770e+00 -2.89902955e-01 5.40336967e-01 -6.41701162e-01
6.82449281e-01 -4.01746929e-01 -1.70248902e+00 2.38031864e-01
2.24027053e-01 -1.20397449e-01 -2.23544106e-01 4.32646811e-01
4.34753329e-01 -3.86471629e-01 -1.44411981e+00 6.00628197e-01
2.83302993e-01 -4.98327762e-01 5.87078869e-01 2.50014439e-02
-3.06301832e-01 -5.99715590e-01 -2.51133204e-01 -1.99624741e+00
-3.66091162e-01 -7.97602236e-01 -3.51718247e-01 1.40736365e+00
6.62342981e-02 -9.34675217e-01 2.88817585e-01 1.13225687e+00
-3.42718482e-01 -1.04975557e+00 -1.14826238e+00 -5.90624869e-01
-1.11539744e-01 4.71339405e-01 1.24923599e+00 9.58632529e-01
1.97342932e-01 -3.20996642e-02 -2.62716785e-02 6.14663899e-01
7.90138245e-01 3.51752341e-01 2.34591872e-01 -6.19153976e-01
-3.27640653e-01 -4.34351414e-01 1.96438596e-01 1.27986282e-01
-2.29190260e-01 -3.07099402e-01 1.31260995e-02 -1.85536671e+00
-1.01984560e-01 -2.48215929e-01 -6.51508272e-01 5.09847462e-01
1.29829526e-01 -4.11926091e-01 2.77608931e-01 -1.30688064e-02
2.33882349e-02 8.62406909e-01 1.02017343e+00 -1.91725835e-01
-3.80086631e-01 1.00185566e-01 -8.68171751e-02 1.59550369e-01
1.09882689e+00 1.81869179e-01 -1.00500762e+00 -1.18808141e-02
-2.87788771e-02 4.02103126e-01 -1.31022558e-01 -9.99362826e-01
2.21317515e-01 -9.87909675e-01 3.45572889e-01 -9.01764274e-01
-1.89989105e-01 -1.52107608e+00 9.01406705e-01 1.04132009e+00
-7.70963430e-02 8.76552582e-01 1.37245014e-01 4.61866707e-01
8.51336643e-02 3.29859406e-01 8.04605901e-01 -1.22017741e-01
-5.74354112e-01 8.44380409e-02 -3.37866902e-01 -3.90439302e-01
1.72850001e+00 5.58306277e-02 -4.69210178e-01 -2.52222091e-01
-3.42492491e-01 1.38191307e+00 2.54827291e-01 2.99924314e-01
-8.04771408e-02 -1.35177362e+00 -3.16052437e-01 -6.69704890e-03
-3.35365593e-01 1.77096836e-02 4.79248971e-01 5.21356642e-01
-2.72002101e-01 5.92557192e-01 -3.08405042e-01 -4.44065124e-01
-8.50221038e-01 6.55755222e-01 6.95315957e-01 -5.23027480e-01
-1.41939715e-01 -1.57422870e-01 -1.74459323e-01 -2.53887754e-02
-2.23414883e-01 -6.53965056e-01 1.15213040e-02 3.91877472e-01
1.61668837e-01 8.15613270e-01 1.62811115e-01 -2.09942639e-01
-2.21227124e-01 5.57938993e-01 5.80052614e-01 2.13399127e-01
1.32358229e+00 -4.03335005e-01 -5.18506654e-02 1.81771383e-01
7.49419749e-01 -5.16959906e-01 -1.10651934e+00 3.90971273e-01
-2.55216271e-01 -2.23388910e-01 4.06769305e-01 -1.31392539e+00
-1.24344516e+00 3.97329368e-02 7.16838419e-01 9.51910853e-01
1.84295464e+00 -6.16727591e-01 8.24947119e-01 -4.41447049e-02
3.13580692e-01 -1.91788864e+00 -3.40940028e-01 -5.07684909e-02
7.03464568e-01 -7.13562667e-01 2.24269927e-01 2.42962331e-01
-5.03025234e-01 9.92056906e-01 6.50464654e-01 8.34977441e-03
6.89436316e-01 2.91042298e-01 -4.58544433e-01 1.22950524e-01
-9.79236186e-01 3.74016911e-01 -1.40122831e-01 2.07756847e-01
1.62661985e-01 7.58197784e-01 -5.46020806e-01 1.12699680e-01
-1.33765161e-01 -8.96304622e-02 4.87734348e-01 1.15213370e+00
-5.46621740e-01 -8.28728497e-01 -5.15836000e-01 5.33067882e-01
-5.34209386e-02 5.63684821e-01 5.68254352e-01 7.04555631e-01
3.63910526e-01 9.25279677e-01 8.10886398e-02 2.59926140e-01
6.54852569e-01 1.10339165e-01 -9.61055160e-02 -1.25236228e-01
-5.90639889e-01 1.71310738e-01 -5.19170314e-02 -1.53920755e-01
-1.66896328e-01 -6.26825511e-01 -1.64851689e+00 -5.52898109e-01
-3.35232615e-01 6.70524240e-01 9.39642429e-01 1.04596972e+00
1.68758631e-01 8.51269543e-01 1.58665359e+00 -7.54623652e-01
-1.09975052e+00 -7.60670662e-01 -1.01257145e+00 9.70376879e-02
2.99628794e-01 -6.87722623e-01 -5.75436413e-01 -4.84325469e-01]
|
[5.639039993286133, 2.472252130508423]
|
269e6093-4f2c-44ee-82b1-b370b683a4c9
|
on-influence-functions-classification
|
2305.16094
| null |
https://arxiv.org/abs/2305.16094v1
|
https://arxiv.org/pdf/2305.16094v1.pdf
|
On Influence Functions, Classification Influence, Relative Influence, Memorization and Generalization
|
Machine learning systems such as large scale recommendation systems or natural language processing systems are usually trained on billions of training points and are associated with hundreds of billions or trillions of parameters. Improving the learning process in such a way that both the training load is reduced and the model accuracy improved is highly desired. In this paper we take a first step toward solving this problem, studying influence functions from the perspective of simplifying the computations they involve. We discuss assumptions, under which influence computations can be performed on significantly fewer parameters. We also demonstrate that the sign of the influence value can indicate whether a training point is to memorize, as opposed to generalize upon. For this purpose we formally define what memorization means for a training point, as opposed to generalization. We conclude that influence functions can be made practical, even for large scale machine learning systems, and that influence values can be taken into account by algorithms that selectively remove training points, as part of the learning process.
|
['Ilqar Ramazanli', 'Ousmane Dia', 'Michael Kounavis']
|
2023-05-25
| null | null | null | null |
['memorization']
|
['natural-language-processing']
|
[ 2.58440316e-01 1.30495295e-01 -7.20994025e-02 -3.93583238e-01
-2.12056592e-01 -6.58329725e-01 5.95534265e-01 6.31996751e-01
-7.44370461e-01 7.76828408e-01 -1.25381634e-01 -5.83703578e-01
-2.06494525e-01 -1.10527158e+00 -9.47174251e-01 -6.41415119e-01
-1.07850187e-01 6.20773077e-01 9.91582796e-02 -6.39505610e-02
6.38189912e-01 6.63252711e-01 -1.67898893e+00 2.69401222e-01
7.46252537e-01 7.92854011e-01 -3.95464040e-02 8.53845477e-01
-1.74297318e-01 5.84048092e-01 -7.89719880e-01 -2.71078020e-01
3.21347892e-01 -2.81475157e-01 -7.79594243e-01 1.63392890e-02
5.15634835e-01 6.08427003e-02 2.57319689e-01 8.10162365e-01
3.76822591e-01 3.00482184e-01 6.51280165e-01 -9.64511812e-01
-1.58088997e-01 5.57923913e-01 -2.90012687e-01 2.78457224e-01
-3.17586935e-03 -6.47421703e-02 8.21085751e-01 -6.67331338e-01
2.97831088e-01 9.97493327e-01 6.82364106e-01 2.62653679e-01
-1.12676835e+00 -3.98225844e-01 4.65275079e-01 -7.27734119e-02
-1.23191822e+00 -5.20153821e-01 4.70909506e-01 -3.68332684e-01
1.17661488e+00 4.41683650e-01 8.20443630e-01 2.29495198e-01
4.41015005e-01 5.63056052e-01 6.45546854e-01 -6.59304917e-01
4.67423707e-01 2.99388766e-01 2.90306449e-01 8.35974455e-01
6.55456126e-01 -2.81060308e-01 -4.30097520e-01 -3.31753224e-01
5.98972321e-01 -4.43535782e-02 7.47027099e-02 -3.09205294e-01
-9.40257430e-01 8.70627463e-01 1.99882194e-01 3.89485687e-01
-2.80834973e-01 2.06938133e-01 3.03226024e-01 7.10730910e-01
5.37781000e-01 9.36103523e-01 -5.98260045e-01 -4.23801616e-02
-9.19724345e-01 1.12509780e-01 1.08043373e+00 6.75579906e-01
9.16875124e-01 -1.92187622e-01 2.14008272e-01 6.79916918e-01
-1.42806619e-01 4.38689798e-01 5.23366690e-01 -8.09301317e-01
2.16311604e-01 7.38046050e-01 3.39777842e-02 -7.96257734e-01
-3.33272338e-01 -6.04079187e-01 -8.47390890e-01 1.49133235e-01
5.28741241e-01 -3.77110720e-01 -7.78733969e-01 1.69735670e+00
4.02950138e-01 7.76352957e-02 -5.44189587e-02 4.22283947e-01
3.24036144e-02 6.51670158e-01 1.31356671e-01 -3.45256537e-01
9.76882875e-01 -6.32920980e-01 -2.50826359e-01 -2.99844503e-01
1.22935581e+00 -9.07634377e-01 1.13759136e+00 6.41041458e-01
-1.26724231e+00 -3.81945014e-01 -9.49139416e-01 1.50851816e-01
-4.19786453e-01 -1.55999094e-01 7.67181575e-01 7.11645603e-01
-1.02280200e+00 1.05161405e+00 -7.92688072e-01 -2.62955904e-01
1.41922683e-01 5.61813176e-01 -4.13063057e-02 -2.69233119e-02
-9.99600291e-01 1.00168312e+00 2.05466866e-01 -1.01986244e-01
-3.16924959e-01 -9.20939684e-01 -4.88167018e-01 2.93502897e-01
3.46216291e-01 -8.11391950e-01 1.02523053e+00 -1.21114576e+00
-1.14336348e+00 4.98405725e-01 -2.29213297e-01 -4.68877554e-01
4.16613251e-01 -3.25994134e-01 -3.46207805e-02 -2.16033444e-01
-3.77798617e-01 1.80713862e-01 1.00418496e+00 -8.02275538e-01
-7.55507648e-01 -4.36326534e-01 1.67983174e-01 2.82827705e-01
-6.76243961e-01 -1.12898059e-01 -3.04384738e-01 -4.09267962e-01
-2.05304066e-04 -1.17087400e+00 -5.15171826e-01 -1.84054017e-01
-1.11330725e-01 -3.50525618e-01 5.47466278e-01 -3.16721164e-02
1.37620008e+00 -1.91686404e+00 4.03959267e-02 7.45414197e-01
1.88958317e-01 3.89424324e-01 -1.68497220e-01 3.54335219e-01
8.74205083e-02 4.27965671e-01 3.71245928e-02 -3.51039544e-02
-2.51347899e-01 2.20017925e-01 -1.76614150e-01 4.79912847e-01
3.37255932e-02 5.17304718e-01 -7.28346109e-01 -1.26509115e-01
1.66717649e-01 3.33621264e-01 -8.83085608e-01 -9.14924592e-02
-1.77448526e-01 6.40072376e-02 -6.16594195e-01 7.11968169e-02
2.69941896e-01 -5.64959705e-01 2.41853639e-01 7.24347755e-02
-6.81001097e-02 4.02366400e-01 -1.33459342e+00 1.09108698e+00
-9.20380831e-01 4.88350332e-01 -9.08288062e-02 -9.77761447e-01
7.93528557e-01 7.10461661e-02 3.88517559e-01 -3.97297174e-01
7.34373853e-02 6.97915852e-02 1.75326034e-01 -1.18762195e-01
5.65914690e-01 -1.68597743e-01 3.10890470e-02 9.75748956e-01
-3.35608333e-01 -3.85439768e-02 4.48950201e-01 2.80875802e-01
1.07656527e+00 -6.11424506e-01 3.52037877e-01 -5.54964602e-01
4.33788240e-01 -2.19612606e-02 3.32261175e-01 9.86544907e-01
4.02431577e-01 9.65625495e-02 4.92090404e-01 -5.12102187e-01
-1.23403549e+00 -5.78216791e-01 -1.24800928e-01 1.29183304e+00
-1.28039539e-01 -5.63857496e-01 -7.15906382e-01 -5.37912011e-01
2.55614728e-01 8.10256422e-01 -6.34402812e-01 -3.57992798e-01
-7.12113619e-01 -7.76895821e-01 2.83598840e-01 5.13627768e-01
1.48151368e-01 -7.66725838e-01 -5.73810935e-01 7.05494434e-02
4.59076196e-01 -5.11009037e-01 -3.05913061e-01 3.83205384e-01
-1.27239835e+00 -1.04992115e+00 -5.31548262e-01 -4.09962356e-01
1.20206404e+00 3.14751208e-01 1.13960254e+00 6.20697021e-01
4.41732816e-02 2.84334987e-01 -2.86119759e-01 -5.70559978e-01
-4.16611671e-01 3.54678333e-01 2.49355003e-01 -2.85709918e-01
6.00988925e-01 -6.89842045e-01 -5.53764284e-01 2.73704678e-01
-8.03722501e-01 -8.82027671e-02 7.79742181e-01 5.85704863e-01
4.23037320e-01 6.13648035e-02 6.86130047e-01 -1.40734410e+00
9.36091006e-01 -3.82871389e-01 -6.80893779e-01 4.84810263e-01
-1.18153512e+00 5.22435606e-01 9.06344533e-01 -5.53053975e-01
-7.66677320e-01 -1.43477712e-02 -1.98490247e-01 -5.35468943e-02
-7.85329938e-02 5.49140215e-01 6.76472252e-03 -3.71859968e-01
8.86453927e-01 9.45973322e-02 -4.01454680e-02 -4.18503761e-01
5.54234684e-01 5.29682636e-01 -1.30395561e-01 -7.51596332e-01
6.78056359e-01 2.10407421e-01 7.31875524e-02 -8.13289225e-01
-7.68625259e-01 -4.21004236e-01 -7.05044329e-01 -1.16966248e-01
8.25608298e-02 -5.36611021e-01 -5.34863532e-01 5.52143157e-02
-8.62857580e-01 -4.18511808e-01 -4.28598970e-01 3.71820152e-01
-3.22794646e-01 7.00275525e-02 -3.66463065e-01 -6.13435924e-01
-3.84156227e-01 -7.30838776e-01 4.47708756e-01 1.91382095e-01
-3.73858958e-01 -1.17454314e+00 5.36253415e-02 8.23857933e-02
4.71836358e-01 -2.73952276e-01 1.10778761e+00 -9.07104790e-01
-3.91779035e-01 -5.39799631e-01 1.66992083e-01 3.31719786e-01
-2.26888452e-02 1.99962151e-03 -7.03289628e-01 -5.12900472e-01
8.80099647e-03 4.28567640e-02 8.20960343e-01 4.25589532e-01
1.20033634e+00 -5.30749202e-01 -3.61348480e-01 4.27700222e-01
1.25413382e+00 -3.15107196e-03 6.20410323e-01 2.53899664e-01
5.58127284e-01 5.33815503e-01 5.11782765e-01 5.03033340e-01
-1.65230945e-01 2.39941686e-01 -1.06859915e-01 8.43600556e-02
2.32071042e-01 -2.33198747e-01 -1.14004888e-01 1.01090610e+00
-1.94200918e-01 -2.35742077e-01 -8.74213815e-01 3.89766514e-01
-1.69635451e+00 -6.83005989e-01 6.33704588e-02 2.61815095e+00
8.04615676e-01 3.65524918e-01 -6.06992133e-02 2.92347938e-01
5.60298324e-01 -8.45123976e-02 -6.67911947e-01 -6.11239314e-01
2.93547004e-01 2.56593823e-01 7.33291864e-01 7.30736971e-01
-6.34456754e-01 8.72510791e-01 6.75879669e+00 8.16599548e-01
-1.39893579e+00 -6.43912852e-02 7.83836067e-01 -5.26626527e-01
-2.91397154e-01 -4.87635992e-02 -8.25660110e-01 2.08477512e-01
1.12247717e+00 -3.57546538e-01 5.07043362e-01 9.36262548e-01
1.43117771e-01 -2.10312933e-01 -1.32624424e+00 5.74151158e-01
-1.25747502e-01 -1.36609352e+00 2.60996670e-01 7.32640252e-02
8.38183045e-01 -7.78722689e-02 5.10668047e-02 1.98500320e-01
3.05607021e-01 -9.84824955e-01 3.61253798e-01 4.25370932e-01
3.84489328e-01 -9.92006421e-01 5.04261434e-01 7.68313706e-01
-8.46674979e-01 -2.69238168e-04 -7.45000839e-01 -4.02475834e-01
-3.50974649e-02 1.06626248e+00 -7.31551588e-01 4.69736345e-02
3.55856806e-01 3.90388757e-01 -5.67637682e-01 1.14948130e+00
-2.66506076e-01 8.35085750e-01 -6.61459804e-01 -4.26215172e-01
5.50549477e-02 -7.91770741e-02 2.01798305e-01 1.09924960e+00
4.26278502e-01 2.83698827e-01 -4.62204404e-02 3.94436955e-01
-2.06608146e-01 4.72053111e-01 -4.97280061e-01 -6.28340393e-02
5.96038580e-01 1.17391956e+00 -7.44820237e-01 -6.37516081e-01
-2.57395297e-01 6.28521562e-01 5.95149875e-01 2.82751918e-01
-3.96874070e-01 -4.13496435e-01 5.98068476e-01 2.61590391e-01
3.03104907e-01 -2.25519955e-01 -5.75989306e-01 -1.14730251e+00
-7.28596747e-02 -8.06913555e-01 2.31194228e-01 -3.30588400e-01
-1.01807249e+00 4.28098500e-01 -1.84532359e-01 -8.52821410e-01
-3.78901780e-01 -4.91826653e-01 -7.22852886e-01 7.85937130e-01
-1.33904934e+00 -5.61199427e-01 8.75610933e-02 4.25589025e-01
2.70335048e-01 4.44658212e-02 8.86696577e-01 3.05084199e-01
-5.33853590e-01 6.85851336e-01 3.66123527e-01 -3.05427641e-01
5.81983745e-01 -9.83520508e-01 2.91074127e-01 6.99236751e-01
2.30773494e-01 1.20002282e+00 8.88518810e-01 -5.73626816e-01
-1.32186282e+00 -9.24657762e-01 1.18474686e+00 -3.79366666e-01
5.49639642e-01 -1.82432666e-01 -1.05880630e+00 5.25530398e-01
-1.57011256e-01 -1.15781918e-01 8.23502302e-01 9.32553172e-01
-2.56652743e-01 -1.44585788e-01 -1.09525573e+00 4.48575050e-01
8.45282972e-01 -3.11857462e-01 -4.38239694e-01 4.79876012e-01
3.57771188e-01 -2.48394371e-03 -9.81588185e-01 2.29309633e-01
5.37580609e-01 -7.92391241e-01 8.11539471e-01 -9.28717196e-01
2.08712846e-01 -5.77097908e-02 1.87400058e-01 -1.48998117e+00
-3.78873259e-01 -5.23892343e-01 -2.85863817e-01 7.22892225e-01
6.54856920e-01 -7.31157959e-01 9.76082921e-01 1.06209970e+00
2.13906661e-01 -8.88221622e-01 -6.07242286e-01 -7.83179522e-01
4.18719471e-01 -4.13154244e-01 6.69283509e-01 9.04236019e-01
2.20582381e-01 6.00668788e-01 -2.68455356e-01 2.22185086e-02
4.13894564e-01 1.86350271e-01 7.74827838e-01 -1.43032837e+00
-3.28038156e-01 -4.29880857e-01 -2.94509977e-01 -1.25302541e+00
-1.46622628e-01 -6.91456318e-01 -2.58171052e-01 -1.30330968e+00
2.89821867e-02 -9.52209651e-01 -4.70275611e-01 4.99236584e-01
-1.51444867e-01 1.17429771e-01 4.68858853e-02 3.88893783e-01
-5.51237464e-01 -4.46352065e-02 1.14869189e+00 2.15761036e-01
-3.11723262e-01 3.18019837e-01 -8.16751182e-01 8.78492296e-01
9.82144892e-01 -5.94687462e-01 -6.09489799e-01 -4.60741729e-01
8.26888025e-01 -2.52008557e-01 -1.76702335e-01 -8.04263830e-01
5.25373697e-01 -2.40263939e-01 4.68578786e-01 -1.79515645e-01
2.69824509e-02 -6.04440570e-01 1.21397218e-02 6.31221175e-01
-6.84570551e-01 -5.10865338e-02 1.93505943e-01 4.61319983e-01
1.13525912e-01 -5.30397534e-01 7.76350260e-01 -1.98924899e-01
-3.97051543e-01 1.23223200e-01 -4.84091341e-01 -1.30782589e-01
9.91618156e-01 3.78834270e-02 -2.31256962e-01 -3.89731526e-01
-6.52402341e-01 5.02080396e-02 5.02169251e-01 1.01025060e-01
2.40054876e-01 -9.86668229e-01 -5.00171006e-01 3.12947989e-01
-7.74245039e-02 -1.49423987e-01 2.85470132e-02 6.18302226e-01
-4.95130658e-01 5.39123595e-01 4.62633558e-02 -1.56923532e-01
-1.51347530e+00 6.11377716e-01 8.60407352e-02 -3.35660011e-01
-4.18822616e-01 8.64771187e-01 -9.44255199e-03 -2.43142635e-01
1.87605917e-01 -3.56897026e-01 -1.35146603e-01 -1.20980144e-02
7.64413357e-01 2.62415439e-01 2.96892464e-01 -1.28545731e-01
-1.54019788e-01 5.03438354e-01 -4.77241397e-01 1.26004189e-01
1.42069626e+00 -1.30965468e-02 -1.11749284e-01 2.52531439e-01
1.04867685e+00 1.81997851e-01 -9.59904313e-01 -2.97151327e-01
-4.65726219e-02 -4.99774754e-01 2.14469433e-01 -7.26862907e-01
-9.79660332e-01 8.82832229e-01 3.21636438e-01 5.76578200e-01
9.76792574e-01 -9.89483297e-02 6.37606800e-01 1.03181612e+00
4.29217994e-01 -1.26336706e+00 -1.99378043e-01 5.55392921e-01
4.62842256e-01 -1.07200634e+00 3.22837830e-01 -2.24964246e-01
-2.97013670e-01 1.03876638e+00 3.17897439e-01 -4.06861573e-01
7.54069865e-01 4.10968333e-01 -1.82770878e-01 -1.15545057e-01
-1.07352757e+00 9.40629542e-02 2.99832046e-01 1.30961671e-01
6.77222311e-01 1.16505669e-02 -6.65044546e-01 3.18448931e-01
-3.30484599e-01 4.90243845e-02 4.05232787e-01 1.06087494e+00
-1.11120760e+00 -1.14846754e+00 -2.44808868e-01 9.51997399e-01
-3.88576239e-01 -3.98702919e-01 -3.41287494e-01 4.44968969e-01
-9.44006443e-02 7.85162687e-01 2.57294506e-01 -4.05170023e-01
1.36704937e-01 2.16355115e-01 5.33339381e-01 -7.46004641e-01
-6.00503147e-01 -1.71276167e-01 1.88428551e-01 -3.52386385e-01
3.52961826e-03 -4.38759744e-01 -1.30371749e+00 -9.33176219e-01
-5.30742168e-01 5.06401896e-01 9.25209641e-01 1.12960100e+00
4.00555849e-01 4.47667450e-01 5.03512025e-01 -5.38644552e-01
-7.70634413e-01 -7.98067272e-01 -5.16551614e-01 2.15090558e-01
1.93106644e-02 -2.50357091e-01 -5.91822505e-01 -2.40544435e-02]
|
[8.38067626953125, 4.30801248550415]
|
a15b10ea-da39-4428-ac44-a1caa49758b9
|
variational-quantum-soft-actor-critic-for
|
2212.11681
| null |
https://arxiv.org/abs/2212.11681v1
|
https://arxiv.org/pdf/2212.11681v1.pdf
|
Variational Quantum Soft Actor-Critic for Robotic Arm Control
|
Deep Reinforcement Learning is emerging as a promising approach for the continuous control task of robotic arm movement. However, the challenges of learning robust and versatile control capabilities are still far from being resolved for real-world applications, mainly because of two common issues of this learning paradigm: the exploration strategy and the slow learning speed, sometimes known as "the curse of dimensionality". This work aims at exploring and assessing the advantages of the application of Quantum Computing to one of the state-of-art Reinforcement Learning techniques for continuous control - namely Soft Actor-Critic. Specifically, the performance of a Variational Quantum Soft Actor-Critic on the movement of a virtual robotic arm has been investigated by means of digital simulations of quantum circuits. A quantum advantage over the classical algorithm has been found in terms of a significant decrease in the amount of required parameters for satisfactory model training, paving the way for further promising developments.
|
['Antonio Policicchio', 'Mattia Pavese', 'Matteo Conterno', 'Ludovico Bozzolo', 'Paola Barillà', 'Alberto Acuto']
|
2022-12-20
| null | null | null | null |
['continuous-control']
|
['playing-games']
|
[ 2.09876567e-01 2.14862570e-01 -2.02405974e-01 1.64540157e-01
-5.58180273e-01 -2.12632373e-01 8.27677727e-01 8.91726464e-02
-7.06864476e-01 8.01103055e-01 -3.74213487e-01 -2.44349718e-01
-3.79675299e-01 -6.92153990e-01 -5.93297839e-01 -1.26052380e+00
3.17686871e-02 4.21639472e-01 -3.07204090e-02 -7.01503873e-01
3.00993949e-01 5.53747296e-01 -1.44447732e+00 -3.84397030e-01
6.15429997e-01 7.99654722e-01 -1.20985515e-01 2.52116948e-01
7.47991353e-02 4.51400161e-01 -1.82091504e-01 -2.42955774e-01
1.99716866e-01 -8.10933113e-01 -5.59283257e-01 -2.02259108e-01
-2.59557247e-01 1.40768081e-01 -4.93668586e-01 1.23646140e+00
5.52843809e-01 3.18344504e-01 5.17152548e-01 -5.85554421e-01
-7.31728524e-02 5.32245636e-01 -1.32626846e-01 -5.75411739e-03
-3.86571735e-02 4.73527819e-01 1.04211152e+00 -2.16159552e-01
7.65998840e-01 1.00500643e+00 1.91678777e-02 7.46403992e-01
-1.26571393e+00 -2.71329373e-01 -4.87602562e-01 4.65981007e-01
-1.27309406e+00 -2.35950768e-01 7.85007715e-01 -2.55735189e-01
8.12765956e-01 -2.43346527e-01 7.51255810e-01 1.04978013e+00
5.26238799e-01 2.89169282e-01 1.20037401e+00 -6.86830819e-01
9.68834758e-01 7.35466182e-02 1.08326962e-02 7.53162563e-01
2.47489333e-01 8.10042143e-01 -5.10813117e-01 3.30686457e-02
6.09188020e-01 -5.27743042e-01 3.56747881e-02 -8.21555495e-01
-1.05790424e+00 9.94679511e-01 6.94350541e-01 6.20207310e-01
-5.47401965e-01 4.75149781e-01 4.25998062e-01 1.08540729e-01
-5.28286994e-02 7.09780097e-01 3.31186131e-02 -4.31555510e-01
-5.16754329e-01 3.92182082e-01 3.39026362e-01 2.18014106e-01
6.14495754e-01 3.21267635e-01 1.33322179e-01 2.35521980e-02
3.02677184e-01 4.66253072e-01 1.69699878e-01 -9.57754374e-01
9.64014083e-02 4.93493110e-01 2.31909767e-01 -3.92526031e-01
-5.61730802e-01 -5.09119153e-01 -7.93755591e-01 9.29633021e-01
5.73677003e-01 -3.18956167e-01 -4.15123075e-01 1.69628024e+00
3.73506546e-01 -2.60011971e-01 2.00265482e-01 9.27306294e-01
-3.95185389e-02 6.03815138e-01 -1.32865561e-02 -2.72203624e-01
1.04650366e+00 -4.95566696e-01 -6.79475307e-01 -6.04888760e-02
4.25390542e-01 -2.71040678e-01 6.97932839e-01 6.13650739e-01
-1.06075120e+00 -4.33385849e-01 -1.20150399e+00 3.45449179e-01
-2.98030853e-01 3.53486501e-02 7.89500773e-01 6.89682066e-01
-4.78496224e-01 1.07937419e+00 -1.04945862e+00 -1.59614250e-01
3.43755633e-01 6.48526847e-01 -3.37848276e-01 2.08451718e-01
-1.03658891e+00 1.24589574e+00 6.48627639e-01 1.02241039e-01
-7.80095816e-01 -2.62584388e-01 -2.74174124e-01 1.72005087e-01
3.66038680e-01 -5.40769994e-01 1.13076293e+00 -6.87865436e-01
-2.37877870e+00 4.90104288e-01 1.25205144e-01 -6.17516518e-01
4.71013308e-01 1.32545620e-01 -6.39096424e-02 2.39277184e-01
-3.96709442e-01 3.33169430e-01 1.00142741e+00 -7.45583773e-01
-1.39287889e-01 -6.74321473e-01 3.68899256e-02 -1.08094260e-01
-1.43027917e-01 -3.90945733e-01 3.43195975e-01 2.08320352e-03
1.28535926e-01 -1.34726059e+00 -6.15175605e-01 -2.94460714e-01
1.06876105e-01 -3.99496198e-01 3.65953177e-01 1.86800346e-01
7.05098033e-01 -2.01429415e+00 8.48723292e-01 2.09489733e-01
-1.20071054e-01 5.21572948e-01 1.17414884e-01 7.39283264e-01
-4.58030105e-02 -3.74908417e-01 -1.12213558e-02 2.01496214e-01
5.61362468e-02 1.46915436e-01 -2.88332582e-01 6.13379061e-01
2.47869834e-01 8.74595106e-01 -9.51126635e-01 -6.45467713e-02
5.11309147e-01 5.45295238e-01 -4.50470805e-01 -1.34430379e-01
-4.82144415e-01 8.57123077e-01 -7.42686987e-01 -1.42259859e-02
2.51695096e-01 1.06684472e-02 3.56697261e-01 -1.27856052e-02
-5.78554213e-01 2.98919708e-01 -1.11835885e+00 1.74930906e+00
-2.99864501e-01 2.54031241e-01 3.59894261e-02 -1.19796598e+00
8.11585665e-01 2.22355336e-01 4.71401244e-01 -1.07564652e+00
5.47357500e-01 5.42568326e-01 3.46167892e-01 -4.65249032e-01
4.52576518e-01 -5.14158249e-01 -8.54679942e-02 1.51410520e-01
1.89327255e-01 -5.04229486e-01 2.09568173e-01 -1.36979103e-01
9.19832945e-01 4.38456118e-01 1.19514443e-01 -3.08176666e-01
5.73399246e-01 1.91004440e-01 5.76186851e-02 6.43545091e-01
-1.61751404e-01 -1.52423695e-01 6.03444695e-01 -2.48402447e-01
-1.11186969e+00 -8.42888832e-01 -1.54345289e-01 4.41975147e-01
2.19735280e-01 -2.86190987e-01 -8.41202855e-01 -8.79851356e-02
-1.12013750e-01 6.64864123e-01 -4.01436448e-01 -4.43859696e-01
-6.38999343e-01 -8.14461291e-01 2.14825094e-01 -1.77587360e-01
1.59532860e-01 -1.07516909e+00 -1.18797219e+00 3.63458425e-01
4.07986075e-01 -1.05903780e+00 6.70633852e-01 5.62382102e-01
-9.88588393e-01 -9.56826091e-01 -5.05497336e-01 -1.68000668e-01
7.96150863e-02 -1.99665338e-01 3.84148479e-01 -2.65571088e-01
-4.72470403e-01 2.48787373e-01 -2.46245652e-01 -8.79515782e-02
-9.23139095e-01 2.41901785e-01 2.33775675e-01 -1.89783782e-01
4.23418358e-02 -4.68162268e-01 -4.87948596e-01 -1.27985567e-01
-6.93420947e-01 -2.84005702e-01 7.79968798e-01 9.70558822e-01
4.12327498e-01 2.41570368e-01 5.93803585e-01 -2.28005946e-01
4.03409660e-01 -1.61191210e-01 -1.11939359e+00 -1.97581630e-02
-6.88414037e-01 6.64227962e-01 6.96670651e-01 -3.01344037e-01
-8.50687087e-01 1.35677338e-01 -3.16239327e-01 2.03815296e-01
5.39578348e-02 3.34025800e-01 3.41362715e-01 -4.68080610e-01
7.10978210e-01 1.68489695e-01 3.23121399e-01 -8.55419859e-02
4.34223443e-01 4.75039899e-01 2.26634026e-01 -5.45404792e-01
7.23861337e-01 3.98762554e-01 8.48630965e-01 -1.05467653e+00
-4.15640086e-01 -1.17252074e-01 -3.89980406e-01 -3.06127161e-01
9.87433374e-01 -5.12269855e-01 -1.35976601e+00 3.76728386e-01
-8.62681925e-01 -3.44466329e-01 -5.29940605e-01 8.39696229e-01
-9.09733355e-01 5.45269668e-01 -4.99389321e-01 -1.13401079e+00
-1.20834626e-01 -1.48831022e+00 5.18490493e-01 4.21986610e-01
2.37933859e-01 -6.15825713e-01 2.63752848e-01 2.43119493e-01
5.28993666e-01 2.17803106e-01 9.99881387e-01 3.07854544e-02
-6.93550050e-01 -4.52172309e-01 2.87313521e-01 2.34837934e-01
-5.20819247e-01 1.42151313e-02 -8.85679722e-01 -3.63306642e-01
5.42498268e-02 -5.71086884e-01 6.24561965e-01 1.20342135e-01
5.98637879e-01 4.66396034e-01 -8.79966393e-02 2.82238424e-02
1.56349146e+00 1.34394258e-01 6.16748452e-01 3.23766291e-01
-1.31535344e-02 3.92112911e-01 4.45794076e-01 5.05827427e-01
-2.84053057e-01 1.01614368e+00 8.97109449e-01 6.00604177e-01
8.70910361e-02 2.21253056e-02 3.93110305e-01 5.82525611e-01
-3.77220154e-01 2.04650536e-01 -5.58581173e-01 4.56934236e-02
-1.81810129e+00 -1.03715062e+00 -1.63427174e-01 2.44477224e+00
3.73169482e-01 4.50036615e-01 8.85338634e-02 2.84745097e-01
3.69530916e-01 9.86295193e-02 -5.27807534e-01 -5.77699423e-01
7.91411325e-02 6.61165535e-01 3.38361710e-01 3.28042150e-01
-6.47358298e-01 8.60089600e-01 5.30100727e+00 8.50814283e-01
-1.24154925e+00 3.87431793e-02 -1.40955716e-01 1.16229504e-01
7.52753392e-02 3.31722796e-01 -6.59249187e-01 3.47942322e-01
1.23126543e+00 1.75638467e-01 8.62977684e-01 7.00330913e-01
1.93892419e-01 -5.38191319e-01 -6.67738616e-01 9.31642294e-01
-3.68936330e-01 -1.43725586e+00 -2.81572849e-01 2.59389609e-01
4.87331569e-01 1.56930819e-01 1.53044000e-01 4.11257774e-01
-4.26843733e-01 -6.61872923e-01 7.38751888e-01 4.85084653e-01
4.54438955e-01 -6.99314773e-01 5.43610632e-01 4.59766984e-01
-4.71340686e-01 -4.07960206e-01 -4.68133301e-01 -2.84164369e-01
2.04107270e-01 2.29830593e-01 -3.32932830e-01 4.83380556e-01
1.72662333e-01 2.28296146e-01 -1.65689513e-01 8.93702865e-01
-2.43857652e-01 4.56018656e-01 -4.96374756e-01 -8.48067224e-01
6.38067365e-01 -6.62400663e-01 5.89160085e-01 4.93371487e-01
2.01199114e-01 -7.89087862e-02 -3.69393587e-01 1.15332603e+00
1.88076571e-01 -3.90446335e-02 -4.59002733e-01 -5.28758168e-01
4.97348271e-02 1.23653841e+00 -7.22506523e-01 4.38006371e-02
2.68448703e-02 5.59948027e-01 2.48427376e-01 -1.02210194e-01
-6.81304812e-01 -2.74043500e-01 1.60879120e-01 -1.05296709e-01
4.89738375e-01 -7.21976280e-01 -1.02924071e-01 -9.46038663e-01
-1.75001949e-01 -6.11877561e-01 -2.42840201e-01 -2.81491429e-01
-5.13833940e-01 2.65363693e-01 -1.36484161e-01 -9.49122727e-01
-4.35137004e-01 -9.13146377e-01 -4.37148839e-01 6.05005503e-01
-1.32384741e+00 -6.88812673e-01 9.96928778e-04 3.84046823e-01
1.30639687e-01 -3.32243145e-01 1.22326469e+00 -2.38818116e-02
-4.68413979e-01 1.93831041e-01 6.50541246e-01 -4.43051875e-01
1.94176018e-01 -1.04993582e+00 2.51624379e-02 4.51601148e-01
1.69374287e-01 3.09379399e-01 1.09075153e+00 -1.10857636e-01
-2.03321671e+00 -1.60844475e-01 3.71943712e-01 -6.57108426e-02
9.22683775e-01 -1.72831744e-01 -6.28159583e-01 1.77777605e-03
1.45016834e-01 -2.50491071e-02 1.65677577e-01 1.05498984e-01
-3.58882215e-04 -8.39332193e-02 -9.16367412e-01 5.46669781e-01
4.02506262e-01 -4.96361166e-01 -4.89302665e-01 2.84907430e-01
2.36285001e-01 -2.09275946e-01 -8.49715233e-01 2.40404144e-01
7.22900927e-01 -1.09482884e+00 7.49542058e-01 -4.27742004e-01
2.27403224e-01 -5.26561961e-02 6.61252886e-02 -1.21571219e+00
-8.55243802e-02 -9.83668745e-01 -1.17219619e-01 5.39165914e-01
1.32670388e-01 -6.17944777e-01 8.78027856e-01 6.45833686e-02
1.01806354e-02 -7.88850904e-01 -1.48434460e+00 -7.57490098e-01
5.08065522e-01 -1.71884224e-01 1.13121653e-02 4.08991247e-01
4.39209759e-01 4.89458203e-01 -9.94856507e-02 -1.18047744e-01
6.78435326e-01 -1.79638695e-02 5.33195436e-01 -8.82652879e-01
-6.19779468e-01 -6.78942144e-01 -5.65752864e-01 -4.41027910e-01
1.91269472e-01 -7.98783422e-01 -2.33757626e-02 -9.53976870e-01
-1.15324631e-01 -2.26772219e-01 -2.50203788e-01 -1.44191965e-01
2.12639049e-01 -6.07990809e-02 3.33291620e-01 7.01975375e-02
-4.64085609e-01 8.31480145e-01 1.12105811e+00 6.93396553e-02
-2.53086567e-01 2.71491528e-01 1.46268345e-02 3.40539992e-01
1.09074712e+00 -6.08938038e-01 -1.44991368e-01 1.09402075e-01
5.65941215e-01 5.89615405e-01 4.65660095e-01 -1.38913596e+00
1.79689065e-01 6.79298490e-02 -1.88813314e-01 -1.68419421e-01
5.63210189e-01 -6.42350793e-01 -8.12466443e-02 1.06537545e+00
-4.87914264e-01 -1.49239317e-01 -7.28936642e-02 7.29617000e-01
-7.91491196e-02 -6.62546635e-01 1.17967939e+00 -9.71519873e-02
-4.87072080e-01 -1.39380723e-01 -5.39546430e-01 -2.08690450e-01
1.10860455e+00 1.95958585e-01 -9.29649323e-02 5.84821552e-02
-6.47800624e-01 -2.89242864e-01 2.95452625e-01 1.60131603e-01
1.20224960e-01 -8.31232786e-01 -2.11099580e-01 2.60884315e-02
5.86395245e-03 -5.02440453e-01 3.02977622e-01 9.72922087e-01
-4.65402722e-01 6.49765968e-01 -6.04552865e-01 -4.62852210e-01
-7.86825716e-01 6.69249952e-01 5.36284387e-01 -3.26154441e-01
-6.53119564e-01 3.16823661e-01 -6.50574207e-01 -6.89899623e-02
1.17474385e-01 -8.54991004e-03 2.01499939e-01 -1.68118536e-01
1.81457788e-01 4.68978286e-01 1.46380365e-01 -3.93893063e-01
-1.45701393e-01 6.54802859e-01 2.21601784e-01 -3.43169153e-01
1.24740851e+00 2.77066648e-01 -9.97851044e-02 4.99665707e-01
8.35323274e-01 -2.98187733e-01 -1.07216513e+00 -6.79107904e-02
1.72852710e-01 1.36919051e-01 3.58018607e-01 -6.30319297e-01
-3.83291721e-01 1.27431107e+00 1.04002929e+00 3.33814204e-01
3.67985874e-01 -1.62963375e-01 5.70413053e-01 8.85494292e-01
9.17502582e-01 -1.34925878e+00 1.19922869e-01 4.67680693e-01
4.88118917e-01 -1.26281428e+00 4.43042703e-02 1.71102330e-01
-3.30684483e-01 1.53118765e+00 7.79345282e-04 -5.09187162e-01
4.45354462e-01 -1.77728340e-01 -3.59470487e-01 -1.87098399e-01
-4.07835722e-01 -4.50662941e-01 -7.16173723e-02 1.80229381e-01
2.55227625e-01 2.49121010e-01 -6.73025370e-01 -1.02335900e-01
1.66259646e-01 2.36829206e-01 5.72152138e-01 8.12754631e-01
-5.30168891e-01 -1.44294584e+00 -7.27107450e-02 -1.69350132e-01
-3.83435994e-01 5.32028317e-01 6.33862019e-02 9.20933425e-01
6.43506050e-02 6.22705400e-01 -5.19081056e-01 -9.52458456e-02
1.76890060e-01 2.09045112e-01 1.00342607e+00 -3.95779669e-01
-6.23452187e-01 -2.06727177e-01 -2.94279099e-01 -4.83760357e-01
-4.38361436e-01 -6.33545160e-01 -1.43445504e+00 -8.24131966e-02
-5.85288048e-01 4.77862507e-01 1.31862319e+00 1.10007918e+00
1.34407640e-01 4.87632453e-01 4.89073277e-01 -9.88540113e-01
-1.50121391e+00 -6.82658255e-01 -4.70677257e-01 2.93965861e-02
2.73677707e-01 -8.37154746e-01 -2.27423519e-01 -8.44085097e-01]
|
[5.592259407043457, 4.856123924255371]
|
02c0cd07-40dc-42a6-8814-b3f70de2acfd
|
dp-ssl-towards-robust-semi-supervised
|
2110.13740
| null |
https://arxiv.org/abs/2110.13740v1
|
https://arxiv.org/pdf/2110.13740v1.pdf
|
DP-SSL: Towards Robust Semi-supervised Learning with A Few Labeled Samples
|
The scarcity of labeled data is a critical obstacle to deep learning. Semi-supervised learning (SSL) provides a promising way to leverage unlabeled data by pseudo labels. However, when the size of labeled data is very small (say a few labeled samples per class), SSL performs poorly and unstably, possibly due to the low quality of learned pseudo labels. In this paper, we propose a new SSL method called DP-SSL that adopts an innovative data programming (DP) scheme to generate probabilistic labels for unlabeled data. Different from existing DP methods that rely on human experts to provide initial labeling functions (LFs), we develop a multiple-choice learning~(MCL) based approach to automatically generate LFs from scratch in SSL style. With the noisy labels produced by the LFs, we design a label model to resolve the conflict and overlap among the noisy labels, and finally infer probabilistic labels for unlabeled samples. Extensive experiments on four standard SSL benchmarks show that DP-SSL can provide reliable labels for unlabeled data and achieve better classification performance on test sets than existing SSL methods, especially when only a small number of labeled samples are available. Concretely, for CIFAR-10 with only 40 labeled samples, DP-SSL achieves 93.82% annotation accuracy on unlabeled data and 93.46% classification accuracy on test data, which are higher than the SOTA results.
|
['Shuigeng Zhou', 'Lu Zhang', 'Jiandong Ding', 'Yi Xu']
|
2021-10-26
| null |
http://proceedings.neurips.cc/paper/2021/hash/854d6fae5ee42911677c739ee1734486-Abstract.html
|
http://proceedings.neurips.cc/paper/2021/file/854d6fae5ee42911677c739ee1734486-Paper.pdf
|
neurips-2021-12
|
['semi-supervised-image-classification']
|
['computer-vision']
|
[-2.94146650e-02 3.37508082e-01 -5.41662157e-01 -9.68758583e-01
-1.37763774e+00 -7.84807384e-01 3.79200667e-01 -6.28959686e-02
-3.34985793e-01 1.12884676e+00 -8.94260406e-03 -3.38763416e-01
1.30819842e-01 -5.72403371e-01 -7.47542024e-01 -9.31114137e-01
4.80944514e-01 7.24677801e-01 -1.24801114e-01 3.38854223e-01
-7.69570172e-02 1.04423508e-01 -1.44050217e+00 3.17009419e-01
9.87132490e-01 1.01837909e+00 -1.04682848e-01 9.58247408e-02
-5.66732943e-01 7.41599143e-01 -6.01929247e-01 -2.31776729e-01
3.10008734e-01 -3.87374997e-01 -8.79853249e-01 2.28122354e-01
3.46542746e-01 -2.16149479e-01 1.76805139e-01 1.04405093e+00
3.74954909e-01 -5.84978089e-02 9.84013259e-01 -1.34406304e+00
-7.49768198e-01 1.24927378e+00 -7.29873240e-01 -2.84527779e-01
8.72212090e-03 4.74946462e-02 1.10402262e+00 -1.37195253e+00
3.54433775e-01 1.34507024e+00 6.61078334e-01 8.72375250e-01
-1.52912056e+00 -9.18222308e-01 2.57630467e-01 -1.34214625e-01
-1.47521138e+00 -2.49876007e-01 7.71939456e-01 -4.57597375e-01
1.68849856e-01 -1.05319157e-01 3.86389829e-02 1.23021007e+00
-3.88613969e-01 1.23125541e+00 1.54138005e+00 -7.20103562e-01
5.83984137e-01 3.21605206e-01 7.61229098e-01 4.35554117e-01
2.97419608e-01 1.82830349e-01 -5.10052383e-01 -1.92760095e-01
2.84158081e-01 -1.27470955e-01 -6.83094189e-02 -1.38088167e-01
-1.21812809e+00 9.39017117e-01 2.45675907e-01 -4.70746011e-02
6.50914945e-03 -9.73621160e-02 1.95351496e-01 1.80770695e-01
5.89212239e-01 3.43841165e-01 -7.95960963e-01 1.97629556e-01
-8.54831815e-01 -1.03414971e-02 7.66144931e-01 1.18078828e+00
9.77601409e-01 1.00819953e-01 -5.58203101e-01 1.00912488e+00
6.27529919e-01 6.62330091e-01 4.37872529e-01 -9.04298902e-01
5.22307932e-01 6.09274685e-01 3.48391503e-01 -4.39822644e-01
-1.88696012e-01 -3.50839376e-01 -8.04070830e-01 1.17902514e-02
6.21537805e-01 -4.77992535e-01 -1.21688044e+00 1.86985552e+00
3.26918691e-01 2.07742453e-01 1.50873944e-01 8.35050642e-01
1.01443982e+00 7.14786053e-01 2.04178452e-01 -2.57704347e-01
9.25098956e-01 -1.04238391e+00 -6.99901581e-01 -3.07733208e-01
9.47254300e-01 -4.23033059e-01 1.37045705e+00 5.66670477e-01
-5.35118341e-01 -6.64667785e-01 -9.67524469e-01 8.34016725e-02
-1.74123287e-01 5.22025943e-01 4.88176465e-01 9.60583091e-01
-7.80251622e-01 2.95921832e-01 -6.43354475e-01 2.70653337e-01
6.40986621e-01 3.66736025e-01 -1.48684368e-01 -4.38300401e-01
-1.26895392e+00 3.94510716e-01 7.64379382e-01 2.07488313e-01
-9.26930010e-01 -5.19055307e-01 -8.67015362e-01 7.41068181e-03
4.91373420e-01 4.87332009e-02 1.41262162e+00 -8.61163557e-01
-1.51140904e+00 7.70849347e-01 -1.99399129e-01 -1.32933095e-01
4.11801815e-01 -6.02489151e-02 -3.33329260e-01 -2.08331734e-01
2.74843931e-01 1.10601819e+00 6.28910184e-01 -1.73429382e+00
-5.92881799e-01 -1.73797473e-01 -9.59942639e-02 1.88209340e-02
-2.65913695e-01 -2.45383874e-01 -2.72309810e-01 -5.92363298e-01
3.41411233e-01 -1.05937028e+00 -3.39114070e-01 -2.23649263e-01
-8.00078988e-01 -7.87610412e-01 6.31045043e-01 8.83146301e-02
9.81769323e-01 -2.19221616e+00 -2.56415546e-01 2.49608353e-01
2.67090172e-01 5.20588636e-01 -2.43651643e-01 -6.09078854e-02
-1.33575678e-01 3.38759214e-01 -3.75967920e-01 -5.33737004e-01
2.31196508e-01 4.39191580e-01 -4.68930483e-01 2.17547119e-01
3.87350202e-01 8.38530600e-01 -1.16699636e+00 -6.91912413e-01
-7.71591440e-03 -4.13414976e-03 -3.40766490e-01 3.04206282e-01
-5.14761329e-01 5.92013836e-01 -5.31435668e-01 9.57371354e-01
8.30325603e-01 -4.62117940e-01 1.51426375e-01 2.23555155e-02
3.69416744e-01 1.12019144e-01 -1.41985190e+00 1.40793931e+00
-3.63817841e-01 2.77625114e-01 -4.61908609e-01 -9.42733526e-01
1.30296254e+00 3.24682713e-01 2.22407773e-01 -1.28161982e-01
2.47117937e-01 3.98801953e-01 -4.34384733e-01 -3.28150362e-01
-9.13548917e-02 -2.16747656e-01 -3.28542233e-01 7.97538221e-01
2.85578072e-01 -1.31740853e-01 3.13293308e-01 9.22579020e-02
7.95077503e-01 2.77476519e-01 4.03943956e-02 -1.38326123e-01
3.81001621e-01 7.97684267e-02 1.06088579e+00 8.84828269e-01
-2.85947323e-01 7.23497331e-01 5.45785367e-01 -4.57999140e-01
-6.93844140e-01 -1.11234796e+00 -3.26905996e-01 1.09981275e+00
3.19529362e-02 -2.16059655e-01 -7.17269480e-01 -1.35161674e+00
-9.93368179e-02 8.50493670e-01 -4.83584374e-01 -1.47243306e-01
-1.86547741e-01 -1.04466712e+00 4.32704210e-01 6.24143779e-01
5.00706077e-01 -1.11302042e+00 2.58290887e-01 1.82905570e-01
-1.94722444e-01 -1.11735857e+00 -3.45720202e-01 5.85809469e-01
-7.80923247e-01 -8.80394101e-01 -5.26194513e-01 -1.01689994e+00
1.05074072e+00 1.50774136e-01 1.21285343e+00 -2.50212073e-01
2.29438588e-01 -2.16219023e-01 -6.00509584e-01 -4.18366879e-01
-5.17721057e-01 8.37172419e-02 1.21381715e-01 4.92220819e-02
8.17714870e-01 -2.58646935e-01 -2.02225298e-01 5.75549543e-01
-8.76267791e-01 3.92900817e-02 5.60016513e-01 1.22775769e+00
8.71384144e-01 1.07796520e-01 1.00096011e+00 -1.60083079e+00
2.83060461e-01 -8.00147891e-01 -6.97793007e-01 3.49207342e-01
-7.94974387e-01 2.88451046e-01 9.17886496e-01 -6.85026586e-01
-1.17963529e+00 5.59066057e-01 -3.24419439e-02 -4.07187790e-01
-3.72067958e-01 5.89744925e-01 -4.33784783e-01 2.30961576e-01
9.74623084e-01 -5.80761023e-02 -3.42895895e-01 -6.44916832e-01
4.28700686e-01 1.02527702e+00 2.69632131e-01 -9.74254072e-01
5.90494454e-01 2.30144352e-01 -2.92388290e-01 -7.93581009e-02
-1.79461169e+00 -2.85649627e-01 -7.60175645e-01 -8.91986415e-02
3.78704786e-01 -1.04362309e+00 -3.66614193e-01 4.54074591e-01
-7.74411917e-01 -4.53237474e-01 -5.07825136e-01 5.85341275e-01
-2.55617708e-01 1.69205830e-01 -6.71307087e-01 -7.18370497e-01
-1.46742091e-01 -1.35588849e+00 1.12892723e+00 3.24639201e-01
-1.17935285e-01 -8.02073121e-01 -6.82300404e-02 5.44573307e-01
-1.30506039e-01 2.65044272e-02 8.56752753e-01 -1.07977140e+00
-3.05626720e-01 -3.01644653e-01 -4.16446865e-01 7.57147670e-01
1.11469589e-01 -6.60527125e-02 -1.31384814e+00 -2.29011700e-01
-8.48886743e-02 -1.10730243e+00 6.66883051e-01 2.66764402e-01
1.37879622e+00 1.13606518e-02 -2.71183491e-01 2.81007469e-01
1.28263330e+00 7.64606595e-02 1.03110172e-01 -1.98280007e-01
9.30712879e-01 5.97206831e-01 1.05084395e+00 5.20099938e-01
3.83522749e-01 1.53762892e-01 2.20255047e-01 -4.03997600e-02
-7.30573535e-02 -4.53461915e-01 3.15778792e-01 9.84977007e-01
4.84975934e-01 -1.93363741e-01 -1.05119371e+00 2.78301716e-01
-1.76829839e+00 -3.23600709e-01 -3.91239703e-01 2.07449198e+00
1.37531137e+00 3.43349516e-01 -1.63172275e-01 3.62731874e-01
9.27772701e-01 -1.42821684e-01 -7.87410080e-01 2.20255796e-02
-9.21304002e-02 2.55777240e-01 5.55235088e-01 4.23628509e-01
-1.25383770e+00 1.03667426e+00 6.68358755e+00 1.20556116e+00
-8.12316358e-01 1.45301148e-01 1.12490034e+00 1.88292995e-01
-4.43675995e-01 -8.06003809e-02 -1.35814273e+00 6.03814483e-01
9.43916619e-01 2.42637739e-01 8.01322833e-02 1.17066312e+00
1.34354845e-01 -1.47964120e-01 -1.26051402e+00 1.09389985e+00
-1.03819005e-01 -1.12159848e+00 -4.81914133e-02 -1.14071734e-01
1.21950042e+00 6.13790825e-02 -5.07787578e-02 6.94168031e-01
1.02350223e+00 -9.43140924e-01 6.50370419e-01 1.46461084e-01
9.64038312e-01 -7.71375000e-01 9.52255428e-01 7.76692629e-01
-8.37280452e-01 -1.11240581e-01 -5.09165406e-01 1.69049695e-01
-1.12240799e-01 1.19682431e+00 -8.59857380e-01 1.88213333e-01
4.01265562e-01 6.94310188e-01 -5.43024361e-01 8.54290128e-01
-8.06512475e-01 1.18804669e+00 -3.42003942e-01 -6.82694241e-02
3.20741832e-01 -1.03311554e-01 -1.41066253e-01 1.03206444e+00
7.20103309e-02 7.83150569e-02 6.33604765e-01 1.00759745e+00
-3.37393522e-01 -1.74153689e-03 -3.95097703e-01 4.91135828e-02
7.96585143e-01 1.20302927e+00 -8.70580435e-01 -4.40511703e-01
-1.95778772e-01 4.12254721e-01 5.31636357e-01 3.94910753e-01
-8.58405828e-01 -9.04347971e-02 -1.36422411e-01 -3.37073356e-01
-6.38470203e-02 2.64916494e-02 -5.21823466e-01 -1.12673020e+00
-1.01632304e-01 -8.18236351e-01 3.75369996e-01 -6.39273465e-01
-1.68927729e+00 5.91847301e-01 6.70760944e-02 -1.43021691e+00
-2.31023878e-01 -5.61978996e-01 -2.19334796e-01 8.16277325e-01
-1.40266001e+00 -1.09395671e+00 -5.98942153e-02 3.10815692e-01
6.93115771e-01 -2.54505306e-01 8.78297925e-01 1.93957731e-01
-7.76859164e-01 7.06430197e-01 3.56978387e-01 2.51648933e-01
8.01797628e-01 -1.48762894e+00 2.37964302e-01 7.15730488e-01
4.01113957e-01 4.23743308e-01 3.30992341e-01 -5.30121565e-01
-1.07224917e+00 -1.23909461e+00 9.06011224e-01 -4.46215063e-01
2.62992263e-01 -4.15601254e-01 -8.54113698e-01 5.90337515e-01
-2.87499398e-01 5.66645920e-01 1.14396155e+00 1.24482363e-01
-5.97327292e-01 -1.15394697e-01 -1.33251321e+00 3.88078690e-01
7.24072635e-01 -3.39502543e-01 -6.36357248e-01 6.05859578e-01
8.70606542e-01 -3.14755023e-01 -5.49827754e-01 5.79321921e-01
1.75473079e-01 -7.19618082e-01 5.59361100e-01 -3.89526606e-01
3.51562798e-01 -5.39284170e-01 -2.13271201e-01 -1.43022537e+00
-9.07522365e-02 -1.45671234e-01 6.08235672e-02 1.65730059e+00
7.18574822e-01 -4.39685494e-01 1.13265145e+00 9.57371652e-01
-1.20943775e-02 -7.63820112e-01 -5.57959080e-01 -7.68072069e-01
-1.98078230e-02 -6.63441539e-01 5.55102110e-01 1.16985762e+00
-3.01778853e-01 4.55246091e-01 -2.16696441e-01 4.33169715e-02
8.80324543e-01 2.69794732e-01 5.43547809e-01 -1.43745077e+00
-3.00305486e-01 -2.94200927e-02 2.50628501e-01 -1.04510462e+00
6.74946129e-01 -9.74595428e-01 5.17217994e-01 -1.20266843e+00
2.10530177e-01 -1.25407696e+00 -5.06606162e-01 9.83527243e-01
-3.88117433e-01 5.05496144e-01 -1.23184122e-01 3.01275223e-01
-7.08896458e-01 4.70122963e-01 1.23029113e+00 -1.70480445e-01
-1.41531512e-01 3.29806358e-01 -8.07879627e-01 7.82048523e-01
7.66193390e-01 -8.33216190e-01 -6.73096716e-01 -3.63064975e-01
1.13263637e-01 -1.30787045e-01 -2.48081788e-01 -7.51069009e-01
-3.47122438e-02 -3.58867824e-01 2.97760844e-01 -7.11663842e-01
2.75369994e-02 -7.50712454e-01 -9.07940790e-02 1.08699307e-01
-7.60976613e-01 -7.15689242e-01 -2.92211711e-01 6.21533096e-01
-3.28302592e-01 -6.74685478e-01 8.36548090e-01 -3.56949605e-02
-6.10067248e-01 2.52955556e-01 -2.35583708e-01 4.13539946e-01
9.64475155e-01 1.01316214e-01 -2.29919493e-01 -4.22370844e-02
-8.43575358e-01 5.09538949e-01 1.43656433e-01 1.77356005e-01
4.40258950e-01 -1.45884597e+00 -6.14200830e-01 2.20344782e-01
3.37569267e-01 6.66806281e-01 -3.55119221e-02 1.27472326e-01
-1.98085010e-01 2.75656074e-01 2.25613266e-01 -7.64116764e-01
-6.90104604e-01 5.11949420e-01 -1.20928623e-01 -2.78178394e-01
-2.39826560e-01 1.12219012e+00 -2.23643258e-02 -9.27475691e-01
5.71465611e-01 -2.80557632e-01 -1.63626015e-01 1.36673510e-01
4.15289700e-01 1.61964610e-01 5.37995882e-02 -3.98860097e-01
-4.26277220e-02 3.37393880e-01 -1.97055891e-01 -2.99185701e-02
1.13270950e+00 3.27816494e-02 1.38286516e-01 7.50225902e-01
1.07298243e+00 -1.43905476e-01 -1.47802258e+00 -6.78323030e-01
3.07476103e-01 -4.08578128e-01 6.44577853e-03 -8.62113535e-01
-1.12328577e+00 8.79127741e-01 2.84338534e-01 -2.71000993e-02
8.56513977e-01 -6.89790957e-03 5.72788477e-01 4.97215033e-01
6.32317126e-01 -1.05986023e+00 1.83526397e-01 4.78265226e-01
3.86073977e-01 -1.69620228e+00 -1.76537007e-01 -6.35385990e-01
-9.60696578e-01 8.56738091e-01 9.13484275e-01 6.80487007e-02
8.01509142e-01 3.12277317e-01 2.85316378e-01 1.42370000e-01
-7.07664549e-01 -3.41148525e-02 1.40311763e-01 4.52787697e-01
3.81758422e-01 2.34549880e-01 -1.82765901e-01 9.24453557e-01
9.73435417e-02 1.65570796e-01 4.33134675e-01 1.00923371e+00
-4.58487749e-01 -1.45101333e+00 -4.16154236e-01 6.26947582e-01
-2.65871793e-01 -9.04318020e-02 -1.80428728e-01 2.99813509e-01
3.49127293e-01 1.13500011e+00 -3.15162808e-01 -3.76075000e-01
-8.82310644e-02 3.45481098e-01 7.42527097e-02 -1.24157655e+00
-2.31203660e-01 7.48521164e-02 -1.59461881e-04 -6.50390983e-02
-6.30296111e-01 -2.82050908e-01 -1.41262972e+00 1.41038179e-01
-6.16163552e-01 3.17930728e-01 5.12661874e-01 1.16108572e+00
1.48750126e-01 1.02482781e-01 1.14353108e+00 -6.22905433e-01
-8.54619801e-01 -9.87634718e-01 -9.23208654e-01 3.63503963e-01
-7.51301274e-02 -8.20371151e-01 -4.67533141e-01 1.57422468e-01]
|
[9.503409385681152, 3.821488380432129]
|
1c6194e8-a676-4ab1-8039-90a3151a6e7b
|
camouflaged-object-detection
| null | null |
http://openaccess.thecvf.com/content_CVPR_2020/html/Fan_Camouflaged_Object_Detection_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Fan_Camouflaged_Object_Detection_CVPR_2020_paper.pdf
|
Camouflaged Object Detection
|
We present a comprehensive study on a new task named camouflaged object detection (COD), which aims to identify objects that are "seamlessly" embedded in their surroundings. The high intrinsic similarities between the target object and the background make COD far more challenging than the traditional object detection task. To address this issue, we elaborately collect a novel dataset, called COD10K, which comprises 10,000 images covering camouflaged objects in various natural scenes, over 78 object categories. All the images are densely annotated with category, bounding-box, object-/instance-level, and matting-level labels. This dataset could serve as a catalyst for progressing many vision tasks, e.g., localization, segmentation, and alpha-matting, etc. In addition, we develop a simple but effective framework for COD, termed Search Identification Network (SINet). Without any bells and whistles, SINet outperforms various state-of-the-art object detection baselines on all datasets tested, making it a robust, general framework that can help facilitate future research in COD. Finally, we conduct a large-scale COD study, evaluating 13 cutting-edge models, providing some interesting findings, and showing several potential applications. Our research offers the community an opportunity to explore more in this new field. The code will be available at https://github.com/DengPingFan/SINet/.
|
[' Ling Shao', ' Jianbing Shen', ' Ming-Ming Cheng', ' Guolei Sun', ' Ge-Peng Ji', 'Deng-Ping Fan']
|
2020-06-01
| null | null | null |
cvpr-2020-6
|
['camouflaged-object-segmentation']
|
['computer-vision']
|
[ 2.17209503e-01 -4.58184749e-01 -3.57960165e-01 6.44253567e-02
-5.84927440e-01 -7.19989359e-01 5.59568405e-01 -3.19789857e-01
-1.68642089e-01 4.64040846e-01 -1.33960798e-01 -2.37612382e-01
2.43113771e-01 -3.88357043e-01 -6.39124811e-01 -9.47538853e-01
1.60208255e-01 6.80301636e-02 7.62752712e-01 1.75174475e-02
4.58678156e-01 4.91933435e-01 -1.63315618e+00 2.69452542e-01
7.67961800e-01 1.09022677e+00 5.09533644e-01 4.20885831e-01
1.90558180e-01 3.89208049e-01 -7.24503577e-01 -5.05033970e-01
3.90380502e-01 -1.43687084e-01 -5.08634031e-01 1.41392559e-01
8.44956219e-01 -2.71570891e-01 -3.84764701e-01 1.47278917e+00
4.28595394e-01 8.86183232e-02 5.86528480e-01 -1.45669186e+00
-9.37896907e-01 1.47435501e-01 -9.02192593e-01 3.55422884e-01
-1.45748667e-02 5.54599166e-01 7.67321527e-01 -1.29339898e+00
5.67195714e-01 1.24751759e+00 6.23324871e-01 5.93999028e-01
-8.80332291e-01 -9.51788962e-01 2.33121499e-01 3.41407508e-01
-1.32443368e+00 -2.73365140e-01 7.87199736e-01 -4.31254625e-01
3.70709121e-01 5.37914157e-01 6.78326368e-01 1.21900821e+00
3.13645974e-02 1.22767866e+00 1.18681574e+00 -2.64170051e-01
8.11572671e-02 1.04663931e-01 3.04310948e-01 7.63154268e-01
6.07329309e-01 2.84433931e-01 -3.25479060e-01 -3.40832137e-02
7.75710821e-01 5.56265377e-02 -4.82912123e-01 -2.47753456e-01
-1.29170001e+00 6.53600156e-01 7.58087397e-01 2.29336977e-01
4.92240302e-02 2.35843733e-01 3.91834915e-01 -1.54259369e-01
5.47170341e-01 4.97024953e-01 -2.54670888e-01 1.36799023e-01
-6.67312622e-01 1.53178617e-01 4.70303446e-01 9.39086914e-01
4.82851863e-01 -4.92092893e-02 -3.57616603e-01 7.95353234e-01
1.60924420e-01 5.69662690e-01 3.16121131e-01 -8.71813297e-01
1.52748972e-01 5.79808474e-01 2.29648843e-01 -1.16474557e+00
-6.32857233e-02 -5.05768359e-01 -8.05514157e-01 3.26114178e-01
3.56180161e-01 9.75564681e-03 -1.18361306e+00 1.34621489e+00
7.02885985e-01 7.38802135e-01 -4.03200239e-01 1.33526218e+00
1.06800747e+00 6.21747494e-01 -1.37873724e-01 3.56243812e-02
1.80355275e+00 -1.54450893e+00 -6.08977258e-01 -4.04444009e-01
2.81660557e-01 -1.06415975e+00 1.26506400e+00 5.43908894e-01
-8.10335040e-01 -6.45919859e-01 -9.45406318e-01 -1.20739423e-01
-6.68472886e-01 4.43875253e-01 9.54268277e-01 5.79237521e-01
-7.71857381e-01 1.80611536e-01 -6.49538934e-01 -2.62108535e-01
8.57150018e-01 -2.92615891e-02 -1.43494934e-01 -2.46291384e-01
-6.87347114e-01 7.90450454e-01 3.58795583e-01 3.19326192e-01
-1.19784021e+00 -5.69847763e-01 -7.44903982e-01 -2.02761769e-01
7.75501847e-01 -5.20189583e-01 1.17409861e+00 -8.26258242e-01
-9.41978395e-01 1.10171115e+00 -1.87814415e-01 -2.27895990e-01
4.80817944e-01 -3.21851403e-01 -5.43354094e-01 4.56932299e-02
2.83609301e-01 8.03137541e-01 1.06203163e+00 -1.51975060e+00
-7.19016433e-01 -2.20251948e-01 -2.46875975e-02 -6.53470084e-02
-2.84789711e-01 3.74428898e-01 -8.90028775e-01 -1.08176708e+00
-1.73536003e-01 -8.72867286e-01 -1.23809010e-01 4.35450137e-01
-7.53268540e-01 -3.79533201e-01 1.23619652e+00 -4.87820387e-01
1.19012141e+00 -2.34257293e+00 -3.59499292e-03 -2.41726264e-01
4.08640295e-01 8.27668965e-01 -3.25430274e-01 7.06473961e-02
3.61005031e-02 1.62358180e-01 -1.25303626e-01 -3.31860423e-01
-1.43058419e-01 -9.10632983e-02 -4.66425329e-01 6.12021744e-01
3.05009454e-01 1.24960721e+00 -8.95608902e-01 -4.48072821e-01
3.46974075e-01 3.00249308e-01 -1.19243339e-01 7.75759295e-02
-4.31541741e-01 3.88683915e-01 -4.66849506e-01 1.26819050e+00
9.64770079e-01 -3.69728953e-01 -5.36901474e-01 -3.63091111e-01
-2.39776999e-01 -1.20046087e-01 -1.01172256e+00 1.37875843e+00
8.66182595e-02 1.01198435e+00 1.75051168e-01 -7.62971401e-01
6.80418789e-01 -2.49607116e-01 8.90963972e-02 -5.91130018e-01
2.40169182e-01 2.46626899e-01 5.96248871e-03 -6.00455582e-01
4.61463004e-01 5.09788275e-01 1.01401269e-01 1.53638661e-01
-3.07704210e-01 -1.68038122e-02 2.32286677e-01 1.31017327e-01
8.75398815e-01 -2.51659423e-01 1.26017272e-01 -3.35604817e-01
3.33724141e-01 2.92555869e-01 5.39163768e-01 8.84350240e-01
-5.02331495e-01 6.99956894e-01 2.00404152e-01 -4.29347694e-01
-7.30324030e-01 -1.03632438e+00 -3.03055704e-01 8.71460795e-01
8.76746476e-01 -2.56748617e-01 -7.82974124e-01 -7.52595067e-01
2.44737476e-01 3.69524390e-01 -7.99583077e-01 -1.41895398e-01
-5.27786613e-01 -6.68631732e-01 6.36221349e-01 5.16091347e-01
1.02388060e+00 -1.18271959e+00 -4.13927346e-01 -3.78079206e-01
-2.92253494e-01 -1.05648553e+00 -1.01793087e+00 -7.66196772e-02
-5.43404937e-01 -1.32966375e+00 -8.86202633e-01 -1.10369027e+00
6.55666530e-01 1.14577997e+00 9.79316533e-01 4.32745904e-01
-7.78224647e-01 2.74590433e-01 -3.13570023e-01 -6.05137646e-01
2.77559515e-02 -2.63388753e-01 3.56454775e-02 2.11710762e-02
2.84209877e-01 8.11940953e-02 -9.24350679e-01 7.83231556e-01
-9.84353542e-01 9.72859412e-02 6.45950258e-01 7.90578067e-01
6.12639368e-01 1.53164238e-01 3.02949011e-01 -5.92496574e-01
3.91147345e-01 -3.73588592e-01 -8.56659651e-01 3.55251133e-01
-2.51549244e-01 -5.96846282e-01 1.53737575e-01 -8.25914264e-01
-8.88690650e-01 -2.48806238e-01 1.61238797e-02 -6.36250496e-01
-2.52868533e-01 8.36087987e-02 -2.42862463e-01 -5.58445752e-01
5.50737441e-01 2.06481442e-01 -2.50004679e-01 -6.18906140e-01
4.22104836e-01 6.91344798e-01 8.66922200e-01 -3.71585131e-01
9.98334289e-01 6.98768795e-01 -1.87713668e-01 -8.46403480e-01
-1.12611759e+00 -7.91533172e-01 -3.68997574e-01 -2.06220403e-01
7.51674950e-01 -8.11471581e-01 -6.68703854e-01 8.84291887e-01
-1.20881605e+00 -4.94318694e-01 -8.67116228e-02 1.53399512e-01
-2.09824443e-02 4.81410265e-01 -6.50079846e-01 -7.62917280e-01
-2.61880189e-01 -1.09009838e+00 1.34353411e+00 5.54927468e-01
1.79316521e-01 -8.50640953e-01 -2.73683876e-01 6.28936172e-01
2.79889524e-01 1.42433718e-01 5.62019229e-01 -3.86308372e-01
-1.02279830e+00 -5.63649014e-02 -7.04966247e-01 3.95568222e-01
9.25294831e-02 1.38280779e-01 -9.63873923e-01 -4.47845668e-01
-5.03234863e-02 -2.42003411e-01 1.33257997e+00 4.48838115e-01
1.58604527e+00 -1.99918911e-01 -7.60882616e-01 7.66661704e-01
1.31078815e+00 2.79433370e-01 6.68486118e-01 3.94461513e-01
8.03075671e-01 3.59957010e-01 1.06377923e+00 9.13662091e-02
-2.16379091e-02 8.52841616e-01 7.54631102e-01 -5.01786828e-01
-6.34955168e-01 -1.08428270e-01 3.54650080e-01 2.78994620e-01
1.42076295e-02 -4.96296167e-01 -7.12451816e-01 5.83457053e-01
-1.70154679e+00 -8.20499182e-01 -4.29778755e-01 1.85406935e+00
5.17798722e-01 1.01615131e-01 1.14235952e-01 -2.07728505e-01
9.74271297e-01 1.50282547e-01 -8.26265633e-01 2.41789922e-01
-3.78286391e-01 3.01211346e-02 4.72158939e-01 1.40702844e-01
-1.58057332e+00 1.25600719e+00 6.25734711e+00 1.41381180e+00
-1.18212140e+00 2.00408474e-01 7.58807838e-01 2.50269949e-01
6.60278425e-02 -1.05391078e-01 -1.03138709e+00 8.06765378e-01
8.54171216e-02 1.28688410e-01 3.52192223e-01 9.73231912e-01
1.71119735e-01 -3.61477852e-01 -7.64467120e-01 1.10030520e+00
2.36101598e-01 -1.56583226e+00 -1.24445543e-01 9.47073288e-03
8.99831772e-01 1.13431074e-01 4.38468784e-01 8.18457156e-02
1.56817943e-01 -9.93946612e-01 6.57762885e-01 8.77406076e-02
8.80989254e-01 -1.73736006e-01 6.39811993e-01 2.40536898e-01
-1.37266338e+00 -8.39813352e-02 -4.80150700e-01 1.14095077e-01
8.20760727e-02 6.43249571e-01 -5.05019903e-01 2.86010057e-01
9.68569815e-01 8.49430978e-01 -8.66455436e-01 1.74950588e+00
-2.64177173e-01 6.91874027e-01 -1.54380843e-01 -7.41894543e-03
1.52359307e-01 -1.60775781e-01 6.58168375e-01 1.31198478e+00
1.17102861e-01 -2.76592467e-02 2.87218928e-01 1.01212335e+00
-1.17608041e-01 -1.97331101e-01 -4.38412219e-01 -6.05039783e-02
5.49039900e-01 1.54622805e+00 -1.19028878e+00 -4.39295560e-01
-3.66564989e-01 1.03639340e+00 -3.34715620e-02 4.45924103e-01
-1.12441909e+00 -2.82835573e-01 8.85693789e-01 -8.67946967e-02
3.64959747e-01 -1.86362371e-01 -3.16732347e-01 -1.28615975e+00
1.20013736e-01 -9.46127117e-01 3.87665808e-01 -1.11322987e+00
-1.54713571e+00 3.73281837e-01 -5.36256284e-02 -1.31090462e+00
5.63401520e-01 -1.07494330e+00 -7.82857239e-01 4.64901477e-01
-1.51274598e+00 -1.23096943e+00 -6.93999350e-01 4.66578990e-01
8.14973295e-01 -6.10040091e-02 2.69607335e-01 4.24986392e-01
-9.11032140e-01 5.38444698e-01 7.80036524e-02 3.42198581e-01
8.11512530e-01 -9.39440250e-01 4.76500630e-01 1.01717782e+00
3.29520673e-01 5.61738908e-01 4.94008213e-01 -7.75681138e-01
-1.38127053e+00 -1.37081385e+00 2.60116011e-01 -6.63758099e-01
7.54426837e-01 -5.00166714e-01 -9.16550398e-01 4.73983139e-01
1.40286982e-01 2.51148552e-01 1.32383510e-01 -2.15289831e-01
-4.15183157e-01 2.05428712e-02 -9.23678935e-01 7.14689374e-01
1.46324110e+00 -2.51830637e-01 -3.28257561e-01 6.05965972e-01
1.00885046e+00 -5.76651931e-01 -3.02015245e-01 6.38498485e-01
3.95314157e-01 -9.42290664e-01 1.20704651e+00 -3.71394694e-01
2.65729427e-01 -5.00876367e-01 5.90673164e-02 -1.09534431e+00
-4.12132323e-01 -5.76122463e-01 -2.73004174e-01 1.27815795e+00
-1.43597303e-02 -7.29744136e-01 8.74647796e-01 1.88895926e-01
-5.10301590e-01 -8.62965941e-01 -7.64658391e-01 -9.48253632e-01
-1.63345128e-01 -3.64906609e-01 4.02796477e-01 9.26463246e-01
-6.45650208e-01 1.04006693e-01 -4.22947288e-01 2.33441457e-01
6.63308859e-01 3.06897104e-01 8.61935258e-01 -1.06247616e+00
-9.30776373e-02 -6.95396483e-01 -3.79133433e-01 -1.41588879e+00
-7.21718296e-02 -7.52525687e-01 3.09140924e-02 -1.20792150e+00
4.84880418e-01 -5.00914216e-01 -1.56178743e-01 5.65818131e-01
-4.05504286e-01 7.95078278e-01 9.49255228e-02 4.09478307e-01
-8.08919787e-01 5.10898769e-01 1.66865754e+00 -4.20062125e-01
9.80788842e-02 8.08659121e-02 -8.27063918e-01 8.44753504e-01
7.53591776e-01 -2.80069560e-01 -1.23001866e-01 -4.56361175e-01
-3.68380517e-01 -5.82707226e-01 9.67845380e-01 -7.58278668e-01
3.42362016e-01 -4.21459794e-01 4.34704989e-01 -8.13242853e-01
4.59726661e-01 -6.68834507e-01 2.68903263e-02 5.45814037e-01
1.21407486e-01 -1.82480007e-01 3.89543384e-01 5.63135028e-01
-1.75224051e-01 -2.40286633e-01 8.69125128e-01 -6.07093312e-02
-1.31124043e+00 3.99834365e-01 -5.63464016e-02 8.71281624e-02
1.37946486e+00 -3.79672885e-01 -1.00407732e+00 2.18908098e-02
-2.88631111e-01 2.90089190e-01 5.73286355e-01 6.10349715e-01
7.03095198e-01 -1.32638121e+00 -5.25055170e-01 2.52386868e-01
3.02844703e-01 9.81696248e-02 2.34457374e-01 8.68271291e-01
-4.01749581e-01 4.18830454e-01 -3.84769961e-02 -7.96006680e-01
-1.49682570e+00 7.51394272e-01 3.48489881e-01 2.53360569e-01
-6.55686557e-01 1.21264756e+00 6.57838643e-01 -1.14794690e-02
3.55490863e-01 -2.83307612e-01 1.04597569e-01 -2.08570391e-01
6.49804950e-01 5.10327518e-01 -1.54950097e-01 -5.14722645e-01
-3.62952799e-01 6.85511291e-01 -1.05399907e-01 3.61280799e-01
8.62935901e-01 1.56031242e-02 -3.11214775e-01 5.63907288e-02
9.30755615e-01 1.33656315e-03 -1.38126051e+00 -2.56671667e-01
-2.32645813e-02 -1.00567937e+00 -7.21942037e-02 -8.55545878e-01
-1.27048528e+00 7.61953115e-01 6.47337437e-01 1.46679834e-01
1.21976626e+00 4.36402410e-01 7.73329198e-01 2.14239314e-01
3.37522268e-01 -7.76332974e-01 6.66195333e-01 2.72225380e-01
9.31389630e-01 -1.46167409e+00 -1.22732513e-01 -8.88141632e-01
-4.67277795e-01 6.55541062e-01 9.84009326e-01 -2.31501579e-01
4.69354838e-01 2.73168951e-01 5.82757927e-02 -3.07588041e-01
-5.44634581e-01 -4.36476827e-01 5.89165211e-01 7.52069890e-01
-5.85004203e-02 8.42711851e-02 -1.13064334e-01 4.46694881e-01
1.10758737e-01 -2.62235790e-01 1.80123284e-01 5.12206912e-01
-6.69609845e-01 -8.63420784e-01 -6.76762879e-01 4.87202555e-01
-3.14622730e-01 -1.86518639e-01 -6.76497996e-01 9.01664138e-01
4.94928062e-01 9.75021362e-01 -4.94135208e-02 -3.56427521e-01
1.03482410e-01 -4.90461707e-01 3.73452693e-01 -5.70226789e-01
-2.64559805e-01 8.50562379e-02 -9.51422900e-02 -6.03023410e-01
-2.95721352e-01 -5.60096800e-01 -8.35511565e-01 -2.26527572e-01
-7.44887471e-01 -9.60901901e-02 4.96682614e-01 6.51963294e-01
3.61504763e-01 3.05189490e-01 3.27246100e-01 -1.03832173e+00
-2.95192361e-01 -8.00659716e-01 -4.33824182e-01 3.84896040e-01
4.20570791e-01 -1.05774128e+00 -4.73096043e-01 -1.96433510e-03]
|
[9.653617858886719, -0.19412729144096375]
|
7150c21f-dab0-4d1c-8910-cc8667d20dbb
|
androdet-an-adaptive-android-obfuscation
| null | null |
https://0m1d.com/assets/pdf/J5.pdf
|
https://0m1d.com/assets/pdf/J5.pdf
|
AndrODet: An Adaptive Android Obfuscation Detector
|
Obfuscation techniques modify an app’s source (or machine) code in order to make it more difficult to analyze. This is typically applied to protect intellectual property in benign apps, or to hinder the process of extracting actionable information in the case malware. Since malware analysis often requires considerable resource investment, detecting the particular obfuscation technique used may contribute to apply the right analysis tools, thus leading to some savings. In this paper, we propose AndrODet, a mechanism to detect three popular types of obfuscation in Android applications, namely identifier renaming, string encryption, and control flow obfuscation.
AndrODet leverages online learning techniques, thus being suitable for resource-limited environments that need to operate in a continuous manner. We compare our results with a batch learning algorithm using a dataset of 34,962 apps from both malware and benign apps. Experimental results show that online learning approaches are not only able to compete with batch learning methods in terms of accuracy, but they also save significant amount of time and computational resources. Particularly, AndrODet achieves an accuracy of 92.02% for identifier renaming detection, 81.41% for string encryption detection, and 68.32% for control flow obfuscation detection, on average. Also, the overall accuracy of the system when apps might be obfuscated with more than one technique is around 80.66%.
|
['Lorena Gonzáles-Manzano', 'Juan Tapiador', 'Jose Maria de Fuentes', 'Omid Mirzaei']
|
2019-01-01
| null | null | null |
future-generation-computer-systems-2019-1
|
['android-malware-detection']
|
['miscellaneous']
|
[ 2.44234726e-01 -3.03095520e-01 -8.62134516e-01 2.19579428e-01
-6.42211556e-01 -8.84738803e-01 3.94975543e-01 2.30176613e-01
-1.60534650e-01 4.89510864e-01 -4.98210371e-01 -1.09727204e+00
3.11405659e-01 -7.57138789e-01 -8.32410812e-01 -3.78330469e-01
-3.42443287e-01 -9.84307304e-02 3.08372736e-01 2.41239205e-01
4.17853326e-01 4.30745244e-01 -1.52339280e+00 6.04551554e-01
7.59892225e-01 1.10998106e+00 -2.22551242e-01 9.19445753e-01
-3.78723294e-01 9.66095507e-01 -9.87956762e-01 -8.76134932e-01
3.61382484e-01 -1.92970037e-01 -6.35692358e-01 -3.93283308e-01
4.85446334e-01 -6.84960127e-01 -6.90588430e-02 1.14170456e+00
1.25538051e-01 -4.64752197e-01 4.70916480e-01 -1.26725686e+00
-2.35286608e-01 6.03805244e-01 -8.53349209e-01 4.22038525e-01
5.03270030e-01 2.93305993e-01 8.61774445e-01 -3.01635802e-01
3.22830439e-01 8.51146698e-01 6.75928533e-01 6.15206897e-01
-1.07405257e+00 -1.11450195e+00 -2.71430612e-01 3.55723709e-01
-1.09237242e+00 -5.48494399e-01 6.69330060e-01 -6.41051054e-01
7.75814056e-01 6.01861835e-01 5.09647191e-01 1.17533004e+00
5.44988453e-01 6.25203967e-01 1.18492746e+00 -1.92190722e-01
4.75167394e-01 2.83486217e-01 1.35228321e-01 6.36926532e-01
6.48163319e-01 1.57586396e-01 -2.43434265e-01 -7.81549394e-01
9.17919353e-02 4.51511025e-01 7.28386119e-02 -6.08777590e-02
-5.41054785e-01 8.77796888e-01 -1.38473660e-02 5.90780303e-02
1.72414444e-02 -1.85711518e-01 8.86995614e-01 3.80356193e-01
2.28776932e-01 7.16712356e-01 -6.84176922e-01 -7.64078677e-01
-9.88452375e-01 -1.04323268e-01 9.93794322e-01 6.06850386e-01
8.72735083e-01 -2.65388354e-03 2.03395396e-01 5.90046048e-01
6.34349361e-02 4.96324599e-01 7.88166583e-01 -6.58309579e-01
5.55833697e-01 8.53261530e-01 -1.58029497e-01 -1.05797946e+00
1.66851148e-01 1.53825358e-01 -4.83200401e-01 2.43837014e-01
6.09169304e-01 -1.76179171e-01 -6.59464836e-01 1.05787730e+00
2.44252369e-01 3.11848789e-01 -2.24335983e-01 1.32257402e-01
2.84134984e-01 7.36456394e-01 -1.96130690e-03 -2.50106603e-01
1.59146190e+00 -7.60113001e-01 -6.03724241e-01 -2.24587172e-01
9.56289053e-01 -8.08175802e-01 1.10502684e+00 4.68219638e-01
-6.72311664e-01 -3.16853940e-01 -1.09516692e+00 3.89477879e-01
-7.29130983e-01 3.55979763e-02 5.86311221e-01 1.40879488e+00
-5.61989069e-01 4.06263471e-01 -9.57987249e-01 1.88965276e-01
8.59053016e-01 5.11375546e-01 -9.73285437e-02 2.52758622e-01
-6.65105462e-01 4.91158068e-01 1.40203416e-01 -5.23397446e-01
-9.09285009e-01 -9.58307683e-01 -6.24078751e-01 1.61005914e-01
6.31195962e-01 2.70627379e-01 1.32076156e+00 -9.94911849e-01
-1.46961701e+00 6.12591207e-01 -1.52547732e-01 -7.60485351e-01
4.51360106e-01 -1.76174402e-01 -6.92054212e-01 3.69244032e-02
-1.76252544e-01 -1.60446614e-02 1.45098162e+00 -7.89202034e-01
-6.01513147e-01 -2.53696114e-01 2.22098991e-01 -7.72377014e-01
-8.50704193e-01 2.67297775e-01 8.34494829e-02 -5.82949340e-01
-7.79024661e-01 -1.08028245e+00 2.81750500e-01 -3.18352044e-01
-3.49821955e-01 5.96554251e-03 1.54975271e+00 -1.08478212e+00
1.67953026e+00 -2.19328952e+00 -6.15364492e-01 1.21850900e-01
3.53639811e-01 1.10430479e+00 1.72318250e-01 3.04337263e-01
-4.03601937e-02 7.78464973e-01 -2.58350611e-01 -8.06553513e-02
-4.19232219e-01 -4.21150401e-02 -5.83960652e-01 3.83630276e-01
1.37784883e-01 8.13447475e-01 -7.32777178e-01 -3.34396422e-01
2.66228197e-03 2.19188005e-01 -5.33784747e-01 1.58393383e-01
-3.50249857e-01 9.92810056e-02 -4.62628216e-01 9.90031481e-01
7.27683902e-01 -2.32397765e-01 4.43163425e-01 3.06606412e-01
-6.07089475e-02 5.49060345e-01 -7.63695300e-01 7.18532324e-01
-8.93257797e-01 8.76457334e-01 -4.09103185e-02 -5.92415512e-01
7.42470503e-01 5.83760217e-02 4.03149843e-01 -5.93106151e-01
2.93004930e-01 3.28316838e-01 2.86003262e-01 -5.33118486e-01
3.13662529e-01 6.22879744e-01 -2.13667620e-02 8.72665286e-01
-2.75397032e-01 4.97937739e-01 -7.50769451e-02 -9.95051581e-04
1.56745219e+00 -4.59671170e-01 6.92503929e-01 1.02969207e-01
8.83616745e-01 -3.18543524e-01 3.03429067e-01 7.31618166e-01
-3.31653118e-01 -2.54125118e-01 8.10762703e-01 -4.38673794e-01
-8.90404046e-01 -7.34898806e-01 1.36881322e-01 1.16005111e+00
-2.32533365e-02 -7.91999698e-01 -1.00156188e+00 -1.49340487e+00
4.62983958e-02 3.18944663e-01 -4.11153615e-01 -5.51126778e-01
-9.24624562e-01 -3.15395087e-01 8.57758105e-01 2.31106758e-01
7.53333211e-01 -8.86648476e-01 -7.23099887e-01 -1.18563071e-01
1.32504418e-01 -1.08246017e+00 -4.99070853e-01 -7.69046843e-02
-9.22069013e-01 -1.33056796e+00 -2.54620194e-01 -4.13367212e-01
3.55846256e-01 4.50229436e-01 7.33983457e-01 3.76173437e-01
-3.67557108e-01 1.00234449e-01 -4.36522365e-01 -4.25058037e-01
-9.69291031e-01 4.78657246e-01 -1.09251045e-01 1.32519186e-01
4.63287145e-01 -4.36114043e-01 -2.94564366e-01 5.27327478e-01
-8.87725294e-01 -8.44437838e-01 5.60788453e-01 5.90245962e-01
1.83164656e-01 3.08996141e-01 3.51851732e-01 -1.01674163e+00
4.60775375e-01 -6.89860582e-01 -8.83761764e-01 1.17606059e-01
-7.82929540e-01 -1.65268257e-01 1.17804873e+00 -1.09857070e+00
-7.32268691e-01 7.21879210e-03 -2.35843584e-01 -4.48696673e-01
-8.48103091e-02 -1.45847112e-01 -4.45850044e-01 -6.19130015e-01
7.33171940e-01 2.42390975e-01 3.50388318e-01 -4.38914984e-01
5.14478050e-02 1.07990015e+00 -3.15555334e-02 -2.03185096e-01
9.25497293e-01 4.69377190e-01 -2.76135325e-01 -9.26450849e-01
-3.39644402e-01 -4.69702870e-01 -2.54229873e-01 1.03162616e-01
5.88555336e-01 -6.00358963e-01 -1.07173109e+00 6.84346139e-01
-9.29597616e-01 -3.76673073e-01 6.10545464e-02 -4.34793979e-02
-1.56748161e-01 6.94743633e-01 -4.12411064e-01 -8.58921826e-01
-3.96080285e-01 -1.32676804e+00 9.03835952e-01 1.69315621e-01
-2.66170233e-01 -9.20040250e-01 -1.68839589e-01 4.76991326e-01
4.46848929e-01 5.70386834e-02 8.38660002e-01 -1.08059943e+00
-5.46355486e-01 -5.73931158e-01 -1.33379608e-01 5.46824813e-01
5.55828929e-01 1.94026381e-01 -1.07168329e+00 -3.78616333e-01
2.18519866e-01 1.19821699e-02 4.19481069e-01 -1.98515564e-01
1.73656237e+00 -1.04785430e+00 -4.15014803e-01 5.32350779e-01
1.14884174e+00 5.96858084e-01 9.01023865e-01 3.81453663e-01
8.54236662e-01 2.88284779e-01 6.65157855e-01 3.86633039e-01
-4.66569662e-02 6.66111052e-01 5.93699098e-01 4.46723521e-01
2.26006247e-02 -4.77882713e-01 9.10360992e-01 3.68125856e-01
2.13714764e-01 2.66865082e-02 -8.69556725e-01 2.69624978e-01
-1.33078742e+00 -1.00924706e+00 -2.17042323e-02 2.55786562e+00
9.42854047e-01 3.17683876e-01 6.65396929e-01 4.14119571e-01
7.28971183e-01 2.52026439e-01 -5.62014163e-01 -1.08579969e+00
5.06028831e-01 4.50776309e-01 1.00200462e+00 1.89617962e-01
-1.13106215e+00 6.93824053e-01 5.68510818e+00 1.25081813e+00
-1.49153292e+00 3.42528611e-01 6.43059611e-01 2.13041484e-01
1.27360687e-01 4.11660364e-03 -1.01096904e+00 1.21668577e+00
1.46725523e+00 9.61716846e-02 7.20599830e-01 1.29539371e+00
-1.28000854e-02 -9.70172882e-02 -8.15818071e-01 1.15648401e+00
3.11154891e-02 -1.43589556e+00 -2.73802191e-01 6.25743687e-01
5.30134916e-01 -3.84687006e-01 9.11947489e-02 3.95635545e-01
-5.56183606e-02 -8.48397970e-01 3.27837825e-01 -2.29922831e-01
7.03440368e-01 -8.52099061e-01 5.85456967e-01 5.45969129e-01
-1.13566375e+00 -6.60302579e-01 -1.65473148e-01 -1.46538049e-01
-2.04219744e-01 5.71467519e-01 -1.15363765e+00 3.13144289e-02
7.83140063e-01 5.26362002e-01 -9.15051997e-01 9.30445969e-01
-2.11947620e-01 1.23276591e+00 -3.00384294e-02 -4.01405543e-01
-1.65364951e-01 9.74152312e-02 4.98674750e-01 1.13456964e+00
1.65859818e-01 -5.09841084e-01 -2.28722114e-02 4.91882861e-01
-3.45345140e-01 1.64653823e-01 -8.49082649e-01 -4.95216221e-01
7.00753033e-01 1.31803179e+00 -5.31161129e-01 -3.28337580e-01
-2.78848052e-01 7.72536933e-01 1.34238303e-01 -1.16488114e-01
-1.05801582e+00 -4.80629593e-01 1.00075018e+00 4.78980899e-01
6.37644529e-01 -2.85571218e-01 -3.42226177e-01 -9.90445375e-01
1.95839822e-01 -1.48222375e+00 3.68340284e-01 9.87051651e-02
-9.77446139e-01 1.97231606e-01 -1.47078246e-01 -1.29576540e+00
-4.01356220e-01 -7.94472814e-01 -8.59087169e-01 2.63393730e-01
-1.38243580e+00 -9.23998833e-01 -3.72637734e-02 3.29320103e-01
3.94205511e-01 -3.68595272e-01 5.64464986e-01 5.25138199e-01
-6.00884140e-01 1.15011990e+00 1.33806132e-02 2.48690397e-01
5.23971319e-01 -1.00417817e+00 5.74758291e-01 8.22607219e-01
1.21511512e-01 9.65445518e-01 2.84504175e-01 -9.54291344e-01
-1.77606583e+00 -1.20868516e+00 4.45609629e-01 -4.85974252e-01
9.75754023e-01 -7.00143099e-01 -1.10455811e+00 5.01843691e-01
-1.73008412e-01 2.06918940e-02 9.74043190e-01 -6.03427477e-02
-8.49332690e-01 -2.70268023e-01 -1.31912458e+00 5.45507073e-01
7.17463732e-01 -7.18116879e-01 8.96894634e-02 2.17851400e-01
5.62310994e-01 -1.82290897e-01 -7.33661354e-01 1.41140804e-01
8.02949607e-01 -9.15033579e-01 9.84620214e-01 -6.36483610e-01
2.71659225e-01 -4.21056598e-02 1.36288404e-01 -5.52724183e-01
2.19318405e-01 -1.06412077e+00 -1.20476246e+00 1.40455830e+00
4.14406806e-01 -9.95647013e-01 9.18737829e-01 1.37665853e-01
4.98624682e-01 -8.41900289e-01 -8.78742993e-01 -1.30453551e+00
-1.59561802e-02 -5.30094683e-01 9.43666220e-01 9.13165927e-01
-3.25173914e-01 -1.22353286e-01 -4.86764491e-01 5.36835194e-03
4.06241775e-01 -1.71718895e-01 1.09706032e+00 -8.76432836e-01
-5.65055728e-01 -4.52202469e-01 -4.80356216e-01 -6.99984789e-01
3.96275192e-01 -5.84394693e-01 -4.17482525e-01 -4.00214851e-01
-6.52181134e-02 -5.35305977e-01 6.78994507e-02 6.50850713e-01
-1.83303021e-02 3.30367655e-01 2.58629899e-02 4.09682870e-01
-2.34702110e-01 -1.45048007e-01 6.44688845e-01 -2.61498123e-01
-4.82389659e-01 6.44601882e-01 -7.11465955e-01 8.55663538e-01
1.07123101e+00 -5.28951406e-01 -3.54458481e-01 4.19769101e-02
1.89929157e-01 -4.09243524e-01 4.25590634e-01 -7.75783002e-01
-1.91966504e-01 -1.24197856e-01 7.00208545e-02 -1.11601830e-01
6.90371394e-02 -9.16401625e-01 -1.96082294e-01 8.15680385e-01
4.67644073e-02 1.18131027e-01 3.65904182e-01 6.28499746e-01
2.38923848e-01 -4.08480912e-01 8.23369026e-01 5.32487184e-02
-6.46650970e-01 1.54586568e-01 -6.03253126e-01 -1.32804796e-01
1.34938276e+00 -4.97683108e-01 -5.46253204e-01 -1.42523974e-01
-1.24754220e-01 -4.16355550e-01 7.27218747e-01 5.39961100e-01
3.98945689e-01 -7.66013920e-01 1.13626830e-02 3.65150064e-01
4.93798219e-02 -8.02597046e-01 9.38560143e-02 6.89647794e-01
-4.80877042e-01 1.41515344e-01 -2.03686610e-01 -4.61415917e-01
-1.79258049e+00 9.38173056e-01 2.28291899e-02 -4.48551655e-01
-3.62526357e-01 3.74380708e-01 -4.90960538e-01 -1.01303555e-01
1.43777132e-01 -1.43655434e-01 -8.33621696e-02 7.09943548e-02
1.00438988e+00 8.61887634e-01 3.19478899e-01 -4.18522328e-01
-5.94441116e-01 3.18820566e-01 -5.18561900e-01 4.91187930e-01
7.05788374e-01 2.48108923e-01 -2.70318091e-01 6.90877512e-02
1.57129705e+00 8.34144711e-01 -8.59101534e-01 2.48704940e-01
2.64153361e-01 -8.39161038e-01 -2.96119273e-01 -7.34892428e-01
-9.36688840e-01 8.75311255e-01 8.57189178e-01 7.52126575e-01
1.11726379e+00 -1.11582607e-01 1.20547354e+00 2.08766565e-01
5.28407633e-01 -6.15689695e-01 3.37992236e-02 1.91142201e-01
8.46357197e-02 -1.28933620e+00 1.92061543e-01 -6.14013016e-01
-1.89246923e-01 1.09157670e+00 5.24972200e-01 -5.81224896e-02
6.15065157e-01 4.38494086e-01 -2.65005022e-01 3.93213183e-02
-4.45418984e-01 3.84422421e-01 2.15915591e-01 7.45727479e-01
1.56107461e-02 1.76441193e-01 -2.94846028e-01 3.63049865e-01
-1.50523409e-01 -3.30762565e-01 6.74154699e-01 1.16553009e+00
-2.74011612e-01 -1.51727164e+00 -4.33522999e-01 7.44339347e-01
-9.69679952e-01 8.34152848e-02 -7.39011645e-01 6.82674527e-01
3.14904809e-01 1.08106947e+00 -1.43421441e-01 -7.96462059e-01
-1.19547606e-01 -9.10853893e-02 7.80608803e-02 -6.19172215e-01
-8.03981423e-01 -3.46521169e-01 5.27244024e-02 -5.45967340e-01
1.43739998e-01 -6.13384902e-01 -8.92323792e-01 -5.28302193e-01
-3.47854674e-01 -4.06624004e-02 8.80647540e-01 8.60394239e-01
7.70822942e-01 2.04586446e-01 1.03915155e+00 -4.72698569e-01
-5.72037041e-01 -6.02336168e-01 -3.11374664e-02 -7.37638026e-02
4.86737370e-01 -7.58772016e-01 -5.29210746e-01 1.36074990e-01]
|
[14.417659759521484, 9.678044319152832]
|
2d94aea9-ca92-4318-9ff1-80be7ecf4254
|
pushing-one-pair-of-labels-apart-each-time-in
|
2302.14695
| null |
https://arxiv.org/abs/2302.14695v1
|
https://arxiv.org/pdf/2302.14695v1.pdf
|
Pushing One Pair of Labels Apart Each Time in Multi-Label Learning: From Single Positive to Full Labels
|
In Multi-Label Learning (MLL), it is extremely challenging to accurately annotate every appearing object due to expensive costs and limited knowledge. When facing such a challenge, a more practical and cheaper alternative should be Single Positive Multi-Label Learning (SPMLL), where only one positive label needs to be provided per sample. Existing SPMLL methods usually assume unknown labels as negatives, which inevitably introduces false negatives as noisy labels. More seriously, Binary Cross Entropy (BCE) loss is often used for training, which is notoriously not robust to noisy labels. To mitigate this issue, we customize an objective function for SPMLL by pushing only one pair of labels apart each time to prevent the domination of negative labels, which is the main culprit of fitting noisy labels in SPMLL. To further combat such noisy labels, we explore the high-rankness of label matrix, which can also push apart different labels. By directly extending from SPMLL to MLL with full labels, a unified loss applicable to both settings is derived. Experiments on real datasets demonstrate that the proposed loss not only performs more robustly to noisy labels for SPMLL but also works well for full labels. Besides, we empirically discover that high-rankness can mitigate the dramatic performance drop in SPMLL. Most surprisingly, even without any regularization or fine-tuned label correction, only adopting our loss defeats state-of-the-art SPMLL methods on CUB, a dataset that severely lacks labels.
|
['Songcan Chen', 'Xinrui Wang', 'Xiang Li']
|
2023-02-28
| null | null | null | null |
['multi-label-learning']
|
['methodology']
|
[ 4.10544097e-01 -1.17130987e-01 -2.73254752e-01 -5.71764827e-01
-1.19465911e+00 -6.66332662e-01 2.71449327e-01 2.90610760e-01
-4.45781380e-01 7.82689512e-01 -2.46373162e-01 -4.40599546e-02
-1.40097380e-01 -5.15421689e-01 -5.44268787e-01 -1.07172704e+00
3.39776933e-01 2.97243178e-01 1.22420117e-01 2.25479245e-01
-1.70673624e-01 1.39017418e-01 -1.39243245e+00 2.30272010e-01
8.08384478e-01 1.07405913e+00 7.81051666e-02 4.36181501e-02
2.77899280e-02 6.84051096e-01 -5.56266546e-01 -6.43525720e-01
3.83567929e-01 -3.04184586e-01 -8.29204917e-01 2.43927911e-01
5.60276747e-01 -6.53168410e-02 1.48375273e-01 1.34163058e+00
4.51355040e-01 -2.92550772e-02 6.85474217e-01 -1.31838739e+00
-4.25621897e-01 6.72556818e-01 -1.04390526e+00 -2.64232725e-01
-9.87343024e-03 9.61151719e-02 1.42084634e+00 -9.50989902e-01
1.51727289e-01 1.46925712e+00 9.21945035e-01 4.70065385e-01
-1.44615400e+00 -9.45621490e-01 2.89199531e-01 -8.12169984e-02
-1.58631277e+00 -9.68111828e-02 7.42517114e-01 -2.77152568e-01
2.04726070e-01 4.48746592e-01 -5.19321896e-02 1.08258426e+00
-8.91151931e-03 1.00232041e+00 1.46915638e+00 -3.48592550e-01
-6.60970341e-03 2.98951954e-01 3.61665130e-01 5.18392742e-01
2.25745216e-01 -1.91153616e-01 -2.12407649e-01 -3.68281245e-01
2.71524817e-01 8.29852000e-02 -2.16810167e-01 -2.59917587e-01
-1.16090238e+00 7.20355093e-01 1.46837637e-01 1.96613908e-01
6.79701716e-02 1.02944188e-01 6.09313130e-01 2.22687826e-01
5.67591310e-01 2.96904385e-01 -6.82574868e-01 2.26916224e-01
-7.99645960e-01 7.58330524e-02 4.43924516e-01 7.89743662e-01
9.07908440e-01 -4.51036662e-01 -4.06950831e-01 1.22712195e+00
3.33502620e-01 4.35732067e-01 3.20732027e-01 -8.15380871e-01
5.56014895e-01 5.21568656e-01 1.80056661e-01 -9.92715418e-01
-6.05552912e-01 -7.54233539e-01 -1.08447886e+00 -8.98314491e-02
5.93949676e-01 -3.59823965e-02 -6.02951527e-01 1.96568918e+00
4.72710520e-01 7.89324790e-02 -2.27437526e-01 9.01516259e-01
6.08392179e-01 3.44207853e-01 2.96892792e-01 -4.62566763e-01
1.42112637e+00 -9.13891494e-01 -6.87551022e-01 -2.95257390e-01
1.02082038e+00 -6.13390505e-01 1.36075580e+00 4.25220817e-01
-5.54336250e-01 -2.53892243e-01 -8.33511233e-01 2.33248454e-02
-1.41511545e-01 3.28788877e-01 5.71033299e-01 7.88947463e-01
-6.19130611e-01 4.83363271e-01 -4.86137956e-01 1.53558617e-02
5.32028377e-01 3.78246546e-01 -3.24675113e-01 -3.71220350e-01
-1.27552700e+00 6.71371818e-01 3.49528193e-01 2.01882005e-01
-4.91927147e-01 -6.12291574e-01 -6.77293479e-01 -7.47332210e-03
8.22741151e-01 -1.35491073e-01 1.07741237e+00 -6.94407344e-01
-1.08881426e+00 8.44519615e-01 -1.23539500e-01 -1.10186443e-01
7.36619651e-01 -8.32863972e-02 -3.57499897e-01 -2.14434892e-01
3.84525180e-01 4.81608510e-01 7.40033269e-01 -1.69002438e+00
-5.99407673e-01 -1.47247404e-01 8.84943828e-02 8.19823518e-02
-6.37272656e-01 -5.64467832e-02 -2.77491599e-01 -6.74701333e-01
2.70794809e-01 -9.29633617e-01 -2.38493010e-01 -6.76542521e-02
-6.74354374e-01 -6.59770548e-01 8.44465137e-01 -2.78737247e-01
1.25942731e+00 -2.19340396e+00 -1.86260685e-01 2.86344975e-01
1.91681013e-01 2.00316921e-01 -2.93151408e-01 -1.88891254e-02
-8.24326351e-02 3.43920082e-01 -4.05263186e-01 -7.54952133e-01
2.11714488e-02 3.67025554e-01 -1.08259000e-01 5.72984397e-01
2.77653992e-01 7.18617201e-01 -1.18634236e+00 -6.66585863e-01
-9.36765075e-02 2.55743831e-01 -3.70543271e-01 -1.98065504e-01
-1.35701805e-01 6.27833188e-01 -4.45217401e-01 7.78915823e-01
9.03330505e-01 -7.64078140e-01 1.71762004e-01 -3.47962916e-01
3.23436320e-01 -1.05682924e-01 -1.48876882e+00 1.27433181e+00
-6.27833664e-01 1.41776714e-03 1.67250093e-02 -1.08321655e+00
7.09585309e-01 2.83224642e-01 5.76108456e-01 -3.60360414e-01
1.38399258e-01 3.34632009e-01 -3.96301985e-01 -2.94782430e-01
4.13722955e-02 -4.74383831e-01 -2.51959175e-01 3.81575435e-01
-1.98823258e-01 2.13629514e-01 2.10035086e-01 -4.70453985e-02
1.05680346e+00 -5.04818112e-02 2.62763590e-01 -1.15127616e-01
6.19800746e-01 -5.28695405e-01 1.17188966e+00 7.46004522e-01
-3.76478255e-01 7.04282224e-01 5.34516573e-01 -3.90672125e-02
-6.66725397e-01 -7.57259488e-01 -5.28042495e-01 1.24642217e+00
2.82619387e-01 -2.77083784e-01 -3.74200463e-01 -1.25138116e+00
8.22669417e-02 4.47164625e-01 -3.38883281e-01 -5.01304530e-02
-4.09122199e-01 -1.45619786e+00 5.73001683e-01 1.54496342e-01
4.18872446e-01 -6.24905050e-01 3.78451287e-03 2.08494753e-01
-5.11595786e-01 -1.19394886e+00 -6.32299662e-01 4.68116999e-01
-5.06183624e-01 -1.02344000e+00 -6.03543937e-01 -6.38688147e-01
8.02810788e-01 2.75254369e-01 9.59796011e-01 4.41109166e-02
2.60674153e-02 -2.34825630e-02 -3.65647435e-01 -1.63763598e-01
-4.95404571e-01 9.94009525e-02 8.93958062e-02 3.61150533e-01
3.24582905e-01 -4.51670855e-01 -3.23147386e-01 6.37734294e-01
-1.03396142e+00 -1.13485679e-01 5.28591037e-01 1.20279479e+00
8.01315486e-01 4.21330392e-01 9.31701064e-01 -1.19877779e+00
3.18341434e-01 -5.64657092e-01 -4.89053190e-01 5.30506551e-01
-7.47642457e-01 2.88641802e-03 7.91986406e-01 -7.07234085e-01
-7.53489912e-01 2.71427453e-01 -1.38439104e-01 -3.77368301e-01
-1.09187298e-01 2.84886360e-01 -3.63975972e-01 -1.44425184e-01
4.25131679e-01 -1.15543820e-01 -2.79728383e-01 -6.53417528e-01
2.47964159e-01 7.95660198e-01 1.37109756e-01 -5.28907359e-01
6.50362134e-01 4.41265643e-01 3.54116827e-01 -5.10289788e-01
-1.56262648e+00 -8.32655370e-01 -4.96702492e-01 -7.66081288e-02
5.15338242e-01 -9.19008851e-01 -7.96014965e-01 5.01446605e-01
-9.72260714e-01 -2.97946744e-02 -1.41731501e-01 3.07068110e-01
-2.00344846e-01 7.69709527e-01 -7.37877488e-01 -9.03388619e-01
-1.17680691e-01 -1.29007256e+00 1.17342937e+00 -9.36140586e-03
-4.54751477e-02 -9.58337843e-01 -2.38289967e-01 6.27564549e-01
7.24924505e-02 2.56716311e-01 8.78648043e-01 -6.57972753e-01
-2.18397379e-01 -2.30792880e-01 -5.09345531e-01 7.85511017e-01
1.99456692e-01 -3.84810865e-01 -1.13513362e+00 -5.43772221e-01
1.43370271e-01 -7.94362426e-01 9.87460315e-01 5.30644208e-02
1.27949595e+00 -2.38830566e-01 -3.53842288e-01 2.82969862e-01
1.48876560e+00 -1.50268778e-01 7.82529786e-02 6.60192072e-02
1.02179980e+00 6.64860666e-01 8.43940854e-01 4.44827110e-01
5.64678073e-01 6.90340757e-01 5.48713744e-01 -1.29487097e-01
3.23920213e-02 2.49544680e-02 2.30891749e-01 9.13430512e-01
3.80306661e-01 -4.32031095e-01 -7.59889424e-01 3.48248452e-01
-1.78310430e+00 -5.39027035e-01 -3.29477221e-01 2.26687813e+00
1.27944803e+00 3.04015204e-02 -6.84589194e-03 4.43450421e-01
8.57238889e-01 1.24872953e-01 -6.38140321e-01 2.24439055e-01
-4.27846342e-01 -3.37815471e-02 6.66702271e-01 2.42818773e-01
-1.58129251e+00 6.50610209e-01 5.49454737e+00 1.37296271e+00
-9.66212392e-01 5.96499801e-01 8.48673820e-01 -7.32580721e-02
-2.05420256e-01 -3.68247852e-02 -1.07578921e+00 7.14358985e-01
6.45613372e-01 3.61603886e-01 6.37486950e-02 7.84920752e-01
7.73805305e-02 -1.16531812e-01 -9.71894443e-01 1.01456511e+00
-7.72788078e-02 -6.96193576e-01 -1.98574528e-01 -1.09500615e-02
8.23087096e-01 -1.27954155e-01 1.06399901e-01 3.69690597e-01
3.76091301e-01 -7.88593650e-01 7.42752254e-01 1.54819772e-01
8.70564938e-01 -6.73471093e-01 9.17562902e-01 5.95631480e-01
-1.20684588e+00 -2.55930603e-01 -4.97875035e-01 9.82226729e-02
1.29438668e-01 1.27188957e+00 -3.52110624e-01 5.35951138e-01
5.10911822e-01 6.71235263e-01 -6.20598793e-01 1.03745854e+00
-2.09362239e-01 5.77372909e-01 -4.80467439e-01 2.21651942e-01
1.33949950e-01 -5.00782616e-02 2.64359206e-01 1.13426208e+00
1.84939250e-01 -1.43619254e-01 6.92077160e-01 5.29089332e-01
-3.80388260e-01 3.21116745e-01 -3.09429646e-01 2.67685235e-01
5.23961902e-01 1.49759138e+00 -7.73702383e-01 -1.66402489e-01
-6.20985270e-01 9.23652887e-01 4.28378671e-01 3.04835796e-01
-8.38729024e-01 -1.41408714e-02 4.00693089e-01 -1.47084013e-01
2.94421054e-02 1.95469260e-01 -5.35538554e-01 -1.17142236e+00
2.09779456e-01 -7.37328827e-01 3.88569444e-01 -1.30194783e-01
-1.87553310e+00 2.44798779e-01 -1.71070367e-01 -1.44948435e+00
2.16166615e-01 -3.00041199e-01 -1.17884062e-01 7.54270732e-01
-1.64473474e+00 -1.29996693e+00 -7.83280283e-02 1.70742214e-01
3.02285969e-01 3.61418813e-01 6.20200276e-01 8.64497721e-01
-8.38281989e-01 1.06593239e+00 3.72202069e-01 -1.12508573e-01
1.14098859e+00 -1.23126864e+00 -1.62454143e-01 6.03164554e-01
8.21151510e-02 2.67513037e-01 5.52928269e-01 -4.75636452e-01
-8.02168667e-01 -1.45601428e+00 8.84202242e-01 -4.54084694e-01
6.19504869e-01 -4.63003546e-01 -1.13558125e+00 4.32442218e-01
-4.76438761e-01 3.49460036e-01 7.95197248e-01 3.55389901e-02
-5.60734808e-01 -2.43555784e-01 -1.27698815e+00 3.63851696e-01
8.71609926e-01 -4.95709538e-01 -5.42640574e-02 8.07366192e-01
7.57466912e-01 -2.80444622e-01 -9.40516412e-01 6.94119513e-01
3.94938946e-01 -8.03568423e-01 8.39895487e-01 -7.05116317e-02
1.94682792e-01 -3.94707680e-01 -1.17865734e-01 -1.17831540e+00
-2.09783092e-01 -2.93496698e-01 1.09650642e-01 1.58997786e+00
5.03982127e-01 -6.49847865e-01 6.62639499e-01 5.23784637e-01
7.22995400e-02 -1.03130639e+00 -1.02472031e+00 -9.95505810e-01
1.51952848e-01 -4.80090529e-01 4.18687284e-01 1.27816117e+00
-3.19949538e-01 3.23704034e-01 -7.47793078e-01 2.88026154e-01
8.49452257e-01 -7.69645274e-02 4.03409719e-01 -1.47687113e+00
-3.48616898e-01 -3.28098506e-01 -8.82528797e-02 -1.00535953e+00
4.20708269e-01 -1.01836658e+00 2.64063239e-01 -1.24023867e+00
5.68093002e-01 -1.07461882e+00 -5.71752071e-01 8.50253284e-01
-4.94042397e-01 7.13403583e-01 2.06572428e-01 4.19404656e-01
-8.15368056e-01 4.93192762e-01 1.37614870e+00 -2.13059783e-01
5.47067076e-02 1.74913853e-01 -7.78035820e-01 7.85415530e-01
6.10044241e-01 -7.91702211e-01 -2.64726788e-01 -1.04589194e-01
3.58091474e-01 -2.36467734e-01 2.33744130e-01 -7.17483044e-01
3.54814753e-02 -1.80833399e-01 -6.40119314e-02 -4.72435743e-01
1.72975972e-01 -8.13526332e-01 5.82293719e-02 1.28838792e-01
-3.70606065e-01 -4.50977236e-01 -2.04916775e-01 7.60845184e-01
-2.42086545e-01 -4.23635274e-01 9.54827607e-01 3.34926434e-02
-3.67662936e-01 2.76205480e-01 9.54356343e-02 2.20965251e-01
1.06910217e+00 2.01484650e-01 -2.42343754e-01 2.69780625e-02
-6.34393394e-01 6.56888306e-01 3.60210836e-01 2.15439573e-01
1.22993283e-01 -1.48629212e+00 -5.50768077e-01 5.82247972e-02
2.13177264e-01 2.54130751e-01 3.36947113e-01 1.03392231e+00
1.74171895e-01 2.66674459e-01 4.04193282e-01 -6.18335128e-01
-1.33320522e+00 5.87786973e-01 1.48183778e-01 -7.02026248e-01
-3.37559730e-01 1.01301730e+00 3.72248173e-01 -6.34220302e-01
4.42410916e-01 -1.66408986e-01 -1.96658656e-01 3.74930590e-01
3.46334487e-01 3.48804206e-01 1.68024957e-01 -7.91369319e-01
-3.19669604e-01 7.11358011e-01 -2.69491136e-01 3.35958213e-01
9.35345769e-01 -4.21930373e-01 -2.08081916e-01 7.94725120e-01
1.29673290e+00 -1.04348205e-01 -1.29438293e+00 -5.32087326e-01
1.48728222e-01 -3.34714711e-01 -4.01086826e-03 -7.04395771e-01
-1.13101161e+00 8.39185357e-01 4.72851187e-01 2.74362534e-01
1.19497061e+00 -3.98178361e-02 9.67623770e-01 2.81729519e-01
6.12440586e-01 -1.08252585e+00 2.12851524e-01 3.90056044e-01
4.12022889e-01 -1.75726604e+00 -2.20823642e-02 -7.75658786e-01
-5.81322372e-01 6.95223987e-01 5.25021970e-01 3.65517408e-01
6.86783016e-01 1.02180272e-01 7.79347047e-02 7.22976625e-02
-4.69412386e-01 -2.29579229e-02 2.00705871e-01 1.57232910e-01
2.98032999e-01 1.51730433e-01 -6.01321101e-01 6.62494302e-01
2.87052900e-01 -3.30133766e-01 3.42947453e-01 7.22361863e-01
-1.14176467e-01 -1.43774760e+00 -4.34520960e-01 7.69782543e-01
-8.13158453e-01 6.36316240e-02 -1.82577576e-02 4.36780602e-01
5.43045223e-01 1.10358167e+00 -3.49842191e-01 -3.74351054e-01
3.01310867e-01 1.10877387e-01 1.92617074e-01 -7.06413507e-01
-6.11751676e-01 1.42029211e-01 -1.32255748e-01 -3.81948352e-01
-6.95448816e-01 -5.89754641e-01 -1.18192363e+00 -1.12910897e-01
-8.45897675e-01 -6.27192557e-02 3.11610103e-01 1.09765744e+00
-6.00826740e-03 3.61978859e-01 9.08266664e-01 -5.51499069e-01
-1.07370901e+00 -9.65269268e-01 -8.60186219e-01 6.78511739e-01
4.14288968e-01 -9.89199102e-01 -7.66113818e-01 -2.39653006e-01]
|
[9.439359664916992, 4.006559371948242]
|
1dc04d6f-783f-460f-9802-f89d582eb774
|
extracting-label-specific-key-input-features
|
2202.06474
| null |
https://arxiv.org/abs/2202.06474v1
|
https://arxiv.org/pdf/2202.06474v1.pdf
|
Extracting Label-specific Key Input Features for Neural Code Intelligence Models
|
The code intelligence (CI) models are often black-box and do not offer any insights on the input features that they learn for making correct predictions. This opacity may lead to distrust in their prediction and hamper their wider adoption in safety-critical applications. In recent, the program reduction technique is widely being used to identify key input features in order to explain the prediction of CI models. The approach removes irrelevant parts from an input program and keeps the minimal snippets that a CI model needs to maintain its prediction. However, the state-of-the-art approaches mainly use a syntax-unaware program reduction technique that does not follow the syntax of programs, which adds significant overhead to the reduction of input programs and explainability of models. In this paper, we apply a syntax-guided program reduction technique that follows the syntax of input programs during reduction. Our experiments on multiple models across different types of input programs show that the syntax-guided program reduction technique significantly outperforms the syntax-unaware program reduction technique in reducing the size of input programs. Extracting key input features from reduced programs reveals that the syntax-guided reduced programs contain more label-specific key input features and are more vulnerable to adversarial transformation when renaming the key tokens in programs. These label-specific key input features may help to understand the reasoning of models' prediction from different perspectives and increase the trustworthiness to correct classification given by CI models.
|
['Md Rafiqul Islam Rabin']
|
2022-02-14
| null | null | null | null |
['method-name-prediction']
|
['natural-language-processing']
|
[ 3.61966908e-01 2.86209822e-01 -5.70213258e-01 -3.92659456e-01
-3.25601399e-01 -7.72233069e-01 2.36430988e-01 3.75285178e-01
7.33937696e-02 1.88942656e-01 -7.03856573e-02 -8.45923424e-01
1.11122243e-01 -1.04756248e+00 -1.02766311e+00 -2.52308488e-01
1.66169256e-01 -1.11019969e-01 5.73367536e-01 -1.94512844e-01
6.37605190e-01 2.24211007e-01 -1.68636668e+00 9.64798927e-01
9.85672832e-01 6.59000933e-01 -7.83281401e-02 7.97495246e-01
-1.98754251e-01 1.44803631e+00 -5.89570582e-01 -5.41771173e-01
3.93221259e-01 -2.45441377e-01 -9.50326324e-01 -7.05070496e-01
-1.95209179e-02 -2.57760435e-01 -7.72594586e-02 1.38771248e+00
-3.76661807e-01 -2.67284185e-01 1.61347851e-01 -1.70568144e+00
-5.85377455e-01 9.49949026e-01 -3.75554323e-01 2.04098504e-02
4.82253075e-01 1.40938580e-01 9.65253472e-01 -4.80766296e-01
3.46590281e-01 1.16793644e+00 7.22003639e-01 5.56719244e-01
-1.28925335e+00 -7.98315585e-01 1.45836696e-01 2.48946816e-01
-1.36034572e+00 -2.21331697e-02 7.41344869e-01 -6.22078896e-01
1.33769536e+00 8.87204468e-01 3.08094889e-01 5.30324399e-01
8.55332255e-01 5.19967616e-01 1.08174551e+00 -5.70086062e-01
2.24045411e-01 7.53073752e-01 7.25280583e-01 7.33404160e-01
4.96597081e-01 5.08947134e-01 -1.85567215e-01 -8.18895340e-01
-5.72411120e-02 4.66518253e-01 -1.66226670e-01 -2.95502305e-01
-9.66655254e-01 1.12236476e+00 4.28001165e-01 2.38347098e-01
1.30939752e-01 1.62954092e-01 8.19940388e-01 4.25256133e-01
-3.99229936e-02 6.48626149e-01 -7.86458552e-01 -6.78495169e-02
-6.47906840e-01 4.88859594e-01 1.00029898e+00 1.18014681e+00
1.08146214e+00 -8.02667886e-02 6.32485151e-02 -1.16922125e-01
2.88772732e-01 5.36562979e-01 5.19176781e-01 -7.29683757e-01
4.81274605e-01 1.42487550e+00 -4.00848925e-01 -1.41043603e+00
-1.21358633e-01 -8.73354971e-02 -4.72589076e-01 5.74890971e-01
8.59277174e-02 1.65046617e-01 -5.09366214e-01 1.40644133e+00
2.34387107e-02 -2.98781931e-01 1.31700709e-01 3.24935138e-01
5.39989471e-01 5.99534392e-01 -1.71109751e-01 1.32878780e-01
1.12450397e+00 -1.04836893e+00 -1.86621144e-01 -3.25147659e-01
1.25086927e+00 -5.09125054e-01 1.21198726e+00 4.52433079e-01
-3.40133429e-01 -5.94730556e-01 -1.20069313e+00 3.98113132e-01
-4.57755387e-01 -1.87231496e-01 7.39235997e-01 9.67441201e-01
-7.43492424e-01 7.22259521e-01 -6.63564801e-01 1.17653616e-01
4.13848132e-01 8.21717143e-01 -5.42698085e-01 -7.48375952e-02
-8.33888113e-01 7.45362103e-01 4.47238445e-01 -5.27110517e-01
-7.16330111e-01 -8.52284729e-01 -9.52629626e-01 1.88689187e-01
5.88155806e-01 -3.10972363e-01 1.26126039e+00 -1.17677474e+00
-9.49435890e-01 3.85070950e-01 -4.05131698e-01 -5.62067032e-01
1.60133690e-01 1.08043291e-02 -3.50542635e-01 -2.43515790e-01
-1.11575276e-01 8.75427052e-02 9.93613899e-01 -1.31120813e+00
-8.12634885e-01 -2.12468967e-01 3.84999424e-01 -6.50891423e-01
-1.74279630e-01 1.95204809e-01 1.01047888e-01 -4.36476827e-01
-2.05720603e-01 -1.28699505e+00 -3.61260384e-01 -2.87670076e-01
-5.81965566e-01 1.65676996e-01 1.10218894e+00 -5.52774489e-01
1.74982953e+00 -2.14816117e+00 -1.53546497e-01 4.96008635e-01
5.37054181e-01 4.55198914e-01 6.48723170e-02 3.18180025e-01
-2.62814492e-01 8.09411943e-01 -1.98284283e-01 2.52167434e-01
-8.96438137e-02 2.40278482e-01 -6.79147184e-01 1.73120230e-01
1.44859537e-01 7.99798548e-01 -7.06953943e-01 -3.28428298e-01
1.65682852e-01 -3.89092602e-02 -1.12249410e+00 2.93509901e-01
-3.04181576e-01 1.54884443e-01 -6.33316696e-01 5.72447240e-01
6.81826711e-01 6.97291177e-03 -1.37946337e-01 -3.62000205e-02
7.35730231e-02 2.28684068e-01 -8.12992454e-01 8.72054696e-01
-4.64760512e-01 5.71778834e-01 -4.37359035e-01 -7.42684782e-01
1.10470259e+00 8.70420709e-02 -1.14601471e-01 -3.37576866e-01
-4.07658964e-02 1.31388158e-01 3.04724514e-01 -6.26713932e-01
3.67687523e-01 7.88666382e-02 -4.32746500e-01 8.16649914e-01
-5.83248436e-01 -1.84353784e-01 -1.64796561e-01 2.26396099e-01
1.41939473e+00 -6.91547245e-02 7.60820985e-01 -2.80634135e-01
1.04904413e+00 4.36511993e-01 9.28690970e-01 8.14954519e-01
-1.29874572e-01 3.42624396e-01 8.20104659e-01 -9.14440453e-01
-9.61877763e-01 -2.70989150e-01 2.97343791e-01 8.45115066e-01
-5.52163981e-02 -9.78554308e-01 -8.91009867e-01 -1.41062796e+00
4.39061224e-02 1.21544325e+00 -8.43125105e-01 -7.10627079e-01
-5.27591765e-01 -3.01035225e-01 7.17456520e-01 5.76869667e-01
1.32040456e-01 -8.84352863e-01 -9.57295835e-01 -1.15071528e-01
3.99849378e-03 -4.95824099e-01 -4.83518839e-01 3.86627942e-01
-8.16727221e-01 -1.49812412e+00 4.62991625e-01 -3.81438792e-01
1.09291518e+00 1.10980786e-01 9.48959470e-01 7.36749947e-01
5.34325913e-02 -1.68421417e-02 -5.22007942e-01 -7.59388268e-01
-1.30160797e+00 3.34848091e-02 -2.26804972e-01 -3.63268673e-01
8.13686073e-01 -1.94931462e-01 1.53290063e-01 3.72254938e-01
-1.01468003e+00 5.10475673e-02 2.81753242e-01 9.05356586e-01
3.14645797e-01 4.07498121e-01 1.83308333e-01 -1.61599708e+00
4.67894971e-01 -4.29922312e-01 -6.84474230e-01 4.70889390e-01
-9.83187377e-01 4.97985244e-01 1.32291186e+00 -5.84489167e-01
-7.56763399e-01 2.00510964e-01 1.79880425e-01 -3.76236975e-01
-1.68332592e-01 4.68379438e-01 -3.27233016e-01 -5.62106609e-01
1.00278080e+00 1.53185919e-01 2.67355535e-02 -2.13366926e-01
-5.69939017e-02 5.05791247e-01 1.69882625e-01 -5.22214234e-01
1.14402175e+00 1.04626663e-01 1.49632571e-02 1.12690993e-01
-2.43910238e-01 1.08827338e-01 -5.47383547e-01 2.60136336e-01
4.13436472e-01 -3.90370607e-01 -7.54719675e-01 2.28132710e-01
-1.31807852e+00 -9.73970890e-02 -1.97987273e-01 -2.99055483e-02
-3.51800084e-01 4.45518911e-01 -2.74221450e-01 -6.29668891e-01
-5.13186276e-01 -1.78315926e+00 5.52220643e-01 6.03006668e-02
-7.37701297e-01 -6.90171540e-01 3.83272395e-02 1.47065356e-01
4.99948055e-01 1.93405241e-01 1.83141506e+00 -1.23053229e+00
-5.29557407e-01 -7.50303388e-01 1.40952557e-01 6.48378730e-01
1.70332775e-01 3.75699639e-01 -1.06559372e+00 -6.02491014e-02
3.31462801e-01 2.00846761e-01 2.59607494e-01 -7.78490677e-02
1.22059476e+00 -8.74266088e-01 -4.92372483e-01 5.82841039e-01
1.35634947e+00 5.49324155e-01 5.64020813e-01 3.85479361e-01
8.07345450e-01 5.97979069e-01 8.58221769e-01 3.31236511e-01
1.99411869e-01 4.09421563e-01 6.01207256e-01 2.00576156e-01
4.38065201e-01 -4.88030970e-01 6.42797649e-01 3.74380887e-01
2.13743195e-01 2.53288805e-01 -1.19940495e+00 2.79040813e-01
-1.79161263e+00 -8.82100463e-01 -2.87342221e-01 2.28293371e+00
7.63980150e-01 3.41600060e-01 -1.08462162e-01 5.74272454e-01
5.58243394e-01 -2.84534574e-01 -3.96039307e-01 -1.12766099e+00
3.37057948e-01 1.15319975e-01 7.02164888e-01 4.62863326e-01
-7.91102767e-01 7.93929517e-01 6.02688837e+00 7.60488987e-01
-1.11191547e+00 1.30148660e-02 4.31585670e-01 4.03957099e-01
-7.80224144e-01 7.27830410e-01 -7.68596351e-01 3.38131040e-01
1.02763927e+00 -8.93193126e-01 4.82283324e-01 1.60430241e+00
-1.69290960e-01 -1.85714692e-01 -1.70793808e+00 4.97668087e-01
-9.38216746e-02 -1.36563158e+00 5.94598018e-02 -6.55588061e-02
5.17243803e-01 -3.72347653e-01 -8.32035467e-02 5.41593134e-01
2.43425772e-01 -1.15503275e+00 9.85643208e-01 3.39669704e-01
3.87496144e-01 -9.35410857e-01 1.17431796e+00 6.47421598e-01
-1.08450413e+00 -6.38141811e-01 -3.50126177e-01 -4.45455611e-01
-7.21544921e-01 3.23584855e-01 -1.08786976e+00 3.02334517e-01
8.01075280e-01 3.31725150e-01 -1.16909385e+00 5.71809173e-01
-2.72751600e-01 7.05222011e-01 1.41109610e-02 -2.72063941e-01
-2.14259833e-01 2.59113908e-01 4.75830823e-01 9.85957801e-01
1.15224466e-01 -1.41333356e-01 4.65837643e-02 1.24139595e+00
2.51712680e-01 -7.74854049e-02 -9.40061212e-01 -1.07855551e-01
4.74165797e-01 8.86300206e-01 -4.64785457e-01 -3.82223964e-01
-7.09376574e-01 3.48988235e-01 5.39179938e-03 8.33271369e-02
-8.50156307e-01 -4.87648964e-01 5.85516393e-01 2.51446724e-01
4.49269973e-02 1.74516246e-01 -7.30814397e-01 -8.19435358e-01
1.41899258e-01 -1.46771932e+00 2.13356420e-01 -3.40717107e-01
-7.10417986e-01 9.89781022e-01 1.58719406e-01 -1.29213333e+00
-3.45405698e-01 -3.69982779e-01 -9.32357132e-01 8.97942781e-01
-1.07922602e+00 -1.08065844e+00 -1.45105496e-01 5.82263470e-01
2.09099114e-01 -3.00546199e-01 1.06249249e+00 -2.34139189e-01
-4.24175054e-01 1.00394368e+00 -3.32445681e-01 2.98393145e-02
4.33290571e-01 -9.04184699e-01 5.21402836e-01 1.20011044e+00
-2.29773641e-01 1.09028220e+00 8.54024529e-01 -8.03465188e-01
-1.72774494e+00 -1.21971047e+00 9.49383497e-01 -6.88470602e-01
5.67352533e-01 -1.67837471e-01 -1.15617371e+00 9.03700948e-01
-2.36837789e-01 1.64779499e-01 8.27100754e-01 -2.32312948e-01
-7.99658835e-01 -2.22942457e-01 -1.47962892e+00 4.48849410e-01
3.64482909e-01 -7.19601035e-01 -7.22029984e-01 -8.87250379e-02
1.10318804e+00 -7.80615881e-02 -6.89030230e-01 3.72762471e-01
4.79739755e-01 -1.03252935e+00 5.94651222e-01 -7.21705914e-01
5.37322998e-01 -5.50100088e-01 -2.77207166e-01 -8.50284517e-01
-3.48634750e-01 -4.51019406e-01 -2.51230329e-01 1.25653625e+00
6.33441091e-01 -7.94962466e-01 6.89224124e-01 1.34692526e+00
3.90900262e-02 -6.54991210e-01 -4.91969377e-01 -5.73915601e-01
1.45341322e-01 -7.04662621e-01 1.11093915e+00 7.66454875e-01
4.41766828e-01 -1.71825051e-01 -2.07764283e-01 2.02570081e-01
3.44381154e-01 4.08440530e-01 1.00984752e+00 -1.06701720e+00
-4.81921673e-01 -2.67937928e-01 -6.20744169e-01 -9.24688280e-02
4.05085236e-01 -1.07710838e+00 -9.80985686e-02 -6.95180535e-01
4.60649729e-01 -5.28784215e-01 -1.27362639e-01 1.00876641e+00
-2.64942974e-01 -4.01876032e-01 4.41249192e-01 2.18695238e-01
-1.49648085e-01 -5.81794940e-02 5.72336435e-01 -4.52432543e-01
-2.97451138e-01 4.55941975e-01 -1.02720737e+00 8.81385565e-01
6.95101857e-01 -1.23328555e+00 -5.31277061e-01 -4.95397113e-02
5.27092636e-01 -1.37885243e-01 3.75412107e-01 -1.01817930e+00
3.36687714e-01 -4.87231731e-01 -1.65595949e-01 -2.03610837e-01
-6.10966980e-01 -1.32063663e+00 6.26535892e-01 1.02992249e+00
-5.47943890e-01 1.67395741e-01 3.81232649e-01 4.09704894e-01
-2.71051645e-01 -7.98695207e-01 6.93469822e-01 -2.13392794e-01
-6.66272819e-01 4.81893728e-03 -5.78205168e-01 -2.43521333e-01
1.29455984e+00 -3.11148137e-01 -2.83689320e-01 -3.59794721e-02
-2.05788791e-01 -2.00024530e-01 9.83407438e-01 5.81948161e-01
5.68529606e-01 -9.68152761e-01 -2.68372655e-01 7.11019695e-01
4.07011986e-01 -1.39943406e-01 -3.62473615e-02 6.84108198e-01
-5.96671045e-01 4.66393590e-01 -1.85918033e-01 -4.45375651e-01
-1.70713615e+00 1.19807220e+00 1.43834099e-01 -4.33429480e-01
-3.88407171e-01 6.46469653e-01 4.42501366e-01 -6.52399600e-01
7.67863821e-03 -7.56412208e-01 -3.56795229e-02 -6.80909336e-01
7.13511944e-01 2.90424407e-01 1.29212990e-01 -4.55284715e-01
-6.62220895e-01 3.94850969e-01 -4.12823647e-01 4.88473654e-01
1.23296797e+00 2.63454467e-01 -4.10927415e-01 1.55914843e-01
1.15190172e+00 2.94728726e-01 -7.44406164e-01 -1.59692958e-01
2.83573419e-01 -8.19489241e-01 -6.47764876e-02 -7.15004385e-01
-1.09672272e+00 1.07262957e+00 3.56803358e-01 1.57392040e-01
1.17624080e+00 -3.23494911e-01 5.59364080e-01 5.37396729e-01
7.32362688e-01 -6.24154806e-01 -3.93062800e-01 3.73261511e-01
6.33160472e-01 -1.22908711e+00 -7.22174048e-02 -7.13576198e-01
-7.16168582e-01 1.35545707e+00 8.79840672e-01 1.72160957e-02
5.31379819e-01 8.39032352e-01 -2.39650831e-01 1.23073980e-01
-8.24324131e-01 6.64962471e-01 1.62482575e-01 8.18405747e-01
1.27774358e-01 1.58878639e-01 -4.74004596e-02 1.34609759e+00
-2.04140365e-01 6.89729452e-02 9.47764635e-01 1.32848024e+00
-3.03301781e-01 -1.38172591e+00 -5.71940899e-01 6.27666831e-01
-4.56008524e-01 -2.53423661e-01 -5.80721438e-01 7.20001161e-01
2.60881573e-01 1.12300503e+00 -6.57106638e-01 -1.30069208e+00
3.17137212e-01 8.99663866e-02 1.44981714e-02 -8.10049891e-01
-1.23914266e+00 -5.77944636e-01 -8.72626230e-02 -7.98207164e-01
7.81356990e-02 -4.11929667e-01 -1.52627826e+00 -8.26148570e-01
-5.60995221e-01 2.31943592e-01 3.26258987e-01 8.49952579e-01
4.90403622e-01 4.77409035e-01 7.75264800e-01 -3.91478360e-01
-6.05331123e-01 -3.53172839e-01 -1.08107492e-01 3.64148676e-01
4.27915871e-01 -3.57055604e-01 -5.08917391e-01 2.30926380e-01]
|
[7.338871955871582, 7.741972923278809]
|
da7906ba-efc1-4dd2-887f-e1e04f4d1176
|
recognition-of-they-them-as-singular-personal
| null | null |
https://aclanthology.org/2022.naacl-main.250
|
https://aclanthology.org/2022.naacl-main.250.pdf
|
Recognition of They/Them as Singular Personal Pronouns in Coreference Resolution
|
As using they/them as personal pronouns becomes increasingly common in English, it is important that coreference resolution systems work as well for individuals who use personal “they” as they do for those who use gendered personal pronouns. We introduce a new benchmark for coreference resolution systems which evaluates singular personal “they” recognition. Using these WinoNB schemas, we evaluate a number of publicly available coreference resolution systems and confirm their bias toward resolving “they” pronouns as plural.
|
['Rachel Rudinger', 'Connor Baumler']
| null | null | null | null |
naacl-2022-7
|
['coreference-resolution']
|
['natural-language-processing']
|
[-7.16138780e-02 3.97038460e-01 -6.36989415e-01 -3.10043871e-01
-9.08531964e-01 -9.94937420e-01 8.58030856e-01 4.17992741e-01
-9.44261014e-01 1.11068559e+00 8.31466854e-01 -5.42426780e-02
-2.90637612e-01 -5.77756822e-01 9.31123924e-03 -4.53256071e-01
5.43453395e-01 1.38645732e+00 1.89085126e-01 -8.59856963e-01
1.65535361e-01 6.96414649e-01 -1.65400469e+00 1.14814527e-01
4.66681451e-01 4.11611833e-02 -1.02302514e-01 2.58214772e-01
-3.55227739e-01 2.45680511e-01 -9.57024455e-01 -9.61043596e-01
-1.19479544e-01 3.39350164e-01 -1.37653935e+00 -1.04321647e+00
8.41525316e-01 7.39885271e-02 -2.03916907e-01 1.22941911e+00
6.22837424e-01 3.65604401e-01 4.93113279e-01 -9.88692522e-01
-1.32303804e-01 1.19448841e+00 -4.84147161e-01 7.02658594e-01
1.10154164e+00 -3.96576703e-01 1.53059781e+00 -4.23586845e-01
1.24225116e+00 2.02991939e+00 6.89101815e-01 1.14108396e+00
-1.42009282e+00 -1.07631505e+00 -2.11341441e-01 3.02253276e-01
-1.48262715e+00 -7.06220627e-01 4.34437096e-01 -1.22226011e-02
1.42315888e+00 8.91689658e-01 1.58850819e-01 1.47851825e+00
2.22895481e-02 4.97876853e-01 6.74716651e-01 -4.27664459e-01
-4.95892107e-01 -1.68410078e-01 8.01223934e-01 -1.86150506e-01
5.69148242e-01 1.43705755e-01 -6.60998821e-01 -3.85914922e-01
1.58020139e-01 -5.37955880e-01 -1.53418526e-01 2.62283117e-01
-9.89220500e-01 7.08184898e-01 -1.37768745e-01 7.64278650e-01
-1.37287304e-02 -1.27166525e-01 3.66345376e-01 9.46757495e-02
-3.96381468e-01 8.79209757e-01 -2.68649757e-01 -3.35108519e-01
-6.77226543e-01 8.91918182e-01 1.15111530e+00 1.22465324e+00
2.86648989e-01 -7.49719381e-01 -2.20456600e-01 1.11600077e+00
-5.96492961e-02 5.63003778e-01 2.95095414e-01 -1.69070494e+00
3.44178557e-01 6.32322013e-01 5.75689137e-01 -8.49675775e-01
-7.93545544e-01 2.87181232e-02 -2.89771229e-01 -3.13369215e-01
5.95723748e-01 1.53734803e-01 1.76593363e-01 1.97100437e+00
8.79177973e-02 -6.41808152e-01 4.30196464e-01 7.43257999e-01
1.31199431e+00 1.17370531e-01 7.45330632e-01 -6.16141081e-01
1.78587794e+00 -3.12697977e-01 -1.08403456e+00 -3.27272028e-01
2.68524379e-01 -9.59027171e-01 8.54809701e-01 -8.72348323e-02
-1.31290746e+00 4.84119616e-02 -7.79873192e-01 -4.93164986e-01
-1.21264994e-01 -6.00809872e-01 7.79538929e-01 6.47842169e-01
-5.91310263e-01 7.18086600e-01 -5.92513263e-01 -1.01359022e+00
-1.82471618e-01 7.02112257e-01 -6.87283635e-01 6.75096214e-01
-1.50114322e+00 1.19648945e+00 4.31439251e-01 -6.33725047e-01
-6.87165782e-02 -6.25981987e-01 -7.30380118e-01 -8.07108954e-02
6.14996925e-02 -5.12662530e-01 1.54856539e+00 -6.12081349e-01
-9.35955822e-01 1.77273571e+00 -5.42377293e-01 -1.80347830e-01
1.71654284e-01 -4.95152354e-01 -1.02905440e+00 -4.82701510e-02
5.87109864e-01 7.66790569e-01 9.91242379e-02 -1.16346645e+00
-1.43925428e+00 -7.11026907e-01 4.50538486e-01 4.77419496e-01
-1.20287240e-01 9.56541657e-01 -1.50692001e-01 -2.65591979e-01
1.69025913e-01 -9.07837093e-01 4.99203175e-01 -8.08251202e-01
-5.43501377e-01 -1.21117759e+00 5.99492908e-01 -1.02074489e-01
1.52741170e+00 -2.02593684e+00 1.72545224e-01 1.32032886e-01
4.86067720e-02 2.09741727e-01 1.08207248e-01 5.89296281e-01
-3.06580126e-01 1.94731101e-01 1.46068320e-01 -2.16152117e-01
3.43679577e-01 8.69331360e-01 -5.34080803e-01 9.27607417e-02
-1.04228526e-01 4.81272459e-01 -1.13797140e+00 -9.10723090e-01
-2.20395148e-01 2.50664234e-01 -4.64605212e-01 1.22099780e-02
1.77043766e-01 4.17516716e-02 -1.41445205e-01 5.14776528e-01
6.90905154e-01 4.55224156e-01 9.85803783e-01 -3.65407795e-01
-5.63071311e-01 5.98291755e-01 -1.26009357e+00 1.17748868e+00
1.28746703e-01 2.20859528e-01 4.56767261e-01 -3.38281065e-01
7.95165300e-01 4.38439876e-01 8.43883902e-02 -6.18470669e-01
-1.44273490e-01 4.51875538e-01 1.77447021e-01 -1.28173172e-01
1.08038294e+00 -4.66585755e-01 -6.27930820e-01 7.26451129e-02
5.94163239e-02 -4.71674167e-02 6.61005557e-01 3.05426151e-01
8.70344520e-01 3.87799069e-02 5.94507396e-01 -9.76768315e-01
1.01323032e+00 3.46522838e-01 1.25595009e+00 3.46397251e-01
-3.39337528e-01 3.18403214e-01 6.67950571e-01 -3.17400426e-01
-5.19304514e-01 -1.31773472e+00 -8.08617532e-01 1.43388522e+00
4.49343652e-01 -9.94927526e-01 -6.31361127e-01 -3.96488279e-01
9.31719989e-02 1.28302050e+00 -4.11582589e-01 2.68683344e-01
-1.12357485e+00 -5.85107267e-01 1.16640019e+00 6.48801267e-01
-7.77712837e-02 -1.12090826e+00 -6.80923164e-01 2.59353608e-01
-5.42178392e-01 -9.66145337e-01 -4.25200462e-01 1.83949172e-01
-2.67744452e-01 -1.52117229e+00 -1.01771317e-01 -8.48759592e-01
-1.18317440e-01 -2.98589885e-01 1.73022175e+00 3.43264431e-01
1.09437764e-01 4.69100028e-01 -1.22794725e-01 -4.55003917e-01
-3.57048601e-01 3.18993568e-01 4.68703002e-01 -8.87385130e-01
1.42037439e+00 -7.41201222e-01 -1.84009701e-01 9.87603962e-02
-3.07928354e-01 -6.23206437e-01 -2.37067297e-01 7.03640163e-01
2.23047733e-01 -6.78282619e-01 7.03174174e-02 -1.42155576e+00
9.81759429e-01 -1.15579151e-01 -2.46858686e-01 3.30276042e-02
-6.29135430e-01 -4.53317160e-04 2.91436315e-01 -1.66217145e-02
-1.38399351e+00 -5.72168589e-01 -5.21816134e-01 2.46782795e-01
-2.39961430e-01 -9.35425833e-02 -3.83507490e-01 6.44446731e-01
8.27030957e-01 -5.81136465e-01 -3.13138306e-01 -6.81443989e-01
-7.40753561e-02 6.84721351e-01 1.39201200e+00 -1.41759658e+00
6.96107090e-01 3.01522553e-01 7.82409161e-02 -9.12556887e-01
-7.25364506e-01 -6.45555496e-01 -6.15564048e-01 4.92555797e-01
8.43844473e-01 -5.66896141e-01 -1.51534426e+00 3.87635566e-02
-1.56969166e+00 3.62232447e-01 -3.56145024e-01 -1.46973804e-02
-4.07248080e-01 6.15850449e-01 -7.59225130e-01 -4.54368740e-01
-7.15882123e-01 -8.89400005e-01 8.90362978e-01 5.23848772e-01
-1.50825894e+00 -9.60049868e-01 1.85670897e-01 1.17592663e-01
-8.86543691e-02 1.08628133e-02 1.13198733e+00 -1.01292908e+00
4.04720753e-01 1.88545272e-01 -1.27140775e-01 -3.55366588e-01
-1.03289358e-01 8.04045200e-02 -4.64492977e-01 -1.11863159e-01
-5.25537968e-01 8.54139999e-02 2.38531575e-01 -2.60861367e-01
2.37224802e-01 -1.79770723e-01 -9.06810820e-01 6.30545855e-01
1.07702923e+00 2.17252508e-01 6.00434780e-01 7.64120936e-01
3.94384474e-01 8.43850255e-01 6.72724307e-01 6.65585101e-02
9.63257015e-01 9.85804439e-01 -8.04914981e-02 7.20612407e-01
-1.97048455e-01 -1.48173079e-01 4.10305988e-03 4.49808717e-01
-4.73541409e-01 2.90198684e-01 -1.22456014e+00 6.63656831e-01
-1.70699370e+00 -1.36514771e+00 -4.24922794e-01 1.85803008e+00
1.42839468e+00 -6.60363883e-02 1.43595068e-02 8.88651460e-02
8.54996860e-01 1.16030142e-01 -1.11819087e-02 -6.53096437e-01
-7.69882679e-01 6.50366843e-01 1.90079883e-01 9.73403215e-01
-1.12389457e+00 1.34617186e+00 7.12287235e+00 3.78823161e-01
-6.31350994e-01 5.71160018e-02 -2.86662132e-01 -7.34239891e-02
-5.26492774e-01 1.12578452e-01 -1.58805573e+00 1.67130277e-01
7.30920255e-01 -6.70417249e-01 1.87210351e-01 3.54212672e-01
-3.82713497e-01 -8.98549557e-02 -1.69758785e+00 1.30560243e+00
-2.18144611e-01 -1.11256540e+00 1.71049297e-01 -3.03319514e-01
2.28650138e-01 -2.55713969e-01 -4.87972021e-01 3.36582869e-01
6.80610657e-01 -1.11020041e+00 7.93620884e-01 5.57659626e-01
6.46189094e-01 -8.36205304e-01 7.22719193e-01 -7.82312546e-03
-8.92746687e-01 -1.50621654e-02 -3.60134274e-01 -1.72602922e-01
1.91411197e-01 -2.94243723e-01 -8.93038809e-02 3.66539150e-01
9.98012841e-01 1.57007694e-01 -2.51748323e-01 3.40766907e-01
-6.36125505e-01 1.50834978e-01 -5.52150786e-01 1.28374249e-01
-3.09657276e-01 -4.12049005e-03 1.24200690e+00 1.59827876e+00
8.60403175e-04 6.10971510e-01 -1.19318716e-01 4.90602881e-01
1.05805852e-01 2.74824888e-01 -2.83415318e-02 2.25665703e-01
1.48678589e+00 1.38581812e+00 -1.40995890e-01 -9.32209566e-02
-2.86307752e-01 5.96003115e-01 3.50190073e-01 4.60183062e-02
-5.87251745e-02 -2.89469689e-01 1.36238527e+00 1.70053333e-01
-4.27929282e-01 1.61407053e-01 -5.82795814e-02 -7.97543168e-01
-4.55233604e-01 -1.07593381e+00 1.21956193e+00 -5.07907510e-01
-1.30969632e+00 4.24863726e-01 3.39843661e-01 -5.19386768e-01
-7.42467940e-01 -5.89984357e-01 -5.63928127e-01 1.10330021e+00
-1.16567981e+00 -1.13910413e+00 -5.87194823e-02 7.27362812e-01
-7.22703487e-02 -2.17404559e-01 1.57044649e+00 1.90573901e-01
-4.65352327e-01 1.09084153e+00 -5.02011538e-01 2.31954798e-01
1.30763638e+00 -1.43823552e+00 1.40173569e-01 5.04119039e-01
-1.75246432e-01 1.51221991e+00 1.38524103e+00 -4.26513940e-01
-1.24923885e+00 -2.49050602e-01 1.68543851e+00 -9.48168933e-01
6.20325923e-01 2.93592602e-01 -7.11744905e-01 1.25016463e+00
5.55097342e-01 -5.06625295e-01 7.53447592e-01 9.47079420e-01
-6.47321224e-01 -2.16571137e-01 -1.27964938e+00 7.81386554e-01
1.55721354e+00 -3.97927582e-01 -1.75821412e+00 1.90076826e-03
4.77737457e-01 -6.88675404e-01 -1.16769087e+00 5.28012872e-01
3.62908006e-01 -8.73555541e-01 1.05783856e+00 -7.71125972e-01
1.91405565e-02 -1.51131332e-01 -4.77989048e-01 -1.07524633e+00
-4.46071863e-01 -8.48045886e-01 1.09499479e-02 1.72598445e+00
2.11781591e-01 -6.95861995e-01 3.94271761e-01 1.02734971e+00
2.28787288e-02 4.25067753e-01 -1.31923437e+00 -4.12214518e-01
7.12735832e-01 -1.28479198e-01 9.69897449e-01 1.24375331e+00
9.12061989e-01 7.73128867e-01 3.90412509e-01 -1.46919698e-01
9.60472465e-01 1.94065318e-01 5.69492877e-01 -2.05902076e+00
2.78809279e-01 -6.30616546e-01 -1.67980343e-01 -4.32886720e-01
9.82937753e-01 -1.06938672e+00 -5.76647937e-01 -1.12172472e+00
3.70312870e-01 -3.94802779e-01 4.62407880e-02 5.67064404e-01
-1.75654978e-01 1.62736431e-01 2.24639371e-01 4.39584225e-01
-5.55855751e-01 -3.65526229e-02 7.02468276e-01 -1.34523422e-01
6.57306463e-02 -4.53949660e-01 -1.21968555e+00 1.08448136e+00
2.39844888e-01 -3.71679068e-01 5.12588263e-01 -2.56924897e-01
5.81013739e-01 -4.19431105e-02 -5.70779368e-02 -7.24950433e-01
5.02278805e-01 -3.41569304e-01 -1.62398964e-02 -4.41869229e-01
2.52806574e-01 -4.27982152e-01 2.52317607e-01 1.98178515e-01
3.99955874e-03 3.54278326e-01 2.67343163e-01 -4.11627918e-01
-1.75301701e-01 -4.83721495e-01 7.48202443e-01 -2.24029243e-01
-1.18295515e+00 -2.33350828e-01 -4.99064736e-02 6.36085093e-01
7.00026870e-01 7.69816414e-02 -7.74566174e-01 -2.43414864e-02
-7.80099332e-01 4.12181556e-01 7.23987639e-01 7.16711104e-01
4.73843180e-02 -1.29430652e+00 -7.42218912e-01 -2.94990510e-01
3.78048062e-01 -4.36212629e-01 -1.88799471e-01 5.44102490e-01
-2.86358118e-01 6.43039405e-01 -5.19516766e-01 -2.04416752e-01
-1.75695527e+00 4.17143524e-01 4.07947928e-01 6.06403649e-02
-5.49627244e-01 8.25309575e-01 -2.31675714e-01 -7.14848757e-01
1.29411176e-01 5.33080161e-01 -8.45082343e-01 5.82723320e-01
7.66269505e-01 5.83220661e-01 -1.12743810e-01 -1.47545946e+00
-1.01883531e+00 2.90255815e-01 -1.19975738e-01 -2.33819738e-01
1.14832461e+00 -1.97761193e-01 -9.07119036e-01 4.13646936e-01
4.96130079e-01 7.87337661e-01 -2.74188459e-01 -5.90194911e-02
5.53775012e-01 -4.21899796e-01 -5.26800573e-01 -5.45338750e-01
-3.58105808e-01 2.31781468e-01 3.19681227e-01 -2.94469600e-03
5.05401134e-01 2.62179762e-01 5.87149978e-01 4.14593726e-01
6.49843037e-01 -1.44194853e+00 -9.69925523e-01 1.06678212e+00
9.56984460e-01 -5.38312852e-01 -6.75097480e-03 -5.63960671e-01
-2.59354562e-01 1.05847454e+00 7.47845590e-01 -7.90106356e-02
1.90018043e-01 6.60024285e-01 4.91970420e-01 -1.62231565e-01
-7.16134310e-01 -3.69108051e-01 -8.23456645e-02 9.40259218e-01
1.17166817e+00 5.81146836e-01 -1.52801418e+00 1.34052217e+00
-1.15967774e+00 -4.39750701e-01 5.90261698e-01 4.24747914e-01
-5.94801679e-02 -1.93040371e+00 -6.80285633e-01 1.00536421e-01
-1.09348583e+00 -2.08177999e-01 -4.62420732e-01 1.16830397e+00
1.21523239e-01 1.12017536e+00 6.47855043e-01 -5.11042885e-02
8.76314461e-01 1.99224666e-01 8.02335918e-01 -6.50810421e-01
-9.87232983e-01 -4.27528381e-01 9.30661559e-01 -6.17152810e-01
-7.18967736e-01 -1.46107471e+00 -1.69094527e+00 -8.43916953e-01
2.57634670e-01 6.36242032e-01 -1.44065127e-01 9.28269625e-01
-1.68802202e-01 -8.64665285e-02 -3.00542563e-01 -4.51327980e-01
-6.62409127e-01 -9.83934760e-01 -7.02368498e-01 1.16305864e+00
-1.13795340e-01 -8.44187260e-01 -2.58540988e-01 -5.62197328e-01]
|
[9.305306434631348, 9.541956901550293]
|
1d1a1615-7099-4e08-8d5a-5ee3d18ff6f9
|
setsum-summarization-and-visualization-of-1
|
2207.03640
| null |
https://arxiv.org/abs/2207.03640v1
|
https://arxiv.org/pdf/2207.03640v1.pdf
|
SETSum: Summarization and Visualization of Student Evaluations of Teaching
|
Student Evaluations of Teaching (SETs) are widely used in colleges and universities. Typically SET results are summarized for instructors in a static PDF report. The report often includes summary statistics for quantitative ratings and an unsorted list of open-ended student comments. The lack of organization and summarization of the raw comments hinders those interpreting the reports from fully utilizing informative feedback, making accurate inferences, and designing appropriate instructional improvements. In this work, we introduce a novel system, SETSum, that leverages sentiment analysis, aspect extraction, summarization, and visualization techniques to provide organized illustrations of SET findings to instructors and other reviewers. Ten university professors from diverse departments serve as evaluators of the system and all agree that SETSum helps them interpret SET results more efficiently; and 6 out of 10 instructors prefer our system over the standard static PDF report (while the remaining 4 would like to have both). This demonstrates that our work holds the potential to reform the SET reporting conventions in the future. Our code is available at https://github.com/evahuyn/SETSum
|
['Mohit Bansal', 'A. T. Panter', 'Viji Sathy', 'Shiyue Zhang', 'Yinuo Hu']
|
2022-07-08
|
setsum-summarization-and-visualization-of
|
https://aclanthology.org/2022.naacl-demo.9
|
https://aclanthology.org/2022.naacl-demo.9.pdf
|
naacl-acl-2022-7
|
['aspect-extraction']
|
['natural-language-processing']
|
[-3.64078283e-01 2.16093257e-01 -3.98676306e-01 -6.47711396e-01
-1.22397184e+00 -1.26798093e+00 4.76067923e-02 9.69185412e-01
-1.40071288e-01 5.79982340e-01 4.93802875e-01 -1.07756484e+00
-2.88932979e-01 -5.76570868e-01 -5.42976022e-01 -1.55571193e-01
4.06632245e-01 2.39293519e-02 9.59909260e-02 -1.57722786e-01
1.12742639e+00 3.06022555e-01 -1.54777682e+00 3.68014276e-01
1.09834909e+00 4.38489676e-01 -2.18994901e-01 6.87018812e-01
-4.21607792e-01 9.43131745e-01 -1.12281954e+00 -7.27077127e-01
-1.59957498e-01 -2.50690252e-01 -5.96690774e-01 -1.26284093e-01
1.05183351e+00 -3.71056944e-01 4.01502550e-01 9.29160297e-01
5.66160262e-01 3.58909965e-01 5.41154504e-01 -1.23240662e+00
-8.81118059e-01 7.46317446e-01 -6.35964036e-01 5.16457140e-01
6.17495537e-01 -2.31500585e-02 1.12686121e+00 -9.70800519e-01
5.27733743e-01 1.02669358e+00 3.37095439e-01 1.28392920e-01
-7.30021238e-01 -9.41816986e-01 3.88299048e-01 -4.73030470e-02
-8.94243956e-01 -4.04649436e-01 3.87679428e-01 -7.58564889e-01
5.87237418e-01 4.27608997e-01 7.12537944e-01 7.64654875e-01
1.81647643e-01 4.55031246e-01 1.28511572e+00 -3.32484931e-01
3.63984972e-01 7.97633946e-01 6.64856911e-01 6.31000698e-01
9.25156593e-01 -7.18578935e-01 -6.92389727e-01 -2.27638111e-01
3.72603238e-01 7.08241016e-02 -1.90374866e-01 -6.50710985e-02
-8.47709358e-01 5.90056539e-01 1.47949740e-01 -2.24731192e-02
1.02558523e-01 -2.49854937e-01 4.45258051e-01 3.01960886e-01
4.56434488e-01 6.13617122e-01 -2.48510793e-01 -7.06693947e-01
-1.09531009e+00 3.37024599e-01 9.85194385e-01 9.51430261e-01
3.88963670e-01 -7.66644701e-02 -1.14162721e-01 6.92249835e-01
2.89860755e-01 4.83724087e-01 3.51069838e-01 -1.19002211e+00
3.63235623e-01 9.04400885e-01 3.00074071e-01 -1.11294615e+00
1.75324559e-01 -5.57658851e-01 2.70052403e-01 4.14308250e-01
2.53780812e-01 -1.99981228e-01 -6.21456563e-01 1.11101437e+00
9.52361897e-02 -5.64825952e-01 -2.43549690e-01 6.19288385e-01
1.47727323e+00 3.23119819e-01 3.85040902e-02 4.40314263e-02
1.50531840e+00 -6.36413932e-01 -9.72439647e-01 1.17412761e-01
7.64352441e-01 -1.24853361e+00 1.43093956e+00 7.17347324e-01
-1.41953599e+00 -2.68937826e-01 -1.30946517e+00 -8.65133107e-02
-4.80513692e-01 1.85305193e-01 1.98511511e-01 8.69305611e-01
-1.00208378e+00 6.40202761e-01 -8.93658519e-01 -3.91049623e-01
6.30470872e-01 1.67024598e-01 -2.98166454e-01 -2.45795660e-02
-4.14239645e-01 6.62715018e-01 -5.66483021e-01 -3.06109995e-01
-1.45006463e-01 -9.79472518e-01 -8.42059195e-01 2.03634933e-01
2.50780374e-01 -3.39332789e-01 1.86017692e+00 -1.22931704e-01
-1.22996378e+00 6.90723240e-01 -1.85971007e-01 4.77989167e-01
1.79804370e-01 -5.65301239e-01 -2.45783366e-02 9.08522680e-02
2.06427008e-01 9.28700343e-02 6.50501400e-02 -1.09053564e+00
-4.66192871e-01 -4.11775500e-01 1.75094172e-01 2.87517130e-01
-8.94187093e-01 2.91882664e-01 -2.26821929e-01 -5.08486331e-01
1.77079216e-02 -6.35941386e-01 -7.16033280e-02 -2.72960424e-01
-2.82400787e-01 -3.41212779e-01 7.53853381e-01 -4.15318161e-01
1.81111205e+00 -1.99061179e+00 -5.23675740e-01 2.45504498e-01
4.24786478e-01 -3.28208841e-02 2.67823189e-01 6.70610070e-01
-7.74448290e-02 8.99786949e-01 2.86091238e-01 -3.24487597e-01
1.02011509e-01 -2.39903048e-01 -4.23559278e-01 2.80092210e-01
8.42652395e-02 3.42683494e-01 -1.17224514e+00 -6.51425242e-01
-2.48346757e-02 2.72349775e-01 -4.58932877e-01 9.05769318e-02
2.29802236e-01 -4.41898257e-02 -6.53514028e-01 8.13182771e-01
5.99703968e-01 -4.62253988e-01 1.10879399e-01 1.87518299e-01
-7.20025778e-01 9.64451671e-01 -1.19498479e+00 1.08696651e+00
-3.54653180e-01 9.97102916e-01 -1.65800508e-02 -3.38629574e-01
1.08374441e+00 1.82757363e-01 2.76227832e-01 -2.43621811e-01
1.42425448e-02 2.89798051e-01 -2.65239507e-01 -4.35347885e-01
8.47559333e-01 3.74140203e-01 -7.98875540e-02 1.17424786e+00
-4.78902161e-02 -6.08680785e-01 7.23997474e-01 8.20684791e-01
1.16271996e+00 6.87483475e-02 1.31393850e-01 -2.27682427e-01
-9.57244262e-02 1.08271286e-01 3.01988423e-01 6.51322126e-01
-3.39245945e-01 7.13741899e-01 7.63785362e-01 -6.19857237e-02
-8.38716269e-01 -9.59744275e-01 -1.12992972e-01 1.24138141e+00
-5.32538891e-01 -1.14822042e+00 -4.94419783e-01 -6.77115977e-01
3.00742909e-02 8.74736488e-01 -4.09681708e-01 2.33648285e-01
-8.63974839e-02 -1.66115999e-01 1.02942206e-01 5.49791634e-01
-1.38858154e-01 -7.22152889e-01 -6.73353314e-01 -2.26270482e-01
-1.38270706e-01 -6.85616136e-01 -5.41616797e-01 1.98372036e-01
-7.94135630e-01 -9.15974677e-01 -2.76413947e-01 -5.41275620e-01
9.22668457e-01 6.59827650e-01 1.11053669e+00 2.21458584e-01
-7.91831687e-02 7.69829988e-01 -4.70135778e-01 -9.97178078e-01
-2.96984732e-01 -4.21898551e-02 -8.24611112e-02 -9.94352877e-01
6.03838146e-01 -5.06588399e-01 -5.84711313e-01 1.49186984e-01
-8.79240692e-01 -2.61139125e-01 4.58050460e-01 3.25587988e-01
2.91845292e-01 -2.74792194e-01 6.43219829e-01 -1.31047964e+00
9.96678412e-01 -4.66654330e-01 -4.76644158e-01 5.35261407e-02
-9.88839984e-01 -2.40030244e-01 4.97467041e-01 6.30886406e-02
-8.36460888e-01 -6.97499216e-01 1.83812812e-01 2.19313614e-02
2.59895641e-02 7.11820960e-01 3.58598292e-01 -3.08337528e-02
8.02222848e-01 -5.74290693e-01 -1.23209044e-01 -2.94283152e-01
-1.42641515e-01 8.54421854e-01 2.56723344e-01 -5.78604460e-01
8.17877352e-01 -1.44166082e-01 -4.05250043e-01 -3.73184115e-01
-1.17876947e+00 -5.59004128e-01 -3.63139749e-01 -5.58457017e-01
2.16687247e-01 -1.05430186e+00 -9.26156759e-01 -4.78957921e-01
-6.76940978e-01 -1.59074441e-01 -3.21797460e-01 5.76462507e-01
-4.18000221e-02 1.64017707e-01 -6.70364797e-01 -7.08432436e-01
-2.26039469e-01 -1.25047874e+00 7.33883083e-01 9.55494463e-01
-8.59580040e-01 -1.01785958e+00 7.99129978e-02 9.44897056e-01
4.51055110e-01 1.49176851e-01 6.99742138e-01 -9.29238379e-01
-2.78976679e-01 -4.50078249e-01 2.68545933e-02 2.49311313e-01
1.97617412e-01 1.15090656e+00 -1.13900352e+00 -1.58435270e-01
-3.52771580e-01 -4.31824714e-01 3.95607412e-01 4.01237428e-01
1.33393610e+00 -5.64321995e-01 -4.65601450e-03 2.64603347e-02
1.13351750e+00 7.32796490e-02 6.25746846e-02 4.81372565e-01
4.04228836e-01 6.53652906e-01 8.12402785e-01 7.93644786e-01
5.15789807e-01 -1.17113024e-01 2.28427544e-01 2.33146176e-01
2.34135672e-01 -2.68521041e-01 7.04873741e-01 1.49319613e+00
9.60168168e-02 -2.80080587e-01 -9.34289694e-01 6.24758244e-01
-1.47769392e+00 -8.82255435e-01 -2.28187740e-01 2.24526167e+00
7.42844462e-01 5.08105755e-01 4.67828736e-02 -4.51922417e-02
3.49335611e-01 1.32517457e-01 -2.27894381e-01 -1.08103275e+00
3.88402671e-01 5.40020347e-01 3.23228985e-01 4.05059129e-01
-5.04268408e-01 3.67089778e-01 6.31576824e+00 4.02153641e-01
-9.63308454e-01 -3.47287714e-01 6.84902906e-01 -3.08985353e-01
-9.03452933e-01 2.60122985e-01 -1.02088964e+00 1.20016254e-01
1.30526924e+00 -8.22225273e-01 -4.47350800e-01 9.86596704e-01
5.00254989e-01 -3.05401206e-01 -9.30686772e-01 5.36366940e-01
5.56433224e-04 -1.41519737e+00 -3.19025479e-02 9.81928408e-02
8.11914325e-01 -4.55060065e-01 5.72027802e-01 3.43913525e-01
7.79122353e-01 -1.01893699e+00 4.87473518e-01 2.15642408e-01
4.52746511e-01 -7.06238091e-01 7.76719928e-01 -1.63374245e-01
-8.32036257e-01 1.29624279e-02 -1.41181946e-01 -3.47175181e-01
-4.10965204e-01 6.55278981e-01 -1.44543600e+00 3.47326547e-01
8.77942741e-01 8.32542539e-01 -9.63107169e-01 1.05377877e+00
-3.17454636e-01 1.06610823e+00 -8.55842489e-04 -4.37891901e-01
-5.68363704e-02 -1.22353747e-01 4.19266224e-01 1.41388381e+00
5.12022555e-01 3.86086889e-02 1.62925050e-01 5.97140372e-01
-1.50554314e-01 4.62678820e-01 -5.73064804e-01 -6.05572701e-01
9.81066346e-01 1.81997156e+00 -9.46320593e-01 -3.08935404e-01
-3.86082321e-01 5.43910675e-02 1.52549207e-01 2.83797592e-01
-2.84687072e-01 -7.54257083e-01 5.27909040e-01 4.40076023e-01
7.98604712e-02 -3.22147369e-01 -8.63869548e-01 -6.98839068e-01
1.29236564e-01 -9.43277180e-01 4.37354386e-01 -1.02509570e+00
-9.74636674e-01 3.07527930e-01 -4.79168296e-02 -1.43742704e+00
2.88376845e-02 -6.25046730e-01 -9.28644538e-01 7.07266927e-01
-1.17981541e+00 -2.32661128e-01 -7.33208179e-01 -1.44434690e-01
5.66337049e-01 1.63147435e-01 6.42439008e-01 1.86153412e-01
-9.70777869e-01 7.64311552e-01 -8.00220668e-02 4.28190548e-03
1.66385901e+00 -1.76533198e+00 3.46109876e-03 6.17784798e-01
-2.06927374e-01 1.36450994e+00 1.04888129e+00 -4.47423011e-01
-1.15911508e+00 -5.92069030e-01 8.74736488e-01 -8.46332550e-01
8.16341043e-01 5.89269996e-02 -8.64914834e-01 7.23335326e-01
6.29422903e-01 -3.27539146e-01 1.64005589e+00 3.27094615e-01
-2.60521173e-01 -2.09831055e-02 -7.20701754e-01 7.64822960e-01
3.15539002e-01 -3.29965204e-01 -5.33549726e-01 5.14129162e-01
6.19171441e-01 -6.78976595e-01 -1.33419776e+00 5.46720326e-02
7.06119597e-01 -9.35760200e-01 3.46639842e-01 -5.71454465e-01
1.14003718e+00 -2.36700043e-01 2.05196455e-01 -1.37487566e+00
3.34282150e-03 -5.41414976e-01 1.61261335e-01 1.41243768e+00
5.64692199e-01 -3.24561685e-01 7.25076556e-01 1.04615819e+00
-5.76376915e-01 -1.02910495e+00 4.14513759e-02 -1.68727279e-01
3.61013114e-02 -3.77103180e-01 3.60693634e-01 1.08131289e+00
5.88148832e-01 3.55641305e-01 5.02875686e-01 -1.54671699e-01
4.81599897e-01 3.12903494e-01 8.92174065e-01 -1.14266813e+00
1.54174730e-01 -7.59906173e-01 -1.35249972e-01 -6.52636349e-01
-2.69461483e-01 -6.47614241e-01 -3.45202625e-01 -2.00864553e+00
3.68055105e-01 -3.03399146e-01 -2.88684130e-01 6.66591525e-01
-3.65636379e-01 2.13105306e-02 -1.30922450e-02 3.67868552e-03
-8.03465962e-01 5.67946956e-02 1.47724497e+00 2.74296790e-01
-2.77741581e-01 1.50315151e-01 -1.62789524e+00 6.43796563e-01
7.37767875e-01 -3.22235793e-01 -3.35259169e-01 -1.03860430e-01
6.22577190e-01 -2.98874587e-01 1.08592557e-02 -8.85339379e-01
5.49861848e-01 -4.60477114e-01 5.80352247e-01 -7.11505353e-01
-1.74570903e-01 -3.08303475e-01 -3.94639105e-01 -7.71039203e-02
-4.97490078e-01 7.74902344e-01 5.89083016e-01 -1.76801756e-01
-4.82030720e-01 -3.41975540e-01 2.96608120e-01 -2.14590877e-01
-9.98405442e-02 -6.21298254e-02 -5.63954055e-01 3.34274136e-02
7.12431610e-01 -3.99022549e-01 -1.01953065e+00 -8.37166131e-01
-4.21388805e-01 3.85855556e-01 6.26465976e-01 2.65658259e-01
6.38195515e-01 -9.09303844e-01 -5.05225599e-01 -1.57852590e-01
1.24593668e-01 1.89668521e-01 2.29547009e-01 8.84179413e-01
-5.73682964e-01 4.29666609e-01 -2.46229321e-01 -2.98459768e-01
-1.55434132e+00 -3.10193568e-01 -3.14343780e-01 -1.76453501e-01
-2.17299238e-01 9.08387125e-01 -1.95976913e-01 -4.70286459e-01
2.24381760e-01 -5.40206015e-01 -6.47666395e-01 4.80514914e-01
6.68532491e-01 4.87288177e-01 5.14855981e-01 -9.44987983e-02
-1.46623507e-01 1.68773785e-01 -5.07141173e-01 -3.06245208e-01
1.56670606e+00 -1.42760754e-01 1.27926484e-01 1.01905572e+00
9.35343802e-01 9.43836987e-01 -6.59185648e-01 2.00086832e-01
-1.11687988e-01 -5.96477449e-01 -8.22834373e-02 -7.87170589e-01
-6.49525225e-01 8.35879147e-01 3.30235027e-02 2.88341820e-01
8.78732085e-01 -3.59215230e-01 3.18931580e-01 3.56575608e-01
-2.32795537e-01 -1.03382254e+00 4.08709168e-01 4.47606236e-01
6.07088685e-01 -1.31488621e+00 7.38006115e-01 -4.24303204e-01
-6.92235887e-01 1.41535640e+00 9.48589623e-01 2.17614323e-01
5.04840851e-01 5.49884260e-01 2.92943120e-01 -4.94384259e-01
-1.26201177e+00 4.54456270e-01 1.88250631e-01 1.04392208e-01
1.35279787e+00 1.02983303e-01 -4.92267221e-01 7.60600030e-01
-7.51889884e-01 -2.11439788e-01 1.51423240e+00 1.54930127e+00
-7.68112123e-01 -1.09960747e+00 -6.73167408e-01 7.93869734e-01
-7.55863011e-01 6.65376056e-03 -6.05486631e-01 7.13678241e-01
-5.19175112e-01 1.13294792e+00 4.67754528e-02 -4.31768239e-01
4.80978400e-01 1.43411383e-01 2.35656321e-01 -1.12319541e+00
-9.02636051e-01 -7.39979520e-02 7.18617812e-02 -2.55681127e-01
-1.49447218e-01 -8.42501640e-01 -1.25826061e+00 -6.61287487e-01
-2.61966348e-01 5.78706563e-01 9.53739166e-01 5.24951875e-01
6.37519896e-01 8.41004133e-01 4.68499273e-01 -1.60465792e-01
-7.43308723e-01 -9.83200073e-01 -4.38442856e-01 1.29287854e-01
3.53676319e-01 -7.20897138e-01 -4.35196519e-01 1.49588659e-01]
|
[11.18919563293457, 9.262994766235352]
|
ebc66325-f652-448e-8d53-f65519f601ba
|
reinforced-axial-refinement-network-for
|
2008.13748
| null |
https://arxiv.org/abs/2008.13748v1
|
https://arxiv.org/pdf/2008.13748v1.pdf
|
Reinforced Axial Refinement Network for Monocular 3D Object Detection
|
Monocular 3D object detection aims to extract the 3D position and properties of objects from a 2D input image. This is an ill-posed problem with a major difficulty lying in the information loss by depth-agnostic cameras. Conventional approaches sample 3D bounding boxes from the space and infer the relationship between the target object and each of them, however, the probability of effective samples is relatively small in the 3D space. To improve the efficiency of sampling, we propose to start with an initial prediction and refine it gradually towards the ground truth, with only one 3d parameter changed in each step. This requires designing a policy which gets a reward after several steps, and thus we adopt reinforcement learning to optimize it. The proposed framework, Reinforced Axial Refinement Network (RAR-Net), serves as a post-processing stage which can be freely integrated into existing monocular 3D detection methods, and improve the performance on the KITTI dataset with small extra computational costs.
|
['Jie zhou', 'Jiwen Lu', 'Lijie Liu', 'Chufan Wu', 'Qi Tian', 'Lingxi Xie']
|
2020-08-31
| null |
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2822_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123620528.pdf
|
eccv-2020-8
|
['vehicle-pose-estimation']
|
['computer-vision']
|
[ 1.20610058e-01 7.41790235e-02 -9.61359888e-02 -1.50580823e-01
-4.91771162e-01 -4.17222649e-01 4.28246230e-01 -3.07996184e-01
-7.08551347e-01 6.05549455e-01 -2.07098544e-01 -1.34438068e-01
-6.04806235e-04 -7.08445191e-01 -8.06581914e-01 -7.75885403e-01
1.95462093e-01 8.28882337e-01 6.88152850e-01 4.72064108e-01
4.96319056e-01 8.41859519e-01 -1.49098837e+00 -1.74777985e-01
6.66640878e-01 1.24047530e+00 6.21987998e-01 6.28468871e-01
7.58772418e-02 5.29469311e-01 -2.60385633e-01 -8.94117728e-02
5.48962891e-01 -1.90436006e-01 -5.19384146e-01 4.83516246e-01
2.30704218e-01 -7.16655612e-01 -2.69912958e-01 1.26452780e+00
4.81600702e-01 1.51172094e-02 7.56174684e-01 -1.02876735e+00
-1.05536178e-01 1.34160757e-01 -9.76445436e-01 -1.58558153e-02
3.05731326e-01 3.46167952e-01 8.26322734e-01 -1.10190797e+00
6.80965066e-01 1.27086282e+00 4.15512621e-01 6.37512684e-01
-1.10438967e+00 -6.30275428e-01 4.39910412e-01 7.97306448e-02
-1.25749242e+00 -3.41607332e-01 9.67108548e-01 -4.45018351e-01
6.05343997e-01 -6.47531226e-02 1.04918432e+00 9.36498165e-01
-9.36758742e-02 1.08927464e+00 1.16357040e+00 -3.17815006e-01
3.68018031e-01 1.61233783e-01 -1.87297538e-01 6.21123254e-01
3.69633257e-01 3.82490039e-01 -3.23587716e-01 -6.73919870e-03
1.17713869e+00 1.85582891e-01 -2.49313936e-01 -1.06634736e+00
-9.71678436e-01 6.02895439e-01 6.26179099e-01 -1.57835886e-01
-3.56262177e-01 2.21004084e-01 -7.38827288e-02 -2.90532764e-02
4.83458698e-01 3.13492358e-01 -2.85164833e-01 1.68550253e-01
-7.35838115e-01 5.05459845e-01 2.37869829e-01 1.12017381e+00
1.02977061e+00 -2.20947072e-01 -4.42091376e-02 6.94454610e-01
4.46987599e-01 5.34832001e-01 5.20702936e-02 -8.62683594e-01
4.79113221e-01 9.14268792e-01 5.56584716e-01 -6.91954792e-01
-2.69309521e-01 -3.64115685e-01 -5.40209115e-01 6.98936760e-01
4.63611126e-01 -3.11443478e-01 -9.39878523e-01 1.28341627e+00
9.77659047e-01 6.38029864e-03 -3.34457159e-01 1.37426257e+00
4.80915904e-01 4.82040167e-01 -3.08699667e-01 -6.35462850e-02
1.06894815e+00 -8.10257196e-01 -2.01275900e-01 -6.63135409e-01
1.80269971e-01 -6.25956833e-01 7.58790314e-01 4.09540862e-01
-1.08892775e+00 -5.04622459e-01 -9.89511549e-01 8.75195675e-03
-9.52469781e-02 4.05582100e-01 4.82242107e-01 3.50713909e-01
-6.33439064e-01 2.91728050e-01 -9.09277618e-01 -4.16891500e-02
6.91279590e-01 4.46569771e-01 -2.86531687e-01 -2.54470825e-01
-7.14359939e-01 8.06226015e-01 6.24739468e-01 2.66156226e-01
-1.09809017e+00 -4.85264361e-01 -8.41222644e-01 -2.06541598e-01
9.75848198e-01 -7.59804070e-01 1.17959893e+00 -5.84494412e-01
-1.62610292e+00 9.44615543e-01 -8.60714242e-02 -3.76349866e-01
9.08480287e-01 -5.36951661e-01 4.32221651e-01 2.67213106e-01
-3.88977677e-02 9.42419231e-01 1.17691660e+00 -1.37902200e+00
-9.23694611e-01 -7.60981083e-01 2.24893257e-01 6.45897746e-01
1.67653352e-01 -1.28712282e-01 -6.82955205e-01 -1.25006065e-01
5.40208578e-01 -8.48828077e-01 -5.12764931e-01 5.36599994e-01
-4.31412995e-01 -2.56284267e-01 6.38394833e-01 -2.64384121e-01
4.63558346e-01 -1.99904811e+00 3.25641572e-01 -9.70212296e-02
2.17456236e-01 2.09803671e-01 2.27934346e-01 -1.82332575e-01
3.46120447e-01 -3.76396745e-01 -1.70445204e-01 -4.72494513e-01
-1.74699277e-01 -1.29000798e-01 -1.05895020e-01 8.49998951e-01
5.19534588e-01 8.06318820e-01 -9.90196645e-01 -5.37350714e-01
5.04668057e-01 3.29662889e-01 -7.51902878e-01 5.01161575e-01
-5.25347352e-01 4.10477072e-01 -7.97379553e-01 6.41730130e-01
1.06684649e+00 -2.51174271e-01 -2.11685613e-01 -4.84639779e-02
-3.23174864e-01 1.49569243e-01 -1.52614462e+00 1.59875393e+00
-2.48472378e-01 3.78003746e-01 2.73100287e-01 -8.14965427e-01
1.06955874e+00 -2.82472931e-02 3.34656119e-01 -3.14879179e-01
3.43444735e-01 1.14166394e-01 -1.62909985e-01 -4.24013048e-01
4.58986729e-01 -1.51205175e-02 2.61106551e-01 3.37508291e-01
-1.95511311e-01 -5.66786945e-01 -2.09236220e-01 -2.47907013e-01
7.72384107e-01 5.66717923e-01 3.26734364e-01 2.72105545e-01
6.02533281e-01 7.05158785e-02 6.45513237e-01 7.30431736e-01
-1.21680208e-01 7.34439313e-01 4.73009706e-01 -4.11527097e-01
-1.07498753e+00 -8.66866350e-01 -1.38836533e-01 3.32336724e-01
6.62762344e-01 2.00666711e-01 -5.02687931e-01 -9.22832489e-01
1.62654042e-01 3.21229666e-01 -5.95262647e-01 9.51342378e-03
-5.61211586e-01 -4.36232597e-01 -1.76403090e-01 3.70781869e-01
5.70000231e-01 -1.14412487e+00 -1.27173483e+00 1.50959536e-01
1.16733782e-01 -1.24316657e+00 -2.58869737e-01 5.15059829e-01
-9.56216753e-01 -1.19481504e+00 -8.38528097e-01 -5.30631006e-01
8.90793025e-01 5.39165139e-01 7.45731890e-01 -1.59526005e-01
-3.02356899e-01 1.40439257e-01 -2.46373564e-01 -3.99645805e-01
-3.33364122e-03 1.09739073e-01 -1.45484298e-01 4.59611975e-03
3.90413582e-01 -2.55561143e-01 -8.50374162e-01 4.66829211e-01
-5.43315887e-01 1.81298614e-01 7.57356763e-01 8.95288587e-01
7.65735149e-01 -4.49595749e-02 1.66192740e-01 -6.95311904e-01
-7.60848494e-03 -1.47214413e-01 -1.28451347e+00 -2.17277572e-01
-3.60138983e-01 -4.55209091e-02 4.48168308e-01 -4.46744889e-01
-9.95903671e-01 6.59491658e-01 1.33500874e-01 -8.34955812e-01
-2.73954302e-01 -2.25701146e-02 -2.05674425e-01 -5.34113236e-02
4.34518725e-01 2.82340258e-01 2.19787806e-02 -5.23659468e-01
2.28338838e-01 5.88580668e-01 3.63771558e-01 -4.04087812e-01
1.05705357e+00 6.88430190e-01 1.10028379e-01 -6.02877975e-01
-1.10201347e+00 -6.65148139e-01 -9.14483547e-01 -3.82134795e-01
6.99809730e-01 -1.08476162e+00 -8.42712402e-01 4.24559921e-01
-1.23953974e+00 -2.47074977e-01 -2.68073440e-01 6.53498173e-01
-6.88410819e-01 2.10194886e-01 -1.86214775e-01 -1.19629979e+00
-1.22092076e-01 -1.16347528e+00 1.28572381e+00 2.07130820e-01
1.40739039e-01 -4.34345812e-01 -1.35818273e-01 2.45323837e-01
-7.01351389e-02 1.32997129e-02 6.72027946e-01 -2.13977601e-02
-1.18235373e+00 -3.43976706e-01 -4.30927634e-01 3.12626332e-01
-6.78425059e-02 -2.89905638e-01 -1.02908742e+00 -2.99775094e-01
2.49432176e-01 -5.37746012e-01 8.26736450e-01 6.13319278e-01
1.31007683e+00 -3.64968218e-02 -5.15213966e-01 6.28807962e-01
1.36160266e+00 1.79232582e-01 2.21105173e-01 2.71730065e-01
6.12113595e-01 5.84586978e-01 1.10024822e+00 5.51312685e-01
-4.93353494e-02 7.40653515e-01 9.69578624e-01 1.14951283e-01
4.62396257e-03 -4.05300319e-01 1.24652669e-01 9.36653316e-02
-2.55560707e-02 8.28754250e-03 -5.92081785e-01 4.32076007e-01
-1.73285782e+00 -7.12653160e-01 2.92682350e-01 2.32735538e+00
7.53431857e-01 5.71095586e-01 1.79315075e-01 1.53346509e-01
6.40436471e-01 1.43199325e-01 -1.06344378e+00 3.58788490e-01
1.39306277e-01 -2.10787505e-01 5.63185155e-01 6.25419557e-01
-1.08017743e+00 1.11723721e+00 5.57858038e+00 6.72352970e-01
-1.13143039e+00 -3.74446332e-01 5.57182789e-01 -3.00876230e-01
3.80948484e-02 4.31351215e-02 -1.30773568e+00 2.93061644e-01
1.34788960e-01 1.77461594e-01 4.89690959e-01 1.06388462e+00
2.09979251e-01 -4.49881464e-01 -1.37302363e+00 1.07252669e+00
-1.14608452e-01 -1.04224420e+00 -1.15353093e-01 4.96266894e-02
5.01835763e-01 4.10458222e-02 -2.60852315e-02 -8.82481970e-03
2.72124469e-01 -7.60938168e-01 9.08415794e-01 3.26879680e-01
6.22424781e-01 -8.27553093e-01 5.11221588e-01 8.88095737e-01
-9.68536317e-01 -1.56903014e-01 -7.59775400e-01 -8.11453760e-02
3.98695879e-02 6.73686326e-01 -1.09612679e+00 2.11277544e-01
7.43255377e-01 8.28423619e-01 -3.26356679e-01 1.46827686e+00
-4.13307309e-01 1.08550109e-01 -3.55972052e-01 -3.27309340e-01
2.04186887e-01 -4.27705884e-01 8.33566248e-01 6.66119337e-01
2.54967034e-01 -2.83702668e-02 2.96904981e-01 1.22263646e+00
7.25299641e-02 -2.55062193e-01 -5.36929309e-01 4.01902556e-01
5.60398042e-01 1.24957621e+00 -7.57515907e-01 -1.86308503e-01
-2.79428273e-01 8.59858036e-01 5.16008079e-01 2.01468721e-01
-6.26950741e-01 -2.12227583e-01 3.75628412e-01 1.99873894e-01
7.39442825e-01 -1.58094823e-01 -2.60754734e-01 -1.13170922e+00
2.07403719e-01 -7.51989603e-01 4.14799340e-02 -9.26075995e-01
-1.12204075e+00 4.49571341e-01 3.51966992e-02 -1.50247610e+00
-1.56587660e-01 -7.96055198e-01 -1.04082391e-01 9.46829140e-01
-1.69406950e+00 -7.59591997e-01 -5.24227262e-01 3.54701072e-01
6.30723119e-01 1.48674726e-01 2.79447377e-01 3.08684520e-02
-4.23275083e-01 1.81448713e-01 -2.94519633e-01 2.12824624e-02
4.32557791e-01 -1.26435876e+00 2.35367686e-01 6.91644132e-01
1.37915462e-02 1.57155260e-01 5.71699440e-01 -5.60038865e-01
-1.33360052e+00 -1.05399394e+00 4.33393568e-01 -4.72341388e-01
3.72584850e-01 -6.78133726e-01 -6.58188581e-01 4.47757751e-01
-2.25353003e-01 2.89671779e-01 -1.41293675e-01 -1.45099849e-01
-6.37314934e-03 -9.56021100e-02 -1.21149123e+00 6.89729333e-01
1.31294072e+00 -3.36043060e-01 -4.29484397e-01 2.22108424e-01
6.91730022e-01 -8.15965474e-01 -3.70607883e-01 3.55838150e-01
4.36803907e-01 -1.06361306e+00 1.01164901e+00 -2.52625853e-01
3.81771505e-01 -5.68290830e-01 6.22846670e-02 -1.23495948e+00
-1.82977304e-01 -5.89537978e-01 -2.46274084e-01 7.58694053e-01
1.59102961e-01 -4.24613863e-01 1.32451582e+00 3.62814724e-01
5.54536059e-02 -9.06970024e-01 -1.00408506e+00 -5.67084491e-01
-2.11333916e-01 -4.58710253e-01 4.73327219e-01 2.76872128e-01
-4.14847016e-01 4.24637616e-01 -3.77856225e-01 4.84957933e-01
8.98837566e-01 5.71700275e-01 1.06006825e+00 -1.29914951e+00
-4.37871724e-01 -2.87244022e-01 -3.71220469e-01 -1.99212718e+00
-8.88682529e-02 -4.77307618e-01 2.48980567e-01 -1.09495974e+00
2.58887380e-01 -5.65731943e-01 5.81125058e-02 1.81736752e-01
-2.28680059e-01 1.02447346e-01 5.02675101e-02 1.59568682e-01
-5.73111594e-01 1.00274563e+00 1.78236604e+00 -6.84327558e-02
-2.81489104e-01 4.66472983e-01 -3.14556986e-01 9.12241518e-01
3.90269130e-01 -5.26247740e-01 -4.43251938e-01 -3.98863882e-01
1.62778959e-01 3.04494292e-01 5.74648857e-01 -9.52212870e-01
2.55833417e-01 -2.77722150e-01 8.39718282e-01 -1.33481753e+00
6.56545162e-01 -1.28009212e+00 -3.36701900e-01 6.05628133e-01
-1.84258401e-01 -4.17934507e-01 -1.69643193e-01 6.85536861e-01
5.42044193e-02 -5.25929332e-01 1.00873017e+00 -3.19968909e-01
-5.57792723e-01 6.71863079e-01 -8.24723914e-02 -3.83072682e-02
1.19837940e+00 -4.61493939e-01 9.37475190e-02 -1.75104365e-02
-6.31083250e-01 4.07881021e-01 6.05674148e-01 1.44758016e-01
8.87897313e-01 -1.20719373e+00 -4.14956212e-01 3.95363122e-01
2.59676222e-02 8.35607409e-01 -1.21291108e-01 4.26898092e-01
-4.94622558e-01 3.74315053e-01 -6.02014028e-02 -9.54072773e-01
-1.05764711e+00 6.81775928e-01 5.62778354e-01 5.10983169e-02
-8.35399389e-01 8.22622836e-01 3.47645789e-01 -3.85455370e-01
5.49420357e-01 -4.72717464e-01 -3.29466581e-01 -1.11782625e-01
3.75868201e-01 3.64126563e-01 -2.51978368e-01 -1.23815626e-01
-2.21230000e-01 7.95377791e-01 -3.04382503e-01 -9.23925266e-02
1.43358588e+00 -2.56804287e-01 1.83773413e-01 4.78518486e-01
9.22232747e-01 -2.74131030e-01 -2.13525248e+00 -4.58786935e-01
-2.08169371e-01 -9.09582019e-01 1.57428458e-01 -4.68522280e-01
-9.64295208e-01 8.62190962e-01 5.93577564e-01 7.61395842e-02
8.49258661e-01 1.21365003e-01 4.37493801e-01 4.43247408e-01
5.58104217e-01 -8.47881675e-01 2.21363053e-01 3.68745059e-01
8.81017566e-01 -1.40406668e+00 1.56044945e-01 -5.67918122e-01
-2.87761658e-01 1.02562654e+00 9.27863717e-01 -5.09718776e-01
4.92591977e-01 3.74604836e-02 -2.18762115e-01 -1.74840778e-01
-6.02914155e-01 -2.18043610e-01 2.76876420e-01 5.60874104e-01
-1.01135090e-01 -2.51915514e-01 9.86419693e-02 1.02702610e-01
6.65448830e-02 -7.34619703e-03 3.23163778e-01 6.83386505e-01
-6.46146357e-01 -8.80154908e-01 -5.01944840e-01 2.88062960e-01
-1.35672718e-01 2.45269641e-01 -3.54710847e-01 7.79166758e-01
1.66149452e-01 5.69993854e-01 6.90121949e-02 -5.48421107e-02
3.57013136e-01 -3.56012344e-01 6.44266844e-01 -7.75417447e-01
-1.44762238e-02 3.48199278e-01 -2.79043138e-01 -5.92296183e-01
-4.22598511e-01 -8.58212233e-01 -1.02885735e+00 7.26934448e-02
-5.96697450e-01 -1.49801761e-01 5.87236106e-01 7.91736364e-01
9.14342627e-02 1.77092463e-01 9.85615969e-01 -1.26195073e+00
-7.77130604e-01 -8.37379277e-01 -3.79046947e-01 -1.80199984e-02
5.33250034e-01 -9.37067091e-01 -4.57479745e-01 -3.01674575e-01]
|
[7.794862747192383, -2.525606155395508]
|
50132051-c75a-4144-ad5e-c4d751daad2c
|
masked-trajectory-models-for-prediction
|
2305.02968
| null |
https://arxiv.org/abs/2305.02968v1
|
https://arxiv.org/pdf/2305.02968v1.pdf
|
Masked Trajectory Models for Prediction, Representation, and Control
|
We introduce Masked Trajectory Models (MTM) as a generic abstraction for sequential decision making. MTM takes a trajectory, such as a state-action sequence, and aims to reconstruct the trajectory conditioned on random subsets of the same trajectory. By training with a highly randomized masking pattern, MTM learns versatile networks that can take on different roles or capabilities, by simply choosing appropriate masks at inference time. For example, the same MTM network can be used as a forward dynamics model, inverse dynamics model, or even an offline RL agent. Through extensive experiments in several continuous control tasks, we show that the same MTM network -- i.e. same weights -- can match or outperform specialized networks trained for the aforementioned capabilities. Additionally, we find that state representations learned by MTM can significantly accelerate the learning speed of traditional RL algorithms. Finally, in offline RL benchmarks, we find that MTM is competitive with specialized offline RL algorithms, despite MTM being a generic self-supervised learning method without any explicit RL components. Code is available at https://github.com/facebookresearch/mtm
|
['Aravind Rajeswaran', 'Pieter Abbeel', 'Igor Mordatch', 'Yixin Lin', 'Kevin Stone', 'Arjun Majumdar', 'Philipp Wu']
|
2023-05-04
| null | null | null | null |
['offline-rl', 'continuous-control']
|
['playing-games', 'playing-games']
|
[ 1.59249291e-01 3.14398617e-01 -9.32182431e-01 -5.11659384e-02
-4.44516599e-01 -8.89283836e-01 8.18656623e-01 -4.05769914e-01
-3.12814504e-01 6.66732192e-01 2.74895191e-01 -6.28550708e-01
-6.64390400e-02 -3.55889559e-01 -1.03080761e+00 -7.95865059e-01
-3.40970188e-01 5.83611965e-01 1.20917857e-01 -3.00786525e-01
2.55996704e-01 3.92161965e-01 -1.35903788e+00 1.31460041e-01
4.38461721e-01 7.62755215e-01 2.88692623e-01 7.10688949e-01
1.48277253e-01 1.27760601e+00 -3.99487317e-01 1.65035009e-01
4.97825623e-01 -4.79064375e-01 -8.05571318e-01 -7.02897459e-02
7.24286139e-02 -4.25048292e-01 -7.94984400e-01 6.64253891e-01
3.42895836e-01 4.24144506e-01 4.12486881e-01 -1.38981140e+00
-2.81461030e-01 1.21030927e+00 4.54117022e-02 8.54158923e-02
2.43525684e-01 8.95084679e-01 9.45241332e-01 -6.27308071e-01
7.23703861e-01 1.24338388e+00 5.35287678e-01 1.04462087e+00
-1.62438047e+00 -6.78868234e-01 7.09974468e-01 -3.04633919e-02
-9.10176635e-01 -7.83768415e-01 6.08443856e-01 -2.41736680e-01
9.78699744e-01 -3.82874832e-02 6.39976501e-01 1.82879150e+00
4.21206236e-01 1.18961871e+00 1.12259841e+00 6.41289130e-02
4.61617798e-01 -2.89935648e-01 -4.35061045e-02 9.23593044e-01
-1.15883984e-02 5.84472656e-01 -6.77437186e-01 -5.86528853e-02
9.31660175e-01 2.00124890e-01 -2.74315804e-01 -5.23102283e-01
-1.66578078e+00 6.71230733e-01 3.21791261e-01 1.27334371e-01
-8.27133432e-02 7.47658849e-01 3.66786242e-01 7.57249355e-01
1.61699206e-01 6.58504009e-01 -7.20581651e-01 -1.93303034e-01
-6.56191885e-01 4.18158621e-01 9.12385762e-01 8.04543436e-01
7.63687074e-01 3.71700078e-01 -3.92587870e-01 6.02139905e-02
2.79675126e-02 3.73481512e-01 6.33712113e-01 -1.65664327e+00
4.82068568e-01 3.73922378e-01 4.80030417e-01 -4.11576331e-01
-6.33416057e-01 -4.46394652e-01 -9.27569211e-01 4.95448828e-01
5.59561074e-01 -5.09404421e-01 -8.85863960e-01 2.22197413e+00
1.98220909e-01 5.65010071e-01 2.46688545e-01 7.26788819e-01
1.15235597e-01 7.48062015e-01 -2.02602729e-01 -8.75484496e-02
6.64761841e-01 -1.21473825e+00 -6.00545466e-01 -4.23195541e-01
6.50455952e-01 -8.25939700e-02 1.19803059e+00 4.79460090e-01
-1.23407078e+00 -5.00449181e-01 -1.02953160e+00 3.14483106e-01
-8.43977779e-02 3.05236071e-01 6.43159330e-01 1.49854317e-01
-1.24080825e+00 1.37532115e+00 -1.24348307e+00 -1.57662019e-01
1.88728869e-01 5.48389137e-01 -4.11183983e-02 2.16366962e-01
-9.99920607e-01 1.02276075e+00 3.89105588e-01 1.56757951e-01
-1.84299946e+00 -7.65829861e-01 -6.99520051e-01 -2.21147150e-01
5.84671557e-01 -7.26798832e-01 1.79559219e+00 -1.11799788e+00
-2.23480892e+00 4.37933922e-01 -2.60845542e-01 -9.70414042e-01
6.94448709e-01 -2.33210579e-01 -2.94071019e-01 -3.65887210e-02
-5.69014847e-02 6.27525270e-01 1.40601194e+00 -1.01883471e+00
-5.01027882e-01 7.37149492e-02 3.15272063e-01 1.32418439e-01
4.42064613e-01 -2.18473077e-01 4.93426733e-02 -4.37280357e-01
-8.03493261e-02 -1.24988008e+00 -5.93707740e-01 -2.14988384e-02
-5.73967457e-01 -2.37926900e-01 6.91472709e-01 -3.62405241e-01
1.12468982e+00 -2.02041841e+00 6.44080818e-01 -2.42421608e-02
3.73808742e-01 1.23988397e-01 -2.64893770e-01 6.11451030e-01
-2.54104156e-02 1.00803245e-02 -3.47514331e-01 -5.84930539e-01
3.05071950e-01 4.25891340e-01 -7.54407465e-01 6.22737527e-01
2.36630276e-01 1.15694904e+00 -1.08474636e+00 1.81029062e-03
2.42982775e-01 -6.94957003e-02 -3.16697508e-01 2.39308581e-01
-8.87319267e-01 1.06814837e+00 -4.04652089e-01 3.70441258e-01
7.58638419e-03 -4.75216210e-01 5.60142696e-01 2.43909925e-01
-1.66417077e-01 6.17405295e-01 -1.05915916e+00 2.01344728e+00
-6.24653637e-01 6.99601948e-01 1.99951753e-01 -1.04090929e+00
5.11440039e-01 4.76813227e-01 4.56746936e-01 -4.38775897e-01
1.60506323e-01 1.26166150e-01 1.50994658e-01 -1.70506001e-01
2.56933421e-01 2.11194247e-01 -2.22960263e-01 9.23854649e-01
1.91286504e-01 2.59996891e-01 8.12725648e-02 7.56626949e-02
1.33074820e+00 5.75830579e-01 1.92852929e-01 -9.75770056e-02
1.81562245e-01 1.82154298e-01 6.82873249e-01 1.03790319e+00
-2.03638539e-01 7.48521164e-02 6.24314845e-01 -6.86749458e-01
-8.14577222e-01 -1.21124649e+00 3.72853488e-01 1.47638834e+00
-5.26132137e-02 -3.13088059e-01 -3.56259704e-01 -7.44114816e-01
1.91959351e-01 8.02816093e-01 -7.45837212e-01 -4.67381179e-01
-9.41774964e-01 -3.28641444e-01 6.20481670e-01 4.48683500e-01
3.38090509e-01 -1.36311138e+00 -7.27636218e-01 3.56470793e-01
6.00888245e-02 -8.09298217e-01 -6.49940670e-01 4.74075794e-01
-1.09834015e+00 -9.90943849e-01 -3.27129662e-01 -6.99488401e-01
5.94508350e-01 2.59193599e-01 1.19675970e+00 5.15971892e-02
1.91221237e-01 4.28604394e-01 -9.93915349e-02 5.58382347e-02
-7.44477391e-01 4.19801176e-01 5.08147001e-01 8.33707079e-02
-3.80688012e-01 -9.48417306e-01 -6.30132735e-01 3.93652946e-01
-5.24300098e-01 2.58881360e-01 5.43276548e-01 6.65425122e-01
3.38772774e-01 -2.30414450e-01 5.08193672e-01 -8.18514466e-01
5.77230215e-01 -6.48130298e-01 -6.92825675e-01 -1.26124527e-02
-6.91647887e-01 5.70125639e-01 9.15218651e-01 -1.09793639e+00
-7.87730098e-01 3.09571862e-01 2.30870694e-01 -7.64773905e-01
-7.73913041e-02 1.26912937e-01 1.28727257e-01 1.03648059e-01
6.33290112e-01 5.84763229e-01 3.59504342e-01 -4.51042891e-01
6.89096689e-01 -5.79052158e-02 4.11456972e-01 -8.34895015e-01
8.29541385e-01 5.00601232e-01 -4.98028137e-02 -1.51168779e-01
-9.81219947e-01 -1.25768900e-01 -5.25324821e-01 -1.94432616e-01
6.62749767e-01 -9.41662431e-01 -1.14911103e+00 4.29295540e-01
-9.13526595e-01 -1.53411138e+00 -5.33700645e-01 2.13626280e-01
-9.31197166e-01 -2.56249815e-01 -8.38533044e-01 -6.49703562e-01
-3.67549658e-02 -1.17427254e+00 7.96169758e-01 4.78192465e-03
-3.61802846e-01 -9.90951061e-01 2.43839532e-01 -1.69664100e-01
4.46202606e-01 1.58222556e-01 7.27702498e-01 -5.52099943e-01
-7.70727456e-01 1.75972283e-01 4.80929554e-01 9.96968448e-02
3.37113179e-02 -6.48795888e-02 -7.31896281e-01 -5.13080239e-01
-1.49194747e-01 -6.37079716e-01 1.06559551e+00 2.71977574e-01
1.04748416e+00 -7.94166088e-01 -3.98250610e-01 6.71965182e-01
1.11429882e+00 1.57527462e-01 3.16147596e-01 7.84831569e-02
6.87581182e-01 3.62414062e-01 3.36587012e-01 4.43275452e-01
3.59514594e-01 3.44084948e-01 4.90141541e-01 4.16472435e-01
7.76158199e-02 -7.68343508e-01 1.22426057e+00 6.89106464e-01
-7.51969889e-02 -5.12421876e-02 -6.09834433e-01 2.96194017e-01
-2.33935261e+00 -1.11868763e+00 4.07148778e-01 2.20789528e+00
9.17823076e-01 2.27498114e-01 3.89559239e-01 -1.19761303e-01
4.50842798e-01 5.09749711e-01 -1.36798668e+00 -2.84810901e-01
1.18449844e-01 2.86767066e-01 7.47340322e-01 5.26722074e-01
-1.05250490e+00 1.14752328e+00 6.83441114e+00 7.86997914e-01
-1.22027719e+00 2.94363111e-01 3.27287912e-01 -3.43727559e-01
-1.58049390e-01 1.78118512e-01 -7.73732960e-01 3.79613221e-01
1.28644609e+00 -4.71753366e-02 1.11240017e+00 7.40644217e-01
4.34218884e-01 1.65871918e-01 -1.42913306e+00 5.88982224e-01
-4.77565348e-01 -1.66243446e+00 -2.07016408e-01 -5.26080020e-02
8.58231544e-01 3.15825820e-01 2.63593465e-01 7.47244477e-01
1.15791333e+00 -1.14956200e+00 1.04200423e+00 5.61308622e-01
6.07077181e-01 -5.52525759e-01 -6.72292039e-02 7.99822628e-01
-1.28289211e+00 -3.78286839e-01 -8.73981342e-02 -2.82322317e-01
-4.00041528e-02 -5.36924377e-02 -5.78320801e-01 1.49986222e-01
2.75340408e-01 1.10035837e+00 -8.30792636e-02 4.54492211e-01
-6.22546017e-01 9.36153173e-01 -1.51987061e-01 -8.13000575e-02
3.89509559e-01 -2.90452808e-01 5.80447733e-01 8.20211887e-01
-1.00309588e-01 -2.89060235e-01 5.35835564e-01 1.06519043e+00
-8.24697539e-02 -5.54513454e-01 -8.05852056e-01 -2.27125764e-01
3.33757192e-01 1.09051776e+00 -4.46206331e-01 -4.00067031e-01
-8.27607661e-02 1.17596149e+00 6.41090333e-01 6.96418285e-01
-9.53005791e-01 2.24111646e-01 1.13831413e+00 -9.49804559e-02
3.37071657e-01 -6.27200723e-01 5.29198684e-02 -1.32290447e+00
-2.98056483e-01 -1.08154821e+00 5.12592718e-02 -5.65921307e-01
-9.06405687e-01 1.79355890e-01 -2.18437836e-01 -1.43949854e+00
-5.80274463e-01 -4.41111505e-01 -8.35987449e-01 4.51465279e-01
-1.14076293e+00 -7.69550920e-01 2.95679152e-01 7.45051861e-01
8.18475425e-01 -2.08364978e-01 6.64871633e-01 -2.30240807e-01
-1.02348149e+00 2.93437928e-01 9.42438319e-02 1.12969078e-01
3.88201177e-01 -1.33796966e+00 7.23712444e-01 6.73383892e-01
1.65752545e-01 7.89331019e-01 5.78234851e-01 -5.70396900e-01
-1.93543935e+00 -1.41694808e+00 3.68391395e-01 -6.67766094e-01
9.64562893e-01 -4.99473125e-01 -5.34231126e-01 1.33573389e+00
2.81286865e-01 -5.79157285e-02 8.31654295e-02 -9.73202512e-02
-2.76616186e-01 4.45907153e-02 -7.97266066e-01 1.08750999e+00
1.48680449e+00 -5.42112947e-01 -4.04355645e-01 4.76355702e-01
1.20370388e+00 -6.92166746e-01 -6.43253207e-01 -2.38404311e-02
5.65433264e-01 -7.21186996e-01 9.66791213e-01 -1.13637602e+00
2.96399951e-01 -4.39718843e-01 3.12458840e-03 -1.47207046e+00
-3.55488032e-01 -1.45790517e+00 -9.45137441e-01 6.08365059e-01
6.02299571e-01 -1.03485441e+00 7.12214530e-01 2.41539001e-01
-1.12280756e-01 -9.42112684e-01 -8.18808794e-01 -1.05109942e+00
1.82834893e-01 -5.80438018e-01 6.87733591e-01 6.44556344e-01
-2.11996421e-01 2.98973083e-01 -6.07321024e-01 2.32675031e-01
5.06262422e-01 3.17039162e-01 7.43986845e-01 -8.77985597e-01
-7.29619384e-01 -7.12948442e-01 3.61520290e-01 -1.42948329e+00
7.59213805e-01 -1.12059343e+00 1.74739033e-01 -1.18410861e+00
-2.78417885e-01 -4.02424127e-01 -4.13381785e-01 7.20009446e-01
1.80724561e-01 -4.03891921e-01 3.53727132e-01 4.86108541e-01
-7.50238240e-01 6.91140294e-01 1.39487433e+00 -9.46688578e-02
-4.90258336e-01 4.50886697e-01 -5.42511046e-01 7.49403954e-01
1.30849802e+00 -5.16555727e-01 -7.06096351e-01 -2.59537488e-01
2.72024781e-01 5.50061047e-01 4.43824291e-01 -9.76173878e-01
3.35047126e-01 -4.13652301e-01 4.40037325e-02 -3.38540107e-01
3.73755455e-01 -5.68375051e-01 5.79306856e-02 8.69586766e-01
-6.80215478e-01 2.86432922e-01 3.86016332e-02 7.87487745e-01
3.42652380e-01 8.50925669e-02 4.27985340e-01 -3.64987999e-01
-7.21150219e-01 4.63572592e-01 -8.42342973e-01 1.70478389e-01
7.87941039e-01 1.01382688e-01 -3.01767319e-01 -4.92453665e-01
-8.34321439e-01 5.15187144e-01 4.44568038e-01 3.36659938e-01
4.46753502e-01 -1.14788759e+00 -3.03197235e-01 7.16910884e-02
-2.19380677e-01 3.98237035e-02 -1.50333047e-01 9.89389539e-01
-6.03179522e-02 2.09870502e-01 -5.56751387e-03 -4.21103001e-01
-3.95029128e-01 7.02062547e-01 5.95006287e-01 -5.93428254e-01
-7.71401584e-01 3.57092142e-01 -1.19155757e-01 -8.78426075e-01
4.27491188e-01 -9.55359995e-01 2.48407513e-01 -1.60678729e-01
4.66440856e-01 4.29589838e-01 -4.29188430e-01 -3.08141410e-02
-1.75686091e-01 3.87087464e-01 2.92533934e-01 -4.87636477e-01
1.09057152e+00 5.03575616e-02 1.03671022e-01 9.83366191e-01
8.21717143e-01 -2.56035805e-01 -2.02089119e+00 -2.58160084e-01
4.75488156e-02 2.08265394e-01 -2.42581129e-01 -7.72692144e-01
-1.06726313e+00 6.10158086e-01 7.88882449e-02 -6.28625005e-02
8.85717630e-01 -1.63695261e-01 6.98218822e-01 8.35552931e-01
8.23899984e-01 -9.49383497e-01 2.03109846e-01 7.60574818e-01
8.75064790e-01 -1.13146949e+00 -3.14590782e-01 4.15555298e-01
-8.26305091e-01 9.62527514e-01 4.39592004e-01 -6.94898188e-01
5.69503069e-01 2.43966654e-01 -2.43357137e-01 1.95139021e-01
-1.48382628e+00 -3.44363987e-01 2.95315739e-02 4.64231849e-01
-8.97917598e-02 2.72286534e-01 3.26594204e-01 4.72703040e-01
-3.89224440e-02 6.16603009e-02 3.89790386e-01 1.01368201e+00
-4.17472363e-01 -1.12373805e+00 -2.74971146e-02 3.70771110e-01
-5.69414021e-03 1.58266246e-01 -2.52030104e-01 8.40272486e-01
-1.56718165e-01 8.44354868e-01 -3.03949509e-02 -7.96091795e-01
-3.20097953e-02 2.16754928e-01 5.50940156e-01 -5.13239622e-01
-8.29914033e-01 -4.66682106e-01 1.48466840e-01 -1.29515207e+00
-2.40428299e-01 -7.68784940e-01 -1.47070634e+00 -5.72203398e-01
3.94757271e-01 -2.17926368e-01 2.94889033e-01 9.92516696e-01
5.28812051e-01 4.22991395e-01 7.88959086e-01 -1.25115478e+00
-9.66304839e-01 -8.70012522e-01 -2.72250026e-01 -8.36065784e-03
8.69412839e-01 -6.13376796e-01 -4.93093252e-01 -6.68617338e-02]
|
[4.12246561050415, 1.7543448209762573]
|
167da35e-192c-4092-be5d-761c3e1cc4c5
|
a-multi-gate-encoder-for-joint-entity-and
| null | null |
https://aclanthology.org/2022.ccl-1.75
|
https://aclanthology.org/2022.ccl-1.75.pdf
|
A Multi-Gate Encoder for Joint Entity and Relation Extraction
|
“Named entity recognition and relation extraction are core sub-tasks of relational triple extraction. Recent studies have used parameter sharing or joint decoding to create interaction between these two tasks. However, ensuring the specificity of task-specific traits while the two tasks interact properly is a huge difficulty. We propose a multi-gate encoder that models bidirectional task interaction while keeping sufficient feature specificity based on gating mechanism in this paper. Precisely, we design two types of independent gates: task gates to generate task-specific features and interaction gates to generate instructive features to guide the opposite task. Our experiments show that our method increases the state-of-the-art (SOTA) relation F1 scores on ACE04, ACE05 and SciERC datasets to 63.8% (+1.3%), 68.2% (+1.4%), 39.4% (+1.0%), respectively, with higher inference speed over previous SOTA model.”
|
['Li Shengyang', 'Gong Shuai', 'Liu Anqi', 'Liu Yunfei', 'Xiong Xiong']
| null | null | null | null |
ccl-2022-10
|
['joint-entity-and-relation-extraction']
|
['natural-language-processing']
|
[ 8.08354188e-03 5.59269726e-01 -3.75253737e-01 -6.87806964e-01
-8.12251508e-01 -4.84746873e-01 5.89749813e-01 3.89363281e-02
-3.70768934e-01 1.09126568e+00 1.24633655e-01 -4.54804629e-01
1.13555692e-01 -1.06243861e+00 -1.01920152e+00 -2.54820466e-01
-4.11822723e-04 4.55279440e-01 1.49158269e-01 -3.38055909e-01
-4.85354662e-02 1.28534302e-01 -1.14759171e+00 5.86185992e-01
9.09934103e-01 7.85857677e-01 -4.19113562e-02 6.05473280e-01
-2.28330791e-01 1.05533576e+00 -4.26503688e-01 -7.65499771e-01
-8.74239430e-02 -1.47281915e-01 -1.03749883e+00 -4.54053223e-01
-8.84549022e-02 -3.86837348e-02 -6.16980612e-01 8.76985431e-01
5.03727376e-01 -1.11149870e-01 4.36798334e-01 -1.25615358e+00
-9.63917553e-01 1.39955437e+00 -6.69240594e-01 1.33486735e-02
1.83493897e-01 -1.46917567e-01 1.39857662e+00 -6.20544732e-01
6.09424055e-01 8.80602062e-01 4.51937526e-01 6.94566607e-01
-1.22053039e+00 -1.11624587e+00 1.39296532e-01 -2.96209212e-02
-1.57854545e+00 -6.02309406e-01 3.52315903e-01 -2.15387225e-01
1.57758248e+00 4.44295883e-01 3.78437459e-01 9.88565266e-01
3.12458634e-01 7.24103034e-01 7.15858102e-01 -3.06074679e-01
-1.92364112e-01 1.16260760e-01 3.89548361e-01 7.56406844e-01
5.31359792e-01 -8.20291862e-02 -6.76196277e-01 -8.49717036e-02
7.36099601e-01 -3.67704511e-01 -2.03760743e-01 1.92134723e-01
-1.06310451e+00 6.03621364e-01 4.95147169e-01 4.03476685e-01
-2.81854123e-01 2.66097277e-01 4.48636025e-01 2.31711239e-01
4.08929050e-01 6.74721181e-01 -1.01050031e+00 -2.07170844e-01
-3.54945600e-01 2.04496816e-01 9.82547641e-01 1.56595695e+00
6.04415715e-01 -2.56968588e-01 -4.67521399e-01 6.42745078e-01
1.98471457e-01 3.23042363e-01 1.67471975e-01 -3.76475990e-01
8.97725821e-01 9.19099152e-01 -9.69266370e-02 -7.33975172e-01
-6.13814831e-01 -6.80327535e-01 -9.57870483e-01 -6.48083925e-01
4.15998518e-01 -3.81089181e-01 -1.11924660e+00 2.12299275e+00
1.67610303e-01 -6.20323978e-02 3.21386337e-01 5.55003822e-01
1.37317789e+00 5.11539817e-01 4.15066868e-01 -1.87645584e-01
1.72234488e+00 -8.72264564e-01 -1.03324878e+00 -4.76354271e-01
1.12380612e+00 -4.98224765e-01 7.62740016e-01 2.50812713e-02
-1.14997482e+00 -5.09011567e-01 -1.13679874e+00 -4.88350958e-01
-2.82257646e-01 4.32661444e-01 1.13694060e+00 6.97451115e-01
-7.42498577e-01 4.27660137e-01 -8.04931581e-01 -1.23799533e-01
5.41493952e-01 7.63298988e-01 -6.30332351e-01 2.58307129e-01
-1.55710626e+00 8.87600660e-01 4.29500043e-01 2.75036335e-01
-6.44863784e-01 -9.06470001e-01 -8.60897005e-01 2.87618369e-01
5.67381382e-01 -7.81776369e-01 1.12018716e+00 -2.96110958e-02
-1.29425573e+00 8.64488482e-01 -4.23095316e-01 -4.57142204e-01
8.45235661e-02 -5.50769567e-01 -4.60005462e-01 -5.42725325e-01
5.00671454e-02 5.91358066e-01 -9.10669416e-02 -7.66814113e-01
-5.80746770e-01 -3.89289498e-01 1.24672621e-01 3.77490930e-02
-1.71255663e-01 2.11654007e-01 -6.64470375e-01 -4.00136650e-01
2.84286607e-02 -7.77536750e-01 -1.62928492e-01 -6.49545252e-01
-8.20239842e-01 -5.08055687e-01 3.07198524e-01 -7.38566458e-01
1.42440689e+00 -1.94791472e+00 2.01011032e-01 -1.91755109e-02
4.43369597e-01 2.43952870e-01 3.49926539e-02 2.16516316e-01
-1.69456050e-01 5.57177484e-01 -1.16959579e-01 -7.87792951e-02
-1.54474527e-01 1.88686457e-02 5.69611378e-02 -2.72007450e-03
8.28647673e-01 1.36455560e+00 -8.25102508e-01 -3.92801285e-01
-3.98053974e-01 3.21640939e-01 -6.37206495e-01 1.87093183e-01
7.78897526e-03 1.66160077e-01 -6.37497127e-01 6.02637827e-01
5.38249373e-01 -3.24725956e-01 6.89902842e-01 -4.33143884e-01
1.32824481e-01 7.45344043e-01 -9.43314672e-01 1.55921125e+00
-4.95364130e-01 4.23515052e-01 -1.89721793e-01 -8.82238090e-01
1.22358763e+00 4.78195190e-01 9.23459902e-02 -7.86622465e-01
1.83332607e-01 2.37288192e-01 3.83483529e-01 -4.91141379e-01
4.82934862e-01 -1.33529484e-01 -4.69899446e-01 1.59708589e-01
3.98634017e-01 7.01121688e-02 1.10694297e-01 1.42723233e-01
1.30872548e+00 2.04165071e-01 4.88076895e-01 -2.99310356e-01
4.71477062e-01 -2.11602733e-01 8.82252336e-01 5.18294811e-01
9.28441286e-02 3.47251773e-01 9.05261874e-01 -3.22014421e-01
-7.98635006e-01 -6.67195737e-01 -1.10097088e-01 9.85166311e-01
-2.51782149e-01 -7.43168533e-01 -4.32937920e-01 -8.67088675e-01
-3.51765677e-02 7.64966309e-01 -8.67459774e-01 -4.74566430e-01
-8.71114910e-01 -1.16433716e+00 9.21502292e-01 9.25891519e-01
6.74989343e-01 -9.83758926e-01 -6.04459457e-02 1.31321386e-01
-1.46207243e-01 -1.38817179e+00 -1.24441206e-01 6.46772265e-01
-6.40491784e-01 -9.29837465e-01 -1.08426459e-01 -6.75001979e-01
4.92065847e-01 -2.01260149e-01 1.37059271e+00 -5.36379255e-02
9.36171860e-02 -7.74856031e-01 -2.07571924e-01 -3.24222893e-01
-1.24879733e-01 6.61205411e-01 -2.79662162e-01 -2.13024408e-01
5.23994088e-01 -4.24827814e-01 -4.90142591e-02 2.42500216e-01
-5.61972678e-01 3.05398554e-01 7.71610618e-01 9.31631923e-01
4.31620806e-01 -1.99878305e-01 8.37694645e-01 -1.51751351e+00
5.62153995e-01 -5.10511279e-01 -4.51363355e-01 5.55538416e-01
-7.12268174e-01 3.21179777e-01 5.32794535e-01 -1.27216071e-01
-1.14051914e+00 1.25942409e-01 -3.57116051e-02 6.80879131e-02
3.56598645e-02 5.16118824e-01 -7.54217565e-01 3.66894126e-01
5.13641596e-01 -2.56620105e-02 -4.40044314e-01 -2.59607434e-01
4.29932564e-01 7.73206711e-01 5.25571823e-01 -7.18410432e-01
4.94895428e-01 7.95253273e-03 -2.83298977e-02 -7.48962685e-02
-9.75921452e-01 -4.94111981e-03 -5.89356422e-01 5.23700297e-01
9.80569124e-01 -1.10834706e+00 -8.82767856e-01 3.68445188e-01
-1.27154064e+00 -4.11147982e-01 -7.77696893e-02 2.99447060e-01
-1.15852222e-01 -9.20652300e-02 -7.25936949e-01 -5.94879031e-01
-6.99563205e-01 -1.29059947e+00 9.64340806e-01 4.39378291e-01
-3.23237062e-01 -7.37005174e-01 -4.46413577e-01 4.42264080e-01
2.25675046e-01 -3.29666361e-02 9.37216818e-01 -8.48745346e-01
-4.56959367e-01 -1.92571461e-01 -5.67625999e-01 1.29001558e-01
1.99538171e-01 -2.90914979e-02 -1.09213126e+00 1.81986228e-01
-3.53719383e-01 -1.88329205e-01 7.95839190e-01 6.37743548e-02
1.29948640e+00 -3.87994647e-02 -7.64786065e-01 6.21196091e-01
1.05926895e+00 3.29987437e-01 8.03593695e-01 -1.65223867e-01
8.99283290e-01 4.04271543e-01 4.61127460e-01 1.74300477e-01
7.38058031e-01 7.02901900e-01 8.18629712e-02 1.07718028e-01
-2.33745828e-01 -4.62985039e-01 1.03131443e-01 7.54367650e-01
-1.38183251e-01 -3.92476648e-01 -1.09528577e+00 6.81770623e-01
-1.98415816e+00 -5.87314069e-01 -5.70492148e-01 2.05183458e+00
1.42126012e+00 3.49009365e-01 -2.24805921e-01 1.21687800e-01
7.30602920e-01 -1.94975570e-01 -2.66688049e-01 -4.18782145e-01
3.27370968e-03 6.65435553e-01 6.49452031e-01 3.28738362e-01
-1.19836199e+00 1.26934230e+00 5.63384867e+00 7.44634151e-01
-8.01494479e-01 4.54749763e-02 7.55298436e-01 7.38548115e-02
-3.55234832e-01 1.95773765e-01 -1.30243587e+00 2.30666474e-01
1.18345463e+00 -2.42101818e-01 8.93153474e-02 3.76899332e-01
-4.00418669e-01 4.06671911e-02 -1.38993859e+00 7.96256840e-01
-3.75147283e-01 -1.38771868e+00 -7.10125715e-02 -8.50989446e-02
4.97802645e-01 -1.64233714e-01 -2.95784056e-01 7.20323265e-01
5.78723371e-01 -1.40376651e+00 4.84830201e-01 4.60352004e-01
9.59135115e-01 -7.44128168e-01 9.77397025e-01 6.00763932e-02
-1.22718287e+00 1.13781400e-01 -1.18112013e-01 -2.00175464e-01
2.56451100e-01 8.66695046e-01 -7.46256113e-01 1.01383078e+00
5.60660481e-01 5.73467970e-01 -5.63699305e-01 5.81742823e-01
-5.45499980e-01 7.47470319e-01 -2.79318899e-01 -1.87025532e-01
-6.53530061e-02 2.66317546e-01 1.10150672e-01 1.16585493e+00
-5.23847155e-02 2.91938037e-01 -1.01199865e-01 8.88592899e-01
-4.79150206e-01 -6.99569359e-02 -4.23214585e-01 -2.79410630e-01
7.58383334e-01 1.30223691e+00 -2.64223546e-01 -3.44539851e-01
-3.33587259e-01 7.53514171e-01 8.02449346e-01 3.50401402e-02
-1.09045446e+00 -8.73738229e-01 4.74248976e-01 -1.40931517e-01
2.82634884e-01 -1.35077775e-01 -7.25233257e-01 -1.09564734e+00
1.12677440e-01 -5.17758369e-01 2.86039531e-01 -5.93541861e-01
-1.18516004e+00 7.26484895e-01 -1.42955109e-01 -6.46170378e-01
-6.30624667e-02 -4.59350199e-01 -3.01834434e-01 1.10755408e+00
-1.18398356e+00 -1.35666716e+00 -1.32093117e-01 1.03844509e-01
-2.40786355e-02 -5.63487671e-02 1.07203817e+00 7.58002579e-01
-9.94985342e-01 1.18972683e+00 -5.26654124e-01 7.06807554e-01
6.63658798e-01 -1.14289558e+00 8.78541887e-01 6.71637654e-01
4.09962721e-02 8.82881045e-01 2.60713011e-01 -7.61082292e-01
-1.51182163e+00 -1.21052539e+00 1.80394411e+00 -4.66456443e-01
5.89519501e-01 -8.01274419e-01 -8.37502182e-01 1.10745096e+00
1.50309309e-01 1.54328570e-01 7.49171555e-01 8.27541351e-01
-6.62263870e-01 -2.15824217e-01 -1.11968803e+00 5.87056339e-01
1.55284131e+00 -4.71620828e-01 -3.35460961e-01 5.37196621e-02
1.05135083e+00 -6.54663265e-01 -1.20901346e+00 6.88873291e-01
5.29978812e-01 -5.59544206e-01 7.31473684e-01 -1.12273955e+00
6.80223167e-01 -1.93234339e-01 -2.40364343e-01 -1.07904816e+00
-5.50817966e-01 -7.24887669e-01 -2.98015177e-01 1.55010867e+00
1.12020898e+00 -7.20031500e-01 5.66404939e-01 6.89143777e-01
-2.56713003e-01 -9.51350272e-01 -6.91495478e-01 -4.89034891e-01
1.24091208e-01 -4.85103548e-01 9.33271229e-01 1.09351277e+00
2.56361514e-01 1.07422173e+00 -2.79962182e-01 5.51283509e-02
3.98289599e-02 1.51493967e-01 7.26808429e-01 -1.16733396e+00
-1.79787457e-01 -2.38851085e-01 -1.13477580e-01 -8.26030850e-01
2.68656373e-01 -1.15895951e+00 -2.35684752e-01 -1.43580067e+00
4.44591373e-01 -8.62167418e-01 -3.89353305e-01 1.11517537e+00
-5.13566375e-01 1.12626543e-02 -6.30147755e-02 -2.79110283e-01
-2.66522765e-01 4.40504789e-01 1.21892738e+00 1.63419127e-01
-1.99548990e-01 -1.82460964e-01 -1.28497171e+00 1.15879633e-01
8.58931780e-01 -5.07032156e-01 -3.43599647e-01 -8.68373036e-01
7.47976542e-01 2.51037739e-02 5.83422817e-02 -6.20143652e-01
1.28034100e-01 -2.79528946e-02 2.88913906e-01 -3.54549110e-01
2.08561227e-01 -2.71741837e-01 1.45923689e-01 1.73534021e-01
-5.28706968e-01 -5.26937991e-02 3.89163345e-01 3.44658524e-01
-1.50608957e-01 7.82251954e-02 2.51691371e-01 -8.95388890e-03
-4.89946574e-01 1.37468934e-01 9.80815887e-02 2.21982151e-01
9.05325413e-01 3.06758255e-01 -5.17284870e-01 -1.07893184e-01
-7.81306446e-01 4.86298561e-01 -2.82490760e-01 5.57964742e-01
2.73611307e-01 -1.30590308e+00 -9.43627000e-01 2.46454775e-01
5.23522086e-02 3.84950966e-01 5.43108992e-02 8.53978217e-01
-1.37918547e-01 7.49488473e-01 -4.97455150e-02 -9.39744189e-02
-9.39233601e-01 3.05006951e-01 2.23679215e-01 -7.33875513e-01
-2.54553318e-01 1.47477949e+00 1.87675238e-01 -6.91230297e-01
-1.97604299e-01 -5.45364201e-01 -2.12324575e-01 3.67743894e-02
1.76963568e-01 1.08070977e-01 4.50809240e-01 -2.88146853e-01
-6.88053668e-01 8.59610587e-02 -4.70981240e-01 1.61320612e-01
1.32382107e+00 2.64086217e-01 -2.41837278e-01 2.33300298e-01
1.03064263e+00 3.60500324e-03 -5.38187742e-01 -2.63895988e-01
2.78684109e-01 -1.09514005e-01 8.41647685e-02 -1.18197834e+00
-1.32403874e+00 8.28043580e-01 2.83562765e-02 1.45421952e-01
9.84680414e-01 2.77011395e-01 8.83343279e-01 2.35920131e-01
3.32856506e-01 -6.93222880e-01 -6.06458724e-01 7.07335651e-01
7.38673151e-01 -9.66524720e-01 8.59391466e-02 -1.10196221e+00
-5.61233699e-01 7.05711246e-01 9.78016555e-01 2.71976832e-02
6.24001265e-01 7.82125235e-01 -3.15695316e-01 -2.94627905e-01
-1.16411090e+00 -1.97599962e-01 3.00459862e-01 4.10927474e-01
1.17386377e+00 3.25036287e-01 -7.09700227e-01 1.41081309e+00
-3.14142883e-01 1.18917033e-01 2.45193988e-01 9.30157363e-01
2.77720064e-01 -1.27836621e+00 3.05986315e-01 7.40385473e-01
-7.48846114e-01 -4.78570461e-01 -3.62252921e-01 7.33735979e-01
4.30840105e-02 1.04004157e+00 1.36483923e-01 -8.20333898e-01
5.01105428e-01 2.67039388e-01 6.11342192e-01 -8.36771846e-01
-1.03332663e+00 -2.82178938e-01 9.41585004e-01 -3.71877760e-01
-1.25508443e-01 -3.41400892e-01 -1.58747053e+00 -2.46402755e-01
-7.23909080e-01 1.11131869e-01 3.66590381e-01 8.38312268e-01
8.17711353e-01 9.48790431e-01 2.39091575e-01 -2.80539636e-02
-4.12744433e-01 -1.19108415e+00 -3.04777235e-01 3.35307829e-02
-2.26152256e-01 -7.75967181e-01 1.26884460e-01 -1.89152390e-01]
|
[9.277892112731934, 8.716504096984863]
|
ba00e79b-8822-4b8f-8dd0-45d74256a765
|
simplifying-models-with-unlabeled-output-data-1
| null | null |
https://openreview.net/forum?id=GXJPLbB5P-y
|
https://openreview.net/pdf?id=GXJPLbB5P-y
|
Simplifying Models with Unlabeled Output Data
|
We focus on prediction problems with high-dimensional outputs that are subject to output validity constraints, e.g. a pseudocode-to-code translation task where the code must compile. For these problems, labeled input-output pairs are expensive to obtain, but "unlabeled" outputs, i.e. outputs without corresponding inputs, are freely available and provide information about output validity (e.g. code on GitHub). In this paper, we present predict-and-denoise, a framework that can leverage unlabeled outputs. Specifically, we first train a denoiser to map possibly invalid outputs to valid outputs using synthetic perturbations of the unlabeled outputs. Second, we train a predictor composed with this fixed denoiser. We show theoretically that for a family of functions with a high-dimensional discrete valid output space, composing with a denoiser reduces the complexity of a 2-layer ReLU network needed to represent the function and that this complexity gap can be arbitrarily large. We evaluate the framework empirically on several datasets, including image generation from attributes and pseudocode-to-code translation. On the SPoC pseudocode-to-code dataset, our framework improves the proportion of code outputs that pass all test cases by 3-5% over a baseline Transformer.
|
['Percy Liang', 'Tengyu Ma', 'Sang Michael Xie']
|
2020-09-28
| null | null | null | null |
['code-translation']
|
['computer-code']
|
[ 5.52007139e-01 2.83701777e-01 -1.36532605e-01 -5.88680625e-01
-1.34587145e+00 -1.09953022e+00 3.80159229e-01 -3.13042521e-01
3.09889950e-02 6.70732260e-01 1.48399904e-01 -6.56115592e-01
4.44647312e-01 -5.73311210e-01 -1.52440453e+00 -5.29996872e-01
2.40317345e-01 4.30772781e-01 -2.84182787e-01 1.44503102e-01
2.61378605e-02 -1.19579220e-02 -1.57116151e+00 7.34387517e-01
1.02704191e+00 7.73915648e-01 8.82276818e-02 1.10303950e+00
-2.23724525e-02 8.30017686e-01 -8.02141547e-01 -6.82397008e-01
5.03035426e-01 -8.43926907e-01 -8.28805923e-01 -2.84943543e-03
3.18120778e-01 -4.84238595e-01 -6.94816709e-02 1.27945912e+00
1.56137303e-01 -3.19389462e-01 8.70429039e-01 -1.47994554e+00
-1.25478995e+00 8.76494527e-01 1.52758379e-02 -3.83962035e-01
3.66312563e-01 5.28851271e-01 9.90810812e-01 -1.08791971e+00
8.25479388e-01 1.06080556e+00 5.17492354e-01 9.78424549e-01
-1.81268096e+00 -6.37324452e-01 -3.90702486e-01 -2.19303861e-01
-1.11386013e+00 -7.03806639e-01 2.07119763e-01 -7.25331426e-01
1.14625120e+00 3.66588295e-01 1.28459692e-01 1.53173530e+00
2.67214654e-03 5.67075670e-01 1.06430650e+00 -1.92947447e-01
1.56881481e-01 2.85302967e-01 -2.57390141e-01 8.30450773e-01
-3.20474543e-02 7.98394531e-02 -1.54702246e-01 -1.59730449e-01
4.94158983e-01 -5.78304641e-02 -2.71645784e-01 -4.49860811e-01
-1.20693707e+00 7.40952492e-01 3.71896923e-01 -1.40820608e-01
1.85902029e-01 4.95898485e-01 3.57245564e-01 9.05424118e-01
2.03724280e-01 7.48478651e-01 -5.21324158e-01 -3.62169772e-01
-8.18882704e-01 2.23818421e-01 9.49680507e-01 1.40066195e+00
9.73624110e-01 1.52005315e-01 -1.47302777e-01 6.26873791e-01
-6.06458113e-02 6.89959049e-01 5.45815349e-01 -1.25019646e+00
8.73059928e-01 4.51056153e-01 8.22164267e-02 -1.42187566e-01
1.75969020e-01 -1.07117139e-01 -7.56531298e-01 2.88247973e-01
6.85415089e-01 -3.44281137e-01 -1.17412913e+00 1.85963869e+00
-1.09374493e-01 -3.14422175e-02 3.92300129e-01 8.91127944e-01
4.04027343e-01 7.52761304e-01 -4.69842792e-01 -3.22200358e-02
9.31807518e-01 -1.06525517e+00 -3.28044295e-01 -1.66134596e-01
8.58650863e-01 -8.92493665e-01 1.40562606e+00 2.04237729e-01
-1.14397073e+00 -4.32202965e-01 -8.97708774e-01 -4.06906605e-01
-2.85369724e-01 1.68872446e-01 1.79171473e-01 5.18637300e-01
-1.32190752e+00 6.68196201e-01 -7.97735929e-01 -9.35675018e-03
2.47182846e-01 5.16099095e-01 -4.72125471e-01 -1.34634644e-01
-7.97343850e-01 7.59057164e-01 2.80643404e-01 -2.61845529e-01
-1.46408868e+00 -4.73752022e-01 -1.02266479e+00 3.61759394e-01
1.77404702e-01 -5.84233999e-01 1.51187515e+00 -1.45640230e+00
-1.38728476e+00 7.66494870e-01 -2.64172286e-01 -4.87778604e-01
7.27277100e-01 3.45495567e-02 -2.59145707e-01 -1.19336471e-01
2.21959949e-01 7.46887684e-01 1.09298193e+00 -1.20094538e+00
-2.43612036e-01 -4.10796180e-02 1.04476064e-01 -8.83698016e-02
-2.45125473e-01 3.98996249e-02 -5.04261971e-01 -6.59924269e-01
-1.35462537e-01 -1.23077703e+00 1.48547813e-04 5.36864512e-02
-5.77608705e-01 6.39043748e-02 5.15072703e-01 -7.32154965e-01
9.71818089e-01 -2.25122237e+00 4.02229458e-01 1.76804245e-01
7.12087527e-02 -5.68435639e-02 -3.38826656e-01 1.94011495e-01
-1.76191717e-01 4.06659245e-01 -6.63564444e-01 -4.20037091e-01
1.19269922e-01 3.62176538e-01 -5.58928609e-01 4.39760029e-01
7.38692045e-01 1.06437659e+00 -6.90694273e-01 -2.43099675e-01
-4.42938209e-01 2.47500375e-01 -9.55070436e-01 5.61734557e-01
-5.14125526e-01 4.25520927e-01 -1.39914453e-01 7.61377275e-01
3.18359315e-01 -6.21219099e-01 1.46437407e-01 4.81807172e-01
1.16052717e-01 3.86602938e-01 -8.13047051e-01 1.77667117e+00
-7.49753654e-01 8.32133651e-01 1.49755832e-03 -8.33647430e-01
9.38262582e-01 3.74555945e-01 -7.85308182e-02 -2.59976745e-01
-1.29262313e-01 6.15177512e-01 -4.16362248e-02 -3.83936942e-01
3.33625972e-01 1.86370552e-01 -4.93369788e-01 6.64173961e-01
2.94205010e-01 -3.00520182e-01 6.22280389e-02 1.67947441e-01
1.38075376e+00 4.54361439e-01 -8.99707228e-02 -1.11746028e-01
3.19589287e-01 1.03222519e-01 5.29298604e-01 7.27393687e-01
5.05442098e-02 1.00545168e+00 1.10353792e+00 -2.60163486e-01
-1.78708243e+00 -1.01074290e+00 -7.57557750e-02 9.88954127e-01
-1.23489201e-01 -2.86282361e-01 -1.18265879e+00 -9.03137207e-01
1.57655537e-01 7.12659538e-01 -6.45453155e-01 -1.27193570e-01
-6.69874310e-01 -2.97754765e-01 8.24235559e-01 5.28143108e-01
1.71135366e-03 -9.53958690e-01 -2.98778117e-01 1.19440570e-01
-2.37569720e-01 -8.83234203e-01 -6.33931518e-01 6.40839338e-01
-7.23347723e-01 -1.02726519e+00 -7.35567927e-01 -9.91477609e-01
1.14297259e+00 -3.16219360e-01 1.27740240e+00 2.49717101e-01
1.11449920e-01 -6.39188439e-02 -9.96667892e-02 1.42218500e-01
-1.25639927e+00 2.55010575e-01 -6.00992404e-02 -2.50514507e-01
1.06718436e-01 -4.88367409e-01 -2.67596811e-01 2.49874085e-01
-9.89390850e-01 4.74086285e-01 5.97357452e-01 1.17741346e+00
4.69687909e-01 -5.17126322e-01 4.17364091e-01 -1.02439117e+00
5.24727881e-01 -7.78523803e-01 -9.35685039e-01 3.22591037e-01
-7.37635434e-01 7.29069889e-01 1.38711226e+00 -5.59599996e-01
-8.07636976e-01 5.02728701e-01 -2.77936421e-02 -6.90495849e-01
4.63713743e-02 1.00831032e-01 -1.50637001e-01 2.55264103e-01
1.09799671e+00 3.82861197e-01 8.85588378e-02 -4.60583687e-01
5.24379849e-01 9.47117269e-01 8.82062078e-01 -8.37297440e-01
1.06750369e+00 -4.81745265e-02 -2.65645117e-01 1.48169875e-01
-4.66887772e-01 6.95669129e-02 -2.78611302e-01 1.87500834e-01
5.49583197e-01 -1.00874925e+00 -6.09190822e-01 1.25101343e-01
-1.33747315e+00 -7.08938956e-01 -2.07684845e-01 1.80000827e-01
-9.53189850e-01 -1.53448647e-02 -8.44351232e-01 -5.14407396e-01
-3.38504076e-01 -1.80231142e+00 1.15098715e+00 -1.51648432e-01
-5.99212050e-02 -4.89331484e-01 -2.08275542e-01 2.18158606e-02
4.09958363e-01 2.29327410e-01 1.13144422e+00 -6.74899757e-01
-8.83251548e-01 7.78132603e-02 -2.01262236e-01 6.93845987e-01
-1.64272755e-01 -7.61297066e-03 -1.03072405e+00 -1.77405894e-01
-4.21254709e-02 -8.41895461e-01 8.14260364e-01 -1.86026260e-01
1.41707921e+00 -8.20491076e-01 -1.01643279e-02 1.00254798e+00
1.43491161e+00 -1.17525086e-01 4.97070074e-01 -1.19760305e-01
5.74425340e-01 3.93187404e-01 1.92363530e-01 2.94063330e-01
1.70470834e-01 3.85714680e-01 5.61628044e-01 1.70297325e-01
-1.65058076e-01 -6.40683651e-01 7.90437460e-01 8.65893960e-01
2.92454004e-01 -2.15896815e-01 -9.25259233e-01 6.45215631e-01
-1.68095422e+00 -5.88017106e-01 -5.47808446e-02 2.12948084e+00
1.40342903e+00 1.22095160e-01 -2.12620318e-01 -1.14140749e-01
8.56776595e-01 -3.61626923e-01 -9.11233902e-01 -6.24138832e-01
-1.18261501e-01 3.51714343e-01 6.35201931e-01 5.41338682e-01
-9.21158314e-01 8.61412764e-01 6.29835272e+00 5.43535709e-01
-8.84751320e-01 2.60401309e-01 9.28757668e-01 -7.26031587e-02
-6.61562204e-01 1.84746280e-01 -5.70290327e-01 9.12386298e-01
1.46950495e+00 -1.72233820e-01 1.20936346e+00 1.20565605e+00
-2.60561466e-01 3.59878689e-01 -1.64345288e+00 8.64125788e-01
-1.07169054e-01 -1.15423930e+00 -2.83375442e-01 4.13040482e-02
1.20866680e+00 1.81951746e-01 1.89718649e-01 5.81170440e-01
8.61550093e-01 -1.21464264e+00 9.23010111e-01 3.52668822e-01
1.61359334e+00 -5.66472054e-01 4.62344676e-01 3.36900830e-01
-6.06759310e-01 -1.77495420e-01 -4.49259520e-01 3.97482514e-02
-3.51798922e-01 3.86793852e-01 -1.01758003e+00 -2.94549968e-02
4.62811500e-01 5.29717088e-01 -6.25473201e-01 6.11310005e-01
-5.74486375e-01 6.10195994e-01 -1.33578986e-01 2.86068898e-02
8.17380175e-02 6.39236420e-02 7.03249201e-02 1.16712105e+00
8.51033270e-01 -3.20416152e-01 -1.90401107e-01 1.60392630e+00
-7.53827751e-01 -2.36062080e-01 -9.69005823e-01 -1.50031567e-01
3.97754610e-01 9.49337721e-01 -2.66952991e-01 -7.14037299e-01
-3.86736989e-01 1.43695569e+00 5.10590434e-01 5.61010122e-01
-1.07663822e+00 -4.65077847e-01 8.53960872e-01 -1.79990456e-01
3.35954666e-01 3.15215290e-02 -4.26288962e-01 -1.53389347e+00
3.02098751e-01 -1.19480240e+00 5.38402051e-03 -8.13276827e-01
-9.32439029e-01 6.47253633e-01 -4.82764006e-01 -1.34588611e+00
-9.02142048e-01 -5.06574154e-01 -1.63891897e-01 1.15783858e+00
-1.32478106e+00 -7.38155603e-01 -1.42409340e-01 4.53887850e-01
5.99412560e-01 -2.18195587e-01 1.12831604e+00 1.69960886e-01
-3.94811273e-01 8.67048562e-01 5.50905347e-01 4.17171180e-01
6.50874376e-01 -1.48314703e+00 1.03816020e+00 1.04028738e+00
1.23847529e-01 6.73612118e-01 5.45798540e-01 -5.37011325e-01
-1.84294009e+00 -1.40357375e+00 8.31396937e-01 -7.59849191e-01
5.65079451e-01 -8.77241611e-01 -7.91917682e-01 1.02908039e+00
4.69446257e-02 5.35307348e-01 3.83856833e-01 -3.51294637e-01
-8.14089417e-01 2.06937224e-01 -1.10959697e+00 4.60853130e-01
1.12293279e+00 -9.34675157e-01 -3.08320880e-01 4.13677931e-01
9.99380469e-01 -7.95954943e-01 -8.42160106e-01 2.89333630e-02
5.62104940e-01 -8.47544551e-01 6.89506948e-01 -8.48825216e-01
1.13483238e+00 -3.18285555e-01 -1.97187617e-01 -1.45324910e+00
3.35735343e-02 -7.96813667e-01 -3.57606649e-01 1.03398705e+00
7.85530329e-01 -3.25676888e-01 6.17344856e-01 7.32026815e-01
-2.61885852e-01 -5.28691232e-01 -7.88110077e-01 -8.73685002e-01
2.43796095e-01 -4.17390555e-01 8.39844942e-01 7.76088119e-01
-3.32212895e-02 2.60667205e-01 -4.35142040e-01 3.94263193e-02
2.18590066e-01 2.73332864e-01 9.00824726e-01 -6.66433692e-01
-7.94132173e-01 -2.54405975e-01 -1.96796030e-01 -1.15172684e+00
4.62223351e-01 -1.49882913e+00 1.88871518e-01 -1.11048400e+00
1.88564688e-01 -4.43192393e-01 1.41862547e-02 8.61908555e-01
-2.14062244e-01 3.60738844e-01 1.81515053e-01 5.90809643e-01
-3.49752069e-01 2.52303421e-01 1.00115454e+00 -2.87445784e-01
1.22619174e-01 -1.62025452e-01 -5.67730904e-01 2.84896523e-01
6.71175361e-01 -8.58703971e-01 -2.21339583e-01 -7.82996774e-01
3.41391236e-01 4.47686911e-01 3.63918513e-01 -7.76151299e-01
-2.02038407e-01 -9.87336040e-02 3.80613148e-01 4.72684056e-02
2.54933149e-01 -8.21530461e-01 3.19466501e-01 4.59602594e-01
-7.44283676e-01 1.03511050e-01 -1.80911601e-01 4.17197555e-01
-6.85692281e-02 -4.57278192e-01 7.26376772e-01 -1.13630369e-01
-2.10500553e-01 -1.38989925e-01 -9.15231258e-02 2.11715311e-01
9.68500614e-01 7.19524035e-03 -5.46776533e-01 -4.04995084e-01
-5.23260355e-01 -1.80467051e-02 9.44982350e-01 4.81257617e-01
6.20218575e-01 -1.44525886e+00 -7.12161422e-01 5.77633858e-01
2.01651841e-01 -7.08001703e-02 -4.68029618e-01 4.76733118e-01
-5.20580351e-01 2.01322973e-01 -5.94211780e-02 -5.91273069e-01
-1.00854254e+00 8.05014849e-01 3.11358243e-01 -3.07383426e-02
-4.37261224e-01 7.79509783e-01 1.38405100e-01 -8.25416446e-01
1.23334691e-01 -6.84175014e-01 5.04152715e-01 -4.41967309e-01
2.30688453e-01 3.85577120e-02 -3.80363576e-02 -3.80253792e-01
-5.50113432e-02 1.54538512e-01 2.52605945e-01 -1.51709303e-01
1.22984612e+00 2.29801595e-01 -2.13096887e-01 2.87485808e-01
1.78585148e+00 -2.20935017e-01 -1.60021341e+00 -1.97013959e-01
-2.47285143e-02 -5.77065349e-01 -5.49856901e-01 -8.67568731e-01
-1.05891919e+00 1.03400791e+00 4.26856846e-01 1.04595944e-01
1.12019730e+00 -8.18407238e-02 6.31312728e-01 5.74085712e-01
2.37671420e-01 -7.75911689e-01 -1.99982803e-03 4.87585634e-01
7.32004642e-01 -1.24536896e+00 -6.06592298e-01 -1.77266598e-02
-6.94633126e-01 1.26166606e+00 6.36861861e-01 5.72717264e-02
-8.10207948e-02 5.26175320e-01 -9.41154957e-02 2.98911512e-01
-1.17469680e+00 7.21569918e-03 2.91987322e-03 3.77400935e-01
3.60687405e-01 2.73862362e-01 8.34377930e-02 4.65149045e-01
-4.64351714e-01 9.21906531e-02 8.46501946e-01 7.62593389e-01
-1.80473045e-01 -1.23128200e+00 -5.39333820e-01 7.02843547e-01
-5.35370171e-01 -3.99059743e-01 -3.19004655e-01 2.58425504e-01
4.97661382e-02 8.40782404e-01 1.09320991e-01 -5.81968904e-01
5.72252832e-02 5.28552175e-01 1.27543196e-01 -7.96465099e-01
-7.80685067e-01 -1.09198950e-01 8.07731226e-02 -4.83844787e-01
1.64358526e-01 -3.86335045e-01 -1.21993375e+00 -3.01401258e-01
-1.98413476e-01 -2.41838396e-04 8.99710536e-01 4.17117864e-01
6.33748055e-01 3.21687907e-01 7.80452490e-01 -3.20896626e-01
-1.16031182e+00 -9.14198875e-01 -1.78537220e-01 8.22499514e-01
6.41962528e-01 -6.12465367e-02 -4.42214191e-01 6.01724803e-01]
|
[7.894435882568359, 7.722781181335449]
|
70d81846-2aac-41d5-a5fb-04808dc1cf8e
|
rapid-and-robust-endoscopic-content-area
|
2210.14771
| null |
https://arxiv.org/abs/2210.14771v1
|
https://arxiv.org/pdf/2210.14771v1.pdf
|
Rapid and robust endoscopic content area estimation: A lean GPU-based pipeline and curated benchmark dataset
|
Endoscopic content area refers to the informative area enclosed by the dark, non-informative, border regions present in most endoscopic footage. The estimation of the content area is a common task in endoscopic image processing and computer vision pipelines. Despite the apparent simplicity of the problem, several factors make reliable real-time estimation surprisingly challenging. The lack of rigorous investigation into the topic combined with the lack of a common benchmark dataset for this task has been a long-lasting issue in the field. In this paper, we propose two variants of a lean GPU-based computational pipeline combining edge detection and circle fitting. The two variants differ by relying on handcrafted features, and learned features respectively to extract content area edge point candidates. We also present a first-of-its-kind dataset of manually annotated and pseudo-labelled content areas across a range of surgical indications. To encourage further developments, the curated dataset, and an implementation of both algorithms, has been made public (https://doi.org/10.7303/syn32148000, https://github.com/charliebudd/torch-content-area). We compare our proposed algorithm with a state-of-the-art U-Net-based approach and demonstrate significant improvement in terms of both accuracy (Hausdorff distance: 6.3 px versus 118.1 px) and computational time (Average runtime per frame: 0.13 ms versus 11.2 ms).
|
['Tom Vercauteren', 'Sebastien Ourselin', 'Martin Huber', 'Luis C. Garcia-Peraza-Herrera', 'Charlie Budd']
|
2022-10-26
| null | null | null | null |
['edge-detection']
|
['computer-vision']
|
[ 1.26744658e-01 2.72601813e-01 -1.02511505e-02 -1.09214978e-02
-9.51979160e-01 -6.37084126e-01 3.40250522e-01 6.14458263e-01
-7.72826731e-01 4.73022968e-01 1.67851925e-01 -3.15583467e-01
-1.29318118e-01 -4.37004298e-01 -5.93976557e-01 -6.20516777e-01
-3.15564007e-01 2.30692953e-01 4.39217985e-01 1.40783247e-02
3.55723828e-01 3.23335052e-01 -1.23802066e+00 1.23581007e-01
7.64566660e-01 1.18421626e+00 2.21961364e-01 7.66127288e-01
2.78474540e-01 1.04661085e-01 -1.24958575e-01 -2.16011405e-01
5.86198092e-01 -3.75660211e-01 -5.76995671e-01 -9.79958698e-02
1.09119825e-01 -1.41960964e-01 -8.95320401e-02 9.80013251e-01
7.65807748e-01 -7.30190054e-02 4.57973927e-01 -5.59877932e-01
4.77796718e-02 4.87489849e-02 -6.07215881e-01 3.55479747e-01
3.40440363e-01 1.89324066e-01 6.21322632e-01 -5.51360071e-01
1.14157426e+00 3.20504427e-01 1.02405488e+00 1.63607880e-01
-9.86241758e-01 -1.16035052e-01 -4.31682229e-01 -2.33433411e-01
-1.18232095e+00 -1.32224197e-02 3.62182468e-01 -4.90421236e-01
7.38164485e-01 4.46363181e-01 9.03549910e-01 5.87043881e-01
4.11541879e-01 6.79380536e-01 1.21190894e+00 -7.41241276e-01
-4.91615161e-02 3.35815787e-01 -2.33391806e-01 9.36768413e-01
4.59867954e-01 3.57882947e-01 9.18703601e-02 -1.62387773e-01
1.07796347e+00 7.50597790e-02 -5.71708143e-01 -7.47318268e-01
-1.38425028e+00 7.85352588e-01 4.80465382e-01 3.29746544e-01
-5.31796038e-01 -1.16554618e-01 6.79918945e-01 8.00537020e-02
2.30688244e-01 4.22336012e-01 -3.60697955e-01 -5.51733494e-01
-8.38700891e-01 3.36673409e-02 1.01660383e+00 9.80374336e-01
2.61471719e-01 -6.63980663e-01 2.39327133e-01 6.09167099e-01
-3.84429693e-02 -2.42145821e-01 7.63569236e-01 -5.24056017e-01
-6.99402839e-02 6.47081614e-01 1.44376531e-01 -8.27163994e-01
-7.22737253e-01 -4.39547330e-01 -6.25903189e-01 3.46834302e-01
8.12779307e-01 -2.80652970e-01 -7.23641038e-01 1.05282450e+00
6.01774931e-01 -1.39374897e-01 -2.99144477e-01 1.13040996e+00
8.78273606e-01 9.56058875e-02 -2.94851571e-01 -8.92784521e-02
1.68493295e+00 -1.09914327e+00 -2.82880247e-01 1.22886896e-01
8.87403786e-01 -1.20063674e+00 9.91611958e-01 5.22391260e-01
-1.21196985e+00 -4.84695546e-02 -9.67970788e-01 -1.14641137e-01
-3.35981131e-01 4.99250174e-01 7.29349971e-01 5.36803126e-01
-9.53004837e-01 6.29921317e-01 -1.03397036e+00 -5.68949461e-01
2.99411207e-01 5.10769963e-01 -5.94319701e-01 1.17947146e-01
-4.84540254e-01 9.25797343e-01 2.73939431e-01 -5.73432669e-02
-2.45969445e-01 -7.18024194e-01 -8.41715217e-01 -2.12595671e-01
4.16356891e-01 -7.98915148e-01 1.38501048e+00 -6.94722593e-01
-1.45152986e+00 1.30039227e+00 2.08670139e-01 -5.38119614e-01
1.05376589e+00 -2.03003868e-01 -2.28892360e-02 3.34226578e-01
-1.31134048e-01 5.46021938e-01 2.85749227e-01 -8.96931946e-01
-8.90126765e-01 -3.54060888e-01 9.23511758e-02 2.46687010e-01
-8.82016271e-02 1.10302269e-02 -6.73745871e-01 -7.20612764e-01
7.70904571e-02 -1.26184165e+00 -4.76880252e-01 5.89105546e-01
-3.51247519e-01 2.60454684e-01 1.80763602e-01 -8.01876903e-01
1.21321142e+00 -1.93395793e+00 -3.76322776e-01 3.99370305e-02
6.47434443e-02 2.76260734e-01 4.21032876e-01 3.14282775e-01
-1.60004441e-02 -1.84861809e-01 -2.06695020e-01 -1.78104982e-01
-2.85080403e-01 -2.14087829e-01 2.58131206e-01 1.13420331e+00
-2.47973666e-01 7.49183655e-01 -1.09409177e+00 -7.02858150e-01
5.81442595e-01 6.10444009e-01 -5.36634624e-01 2.73712985e-02
2.11834610e-01 5.50984383e-01 -2.80417025e-01 8.49982023e-01
5.87139606e-01 -4.43640951e-04 1.03136584e-01 -4.69664067e-01
-6.26143754e-01 7.06385002e-02 -1.24033582e+00 2.27228117e+00
-5.57004511e-01 5.56608498e-01 1.12730980e-01 -6.84130490e-01
6.52739644e-01 3.48749489e-01 7.85579681e-01 -4.92999256e-01
5.91616035e-01 5.83689392e-01 1.40961662e-01 -6.09054863e-01
4.95438278e-01 -3.29244100e-02 1.23503124e-02 1.80500433e-01
1.08639792e-01 -9.34454948e-02 3.88898015e-01 -3.61279726e-01
9.07031655e-01 4.67367560e-01 9.68961775e-01 -5.11787534e-01
3.25833648e-01 5.16130269e-01 3.34129214e-01 5.03215134e-01
-6.25658393e-01 8.35663915e-01 4.97421563e-01 -6.56371951e-01
-1.06714749e+00 -6.87207758e-01 -5.64475060e-01 3.16447645e-01
4.51305479e-01 -2.46681362e-01 -8.04283619e-01 -6.42761052e-01
-2.14133993e-01 2.57770270e-01 -6.89520061e-01 2.95300841e-01
-5.92850745e-01 -7.43099868e-01 2.69101292e-01 2.70514369e-01
7.20893964e-02 -8.57327402e-01 -1.36579287e+00 1.61032870e-01
1.30514100e-01 -1.02554643e+00 -4.85555410e-01 9.59638953e-02
-1.00707746e+00 -1.36433923e+00 -9.19794321e-01 -9.86187875e-01
9.98795986e-01 1.14696145e-01 1.04811430e+00 -3.70821878e-02
-1.03268087e+00 3.18699270e-01 -4.18541968e-01 -4.36546624e-01
-2.07721218e-01 -8.40587243e-02 -2.14977652e-01 -5.28034151e-01
2.22021684e-01 -2.19029129e-01 -1.26009011e+00 4.10373718e-01
-8.31543148e-01 1.44468993e-01 8.02151501e-01 1.01913023e+00
8.55709314e-01 -5.00381291e-01 -1.06464289e-01 -7.83805668e-01
4.51664239e-01 -2.25589186e-01 -7.60680974e-01 -1.21883176e-01
-5.42232752e-01 -1.79402694e-01 3.94928455e-01 -4.28250313e-01
-6.96108580e-01 3.09343874e-01 -9.55662429e-02 -1.55319184e-01
-2.38543853e-01 4.03135359e-01 8.31069648e-01 -3.23831409e-01
8.62045884e-01 -3.05391662e-02 4.78107810e-01 -1.93259105e-01
3.62014592e-01 5.88334978e-01 6.46972716e-01 -9.11428183e-02
2.82028645e-01 8.34905207e-01 -1.63365275e-01 -7.09510982e-01
-4.16031450e-01 -1.07676899e+00 -5.13702273e-01 -6.17821999e-02
6.80709243e-01 -7.81653345e-01 -6.34527564e-01 1.23838425e-01
-9.35687840e-01 -1.87595218e-01 -3.88942480e-01 9.20985937e-01
-7.49273479e-01 4.15414304e-01 -7.76054442e-01 -4.80706573e-01
-7.68450320e-01 -1.35280716e+00 9.54406619e-01 3.61425668e-01
-2.68414915e-01 -1.16538727e+00 1.75001517e-01 -8.29867497e-02
3.78499746e-01 8.43109667e-01 2.02078149e-01 -6.10834837e-01
-4.45452571e-01 -8.67594421e-01 -2.76405871e-01 -9.59300250e-02
8.54968131e-02 -9.55105498e-02 -7.74530828e-01 -3.86008173e-01
8.44245255e-02 1.64716288e-01 6.19708776e-01 8.58686924e-01
9.81318772e-01 -3.48559059e-02 -5.50778747e-01 9.00732219e-01
1.71100664e+00 -2.60190144e-02 5.43584168e-01 7.33855486e-01
-3.08532938e-02 3.15375775e-01 9.83499229e-01 6.11045241e-01
1.45963416e-01 8.58841360e-01 6.14575446e-01 -4.68879908e-01
-1.30717278e-01 7.82210752e-02 -2.01005936e-01 6.40636981e-01
-3.84928256e-01 2.59999841e-01 -7.75925279e-01 7.56097376e-01
-1.54230642e+00 -5.99788189e-01 -1.29376605e-01 2.56861567e+00
7.75730371e-01 1.59680352e-01 2.35534415e-01 -1.80294260e-01
6.27962947e-01 -4.27459627e-01 -1.74383029e-01 -3.23148400e-01
3.58062565e-01 -3.00644487e-02 8.57554495e-01 4.02439803e-01
-1.43882644e+00 3.22043777e-01 4.93679094e+00 7.75430381e-01
-1.23697555e+00 -8.92108902e-02 5.96241117e-01 -1.29853442e-01
3.19510102e-01 -2.31340267e-02 -4.93765980e-01 3.52733254e-01
5.04236162e-01 -2.14537814e-01 4.73994724e-02 9.60171282e-01
9.89612564e-02 -5.30516446e-01 -7.18404055e-01 1.04541898e+00
1.11562751e-01 -1.46988606e+00 -7.85428286e-01 3.12156975e-01
4.79522973e-01 3.04150641e-01 -1.72234446e-01 -6.70467541e-02
-2.18429670e-01 -6.65365398e-01 4.36735272e-01 4.39141542e-01
1.14222503e+00 -3.77176553e-01 1.10827613e+00 1.40445426e-01
-1.04505491e+00 1.00705557e-01 -1.84424102e-01 2.42993593e-01
3.47695768e-01 5.63128054e-01 -1.21085739e+00 5.06724775e-01
6.93742871e-01 4.28549200e-01 -2.98192322e-01 1.92373860e+00
8.22325423e-02 8.08245167e-02 -6.28814816e-01 -1.79273173e-01
3.31204146e-01 -4.04333204e-01 7.49207795e-01 1.47460866e+00
5.31569481e-01 -1.59493491e-01 -1.18933819e-01 5.37201107e-01
1.95467383e-01 5.32526612e-01 -4.45539087e-01 2.81462789e-01
1.17122322e-01 1.91820765e+00 -1.26151013e+00 -1.24885358e-01
-5.86983323e-01 9.54854131e-01 -3.84840332e-02 -3.10795993e-01
-9.08096492e-01 -5.70821047e-01 2.23613992e-01 3.50221813e-01
2.46119365e-01 -9.63368267e-02 -3.12510669e-01 -9.39633191e-01
2.85995245e-01 -4.32728320e-01 5.25526822e-01 -3.70556295e-01
-8.00557911e-01 7.64368594e-01 -3.13387960e-01 -1.92648625e+00
-2.70534366e-01 -8.39298606e-01 -6.65511847e-01 6.50804102e-01
-1.53046942e+00 -1.01609564e+00 -7.70405710e-01 1.22591034e-01
4.48606879e-01 2.11846977e-01 1.08769405e+00 2.69060671e-01
-1.71730861e-01 5.21199524e-01 3.30702543e-01 -9.53361217e-04
7.83801496e-01 -1.42148662e+00 1.66026682e-01 4.71669525e-01
-1.00274391e-01 7.30399549e-01 7.80583382e-01 -3.09089959e-01
-1.34559369e+00 -5.71208000e-01 4.72408652e-01 -2.92194426e-01
7.60076702e-01 -7.40647018e-02 -4.26489472e-01 5.53179145e-01
1.86897859e-01 3.54975373e-01 7.22946703e-01 -2.39978865e-01
3.72788072e-01 3.08286220e-01 -1.43467236e+00 7.57459998e-01
6.35504603e-01 1.32011667e-01 -3.88534993e-01 3.86160851e-01
2.76845843e-01 -1.17475379e+00 -1.26445091e+00 4.40548301e-01
7.53185332e-01 -1.31998694e+00 9.50715303e-01 8.62616003e-02
2.75635779e-01 -1.49729297e-01 4.96881723e-01 -1.09005785e+00
1.75768808e-01 -8.09178710e-01 1.58894911e-01 4.54153538e-01
4.28186446e-01 -6.91138923e-01 9.12473440e-01 3.10162723e-01
-6.45492017e-01 -1.27054048e+00 -9.25286055e-01 -3.56200248e-01
-1.44784495e-01 -1.37654364e-01 -1.60568193e-01 6.57354236e-01
5.43496013e-01 -4.14845049e-01 2.00195089e-01 -1.75417349e-01
3.44128549e-01 3.49030256e-01 6.89238667e-01 -9.58855569e-01
-1.36411414e-01 -6.04077339e-01 -7.75696516e-01 -7.89829314e-01
-8.14468086e-01 -8.58887255e-01 -8.86036605e-02 -1.61093545e+00
-2.85206791e-02 -4.70709532e-01 5.94072863e-02 1.86723277e-01
5.67827225e-02 2.28006393e-01 -1.20540440e-01 2.89673984e-01
-2.78846771e-01 -1.35856271e-01 1.31131387e+00 4.68012691e-01
-4.04757500e-01 2.15695113e-01 -3.64499927e-01 8.52714717e-01
8.95133138e-01 -2.83649534e-01 1.37432786e-02 3.09157316e-02
-6.34078607e-02 5.01415394e-02 3.78099799e-01 -1.00597048e+00
4.01395947e-01 4.40907598e-01 1.85516581e-01 -4.66929734e-01
3.08227748e-01 -7.75961399e-01 5.37827685e-02 9.08267736e-01
-4.73285764e-02 1.49792045e-01 3.62890333e-01 3.23733866e-01
-2.69823432e-01 -4.52172339e-01 8.93785834e-01 -5.04277289e-01
-6.39735937e-01 2.44672164e-01 9.62786451e-02 -5.66396452e-02
1.40795374e+00 -7.09831953e-01 -9.60297808e-02 9.92328022e-03
-8.22581172e-01 -1.93551436e-01 7.64001489e-01 5.70431091e-02
4.81358141e-01 -7.66335070e-01 -6.18624032e-01 1.23417586e-01
3.17885637e-01 1.96706295e-01 4.03367370e-01 1.78709817e+00
-1.36060715e+00 4.71776217e-01 -1.49484560e-01 -6.29765451e-01
-1.19659853e+00 5.71753442e-01 5.12478828e-01 -3.74822468e-01
-1.09519923e+00 5.75287282e-01 -3.01942346e-03 -2.34639302e-01
-1.97360362e-03 -6.32477105e-01 -5.67067973e-02 -7.74682462e-02
4.01813477e-01 3.60568941e-01 3.93367589e-01 -4.33542222e-01
-1.20715231e-01 5.55784225e-01 -2.49866322e-02 2.55838841e-01
1.29242742e+00 -1.28272280e-01 -1.39418254e-02 3.31407622e-03
1.20348358e+00 2.24923253e-01 -1.01649857e+00 3.99231873e-02
6.36735633e-02 -4.58141983e-01 1.95392638e-01 -8.61943781e-01
-8.41456175e-01 5.78472733e-01 1.03124058e+00 1.69763014e-01
1.06958950e+00 -2.24919587e-01 8.52766573e-01 -4.33787286e-01
3.33784878e-01 -9.72650528e-01 -4.13046867e-01 4.59270738e-02
8.03675115e-01 -1.45716333e+00 1.88680530e-01 -7.88649976e-01
-4.78693336e-01 1.24614882e+00 2.90922850e-01 -2.73444682e-01
6.12485588e-01 3.76343787e-01 1.95533305e-01 -1.96652755e-01
-1.45658359e-01 -2.60644525e-01 3.19172293e-01 3.34909558e-01
6.56424046e-01 -1.94213353e-02 -9.20589209e-01 3.97553116e-01
-1.36541337e-01 3.30641031e-01 6.15615547e-01 1.15566218e+00
-2.15603918e-01 -7.87915170e-01 -1.89333871e-01 4.44719762e-01
-8.45093608e-01 -2.21492946e-01 3.99145275e-01 1.08179998e+00
-1.31474257e-01 3.43251050e-01 -8.09912309e-02 2.71332383e-01
4.45543319e-01 -5.07671893e-01 4.30971175e-01 -2.36445487e-01
-7.66181111e-01 4.05954242e-01 1.39200792e-01 -6.47448301e-01
-2.49985710e-01 -6.79509580e-01 -1.25085819e+00 1.10700406e-01
-5.18179834e-01 -1.88264810e-02 1.18008947e+00 2.29832098e-01
3.79999071e-01 2.52127588e-01 2.16650233e-01 -1.01417816e+00
-4.71957207e-01 -9.20025826e-01 -4.72521454e-01 5.08984149e-01
2.09676921e-01 -7.40134299e-01 -5.48883498e-01 3.27342972e-02]
|
[14.049025535583496, -3.1598422527313232]
|
80ccd3d4-000a-4e50-a926-b2ecf4709803
|
comma-deer-common-sense-aware-multimodal
| null | null |
https://aclanthology.org/2022.coling-1.608
|
https://aclanthology.org/2022.coling-1.608.pdf
|
COMMA-DEER: COmmon-sense Aware Multimodal Multitask Approach for Detection of Emotion and Emotional Reasoning in Conversations
|
Mental health is a critical component of the United Nations’ Sustainable Development Goals (SDGs), particularly Goal 3, which aims to provide “good health and well-being”. The present mental health treatment gap is exacerbated by stigma, lack of human resources, and lack of research capability for implementation and policy reform. We present and discuss a novel task of detecting emotional reasoning (ER) and accompanying emotions in conversations. In particular, we create a first-of-its-kind multimodal mental health conversational corpus that is manually annotated at the utterance level with emotional reasoning and related emotion. We develop a multimodal multitask framework with a novel multimodal feature fusion technique and a contextuality learning module to handle the two tasks. Leveraging multimodal sources of information, commonsense reasoning, and through a multitask framework, our proposed model produces strong results. We achieve performance gains of 6% accuracy and 4.62% F1 on the emotion detection task and 3.56% accuracy and 3.31% F1 on the ER detection task, when compared to the existing state-of-the-art model.
|
['Pushpak Bhattacharyya', 'Asif Ekbal', 'Gopendra Vikram Singh', 'Soumitra Ghosh']
| null | null | null | null |
coling-2022-10
|
['common-sense-reasoning']
|
['reasoning']
|
[ 2.55833179e-01 5.87747276e-01 -1.97044104e-01 -5.29559791e-01
-1.29591584e+00 3.46653499e-02 3.42220962e-01 4.28599536e-01
-3.43218416e-01 5.58108807e-01 6.54824615e-01 2.14562014e-01
5.57349548e-02 -4.00153875e-01 6.88832104e-02 -4.18882221e-01
1.51451260e-01 2.51423031e-01 -5.66909432e-01 -5.08131742e-01
-8.90769809e-02 -1.08681686e-01 -1.22593904e+00 7.91893780e-01
1.18941152e+00 9.72094357e-01 -1.44653782e-01 5.66436112e-01
-1.37194738e-01 9.22836661e-01 -4.06146556e-01 -8.13093305e-01
-6.42829180e-01 -4.69874471e-01 -1.09505975e+00 1.01473719e-01
-1.87104866e-01 -3.04730058e-01 2.16737971e-01 1.08986950e+00
8.26079369e-01 -7.54130408e-02 5.10824680e-01 -1.39663303e+00
-6.17636144e-01 4.49350208e-01 -4.74250048e-01 -2.33471349e-01
8.15869391e-01 -9.21606123e-02 7.30612040e-01 -8.02288651e-01
5.97517729e-01 1.43970740e+00 7.10839510e-01 8.92756462e-01
-9.76018250e-01 -7.44730294e-01 -8.25573504e-02 1.06560297e-01
-9.53131855e-01 -8.62628222e-01 6.16015911e-01 -4.78490949e-01
1.22632599e+00 2.20495939e-01 4.28843617e-01 1.36335516e+00
-1.81814581e-02 6.16580665e-01 1.17212641e+00 -3.26899529e-01
1.40445465e-02 3.04210514e-01 5.67418709e-02 7.20020533e-01
-2.93181390e-01 -7.25412905e-01 -3.52689862e-01 -5.40485561e-01
8.97737890e-02 -1.15581264e-03 -1.05966680e-01 5.70309401e-01
-1.04430628e+00 9.45086300e-01 3.08357179e-02 5.12480617e-01
-7.16568589e-01 -2.48606220e-01 7.74831057e-01 2.82148551e-02
6.91190302e-01 7.99337476e-02 -3.05373877e-01 -5.47881186e-01
-5.31949520e-01 4.54896614e-02 6.88519657e-01 3.15896928e-01
3.62157643e-01 -2.39042178e-01 -4.04864401e-01 1.47104037e+00
3.03483278e-01 7.02935755e-01 3.94745946e-01 -1.09970927e+00
4.92797166e-01 8.63750994e-01 -2.17611507e-01 -1.20745850e+00
-8.16185117e-01 6.87355846e-02 -1.01278746e+00 -5.35168767e-01
-6.03809915e-02 -6.68913007e-01 -4.19864327e-01 2.07210732e+00
5.81450224e-01 1.44632738e-02 4.22733575e-01 5.56538105e-01
1.22418547e+00 5.45278966e-01 5.22407174e-01 -5.28804064e-01
1.80124700e+00 -7.32618272e-01 -1.47243702e+00 -3.09206486e-01
7.95974672e-01 -6.48015201e-01 6.99718118e-01 2.68011928e-01
-1.20731831e+00 -4.02272772e-03 -4.84258771e-01 -6.28928002e-03
-2.94393659e-01 8.28544348e-02 7.99675286e-01 8.88764799e-01
-1.01623499e+00 -1.27623454e-01 -4.99904543e-01 -6.39611781e-01
4.48094964e-01 3.89138795e-02 -6.98625028e-01 -1.52970552e-01
-1.31104994e+00 1.13590622e+00 6.69931024e-02 8.36919695e-02
-4.64363605e-01 -3.88219953e-01 -1.02049243e+00 -4.37171794e-02
3.46933097e-01 -5.49697459e-01 1.14529157e+00 -8.89348984e-01
-1.42785859e+00 1.05664253e+00 -3.09604377e-01 1.36407420e-01
6.17464073e-02 -1.93594396e-01 -9.19733703e-01 4.95145172e-01
1.74807742e-01 6.34864450e-01 4.02851313e-01 -9.49390590e-01
-5.36297441e-01 -7.29632854e-01 -1.81371376e-01 3.49598706e-01
-6.67601049e-01 6.95999980e-01 -1.26098484e-01 -1.75277829e-01
-6.68896362e-02 -7.72393167e-01 -2.18310192e-01 -4.90051746e-01
-4.96254385e-01 -3.68415117e-01 5.64977109e-01 -1.05110395e+00
1.24883676e+00 -2.10850096e+00 1.79687932e-01 -6.51804209e-02
2.59418875e-01 1.87017575e-01 -5.26464023e-02 5.55769622e-01
-1.15104757e-01 2.79009998e-01 -1.05601680e-02 -6.66080356e-01
1.94237679e-01 3.93725671e-02 1.52480528e-01 2.90462047e-01
4.90214318e-01 8.75570238e-01 -9.79897261e-01 -4.87015367e-01
-2.62220670e-02 7.83545554e-01 -5.49137652e-01 2.86345035e-01
2.90069163e-01 4.49994683e-01 -4.19735730e-01 9.28178906e-01
5.76274812e-01 -1.87023357e-01 3.96743625e-01 -1.36440873e-01
4.62110974e-02 8.57975259e-02 -7.95598209e-01 1.69200015e+00
-3.39590609e-01 6.88144863e-02 5.11094391e-01 -9.45834100e-01
8.07760477e-01 8.91305804e-01 7.15567529e-01 -7.96041846e-01
4.07594383e-01 3.73698510e-02 -6.50144219e-02 -1.18896234e+00
2.84381777e-01 -4.93686885e-01 -4.66234893e-01 2.48075843e-01
1.54742424e-03 1.76592618e-01 -2.85776734e-01 2.69090325e-01
1.10790801e+00 -3.54517907e-01 5.24926007e-01 1.26993492e-01
6.29491806e-01 -3.49654943e-01 7.90174901e-01 1.74940407e-01
-5.91133833e-01 8.22795928e-02 7.25387514e-01 6.38802722e-03
-5.41214764e-01 -3.20735842e-01 -1.27672553e-01 1.34254301e+00
-4.75511163e-01 -2.17415169e-01 -9.16219413e-01 -3.31614852e-01
-3.52308124e-01 5.89762449e-01 -7.09838212e-01 -2.55494237e-01
1.31264210e-01 -9.17618573e-01 9.64532137e-01 1.42680913e-01
8.46919358e-01 -1.13842773e+00 -5.01743853e-01 2.26610631e-01
-8.89785707e-01 -1.57748246e+00 -3.16620022e-02 -3.99414450e-01
-2.82870770e-01 -1.04300702e+00 -6.09061599e-01 -5.13077617e-01
2.90130943e-01 -1.46821722e-01 7.58426487e-01 -1.16195388e-01
-2.95421958e-01 5.98955035e-01 -4.71922874e-01 -2.15535969e-01
-4.33717251e-01 -3.27148646e-01 1.43392739e-04 9.03968960e-02
4.96376485e-01 -2.76012361e-01 -4.66477334e-01 -1.34253725e-01
-8.42843890e-01 1.77301303e-01 4.61409628e-01 7.39188552e-01
-3.94338183e-02 -1.38599589e-01 1.14876032e+00 -6.76118135e-01
9.20564413e-01 -9.59155262e-01 4.12987024e-01 3.41853261e-01
-2.38776028e-01 -5.61744869e-01 -1.66058622e-03 -3.10670465e-01
-1.41481078e+00 -2.59204268e-01 -5.74565411e-01 2.07551777e-01
-5.45445383e-01 6.37911975e-01 -3.06142151e-01 3.86273295e-01
2.16487423e-01 -1.98070616e-01 2.95498688e-02 -2.01862991e-01
2.71607280e-01 1.26460743e+00 3.91811013e-01 -4.76026058e-01
-4.21381593e-02 3.41381788e-01 -2.94497341e-01 -8.57456446e-01
-9.31540251e-01 -5.06242335e-01 1.65899191e-02 -5.14843643e-01
1.31789064e+00 -1.07629478e+00 -1.19824386e+00 5.02597153e-01
-1.34745479e+00 -4.82225455e-02 6.01535022e-01 4.98463511e-01
-3.92539948e-01 3.06886375e-01 -8.50457132e-01 -1.48766565e+00
-8.06422114e-01 -1.06345546e+00 1.21560466e+00 8.28125104e-02
-4.77666050e-01 -8.54875624e-01 9.94295701e-02 9.84507978e-01
4.73004162e-01 6.92061782e-01 9.66950774e-01 -7.67455697e-01
6.35915458e-01 -7.33830631e-02 -4.13951933e-01 3.04031640e-01
1.63605973e-01 -3.08040291e-01 -1.20288193e+00 1.23683535e-01
2.11529374e-01 -7.71896422e-01 5.27910173e-01 1.51531056e-01
6.10781610e-01 -5.02266824e-01 -1.80710971e-01 3.18638841e-03
1.14684987e+00 3.89279783e-01 6.33160949e-01 -9.14735533e-03
5.73989451e-01 1.11922681e+00 4.58618313e-01 9.23277974e-01
1.19440973e+00 4.80574459e-01 2.96396434e-01 -2.12857261e-01
1.56525016e-01 2.05975860e-01 3.75322819e-01 8.78292739e-01
-4.43131477e-02 2.39626765e-02 -1.28683352e+00 5.86382508e-01
-2.07580686e+00 -1.13090873e+00 -8.47288594e-02 1.77180898e+00
7.98132241e-01 -4.89357203e-01 1.85361847e-01 1.33879334e-01
8.54837239e-01 -2.50027120e-01 -3.38230282e-01 -7.06605375e-01
3.24989930e-02 -1.09191366e-01 -3.03068221e-01 3.84650946e-01
-1.08612716e+00 7.32515574e-01 5.80666590e+00 5.58049858e-01
-9.08773720e-01 4.99482572e-01 9.04229522e-01 1.74136847e-01
-2.41777450e-01 -6.29076600e-01 -3.40661973e-01 2.32244655e-01
1.13669312e+00 1.59850121e-01 3.61797452e-01 5.72785079e-01
4.48804229e-01 -1.04629956e-01 -7.16631532e-01 1.14087987e+00
4.05862331e-01 -7.67184019e-01 -4.63862807e-01 1.31651714e-01
6.06069505e-01 -2.81815439e-01 1.32331744e-01 5.40974855e-01
-1.10550642e-01 -1.14627576e+00 2.38432467e-01 6.39147043e-01
6.59109473e-01 -1.05345953e+00 1.00728035e+00 2.76000559e-01
-9.06521082e-01 -2.51373500e-01 1.72986776e-01 -5.92951924e-02
2.79226422e-01 6.06195092e-01 -6.62326872e-01 6.16055131e-01
6.09489083e-01 5.28265476e-01 -7.71453157e-02 4.03399199e-01
-1.22537313e-03 3.05886686e-01 1.99965358e-01 -5.51721863e-02
1.79162502e-01 -1.00605544e-02 2.12786913e-01 1.63861024e+00
3.95244241e-01 5.81781209e-01 2.91443348e-01 5.00616550e-01
-2.70068616e-01 5.17614245e-01 -6.11541450e-01 -4.51202065e-01
4.68837798e-01 1.61619031e+00 -3.30802947e-01 -4.57534164e-01
-4.91573304e-01 1.02534080e+00 3.96324784e-01 1.72495082e-01
-9.68011498e-01 -2.89153844e-01 6.28323734e-01 -6.36762977e-01
-2.78992057e-01 3.74975771e-01 -1.56934321e-01 -1.02914453e+00
-3.39247435e-01 -1.13953829e+00 4.65075731e-01 -5.77839613e-01
-1.34547007e+00 5.10750473e-01 -2.26264104e-01 -4.87807453e-01
-3.95360291e-01 -2.11986989e-01 -4.27142888e-01 5.93320847e-01
-1.41286957e+00 -1.39101696e+00 -2.73533046e-01 6.39558911e-01
2.44058102e-01 6.04185984e-02 1.37619102e+00 6.02582991e-01
-1.01616859e+00 4.84197378e-01 -3.52359056e-01 9.58726034e-02
8.42149556e-01 -8.88448536e-01 -4.68659371e-01 3.12875748e-01
-1.01528192e+00 4.57155883e-01 5.35865247e-01 -6.55772030e-01
-1.57982492e+00 -9.01887238e-01 1.27723432e+00 -8.13910216e-02
6.35356247e-01 -1.64870769e-01 -8.29408705e-01 5.39063454e-01
5.30536354e-01 -7.01638103e-01 1.40129709e+00 4.91370022e-01
-4.24799889e-01 2.92267680e-01 -1.93444633e+00 5.80996394e-01
8.22998464e-01 -6.58379674e-01 -6.33295715e-01 4.58438307e-01
6.70360804e-01 -1.32075846e-01 -1.32161176e+00 4.69550610e-01
7.10625768e-01 -9.66815889e-01 8.24065506e-01 -6.14210606e-01
7.34325230e-01 4.51096743e-01 -4.48765099e-01 -1.19307637e+00
-7.55029768e-02 -5.61618268e-01 9.45808515e-02 1.54532635e+00
2.96718299e-01 -6.61161959e-01 1.14238515e-01 1.21310627e+00
-1.41316131e-01 -6.31714880e-01 -9.89449263e-01 -9.55621079e-02
-2.46826857e-01 -6.71074688e-01 4.69707280e-01 1.52501523e+00
9.86733556e-01 6.71941400e-01 -5.44667661e-01 1.20421454e-01
3.47543001e-01 -3.02970707e-01 4.83469903e-01 -1.02771938e+00
9.63340402e-02 -6.07426524e-01 -1.40058845e-01 1.03198409e-01
4.82142627e-01 -6.45729840e-01 -3.49155851e-02 -1.78677130e+00
7.05539048e-01 1.57549843e-01 -8.43109488e-02 9.23118532e-01
-3.31156373e-01 3.30746025e-02 1.12365246e-01 -3.77894640e-01
-7.95033932e-01 6.45334423e-01 9.27066684e-01 3.92308421e-02
-2.01635092e-01 -6.18787825e-01 -1.09237564e+00 8.39828968e-01
8.86814117e-01 -2.12316334e-01 -1.74558014e-01 -1.14653736e-01
3.65916640e-01 5.35845220e-01 1.54328585e-01 -6.79278314e-01
2.76493188e-02 -2.63452619e-01 2.60308869e-02 -3.03904444e-01
8.59051466e-01 -5.96036792e-01 1.46851420e-01 4.11894858e-01
-4.53681290e-01 -9.76984948e-02 3.73997271e-01 1.50408745e-01
-7.65567943e-02 -4.78995442e-02 6.58458233e-01 2.20371764e-02
-3.24618280e-01 -1.36576384e-01 -5.20602465e-01 1.53605947e-02
1.00819027e+00 4.08809602e-01 -7.15164244e-01 -6.09065592e-01
-8.45269442e-01 4.26383734e-01 -1.60458371e-01 4.17132527e-01
8.30853164e-01 -1.34154403e+00 -9.32004392e-01 -2.19061792e-01
2.79487252e-01 -6.20783567e-01 8.44356775e-01 1.35093725e+00
3.51382673e-01 2.73598462e-01 -2.09913731e-01 -1.70386478e-01
-1.46664834e+00 3.33822787e-01 1.67029381e-01 -3.05403233e-01
-1.79721400e-01 5.15658259e-01 -2.68955827e-01 -7.05612481e-01
1.69257239e-01 -7.78299496e-02 -5.82966745e-01 5.18349588e-01
9.46480811e-01 5.86033344e-01 -1.81478798e-01 -1.03037596e+00
-5.78183889e-01 1.40449136e-01 2.44131565e-01 -3.89438838e-01
1.43132138e+00 -3.04155707e-01 -5.83715081e-01 4.41080689e-01
1.11032617e+00 -1.38853297e-01 -2.10047022e-01 2.63510421e-02
7.91895539e-02 -9.21550989e-02 1.63806021e-01 -1.18556499e+00
-5.67288876e-01 8.15045834e-01 6.48214161e-01 3.54828298e-01
1.19124734e+00 7.92247504e-02 8.01869571e-01 2.94936389e-01
-5.17482404e-03 -1.19591987e+00 1.63138926e-01 5.39405048e-01
8.94142032e-01 -1.48369277e+00 -4.67570186e-01 -3.97211045e-01
-1.14994466e+00 8.72858405e-01 4.39474344e-01 5.27105451e-01
3.59719694e-01 5.59119247e-02 2.60529786e-01 -4.48490024e-01
-9.34610963e-01 -4.44844157e-01 1.79687455e-01 5.29551744e-01
8.07552993e-01 2.31254235e-01 -4.70456451e-01 9.89587843e-01
2.45344996e-01 6.55260608e-02 1.10016927e-01 6.92665696e-01
-5.76734066e-01 -8.20965707e-01 -3.79476458e-01 1.66721448e-01
-7.98426151e-01 -1.18492641e-01 -5.46434343e-01 4.35413510e-01
2.38663778e-01 1.61424232e+00 -1.74877256e-01 -4.45672065e-01
3.52595627e-01 5.79257071e-01 6.48841485e-02 -4.95589942e-01
-7.92967439e-01 2.61261046e-01 7.41345584e-01 -6.96876168e-01
-8.83219600e-01 -5.78910708e-01 -1.47999334e+00 -2.08283037e-01
4.55167592e-02 2.30803085e-03 9.37689126e-01 1.24170220e+00
6.90169334e-01 5.83293617e-01 5.71471512e-01 -6.05216444e-01
-7.50220940e-02 -1.22930825e+00 -3.01615030e-01 3.92237246e-01
2.56450623e-01 -3.90207887e-01 -7.51700029e-02 -2.58359760e-01]
|
[13.125397682189941, 5.66584587097168]
|
94858475-4bf6-4a43-b973-3451289b3e09
|
meta-variational-monte-carlo
|
2011.10614
| null |
https://arxiv.org/abs/2011.10614v1
|
https://arxiv.org/pdf/2011.10614v1.pdf
|
Meta Variational Monte Carlo
|
An identification is found between meta-learning and the problem of determining the ground state of a randomly generated Hamiltonian drawn from a known ensemble. A model-agnostic meta-learning approach is proposed to solve the associated learning problem and a preliminary experimental study of random Max-Cut problems indicates that the resulting Meta Variational Monte Carlo accelerates training and improves convergence.
|
['Shravan Veerapaneni', 'Brian Chen', 'Oliver Knitter', 'James Stokes', 'Tianchen Zhao']
|
2020-11-20
| null | null | null | null |
['variational-monte-carlo']
|
['miscellaneous']
|
[ 1.54915035e-01 9.84966084e-02 -5.32758355e-01 -1.56045347e-01
-1.37957752e+00 -2.83823639e-01 5.38392305e-01 -1.73628598e-01
-5.84004104e-01 1.43088853e+00 -3.91167045e-01 -2.25297585e-01
-4.72246587e-01 -9.03095543e-01 -7.14937747e-01 -1.08807898e+00
2.42228545e-02 9.09693182e-01 -4.40899372e-01 -2.42565736e-01
6.73424065e-01 3.14627200e-01 -1.70095694e+00 1.71896517e-01
1.16254652e+00 4.54046607e-01 1.40504435e-01 8.23828995e-01
1.29849836e-01 5.50883412e-01 -6.48538291e-01 -1.91225469e-01
2.18912348e-01 -1.07023501e+00 -1.26613319e+00 -7.42325485e-02
4.52618539e-01 5.12871683e-01 -4.98023629e-01 1.35023510e+00
3.69259536e-01 8.41568351e-01 1.03937173e+00 -1.17134035e+00
-8.13626200e-02 7.39593506e-01 -3.95531535e-01 5.27349710e-01
3.10707483e-02 3.42423052e-01 1.17453194e+00 -4.70672280e-01
6.38307571e-01 9.01630700e-01 7.43316472e-01 8.30970764e-01
-1.53801370e+00 -4.89616662e-01 -3.50910932e-01 4.03276145e-01
-1.67911065e+00 -1.36152670e-01 7.33594179e-01 -6.99560270e-02
1.37642646e+00 2.52337694e-01 9.76352930e-01 1.21870923e+00
7.66793370e-01 4.68003601e-01 1.43504381e+00 -8.41663241e-01
8.16703498e-01 1.82724029e-01 1.49391443e-01 1.03186309e+00
3.22394788e-01 7.39905477e-01 -4.86559123e-01 -6.54552042e-01
2.64327586e-01 -4.95621473e-01 -4.30040434e-02 -4.02376115e-01
-7.43312538e-01 1.19951904e+00 2.22942427e-01 1.79532707e-01
-4.02695566e-01 2.54977643e-01 6.12115324e-01 2.86506861e-01
3.07418287e-01 1.08437681e+00 -4.14084733e-01 -8.25912952e-02
-1.42880678e+00 5.99873841e-01 1.00127709e+00 3.76479685e-01
9.68956411e-01 4.08808321e-01 4.24089096e-02 3.96009564e-01
1.79840013e-01 5.48158407e-01 3.50293100e-01 -1.52268505e+00
2.93807447e-01 2.39464924e-01 9.88995209e-02 -1.08650967e-01
-3.46624583e-01 -5.26683271e-01 -8.43834460e-01 5.22423685e-01
1.06384501e-01 -6.13113165e-01 -8.77662361e-01 1.65891397e+00
3.83050382e-01 2.81755388e-01 1.92315914e-02 4.88824040e-01
3.70845646e-01 1.17713642e+00 -5.96407726e-02 -1.02455986e+00
5.06078720e-01 -1.08201861e+00 -5.19600213e-01 -1.19774528e-01
6.07460976e-01 -1.54337183e-01 3.52605611e-01 6.57095790e-01
-1.41765451e+00 -6.44807696e-01 -1.26428497e+00 3.94571722e-01
-3.85234416e-01 -4.37192559e-01 7.55527735e-01 9.26594853e-01
-8.34739029e-01 1.23784673e+00 -7.74441183e-01 4.06670198e-02
1.46474823e-01 5.97352743e-01 2.48970836e-02 2.30580226e-01
-1.08354270e+00 1.27985060e+00 1.09943771e+00 2.93817390e-02
-1.01571727e+00 -6.39540553e-01 -5.04901826e-01 -4.16621804e-01
1.82354763e-01 -1.01920533e+00 1.10391200e+00 -9.45064783e-01
-1.50271940e+00 7.22222269e-01 -3.35648656e-01 -4.85328943e-01
4.57027376e-01 6.08738661e-01 -3.17063272e-01 -7.23044574e-02
-1.45574123e-01 3.26538503e-01 1.19571495e+00 -1.31534100e+00
-4.79426444e-01 -1.60087109e-01 -4.37285975e-02 1.92230836e-01
2.59261250e-01 -6.51448011e-01 3.44959319e-01 -1.36062786e-01
-2.10679129e-01 -1.12009466e+00 -6.78026438e-01 -1.13235521e+00
-4.10469294e-01 -1.95576832e-01 4.11434442e-01 -1.54934466e-01
1.23346853e+00 -1.43330848e+00 7.96896517e-01 6.58557177e-01
1.09438226e-01 2.05037445e-02 1.51146784e-01 7.30224848e-01
-4.06207532e-01 1.01037428e-01 -4.79849398e-01 -6.62315488e-02
-1.90695614e-01 2.95061886e-01 -2.26099327e-01 4.18947101e-01
-4.20052260e-01 7.40866959e-01 -9.48089182e-01 -8.21189106e-01
2.50790358e-01 7.44716749e-02 -7.86734819e-01 1.59187958e-01
-5.16920209e-01 4.09978330e-01 -4.02362943e-01 1.15406722e-01
4.68135297e-01 -3.41673732e-01 5.02058506e-01 -1.81421131e-01
-2.44322002e-01 -3.05988695e-02 -1.24485266e+00 1.89959550e+00
-6.01976812e-01 3.77708375e-01 -2.86430687e-01 -1.33437443e+00
4.25364614e-01 2.89704531e-01 6.64803326e-01 -4.28851336e-01
4.33453202e-01 6.50748089e-02 1.58802345e-01 -4.13669586e-01
7.09129989e-01 -5.37311912e-01 -1.50715068e-01 7.61260986e-01
4.98055518e-01 -5.00339985e-01 4.04827833e-01 1.91354617e-01
7.47452199e-01 7.05966875e-02 5.08962095e-01 -6.67829871e-01
5.54745197e-01 5.40489197e-01 3.19215119e-01 1.41582656e+00
-3.59377414e-02 2.91197687e-01 9.63108335e-03 -4.79572177e-01
-1.21140885e+00 -9.64401126e-01 -3.57780576e-01 8.68383825e-01
-9.50804427e-02 -6.03054881e-01 -1.23757768e+00 -6.52850866e-01
-1.38723820e-01 1.15501547e+00 -7.79894054e-01 -4.33495939e-01
-7.03478277e-01 -1.45835662e+00 3.25853862e-02 1.13326073e-01
4.68496323e-01 -1.00666153e+00 -4.57791924e-01 3.06769490e-01
-1.86623901e-01 -4.79877204e-01 -3.25481920e-03 5.34455776e-01
-1.21423042e+00 -1.17704332e+00 -1.44697726e-01 -6.07557058e-01
5.12063444e-01 -2.09244788e-01 1.05886817e+00 1.63668264e-02
-6.35839701e-01 4.50819165e-01 1.99967369e-01 6.96396530e-02
-1.03642404e+00 4.23599362e-01 2.44662210e-01 -4.17127252e-01
2.13484988e-02 -5.99441290e-01 -2.90110916e-01 -2.59731263e-01
-4.39568579e-01 -3.21972952e-03 2.26327628e-01 1.37377226e+00
5.88062704e-01 6.33424878e-01 3.28227699e-01 -9.79785085e-01
4.96988922e-01 -4.42331642e-01 -6.75867498e-01 4.33992088e-01
-1.08855724e+00 6.08322382e-01 6.26673818e-01 -3.63755137e-01
-1.28672791e+00 -4.16091681e-02 -8.93579572e-02 -2.61292785e-01
-3.43841724e-02 4.99298900e-01 1.69729516e-01 -3.79660279e-01
9.11129415e-01 2.83101648e-01 -2.40219697e-01 -2.31272802e-02
5.18270910e-01 2.27652177e-01 4.54821527e-01 -1.10065591e+00
6.67297781e-01 2.71546960e-01 4.65427935e-01 -9.00601149e-01
-1.11345530e+00 -7.22467229e-02 -6.17944717e-01 -3.39621842e-01
8.04719925e-01 -4.98903275e-01 -9.03994799e-01 2.05342159e-01
-1.11811578e+00 -4.54952240e-01 -6.22465253e-01 3.61072600e-01
-1.15691960e+00 9.94134471e-02 -4.26350176e-01 -9.98441815e-01
-3.19693208e-01 -9.41705823e-01 5.52479446e-01 3.51100922e-01
-1.39204770e-01 -1.36211991e+00 7.64095068e-01 4.21798557e-01
-1.01668701e-01 3.60777438e-01 1.21582460e+00 -3.42551053e-01
-5.06845295e-01 -2.62850612e-01 4.71016347e-01 1.84932694e-01
-3.64670396e-01 3.87274593e-01 -1.01607859e+00 -6.94826543e-01
1.15596943e-01 -5.93817711e-01 9.34680223e-01 7.54053831e-01
1.04632390e+00 -1.93969011e-01 -4.84801084e-01 5.46836615e-01
1.90221155e+00 1.59782112e-01 3.11799288e-01 2.51100421e-01
4.36937511e-01 9.51404199e-02 3.26983571e-01 3.91403198e-01
-1.50066376e-01 3.23286533e-01 1.62235975e-01 2.99954325e-01
3.00727457e-01 -1.65837899e-01 -7.65231578e-03 8.33636701e-01
-4.39174712e-01 -1.75806746e-01 -8.86479914e-01 1.47775277e-01
-1.89299369e+00 -1.60855353e+00 5.10125346e-02 1.94502091e+00
9.05361414e-01 2.91902632e-01 1.78163514e-01 3.20968270e-01
8.57770801e-01 3.89253795e-01 -7.78265238e-01 -7.14603662e-01
1.49672806e-01 8.61748397e-01 5.00383854e-01 8.96462440e-01
-8.98862839e-01 9.13770020e-01 9.06955719e+00 1.05267823e+00
-9.28467035e-01 3.89169693e-01 4.16699141e-01 -3.83699626e-01
-7.59841502e-03 8.41246173e-02 -7.69003570e-01 2.44649783e-01
1.48347223e+00 -5.98573804e-01 8.81616890e-01 7.04782128e-01
1.45494223e-01 -2.22883284e-01 -1.07345200e+00 8.52396250e-01
-4.70262356e-02 -1.89554679e+00 1.11719519e-01 3.23801517e-01
1.62766099e+00 1.45204723e-01 8.26470703e-02 5.47146857e-01
5.35363615e-01 -9.66573060e-01 7.45029449e-01 6.43939555e-01
4.83572423e-01 -1.37203336e+00 4.45706427e-01 6.23961508e-01
-8.65959585e-01 -2.36025333e-01 -4.67040807e-01 -1.46300584e-01
-8.41588154e-03 3.52064490e-01 -2.95576423e-01 6.56589329e-01
1.27413258e-01 4.56699491e-01 -2.38794357e-01 9.79657888e-01
1.40596047e-01 7.09631145e-01 -1.44272707e-02 -1.01689070e-01
3.49586219e-01 -4.90428805e-01 5.49338877e-01 1.11454809e+00
-8.09943378e-02 2.86716461e-01 3.04349571e-01 1.14174700e+00
1.12574566e-02 -1.81371272e-01 -5.07502437e-01 -4.08105068e-02
3.64047170e-01 1.16045618e+00 -6.85757101e-01 -2.90920317e-01
1.41198903e-01 7.63603926e-01 4.83932614e-01 4.78976130e-01
-9.38895941e-01 -2.71784574e-01 3.30854088e-01 -4.70571309e-01
3.48713726e-01 -1.60280287e-01 -2.58251548e-01 -1.15144491e+00
-8.07049930e-01 -9.75366175e-01 5.49152792e-01 -4.26804066e-01
-1.24897552e+00 3.85603398e-01 4.46264148e-01 -6.66377068e-01
-6.61886334e-01 -5.01640320e-01 -9.65683281e-01 8.75886321e-01
-8.02686155e-01 -5.67304194e-01 5.55858426e-02 3.84683251e-01
5.67549944e-01 -3.27611774e-01 9.65883136e-01 -4.18477416e-01
-9.24701035e-01 4.69346464e-01 6.40538752e-01 -4.97895867e-01
-1.44197732e-01 -1.45551205e+00 2.53287017e-01 5.37181318e-01
2.40809813e-01 5.64111948e-01 1.23154318e+00 -5.04609466e-01
-1.63260794e+00 -7.84290969e-01 3.55363488e-01 -4.66761649e-01
4.81998384e-01 4.31110077e-02 -7.74718821e-01 2.97986448e-01
3.62369508e-01 -2.30175182e-01 5.82407892e-01 2.36451894e-01
-1.45825684e-01 2.13480443e-01 -1.38636005e+00 2.17105627e-01
9.78346825e-01 -7.91301727e-01 -7.82452881e-01 4.97642875e-01
-2.01410279e-02 -5.25783777e-01 -7.12101161e-01 7.79805034e-02
3.95368785e-01 -9.48760152e-01 1.04444647e+00 -1.09603381e+00
2.72516727e-01 1.31504700e-01 2.84708943e-02 -1.59628212e+00
-3.54895830e-01 -8.63077879e-01 -5.21162510e-01 6.26861870e-01
3.53802681e-01 -3.71232271e-01 1.22305083e+00 1.74490541e-01
7.49972239e-02 -6.41281426e-01 -1.22639692e+00 -9.24411356e-01
6.75478399e-01 -3.41734529e-01 1.34700134e-01 9.13611948e-01
1.89354852e-01 3.48923683e-01 -1.96591541e-01 1.14864692e-01
1.30466080e+00 4.41258907e-01 1.53251201e-01 -1.15233743e+00
-6.32716477e-01 -3.90636712e-01 -4.27690521e-02 -2.86548078e-01
8.97714496e-01 -1.23947775e+00 1.05280742e-01 -1.16474450e+00
5.97414732e-01 -2.58058220e-01 -3.60998511e-01 -1.34271070e-01
-2.53965463e-05 1.36629231e-02 -2.03266278e-01 -4.19674851e-02
-5.70686638e-01 6.29109800e-01 9.16333497e-01 -2.86572605e-01
-2.26129696e-01 9.07181278e-02 -3.16899568e-01 5.74671626e-01
9.18874741e-01 -9.64867592e-01 -5.61580956e-01 2.75648952e-01
3.76828164e-01 4.57555652e-01 2.26496220e-01 -1.15952432e+00
2.68235415e-01 -5.63823462e-01 1.41578257e-01 -5.06162763e-01
3.74191761e-01 -1.58199310e-01 2.73398399e-01 7.05772638e-01
-5.00469923e-01 6.72557354e-02 1.22273766e-01 6.72776937e-01
1.59521788e-01 -1.06321776e+00 1.27829742e+00 -6.42423332e-01
-3.50333333e-01 1.16216287e-01 -5.44362187e-01 7.90795982e-01
7.82371104e-01 -7.58462921e-02 9.23746526e-02 -3.68743129e-02
-7.53185570e-01 -8.73876661e-02 4.05680954e-01 -4.12738174e-01
2.70105183e-01 -9.72027063e-01 -4.80778694e-01 6.87138289e-02
-2.90008247e-01 -5.52917242e-01 2.94350088e-01 3.96651149e-01
-3.28067780e-01 2.35996887e-01 -2.16945216e-01 -4.52339709e-01
-8.85080218e-01 4.56189781e-01 9.58973050e-01 -2.70461500e-01
-3.49930286e-01 9.20013189e-01 -7.15341985e-01 -3.62129211e-01
-4.15947288e-02 1.53825894e-01 3.47866416e-01 4.12797667e-02
2.20288187e-01 1.00478041e+00 -9.39075723e-02 -3.36458206e-01
-6.25158250e-02 3.69838715e-01 -8.59012455e-02 -4.51973498e-01
1.14686143e+00 2.34597415e-01 -1.78366244e-01 6.87153578e-01
1.27316856e+00 -4.85794336e-01 -1.08517408e+00 6.84597343e-02
-5.66804558e-02 6.95304498e-02 6.09992266e-01 -5.97714901e-01
-8.56448352e-01 6.92405164e-01 6.67590380e-01 1.77957609e-01
5.47143936e-01 -2.16040820e-01 5.30295908e-01 1.00662994e+00
7.32787371e-01 -1.57326734e+00 3.46959531e-02 4.87411499e-01
4.15173739e-01 -1.23889399e+00 5.18604696e-01 1.90450966e-01
-3.68508160e-01 1.25568926e+00 5.12850642e-01 -3.87929618e-01
6.42871559e-01 1.61394939e-01 -6.59555316e-01 -2.67481446e-01
-9.12608027e-01 1.00603953e-01 1.84150577e-01 5.61062157e-01
-3.20463628e-02 -1.38864219e-01 -1.75104722e-01 4.18668501e-02
-2.49241382e-01 -7.75865242e-02 6.13292456e-01 1.03151476e+00
-6.06982350e-01 -1.17040300e+00 -1.88573197e-01 4.48262334e-01
-6.10118061e-02 1.00658588e-01 -5.36331050e-02 8.13595653e-01
2.17483670e-01 7.80212164e-01 -7.99144730e-02 -2.18803853e-01
-1.45435497e-01 5.40441811e-01 1.33134222e+00 -5.60408592e-01
-8.34315956e-01 -4.38305169e-01 1.00274617e-02 -3.94625783e-01
-5.43340743e-01 -8.71032059e-01 -1.25442219e+00 -7.09689021e-01
-6.58790350e-01 9.75993454e-01 5.76551616e-01 1.25693142e+00
-2.95584083e-01 3.77406090e-01 8.90255630e-01 -1.05053651e+00
-9.05842543e-01 -5.03826559e-01 -7.14686871e-01 -1.65372595e-01
2.83556581e-01 -6.26759231e-01 -6.73549473e-01 -2.68252254e-01]
|
[5.609994888305664, 4.857828617095947]
|
23cde788-4385-4810-94ca-702c00ae2ba0
|
co-learning-of-word-representations-and
| null | null |
https://aclanthology.org/C14-1015
|
https://aclanthology.org/C14-1015.pdf
|
Co-learning of Word Representations and Morpheme Representations
| null |
['Tie-Yan Liu', 'Bin Gao', 'Siyu Qiu', 'Qing Cui', 'Jiang Bian']
|
2014-08-01
|
co-learning-of-word-representations-and-1
|
https://aclanthology.org/C14-1015
|
https://aclanthology.org/C14-1015.pdf
|
coling-2014-8
|
['learning-word-embeddings']
|
['methodology']
|
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
|
[-7.416479587554932, 3.5920908451080322]
|
1c9bced2-5170-4a91-97f8-9d00222d1b98
|
learning-to-caricature-via-semantic-shape
|
2008.05090
| null |
https://arxiv.org/abs/2008.05090v2
|
https://arxiv.org/pdf/2008.05090v2.pdf
|
Learning to Caricature via Semantic Shape Transform
|
Caricature is an artistic drawing created to abstract or exaggerate facial features of a person. Rendering visually pleasing caricatures is a difficult task that requires professional skills, and thus it is of great interest to design a method to automatically generate such drawings. To deal with large shape changes, we propose an algorithm based on a semantic shape transform to produce diverse and plausible shape exaggerations. Specifically, we predict pixel-wise semantic correspondences and perform image warping on the input photo to achieve dense shape transformation. We show that the proposed framework is able to render visually pleasing shape exaggerations while maintaining their facial structures. In addition, our model allows users to manipulate the shape via the semantic map. We demonstrate the effectiveness of our approach on a large photograph-caricature benchmark dataset with comparisons to the state-of-the-art methods.
|
['Ming-Hsuan Yang', 'Yijun Li', 'Yu-Ting Chang', 'Yi-Hsuan Tsai', 'Wei-Chih Hung', 'Deng Cai', 'Wenqing Chu']
|
2020-08-12
| null | null | null | null |
['caricature']
|
['computer-vision']
|
[ 6.05785370e-01 4.20528263e-01 4.95523036e-01 -5.31494737e-01
-1.62838936e-01 -5.30425310e-01 6.59363031e-01 -4.14828032e-01
1.11836590e-01 5.62278986e-01 8.00197721e-02 4.38778624e-02
3.22350860e-01 -8.96169603e-01 -8.81164551e-01 -1.73306197e-01
5.97544432e-01 2.99985886e-01 -1.12272188e-01 -2.88710356e-01
4.69020724e-01 1.23114574e+00 -1.63792539e+00 3.81343424e-01
9.68037665e-01 9.04330134e-01 -3.01151723e-02 3.64510715e-01
-4.09246027e-01 3.36105615e-01 -6.10999107e-01 -9.33092952e-01
3.82302523e-01 -4.62981045e-01 -4.51889336e-01 5.78623950e-01
9.14437890e-01 -1.89982593e-01 5.69081353e-03 1.13535774e+00
3.29359680e-01 -1.66751906e-01 7.81504810e-01 -1.38077378e+00
-1.09745777e+00 1.08688883e-01 -9.01026011e-01 -6.36047542e-01
3.67501438e-01 2.35005245e-01 6.44862533e-01 -1.08962727e+00
9.58394825e-01 1.65784502e+00 7.42734790e-01 7.75780678e-01
-1.51354945e+00 -6.65311038e-01 4.21968400e-02 -1.61188290e-01
-1.31534588e+00 -4.20498699e-01 1.33600819e+00 -2.31836766e-01
3.13083351e-01 5.84503055e-01 1.06033957e+00 9.26450074e-01
2.26964448e-02 8.44396710e-01 1.28800130e+00 -3.79899442e-01
3.71570081e-01 3.79800908e-02 -6.55499697e-01 8.15523446e-01
9.34873819e-02 -6.47478923e-02 -2.45819047e-01 -1.15979940e-01
1.39649975e+00 5.88154308e-02 -1.92855000e-01 -6.57010138e-01
-1.00565505e+00 6.02606475e-01 7.09024310e-01 1.03248671e-01
-4.46475506e-01 6.14196897e-01 -9.91996825e-02 5.08103408e-02
5.32130361e-01 8.25145185e-01 7.52895549e-02 7.81842694e-02
-9.71326530e-01 4.87416297e-01 4.56373364e-01 9.49083090e-01
7.06045747e-01 2.61397094e-01 -2.02378646e-01 8.41933429e-01
6.95028389e-03 5.88793695e-01 9.12742782e-03 -1.16162586e+00
8.29907209e-02 7.80690432e-01 2.14364022e-01 -1.26341403e+00
-1.29676074e-01 4.28223573e-02 -9.42991555e-01 9.02983904e-01
1.78493142e-01 8.44748095e-02 -9.41763043e-01 1.50029516e+00
3.01514238e-01 1.91827014e-01 -7.76075870e-02 9.00164723e-01
7.25997567e-01 6.05102956e-01 9.70051959e-02 1.96541876e-01
1.22009420e+00 -9.47353840e-01 -8.48336995e-01 -1.71271950e-01
1.12740040e-01 -1.00016832e+00 1.45177937e+00 2.02295601e-01
-1.44016051e+00 -5.11831999e-01 -8.30630779e-01 -4.04478669e-01
-1.60836011e-01 3.89578193e-01 6.52177513e-01 6.95439935e-01
-1.03696072e+00 7.26001084e-01 -3.45348835e-01 -1.11220643e-01
8.00701976e-01 3.75243574e-02 -3.97796631e-01 7.07004219e-02
-7.67598033e-01 7.10255027e-01 2.32785176e-02 7.52888024e-02
-3.55556786e-01 -1.02719712e+00 -9.33371484e-01 9.05967429e-02
-1.03868609e-02 -1.01501501e+00 9.94390428e-01 -1.41430330e+00
-1.78089893e+00 1.03290021e+00 -5.84188029e-02 6.69335015e-03
9.48433936e-01 1.35111004e-01 -8.54431018e-02 3.10216416e-02
-3.76758188e-01 1.08026814e+00 1.34246266e+00 -1.62982178e+00
-2.47954324e-01 -3.27162892e-01 1.23396059e-02 2.29835331e-01
-3.95427316e-01 -7.25377351e-02 -3.49671215e-01 -1.10700119e+00
-3.29287052e-02 -9.14684474e-01 -2.72374988e-01 9.02237356e-01
-4.47878629e-01 8.22479725e-02 1.10395086e+00 -6.30840123e-01
7.51993895e-01 -2.10847068e+00 1.66954607e-01 3.56356859e-01
2.26426318e-01 2.62192607e-01 -3.56400102e-01 2.99816996e-01
-8.91464949e-02 2.30027199e-01 -4.60541129e-01 -6.30521119e-01
4.30493392e-02 4.57728505e-02 -2.78013319e-01 1.62361845e-01
3.88872176e-01 1.19026113e+00 -7.42047548e-01 -4.36034858e-01
2.53946811e-01 8.65750611e-01 -6.94105148e-01 1.84261680e-01
-2.20090568e-01 4.05631274e-01 -4.10372049e-01 5.87725401e-01
1.03630292e+00 2.69810166e-02 -5.90993017e-02 -4.68084514e-01
-2.95648798e-02 -4.39212501e-01 -1.05091226e+00 1.81771231e+00
-5.87766707e-01 6.71092808e-01 -6.02771901e-02 -3.59033406e-01
1.34522378e+00 2.31069122e-02 1.41825020e-01 -5.34419179e-01
1.80704758e-01 2.73306906e-01 -4.45492774e-01 -3.18505526e-01
5.23460090e-01 -4.05708313e-01 1.04299814e-01 3.46423298e-01
-6.92883074e-01 -9.89764452e-01 -2.37151697e-01 -5.28645366e-02
5.94321728e-01 3.81853223e-01 2.01960355e-01 -1.63607702e-01
5.64746022e-01 -1.98724687e-01 5.88728972e-02 3.72983590e-02
3.08769763e-01 1.04695594e+00 4.39898282e-01 -8.54463816e-01
-1.41060770e+00 -1.15591788e+00 1.54696703e-01 4.76925761e-01
1.65477902e-01 3.12038437e-02 -1.23401952e+00 -4.54989582e-01
2.48093352e-01 8.77189696e-01 -8.57836604e-01 -1.76245153e-01
-6.76276386e-01 -1.27974376e-01 3.49652648e-01 5.77125251e-01
6.01414859e-01 -1.41882145e+00 -6.64702475e-01 -7.64875207e-03
4.27635014e-03 -9.63228941e-01 -8.85781765e-01 -9.16293621e-01
-7.45985687e-01 -6.92179322e-01 -1.11736858e+00 -8.34994078e-01
1.23675299e+00 2.11650655e-01 1.05876708e+00 2.50674695e-01
-5.94733119e-01 2.30728596e-01 1.69651434e-02 -5.17456234e-01
-7.03243077e-01 -3.68799895e-01 -2.57511944e-01 4.98750180e-01
-2.12935999e-01 -8.18413377e-01 -8.35675180e-01 2.42610335e-01
-1.18769741e+00 7.69448161e-01 5.91804206e-01 4.49245572e-01
7.48074293e-01 -3.08801502e-01 4.14966047e-01 -9.28222895e-01
8.74198258e-01 2.57328242e-01 -6.14945710e-01 3.59631181e-01
-1.41390756e-01 2.13757992e-01 7.10302532e-01 -5.73811889e-01
-1.29761219e+00 3.41358870e-01 -1.04899138e-01 -7.89440870e-01
-7.17808753e-02 -1.92394078e-01 -2.76624650e-01 -5.26448369e-01
6.12538695e-01 1.58631161e-01 2.47174740e-01 -5.81219375e-01
8.38083684e-01 4.82426167e-01 7.56367147e-01 -5.90435207e-01
1.12146294e+00 7.84709513e-01 2.31273338e-01 -8.50657284e-01
-3.68255049e-01 3.68698001e-01 -6.88026726e-01 -4.43124503e-01
6.13210320e-01 -4.65983480e-01 -6.19951844e-01 3.77090096e-01
-1.46183491e+00 -1.90433726e-01 -5.54166734e-01 -1.64791256e-01
-1.03347230e+00 3.99964184e-01 -1.59521058e-01 -6.24170661e-01
-5.48812330e-01 -1.00938368e+00 1.45873499e+00 2.12526798e-01
-3.93786520e-01 -7.06520915e-01 -9.21796635e-02 3.34834486e-01
4.67954338e-01 8.53997052e-01 8.99407506e-01 4.11934257e-01
-6.65861905e-01 -1.25104845e-01 -5.93818247e-01 1.16954319e-01
3.90004933e-01 3.81357044e-01 -9.79120314e-01 -3.05436254e-02
-5.79626143e-01 -1.62828088e-01 4.38543290e-01 1.12832591e-01
1.68544900e+00 -5.01666248e-01 -2.08506763e-01 7.62898028e-01
1.27885449e+00 2.11517629e-03 1.15429842e+00 -9.12316591e-02
8.77196372e-01 6.93281174e-01 4.83346939e-01 5.04864931e-01
6.26570731e-02 8.85088205e-01 4.35362965e-01 -5.15762150e-01
-5.30206263e-01 -7.88251221e-01 -1.02234811e-01 1.66583374e-01
-4.48810726e-01 1.79525286e-01 -4.03221607e-01 3.89769375e-01
-1.56592083e+00 -8.52385581e-01 -1.51963726e-01 2.00915289e+00
9.36853647e-01 -3.26804757e-01 -4.76588160e-02 6.49067163e-02
6.66629791e-01 1.49368886e-02 -4.86806601e-01 -8.01554441e-01
-4.68391692e-03 3.87010247e-01 1.98781267e-01 4.24665868e-01
-6.08333051e-01 1.19529080e+00 6.04159832e+00 9.15523708e-01
-1.16013563e+00 -2.66336352e-01 7.52827525e-01 1.27482966e-01
-8.83268058e-01 -3.23298544e-01 -1.36303648e-01 3.52771789e-01
4.76147868e-02 -3.10624748e-01 5.34625053e-01 9.01548684e-01
3.52259964e-01 2.91855663e-01 -9.37677264e-01 1.31551325e+00
2.19666675e-01 -1.61216927e+00 6.62365615e-01 -1.07949294e-01
1.13824868e+00 -1.08101797e+00 4.38286930e-01 -3.69660735e-01
1.58738226e-01 -1.26552820e+00 9.14847672e-01 8.20406497e-01
1.24385870e+00 -8.78902972e-01 7.70933777e-02 4.74290587e-02
-1.08061135e+00 3.45776320e-01 -4.93716478e-01 1.68264791e-01
2.55063981e-01 4.04870629e-01 -6.50991976e-01 7.75730908e-02
3.23069930e-01 3.11777592e-01 -5.81648707e-01 1.08602750e+00
-3.24584484e-01 -2.55328178e-01 -5.94313890e-02 -9.62085649e-02
6.83358833e-02 -4.83990699e-01 3.16669315e-01 9.55865681e-01
7.22935855e-01 1.05195478e-01 -3.32701385e-01 1.37663388e+00
-4.02184635e-01 3.76045614e-01 -8.61879587e-01 -1.02652818e-01
3.04805547e-01 1.32338071e+00 -5.43173671e-01 -3.00320148e-01
-7.54403025e-02 1.62564051e+00 4.15037572e-01 2.40527958e-01
-9.46750104e-01 -3.74659985e-01 8.61969709e-01 3.54438722e-01
6.84681162e-02 -1.18852966e-01 -6.44072771e-01 -8.79666805e-01
1.08768508e-01 -8.03521216e-01 -2.34225526e-01 -1.35612881e+00
-1.10883701e+00 8.11563134e-01 -3.48775387e-01 -1.16432357e+00
-3.12780328e-02 -3.56357366e-01 -6.55443490e-01 8.86824608e-01
-1.41459048e+00 -1.63118994e+00 -8.25711668e-01 5.62331438e-01
5.52414477e-01 9.32461172e-02 7.45896697e-01 -3.44724543e-02
3.33013423e-02 6.44687891e-01 -3.23540896e-01 -1.13116272e-01
5.53540707e-01 -1.01892984e+00 9.04904783e-01 4.96495545e-01
1.65849626e-01 3.19578767e-01 8.23478818e-01 -5.65756083e-01
-1.26682711e+00 -1.30150557e+00 8.03669810e-01 -3.16540867e-01
1.06174186e-01 -3.19268405e-01 -8.01614761e-01 5.27969778e-01
1.21304885e-01 5.49856424e-02 2.28523791e-01 -6.50877953e-01
-4.47396815e-01 -2.57574953e-02 -1.60433912e+00 1.11360812e+00
1.50345373e+00 -1.55044794e-01 -5.20388842e-01 1.45880416e-01
4.28761572e-01 -3.95354003e-01 -6.02253973e-01 2.31829137e-01
8.16272736e-01 -1.03646326e+00 1.20248222e+00 -5.08079350e-01
6.26776099e-01 -3.01909536e-01 3.18595260e-01 -1.69316638e+00
-3.06079566e-01 -9.13733006e-01 2.79472709e-01 9.77145970e-01
2.02768028e-01 -4.34981078e-01 9.43993866e-01 8.54788184e-01
3.15974616e-02 -5.17520249e-01 -5.80413461e-01 -7.45798647e-01
1.02159602e-03 -4.48867790e-02 1.16364837e+00 8.35886896e-01
-3.32609862e-01 -2.49697208e-01 -5.01812518e-01 -1.53658524e-01
7.10905373e-01 4.81637686e-01 1.10538006e+00 -1.29355931e+00
1.16691031e-01 -6.93697214e-01 -4.16191131e-01 -7.93656349e-01
2.44501278e-01 -6.51328385e-01 -3.00467551e-01 -1.47665834e+00
-9.24840104e-03 -5.16030967e-01 4.63125139e-01 4.77921844e-01
-9.02159140e-02 7.95247614e-01 4.52935249e-01 -7.27444291e-02
6.23071156e-02 8.56602132e-01 2.05973482e+00 -1.11080498e-01
-2.00334519e-01 -5.06028682e-02 -9.14971173e-01 8.53413820e-01
7.90083408e-01 -1.36059225e-01 -5.12015820e-01 -4.43276823e-01
1.87282696e-01 -7.25410506e-02 5.11796892e-01 -8.78829539e-01
-2.28613347e-01 -4.63600814e-01 6.46459877e-01 -3.20803046e-01
6.83299482e-01 -9.40225840e-01 6.83444977e-01 4.70831335e-01
-3.93286347e-01 1.94071457e-02 2.01996043e-01 2.91351348e-01
1.06150825e-02 -1.27908271e-02 1.15760350e+00 -3.11621483e-02
-4.64224547e-01 4.50730294e-01 9.01310295e-02 -2.34803066e-01
1.15172672e+00 -3.67311865e-01 -2.04657428e-02 -5.80924928e-01
-4.82080966e-01 -2.41004720e-01 1.00243020e+00 6.25119448e-01
1.14130151e+00 -1.86082745e+00 -8.26254070e-01 4.90031332e-01
2.23187879e-01 -2.76599079e-01 2.82070637e-01 1.89768031e-01
-1.07949615e+00 7.49160349e-02 -7.53769875e-01 -1.24119245e-01
-1.49735606e+00 6.43854022e-01 3.09205711e-01 2.74050266e-01
-8.22660089e-01 6.35923386e-01 6.43045485e-01 -3.07952940e-01
-1.22910760e-01 -4.39161390e-01 -8.71049762e-02 -3.73144507e-01
6.80773914e-01 2.34399527e-01 -2.31262818e-01 -5.60583949e-01
-1.17284628e-02 1.08750463e+00 4.07007068e-01 -7.16128945e-02
1.19609153e+00 2.75441315e-02 -9.08754915e-02 -2.41772160e-01
1.18188226e+00 2.01517627e-01 -1.53774405e+00 1.08012315e-02
-4.66927797e-01 -1.04212379e+00 -1.97289899e-01 -6.84156418e-01
-1.32219613e+00 9.05146956e-01 3.40112269e-01 -9.31274071e-02
1.17609203e+00 -2.09719613e-01 8.24165821e-01 4.56095785e-02
5.04650414e-01 -8.32904100e-01 3.58534008e-01 -2.36756936e-01
1.84030521e+00 -8.40793133e-01 -1.63233742e-01 -1.00936711e+00
-7.75322914e-01 1.19632947e+00 4.85250384e-01 -3.63961637e-01
2.96045870e-01 3.08969259e-01 3.96615341e-02 -2.20716968e-01
-1.98346272e-01 1.04640387e-01 5.41831374e-01 7.76665509e-01
1.85831651e-01 1.90422460e-01 -4.59469706e-01 1.72732830e-01
-4.82363254e-01 2.42089659e-01 5.35861075e-01 5.05282581e-01
-2.29670286e-01 -1.09197664e+00 -5.37380040e-01 5.42338714e-02
7.89255872e-02 3.30702998e-02 -8.40690315e-01 7.12036848e-01
-1.04343250e-01 4.83965874e-01 1.12606928e-01 3.49555202e-02
6.85940564e-01 -1.25212967e-01 7.87032902e-01 -2.77986825e-01
-2.83889383e-01 -2.05018103e-01 -8.67666826e-02 -7.96485484e-01
-2.51934826e-01 -3.85498941e-01 -1.11992526e+00 -5.53390861e-01
3.29934478e-01 -3.66221994e-01 7.79055357e-01 3.46844971e-01
5.57610452e-01 3.54727477e-01 7.94769049e-01 -1.19634914e+00
-3.82411629e-01 -5.83603144e-01 -6.60179019e-01 9.89768744e-01
-7.69307688e-02 -6.78492963e-01 -2.10072950e-01 3.24138105e-01]
|
[12.005443572998047, -0.44594117999076843]
|
178edf45-0325-40c1-860f-403fdac4ba58
|
d2net-a-denoising-and-dereverberation-network
| null | null |
https://ieeexplore.ieee.org/abstract/document/9979863
|
http://www.apsipa.org/proceedings/2022/APSIPA%202022/ThPM1-2/1570833515.pdf
|
D²Net: A Denoising and Dereverberation Network Based on Two-branch Encoder and Dual-path Transformer
|
The simultaneous denoising and dereverberation for single-channel mixture speech under the complicated acoustic environment is considered to be a challengeable task. In this paper, we propose a denoising and dereverberation network named as D²Net in which a two-branch encoder (TBE) is designed to extract and selectively fuse features with different granularity. In addition, we design a global-local dual-path transformer (GLDPT) which introduces the local dense synthesizer attention (LDSA) in the dual-path transformer to improve the perception of local information. We evaluated our proposed D²Net and conducted ablation studies on the VoiceBank+DEMAND and WHAMR! datasets. Meanwhile, we chose three types of data in the WHAMR! dataset to verify the ability of the D²Net on the tasks of denoising-only, dereverberation-only, and simultaneous denoising and dereverberation, respectively. Experimental results show that our proposed model outperforms the comparative models, and all achieve better performance on the tasks of simultaneous denoising and dereverberation, dereverberation-only, and denoising-only, while keeping a small number of network parameters.
|
['and Ying Hu', 'Yadong Chen', 'Wenbing Wei', 'Liusong Wang']
|
2022-11-21
| null | null | null |
apsipa-asc-2022-11
|
['speech-enhancement']
|
['speech']
|
[-2.57415771e-01 -3.82893354e-01 4.87650424e-01 -3.66944999e-01
-1.23424852e+00 -4.28866446e-01 1.65155485e-01 -6.85455084e-01
-3.70094240e-01 3.49187523e-01 5.67417085e-01 -4.72218633e-01
1.06223419e-01 -2.19890103e-01 -4.34915513e-01 -1.02798712e+00
3.73961449e-01 -2.10801288e-01 -3.12577486e-02 -2.84138858e-01
-4.69218194e-01 1.60494208e-01 -1.31675088e+00 4.36779082e-01
1.04813898e+00 1.05586338e+00 9.66374993e-01 7.31622279e-01
2.63592154e-01 5.71062684e-01 -8.82691681e-01 -5.09069562e-01
3.19136739e-01 -6.28947198e-01 6.88141659e-02 -1.39114603e-01
4.22029734e-01 -4.78563160e-01 -7.00635612e-01 1.22536981e+00
1.36117887e+00 2.79976457e-01 1.49765372e-01 -7.90182531e-01
-7.38325953e-01 9.31890845e-01 -4.52168584e-01 4.80253309e-01
9.49361622e-02 4.61173326e-01 1.24500382e+00 -1.27496779e+00
-1.45811260e-01 1.60634804e+00 7.91214645e-01 4.21748757e-01
-1.07168746e+00 -1.21221554e+00 2.94505298e-01 4.42479104e-01
-1.44513571e+00 -1.16846228e+00 8.32537115e-01 6.90793172e-02
1.12446320e+00 3.37372065e-01 3.33418727e-01 1.21075201e+00
-1.09934747e-01 9.61814702e-01 1.10988402e+00 -8.70971978e-02
-3.25718634e-02 -2.18946904e-01 -8.98366570e-02 3.25286150e-01
-1.97719470e-01 4.16586578e-01 -9.89737570e-01 2.78556764e-01
6.76614404e-01 -4.11610067e-01 -7.53621042e-01 7.36581504e-01
-1.19673443e+00 4.20384645e-01 2.45953709e-01 1.80887237e-01
-3.16377461e-01 1.90908551e-01 3.75510722e-01 7.41340399e-01
7.82665670e-01 2.00871035e-01 -6.91749990e-01 -2.44696632e-01
-8.63249898e-01 4.58322428e-02 6.72821820e-01 9.97707546e-01
3.55476975e-01 6.97535336e-01 -3.92872125e-01 1.27185786e+00
4.70038950e-01 7.48521328e-01 7.11092114e-01 -6.87747180e-01
8.13900530e-01 -4.89849031e-01 -1.58728361e-01 -6.13883197e-01
-2.35853642e-01 -1.01855457e+00 -1.13714647e+00 -1.86686844e-01
-5.31437770e-02 -5.32129705e-01 -9.89021897e-01 2.07378411e+00
6.73572868e-02 5.50921261e-01 2.10097983e-01 1.10280073e+00
1.02630305e+00 9.70212996e-01 -1.59578726e-01 -3.38154227e-01
1.25687563e+00 -1.24663913e+00 -1.38891447e+00 -3.32406461e-01
2.70428360e-01 -1.04933619e+00 1.15781975e+00 5.69768012e-01
-1.09764743e+00 -9.79787827e-01 -1.09619558e+00 -3.80175918e-01
1.41890630e-01 3.61143202e-01 2.27952510e-01 6.72841430e-01
-1.04408956e+00 3.74610811e-01 -8.62107813e-01 2.25190222e-01
1.12277426e-01 4.28060293e-02 -2.58209080e-01 -1.65847734e-01
-1.47071433e+00 7.98233867e-01 -2.12779213e-02 6.75203919e-01
-1.21641266e+00 -8.29466641e-01 -8.53988528e-01 2.58511484e-01
2.69269288e-01 -4.56021488e-01 1.33702111e+00 -4.55265611e-01
-1.58990526e+00 3.19313049e-01 -5.05762041e-01 -3.59283894e-01
2.90525675e-01 -6.20102167e-01 -7.64651418e-01 -2.29791120e-01
-1.42961428e-01 2.29981095e-01 9.78972733e-01 -1.03694105e+00
-5.73977470e-01 -2.62184680e-01 -3.10195446e-01 6.03076458e-01
-4.87241186e-02 1.37947902e-01 -2.43936002e-01 -1.26060164e+00
2.48853967e-01 -4.14829671e-01 1.14263080e-01 -4.99534369e-01
-7.26149321e-01 -1.12466104e-01 8.09199870e-01 -1.42186940e+00
1.47007644e+00 -2.82534075e+00 3.15361470e-01 -2.39708766e-01
2.55933013e-02 4.45178509e-01 -5.16411602e-01 2.83927351e-01
-1.25759259e-01 1.62698761e-01 -1.79821089e-01 -8.02956700e-01
2.17351034e-01 1.01167403e-01 -4.69463259e-01 3.78627449e-01
1.02625333e-01 6.75247073e-01 -6.99234545e-01 -1.39413789e-01
2.26432070e-01 7.37696171e-01 -6.03809655e-01 5.42471051e-01
1.24056399e-01 5.32650173e-01 1.08537965e-01 6.22715950e-01
8.37248802e-01 6.21168911e-01 -6.63555041e-02 -6.01781368e-01
-9.08811241e-02 7.85728276e-01 -1.23288143e+00 1.75646245e+00
-9.06011522e-01 5.30646443e-01 7.49548912e-01 -6.29893959e-01
9.29440558e-01 7.39260674e-01 -1.29455030e-01 -8.90228033e-01
-2.19079875e-03 6.23921752e-01 5.51426053e-01 -4.20476735e-01
2.20528439e-01 -3.24780554e-01 1.13216490e-01 1.76162705e-01
3.64023089e-01 -2.92493880e-01 -2.49551922e-01 -1.76112965e-01
1.13683331e+00 -2.05837831e-01 -7.21223000e-03 -2.02276379e-01
2.84106463e-01 -1.01726401e+00 8.76344144e-01 7.32488394e-01
-3.03347528e-01 8.46222341e-01 1.80349186e-01 2.92060196e-01
-6.43963277e-01 -1.27004278e+00 1.15018308e-01 1.30786276e+00
2.08191574e-02 -5.18522561e-01 -7.47376859e-01 -2.41918415e-01
-2.69512296e-01 1.01374471e+00 1.05653387e-02 -4.53994960e-01
-6.70912147e-01 -7.51390755e-01 6.58298671e-01 3.21371913e-01
8.98460567e-01 -8.91397595e-01 6.04017004e-02 1.86886743e-01
-5.43223262e-01 -1.01740992e+00 -1.14232159e+00 4.37232763e-01
-2.22167507e-01 -4.41272348e-01 -6.60440028e-01 -9.09153759e-01
-2.24427320e-02 5.17224848e-01 8.59913766e-01 -1.67393386e-01
4.74782139e-01 -1.42449468e-01 -5.25422573e-01 -3.68918359e-01
-2.29011893e-01 2.07868651e-01 3.75737548e-02 3.14228952e-01
-7.52897188e-02 -1.03742099e+00 -6.94894016e-01 4.52892244e-01
-6.98932350e-01 -6.96384236e-02 7.52156436e-01 9.81559157e-01
5.69335938e-01 3.70940238e-01 1.03547418e+00 -4.22309875e-01
1.15655100e+00 -3.93204540e-01 -1.17254943e-01 4.43236865e-02
-2.27970973e-01 -3.05148840e-01 6.93458319e-01 -5.46930552e-01
-1.30758190e+00 -4.78392065e-01 -8.86596501e-01 -5.55883169e-01
5.80221377e-02 5.33350289e-01 -1.10539389e+00 5.27315557e-01
1.57967165e-01 3.92646760e-01 -2.81292766e-01 -1.12386727e+00
4.92826998e-01 1.14092004e+00 6.18586600e-01 -2.22253114e-01
6.86265111e-01 6.43350035e-02 -7.15559006e-01 -9.35024142e-01
-8.30634415e-01 -2.97734112e-01 -5.54250777e-02 8.47015530e-02
6.17619336e-01 -1.32145214e+00 -3.83516550e-01 1.07680488e+00
-1.24200022e+00 -4.39391404e-01 -2.99281776e-01 6.90987766e-01
-1.78296566e-01 2.12070331e-01 -7.06873000e-01 -8.05768788e-01
-4.82451439e-01 -1.58275378e+00 9.42295551e-01 -5.34520485e-02
1.24043003e-01 -4.99399364e-01 -3.36737752e-01 2.33063057e-01
8.28044534e-01 -5.03051996e-01 6.95766270e-01 -6.63557112e-01
-3.25761944e-01 1.47286251e-01 9.26353969e-03 1.06668150e+00
3.60478848e-01 -5.00352919e-01 -1.44417882e+00 -4.04128760e-01
5.05146563e-01 -2.04881374e-02 1.08452034e+00 4.07859117e-01
1.24427497e+00 -3.24732751e-01 1.53929651e-01 9.19770360e-01
6.43891156e-01 4.11563516e-01 7.43695915e-01 -3.85004014e-01
7.48132944e-01 2.38952860e-01 3.85984182e-01 1.74023062e-01
3.47720414e-01 6.66323125e-01 2.75327802e-01 -3.42517048e-01
-9.23973143e-01 -3.10679585e-01 7.21296251e-01 1.79705012e+00
2.88323820e-01 -5.59358537e-01 -2.39711806e-01 6.32931590e-01
-1.47705102e+00 -7.07275629e-01 1.06340684e-01 2.02649212e+00
1.16791618e+00 -7.10069835e-02 -4.25843656e-01 2.28403106e-01
7.48004496e-01 6.04912460e-01 -5.74668705e-01 -3.00054431e-01
-3.96463364e-01 6.56584501e-01 1.12952158e-01 7.61663198e-01
-8.64578843e-01 9.18536842e-01 5.24509525e+00 1.45633399e+00
-1.02194583e+00 5.72921515e-01 6.37743831e-01 -2.11313650e-01
-4.58157688e-01 -2.15706393e-01 -7.14274824e-01 6.73674226e-01
9.07246649e-01 1.41977459e-01 8.82097304e-01 3.66674602e-01
5.22612929e-01 2.11579382e-01 -9.81857300e-01 1.07348919e+00
7.03732192e-04 -5.97649872e-01 -2.49013826e-01 -3.57129484e-01
3.63558739e-01 1.15461081e-01 3.13535303e-01 5.82525969e-01
2.09845155e-01 -9.47439790e-01 9.86452699e-01 3.38423908e-01
9.60783243e-01 -8.86034608e-01 6.23538136e-01 3.85479540e-01
-1.34000909e+00 -8.17907602e-02 -1.10151894e-01 1.15457453e-01
4.15706933e-01 9.51236665e-01 -3.81975204e-01 5.55806458e-01
7.87551105e-01 5.67744315e-01 -1.03627026e-01 9.36005175e-01
-7.31841505e-01 1.04585862e+00 -3.81072730e-01 3.64174932e-01
-1.68810233e-01 4.48162779e-02 7.90689409e-01 1.20636344e+00
5.27343512e-01 -6.22849651e-02 -1.60647601e-01 6.88450098e-01
-3.51627439e-01 -2.81854928e-01 8.32023192e-03 1.92980528e-01
1.05164504e+00 1.08408713e+00 1.75529867e-01 -2.86011733e-02
-2.96656579e-01 9.90498066e-01 4.80065718e-02 6.98380828e-01
-8.60672832e-01 -7.16608524e-01 1.04437637e+00 -2.62888819e-01
5.47752321e-01 -3.05518329e-01 1.49640189e-02 -1.24597466e+00
2.31202282e-02 -1.29971468e+00 -1.23483077e-01 -8.76311541e-01
-1.41372311e+00 8.76049280e-01 -2.69895136e-01 -8.57406199e-01
1.32834300e-01 -4.39708292e-01 -6.68651342e-01 1.38387370e+00
-1.76371872e+00 -1.05683661e+00 4.79624607e-02 6.01792157e-01
9.08522367e-01 -7.04295784e-02 6.53043509e-01 8.56956303e-01
-9.41794276e-01 8.79974842e-01 6.77461624e-02 2.64445450e-02
9.15010631e-01 -1.08963859e+00 5.43555081e-01 1.22966146e+00
1.74125955e-01 4.32753474e-01 5.22641897e-01 -4.84593004e-01
-1.20116878e+00 -1.21818411e+00 7.79009819e-01 1.97441638e-01
4.31358069e-01 -8.83380711e-01 -9.49551046e-01 7.02850044e-01
5.35982549e-01 4.69713248e-02 4.79146898e-01 3.73786911e-02
-4.17900473e-01 -5.89827836e-01 -8.41613114e-01 6.47750080e-01
1.23739707e+00 -8.41850102e-01 -4.65249687e-01 3.57179157e-02
1.35587370e+00 -5.55367172e-01 -6.30576134e-01 4.47073191e-01
2.99959064e-01 -9.66704667e-01 9.51631069e-01 -9.15692002e-02
6.48836941e-02 -3.78461063e-01 -7.07930565e-01 -2.07854033e+00
-1.33112833e-01 -1.15014052e+00 -2.11978719e-01 1.72895801e+00
4.69963998e-01 -8.53966236e-01 -2.36802503e-01 -9.20428261e-02
-9.19461131e-01 -5.22545159e-01 -1.24991322e+00 -7.19342887e-01
-7.25010931e-02 -7.32726514e-01 7.27660358e-01 5.90491414e-01
-4.93651360e-01 4.48603749e-01 -6.90949202e-01 4.15647119e-01
3.05217117e-01 -1.87295437e-01 4.48179066e-01 -4.78068322e-01
-6.25187218e-01 -1.89233750e-01 2.12278292e-01 -1.66415131e+00
1.53861240e-01 -7.34000027e-01 4.37079787e-01 -1.32552111e+00
-3.54733139e-01 -1.90155521e-01 -5.17348409e-01 2.42021829e-01
-3.97929847e-01 -8.08677897e-02 2.06049830e-01 3.21130641e-02
-2.11481795e-01 9.01222229e-01 1.51913869e+00 -3.25178653e-02
-1.20199107e-01 9.74649191e-02 -9.32543397e-01 6.26408041e-01
5.74629188e-01 -2.48744845e-01 -5.12071073e-01 -9.59679782e-01
-1.80153310e-01 2.52006292e-01 2.01672614e-01 -7.86970496e-01
2.01646358e-01 3.60959917e-01 -9.95654240e-02 -6.00938857e-01
7.56254792e-01 -5.52167296e-01 -5.56155071e-02 3.91137823e-02
-3.42167944e-01 -6.88408017e-02 1.98586583e-01 4.52837229e-01
-5.26495576e-01 1.64540529e-01 7.76173532e-01 -3.05508431e-02
-4.16941077e-01 2.27001116e-01 -2.79152393e-01 9.99782681e-02
3.32552850e-01 3.67468297e-01 -3.50964278e-01 -6.39656067e-01
-8.23076844e-01 2.72576511e-02 -3.31495613e-01 4.28182423e-01
7.66604781e-01 -1.35391521e+00 -1.07819963e+00 4.31463271e-01
-5.09280264e-01 1.74739152e-01 5.53989589e-01 9.38173056e-01
5.83376549e-03 -4.78731059e-02 4.08860445e-01 -2.43254617e-01
-9.36470747e-01 2.53034711e-01 8.08101237e-01 -1.08487613e-01
-4.42636609e-01 1.37580836e+00 5.68060577e-01 -4.74695385e-01
5.99442005e-01 -4.49702203e-01 1.46541595e-01 -9.94604547e-03
3.74994904e-01 4.76213753e-01 2.46695876e-01 -6.88232839e-01
-2.33495697e-01 1.38667867e-01 -8.13223794e-02 -3.34770411e-01
1.20529461e+00 -6.71378970e-01 -1.66374482e-02 4.78314906e-01
1.15417564e+00 5.09418845e-01 -1.22190857e+00 -4.31777030e-01
-5.77776253e-01 -4.40619171e-01 4.89135265e-01 -1.06733727e+00
-1.48240805e+00 1.26911986e+00 4.91421819e-01 1.67602479e-01
1.55433989e+00 -1.65127426e-01 1.07796919e+00 4.63278852e-02
2.62810793e-02 -9.05013502e-01 6.92991540e-02 5.74799001e-01
1.32744002e+00 -8.65121067e-01 -5.58828592e-01 -4.32363540e-01
-5.06110668e-01 6.44613326e-01 6.64405882e-01 7.87670463e-02
9.18812752e-01 6.04554296e-01 3.33149105e-01 1.50061563e-01
-8.83401155e-01 -2.98852712e-01 6.18584752e-02 5.58545828e-01
4.21095878e-01 8.78900960e-02 6.49269894e-02 1.09418058e+00
-6.58387363e-01 -7.51965642e-01 1.36119202e-01 3.37251365e-01
-3.21070433e-01 -8.77905667e-01 -3.37215394e-01 3.34850848e-01
-4.93290931e-01 -7.08486915e-01 -1.42346933e-01 5.11688471e-01
3.25532168e-01 1.53374660e+00 2.04047766e-02 -7.78170049e-01
6.84176445e-01 -1.25807999e-02 1.14805311e-01 -5.13177335e-01
-6.70796394e-01 7.73024797e-01 2.69008726e-01 -4.40596282e-01
-8.40507373e-02 -4.95483011e-01 -8.59415531e-01 -1.19500503e-01
-7.98546314e-01 6.83197007e-02 4.52649593e-01 9.33604181e-01
4.25991505e-01 1.27551818e+00 8.50673795e-01 -6.92959249e-01
-9.28206205e-01 -1.62840736e+00 -9.34263825e-01 2.28661373e-01
8.08170438e-01 -4.50382560e-01 -7.25346625e-01 -1.53129250e-01]
|
[14.900410652160645, 5.96978235244751]
|
3e9c78e2-8022-4c76-b044-1870c4b2c5f3
|
data-efficient-french-language-modeling-with
|
2306.01497
| null |
https://arxiv.org/abs/2306.01497v1
|
https://arxiv.org/pdf/2306.01497v1.pdf
|
Data-Efficient French Language Modeling with CamemBERTa
|
Recent advances in NLP have significantly improved the performance of language models on a variety of tasks. While these advances are largely driven by the availability of large amounts of data and computational power, they also benefit from the development of better training methods and architectures. In this paper, we introduce CamemBERTa, a French DeBERTa model that builds upon the DeBERTaV3 architecture and training objective. We evaluate our model's performance on a variety of French downstream tasks and datasets, including question answering, part-of-speech tagging, dependency parsing, named entity recognition, and the FLUE benchmark, and compare against CamemBERT, the state-of-the-art monolingual model for French. Our results show that, given the same amount of training tokens, our model outperforms BERT-based models trained with MLM on most tasks. Furthermore, our new model reaches similar or superior performance on downstream tasks compared to CamemBERT, despite being trained on only 30% of its total number of input tokens. In addition to our experimental results, we also publicly release the weights and code implementation of CamemBERTa, making it the first publicly available DeBERTaV3 model outside of the original paper and the first openly available implementation of a DeBERTaV3 training objective. https://gitlab.inria.fr/almanach/CamemBERTa
|
['Djamé Seddah', 'Benoît Sagot', 'Wissam Antoun']
|
2023-06-02
| null | null | null | null |
['dependency-parsing', 'part-of-speech-tagging']
|
['natural-language-processing', 'natural-language-processing']
|
[-4.62489665e-01 -1.42566571e-02 -2.13309318e-01 -6.38989031e-01
-1.21144605e+00 -1.04248452e+00 6.47474051e-01 2.43944541e-01
-5.70206285e-01 8.69531572e-01 3.70141268e-01 -7.71383405e-01
1.99480817e-01 -5.36829293e-01 -8.11413407e-01 -1.35872766e-01
2.31995389e-01 7.01549709e-01 2.74018198e-01 -3.20199937e-01
-4.27491844e-01 1.82961106e-01 -5.90138376e-01 7.02853918e-01
9.19459522e-01 7.04436600e-01 3.19792420e-01 7.37398326e-01
-3.05941820e-01 9.07073677e-01 -4.22231585e-01 -8.16119313e-01
-3.32516953e-02 -2.86979049e-01 -1.29193115e+00 -5.02037287e-01
3.91271740e-01 1.82263762e-01 -1.30702659e-01 6.30853832e-01
4.03202981e-01 6.38878345e-02 3.37274730e-01 -6.73647523e-01
-9.66762125e-01 1.15263283e+00 -8.55487213e-03 3.41313243e-01
4.21114296e-01 -2.63010953e-02 1.25016510e+00 -1.00349295e+00
7.98742950e-01 1.24006164e+00 9.54910100e-01 6.35874569e-01
-1.27954400e+00 -4.21507865e-01 4.58064616e-01 -1.39054544e-02
-1.05889261e+00 -6.40731871e-01 1.03581443e-01 -4.05881822e-01
1.66637194e+00 -9.24641360e-03 3.70666683e-01 1.11364770e+00
2.05757543e-01 1.17026412e+00 1.15681148e+00 -5.96530378e-01
-1.12705424e-01 2.27459922e-01 4.26215768e-01 7.91694462e-01
-1.56358361e-01 -3.49488519e-02 -3.66833657e-01 -5.45392372e-02
3.17504823e-01 -5.00899076e-01 -2.57907867e-01 8.11001137e-02
-1.25622034e+00 8.43583941e-01 4.02488649e-01 8.01615298e-01
-1.63612127e-01 -4.45690798e-03 3.02917540e-01 3.95128399e-01
6.57383919e-01 5.41666210e-01 -1.35741436e+00 -3.72449964e-01
-7.97845244e-01 1.12770706e-01 1.18465710e+00 9.99602318e-01
6.79071844e-01 -1.23648062e-01 -1.68656215e-01 1.03769815e+00
3.22986037e-01 6.34287596e-01 4.69287485e-01 -6.83409035e-01
8.99025619e-01 5.16347945e-01 8.23149383e-02 -3.02260727e-01
-3.95880282e-01 -2.19489723e-01 -4.07439917e-01 -3.07305366e-01
8.99464369e-01 -4.71699566e-01 -1.03234100e+00 1.87900662e+00
4.72609997e-02 -3.49512666e-01 3.57694089e-01 5.80189884e-01
6.81537449e-01 8.78481388e-01 3.14383358e-01 2.48540908e-01
1.31673360e+00 -1.35939610e+00 -4.84788120e-01 -6.01697385e-01
1.10537136e+00 -1.11187720e+00 9.83873487e-01 2.45843902e-01
-1.17335141e+00 -6.46703422e-01 -5.75011432e-01 -3.39586049e-01
-6.47679210e-01 4.17893320e-01 8.38149369e-01 6.42386138e-01
-1.12695229e+00 6.33988738e-01 -1.01386893e+00 -6.56601787e-01
8.36484134e-02 3.08515668e-01 -5.70061326e-01 -2.79926240e-01
-1.25245416e+00 1.06659198e+00 4.32881624e-01 1.53633669e-01
-7.43457079e-01 -7.43602812e-01 -9.80565608e-01 -6.01074025e-02
1.59303024e-01 -4.10222173e-01 1.73924136e+00 -8.66410553e-01
-1.51492333e+00 9.43933666e-01 -4.13395017e-01 -5.08394182e-01
2.74464071e-01 -7.25280225e-01 -6.32336736e-01 -3.25850964e-01
2.42972270e-01 7.18137681e-01 5.07249273e-02 -7.29957044e-01
-7.40440845e-01 -4.13028710e-02 5.87302633e-02 -1.49194986e-01
-2.03529391e-02 4.04365718e-01 -3.76077503e-01 -4.07985270e-01
-3.87427837e-01 -9.06246781e-01 -1.74552754e-01 -7.98166811e-01
-2.32162043e-01 -4.93007898e-01 4.75575417e-01 -8.54898036e-01
1.18505359e+00 -2.12086248e+00 1.21499099e-01 -4.01689470e-01
-1.36845931e-01 7.94216514e-01 -5.22072554e-01 8.09896529e-01
-1.64110571e-01 3.60056609e-01 -4.27890062e-01 -4.19395715e-01
3.55226435e-02 2.86862016e-01 -1.97714895e-01 1.35160446e-01
4.41735417e-01 1.13690186e+00 -1.06968796e+00 -2.28918999e-01
2.35528037e-01 3.72002751e-01 -6.19084060e-01 2.93104649e-01
-4.65350807e-01 5.83316147e-01 -3.93578380e-01 5.84491253e-01
5.37118912e-01 -5.87892234e-02 2.80210108e-01 1.74356610e-01
-3.94735694e-01 9.26978290e-01 -5.53439021e-01 1.96010625e+00
-8.76487315e-01 3.88400644e-01 2.58703917e-01 -6.89789474e-01
7.09702849e-01 5.94554901e-01 3.82154286e-02 -5.71514189e-01
7.30035976e-02 6.19319260e-01 2.16207281e-01 -2.09702373e-01
3.64084154e-01 -8.07521939e-02 -3.80752653e-01 3.67181569e-01
5.66947579e-01 -1.72613934e-01 5.52092373e-01 2.32920602e-01
1.26189721e+00 5.56875646e-01 5.53907812e-01 -4.30141449e-01
6.52299106e-01 3.00672501e-01 6.15242243e-01 5.92612028e-01
-2.36839324e-01 2.74783790e-01 4.66105670e-01 -4.75295216e-01
-6.72651947e-01 -1.14725304e+00 -3.51347804e-01 1.46715724e+00
-5.33280849e-01 -6.49044394e-01 -8.25030506e-01 -1.29043949e+00
7.12622628e-02 9.29891586e-01 -5.36929965e-01 4.32290405e-01
-1.06184316e+00 -9.11414504e-01 8.41811776e-01 4.88395333e-01
4.46575552e-01 -1.44600153e+00 1.65601566e-01 4.90906417e-01
-2.17295989e-01 -1.21241271e+00 -4.76092964e-01 2.91264057e-01
-6.78742111e-01 -1.06561565e+00 -7.05278397e-01 -1.13111019e+00
1.56769216e-01 -2.02589646e-01 1.69617736e+00 -2.13132575e-01
1.19327940e-01 3.40560704e-01 -6.60338700e-01 -1.80252224e-01
-8.16432297e-01 6.77295208e-01 -2.90398151e-01 -3.94886136e-01
6.66808426e-01 -3.24339569e-01 -1.69732228e-01 -4.20435853e-02
-4.39283967e-01 -1.55263260e-01 6.24886990e-01 8.17973554e-01
2.84001052e-01 -6.62107766e-01 7.77626097e-01 -1.37450457e+00
5.73989928e-01 -6.30739510e-01 -5.99156380e-01 3.36198747e-01
-4.10990596e-01 1.20765537e-01 9.39588249e-01 -7.25201741e-02
-1.25596869e+00 -6.28437102e-02 -8.36290479e-01 1.57868460e-01
-4.97030735e-01 5.55241704e-01 -1.15957879e-01 3.76185775e-01
5.23754597e-01 -1.50184304e-01 -6.31994307e-01 -9.20373201e-01
8.31983447e-01 8.19958210e-01 2.83377647e-01 -6.82590961e-01
2.30118126e-01 -7.62222707e-02 -6.95038140e-01 -6.62205577e-01
-1.18001294e+00 -5.53068697e-01 -9.15458143e-01 3.00133020e-01
9.51580644e-01 -8.75176728e-01 -4.25087482e-01 5.14281690e-01
-1.39027536e+00 -8.53091180e-01 -2.39781737e-01 5.24512351e-01
-2.90713042e-01 1.61280617e-01 -1.26764643e+00 -5.17745137e-01
-2.42187232e-01 -1.05375910e+00 8.97978783e-01 -1.70857966e-01
-2.62432724e-01 -1.38495135e+00 3.89265299e-01 4.24008340e-01
2.98408926e-01 -2.15028048e-01 1.11508882e+00 -9.20588672e-01
-4.13494676e-01 -6.76525459e-02 -1.32172167e-01 6.24347627e-01
9.66674089e-02 -2.62072593e-01 -8.96990359e-01 -2.38945082e-01
-2.69313127e-01 -4.57071990e-01 1.00232124e+00 2.54903048e-01
4.20426011e-01 -1.27419591e-01 -3.07994872e-01 5.69342375e-01
1.25568843e+00 2.57228643e-01 1.22521847e-01 3.25291425e-01
5.13041615e-01 6.09698772e-01 5.48200727e-01 -2.34457105e-01
6.59249187e-01 4.26409304e-01 1.66247785e-01 -8.84140134e-02
-3.89244109e-01 -4.33654457e-01 6.96181536e-01 1.41853440e+00
1.38730183e-01 -5.49698591e-01 -1.11933684e+00 9.23033118e-01
-1.82339072e+00 -4.29979682e-01 -4.10164326e-01 1.92303777e+00
9.68378127e-01 -2.28256360e-01 -9.72115621e-02 -5.14645517e-01
4.84481752e-01 1.94025964e-01 -1.83382139e-01 -8.73204887e-01
-1.91224411e-01 5.51471591e-01 2.63993353e-01 8.81928444e-01
-1.19311333e+00 1.81494594e+00 6.44672966e+00 6.05864942e-01
-8.53055537e-01 6.12849414e-01 4.91220504e-01 2.07837686e-01
-8.45971406e-02 5.74328527e-02 -1.27823365e+00 1.97441712e-01
1.50633371e+00 5.91372289e-02 3.65388006e-01 7.29871750e-01
7.40968511e-02 2.53908355e-02 -1.25332344e+00 3.35637361e-01
-7.25682452e-02 -1.04443276e+00 -2.72329569e-01 -2.17477143e-01
7.50385523e-01 7.73621917e-01 -1.44303635e-01 8.16699922e-01
9.96608555e-01 -9.36919391e-01 7.62175858e-01 7.79066384e-02
5.61501741e-01 -5.73189080e-01 9.73634779e-01 2.64816314e-01
-1.11948061e+00 1.19106527e-02 -4.36365902e-01 -1.53939305e-02
5.80432534e-01 5.97810626e-01 -8.79212856e-01 6.31465733e-01
6.40386939e-01 8.60610306e-01 -5.61966479e-01 6.49506688e-01
-7.00076401e-01 1.27041721e+00 -5.21687828e-02 -7.91794900e-03
6.76137686e-01 -9.06030908e-02 3.14027965e-01 1.88622248e+00
2.15616658e-01 -1.01504549e-01 2.37984106e-01 5.84826112e-01
-3.69005799e-01 5.66414118e-01 -4.26257968e-01 -2.93477178e-01
2.33037308e-01 1.24064589e+00 -2.91264385e-01 -2.68774867e-01
-7.89287984e-01 9.94047463e-01 9.37311232e-01 2.89500058e-01
-5.80980957e-01 -4.12639409e-01 6.49503946e-01 -7.87297040e-02
3.84205073e-01 -4.16779816e-01 1.43264636e-01 -1.30535471e+00
-2.69008815e-01 -8.80586982e-01 5.40089488e-01 -7.03844190e-01
-1.40850282e+00 1.14672816e+00 -1.32499129e-01 -4.57145393e-01
-5.38251102e-01 -1.02325630e+00 -3.41307282e-01 1.14353704e+00
-1.69759750e+00 -1.42956221e+00 3.97694021e-01 5.12179017e-01
6.97257400e-01 -2.91211933e-01 1.24041355e+00 3.19886327e-01
-5.13880968e-01 4.39811587e-01 1.70630783e-01 5.41640162e-01
1.00615847e+00 -1.43236709e+00 9.23491001e-01 9.82728362e-01
5.13146996e-01 5.75975597e-01 1.62372336e-01 -6.22314692e-01
-9.72369432e-01 -1.24325311e+00 1.64831829e+00 -8.03958118e-01
1.03226566e+00 -7.29786932e-01 -7.16387451e-01 1.48798203e+00
6.28910184e-01 -1.67504117e-01 6.08233154e-01 6.85409606e-01
-3.04798305e-01 -2.66365316e-02 -8.03303361e-01 3.59585136e-01
9.40790057e-01 -3.18762362e-01 -8.80937219e-01 5.05975127e-01
8.64362061e-01 -3.11795354e-01 -1.09403980e+00 2.15901926e-01
2.85546362e-01 -9.59971488e-01 5.31099141e-01 -7.70211756e-01
1.48527697e-01 1.35352507e-01 -2.36759901e-01 -1.63140476e+00
-5.28015375e-01 -6.36443198e-01 2.68032432e-01 1.56623662e+00
1.06282449e+00 -8.05728316e-01 3.94306451e-01 1.40015051e-01
-4.95262593e-01 -5.90992153e-01 -8.51002216e-01 -6.72783017e-01
7.02965021e-01 -6.97083175e-01 4.09672409e-01 7.92992055e-01
-1.17448799e-01 7.37013876e-01 -5.25840484e-02 -2.26128530e-02
2.78354883e-02 9.95954722e-02 5.15618265e-01 -9.70361114e-01
-4.80829328e-01 -1.56053320e-01 7.41387531e-02 -1.26456547e+00
4.57499117e-01 -1.48709416e+00 1.82795286e-01 -1.66564596e+00
-1.97601303e-01 -6.96387768e-01 -1.89718097e-01 9.45161104e-01
-1.04036473e-01 1.82565600e-01 4.43727642e-01 5.60687780e-02
-5.24370611e-01 2.40196750e-01 8.13740790e-01 1.34997442e-01
-1.92531332e-01 2.36078233e-01 -6.11571968e-01 7.97874987e-01
8.10809672e-01 -5.68787098e-01 5.49908653e-02 -1.15296936e+00
2.84735262e-01 -8.34378153e-02 -1.25305414e-01 -7.76172757e-01
-1.90489307e-01 2.03998849e-01 2.20710859e-01 -4.01083410e-01
3.68221134e-01 -3.65916103e-01 -4.47746605e-01 3.84333551e-01
-1.38164729e-01 4.38724816e-01 5.42410970e-01 1.17284700e-01
-3.14565867e-01 -3.27261329e-01 7.26808906e-01 -3.65757734e-01
-7.63671160e-01 1.37959957e-01 -6.05282903e-01 4.87290233e-01
5.71067214e-01 4.60263222e-01 -4.54120636e-01 -6.54770508e-02
-9.34448361e-01 1.64541662e-01 1.22885868e-01 6.32720113e-01
-3.77685279e-02 -8.92962873e-01 -9.93885636e-01 3.29157293e-01
-7.88262021e-03 -2.83115804e-01 2.48248130e-03 8.31030726e-01
-5.85639358e-01 1.13869882e+00 2.63627559e-01 -2.89410412e-01
-1.03539860e+00 4.16514546e-01 1.99847206e-01 -8.45021427e-01
-4.84115869e-01 9.93489683e-01 1.57273233e-01 -1.13547504e+00
-2.85725236e-01 -5.17527640e-01 -9.55140144e-02 -1.58317178e-01
3.38059723e-01 1.51419044e-01 2.77347833e-01 -6.81317747e-01
-5.59662402e-01 3.09627831e-01 -2.40276128e-01 -1.89557999e-01
1.28500795e+00 5.63918874e-02 -1.99165151e-01 4.78043735e-01
9.94135439e-01 5.81460238e-01 -8.27141404e-01 -5.71456030e-02
3.05467606e-01 1.73638370e-02 -2.04438075e-01 -1.54127264e+00
-7.97420323e-01 1.04550862e+00 5.59427068e-02 9.61233824e-02
7.72240698e-01 3.47457141e-01 9.97538626e-01 4.55517113e-01
3.60084146e-01 -7.18585491e-01 -5.33492029e-01 1.16813362e+00
6.30128443e-01 -1.11235952e+00 -5.57825387e-01 -4.08392787e-01
-6.66160405e-01 7.30650902e-01 4.43850219e-01 -1.44552812e-01
7.04023480e-01 4.25022125e-01 5.93534946e-01 1.79340363e-01
-9.78066206e-01 -3.40068102e-01 1.29566401e-01 5.94352067e-01
1.14463735e+00 1.84960991e-01 -5.47588825e-01 9.75727439e-01
-4.47320402e-01 -4.80090901e-02 1.89040869e-01 8.31099808e-01
-1.95866019e-01 -1.71412325e+00 1.99883897e-02 8.02743882e-02
-9.76085484e-01 -7.49066591e-01 -4.12092000e-01 1.21533549e+00
2.17890948e-01 1.20777118e+00 -1.27120286e-01 -3.50756161e-02
4.75747228e-01 5.78462839e-01 4.61797744e-01 -1.04091620e+00
-1.10853291e+00 3.49398615e-04 4.69466090e-01 -5.04474699e-01
-2.70460188e-01 -6.69340611e-01 -1.07025230e+00 -2.40599722e-01
-2.00201169e-01 7.13530183e-01 7.04163313e-01 9.35099781e-01
3.91957730e-01 3.06354940e-01 1.26838580e-01 -4.79353398e-01
-3.19753468e-01 -1.34318006e+00 -3.71355444e-01 7.06137195e-02
-3.16334739e-02 -9.71323177e-02 -1.89700335e-01 -3.32762226e-02]
|
[10.597936630249023, 9.878244400024414]
|
ff049a98-3edb-445c-aec8-7ae35e14caf8
|
exploration-of-unranked-items-in-safe-online
|
2305.01202
| null |
https://arxiv.org/abs/2305.01202v1
|
https://arxiv.org/pdf/2305.01202v1.pdf
|
Exploration of Unranked Items in Safe Online Learning to Re-Rank
|
Bandit algorithms for online learning to rank (OLTR) problems often aim to maximize long-term revenue by utilizing user feedback. From a practical point of view, however, such algorithms have a high risk of hurting user experience due to their aggressive exploration. Thus, there has been a rising demand for safe exploration in recent years. One approach to safe exploration is to gradually enhance the quality of an original ranking that is already guaranteed acceptable quality. In this paper, we propose a safe OLTR algorithm that efficiently exchanges one of the items in the current ranking with an item outside the ranking (i.e., an unranked item) to perform exploration. We select an unranked item optimistically to explore based on Kullback-Leibler upper confidence bounds (KL-UCB) and safely re-rank the items including the selected one. Through experiments, we demonstrate that the proposed algorithm improves long-term regret from baselines without any safety violation.
|
['Togashi Riku', 'Kenshi Abe', 'Kaito Ariu', 'Hiroaki Shiino']
|
2023-05-02
| null | null | null | null |
['safe-exploration']
|
['robots']
|
[-1.73529521e-01 1.49252862e-01 -8.56276691e-01 -2.85032272e-01
-1.19155800e+00 -7.05788612e-01 -4.23031934e-02 3.88862222e-01
-5.58354616e-01 1.22641158e+00 2.81243861e-01 -4.78424788e-01
-6.25869930e-01 -5.29843211e-01 -9.66907740e-01 -6.85208082e-01
-3.98331165e-01 4.60890859e-01 -2.28016928e-01 8.85231644e-02
2.12254092e-01 2.62408823e-01 -1.25868142e+00 1.99080914e-01
1.38036287e+00 1.41500413e+00 -4.57481369e-02 8.04702342e-02
6.11262508e-02 5.40852606e-01 -2.85399199e-01 -5.55514455e-01
6.29082561e-01 -3.50825846e-01 -8.13778758e-01 -3.61864299e-01
-1.53645784e-01 -8.37258816e-01 6.52575046e-02 1.22106409e+00
1.11664034e-01 3.78527194e-01 -1.59559119e-02 -1.16970778e+00
-3.00168514e-01 1.03399158e+00 -8.29178572e-01 -6.28940016e-02
2.94633150e-01 -3.60687643e-01 1.64239466e+00 -5.79667628e-01
5.84870219e-01 1.06925976e+00 8.16253647e-02 4.43026632e-01
-1.20407212e+00 -7.70902991e-01 7.54328132e-01 -3.52224819e-02
-1.02390349e+00 -5.76590672e-02 5.79561710e-01 1.43545419e-01
1.27112970e-01 6.86842442e-01 8.26709509e-01 6.78939223e-01
-6.20936565e-02 1.09004974e+00 1.16911197e+00 -3.61390769e-01
6.81010664e-01 3.95821273e-01 1.28964230e-01 3.11868429e-01
5.87017238e-01 4.34855163e-01 -7.95516968e-01 -4.51969743e-01
3.70146662e-01 1.44920841e-01 -1.50980175e-01 -9.81543541e-01
-6.78551018e-01 9.87490654e-01 6.01307333e-01 -2.84168273e-01
-6.47644341e-01 3.17904174e-01 3.87156308e-01 2.16614157e-01
3.83950055e-01 4.16009039e-01 -5.21795094e-01 -4.84972268e-01
-7.46451080e-01 2.87420660e-01 6.23915315e-01 7.31922925e-01
4.51116532e-01 -5.01989424e-01 -4.94077206e-01 7.70492613e-01
2.86448121e-01 2.16260985e-01 1.14183620e-01 -8.31779242e-01
6.71923101e-01 4.47123885e-01 7.46501505e-01 -5.19241631e-01
1.23629630e-01 -8.13660979e-01 -4.08809274e-01 4.11020398e-01
1.13496177e-01 -2.27662489e-01 -5.94213128e-01 1.84876764e+00
2.40035132e-01 -3.00811261e-01 -6.52280226e-02 1.23012841e+00
-1.73763409e-01 5.86603820e-01 1.23069361e-02 -7.22634196e-01
7.79920101e-01 -8.47478747e-01 -6.13062501e-01 -1.96033139e-02
4.39183801e-01 -1.41351566e-01 1.07619596e+00 9.56035018e-01
-1.15337276e+00 9.71732810e-02 -1.10820305e+00 4.81266022e-01
1.40614286e-01 -1.42803237e-01 9.12903786e-01 7.23155677e-01
-5.72889328e-01 7.86609113e-01 -6.67645693e-01 9.57222730e-02
4.76033121e-01 4.73202735e-01 1.03934139e-01 1.12812817e-01
-1.29262221e+00 4.53758329e-01 5.78233421e-01 6.93206117e-02
-1.02885377e+00 -6.55505657e-01 -2.23049909e-01 2.57942349e-01
1.01556575e+00 -3.77116412e-01 1.44431400e+00 -9.79600251e-01
-1.47154903e+00 2.05409676e-01 1.19902588e-01 -7.98954427e-01
8.92354548e-01 -7.60142744e-01 -1.55174404e-01 -2.90566921e-01
-7.74403885e-02 2.69230187e-01 4.55123991e-01 -1.30234337e+00
-1.03522229e+00 -4.16273475e-01 5.23082554e-01 6.19974732e-01
-7.45990515e-01 -1.13829106e-01 -2.80397832e-01 -4.09872472e-01
9.99002382e-02 -1.13872969e+00 -4.11593616e-01 -1.17672354e-01
-4.14473444e-01 5.25385439e-02 2.30725676e-01 -3.67364585e-01
1.60557592e+00 -1.93368685e+00 -1.09361753e-01 6.09718025e-01
-1.47826642e-01 3.03719211e-02 1.32098287e-01 3.18209529e-01
7.79140070e-02 4.42009866e-01 2.83363432e-01 1.23366145e-02
3.82676348e-02 3.86754647e-02 -6.31929874e-01 2.53092855e-01
-6.66045129e-01 6.37857139e-01 -1.21391213e+00 -3.69939879e-02
-1.16404630e-01 -1.95172951e-01 -6.74596846e-01 1.12586312e-01
-3.72914374e-01 2.09132209e-01 -6.96998894e-01 5.83264709e-01
5.57246983e-01 -1.84093162e-01 1.99454710e-01 1.79317862e-01
-1.54546916e-01 3.39762121e-01 -1.31816101e+00 1.32876503e+00
-6.96233273e-01 -6.77804574e-02 1.32663906e-01 -5.15175462e-01
5.75250208e-01 1.03705600e-01 4.97816056e-01 -6.92355394e-01
-1.45928666e-01 4.57117945e-01 -1.79613292e-01 1.77218556e-01
4.36365098e-01 -9.80009958e-02 -1.73128888e-01 6.18007064e-01
-6.15116119e-01 6.26526892e-01 1.42216057e-01 3.66409034e-01
6.47681475e-01 7.22625330e-02 4.16971862e-01 -1.77505881e-01
2.82011181e-01 -3.49384934e-01 7.50561178e-01 1.03931952e+00
-1.11162506e-01 -1.40738199e-02 6.62470818e-01 -3.46373886e-01
-6.44095302e-01 -1.21432614e+00 4.10212018e-02 1.42936146e+00
3.13289046e-01 -2.78225929e-01 -3.49365205e-01 -9.55047309e-01
4.25481141e-01 9.45221245e-01 -4.91012335e-01 -2.30282485e-01
-5.12072369e-02 -3.87688160e-01 -1.28381690e-02 3.99874598e-01
2.59123385e-01 -6.87469065e-01 -9.08930898e-01 2.28340358e-01
-7.55353570e-02 -2.44039536e-01 -7.01667488e-01 3.98107946e-01
-1.05986798e+00 -6.24749541e-01 -6.29158258e-01 9.42650214e-02
6.14459693e-01 2.61577547e-01 7.64688194e-01 -3.32397670e-01
2.23436102e-01 -1.43681258e-01 -4.89205897e-01 -4.54176307e-01
1.15582794e-01 -3.14381905e-02 3.29637408e-01 1.15196936e-01
-9.94377732e-02 -1.13686502e-01 -8.25402021e-01 3.81860524e-01
-5.47379315e-01 -1.08200274e-01 6.42226934e-01 7.69691646e-01
8.65558684e-01 3.13071668e-01 8.09199214e-01 -9.91385460e-01
8.49549353e-01 -4.33617055e-01 -1.08620870e+00 4.45053637e-01
-1.09716249e+00 4.56826746e-01 5.78777134e-01 -3.88679415e-01
-1.19602406e+00 -1.56477720e-01 3.57139468e-01 -3.43005747e-01
5.83426118e-01 8.96932960e-01 -1.63301319e-01 1.75593510e-01
3.49811912e-01 -9.09582525e-02 -4.30774450e-01 -5.27614653e-01
4.83848035e-01 4.87109900e-01 1.95463330e-01 -6.69376910e-01
5.95213592e-01 4.46786076e-01 -1.10484198e-01 -6.28371313e-02
-1.38955903e+00 -4.59999502e-01 7.29006827e-02 -2.23021105e-01
1.58067361e-01 -7.99540460e-01 -9.03652608e-01 -2.74657488e-01
-5.29285908e-01 -2.77096592e-02 -3.21135610e-01 7.54917204e-01
-5.68505168e-01 -1.08852675e-02 -1.51436105e-01 -1.48869205e+00
-5.48297286e-01 -8.38694692e-01 5.00562191e-01 4.10355330e-01
-1.52402058e-01 -5.47065914e-01 -1.65918812e-01 3.10169935e-01
1.17641628e-01 1.89972818e-01 6.68273330e-01 -5.66562712e-01
-8.37687016e-01 -3.03228468e-01 -4.32753079e-02 1.97302610e-01
3.28733288e-02 -3.74912173e-01 -6.02534294e-01 -6.98418260e-01
-2.31657282e-01 -5.08175910e-01 8.46969664e-01 4.60651249e-01
1.45629358e+00 -8.05992424e-01 -4.02650595e-01 4.53548431e-01
1.33112478e+00 4.72069174e-01 3.03711742e-01 5.94259679e-01
1.43606037e-01 3.71564239e-01 1.46205854e+00 1.01218307e+00
-1.64652959e-01 6.58847868e-01 7.42791653e-01 3.46829325e-01
7.57660389e-01 -6.56373143e-01 5.30087948e-01 3.06949355e-02
1.64019793e-01 -3.11143517e-01 -4.00071681e-01 4.08924580e-01
-2.18244672e+00 -7.57522881e-01 4.08883691e-01 3.19281507e+00
9.71883416e-01 6.89715624e-01 3.31260264e-01 4.44409857e-03
5.98632991e-01 -4.79967073e-02 -1.08708775e+00 -6.87092423e-01
2.72576988e-01 -3.94766271e-01 8.35275292e-01 3.88723105e-01
-8.98329973e-01 5.05894601e-01 5.17290115e+00 8.43766987e-01
-8.88218939e-01 -1.17847517e-01 9.76487696e-01 -8.23704779e-01
-6.08916879e-01 3.30379486e-01 -8.62476349e-01 6.02994025e-01
8.60614240e-01 -7.56170630e-01 8.53876531e-01 1.17435002e+00
4.44502473e-01 -3.68208528e-01 -1.20974517e+00 7.37783849e-01
-3.48529100e-01 -1.12918663e+00 -5.69497086e-02 4.01769161e-01
9.40752327e-01 -3.75754446e-01 2.38999888e-01 4.10891712e-01
6.19491518e-01 -8.11168253e-01 8.56478214e-01 3.90626669e-01
7.81510949e-01 -1.45989060e+00 6.49669886e-01 5.76605499e-01
-7.93182373e-01 -5.39818525e-01 -1.63424492e-01 -7.84645975e-03
2.14376882e-01 8.15358758e-01 -6.72163188e-01 5.10710478e-01
7.30822921e-01 6.39923885e-02 -3.34216468e-02 1.43625939e+00
-3.37100238e-01 4.32751626e-01 -2.83940136e-01 -2.50185043e-01
4.85134214e-01 -4.56433594e-01 5.69439888e-01 3.55944276e-01
4.03888851e-01 -4.32551987e-02 3.32338452e-01 6.32321596e-01
-4.36369210e-01 2.78519690e-01 -2.22702175e-01 -1.26557305e-01
6.44489110e-01 1.11055493e+00 -3.75137806e-01 -2.10086584e-01
-3.67676467e-02 7.97660053e-01 3.93671393e-01 1.47045955e-01
-9.61000800e-01 -2.06632510e-01 6.14525616e-01 9.10222605e-02
1.42785192e-01 4.12393779e-01 -4.82900113e-01 -6.87765539e-01
2.87643075e-01 -5.62888384e-01 8.72904241e-01 -4.92739767e-01
-9.72825289e-01 1.49093837e-01 1.12216793e-01 -1.22574115e+00
-2.04219893e-01 3.73650678e-02 -3.98899734e-01 7.35267401e-01
-1.26663578e+00 -5.54280818e-01 9.88497511e-02 -4.19021025e-02
2.23752663e-01 2.76938915e-01 4.08866167e-01 5.89017980e-02
-2.83604294e-01 8.10968518e-01 7.29747951e-01 -5.12914836e-01
6.25163257e-01 -1.31855094e+00 -3.56628418e-01 5.64785779e-01
6.71051145e-02 8.53245497e-01 7.53623903e-01 -8.15304101e-01
-1.24345839e+00 -1.01301253e+00 5.52535474e-01 1.90970168e-01
5.77634335e-01 -3.15101922e-01 -5.38710058e-01 5.35542548e-01
-2.75253534e-01 -2.08930448e-01 5.73544741e-01 6.64988160e-01
-1.90120056e-01 -5.90761065e-01 -1.17514753e+00 9.12342131e-01
9.43527818e-01 -1.63159087e-01 -1.22947767e-01 2.81109750e-01
7.68853724e-01 -3.58692706e-01 -5.12867153e-01 4.97309089e-01
9.96430039e-01 -9.48075950e-01 7.68211424e-01 -6.63556457e-01
6.40475526e-02 -7.20681921e-02 3.91664058e-02 -1.46669352e+00
-9.18990523e-02 -1.08595037e+00 -5.38978934e-01 9.27591741e-01
5.55016875e-01 -5.07394373e-01 1.20698440e+00 8.98974657e-01
2.07972139e-01 -1.21322167e+00 -9.57945347e-01 -1.16328108e+00
-9.43978578e-02 -7.48202056e-02 5.29175043e-01 1.67880416e-01
5.55463910e-01 5.70745356e-02 -9.06083405e-01 -5.24239540e-02
8.12303722e-01 6.18845940e-01 4.07236844e-01 -1.16948652e+00
-4.77984279e-01 -3.43743145e-01 4.04534250e-01 -1.13910186e+00
-2.38822103e-01 -6.37459219e-01 9.57717225e-02 -1.38911343e+00
6.40031219e-01 -7.01101363e-01 -1.08699214e+00 3.78924906e-01
-1.42808318e-01 -2.46622831e-01 1.51402801e-01 6.68489560e-02
-1.19273853e+00 6.17732346e-01 1.09161270e+00 2.77976483e-01
-6.80320978e-01 4.49379712e-01 -1.18040907e+00 5.55420578e-01
8.34655702e-01 -4.51089740e-01 -5.94598174e-01 -4.06388938e-02
7.41403878e-01 4.66107577e-01 -1.19283654e-01 -5.25084972e-01
3.04189678e-02 -5.26646376e-01 1.29070655e-01 -9.27271783e-01
1.34839609e-01 -9.17167485e-01 2.19362497e-01 6.96017444e-01
-9.26983833e-01 -2.44492009e-01 -1.43233389e-01 9.93470073e-01
5.33011854e-02 -2.68462896e-01 8.21573079e-01 1.04887187e-01
-3.28028828e-01 3.57671380e-01 2.14220863e-02 -1.16398498e-01
1.18114161e+00 -1.80558376e-02 -3.51396874e-02 -7.28781641e-01
-5.32902122e-01 7.10148275e-01 4.16042179e-01 4.21574563e-01
4.80009973e-01 -1.34806418e+00 -3.51091802e-01 -2.20738530e-01
2.52881497e-01 -3.37671936e-01 1.12761885e-01 5.57320595e-01
4.33809757e-02 8.30299854e-01 6.24129102e-02 3.26692536e-02
-1.22763026e+00 7.03379750e-01 -9.51952562e-02 -8.39828789e-01
-2.84353882e-01 9.30327415e-01 1.94320530e-01 5.43651469e-02
6.96799040e-01 -2.07684845e-01 9.63012800e-02 1.40533566e-01
4.66506571e-01 6.17714643e-01 2.63622832e-02 5.87978959e-02
-3.81547332e-01 -1.37789860e-01 -5.26159704e-01 -4.14051294e-01
1.12979829e+00 -2.60012090e-01 1.32892877e-01 4.03929383e-01
9.74216640e-01 3.49642962e-01 -1.37857473e+00 -1.12290345e-01
3.41651946e-01 -1.06085896e+00 2.91460186e-01 -1.28596234e+00
-8.13392103e-01 2.39473775e-01 6.14257574e-01 6.08930551e-02
1.00800967e+00 -1.77729458e-01 7.50534058e-01 5.30240357e-01
8.49848688e-01 -1.47178566e+00 9.46788676e-03 1.30300656e-01
8.34988654e-01 -1.03863490e+00 2.75706142e-01 -1.17679924e-01
-9.07076716e-01 6.22822404e-01 5.56470335e-01 1.24614105e-01
3.54076594e-01 -2.89253801e-01 -3.36603761e-01 2.81097323e-01
-9.89100099e-01 -2.83138771e-02 1.71685174e-01 -5.27699329e-02
1.59955397e-01 4.20691013e-01 -6.79627776e-01 8.55433702e-01
8.33775569e-03 9.62136537e-02 4.49536145e-01 7.88790643e-01
-6.20419025e-01 -1.23868072e+00 -3.71329725e-01 9.76475954e-01
-7.78676271e-01 7.15876669e-02 -1.76124454e-01 2.92857915e-01
-3.13556254e-01 8.89814198e-01 -2.03009620e-01 -1.65835366e-01
2.14830130e-01 -1.99042946e-01 2.58860946e-01 -3.89290571e-01
-2.86365926e-01 3.03237051e-01 3.21006924e-01 -7.13196397e-01
3.20576251e-01 -7.00585127e-01 -1.27671778e+00 -1.54812261e-01
-6.14984810e-01 7.27037370e-01 6.15897834e-01 5.31763494e-01
2.79396027e-01 1.81597412e-01 1.04103017e+00 -1.03095651e-01
-1.35413170e+00 -3.89610082e-01 -9.30568635e-01 2.91227162e-01
2.62301147e-01 -8.61964941e-01 -3.40165675e-01 -7.28816390e-01]
|
[4.604611396789551, 3.3223555088043213]
|
16985df8-995e-41f0-82b7-e15da1372e06
|
incorporating-structured-sentences-with-time
|
2304.04717
| null |
https://arxiv.org/abs/2304.04717v1
|
https://arxiv.org/pdf/2304.04717v1.pdf
|
Incorporating Structured Sentences with Time-enhanced BERT for Fully-inductive Temporal Relation Prediction
|
Temporal relation prediction in incomplete temporal knowledge graphs (TKGs) is a popular temporal knowledge graph completion (TKGC) problem in both transductive and inductive settings. Traditional embedding-based TKGC models (TKGE) rely on structured connections and can only handle a fixed set of entities, i.e., the transductive setting. In the inductive setting where test TKGs contain emerging entities, the latest methods are based on symbolic rules or pre-trained language models (PLMs). However, they suffer from being inflexible and not time-specific, respectively. In this work, we extend the fully-inductive setting, where entities in the training and test sets are totally disjoint, into TKGs and take a further step towards a more flexible and time-sensitive temporal relation prediction approach SST-BERT, incorporating Structured Sentences with Time-enhanced BERT. Our model can obtain the entity history and implicitly learn rules in the semantic space by encoding structured sentences, solving the problem of inflexibility. We propose to use a time masking MLM task to pre-train BERT in a corpus rich in temporal tokens specially generated for TKGs, enhancing the time sensitivity of SST-BERT. To compute the probability of occurrence of a target quadruple, we aggregate all its structured sentences from both temporal and semantic perspectives into a score. Experiments on the transductive datasets and newly generated fully-inductive benchmarks show that SST-BERT successfully improves over state-of-the-art baselines.
|
['Yong Dou', 'Zhen Huang', 'Fenglong Su', 'Chengjin Xu', 'Zhongwu Chen']
|
2023-04-10
| null | null | null | null |
['knowledge-graph-completion', 'temporal-knowledge-graph-completion']
|
['knowledge-base', 'knowledge-base']
|
[ 9.38421264e-02 3.98377389e-01 -8.25788856e-01 -3.20915222e-01
-4.11609381e-01 -6.71624124e-01 8.27731729e-01 9.68882665e-02
-2.78216690e-01 8.29490900e-01 4.25922751e-01 -3.64354998e-01
-3.97842735e-01 -1.15403664e+00 -9.19047117e-01 -6.55467510e-01
-5.09603083e-01 7.16287792e-01 4.80795026e-01 -4.71262783e-01
-5.31173527e-01 7.90339336e-02 -1.08720386e+00 2.76423603e-01
9.82694507e-01 8.33835661e-01 -3.41705471e-01 4.32049751e-01
2.63739675e-02 1.30563664e+00 -9.98547450e-02 -6.43149495e-01
-1.02849610e-01 -1.94373146e-01 -1.16775119e+00 -2.73013234e-01
-1.27123475e-01 5.75627945e-02 -9.83763814e-01 4.93859917e-01
7.79739097e-02 4.02056158e-01 4.76823032e-01 -1.39146960e+00
-9.08012033e-01 1.22048664e+00 7.46232877e-03 2.94243664e-01
3.51802945e-01 -9.12331641e-02 1.34130287e+00 -6.95275843e-01
9.10244644e-01 1.23709965e+00 6.94390833e-01 4.23944294e-01
-1.52028179e+00 -4.75324541e-01 6.41711593e-01 7.15100706e-01
-1.22225225e+00 -1.73139483e-01 1.04920173e+00 -3.39015156e-01
1.37499666e+00 2.59086430e-01 8.45733643e-01 1.57202244e+00
-1.59043688e-02 1.02285659e+00 8.90222728e-01 -3.64503890e-01
2.39628375e-01 -1.71433181e-01 3.40227693e-01 6.67727590e-01
-1.40041336e-01 4.25656319e-01 -7.79993176e-01 7.90283307e-02
5.45252979e-01 -9.42727700e-02 -2.36625269e-01 -4.09240246e-01
-1.52995920e+00 6.99094594e-01 5.33055365e-01 5.44153929e-01
-1.59960926e-01 3.11674863e-01 6.32453799e-01 5.37980676e-01
8.71598601e-01 1.56061679e-01 -8.96495461e-01 -4.87184711e-02
-5.47660530e-01 1.59875557e-01 8.52175772e-01 1.01524794e+00
6.76323175e-01 -1.19607002e-01 -5.61026037e-01 5.13481617e-01
4.43332829e-02 1.84751004e-01 4.69917715e-01 -4.57479388e-01
7.68495381e-01 8.11013043e-01 -7.19023570e-02 -7.76980639e-01
-4.15185928e-01 -3.91261041e-01 -6.26592159e-01 -6.86312914e-01
2.44716421e-01 7.72853717e-02 -1.14449310e+00 2.22422218e+00
4.86553222e-01 9.11213994e-01 3.67095411e-01 6.08368576e-01
7.63215661e-01 8.69614482e-01 3.31382960e-01 -5.34288585e-01
1.16872680e+00 -9.52591062e-01 -7.94403195e-01 -3.15116078e-01
1.23933709e+00 1.57382101e-01 9.74533081e-01 3.01923789e-02
-6.88901067e-01 -1.49369970e-01 -9.02375400e-01 -9.99979973e-02
-8.15686643e-01 -8.54715332e-02 1.10905647e+00 9.84228253e-02
-1.01667321e+00 8.32396567e-01 -1.07586801e+00 -3.18709880e-01
1.35374874e-01 4.27209556e-01 -4.51213151e-01 -1.58634961e-01
-2.11695027e+00 8.73568356e-01 9.43830252e-01 2.46334568e-01
-8.99112165e-01 -8.17413986e-01 -1.19591081e+00 -1.64922655e-01
9.17042077e-01 -5.99399209e-01 1.00259542e+00 -5.32201052e-01
-1.37367392e+00 7.86785781e-01 -6.00643270e-02 -7.04192579e-01
2.31409192e-01 -4.45678905e-02 -8.92696202e-01 2.32915375e-02
1.26532927e-01 2.90715307e-01 6.57979429e-01 -8.38341117e-01
-3.73441517e-01 -2.48516724e-01 2.68474966e-01 4.44174148e-02
-3.92006934e-01 -2.62878031e-01 -5.65615416e-01 -7.43736982e-01
-7.35142305e-02 -1.12234294e+00 -9.47575569e-02 -7.12283432e-01
-5.91383040e-01 -7.73206115e-01 6.91901565e-01 -6.40525579e-01
1.56037152e+00 -1.97929907e+00 5.96701622e-01 2.24812776e-02
8.79815668e-02 1.94925606e-01 -1.27053604e-01 5.74393749e-01
-2.93713510e-01 8.34649205e-02 -1.71065003e-01 -4.85740006e-01
2.01429963e-01 7.03813791e-01 -5.86841285e-01 1.53193653e-01
3.48696977e-01 1.45516157e+00 -1.51953447e+00 -7.74008214e-01
1.10951312e-01 2.38131404e-01 -2.81484604e-01 -1.16409816e-01
-6.62322283e-01 3.52833807e-01 -5.96010566e-01 6.22237563e-01
5.62913679e-02 -4.00801748e-01 5.07310808e-01 -2.19112828e-01
1.82751060e-01 7.35226512e-01 -7.44062603e-01 1.95507038e+00
-5.85276127e-01 3.08027744e-01 -7.82610297e-01 -1.19251513e+00
7.15181053e-01 6.76917315e-01 6.30389929e-01 -6.70361936e-01
-1.91830799e-01 1.61483333e-01 -2.92122334e-01 -4.85712767e-01
4.71040159e-01 -3.78554285e-01 -3.83249789e-01 3.45439166e-01
4.39505249e-01 2.28358224e-01 3.20740908e-01 4.97314602e-01
1.52430964e+00 5.79877913e-01 8.76209885e-03 1.66544646e-01
2.79166460e-01 4.89889905e-02 1.01908815e+00 3.08352053e-01
-1.56050503e-01 -2.04974506e-02 6.56378686e-01 -5.04184961e-01
-6.99268699e-01 -1.18621814e+00 1.12588704e-01 1.14126670e+00
2.22382873e-01 -7.23462462e-01 -4.92947362e-03 -1.28016114e+00
-1.46594653e-02 9.32551861e-01 -9.06852245e-01 -7.09849954e-01
-8.28818738e-01 -7.28648782e-01 6.29158080e-01 7.60107815e-01
2.71348864e-01 -1.14224279e+00 1.07627615e-01 5.35135269e-01
-6.05957508e-01 -1.39216292e+00 -3.63942713e-01 4.10376370e-01
-9.25024450e-01 -8.40478361e-01 -1.65698051e-01 -8.69227469e-01
3.37049335e-01 -2.09240064e-01 1.24868429e+00 -4.61788356e-01
1.80005729e-01 3.39991122e-01 -5.71788073e-01 -3.37918252e-02
-1.53071135e-01 1.29167870e-01 1.30596697e-01 2.22572401e-01
4.42847550e-01 -9.95499194e-01 -4.77345198e-01 3.14266682e-01
-8.40457082e-01 3.19718361e-01 2.57820129e-01 8.18384409e-01
6.09219551e-01 2.52093226e-01 8.39862764e-01 -8.92072976e-01
2.68674165e-01 -6.55848861e-01 -3.66360903e-01 6.67008698e-01
-8.38704288e-01 4.54153240e-01 7.05946267e-01 -9.23245609e-01
-1.24547160e+00 -8.80123079e-02 5.00859320e-01 -6.35895252e-01
3.78856212e-01 1.03700149e+00 -1.93892382e-02 3.41299564e-01
7.38260448e-01 2.80605763e-01 -5.96486807e-01 -2.09148005e-01
7.81271398e-01 -1.06760614e-01 5.36370575e-01 -8.61517608e-01
8.96519184e-01 5.09813666e-01 3.27929817e-02 -2.09562823e-01
-1.14282775e+00 -4.90939856e-01 -6.01215839e-01 -1.77678823e-01
6.34066105e-01 -1.01137185e+00 -5.42494714e-01 9.21698660e-02
-1.10030079e+00 -7.58554816e-01 -6.14341676e-01 7.30621159e-01
-6.59184098e-01 2.20546037e-01 -8.09840858e-01 -6.74737513e-01
-2.16581747e-01 -5.40420234e-01 1.10390496e+00 -2.32156441e-01
-9.19932872e-02 -1.58009064e+00 4.32909101e-01 2.10548952e-01
-3.93569544e-02 5.44952214e-01 1.09528136e+00 -6.59737587e-01
-6.44251704e-01 -2.14360729e-01 1.52149454e-01 1.86006390e-02
1.08930357e-01 -2.55613655e-01 -8.41246426e-01 4.47518677e-02
-3.31046194e-01 -5.74136972e-01 1.17761278e+00 1.68875694e-01
6.59049094e-01 -5.00865757e-01 -7.70920873e-01 3.56888741e-01
1.26773322e+00 5.13252430e-02 6.52709067e-01 1.39427826e-01
9.05365169e-01 6.43362463e-01 9.91377056e-01 2.46866599e-01
8.74134064e-01 8.20172668e-01 1.11568592e-01 2.61807323e-01
4.48065251e-02 -6.92510247e-01 6.94869637e-01 1.13919270e+00
-4.00194079e-01 -3.94181103e-01 -9.24530566e-01 1.03182566e+00
-2.45468903e+00 -1.06200254e+00 -3.24863866e-02 2.04977965e+00
1.43437505e+00 1.41902104e-01 -8.75010807e-03 3.01225126e-01
5.39727330e-01 3.98632139e-01 -4.61430848e-01 -1.45100862e-01
-2.61871308e-01 1.73966736e-01 4.28984284e-01 3.93632531e-01
-1.20326281e+00 1.40830350e+00 4.87170315e+00 9.89531457e-01
-8.85109901e-01 1.83498159e-01 2.70451784e-01 -8.49206895e-02
-5.61950088e-01 4.26571250e-01 -6.16848290e-01 2.96354234e-01
1.29874396e+00 -3.42665613e-01 5.08026659e-01 5.20275116e-01
2.91293971e-02 3.00476432e-01 -1.52137518e+00 7.62974024e-01
-2.63346612e-01 -1.25228107e+00 -1.84956521e-01 -1.86739191e-01
6.77745819e-01 -4.67650816e-02 -1.37500197e-01 9.84553993e-01
6.60322070e-01 -8.47704709e-01 7.00662792e-01 6.35100007e-01
8.15570295e-01 -3.31457615e-01 4.97065336e-01 2.58106202e-01
-1.61589408e+00 2.31631193e-02 7.01985806e-02 1.02768958e-01
4.11888540e-01 6.32213295e-01 -1.00044298e+00 1.09444952e+00
5.19706905e-01 1.29950249e+00 -5.15404999e-01 4.58272040e-01
-5.57524800e-01 9.47922587e-01 -4.11543518e-01 3.71990763e-02
2.23906189e-01 -1.67586058e-01 5.59462070e-01 9.99979913e-01
9.21418294e-02 1.68556854e-01 4.08543833e-02 8.09594154e-01
-1.56287998e-01 -1.56258136e-01 -8.37864399e-01 -4.13903058e-01
4.43513393e-01 1.03881514e+00 -5.24845719e-01 -3.68839085e-01
-4.48889613e-01 9.65010464e-01 6.91059172e-01 6.97281420e-01
-1.13455367e+00 2.76566762e-03 2.94916838e-01 -2.29678214e-01
3.46165657e-01 -2.11934805e-01 5.16349003e-02 -1.46573055e+00
3.11574042e-01 -3.50811243e-01 9.45572078e-01 -6.00146592e-01
-1.12402380e+00 3.46649468e-01 3.53227407e-01 -1.17004216e+00
-4.45052087e-01 -3.33216310e-01 -3.91373903e-01 4.29139972e-01
-1.41784513e+00 -1.70392168e+00 1.60872757e-01 8.85427475e-01
2.76226878e-01 3.34919870e-01 7.06924379e-01 3.97468358e-01
-6.32100046e-01 4.14250821e-01 -2.52776027e-01 1.51510000e-01
5.18607914e-01 -1.37633359e+00 3.38210046e-01 9.10488844e-01
4.12847638e-01 4.90570128e-01 4.53515470e-01 -8.33959877e-01
-1.45355308e+00 -1.66701210e+00 1.47268057e+00 -7.01125562e-01
1.20375931e+00 -5.39398491e-01 -9.23748434e-01 1.38622797e+00
-2.78793305e-01 3.84931266e-01 4.26968127e-01 6.54153109e-01
-6.16656184e-01 -1.94503814e-01 -6.01584017e-01 8.08911741e-01
1.67002308e+00 -8.95717323e-01 -8.88509274e-01 6.78716540e-01
1.47913682e+00 -2.05993742e-01 -1.24847519e+00 7.23330021e-01
6.53208643e-02 -3.20707053e-01 1.13964164e+00 -1.00480354e+00
3.23097199e-01 -2.83492565e-01 9.47032869e-02 -1.34633398e+00
-3.17049295e-01 -8.62605453e-01 -8.86598945e-01 1.35319567e+00
5.49456358e-01 -6.49273276e-01 7.16162622e-01 4.72913861e-01
-2.24528298e-01 -7.70965576e-01 -1.13832772e+00 -1.29147851e+00
-1.34455860e-01 -8.69638205e-01 5.30819058e-01 1.30038631e+00
6.57462180e-01 7.11059928e-01 -5.75590849e-01 1.24250479e-01
3.54962558e-01 3.60974729e-01 1.62947014e-01 -1.11872995e+00
-2.83420175e-01 1.18076308e-02 -4.44921553e-01 -6.60709023e-01
6.26401782e-01 -1.22206640e+00 -6.32998720e-02 -1.54282594e+00
-7.42200315e-02 -6.22676551e-01 -5.42648077e-01 9.52013969e-01
-1.72388285e-01 -1.87966466e-01 -4.02384907e-01 6.08763285e-02
-9.67292547e-01 1.03630173e+00 1.21705294e+00 -3.42435449e-01
-4.49239284e-01 -1.60018951e-01 -2.67863840e-01 2.75377631e-01
6.03791535e-01 -4.73569125e-01 -8.75396013e-01 -1.54930353e-01
6.60617828e-01 2.95084298e-01 5.14874876e-01 -4.77238297e-01
3.33318770e-01 -3.16409051e-01 -3.37403923e-01 -5.21554291e-01
4.89935607e-01 -6.87811196e-01 3.35677296e-01 2.33892962e-01
-4.06765997e-01 -4.31961477e-01 1.05913036e-01 1.19145977e+00
-4.45929110e-01 3.69026214e-01 -5.26764570e-03 1.72913715e-01
-1.00980186e+00 7.31640577e-01 7.70816952e-02 2.09911987e-01
1.08459747e+00 4.68357094e-02 -5.35280585e-01 -1.72548100e-01
-1.04551733e+00 5.63421845e-01 -5.91068901e-02 5.90785980e-01
6.13126934e-01 -1.85891092e+00 -4.16061074e-01 -3.90685230e-01
6.47529125e-01 1.41635925e-01 2.31327340e-01 1.30370772e+00
1.61183685e-01 6.74849570e-01 5.92010498e-01 -3.49334061e-01
-6.91062272e-01 1.12120950e+00 2.01596200e-01 -8.84250939e-01
-8.29829991e-01 7.31717706e-01 2.59279460e-01 -5.25657713e-01
1.11724749e-01 -8.16907823e-01 -2.31310979e-01 6.62024394e-02
-1.16997108e-01 1.07202150e-01 4.27497886e-02 -4.43755597e-01
-5.53055227e-01 2.41224200e-01 2.77843159e-02 -2.41371825e-01
1.43205559e+00 1.53234616e-01 -4.86102961e-02 8.41851532e-01
1.18937600e+00 -3.05328339e-01 -9.40517843e-01 -9.33444440e-01
3.42928231e-01 5.64162508e-02 -5.07038310e-02 -6.99002087e-01
-9.52473462e-01 3.95907581e-01 -6.48200810e-02 2.39408016e-01
1.06955409e+00 4.45569485e-01 8.35952222e-01 4.31609690e-01
5.60329258e-01 -1.12270916e+00 9.57402065e-02 6.96771920e-01
8.50410759e-01 -1.11060894e+00 -3.07001621e-01 -6.47620797e-01
-6.78805232e-01 7.49753177e-01 4.75823939e-01 1.77497864e-01
7.48459756e-01 -1.76280200e-01 -5.28818965e-01 -4.43355888e-01
-1.34309280e+00 -4.58626091e-01 4.71687764e-01 5.03896713e-01
1.48162887e-01 4.00013387e-01 -2.99399197e-01 7.73405135e-01
6.71780529e-03 2.22325578e-01 -5.65280020e-02 9.78323936e-01
2.83458352e-01 -1.13522136e+00 1.50476933e-01 3.57609332e-01
8.78060411e-04 -2.78040916e-01 -4.55808610e-01 7.78993964e-01
6.88905865e-02 9.66541588e-01 -3.13126415e-01 -8.72892320e-01
3.05329174e-01 3.38232428e-01 5.05409002e-01 -5.59707224e-01
-3.69639546e-01 -2.99551994e-01 7.43818939e-01 -7.41834521e-01
-6.38050795e-01 -5.57113051e-01 -1.38542998e+00 -1.36903062e-01
-4.12280083e-01 2.73856491e-01 -1.23060504e-02 1.11815345e+00
3.36206615e-01 6.38835013e-01 4.87565815e-01 -3.39498699e-01
-1.93504736e-01 -8.90433490e-01 -6.49394035e-01 6.09137297e-01
2.62394488e-01 -8.89466226e-01 -3.76126975e-01 1.76738873e-01]
|
[8.626049041748047, 8.061673164367676]
|
763ec004-67ca-4b92-9ace-20feeb175575
|
towards-accurate-translation-via-semantically
|
2306.12089
| null |
https://arxiv.org/abs/2306.12089v1
|
https://arxiv.org/pdf/2306.12089v1.pdf
|
Towards Accurate Translation via Semantically Appropriate Application of Lexical Constraints
|
Lexically-constrained NMT (LNMT) aims to incorporate user-provided terminology into translations. Despite its practical advantages, existing work has not evaluated LNMT models under challenging real-world conditions. In this paper, we focus on two important but under-studied issues that lie in the current evaluation process of LNMT studies. The model needs to cope with challenging lexical constraints that are "homographs" or "unseen" during training. To this end, we first design a homograph disambiguation module to differentiate the meanings of homographs. Moreover, we propose PLUMCOT, which integrates contextually rich information about unseen lexical constraints from pre-trained language models and strengthens a copy mechanism of the pointer network via direct supervision of a copying score. We also release HOLLY, an evaluation benchmark for assessing the ability of a model to cope with "homographic" and "unseen" lexical constraints. Experiments on HOLLY and the previous test setup show the effectiveness of our method. The effects of PLUMCOT are shown to be remarkable in "unseen" constraints. Our dataset is available at https://github.com/papago-lab/HOLLY-benchmark
|
['Jaegul Choo', 'Cheonbok Park', 'Hyoung-Gyu Lee', 'Dayeon Ki', 'Koanho Lee', 'Yujin Baek']
|
2023-06-21
| null | null | null | null |
['nmt']
|
['computer-code']
|
[ 2.25704923e-01 1.03284582e-01 -4.14304882e-01 -3.38037372e-01
-7.71495342e-01 -8.12532008e-01 8.19138467e-01 -1.83937147e-01
-5.91978312e-01 7.48759747e-01 2.64581561e-01 -6.35876000e-01
3.84148248e-02 -3.71528953e-01 -7.64579654e-01 -2.67416239e-01
3.17243785e-01 6.91440284e-01 1.01470537e-01 -4.48749840e-01
1.85613021e-01 1.32034600e-01 -1.18840742e+00 3.63669604e-01
9.56572711e-01 2.58458227e-01 4.65489179e-01 2.01311946e-01
-7.56891668e-02 1.39648989e-01 -6.43998086e-01 -8.47712517e-01
3.44656795e-01 -3.30781430e-01 -8.54724169e-01 -2.32185215e-01
7.16099560e-01 3.28182280e-02 3.68665531e-02 1.17475021e+00
5.46521962e-01 -4.86702323e-02 4.31636453e-01 -9.70194817e-01
-7.33424783e-01 1.14093840e+00 -2.13577077e-01 4.36363667e-01
2.70277172e-01 4.93705198e-02 1.28463995e+00 -1.12519658e+00
1.02996564e+00 1.19373643e+00 4.46353942e-01 7.21687257e-01
-1.26479149e+00 -5.86549878e-01 3.40652794e-01 3.25267792e-01
-1.35891247e+00 -5.29818058e-01 5.82932353e-01 -2.37823769e-01
1.27393103e+00 4.56646621e-01 1.95700452e-01 1.51953936e+00
3.07724446e-01 6.19502366e-01 1.17683041e+00 -7.12328196e-01
-1.11992480e-02 2.69771516e-01 1.04175881e-01 4.05433327e-01
2.10396245e-01 2.08420500e-01 -6.71583831e-01 1.58670083e-01
5.21905243e-01 -5.01416147e-01 -4.43974614e-01 -3.12647223e-01
-1.36539304e+00 6.55436635e-01 2.11987078e-01 6.94360137e-01
1.23647086e-01 -1.01578660e-01 3.58128905e-01 4.71580565e-01
3.76017213e-01 7.11466908e-01 -6.08074844e-01 -1.89717978e-01
-9.56181407e-01 -7.48816431e-02 7.95348048e-01 1.30393255e+00
6.11758888e-01 -2.46612430e-01 -1.66897237e-01 9.88254964e-01
8.51697847e-02 4.43080992e-01 5.94907284e-01 -6.26314759e-01
8.57364893e-01 4.97956276e-01 -1.98677599e-01 -7.44682610e-01
-4.30129468e-01 -7.28190243e-01 -5.89764178e-01 -3.24527264e-01
4.37569410e-01 -5.31119108e-02 -9.15314734e-01 2.15288639e+00
2.47357544e-02 -1.32888541e-01 7.50635425e-03 9.49550986e-01
6.06424391e-01 3.88461351e-01 5.06052598e-02 -1.77544579e-01
1.27272248e+00 -1.11107135e+00 -8.41978610e-01 -4.81023520e-01
9.65886891e-01 -1.19535136e+00 1.60557437e+00 2.06854224e-01
-1.27702868e+00 -4.19228524e-01 -1.01852548e+00 -3.05242747e-01
-6.64486229e-01 1.08394183e-01 4.29229677e-01 7.26243019e-01
-1.11163032e+00 4.25600976e-01 -7.13350713e-01 -5.61700642e-01
-4.05522846e-02 3.56231689e-01 -2.88085431e-01 -1.22372337e-01
-1.63548803e+00 1.32732677e+00 5.09084582e-01 1.51003197e-01
-4.93276477e-01 -6.17679954e-01 -8.57951522e-01 -4.78157960e-02
7.01524556e-01 -8.75348866e-01 1.33659875e+00 -8.53177130e-01
-1.11381447e+00 1.01039171e+00 -3.09167981e-01 -2.56151080e-01
5.79444945e-01 -2.09306598e-01 -4.79087651e-01 -1.79167047e-01
2.52392590e-01 7.12612450e-01 4.11404192e-01 -1.16018319e+00
-3.17604005e-01 -2.46576294e-01 1.68243662e-01 4.22792822e-01
-3.48273665e-01 2.21233025e-01 -8.44197631e-01 -9.58218277e-01
1.67296574e-01 -1.07673776e+00 6.50119185e-02 -5.74288487e-01
-6.04885876e-01 -2.33497590e-01 4.53548104e-01 -5.77068627e-01
1.53846467e+00 -1.92387128e+00 3.60052317e-01 1.97665632e-01
-1.91555843e-02 2.16629595e-01 -2.39619642e-01 6.51413083e-01
-6.99244738e-02 5.51654637e-01 -1.83265656e-01 -5.34294128e-01
2.18266681e-01 5.07761538e-01 -1.93102062e-01 2.17189882e-02
1.25598937e-01 1.06573188e+00 -6.78687334e-01 -4.58111525e-01
-2.74339993e-03 2.74113178e-01 -5.36832154e-01 -1.95139393e-01
-4.12434042e-01 3.96700144e-01 -4.72923629e-02 6.62057221e-01
4.92465436e-01 -1.38357401e-01 4.85189080e-01 -1.67587921e-01
-1.32147735e-02 7.03560293e-01 -9.74064529e-01 1.96187794e+00
-5.21768808e-01 3.83039415e-01 -2.29387701e-01 -4.67237681e-01
4.96342063e-01 2.84471661e-01 -1.94173098e-01 -1.01925242e+00
1.93369359e-01 4.87001032e-01 1.51543379e-01 -3.34866792e-01
7.72578776e-01 -1.87924698e-01 -1.26591772e-01 3.53818983e-01
1.25987440e-01 4.79402998e-03 4.75506604e-01 3.22158664e-01
9.24208701e-01 1.39911681e-01 1.53196067e-01 -4.99718219e-01
4.03242260e-01 4.26062196e-03 6.35991752e-01 6.48852408e-01
4.25237007e-02 5.96994579e-01 4.65742499e-01 -2.87941396e-02
-9.36446905e-01 -1.05109513e+00 -2.25279868e-01 1.22260988e+00
2.42270842e-01 -8.39352012e-01 -7.16871202e-01 -8.59365344e-01
-2.71040678e-01 1.04668200e+00 -5.55013597e-01 -1.42222658e-01
-9.46090341e-01 -8.37377012e-01 6.85711861e-01 4.77746606e-01
1.81218684e-01 -1.08009481e+00 -1.94883108e-01 1.01567782e-01
-4.29276139e-01 -1.32046902e+00 -6.85925126e-01 2.80141205e-01
-7.36972153e-01 -8.55738997e-01 -3.74133468e-01 -9.42899466e-01
5.83908916e-01 2.86944564e-02 1.39240503e+00 2.00498760e-01
7.22844601e-02 1.73850469e-02 -4.11404997e-01 -1.24923661e-01
-5.08873224e-01 5.38102329e-01 -1.21900430e-02 -5.09907544e-01
5.02660871e-01 -6.03540242e-01 -3.43881667e-01 6.73258364e-01
-9.92821038e-01 1.17560521e-01 7.05916584e-01 9.40398455e-01
5.78030527e-01 -1.86373115e-01 4.61939156e-01 -1.20307982e+00
6.99850500e-01 -3.73124957e-01 -5.17013013e-01 3.90482455e-01
-7.65506983e-01 1.99919522e-01 5.56849122e-01 -4.54370797e-01
-9.85657215e-01 -3.21874261e-01 -2.05432296e-01 -2.08337933e-01
5.35962544e-02 7.74141192e-01 -4.39164668e-01 2.75907815e-01
5.81504107e-01 -2.98430473e-02 -3.93215656e-01 -7.40764022e-01
6.01666927e-01 3.88368845e-01 5.94141424e-01 -1.08921528e+00
8.66665900e-01 1.48951560e-01 -3.40517521e-01 -4.35266852e-01
-7.86141336e-01 -4.36701775e-01 -7.34034836e-01 2.74034858e-01
5.97515643e-01 -7.68974245e-01 -5.35358526e-02 1.96711689e-01
-1.18952966e+00 -4.29812282e-01 -1.05331644e-01 3.58986586e-01
-3.09964091e-01 5.63681781e-01 -7.11810887e-01 -2.30198264e-01
-1.89357474e-01 -1.18428862e+00 8.76137197e-01 -5.84887564e-02
-5.73047996e-01 -1.27579463e+00 -6.26819115e-03 5.21078110e-01
4.65006322e-01 -2.06758887e-01 1.23734009e+00 -8.09240818e-01
-5.39584696e-01 2.18956918e-01 -1.14436410e-01 3.27394217e-01
-1.12975940e-01 -1.76039845e-01 -9.25408900e-01 -2.77170658e-01
2.98787095e-02 -2.49445081e-01 7.13902354e-01 -8.16446245e-02
6.95578814e-01 -3.57367963e-01 -2.92005420e-01 6.30287409e-01
1.35322785e+00 -3.33624110e-02 5.81622779e-01 5.87002695e-01
6.12722099e-01 6.14027977e-01 4.56229955e-01 -9.01282132e-02
5.07596731e-01 9.70926762e-01 3.42729717e-01 -8.23746994e-02
-4.69607770e-01 -4.36346531e-01 4.56536531e-01 1.31170022e+00
2.67287254e-01 -5.75838387e-01 -1.04566360e+00 5.95974922e-01
-1.78203344e+00 -5.43090582e-01 -1.65191844e-01 2.02923703e+00
1.21748495e+00 5.12413800e-01 -3.02322954e-01 -1.23512991e-01
7.83773899e-01 -3.08742765e-02 -2.92057961e-01 -5.41689694e-01
-5.43190837e-01 3.70507538e-01 3.47706050e-01 8.11929345e-01
-8.20855200e-01 1.51712239e+00 5.58682108e+00 9.89181161e-01
-9.96696949e-01 3.03909987e-01 1.20728940e-01 2.01557614e-02
-5.83707988e-01 3.00472558e-01 -9.51601923e-01 4.71626759e-01
8.26256514e-01 -1.86441213e-01 3.80931318e-01 4.77290958e-01
8.38666707e-02 4.21504350e-03 -1.47725582e+00 5.80267131e-01
1.64236128e-01 -1.00719070e+00 2.77994514e-01 1.07978173e-01
6.73673511e-01 2.78381795e-01 2.00984329e-01 4.63207215e-01
2.84687817e-01 -8.93022776e-01 7.61936426e-01 1.13190569e-01
8.29193234e-01 -5.05675375e-01 7.52468884e-01 4.13938165e-01
-8.48888040e-01 3.19494426e-01 -4.34038579e-01 -1.00251652e-01
1.90797001e-01 4.67222363e-01 -8.00555646e-01 7.30555952e-01
3.96662623e-01 5.33418238e-01 -8.89942408e-01 8.25674891e-01
-7.59627044e-01 6.63159311e-01 -3.86557609e-01 1.98985860e-01
3.24501038e-01 2.03240868e-02 6.05281591e-01 1.42024267e+00
2.87658483e-01 -2.46375054e-01 9.15439874e-02 9.34010208e-01
-2.97128230e-01 3.13820720e-01 -5.66203594e-01 4.01279796e-03
6.09998345e-01 1.18348908e+00 -6.44112051e-01 -2.26112649e-01
-3.51071715e-01 1.01087821e+00 4.65595037e-01 3.49113226e-01
-7.69020259e-01 9.34460461e-02 3.41335267e-01 8.24425742e-02
1.11132473e-01 -2.96462446e-01 -4.35091823e-01 -1.47251487e+00
2.59565502e-01 -1.09698784e+00 4.86676335e-01 -7.23671675e-01
-1.23449278e+00 6.47139251e-01 1.23472139e-01 -9.48833406e-01
-2.26612136e-01 -6.10080838e-01 -4.90286380e-01 9.62573588e-01
-1.61390436e+00 -1.20298064e+00 1.56894848e-01 5.26716411e-01
6.39230013e-01 5.09690791e-02 8.92831326e-01 5.49379110e-01
-7.36810744e-01 9.42900896e-01 -1.53231695e-01 9.95750874e-02
1.01220083e+00 -1.34261894e+00 7.02508748e-01 1.07769871e+00
3.73515218e-01 1.14718509e+00 7.11828709e-01 -7.81167567e-01
-1.13827300e+00 -8.79031956e-01 1.60742593e+00 -7.69683063e-01
9.27224398e-01 -6.41556919e-01 -9.72695947e-01 9.17019427e-01
5.29912531e-01 -5.12372732e-01 6.72203124e-01 4.97001320e-01
-6.06148958e-01 3.76581311e-01 -6.46090567e-01 8.92658710e-01
1.40024078e+00 -4.27510858e-01 -9.52189684e-01 4.02549326e-01
7.63478160e-01 -6.71740055e-01 -7.78083980e-01 5.24018824e-01
2.84555852e-01 -6.66568816e-01 7.14708209e-01 -5.44868886e-01
3.78450185e-01 -1.41158462e-01 -2.71671474e-01 -1.45563710e+00
-1.50076404e-01 -6.83467507e-01 2.98601955e-01 1.27607489e+00
9.21551168e-01 -6.37053490e-01 5.33739269e-01 3.85651708e-01
-4.29783285e-01 -5.92417955e-01 -1.04777694e+00 -1.00511050e+00
4.61076558e-01 -4.43298340e-01 4.43858385e-01 1.31355762e+00
2.87224352e-01 7.82102644e-01 -2.91361153e-01 3.38393599e-02
2.93084025e-01 -5.24031706e-02 5.25440097e-01 -8.79655600e-01
-3.85568857e-01 -5.34807086e-01 -6.64175674e-02 -9.44713354e-01
2.87933022e-01 -1.40411603e+00 -2.26930872e-01 -1.34097111e+00
1.97235480e-01 -4.56715465e-01 -1.55060574e-01 6.50004208e-01
-2.17329025e-01 2.15975925e-01 3.22206795e-01 3.94793838e-01
-5.71155250e-01 3.79595369e-01 1.21328926e+00 3.42830792e-02
-2.93021314e-02 -2.49660835e-01 -7.52480149e-01 6.08398497e-01
1.16381907e+00 -4.50164348e-01 -5.88886499e-01 -1.07314074e+00
5.95964193e-01 -2.40675241e-01 1.66383848e-01 -5.95538855e-01
3.28470200e-01 -2.34772433e-02 -1.07967757e-01 -5.05462229e-01
1.62679940e-01 -7.30569661e-01 5.99939972e-02 1.85098574e-01
-3.04014087e-01 6.11757159e-01 4.02752906e-01 6.72577918e-02
-4.28949855e-02 -2.65886962e-01 4.42915857e-01 -7.89877698e-02
-6.89826250e-01 -1.74851865e-01 -1.17802948e-01 5.22168636e-01
5.26085675e-01 -4.57876660e-02 -6.95416093e-01 -8.06594267e-02
-8.68327498e-01 4.38519657e-01 6.50950015e-01 6.63922548e-01
3.25304955e-01 -1.13755286e+00 -5.10133564e-01 2.47506708e-01
2.54924566e-01 -1.73018649e-01 -7.32852221e-02 9.14785445e-01
-2.71126032e-01 4.83229637e-01 -1.38930172e-01 -4.59805310e-01
-1.12442577e+00 6.27382159e-01 2.50907481e-01 -5.36862850e-01
-4.11167264e-01 7.03479052e-01 8.02047104e-02 -7.04060256e-01
3.58281434e-01 -4.95039999e-01 1.24285750e-01 2.20930669e-03
-1.68022849e-02 1.06472380e-01 4.70719039e-01 -6.50899351e-01
-3.26190352e-01 4.67802942e-01 -4.63815510e-01 -3.35574389e-01
8.60170007e-01 -3.38353336e-01 -1.46282092e-01 3.68105501e-01
9.75571036e-01 3.07191193e-01 -5.51311076e-01 -4.70698267e-01
4.69378859e-01 -1.60267383e-01 -2.77414948e-01 -1.41382873e+00
-7.42773235e-01 9.51879084e-01 1.97267473e-01 -8.59048516e-02
9.03894365e-01 -7.03332126e-02 7.47722208e-01 4.50560302e-01
2.60475397e-01 -1.28696859e+00 -1.21914141e-01 7.56135583e-01
8.85207295e-01 -9.81494606e-01 -3.08076024e-01 -6.66742086e-01
-5.35117805e-01 7.52898872e-01 7.49291778e-01 3.45839262e-01
3.59925330e-01 3.69713694e-01 3.62763196e-01 -2.39146248e-01
-9.12751377e-01 -2.12635830e-01 3.99504662e-01 2.74015129e-01
5.49200296e-01 2.16648802e-02 -8.35633457e-01 5.81654727e-01
-6.79850519e-01 -2.74778098e-01 5.02297997e-01 8.40400100e-01
-2.15000525e-01 -1.66176105e+00 -1.67145073e-01 -4.30487469e-02
-5.24448752e-01 -6.88199520e-01 -7.89761841e-01 1.15458369e+00
1.93380371e-01 8.06229651e-01 -3.57171625e-01 -3.08343261e-01
3.35757166e-01 3.50292921e-01 4.54968363e-01 -9.23546791e-01
-7.33917356e-01 4.01815385e-01 3.37667644e-01 -4.93075162e-01
-2.43601620e-01 -5.06999016e-01 -1.10112953e+00 -2.75102347e-01
-3.21895391e-01 9.67231616e-02 6.46833181e-01 9.38988209e-01
2.67272264e-01 4.30685669e-01 6.45495728e-02 -4.47632164e-01
-6.29561007e-01 -9.52681303e-01 -1.63529351e-01 5.80805480e-01
-1.99469000e-01 -6.03299618e-01 -4.56233501e-01 -4.86263558e-02]
|
[11.281661033630371, 9.958633422851562]
|
94a4ae65-5002-47c2-8b2b-b79b3eb48132
|
auto-encoding-progressive-generative
|
1903.03477
| null |
http://arxiv.org/abs/1903.03477v1
|
http://arxiv.org/pdf/1903.03477v1.pdf
|
Auto-Encoding Progressive Generative Adversarial Networks For 3D Multi Object Scenes
|
3D multi object generative models allow us to synthesize a large range of
novel 3D multi object scenes and also identify objects, shapes, layouts and
their positions. But multi object scenes are difficult to create because of the
dataset being multimodal in nature. The conventional 3D generative adversarial
models are not efficient in generating multi object scenes, they usually tend
to generate either one object or generate fuzzy results of multiple objects.
Auto-encoder models have much scope in feature extraction and representation
learning using the unsupervised paradigm in probabilistic spaces. We try to
make use of this property in our proposed model. In this paper we propose a
novel architecture using 3DConvNets trained with the progressive training
paradigm that has been able to generate realistic high resolution 3D scenes of
rooms, bedrooms, offices etc. with various pieces of furniture and objects. We
make use of the adversarial auto-encoder along with the WGAN-GP loss parameter
in our discriminator loss function. Finally this new approach to multi object
scene generation has also been able to generate more number of objects per
scene.
|
['Pratik Kanani', 'Manan Oza', 'Vedant Singh', 'Himanshu Vaghela']
|
2019-03-08
| null | null | null | null |
['scene-generation']
|
['computer-vision']
|
[ 2.31116757e-01 1.75155714e-01 8.64109874e-01 -2.75133252e-01
-7.29258955e-01 -6.45047665e-01 9.49397445e-01 -2.58915633e-01
1.15724958e-01 1.04027164e+00 7.99347758e-02 3.37370038e-01
-2.23139897e-01 -1.21009362e+00 -9.81219530e-01 -5.96958399e-01
2.02443153e-01 1.10098541e+00 1.24190114e-01 -2.29729474e-01
7.71568250e-03 7.98642099e-01 -1.74395418e+00 3.57517511e-01
5.83713949e-01 4.50662524e-01 5.81053972e-01 1.07549143e+00
-6.65520430e-02 6.67002261e-01 -1.03353751e+00 -4.01300550e-01
4.72205192e-01 -5.44387400e-01 -5.37768722e-01 4.26923692e-01
5.98276734e-01 -1.23619340e-01 -6.12655915e-02 8.11974227e-01
8.46262932e-01 4.05396521e-01 1.33312261e+00 -1.47843337e+00
-9.94389832e-01 5.08335173e-01 -4.56298470e-01 -9.44015458e-02
6.02119684e-01 2.84589417e-02 5.60967922e-01 -7.37244964e-01
8.28563094e-01 1.75584543e+00 5.05809486e-01 6.60637140e-01
-1.38917553e+00 -3.32790047e-01 -4.27706718e-01 -5.26316762e-02
-1.48138344e+00 -9.99540463e-02 9.84801412e-01 -3.94720018e-01
9.79049385e-01 4.89660978e-01 5.64771533e-01 1.48395669e+00
2.15099230e-01 7.84483552e-01 1.40065098e+00 -5.08359075e-01
1.11216186e-02 5.52134216e-01 -5.24870217e-01 5.23227513e-01
8.01500082e-02 4.52647470e-02 1.28095016e-01 9.34253111e-02
1.06854272e+00 1.05242044e-01 1.51472554e-01 -4.09219176e-01
-1.11423934e+00 9.93137002e-01 4.57454622e-01 4.77390081e-01
-4.13440466e-01 4.29952711e-01 1.07537121e-01 -9.12168697e-02
5.05438335e-02 7.46143878e-01 -6.62135482e-02 3.06776106e-01
-7.74092317e-01 8.92652392e-01 5.68150938e-01 1.40595496e+00
4.95891362e-01 3.45097125e-01 -5.26478112e-01 8.78403544e-01
3.16246301e-01 6.80616438e-01 5.33024549e-01 -6.19715393e-01
4.76252586e-01 4.72808868e-01 6.82248846e-02 -8.64605010e-01
-1.71656832e-01 -1.75106511e-01 -9.82981920e-01 6.43261492e-01
-3.85282971e-02 -1.93520576e-01 -1.34240162e+00 1.44442940e+00
4.66129512e-01 -1.37529634e-02 2.71400124e-01 5.00776172e-01
1.11253178e+00 9.26823318e-01 2.41445620e-02 2.14935720e-01
1.20957935e+00 -7.35616863e-01 -5.77244103e-01 2.14633256e-01
-2.26983383e-01 -1.19005418e+00 8.20493937e-01 1.58106044e-01
-1.31481504e+00 -1.01335013e+00 -9.20002818e-01 6.09163158e-02
-8.78255308e-01 -3.93786728e-02 5.65072238e-01 8.23601723e-01
-9.19580221e-01 4.34800535e-01 -4.43132043e-01 -2.82174885e-01
6.53914332e-01 3.17931116e-01 -3.09116900e-01 -4.01162095e-02
-8.34337592e-01 1.10582197e+00 8.26296329e-01 -1.18713476e-01
-1.33432305e+00 -4.04856056e-01 -9.64916706e-01 -9.50801894e-02
-1.33872867e-01 -1.16703773e+00 7.86852121e-01 -7.13692963e-01
-1.40258861e+00 8.56385112e-01 2.91336626e-01 -2.57267416e-01
6.99052989e-01 2.11867783e-02 -3.24313074e-01 -1.39277562e-01
1.79458812e-01 1.10320783e+00 9.74191904e-01 -1.77168810e+00
-4.55279112e-01 -1.52445316e-01 8.09072256e-02 2.74542987e-01
3.75982463e-01 -3.17371130e-01 2.23707676e-01 -9.46938455e-01
2.05800254e-02 -8.36967468e-01 -4.46048081e-01 -5.30155659e-01
-8.97082984e-01 -2.08565965e-02 1.05247474e+00 -1.21642120e-01
2.90537953e-01 -1.76501870e+00 3.92909080e-01 7.56932348e-02
-1.55782416e-01 -4.93024588e-02 -7.19938204e-02 4.64876086e-01
-6.93800598e-02 3.50240439e-01 -1.21317878e-01 -6.45052791e-01
2.69945413e-01 2.70240247e-01 -2.46610701e-01 1.38851088e-02
3.94685179e-01 9.89627540e-01 -6.23504937e-01 -7.17318654e-01
6.50712729e-01 7.95780838e-01 -6.30091071e-01 3.46845150e-01
-4.49093610e-01 4.12758887e-01 -5.32708347e-01 6.12852097e-01
8.57900500e-01 6.64177313e-02 -4.39908266e-01 -1.72385827e-01
3.39129597e-01 -4.61985976e-01 -1.44379199e+00 1.76323509e+00
-5.30287743e-01 4.17918563e-01 -5.59753954e-01 -6.02696180e-01
1.19172430e+00 4.27501291e-01 2.67711669e-01 -1.21606827e-01
1.41591415e-01 7.30754016e-03 -2.65526831e-01 -4.62787807e-01
7.19398558e-01 -5.55097163e-01 -2.64631242e-01 1.42751172e-01
3.31202656e-01 -1.17778242e+00 1.80172637e-01 -1.38818949e-01
7.67699063e-01 5.33353627e-01 1.09919965e-01 4.19721454e-02
2.85189509e-01 -4.99933586e-02 -2.67587304e-02 8.96933436e-01
2.91667163e-01 1.18250716e+00 1.34039357e-01 -3.18535239e-01
-1.67510879e+00 -1.44489276e+00 -1.75069273e-01 3.50004375e-01
-1.30498067e-01 3.64289373e-01 -4.02243882e-01 -5.74976504e-01
-1.19417563e-01 9.66602147e-01 -7.44898498e-01 3.36238183e-02
-5.35109997e-01 -9.39290881e-01 4.83496428e-01 2.54953623e-01
4.89233464e-01 -1.55934680e+00 -6.23638034e-01 3.97470921e-01
1.06139518e-01 -1.04229009e+00 -6.73076883e-02 1.73387229e-01
-8.12598526e-01 -6.99821711e-01 -1.06657159e+00 -8.92352045e-01
7.86663353e-01 -2.71468848e-01 1.21797419e+00 -6.75416291e-01
-6.91827357e-01 2.76801586e-01 -5.69254160e-01 -7.71279931e-01
-9.46711957e-01 -9.57902446e-02 -1.34651944e-01 -1.24469638e-01
-9.30717662e-02 -8.03534150e-01 -1.57844931e-01 2.43299864e-02
-1.23107016e+00 1.37011960e-01 6.37549460e-01 7.86611438e-01
6.17338240e-01 4.52181697e-01 6.65586710e-01 -1.03696001e+00
6.08681083e-01 -4.35421526e-01 -5.03242075e-01 1.75195754e-01
3.84901129e-02 1.69449240e-01 6.43106997e-01 -5.27274430e-01
-1.10041153e+00 2.37104952e-01 -3.13191712e-01 -6.90791011e-01
-6.73381150e-01 -3.15853685e-01 -4.36144143e-01 1.75160840e-01
7.24440157e-01 2.78005034e-01 -4.54459727e-01 -3.31902176e-01
5.50439119e-01 2.73236096e-01 2.91564256e-01 -5.40828764e-01
1.13018560e+00 2.52181768e-01 3.94591957e-01 -6.40469432e-01
-5.04963338e-01 7.26962909e-02 -6.45660341e-01 -1.05254531e-01
1.29918313e+00 -7.41237521e-01 -4.32234943e-01 3.01537335e-01
-1.46839750e+00 -2.16458682e-02 -6.22205138e-01 2.37902224e-01
-9.93114352e-01 -5.32831065e-02 -2.47464493e-01 -1.00593829e+00
-2.11830944e-01 -1.22177470e+00 1.29463470e+00 3.04970413e-01
-5.36681451e-02 -9.32748616e-01 4.77078035e-02 2.78168738e-01
2.98414379e-01 1.25592875e+00 1.06401980e+00 -3.85272741e-01
-8.88831258e-01 -2.69731432e-01 1.11686975e-01 3.93598586e-01
3.22095007e-01 3.21322829e-02 -1.01909447e+00 -2.94537609e-03
-1.15171708e-02 -2.66218662e-01 4.92757231e-01 4.64993864e-01
1.02711213e+00 -4.01357979e-01 -2.33666092e-01 4.44292605e-01
1.95668364e+00 4.02512521e-01 9.96095300e-01 1.38346395e-02
8.34568262e-01 3.51861537e-01 4.27857071e-01 4.50856477e-01
-1.53380055e-02 7.83605754e-01 7.20793188e-01 9.45953652e-02
-4.55047876e-01 -3.71544242e-01 8.22316185e-02 2.96626598e-01
-3.04999620e-01 -8.40995133e-01 -6.14625216e-01 5.33551097e-01
-1.37805629e+00 -1.37826777e+00 -2.35590652e-01 1.87786782e+00
5.90636730e-01 1.96600765e-01 2.62812078e-02 1.66031271e-01
7.27846742e-01 -4.87871505e-02 -6.65278062e-02 -4.50528055e-01
-4.51985836e-01 5.20667017e-01 3.03328335e-01 3.92309159e-01
-1.13885832e+00 8.25497270e-01 6.29707670e+00 9.23604429e-01
-6.68886244e-01 1.47838471e-02 5.00053346e-01 -1.29132509e-01
-4.18763578e-01 -3.70448828e-01 -1.02123475e+00 5.12658179e-01
5.39190710e-01 1.78569660e-01 1.12774283e-01 9.57168996e-01
-9.48860794e-02 -1.06730767e-01 -1.07012522e+00 1.08243084e+00
4.00093824e-01 -1.36568189e+00 5.93427539e-01 -1.23816999e-02
1.21220100e+00 -4.48923141e-01 2.94446677e-01 2.43267134e-01
5.11946380e-01 -1.43457615e+00 9.31765437e-01 9.17263210e-01
6.01905704e-01 -9.61551189e-01 6.96238220e-01 2.23571673e-01
-7.63947248e-01 1.82530195e-01 -5.37455499e-01 3.58095914e-01
4.10380155e-01 2.65143305e-01 -1.40769494e+00 6.37489378e-01
5.21138906e-01 9.51462463e-02 -6.08799279e-01 1.16327965e+00
-4.10534106e-02 -2.62811244e-01 -3.29752028e-01 -3.86938930e-01
3.10381293e-01 -8.06865022e-02 8.62252653e-01 9.98730719e-01
8.04512024e-01 -2.52222091e-01 6.37595132e-02 1.41774678e+00
1.25390127e-01 -7.31568858e-02 -1.14734876e+00 2.00940847e-01
1.05161935e-01 1.15711880e+00 -6.78112090e-01 -3.27488959e-01
1.61169127e-01 1.19322288e+00 -7.67471418e-02 5.99510632e-02
-1.00363731e+00 -5.59061050e-01 2.28023008e-01 2.66876906e-01
5.05524039e-01 -2.27089718e-01 -7.27808401e-02 -7.72001326e-01
-2.32347354e-01 -7.04602599e-01 -1.34983793e-01 -1.30675995e+00
-1.38092232e+00 9.92435753e-01 2.34583542e-01 -1.16474974e+00
-5.63379824e-01 -5.58574975e-01 -4.00219142e-01 9.77217019e-01
-8.48183572e-01 -1.85604608e+00 -3.35592300e-01 6.28446937e-01
8.45695078e-01 -4.89613950e-01 1.02137816e+00 2.95293272e-01
-7.90121108e-02 2.34061360e-01 5.59504703e-02 -1.18439637e-01
4.34888363e-01 -1.60958076e+00 9.18472707e-02 6.91493034e-01
4.44177300e-01 1.86231032e-01 1.03896773e+00 -5.71426630e-01
-9.81987059e-01 -1.34179938e+00 4.23390627e-01 -1.05486155e+00
-1.05312854e-01 -6.20180905e-01 -3.91871125e-01 7.22704411e-01
3.62288594e-01 -3.68760645e-01 5.71825802e-01 -3.51364136e-01
1.07574865e-01 1.39846638e-01 -1.74032283e+00 5.41677654e-01
7.54329443e-01 3.94198485e-02 -7.11000383e-01 6.24093592e-01
7.25325048e-01 -5.69615066e-01 -9.89848852e-01 3.57858151e-01
1.11612454e-01 -1.02691817e+00 1.29263890e+00 -5.18039823e-01
8.36981893e-01 -3.23085397e-01 -3.52089942e-01 -1.36526179e+00
-3.56440127e-01 -3.54795814e-01 1.09613821e-01 1.45955801e+00
4.47283626e-01 -2.49628708e-01 7.02100515e-01 1.97506770e-01
-2.83302277e-01 -4.44789857e-01 -7.04512715e-01 -6.63197815e-01
-5.46090631e-03 -9.78106409e-02 9.34829712e-01 5.78047812e-01
-1.05509937e+00 4.65484470e-01 -5.52467406e-01 1.27898306e-01
8.51049185e-01 2.67293751e-01 9.85331118e-01 -1.05971432e+00
-6.02076650e-01 -3.35922301e-01 -6.77769005e-01 -3.21690619e-01
-6.25270009e-02 -8.47702086e-01 -6.23976514e-02 -1.68276727e+00
2.38163978e-01 -7.89380789e-01 1.91487893e-01 1.90636247e-01
1.68376672e-03 4.92971510e-01 3.08051914e-01 -1.50215849e-01
-2.32469931e-01 6.28752947e-01 1.77098548e+00 -2.59932786e-01
-1.87884241e-01 2.85954148e-01 -4.04069811e-01 4.63088512e-01
6.94141209e-01 -5.58434606e-01 -5.49335659e-01 -3.72969180e-01
-1.15588279e-02 -1.19568460e-01 7.99429893e-01 -1.33893359e+00
-3.85816157e-01 -8.39684829e-02 1.16202736e+00 -9.23202157e-01
8.96585047e-01 -9.65273798e-01 8.76478434e-01 1.93977982e-01
8.19814652e-02 5.06962761e-02 3.40096861e-01 4.33471531e-01
-1.41217440e-01 -5.21269500e-01 1.03142083e+00 -8.11991334e-01
-4.12278831e-01 2.44665176e-01 -5.01165539e-02 -2.00320274e-01
1.36478615e+00 -4.96971905e-01 6.78783804e-02 -6.93139732e-02
-1.05577326e+00 -2.82373577e-01 4.31080669e-01 6.59506679e-01
6.83119774e-01 -1.72960722e+00 -8.88286293e-01 1.31792858e-01
-1.53534979e-01 4.07090992e-01 4.91619378e-01 -9.52730849e-02
-8.92551363e-01 3.36866528e-01 -8.08963239e-01 -5.00339389e-01
-1.10882771e+00 7.54052281e-01 3.72927427e-01 -3.11432123e-01
-3.73883665e-01 8.95045877e-01 1.18663244e-01 -5.50509334e-01
-2.15310037e-01 -2.33410195e-01 -2.25338548e-01 -8.19784924e-02
-1.03981802e-02 2.55252898e-01 -3.54218215e-01 -6.72792673e-01
3.96815687e-02 5.83998382e-01 1.97086856e-01 -3.23629469e-01
1.47151780e+00 2.35123143e-01 1.16706200e-01 4.18962508e-01
1.23019075e+00 9.79641359e-03 -1.01907861e+00 3.25018644e-01
-4.92116690e-01 -5.56502223e-01 -4.28952426e-01 -8.43027234e-01
-7.58573592e-01 7.01331973e-01 8.92171204e-01 2.95166820e-01
9.06661570e-01 6.78741857e-02 4.55557376e-01 1.93710290e-02
4.91249442e-01 -7.18909085e-01 4.21949804e-01 1.78661749e-01
1.37657642e+00 -1.28404832e+00 -6.99561387e-02 -3.01735371e-01
-5.40360272e-01 8.94663274e-01 6.04623079e-01 -5.03268600e-01
4.34540182e-01 2.53802747e-01 -2.90391266e-01 -3.14338893e-01
-3.95351857e-01 -2.60178953e-01 2.71288186e-01 1.14363444e+00
2.55673081e-01 2.74314940e-01 2.30204806e-01 1.79365382e-01
-5.72215319e-01 -2.96178848e-01 6.94100738e-01 7.02120006e-01
-1.33498535e-01 -1.33238709e+00 -6.14527225e-01 3.37586612e-01
-5.25966287e-01 4.21235859e-02 2.03949288e-02 9.24776137e-01
6.56177878e-01 5.94831288e-01 7.43541354e-03 -1.33064210e-01
5.45899212e-01 -5.82684902e-03 1.02446055e+00 -7.75393784e-01
-5.35214722e-01 -5.75071946e-02 -1.61425009e-01 1.27702624e-01
-4.59291935e-01 -7.45665669e-01 -8.46866071e-01 1.91675685e-02
-2.35072806e-01 -1.38540238e-01 8.40383112e-01 3.94652963e-01
1.85606301e-01 1.00956547e+00 6.49757624e-01 -1.03295660e+00
-1.89144060e-01 -1.26219177e+00 -7.53538787e-01 5.81168711e-01
5.93647361e-02 -8.33884120e-01 2.90993731e-02 2.89321065e-01]
|
[11.664518356323242, -0.36715275049209595]
|
1b95587b-b83f-44c4-a386-dc29a104f58f
|
reinforcement-learning-with-demonstrations
|
2212.01509
| null |
https://arxiv.org/abs/2212.01509v2
|
https://arxiv.org/pdf/2212.01509v2.pdf
|
Reinforcement learning with Demonstrations from Mismatched Task under Sparse Reward
|
Reinforcement learning often suffer from the sparse reward issue in real-world robotics problems. Learning from demonstration (LfD) is an effective way to eliminate this problem, which leverages collected expert data to aid online learning. Prior works often assume that the learning agent and the expert aim to accomplish the same task, which requires collecting new data for every new task. In this paper, we consider the case where the target task is mismatched from but similar with that of the expert. Such setting can be challenging and we found existing LfD methods can not effectively guide learning in mismatched new tasks with sparse rewards. We propose conservative reward shaping from demonstration (CRSfD), which shapes the sparse rewards using estimated expert value function. To accelerate learning processes, CRSfD guides the agent to conservatively explore around demonstrations. Experimental results of robot manipulation tasks show that our approach outperforms baseline LfD methods when transferring demonstrations collected in a single task to other different but similar tasks.
|
['Jianyu Chen', 'Chengming Shi', 'Zheng Wu', 'Jingyue Gao', 'Yanjiang Guo']
|
2022-12-03
| null | null | null | null |
['robot-manipulation']
|
['robots']
|
[ 5.47184199e-02 9.07786638e-02 -1.99614525e-01 -2.54727036e-01
-6.55332446e-01 -5.47190666e-01 5.21365225e-01 -2.29315124e-02
-7.84668028e-01 1.20245159e+00 1.07454829e-01 -4.70230021e-02
-1.66493967e-01 -2.78370023e-01 -9.74002957e-01 -6.11763895e-01
-3.44331115e-01 6.09190643e-01 3.19735974e-01 -2.09157541e-01
5.42809546e-01 3.99691880e-01 -1.44594073e+00 -7.72985518e-02
1.19617522e+00 6.80979788e-01 1.01639915e+00 5.50287724e-01
1.56379953e-01 8.70331585e-01 -7.01661229e-01 3.39816689e-01
8.14236701e-01 -2.81119704e-01 -7.41654634e-01 -1.53525397e-02
9.52493697e-02 -1.02361155e+00 -5.21241009e-01 9.94840741e-01
4.21334594e-01 6.50924027e-01 4.62229103e-01 -1.63748467e+00
-4.89062160e-01 7.61695683e-01 -5.11162996e-01 7.64751509e-02
4.37426358e-01 6.22176290e-01 6.29211247e-01 -6.72646105e-01
6.30565286e-01 1.51175082e+00 3.31189513e-01 7.60775387e-01
-1.00726449e+00 -6.75636649e-01 5.65118194e-01 2.03389093e-01
-5.68085730e-01 -2.04059705e-01 4.83061641e-01 -4.48426902e-01
9.05818582e-01 -3.95355314e-01 8.11287582e-01 1.21925998e+00
1.49319142e-01 1.08848238e+00 1.35785854e+00 -1.28205895e-01
5.65179706e-01 -9.72620770e-02 -2.02265322e-01 5.16032517e-01
1.30453467e-01 7.13179231e-01 -6.25094891e-01 -1.29173938e-02
8.54806840e-01 3.04657459e-01 -2.51441717e-01 -9.62698042e-01
-1.53171718e+00 7.14620531e-01 5.19433141e-01 -1.46257982e-01
-7.99501538e-01 2.93227524e-01 4.77220029e-01 9.63274777e-01
-3.21760416e-01 9.04682338e-01 -4.91919339e-01 -5.46791971e-01
-2.71327823e-01 7.40585089e-01 8.50523949e-01 1.28250289e+00
8.81459892e-01 1.31135136e-01 -3.70000392e-01 6.38810813e-01
9.91803259e-02 3.92900944e-01 5.57162583e-01 -1.59379947e+00
7.74117351e-01 3.63756269e-01 7.05916703e-01 -3.23354542e-01
-1.91678017e-01 -4.12754565e-02 -4.93425727e-02 9.59594727e-01
4.95193481e-01 -3.60523432e-01 -9.31695104e-01 1.60163641e+00
6.26003325e-01 -2.11123936e-03 3.21585715e-01 1.25308144e+00
2.85369486e-01 3.35242808e-01 -1.37507141e-01 -9.72507149e-02
5.30298650e-01 -1.41013086e+00 -6.10894263e-01 -4.61759090e-01
5.72961807e-01 -3.57773572e-01 1.38281107e+00 5.38870811e-01
-9.19224679e-01 -6.15051448e-01 -9.01103556e-01 8.55372921e-02
7.54282298e-03 -2.01146081e-02 6.52440071e-01 -1.32568151e-01
-8.21049631e-01 9.77836251e-01 -9.04731452e-01 -2.56745726e-01
5.88804483e-01 3.28549355e-01 -3.11714709e-01 -3.79659534e-01
-7.44128048e-01 1.18727803e+00 3.88163984e-01 -1.82706360e-02
-1.93498147e+00 -5.99961817e-01 -8.38653743e-01 -8.44848230e-02
9.45358992e-01 -4.01552498e-01 1.91284513e+00 -7.41587937e-01
-1.93198740e+00 2.15180576e-01 3.50745499e-01 -5.69047034e-01
5.38665295e-01 -6.69364870e-01 4.42666322e-01 -9.70533937e-02
4.86888975e-01 8.71303439e-01 1.15146911e+00 -1.37732279e+00
-9.26127255e-01 -2.14530021e-01 4.23565328e-01 7.73959458e-01
-2.48457789e-02 -5.36127388e-01 1.09668784e-01 -2.72608370e-01
-2.22518340e-01 -1.08704495e+00 -3.13920051e-01 1.93896934e-01
5.39632253e-02 -5.55909932e-01 9.55208480e-01 -1.60313651e-01
4.40121651e-01 -2.13169050e+00 4.77316141e-01 -1.19268231e-01
2.52312183e-01 1.62033975e-01 -4.79367912e-01 5.87988377e-01
2.72887945e-01 -6.29134119e-01 -2.30929051e-02 -2.49728277e-01
2.63584703e-01 7.26450741e-01 -5.85072160e-01 2.68340707e-01
2.53432095e-02 8.69933903e-01 -1.64172208e+00 -1.50449723e-01
1.16697073e-01 -9.24300477e-02 -6.47597730e-01 7.09024608e-01
-4.00045216e-01 8.24581802e-01 -7.36994684e-01 5.57398379e-01
4.07893777e-01 3.56480777e-02 1.01493679e-01 2.75586456e-01
-2.45542511e-01 3.22863817e-01 -1.00593019e+00 2.15641284e+00
-4.37411666e-01 3.41486216e-01 3.58981013e-01 -1.19029045e+00
8.81780446e-01 4.07573022e-02 4.87848371e-01 -6.12373650e-01
-8.71852711e-02 3.30003619e-01 4.46950734e-01 -7.70161152e-01
5.09539068e-01 1.31493688e-01 4.15192172e-02 5.38943529e-01
2.30344236e-01 -7.26353049e-01 3.09563935e-01 1.86420932e-01
1.45835149e+00 7.91273475e-01 3.45748514e-01 1.06057175e-01
-1.62305534e-01 3.44623268e-01 6.13246918e-01 1.22111821e+00
-7.08149314e-01 -5.83032072e-02 3.02291751e-01 -2.98152894e-01
-8.96222234e-01 -1.14365458e+00 5.17931044e-01 1.25363302e+00
3.16013187e-01 -4.99790721e-02 -3.13274086e-01 -1.09792209e+00
4.52002615e-01 5.92448771e-01 -4.70569700e-01 -2.09819973e-01
-7.11601794e-01 9.33828801e-02 -2.38250848e-02 5.05191207e-01
4.17354316e-01 -1.46741712e+00 -1.32602882e+00 3.89623553e-01
1.00571185e-01 -7.85128236e-01 -6.30550623e-01 6.48036242e-01
-9.17201817e-01 -1.13830328e+00 -9.78469670e-01 -9.22758818e-01
7.84740269e-01 6.06322348e-01 7.33210921e-01 -2.05740243e-01
-1.80638596e-01 7.61954904e-01 -4.82137471e-01 -4.69303071e-01
-4.15038437e-01 -1.55294478e-01 3.93154502e-01 -7.09357023e-01
9.14406106e-02 -6.50866687e-01 -5.86926043e-01 2.15040222e-01
-4.67300445e-01 -2.11513221e-01 9.14646626e-01 1.21495187e+00
3.40613455e-01 -2.05875531e-01 8.52265656e-01 -4.29523051e-01
1.04012513e+00 -5.53154171e-01 -7.81205654e-01 -1.29645407e-01
-7.63887227e-01 2.56281376e-01 7.53533840e-01 -9.19949889e-01
-9.94883895e-01 3.52781564e-01 5.08830011e-01 -6.82163179e-01
-1.64761633e-01 4.23844218e-01 3.48031580e-01 -5.91109656e-02
6.85299277e-01 9.48926210e-02 2.43843883e-01 -3.11624616e-01
3.21717322e-01 3.86627167e-01 4.70790833e-01 -9.20888424e-01
7.95852125e-01 1.46328777e-01 -2.04592407e-01 -7.02868327e-02
-5.53384304e-01 -4.03992832e-01 -4.38310295e-01 -2.89085776e-01
3.09793860e-01 -9.16282296e-01 -1.07996452e+00 3.26499790e-01
-1.00505507e+00 -1.17312860e+00 -7.79490769e-01 8.68383348e-01
-1.03545511e+00 1.99357957e-01 -3.92437756e-01 -9.14332926e-01
2.33076572e-01 -1.32822669e+00 9.55882072e-01 4.13893938e-01
-4.36219536e-02 -5.46062350e-01 1.56106904e-01 -2.84115374e-02
3.99436444e-01 -4.15250771e-02 4.47372198e-01 -4.66322452e-01
-6.77126527e-01 2.76169091e-01 -8.87419060e-02 2.19503045e-01
3.64866763e-01 -7.17296898e-01 -4.88809943e-01 -7.10700095e-01
1.46731779e-01 -1.17720628e+00 6.43463612e-01 2.03128487e-01
8.12871277e-01 -2.69008011e-01 -2.91839838e-01 9.93056446e-02
9.36203420e-01 4.47434723e-01 2.95404822e-01 5.36775231e-01
4.30852503e-01 5.53735137e-01 1.53786790e+00 7.23086298e-01
4.22749102e-01 4.77031618e-01 7.63862014e-01 3.84167641e-01
-1.35925606e-01 -4.37218606e-01 7.78501511e-01 4.75958824e-01
9.13311094e-02 1.56343907e-01 -4.94488358e-01 7.16369927e-01
-2.36793661e+00 -8.67754042e-01 4.60947424e-01 1.97615814e+00
1.06642711e+00 7.69548118e-02 2.06118584e-01 -3.08064431e-01
2.74752855e-01 -2.19307110e-01 -1.33114517e+00 -1.06339216e-01
5.02484381e-01 -4.72687930e-02 2.66512930e-01 5.76477587e-01
-6.16202950e-01 9.85592663e-01 6.14311218e+00 4.28600907e-01
-8.89252782e-01 3.13997385e-03 -2.46946722e-01 -3.58112305e-01
2.24627435e-01 1.49324387e-01 -5.98562300e-01 4.04532552e-01
3.25204909e-01 -3.35556895e-01 9.54991877e-01 1.26622295e+00
2.61537313e-01 -6.73268020e-01 -1.70147300e+00 9.07625496e-01
-2.17400700e-01 -6.50692105e-01 -5.16205966e-01 -2.73277536e-02
7.87669957e-01 2.54404455e-01 1.33614406e-01 1.00095165e+00
1.08089459e+00 -7.39587009e-01 5.50797045e-01 4.58416969e-01
4.60395306e-01 -2.80541807e-01 3.68200451e-01 7.90943623e-01
-8.96709561e-01 -5.76747954e-01 -4.22991544e-01 -5.13379872e-01
8.82695094e-02 -5.02204150e-03 -1.53038752e+00 1.57210067e-01
7.70680726e-01 8.96606743e-01 -8.35985970e-03 1.01257932e+00
-3.96346331e-01 -2.54423311e-03 -1.55188411e-01 -2.76387215e-01
4.98849183e-01 -2.72022277e-01 6.60108030e-01 3.75570625e-01
4.44662213e-01 -1.32647485e-01 8.26632440e-01 9.43760931e-01
1.85825348e-01 -3.81591290e-01 -1.06156945e+00 -4.25299965e-02
7.31120408e-01 1.01489198e+00 -2.69592047e-01 -2.83818811e-01
-1.06151201e-01 1.15490377e+00 8.92000973e-01 4.56001133e-01
-7.09384143e-01 -3.85149509e-01 5.54193020e-01 -3.15212667e-01
5.01350820e-01 -4.87139732e-01 4.21833128e-01 -8.67550492e-01
2.14002594e-01 -1.12112474e+00 3.22386920e-02 -9.32904363e-01
-1.34731174e+00 7.53160566e-02 4.77364808e-02 -1.52894032e+00
-4.78592336e-01 -4.90944266e-01 -3.91609967e-01 6.94548011e-01
-1.91223407e+00 -5.53543985e-01 -4.56598163e-01 6.55611038e-01
1.00011206e+00 -4.01836842e-01 6.26216471e-01 -1.08175248e-01
-1.45130694e-01 1.61477253e-01 2.13840693e-01 -3.21843147e-01
1.06224501e+00 -1.35123539e+00 -4.74093407e-02 3.22445810e-01
-2.84871548e-01 5.12855470e-01 6.78627372e-01 -8.25506449e-01
-1.67326188e+00 -6.97150350e-01 -3.95233557e-02 -2.31429920e-01
7.14276910e-01 -1.26730263e-01 -7.91163504e-01 6.59681559e-01
1.06041111e-01 3.87422964e-02 9.68337804e-02 -1.41187623e-01
-8.63498971e-02 2.20360700e-02 -1.12864721e+00 7.71754503e-01
1.21079648e+00 -2.56003052e-01 -9.25308704e-01 4.76344496e-01
8.27461779e-01 -7.21917570e-01 -7.26217091e-01 2.16548905e-01
5.08781612e-01 -7.65474200e-01 7.24364817e-01 -8.04100931e-01
3.85803729e-01 -3.64907056e-01 -5.28293720e-04 -2.18688202e+00
-1.54833585e-01 -8.14720571e-01 -5.41100740e-01 4.72683668e-01
5.72035760e-02 -6.79880083e-01 6.03677750e-01 4.28822398e-01
-5.67064226e-01 -6.34286523e-01 -7.61792123e-01 -1.28990912e+00
-3.83375771e-02 1.72516659e-01 4.01789576e-01 7.29317307e-01
4.64324832e-01 2.71966219e-01 -4.83245313e-01 -5.69861867e-02
6.65605962e-01 2.65459448e-01 1.26131094e+00 -1.05902016e+00
-4.84836221e-01 -1.99468210e-01 1.58722371e-01 -1.65927720e+00
4.70236242e-01 -7.33336985e-01 7.70456016e-01 -1.51555145e+00
1.08430386e-01 -8.04475129e-01 -1.18222624e-01 5.75595915e-01
-1.87191218e-01 -6.72008038e-01 4.77074325e-01 3.21652740e-01
-8.32108378e-01 9.95238006e-01 1.92402470e+00 -1.25377983e-01
-4.38603401e-01 7.80340657e-02 -3.57841849e-01 5.87251186e-01
9.14725661e-01 -5.86649656e-01 -7.73485720e-01 -5.87830365e-01
-1.05205089e-01 2.02943534e-01 3.26855808e-01 -9.25935566e-01
4.01949644e-01 -5.47360897e-01 1.19253136e-01 -3.52018148e-01
4.86329854e-01 -1.05014408e+00 -5.19105256e-01 8.77386332e-01
-5.74839056e-01 4.36767191e-02 4.02676836e-02 9.90733743e-01
5.17525300e-02 -4.63196307e-01 3.40437204e-01 -3.37282389e-01
-9.03481364e-01 2.42121100e-01 -4.68539596e-01 1.12938322e-01
1.28283203e+00 -1.20557070e-01 -4.45968628e-01 -5.39605379e-01
-6.58783138e-01 9.22868907e-01 4.01725799e-01 4.34682220e-01
9.32126880e-01 -1.29874241e+00 -4.38962489e-01 -5.34853414e-02
3.01899258e-02 3.22132885e-01 -1.01542450e-01 8.10027301e-01
1.54898494e-01 8.03160965e-02 -6.42313004e-01 -4.95102316e-01
-9.49699700e-01 7.22303510e-01 7.17279539e-02 -3.22454661e-01
-8.07204247e-01 6.88343048e-01 1.86152995e-01 -8.33299220e-01
7.77743816e-01 -6.32022858e-01 -2.34430246e-02 -3.07142258e-01
2.25520670e-01 5.37958622e-01 -3.88458222e-01 2.58238971e-01
9.14371088e-02 1.09016985e-01 -4.82658356e-01 -2.28094563e-01
1.44343138e+00 -1.62797347e-01 3.85348439e-01 7.29513407e-01
6.63593233e-01 -4.32822585e-01 -2.28716826e+00 -4.45454627e-01
-8.09679478e-02 -8.56325150e-01 -3.44820231e-01 -9.71013725e-01
-5.75415492e-01 7.01809525e-01 5.06247580e-01 -2.82302815e-02
6.75802290e-01 -1.03600234e-01 6.76244438e-01 1.19943655e+00
9.61618721e-01 -1.58701360e+00 9.83695149e-01 8.72632563e-01
1.19167948e+00 -1.62095606e+00 5.83421215e-02 9.33362320e-02
-9.91153061e-01 9.72757757e-01 1.26061821e+00 -4.28896964e-01
2.01710224e-01 8.25720280e-02 -1.89117163e-01 -5.78495525e-02
-8.20359707e-01 -2.10381806e-01 -3.26207817e-01 1.11724222e+00
-2.40301594e-01 -9.36574116e-02 1.33749530e-01 2.53493994e-01
3.77556309e-03 2.13929325e-01 7.09729970e-01 1.57943714e+00
-7.28638768e-01 -9.73645508e-01 -3.63987148e-01 5.21760345e-01
2.82030940e-01 1.74289957e-01 -2.43779331e-01 7.31980145e-01
-1.67755291e-01 7.18990564e-01 -1.12387277e-01 -1.38656914e-01
4.01834875e-01 -1.90746799e-01 9.26937461e-01 -9.63353395e-01
-3.88945401e-01 -5.14557622e-02 -1.72538802e-01 -8.57625186e-01
-3.83640379e-01 -7.36320913e-01 -1.71225214e+00 1.27643093e-01
-1.51183918e-01 7.16121495e-02 5.72398543e-01 7.89451301e-01
3.54824275e-01 6.96562707e-01 8.37452888e-01 -1.30800843e+00
-1.48489797e+00 -1.07185006e+00 -6.35428250e-01 2.74406165e-01
7.85977721e-01 -1.31919289e+00 -3.60956758e-01 -1.50856853e-01]
|
[4.29988431930542, 1.3532452583312988]
|
7a10f4bb-5dc4-4643-8f3e-febb18681f34
|
incremental-generalized-category-discovery
|
2304.14310
| null |
https://arxiv.org/abs/2304.14310v1
|
https://arxiv.org/pdf/2304.14310v1.pdf
|
Incremental Generalized Category Discovery
|
We explore the problem of Incremental Generalized Category Discovery (IGCD). This is a challenging category incremental learning setting where the goal is to develop models that can correctly categorize images from previously seen categories, in addition to discovering novel ones. Learning is performed over a series of time steps where the model obtains new labeled and unlabeled data, and discards old data, at each iteration. The difficulty of the problem is compounded in our generalized setting as the unlabeled data can contain images from categories that may or may not have been observed before. We present a new method for IGCD which combines non-parametric categorization with efficient image sampling to mitigate catastrophic forgetting. To quantify performance, we propose a new benchmark dataset named iNatIGCD that is motivated by a real-world fine-grained visual categorization task. In our experiments we outperform existing related methods
|
['Oisin Mac Aodha', 'Bingchen Zhao']
|
2023-04-27
| null | null | null | null |
['fine-grained-visual-categorization', 'incremental-learning']
|
['computer-vision', 'methodology']
|
[ 5.12187064e-01 -9.15466398e-02 -2.04789504e-01 -3.99387956e-01
-4.70532328e-01 -6.45714760e-01 6.30833745e-01 4.05582011e-01
-4.01103675e-01 6.32105768e-01 -2.01179951e-01 -2.14625418e-01
-1.72818631e-01 -5.44456422e-01 -7.63471901e-01 -6.48337066e-01
-2.86004126e-01 5.98647296e-01 4.09085959e-01 4.80337948e-01
4.23086971e-01 2.81444430e-01 -2.03497720e+00 3.13568503e-01
8.69903564e-01 9.67680633e-01 3.33446980e-01 5.82208276e-01
-7.69420639e-02 7.95592189e-01 -2.40848392e-01 -1.42075494e-01
1.87050954e-01 -2.19894186e-01 -1.01917768e+00 6.77998602e-01
7.21520662e-01 -2.86460310e-01 -6.29111603e-02 1.24012947e+00
4.58602421e-02 3.42144817e-01 9.28405225e-01 -1.50194955e+00
-8.55532467e-01 5.26457965e-01 -5.70260227e-01 2.03910157e-01
-2.22139373e-01 1.06715351e-01 7.33695388e-01 -1.09539580e+00
5.60309589e-01 1.39784682e+00 5.22556961e-01 7.48919547e-01
-1.58616841e+00 -6.11019850e-01 8.37813199e-01 4.64530975e-01
-1.43052483e+00 -2.58781463e-01 5.50821602e-01 -7.30312943e-01
3.70849282e-01 -1.89903274e-01 1.58926696e-01 8.68504941e-01
-9.04798433e-02 7.57208228e-01 1.28651083e+00 -5.53741038e-01
7.19372332e-01 1.96372211e-01 8.58862638e-01 5.94408631e-01
4.18466628e-01 5.53640500e-02 -3.42473775e-01 -1.43599525e-01
2.80490488e-01 4.77185309e-01 8.99243802e-02 -7.51846135e-01
-1.21305013e+00 6.90909207e-01 5.06338418e-01 1.72312051e-01
-3.23318690e-01 8.86729583e-02 3.47821981e-01 6.40122771e-01
3.72130096e-01 2.02069685e-01 -7.16046035e-01 4.72268075e-01
-7.23698795e-01 6.60427138e-02 4.39621449e-01 8.77131045e-01
1.20209825e+00 -1.94772959e-01 -3.37436795e-01 1.04935133e+00
-6.34910986e-02 3.22703749e-01 7.10471809e-01 -9.88175213e-01
-1.04587145e-01 6.60020530e-01 7.04081655e-02 -4.80734348e-01
-6.38763532e-02 -4.59537268e-01 -9.23870265e-01 3.27151150e-01
3.62099975e-01 2.90776759e-01 -1.61274314e+00 1.97093928e+00
3.89156759e-01 6.25326931e-01 -5.06859422e-02 5.62423289e-01
5.09670734e-01 3.28828126e-01 3.70090008e-01 -3.30224365e-01
1.13285506e+00 -1.06506228e+00 -2.93895036e-01 -3.18865955e-01
1.57096416e-01 -2.56711155e-01 1.25907743e+00 5.92827082e-01
-5.33277869e-01 -9.10969317e-01 -9.40236032e-01 8.78918394e-02
-4.67742383e-01 -1.23611568e-02 5.84384501e-01 2.51967847e-01
-9.83602583e-01 4.81761187e-01 -7.15631127e-01 -4.23499137e-01
6.14628613e-01 1.58596367e-01 -2.23520473e-01 -5.23177922e-01
-6.21920764e-01 3.91138613e-01 7.09995508e-01 -3.01155478e-01
-1.45335412e+00 -7.22278476e-01 -5.18817723e-01 1.57631245e-02
7.17334211e-01 -6.40943110e-01 1.46510005e+00 -1.23502314e+00
-7.92873621e-01 9.22632933e-01 -4.99359518e-01 -7.27086186e-01
3.70384067e-01 -2.37987451e-02 -3.79830211e-01 -2.99092121e-02
4.19505328e-01 8.01506519e-01 1.29341817e+00 -1.51679230e+00
-1.06490159e+00 -5.59565663e-01 9.29925069e-02 -1.75343547e-02
-2.33738542e-01 -6.21423185e-01 -4.26081687e-01 -6.82536602e-01
4.74366844e-01 -1.08570278e+00 -1.45052016e-01 6.12113299e-03
-3.53978842e-01 -5.60167074e-01 1.00904584e+00 -3.19169998e-01
8.01712930e-01 -2.13910484e+00 4.85603996e-02 -8.97959247e-02
5.06116211e-01 4.27861698e-02 -2.06867203e-01 -6.92323223e-02
-1.15739435e-01 1.29518703e-01 -2.40809679e-01 -4.62423861e-01
-2.43168563e-01 2.60971695e-01 -7.37116516e-01 1.12751432e-01
-7.16582835e-02 7.17269003e-01 -8.78730178e-01 -3.34824443e-01
-2.53222343e-02 -1.12914495e-01 -5.39503098e-01 1.71349496e-01
-5.36505938e-01 3.52911204e-01 2.15111356e-02 8.90271306e-01
8.72237980e-01 -5.42101204e-01 4.88054082e-02 3.54685374e-02
8.78370851e-02 -2.85020739e-01 -1.21002495e+00 1.50018466e+00
-3.72122616e-01 3.04548532e-01 -4.21608895e-01 -1.23380458e+00
5.24378240e-01 -1.62151292e-01 6.89077079e-02 -4.57977682e-01
-2.09234029e-01 -8.81491601e-02 -2.25745082e-01 -1.83356717e-01
2.39208207e-01 -2.58146644e-01 -1.49473980e-01 5.90482175e-01
3.91000003e-01 1.89103648e-01 1.93039030e-01 4.03248161e-01
9.79541063e-01 -3.57806146e-01 3.68418962e-01 -2.04805166e-01
4.18102682e-01 1.33948401e-01 7.16703415e-01 1.49910522e+00
-2.10788831e-01 3.62559199e-01 2.49789521e-01 -4.78855461e-01
-8.75776172e-01 -1.62497354e+00 -5.99027835e-02 1.38789213e+00
3.18812490e-01 8.37673619e-02 -2.24935293e-01 -1.15459120e+00
2.73169100e-01 7.10597336e-01 -8.66217971e-01 -3.96059453e-01
-4.71577011e-02 -5.07848561e-01 -6.33193254e-02 4.06717181e-01
4.37442124e-01 -9.51608956e-01 -3.23530644e-01 2.48184785e-01
-9.71291959e-02 -7.82165349e-01 -5.46514928e-01 2.60896206e-01
-1.01858044e+00 -1.31912446e+00 -4.44915235e-01 -1.08681035e+00
9.43668723e-01 7.55581856e-01 1.12163091e+00 1.24018207e-01
-4.80717003e-01 5.98736763e-01 -4.34518099e-01 -3.93207759e-01
-3.64049226e-01 -1.96898691e-02 3.36494297e-01 2.46775448e-01
3.87194723e-01 -4.59939688e-01 -4.60279644e-01 2.99698114e-01
-9.97725487e-01 1.62681729e-01 7.09387004e-01 1.28208840e+00
9.55855072e-01 4.27166671e-01 9.86034036e-01 -1.23821783e+00
1.15378216e-01 -7.24774182e-01 -6.07452869e-01 4.13253397e-01
-8.79847646e-01 2.74459183e-01 6.45350456e-01 -9.74606276e-01
-1.17099607e+00 2.03911588e-01 3.94342750e-01 -6.46143436e-01
-3.66874307e-01 4.02188480e-01 -1.53116109e-02 9.14810598e-02
4.85635400e-01 4.33371216e-01 -5.17173000e-02 -7.03767359e-01
5.12893260e-01 5.43438196e-01 1.01860607e+00 -4.87527847e-01
8.60821486e-01 6.12718523e-01 -2.62871265e-01 -6.17319405e-01
-1.12473202e+00 -5.46446860e-01 -1.10960448e+00 -5.62607497e-02
3.42808008e-01 -1.06186020e+00 -4.70128655e-01 6.03320599e-01
-7.06628919e-01 -4.54020590e-01 -4.98380244e-01 1.58494070e-01
-4.33063626e-01 3.79675180e-01 -3.13296050e-01 -6.06493831e-01
-1.33118123e-01 -7.71479189e-01 6.93579018e-01 2.51046121e-01
1.86528698e-01 -7.84007490e-01 -1.45293817e-01 -1.20050140e-01
1.72003135e-01 -5.54355085e-02 1.33381629e+00 -6.34475112e-01
-6.83961987e-01 1.11637525e-02 -2.99943209e-01 6.38611019e-01
3.68381053e-01 -3.86886895e-01 -9.17790174e-01 -7.73574054e-01
-2.80902773e-01 -5.79167604e-01 1.42099357e+00 2.24549010e-01
1.55960083e+00 -3.77080232e-01 -5.48556030e-01 2.89497584e-01
1.53675497e+00 3.74668330e-01 2.15326831e-01 1.08221591e-01
4.87303972e-01 3.91453981e-01 7.84424543e-01 3.96441460e-01
3.81269574e-01 4.07257974e-01 2.98173666e-01 1.20815009e-01
-4.51213628e-01 -3.42396051e-01 -1.03742592e-01 4.87874806e-01
4.51507658e-01 -5.41697331e-02 -7.41585255e-01 9.86706138e-01
-1.87265623e+00 -8.97598207e-01 2.90430039e-01 2.40011764e+00
7.19342351e-01 2.17658222e-01 8.66083652e-02 1.46794885e-01
8.70611727e-01 -4.46853191e-01 -1.25583529e+00 4.72271740e-02
3.60990912e-02 1.48112670e-01 1.33100316e-01 3.67528111e-01
-1.42956150e+00 9.09527719e-01 6.32649612e+00 6.76694870e-01
-1.07750511e+00 2.47599736e-01 8.90920997e-01 1.16357878e-01
-2.94643119e-02 1.40497267e-01 -7.33053088e-01 5.04637539e-01
6.57417238e-01 -6.13967657e-01 4.24918830e-01 1.18579745e+00
-3.18292052e-01 -2.92561054e-01 -1.37663996e+00 8.87064278e-01
7.77477399e-02 -1.09462988e+00 3.24029326e-01 -2.46965244e-01
1.02218139e+00 -2.84777052e-04 4.24246937e-01 6.91426218e-01
7.46672392e-01 -6.30571842e-01 6.97628140e-01 5.55439293e-01
1.06576729e+00 -5.62730312e-01 3.77954572e-01 4.12874222e-01
-1.24956560e+00 -6.53899193e-01 -4.89981860e-01 5.76893315e-02
-4.27941918e-01 6.98490441e-01 -1.10784924e+00 1.46003634e-01
9.43880320e-01 8.28338265e-01 -1.12336659e+00 1.36874545e+00
-6.89949244e-02 7.76245117e-01 -1.42848104e-01 3.66558880e-01
-2.36677565e-03 2.85956502e-01 2.69242436e-01 8.95695925e-01
2.94402957e-01 -2.07482837e-02 4.36563730e-01 7.75813401e-01
-2.88133621e-01 -2.35193089e-01 -5.27374208e-01 3.53528619e-01
8.69860113e-01 1.05988193e+00 -9.98235226e-01 -7.31056988e-01
-3.80300522e-01 1.17433906e+00 5.27587652e-01 3.44174862e-01
-5.04116774e-01 -2.85225034e-01 5.07605255e-01 3.34182847e-03
5.00110090e-01 -1.01412296e-01 -7.83177763e-02 -1.43254292e+00
-2.49313265e-01 -5.66532969e-01 9.94985580e-01 -5.48970461e-01
-1.63938093e+00 4.59069759e-01 1.35653004e-01 -1.17304015e+00
-2.52607435e-01 -2.87082076e-01 -3.96778166e-01 4.09800559e-01
-1.44568527e+00 -1.08073759e+00 -5.86107731e-01 6.51324093e-01
1.03596509e+00 -1.22027032e-01 5.82368553e-01 6.89007938e-02
-3.64659131e-01 6.68294966e-01 4.35194582e-01 -2.11039230e-01
7.12623477e-01 -1.37260604e+00 3.32689732e-01 1.14353895e+00
1.03255577e-01 5.20372272e-01 4.97361481e-01 -6.40627921e-01
-1.01828337e+00 -1.75780821e+00 5.71345389e-01 -3.46175402e-01
4.18324530e-01 -5.48322201e-01 -1.26590037e+00 9.26735520e-01
-4.43832397e-01 8.74984339e-02 2.42798880e-01 4.08911631e-02
-8.28401625e-01 -2.20363721e-01 -1.31257820e+00 3.72386724e-01
1.27410698e+00 -4.11023676e-01 -7.75269151e-01 2.52042174e-01
8.91188622e-01 1.89401120e-01 -1.78213343e-01 3.54092091e-01
3.12775493e-01 -5.19968092e-01 1.04121268e+00 -6.10747337e-01
-1.71736442e-02 -4.96614516e-01 -1.34693384e-01 -1.45976555e+00
-6.14002645e-01 -5.81273586e-02 -2.03898206e-01 1.31621873e+00
4.72770669e-02 -6.61515951e-01 6.92379415e-01 3.05575132e-01
9.94891375e-02 -1.50579035e-01 -9.52202141e-01 -1.28733909e+00
1.68734372e-01 -3.51436555e-01 5.67853630e-01 7.99036324e-01
-3.69210213e-01 3.93073201e-01 -3.01925898e-01 2.13945717e-01
1.22625744e+00 5.33831298e-01 5.36432207e-01 -1.66307330e+00
-1.13934465e-01 -1.45921513e-01 -2.81667799e-01 -1.00605261e+00
2.17065781e-01 -8.58724058e-01 3.76593977e-01 -1.17190409e+00
7.30096757e-01 -7.05379546e-01 -7.03009903e-01 8.82442176e-01
-3.64714295e-01 4.24043030e-01 1.88406765e-01 6.06810689e-01
-1.01040649e+00 2.86062241e-01 7.18934178e-01 -5.22449970e-01
-2.10598931e-01 8.68453011e-02 -1.00093222e+00 6.79668128e-01
5.37859619e-01 -6.14503860e-01 -7.05205262e-01 -1.83431923e-01
-4.19099420e-01 -3.11523616e-01 5.74419141e-01 -1.15311646e+00
4.10174459e-01 -2.04484880e-01 3.09697151e-01 -9.26920116e-01
-3.17068808e-02 -6.49399161e-01 1.48768470e-01 6.98401451e-01
-5.06240129e-01 -2.71937370e-01 -1.97167117e-02 1.11041975e+00
-2.19406802e-02 -3.41983549e-02 1.19164646e+00 -2.32593730e-01
-1.50910783e+00 6.25159442e-01 -3.36030126e-01 -7.14693442e-02
1.15540266e+00 7.79281706e-02 -3.97754908e-01 -7.47521147e-02
-1.17172015e+00 2.20677778e-01 6.35785222e-01 6.35288239e-01
7.22894967e-01 -1.35022700e+00 -5.47787964e-01 2.30624333e-01
6.66503787e-01 -7.88572207e-02 5.54273844e-01 2.76401997e-01
1.69494569e-01 3.21854442e-01 -1.13176659e-01 -7.87384391e-01
-1.35034370e+00 1.40709388e+00 1.49308398e-01 -1.34479761e-01
-5.45678020e-01 7.72142291e-01 6.06066048e-01 -4.58828211e-01
3.39275420e-01 -1.50895894e-01 -2.75116563e-01 4.44306321e-02
7.29613006e-01 2.65259266e-01 9.44022015e-02 -2.89135247e-01
-2.26125523e-01 3.35451335e-01 -6.30000412e-01 1.97432086e-01
1.08258235e+00 -6.04278386e-01 -1.01933584e-01 8.30458343e-01
1.16965139e+00 -6.27601445e-01 -1.45855296e+00 -8.83965135e-01
2.86953360e-01 -5.10458887e-01 -2.01620445e-01 -1.17637694e+00
-7.52439797e-01 6.19369566e-01 1.23879361e+00 2.83618998e-02
1.30273950e+00 3.75157148e-01 3.09012800e-01 5.98960578e-01
8.22219789e-01 -1.04183269e+00 6.15118623e-01 6.13466918e-01
7.34912694e-01 -1.36445844e+00 -2.68793285e-01 -2.29572743e-01
-5.59664786e-01 8.43366504e-01 8.07915568e-01 -8.06775242e-02
7.71868348e-01 -2.89030373e-01 -1.78782061e-01 1.52729407e-01
-1.08421481e+00 -4.75503832e-01 1.58368334e-01 7.89969087e-01
-2.78910071e-01 1.75298437e-01 7.34978169e-02 7.61945844e-01
2.49060959e-01 2.02873453e-01 5.93943477e-01 9.21054959e-01
-5.70407510e-01 -9.05158520e-01 -3.24358940e-01 8.03077281e-01
-3.57027538e-02 4.94305752e-02 -1.50435761e-01 4.30599183e-01
4.88257974e-01 9.58602309e-01 3.32966536e-01 -4.77744937e-01
1.58073440e-01 1.68653086e-01 3.03662270e-01 -9.60912883e-01
2.28643090e-01 -2.10839704e-01 -5.33721149e-01 -2.08557129e-01
-1.88501686e-01 -8.70084703e-01 -1.02910757e+00 -1.16995625e-01
-1.84594929e-01 2.01183945e-01 3.58428866e-01 8.03021371e-01
5.15001953e-01 4.20347333e-01 9.88786995e-01 -7.07938492e-01
-4.82775539e-01 -8.94846261e-01 -7.12812364e-01 6.97885811e-01
6.05449557e-01 -1.12376273e+00 -5.15110314e-01 5.33083439e-01]
|
[9.869730949401855, 3.217721462249756]
|
78f0683c-e302-4d6b-b3d6-5f9e5d0947f1
|
decoding-finger-movements-from-ecog-signals
| null | null |
https://www.frontiersin.org/articles/10.3389/fnins.2012.00029/full
|
https://www.frontiersin.org/articles/10.3389/fnins.2012.00029/full
|
Decoding finger movements from ECoG signals using switching linear models
|
One of the most interesting challenges in ECoG-based Brain-Machine Interface is movement prediction. Being able to perform such a prediction paves the way to high-degree precision command for a machine such as a robotic arm or robotic hands. As a witness of the BCI community increasing interest toward such a problem, the fourth BCI Competition provides a dataset which aim is to predict individual finger movements from ECoG signals. The difficulty of the problem relies on the fact that there is no simple relation between ECoG signals and finger movements. We propose in this paper, to estimate and decode these finger flexions using switching models controlled by an hidden state. Switching models can integrate prior knowledge about the decoding problem and helps in predicting fine and precise movements. Our model is thus based on a first block which estimates which finger is moving and another block which, knowing which finger is moving, predicts the movements of all other fingers. Numerical results that have been submitted to the Competition show that the model yields high decoding performances when the hidden state is well estimated. This approach achieved the second place in the BCI competition with a correlation measure between real and predicted movements of 0.42.
|
['Alain Rakotomamonjy', 'Rémi Flamary']
|
2012-03-06
| null | null | null |
front-neurosci-sec-neuroprosthetics-2012-3
|
['brain-decoding', 'brain-decoding']
|
['medical', 'miscellaneous']
|
[ 1.22381352e-01 1.79837167e-01 -1.44709066e-01 -5.53975776e-02
-3.30703229e-01 -2.97488272e-01 6.05067730e-01 -5.17215490e-01
-5.00908017e-01 8.06079090e-01 1.37250215e-01 -9.22278613e-02
-4.56760079e-01 -3.80854249e-01 -7.22413123e-01 -7.16660798e-01
-3.16077352e-01 5.41015625e-01 3.46904576e-01 -4.04153883e-01
3.90767276e-01 7.03706264e-01 -1.44842577e+00 3.59247029e-01
6.63956404e-01 1.09813821e+00 5.94690859e-01 6.98086739e-01
3.33628178e-01 4.59195465e-01 -4.78416800e-01 -1.01983361e-01
1.43971115e-01 -4.91349816e-01 -8.58814716e-01 -5.25721490e-01
-1.62013635e-01 -7.93258026e-02 -3.79128724e-01 8.06193948e-01
7.51143038e-01 -2.91550189e-01 8.50976646e-01 -1.13881052e+00
1.70350417e-01 6.21289194e-01 -2.84564883e-01 3.20107758e-01
3.56683731e-01 1.16649963e-01 6.60464585e-01 -2.87211865e-01
7.68263996e-01 7.18597531e-01 4.61944789e-01 5.98459482e-01
-1.12077534e+00 -5.76077998e-01 7.94915333e-02 7.73657978e-01
-1.34155869e+00 -4.52077627e-01 7.80790687e-01 -6.26674354e-01
9.28723931e-01 4.70366836e-01 1.01883757e+00 1.27243674e+00
5.12439668e-01 7.15729773e-01 9.87189710e-01 -6.10327601e-01
2.00952008e-01 3.31627801e-02 6.97879419e-02 8.33859071e-02
-2.24029813e-02 1.25340119e-01 -7.91966558e-01 2.66331375e-01
7.57464707e-01 -4.21545863e-01 -6.32698476e-01 -7.42016360e-02
-1.35369146e+00 4.00177985e-01 5.05244434e-01 8.71919930e-01
-8.85314405e-01 2.32692584e-01 -6.89801350e-02 3.16421181e-01
7.05170408e-02 3.30250561e-01 -3.34508121e-01 -6.94466710e-01
-1.14750481e+00 1.47753209e-01 8.66989613e-01 7.63920188e-01
9.54706222e-02 -4.28171217e-01 -4.07793462e-01 4.58053321e-01
1.91366449e-01 3.48253578e-01 5.65668225e-01 -6.84921086e-01
6.84846938e-01 3.27354133e-01 3.69015634e-01 -8.09295654e-01
-7.69016981e-01 -4.63528126e-01 -7.51203179e-01 4.60692912e-01
9.12126899e-01 -2.38120750e-01 -6.83459222e-01 1.83008862e+00
-1.50410011e-01 -3.56942303e-02 -4.89285201e-01 1.13195813e+00
1.23045333e-02 2.33705774e-01 -1.65487781e-01 -2.12329105e-01
1.12187564e+00 -3.28767240e-01 -7.80998409e-01 -1.91170290e-01
3.84890825e-01 -4.22739685e-01 6.82217121e-01 1.09968984e+00
-1.04237902e+00 -3.90498161e-01 -9.09013927e-01 5.09952247e-01
-1.23763181e-01 4.57424223e-01 6.30244195e-01 6.07748210e-01
-1.10273206e+00 9.81358588e-01 -1.23930895e+00 -4.04316485e-01
3.00807536e-01 9.65764463e-01 -4.18540746e-01 4.97169793e-01
-1.18058932e+00 1.50052416e+00 3.63520443e-01 5.10873437e-01
-2.45941818e-01 -1.58288896e-01 8.18450004e-02 5.12907989e-02
-1.45276040e-01 -3.29397261e-01 6.67709947e-01 -8.30367029e-01
-1.74520087e+00 6.78864539e-01 -2.34203264e-01 -3.73885900e-01
1.00732267e+00 -1.19175069e-01 -5.34239292e-01 -6.51576892e-02
-1.95365310e-01 7.26732433e-01 9.78193164e-01 -7.83546329e-01
-5.24898648e-01 -8.25253248e-01 -4.66284364e-01 -7.57012144e-02
-1.29386902e-01 -2.49675989e-01 -3.23291749e-01 -4.81245905e-01
3.16149831e-01 -1.16774690e+00 1.67936802e-01 -3.54278833e-01
-6.76945925e-01 -2.73362041e-01 1.94500446e-01 -1.03234363e+00
1.29518819e+00 -1.84518754e+00 6.86878979e-01 6.41700268e-01
2.69769162e-01 8.72622128e-04 -2.93795653e-02 3.55419844e-01
-2.82029450e-01 -3.52791309e-01 1.58944666e-01 2.00687915e-01
-8.76179524e-03 -2.47767419e-01 -1.75795823e-01 5.49924910e-01
-1.03975289e-01 8.74823153e-01 -4.72588956e-01 3.71400546e-03
4.16430533e-02 4.42613006e-01 -4.86219227e-01 4.30134796e-02
1.40678078e-01 8.37510526e-01 -2.91902661e-01 1.98290288e-03
4.00162458e-01 2.48369407e-02 2.59378165e-01 -2.93169320e-01
-3.12967271e-01 3.34208518e-01 -1.17851150e+00 1.79278243e+00
-2.52489120e-01 9.29513097e-01 -1.80514827e-01 -1.11104739e+00
8.43188405e-01 5.00393510e-01 6.44847453e-01 -7.51762331e-01
5.83318472e-01 3.30199450e-01 6.40555680e-01 -5.15852988e-01
-1.06081568e-01 6.33863360e-02 2.15195715e-01 4.25238639e-01
-2.38848142e-02 1.09475933e-01 8.16865414e-02 -1.15851097e-01
1.11207020e+00 2.41381735e-01 -3.48638780e-02 -2.66012043e-01
4.10133034e-01 -2.24944308e-01 -9.25775268e-04 7.97578990e-01
7.73498565e-02 4.03043330e-01 7.02426791e-01 -9.68242809e-02
-5.25690198e-01 -7.96323657e-01 2.30038650e-02 5.86773336e-01
9.07425433e-02 -5.37708290e-02 -1.08910942e+00 -2.77615432e-02
-1.13344617e-01 6.22158647e-01 -7.93823004e-01 -3.93585712e-01
-5.40125489e-01 -3.76125991e-01 2.59823412e-01 5.22947371e-01
2.38952175e-01 -1.29554355e+00 -8.67063046e-01 3.90291631e-01
-4.23607260e-01 -8.71999800e-01 7.60525614e-02 4.63379681e-01
-9.35398519e-01 -1.00800097e+00 -8.83675694e-01 -5.41876674e-01
2.11889908e-01 -5.58698475e-01 4.80193943e-01 -2.92457134e-01
-2.21076414e-01 1.74610123e-01 -3.14337611e-01 -4.83829528e-01
-1.29360020e-01 3.72423053e-01 1.23445675e-01 1.20868415e-01
5.16438723e-01 -8.22564423e-01 -5.72994709e-01 2.58121580e-01
-2.44608939e-01 2.69913554e-01 7.89281070e-01 5.09541392e-01
2.24259213e-01 -3.62622619e-01 4.84717697e-01 -2.44373605e-01
8.02850783e-01 -1.98624372e-01 -4.24348235e-01 3.78579378e-01
-3.74846339e-01 3.34146321e-01 3.83182734e-01 -5.81683517e-01
-5.95425963e-01 5.22953331e-01 -3.17566633e-01 1.55332446e-01
-2.00147688e-01 2.13488519e-01 -3.68633896e-01 -9.15161967e-02
5.68697035e-01 3.97175223e-01 3.69592495e-02 -5.75298667e-01
1.28462493e-01 9.04896438e-01 6.59866512e-01 -1.57764658e-01
2.95486540e-01 2.14447409e-01 3.13997902e-02 -9.06643867e-01
7.38956258e-02 -1.82590991e-01 -1.03951681e+00 -4.79689717e-01
7.93368876e-01 -4.11769986e-01 -1.14152479e+00 7.85860896e-01
-1.46608722e+00 -5.23640037e-01 9.50969383e-02 9.58950341e-01
-8.83762717e-01 -4.39721392e-03 -2.82642335e-01 -7.44471908e-01
-2.18904927e-01 -1.12172174e+00 6.70204103e-01 -1.01297863e-01
-6.33521020e-01 -4.62098360e-01 2.58133225e-02 -2.54022423e-03
4.40166265e-01 8.95793959e-02 6.16608799e-01 -4.27797019e-01
-4.07552540e-01 -6.16012394e-01 1.78326398e-01 1.53357144e-02
-1.92655399e-02 -4.68041837e-01 -7.84704268e-01 -1.16637677e-01
1.00709401e-01 1.98314518e-01 4.96446192e-01 6.84650421e-01
1.11665142e+00 5.25123850e-02 -4.76813018e-01 2.97783643e-01
1.05307496e+00 2.84882277e-01 1.19576442e+00 7.19559342e-02
4.07842398e-01 6.29196882e-01 1.47274449e-01 3.95398349e-01
1.81128934e-01 1.30445874e+00 4.58606958e-01 3.85396689e-01
-3.34857106e-02 1.08581528e-01 1.02524906e-01 6.27726972e-01
-9.12690878e-01 -8.33302960e-02 -8.29263628e-01 1.99975938e-01
-1.70716298e+00 -1.04052436e+00 -4.75273997e-01 2.47542191e+00
6.51379764e-01 2.82508910e-01 2.75690556e-01 7.04076648e-01
6.18004382e-01 -5.55490851e-01 -5.88263988e-01 -1.45584956e-01
2.62216866e-01 5.36841452e-01 5.47841668e-01 5.93890369e-01
-5.34473360e-01 9.04054105e-01 5.80805779e+00 5.20437121e-01
-1.52508533e+00 1.90668449e-01 2.87560429e-02 -2.70853013e-01
2.93883979e-01 -3.23794693e-01 -8.04646194e-01 9.38344061e-01
8.89235198e-01 9.24023613e-02 9.50424373e-01 1.93825483e-01
5.07021785e-01 -4.65328604e-01 -1.24637341e+00 1.30543613e+00
-1.42972305e-01 -9.90045190e-01 -3.35887313e-01 2.61510104e-01
2.05921024e-01 1.08128991e-02 -2.09960029e-01 -3.06289587e-02
-3.90326411e-01 -1.08462763e+00 9.90976393e-01 1.18641698e+00
7.27279484e-01 -4.66895223e-01 4.45927978e-01 7.57962763e-01
-9.23095107e-01 -1.80070609e-01 5.19479848e-02 -3.00340533e-01
1.00169368e-01 1.84584603e-01 -4.56639946e-01 1.64615259e-01
4.67996061e-01 5.30052483e-01 -3.80028903e-01 1.26824653e+00
-5.57424724e-01 5.88544369e-01 -5.27373731e-01 -3.72937977e-01
-1.57519817e-01 1.67140998e-02 6.12506211e-01 9.39930677e-01
5.34256995e-01 1.53491467e-01 -7.79808462e-01 9.03314650e-01
3.92446309e-01 -7.72907734e-02 -3.12388718e-01 3.03689111e-02
2.91738689e-01 8.36676776e-01 -7.40844190e-01 9.99566391e-02
8.86601806e-02 1.38463902e+00 1.80144191e-01 4.14002120e-01
-7.99106419e-01 -5.74752390e-01 2.62315154e-01 2.59899259e-01
1.41141072e-01 -3.65015745e-01 -6.10063374e-01 -1.01700282e+00
3.80430639e-01 -5.75281918e-01 -2.23790243e-01 -8.33256364e-01
-4.95599180e-01 6.75497651e-01 -1.34165004e-01 -1.19453347e+00
-7.07332373e-01 -1.06192005e+00 -3.47155124e-01 1.15360856e+00
-8.57298315e-01 -7.58240402e-01 -4.07240838e-01 9.08683479e-01
-1.66260190e-02 -9.03149247e-02 8.61874282e-01 2.76785791e-01
-2.64119178e-01 4.79203641e-01 -8.99203494e-02 7.11160749e-02
5.30947983e-01 -1.00280166e+00 1.00169018e-01 6.40805602e-01
2.99742162e-01 4.99504387e-01 7.23431408e-01 -3.77128571e-01
-1.33078170e+00 -2.23910600e-01 1.13132632e+00 -4.72691625e-01
3.27807486e-01 -4.03313369e-01 -5.63377619e-01 5.43773592e-01
-1.28826261e-01 -1.98356107e-01 1.04436629e-01 -4.78384867e-02
4.21632916e-01 -1.87118456e-01 -7.34488785e-01 3.82543266e-01
1.07114255e+00 -3.28336298e-01 -6.50878489e-01 1.57917231e-01
-4.41858768e-01 -2.76978046e-01 -6.54053688e-01 9.73329619e-02
1.03813791e+00 -1.04284072e+00 6.77479565e-01 -5.37142336e-01
1.93215728e-01 2.10675016e-01 1.95830196e-01 -1.64554060e+00
-3.00949603e-01 -6.05468988e-01 -2.30020791e-01 5.47922552e-01
3.73074770e-01 -6.31742179e-01 8.65710974e-01 7.13929951e-01
2.83920705e-01 -7.40953982e-01 -1.07684100e+00 -9.09310997e-01
5.28801717e-02 -7.20613420e-01 3.43594283e-01 3.63272429e-01
6.34243608e-01 1.84898511e-01 -4.47474778e-01 -9.67718437e-02
3.84741545e-01 -2.21729442e-01 4.93744612e-01 -1.60070658e+00
-3.96944672e-01 -8.22348952e-01 -8.17259312e-01 -1.24462581e+00
-5.57256937e-02 -8.13192666e-01 4.59018163e-02 -1.70505774e+00
-1.65676296e-01 -2.22289190e-01 -2.97188669e-01 4.38603580e-01
2.11666256e-01 8.79354700e-02 1.20246224e-01 2.12141350e-01
-1.04909256e-01 6.90782741e-02 1.11534524e+00 -4.59845550e-02
-5.85672855e-01 4.15255487e-01 -2.64163166e-01 4.02357519e-01
8.83842409e-01 -4.31152374e-01 -1.06815971e-01 -2.73187518e-01
2.26253510e-01 4.49756116e-01 4.69295442e-01 -1.34968233e+00
7.32462287e-01 3.87324035e-01 6.30759418e-01 -4.58945364e-01
5.19490719e-01 -8.77084017e-01 2.57739037e-01 9.05736029e-01
-2.67209649e-01 -3.57457072e-01 3.67777497e-02 1.74189106e-01
-6.54433668e-02 -5.26320003e-02 5.65121591e-01 6.34834841e-02
-6.73318923e-01 1.08627953e-01 -6.86896503e-01 -1.77342877e-01
8.64110529e-01 -6.03173912e-01 2.20260516e-01 -5.63433051e-01
-1.35191441e+00 -2.03051299e-01 -2.32506573e-01 5.81812680e-01
3.23732764e-01 -1.11838758e+00 -6.57324493e-01 4.46402669e-01
-1.53287977e-01 -8.73104513e-01 3.06488257e-02 1.39881432e+00
-2.83170074e-01 8.29865873e-01 -7.86783159e-01 -5.32348156e-01
-1.24186587e+00 7.24930912e-02 5.07278383e-01 -4.78963405e-02
-5.28051198e-01 9.02958870e-01 -5.58261931e-01 1.96310356e-01
4.42756295e-01 -4.17144001e-01 -4.52764571e-01 1.33186415e-01
5.57854295e-01 4.72261965e-01 4.23206240e-01 -6.43708289e-01
-5.75660825e-01 5.99672914e-01 2.39952251e-01 -4.65290219e-01
1.51295662e+00 1.34234250e-01 -9.57326032e-03 4.28330570e-01
8.11387956e-01 -1.59986615e-01 -1.30228198e+00 3.89090359e-01
2.65969485e-01 -2.28218496e-01 2.62499601e-01 -1.36265945e+00
-1.08857667e+00 1.19309580e+00 1.01416063e+00 3.99472117e-02
1.04469025e+00 -2.04207048e-01 4.68621820e-01 3.50884318e-01
9.79084134e-01 -1.18518376e+00 -3.28973681e-01 2.57296890e-01
1.17499685e+00 -8.92190874e-01 -2.71789253e-01 -1.81626931e-01
-4.99025673e-01 1.27549696e+00 2.53845781e-01 -2.46140629e-01
7.76213109e-01 2.23445818e-01 -2.52248347e-01 5.35089597e-02
-3.25254142e-01 -1.63915038e-01 5.92898488e-01 5.75403154e-01
4.69271690e-01 3.04483265e-01 -8.00243437e-01 9.75989997e-01
-4.01834816e-01 5.81540048e-01 7.71475807e-02 6.48544252e-01
-3.34172159e-01 -1.18968105e+00 -3.28114688e-01 6.13371849e-01
-2.10215017e-01 1.08379625e-01 -4.52282786e-01 9.22550917e-01
1.49845436e-01 8.54684770e-01 -9.00161564e-02 -6.95981443e-01
4.29830402e-01 1.72915205e-01 1.21534503e+00 -1.88791320e-01
-5.29629886e-01 -1.64368287e-01 -2.73138106e-01 -9.31160271e-01
-4.85523185e-03 -6.38481557e-01 -1.16286278e+00 1.90897528e-02
-3.42336982e-01 4.69814464e-02 1.16476095e+00 1.18191969e+00
2.57284701e-01 4.21075970e-01 2.24734783e-01 -1.39310014e+00
-6.84300900e-01 -1.26605666e+00 -9.38179851e-01 1.55810446e-01
1.02368310e-01 -7.22689748e-01 -6.46228567e-02 -8.62604901e-02]
|
[12.960836410522461, 3.392345428466797]
|
d6bcc3b8-0f8b-49b9-977d-b4bff95fb33e
|
distributional-model-equivalence-for-risk
|
2307.01708
| null |
https://arxiv.org/abs/2307.01708v1
|
https://arxiv.org/pdf/2307.01708v1.pdf
|
Distributional Model Equivalence for Risk-Sensitive Reinforcement Learning
|
We consider the problem of learning models for risk-sensitive reinforcement learning. We theoretically demonstrate that proper value equivalence, a method of learning models which can be used to plan optimally in the risk-neutral setting, is not sufficient to plan optimally in the risk-sensitive setting. We leverage distributional reinforcement learning to introduce two new notions of model equivalence, one which is general and can be used to plan for any risk measure, but is intractable; and a practical variation which allows one to choose which risk measures they may plan optimally for. We demonstrate how our framework can be used to augment any model-free risk-sensitive algorithm, and provide both tabular and large-scale experiments to demonstrate its ability.
|
['Amir-Massoud Farahmand', 'Murat A. Erdogdu', 'Tyler Kastner']
|
2023-07-04
| null | null | null | null |
['distributional-reinforcement-learning']
|
['methodology']
|
[ 1.04291372e-01 4.78381962e-01 -8.17832828e-01 -3.73948157e-01
-1.48777986e+00 -7.34605432e-01 3.20620716e-01 2.98077255e-01
-7.08338439e-01 8.82237732e-01 1.66416824e-01 -5.95859289e-01
-6.42956138e-01 -1.09533918e+00 -5.37604809e-01 -6.18592739e-01
-4.62964863e-01 7.84365714e-01 1.26116171e-01 -2.96014637e-01
3.23869348e-01 3.60078156e-01 -1.17071307e+00 9.54962894e-02
8.61973584e-01 7.58582950e-01 -2.20694795e-01 6.89237595e-01
4.22585279e-01 8.76012504e-01 -2.91440070e-01 -3.01302940e-01
5.18317997e-01 -2.91499764e-01 -9.62153018e-01 -2.98616081e-01
9.93515402e-02 -5.79817533e-01 8.85477103e-03 1.12441754e+00
4.97884154e-01 3.63462240e-01 9.04668331e-01 -1.30800438e+00
-1.56588882e-01 8.64650011e-01 -4.05069351e-01 2.62764871e-01
3.09483856e-01 2.11196959e-01 1.29519391e+00 -1.58138841e-01
3.53229880e-01 1.50527477e+00 4.47202414e-01 7.80974448e-01
-1.30284417e+00 -5.86986184e-01 1.87116683e-01 -1.28926888e-01
-6.58881307e-01 -8.00879896e-02 5.25901973e-01 -4.73368436e-01
8.58329594e-01 3.35502148e-01 6.93990409e-01 8.93711150e-01
5.50974965e-01 6.12627506e-01 1.52734280e+00 -4.43348378e-01
7.37442255e-01 -1.94763735e-01 1.47692328e-02 5.42106867e-01
1.87945023e-01 1.21629667e+00 -1.60653681e-01 -6.18208885e-01
4.39232409e-01 1.05327368e-01 3.00169196e-02 -6.95446610e-01
-8.72975886e-01 1.44530368e+00 1.99946523e-01 -1.07909940e-01
-2.21174777e-01 4.13773566e-01 3.75329465e-01 7.34032154e-01
3.52368027e-01 9.23343003e-01 -4.30738002e-01 -9.66612250e-02
-5.48816860e-01 8.42992425e-01 9.16699111e-01 6.41114175e-01
5.18715322e-01 -7.29428083e-02 -5.44852018e-01 4.93964553e-01
3.96773696e-01 5.07318437e-01 2.59984821e-01 -1.37624884e+00
5.65608382e-01 6.80316985e-02 6.29908323e-01 -2.93056250e-01
-6.56679451e-01 -2.25987732e-02 -1.48217514e-01 6.06469154e-01
3.63965839e-01 -5.75040758e-01 -6.29127502e-01 2.17052221e+00
1.43811405e-01 -2.01020036e-02 2.51379848e-01 2.72603482e-01
-1.48987785e-01 5.23337960e-01 2.62655824e-01 -7.35720634e-01
9.62365687e-01 -4.12291586e-01 -2.99485445e-01 -2.23512903e-01
9.71886396e-01 -2.75522053e-01 1.35826325e+00 2.91773200e-01
-1.26721835e+00 1.97753057e-01 -1.18508887e+00 5.26590347e-01
3.03643756e-02 -1.02727938e+00 6.16496623e-01 1.02134347e+00
-9.30046976e-01 9.25286353e-01 -7.17070639e-01 -4.16930020e-02
6.63959265e-01 3.32549870e-01 1.57763705e-01 1.64715592e-02
-1.44041526e+00 1.07857132e+00 6.52666688e-01 -6.54819489e-01
-1.61706281e+00 -7.66811728e-01 -8.13046753e-01 2.44759068e-01
9.92630541e-01 -5.55002332e-01 1.71915495e+00 -5.39037526e-01
-1.45352352e+00 2.96420425e-01 4.21786189e-01 -7.76791632e-01
7.27981865e-01 -3.44243157e-03 -5.93263619e-02 2.02398643e-01
9.53794718e-02 2.26953983e-01 6.05309248e-01 -9.60135400e-01
-7.43356884e-01 -2.55420238e-01 4.91979450e-01 4.73860502e-01
-2.28196844e-01 2.02250168e-01 4.89937633e-01 -5.19477546e-01
-3.58199537e-01 -1.04214513e+00 -9.33892012e-01 -4.34832484e-01
-3.05327713e-01 -2.53562033e-01 -1.14791207e-02 -1.73060894e-01
1.12906992e+00 -1.64811146e+00 -1.11828990e-01 5.82147896e-01
-1.77178502e-01 -2.77071595e-01 -6.83517978e-02 3.96134704e-01
-1.53046608e-01 5.08014560e-01 -6.20216668e-01 6.57783747e-02
4.55481887e-01 2.19549164e-01 -5.32443404e-01 4.80901778e-01
-5.66455349e-02 9.28674161e-01 -8.86400700e-01 -3.80088598e-01
2.95074135e-02 -2.56369621e-01 -9.50184405e-01 3.28327745e-01
-4.01067317e-01 -2.97133829e-02 -5.93961716e-01 4.04832244e-01
3.55888397e-01 1.71078891e-01 3.00494999e-01 6.80423498e-01
2.21177056e-01 1.59463227e-01 -1.15395832e+00 1.00437593e+00
-5.53570390e-01 -2.27593124e-01 -2.29571298e-01 -9.33680952e-01
4.39047098e-01 1.54119298e-01 5.57005644e-01 -4.94400889e-01
3.18854973e-02 2.13535782e-02 -2.08957836e-01 -3.01933438e-01
1.58147439e-01 -1.07039642e+00 -5.84356010e-01 9.77628350e-01
-3.23574305e-01 -4.08158362e-01 -9.96925160e-02 -6.83728382e-02
1.11925232e+00 -2.28401478e-02 5.09030581e-01 -7.50480056e-01
9.21744183e-02 -1.68448314e-01 8.52897346e-01 1.15602624e+00
-3.59469026e-01 5.23283482e-02 1.06837928e+00 -3.86588991e-01
-6.74686015e-01 -1.27089417e+00 -1.43747464e-01 1.35489428e+00
-8.53244122e-03 -1.78403109e-01 -7.04773188e-01 -1.23666978e+00
1.92044690e-01 1.13987291e+00 -8.55588138e-01 -3.28568608e-01
-3.11863959e-01 -9.52255666e-01 2.52310634e-01 6.61591887e-01
-1.57763399e-02 -1.05200398e+00 -9.35286164e-01 -2.73873731e-02
2.30551943e-01 -5.75833134e-02 -6.98051929e-01 5.76943457e-01
-9.19334233e-01 -1.40341663e+00 -3.42384368e-01 -2.96398312e-01
3.70522797e-01 -1.97623447e-01 1.22199523e+00 -1.53039783e-01
3.69349420e-01 7.25567102e-01 1.62914574e-01 -8.20595443e-01
-6.74297273e-01 -1.92174256e-01 1.57605737e-01 -5.61870694e-01
2.03531057e-01 -2.85835475e-01 -5.44671595e-01 1.59287304e-01
-9.24272895e-01 -5.34874380e-01 2.04985246e-01 1.00636065e+00
6.38735414e-01 2.92027354e-01 9.42603588e-01 -1.45172095e+00
9.39158142e-01 -5.88019609e-01 -9.66332972e-01 6.27034724e-01
-1.07279050e+00 4.58692521e-01 5.41451812e-01 -9.10126492e-02
-9.80090201e-01 -7.35402405e-02 -2.03716248e-01 -4.24230844e-02
2.07959279e-01 4.40654188e-01 -4.16909575e-01 -3.95638235e-02
8.99787247e-01 -2.91990340e-01 5.00909761e-02 -2.20908493e-01
4.71577436e-01 2.79544413e-01 1.02413379e-01 -1.07206094e+00
8.40035141e-01 1.09621137e-01 2.70416468e-01 -1.46584496e-01
-8.87170017e-01 9.75301564e-02 -1.50646180e-01 8.98345411e-02
8.06469321e-01 -7.57213593e-01 -8.79094899e-01 -1.30992666e-01
-3.86299044e-01 -7.93161511e-01 -8.09300959e-01 2.81047672e-01
-1.42138267e+00 2.08726779e-01 -3.60213041e-01 -9.35491920e-01
1.96950394e-03 -1.14780819e+00 4.39732999e-01 6.71332702e-02
-7.26478845e-02 -1.37158167e+00 3.77604872e-01 -4.62685786e-02
2.24490687e-01 3.54392946e-01 1.32150507e+00 -8.61436248e-01
-2.34331354e-01 1.67808309e-01 4.34946865e-01 2.56740570e-01
-1.39150277e-01 -2.86480069e-01 -7.54931331e-01 -6.09408617e-01
3.08968127e-01 -7.70915985e-01 1.02254343e+00 7.04505444e-01
1.24347878e+00 -9.81492817e-01 -1.82024479e-01 4.32381392e-01
1.45691311e+00 5.05412221e-01 4.33788449e-01 4.82196242e-01
2.53572017e-01 8.56638134e-01 1.05714464e+00 4.32538718e-01
3.53765815e-01 5.71005106e-01 5.23155391e-01 2.56963909e-01
8.83542120e-01 -5.27279913e-01 3.97720098e-01 -1.65551528e-01
2.16397330e-01 7.95424581e-02 -8.67883384e-01 3.29569757e-01
-1.86068809e+00 -1.22357357e+00 7.46499658e-01 2.56499171e+00
1.04808724e+00 3.50845188e-01 7.98616529e-01 -5.32056876e-02
4.43484247e-01 5.17317913e-02 -7.25500166e-01 -1.01979470e+00
3.64976138e-01 2.83625424e-01 7.47952163e-01 9.63694990e-01
-1.20810580e+00 6.44936860e-01 8.30565739e+00 9.41111922e-01
-4.86910135e-01 1.17662892e-01 1.09318185e+00 -2.41370127e-01
-1.14938462e+00 1.47521466e-01 -4.53892082e-01 3.47506344e-01
1.34480083e+00 -6.55784667e-01 4.17221338e-01 1.20800936e+00
9.49898660e-02 8.62063617e-02 -1.55139399e+00 4.97088492e-01
-5.49885988e-01 -1.22300911e+00 -1.23879835e-01 8.67667720e-02
9.09358799e-01 -4.62950915e-01 2.58679181e-01 7.83286154e-01
1.07840228e+00 -1.42179024e+00 4.53474641e-01 2.43347600e-01
7.29658306e-01 -1.49367702e+00 5.89574099e-01 4.55719203e-01
-6.56268477e-01 -6.49915159e-01 -4.57207441e-01 -3.31467874e-02
-1.25178263e-01 2.02846825e-01 -7.25651920e-01 3.41500342e-01
3.71721357e-01 4.28710133e-01 -2.38035858e-01 6.50455654e-01
-2.07257271e-01 6.10739768e-01 -2.11346462e-01 3.82579044e-02
3.09564620e-01 8.82803500e-02 4.24109429e-01 8.31223726e-01
2.18057707e-01 5.63899940e-03 5.22142291e-01 6.53744519e-01
2.17108116e-01 1.01521216e-01 -9.08564687e-01 2.27402389e-01
5.55921853e-01 7.36263275e-01 -3.84549946e-01 -9.27020833e-02
-2.54087299e-01 2.38530427e-01 3.68948817e-01 2.89587259e-01
-7.50009716e-01 -2.13733539e-01 6.39885128e-01 -4.95773740e-02
4.05649701e-03 2.04350278e-01 -3.63419652e-01 -8.66226435e-01
-3.22113872e-01 -1.04356563e+00 1.25059104e+00 -1.25156730e-01
-1.53589761e+00 1.33356541e-01 5.22760391e-01 -1.05244136e+00
-8.36526453e-01 -4.55948949e-01 -6.96132958e-01 8.90433788e-01
-1.43386078e+00 -6.53982043e-01 5.04663885e-01 7.49789059e-01
1.79711193e-01 -1.80274978e-01 7.80765831e-01 -5.63090086e-01
-3.79403681e-01 7.82794118e-01 4.38097715e-02 -4.46137667e-01
4.79444355e-01 -1.79404867e+00 7.82570913e-02 7.49231517e-01
-6.62883818e-02 2.67234147e-01 6.51709378e-01 -5.41720748e-01
-1.19910443e+00 -1.16971624e+00 4.83199596e-01 -6.67979956e-01
5.27778149e-01 3.85566801e-02 -5.99059045e-01 8.79465878e-01
-1.47231430e-01 -2.19948247e-01 7.46232748e-01 3.49451870e-01
-3.42569500e-01 -2.24309340e-01 -1.65078676e+00 7.51042724e-01
8.24668527e-01 -3.83602768e-01 -9.19525146e-01 4.57333386e-01
1.02322972e+00 -2.67711848e-01 -9.99444425e-01 4.95798528e-01
3.38839889e-01 -1.11645651e+00 9.13592637e-01 -1.10691667e+00
2.26643458e-01 2.18853101e-01 -3.72899055e-01 -1.52635348e+00
-2.69559056e-01 -1.06181860e+00 -1.99500859e-01 7.94163823e-01
4.44372743e-01 -9.43234682e-01 6.87834442e-01 1.03203714e+00
7.56940618e-02 -1.03940451e+00 -1.22462165e+00 -1.04163289e+00
1.05578351e+00 -5.61309814e-01 8.78402710e-01 6.19543195e-01
3.72466356e-01 -1.18428402e-01 -3.19874763e-01 -5.33463014e-03
8.47928524e-01 2.72885896e-02 1.96077943e-01 -1.03553927e+00
-6.13593996e-01 -6.59694731e-01 -2.59630419e-02 -4.35082465e-01
4.85416502e-01 -8.60989451e-01 1.45542070e-01 -1.06805789e+00
4.07920748e-01 -7.74195790e-01 -7.51233518e-01 6.94294691e-01
-3.09123635e-01 -4.09224153e-01 2.20999613e-01 -1.13442726e-01
-4.58881080e-01 5.70214570e-01 1.06874454e+00 -8.56699049e-02
-4.46293354e-01 3.50384116e-01 -1.35164368e+00 6.93956971e-01
1.01360857e+00 -7.10574806e-01 -9.43465769e-01 3.55344415e-01
4.42277253e-01 5.06805539e-01 1.79325670e-01 -6.74671710e-01
-4.08308476e-01 -1.06892860e+00 2.16650039e-01 -4.56253216e-02
-3.20370257e-01 -6.54696822e-01 -1.70496300e-01 9.18692052e-01
-1.04491413e+00 1.04080029e-01 1.04288965e-01 7.61007130e-01
1.77632004e-01 -5.86913109e-01 1.17751813e+00 -1.91157505e-01
-2.35821769e-01 6.74737215e-01 -2.92642146e-01 7.69700944e-01
1.29244947e+00 2.95898765e-01 -2.35660613e-01 -5.96070647e-01
-4.91468728e-01 5.35591125e-01 6.24538302e-01 -6.47186935e-02
5.34955919e-01 -1.49290431e+00 -6.45138502e-01 -7.31735006e-02
1.85173422e-01 -2.52189815e-01 2.02629671e-01 3.88339192e-01
-1.49177775e-01 2.16667354e-01 -1.99618042e-01 -7.74360895e-02
-7.42982566e-01 1.05076981e+00 7.11730182e-01 -7.99178839e-01
-3.78386736e-01 6.23792231e-01 4.47576702e-01 -2.67268628e-01
2.14442879e-01 -3.57934356e-01 -1.40506536e-01 -9.66113061e-02
8.72012615e-01 4.41110313e-01 -2.55790830e-01 -1.96300223e-02
-2.74892479e-01 3.79572928e-01 -9.52148959e-02 -6.64552033e-01
1.26450264e+00 -1.33118322e-02 3.52579981e-01 4.14991826e-01
8.32932293e-01 -1.87101424e-01 -1.64842963e+00 1.14340059e-01
1.97508693e-01 -5.32760918e-01 -1.33984476e-01 -7.77866721e-01
-6.54483795e-01 7.00178623e-01 6.27492726e-01 3.75564098e-01
1.17136395e+00 -3.30998302e-01 3.31996173e-01 5.77767313e-01
7.01751649e-01 -1.39628673e+00 8.77889097e-02 2.99846858e-01
8.90452445e-01 -1.18946826e+00 -1.09192915e-02 4.58023995e-02
-1.00168431e+00 8.91542494e-01 4.21346247e-01 -3.03760260e-01
8.76163125e-01 3.91015708e-01 -2.15658620e-01 6.24562651e-02
-1.04764593e+00 -7.70770982e-02 8.32118317e-02 9.15542185e-01
-9.21066478e-02 4.20519829e-01 -1.60547972e-01 7.38983214e-01
-1.62313864e-01 -1.58917278e-01 6.76374614e-01 9.72934484e-01
-5.32609224e-01 -1.32834077e+00 -1.91734672e-01 7.62074709e-01
-6.77110910e-01 8.56076106e-02 2.63721328e-02 7.55477965e-01
-3.50787759e-01 1.07456028e+00 -7.65369758e-02 -4.10857528e-01
2.41477162e-01 1.66704103e-01 5.81349611e-01 -7.43657649e-01
-4.38863486e-01 -1.16797701e-01 1.18076205e-01 -9.57198799e-01
-2.05529273e-01 -6.90363884e-01 -1.12551594e+00 -3.49890471e-01
2.60660529e-01 1.59511238e-01 -1.31962612e-01 9.37239289e-01
-2.22802281e-01 7.05685467e-02 1.15654564e+00 -3.77227068e-01
-1.65500522e+00 -4.69392061e-01 -9.04463232e-01 2.57601440e-01
4.53198284e-01 -9.06091511e-01 -6.57735825e-01 -6.80175960e-01]
|
[4.226290702819824, 2.5526862144470215]
|
243e5f10-5d1f-4f1f-b777-d6bff3c5cf75
|
design-of-novel-algorithm-and-architecture
|
1409.4043
| null |
http://arxiv.org/abs/1409.4043v1
|
http://arxiv.org/pdf/1409.4043v1.pdf
|
Design of Novel Algorithm and Architecture for Gaussian Based Color Image Enhancement System for Real Time Applications
|
This paper presents the development of a new algorithm for Gaussian based
color image enhancement system. The algorithm has been designed into
architecture suitable for FPGA/ASIC implementation. The color image enhancement
is achieved by first convolving an original image with a Gaussian kernel since
Gaussian distribution is a point spread function which smoothen the image.
Further, logarithm-domain processing and gain/offset corrections are employed
in order to enhance and translate pixels into the display range of 0 to 255.
The proposed algorithm not only provides better dynamic range compression and
color rendition effect but also achieves color constancy in an image. The
design exploits high degrees of pipelining and parallel processing to achieve
real time performance. The design has been realized by RTL compliant Verilog
coding and fits into a single FPGA with a gate count utilization of 321,804.
The proposed method is implemented using Xilinx Virtex-II Pro XC2VP40-7FF1148
FPGA device and is capable of processing high resolution color motion pictures
of sizes of up to 1600x1200 pixels at the real time video rate of 116 frames
per second. This shows that the proposed design would work for not only still
images but also for high resolution video sequences.
|
['M. Ravishankar', 'M. C. Hanumantharaju', 'D. R. Rameshbabu']
|
2014-09-14
| null | null | null | null |
['color-constancy']
|
['computer-vision']
|
[ 4.20034826e-01 -2.91910529e-01 2.16462851e-01 -2.93489277e-01
2.66434941e-02 -4.94651288e-01 2.66314119e-01 1.84273332e-01
-6.96781754e-01 5.57428658e-01 -3.14353883e-01 -6.22414529e-01
5.11014834e-02 -7.85508454e-01 -2.73039103e-01 -4.85397846e-01
-4.45894077e-02 -3.30603868e-01 4.90940243e-01 -2.68283649e-03
5.10718286e-01 5.71684182e-01 -1.36756301e+00 1.06721275e-01
6.70063913e-01 9.99682724e-01 4.56533581e-01 1.30842674e+00
4.15710270e-01 8.24600160e-01 -4.09933686e-01 -1.60230592e-01
6.91731870e-01 -2.37334132e-01 -4.96863842e-01 3.32502186e-01
2.49765873e-01 -6.66204631e-01 -2.97933340e-01 1.29729140e+00
4.08010483e-01 -2.00892445e-02 5.92715025e-01 -1.00088668e+00
-4.18838561e-01 1.41014501e-01 -1.19315910e+00 3.55997652e-01
-1.13425501e-01 1.12491623e-01 3.39051962e-01 -5.33054769e-01
3.12480271e-01 1.00387919e+00 3.82979959e-01 -3.70780146e-03
-9.45374608e-01 -5.90864182e-01 -7.99820840e-01 3.01579565e-01
-1.53964972e+00 -6.54670224e-02 5.78801930e-01 -3.00680436e-02
1.08012748e+00 4.14123774e-01 5.56465030e-01 -1.19075872e-01
7.54520118e-01 5.82575947e-02 1.52970028e+00 -5.70777774e-01
6.74230605e-02 4.22133088e-01 2.86133811e-02 7.02674627e-01
6.76906645e-01 5.84964678e-02 -1.17889620e-01 2.20301598e-01
1.32568157e+00 4.19360511e-02 -2.55973607e-01 1.61281645e-01
-9.53494072e-01 5.61887264e-01 3.85093004e-01 3.55109811e-01
-5.03077507e-01 4.75941956e-01 4.74863648e-01 1.59938946e-01
-3.66562694e-01 9.12180915e-02 -2.94461936e-01 -2.05455318e-01
-9.42048132e-01 -1.07724272e-01 1.85541481e-01 1.03661025e+00
5.91991782e-01 4.53319639e-01 3.94439578e-01 3.30399513e-01
5.18096626e-01 6.82986915e-01 4.98890191e-01 -7.73590028e-01
2.57355183e-01 3.54900718e-01 2.20389724e-01 -9.87787545e-01
-2.44513869e-01 -1.66378319e-01 -8.24611843e-01 1.15896142e+00
1.74261034e-01 -4.14835483e-01 -9.94828939e-01 9.04872715e-01
6.17801510e-02 -2.24023059e-01 3.05057079e-01 8.76177132e-01
2.14483529e-01 1.27109659e+00 6.36109263e-02 -1.63049906e-01
1.79422259e+00 -6.85001314e-01 -8.18990469e-01 3.95113230e-02
-6.02376983e-02 -1.32687414e+00 7.71349132e-01 7.58355081e-01
-1.18553019e+00 -1.05069458e+00 -1.61809123e+00 8.55858997e-03
-1.02748178e-01 6.53182566e-01 3.64047259e-01 1.04516256e+00
-1.09369683e+00 4.55158889e-01 -8.32501471e-01 -2.49611944e-01
-2.17119858e-04 6.47425413e-01 -2.73956627e-01 2.65583217e-01
-8.19672108e-01 6.92704916e-01 6.21777952e-01 8.28439817e-02
-1.65106162e-01 -5.96843302e-01 -4.99791324e-01 1.02096431e-01
-2.65975028e-01 -2.34531462e-01 6.70741439e-01 -1.25843859e+00
-1.59421039e+00 5.78749716e-01 1.65845200e-01 -6.62689328e-01
2.56444275e-01 6.17997572e-02 -7.80661643e-01 5.35086691e-01
-4.92716104e-01 5.59125543e-01 8.86226237e-01 -6.16627753e-01
-1.05621243e+00 -3.87163192e-01 -3.68940502e-01 3.10105234e-01
-2.64590114e-01 2.36242980e-01 -2.34312147e-01 -5.45131624e-01
-6.50607049e-02 -6.39425039e-01 -2.81801760e-01 9.74374264e-02
3.58079821e-02 5.88760078e-01 1.26478410e+00 -7.06681371e-01
1.24225092e+00 -2.00151014e+00 -7.40214050e-01 4.96338397e-01
-2.70738959e-01 6.46416545e-01 4.25662696e-01 3.24861139e-01
-4.39223498e-02 -4.20689821e-01 9.32267308e-02 3.43444020e-01
-3.49570364e-01 -2.51330286e-01 1.65493667e-01 7.58041322e-01
-2.84717698e-03 3.00511897e-01 -1.44462988e-01 -5.12470305e-01
6.33789420e-01 1.08628035e+00 -3.90401989e-01 -1.42128021e-01
4.75438565e-01 -4.01984230e-02 -2.06174731e-01 5.12591362e-01
1.31557298e+00 2.24573493e-01 1.62916422e-01 -6.11510873e-01
-5.79231918e-01 -4.75956172e-01 -1.58602965e+00 1.03600597e+00
-3.68941903e-01 1.04531789e+00 3.20111066e-01 -4.96242821e-01
1.35058880e+00 3.50339353e-01 1.69996396e-01 -6.27962053e-01
6.54147267e-01 1.12458460e-01 2.60621514e-02 -4.08243984e-01
9.66428757e-01 -2.25145757e-01 4.04194206e-01 2.68895775e-01
-2.26724401e-01 -6.94911480e-02 2.62107015e-01 -1.34227872e-01
7.34959424e-01 6.30981401e-02 7.04945803e-01 -6.52389586e-01
7.89056242e-01 2.36348778e-01 2.96553552e-01 1.24640197e-01
-4.37131405e-01 1.49386317e-01 -9.13889334e-02 -9.49681923e-02
-1.68006814e+00 -1.02083337e+00 -3.22110772e-01 4.36695307e-01
4.98779446e-01 3.01749539e-02 -5.61104119e-01 1.73890829e-01
-3.66615415e-01 3.96419078e-01 -9.64470059e-02 1.34374335e-01
-6.03987575e-01 -4.98976678e-01 3.11535716e-01 4.13667411e-01
1.05834723e+00 -8.40300620e-01 -1.53546143e+00 3.45906526e-01
7.96192884e-01 -9.36290860e-01 -3.03139091e-01 9.25706774e-02
-1.19467676e+00 -8.35776627e-01 -5.12108624e-01 -1.13050580e+00
8.32853079e-01 3.24448705e-01 5.30427456e-01 -1.77527025e-01
-8.82686019e-01 1.56629711e-01 -3.56237143e-01 -1.82574391e-01
-2.69686013e-01 -7.34667242e-01 -2.55399376e-01 -1.28328159e-01
5.25933623e-01 -2.00214431e-01 -1.16057503e+00 7.32647348e-03
-9.48935926e-01 1.04932085e-01 7.64754772e-01 7.10999966e-01
3.18274856e-01 7.65607238e-01 2.17038944e-01 -6.85485601e-01
3.73042583e-01 1.84973598e-01 -1.16486979e+00 -2.48569865e-02
-7.26018190e-01 -1.69267878e-02 9.00995314e-01 -1.94289982e-02
-1.41467381e+00 4.35119241e-01 -1.41712008e-02 5.49583845e-02
6.02243282e-03 4.85052727e-02 1.55988812e-01 -4.18626696e-01
6.10890687e-01 2.82038987e-01 4.63503063e-01 -8.50457698e-02
1.25390470e-01 1.09708762e+00 1.17654204e+00 9.72924083e-02
6.23506248e-01 5.48833191e-01 3.79711717e-01 -1.00922155e+00
6.13819778e-01 -5.31427860e-01 -1.79705426e-01 -2.89153874e-01
9.98324931e-01 -1.10459185e+00 -9.03827429e-01 4.27224189e-01
-6.54132962e-01 9.95653123e-02 3.85043323e-01 6.67805254e-01
-4.19564009e-01 4.31924462e-01 -8.79191339e-01 -8.43829870e-01
-9.57089663e-01 -1.21532047e+00 3.53661209e-01 9.14460838e-01
-3.62512134e-02 -9.91070271e-01 -6.16233528e-01 -1.02579832e-01
7.19621778e-01 3.87935817e-01 5.73463619e-01 2.16494888e-01
-5.85015714e-01 -4.28373367e-01 -3.91135782e-01 3.93861145e-01
4.01912957e-01 3.42105597e-01 -5.46219587e-01 -3.53333324e-01
2.82581151e-01 2.03826621e-01 4.20761883e-01 5.48515260e-01
4.67204005e-01 3.07813268e-02 -1.05386227e-01 5.72573543e-01
2.66333771e+00 7.93001533e-01 1.10366476e+00 5.21088421e-01
2.99312085e-01 -6.60351198e-03 1.02311981e+00 6.31445765e-01
-7.17732608e-02 3.16721588e-01 3.84724379e-01 -3.82155448e-01
-8.53915140e-02 5.58745712e-02 1.93390518e-01 1.32294223e-01
8.93859714e-02 -1.36945881e-02 -3.89722109e-01 3.69678140e-01
-9.97202277e-01 -6.91649139e-01 -6.66236520e-01 2.32323909e+00
7.19886422e-01 8.58071595e-02 -1.21096805e-01 6.86219990e-01
8.56367946e-01 -3.56061935e-01 2.73232814e-02 -9.76919651e-01
2.31929541e-01 7.11284399e-01 1.25964379e+00 6.94973528e-01
-1.02952516e+00 6.45556092e-01 4.50716829e+00 6.58122599e-01
-1.66511214e+00 -2.26512104e-01 5.40847600e-01 3.01715940e-01
3.57698619e-01 8.31188783e-02 -6.06074333e-01 4.40110326e-01
9.56475139e-01 -2.63848335e-01 1.42803490e-01 8.43628109e-01
4.74294156e-01 -6.51449621e-01 -2.96076953e-01 1.08247817e+00
-2.26914898e-01 -1.04781580e+00 -1.78156376e-01 -8.79365578e-02
6.50525331e-01 -5.41827321e-01 3.73062760e-01 -4.30227190e-01
-2.12949246e-01 -8.19332421e-01 5.68117917e-01 -6.22394979e-02
8.58950257e-01 -1.18334937e+00 6.48779333e-01 -1.33231834e-01
-1.21922600e+00 -2.38600731e-01 -7.67608225e-01 8.51122066e-02
2.01053604e-01 4.11317348e-02 -1.11135209e+00 2.07668841e-01
5.21454215e-01 1.56932756e-01 -4.40235227e-01 1.21444499e+00
3.76128793e-01 3.59183550e-01 -3.53664070e-01 -1.73368473e-02
3.38613689e-01 -4.08435971e-01 2.48570666e-01 1.41913319e+00
7.42303371e-01 3.83249402e-01 -6.71039820e-01 3.49778473e-01
4.62440491e-01 2.71364123e-01 -1.84492797e-01 5.48458286e-02
3.21759582e-01 1.41106021e+00 -1.05924404e+00 -6.03237450e-01
-4.40025926e-01 1.14282072e+00 -5.82402110e-01 3.46204899e-02
-8.73126984e-01 -1.25911605e+00 4.76404756e-01 3.45405489e-01
6.42137468e-01 -4.22383010e-01 -4.47946221e-01 -2.99455702e-01
-3.93359452e-01 -8.07850659e-01 7.15526864e-02 -8.10889482e-01
-2.05148950e-01 8.07599247e-01 -3.11409771e-01 -1.28661871e+00
1.99825503e-02 -1.02079427e+00 -5.42312562e-01 1.13672817e+00
-1.19476044e+00 -9.48660791e-01 -5.83165407e-01 6.56313896e-01
5.20301282e-01 -3.58191192e-01 4.13126826e-01 3.79348367e-01
2.04848056e-03 4.72322494e-01 4.86411512e-01 -2.04543129e-01
7.58569300e-01 -1.18135655e+00 2.18750276e-02 1.36280942e+00
-4.03286487e-01 7.11938560e-01 1.08666420e+00 -6.37072742e-01
-1.60752273e+00 -9.09422934e-01 5.65594018e-01 5.42692363e-01
9.72280651e-02 1.50748547e-02 -5.61045945e-01 2.63226211e-01
7.73211896e-01 -3.82503681e-02 1.40424624e-01 -1.07906270e+00
1.72518134e-01 -4.47856516e-01 -1.68815470e+00 3.82223010e-01
-2.32094973e-01 -7.68891498e-02 5.17979898e-02 -2.16737568e-01
4.73789731e-03 -5.14814317e-01 -8.45966220e-01 -1.06286183e-02
3.76359969e-01 -1.21922946e+00 7.05847740e-01 3.23587596e-01
3.01455081e-01 -9.36057925e-01 -1.03550576e-01 -6.14861071e-01
-2.25751683e-01 -8.39857042e-01 6.70049965e-01 1.14826679e+00
2.61970550e-01 -4.17100012e-01 7.12108374e-01 4.72773552e-01
1.18063152e-01 -4.41794872e-01 -5.67245781e-01 -4.03594494e-01
-4.83977705e-01 4.31593880e-02 2.18263790e-01 2.68329531e-01
1.35817334e-01 -1.87603906e-02 -4.86587018e-01 4.43889409e-01
7.74117649e-01 -2.29803413e-01 4.61237848e-01 -5.54976344e-01
-3.74697953e-01 -8.70846137e-02 -9.76021588e-01 -6.46938562e-01
-8.39435101e-01 -1.62801102e-01 -3.57228547e-01 -1.21828830e+00
6.69556484e-02 -2.35012412e-01 -4.70939986e-02 -4.16755006e-02
-8.35387327e-05 7.53817201e-01 1.03418499e-01 -2.10783631e-01
-2.44142301e-02 -2.33061895e-01 9.26383197e-01 2.60189176e-01
-1.50640577e-01 1.91098135e-02 -4.99435425e-01 4.72178131e-01
7.76903450e-01 1.08611055e-01 -7.69421577e-01 -5.25561757e-02
-2.22619399e-02 1.85792878e-01 2.87036151e-01 -1.30199957e+00
2.85206586e-01 1.25588313e-01 8.11434984e-01 -6.52716994e-01
2.95659065e-01 -1.38362956e+00 5.81912518e-01 8.82524729e-01
1.14852443e-01 6.63546383e-01 2.53554136e-01 3.06944221e-01
-3.39654446e-01 -4.15267318e-01 1.34690750e+00 1.58471093e-01
-1.19267046e+00 -2.17400357e-01 -6.56812608e-01 -7.93934107e-01
1.55996168e+00 -8.31394017e-01 -2.80347496e-01 -3.00525486e-01
-3.05541396e-01 -2.82657981e-01 7.39789069e-01 -9.72127914e-03
7.68606663e-01 -1.08272910e+00 -4.98496413e-01 4.20654505e-01
-3.94611001e-01 -6.73766434e-01 5.54616690e-01 5.26178658e-01
-1.90025127e+00 5.50378442e-01 -1.01805890e+00 -3.09084654e-01
-1.99906600e+00 5.08221984e-01 1.46054462e-01 2.52468169e-01
-7.74312258e-01 7.13865638e-01 -3.44013393e-01 8.01295161e-01
-2.20331863e-01 -3.06297183e-01 -2.65817881e-01 -5.69927931e-01
7.57858455e-01 6.00994945e-01 -2.09705442e-01 -5.48407793e-01
-2.28770718e-01 7.89247990e-01 2.61211563e-02 -3.95838916e-01
1.04875791e+00 -4.41165686e-01 -1.28619581e-01 -3.81832808e-01
1.42372549e+00 3.13657045e-01 -1.33644009e+00 3.39918107e-01
-1.26144737e-01 -9.10630643e-01 3.06738138e-01 -7.89253771e-01
-1.01398838e+00 5.89310467e-01 1.22910619e+00 -2.00548112e-01
1.66332459e+00 -8.94883275e-01 5.70910156e-01 -2.38160819e-01
1.73532054e-01 -1.34757936e+00 -2.09669068e-01 -4.78791296e-02
2.57089823e-01 -1.04781902e+00 4.75541979e-01 -2.50066340e-01
-9.83477175e-01 1.63665617e+00 4.98414129e-01 -5.06409824e-01
4.84445781e-01 6.43249929e-01 1.69843361e-01 1.85377091e-01
-4.01205003e-01 1.31436437e-03 -1.44137800e-01 6.64834619e-01
6.36484027e-01 1.31212339e-01 -7.66027212e-01 -6.98167458e-02
-1.87413901e-01 2.23788530e-01 1.13072395e+00 1.06990242e+00
-7.68779159e-01 -8.83883119e-01 -7.41490006e-01 1.56433761e-01
-1.08648217e+00 -7.60104209e-02 4.61025089e-01 8.90270591e-01
7.49421269e-02 1.06109440e+00 3.81505489e-01 -2.61186540e-01
-1.79871768e-01 -4.98313725e-01 5.83291233e-01 1.40407950e-01
-6.27612233e-01 4.38031644e-01 -7.50328824e-02 -2.47686461e-01
3.93623300e-02 -4.11132962e-01 -1.60050869e+00 -4.41700459e-01
6.07387088e-02 1.85209990e-01 1.29487193e+00 2.40859091e-01
1.23166278e-01 5.52639604e-01 5.76986372e-01 -2.59183854e-01
-3.39300960e-01 -8.43144059e-01 -8.32120895e-01 2.49718428e-02
2.37611115e-01 -6.92960992e-02 1.33027121e-01 5.49848437e-01]
|
[9.625669479370117, -2.0247974395751953]
|
5cda59aa-6681-4a90-a768-88818f977bb8
|
a-convolutional-transformer-network-for-crack
|
2302.11728
| null |
https://arxiv.org/abs/2302.11728v1
|
https://arxiv.org/pdf/2302.11728v1.pdf
|
A Convolutional-Transformer Network for Crack Segmentation with Boundary Awareness
|
Cracks play a crucial role in assessing the safety and durability of manufactured buildings. However, the long and sharp topological features and complex background of cracks make the task of crack segmentation extremely challenging. In this paper, we propose a novel convolutional-transformer network based on encoder-decoder architecture to solve this challenge. Particularly, we designed a Dilated Residual Block (DRB) and a Boundary Awareness Module (BAM). The DRB pays attention to the local detail of cracks and adjusts the feature dimension for other blocks as needed. And the BAM learns the boundary features from the dilated crack label. Furthermore, the DRB is combined with a lightweight transformer that captures global information to serve as an effective encoder. Experimental results show that the proposed network performs better than state-of-the-art algorithms on two typical datasets. Datasets, code, and trained models are available for research at https://github.com/HqiTao/CT-crackseg.
|
['Hong Zhang', 'Jinqiang Cui', 'Bingxi Liu', 'Huaqi Tao']
|
2023-02-23
| null | null | null | null |
['crack-segmentation']
|
['computer-vision']
|
[-1.05844788e-01 -1.34009838e-01 -9.57882032e-02 -3.51313204e-01
-7.41627395e-01 -1.16567984e-01 2.86202990e-02 1.14320405e-02
1.10134427e-02 2.35594064e-01 2.14135215e-01 -1.10474296e-01
1.65749207e-01 -1.13488102e+00 -8.09869051e-01 -8.23550165e-01
1.54806674e-01 1.03659458e-01 7.28167832e-01 -2.07921386e-01
5.29950261e-01 1.77329034e-01 -1.34660602e+00 4.00477707e-01
8.33093405e-01 1.31518769e+00 3.53902578e-01 2.76688963e-01
1.92497015e-01 6.34777725e-01 -7.52789751e-02 -5.89139797e-02
-2.81180218e-02 3.21968757e-02 -6.92309439e-01 3.60905752e-02
1.45842373e-01 -6.21290147e-01 -5.67960024e-01 6.46173894e-01
8.33611608e-01 -1.37822807e-01 5.03307164e-01 -8.32037032e-01
-5.85995495e-01 6.37869954e-01 -6.23445630e-01 2.61964172e-01
1.90183118e-01 1.86424151e-01 1.01737201e+00 -1.22784281e+00
2.42976084e-01 8.00562918e-01 8.48543227e-01 4.09071356e-01
-7.24004984e-01 -7.99301326e-01 6.42436370e-02 3.37797284e-01
-1.24053955e+00 -3.26368004e-01 1.26223767e+00 -5.95124125e-01
6.09116554e-01 4.91840504e-02 6.26208901e-01 8.32303107e-01
1.98442787e-01 6.24075890e-01 5.62477410e-01 -1.23078607e-01
6.61404878e-02 -3.54155272e-01 1.39779955e-01 1.21166301e+00
1.42158747e-01 -3.07582021e-02 -1.94682628e-01 3.81026208e-01
9.78621781e-01 9.29698050e-02 -3.64455074e-01 -3.45072985e-01
-9.22052801e-01 7.15457916e-01 7.15761781e-01 2.58866012e-01
-1.87119484e-01 5.65954208e-01 4.21578199e-01 -1.85735002e-02
3.44048440e-01 1.44071177e-01 -2.71093667e-01 6.71762973e-02
-6.68404579e-01 7.67272785e-02 2.77707815e-01 7.95537412e-01
8.94813359e-01 -1.36503607e-01 -6.70472309e-02 1.09461367e+00
4.71851379e-01 2.51947403e-01 6.76297396e-02 -9.53074038e-01
7.31266737e-01 8.46042097e-01 -3.37422132e-01 -1.07597613e+00
-5.27334332e-01 -4.29599345e-01 -8.55874419e-01 1.27643207e-02
2.45715436e-02 -3.28201167e-02 -1.08324158e+00 1.15180802e+00
4.36524481e-01 -1.96793955e-02 -5.61064243e-01 9.23338413e-01
9.95576739e-01 7.01015353e-01 -2.80619234e-01 2.97757089e-01
1.32275701e+00 -1.18995523e+00 -4.78957266e-01 -3.91384155e-01
5.05121589e-01 -7.09413588e-01 1.24237466e+00 2.63840616e-01
-1.09875000e+00 -4.83948797e-01 -1.14989913e+00 -2.12828800e-01
-8.50861743e-02 5.28957367e-01 2.51202732e-01 1.52350992e-01
-7.37266839e-01 4.34098154e-01 -9.46315527e-01 -7.39883110e-02
7.60695755e-01 2.75664151e-01 -8.67603198e-02 -2.86662251e-01
-1.22224963e+00 5.71721554e-01 1.98132604e-01 6.91452026e-01
-1.16806245e+00 -5.20808280e-01 -9.40704823e-01 -3.45748030e-02
2.98186779e-01 -2.28378728e-01 1.19006562e+00 -1.33117884e-01
-1.24295676e+00 6.02978945e-01 2.21463144e-01 1.50823206e-01
2.95707524e-01 -2.92769670e-01 -5.45468256e-02 4.26369458e-01
1.25917017e-01 4.06246483e-01 7.32103109e-01 -1.37957275e+00
-4.23040181e-01 -1.22483052e-01 1.94366410e-01 -1.01524614e-01
-2.59081870e-01 -2.40366623e-01 -5.05233824e-01 -8.06930602e-01
3.13292086e-01 -5.73641956e-01 -2.63496280e-01 1.98680311e-01
-5.60827017e-01 -1.50760219e-01 9.32438612e-01 -9.98205304e-01
1.71066892e+00 -2.21278739e+00 6.50260895e-02 1.95681691e-01
2.50233680e-01 3.86044830e-02 1.57693967e-01 5.77814937e-01
-1.62341282e-01 1.39822558e-01 -7.90013790e-01 -2.69297093e-01
-2.12441280e-01 1.36268020e-01 1.40337750e-01 4.84074056e-01
3.81163716e-01 6.33487642e-01 -6.52241290e-01 -8.43769550e-01
2.17929650e-02 3.41747820e-01 -5.54038703e-01 1.62008360e-01
-4.82294038e-02 4.24350053e-01 -6.72138989e-01 9.28127646e-01
7.86083341e-01 -2.52917588e-01 -3.14093143e-01 -4.29552466e-01
-3.51583004e-01 3.18431139e-01 -1.04179227e+00 1.72209156e+00
-4.19652104e-01 3.87699217e-01 1.48588195e-01 -9.98700261e-01
9.82464135e-01 4.47751611e-01 3.99573386e-01 -7.92316198e-01
3.73614311e-01 3.51192772e-01 -3.69982600e-01 -9.59573030e-01
1.31107628e-01 4.50288057e-02 -1.50942788e-01 3.29871982e-01
-4.30545747e-01 -2.50346184e-01 1.23628840e-01 4.14347462e-02
1.23747718e+00 1.98194087e-01 -4.70849872e-01 -1.17379390e-01
6.03018522e-01 -1.58141255e-01 7.96237409e-01 -6.91955388e-02
-3.84504274e-02 9.97344494e-01 3.86899859e-01 -5.99071801e-01
-1.10805225e+00 -9.65060651e-01 -1.34392202e-01 4.58606988e-01
4.48632926e-01 -3.68806273e-01 -8.57179821e-01 -6.20650649e-01
-1.04516350e-01 1.61842182e-01 -6.64882958e-01 -2.86296755e-01
-9.52455044e-01 -3.92006427e-01 3.65967125e-01 1.02695215e+00
8.60987604e-01 -9.48001266e-01 -6.99879467e-01 1.95006028e-01
-4.93906200e-01 -8.22641075e-01 -7.32610822e-01 6.00788184e-02
-8.16907823e-01 -1.24434936e+00 -4.97131616e-01 -1.19075954e+00
8.91135991e-01 1.72546819e-01 7.10284114e-01 6.91569507e-01
-3.86683464e-01 -7.77455494e-02 -5.90408027e-01 1.25239402e-01
-6.94505349e-02 3.29577476e-01 -6.49318993e-01 -6.45198300e-02
-3.09249789e-01 -5.40033042e-01 -1.02845788e+00 5.75279355e-01
-8.80296111e-01 2.96013415e-01 7.13746190e-01 7.82082200e-01
5.93292236e-01 4.55802888e-01 4.06614095e-01 -7.12045670e-01
3.22798491e-01 -4.41460878e-01 -3.30143154e-01 -1.93160139e-02
-6.64002120e-01 -5.85545972e-02 4.39852774e-01 -7.03795701e-02
-1.11137235e+00 9.23108608e-02 -5.83525002e-01 -1.40079737e-01
1.59540981e-01 6.99174702e-01 -3.06974292e-01 5.17822430e-02
2.63708293e-01 -3.11006419e-02 -2.29989171e-01 -7.19217598e-01
-7.85193220e-02 9.22492385e-01 2.85862118e-01 -7.62231469e-01
9.20909524e-01 4.24269259e-01 -2.54223436e-01 -5.90296686e-01
-8.42604816e-01 -3.94874185e-01 -7.21125662e-01 -6.64749384e-01
9.43750918e-01 -8.95518303e-01 -3.63839298e-01 9.76133525e-01
-1.11598170e+00 -5.66885591e-01 -1.72093004e-01 1.25003859e-01
-3.32324892e-01 3.45831096e-01 -1.04550588e+00 -4.73463476e-01
-5.09035528e-01 -1.30656171e+00 1.12416887e+00 3.36145729e-01
2.70159006e-01 -6.05272412e-01 -1.95175149e-02 7.69990265e-01
2.79947370e-01 4.60891336e-01 9.36904192e-01 3.21296930e-01
-7.87405133e-01 -2.66559064e-01 -2.70418614e-01 5.65109968e-01
1.30200475e-01 3.37489396e-01 -8.45321715e-01 -1.58554092e-01
-1.41559139e-01 -3.15547287e-01 1.14813185e+00 1.73680469e-01
1.25342894e+00 -1.04247987e-01 -4.94424194e-01 3.52483988e-01
1.50400734e+00 5.66264726e-02 8.91127229e-01 2.84576625e-01
9.88118827e-01 4.20194328e-01 7.10097611e-01 4.40424830e-01
6.70099974e-01 5.06770730e-01 7.60247588e-01 -1.59159929e-01
-2.33067125e-01 -2.15659529e-01 3.11595142e-01 1.18311083e+00
-1.82314813e-01 3.98277529e-02 -1.21548796e+00 7.90486038e-01
-1.69251573e+00 -7.70793915e-01 -5.47604442e-01 1.76699221e+00
9.94911551e-01 4.18179870e-01 -1.63079679e-01 5.44357240e-01
7.50406563e-01 2.90802389e-01 -3.05350453e-01 -2.83329189e-01
1.69382453e-01 1.16991594e-01 3.62685889e-01 2.33073816e-01
-1.14677739e+00 6.22830749e-01 5.35820961e+00 8.47811997e-01
-9.36636090e-01 2.25869179e-01 5.78088939e-01 2.86016077e-01
-3.77293468e-01 1.18979804e-01 -5.31004846e-01 6.54049277e-01
4.21344578e-01 5.90281188e-01 1.78045303e-01 5.98517001e-01
1.45050094e-01 -1.21577084e-01 -7.45922089e-01 5.89605689e-01
-1.28608048e-01 -1.29345405e+00 -4.19802666e-01 -7.64654726e-02
4.57190424e-01 4.40095477e-02 -2.04287842e-02 -1.03484228e-01
-1.29074484e-01 -6.59309924e-01 1.11016762e+00 5.06285369e-01
9.24044788e-01 -6.40485406e-01 7.17253983e-01 1.18298151e-01
-1.72034383e+00 -4.89580750e-01 -3.05583805e-01 1.18661784e-02
1.37023315e-01 8.64259839e-01 -2.89307952e-01 6.57965779e-01
1.09320867e+00 1.09244967e+00 -5.93004107e-01 1.00634670e+00
-4.34666574e-01 7.94972718e-01 -2.84669638e-01 5.29904544e-01
1.33450940e-01 -1.51804779e-02 6.16553053e-02 1.04778945e+00
3.72344375e-01 1.16847426e-01 9.66715515e-02 6.99277818e-01
-1.91181824e-01 -1.40072450e-01 -2.54450321e-01 2.49528706e-01
5.98571479e-01 1.31545234e+00 -8.87929380e-01 1.05480917e-01
-5.06882906e-01 5.85101902e-01 3.42560410e-01 -3.15141007e-02
-1.17320633e+00 -4.40256596e-01 3.70778412e-01 5.56087017e-01
6.15399301e-01 -1.91656649e-01 -4.68293011e-01 -9.61200416e-01
2.92200059e-01 -5.46982348e-01 2.52403736e-01 -8.35703373e-01
-1.06782043e+00 3.80837083e-01 -2.42580429e-01 -1.30979443e+00
6.77311540e-01 -3.29390734e-01 -9.15023386e-01 3.54463130e-01
-1.50795567e+00 -1.40085340e+00 -6.24116182e-01 2.38718897e-01
7.47332752e-01 5.00978410e-01 2.68011123e-01 8.64240885e-01
-1.13943315e+00 4.71643299e-01 -1.14292145e-01 6.12577677e-01
5.00290692e-01 -9.48553801e-01 3.87864798e-01 9.11222816e-01
-5.99941850e-01 1.70523927e-01 3.79723728e-01 -7.80678749e-01
-1.10180008e+00 -1.18609369e+00 5.15718639e-01 -1.27634406e-01
3.94061148e-01 -3.33814174e-01 -9.81129110e-01 4.57455635e-01
1.12077661e-01 7.32128620e-02 2.18332142e-01 -3.31628293e-01
-2.23258391e-01 -3.82486820e-01 -8.12450886e-01 6.72868192e-02
1.04960310e+00 -4.87514317e-01 -3.21426690e-01 9.78735015e-02
7.15437114e-01 -6.96292222e-01 -1.08193219e+00 4.97682780e-01
6.32760525e-01 -1.00205469e+00 8.92053783e-01 2.73979932e-01
1.03676033e+00 -4.58127230e-01 -1.63717568e-02 -9.82337654e-01
-4.81104702e-01 -8.55771974e-02 -4.36109491e-02 1.37150443e+00
4.33528453e-01 -4.02739733e-01 7.81318545e-01 3.34973216e-01
-8.66704702e-01 -1.54323578e+00 -8.59147787e-01 -2.97158062e-01
1.75272971e-02 -2.79198825e-01 7.54250288e-01 6.40983164e-01
-1.92049295e-01 2.23839864e-01 -3.76032032e-02 4.03297454e-01
4.79550391e-01 2.05416948e-01 2.99697161e-01 -1.30915284e+00
1.12129681e-01 -1.95147976e-01 -2.21542552e-01 -1.18707573e+00
-2.29630068e-01 -5.84181786e-01 3.49981934e-01 -1.91153860e+00
8.94106776e-02 -8.27965617e-01 -2.34582722e-01 6.67662561e-01
-1.12341680e-01 2.97983050e-01 -1.46901235e-01 1.97099894e-01
-3.81144404e-01 6.43178403e-01 1.59501672e+00 -4.04758483e-01
1.61248833e-01 -3.88031974e-02 -4.78170156e-01 6.07080877e-01
1.05873132e+00 -6.29072130e-01 -2.17747658e-01 -8.39513063e-01
4.74977165e-01 -3.62920389e-03 5.55463135e-01 -1.21994495e+00
3.77483904e-01 1.31028116e-01 2.40938425e-01 -9.10162687e-01
2.36591235e-01 -9.47501004e-01 1.11938044e-01 7.17234850e-01
-9.57711637e-02 1.31616324e-01 -6.53278530e-02 4.83621299e-01
-2.28412181e-01 -3.17397505e-01 8.91660869e-01 -9.31706131e-02
-4.04462934e-01 4.59825188e-01 -2.57279694e-01 6.56109378e-02
1.06554759e+00 -3.94101590e-01 -4.91661042e-01 4.23873961e-03
-4.20101136e-01 5.87162018e-01 6.99140489e-01 2.93270141e-01
1.09192288e+00 -1.29775894e+00 -5.40045798e-01 3.72464657e-01
4.50162403e-02 7.73836136e-01 5.31163156e-01 9.16646004e-01
-1.03639340e+00 -9.55785885e-02 -1.43865719e-01 -4.90102649e-01
-9.84660685e-01 3.64638418e-01 5.29665411e-01 -1.61919996e-01
-7.88695216e-01 1.03266144e+00 8.57253373e-02 -2.11596683e-01
1.32424533e-01 -6.94562018e-01 -2.92631477e-01 1.70085207e-01
4.03459743e-02 5.94507277e-01 1.33816615e-01 -5.86946905e-01
-3.02181482e-01 1.10818481e+00 1.51057346e-02 4.35496360e-01
1.62482548e+00 -3.04134458e-01 -3.40976983e-01 3.15547407e-01
1.07115602e+00 -2.27540106e-01 -1.54786646e+00 -1.47476450e-01
-1.25214413e-01 -2.93489903e-01 1.60882041e-01 -4.09381181e-01
-1.83146977e+00 1.13602722e+00 5.10753095e-01 1.11609280e-01
1.41739047e+00 4.68229391e-02 1.55185640e+00 -1.68664977e-01
2.92827517e-01 -1.16907978e+00 4.33512747e-01 3.75073016e-01
8.30208957e-01 -9.70350206e-01 5.98229542e-02 -6.85092568e-01
-3.29815209e-01 1.07584953e+00 7.90292025e-01 -2.58892149e-01
9.84494150e-01 4.09073830e-01 -3.46258469e-02 -6.69559896e-01
-5.91259599e-01 -2.64746808e-02 -8.79658759e-02 3.05735201e-01
1.77029222e-01 -3.00934434e-01 -2.80994207e-01 6.23070300e-01
1.95605680e-01 -7.02679232e-02 4.45048571e-01 1.26297235e+00
-7.02559233e-01 -1.05423713e+00 -2.23476052e-01 4.09129500e-01
-3.45993251e-01 -8.58856644e-03 3.07538398e-02 5.54129183e-01
5.08214056e-01 8.14961493e-01 -1.40438631e-01 -7.88707137e-01
3.93186271e-01 -3.42632979e-01 2.12589338e-01 -6.62491381e-01
-6.26975119e-01 2.53801681e-02 1.23156741e-01 -6.48625135e-01
-3.05436939e-01 -5.96434116e-01 -1.63784957e+00 -6.32059798e-02
-6.43815219e-01 1.75522208e-01 4.40673113e-01 7.43340373e-01
2.38856971e-01 7.89763451e-01 1.09270978e+00 -7.33862638e-01
-2.17833057e-01 -8.65349352e-01 -4.42814916e-01 1.41174018e-01
3.87470514e-01 -9.32698667e-01 -3.14589083e-01 3.30625236e-01]
|
[7.5450568199157715, 1.4479886293411255]
|
ede7dd3e-f852-4fcf-a7db-bc9b9296cf01
|
few-shot-bioacoustic-event-detection-at-the-1
|
2306.09223
| null |
https://arxiv.org/abs/2306.09223v1
|
https://arxiv.org/pdf/2306.09223v1.pdf
|
Few-shot bioacoustic event detection at the DCASE 2023 challenge
|
Few-shot bioacoustic event detection consists in detecting sound events of specified types, in varying soundscapes, while having access to only a few examples of the class of interest. This task ran as part of the DCASE challenge for the third time this year with an evaluation set expanded to include new animal species, and a new rule: ensemble models were no longer allowed. The 2023 few shot task received submissions from 6 different teams with F-scores reaching as high as 63% on the evaluation set. Here we describe the task, focusing on describing the elements that differed from previous years. We also take a look back at past editions to describe how the task has evolved. Not only have the F-score results steadily improved (40% to 60% to 63%), but the type of systems proposed have also become more complex. Sound event detection systems are no longer simple variations of the baselines provided: multiple few-shot learning methodologies are still strong contenders for the task.
|
['Dan Stowell', 'Vincent Lostanlen', 'Hanna Pamuła', 'Lisa Gill', 'Ariana Strandburg-Peshkin', 'Joe Morford', 'Ivan Kiskin', 'Frants Jensen', 'Michael Emmerson', 'Emily Grout', 'Helen Whitehead', 'Ester Vidaña-Vila', 'Shubhr Singh', 'Burooj Ghani', 'Ines Nolasco']
|
2023-06-15
| null | null | null | null |
['sound-event-detection']
|
['audio']
|
[ 1.71545178e-01 -2.12091744e-01 5.64209819e-01 -1.63254872e-01
-1.26445699e+00 -6.83143318e-01 4.89401430e-01 6.91482425e-02
-7.05068707e-01 6.03271246e-01 4.63191837e-01 1.98418677e-01
-1.17003851e-01 -1.94334373e-01 -3.22446525e-01 -4.98422146e-01
-5.87979555e-01 6.65708333e-02 8.24062824e-01 -4.16792601e-01
1.42964557e-01 5.99934906e-02 -2.03015375e+00 4.69350100e-01
8.77047330e-02 7.35329866e-01 1.43009901e-01 1.26127028e+00
1.16916597e-01 6.88139498e-01 -1.06894779e+00 -2.70958811e-01
-1.19236141e-01 -3.76765013e-01 -6.33366108e-01 -7.18845904e-01
5.95025659e-01 5.82598150e-02 -3.99979874e-02 7.34320343e-01
1.17648244e+00 5.03428578e-01 2.85178423e-01 -1.27890527e+00
-1.07278891e-01 8.29000652e-01 -1.33484051e-01 7.72660732e-01
5.95682979e-01 3.04457307e-01 1.27778721e+00 -9.14969265e-01
4.28801864e-01 8.33771348e-01 1.01635313e+00 7.75772035e-01
-1.02724612e+00 -9.31312501e-01 8.00637249e-03 5.04431725e-01
-1.03046346e+00 -8.52275908e-01 3.63829583e-01 -4.77365673e-01
1.52315295e+00 4.30424750e-01 5.70851803e-01 1.37536216e+00
-2.03746781e-01 4.61285025e-01 7.74585485e-01 -4.65711266e-01
5.30072033e-01 -1.46476537e-01 1.39678091e-01 1.26018539e-01
2.31229365e-02 3.57758492e-01 -1.06306243e+00 -3.41647565e-01
1.62467770e-02 -4.59252536e-01 -3.05059582e-01 3.66998404e-01
-1.19297254e+00 5.34165561e-01 -2.10613698e-01 6.46791399e-01
-3.03699046e-01 -5.05538210e-02 8.66710067e-01 3.89979541e-01
5.12650371e-01 8.17153573e-01 -6.39352679e-01 -9.21065032e-01
-1.01217997e+00 4.18997586e-01 1.15106869e+00 6.68194950e-01
2.66060144e-01 4.52001870e-01 -1.45684227e-01 1.10489595e+00
-1.47313446e-01 2.38641754e-01 5.79481065e-01 -9.59471107e-01
2.29731098e-01 -3.88653964e-01 1.69404104e-01 -5.47650754e-01
-5.41450977e-01 -3.03398192e-01 -2.69370914e-01 2.49882087e-01
4.08881068e-01 -6.39989197e-01 -9.23124433e-01 1.78966653e+00
4.56126332e-02 6.15554035e-01 -2.84263998e-01 6.50753558e-01
1.15113842e+00 9.25224364e-01 3.69368821e-01 -2.61988759e-01
1.43866777e+00 -5.59591174e-01 -7.31086969e-01 -2.44932845e-01
1.80923924e-01 -9.83282745e-01 9.77543712e-01 7.12596834e-01
-1.19982386e+00 -7.01261044e-01 -1.32623589e+00 4.41449076e-01
-4.84008551e-01 -5.94685435e-01 3.61778080e-01 8.18280160e-01
-9.50254261e-01 5.75886846e-01 -7.90184200e-01 -4.49917912e-01
1.00661732e-01 1.33342540e-03 -9.81216803e-02 4.94610280e-01
-1.52052879e+00 9.24289465e-01 1.68698162e-01 -3.67009223e-01
-9.76959050e-01 -1.25085986e+00 -6.28118634e-01 1.35718554e-01
3.37871730e-01 -9.95450094e-02 1.93664742e+00 -3.08847100e-01
-1.52641571e+00 7.85069585e-01 2.21016929e-01 -5.34658909e-01
2.77117163e-01 -4.02797788e-01 -1.01669836e+00 6.08130172e-02
1.05734535e-01 2.68788964e-01 4.30882931e-01 -8.19045365e-01
-1.23516154e+00 2.09636420e-01 -1.72294915e-01 -1.47220984e-01
-1.28681228e-01 8.47832501e-01 4.05106917e-02 -6.88986063e-01
-4.51266497e-01 -6.94486141e-01 -1.30455613e-01 -3.14393073e-01
1.07367747e-01 -2.97600269e-01 4.88272190e-01 -4.25364405e-01
1.35053837e+00 -2.35642171e+00 -2.67637670e-01 -4.30546284e-01
2.72924267e-02 3.67323697e-01 -1.76526561e-01 7.22545445e-01
-3.03997129e-01 7.64324740e-02 -3.58714163e-01 -3.37507427e-01
2.32625544e-01 -8.45701545e-02 -5.89696765e-01 6.87453672e-02
1.91037312e-01 2.64695048e-01 -1.21156633e+00 -1.96243286e-01
3.20929959e-02 1.88210189e-01 -2.76514918e-01 1.99773297e-01
6.56816736e-02 -2.50513833e-02 1.42961949e-01 3.56402427e-01
3.35229009e-01 2.04588696e-01 -3.48524272e-01 1.26516476e-01
-6.00424826e-01 5.74546278e-01 -1.39741790e+00 1.74292600e+00
-3.34937930e-01 8.07962894e-01 2.24464536e-01 -6.65600300e-01
5.46400130e-01 8.67459893e-01 6.64319277e-01 -2.90872574e-01
-1.70948669e-01 4.11346465e-01 5.79160392e-01 -7.36512303e-01
4.13909137e-01 -4.64738429e-01 -3.61275762e-01 3.29642504e-01
5.00938416e-01 -3.67766351e-01 4.50357467e-01 8.92217830e-02
1.39211977e+00 5.15808500e-02 3.77177596e-01 -2.27479637e-02
4.97158654e-02 -2.15930473e-02 6.22212112e-01 1.01393127e+00
-5.96208096e-01 8.95559669e-01 1.01456091e-01 -2.85827309e-01
-8.54184330e-01 -1.16038883e+00 -2.92080015e-01 1.58399367e+00
-4.15964395e-01 -7.58360326e-01 -6.04086220e-01 -1.75091088e-01
-2.61762649e-01 8.85812044e-01 -5.34452856e-01 -1.39316827e-01
-6.21785700e-01 -9.01830494e-01 1.09930134e+00 4.96749014e-01
4.47714701e-02 -1.47487056e+00 -1.16299641e+00 5.82656085e-01
-2.10036963e-01 -1.05334055e+00 -2.03865066e-01 7.64899254e-01
-2.64962941e-01 -6.80424988e-01 -8.42440009e-01 -7.02087462e-01
-5.39569557e-01 -1.73935682e-01 1.26388955e+00 -4.72653031e-01
-7.03183770e-01 4.81747091e-01 -7.32760131e-01 -1.22726345e+00
-4.81567144e-01 -3.21443416e-02 1.92036539e-01 -3.01021188e-01
4.54579592e-01 -8.43661785e-01 -5.28685451e-01 1.47523850e-01
-6.17237568e-01 -6.52796209e-01 1.18857361e-01 6.03508413e-01
9.06529725e-02 -4.43169028e-01 1.29885375e+00 -6.90569341e-01
6.10833943e-01 -5.44266522e-01 -1.17462754e-01 -1.20144442e-01
-2.31631875e-01 -5.69879055e-01 4.01471525e-01 -6.31439507e-01
-8.87902200e-01 -1.14074923e-01 -6.01848304e-01 -2.12766081e-01
-4.69148904e-01 1.02933981e-01 2.14688361e-01 3.30714852e-01
1.11821854e+00 -1.70347299e-02 -3.53498578e-01 -6.46651328e-01
1.70613676e-01 7.28223920e-01 6.92312002e-01 -4.20944244e-01
4.76828396e-01 1.15114175e-01 -5.05008399e-01 -1.02679276e+00
-6.53503120e-01 -6.17151916e-01 -2.66107380e-01 -3.71604085e-01
7.88484812e-01 -7.90621042e-01 -3.51569563e-01 5.25993824e-01
-1.11653054e+00 -3.16617638e-01 -7.26073325e-01 7.63402760e-01
-3.65900666e-01 7.73086026e-02 -5.89493036e-01 -9.99557197e-01
-4.32541013e-01 -5.38886666e-01 8.71512711e-01 1.77217364e-01
-5.89874625e-01 -5.42118669e-01 7.90575385e-01 -2.36674666e-01
6.05166972e-01 1.85314223e-01 4.21071291e-01 -1.18654108e+00
3.44467014e-01 -3.44691932e-01 4.15373385e-01 3.01293463e-01
1.37866944e-01 2.29852796e-01 -1.77071512e+00 -2.14357749e-02
7.41977468e-02 -2.57174581e-01 9.86380339e-01 4.00242537e-01
8.44880521e-01 2.36510411e-01 -1.72992975e-01 1.76435903e-01
7.64456868e-01 6.00639045e-01 3.48188132e-01 2.15946883e-01
-6.94992468e-02 6.53097630e-01 5.41015804e-01 6.00434005e-01
3.77589017e-02 6.11725807e-01 1.49222404e-01 2.66971737e-01
-3.81380707e-01 -4.98189926e-02 6.01152837e-01 1.07671070e+00
-1.60955817e-01 -3.05568337e-01 -9.89791512e-01 7.98102260e-01
-1.50129128e+00 -1.46017742e+00 -1.04946144e-01 2.22681260e+00
9.65906262e-01 3.68388236e-01 5.47559917e-01 3.41894120e-01
8.00230980e-01 3.32815409e-01 -3.40479344e-01 -5.26921153e-01
8.11588094e-02 7.64778435e-01 -1.51045069e-01 2.07919434e-01
-1.28363228e+00 6.48609698e-01 7.72191525e+00 7.72905290e-01
-1.04001725e+00 2.86715567e-01 1.32471202e-02 -6.06304526e-01
2.50512540e-01 -1.25127718e-01 -7.78083026e-01 5.75749576e-01
1.53824866e+00 -4.08252627e-01 5.14477640e-02 5.44459462e-01
1.33725265e-02 -5.89708909e-02 -1.08217049e+00 9.33745682e-01
1.06564380e-01 -1.02457380e+00 -5.97850740e-01 -2.75524318e-01
5.04240513e-01 5.83449781e-01 -2.10853085e-01 7.90077507e-01
3.65644634e-01 -6.78311944e-01 9.33389544e-01 4.61561620e-01
8.60189915e-01 -6.01324439e-01 4.30812091e-01 3.30531031e-01
-1.26731110e+00 -8.62230211e-02 -2.61053503e-01 -3.69924933e-01
4.58046675e-01 2.85518646e-01 -8.15814376e-01 3.09451371e-01
1.19177353e+00 3.43486458e-01 -3.67281288e-01 1.73112190e+00
-4.52128006e-03 1.15002394e+00 -5.95263600e-01 -3.90790612e-01
6.64793849e-02 6.72993064e-01 1.16459715e+00 1.75249374e+00
4.67667162e-01 1.39710635e-01 2.31888760e-02 3.37514699e-01
1.26269773e-01 9.98626798e-02 -4.99819666e-01 2.04039216e-01
6.56382382e-01 1.34309602e+00 -5.48522592e-01 -2.73041993e-01
-4.74982470e-01 4.85862017e-01 -1.33367851e-01 1.52706444e-01
-9.03867662e-01 -1.03282630e+00 8.33350182e-01 -1.75972030e-01
4.57834244e-01 2.04539075e-01 1.73044398e-01 -7.42829919e-01
-3.84947836e-01 -8.82173121e-01 6.22749150e-01 -8.30943108e-01
-1.42024052e+00 7.71360934e-01 2.13238060e-01 -1.42670834e+00
-4.87675399e-01 -4.18838531e-01 -8.96458328e-01 7.70177066e-01
-1.11552644e+00 -5.13386250e-01 6.87684044e-02 5.04045486e-02
8.02376091e-01 -2.57177979e-01 1.27538788e+00 6.52017951e-01
-2.80774742e-01 5.31882107e-01 -8.37595686e-02 -3.66575792e-02
1.22565734e+00 -1.33177900e+00 7.20315099e-01 7.47365177e-01
6.59894466e-01 3.54174554e-01 1.07082939e+00 -3.76101911e-01
-6.83871627e-01 -6.65047467e-01 8.58872712e-01 -6.18015826e-01
9.72576320e-01 -5.33624709e-01 -1.00429261e+00 3.29662859e-01
4.35823113e-01 4.06939238e-02 1.02044904e+00 4.31281060e-01
-4.82168496e-01 -1.69591486e-01 -7.63975561e-01 3.11190903e-01
1.00580871e+00 -4.99433935e-01 -8.64402771e-01 9.29960385e-02
7.25049794e-01 -4.64207888e-01 -6.23566508e-01 4.00302559e-01
7.93000042e-01 -7.33572066e-01 8.19101095e-01 -7.90167332e-01
2.43734881e-01 -2.12746531e-01 -2.36684278e-01 -1.41234434e+00
-2.87196934e-01 -8.55646312e-01 -5.28254285e-02 1.34161770e+00
5.32004654e-01 -3.55711460e-01 3.58412057e-01 1.79490462e-01
-6.42917812e-01 -2.69872636e-01 -1.24952948e+00 -9.91882265e-01
7.81777799e-02 -9.52854991e-01 3.51196826e-01 6.49169087e-01
2.18713909e-01 4.78912979e-01 -3.66824269e-01 -3.58650349e-02
4.19402838e-01 -1.81757942e-01 5.67985237e-01 -1.48359776e+00
-5.15641034e-01 -6.55577838e-01 -4.84293103e-01 -4.36375827e-01
-4.25980359e-01 -7.92964280e-01 5.53451478e-01 -1.12588751e+00
7.52986446e-02 2.15170324e-01 -7.68050075e-01 5.83917856e-01
-2.48587772e-01 3.94691020e-01 2.64580637e-01 -6.84369877e-02
-5.38437665e-01 6.64142966e-02 6.03241742e-01 1.31774127e-01
-1.58819541e-01 3.41838539e-01 -7.31082976e-01 8.42016399e-01
6.08247936e-01 -6.07191265e-01 -2.32077092e-01 -9.25892368e-02
2.68814236e-01 -1.02131464e-01 3.39513600e-01 -1.57176864e+00
4.23563451e-01 1.09087393e-01 7.30163977e-02 -4.01874006e-01
5.61613798e-01 -3.21027637e-01 1.56863645e-01 2.39539504e-01
-6.03339195e-01 -1.68924257e-01 7.05275953e-01 6.11736655e-01
-1.59869000e-01 -4.31632280e-01 7.29635775e-01 -9.86407250e-02
-1.06331038e+00 -8.20628405e-02 -7.55657673e-01 4.95307535e-01
8.71456385e-01 -1.01908512e-01 -1.52549058e-01 -3.31690371e-01
-1.11969829e+00 -3.01016308e-02 -4.65229720e-01 6.31525457e-01
2.40937620e-01 -8.81068408e-01 -1.07606697e+00 -1.81458533e-01
3.33070844e-01 -5.20465195e-01 4.63386834e-01 6.54548466e-01
-6.65320754e-02 9.20558050e-02 -2.28214860e-01 -4.24834669e-01
-1.45340681e+00 1.54680600e-02 3.95895034e-01 1.69584423e-01
-8.16512465e-01 1.37624753e+00 -1.01556234e-01 -3.09944421e-01
5.56154191e-01 -2.45889887e-01 -4.57938254e-01 5.61458945e-01
9.22761798e-01 6.17665410e-01 2.58120865e-01 -2.80815125e-01
-6.41799629e-01 2.73279190e-01 6.22681342e-02 -5.50501645e-01
1.70862854e+00 2.07109496e-01 4.84925300e-01 1.20721400e+00
8.59060824e-01 1.74196199e-01 -1.00116372e+00 -2.60859355e-02
1.47795036e-01 3.35588343e-02 -9.81432796e-02 -1.23395407e+00
-5.82173645e-01 1.07183468e+00 7.85387516e-01 6.83010399e-01
9.09815490e-01 3.10882404e-02 8.26132774e-01 2.55240053e-01
2.48679608e-01 -1.20619011e+00 -5.63971587e-02 6.16542041e-01
9.30258870e-01 -9.23261404e-01 -1.47420004e-01 1.69691145e-01
-5.55444062e-01 1.05347908e+00 4.08445954e-01 -1.37500495e-01
9.35774028e-01 6.17667973e-01 1.91791896e-02 -2.27860451e-01
-1.18984413e+00 -4.66764927e-01 2.16900289e-01 7.20007598e-01
6.55983269e-01 -8.22783858e-02 -1.14347681e-01 1.13855755e+00
-4.76350456e-01 -1.84922442e-01 3.23896289e-01 9.15733814e-01
-8.56866419e-01 -8.25276613e-01 -2.70079046e-01 4.75661933e-01
-8.18990648e-01 -1.71537191e-01 -3.33284557e-01 6.49945199e-01
2.88450778e-01 1.30652213e+00 5.32207824e-02 -4.45672363e-01
9.32914793e-01 5.92213929e-01 2.55949259e-01 -9.95168149e-01
-9.88383293e-01 2.69576907e-01 4.68268514e-01 -3.09031665e-01
-2.90965796e-01 -1.07867587e+00 -1.13419926e+00 1.03914477e-01
-4.63161469e-01 4.67678964e-01 5.59727848e-01 6.38641179e-01
1.78011239e-01 8.37312579e-01 4.84956443e-01 -8.70920360e-01
-7.19601333e-01 -1.26880479e+00 -5.67178190e-01 3.59060592e-03
5.26769757e-01 -4.67594236e-01 -7.86178648e-01 2.08882689e-01]
|
[15.144271850585938, 5.107876777648926]
|
a6be53be-ef21-4d03-a042-a5da724db575
|
dkma-uld-domain-knowledge-augmented-multi-1
|
2203.06886
| null |
https://arxiv.org/abs/2203.06886v1
|
https://arxiv.org/pdf/2203.06886v1.pdf
|
DKMA-ULD: Domain Knowledge augmented Multi-head Attention based Robust Universal Lesion Detection
|
Incorporating data-specific domain knowledge in deep networks explicitly can provide important cues beneficial for lesion detection and can mitigate the need for diverse heterogeneous datasets for learning robust detectors. In this paper, we exploit the domain information present in computed tomography (CT) scans and propose a robust universal lesion detection (ULD) network that can detect lesions across all organs of the body by training on a single dataset, DeepLesion. We analyze CT-slices of varying intensities, generated using heuristically determined Hounsfield Unit(HU) windows that individually highlight different organs and are given as inputs to the deep network. The features obtained from the multiple intensity images are fused using a novel convolution augmented multi-head self-attention module and subsequently, passed to a Region Proposal Network (RPN) for lesion detection. In addition, we observed that traditional anchor boxes used in RPN for natural images are not suitable for lesion sizes often found in medical images. Therefore, we propose to use lesion-specific anchor sizes and ratios in the RPN for improving the detection performance. We use self-supervision to initialize weights of our network on the DeepLesion dataset to further imbibe domain knowledge. Our proposed Domain Knowledge augmented Multi-head Attention based Universal Lesion Detection Network DMKA-ULD produces refined and precise bounding boxes around lesions across different organs. We evaluate the efficacy of our network on the publicly available DeepLesion dataset which comprises of approximately 32K CT scans with annotated lesions across all organs of the body. Results demonstrate that we outperform existing state-of-the-art methods achieving an overall sensitivity of 87.16%.
|
['Lovekesh Vig', 'Monika Sharma', 'Meghal Dani', 'Manu Sheoran']
|
2022-03-14
|
dkma-uld-domain-knowledge-augmented-multi
|
https://www.bmvc2021-virtualconference.com/conference/papers/paper_1249.html
|
https://www.bmvc2021-virtualconference.com/assets/papers/1249.pdf
|
british-machine-vision-conference-2021-11
|
['medical-object-detection']
|
['computer-vision']
|
[ 2.11734906e-01 2.54652530e-01 -3.42717826e-01 -1.83583423e-01
-1.27781928e+00 -3.29810828e-01 3.98296535e-01 3.11498165e-01
-7.22005486e-01 4.30795372e-01 2.22335726e-01 -7.26944506e-02
1.32309169e-01 -7.03796506e-01 -7.74249971e-01 -8.56021166e-01
-1.36950031e-01 5.05678356e-01 8.46162021e-01 1.30967414e-02
-2.71182179e-01 8.68016541e-01 -1.00545621e+00 6.30494833e-01
6.34517252e-01 1.13619864e+00 4.89516258e-01 7.94954002e-01
3.14680755e-01 7.74913669e-01 -4.45339859e-01 -1.49780631e-01
3.74595404e-01 -1.58135295e-01 -6.30396903e-01 -1.04351779e-02
4.63291407e-01 -6.75912559e-01 -5.55733621e-01 9.16377902e-01
7.36057341e-01 -1.33562803e-01 1.04422331e+00 -8.33419681e-01
-5.51681399e-01 6.21976376e-01 -7.06179678e-01 8.57837081e-01
-1.65733203e-01 4.52565461e-01 7.99485862e-01 -9.14304495e-01
6.42726302e-01 9.28703487e-01 8.82134914e-01 6.44745469e-01
-8.74014139e-01 -5.93649149e-01 -2.49709666e-01 -3.54995094e-02
-1.37934601e+00 4.77684624e-02 3.89942437e-01 -5.20797074e-01
7.79592335e-01 2.52152205e-01 3.51934701e-01 1.14825916e+00
4.79793042e-01 9.23996627e-01 6.26273632e-01 -2.97454417e-01
-6.22449853e-02 5.41556925e-02 -1.83850706e-01 1.12726402e+00
3.84599864e-01 3.53099406e-02 7.30296224e-02 -1.90647826e-01
1.22894490e+00 3.75389248e-01 -3.62284780e-01 -4.59882945e-01
-1.33191514e+00 9.86596167e-01 1.27345324e+00 4.49488223e-01
-6.55283511e-01 2.34656423e-01 5.80930710e-01 -4.70889628e-01
2.36499310e-01 2.64800400e-01 -1.45422235e-01 6.95136428e-01
-8.10539901e-01 -1.11894652e-01 2.25877568e-01 3.62797350e-01
2.29773954e-01 -3.40039849e-01 -7.96383142e-01 8.86965036e-01
1.38510168e-01 3.21084529e-01 7.98358679e-01 -4.39962178e-01
3.67060959e-01 9.20552731e-01 -2.64829844e-01 -5.23525953e-01
-9.15427864e-01 -7.18764842e-01 -9.63790536e-01 2.11553544e-01
6.64937973e-01 -1.89245388e-01 -1.53648055e+00 1.39306712e+00
5.80318391e-01 7.22382665e-02 -1.36011437e-01 1.28312266e+00
1.10632706e+00 2.51145601e-01 5.07057726e-01 3.03716123e-01
1.75918674e+00 -9.46755469e-01 -9.20388997e-02 -1.18333817e-01
8.33049655e-01 -4.08402473e-01 1.06926572e+00 -2.81925555e-02
-9.71584201e-01 -2.79023886e-01 -1.06902850e+00 -7.52391145e-02
-3.93066466e-01 3.88387084e-01 4.09689814e-01 6.30131721e-01
-7.80767202e-01 3.88926178e-01 -9.93841112e-01 -5.31072855e-01
9.42775011e-01 4.01975095e-01 -2.62503862e-01 -2.59682566e-01
-1.02111876e+00 1.07317257e+00 5.76535940e-01 -7.30225742e-02
-1.41576099e+00 -1.08663261e+00 -7.73402035e-01 -3.31140049e-02
5.10188997e-01 -7.71266520e-01 1.24617159e+00 -7.08143055e-01
-9.18786883e-01 1.08995843e+00 3.53364795e-01 -6.34253144e-01
6.80161834e-01 2.07629517e-01 -2.21473828e-01 6.75784349e-01
2.00080171e-01 1.04003906e+00 5.78203380e-01 -1.02920699e+00
-6.53952420e-01 -1.43411502e-01 -2.53012031e-02 1.19874671e-01
-1.28380418e-01 1.10783093e-01 -6.49697661e-01 -8.34922254e-01
-3.77783589e-02 -8.17884088e-01 -4.96496409e-01 4.21199650e-01
-6.53165460e-01 -1.12560794e-01 6.72781527e-01 -7.02657163e-01
9.30956066e-01 -1.76615024e+00 -5.70651442e-02 3.43430549e-01
3.31035376e-01 2.46202499e-01 -3.12269498e-02 -4.06232566e-01
-2.80944079e-01 -7.83535615e-02 -3.36136818e-01 -2.19899770e-02
-3.02765876e-01 2.49876946e-01 2.66237557e-01 7.72328079e-01
2.86667675e-01 1.14783859e+00 -7.99239933e-01 -9.89564776e-01
5.00807285e-01 5.26324153e-01 -5.15140295e-01 1.20077789e-01
-6.92703351e-02 2.93945253e-01 -4.91651803e-01 9.39001560e-01
4.95054305e-01 -5.77394664e-01 -1.68880060e-01 -5.28466821e-01
1.80702120e-01 -1.51695251e-01 -8.12463105e-01 1.62993217e+00
-4.34141457e-01 2.57839262e-01 -3.77406320e-03 -7.36758530e-01
4.15216506e-01 3.91678751e-01 7.55175650e-01 -5.85755527e-01
5.49572349e-01 1.33695617e-01 2.29299754e-01 -5.43764114e-01
-8.92917737e-02 -2.34553233e-01 -6.76090941e-02 3.61484259e-01
1.69224754e-01 3.09003256e-02 1.62390873e-01 1.95854023e-01
1.44100630e+00 -5.49202979e-01 4.99093711e-01 -2.83852816e-01
6.70402110e-01 9.13100615e-02 3.41748029e-01 9.43785310e-01
-6.07702494e-01 9.73870158e-01 4.04914707e-01 -5.07089555e-01
-1.04002416e+00 -1.33777082e+00 -5.30185640e-01 1.25749373e+00
-7.80080408e-02 5.11409976e-02 -7.17329681e-01 -1.12853575e+00
1.24657713e-01 3.39223683e-01 -1.13697660e+00 -1.53248785e-02
-6.46803677e-01 -1.06913781e+00 8.06288838e-01 9.66716051e-01
5.32063246e-01 -1.09967494e+00 -1.10459423e+00 2.56289065e-01
-4.85899933e-02 -1.09240448e+00 -5.05950868e-01 4.36975837e-01
-7.58267343e-01 -1.27921772e+00 -1.38017416e+00 -7.06874132e-01
8.44850779e-01 -1.39443129e-01 1.07195663e+00 2.02542558e-01
-1.09696949e+00 3.78580272e-01 -1.35278329e-01 -3.50810498e-01
-5.63231409e-01 1.49274290e-01 -3.95009041e-01 -2.66646832e-01
8.16763006e-03 -1.28995702e-01 -9.61237192e-01 4.30950314e-01
-1.09724402e+00 9.61634505e-04 1.03022671e+00 9.34850335e-01
5.54647207e-01 -4.36438292e-01 4.90374148e-01 -8.87605906e-01
3.27890575e-01 -6.91790640e-01 -3.27377260e-01 4.90016103e-01
7.93454349e-02 4.97495160e-02 3.48766744e-01 -4.69864935e-01
-1.02064383e+00 3.36739153e-01 -2.95979585e-02 -4.06516850e-01
-2.62935817e-01 2.20348120e-01 3.35470796e-01 -3.17882627e-01
1.23690760e+00 -2.53575519e-02 -1.88859969e-01 -6.71890303e-02
2.42584988e-01 4.35585260e-01 8.31885457e-01 -4.29928482e-01
5.12719810e-01 7.44196415e-01 2.20216721e-01 -4.76857543e-01
-9.17243004e-01 -6.61495507e-01 -8.10372710e-01 -1.92065939e-01
1.05763376e+00 -8.41985643e-01 -3.30811441e-01 8.53027403e-02
-8.96264255e-01 -2.54851192e-01 -2.78726697e-01 3.95350218e-01
-3.07713866e-01 2.13466555e-01 -1.08015621e+00 -3.57372046e-01
-7.32771695e-01 -1.34311044e+00 1.30967367e+00 2.58403152e-01
1.45232985e-02 -9.48532760e-01 -1.55595526e-01 1.35746047e-01
5.07576287e-01 5.63207626e-01 8.58477950e-01 -6.54092252e-01
-5.09981811e-01 -3.08974326e-01 -7.42950737e-01 6.78912774e-02
1.02837525e-01 -2.25861683e-01 -9.04134572e-01 -1.49741545e-01
-5.33323944e-01 -4.17786598e-01 1.21384585e+00 8.04749787e-01
1.38556015e+00 2.30201721e-01 -5.80645740e-01 7.19798923e-01
1.43553674e+00 -1.84132710e-01 3.35063130e-01 4.60760325e-01
5.89062333e-01 2.82368690e-01 2.26210684e-01 5.11175931e-01
1.15197197e-01 4.11523521e-01 7.66297519e-01 -7.06338882e-01
-5.91713190e-01 1.16589844e-01 -5.92935458e-02 -6.70065731e-02
-4.56451736e-02 -1.54491261e-01 -1.19277430e+00 9.37236011e-01
-1.43988109e+00 -5.11966586e-01 1.47455290e-01 1.72286785e+00
8.35454881e-01 1.30509660e-01 2.01997265e-01 -3.03227782e-01
8.23085725e-01 -1.56769156e-01 -8.34441364e-01 4.48297150e-02
-1.15380017e-02 3.68595183e-01 9.78485227e-01 5.54617345e-02
-1.61257374e+00 6.05480552e-01 5.71005011e+00 7.21485019e-01
-1.19854164e+00 4.09099847e-01 7.31939912e-01 -1.92748517e-01
1.93094224e-01 -6.71779990e-01 -6.17379248e-01 1.04715653e-01
5.35879254e-01 4.75766249e-02 -2.78633744e-01 9.60092247e-01
9.16009583e-03 -2.34506443e-01 -1.09411693e+00 8.37287366e-01
1.30111009e-01 -1.33135712e+00 -8.88634287e-03 -1.70083731e-01
6.94490194e-01 6.21744215e-01 2.31397986e-01 2.78733253e-01
3.86587262e-01 -1.16039312e+00 3.04848164e-01 2.26900592e-01
8.80050838e-01 -5.47384024e-01 8.74925017e-01 1.58591941e-01
-1.12764871e+00 -9.44975540e-02 -5.38308561e-01 7.04556286e-01
2.82386914e-02 4.88485873e-01 -1.68447089e+00 3.49608094e-01
6.33799493e-01 4.21615213e-01 -7.65950501e-01 1.43138218e+00
-6.86737001e-02 4.77831155e-01 -6.76410317e-01 2.46614933e-01
5.78147531e-01 6.50610566e-01 3.06082934e-01 1.60691023e+00
3.33212912e-01 1.52503908e-01 2.75065720e-01 9.49953437e-01
-2.92161435e-01 2.01286599e-01 -1.96950600e-01 4.81272370e-01
9.24616531e-02 1.49097788e+00 -1.06970668e+00 -5.98690927e-01
-2.97489166e-01 8.45955014e-01 2.51252204e-01 1.09676808e-01
-1.14067769e+00 5.44321798e-02 2.99262673e-01 1.84694797e-01
4.54884082e-01 3.00730705e-01 -1.38990477e-01 -1.04394698e+00
-4.29591566e-01 -4.56525683e-01 1.01847744e+00 -6.04140103e-01
-1.41912615e+00 5.69092333e-01 -3.41342911e-02 -1.05246890e+00
-1.63872484e-02 -8.01426470e-01 -7.44028926e-01 7.40173459e-01
-1.62633789e+00 -1.25804007e+00 -5.45319498e-01 8.49803388e-01
4.79192287e-01 2.15029970e-01 6.11507952e-01 4.30853516e-01
-7.48650134e-01 7.40466654e-01 -2.07502499e-01 5.43990970e-01
7.66039431e-01 -1.36504340e+00 5.02563231e-02 7.11593866e-01
-1.94752082e-01 3.01828653e-01 3.03881228e-01 -6.69322908e-01
-9.17223871e-01 -1.47531915e+00 9.58313514e-03 -3.57298642e-01
5.92017591e-01 6.08604169e-03 -9.00066793e-01 7.25767672e-01
-3.25725153e-02 8.77095819e-01 4.78394538e-01 -6.04061604e-01
-2.39852205e-01 1.86277315e-01 -1.56815684e+00 3.65689278e-01
6.66494131e-01 -3.03825319e-01 -6.51458502e-01 6.14623249e-01
5.88347435e-01 -7.74360001e-01 -9.26291764e-01 5.55029631e-01
3.28403741e-01 -6.97670639e-01 1.37805641e+00 -5.48833847e-01
4.14923310e-01 -1.35083854e-01 -8.28268230e-02 -1.15814412e+00
-3.30232054e-01 3.18481147e-01 8.06092694e-02 5.05178213e-01
4.19371009e-01 -4.31132853e-01 1.08914232e+00 4.78610337e-01
-3.95876944e-01 -7.74428725e-01 -1.08129239e+00 -3.69020075e-01
3.52286398e-01 -4.51092094e-01 2.97626823e-01 6.97595239e-01
-2.22744629e-01 -2.70716786e-01 9.24923345e-02 4.63677406e-01
8.82501781e-01 -1.60617068e-01 2.45969519e-01 -9.65328097e-01
-2.14029789e-01 -4.97397900e-01 -4.43740755e-01 -7.02484727e-01
-3.25289488e-01 -1.26504409e+00 -8.18424043e-04 -1.47097743e+00
5.88315964e-01 -3.78584415e-01 -7.45046854e-01 6.69079959e-01
-2.09334359e-01 5.95472813e-01 1.68893546e-01 -5.93959540e-03
-6.88670039e-01 1.39480516e-01 1.53521550e+00 -2.42695466e-01
1.61131144e-01 -2.29930952e-01 -4.18267250e-01 9.45398808e-01
5.54486334e-01 -4.71350640e-01 1.36522099e-01 -1.81298509e-01
-5.12361228e-01 7.17863813e-02 8.02328706e-01 -1.40095651e+00
3.49869996e-01 1.49977341e-01 9.56620753e-01 -7.87578046e-01
2.45159835e-01 -8.13778043e-01 -4.46599752e-01 8.36235523e-01
-2.46477097e-01 -3.44080478e-01 3.15365076e-01 4.30680066e-01
7.40153790e-02 -1.79533318e-01 1.16110861e+00 -4.73447591e-01
-7.95576692e-01 3.96881491e-01 -2.36423090e-01 1.57876611e-02
1.40619731e+00 -4.37053666e-02 -3.35216939e-01 1.20820999e-01
-8.57591808e-01 3.09625179e-01 1.78644508e-01 1.20667681e-01
7.00236499e-01 -1.20084858e+00 -7.87787855e-01 5.50009832e-02
3.05828720e-01 1.98060140e-01 5.35392880e-01 1.04698384e+00
-8.31026077e-01 6.01240814e-01 -2.92711079e-01 -1.05520809e+00
-1.08526897e+00 4.67073262e-01 9.49965894e-01 -5.62226355e-01
-7.80614674e-01 1.09834933e+00 6.23053253e-01 -3.58017802e-01
1.88261345e-01 -1.01251590e+00 -8.03706050e-02 -1.55274689e-01
5.83873153e-01 -3.00958920e-02 2.96881914e-01 -6.03380799e-01
-5.55770218e-01 4.68432099e-01 -3.70982707e-01 3.94634604e-01
1.22069275e+00 2.72531688e-01 4.02997464e-01 -2.30204269e-01
1.25575185e+00 -4.54716235e-01 -1.40600848e+00 -5.10227978e-01
-4.62880358e-02 -1.89151809e-01 4.64175135e-01 -1.08084762e+00
-1.34211540e+00 7.23382711e-01 1.10220981e+00 -3.17456096e-01
1.04127204e+00 5.15392840e-01 8.87812376e-01 1.05515726e-01
5.07977791e-02 -6.86215222e-01 2.54612148e-01 1.74696192e-01
8.03548276e-01 -1.60275877e+00 3.16028669e-02 -3.72690231e-01
-7.06801891e-01 1.16559982e+00 8.23276579e-01 -2.65864611e-01
4.23365563e-01 4.78701651e-01 1.14989460e-01 -3.72869998e-01
-5.36825299e-01 -4.80723679e-01 6.16092145e-01 6.23203635e-01
3.28986794e-01 1.34837732e-01 1.60019487e-01 6.14928067e-01
4.00823712e-01 -9.31721628e-02 4.52743530e-01 8.39189231e-01
-7.52667725e-01 -6.34900868e-01 -7.78643548e-01 6.55422390e-01
-6.77752733e-01 -7.85672851e-03 5.05741592e-03 1.14524806e+00
2.98434287e-01 2.39992917e-01 -1.26791969e-01 2.21040919e-01
3.18071097e-01 -1.70146078e-01 4.77742195e-01 -7.42462814e-01
-8.71666431e-01 -2.06783880e-02 -3.37064236e-01 -4.97595042e-01
-8.72845203e-02 -4.88841116e-01 -1.64465439e+00 2.20878199e-01
-9.83607247e-02 -4.03128386e-01 4.80552405e-01 8.38299990e-01
-4.91540097e-02 9.06069696e-01 1.36611894e-01 -1.02480364e+00
-5.28927386e-01 -9.48810041e-01 -4.17535394e-01 5.13641715e-01
4.29973930e-01 -9.63637769e-01 -8.54106843e-02 -1.87451169e-01]
|
[15.098742485046387, -2.2968008518218994]
|
560e1ce1-5584-4f62-9eb8-574aeb4336c0
|
reconstructing-vechicles-from-a-single-image
|
1609.09468
| null |
http://arxiv.org/abs/1609.09468v1
|
http://arxiv.org/pdf/1609.09468v1.pdf
|
Reconstructing Vechicles from a Single Image: Shape Priors for Road Scene Understanding
|
We present an approach for reconstructing vehicles from a single (RGB) image,
in the context of autonomous driving. Though the problem appears to be
ill-posed, we demonstrate that prior knowledge about how 3D shapes of vehicles
project to an image can be used to reason about the reverse process, i.e., how
shapes (back-)project from 2D to 3D. We encode this knowledge in \emph{shape
priors}, which are learnt over a small keypoint-annotated dataset. We then
formulate a shape-aware adjustment problem that uses the learnt shape priors to
recover the 3D pose and shape of a query object from an image. For shape
representation and inference, we leverage recent successes of Convolutional
Neural Networks (CNNs) for the task of object and keypoint localization, and
train a novel cascaded fully-convolutional architecture to localize vehicle
\emph{keypoints} in images. The shape-aware adjustment then robustly recovers
shape (3D locations of the detected keypoints) while simultaneously filling in
occluded keypoints. To tackle estimation errors incurred due to erroneously
detected keypoints, we use an Iteratively Re-weighted Least Squares (IRLS)
scheme for robust optimization, and as a by-product characterize noise models
for each predicted keypoint. We evaluate our approach on autonomous driving
benchmarks, and present superior results to existing monocular, as well as
stereo approaches.
|
['Falak Chhaya', 'G. V. Sai Krishna', 'K. Madhava Krishna', 'J. Krishna Murthy']
|
2016-09-29
| null | null | null | null |
['road-scene-understanding']
|
['computer-vision']
|
[ 1.42336592e-01 2.15457320e-01 5.63359028e-03 -6.06224537e-01
-9.01210546e-01 -9.90557432e-01 8.11947405e-01 -2.87576109e-01
-3.79472971e-01 1.70426160e-01 -4.76410501e-02 -3.00118357e-01
1.68341130e-01 -5.71366608e-01 -1.56101155e+00 -4.42306101e-01
3.62409592e-01 7.58101583e-01 3.70801091e-01 -1.33168310e-01
3.73735428e-01 1.02904999e+00 -1.81395721e+00 6.35225773e-02
4.45141822e-01 1.16876853e+00 2.84116566e-01 7.94933379e-01
2.24789567e-02 4.76712495e-01 -7.72389024e-02 -3.72637033e-01
5.85628808e-01 4.33350354e-01 -5.79420149e-01 3.24825227e-01
1.21032667e+00 -5.72904587e-01 -6.71255827e-01 9.18973386e-01
-1.03057824e-01 -2.82576848e-02 7.18912065e-01 -1.60375118e+00
-3.11396420e-01 -3.84502083e-01 -4.41678464e-01 4.61695231e-02
2.93225884e-01 4.48439538e-01 9.45641935e-01 -1.38801098e+00
7.95208991e-01 1.45544970e+00 8.75130236e-01 2.72439718e-01
-1.41512227e+00 -6.33797467e-01 2.70311743e-01 2.42579013e-01
-1.35878253e+00 -9.02768791e-01 7.87322938e-01 -6.93430066e-01
9.57831502e-01 -1.28012048e-02 5.69990814e-01 7.83259571e-01
7.68056437e-02 9.28304315e-01 6.85417354e-01 1.54103056e-01
7.36999065e-02 8.86744708e-02 -1.10593192e-01 7.50302613e-01
1.39600873e-01 4.80159163e-01 -5.26960075e-01 -2.02064477e-02
6.44192696e-01 1.14269458e-01 -3.69984619e-02 -1.07023776e+00
-1.21375990e+00 6.97893739e-01 6.13771915e-01 -5.18952489e-01
-2.68432349e-01 6.00140631e-01 -1.25256971e-01 -3.51103097e-02
3.73583168e-01 1.36580169e-01 -6.11184597e-01 1.68576971e-01
-7.20645487e-01 6.56885564e-01 4.71936494e-01 1.25682163e+00
1.33287847e+00 -1.18139498e-01 7.29372054e-02 4.57126379e-01
5.84097385e-01 1.07001162e+00 -2.74484187e-01 -1.48237693e+00
6.59090877e-01 5.59758306e-01 3.85695815e-01 -9.48429763e-01
-4.46701854e-01 -1.26868770e-01 -2.51379907e-01 5.53407907e-01
3.62988979e-01 1.39549017e-01 -1.12677419e+00 1.52471685e+00
6.02936685e-01 3.24966311e-01 8.05346891e-02 9.35300171e-01
9.73280311e-01 5.71546972e-01 -3.58316034e-01 3.37915123e-01
1.12670934e+00 -8.48475575e-01 -1.18072458e-01 -8.46512020e-01
2.25087926e-01 -6.49883509e-01 5.24567366e-01 8.30191672e-02
-9.93689358e-01 -6.36674047e-01 -8.59357297e-01 -3.09377909e-01
-3.53180617e-01 1.28095910e-01 3.17675799e-01 2.09250480e-01
-1.18301189e+00 1.48832440e-01 -7.46396482e-01 -2.41337895e-01
6.66289330e-01 3.93728137e-01 -7.68379509e-01 -1.82441130e-01
-4.05660421e-01 9.54700649e-01 3.93943898e-02 1.22187182e-01
-1.27745974e+00 -1.06317270e+00 -1.33559549e+00 -5.08198261e-01
4.64266509e-01 -6.47206962e-01 1.25490212e+00 -5.93746245e-01
-1.05621386e+00 1.16247880e+00 -6.26125932e-01 -5.38459897e-01
4.39196020e-01 -2.93813527e-01 3.49524692e-02 4.40605804e-02
4.13483024e-01 1.34512079e+00 1.22373366e+00 -1.65927947e+00
-9.05730903e-01 -6.03941321e-01 1.00993656e-01 2.47821182e-01
4.89682943e-01 -4.47022229e-01 -8.53270292e-01 -1.62158906e-01
6.27873719e-01 -1.20918894e+00 -1.42515570e-01 5.02293408e-01
-3.01741332e-01 -7.18316436e-02 1.05157328e+00 -5.53307533e-01
1.26174554e-01 -2.12900162e+00 1.84018508e-01 1.93737552e-01
2.32545674e-01 -2.30030194e-02 -1.72883883e-01 -9.29033980e-02
1.52237713e-01 -3.54553640e-01 -3.26457709e-01 -7.66726911e-01
1.43141285e-01 5.98718703e-01 -5.25620639e-01 7.81767428e-01
5.95050097e-01 1.25901258e+00 -7.10847139e-01 -1.17722765e-01
5.39367795e-01 6.46021128e-01 -6.21976793e-01 2.83300787e-01
-3.77872765e-01 4.22116071e-01 -3.22908163e-01 6.59904957e-01
1.05544198e+00 1.32858068e-01 -5.37655950e-01 -5.90493858e-01
-3.04344475e-01 1.40306260e-02 -1.14179802e+00 1.62050700e+00
-4.26205724e-01 7.57093072e-01 3.05315018e-01 -8.60070229e-01
9.43253040e-01 -1.60148442e-01 5.43724537e-01 -7.28419185e-01
-8.42018202e-02 1.49316072e-01 -5.98469555e-01 -4.80249763e-01
6.13095164e-01 2.85924941e-01 -1.46782473e-02 6.20765276e-02
1.79213695e-02 -8.13118517e-01 -1.79760814e-01 1.46802887e-01
8.95422935e-01 4.37234819e-01 -1.03801191e-01 -3.64428647e-02
4.67062056e-01 2.36940145e-01 3.85838032e-01 5.68529844e-01
-9.58623961e-02 8.23403418e-01 2.76322484e-01 -7.43106902e-01
-1.33232915e+00 -1.03277242e+00 -2.81910688e-01 8.78664017e-01
4.78299439e-01 7.72366375e-02 -4.35101777e-01 -6.40567183e-01
7.40101576e-01 6.06250107e-01 -6.47901058e-01 -5.84629215e-02
-6.85099185e-01 -7.36777335e-02 2.89248109e-01 6.60764337e-01
3.05671990e-01 -7.54486322e-01 -9.23286796e-01 2.22640913e-02
-8.49848017e-02 -1.71072185e+00 -4.86973822e-01 2.53554672e-01
-5.45270324e-01 -1.17041242e+00 -3.48514140e-01 -5.97978890e-01
8.14818144e-01 5.79594612e-01 1.09603870e+00 -9.66556817e-02
-2.38625497e-01 8.33860040e-01 1.67084754e-01 -4.46962893e-01
-2.62257934e-01 -4.24095541e-01 1.19667584e-02 2.17628866e-01
3.67877632e-01 -2.95137346e-01 -5.91171503e-01 4.92493540e-01
-5.45831800e-01 1.28899440e-01 6.08948469e-01 3.70048344e-01
9.81337488e-01 -3.70180547e-01 -5.41637950e-02 -4.09624040e-01
-2.32051492e-01 -2.51003504e-01 -1.11787224e+00 -2.02978313e-01
-2.09717855e-01 2.39196345e-01 1.38335124e-01 -8.85689780e-02
-7.61551678e-01 8.37532103e-01 -1.92406923e-01 -9.26347673e-01
-3.85947257e-01 -1.11509316e-01 -1.28075123e-01 -4.90623564e-01
4.73964572e-01 3.23965281e-01 2.04368830e-01 -3.18596363e-01
6.39898837e-01 3.29779565e-01 9.52779651e-01 -4.25174832e-01
1.15828061e+00 1.02496767e+00 3.09620798e-01 -6.37279749e-01
-8.98668110e-01 -7.35300481e-01 -9.24093843e-01 -2.58677214e-01
9.91375744e-01 -1.31731057e+00 -7.97154963e-01 2.66947836e-01
-1.37279475e+00 -3.90593320e-01 -2.03167126e-01 2.09310785e-01
-9.23296869e-01 1.42692804e-01 9.92606357e-02 -6.46192968e-01
7.08702579e-02 -1.42385411e+00 1.78781843e+00 -6.65987581e-02
5.91019168e-02 -6.59805238e-01 -1.88080266e-01 5.53058267e-01
2.12136209e-01 1.35065123e-01 4.78542000e-01 -2.38638788e-01
-1.17779970e+00 -2.74846852e-01 -5.30023158e-01 1.46029681e-01
-3.50775748e-01 -1.72979355e-01 -1.29544234e+00 -2.06433967e-01
-3.82990301e-01 -2.73614496e-01 1.11716723e+00 4.31898594e-01
9.29127634e-01 -1.18641235e-01 -5.77492416e-01 9.85197067e-01
1.18826056e+00 -2.14552298e-01 2.58481860e-01 1.89851284e-01
1.00146735e+00 6.37203217e-01 5.94339311e-01 1.96528748e-01
1.00009859e+00 8.80986571e-01 1.01972473e+00 4.95048761e-02
-2.50352591e-01 -4.97291446e-01 2.92666644e-01 -8.07207450e-02
2.17422068e-01 3.32834393e-01 -9.74095583e-01 7.07806647e-01
-1.74559951e+00 -6.60477281e-01 -1.64497182e-01 2.04437757e+00
4.88613695e-01 2.14936465e-01 2.10010689e-02 -2.56290615e-01
2.88022339e-01 7.31031299e-02 -1.00264084e+00 -4.45501041e-03
-1.57134846e-01 -2.51954734e-01 1.03910804e+00 8.34146559e-01
-1.32161677e+00 1.05378938e+00 6.13430595e+00 3.04597765e-01
-1.04101431e+00 -5.10132946e-02 5.50228834e-01 2.03714862e-01
-5.86796165e-01 1.91338494e-01 -1.14077091e+00 3.88652273e-02
4.20240402e-01 2.64190078e-01 5.42560816e-01 8.82179141e-01
1.10979363e-01 -1.86889306e-01 -1.42250216e+00 1.29140401e+00
2.43775576e-01 -1.45653164e+00 -1.64552093e-01 1.95136458e-01
8.90549242e-01 6.24353647e-01 1.76918805e-01 1.26240328e-01
2.93497562e-01 -1.02146244e+00 1.29647100e+00 7.38218963e-01
6.60478055e-01 -5.24495602e-01 4.41599488e-01 4.78007734e-01
-1.43900502e+00 -2.01932922e-01 -4.69858587e-01 1.39743656e-01
3.32104303e-02 3.10532719e-01 -9.12652075e-01 1.92680463e-01
8.71633828e-01 9.44686651e-01 -5.97340763e-01 8.47041786e-01
-1.56099543e-01 6.78988099e-02 -6.45877302e-01 4.20785844e-01
2.25892201e-01 -6.09633029e-02 6.95898592e-01 9.52651083e-01
2.39279896e-01 2.52363961e-02 1.43124565e-01 1.33549690e+00
-3.10084615e-02 -4.59047645e-01 -9.01415765e-01 4.76749212e-01
5.02186000e-01 1.35266781e+00 -4.96671826e-01 -1.80343807e-01
-4.86433238e-01 7.14820147e-01 2.38292366e-01 5.93321204e-01
-6.76543653e-01 2.25020856e-01 9.98025179e-01 3.48517418e-01
8.06210637e-01 -4.92575616e-01 -3.18717629e-01 -7.89243042e-01
2.48318613e-01 -4.29878533e-01 -1.39469147e-01 -1.20430553e+00
-9.17132556e-01 1.34101018e-01 7.91879743e-02 -1.10866773e+00
-2.10723743e-01 -7.92201757e-01 -3.74823421e-01 7.84248114e-01
-1.92994440e+00 -1.43188226e+00 -5.19836903e-01 5.71392715e-01
6.50834143e-01 -8.90885666e-03 3.07310641e-01 -1.19952029e-02
-3.58985527e-03 2.59789407e-01 -2.38367692e-01 -5.76433502e-02
2.92822272e-01 -1.10820150e+00 7.59534240e-01 7.09551752e-01
2.38394499e-01 1.87633425e-01 7.04259634e-01 -4.59353030e-01
-2.04427958e+00 -1.49311042e+00 5.79913199e-01 -1.08800364e+00
4.07402575e-01 -5.15362978e-01 -5.50375879e-01 9.78333771e-01
-3.10545653e-01 6.92931473e-01 -2.97818452e-01 -4.55232948e-01
-5.84968507e-01 -3.56134921e-01 -1.05391848e+00 3.30593884e-01
1.18300974e+00 -7.20122159e-01 -4.34282392e-01 3.00793052e-01
7.44809568e-01 -7.75296152e-01 -3.71379256e-01 5.45176685e-01
5.17881691e-01 -8.94470334e-01 1.48914230e+00 -2.77541518e-01
1.36534527e-01 -6.16121590e-01 -5.10561407e-01 -1.08943224e+00
-7.07294121e-02 -4.58064377e-01 -9.20204297e-02 7.96186149e-01
3.39032948e-01 -3.64977747e-01 1.01453018e+00 7.45446026e-01
-4.75443512e-01 -4.63362694e-01 -1.11827612e+00 -4.35510784e-01
-1.22095935e-01 -9.55183804e-01 5.83837748e-01 2.46453613e-01
-7.33847082e-01 1.40500113e-01 -1.09148055e-01 7.13276446e-01
8.92836392e-01 2.37924561e-01 1.35421741e+00 -1.25742507e+00
9.00694802e-02 -2.59944588e-01 -8.43897998e-01 -1.54263830e+00
5.49705505e-01 -8.12135994e-01 4.60372716e-01 -1.44601500e+00
-5.30201234e-02 -4.94512528e-01 3.47553372e-01 5.18092871e-01
6.17453940e-02 5.36527336e-01 1.29510328e-01 8.92623737e-02
-6.41438901e-01 4.48124647e-01 1.11666954e+00 -3.88380587e-01
1.19755063e-02 1.02169700e-01 -3.88265908e-01 8.63660455e-01
2.87775666e-01 -3.26551408e-01 -2.43075803e-01 -7.11592853e-01
4.77489710e-01 -4.85922433e-02 1.17092562e+00 -8.39659035e-01
5.57674408e-01 -1.50530234e-01 5.68197370e-01 -1.20912373e+00
7.99743235e-01 -1.20443809e+00 8.01553577e-02 1.43957794e-01
-8.96963254e-02 -1.93444695e-02 3.45947385e-01 7.02263534e-01
7.38829076e-02 5.41048646e-02 6.18293285e-01 -1.97027884e-02
-1.07971501e+00 6.63509488e-01 -4.89072427e-02 5.55844046e-02
1.03170455e+00 -5.18424988e-01 8.75446871e-02 -3.75621140e-01
-5.05072951e-01 4.07444179e-01 6.18009031e-01 6.57478452e-01
1.02333403e+00 -1.33770478e+00 -6.56893909e-01 7.17186570e-01
5.04857779e-01 5.70842564e-01 3.15378085e-02 7.39445984e-01
-4.83253181e-01 5.70636511e-01 4.40673117e-04 -1.34078896e+00
-1.09131718e+00 4.41265345e-01 5.94184875e-01 6.55412972e-01
-7.55818248e-01 9.98307288e-01 5.10732412e-01 -9.51628029e-01
3.68888378e-01 -5.05976796e-01 -7.15009719e-02 -1.49835587e-01
2.91052997e-01 6.48973659e-02 3.39706540e-01 -1.28591144e+00
-4.47721303e-01 1.25719237e+00 2.08224580e-01 -1.88036412e-02
1.30368972e+00 -3.68565619e-01 3.47853974e-02 2.58661985e-01
1.37135851e+00 -1.35834187e-01 -2.11268234e+00 -5.27153194e-01
-1.99246988e-01 -5.61418593e-01 1.10966749e-01 -3.45243812e-01
-1.04713202e+00 8.16668808e-01 4.96057600e-01 -3.40757936e-01
4.66240317e-01 4.24820304e-01 5.36068201e-01 6.26799822e-01
3.03898841e-01 -8.66519332e-01 -3.35272849e-02 7.21611857e-01
9.13343668e-01 -1.61375082e+00 -1.26899540e-01 -3.32616299e-01
-4.21481609e-01 1.20801973e+00 5.18041968e-01 -1.82035387e-01
7.33007252e-01 3.48379731e-01 5.23241013e-02 -3.72837454e-01
-5.67366242e-01 -2.90247202e-01 6.04785323e-01 6.98337913e-01
-3.86818051e-01 -7.19311386e-02 9.02592659e-01 1.42008245e-01
-2.91594982e-01 -3.02147269e-01 2.34561533e-01 6.27223551e-01
-6.35646462e-01 -4.33961838e-01 -6.77366614e-01 1.89632326e-01
2.14878947e-01 1.66195467e-01 -3.56142730e-01 5.00274241e-01
5.49026310e-01 8.33290517e-01 4.21296328e-01 -2.14130983e-01
5.93547642e-01 -1.73036724e-01 4.34744269e-01 -4.06055689e-01
1.57100230e-01 -1.34771600e-01 -1.48215324e-01 -9.99683678e-01
-4.07770395e-01 -8.60866129e-01 -1.30010664e+00 -7.57003501e-02
4.54805084e-02 -3.75508100e-01 9.41140294e-01 1.16471410e+00
3.63438040e-01 1.40499115e-01 6.16667330e-01 -1.44858539e+00
-3.37118715e-01 -3.85171503e-01 -5.43443635e-02 4.36497450e-01
9.20087039e-01 -7.82290339e-01 -3.37537557e-01 2.34815136e-01]
|
[7.836769104003906, -2.562614917755127]
|
8eab834b-7840-400b-af32-d465ee387a37
|
a-combined-cnn-and-lstm-model-for-arabic
|
1807.02911
| null |
http://arxiv.org/abs/1807.02911v3
|
http://arxiv.org/pdf/1807.02911v3.pdf
|
A Combined CNN and LSTM Model for Arabic Sentiment Analysis
|
Deep neural networks have shown good data modelling capabilities when dealing
with challenging and large datasets from a wide range of application areas.
Convolutional Neural Networks (CNNs) offer advantages in selecting good
features and Long Short-Term Memory (LSTM) networks have proven good abilities
of learning sequential data. Both approaches have been reported to provide
improved results in areas such image processing, voice recognition, language
translation and other Natural Language Processing (NLP) tasks. Sentiment
classification for short text messages from Twitter is a challenging task, and
the complexity increases for Arabic language sentiment classification tasks
because Arabic is a rich language in morphology. In addition, the availability
of accurate pre-processing tools for Arabic is another current limitation,
along with limited research available in this area. In this paper, we
investigate the benefits of integrating CNNs and LSTMs and report obtained
improved accuracy for Arabic sentiment analysis on different datasets.
Additionally, we seek to consider the morphological diversity of particular
Arabic words by using different sentiment classification levels.
|
['Matthew England', 'Vasile Palade', 'Abdulaziz M. Alayba', 'Rahat Iqbal']
|
2018-07-09
| null | null | null | null |
['arabic-sentiment-analysis']
|
['natural-language-processing']
|
[-1.02645829e-02 -5.19845188e-01 -9.82825831e-02 -4.67460722e-01
-3.20980340e-01 -4.47726309e-01 6.30486250e-01 3.48656207e-01
-7.84784555e-01 5.54639339e-01 1.16233900e-01 -3.22050482e-01
3.52531821e-02 -7.93550789e-01 -2.38639593e-01 -6.61287546e-01
-5.91678396e-02 2.12432027e-01 -2.51859039e-01 -8.02438915e-01
6.15015268e-01 6.56401575e-01 -1.44256628e+00 7.48490214e-01
8.23511124e-01 1.11741841e+00 2.63228804e-01 5.94766319e-01
-5.36434114e-01 9.16891396e-01 -6.90033257e-01 -6.16308272e-01
-2.87434787e-01 -7.34867677e-02 -9.02396917e-01 3.11677135e-03
-7.69883208e-03 2.29815859e-02 2.72457302e-01 7.28240013e-01
6.97084188e-01 1.59714401e-01 5.04989982e-01 -9.13729012e-01
-7.13311851e-01 6.20712340e-01 -3.33021760e-01 2.82092810e-01
2.76447982e-01 -2.26764888e-01 5.96728742e-01 -1.13250661e+00
2.98003107e-01 1.25136471e+00 8.90307188e-01 3.57905835e-01
-3.01741242e-01 -5.50097883e-01 -1.49349375e-02 2.24570572e-01
-9.35014844e-01 -4.18085992e-01 5.83823740e-01 -2.44395241e-01
1.36903059e+00 -5.53707257e-02 3.71944547e-01 9.32304978e-01
5.97426534e-01 6.75802112e-01 1.15590394e+00 -6.40209317e-01
-1.69259638e-01 4.62323785e-01 1.66628763e-01 7.12567866e-01
-1.18559696e-01 -4.21833932e-01 -7.13342071e-01 2.02893183e-01
4.20869738e-02 8.03075880e-02 3.56576324e-01 4.97354984e-01
-1.11902499e+00 1.02937305e+00 6.38882279e-01 7.89145589e-01
-4.69223022e-01 -2.41321623e-01 9.15041387e-01 7.70617843e-01
6.40868783e-01 4.39249754e-01 -9.34179068e-01 -1.41008303e-01
-7.58615673e-01 -1.65953562e-01 7.81374872e-01 5.97159207e-01
2.52772599e-01 5.78677595e-01 3.16922188e-01 1.07457459e+00
4.15431798e-01 6.18755639e-01 1.21672654e+00 -9.51893926e-02
6.31457269e-01 8.68212938e-01 -3.30550581e-01 -1.55339193e+00
-8.32071960e-01 -2.42289335e-01 -9.49349582e-01 2.60206982e-02
4.34533089e-01 -3.57177615e-01 -8.65520656e-01 1.19174767e+00
-1.21516712e-01 -8.04876924e-01 4.94302273e-01 5.07323384e-01
1.12094235e+00 1.05262697e+00 2.40026549e-01 -7.57955238e-02
1.59070194e+00 -8.58423948e-01 -7.38877475e-01 -5.21169662e-01
7.77692258e-01 -1.06638396e+00 1.09344554e+00 5.68211615e-01
-8.38338673e-01 -4.41608787e-01 -9.77487504e-01 -8.07202160e-02
-1.31871021e+00 3.69457155e-01 7.60456026e-01 9.34143245e-01
-1.01657152e+00 3.53684932e-01 -6.39698505e-01 -7.14205980e-01
4.43442881e-01 7.57939577e-01 -4.24164772e-01 1.70725122e-01
-1.28852201e+00 1.19218004e+00 2.65281051e-01 5.27123749e-01
-3.70771617e-01 -6.45691380e-02 -9.08618033e-01 -1.03817783e-01
1.41433971e-02 1.59820747e-02 9.91665304e-01 -1.64708483e+00
-1.54556584e+00 7.64950693e-01 -2.01358408e-01 -5.77827156e-01
5.88722876e-04 5.54777762e-05 -8.06981087e-01 -1.08023807e-01
-1.82495296e-01 6.84454858e-01 8.48283529e-01 -7.65527606e-01
-5.69072545e-01 -5.86257577e-01 -1.36067241e-01 3.37890744e-01
-1.01726818e+00 6.72257543e-01 2.87913442e-01 -6.94147587e-01
7.79196396e-02 -8.60848010e-01 -1.61806524e-01 -5.34133971e-01
-1.15861081e-01 -2.22295552e-01 9.47079659e-01 -7.78733075e-01
1.12102187e+00 -2.04108357e+00 -2.57727623e-01 7.84168616e-02
-3.39260459e-01 5.35033047e-01 -1.89311206e-01 6.23353601e-01
1.24036551e-01 3.59010637e-01 -5.94668090e-02 -5.37999459e-02
-1.80272922e-01 1.67618260e-01 -2.28856668e-01 3.09660375e-01
4.16887522e-01 8.17480445e-01 -5.10508478e-01 -4.66346592e-01
7.44952783e-02 5.95530987e-01 1.81516975e-01 -3.55169892e-01
-1.01176962e-01 1.98537678e-01 -2.81033665e-01 8.93201709e-01
4.22877073e-01 7.36738592e-02 -5.02038142e-03 -1.51638195e-01
-2.99615920e-01 2.54440814e-01 -7.55633354e-01 1.42497146e+00
-8.29704762e-01 1.00079834e+00 -1.29015058e-01 -1.07320845e+00
1.07383049e+00 6.23583734e-01 1.47791475e-01 -8.55391204e-01
7.31124401e-01 5.03682911e-01 2.59554923e-01 -6.48622811e-01
7.48706698e-01 -2.86903381e-01 -5.11988439e-02 4.81876552e-01
2.12238237e-01 -9.74717066e-02 2.81947017e-01 -2.07678020e-01
2.11571902e-01 -2.60307133e-01 1.94195405e-01 -2.82838464e-01
8.56419027e-01 9.29171219e-02 2.41977442e-02 2.99343854e-01
-2.40370825e-01 4.40914512e-01 3.12394857e-01 -7.17743874e-01
-7.49419391e-01 -9.11770836e-02 -1.36996239e-01 1.66535366e+00
-5.03947616e-01 -1.78749077e-02 -5.61810434e-01 -4.09219682e-01
-4.41627204e-01 2.92471141e-01 -5.77094734e-01 8.49002972e-02
-6.05445147e-01 -1.36504877e+00 7.37423658e-01 4.85233188e-01
5.77779889e-01 -1.77899754e+00 -6.60278320e-01 2.51024306e-01
-7.35064000e-02 -1.05073845e+00 6.42839298e-02 4.82675731e-01
-1.05452585e+00 -8.03547382e-01 -6.34109914e-01 -1.40883660e+00
4.81426746e-01 -3.22894678e-02 9.77856696e-01 5.10386899e-02
4.89642322e-02 -2.26676285e-01 -6.51180565e-01 -1.00204539e+00
-4.38628405e-01 6.84498012e-01 6.74048066e-02 6.80741817e-02
6.59384847e-01 4.30761836e-02 -1.10338040e-01 -1.12553556e-02
-1.07324791e+00 -3.21168751e-01 5.62469244e-01 8.27322304e-01
-3.26797226e-03 1.52935550e-01 1.11417973e+00 -7.42559493e-01
1.06590474e+00 -5.00590920e-01 -7.74555802e-02 1.46265015e-01
-5.16945481e-01 -1.66188553e-01 9.07369733e-01 -2.25249633e-01
-1.15388691e+00 -2.05637127e-01 -5.26142061e-01 5.83617210e-01
-3.81917119e-01 1.11912978e+00 2.99168646e-01 -1.95023879e-01
6.02399051e-01 2.51401216e-01 3.62849027e-01 -1.97184354e-01
-1.37066588e-01 1.17178881e+00 -1.35281488e-01 -1.13688260e-01
4.68542986e-02 3.73125494e-01 -1.48238152e-01 -1.15598559e+00
-9.41449881e-01 -1.04745351e-01 -8.63122165e-01 -9.27417055e-02
8.33375871e-01 -7.30625331e-01 -6.58465147e-01 1.05903840e+00
-8.71327102e-01 -1.75721481e-01 2.65260071e-01 3.35057348e-01
-1.87950768e-02 1.58110429e-02 -8.96614015e-01 -8.01948309e-01
-8.20569456e-01 -1.33909988e+00 6.57081366e-01 2.86219001e-01
-2.30444863e-01 -1.36296928e+00 -3.84854138e-01 2.33981490e-01
7.47773349e-01 1.40867829e-01 9.06568229e-01 -9.32274401e-01
3.51577282e-01 -5.25070190e-01 -9.28626582e-02 5.36096752e-01
2.52055943e-01 3.19816560e-01 -1.11227882e+00 -2.17574015e-01
8.05682391e-02 -6.12904549e-01 8.15160453e-01 5.26948273e-01
8.65104914e-01 -2.91134953e-01 1.89225554e-01 1.69263855e-01
1.25685024e+00 4.88989472e-01 5.29400766e-01 8.67954612e-01
5.56905270e-01 1.10145640e+00 7.48619914e-01 2.44861156e-01
4.48328733e-01 1.37532772e-02 3.28689843e-01 -1.42559782e-01
1.84001341e-01 5.54883957e-01 6.75792992e-01 1.27487850e+00
2.69875526e-01 -3.47428590e-01 -1.24186325e+00 6.61269963e-01
-1.41600478e+00 -7.34639645e-01 -3.96002799e-01 1.60936964e+00
7.39884138e-01 4.34387513e-02 6.76629171e-02 6.12658799e-01
3.93337160e-01 1.84562072e-01 -1.74586833e-01 -1.24467516e+00
-4.37496722e-01 2.95263112e-01 3.24350804e-01 3.55308682e-01
-1.17513442e+00 1.21015131e+00 6.03352594e+00 7.72643089e-01
-1.79704118e+00 9.02841389e-02 8.70370209e-01 2.56436709e-02
1.40235662e-01 -6.24891043e-01 -6.55036330e-01 3.68552625e-01
1.30897570e+00 3.04628074e-01 4.49634716e-02 4.00387168e-01
2.78176278e-01 -3.27052951e-01 -3.66442323e-01 6.31622255e-01
4.06002194e-01 -1.07562482e+00 3.15664202e-01 -3.24205220e-01
6.53113484e-01 3.66025895e-01 5.76254189e-01 2.36387730e-01
-2.91190714e-01 -1.41672444e+00 4.96466964e-01 2.77120352e-01
5.87671041e-01 -1.29279792e+00 1.29544592e+00 1.79250672e-01
-8.47880483e-01 -4.02753770e-01 -2.29068667e-01 -3.81009072e-01
-1.68465585e-01 2.30121002e-01 -9.38907146e-01 2.45012701e-01
9.34445441e-01 7.85861015e-01 -8.47866654e-01 4.73640740e-01
1.31688759e-01 6.80885315e-01 -1.66258514e-01 -5.25668263e-01
7.84535468e-01 -3.04763377e-01 -1.51965078e-02 1.48310161e+00
2.72314370e-01 -3.24420780e-01 -2.01094136e-01 -2.26030778e-02
8.85570571e-02 6.86551332e-01 -6.13232374e-01 -4.70340550e-01
-9.72991139e-02 1.34942102e+00 -1.23873341e+00 -1.76651284e-01
-4.31227714e-01 6.01330876e-01 -1.29780918e-01 2.00516820e-01
-3.48413706e-01 -6.79201365e-01 2.86191285e-01 -3.20591420e-01
-1.58630069e-02 -2.58836508e-01 -7.54554808e-01 -7.74311900e-01
-1.71590015e-01 -1.20308447e+00 4.89858776e-01 -6.99744940e-01
-1.03037620e+00 1.07059312e+00 -6.32769167e-01 -1.07882655e+00
-9.75944623e-02 -1.14001501e+00 -3.66851568e-01 7.23406911e-01
-1.59330702e+00 -1.60904825e+00 -5.08673750e-02 6.40490770e-01
8.91899645e-01 -6.98421955e-01 1.03986669e+00 3.24752390e-01
-5.69127381e-01 4.62328643e-01 2.23044276e-01 2.53871560e-01
6.61923349e-01 -1.01883519e+00 2.40091216e-02 6.67723000e-01
6.89683761e-03 5.59561014e-01 4.00396973e-01 -3.23022276e-01
-1.35743666e+00 -8.60940516e-01 1.30020583e+00 -2.35547841e-01
7.22033501e-01 -1.64553896e-01 -6.62972510e-01 5.49818337e-01
6.30427599e-01 -4.92834538e-01 9.84889627e-01 1.35141104e-01
-1.02781206e-02 -2.27300793e-01 -1.15367293e+00 6.24146998e-01
2.27289964e-02 -5.14742851e-01 -3.02305013e-01 3.53608668e-01
2.62851059e-01 -1.79762885e-01 -7.37444818e-01 2.50604123e-01
7.01190233e-01 -8.16189408e-01 7.05643475e-01 -5.64396620e-01
6.54628396e-01 7.54359514e-02 -2.21136417e-02 -1.34060240e+00
3.59745532e-01 -1.18455194e-01 2.88939327e-01 1.14112270e+00
7.67586470e-01 -8.09854090e-01 5.00066102e-01 1.67777807e-01
-6.15661219e-02 -8.51682425e-01 -4.18674886e-01 -2.39445925e-01
3.58212084e-01 -6.12764239e-01 4.82577980e-01 1.04984450e+00
1.34816989e-01 4.02900338e-01 -3.40981930e-01 -2.55247533e-01
-1.31656572e-01 -3.55341248e-02 2.67853528e-01 -1.18145370e+00
6.35521293e-01 -7.03779757e-01 -2.91264594e-01 -1.71233982e-01
2.64755338e-01 -7.56984532e-01 -1.37064546e-01 -1.64861119e+00
-2.89261013e-01 -2.66676039e-01 -2.17213482e-01 5.65381050e-01
1.90683976e-01 4.55824405e-01 1.00067608e-01 -1.05438501e-01
-2.94420093e-01 2.77857721e-01 1.16772974e+00 -2.06585780e-01
-2.64296144e-01 1.07593440e-01 -7.86936820e-01 8.68055046e-01
1.32595682e+00 -4.70988274e-01 -1.70167997e-01 -7.24423647e-01
8.31625760e-01 -2.41837457e-01 -3.47939938e-01 -8.09250832e-01
2.41039887e-01 3.32073644e-02 5.57693779e-01 -6.86300874e-01
3.14547598e-01 -7.95354962e-01 -5.69222808e-01 6.27862155e-01
-3.05822790e-01 5.76543748e-01 5.31956017e-01 -4.67695035e-02
-6.37122095e-01 -4.47036237e-01 6.87325060e-01 -2.61921227e-01
-9.07040119e-01 2.10214570e-01 -1.10926402e+00 -4.24987555e-01
6.53984785e-01 -4.55489874e-01 -7.57358894e-02 -4.11789358e-01
-7.35867262e-01 7.41218403e-02 6.56920075e-02 7.69006670e-01
5.90282381e-01 -9.94822502e-01 -6.32956445e-01 2.17252061e-01
3.43738832e-02 -2.97141612e-01 -3.88767384e-02 1.06542444e+00
-1.00745392e+00 6.39044583e-01 -5.26598394e-01 -2.33259603e-01
-1.42939854e+00 1.93311676e-01 2.92196900e-01 -1.12653151e-01
3.97568941e-02 9.45735037e-01 -5.61099291e-01 -9.04489815e-01
1.87188357e-01 -3.76104504e-01 -1.14584708e+00 9.16485190e-01
4.85814869e-01 3.61975819e-01 6.81657374e-01 -1.12976086e+00
-3.97573650e-01 4.33339059e-01 -8.68806466e-02 -3.55990715e-02
1.45688760e+00 -1.35646641e-01 -6.11442804e-01 7.78433502e-01
1.23757100e+00 -5.72581738e-02 -3.17880839e-01 2.38419790e-02
5.25451839e-01 1.56354066e-02 4.49525677e-02 -9.47281420e-01
-1.10583043e+00 1.20496428e+00 6.50683761e-01 4.02871877e-01
1.16317832e+00 -6.57147408e-01 8.23892593e-01 8.39516401e-01
3.35919298e-02 -1.43528450e+00 1.11422516e-01 1.27567434e+00
8.27451050e-01 -1.56392789e+00 -3.02442133e-01 2.05373213e-01
-8.82579327e-01 1.61088717e+00 4.88104314e-01 -5.72255068e-02
8.97563815e-01 2.48164937e-01 6.89522147e-01 -3.77867192e-01
-5.15076518e-01 -2.03739360e-01 2.07197368e-01 5.16816139e-01
1.05701637e+00 -8.91754106e-02 -4.63922560e-01 5.19004524e-01
-5.69327593e-01 -2.53485888e-01 7.39474595e-01 1.06296575e+00
-4.11275417e-01 -9.46291089e-01 -4.93306041e-01 7.00446725e-01
-1.09731197e+00 -4.04160887e-01 -5.28135300e-01 5.27919412e-01
-3.61611247e-02 1.37562895e+00 6.61745071e-02 -2.66795635e-01
8.58610123e-03 2.35437527e-01 1.29448757e-01 -5.60152769e-01
-1.29351068e+00 -1.87222824e-01 1.21775098e-01 -2.22774390e-02
-7.58947790e-01 -5.94229639e-01 -1.27991772e+00 -3.84530395e-01
-5.20858586e-01 1.16584904e-01 1.24194300e+00 1.24022233e+00
2.18963340e-01 4.95910883e-01 5.78659892e-01 -7.32537389e-01
-2.69706279e-01 -1.35430872e+00 -5.77158809e-01 1.86578497e-01
4.11285877e-01 -1.89274281e-01 -4.36456427e-02 1.51507184e-01]
|
[11.15522289276123, 7.060948371887207]
|
d7ee4113-c9f0-4a60-9588-70ff623ca589
|
on-metrics-to-assess-the-transferability-of
|
1912.06200
| null |
https://arxiv.org/abs/1912.06200v1
|
https://arxiv.org/pdf/1912.06200v1.pdf
|
On Metrics to Assess the Transferability of Machine Learning Models in Non-Intrusive Load Monitoring
|
To assess the performance of load disaggregation algorithms it is common practise to train a candidate algorithm on data from one or multiple households and subsequently apply cross-validation by evaluating the classification and energy estimation performance on unseen portions of the dataset derived from the same households. With an emerging discussion of transferability in Non-Intrusive Load Monitoring (NILM), there is a need for domain-specific metrics to assess the performance of NILM algorithms on new test scenarios being unseen buildings. In this paper, we discuss several metrics to assess the generalisation ability of NILM algorithms. These metrics target different aspects of performance evaluation in NILM and are meant to complement the traditional performance evaluation approach. We demonstrate how our metrics can be utilised to evaluate NILM algorithms by means of two case studies. We conduct our studies on several energy consumption datasets and take into consideration five state-of-the-art as well as four baseline NILM solutions. Finally, we formulate research challenges for future work.
|
['Wilfried Elmenreich', 'Stephen Makonin', 'Christoph Klemenjak', 'Anthony Faustine']
|
2019-12-12
| null | null | null | null |
['non-intrusive-load-monitoring', 'non-intrusive-load-monitoring', 'non-intrusive-load-monitoring']
|
['knowledge-base', 'miscellaneous', 'time-series']
|
[ 2.20177829e-01 -1.21519841e-01 4.26769964e-02 -4.87902403e-01
-8.05035710e-01 -6.61948144e-01 8.02871883e-01 4.11629051e-01
-2.25161895e-01 8.54277372e-01 4.12530266e-02 -1.52432650e-01
-3.60482842e-01 -1.00466239e+00 -3.00417066e-01 -7.82511830e-01
-3.52381170e-01 5.96140325e-01 4.74366471e-02 6.41419459e-03
-1.27603039e-01 4.30118859e-01 -1.81213748e+00 9.14284438e-02
7.88151860e-01 9.90578294e-01 -2.94384416e-02 6.17088914e-01
6.12187147e-01 6.42299354e-01 -9.81305242e-01 -1.26579612e-01
9.79973748e-02 -3.14927161e-01 -9.33039367e-01 -8.00713003e-02
3.82867008e-02 -2.78769910e-01 2.09880814e-01 7.06588387e-01
8.88591528e-01 1.71399400e-01 7.82048166e-01 -1.86907804e+00
2.09398940e-02 6.17110550e-01 -1.27081886e-01 4.19351876e-01
7.21034646e-01 1.73486874e-01 8.10918689e-01 -1.60355181e-01
-8.31453130e-02 8.24030638e-01 9.08691585e-01 7.99945593e-02
-1.58840394e+00 -6.85570657e-01 -2.36920565e-01 4.08662140e-01
-1.44078469e+00 -5.90935349e-01 6.80043399e-01 -1.59722552e-01
1.47476602e+00 1.00469983e+00 5.61608911e-01 1.07585561e+00
-1.15905970e-01 8.20342183e-01 1.47542500e+00 -5.72039664e-01
8.08357954e-01 2.44711712e-01 1.48237526e-01 2.43360281e-01
5.13360322e-01 3.49140555e-01 -2.16305941e-01 -4.17900831e-01
-2.71323293e-01 -4.53501344e-01 -1.11343406e-01 -4.38105941e-01
-9.07455742e-01 8.60003293e-01 1.76656544e-01 5.15263796e-01
-1.70364454e-01 9.84743014e-02 5.61709523e-01 -4.54542115e-02
4.06540722e-01 3.34204823e-01 -5.23105145e-01 -3.18844616e-01
-1.55500424e+00 1.35472134e-01 1.13982022e+00 6.51252389e-01
8.07402074e-01 -1.09491616e-01 -1.30576193e-01 5.82909226e-01
3.00295860e-01 4.85768497e-01 6.04197502e-01 -8.14695418e-01
1.61836714e-01 6.72426164e-01 1.11453824e-01 -4.75545555e-01
-9.45836008e-01 -1.02249421e-01 -8.41106176e-01 1.77121744e-01
4.12664935e-02 -3.23876411e-01 -5.14795959e-01 1.73531806e+00
1.67480469e-01 6.23625293e-02 1.44662827e-01 2.46480688e-01
6.09901965e-01 8.24314773e-01 3.67981672e-01 -5.80878556e-01
8.65987420e-01 -6.45515680e-01 -5.57362974e-01 1.81814656e-02
7.29864955e-01 -4.75087911e-01 8.22639287e-01 3.01553667e-01
-1.02289808e+00 -3.96245390e-01 -1.17719054e+00 3.91712159e-01
-9.37226236e-01 -1.79198608e-01 1.99736670e-01 1.15067804e+00
-9.78447974e-01 7.53570259e-01 -1.01491499e+00 -1.12767279e+00
1.88870504e-01 5.07631302e-01 3.07055503e-01 3.44845057e-01
-1.04356337e+00 1.25215292e+00 9.85974371e-01 -2.02114508e-01
-9.26630020e-01 -6.70536041e-01 -7.11282253e-01 -1.21729799e-01
-3.30730751e-02 -4.21475381e-01 1.27144980e+00 -3.46330285e-01
-1.05134380e+00 7.65097260e-01 2.95941144e-01 -5.94130456e-01
5.57778060e-01 6.04207963e-02 -9.52004910e-01 -3.33309352e-01
1.56145245e-01 2.87503660e-01 1.22659579e-01 -1.39680445e+00
-1.13336170e+00 -1.44096300e-01 -7.36381710e-02 -1.85032323e-01
-2.07149655e-01 -3.13726962e-01 2.86841929e-01 -4.59201097e-01
-4.16733384e-01 -8.06749165e-01 1.82342455e-01 -8.33775818e-01
-5.95176518e-01 -2.52546340e-01 1.21674919e+00 -5.36725998e-01
1.63357806e+00 -1.86497533e+00 -2.31771439e-01 4.92807865e-01
-4.16979432e-01 1.64690167e-01 2.17300847e-01 6.37627125e-01
8.59339535e-03 1.99586861e-02 -5.00894547e-01 -1.41185746e-01
6.86773658e-01 5.42842150e-01 1.45550445e-01 4.35233206e-01
3.57315727e-02 1.05801034e+00 -1.01975358e+00 -3.86188626e-01
8.97902131e-01 3.87807012e-01 1.10724613e-01 3.23711067e-01
-2.44327020e-02 1.06384598e-01 -1.75825581e-01 6.91393554e-01
3.72800082e-01 -8.93779472e-02 3.07129025e-01 -6.03639305e-01
-8.58217012e-03 2.98822522e-01 -1.34649527e+00 1.36649275e+00
-6.53048694e-01 4.54799533e-01 -3.69617701e-01 -1.11963034e+00
5.43300033e-01 3.55511785e-01 6.64892018e-01 -8.01605999e-01
2.48948619e-01 3.42655987e-01 -2.64966190e-01 -2.79502779e-01
1.72243685e-01 9.06615853e-02 -5.31518757e-01 5.55327058e-01
9.73545313e-02 3.23853940e-02 4.66561377e-01 -4.95649725e-01
1.26818693e+00 4.31270376e-02 6.63882077e-01 -8.49164963e-01
6.26762331e-01 1.58139016e-03 1.85123786e-01 6.32596195e-01
-3.78514200e-01 -8.45018309e-04 -2.25101322e-01 -4.67323840e-01
-8.03046107e-01 -1.24720144e+00 -4.58622843e-01 1.11635661e+00
-1.19225614e-01 -4.26283777e-01 -1.00596654e+00 -7.24066913e-01
3.54562886e-02 1.53643239e+00 -7.08249390e-01 -3.07869554e-01
-4.32893127e-01 -1.87061143e+00 5.11302948e-01 5.21892667e-01
8.09701324e-01 -9.74080265e-01 -1.47273099e+00 4.57859993e-01
-3.23644310e-01 -8.46265972e-01 2.54916400e-01 8.89727175e-01
-5.46407342e-01 -1.19503915e+00 -8.06635022e-02 -4.19720411e-01
2.09304377e-01 -1.36960045e-01 1.97792518e+00 -3.10670063e-02
-3.04176331e-01 6.66012287e-01 -4.33498263e-01 -5.83520770e-01
-7.51639724e-01 6.61553681e-01 2.96138991e-02 -3.95033777e-01
1.04051065e+00 -9.49254036e-01 -7.02376962e-01 6.59681797e-01
-8.31464648e-01 -3.10987860e-01 2.92548448e-01 1.23022206e-01
4.08193946e-01 8.13502669e-01 5.37561774e-01 -5.51223159e-01
4.43618685e-01 -7.23677397e-01 -6.09363675e-01 4.73707676e-01
-1.16777086e+00 -2.16553267e-02 4.23933864e-01 -1.89854160e-01
-7.23206460e-01 -6.03267225e-03 -1.35995895e-01 3.11854810e-01
-4.38856214e-01 -9.61987898e-02 -5.81958532e-01 2.34245673e-01
5.45192420e-01 5.79743972e-03 -7.07189620e-01 -4.67261434e-01
2.71539897e-01 4.05390799e-01 6.12093866e-01 -8.08814347e-01
9.52367961e-01 2.63510227e-01 1.07264154e-01 -6.09642982e-01
-6.74451292e-01 -3.91275853e-01 -6.24914467e-01 -2.50972867e-01
9.66988325e-01 -5.89945316e-01 -6.66589677e-01 3.94422680e-01
-5.37775517e-01 -6.79052651e-01 -6.88384533e-01 -1.15533307e-01
-6.21691048e-01 -2.23186556e-02 -5.17401025e-02 -9.86271858e-01
-4.57981735e-01 -1.08406472e+00 1.18370485e+00 7.92043209e-02
-8.28646541e-01 -1.28006482e+00 3.41927409e-01 1.41551480e-01
3.77851278e-01 9.11707938e-01 9.68028128e-01 -8.32976699e-01
-1.35255277e-01 -1.25480935e-01 2.06613213e-01 2.41955295e-01
3.79673243e-01 -1.95066839e-01 -1.52534997e+00 -8.27551544e-01
3.99227031e-02 -1.87237754e-01 4.59276050e-01 1.49909079e-01
9.49851274e-01 -3.00630569e-01 -5.76560020e-01 3.29987735e-01
1.99380028e+00 2.44688347e-01 6.19411588e-01 6.89054430e-01
2.81079233e-01 4.99018908e-01 3.61971766e-01 3.64340872e-01
4.67869461e-01 6.05413496e-01 4.55063254e-01 -1.47562817e-01
1.29726201e-01 5.36133461e-02 3.27938676e-01 8.34568024e-01
-4.31898469e-03 -7.15217948e-01 -8.35166037e-01 6.49141073e-01
-1.78054857e+00 -8.64548504e-01 -1.64097976e-02 2.16897440e+00
5.43167710e-01 2.80024588e-01 6.77708864e-01 9.76588786e-01
3.65693569e-01 9.61647406e-02 -6.09382808e-01 -5.48831522e-01
-1.00081086e-01 3.61266345e-01 5.86398423e-01 1.53968766e-01
-1.29117560e+00 -1.90930024e-01 7.25060654e+00 7.63596237e-01
-3.91650081e-01 3.86568844e-01 7.48507380e-01 -1.96316093e-01
2.07464606e-01 -4.35038447e-01 -6.15973949e-01 7.82342732e-01
1.71780682e+00 -1.99580684e-01 5.20980716e-01 7.06981778e-01
1.59361422e-01 -5.50768435e-01 -1.76589739e+00 8.85546327e-01
-1.40093610e-01 -7.41098285e-01 -4.44164604e-01 2.28300825e-01
9.46287334e-01 1.45952880e-01 -4.57874268e-01 5.38803577e-01
6.41901553e-01 -9.11419034e-01 4.48199868e-01 3.55973512e-01
3.18281472e-01 -8.14395130e-01 8.95576715e-01 3.33577693e-01
-1.61689222e+00 -1.66759506e-01 6.08639512e-03 -6.12323806e-02
-1.10860029e-02 5.84630430e-01 -6.73639119e-01 8.80181313e-01
1.16175711e+00 1.81106299e-01 -1.00431347e+00 8.61176610e-01
8.92332345e-02 7.91767180e-01 -9.02587116e-01 1.56912223e-01
2.19743960e-02 1.06349833e-01 1.43388519e-02 1.58909762e+00
3.46896619e-01 -1.69461831e-01 2.38263682e-01 6.56714916e-01
2.39086479e-01 -2.71166209e-02 -3.39265198e-01 6.19352341e-01
5.28726935e-01 1.40630341e+00 -8.52499306e-01 -3.26640338e-01
-5.20673394e-01 8.99733603e-01 -1.72944427e-01 6.40627667e-02
-9.56241488e-01 5.13796844e-02 5.08448720e-01 1.22532941e-01
9.00924280e-02 4.08086509e-01 -7.14066252e-02 -7.03761339e-01
-9.97417718e-02 -5.99466205e-01 7.13286281e-01 -5.91941357e-01
-1.48432827e+00 1.65157899e-01 6.65789485e-01 -8.83803189e-01
-6.12059176e-01 -2.56114662e-01 -8.45586896e-01 5.09143829e-01
-1.33715892e+00 -1.01651680e+00 -6.34857416e-01 2.25554839e-01
4.14166361e-01 5.58669157e-02 1.20586205e+00 5.66213608e-01
-8.34538400e-01 4.66182262e-01 5.24091780e-01 -1.25413939e-01
1.01554662e-01 -1.77479291e+00 5.25334775e-01 7.45861292e-01
5.71258739e-03 -3.53319526e-01 9.70436692e-01 -2.82555640e-01
-8.22478175e-01 -1.59970558e+00 4.67026353e-01 -9.82321978e-01
3.71411532e-01 -4.71583962e-01 -7.14769483e-01 6.59162402e-01
4.40404266e-01 -3.00090849e-01 1.04747057e+00 -2.43243530e-01
3.13666821e-01 -3.44095796e-01 -1.61895537e+00 -1.03998937e-01
8.47600818e-01 -2.42586359e-01 -6.22936964e-01 2.78732330e-01
-3.02689429e-03 9.15365145e-02 -1.45460832e+00 7.48535395e-01
4.45411384e-01 -1.35098767e+00 9.86637771e-01 -1.29349073e-02
-3.55153590e-01 -4.59419966e-01 -8.61199439e-01 -1.31934834e+00
-3.39674473e-01 -5.37004888e-01 -6.96279347e-01 1.94422841e+00
1.88214198e-01 -5.45789719e-01 5.95227480e-01 7.64109433e-01
3.69728684e-01 -5.27116060e-01 -9.42460835e-01 -1.02224934e+00
1.78299069e-01 -4.45292681e-01 1.23021221e+00 7.48873949e-01
-1.53158575e-01 3.60346615e-01 8.73938501e-02 3.36334616e-01
9.03421104e-01 -8.18220340e-03 5.37071884e-01 -1.39031041e+00
-2.55766455e-02 -4.61192638e-01 -5.89998364e-01 -1.01754509e-01
-3.75594608e-02 -8.11669171e-01 -3.60256769e-02 -1.54458213e+00
3.23946029e-01 -1.04687870e-01 -7.82358348e-01 4.75673378e-01
1.78855777e-01 6.05254531e-01 2.00663358e-01 3.57017182e-02
-6.89529419e-01 3.55131030e-01 4.57672514e-02 -3.44224960e-01
-1.39183506e-01 -1.47597892e-02 -3.39781880e-01 7.29912758e-01
1.24324977e+00 -3.76296252e-01 -7.56235480e-01 1.81914475e-02
4.51332256e-02 -6.12764835e-01 5.58579624e-01 -1.71407485e+00
-2.01308697e-01 -1.69995632e-02 8.13398063e-01 -9.76879060e-01
1.20689450e-02 -1.31738794e+00 8.43421042e-01 5.76196015e-01
1.66151911e-01 3.89512539e-01 3.17540169e-01 1.37882143e-01
4.43923444e-01 -3.37119490e-01 8.46924424e-01 -1.42407298e-01
-9.88735437e-01 -2.42902324e-01 -3.72449279e-01 1.22435838e-01
1.32947052e+00 -2.48931542e-01 -1.36580601e-01 7.39879236e-02
-6.36558414e-01 4.41161573e-01 7.05010474e-01 3.02374780e-01
-1.86659262e-01 -1.63450980e+00 -6.15495265e-01 -5.09500876e-03
2.27806643e-01 -2.46919990e-01 -3.81622583e-01 4.31176186e-01
-1.93075761e-01 3.13056290e-01 7.77967274e-03 -7.84516692e-01
-9.26356912e-01 8.08535576e-01 7.48433113e-01 -5.13189375e-01
-1.74057901e-01 -4.90613701e-03 -3.37893218e-02 -4.59912896e-01
2.10056663e-01 -6.20986581e-01 1.34358853e-01 2.70951718e-01
2.58230835e-01 1.13582742e+00 5.99560738e-01 -6.87682509e-01
-7.52053916e-01 3.69788170e-01 6.68152452e-01 2.44653866e-01
1.25338924e+00 -4.12554562e-01 3.13688144e-02 8.64138722e-01
1.35396302e+00 -5.12747645e-01 -8.80078733e-01 2.26093769e-01
4.71865892e-01 1.59656182e-01 8.89406055e-02 -1.37594497e+00
-9.34432626e-01 3.48699778e-01 1.49708867e+00 9.19894218e-01
1.68132138e+00 -1.06460191e-02 6.98903322e-01 1.90433085e-01
5.32717168e-01 -1.47923410e+00 -4.58616883e-01 -3.96569431e-01
5.14275312e-01 -1.21639037e+00 4.14646626e-01 1.41143754e-01
2.40452513e-01 8.80133033e-01 3.37694228e-01 2.70697892e-01
9.35496747e-01 3.62522364e-01 -1.60722196e-01 -2.66283393e-01
-5.17506421e-01 -2.38657534e-01 1.70854002e-01 8.69689703e-01
2.36436605e-01 3.35679382e-01 -7.40657002e-02 3.29092264e-01
-5.31199455e-01 -4.79214117e-02 1.53750420e-01 1.18195522e+00
-3.55700523e-01 -1.06230569e+00 -5.26535869e-01 8.07215154e-01
-3.79046261e-01 4.38878447e-01 -1.82985216e-01 1.17876458e+00
4.69905883e-01 1.49881363e+00 1.45554887e-02 -4.02828991e-01
8.37039649e-01 4.05303121e-01 5.01889825e-01 -1.46430403e-01
-6.41443670e-01 -3.19837362e-01 3.16993535e-01 -5.36725760e-01
-9.97477353e-01 -1.04185140e+00 -7.85387397e-01 -5.72013497e-01
-3.89416665e-01 2.03804284e-01 5.10150075e-01 8.53017628e-01
-1.45599648e-01 8.28101337e-01 8.36464942e-01 -9.48805332e-01
-3.98203611e-01 -9.01110828e-01 -4.98173803e-01 7.02941000e-01
1.39580637e-01 -7.59375036e-01 -3.85476828e-01 6.00720122e-02]
|
[6.0477752685546875, 2.6108806133270264]
|
00c92588-b4d1-44b9-95db-8c20fd7e4904
|
segmenting-transparent-object-in-the-wild
|
2101.08461
| null |
https://arxiv.org/abs/2101.08461v3
|
https://arxiv.org/pdf/2101.08461v3.pdf
|
Segmenting Transparent Object in the Wild with Transformer
|
This work presents a new fine-grained transparent object segmentation dataset, termed Trans10K-v2, extending Trans10K-v1, the first large-scale transparent object segmentation dataset. Unlike Trans10K-v1 that only has two limited categories, our new dataset has several appealing benefits. (1) It has 11 fine-grained categories of transparent objects, commonly occurring in the human domestic environment, making it more practical for real-world application. (2) Trans10K-v2 brings more challenges for the current advanced segmentation methods than its former version. Furthermore, a novel transformer-based segmentation pipeline termed Trans2Seg is proposed. Firstly, the transformer encoder of Trans2Seg provides the global receptive field in contrast to CNN's local receptive field, which shows excellent advantages over pure CNN architectures. Secondly, by formulating semantic segmentation as a problem of dictionary look-up, we design a set of learnable prototypes as the query of Trans2Seg's transformer decoder, where each prototype learns the statistics of one category in the whole dataset. We benchmark more than 20 recent semantic segmentation methods, demonstrating that Trans2Seg significantly outperforms all the CNN-based methods, showing the proposed algorithm's potential ability to solve transparent object segmentation.
|
['Ping Luo', 'Ding Liang', 'Hang Xu', 'Peize Sun', 'Wenhai Wang', 'Wenjia Wang', 'Enze Xie']
|
2021-01-21
| null | null | null | null |
['transparent-objects']
|
['computer-vision']
|
[ 1.77955609e-02 3.49639416e-01 -6.13647513e-03 -4.48564351e-01
-5.57499051e-01 -6.07975721e-01 2.62983322e-01 -2.88947344e-01
-1.72440216e-01 3.40845972e-01 7.94371143e-02 -1.40902251e-01
3.14200133e-01 -1.02508402e+00 -7.17177570e-01 -8.03960264e-01
3.18597645e-01 4.24923658e-01 8.50854814e-01 -1.67030931e-01
2.34672591e-01 1.07377626e-01 -1.25542068e+00 3.29997391e-01
1.03962171e+00 1.61804652e+00 6.36359632e-01 2.27482066e-01
-3.58262420e-01 3.12177837e-01 -6.28312409e-01 -4.92724478e-01
5.33163488e-01 -6.10657297e-02 -1.17138314e+00 -6.02536052e-02
8.20085108e-01 -2.61971563e-01 4.03278433e-02 1.05623424e+00
2.99090683e-01 6.08368739e-02 7.09775329e-01 -7.21282005e-01
-9.63074446e-01 6.30786538e-01 -3.32260489e-01 -2.47645993e-02
-8.81192535e-02 1.40114546e-01 1.09177327e+00 -7.47315526e-01
6.61742270e-01 1.21559393e+00 7.64727235e-01 6.68716252e-01
-8.82869959e-01 -4.36994582e-01 5.61833680e-01 -5.28690815e-02
-1.17737889e+00 -1.33643433e-01 5.00423074e-01 -5.52659392e-01
1.00993073e+00 2.54286796e-01 1.08016551e+00 7.12440372e-01
3.50001790e-02 1.18170094e+00 1.11950827e+00 1.29744977e-01
3.51604581e-01 2.41688229e-02 2.54138827e-01 7.94749796e-01
4.40037716e-03 -2.12454796e-01 -2.51321644e-01 2.93678731e-01
1.00781441e+00 -2.88930386e-01 -1.16027795e-01 -2.43575215e-01
-1.32266998e+00 7.45412409e-01 8.27065945e-01 2.83904225e-01
-4.84506264e-02 4.12275374e-01 5.03261089e-01 2.41328076e-01
5.62308669e-01 3.75674695e-01 -6.77990258e-01 1.16294965e-01
-9.15890694e-01 1.48107782e-01 6.16381049e-01 1.35777795e+00
8.86989951e-01 1.86325476e-01 -3.53989959e-01 9.83538985e-01
4.81058925e-01 4.10899401e-01 6.24703705e-01 -7.92109787e-01
5.66853702e-01 6.77146614e-01 -3.00693393e-01 -5.94137728e-01
-1.82334989e-01 -2.91474283e-01 -6.56767428e-01 -2.46617034e-01
4.12844300e-01 3.08141351e-01 -1.36235392e+00 1.37699962e+00
5.08997262e-01 -1.84204981e-01 1.92641407e-01 9.88905549e-01
1.50768805e+00 7.24583149e-01 -7.10571790e-03 2.64232367e-01
1.41702962e+00 -1.48826575e+00 -3.23864728e-01 -3.18498433e-01
4.55913037e-01 -8.25514495e-01 1.17646945e+00 3.75611544e-01
-8.91871393e-01 -6.47700608e-01 -6.40732765e-01 -4.53312367e-01
-6.71208680e-01 7.97216818e-02 1.20825613e+00 7.51496375e-01
-1.41951644e+00 2.50475377e-01 -5.33932269e-01 -4.88134950e-01
8.29082668e-01 5.84535241e-01 -4.04389314e-02 1.11550801e-01
-8.44721675e-01 4.46816415e-01 5.93591213e-01 3.45546216e-01
-9.80283618e-01 -7.08841980e-01 -8.81772578e-01 -1.53252095e-01
3.92579377e-01 -5.75919569e-01 1.25726950e+00 -6.67149186e-01
-1.56074107e+00 1.15294552e+00 -7.83958212e-02 -2.15929821e-01
4.51419055e-01 -1.23382032e-01 -1.98225111e-01 3.20971578e-01
3.51263136e-01 1.15291572e+00 1.09366882e+00 -1.34742689e+00
-5.91430783e-01 -1.23810276e-01 2.04596132e-01 6.62499517e-02
-2.53398746e-01 -2.79306233e-01 -1.07186687e+00 -9.42231536e-01
3.89591902e-01 -6.63717210e-01 -2.79275030e-01 1.05539523e-01
-8.85653675e-01 -4.06352192e-01 9.85391498e-01 -5.30887485e-01
7.25643635e-01 -2.18686342e+00 1.30641162e-01 1.94517970e-01
4.21333760e-01 3.63514274e-01 -1.70356259e-01 9.31891650e-02
2.90813535e-01 1.12793565e-01 -6.17047191e-01 -4.28626001e-01
1.17066659e-01 2.58014381e-01 -3.60176533e-01 1.31541654e-01
-8.84515513e-03 1.39716029e+00 -6.19044304e-01 -8.08515012e-01
3.12627882e-01 1.79913461e-01 -6.39629126e-01 8.84567425e-02
-5.49233675e-01 4.36121017e-01 -7.30496109e-01 9.45212781e-01
8.77984345e-01 -1.04505219e-01 -3.95440549e-01 -4.86655563e-01
-1.54859230e-01 1.15915753e-01 -7.02675998e-01 1.92174888e+00
-2.94587433e-01 6.01523340e-01 6.29661754e-02 -1.11197531e+00
1.02578437e+00 -1.11791296e-02 3.35135460e-01 -1.14767897e+00
1.20518990e-01 3.24028432e-01 -3.91376346e-01 -4.30700630e-01
7.35121667e-01 4.33720388e-02 -2.35931516e-01 1.66328087e-01
1.92571595e-01 -8.35452676e-01 2.00536966e-01 2.22934633e-01
5.88398993e-01 3.32835406e-01 -2.62805969e-01 -5.87778866e-01
1.26658231e-01 7.09452182e-02 8.82339597e-01 7.27423012e-01
-2.04587027e-01 9.16666865e-01 2.01031461e-01 -4.95945066e-01
-7.08896935e-01 -1.22654486e+00 -3.27616423e-01 7.79327631e-01
7.95626879e-01 -2.45767683e-01 -1.17151034e+00 -7.28394151e-01
-1.88312121e-02 1.80292383e-01 -7.25115418e-01 1.11399308e-01
-5.74751198e-01 -7.42073953e-01 4.11541641e-01 6.97334170e-01
1.23003709e+00 -1.28462076e+00 -5.37715554e-01 1.28862694e-01
-3.35571468e-01 -1.22339058e+00 -7.81105697e-01 2.80531436e-01
-1.32909846e+00 -1.07102859e+00 -9.71362531e-01 -1.14462912e+00
6.09551907e-01 3.04991573e-01 1.17929494e+00 2.36373078e-02
-2.88427889e-01 2.69576818e-01 -3.37669581e-01 -9.22970921e-02
-7.69339129e-02 1.44318163e-01 -3.42294514e-01 -2.18163133e-01
1.19803600e-01 -1.98543787e-01 -8.11843872e-01 3.59694123e-01
-7.44283557e-01 1.69878215e-01 5.10403514e-01 5.59784114e-01
1.02140808e+00 2.38241162e-02 3.26784194e-01 -1.13095415e+00
1.83489278e-01 -1.79549873e-01 -6.05493605e-01 3.38663816e-01
-1.60741046e-01 -4.55533326e-01 5.99488676e-01 -2.48903185e-01
-1.16627157e+00 -5.94432577e-02 -4.06504601e-01 -1.39893636e-01
1.43564781e-02 1.01208165e-01 -3.79529178e-01 -2.54117131e-01
7.55569488e-02 3.15469474e-01 -6.31625354e-01 -6.61389887e-01
5.45445204e-01 5.98148942e-01 6.36560917e-01 -7.19201922e-01
7.96668410e-01 6.09441638e-01 -5.90004742e-01 -1.10302782e+00
-8.92610908e-01 -6.09036326e-01 -8.96846831e-01 -3.77281569e-02
1.31741667e+00 -8.89990687e-01 -6.34837091e-01 1.06505179e+00
-9.42874610e-01 -9.77364600e-01 -6.95773959e-01 9.48278829e-02
-6.73128128e-01 4.04500365e-01 -7.95639694e-01 -2.64371514e-01
-5.36925614e-01 -1.45762014e+00 1.55011404e+00 3.84813666e-01
8.09252933e-02 -1.16262555e+00 -4.25539792e-01 7.15696394e-01
2.31728256e-01 -6.35280311e-02 1.00147438e+00 -2.43542120e-01
-9.62476432e-01 3.13711017e-01 -5.28914392e-01 4.59439397e-01
2.32818931e-01 -1.20164886e-01 -1.20057034e+00 -2.86240131e-01
-1.52078629e-01 -4.50716019e-01 1.17267990e+00 4.99948233e-01
1.46783864e+00 -1.89447090e-01 -3.11325312e-01 9.15950775e-01
1.45474494e+00 2.22884879e-01 6.65470839e-01 3.23538363e-01
1.17897832e+00 4.50506389e-01 6.94619000e-01 -1.03589624e-01
6.48764193e-01 6.25950336e-01 5.98156989e-01 -4.95395929e-01
-4.52243984e-01 -1.86850399e-01 5.59331998e-02 1.23557043e+00
2.49401890e-02 -2.58930117e-01 -4.94126230e-01 9.02972877e-01
-1.56842721e+00 -3.90327841e-01 -2.69736618e-01 1.85113263e+00
7.08296001e-01 2.07374439e-01 1.39180884e-01 -7.28798062e-02
5.03974438e-01 1.67553574e-01 -7.23585725e-01 -6.31049037e-01
-2.20876202e-01 3.79701674e-01 5.09835899e-01 9.54946429e-02
-1.24131000e+00 1.62747586e+00 6.85558414e+00 1.22956657e+00
-1.11154544e+00 2.13668436e-01 6.16065800e-01 4.90855426e-01
-5.88127613e-01 -6.64473176e-02 -8.57688725e-01 4.65677172e-01
2.26108104e-01 4.74812865e-01 5.46726212e-02 7.67361760e-01
-1.04942963e-01 -2.95657098e-01 -9.96629179e-01 9.40508187e-01
-1.63963377e-01 -1.41926527e+00 2.74045587e-01 8.12042132e-02
9.02130604e-01 3.10281634e-01 4.33568507e-01 1.44382179e-01
3.37701887e-01 -9.53536093e-01 1.25799012e+00 2.18836114e-01
9.17109370e-01 -2.01225519e-01 3.70260030e-01 -7.35079870e-02
-1.55150020e+00 -8.65455437e-03 -6.35829031e-01 5.33747733e-01
2.64535844e-02 7.41376102e-01 -5.71448743e-01 4.21454728e-01
1.27014291e+00 1.00194860e+00 -4.78068322e-01 1.20928049e+00
-7.89153501e-02 8.75523686e-01 -5.12919605e-01 1.68558642e-01
5.83169997e-01 -5.16138196e-01 1.55078053e-01 1.26034844e+00
9.73529741e-02 1.53643987e-03 2.00892389e-01 1.09830964e+00
-2.83743799e-01 1.36231333e-01 -2.81335533e-01 -1.87070128e-02
3.95468444e-01 1.16540730e+00 -1.50580227e+00 -6.92405224e-01
-4.15239096e-01 1.17611110e+00 1.48244172e-01 3.65563571e-01
-5.78553796e-01 -3.27888280e-01 6.19857371e-01 -4.01662558e-01
7.62567163e-01 1.43431379e-02 -3.93094152e-01 -1.14429474e+00
1.37017697e-01 -7.04842269e-01 1.61174595e-01 -7.98120499e-01
-1.09478176e+00 6.01076722e-01 -2.39248395e-01 -1.06379342e+00
5.51797450e-01 -6.54005110e-01 -5.42496383e-01 6.24930561e-01
-1.60689533e+00 -1.60580575e+00 -3.05618405e-01 5.94064116e-01
9.17214692e-01 4.86985082e-03 6.47417605e-01 2.46805057e-01
-4.77662325e-01 6.19080245e-01 1.99178070e-01 2.46295378e-01
2.39685863e-01 -1.44658768e+00 9.40421522e-01 6.81757927e-01
6.95112422e-02 3.89033109e-01 1.44861564e-01 -6.72968686e-01
-1.15365303e+00 -1.41540480e+00 3.66517693e-01 -3.83995086e-01
3.03956032e-01 -7.22017646e-01 -8.03263903e-01 6.41507208e-01
-1.42174721e-01 1.86997615e-02 3.70786250e-01 -2.61435777e-01
-4.91441309e-01 -1.79518521e-01 -1.40645444e+00 4.76158649e-01
1.21680725e+00 -4.97114778e-01 -7.70057082e-01 4.35949683e-01
1.26994669e+00 -6.34024203e-01 -1.19157314e+00 2.29819536e-01
5.19180000e-01 -8.92161071e-01 1.07740939e+00 1.38399035e-01
3.25404406e-01 -6.28574789e-02 -3.75005990e-01 -1.04604518e+00
-7.41327405e-02 -4.27937955e-01 9.14684534e-02 1.34454179e+00
3.41042787e-01 -8.30084085e-01 1.06577730e+00 2.51828253e-01
-7.80901611e-01 -8.83066893e-01 -8.95618260e-01 -6.46313012e-01
3.26869518e-01 -4.34869498e-01 8.09374273e-01 5.95391452e-01
-6.67832196e-01 -2.00398460e-01 -7.85137992e-03 -5.31941876e-02
7.29544759e-01 6.58972204e-01 4.50709760e-01 -1.17069781e+00
-1.61334708e-01 -4.07704830e-01 -2.94248104e-01 -1.93801367e+00
-7.76728317e-02 -8.88969660e-01 3.53039354e-01 -1.95398164e+00
1.56733915e-01 -1.18653727e+00 -1.10019837e-02 7.02809393e-01
9.12181139e-02 9.42852318e-01 1.15423091e-01 2.96738029e-01
-7.88292229e-01 7.50634193e-01 2.04332161e+00 -6.52828813e-01
-2.57365435e-01 4.76409085e-02 -7.47945547e-01 7.14827240e-01
5.88989973e-01 -1.20903835e-01 -5.03971040e-01 -9.44665611e-01
5.14929518e-02 -3.48797798e-01 2.24901959e-01 -9.21915472e-01
1.69020630e-02 -7.21836090e-02 1.12818152e-01 -9.33681309e-01
5.65502465e-01 -6.23782158e-01 -2.77805746e-01 1.84289306e-01
7.58912936e-02 -5.68803191e-01 1.67257428e-01 4.48981792e-01
-4.01597500e-01 -9.76340547e-02 7.93558776e-01 -4.67840791e-01
-1.34786236e+00 7.40018606e-01 -4.71271127e-01 4.05388862e-01
8.40876400e-01 -9.45692420e-01 -2.16595709e-01 3.48303825e-01
-6.95730627e-01 3.44410688e-01 5.84005535e-01 4.75586176e-01
7.58396983e-01 -9.14246559e-01 -1.51830763e-01 5.68341129e-02
2.17107430e-01 7.72315919e-01 3.48706424e-01 5.82076967e-01
-7.16601610e-01 4.35562879e-01 -1.20825000e-01 -7.40963697e-01
-9.06650186e-01 1.85747594e-01 3.45237345e-01 2.01803625e-01
-1.14408529e+00 1.53999007e+00 9.15079117e-01 -8.82388353e-01
1.08295381e-01 -8.69155645e-01 -2.91906685e-01 -6.53761327e-02
-1.13820754e-01 6.16164207e-02 1.86277360e-01 -8.31714094e-01
-2.38737851e-01 1.12779522e+00 -9.58731547e-02 3.60647589e-01
1.30023038e+00 -2.35358536e-01 -4.24242854e-01 3.35630536e-01
1.22450244e+00 -2.93568879e-01 -1.31165397e+00 -2.25937963e-01
-1.30150110e-01 -2.83284873e-01 -7.05443099e-02 -7.23138034e-01
-1.62292993e+00 1.07892847e+00 5.21389365e-01 -4.34415638e-02
1.18103051e+00 3.14068198e-01 1.44638407e+00 4.28035736e-01
5.62287450e-01 -1.31278634e+00 1.09386533e-01 4.86619413e-01
5.21847248e-01 -1.19235504e+00 -2.26226509e-01 -1.17082620e+00
-5.62505841e-01 8.38776052e-01 6.98016584e-01 1.04727589e-01
6.04269803e-01 -5.37755862e-02 4.53415573e-01 -4.98294711e-01
-1.57158941e-01 -2.28207111e-01 1.41909778e-01 8.74667287e-01
5.15211001e-02 2.64052778e-01 1.05330199e-01 4.84612793e-01
-6.29323125e-01 -2.99822122e-01 1.91484213e-01 6.26105428e-01
-4.56290871e-01 -8.39811027e-01 -3.52050215e-02 6.85624480e-01
-2.13692337e-01 -2.08674327e-01 -2.94664055e-01 6.77731872e-01
4.78764355e-01 8.32296312e-01 1.63946182e-01 -1.33900762e-01
1.64998665e-01 -5.20933986e-01 4.34380770e-01 -8.45205247e-01
-7.25707650e-01 2.45320126e-02 -1.66108117e-01 -8.74084711e-01
-5.07963002e-01 -3.45679671e-01 -1.42093241e+00 1.25252947e-01
-4.05147046e-01 3.52689922e-01 7.57229686e-01 1.01871884e+00
2.78674290e-02 5.30907571e-01 3.19518924e-01 -8.48833382e-01
-2.90258974e-02 -7.20837653e-01 -7.32469976e-01 2.09986344e-01
2.95815974e-01 -5.96688211e-01 1.12398647e-01 2.36349463e-01]
|
[9.542881965637207, 0.1917513608932495]
|
e7705b23-4278-4263-8680-fed0563780cb
|
unsupervised-translation-of-programming
|
2006.03511
| null |
https://arxiv.org/abs/2006.03511v3
|
https://arxiv.org/pdf/2006.03511v3.pdf
|
Unsupervised Translation of Programming Languages
|
A transcompiler, also known as source-to-source translator, is a system that converts source code from a high-level programming language (such as C++ or Python) to another. Transcompilers are primarily used for interoperability, and to port codebases written in an obsolete or deprecated language (e.g. COBOL, Python 2) to a modern one. They typically rely on handcrafted rewrite rules, applied to the source code abstract syntax tree. Unfortunately, the resulting translations often lack readability, fail to respect the target language conventions, and require manual modifications in order to work properly. The overall translation process is timeconsuming and requires expertise in both the source and target languages, making code-translation projects expensive. Although neural models significantly outperform their rule-based counterparts in the context of natural language translation, their applications to transcompilation have been limited due to the scarcity of parallel data in this domain. In this paper, we propose to leverage recent approaches in unsupervised machine translation to train a fully unsupervised neural transcompiler. We train our model on source code from open source GitHub projects, and show that it can translate functions between C++, Java, and Python with high accuracy. Our method relies exclusively on monolingual source code, requires no expertise in the source or target languages, and can easily be generalized to other programming languages. We also build and release a test set composed of 852 parallel functions, along with unit tests to check the correctness of translations. We show that our model outperforms rule-based commercial baselines by a significant margin.
|
['Marie-Anne Lachaux', 'Lowik Chanussot', 'Guillaume Lample', 'Baptiste Roziere']
|
2020-06-05
| null |
http://proceedings.neurips.cc/paper/2020/hash/ed23fbf18c2cd35f8c7f8de44f85c08d-Abstract.html
|
http://proceedings.neurips.cc/paper/2020/file/ed23fbf18c2cd35f8c7f8de44f85c08d-Paper.pdf
|
neurips-2020-12
|
['code-translation', 'unsupervised-machine-translation']
|
['computer-code', 'natural-language-processing']
|
[ 1.88628267e-02 -2.73268044e-01 -4.17380959e-01 -4.01641369e-01
-9.47969198e-01 -1.05196714e+00 3.32790315e-01 -2.76116114e-02
-1.78899795e-01 5.54028392e-01 2.87549458e-02 -9.94205534e-01
4.51556295e-01 -8.20962727e-01 -1.22856832e+00 4.90272120e-02
3.20877194e-01 2.74600953e-01 2.05690376e-02 -6.11252010e-01
3.64597775e-02 -2.45991107e-02 -1.25282669e+00 3.85665715e-01
1.27849972e+00 1.78311139e-01 5.83908670e-02 6.60788715e-01
-4.53174293e-01 5.17728150e-01 -5.48628032e-01 -7.51898408e-01
3.83034855e-01 -5.88775039e-01 -9.03953791e-01 -6.51569843e-01
3.28978866e-01 -1.39102593e-01 9.02542025e-02 1.13692999e+00
2.97906220e-01 -5.86778045e-01 2.96609372e-01 -1.32231736e+00
-9.47753906e-01 9.06183422e-01 -3.38919878e-01 -2.29892358e-01
3.21612120e-01 2.45363876e-01 9.16098237e-01 -8.29256713e-01
6.68533206e-01 7.50350595e-01 9.46046770e-01 6.05185390e-01
-1.58630741e+00 -5.70405245e-01 -3.62131953e-01 -3.57381970e-01
-1.30794311e+00 -5.51292658e-01 3.46225291e-01 -6.71452641e-01
1.55502260e+00 1.34183556e-01 3.75811070e-01 1.14248013e+00
5.21007836e-01 2.31372908e-01 7.49075413e-01 -6.31043077e-01
1.68170054e-02 2.82973379e-01 8.71476158e-02 8.66092563e-01
2.74613857e-01 -4.17778268e-02 -2.09623605e-01 -3.96402061e-01
3.34093899e-01 -1.19672671e-01 -2.74581552e-01 -3.64171267e-01
-1.49690902e+00 6.40948117e-01 3.97151321e-01 4.23291028e-01
7.83285350e-02 3.76811534e-01 7.13032663e-01 8.46980214e-01
-7.09501728e-02 7.57625997e-01 -8.39608490e-01 -4.64109659e-01
-8.18169415e-01 5.63709959e-02 1.30260825e+00 1.31108880e+00
1.06499910e+00 7.59173781e-02 3.07319909e-01 7.70112336e-01
1.61483228e-01 5.85542440e-01 7.04822540e-01 -7.67892480e-01
8.26543093e-01 7.36727655e-01 -7.83400461e-02 -7.63306975e-01
5.08575626e-02 -2.71208435e-01 -3.86591285e-01 1.68485239e-01
2.29619384e-01 -1.80183962e-01 -4.36669916e-01 1.64232183e+00
-8.25841054e-02 -6.62509501e-01 2.69407183e-01 7.30190516e-01
5.18570840e-01 6.10541463e-01 -2.10203037e-01 2.00014099e-01
8.73606741e-01 -1.18467522e+00 2.80735362e-02 -3.35819393e-01
9.79011297e-01 -1.09502017e+00 1.50728440e+00 1.28450468e-01
-1.14912260e+00 -3.72421324e-01 -1.08318436e+00 -2.53367573e-01
-5.06390154e-01 2.12121114e-01 8.15819919e-01 8.07939768e-01
-1.33895433e+00 7.04414666e-01 -8.95886481e-01 -6.79214537e-01
-8.48502095e-04 2.98403382e-01 -5.33973098e-01 2.17340082e-01
-8.10290873e-01 9.98804808e-01 4.90531176e-01 -2.53887504e-01
-7.38839924e-01 -9.61838484e-01 -8.76494646e-01 1.13358147e-01
-2.41726208e-02 -8.12385798e-01 1.64063811e+00 -1.54066825e+00
-1.69337153e+00 8.56975734e-01 9.86699164e-02 -1.92278966e-01
2.86916643e-01 -2.82902028e-02 -4.07271147e-01 -4.96430427e-01
2.97172278e-01 3.33839059e-01 7.02868104e-01 -7.71360993e-01
-4.32939470e-01 6.89017475e-02 3.14746767e-01 -2.64022171e-01
-3.85814041e-01 4.89634573e-01 -4.71350044e-01 -4.78737116e-01
-5.88455081e-01 -1.03736198e+00 8.72917175e-02 -6.59677759e-02
-1.77027702e-01 2.48243108e-01 3.75480801e-01 -7.98635602e-01
9.23053324e-01 -2.20950127e+00 3.72576237e-01 9.72506329e-02
1.76680982e-01 5.99606372e-02 -3.69937837e-01 6.13868833e-01
-2.45895892e-01 4.41154152e-01 -6.49827361e-01 1.17563367e-01
2.45132536e-01 1.79770008e-01 -4.29489285e-01 3.35741818e-01
1.72088385e-01 1.08618104e+00 -9.51110363e-01 -3.60756159e-01
-2.39015386e-01 2.52593666e-01 -9.46885169e-01 1.88867465e-01
-4.71053600e-01 1.72654897e-01 -1.27293319e-01 6.21560574e-01
3.56028527e-01 -2.10947514e-01 2.60141224e-01 1.26698181e-01
-3.95355880e-01 7.26654649e-01 -5.17460883e-01 2.14787602e+00
-1.26522839e+00 5.61943769e-01 -2.87044439e-02 -7.30035841e-01
8.13805223e-01 2.75542527e-01 2.80594043e-02 -5.01310885e-01
-1.65042698e-01 8.40785563e-01 1.88175306e-01 -2.95070350e-01
3.94505411e-01 1.35004684e-01 -4.05352324e-01 9.10472453e-01
1.77139670e-01 -4.94897515e-01 5.04219234e-01 1.41476542e-01
1.23135149e+00 5.51853061e-01 3.46726060e-01 -2.55137205e-01
2.91738272e-01 3.47924948e-01 4.67185885e-01 6.21605515e-01
2.76339382e-01 4.82535809e-01 6.10173643e-01 -3.41067910e-01
-1.31795287e+00 -1.10426784e+00 3.07925977e-02 1.21631312e+00
-4.19221878e-01 -8.05238307e-01 -8.49117219e-01 -7.89879441e-01
-2.25229695e-01 7.47760355e-01 -2.47768611e-01 -3.58167201e-01
-7.85588503e-01 -4.39704239e-01 1.15990591e+00 6.07314944e-01
3.58587533e-01 -8.09688509e-01 -5.27607143e-01 3.56555343e-01
-1.14139363e-01 -8.16953361e-01 -9.06521499e-01 2.44431376e-01
-6.88576519e-01 -1.02222192e+00 -4.31940913e-01 -9.69992280e-01
6.47041619e-01 -5.37693501e-02 1.55983114e+00 2.95493335e-01
-3.73192690e-03 1.60623074e-01 -2.25744814e-01 -9.73612219e-02
-1.18597269e+00 6.25318110e-01 -1.33210689e-01 -5.23584306e-01
2.73418069e-01 -8.37551594e-01 -1.00735195e-01 3.04163635e-01
-8.68937433e-01 2.21554294e-01 6.34625137e-01 9.09166753e-01
-4.34503378e-03 -4.30776685e-01 2.73497522e-01 -9.78293777e-01
6.36418879e-01 -4.98704076e-01 -1.00982404e+00 3.09736252e-01
-6.72526121e-01 1.59497678e-01 1.24890816e+00 -2.26936579e-01
-8.29412103e-01 -6.39069229e-02 -2.41190284e-01 -8.29015896e-02
4.12122086e-02 1.06258512e+00 -8.78063217e-03 -2.52313733e-01
1.28497446e+00 3.45677495e-01 -1.37907416e-01 -4.30493623e-01
4.04499918e-01 7.24749982e-01 7.70525634e-01 -1.05161572e+00
9.97738004e-01 -1.98886797e-01 -6.00397348e-01 -1.61366701e-01
-4.95006405e-02 9.04274508e-02 -4.50625598e-01 3.62570822e-01
3.68216604e-01 -8.34792256e-01 -1.95981994e-01 2.49615058e-01
-1.51574826e+00 -6.38544738e-01 -5.03589287e-02 1.17991537e-01
-4.97717232e-01 2.52196759e-01 -8.28367829e-01 2.12437257e-01
-5.81220627e-01 -1.46813107e+00 8.20516527e-01 -3.11687235e-02
-5.28031409e-01 -9.14702654e-01 3.77175778e-01 1.35828897e-01
9.76664245e-01 -1.15386963e-01 1.45377922e+00 -4.39901173e-01
-5.36439121e-01 -1.74093530e-01 -1.93375722e-01 4.43805128e-01
2.47245282e-01 4.97273386e-01 -4.06765431e-01 -2.65888780e-01
-3.01655173e-01 -4.68977898e-01 3.31368893e-01 -5.19558787e-01
7.33801842e-01 -3.93048614e-01 -9.13587958e-02 8.39147687e-01
1.40233839e+00 -1.66549519e-01 5.89895010e-01 4.95963693e-01
6.36434913e-01 2.82581985e-01 -1.13047443e-01 1.46783283e-02
5.29412866e-01 6.57387674e-01 1.25054464e-01 7.84555376e-02
-6.46410659e-02 -3.82963747e-01 9.26118970e-01 1.37465918e+00
1.72725394e-01 3.26124251e-01 -1.38491201e+00 7.18806386e-01
-1.63620389e+00 -4.37960625e-01 -2.64059991e-01 2.29947186e+00
1.51290929e+00 -9.56927612e-02 -1.21368572e-01 -3.97388756e-01
4.51481342e-01 -4.59266335e-01 -3.48634779e-01 -7.69387007e-01
2.84862399e-01 5.73151350e-01 3.51737112e-01 3.49360257e-01
-6.46243751e-01 1.11006486e+00 6.06240368e+00 3.61162692e-01
-1.59637082e+00 3.53267372e-01 6.62757130e-03 1.14118621e-01
-5.73631585e-01 5.42521298e-01 -4.84026372e-01 4.80188459e-01
1.39538109e+00 -6.17703855e-01 1.04284799e+00 1.15666020e+00
-1.30221188e-01 2.79992461e-01 -1.67286193e+00 7.46667325e-01
-5.74021675e-02 -1.35566521e+00 -2.15692550e-01 -3.34400982e-01
8.17706764e-01 5.97032070e-01 -2.43671715e-01 9.15294290e-01
5.91092288e-01 -1.07864666e+00 9.02442336e-01 4.80442755e-02
1.07389879e+00 -5.46450913e-01 5.90179384e-01 2.20140442e-01
-9.59264517e-01 3.85759979e-01 -3.53162080e-01 -1.28145978e-01
-2.80738920e-01 3.28907073e-01 -7.23588347e-01 5.18745363e-01
6.79511905e-01 6.60861194e-01 -7.82865763e-01 7.71985769e-01
-4.23247814e-01 5.91220617e-01 -2.66361117e-01 2.90602483e-02
-5.79368584e-02 -9.23206061e-02 1.51191130e-01 1.34823179e+00
5.31892121e-01 -6.26027346e-01 2.05288947e-01 1.48180473e+00
-4.71391886e-01 4.31640446e-01 -7.57935584e-01 -5.33175468e-01
2.13761300e-01 1.22328508e+00 -2.96568573e-01 -2.35498711e-01
-8.69721413e-01 1.08539021e+00 5.45856714e-01 4.40833986e-01
-1.06851280e+00 -9.38986659e-01 6.78771794e-01 -9.32366401e-02
1.48727372e-01 -2.69716203e-01 -2.51520365e-01 -1.65421033e+00
5.13955712e-01 -1.50256586e+00 -1.00303106e-01 -6.46825433e-01
-1.11968684e+00 8.23316932e-01 -1.92936480e-01 -1.04263794e+00
-4.64366615e-01 -4.12954897e-01 -6.98192596e-01 1.17448616e+00
-1.34661901e+00 -1.06795752e+00 -3.40496264e-02 4.54015791e-01
2.39379928e-01 -3.94776404e-01 1.10392165e+00 7.37136960e-01
-4.39236760e-01 8.50446463e-01 1.79769531e-01 3.56083721e-01
9.87211108e-01 -1.14133668e+00 9.98836935e-01 1.06168580e+00
-1.35351932e-02 1.41892231e+00 5.46348035e-01 -4.88947034e-01
-1.90966153e+00 -1.32587552e+00 1.09851134e+00 -5.53148508e-01
9.98001933e-01 -5.68415642e-01 -9.15562093e-01 9.35081899e-01
4.03529227e-01 1.67826470e-02 6.63327098e-01 1.04329996e-01
-9.37692463e-01 -9.08477753e-02 -7.71131814e-01 8.49157274e-01
9.28379297e-01 -8.93289089e-01 -5.47520697e-01 4.32150871e-01
7.85130918e-01 -5.59573054e-01 -1.17057276e+00 -4.55891006e-02
6.19427681e-01 -7.73028910e-01 6.29015207e-01 -7.20540285e-01
9.62501824e-01 -4.69351381e-01 -3.04221034e-01 -1.52914131e+00
-2.15133987e-02 -7.71247745e-01 2.17160150e-01 1.30524409e+00
9.84205782e-01 -9.33100820e-01 2.76801199e-01 6.15426183e-01
-2.86671221e-01 -3.05230469e-01 -5.01883268e-01 -9.37508881e-01
5.85575521e-01 -4.49625105e-01 8.05556893e-01 1.22689092e+00
4.11377281e-01 6.72143996e-01 -9.86095518e-02 -8.11850503e-02
3.36315930e-02 5.28338015e-01 1.34557879e+00 -8.30188572e-01
-9.73854184e-01 -6.10731542e-01 -2.20602173e-02 -7.54258931e-01
5.24780571e-01 -1.61282778e+00 2.58838564e-01 -1.15954280e+00
1.48810700e-01 -5.32071710e-01 1.91015527e-01 1.05874574e+00
1.30907819e-01 1.62063420e-01 1.67819589e-01 4.90196228e-01
-1.55657455e-01 4.74662304e-01 4.59180236e-01 -3.99434865e-01
-1.84240222e-01 -3.28973472e-01 -9.19620395e-01 5.70611477e-01
8.72584701e-01 -8.40669394e-01 -1.67709380e-01 -9.45834994e-01
6.79239810e-01 6.37458712e-02 2.56946176e-01 -9.07656610e-01
5.92087023e-02 -1.82449892e-01 -2.55362600e-01 3.47981721e-01
-4.01235551e-01 -7.47030079e-01 4.11192656e-01 3.10336798e-01
-3.07608783e-01 5.25776505e-01 2.95848995e-01 -6.45994470e-02
-1.89783528e-01 -5.33797383e-01 7.05193818e-01 -2.78981209e-01
-4.96242493e-01 1.29482308e-02 -3.71645182e-01 2.10631996e-01
6.93436682e-01 1.64388880e-01 -6.18789375e-01 3.04657798e-02
-4.90054674e-02 -1.29724726e-01 1.31030345e+00 7.14404941e-01
7.36184567e-02 -1.14351320e+00 -6.15349889e-01 3.13066483e-01
5.56616604e-01 -4.69418794e-01 -5.95743239e-01 9.97319102e-01
-9.76481497e-01 3.06561798e-01 -4.15646225e-01 -4.89102036e-01
-7.67052948e-01 5.20800948e-01 4.98723984e-01 -1.79840043e-01
-4.65648592e-01 2.33123451e-01 -1.29929125e-01 -1.34002745e+00
-3.50455821e-01 -5.87150633e-01 7.32233882e-01 -7.60770500e-01
3.56103539e-01 -1.57000273e-01 5.65197170e-01 -3.99479479e-01
-4.56648111e-01 4.44405884e-01 -2.73312721e-02 -7.69914016e-02
1.22441292e+00 5.04914224e-01 -8.17697406e-01 3.32755446e-01
1.48561811e+00 2.28750154e-01 -5.73365569e-01 -1.09948248e-01
3.79305705e-02 -4.19070810e-01 -2.78680235e-01 -8.92845690e-01
-9.51288223e-01 7.61209369e-01 8.99233818e-02 -1.88041717e-01
7.58488119e-01 -2.79609472e-01 7.93676555e-01 8.81369770e-01
8.19166124e-01 -7.10143030e-01 -3.63455743e-01 8.08918536e-01
8.20021927e-01 -1.11421216e+00 -3.90051872e-01 -2.13781834e-01
-1.78142264e-01 1.36722279e+00 6.78971827e-01 -4.98018553e-03
1.70195192e-01 6.18340671e-01 1.53924718e-01 3.02729577e-01
-8.10847402e-01 2.73423553e-01 -1.21601172e-01 5.60064077e-01
1.00781631e+00 -1.26668185e-01 -4.23869222e-01 5.58017313e-01
-4.54305023e-01 3.96557897e-01 8.57101560e-01 1.24306607e+00
1.81019783e-01 -1.68078911e+00 -3.96122247e-01 4.23155963e-01
-4.99652565e-01 -6.02439821e-01 -3.76367658e-01 6.79272652e-01
-5.79311438e-02 7.12318420e-01 -2.26332441e-01 -2.83174723e-01
2.74587929e-01 1.92575023e-01 6.07327998e-01 -9.85296428e-01
-1.20016861e+00 -4.53869075e-01 2.56696790e-01 -5.67335308e-01
6.95583448e-02 -3.86719525e-01 -1.13885176e+00 -6.77401483e-01
1.07206361e-04 2.51111418e-01 8.80076826e-01 7.93305278e-01
7.47783959e-01 3.02766174e-01 1.56577900e-01 -5.69326580e-01
-6.42212212e-01 -6.35011911e-01 1.88831314e-01 2.10036725e-01
9.52857658e-02 -2.59371735e-02 2.95422692e-02 5.44657290e-01]
|
[7.695737838745117, 7.87480354309082]
|
0bfae0d2-7fe5-4b88-b5a7-25ce8fefecbb
|
salient-region-segmentation
|
1803.05759
| null |
http://arxiv.org/abs/1803.05759v1
|
http://arxiv.org/pdf/1803.05759v1.pdf
|
Salient Region Segmentation
|
Saliency prediction is a well studied problem in computer vision. Early
saliency models were based on low-level hand-crafted feature derived from
insights gained in neuroscience and psychophysics. In the wake of deep learning
breakthrough, a new cohort of models were proposed based on neural network
architectures, allowing significantly higher gaze prediction than previous
shallow models, on all metrics.
However, most models treat the saliency prediction as a \textit{regression}
problem, and accurate regression of high-dimensional data is known to be a hard
problem. Furthermore, it is unclear that intermediate levels of saliency (ie,
neither very high, nor very low) are meaningful: Something is either salient,
or it is not.
Drawing from those two observations, we reformulate the saliency prediction
problem as a salient region \textit{segmentation} problem. We demonstrate that
the reformulation allows for faster convergence than the classical regression
problem, while performance is comparable to state-of-the-art.
We also visualise the general features learned by the model, which are showed
to be consistent with insights from psychophysics.
|
['Sen He', 'Nicolas Pugeault']
|
2018-03-15
| null | null | null | null |
['eye-tracking']
|
['computer-vision']
|
[ 4.29732174e-01 4.03965175e-01 -2.68682122e-01 -2.71499425e-01
-5.03835022e-01 -2.29264200e-01 6.47369504e-01 3.97094011e-01
-4.09814954e-01 7.20497429e-01 1.89116120e-01 -4.42900509e-02
-1.55043751e-01 -2.06691101e-01 -9.68625426e-01 -7.27485478e-01
-4.84287553e-02 2.64479697e-01 6.53951943e-01 -3.63865674e-01
7.40828216e-01 2.73852110e-01 -1.90241539e+00 1.14593074e-01
8.02471817e-01 1.13440621e+00 6.08257115e-01 4.69490767e-01
1.42690688e-01 6.85892105e-01 -2.51950592e-01 -1.09826270e-02
2.60126144e-01 -4.32781845e-01 -8.17106485e-01 -3.50446343e-01
6.25118673e-01 7.92547911e-02 -2.91411635e-02 1.13303506e+00
2.23812506e-01 4.78179008e-02 6.40181303e-01 -1.39086938e+00
-8.51513505e-01 2.79401928e-01 -6.91877127e-01 4.96700317e-01
1.50518730e-01 -1.11186981e-01 1.41333711e+00 -1.07741380e+00
5.06833136e-01 1.03006506e+00 6.80382788e-01 4.78569269e-01
-1.31598008e+00 -1.83082655e-01 3.97114545e-01 4.47546691e-01
-9.33190763e-01 -2.68123865e-01 7.94278800e-01 -4.81667668e-01
7.74267435e-01 3.98349136e-01 8.65323663e-01 8.56306672e-01
3.91653866e-01 9.87592578e-01 1.29357183e+00 -3.77795190e-01
3.62144291e-01 2.55240232e-01 1.18620075e-01 4.05405134e-01
3.05746824e-01 3.24709862e-01 -7.62832165e-01 2.76359379e-01
8.28200698e-01 -3.08242859e-04 -3.31867546e-01 -8.16643715e-01
-1.38488424e+00 9.80578363e-01 9.09010887e-01 2.59424984e-01
-4.05050278e-01 8.62892643e-02 1.41916752e-01 1.08186558e-01
4.38519150e-01 8.75786364e-01 -5.09623051e-01 -1.53136570e-02
-1.33145845e+00 4.36120629e-01 5.18038392e-01 7.40547121e-01
8.35264862e-01 1.89248368e-01 -8.72238651e-02 4.24821049e-01
2.86281854e-01 3.01053971e-01 6.06948733e-01 -8.36624503e-01
-1.08202212e-01 4.68146473e-01 2.62994647e-01 -1.01568413e+00
-8.24830234e-01 -6.11710608e-01 -7.60831654e-01 4.88193780e-01
4.92794424e-01 8.11152756e-02 -1.04335880e+00 1.83420312e+00
-2.49243658e-02 1.79008663e-01 -1.79856911e-01 1.39133048e+00
8.28633428e-01 2.39222348e-01 4.69134822e-02 -2.08462864e-01
1.16309643e+00 -1.03202415e+00 -4.49485421e-01 -6.01829410e-01
2.57632703e-01 -5.83095312e-01 8.75185609e-01 4.11461294e-01
-1.18623972e+00 -6.19923174e-01 -1.13148832e+00 -3.47400516e-01
-4.65238899e-01 -2.16094390e-01 8.19757521e-01 2.54412055e-01
-1.55051970e+00 7.66541064e-01 -4.78308916e-01 -5.25208354e-01
5.54111779e-01 4.74045008e-01 -1.28800929e-01 4.80902255e-01
-1.04409111e+00 1.43164575e+00 3.03364694e-01 6.30217195e-02
-8.42510819e-01 -7.52635896e-01 -8.57692540e-01 6.35496750e-02
3.68360221e-01 -6.35072887e-01 1.24711633e+00 -1.46538401e+00
-1.17219436e+00 1.06477594e+00 -4.11507994e-01 -8.15754056e-01
3.45193297e-01 -3.55699748e-01 -5.37957763e-03 3.45908962e-02
1.14636958e-01 1.01264095e+00 1.03616929e+00 -1.41599000e+00
-6.73788488e-01 -3.09512973e-01 7.97190070e-02 2.91905522e-01
1.33562848e-01 1.62518118e-02 1.12369880e-01 -4.91444141e-01
2.70319015e-01 -8.57680678e-01 -4.76978987e-01 2.86653899e-02
-3.43060106e-01 -1.03048205e-01 5.35146534e-01 -5.75215816e-01
8.65776896e-01 -1.97651970e+00 2.47060850e-01 2.32246313e-02
6.23021305e-01 2.33191237e-01 5.05618472e-03 1.15577579e-01
-4.92967159e-01 7.17721283e-02 -2.67628789e-01 -2.77158413e-02
6.31515160e-02 -1.77037403e-01 -5.57735443e-01 5.59190452e-01
5.03819227e-01 1.14175606e+00 -1.02928412e+00 -3.73970687e-01
2.06776142e-01 2.29280591e-01 -5.66930354e-01 2.70577427e-02
-2.55473018e-01 3.35741490e-01 -4.34779733e-01 4.79000598e-01
5.09855747e-01 -3.67770612e-01 -2.76417613e-01 -7.85442144e-02
-4.02413279e-01 2.27895334e-01 -7.87006021e-01 1.66717589e+00
9.03050154e-02 1.03781235e+00 -1.63679272e-01 -1.39411569e+00
9.63325441e-01 3.34284455e-02 5.44508040e-01 -8.33290279e-01
2.01625019e-01 1.35568276e-01 1.97072238e-01 -2.00442091e-01
7.61071384e-01 -2.34027773e-01 1.27774522e-01 2.10868910e-01
3.31074670e-02 -2.63237983e-01 -2.93897260e-02 -6.69689327e-02
8.86189103e-01 4.81007993e-01 4.46484953e-01 -7.32148051e-01
1.94241866e-01 3.31311971e-01 5.08468688e-01 9.04182971e-01
-4.62306917e-01 1.07963943e+00 7.41366744e-01 -4.69063818e-01
-1.08532214e+00 -1.01896203e+00 -1.45759955e-01 1.28934753e+00
6.61673307e-01 -2.46272027e-01 -8.90157282e-01 -3.22134584e-01
-1.57402664e-01 4.28602219e-01 -1.05839646e+00 -2.83241451e-01
-3.94374311e-01 -7.60299265e-01 -3.57034355e-02 5.33866823e-01
2.01280519e-01 -1.41158795e+00 -1.27925670e+00 9.70422253e-02
1.85982168e-01 -8.50801170e-01 -6.29435107e-02 3.97935838e-01
-9.08360660e-01 -9.22090828e-01 -9.20144737e-01 -8.39361846e-01
4.79952574e-01 4.90267128e-01 1.31299675e+00 2.59597957e-01
-3.88069451e-01 1.33184150e-01 -1.95728272e-01 -8.47250223e-01
9.62983221e-02 1.33877769e-02 4.86891232e-02 -1.32072866e-01
5.82504451e-01 -4.03960973e-01 -8.23171079e-01 1.38959706e-01
-7.78398335e-01 3.61798495e-01 7.89299190e-01 7.01033235e-01
6.10689878e-01 -4.70122457e-01 8.76383722e-01 -6.04269385e-01
4.51158375e-01 -4.37385142e-01 -3.79945517e-01 1.18042529e-01
-5.58257520e-01 1.29149839e-01 4.40652639e-01 -2.57046282e-01
-6.19433641e-01 1.01712428e-01 -4.78132814e-03 -4.33959335e-01
-3.24156016e-01 3.83324325e-01 3.67297828e-01 -1.08907498e-01
6.28139377e-01 2.10194826e-01 -1.36754438e-02 -2.31840491e-01
3.07530820e-01 1.34674609e-01 4.31978434e-01 -2.57244080e-01
6.58363283e-01 4.39804226e-01 7.42976144e-02 -9.70793426e-01
-1.26721442e+00 -3.50266337e-01 -8.57226610e-01 -1.79746151e-01
9.73449588e-01 -7.64631391e-01 -6.57019556e-01 1.22809991e-01
-1.00872636e+00 -3.71191710e-01 -5.75009644e-01 3.19627374e-01
-8.69788647e-01 9.97433215e-02 -9.14898515e-02 -7.10367382e-01
-1.02776974e-01 -1.07219696e+00 1.01625228e+00 5.80229878e-01
-3.16937059e-01 -9.85329032e-01 -5.95008096e-05 -6.26272410e-02
5.94451547e-01 2.66402513e-01 7.92287767e-01 -7.35481501e-01
-5.77777326e-01 6.40905499e-02 -6.00262702e-01 -7.25514768e-03
-2.00587213e-01 -5.57032116e-02 -1.12418222e+00 -1.59939930e-01
1.71433359e-01 -2.99462259e-01 1.20080423e+00 8.14484179e-01
1.12198818e+00 8.32513347e-02 -2.48148486e-01 4.65851396e-01
1.36977077e+00 -2.32402474e-01 5.42310417e-01 5.25206745e-01
4.85312849e-01 8.06969821e-01 7.30021298e-01 1.64600402e-01
4.28358614e-01 6.03797436e-01 9.33076560e-01 -4.28535432e-01
-8.93031829e-04 -1.11090802e-01 1.10685550e-01 4.31038916e-01
-2.74357229e-01 3.04914117e-01 -9.50067639e-01 8.04826260e-01
-2.19336271e+00 -1.00520277e+00 -2.58762002e-01 2.26404476e+00
7.43158102e-01 4.59176511e-01 3.63810062e-01 3.04115321e-02
6.82016432e-01 1.29852712e-01 -7.76393950e-01 -5.83516002e-01
-1.75561547e-01 1.74759775e-01 3.34686637e-01 3.31622243e-01
-1.14747036e+00 1.15977347e+00 7.13424683e+00 5.34471273e-01
-1.43399322e+00 -1.85799710e-02 6.61212862e-01 -9.33514386e-02
-3.62302929e-01 6.19457625e-02 -7.47672856e-01 3.11521977e-01
7.07868218e-01 -2.43365407e-01 2.62507558e-01 9.52193856e-01
1.35316998e-01 -4.85804290e-01 -1.25687587e+00 8.61220837e-01
3.49637032e-01 -1.36245501e+00 -2.28390723e-01 -9.07767490e-02
6.88709736e-01 3.16346794e-01 4.45999801e-01 4.39705551e-01
8.87128562e-02 -1.35030568e+00 9.43020105e-01 5.97967505e-01
2.90618837e-01 -3.59936476e-01 4.65536267e-01 3.96112472e-01
-8.50279391e-01 -5.77354208e-02 -6.79815114e-01 -3.30939382e-01
-6.84085637e-02 4.72150683e-01 -6.02383137e-01 1.77954242e-01
8.58217657e-01 1.00459194e+00 -1.01440322e+00 1.36949515e+00
-6.60754144e-02 3.21570486e-01 -9.65922922e-02 -1.97583750e-01
4.53964859e-01 -8.75866264e-02 5.39543271e-01 1.19753182e+00
5.59758767e-02 -1.60204709e-01 -2.55385228e-02 1.07529294e+00
2.72447407e-01 -2.03312794e-03 -5.88055909e-01 3.18685025e-01
-1.05804794e-01 1.34676373e+00 -8.87826920e-01 -5.24678901e-02
-3.46045732e-01 6.95701778e-01 4.82079178e-01 4.75546539e-01
-8.49422395e-01 -9.33694392e-02 5.30229151e-01 1.47728726e-01
5.15458047e-01 6.31075120e-03 -7.46713698e-01 -1.06452429e+00
-1.20397225e-01 -4.87045497e-01 -1.19020507e-01 -1.03362966e+00
-1.04702568e+00 4.56969529e-01 -1.96962312e-01 -1.10219610e+00
-1.36767268e-01 -6.87074423e-01 -6.43983424e-01 8.46123755e-01
-1.97584319e+00 -1.03391230e+00 -2.55718380e-01 3.65258783e-01
5.38912833e-01 1.07127719e-01 6.25635147e-01 -2.96373636e-01
-2.92127490e-01 1.20545797e-01 -3.74868065e-02 -2.86417693e-01
6.02626324e-01 -1.57942104e+00 3.32776010e-01 7.98032582e-01
2.28905186e-01 4.33917671e-01 1.14539158e+00 -3.19981873e-01
-1.10227311e+00 -7.42750823e-01 7.98264384e-01 -6.94720209e-01
6.95057750e-01 -3.22189420e-01 -1.05664909e+00 5.08402586e-01
3.81829709e-01 2.18520835e-02 2.89885819e-01 1.63412794e-01
-6.42534420e-02 1.86294332e-01 -9.08608913e-01 7.22485006e-01
8.57979417e-01 -2.66542733e-01 -1.04111099e+00 9.87741351e-02
6.45751774e-01 -2.78540671e-01 -3.58377069e-01 4.89192903e-01
5.38245857e-01 -1.37686121e+00 1.08804619e+00 -6.98917449e-01
7.25141704e-01 -1.25957072e-01 1.12689044e-02 -1.36480832e+00
-5.03157198e-01 -4.73325908e-01 -2.02591494e-01 5.82060814e-01
5.90660691e-01 -2.65175909e-01 7.28421450e-01 4.50315118e-01
-3.14607054e-01 -9.35276747e-01 -9.70192194e-01 -5.00421405e-01
1.89549580e-01 -2.14329317e-01 1.38893247e-01 8.15784693e-01
1.45456105e-01 5.23847818e-01 -2.57813215e-01 -2.56812535e-02
6.11988068e-01 4.04890478e-01 4.65175062e-01 -1.67480826e+00
9.56114158e-02 -8.09277833e-01 -5.54790735e-01 -1.05367064e+00
1.55387089e-01 -7.23329127e-01 3.92909467e-01 -1.72473419e+00
1.21131346e-01 -2.32308224e-01 -6.51472151e-01 4.12841797e-01
-3.22493285e-01 4.53633308e-01 4.15442079e-01 1.14393614e-01
-7.66849518e-01 4.33741182e-01 1.32770002e+00 1.14727043e-01
-9.79865417e-02 2.35217297e-03 -1.07587123e+00 1.01080525e+00
8.50730658e-01 -1.61143169e-01 -2.23960742e-01 -1.46284729e-01
4.97025937e-01 -2.07199171e-01 6.68831885e-01 -1.11376500e+00
3.05768073e-01 -3.59461486e-01 4.68636274e-01 -6.10830605e-01
5.62803149e-01 -6.74545407e-01 -5.75697482e-01 2.71879971e-01
-4.55329746e-01 -2.19566211e-01 2.12348521e-01 4.38206732e-01
-1.32610619e-01 -1.82233751e-01 9.19147670e-01 -1.69777170e-01
-1.11513603e+00 1.43384084e-01 -1.90047324e-01 2.43363440e-01
9.96127248e-01 -5.36694944e-01 -2.70459652e-01 -3.76095444e-01
-7.63246417e-01 5.35326190e-02 4.94392425e-01 6.04115069e-01
7.56887853e-01 -1.04723930e+00 -6.24467850e-01 1.26660287e-01
3.42890099e-02 -1.05957035e-02 -5.06989323e-02 1.18874490e+00
-2.19599441e-01 4.77253735e-01 -4.45642442e-01 -8.77531350e-01
-8.53594363e-01 7.55968392e-01 2.10911259e-01 1.37149274e-01
-5.77371478e-01 9.31031942e-01 6.27700210e-01 -3.50144356e-02
2.14837030e-01 -5.09763598e-01 -4.73001391e-01 1.32760853e-01
4.37904805e-01 -5.93086518e-03 -1.16714418e-01 -6.64885759e-01
-3.71076286e-01 7.24225819e-01 -1.13352783e-01 1.35658473e-01
1.54450631e+00 -6.57952130e-02 4.03018631e-02 6.73052728e-01
6.57052994e-01 -4.12009239e-01 -1.58702886e+00 -2.03436255e-01
3.20687264e-01 -1.81749821e-01 1.38396114e-01 -7.46900856e-01
-7.91686594e-01 1.17390740e+00 4.59453821e-01 4.52892601e-01
9.83658671e-01 5.05776368e-02 4.62557584e-01 2.52121747e-01
1.39139086e-01 -1.28805435e+00 1.57862604e-01 5.72117388e-01
1.12124598e+00 -1.66577649e+00 1.30854383e-01 -1.30256608e-01
-9.49720502e-01 9.24389601e-01 6.88522398e-01 -6.66197598e-01
7.24453032e-01 -5.78182042e-02 -1.10903449e-01 -2.01025948e-01
-6.56948388e-01 -6.10906839e-01 6.24521613e-01 5.76246023e-01
4.35889393e-01 -1.15451224e-01 -5.17288670e-02 5.66159427e-01
-3.87906462e-01 1.68529619e-02 6.08265758e-01 7.87815273e-01
-9.69894230e-01 -4.61476952e-01 -2.69896775e-01 5.98828912e-01
-3.75813961e-01 -2.86175311e-01 -3.88462842e-01 7.91664779e-01
4.19318154e-02 7.46887624e-01 1.46062657e-01 -1.19448863e-01
9.55278575e-02 -1.55656949e-01 4.79060233e-01 -6.83726728e-01
-6.06436312e-01 -9.61726345e-03 -4.02075648e-01 -6.14455819e-01
-6.00281954e-01 -5.91392815e-01 -1.17831755e+00 2.30761692e-02
-3.50517303e-01 8.16829056e-02 6.74905896e-01 1.24259138e+00
1.73202872e-01 5.21341920e-01 3.98288906e-01 -1.28665483e+00
-2.88425088e-01 -7.75895774e-01 -6.43811762e-01 2.10015327e-01
8.45843256e-01 -8.40731084e-01 -3.65739912e-01 1.04147196e-01]
|
[10.046239852905273, 1.5379610061645508]
|
5bb59449-98cc-4ca3-930c-1234674121e2
|
a-data-variation-robust-learning-model-based
|
2302.04438
| null |
https://arxiv.org/abs/2302.04438v2
|
https://arxiv.org/pdf/2302.04438v2.pdf
|
An information-theoretic learning model based on importance sampling
|
A crucial assumption underlying the most current theory of machine learning is that the training distribution is identical to the test distribution. However, this assumption may not hold in some real-world applications. In this paper, we develop a learning model based on principles of information theory by minimizing the worst-case loss at prescribed levels of uncertainty. We reformulate the empirical estimation of the risk functional and the distribution deviation constraint based on the importance sampling method. The objective of the proposed approach is to minimize the loss under maximum degradation and hence the resulting problem is a minimax problem which can be converted to an unconstrained minimum problem using the Lagrange method with the Lagrange multiplier $T$. We reveal that the minimization of the objective function under logarithmic transformation is equivalent to the minimization of the p-norm loss with $p=\frac{1}{T}$. We applied the proposed model to the face verification task on Racial Faces in the Wild datasets and showed that the proposed model performs better under large distribution deviations.
|
['Mengyao Li', 'Fei Gao', 'Lizhen Ji', 'Jiangshe Zhang']
|
2023-02-09
| null | null | null | null |
['face-verification']
|
['computer-vision']
|
[ 5.20921290e-01 4.59163934e-01 3.60051319e-02 -6.40290678e-01
-7.82362282e-01 -2.32684791e-01 1.68141276e-01 2.35401645e-01
-5.93407214e-01 8.68494749e-01 -5.17136931e-01 -3.00526202e-01
-5.98199308e-01 -6.83999002e-01 -8.43877792e-01 -9.93987858e-01
1.05836116e-01 1.12343691e-01 -2.44309798e-01 3.18135560e-01
4.52944458e-01 4.82636869e-01 -1.60556722e+00 -2.37197503e-01
9.25030470e-01 1.53454554e+00 -1.18493140e-01 2.47783363e-01
2.08399057e-01 3.88674080e-01 -6.04007959e-01 -6.73803329e-01
5.56459486e-01 -4.83614594e-01 -5.05983055e-01 1.29475400e-01
4.29524183e-01 -1.60622120e-01 2.57376522e-01 1.42701912e+00
6.19386554e-01 2.78268456e-01 1.00760686e+00 -1.37794995e+00
-2.13520899e-01 1.15518324e-01 -7.07794368e-01 9.67506766e-02
-6.12873258e-03 -3.94224405e-01 1.03621089e+00 -7.22873092e-01
2.61571676e-01 1.15103412e+00 4.91087407e-01 4.08572584e-01
-1.30255365e+00 -5.23634374e-01 -1.25024796e-01 4.12912145e-02
-1.61762381e+00 -4.29567844e-01 5.68267465e-01 -4.53640163e-01
2.55929500e-01 6.20163940e-02 1.36390999e-01 4.99647915e-01
2.57815927e-01 5.34408569e-01 1.06905377e+00 -7.21215069e-01
3.93126756e-01 6.42282903e-01 -8.19209069e-02 5.85492373e-01
5.86719871e-01 2.40114078e-01 -3.15776795e-01 -2.66658992e-01
2.68397003e-01 -2.86051154e-01 -2.43125305e-01 -4.35135484e-01
-3.47280830e-01 9.89863753e-01 3.32376473e-02 7.81301856e-02
-2.37230659e-01 3.47233452e-02 1.61983147e-01 2.26953149e-01
7.22512066e-01 6.43184036e-02 -2.85215765e-01 2.64577448e-01
-8.87206256e-01 2.22381085e-01 8.37309837e-01 7.08979785e-01
5.31630635e-01 2.45740525e-02 -5.88326603e-02 7.52855480e-01
7.04656899e-01 4.98001248e-01 -1.16046757e-01 -1.04877913e+00
3.86035889e-01 2.84056485e-01 2.76561320e-01 -1.07660317e+00
1.38925344e-01 -6.13388538e-01 -4.76921678e-01 4.86131549e-01
8.10818493e-01 -4.82156128e-01 -2.61904299e-01 2.11047459e+00
4.20476317e-01 -6.34498000e-02 -4.60633039e-02 4.99561727e-01
-2.90010851e-02 5.44716954e-01 -7.24310279e-02 -8.29255223e-01
8.61359179e-01 -4.75612096e-02 -9.80086207e-01 5.21630310e-02
2.14739174e-01 -6.86361909e-01 8.29039991e-01 6.16416335e-01
-1.01842785e+00 -2.83932865e-01 -1.32426620e+00 4.98936594e-01
1.04903966e-01 4.76900451e-02 8.74818787e-02 1.15134084e+00
-8.24198961e-01 6.87294304e-01 -3.95012498e-01 -2.66796857e-01
4.31824863e-01 5.05668461e-01 -2.39548355e-01 -2.51773791e-03
-8.19807887e-01 7.79521525e-01 4.48318899e-01 3.67432922e-01
-6.69347048e-01 -7.10862815e-01 -6.84250057e-01 7.47011602e-02
4.60079491e-01 -2.76510239e-01 9.55865681e-01 -1.17736661e+00
-1.48756337e+00 8.62672389e-01 -1.45869954e-02 -3.78470689e-01
8.04951429e-01 -2.05203950e-01 6.93339780e-02 -1.07565276e-01
-1.89293832e-01 3.72220203e-02 1.09399486e+00 -1.25708771e+00
-4.33878303e-01 -5.85145950e-01 -4.13309261e-02 1.25807831e-02
-2.69739538e-01 -1.62839487e-01 -4.58292142e-02 -4.82421786e-01
-1.32099494e-01 -7.01542020e-01 7.79507533e-02 2.39265814e-01
-1.83239833e-01 -9.67802107e-02 6.67378366e-01 -7.14609861e-01
1.15527844e+00 -2.30230522e+00 -1.39052823e-01 5.38970351e-01
-3.22538465e-01 6.15686551e-02 2.68983930e-01 3.32361966e-01
-1.28664032e-01 -1.22695938e-02 -7.00973213e-01 -2.41915122e-01
1.02076270e-01 -2.46680472e-02 -1.54324666e-01 8.73647094e-01
-6.59609586e-02 1.03862308e-01 -5.15638471e-01 -4.94469523e-01
-9.01124477e-02 5.01182377e-01 -5.83195210e-01 1.83618397e-01
-2.09985331e-01 2.10414827e-01 -2.59452701e-01 3.04167271e-01
9.19865072e-01 1.17754020e-01 2.43947014e-01 -4.27826084e-02
9.03185010e-02 -3.89942825e-01 -1.50139964e+00 1.39540362e+00
-3.31109881e-01 4.79479760e-01 3.64718586e-01 -1.35445535e+00
9.16304171e-01 2.01500028e-01 5.92652619e-01 -2.84864038e-01
3.71365011e-01 1.37690604e-01 -9.20168962e-03 -2.69197971e-01
-1.83639407e-01 -5.75738192e-01 4.26335894e-02 3.22665453e-01
-1.36547182e-02 -4.50808853e-02 -9.07302126e-02 -1.74924299e-01
5.80551922e-01 -6.32112101e-02 5.60283005e-01 -6.81496143e-01
6.29374981e-01 -6.15540206e-01 7.76028752e-01 6.39628351e-01
-2.27101699e-01 2.67863303e-01 8.81799281e-01 5.24457842e-02
-8.94676149e-01 -9.65004802e-01 -6.47859335e-01 5.06304085e-01
-2.54953235e-01 1.70555875e-01 -1.05159831e+00 -7.42407918e-01
2.31299371e-01 8.95154059e-01 -4.45199609e-01 -9.07916799e-02
-1.44384474e-01 -1.02057266e+00 3.10184598e-01 -1.65270492e-01
5.37570000e-01 -6.12288058e-01 -6.12349629e-01 -3.02766532e-01
7.45990053e-02 -7.19314396e-01 -4.25844163e-01 8.31376179e-04
-6.18991673e-01 -1.01331651e+00 -5.91323256e-01 -4.73251939e-01
8.89058769e-01 -4.88043040e-01 6.92138314e-01 -1.67613566e-01
-3.74035329e-01 4.09166247e-01 -3.67743038e-02 -6.63040400e-01
-2.91411608e-01 -4.02885139e-01 1.72221839e-01 5.43089032e-01
2.06710160e-01 -3.85263652e-01 -5.07424295e-01 2.14863345e-01
-9.60414469e-01 -6.15692377e-01 3.79154474e-01 8.81726801e-01
6.97491586e-01 7.50895560e-01 7.63928115e-01 -8.92295897e-01
5.90468109e-01 -3.53140980e-01 -1.10198116e+00 3.75502735e-01
-8.41087997e-01 3.84832025e-01 4.05669481e-01 -1.66348368e-01
-1.28391075e+00 4.80344221e-02 1.90912560e-02 -9.76118296e-02
1.00706138e-01 3.51640344e-01 -8.65605116e-01 -2.94193745e-01
2.59991378e-01 8.21201801e-02 7.70791844e-02 -4.51102823e-01
1.31226316e-01 7.04450250e-01 1.62399635e-01 -8.48376334e-01
6.02444291e-01 3.47575873e-01 5.46204686e-01 -9.12000656e-01
-9.01609242e-01 -7.46983057e-03 -4.55510587e-01 -3.32317501e-01
5.72986186e-01 -3.29831272e-01 -1.19352782e+00 3.09839725e-01
-8.98853242e-01 3.60446982e-02 -5.23831189e-01 4.55544949e-01
-8.49956393e-01 6.29916012e-01 1.06510241e-02 -1.54235458e+00
-2.97882736e-01 -8.98478150e-01 6.41138792e-01 1.80566266e-01
1.54287785e-01 -9.46891546e-01 1.54949958e-02 2.10381940e-01
1.99461490e-01 6.41086102e-01 1.16340935e+00 -5.22730887e-01
-2.43992761e-01 -4.08182919e-01 -7.67573789e-02 1.00588381e+00
9.87232476e-02 -1.20109670e-01 -9.75798786e-01 -4.48618531e-01
6.06441259e-01 -1.60118714e-01 7.20417976e-01 5.87320626e-01
1.22741127e+00 -5.72322726e-01 9.23999995e-02 5.61004102e-01
1.86705279e+00 4.21343327e-01 4.35416371e-01 -2.32955739e-01
-3.41497138e-02 1.00176644e+00 7.80523479e-01 7.51042902e-01
-1.94769144e-01 7.88181245e-01 4.57153201e-01 4.85820204e-01
4.15077955e-01 -2.24323124e-01 3.20599705e-01 3.16124320e-01
3.51083517e-01 -3.25728774e-01 -7.49457121e-01 3.67015988e-01
-1.79051924e+00 -9.79331791e-01 2.98584431e-01 3.06363678e+00
7.61227131e-01 1.03452317e-01 -1.09201844e-03 2.58975178e-01
8.14975679e-01 -2.16919780e-01 -6.21610284e-01 -5.12015998e-01
4.95020971e-02 1.77701056e-01 4.31294769e-01 7.98761547e-01
-1.00891268e+00 2.98544437e-01 6.28496170e+00 1.01314473e+00
-8.54911268e-01 4.44305167e-02 8.71018648e-01 -5.45783564e-02
-1.49419475e-02 3.27416696e-03 -8.91235352e-01 7.78494060e-01
8.51428747e-01 -5.42493224e-01 2.69193679e-01 7.92906761e-01
2.08062693e-01 -3.55543613e-01 -1.08110774e+00 9.65528905e-01
1.73875749e-01 -5.75006664e-01 -2.49070570e-01 3.31847548e-01
6.07783437e-01 -6.66011989e-01 4.85840023e-01 -8.53028372e-02
-1.13130249e-01 -1.05665529e+00 6.45149887e-01 5.35678148e-01
1.00731170e+00 -1.01311457e+00 7.71728814e-01 5.24962783e-01
-8.69381189e-01 -2.27208763e-01 -3.93356055e-01 3.39982621e-02
7.18278140e-02 9.11173880e-01 -6.90080166e-01 4.41784233e-01
2.90488064e-01 2.99115509e-01 3.70711391e-03 1.19127893e+00
-6.65567815e-02 2.68129379e-01 -5.98553479e-01 1.02495119e-01
-3.04492325e-01 -4.59575534e-01 6.84629977e-01 8.78405929e-01
5.74129045e-01 -1.22762792e-01 -1.25469729e-01 7.69757688e-01
-1.49762824e-01 3.01474631e-01 -5.70223927e-01 1.97701886e-01
5.40545702e-01 8.55252564e-01 -5.18221021e-01 1.21660084e-01
-2.67709821e-01 7.47126579e-01 2.00424761e-01 3.00076127e-01
-5.78534424e-01 -4.75675821e-01 5.56128621e-01 1.63945645e-01
2.72002786e-01 2.10140377e-01 -3.12125891e-01 -7.56900132e-01
3.59895825e-01 -6.79315567e-01 5.68293512e-01 5.44562750e-02
-1.41092765e+00 3.67631257e-01 3.65507454e-01 -1.01213288e+00
-2.19084591e-01 -7.52120197e-01 -5.92567742e-01 1.10908902e+00
-1.33147311e+00 -5.01594663e-01 1.50301263e-01 6.05996370e-01
1.70005292e-01 -1.20133460e-01 5.82379758e-01 4.26379323e-01
-7.35458970e-01 9.35498297e-01 4.61675137e-01 -2.30836347e-01
4.85799253e-01 -1.12193882e+00 -5.12871742e-01 9.98830497e-01
-3.47056925e-01 4.23082650e-01 1.06254923e+00 -5.09312510e-01
-9.47524488e-01 -8.92171562e-01 8.67616117e-01 -3.60903256e-02
3.71604681e-01 -3.06635529e-01 -6.09236598e-01 3.11750799e-01
-1.07158519e-01 1.07622467e-01 6.81552231e-01 -2.89727151e-01
-2.48734236e-01 -5.29788554e-01 -1.93362641e+00 -3.46238464e-02
6.02653027e-01 -4.28230554e-01 -2.13208824e-01 2.51048237e-01
2.34797597e-01 5.94280548e-02 -9.43179727e-01 4.69758958e-01
7.47847140e-01 -7.36431539e-01 7.55435884e-01 -5.73684514e-01
1.16789363e-01 -1.32562280e-01 -6.19596064e-01 -1.10645878e+00
2.54856974e-01 -6.40613198e-01 -1.35711185e-03 1.36399913e+00
4.21461463e-01 -6.28805161e-01 7.28473186e-01 8.37782681e-01
4.59547222e-01 -8.57113361e-01 -1.52699947e+00 -8.39389920e-01
1.83553368e-01 -3.72178674e-01 7.86711648e-02 4.61285591e-01
-2.03344319e-03 -2.13478848e-01 -5.08086622e-01 1.60394683e-01
1.33960032e+00 -2.55346119e-01 2.27637559e-01 -1.22826648e+00
-6.65430784e-01 -1.68222025e-01 -3.90677094e-01 -6.10321820e-01
3.90316427e-01 -6.45740330e-01 3.31252307e-01 -8.78627062e-01
2.13499203e-01 -3.96912515e-01 -4.17925507e-01 -6.84698448e-02
-8.06633309e-02 -2.36284018e-01 2.00206488e-01 -1.92804500e-01
-1.14161614e-02 6.21042252e-01 7.46489704e-01 -1.86689347e-02
4.92091337e-03 5.47293842e-01 -7.04259276e-01 7.28866875e-01
6.38405740e-01 -5.48505962e-01 -7.26162016e-01 -1.34791970e-01
3.69354218e-01 1.43358752e-01 7.41250440e-02 -7.37879634e-01
8.63795262e-03 -1.85550183e-01 2.37424970e-02 -3.21151197e-01
3.74771863e-01 -1.16813958e+00 1.78950429e-01 3.52657497e-01
-3.82996708e-01 -2.26762787e-01 -5.04080467e-02 6.66610658e-01
-6.57137856e-02 -5.97965956e-01 1.26506567e+00 7.63009042e-02
-6.95084706e-02 1.88540637e-01 8.03712159e-02 1.95117993e-03
1.27325904e+00 2.51149782e-03 -8.04870874e-02 -4.11371917e-01
-5.01713276e-01 -1.58097461e-01 2.47927889e-01 -5.85997254e-02
5.73787630e-01 -1.11721253e+00 -7.41546333e-01 1.95217282e-01
-1.94172427e-01 -3.07029307e-01 1.59150511e-01 7.62409031e-01
-3.53805542e-01 3.86322767e-01 -1.43618258e-02 -3.57446998e-01
-1.29543674e+00 4.06417549e-01 5.87889731e-01 -2.04973206e-01
3.69675388e-03 8.05538356e-01 1.83051363e-01 -6.67290092e-02
5.32599807e-01 2.02491835e-01 -3.85450125e-02 7.58880824e-02
5.25308490e-01 4.97613847e-01 1.14182718e-02 -5.59989572e-01
-3.35964918e-01 5.44495404e-01 9.69495997e-02 -4.35464531e-01
1.22924650e+00 -2.16345042e-01 -2.37966776e-01 2.97650605e-01
1.44211626e+00 -6.27191812e-02 -1.30069661e+00 -2.33565852e-01
1.39354438e-01 -7.94582129e-01 1.49450004e-01 -6.91698909e-01
-1.08673155e+00 8.36644292e-01 1.01267111e+00 -9.37765930e-03
1.25874972e+00 -3.31772506e-01 2.90511191e-01 2.21508592e-01
4.94454324e-01 -1.21234763e+00 -2.35599890e-01 1.14623345e-01
7.63359129e-01 -1.13554800e+00 4.26684245e-02 -4.31350470e-01
-3.10586333e-01 8.63799393e-01 3.33894402e-01 -3.76902111e-02
1.15483665e+00 2.19220251e-01 -2.66449213e-01 2.26998582e-01
-5.34703374e-01 1.20891668e-01 3.95287693e-01 4.75809336e-01
4.04136419e-01 1.04351506e-01 -7.66217053e-01 6.13626778e-01
-5.48115447e-02 2.59885434e-02 1.57626092e-01 7.27820218e-01
-3.82764220e-01 -1.10006368e+00 -4.29756910e-01 4.20513153e-01
-8.46841276e-01 2.21003205e-01 -1.02597281e-01 5.51707447e-01
4.64342445e-01 1.06659448e+00 -8.51032361e-02 -6.13804953e-03
2.01619893e-01 3.30620289e-01 6.06341422e-01 -3.71321797e-01
5.75578678e-03 -3.68449651e-02 -2.21774980e-01 -2.62763768e-01
-3.44561011e-01 -8.84238362e-01 -6.97605789e-01 -2.97356129e-01
-3.49464804e-01 3.54114503e-01 8.10847223e-01 9.05609310e-01
-5.51839620e-02 8.33487660e-02 7.94601619e-01 -3.08869272e-01
-1.06414449e+00 -7.32259333e-01 -9.55401897e-01 2.93186039e-01
3.32092404e-01 -7.11355507e-01 -6.91052258e-01 -9.83502045e-02]
|
[7.033913612365723, 4.014965534210205]
|
ee1d0c06-7838-4fa2-866c-753073ba1a28
|
seec-semantic-vector-federation-across-edge
|
2008.13298
| null |
https://arxiv.org/abs/2008.13298v1
|
https://arxiv.org/pdf/2008.13298v1.pdf
|
SEEC: Semantic Vector Federation across Edge Computing Environments
|
Semantic vector embedding techniques have proven useful in learning semantic representations of data across multiple domains. A key application enabled by such techniques is the ability to measure semantic similarity between given data samples and find data most similar to a given sample. State-of-the-art embedding approaches assume all data is available on a single site. However, in many business settings, data is distributed across multiple edge locations and cannot be aggregated due to a variety of constraints. Hence, the applicability of state-of-the-art embedding approaches is limited to freely shared datasets, leaving out applications with sensitive or mission-critical data. This paper addresses this gap by proposing novel unsupervised algorithms called \emph{SEEC} for learning and applying semantic vector embedding in a variety of distributed settings. Specifically, for scenarios where multiple edge locations can engage in joint learning, we adapt the recently proposed federated learning techniques for semantic vector embedding. Where joint learning is not possible, we propose novel semantic vector translation algorithms to enable semantic query across multiple edge locations, each with its own semantic vector-space. Experimental results on natural language as well as graph datasets show that this may be a promising new direction.
|
['Graham Bent', 'Shalisha Witherspoon', 'Nirmit Desai', 'Dean Steuer']
|
2020-08-30
| null | null | null | null |
['learning-semantic-representations']
|
['methodology']
|
[ 2.15272754e-01 9.16478038e-02 -7.83027947e-01 -2.38068432e-01
-7.41357505e-01 -5.75480580e-01 7.90843427e-01 7.97753632e-01
-2.82968372e-01 4.30531502e-01 3.45509559e-01 -3.32817912e-01
-4.74585325e-01 -1.00697184e+00 -6.12804651e-01 -4.11632240e-01
-2.29965672e-01 6.22854352e-01 1.45056486e-01 -3.41245919e-01
2.16958821e-01 3.27052176e-01 -1.26996922e+00 2.32475430e-01
3.32113087e-01 9.71422970e-01 1.79005891e-01 3.66500378e-01
-8.02173078e-01 4.27850962e-01 -3.05134892e-01 -3.94173950e-01
4.05152857e-01 -9.27405730e-02 -9.42087650e-01 -7.44403750e-02
1.80728555e-01 1.22257005e-02 -3.03481340e-01 1.09486997e+00
2.03218609e-01 1.67769566e-01 5.17252028e-01 -1.95620441e+00
-1.02816474e+00 5.21067023e-01 -4.43104267e-01 4.00371403e-02
4.40271437e-01 -3.71523619e-01 1.44500363e+00 -6.21847034e-01
9.57972586e-01 9.72607017e-01 4.15029019e-01 2.85727620e-01
-1.22707248e+00 -3.72909665e-01 2.65425384e-01 4.94065404e-01
-1.13530111e+00 -4.80357185e-02 1.10708249e+00 -3.04952234e-01
9.89140987e-01 2.16543406e-01 4.07764018e-01 1.04428291e+00
-7.95360431e-02 7.86220968e-01 7.83109188e-01 -5.11095643e-01
4.90837753e-01 4.57743913e-01 9.52031240e-02 5.92014134e-01
4.04368222e-01 -2.52211004e-01 -6.03300929e-01 -6.07261181e-01
3.17862302e-01 7.03101397e-01 -1.45519406e-01 -1.08618402e+00
-1.33595681e+00 1.25618517e+00 7.26067424e-01 5.78018725e-01
-5.10870993e-01 2.21294984e-01 7.32051194e-01 6.10057354e-01
6.43402398e-01 5.52609921e-01 -4.67517644e-01 -3.23508829e-02
-7.90785193e-01 1.22262627e-01 9.06743705e-01 1.10300958e+00
1.11447585e+00 -2.19719321e-01 2.32484952e-01 6.51383698e-01
2.05720559e-01 2.78923243e-01 4.51453626e-01 -7.16116250e-01
6.49774551e-01 8.96547735e-01 -6.10129349e-02 -1.37927842e+00
-4.37027998e-02 3.69521081e-02 -5.43237209e-01 -7.12504461e-02
-2.62419824e-02 1.95149928e-01 -4.12702978e-01 1.66168392e+00
4.68401402e-01 4.40997779e-01 2.18124166e-01 8.08899701e-01
3.39818120e-01 3.74063253e-01 3.45295705e-02 9.28215683e-02
1.28146684e+00 -8.35116386e-01 -6.29644513e-01 -3.12450200e-01
1.00329888e+00 -4.48123693e-01 1.01297176e+00 -1.78682789e-01
-4.11789209e-01 -1.48336202e-01 -1.02223384e+00 1.38812140e-01
-1.07852805e+00 -6.21583998e-01 8.90922487e-01 4.24087644e-01
-1.09394681e+00 4.10795778e-01 -7.23116934e-01 -1.04660022e+00
5.43976426e-01 2.19083995e-01 -7.44542778e-01 -5.65088451e-01
-1.23979747e+00 6.09018385e-01 2.31205747e-01 -5.21491766e-01
-4.00073856e-01 -7.55281925e-01 -9.46418941e-01 1.71943143e-01
5.66750705e-01 -7.81504571e-01 7.90842831e-01 -8.17862511e-01
-7.39067674e-01 7.00947464e-01 5.66398576e-02 -5.14875472e-01
2.87356764e-01 1.65838599e-02 -7.32708335e-01 2.53608793e-01
3.86762440e-01 3.56620729e-01 8.19005251e-01 -1.27875352e+00
-6.49152339e-01 -6.81122005e-01 3.28123629e-01 1.59107044e-01
-1.09518647e+00 5.23183011e-02 -3.10181230e-01 -4.02736694e-01
-1.01609364e-01 -7.16044188e-01 -3.01939458e-01 2.98790812e-01
2.31618751e-02 -3.59564781e-01 1.39391303e+00 -5.18573582e-01
1.02888918e+00 -2.12573981e+00 1.83640376e-01 3.83805245e-01
3.53492081e-01 -6.95241690e-02 -3.31073403e-01 1.04840219e+00
7.39619359e-02 2.50520647e-01 -1.84573203e-01 -3.14393222e-01
2.73128748e-01 3.64305645e-01 -2.41413906e-01 5.09915292e-01
4.39963639e-02 9.09221351e-01 -1.27904284e+00 -4.12452430e-01
2.58892089e-01 4.09502745e-01 -5.38186312e-01 7.63985068e-02
-1.06662311e-01 -1.10108413e-01 -7.43746221e-01 7.83103168e-01
4.27575737e-01 -4.98947203e-01 5.77896297e-01 -1.54939249e-01
3.60885471e-01 -1.66796327e-01 -1.14784515e+00 2.23599458e+00
-7.75537074e-01 5.94514728e-01 3.37311104e-02 -1.59446287e+00
9.10843790e-01 2.66319752e-01 8.74607861e-01 -6.09327912e-01
-2.63806701e-01 1.56748548e-01 -5.51323891e-01 -4.70400810e-01
7.28112757e-01 -3.56202982e-02 -3.47274721e-01 8.07064593e-01
4.95365337e-02 1.50415212e-01 -1.72328830e-01 5.66402078e-01
1.39740050e+00 -3.41641992e-01 3.42003942e-01 -8.38094503e-02
1.34518981e-01 1.42439008e-01 2.63850927e-01 5.70617557e-01
-2.45256603e-01 2.30365008e-01 3.53455037e-01 -3.04476947e-01
-1.04132497e+00 -1.02243066e+00 1.64816409e-01 1.19571495e+00
3.49853843e-01 -5.34153402e-01 -3.90993416e-01 -9.92321074e-01
5.92266977e-01 6.65183187e-01 -6.62489772e-01 -3.93538058e-01
-1.52845830e-01 -3.01057577e-01 1.46972343e-01 7.32687771e-01
-5.66864479e-03 -7.15147376e-01 -4.74339128e-01 3.30509752e-01
-8.43273327e-02 -1.18221152e+00 -3.91108751e-01 -1.64115325e-01
-6.68982625e-01 -1.21512854e+00 -4.71333176e-01 -7.71997690e-01
6.55901253e-01 8.62485588e-01 9.57433820e-01 -5.10397404e-02
-5.48274398e-01 1.05671597e+00 -6.38586581e-01 -1.58888489e-01
-3.07353675e-01 1.32902324e-01 2.39113227e-01 3.39339435e-01
7.73886442e-01 -4.83060330e-01 -5.34106970e-01 8.59556273e-02
-1.18925440e+00 -4.48401123e-01 2.43426070e-01 9.24238503e-01
4.80357826e-01 1.93415526e-02 7.74926901e-01 -1.07091594e+00
7.79725075e-01 -1.11331308e+00 -3.15969527e-01 5.00972927e-01
-9.39194798e-01 5.15946858e-02 5.79953551e-01 -4.02162820e-01
-5.26370347e-01 -4.96695161e-01 6.59040868e-01 -7.89590240e-01
1.79920755e-02 8.48604321e-01 -1.59295768e-01 -5.34021072e-02
4.73742425e-01 1.45778358e-01 2.80191451e-01 -4.10973966e-01
7.65615463e-01 8.96835625e-01 1.92017704e-01 -4.61031258e-01
9.31933284e-01 6.94259405e-01 -5.05602658e-02 -8.67807448e-01
-4.19788331e-01 -9.96911407e-01 -3.92515212e-01 6.42125160e-02
8.46726775e-01 -9.24651623e-01 -3.83228719e-01 -1.82355255e-01
-8.09891582e-01 1.96732618e-02 -5.10023475e-01 4.84513104e-01
-6.44745648e-01 4.54764217e-01 -2.35503223e-02 -3.88856679e-01
-2.66473860e-01 -8.82836998e-01 1.05537868e+00 -1.28864795e-01
-3.40830326e-01 -1.54542232e+00 -1.11829080e-02 4.56268251e-01
7.42687404e-01 1.66594148e-01 1.10663080e+00 -1.17028701e+00
-6.09685898e-01 -7.62575567e-01 -2.78352201e-01 7.24957809e-02
4.79411930e-01 -5.93805432e-01 -6.00079000e-01 -7.30587423e-01
-3.56521219e-01 -2.65674829e-01 4.75117445e-01 -1.58629388e-01
1.21712005e+00 -3.73266429e-01 -6.82353079e-01 4.30336237e-01
1.85024166e+00 -3.42003971e-01 6.70421273e-02 4.99302536e-01
7.46905625e-01 6.26902401e-01 7.57602394e-01 6.07661188e-01
6.03398621e-01 7.89607644e-01 5.42964399e-01 -7.44556412e-02
1.53146954e-02 -4.96978313e-01 1.33259743e-01 6.22227490e-01
3.12432796e-01 -2.08335817e-01 -9.36662316e-01 1.09119511e+00
-2.04292321e+00 -1.02930319e+00 3.50634485e-01 2.21714687e+00
1.48880661e-01 -2.40151152e-01 -1.37375970e-03 -1.08415261e-02
6.78315461e-01 4.61411178e-01 -5.89407504e-01 -2.58154094e-01
9.52207521e-02 9.46793333e-02 8.82776260e-01 1.88079670e-01
-8.50239038e-01 7.87889361e-01 5.37499666e+00 6.35269403e-01
-9.31686223e-01 4.86818761e-01 3.99786718e-02 -2.99107153e-02
-8.63077164e-01 3.14719051e-01 -4.03776646e-01 3.64793003e-01
9.75584388e-01 -6.46393239e-01 6.09869599e-01 9.99447405e-01
-2.33413756e-01 3.32976937e-01 -1.26880443e+00 1.10961902e+00
3.68116289e-01 -1.68633640e+00 1.24321364e-01 3.23864222e-01
6.69563770e-01 3.03147435e-01 -8.72184988e-03 3.81741852e-01
4.15154278e-01 -7.78105438e-01 1.74265578e-01 3.04825902e-01
7.97630787e-01 -6.51871562e-01 5.59141159e-01 6.10122122e-02
-1.37850213e+00 -2.54627675e-01 -4.90362883e-01 2.88824737e-01
4.54371944e-02 3.79344553e-01 -8.83417547e-01 1.02499962e+00
6.35012507e-01 9.28601444e-01 -4.98945415e-01 6.63335800e-01
2.09532261e-01 2.01200634e-01 -2.12036580e-01 -8.44253674e-02
3.47029716e-01 -2.56125569e-01 5.16440511e-01 9.99559462e-01
6.02290034e-01 -4.89619553e-01 4.70713437e-01 7.32314885e-01
-4.12050098e-01 4.37640846e-01 -1.30561030e+00 -4.35631990e-01
8.83142173e-01 1.17555106e+00 -5.56600571e-01 -1.78493991e-01
-1.00779200e+00 1.26029718e+00 4.68825608e-01 4.33048397e-01
-5.61548531e-01 -4.65853363e-01 1.05526125e+00 1.45431250e-01
2.23526791e-01 -2.05997661e-01 4.81101684e-02 -1.20174050e+00
2.27900192e-01 -5.87812543e-01 7.33993590e-01 -4.03147370e-01
-1.60340691e+00 1.91374362e-01 2.75808442e-02 -1.29944086e+00
-2.83685356e-01 -4.72805232e-01 -6.29536569e-01 5.66718400e-01
-1.85813713e+00 -1.46370125e+00 -2.03536972e-01 9.00648177e-01
4.61300552e-01 -6.70718610e-01 1.15761817e+00 1.82238847e-01
-4.93245907e-02 5.77840924e-01 5.14433742e-01 -4.45282422e-02
6.83337748e-01 -1.27568173e+00 3.84895861e-01 5.87258518e-01
4.97832596e-01 5.23847640e-01 4.59695935e-01 -4.07459736e-01
-2.00686336e+00 -1.28931677e+00 1.01322675e+00 -3.37696195e-01
1.13769007e+00 -5.35696983e-01 -9.53000963e-01 8.17631185e-01
1.45596951e-01 4.36641663e-01 1.05693686e+00 3.27484667e-01
-5.68470478e-01 -2.05348074e-01 -1.26188207e+00 5.14246941e-01
1.12261784e+00 -1.03046799e+00 -4.44946885e-01 6.17080688e-01
1.04834962e+00 3.74634534e-01 -1.10109282e+00 1.11584924e-02
5.01698777e-02 -4.83865291e-01 1.04367781e+00 -1.01638675e+00
7.63153657e-02 -1.22230686e-01 -8.21213841e-01 -1.50067842e+00
-1.37652501e-01 -4.15103763e-01 -3.72920245e-01 1.27837527e+00
2.12078780e-01 -9.37521517e-01 8.23076010e-01 6.62085235e-01
1.35712475e-01 -3.94556552e-01 -9.17927146e-01 -1.12985659e+00
-7.59236440e-02 -4.25119311e-01 9.64280248e-01 1.47065926e+00
2.33958632e-01 3.31235416e-02 -9.32844505e-02 1.99692458e-01
7.77485132e-01 4.64919597e-01 8.69155586e-01 -1.26381493e+00
-2.06266716e-01 -2.39349917e-01 -9.53662694e-01 -5.43829501e-01
5.10542393e-01 -1.44160366e+00 -6.51893139e-01 -1.86719489e+00
1.29617974e-01 -5.44786990e-01 -7.24017918e-01 5.55056751e-01
1.10344030e-01 -1.41084686e-01 2.81606942e-01 1.87955186e-01
-5.39370835e-01 6.22469246e-01 5.81210673e-01 -4.53232050e-01
1.16309270e-01 -3.32947046e-01 -9.60045695e-01 1.85905382e-01
6.56073570e-01 -5.01852393e-01 -7.98968196e-01 -4.01349604e-01
1.02831282e-01 -2.98333652e-02 5.39679945e-01 -6.57416761e-01
3.34666193e-01 -3.76007199e-01 -2.29147077e-01 -8.33216086e-02
1.79188594e-01 -1.32064188e+00 6.50238767e-02 2.21994743e-01
-3.51067871e-01 1.88466877e-01 -1.20695569e-01 1.22037971e+00
-4.30146635e-01 4.72197421e-02 3.38266820e-01 1.58795100e-02
-1.21150529e+00 6.04974866e-01 1.96448490e-01 1.88761488e-01
1.44797480e+00 -2.47618452e-01 -1.74715355e-01 -4.26009923e-01
-4.16073799e-01 3.31270754e-01 7.54434884e-01 9.16549027e-01
7.67567694e-01 -1.55309331e+00 -4.72767174e-01 2.85154611e-01
9.59227145e-01 -4.23155427e-01 2.01334566e-01 4.90238398e-01
-2.81879064e-02 3.16930622e-01 -1.03223227e-01 -4.49746549e-01
-1.05710649e+00 9.52288091e-01 -2.53161013e-01 -1.70471385e-01
-6.93307638e-01 4.40616876e-01 -1.21712647e-01 -7.13480473e-01
1.34275546e-02 1.80214390e-01 1.05745755e-01 2.96665341e-01
3.10676038e-01 1.83085307e-01 1.66128531e-01 -4.85859543e-01
-3.94566208e-01 4.02012080e-01 -1.41502589e-01 6.49580508e-02
1.54477918e+00 -2.57935852e-01 -4.78957854e-02 5.10662496e-01
1.79842496e+00 -3.59813541e-01 -6.87699020e-01 -7.57469714e-01
2.48994887e-01 -7.74023592e-01 1.54814959e-01 -3.25183570e-01
-1.07446587e+00 7.06930995e-01 6.55434310e-01 2.43358925e-01
8.52183044e-01 2.81410277e-01 9.06085134e-01 4.67927903e-01
7.65516698e-01 -1.28108120e+00 1.21494763e-01 -2.45057736e-02
5.62436700e-01 -1.47204709e+00 -1.92763135e-01 -1.82547420e-01
-6.89568698e-01 8.65312517e-01 2.56659448e-01 3.95719223e-02
9.65845764e-01 -1.81399748e-01 -3.80599909e-02 -4.82831448e-01
-7.93851674e-01 -1.34024918e-01 -1.48372352e-02 8.58911455e-01
8.78394544e-02 3.29113752e-01 -1.45421043e-01 2.01152220e-01
1.67956263e-01 -2.10813552e-01 4.95957255e-01 1.27325714e+00
-1.29716322e-01 -1.41380084e+00 3.80108580e-02 6.61167622e-01
-1.57544851e-01 3.71852960e-03 -9.93822888e-02 7.27937639e-01
-3.19539875e-01 8.70252848e-01 8.81270394e-02 -2.93509603e-01
2.38755882e-01 2.74179846e-01 3.85130458e-02 -7.24978209e-01
-1.45292133e-01 -4.95939076e-01 -8.12946931e-02 -8.25568438e-01
-4.65622783e-01 -8.24305296e-01 -1.18271565e+00 -1.87555328e-01
-5.63759506e-02 1.48027182e-01 9.56445277e-01 6.33399367e-01
7.77198792e-01 2.44782701e-01 8.72889578e-01 -4.17901725e-01
-8.78561318e-01 -4.65611935e-01 -9.18394446e-01 1.04413652e+00
7.92159066e-02 -7.47839212e-01 -3.13263148e-01 -1.71616286e-01]
|
[8.694659233093262, 7.8178181648254395]
|
3fdf5c22-8fb7-4b02-98a7-ed8b5f6f03f9
|
semantic-preserving-linguistic-steganography
|
2203.03795
| null |
https://arxiv.org/abs/2203.03795v1
|
https://arxiv.org/pdf/2203.03795v1.pdf
|
Semantic-Preserving Linguistic Steganography by Pivot Translation and Semantic-Aware Bins Coding
|
Linguistic steganography (LS) aims to embed secret information into a highly encoded text for covert communication. It can be roughly divided to two main categories, i.e., modification based LS (MLS) and generation based LS (GLS). Unlike MLS that hides secret data by slightly modifying a given text without impairing the meaning of the text, GLS uses a trained language model to directly generate a text carrying secret data. A common disadvantage for MLS methods is that the embedding payload is very low, whose return is well preserving the semantic quality of the text. In contrast, GLS allows the data hider to embed a high payload, which has to pay the high price of uncontrollable semantics. In this paper, we propose a novel LS method to modify a given text by pivoting it between two different languages and embed secret data by applying a GLS-like information encoding strategy. Our purpose is to alter the expression of the given text, enabling a high payload to be embedded while keeping the semantic information unchanged. Experimental results have shown that the proposed work not only achieves a high embedding payload, but also shows superior performance in maintaining the semantic consistency and resisting linguistic steganalysis.
|
['Xinpeng Zhang', 'Guorui Feng', 'Biao Yi', 'Hanzhou Wu', 'Tianyu Yang']
|
2022-03-08
| null | null | null | null |
['steganalysis']
|
['computer-vision']
|
[ 9.70661700e-01 5.85097313e-01 3.42738554e-02 -1.26121908e-01
-2.18291968e-01 -6.35794282e-01 6.55202031e-01 1.14211375e-02
-3.28574836e-01 6.68379247e-01 2.05191121e-01 -4.16526705e-01
5.61435640e-01 -1.18133867e+00 -7.59973407e-01 -8.12271774e-01
1.69807732e-01 -2.91536182e-01 4.42765236e-01 -4.78548408e-01
3.32275569e-01 1.07274398e-01 -1.16511106e+00 3.43243301e-01
8.65988791e-01 5.79589963e-01 2.02543110e-01 3.23150218e-01
-2.87593186e-01 6.68974280e-01 -8.71643722e-01 -3.55031699e-01
3.10158700e-01 -8.61173093e-01 -4.80149567e-01 5.47098555e-02
-3.81188691e-01 -2.49318108e-01 -3.60268176e-01 1.41227412e+00
1.44229263e-01 -5.55033326e-01 4.18503612e-01 -1.16851342e+00
-7.96156764e-01 8.27205956e-01 -3.51432860e-01 -5.04602671e-01
3.93135369e-01 8.93319845e-02 5.46001375e-01 -3.43650430e-02
6.56968772e-01 1.38128853e+00 3.13555211e-01 7.85890937e-01
-9.38105464e-01 -9.02986825e-01 -7.94815421e-02 -2.50719815e-01
-1.44201231e+00 -5.23533940e-01 9.69290674e-01 -8.48507602e-03
6.02064371e-01 7.15316355e-01 7.03743219e-01 8.03383827e-01
7.60603786e-01 5.00126541e-01 1.42187357e+00 -6.86421394e-01
2.03310102e-01 8.33396673e-01 -4.73664224e-01 8.00500810e-01
6.97964251e-01 1.60805568e-01 -2.17263386e-01 -1.52434513e-01
2.38054529e-01 3.74563411e-02 -6.32044911e-01 -1.31681785e-01
-1.39254439e+00 9.05643225e-01 2.68478483e-01 6.80550516e-01
1.32036507e-01 2.37824231e-01 3.78760129e-01 7.70334482e-01
3.81624818e-01 2.33804397e-02 -4.43636179e-02 3.28228176e-01
-6.56858027e-01 -7.48822987e-02 1.04969788e+00 8.89333785e-01
6.67257071e-01 2.63802826e-01 2.87217170e-01 8.18463117e-02
7.53938496e-01 8.96461666e-01 7.84992456e-01 -2.53783941e-01
6.17230892e-01 5.90753376e-01 -1.51105717e-01 -1.66012871e+00
1.27388984e-01 3.94847849e-03 -8.22303236e-01 4.04312491e-01
4.39445935e-02 -8.94892290e-02 -7.01871455e-01 1.87253690e+00
5.21647558e-02 -2.66232044e-01 5.51451564e-01 5.66000044e-01
7.57189214e-01 1.01482451e+00 -1.78392515e-01 -3.26663405e-01
1.26763332e+00 -5.28497696e-01 -1.17641866e+00 -3.11012208e-01
9.03353631e-01 -7.91977346e-01 6.34194553e-01 -1.20831199e-01
-6.82488501e-01 -2.05641881e-01 -1.58664572e+00 1.64616242e-01
-7.24953651e-01 -4.54831541e-01 2.31034547e-01 1.25028098e+00
-1.07702577e+00 1.99474126e-01 -4.09874320e-01 -1.48067042e-01
-7.15667731e-04 5.39967775e-01 -5.04055679e-01 2.13782698e-01
-1.85673440e+00 6.36905789e-01 1.12689686e+00 -1.56942621e-01
-4.86435056e-01 -1.71709314e-01 -1.19590306e+00 -6.73520118e-02
3.51698130e-01 -3.59110206e-01 6.05946183e-01 -1.23843241e+00
-1.78857398e+00 7.97066629e-01 -3.96321826e-02 -4.73388195e-01
5.95037341e-01 5.83275139e-01 -9.64637458e-01 1.12121381e-01
-4.75848541e-02 4.58597720e-01 1.46052086e+00 -1.32112622e+00
-4.62922722e-01 -9.83349904e-02 -1.46450726e-02 -3.55204083e-02
-5.55388689e-01 -7.10891113e-02 -1.49021804e-01 -8.58843267e-01
1.81722388e-01 -1.01017165e+00 1.13409705e-01 2.19615642e-03
-6.12381101e-01 3.67572337e-01 1.51122260e+00 -6.45422339e-01
1.33394766e+00 -2.36926389e+00 -7.32902884e-02 3.82970333e-01
2.36082777e-01 6.09694839e-01 1.25743240e-01 8.06357503e-01
4.23922390e-02 5.72920561e-01 -6.72425449e-01 1.54656789e-03
-5.46811298e-02 2.40789905e-01 -4.32984263e-01 5.85991383e-01
-2.98565067e-02 1.08458972e+00 -7.29789376e-01 -4.60026920e-01
1.83671206e-01 5.39977908e-01 -4.02718812e-01 -2.90802084e-02
-2.49379441e-01 3.23800176e-01 -6.00990534e-01 3.45817685e-01
8.22523296e-01 4.38754968e-02 4.44234252e-01 -9.62359831e-02
1.55529976e-02 3.26935761e-02 -1.01589000e+00 1.11718023e+00
-2.92610466e-01 4.93643463e-01 -4.38569263e-02 -6.93520367e-01
1.28097165e+00 5.65021336e-01 7.58478185e-04 -5.16151786e-01
3.17854732e-01 4.51386303e-01 -1.07529938e-01 -5.13388455e-01
4.66717541e-01 -4.51849550e-01 -4.32088733e-01 7.46158838e-01
-6.80899739e-01 -1.82938665e-01 -3.71692181e-01 1.59398556e-01
7.69381881e-01 3.07411086e-02 4.59164500e-01 -2.08983213e-01
1.02088869e+00 -3.20692867e-01 3.75633180e-01 3.84438008e-01
9.37455297e-02 -5.10114655e-02 4.69599068e-01 3.64702381e-02
-1.08696151e+00 -5.07254601e-01 3.22674185e-01 2.32684329e-01
6.54855251e-01 -3.95864069e-01 -9.26136851e-01 -8.87940288e-01
-6.86528161e-02 8.98590326e-01 -4.68184084e-01 -7.55446196e-01
-6.86547041e-01 -5.16533077e-01 9.27198112e-01 -2.49112919e-01
1.21294212e+00 -1.06268919e+00 -5.75641334e-01 1.22921944e-01
-2.42231697e-01 -9.71777439e-01 -5.90401232e-01 -2.47453809e-01
-6.18547857e-01 -7.12633312e-01 -4.65385348e-01 -1.04824293e+00
1.15978980e+00 3.41880888e-01 6.60857558e-02 3.74043286e-01
3.28813672e-01 -1.82173193e-01 -6.07640028e-01 -3.83558154e-01
-1.34725571e+00 2.98856068e-02 -2.83290595e-01 4.06691641e-01
1.92825019e-01 -3.59841615e-01 -2.91160524e-01 1.12677507e-01
-1.74762714e+00 3.41092408e-01 7.42601693e-01 5.70074081e-01
9.61035639e-02 6.28243625e-01 3.93793434e-01 -1.18606329e+00
4.56791312e-01 -3.75799119e-01 -3.85613829e-01 2.12751716e-01
-8.24885607e-01 2.83401906e-01 9.96193349e-01 -4.35210168e-01
-8.82146776e-01 -3.39019358e-01 -9.48825553e-02 2.10949585e-01
1.76126674e-01 2.66649187e-01 -5.26959598e-01 -6.25291467e-01
5.78725114e-02 8.54117811e-01 4.62754995e-01 -2.53738433e-01
1.77867457e-01 1.02511239e+00 1.15760349e-01 3.30270529e-02
1.38219416e+00 8.62453878e-01 3.54884528e-02 -7.54002452e-01
-1.14729673e-01 2.84866482e-01 -2.62719482e-01 5.39062954e-02
7.42888391e-01 -4.83142823e-01 -6.36826396e-01 8.10336411e-01
-1.03795993e+00 8.82068351e-02 -1.81655884e-01 2.77961306e-02
-3.89091790e-01 9.38546181e-01 -3.63097072e-01 -5.64390421e-01
-3.53546023e-01 -1.05252373e+00 7.43591905e-01 -2.21921921e-01
1.33240754e-02 -1.25291002e+00 -3.68852794e-01 2.21059203e-01
4.46224928e-01 8.45491230e-01 1.06756413e+00 -7.15510607e-01
-5.61518967e-01 -5.49202025e-01 -5.72426915e-02 4.10986692e-01
7.35570371e-01 -3.68007660e-01 -5.43334603e-01 -7.21874118e-01
5.19290268e-01 1.54300317e-01 6.27311528e-01 -5.43420792e-01
5.75715065e-01 -1.07058978e+00 -3.39567661e-01 7.85794377e-01
1.59291518e+00 6.68843269e-01 9.54047024e-01 5.99318445e-01
6.21526003e-01 6.12827003e-01 3.89512658e-01 2.77589876e-02
1.97914004e-01 4.39176768e-01 3.57413679e-01 -9.87941921e-02
-4.85118739e-02 -7.20593154e-01 7.24746227e-01 9.96619165e-01
4.40676332e-01 -7.71854997e-01 -3.21335405e-01 8.21209252e-02
-1.53218603e+00 -8.36562395e-01 -3.60832140e-02 1.99996400e+00
1.00375521e+00 2.66197264e-01 -4.86114621e-01 6.12869382e-01
1.02375233e+00 6.46837175e-01 -2.46775791e-01 -6.88026309e-01
-2.70479530e-01 -1.19531140e-01 9.13790762e-01 7.60819972e-01
-8.10333490e-01 1.05449212e+00 5.67841387e+00 1.09692907e+00
-1.44508326e+00 1.08416118e-01 1.05304912e-01 4.57625866e-01
-8.25199962e-01 2.93512821e-01 -6.25381470e-01 1.04598832e+00
8.64613771e-01 -4.78643000e-01 4.04283255e-01 9.08333436e-02
1.69370666e-01 6.30170479e-02 -5.45392334e-01 5.64708233e-01
3.34007591e-01 -1.16087794e+00 5.81688643e-01 2.34610811e-01
5.03891170e-01 -8.69124532e-01 1.05963729e-01 -1.11735791e-01
-1.98746547e-01 -7.69687891e-01 9.41078722e-01 3.24597180e-01
1.08735025e+00 -8.55637133e-01 8.15906405e-01 4.75075096e-01
-8.91339302e-01 1.37467712e-01 -2.01024592e-01 9.05108675e-02
8.80473293e-03 3.75187367e-01 -6.90424860e-01 7.52346694e-01
1.02213107e-01 5.70265591e-01 -3.70802701e-01 1.28483206e-01
-4.72956419e-01 4.12271172e-01 -1.24439947e-01 -3.77833903e-01
4.21931207e-01 -1.99485332e-01 1.08063161e+00 1.14246535e+00
3.96577686e-01 3.40378359e-02 -1.48515124e-03 8.17581594e-01
-8.86135325e-02 1.51453212e-01 -9.33804095e-01 -5.09791434e-01
4.36592996e-01 5.97366691e-01 -7.14340270e-01 -4.71697986e-01
-1.41315192e-01 1.39219189e+00 -4.84434009e-01 1.99750915e-01
-7.43325233e-01 -1.03968978e+00 7.45629892e-02 3.63078266e-01
4.08554524e-01 -2.61054814e-01 -1.02454759e-01 -1.11691070e+00
1.45997807e-01 -9.41628039e-01 -8.81949365e-02 -4.04633433e-01
-4.36615467e-01 4.28515315e-01 -2.32154459e-01 -1.40328908e+00
-3.90196405e-02 -9.20442492e-02 -2.37386286e-01 7.97758341e-01
-1.69172621e+00 -1.40703046e+00 -1.56566855e-02 7.14747846e-01
1.92199677e-01 -3.25124204e-01 8.50398064e-01 4.74901777e-03
-2.44003877e-01 7.76591957e-01 2.43360683e-01 2.36030787e-01
5.14136493e-01 -6.93376005e-01 3.88293326e-01 9.09173846e-01
-4.45558429e-01 6.91194654e-01 9.42543268e-01 -1.08385062e+00
-1.48487163e+00 -1.20051908e+00 1.26798856e+00 9.58183408e-02
5.70028603e-01 -6.03645980e-01 -8.51510406e-01 7.48209357e-01
1.83585256e-01 -5.82009554e-01 5.13909698e-01 -1.14444578e+00
-4.42665160e-01 -1.06456336e-02 -1.75576568e+00 7.22067356e-01
6.94924176e-01 -5.48079252e-01 -6.94487691e-01 1.52269796e-01
1.28923047e+00 -2.16233015e-01 -5.27918279e-01 1.84310004e-01
4.85104829e-01 -9.35711980e-01 5.70856273e-01 -1.38956886e-02
2.74371088e-01 -6.46673560e-01 -1.99047208e-01 -9.79126751e-01
-8.05697590e-02 -9.67473269e-01 -2.20627692e-02 1.34454668e+00
1.58474550e-01 -1.37921417e+00 5.00912488e-01 2.81587958e-01
3.19001555e-01 -2.15046152e-01 -8.07901382e-01 -9.43796217e-01
-1.43582642e-01 6.07015332e-03 9.84043837e-01 1.05194569e+00
4.67463285e-02 -8.83209109e-02 -7.57970572e-01 2.61607468e-01
7.82368362e-01 -8.04279000e-02 5.06219387e-01 -7.50143528e-01
-6.24873675e-02 -9.99512151e-02 -7.15476215e-01 -8.49025905e-01
1.93890125e-01 -1.22358465e+00 -1.19150512e-01 -1.07767367e+00
-1.43376574e-01 -3.59928340e-01 -2.78899074e-02 3.86835575e-01
1.67297542e-01 4.79147941e-01 2.11389124e-01 4.31573302e-01
-1.25679880e-01 4.25494999e-01 1.32210886e+00 -2.94391841e-01
-1.74810946e-01 -1.47285730e-01 -8.99500370e-01 5.06821811e-01
9.31642592e-01 -8.44170153e-01 -5.29810548e-01 -1.68258548e-01
2.22272336e-01 -7.02829510e-02 2.70628273e-01 -7.23400950e-01
3.85812484e-02 -2.70430297e-01 -1.09276205e-01 -8.54011104e-02
-1.62652582e-02 -1.31230640e+00 4.50179964e-01 1.34980118e+00
-3.06577265e-01 -5.71811736e-01 -6.89509436e-02 6.34153068e-01
-1.62630171e-01 -4.28918034e-01 9.17701244e-01 -2.07331032e-01
-6.74953222e-01 -9.67880432e-03 -6.64051771e-01 -3.65754098e-01
1.29135704e+00 -5.59701145e-01 -2.39381582e-01 -4.53523815e-01
-3.32865506e-01 -1.93526790e-01 8.42339039e-01 5.82967281e-01
6.88153982e-01 -1.28123856e+00 -5.27952194e-01 8.40945721e-01
6.04612231e-02 -4.82259274e-01 -4.79388982e-02 3.20307881e-01
-8.23841810e-01 6.27575219e-01 -1.67738907e-02 -1.87478170e-01
-1.34308290e+00 7.52735317e-01 7.55074173e-02 -1.50638759e-01
-8.21714461e-01 1.34752542e-01 1.20140582e-01 -3.16964298e-01
-1.20016448e-01 -7.16722012e-02 -1.06309384e-01 -2.99329400e-01
5.98533750e-01 1.11484759e-01 -3.28616291e-01 -8.11380148e-01
-1.06627993e-01 6.62244499e-01 5.26463017e-02 -1.01012088e-01
8.56057525e-01 -6.32677257e-01 -6.77616596e-01 1.81963980e-01
1.57298172e+00 5.72883546e-01 -5.63569307e-01 -3.00977767e-01
-1.19633071e-01 -6.64881170e-01 -3.86207029e-02 -6.11070931e-01
-9.90037084e-01 4.48188812e-01 3.10867906e-01 6.26346350e-01
1.04354191e+00 -3.79556149e-01 1.44343448e+00 2.60322779e-01
6.93538964e-01 -7.91805387e-01 -1.50239438e-01 1.13291107e-01
7.10495591e-01 -9.21384573e-01 -1.32477269e-01 -7.36870587e-01
-4.89374489e-01 1.16118538e+00 2.09929161e-02 4.11848389e-02
5.91345429e-01 4.13251251e-01 5.17212562e-02 -2.34730784e-02
-2.73538679e-01 2.58009791e-01 -9.04630423e-02 6.73826039e-01
-1.63240179e-01 -2.60712635e-02 -6.76558912e-01 9.26414803e-02
-4.98014510e-01 -1.02693543e-01 8.66267562e-01 1.19457841e+00
-7.83653021e-01 -1.36078346e+00 -6.24004900e-01 -1.97592396e-02
-7.07369268e-01 -7.50068724e-02 -4.79375333e-01 8.01545143e-01
4.33614999e-01 1.14624488e+00 -2.28796646e-01 -6.66198611e-01
-3.95698324e-02 6.46082833e-02 3.47592905e-02 -4.33694929e-01
-4.08732861e-01 -5.66301718e-02 -1.44771216e-02 -1.68214723e-01
-5.44320166e-01 -2.70835578e-01 -1.37954950e+00 -7.94244766e-01
-2.03367397e-01 3.20858449e-01 6.90124035e-01 9.29380834e-01
8.29965994e-02 4.63218927e-01 1.11953676e+00 1.22834668e-01
-3.11690897e-01 -4.91122901e-01 -7.24133492e-01 3.64620388e-01
6.07644558e-01 1.68659370e-02 -7.85694718e-01 2.75663316e-01]
|
[4.328485488891602, 8.050503730773926]
|
d091326f-38ec-450b-8c8d-93e13222df42
|
robustsleepnet-transfer-learning-for
|
2101.02452
| null |
https://arxiv.org/abs/2101.02452v2
|
https://arxiv.org/pdf/2101.02452v2.pdf
|
RobustSleepNet: Transfer learning for automated sleep staging at scale
|
Sleep disorder diagnosis relies on the analysis of polysomnography (PSG) records. As a preliminary step of this examination, sleep stages are systematically determined. In practice, sleep stage classification relies on the visual inspection of 30-second epochs of polysomnography signals. Numerous automatic approaches have been developed to replace this tedious and expensive task. Although these methods demonstrated better performance than human sleep experts on specific datasets, they remain largely unused in sleep clinics. The main reason is that each sleep clinic uses a specific PSG montage that most automatic approaches cannot handle out-of-the-box. Moreover, even when the PSG montage is compatible, publications have shown that automatic approaches perform poorly on unseen data with different demographics. To address these issues, we introduce RobustSleepNet, a deep learning model for automatic sleep stage classification able to handle arbitrary PSG montages. We trained and evaluated this model in a leave-one-out-dataset fashion on a large corpus of 8 heterogeneous sleep staging datasets to make it robust to demographic changes. When evaluated on an unseen dataset, RobustSleepNet reaches 97% of the F1 of a model explicitly trained on this dataset. Hence, RobustSleepNet unlocks the possibility to perform high-quality out-of-the-box automatic sleep staging with any clinical setup. We further show that finetuning RobustSleepNet, using a part of the unseen dataset, increases the F1 by 2% when compared to a model trained specifically for this dataset. Therefore, finetuning might be used to reach a state-of-the-art level of performance on a specific population.
|
['Valentin Thorey', 'Antoine Guillot']
|
2021-01-07
| null | null | null | null |
['sleep-staging', 'automatic-sleep-stage-classification']
|
['medical', 'medical']
|
[ 1.71566736e-02 8.83232355e-02 -8.46587494e-02 -5.32342911e-01
-5.40503860e-01 -4.76251841e-01 4.64056991e-02 3.12725157e-01
-7.85877347e-01 7.59464920e-01 -1.03369161e-01 -3.33017588e-01
-2.50297245e-02 -3.54455501e-01 -1.36022881e-01 -6.78662658e-01
5.10366037e-02 8.68731797e-01 4.19088125e-01 -1.49146497e-01
-2.33535364e-01 4.04061973e-01 -1.59989953e+00 7.51729384e-02
8.39012802e-01 9.00118828e-01 2.81577468e-01 5.53188086e-01
1.03529446e-01 -5.21628112e-02 -7.99330652e-01 -2.07408458e-01
2.22304583e-01 -4.00398612e-01 -5.64089596e-01 -5.57260402e-02
4.11009222e-01 6.87648952e-02 8.75140056e-02 9.90040421e-01
7.62470901e-01 -9.70024541e-02 1.95049152e-01 -9.31935370e-01
4.26320126e-03 2.35003710e-01 8.34542140e-02 7.49019802e-01
2.35890329e-01 2.68370301e-01 8.63735318e-01 -3.33468884e-01
5.20905495e-01 3.48713189e-01 7.67052412e-01 8.15086067e-01
-1.46147430e+00 -6.91958547e-01 -2.65270680e-01 3.22112702e-02
-1.24063826e+00 -5.87028623e-01 3.89476448e-01 -2.48568058e-01
8.44590962e-01 3.74593198e-01 1.22682750e+00 9.96574938e-01
3.76928926e-01 2.40591928e-01 1.36213124e+00 -1.19969100e-01
4.53569710e-01 1.82781622e-01 1.07231565e-01 5.63005269e-01
5.51704586e-01 -2.31446251e-01 -3.16523522e-01 6.67504743e-02
3.95152360e-01 1.56631067e-01 -3.14691782e-01 -2.73192316e-01
-1.03401268e+00 4.42054033e-01 2.95907557e-01 8.23289573e-01
-2.05959767e-01 -2.39019841e-01 3.60450566e-01 2.68434465e-01
4.44377393e-01 7.29665935e-01 -5.14614344e-01 -4.79080260e-01
-1.78799605e+00 -2.03521103e-01 7.21829414e-01 3.97906482e-01
5.03192842e-01 -2.42873311e-01 -3.51061486e-02 7.09919214e-01
-1.02509046e-03 4.07062680e-01 8.89220297e-01 -5.63749135e-01
9.37559903e-02 9.50677156e-01 -7.01215565e-02 -5.30752838e-01
-1.17708790e+00 -8.71973813e-01 -7.00945854e-01 9.00205150e-02
5.59066296e-01 -1.54921757e-02 -1.02918267e+00 1.57309246e+00
1.47963688e-02 -4.88571264e-02 -2.84759074e-01 7.13084638e-01
8.19091976e-01 -2.82908492e-02 2.67752931e-02 -3.52780014e-01
1.64021981e+00 -5.84345520e-01 -4.60740119e-01 -5.79174995e-01
4.53005582e-01 -5.34685969e-01 1.22649765e+00 7.02882290e-01
-1.03988433e+00 -3.84492129e-01 -1.20157599e+00 -3.14739831e-02
-3.76258403e-01 3.29884529e-01 2.55293846e-01 9.33093607e-01
-1.39065742e+00 8.83380473e-01 -1.49520791e+00 -8.08934271e-01
4.45599109e-01 1.01366842e+00 -6.04097724e-01 3.87179822e-01
-8.63823950e-01 9.30628538e-01 2.35738412e-01 2.52041578e-01
-6.71338558e-01 -5.85994005e-01 -5.85916519e-01 2.90686488e-01
2.57531643e-01 -9.81184006e-01 9.63743746e-01 -6.54022336e-01
-1.18102860e+00 1.44014263e+00 -2.87140220e-01 -5.71802974e-01
5.62107205e-01 1.32477563e-03 -6.65364146e-01 2.54145771e-01
1.13242768e-01 3.59797597e-01 9.92600918e-01 -6.24360144e-01
-3.46726060e-01 -4.73589003e-01 1.90864392e-02 -2.76294768e-01
-1.81722686e-01 -7.43499845e-02 -5.37086368e-01 -3.40676069e-01
1.24017403e-01 -1.17465460e+00 -2.57730842e-01 -2.18874678e-01
-3.19848865e-01 5.56772351e-02 2.82083780e-01 -5.32194376e-01
1.38616514e+00 -2.28794765e+00 -7.12277293e-02 -3.75938341e-02
6.50140464e-01 5.72414100e-01 2.77530670e-01 1.26544982e-01
-2.50725418e-01 1.37221768e-01 -1.73896179e-01 -8.49382222e-01
-1.93442851e-01 2.52711803e-01 4.26338464e-01 6.78846002e-01
-1.93869784e-01 7.14143276e-01 -8.61610293e-01 -4.21129107e-01
2.33820796e-01 2.67930478e-01 -6.59290731e-01 9.89112705e-02
2.65306324e-01 6.77011490e-01 -8.33977386e-02 4.62809831e-01
3.17634463e-01 -5.38822293e-01 2.59296060e-01 -1.09337136e-01
8.54680222e-03 4.25208062e-01 -7.14096725e-01 2.01518035e+00
-4.86540675e-01 6.54797614e-01 1.00360783e-02 -7.64438212e-01
7.92130232e-01 3.24836135e-01 4.50959742e-01 -4.42292571e-01
3.14193815e-01 5.73980570e-01 4.55872118e-01 -4.21750069e-01
3.69602621e-01 -6.88668489e-01 2.43834723e-02 2.27445737e-01
2.63850093e-01 7.54177272e-02 4.19067979e-01 -8.20941478e-02
1.45798194e+00 -4.20532763e-01 5.36371946e-01 -4.79166538e-01
6.04778409e-01 -3.86653095e-01 6.24357522e-01 6.68030500e-01
-6.01204634e-01 1.00262141e+00 5.61139643e-01 -4.70584452e-01
-7.16501236e-01 -1.06784582e+00 -3.22335154e-01 6.32124662e-01
-1.32922545e-01 -8.17113280e-01 -7.29985118e-01 -9.87928033e-01
-3.19282800e-01 4.59449053e-01 -9.23584104e-01 -2.10512042e-01
-1.94284692e-01 -9.65488851e-01 4.25423563e-01 4.10653144e-01
2.87612259e-01 -1.09172273e+00 -7.83725858e-01 1.53335273e-01
9.51150209e-02 -1.16115057e+00 -1.34915903e-01 4.12085593e-01
-9.38924074e-01 -1.32167184e+00 -7.44938731e-01 -4.61149991e-01
7.01413453e-01 -1.70693561e-01 1.29689729e+00 2.31643543e-01
-2.84413695e-01 1.84167460e-01 -2.12990865e-01 -9.97420400e-02
-3.78329098e-01 6.05652630e-01 4.43462223e-01 6.51626736e-02
5.78828931e-01 -1.08391285e+00 -1.10224128e+00 4.29408789e-01
-5.98607302e-01 -2.49179572e-01 5.10927796e-01 5.90301573e-01
5.15550077e-01 -2.31182575e-01 5.03612578e-01 -8.10172141e-01
4.20838833e-01 -2.30592370e-01 -5.60163319e-01 4.72117588e-02
-8.29769611e-01 -2.31772903e-02 7.06759632e-01 -2.07129583e-01
-2.79254586e-01 -6.61936030e-02 -4.66563374e-01 -4.46593970e-01
-3.43000531e-01 1.13372870e-01 -6.51832893e-02 1.25293732e-01
6.39556348e-01 2.22152263e-01 1.42634958e-01 -6.56003058e-01
-2.07484305e-01 6.84433699e-01 4.80054528e-01 1.77276768e-02
6.23213351e-01 5.75823903e-01 1.10919289e-01 -8.82002711e-01
-8.78610611e-01 -7.31984019e-01 -8.01229477e-01 9.55232382e-02
1.23844779e+00 -7.40237892e-01 -8.08417737e-01 2.31061861e-01
-4.24372315e-01 -4.94793534e-01 -3.28170687e-01 3.84698927e-01
-3.47630143e-01 1.32337689e-01 -1.13205798e-01 -5.05919933e-01
-5.93818009e-01 -1.34996569e+00 1.07342041e+00 2.60786265e-01
-6.00135624e-01 -1.09006703e+00 2.67315537e-01 3.69575560e-01
4.80567694e-01 -4.54243757e-02 6.62221730e-01 -1.18102646e+00
-9.44795161e-02 -3.20900977e-01 1.34848684e-01 4.66488570e-01
4.14987177e-01 -2.74980873e-01 -1.06094038e+00 -4.81023550e-01
2.79458970e-01 5.37833422e-02 7.76457667e-01 3.71139675e-01
1.00175226e+00 1.80244371e-01 -2.41972804e-01 7.87525952e-01
1.17641568e+00 6.65398315e-02 6.06317222e-01 4.78659689e-01
3.75552982e-01 4.12486903e-02 1.01730786e-01 1.05505034e-01
3.26494277e-01 6.59771144e-01 2.77754217e-01 -1.50408626e-01
-1.28145844e-01 1.38086975e-01 1.42310709e-01 6.80365443e-01
-6.50811568e-02 -2.03545056e-02 -8.29386532e-01 5.68873763e-01
-1.40668178e+00 -6.74250901e-01 3.96416448e-02 2.31324959e+00
5.98953366e-01 6.10175073e-01 5.14338791e-01 3.03063422e-01
5.04114568e-01 2.35535011e-01 -4.73922104e-01 -3.41706008e-01
1.14257731e-01 5.42647839e-01 3.31821203e-01 5.37386686e-02
-7.95754135e-01 5.87120235e-01 6.15667200e+00 4.66874689e-01
-1.59502602e+00 4.22389477e-01 2.93210566e-01 -5.75067043e-01
1.83425710e-01 -1.43190265e-01 -8.21597993e-01 9.55738842e-01
1.19507444e+00 2.07064971e-01 4.53080356e-01 7.89598107e-01
4.80097771e-01 -3.73239040e-01 -1.21567535e+00 1.24782765e+00
7.65929595e-02 -1.13840723e+00 -7.23150253e-01 2.07523078e-01
2.90923893e-01 4.35994238e-01 -1.99960828e-01 2.96733260e-01
-5.54866672e-01 -9.67790842e-01 2.65857816e-01 3.98504823e-01
1.09435689e+00 -2.25856125e-01 9.74498868e-01 1.44438684e-01
-8.41888726e-01 -6.01249933e-03 -4.57014591e-02 3.18611711e-02
3.00122440e-01 5.94857991e-01 -9.84653413e-01 2.96853065e-01
7.22331285e-01 7.62827098e-01 -1.21897066e+00 1.37506425e+00
-3.77319425e-01 7.45527565e-01 -4.57302034e-01 1.34129390e-01
-4.35211174e-02 -1.59127428e-03 5.00655830e-01 8.60521019e-01
4.88877237e-01 -1.37192458e-01 -1.08140856e-01 7.79314637e-01
-8.24911147e-03 -4.71053421e-02 -2.89409727e-01 -1.40166476e-01
2.62129419e-02 1.48633265e+00 -1.21659482e+00 -2.11133614e-01
-3.05708259e-01 8.74323428e-01 1.52113169e-01 -4.16002795e-02
-4.59773213e-01 -2.01594681e-01 5.45701265e-01 5.78705847e-01
3.62088412e-01 4.94967401e-02 -3.18401515e-01 -1.37893331e+00
-2.03724075e-02 -7.51486599e-01 4.26259965e-01 -6.25513911e-01
-1.03752017e+00 9.24372971e-01 -6.64139390e-02 -1.30979788e+00
-1.77694842e-01 -4.83409107e-01 -7.43736923e-01 6.29612625e-01
-1.14041173e+00 -7.17444003e-01 -4.64849889e-01 4.29528505e-01
4.39835221e-01 -7.00136274e-02 8.25191617e-01 5.74898958e-01
-7.47082770e-01 5.50343931e-01 -1.37595788e-01 -1.88182950e-01
7.95763433e-01 -1.57152534e+00 2.71517456e-01 8.36646020e-01
2.96688750e-02 9.49922800e-01 7.97598183e-01 -4.78404760e-01
-7.88955808e-01 -6.61355257e-01 1.03095663e+00 -7.02642560e-01
6.62437141e-01 -5.35597682e-01 -7.13524699e-01 5.63175142e-01
2.47633597e-03 1.47206664e-01 9.40236151e-01 3.74100149e-01
3.51540118e-01 -4.50355709e-01 -1.04957473e+00 2.34731600e-01
8.50823283e-01 -5.37738621e-01 -8.90944600e-01 -3.62304971e-03
1.02990821e-01 -4.49784040e-01 -8.81988883e-01 2.53400505e-01
5.92250109e-01 -1.39994979e+00 4.21581715e-01 -2.43279263e-01
1.58473670e-01 -2.57182837e-01 3.52035582e-01 -1.22267747e+00
7.37054944e-02 -8.76402795e-01 1.85754821e-01 9.95898604e-01
3.91851813e-01 -7.74648070e-01 1.06006575e+00 7.03738570e-01
-4.63063776e-01 -8.22206855e-01 -1.05557668e+00 -6.43188715e-01
-3.29037040e-01 -3.69907707e-01 5.02375722e-01 4.10446674e-01
5.82402293e-03 3.97619724e-01 -1.55560240e-01 -3.56908031e-02
9.56856534e-02 -9.33268387e-03 6.66749775e-01 -1.54903615e+00
-2.14957267e-01 -3.70116204e-01 -8.06513906e-01 -3.57318431e-01
3.17328200e-02 -9.47659731e-01 -9.19453278e-02 -1.63512933e+00
1.46905512e-01 -4.05229688e-01 -5.47432840e-01 6.46779418e-01
-2.34681234e-01 7.34471381e-01 2.35366747e-01 1.00627542e-01
-6.36561871e-01 1.96618274e-01 1.01590455e+00 1.33673385e-01
-5.49928367e-01 4.14985836e-01 -6.07506812e-01 7.50543594e-01
8.14986348e-01 -7.43410826e-01 -5.00193477e-01 1.52382061e-01
3.72066259e-01 -1.88244432e-01 1.95447668e-01 -1.43247914e+00
6.36058897e-02 5.25485873e-01 3.42232287e-01 -4.23452497e-01
6.68803692e-01 -7.74856031e-01 2.47858569e-01 5.67594171e-01
2.11701766e-01 1.63276687e-01 3.18529338e-01 2.26247489e-01
-7.07967160e-03 -3.70921224e-01 6.75693989e-01 -2.30326846e-01
-3.46733898e-01 1.54383913e-01 -4.44128335e-01 2.48971507e-01
4.11597162e-01 -5.00232458e-01 -2.29642481e-01 -1.97354391e-01
-1.32371986e+00 -1.58564243e-02 9.16060150e-01 1.04853876e-01
7.19432980e-02 -6.06489778e-01 -1.12361506e-01 5.93507051e-01
1.24985933e-01 -1.22737899e-01 3.25231791e-01 1.56320834e+00
-6.15393043e-01 4.26038235e-01 -3.95036936e-01 -7.92812586e-01
-1.33430791e+00 5.07439852e-01 5.45269489e-01 -4.85046118e-01
-7.15136170e-01 5.62558234e-01 8.55609104e-02 -3.23674493e-02
-3.98718677e-02 -7.47505903e-01 -2.74039686e-01 3.56560111e-01
4.45368022e-01 1.99501082e-01 7.15129197e-01 -4.29385185e-01
-6.81244552e-01 3.93447161e-01 -1.99320167e-02 4.66357954e-02
1.30146658e+00 -1.51532918e-01 -4.43897545e-02 5.59058130e-01
1.01557398e+00 3.83408725e-01 -9.86369789e-01 5.02071559e-01
-5.67527227e-02 -2.77536482e-01 1.63220819e-02 -8.67067575e-01
-1.13447428e+00 8.82335007e-01 9.22886670e-01 3.88797700e-01
1.31146657e+00 -3.80202420e-02 7.66936183e-01 1.03986830e-01
2.70384699e-01 -7.57717013e-01 -3.47927719e-01 9.05818269e-02
4.09164429e-01 -1.20554543e+00 9.33666006e-02 5.14748394e-02
-4.92297173e-01 9.70043242e-01 2.89651841e-01 -3.00052315e-02
6.44905925e-01 -1.73081130e-01 2.65135676e-01 -4.80848819e-01
-4.56256449e-01 -4.01343167e-01 4.75036561e-01 4.57338572e-01
2.46990964e-01 -5.09036295e-02 -5.81564963e-01 8.79737973e-01
-6.79508984e-01 3.68635535e-01 5.46981931e-01 6.35245383e-01
-1.75060555e-01 -1.34100795e+00 -2.82914490e-02 6.96267605e-01
-9.13535058e-01 -4.95024845e-02 -2.15216562e-01 1.00428391e+00
3.92049044e-01 9.79720771e-01 1.10416636e-01 -4.16208655e-01
2.80182034e-01 2.83063799e-01 4.04877126e-01 -1.03575230e+00
-9.63045359e-01 7.31418654e-02 1.68446720e-01 -7.08340168e-01
-4.10520345e-01 -6.64551258e-01 -9.50911641e-01 9.72572789e-02
-7.09462911e-02 3.61074597e-01 6.13530576e-01 1.17758083e+00
4.71736997e-01 5.39819241e-01 9.29226130e-02 -9.10555959e-01
-9.49137006e-03 -1.02161205e+00 -8.35392594e-01 3.85989815e-01
6.59985602e-01 -7.11180091e-01 -4.83038515e-01 -2.65686184e-01]
|
[13.500575065612793, 3.5295426845550537]
|
875fadbc-6bf0-4662-8e34-155eb1a3b718
|
gazenerf-3d-aware-gaze-redirection-with
|
2212.04823
| null |
https://arxiv.org/abs/2212.04823v2
|
https://arxiv.org/pdf/2212.04823v2.pdf
|
GazeNeRF: 3D-Aware Gaze Redirection with Neural Radiance Fields
|
We propose GazeNeRF, a 3D-aware method for the task of gaze redirection. Existing gaze redirection methods operate on 2D images and struggle to generate 3D consistent results. Instead, we build on the intuition that the face region and eyeballs are separate 3D structures that move in a coordinated yet independent fashion. Our method leverages recent advancements in conditional image-based neural radiance fields and proposes a two-stream architecture that predicts volumetric features for the face and eye regions separately. Rigidly transforming the eye features via a 3D rotation matrix provides fine-grained control over the desired gaze angle. The final, redirected image is then attained via differentiable volume compositing. Our experiments show that this architecture outperforms naively conditioned NeRF baselines as well as previous state-of-the-art 2D gaze redirection methods in terms of redirection accuracy and identity preservation.
|
['Otmar Hilliges', 'Xucong Zhang', 'Hyung Jin Chang', 'Shalini De Mello', 'Gengyan Li', 'Xi Wang', 'Xiangwei Shi', 'Alessandro Ruzzi']
|
2022-12-08
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Ruzzi_GazeNeRF_3D-Aware_Gaze_Redirection_With_Neural_Radiance_Fields_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Ruzzi_GazeNeRF_3D-Aware_Gaze_Redirection_With_Neural_Radiance_Fields_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['gaze-redirection']
|
['computer-vision']
|
[ 1.96235567e-01 1.30152375e-01 -7.09516704e-02 -6.78325236e-01
-3.08903515e-01 -7.54720926e-01 9.83421504e-01 -5.12314677e-01
-2.36741841e-01 2.11015150e-01 5.78281939e-01 -3.10967594e-01
2.11151615e-01 -2.61688590e-01 -6.25571847e-01 -6.30358338e-01
1.69668645e-01 1.42460719e-01 -2.68302828e-01 -2.33969495e-01
7.92227745e-01 6.75197303e-01 -1.97850525e+00 -2.05685589e-02
6.07953906e-01 9.28444922e-01 -2.09391579e-01 7.48931885e-01
-1.65125057e-02 5.75151622e-01 -8.73463377e-02 -1.51150376e-01
3.73605609e-01 -4.61640507e-01 -9.30211008e-01 -1.83613136e-01
1.45886564e+00 -3.63141656e-01 -1.72276303e-01 7.80207813e-01
5.58281422e-01 4.07188118e-01 1.11612248e+00 -1.11016715e+00
-1.34102285e+00 -1.38021618e-01 -1.08752012e+00 5.11256516e-01
6.01068437e-01 2.29807019e-01 1.09802961e+00 -1.01373792e+00
9.30834472e-01 1.37588751e+00 5.23448765e-01 1.00729322e+00
-1.71686006e+00 -7.19857216e-01 4.60125893e-01 -1.34548113e-01
-1.28534293e+00 -9.12374139e-01 6.76282525e-01 -6.24358952e-01
1.16702712e+00 3.09681743e-01 5.19989371e-01 1.12934518e+00
6.99731559e-02 6.00464225e-01 1.40435100e+00 -4.77603197e-01
-2.65194178e-01 -3.16077203e-01 -5.67018650e-02 7.80171156e-01
-2.18809381e-01 1.46210849e-01 -1.05248070e+00 2.97834784e-01
7.21053898e-01 -2.84848154e-01 -6.17755473e-01 -5.43377340e-01
-1.11289644e+00 4.46236104e-01 8.54781926e-01 -1.33371204e-01
-1.11517809e-01 1.34108871e-01 -2.22943276e-01 3.24050970e-02
8.12014997e-01 5.98726213e-01 -2.81688333e-01 1.56597778e-01
-1.01578689e+00 4.61432219e-01 2.39365771e-01 8.96284282e-01
7.43499339e-01 -1.00999221e-01 -4.82980996e-01 3.74219626e-01
6.80795133e-01 7.30293095e-01 2.01121688e-01 -9.23768580e-01
2.28656456e-01 5.43339312e-01 1.43471137e-01 -6.62258446e-01
-6.42377675e-01 1.16234772e-01 -1.60285637e-01 6.55125439e-01
6.10611975e-01 -2.78904252e-02 -1.31279862e+00 2.10888696e+00
6.20207071e-01 1.11849315e-01 -4.54065591e-01 1.05419528e+00
9.07396972e-01 1.38935074e-01 1.71137139e-01 3.36270422e-01
1.47902668e+00 -7.46380210e-01 -3.19638431e-01 -2.39211187e-01
1.98566318e-01 -7.18384266e-01 1.35826313e+00 -6.89194649e-02
-1.22975838e+00 -3.23033810e-01 -7.37961233e-01 -9.18412387e-01
-3.26795995e-01 -1.64662629e-01 7.47700632e-01 7.78927863e-01
-1.64042974e+00 3.05805057e-01 -6.71029389e-01 -3.94115090e-01
7.78847873e-01 4.17745471e-01 -6.10434651e-01 3.59977543e-01
-6.95306420e-01 1.10465157e+00 -5.55821955e-01 1.89126581e-01
-6.99907601e-01 -1.28765535e+00 -1.01090646e+00 -3.76519859e-02
-2.59816021e-01 -9.46053028e-01 1.32255757e+00 -8.59566629e-01
-1.52201605e+00 1.80188501e+00 -7.42043138e-01 9.40862522e-02
2.13398337e-01 -2.10277081e-01 -8.44275057e-02 7.25250021e-02
3.92769128e-02 1.17948067e+00 1.30772805e+00 -1.06347060e+00
-5.23303449e-01 -6.85229301e-01 1.56654984e-01 7.33963013e-01
-1.97124586e-01 8.77529234e-02 -3.10160130e-01 -3.25396657e-01
-1.12919934e-01 -9.20870304e-01 5.28477132e-01 3.40655267e-01
-6.81267381e-01 -1.94799721e-01 5.70955932e-01 -3.84362698e-01
9.38428402e-01 -2.03648734e+00 4.63676095e-01 1.07332937e-01
7.00063825e-01 -4.16480899e-01 -1.48559958e-01 -2.88672239e-01
-5.60007930e-01 2.89302021e-01 1.81691021e-01 -7.48101234e-01
1.69457018e-01 -7.68520296e-01 -3.23847026e-01 7.67225266e-01
4.14080441e-01 1.12438571e+00 -7.79846907e-01 -3.81239682e-01
8.39145109e-02 8.39291811e-01 -7.19706595e-01 6.62719011e-02
7.57044600e-03 6.47509277e-01 -1.57308936e-01 6.26773059e-01
8.76694143e-01 -3.37128103e-01 -7.07558468e-02 -2.54489094e-01
-4.28289056e-01 6.77778423e-01 -2.24536791e-01 1.89248180e+00
-2.62109786e-01 1.08376563e+00 -1.17095690e-02 -1.72972575e-01
6.25240684e-01 -1.32065669e-01 3.27073038e-01 -9.57930028e-01
3.69967729e-01 -3.67583573e-01 -3.43422294e-01 -2.64490515e-01
5.77473819e-01 -1.43319905e-01 1.07489787e-01 9.16976035e-01
8.86629447e-02 -1.98359549e-01 -3.01628947e-01 1.73303708e-02
5.66436648e-01 9.44300115e-01 -1.20792814e-01 -5.78510404e-01
2.27186024e-01 -3.29947501e-01 5.12152091e-02 2.46635437e-01
-4.39227223e-01 9.84639108e-01 6.35507047e-01 -1.63411662e-01
-9.08151805e-01 -1.21803379e+00 -2.70475507e-01 1.43670261e+00
3.69035989e-01 -1.62804395e-01 -1.05434549e+00 -7.07152843e-01
6.02651648e-02 7.52396643e-01 -1.38384318e+00 -9.47220847e-02
-4.58754122e-01 -5.49564481e-01 4.27749991e-01 2.15023831e-01
8.77077356e-02 -7.60774493e-01 -7.08268285e-01 -7.47943342e-01
-1.17692938e-02 -6.45827591e-01 -9.88029599e-01 -1.46028593e-01
-4.33074206e-01 -1.30580509e+00 -8.26674819e-01 -7.49562860e-01
8.97871375e-01 5.86773098e-01 1.35616553e+00 1.09318338e-01
-2.80932128e-01 4.89629149e-01 1.18776679e-01 -3.85011703e-01
3.87621559e-02 2.07523406e-01 2.24044845e-01 3.80469412e-02
8.55425537e-01 -5.51920712e-01 -1.01622427e+00 1.26910150e-01
-3.51134509e-01 1.47619203e-01 -1.62976697e-01 4.01827514e-01
4.82608616e-01 -1.05816436e+00 1.13134518e-01 -8.93473685e-01
7.56415308e-01 -1.93953618e-01 -7.30278611e-01 1.99847579e-01
-6.68341577e-01 3.57267171e-01 1.27193227e-01 -2.17130095e-01
-1.37440646e+00 -1.41008079e-01 1.63915798e-01 -5.46466708e-01
-3.32059860e-01 -3.97121280e-01 8.19897801e-02 -2.54427463e-01
1.21176779e+00 -1.53291523e-01 8.08398798e-02 -3.03355247e-01
8.26322913e-01 4.65162039e-01 5.25479615e-01 -3.16244245e-01
9.30982411e-01 6.98016107e-01 1.95423841e-01 -5.43741524e-01
-1.03025281e+00 -6.50149807e-02 -9.39290047e-01 -3.40109289e-01
1.16335380e+00 -9.04623985e-01 -1.32967627e+00 6.22057438e-01
-9.86568630e-01 -5.36839485e-01 -1.30647644e-01 1.73389390e-01
-5.51765621e-01 -8.72481242e-02 -6.68121204e-02 -5.14452636e-01
-4.40480649e-01 -1.02007055e+00 1.61645722e+00 5.68239927e-01
-4.38895136e-01 -9.41293418e-01 2.26191297e-01 1.78684652e-01
4.19689119e-01 3.08813900e-01 8.28385472e-01 1.94712598e-02
-5.62194467e-01 2.88292438e-01 -3.48194033e-01 -2.54543662e-01
1.49234459e-01 2.48678431e-01 -1.57748163e+00 -1.14844693e-02
-3.90403450e-01 -5.21371901e-01 8.87639403e-01 6.44699037e-01
9.40583825e-01 2.43605658e-01 -6.11729801e-01 1.16344023e+00
8.35709631e-01 -4.94944692e-01 5.20291090e-01 1.43845573e-01
9.17351305e-01 7.67351985e-01 2.18100518e-01 9.91715956e-03
7.22344220e-01 4.16801482e-01 5.42530954e-01 -1.95916772e-01
-5.18852711e-01 -3.76298070e-01 1.62172735e-01 -1.56230927e-01
-3.21767151e-01 -7.47607574e-02 -7.99283266e-01 4.42964077e-01
-1.26751351e+00 -1.09250557e+00 -1.84148848e-02 2.19821000e+00
9.30108130e-01 -1.67358279e-01 5.87589256e-02 -5.36143541e-01
7.04212427e-01 3.89119565e-01 -9.95909512e-01 -3.61209571e-01
-2.34168284e-02 2.98502982e-01 4.35899645e-01 7.51959860e-01
-1.13309109e+00 1.19240677e+00 7.65650320e+00 -3.53515632e-02
-1.52523720e+00 -1.63197726e-01 5.65072775e-01 -8.19919646e-01
-8.60483229e-01 -2.31967241e-01 -1.03641605e+00 1.26593158e-01
6.97458684e-01 7.56356418e-02 7.95769274e-01 3.51295441e-01
9.22502056e-02 3.10628377e-02 -1.37915611e+00 1.18233621e+00
5.20550728e-01 -1.42163384e+00 -2.72438914e-01 3.08782548e-01
6.23127759e-01 4.58660185e-01 7.51793861e-01 -1.59979165e-01
1.79843694e-01 -1.43971610e+00 9.48936164e-01 8.78164768e-01
1.47661352e+00 -3.49248827e-01 -4.49408740e-01 -2.70352572e-01
-8.67425382e-01 2.81175256e-01 3.15479077e-02 4.81699966e-02
7.64400437e-02 -2.20581278e-01 -8.02240074e-01 -1.11125872e-01
1.04482138e+00 7.11585045e-01 -7.66605496e-01 6.76673710e-01
-4.38436925e-01 1.37842759e-01 -4.14641291e-01 -7.97781162e-03
-9.82502103e-02 7.71521498e-03 5.54232180e-01 7.64810264e-01
2.83801239e-02 -9.21830460e-02 -8.79737496e-01 1.28166270e+00
-5.25582790e-01 -1.26638398e-01 -6.45365715e-01 3.51088226e-01
4.38558370e-01 1.30389917e+00 -4.48877603e-01 2.62291789e-01
-4.02900040e-01 1.06094408e+00 7.46800959e-01 6.62497520e-01
-6.70838892e-01 -3.55634868e-01 1.31414378e+00 3.14413577e-01
4.94334012e-01 -8.61992836e-02 -4.17823881e-01 -1.44162881e+00
-1.49363264e-01 -3.77836287e-01 -1.17624234e-02 -1.41380596e+00
-1.20718801e+00 7.55034566e-01 -1.76095501e-01 -9.01473939e-01
-2.52492964e-01 -9.77069318e-01 -5.34662783e-01 1.36484528e+00
-1.97718048e+00 -1.49138236e+00 -6.23154461e-01 9.32439744e-01
5.25977574e-02 -2.26087891e-03 1.07579160e+00 -1.17913015e-01
-4.34013039e-01 9.09975946e-01 -1.61647290e-01 -7.17572719e-02
1.14454591e+00 -1.44931901e+00 9.32743192e-01 8.31095099e-01
2.46035680e-01 9.56400871e-01 5.10086000e-01 -2.67120242e-01
-1.32816362e+00 -6.09709084e-01 9.65062141e-01 -1.32995570e+00
2.45154962e-01 -8.69122148e-01 -5.29234290e-01 9.65651751e-01
3.75272512e-01 9.61209908e-02 7.70117998e-01 7.13482141e-01
-1.01439142e+00 8.59777182e-02 -1.12304056e+00 9.36835527e-01
1.41759491e+00 -9.98032212e-01 -6.19234681e-01 2.90453553e-01
4.52869385e-01 -6.62534356e-01 -5.55862606e-01 -1.48198590e-01
7.58933425e-01 -1.18216562e+00 1.02133000e+00 -1.03660059e+00
3.45646471e-01 -2.98488110e-01 1.51113153e-01 -1.23079562e+00
-4.17476833e-01 -9.39681947e-01 -1.16366699e-01 9.13825512e-01
3.95871878e-01 -3.78887534e-01 7.85053670e-01 9.76862490e-01
2.12120801e-01 -3.98736238e-01 -7.18903065e-01 1.58329278e-01
2.06226245e-01 -2.11965173e-01 8.73020053e-01 8.48389566e-01
-3.82438488e-02 8.11362863e-01 -9.53414217e-02 -4.24914323e-02
6.62950575e-01 3.65794301e-01 8.29689980e-01 -1.33409095e+00
3.05605412e-01 -7.29518890e-01 -1.74608141e-01 -1.26187527e+00
6.47847950e-01 -1.02275383e+00 -1.22630864e-01 -1.11651194e+00
-4.59420867e-03 -3.56215000e-01 -1.31179899e-01 6.63375914e-01
-2.47247711e-01 6.85209870e-01 5.34763373e-02 1.68005511e-01
-2.88762242e-01 4.18722540e-01 1.41025078e+00 1.62283346e-01
-2.42765307e-01 -2.72252887e-01 -1.29784453e+00 5.20983338e-01
4.79506940e-01 -3.88907790e-02 -6.22545421e-01 -9.06523168e-01
4.27456349e-01 -6.25065148e-01 4.67238873e-01 -1.82709530e-01
4.88132983e-02 -1.95884660e-01 7.52391577e-01 -4.12108958e-01
5.46848536e-01 -2.78659910e-01 -5.67337513e-01 -4.45616633e-01
-4.35547382e-01 -1.11125121e-02 1.98566750e-01 3.99884492e-01
2.85750359e-01 2.78116137e-01 9.33522820e-01 8.44585150e-02
-6.02136195e-01 6.35264933e-01 1.16377711e-01 3.68701458e-01
7.82515824e-01 -4.39859271e-01 -5.21148026e-01 -3.53049159e-01
-4.82149899e-01 1.22931011e-01 1.08263373e+00 8.00191164e-01
4.65393841e-01 -1.08883619e+00 -4.96726692e-01 5.34412622e-01
2.34365851e-01 6.06996231e-02 2.43198767e-01 7.95149088e-01
-4.93121594e-01 3.15321565e-01 -4.73845363e-01 -7.36250579e-01
-1.16960704e+00 4.94372219e-01 8.28431308e-01 6.07458770e-01
-5.33481121e-01 1.41457093e+00 6.15328789e-01 -5.41329622e-01
8.84071961e-02 -2.37262234e-01 -4.70030487e-01 1.46459192e-01
7.70421565e-01 -7.29206055e-02 5.23551628e-02 -1.06335759e+00
-6.53866410e-01 1.05533612e+00 -1.29681438e-01 -2.41896555e-01
1.28760433e+00 -5.59207737e-01 -8.56644288e-02 1.53588369e-01
1.26909101e+00 3.48415762e-01 -1.63911974e+00 4.72585075e-02
-5.45309722e-01 -6.89235270e-01 2.53496110e-01 -6.88607454e-01
-9.60033536e-01 9.73979950e-01 5.73165715e-01 6.24850914e-02
1.08574307e+00 1.53999001e-01 3.12739670e-01 -2.27221176e-02
-1.94451615e-01 -7.12511957e-01 -2.96844333e-01 4.39973742e-01
7.57945597e-01 -1.21471035e+00 -4.31447662e-03 -6.30250648e-02
-6.92993164e-01 7.87949204e-01 8.36353183e-01 -1.00111201e-01
8.48102033e-01 8.48771166e-03 2.81626523e-01 -4.53805208e-01
-6.08923018e-01 -2.44631693e-01 8.61662388e-01 9.39644873e-01
8.31391633e-01 -2.89569885e-01 4.47362304e-01 -4.60544005e-02
-5.78002989e-01 -2.57162482e-01 1.28761828e-01 5.19916534e-01
-1.35220304e-01 -6.97450459e-01 -2.66797215e-01 3.28067005e-01
-2.79299498e-01 -4.62335378e-01 -5.29985487e-01 7.91033447e-01
-1.44631311e-01 5.03845155e-01 6.34483695e-01 -2.22242340e-01
2.51916170e-01 2.62585700e-01 1.13817334e+00 -6.47763729e-01
-3.40041071e-01 -1.05859078e-01 -2.29724348e-01 -8.48644674e-01
-5.80191851e-01 -9.18138683e-01 -1.24705839e+00 -5.11004090e-01
-2.05543876e-01 -5.42824090e-01 6.65395141e-01 6.70249999e-01
7.98876286e-01 2.12283432e-01 5.48452675e-01 -1.34403014e+00
-1.54359370e-01 -9.05992985e-01 -4.75498348e-01 6.08848691e-01
8.63061845e-01 -9.75517571e-01 -3.13786387e-01 1.84168145e-01]
|
[14.11436653137207, 0.045289695262908936]
|
25057f50-335d-4981-a07e-bce80081fa3a
|
high-for-low-and-low-for-high-efficient
|
1504.06201
| null |
http://arxiv.org/abs/1504.06201v3
|
http://arxiv.org/pdf/1504.06201v3.pdf
|
High-for-Low and Low-for-High: Efficient Boundary Detection from Deep Object Features and its Applications to High-Level Vision
|
Most of the current boundary detection systems rely exclusively on low-level
features, such as color and texture. However, perception studies suggest that
humans employ object-level reasoning when judging if a particular pixel is a
boundary. Inspired by this observation, in this work we show how to predict
boundaries by exploiting object-level features from a pretrained
object-classification network. Our method can be viewed as a "High-for-Low"
approach where high-level object features inform the low-level boundary
detection process. Our model achieves state-of-the-art performance on an
established boundary detection benchmark and it is efficient to run.
Additionally, we show that due to the semantic nature of our boundaries we
can use them to aid a number of high-level vision tasks. We demonstrate that
using our boundaries we improve the performance of state-of-the-art methods on
the problems of semantic boundary labeling, semantic segmentation and object
proposal generation. We can view this process as a "Low-for-High" scheme, where
low-level boundaries aid high-level vision tasks.
Thus, our contributions include a boundary detection system that is accurate,
efficient, generalizes well to multiple datasets, and is also shown to improve
existing state-of-the-art high-level vision methods on three distinct tasks.
|
['Gedas Bertasius', 'Lorenzo Torresani', 'Jianbo Shi']
|
2015-04-23
|
high-for-low-and-low-for-high-efficient-1
|
http://openaccess.thecvf.com/content_iccv_2015/html/Bertasius_High-for-Low_and_Low-for-High_ICCV_2015_paper.html
|
http://openaccess.thecvf.com/content_iccv_2015/papers/Bertasius_High-for-Low_and_Low-for-High_ICCV_2015_paper.pdf
|
iccv-2015-12
|
['object-proposal-generation']
|
['computer-vision']
|
[ 6.02245331e-01 2.80936778e-01 -1.84040859e-01 -4.65378284e-01
-7.56906271e-01 -5.25271893e-01 6.13629282e-01 3.32906991e-01
-4.75100249e-01 2.23809555e-01 -3.79199147e-01 -2.81759590e-01
2.59923846e-01 -9.86963034e-01 -9.57388699e-01 -3.31540883e-01
1.02894373e-01 7.85319865e-01 9.56070364e-01 -1.51095092e-01
5.55335999e-01 5.33058405e-01 -1.93405879e+00 6.26882792e-01
7.17111170e-01 1.36984742e+00 6.35069758e-02 7.10475683e-01
-3.32633615e-01 2.44774446e-01 -2.93161720e-01 9.88205746e-02
4.88682926e-01 -3.74048710e-01 -1.28841877e+00 1.75447196e-01
1.05209291e+00 -8.35514590e-02 4.14125293e-01 1.21813679e+00
-9.23731029e-02 9.90062952e-03 7.45104492e-01 -1.19949090e+00
-5.77139199e-01 1.66933313e-01 -6.45981848e-01 7.48730153e-02
3.52022380e-01 1.06863320e-01 1.24528968e+00 -9.37048197e-01
8.84881258e-01 1.17583656e+00 8.00774157e-01 5.68264961e-01
-1.44464850e+00 -1.22812249e-01 5.24149656e-01 1.84365779e-01
-1.20898807e+00 -1.63958535e-01 7.49189138e-01 -8.04626584e-01
8.84526789e-01 -7.88777228e-03 8.64663720e-01 4.33950871e-01
-1.29421532e-01 8.82686079e-01 1.51212180e+00 -6.45035148e-01
4.31493402e-01 -1.87932432e-01 5.33180118e-01 1.08495378e+00
4.74898845e-01 3.55349854e-02 -3.17237914e-01 1.05506495e-01
9.04021382e-01 -3.88450593e-01 -2.07292557e-01 -4.71724212e-01
-1.22633314e+00 6.44268870e-01 6.62063360e-01 3.86965603e-01
-2.78139025e-01 2.54781067e-01 4.39262949e-02 -1.41150817e-01
4.20835644e-01 4.16486263e-01 -3.67294014e-01 1.76848993e-01
-1.15210807e+00 2.93904483e-01 9.69832957e-01 6.91243887e-01
1.09613073e+00 -5.10893881e-01 -2.96247393e-01 6.25581443e-01
4.67605084e-01 3.90515067e-02 1.55618235e-01 -1.26535964e+00
3.65425982e-02 8.17389965e-01 2.15966478e-01 -5.56601822e-01
-6.87833011e-01 -1.25482902e-02 -4.17655498e-01 8.94189835e-01
8.04072022e-01 3.02468002e-01 -1.56569827e+00 1.54586470e+00
4.70127940e-01 1.29183233e-01 -1.78880423e-01 1.03624868e+00
9.25234556e-01 3.52723688e-01 9.37938690e-03 3.05066943e-01
1.76934123e+00 -1.15808892e+00 -3.39412093e-01 -4.72487777e-01
4.50248361e-01 -6.90362334e-01 1.16555595e+00 6.00351691e-01
-1.22033370e+00 -6.61279857e-01 -1.01724994e+00 -3.47178459e-01
-6.59357488e-01 1.88721627e-01 7.79056370e-01 5.07393599e-01
-1.47306454e+00 6.28025234e-01 -6.32778049e-01 -6.83514893e-01
7.27633953e-01 3.29628527e-01 -2.91717023e-01 8.23327005e-02
-8.07324111e-01 8.76438916e-01 6.15231931e-01 9.97474864e-02
-8.51652324e-01 -4.61924762e-01 -8.52432549e-01 -6.69995993e-02
5.47947645e-01 -9.88554895e-01 1.36765063e+00 -1.02237988e+00
-1.16406178e+00 1.50592041e+00 -1.23308115e-01 -3.74866128e-01
5.95479488e-01 -4.89526279e-02 2.20065013e-01 6.19243562e-01
3.11046064e-01 1.43002307e+00 7.89818645e-01 -1.63526881e+00
-9.78236616e-01 -2.71245033e-01 1.87512606e-01 -5.92254801e-03
2.54020274e-01 -2.67756730e-01 -5.99338055e-01 -3.64481986e-01
5.50626457e-01 -9.68181431e-01 -2.45705634e-01 4.99311924e-01
-3.86502028e-01 -6.52213633e-01 7.17337310e-01 -4.25722122e-01
6.55539393e-01 -1.93722594e+00 -2.00286746e-01 1.57210752e-01
2.16763645e-01 3.30435932e-01 5.61062433e-02 -9.14215520e-02
2.27548867e-01 5.62520683e-01 -5.76708436e-01 -3.64668757e-01
-2.45839474e-03 1.20184511e-01 6.12926111e-02 3.53818834e-01
4.26863879e-01 8.49149466e-01 -6.94804072e-01 -7.82148600e-01
3.28035384e-01 1.52037114e-01 -5.60802162e-01 1.51504800e-01
-5.87327838e-01 2.43379295e-01 -2.63799369e-01 6.65841877e-01
5.66506267e-01 -4.00619626e-01 -2.57132947e-01 -3.13166916e-01
-2.47433379e-01 1.18728928e-01 -9.85084951e-01 1.72230303e+00
-1.28633782e-01 6.37237310e-01 2.65329361e-01 -1.06391048e+00
7.47750700e-01 4.51039746e-02 2.78050601e-01 -5.14382780e-01
2.53589660e-01 3.48714441e-01 -1.05359079e-02 -3.77001852e-01
3.81139100e-01 -8.65855778e-04 -7.92254880e-03 2.33103752e-01
-1.45425677e-01 -5.54013550e-01 5.02834558e-01 3.69898081e-02
9.12245274e-01 5.99651277e-01 1.89656794e-01 -3.48623127e-01
4.71146911e-01 3.49760801e-01 5.21646976e-01 1.08559322e+00
-5.44832587e-01 9.48205888e-01 5.79763055e-01 -4.43398327e-01
-1.06625402e+00 -1.01396167e+00 -3.03488851e-01 1.05730653e+00
6.42691910e-01 -4.03357595e-02 -1.34122074e+00 -6.47880852e-01
5.67943305e-02 3.89072806e-01 -1.01874888e+00 2.18425050e-01
-4.87220049e-01 -4.46987480e-01 3.38234335e-01 8.36914301e-01
7.37989187e-01 -1.28437459e+00 -8.61204863e-01 3.31836790e-02
8.83320114e-04 -1.39807773e+00 -2.35871404e-01 1.02821521e-01
-8.61888230e-01 -1.13854384e+00 -5.75420558e-01 -9.27662849e-01
8.26250911e-01 2.74472803e-01 1.20690906e+00 2.63990730e-01
-6.91116810e-01 4.72375482e-01 -3.27688634e-01 -1.95992395e-01
-3.16997081e-01 -9.99917090e-02 -4.00936902e-01 -3.61601524e-02
2.99987793e-01 -2.17948541e-01 -7.67453134e-01 2.15072915e-01
-8.58232498e-01 3.68972659e-01 5.50517738e-01 7.67000794e-01
8.56798947e-01 -2.77985424e-01 3.16336572e-01 -8.90544593e-01
2.55596697e-01 8.91644508e-02 -6.81587338e-01 3.43717933e-01
-6.49800420e-01 9.09843892e-02 2.82801270e-01 -6.56245574e-02
-9.89031732e-01 2.12157890e-01 -1.83124274e-01 -9.41466317e-02
-6.28380775e-01 -1.84083115e-02 5.88557683e-02 -3.66581827e-01
4.66593146e-01 -2.12471277e-01 -1.24064825e-01 -3.94982845e-01
6.69517815e-01 5.99548638e-01 5.26682794e-01 -6.67824686e-01
4.06515837e-01 9.63020504e-01 9.45444331e-02 -7.91115820e-01
-1.27062154e+00 -9.34288740e-01 -8.70352805e-01 -2.96802998e-01
1.36220384e+00 -4.95276272e-01 -9.37887490e-01 7.38441467e-01
-1.21895909e+00 -7.00325906e-01 -3.90191257e-01 1.31163940e-01
-1.08328402e+00 2.25683019e-01 -6.56411946e-01 -7.27541924e-01
-2.36825168e-01 -1.12584877e+00 1.42880118e+00 6.00020707e-01
-2.25280523e-02 -9.95841980e-01 -2.55368114e-01 7.70796061e-01
2.68067092e-01 2.09133223e-01 9.36720490e-01 -5.58920145e-01
-8.90821338e-01 2.46126190e-01 -7.42671430e-01 2.75676757e-01
-2.06622303e-01 5.96836247e-02 -1.10506666e+00 8.87549371e-02
-3.43808651e-01 -4.59108770e-01 1.33908832e+00 5.14998198e-01
1.18042362e+00 1.47204965e-01 -5.71534216e-01 5.65038860e-01
1.52760375e+00 -7.69827068e-02 4.57722366e-01 5.08595765e-01
7.71491945e-01 8.44968498e-01 6.90751851e-01 6.05931990e-02
5.85241497e-01 6.87545419e-01 5.71491361e-01 -6.33999765e-01
-6.38282537e-01 3.88493910e-02 -3.49202156e-02 -3.05103473e-02
3.84522639e-02 8.10858086e-02 -1.18449819e+00 8.43165338e-01
-2.03534722e+00 -5.98464370e-01 -4.44266945e-01 1.86232829e+00
8.72857809e-01 5.87559223e-01 2.55379796e-01 7.90078044e-02
9.21303570e-01 -2.46792659e-02 -5.65887332e-01 -5.68043947e-01
9.08089802e-02 2.48276711e-01 4.57884192e-01 5.19683361e-01
-1.40569937e+00 1.40231180e+00 6.32985020e+00 5.12983322e-01
-8.52238238e-01 6.51905835e-02 8.15252304e-01 6.13388896e-01
-1.47222206e-01 2.54280001e-01 -1.00820005e+00 -7.04879537e-02
2.63713956e-01 4.03073430e-01 1.06759071e-01 8.62385213e-01
-8.99491608e-02 -7.43916750e-01 -1.48267066e+00 7.99814939e-01
2.75719106e-01 -1.47047925e+00 -3.00178844e-02 -2.95323506e-02
7.95137703e-01 1.00883670e-01 -1.71540424e-01 -4.12848219e-02
3.91763806e-01 -1.00431180e+00 9.02217329e-01 4.94961649e-01
5.63310504e-01 3.84911485e-02 5.16238570e-01 2.06937850e-01
-1.39066422e+00 1.39497727e-01 -2.35509887e-01 -7.36348182e-02
1.93858296e-01 5.85734487e-01 -8.01144183e-01 2.29561970e-01
7.59430528e-01 4.14438665e-01 -6.21581018e-01 1.25357187e+00
-3.73577684e-01 3.63319486e-01 -4.76360738e-01 2.13746607e-01
5.11176944e-01 -1.37819335e-01 7.12524280e-02 1.22866392e+00
-1.27608821e-01 1.44644797e-01 7.24721670e-01 1.20744658e+00
-4.70054708e-02 -4.77142520e-02 -3.85352939e-01 2.41643131e-01
1.86384007e-01 1.37522686e+00 -1.53910470e+00 -4.82748687e-01
-3.09034139e-01 8.81806552e-01 3.35478425e-01 2.71243632e-01
-4.57415611e-01 -3.61820251e-01 4.09923345e-01 1.33020908e-01
5.31688511e-01 -1.77644283e-01 -6.83918774e-01 -9.07484770e-01
6.87764436e-02 -3.04245055e-01 2.49254823e-01 -9.14045334e-01
-1.24051416e+00 3.28151792e-01 -9.38420296e-02 -7.43798971e-01
1.09318733e-01 -9.05655682e-01 -4.30660427e-01 6.21344566e-01
-1.95220006e+00 -1.38678682e+00 -5.94589531e-01 2.58989632e-01
6.05722427e-01 5.20456314e-01 6.26778662e-01 -2.27972031e-01
-1.10056698e-01 6.87955394e-02 -4.70052391e-01 5.21288037e-01
3.58410597e-01 -1.49639773e+00 7.47277379e-01 7.83053160e-01
2.38646865e-01 1.93977535e-01 4.38864589e-01 -5.16300082e-01
-8.44595134e-01 -7.96022832e-01 6.63281322e-01 -2.86277711e-01
5.37953436e-01 -4.43918914e-01 -9.13457334e-01 4.24065471e-01
-5.70396380e-03 5.31377256e-01 3.09635401e-01 3.38305831e-02
-4.22981560e-01 5.61998896e-02 -1.19803333e+00 4.97663975e-01
1.22658575e+00 -3.51025939e-01 -1.00976360e+00 2.43685335e-01
5.09693682e-01 -2.69296259e-01 -7.37874389e-01 5.97393811e-01
6.29329979e-01 -1.34116697e+00 9.05747890e-01 -3.66504788e-01
1.77880049e-01 -4.20129895e-01 8.97638649e-02 -8.19667161e-01
-1.17195830e-01 -2.29150042e-01 2.45608985e-01 1.04982328e+00
3.08761537e-01 -5.36271274e-01 8.68892550e-01 4.14563060e-01
-2.59823591e-01 -8.02418172e-01 -9.20133352e-01 -7.69911945e-01
5.52554950e-02 -3.67584080e-01 8.00383836e-02 5.83603024e-01
-3.02403390e-01 2.08127871e-01 3.74866754e-01 7.22531006e-02
6.77218437e-01 6.26049817e-01 5.47164500e-01 -1.80869782e+00
-2.99383439e-02 -8.23845863e-01 -5.81289172e-01 -1.24074709e+00
2.75852501e-01 -7.55321085e-01 5.76630175e-01 -2.23029494e+00
4.90384437e-02 -7.16718554e-01 -1.99934557e-01 7.22724736e-01
-8.47799927e-02 7.46172309e-01 3.57685447e-01 2.62948960e-01
-7.50125825e-01 5.92441335e-02 1.33016753e+00 -1.41939625e-01
-2.84305327e-02 -1.77561000e-01 -4.26715761e-01 1.20434761e+00
6.26618564e-01 -4.30143952e-01 -1.45309614e-02 -2.80368309e-02
2.69018888e-01 -1.80725262e-01 5.55558085e-01 -1.16134477e+00
2.01918229e-01 -1.61929727e-01 2.00959384e-01 -4.11745131e-01
4.13613617e-01 -5.85755050e-01 -6.15272284e-01 4.63018894e-01
-2.91916639e-01 -5.06753802e-01 1.17037632e-01 3.55199397e-01
-1.67098567e-01 -5.10429204e-01 8.91989768e-01 -5.24328470e-01
-1.20324421e+00 1.60334870e-01 -2.16696367e-01 3.64654124e-01
1.29174387e+00 -6.84806824e-01 -5.11627853e-01 1.13353454e-01
-8.45854163e-01 2.33547315e-01 8.44998658e-01 2.72179335e-01
5.28856397e-01 -7.77319789e-01 -5.17254651e-01 2.28358671e-01
2.78266758e-01 3.49863231e-01 -1.32733449e-01 8.37750494e-01
-8.17190945e-01 3.83031309e-01 -3.60298485e-01 -1.05135953e+00
-1.09068167e+00 3.63324314e-01 3.89631093e-01 2.39380896e-02
-7.01801121e-01 1.10737741e+00 5.56514323e-01 -1.77449331e-01
1.53505832e-01 -6.97214425e-01 -1.16196200e-01 1.69933841e-01
1.92887157e-01 6.90086856e-02 1.34800076e-02 -5.62007070e-01
-5.90148628e-01 1.30795431e+00 -3.30795050e-02 -2.92410195e-01
8.76029491e-01 -6.63925568e-03 -2.38779292e-01 6.72474086e-01
6.99574947e-01 -4.34437037e-01 -1.57012582e+00 7.06589371e-02
2.81429350e-01 -2.25687012e-01 5.67562021e-02 -9.25201178e-01
-6.76355898e-01 1.10986555e+00 5.12444019e-01 5.12671649e-01
1.10165846e+00 3.78069639e-01 6.90423071e-01 3.57909739e-01
5.45319200e-01 -1.39669275e+00 2.64993697e-01 4.51955259e-01
3.79104286e-01 -1.62452579e+00 -1.08721025e-01 -8.84508848e-01
-3.75252038e-01 1.02269316e+00 6.85472846e-01 -3.05369407e-01
5.59262574e-01 3.65399271e-01 1.88099638e-01 -2.59463489e-01
-5.09213686e-01 -9.06958044e-01 4.15902197e-01 7.34812140e-01
3.23622793e-01 7.16900304e-02 -4.05930012e-01 -1.06160855e-02
2.40168139e-01 2.56277621e-02 6.57586038e-01 8.39694142e-01
-9.72294211e-01 -1.22027063e+00 -4.37685192e-01 3.47443432e-01
-3.41684699e-01 -8.96677375e-02 -6.79685295e-01 8.37423623e-01
4.63059694e-01 8.53636682e-01 3.18471313e-01 1.16297245e-01
1.04306832e-01 1.37223586e-01 7.51703143e-01 -1.02907538e+00
-6.63251758e-01 -1.43406928e-01 8.48607346e-02 -7.91844606e-01
-7.70983100e-01 -4.98393267e-01 -1.77832150e+00 3.48209649e-01
-3.33898634e-01 -9.52235088e-02 8.21868896e-01 1.13504386e+00
2.15248168e-01 3.69703442e-01 3.27103361e-02 -1.10336590e+00
-1.84415013e-01 -7.57290006e-01 -4.19070125e-01 7.40194082e-01
2.54286468e-01 -8.00439835e-01 -2.87726939e-01 3.64201665e-01]
|
[9.497322082519531, 0.28920382261276245]
|
65ff6ac3-ca0f-4490-bdf6-281876fce61a
|
target-aware-generative-augmentations-for
|
2305.13284
| null |
https://arxiv.org/abs/2305.13284v1
|
https://arxiv.org/pdf/2305.13284v1.pdf
|
Target-Aware Generative Augmentations for Single-Shot Adaptation
|
In this paper, we address the problem of adapting models from a source domain to a target domain, a task that has become increasingly important due to the brittle generalization of deep neural networks. While several test-time adaptation techniques have emerged, they typically rely on synthetic toolbox data augmentations in cases of limited target data availability. We consider the challenging setting of single-shot adaptation and explore the design of augmentation strategies. We argue that augmentations utilized by existing methods are insufficient to handle large distribution shifts, and hence propose a new approach SiSTA, which first fine-tunes a generative model from the source domain using a single-shot target, and then employs novel sampling strategies for curating synthetic target data. Using experiments on a variety of benchmarks, distribution shifts and image corruptions, we find that SiSTA produces significantly improved generalization over existing baselines in face attribute detection and multi-class object recognition. Furthermore, SiSTA performs competitively to models obtained by training on larger target datasets. Our codes can be accessed at https://github.com/Rakshith-2905/SiSTA.
|
['Jayaraman J. Thiagarajan', 'Pavan Turaga', 'Rakshith Subramanyam', 'Kowshik Thopalli']
|
2023-05-22
| null | null | null | null |
['object-recognition']
|
['computer-vision']
|
[ 4.45565850e-01 -1.62991494e-01 -9.98119563e-02 -5.34209609e-01
-1.03711808e+00 -5.92671335e-01 6.79325461e-01 -3.82165223e-01
-2.24937648e-01 7.85552502e-01 -4.52164635e-02 -2.16922805e-01
1.94311738e-02 -5.49426377e-01 -8.20748270e-01 -6.44587934e-01
3.35914254e-01 7.26711512e-01 1.01822935e-01 -1.85680941e-01
-4.85454164e-02 5.15938818e-01 -1.55353773e+00 2.13922068e-01
6.36936009e-01 8.87227178e-01 -1.97702035e-01 6.12860501e-01
7.92585984e-02 3.54568690e-01 -8.03851485e-01 -5.71859479e-01
4.45659578e-01 -7.52770305e-01 -7.10226119e-01 2.03820243e-01
8.36434126e-01 -4.01317716e-01 -2.37067804e-01 9.41771269e-01
1.01844335e+00 1.75408423e-01 8.28258812e-01 -1.56351209e+00
-8.20793211e-01 3.68548185e-01 -6.36738956e-01 5.08957148e-01
3.67244408e-02 3.24564278e-01 5.38689017e-01 -1.05732274e+00
5.21868169e-01 1.39754832e+00 7.88979232e-01 1.12235129e+00
-1.71510220e+00 -1.00521410e+00 4.04625349e-02 7.27937147e-02
-1.26826870e+00 -1.00641882e+00 7.24051535e-01 -3.57149214e-01
6.26530468e-01 9.02259350e-02 1.56930521e-01 1.86524248e+00
-2.64980137e-01 7.07948804e-01 1.28542602e+00 -4.24145460e-01
3.52250695e-01 2.38397270e-01 -5.76059632e-02 2.81065345e-01
2.76758313e-01 6.76030070e-02 -4.88510191e-01 -2.87873358e-01
5.44083714e-01 -1.67633638e-01 -1.65422946e-01 -5.47263145e-01
-7.55207956e-01 8.40502203e-01 1.31712526e-01 8.62188116e-02
-1.89211443e-01 -6.63356930e-02 3.81319046e-01 4.57761645e-01
6.18226469e-01 4.43606615e-01 -4.19553816e-01 -2.01290697e-01
-8.32450449e-01 3.24606299e-01 7.55386233e-01 9.84638274e-01
5.69180846e-01 3.65114480e-01 -2.20562562e-01 1.07565773e+00
-7.81628639e-02 6.33293927e-01 6.49892807e-01 -8.18105578e-01
1.88402012e-01 2.78817296e-01 -2.05800012e-02 -4.60733175e-01
-1.55048668e-01 -4.13688213e-01 -6.10760927e-01 3.65498781e-01
7.50337899e-01 -2.34098032e-01 -1.35802448e+00 2.12924767e+00
4.45525050e-01 2.67621636e-01 8.24445337e-02 4.42048997e-01
6.23398185e-01 3.42756122e-01 1.17993832e-01 -5.16968314e-03
9.64547515e-01 -7.38295913e-01 -3.73507470e-01 -4.36388046e-01
3.36897790e-01 -7.05800354e-01 1.42833292e+00 2.83500701e-01
-9.61451530e-01 -5.49646497e-01 -9.29867446e-01 2.55796582e-01
-4.68350351e-01 -1.01381242e-01 3.96472752e-01 8.38962436e-01
-1.01662278e+00 4.65549231e-01 -8.18909347e-01 -7.56068885e-01
9.35814381e-01 3.35719585e-01 -3.51343781e-01 -3.84148091e-01
-8.39474440e-01 7.71328986e-01 2.12751299e-01 -3.14604700e-01
-1.07396066e+00 -8.85969818e-01 -7.32493758e-01 -6.11564517e-02
4.92442757e-01 -5.74320495e-01 1.52452540e+00 -1.18745327e+00
-1.39710867e+00 7.64157474e-01 -4.36039567e-02 -3.15005839e-01
5.56128502e-01 -1.54833078e-01 -4.41110253e-01 -9.89385247e-02
3.38563211e-02 7.73986399e-01 1.15431225e+00 -1.27946198e+00
-2.64706641e-01 -3.33264530e-01 -1.72529250e-01 1.21417800e-02
-7.16207087e-01 9.43987146e-02 -2.96109438e-01 -7.44424760e-01
-3.76192123e-01 -9.51096296e-01 -5.01548909e-02 -1.40575066e-01
-3.68266940e-01 -8.92891828e-03 9.47358668e-01 -3.43956083e-01
1.01266313e+00 -2.16047049e+00 -3.36117782e-02 6.16520233e-02
1.01132067e-02 5.46175122e-01 -4.74523515e-01 2.54384488e-01
-4.39772427e-01 3.86018716e-02 -4.69754457e-01 -4.85275686e-01
7.80339092e-02 1.73847824e-01 -4.47102815e-01 2.36407429e-01
4.23080653e-01 8.82174790e-01 -6.21552587e-01 -2.64333874e-01
-2.01824635e-01 5.26796341e-01 -6.45463586e-01 2.60902435e-01
-2.52640188e-01 4.21734303e-01 -1.49089143e-01 8.92601073e-01
7.37751901e-01 -4.03717279e-01 -1.49897859e-03 -1.37075568e-02
4.48238522e-01 -3.52071188e-02 -1.07081866e+00 1.55275047e+00
-2.49489337e-01 5.43901324e-01 -1.07132509e-01 -9.35502529e-01
8.66871059e-01 1.04536958e-01 3.15371335e-01 -6.27889752e-01
1.64788693e-01 1.83718711e-01 2.21955135e-01 -2.46551946e-01
1.26367614e-01 -2.53734231e-01 4.78086695e-02 5.43201745e-01
4.08101261e-01 1.80418286e-02 2.54654944e-01 1.48989007e-01
1.33276236e+00 2.83226907e-01 2.51662403e-01 -1.50046229e-01
1.46164456e-02 -8.68242159e-02 4.66685951e-01 9.06916618e-01
-4.29543525e-01 8.43865156e-01 4.38244849e-01 -3.08166713e-01
-1.27232242e+00 -1.22956276e+00 -2.30296090e-01 1.38981676e+00
-3.07246745e-01 -1.42169774e-01 -8.76330256e-01 -8.53271425e-01
2.07054287e-01 7.04493046e-01 -9.40666139e-01 -4.95274246e-01
-4.43610251e-01 -1.06624484e+00 6.61150217e-01 6.17494702e-01
3.80159020e-01 -1.06908834e+00 -4.45503056e-01 -1.02582993e-02
1.27758741e-01 -9.95230258e-01 -4.30505872e-01 2.68220544e-01
-6.52399302e-01 -9.59606171e-01 -8.89796317e-01 -5.84387779e-01
6.65919781e-01 8.85733366e-02 1.26594901e+00 -1.47184491e-01
-4.53408390e-01 5.83428562e-01 -1.92064688e-01 -5.26255012e-01
-7.30848014e-01 2.90756762e-01 3.16918135e-01 1.07004374e-01
6.88099504e-01 -7.56385088e-01 -4.78502661e-01 3.63699794e-01
-1.03051114e+00 -9.96829346e-02 6.91554666e-01 1.00813961e+00
3.76248598e-01 -3.70459646e-01 9.80288863e-01 -1.11281073e+00
7.03806460e-01 -6.59958899e-01 -5.28576851e-01 1.15285009e-01
-7.30477870e-01 -5.60402647e-02 5.31345844e-01 -9.57885146e-01
-1.03606880e+00 -1.33881271e-02 -9.93344486e-02 -7.18512118e-01
-5.03620684e-01 7.95411468e-02 -4.32740599e-01 -9.72952992e-02
1.15902960e+00 1.55767202e-01 1.94328696e-01 -5.30804515e-01
3.33307266e-01 6.38449907e-01 5.20766079e-01 -7.39681005e-01
1.05976641e+00 3.31759423e-01 -2.32105434e-01 -7.16934741e-01
-7.88694382e-01 -5.89251667e-02 -6.31876886e-01 -4.70796488e-02
3.90618980e-01 -8.08435917e-01 -1.60383657e-01 6.59631252e-01
-7.96164870e-01 -7.97650516e-01 -4.76275682e-01 8.45579207e-02
-6.12044156e-01 -9.91302263e-03 -3.57706487e-01 -5.56808949e-01
-1.62691012e-01 -9.96979296e-01 9.15740013e-01 3.41048717e-01
-3.03048491e-01 -8.12578499e-01 2.18483850e-01 9.35271308e-02
7.51181602e-01 2.68663615e-01 7.60273278e-01 -1.11501813e+00
-2.13856950e-01 -1.51760459e-01 -1.06951199e-01 5.04755616e-01
3.96939576e-01 -9.47920978e-02 -1.34677041e+00 -5.76878607e-01
-1.75461829e-01 -7.72706032e-01 7.85596013e-01 2.57803589e-01
1.32899559e+00 -2.63550818e-01 -3.51033539e-01 7.86449790e-01
1.14750850e+00 1.91907033e-01 5.96713781e-01 3.43957782e-01
4.91274327e-01 3.38170260e-01 4.19886321e-01 4.77989495e-01
1.04089826e-01 7.09801555e-01 2.79896855e-01 -1.35929540e-01
-4.56636995e-01 -2.52806395e-01 2.60217190e-01 2.17229024e-01
3.16857755e-01 -2.79232621e-01 -1.06740725e+00 6.93789721e-01
-1.48794353e+00 -9.49331284e-01 6.10092103e-01 2.21883321e+00
1.10150385e+00 1.45532131e-01 5.27664900e-01 -6.72179386e-02
6.60196304e-01 -6.83222264e-02 -1.04119492e+00 -1.65388599e-01
-1.94708779e-01 5.01573503e-01 3.35406899e-01 3.40287387e-01
-1.10332215e+00 9.62053895e-01 6.53748465e+00 8.06139708e-01
-1.16937482e+00 3.17375511e-01 9.32946682e-01 -4.29532290e-01
-7.39112571e-02 -3.92085135e-01 -9.89429772e-01 5.71254134e-01
1.15923619e+00 -2.33317256e-01 4.58894849e-01 9.65381861e-01
-2.29138136e-01 1.50452435e-01 -1.37615514e+00 1.01022470e+00
4.00107801e-01 -1.03086543e+00 3.77351381e-02 1.05524480e-01
9.11526561e-01 1.15958229e-01 4.65361208e-01 5.15801847e-01
5.79950690e-01 -1.20120120e+00 3.60099792e-01 2.71692693e-01
9.36142325e-01 -5.07761180e-01 3.99830163e-01 1.31283551e-01
-6.22727692e-01 -1.12637170e-01 -3.03990483e-01 3.10972452e-01
-2.85346478e-01 1.44811660e-01 -1.01121247e+00 1.50263449e-02
7.37937748e-01 4.40033436e-01 -9.99908984e-01 1.10615981e+00
1.13415435e-01 9.54834998e-01 -4.79888827e-01 2.60248274e-01
-2.13912681e-01 1.95859924e-01 4.40466464e-01 1.11868632e+00
4.27672476e-01 -2.46772543e-01 2.64433096e-03 8.68734062e-01
-4.08493638e-01 -1.17595434e-01 -8.58889580e-01 -1.38141051e-01
6.95928097e-01 1.16690207e+00 -5.72439075e-01 -3.64881784e-01
-3.87187034e-01 1.00625384e+00 3.96749973e-01 6.09347463e-01
-9.07224655e-01 -2.82333851e-01 7.59280145e-01 1.55655146e-01
4.21576202e-01 1.17395028e-01 -1.52005941e-01 -1.05809927e+00
7.82557428e-02 -1.32131088e+00 4.52473789e-01 -5.51516235e-01
-1.44428051e+00 7.25795269e-01 2.94429868e-01 -1.16219401e+00
-4.53043103e-01 -5.63994944e-01 -5.42272925e-01 8.21867287e-01
-1.23966861e+00 -1.20071137e+00 -4.11921054e-01 6.75037682e-01
6.59654260e-01 -4.33120757e-01 9.73040581e-01 4.33245718e-01
-7.62445152e-01 1.09089994e+00 1.94409415e-01 7.51314545e-03
1.04240131e+00 -1.11525869e+00 6.45150423e-01 1.00290942e+00
2.09829956e-01 4.37951177e-01 7.81069219e-01 -3.77922773e-01
-1.08009541e+00 -1.13845253e+00 1.85188726e-01 -6.59408987e-01
5.74827552e-01 -6.74793482e-01 -1.20824277e+00 8.72391164e-01
2.07830682e-01 3.98599446e-01 8.46261322e-01 5.08624353e-02
-7.08445132e-01 -2.74727762e-01 -1.32877028e+00 5.90204179e-01
1.04451191e+00 -2.76003987e-01 -2.51091301e-01 2.73032874e-01
3.71583313e-01 -4.59635496e-01 -7.09677994e-01 4.52366471e-01
4.61307496e-01 -7.90318072e-01 9.34591532e-01 -9.96231675e-01
2.75576264e-01 3.86296101e-02 -2.46272892e-01 -1.55478501e+00
-2.46153921e-01 -6.46500647e-01 -2.67601311e-01 1.48959672e+00
4.70294625e-01 -7.13443160e-01 8.41196895e-01 6.62211418e-01
9.00845304e-02 -6.95406914e-01 -9.74269927e-01 -9.88312304e-01
3.43764901e-01 -1.20798312e-01 6.85401678e-01 9.62099493e-01
-5.05599320e-01 3.56799990e-01 -3.37833196e-01 -3.92357334e-02
7.13712156e-01 -1.64505884e-01 1.10474539e+00 -1.18414545e+00
-4.08853978e-01 -4.68709826e-01 -2.60860473e-01 -6.47827566e-01
1.85919479e-01 -7.36430883e-01 4.66743745e-02 -1.08994067e+00
3.34058672e-01 -3.66649598e-01 -3.72951269e-01 8.11536670e-01
-2.06413880e-01 6.48379266e-01 1.80445313e-02 1.23612352e-01
-4.33051348e-01 6.38558924e-01 9.22629595e-01 -4.41885293e-02
-8.34450945e-02 5.99685200e-02 -1.03163624e+00 5.95762253e-01
1.00566578e+00 -4.24487591e-01 -5.86440563e-01 -3.76913428e-01
-1.75496846e-01 -5.57176411e-01 4.76725042e-01 -1.25692606e+00
-1.09636508e-01 -1.63260639e-01 7.00038850e-01 -8.30742419e-02
5.73292732e-01 -5.95319748e-01 -2.22561937e-02 2.17737705e-01
-3.50337476e-01 1.09467976e-01 6.65388227e-01 4.73555148e-01
1.36022657e-01 -8.22099075e-02 1.21237504e+00 1.45882607e-01
-6.48793042e-01 4.89367455e-01 3.79506126e-02 3.95793170e-01
1.07110703e+00 -1.67123571e-01 -5.31316757e-01 -3.47840071e-01
-7.01426923e-01 -1.37691051e-01 6.03060126e-01 6.16617084e-01
5.22703826e-01 -1.50550675e+00 -6.89760864e-01 4.85815376e-01
2.64972627e-01 -1.54154420e-01 8.55042040e-02 6.88024044e-01
6.73893793e-03 -4.93071042e-02 -4.22564059e-01 -5.82823336e-01
-1.35945189e+00 5.42033195e-01 4.70467389e-01 1.52922243e-01
-3.86414856e-01 9.64960873e-01 2.97572792e-01 -4.06891167e-01
3.06819111e-01 1.00946605e-01 1.42177954e-01 -5.40365539e-02
6.42563879e-01 1.30849168e-01 5.67302518e-02 -5.93760610e-01
-2.65984893e-01 2.23619759e-01 -3.98753315e-01 -1.28325775e-01
1.43395996e+00 3.12435478e-02 3.35365117e-01 4.07021552e-01
9.55068588e-01 -1.40782088e-01 -1.49032462e+00 -4.00823325e-01
-1.21093631e-01 -6.27698600e-01 -4.47472870e-01 -1.02272165e+00
-9.92275059e-01 8.33931148e-01 8.85422349e-01 -8.38947296e-03
1.34087920e+00 1.16455384e-01 4.51619685e-01 2.11249158e-01
3.51012647e-02 -9.19829726e-01 4.34013546e-01 2.78338492e-01
8.62428665e-01 -1.40585792e+00 -2.70587921e-01 -2.04258263e-02
-5.67349494e-01 7.24925160e-01 1.11604893e+00 1.28509089e-01
5.15939653e-01 3.85389477e-01 1.34231687e-01 9.20409039e-02
-8.12375367e-01 -2.50210375e-01 4.92389277e-02 9.30615962e-01
2.43701056e-01 -1.57619193e-01 3.61471325e-01 4.16354030e-01
-1.10744230e-01 3.98324542e-02 4.75784838e-01 8.28988612e-01
-2.03261986e-01 -1.36924493e+00 -4.94843215e-01 5.93057990e-01
-4.78252232e-01 -6.51045889e-02 -3.96038473e-01 9.73317564e-01
6.63230121e-02 6.49437428e-01 4.05588448e-02 -3.43325555e-01
3.20293128e-01 4.42970693e-01 6.15281641e-01 -8.25340986e-01
-1.84066504e-01 1.24270061e-03 -9.22189131e-02 -3.80798936e-01
-1.89194262e-01 -9.27590668e-01 -6.03255510e-01 -2.65041381e-01
-1.77521873e-02 -1.31562546e-01 5.19617677e-01 7.56056726e-01
5.75156987e-01 4.47250575e-01 4.31105286e-01 -7.83623576e-01
-9.01932240e-01 -1.27485299e+00 -3.39538813e-01 5.87583542e-01
3.63680542e-01 -8.83721828e-01 -3.50857854e-01 7.70592242e-02]
|
[10.127312660217285, 2.8119099140167236]
|
d393766d-15fe-44cd-b33e-4eb9bd722cfb
|
feature-aggregated-queries-for-transformer
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Cui_Feature_Aggregated_Queries_for_Transformer-Based_Video_Object_Detectors_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Cui_Feature_Aggregated_Queries_for_Transformer-Based_Video_Object_Detectors_CVPR_2023_paper.pdf
|
Feature Aggregated Queries for Transformer-Based Video Object Detectors
|
Video object detection needs to solve feature degradation situations that rarely happen in the image domain. One solution is to use the temporal information and fuse the features from the neighboring frames. With Transformer-based object detectors getting a better performance on the image domain tasks, recent works began to extend those methods to video object detection. However, those existing Transformer-based video object detectors still follow the same pipeline as those used for classical object detectors, like enhancing the object feature representations by aggregation. In this work, we take a different perspective on video object detection. In detail, we improve the qualities of queries for the Transformer-based models by aggregation. To achieve this goal, we first propose a vanilla query aggregation module that weighted averages the queries according to the features of the neighboring frames. Then, we extend the vanilla module to a more practical version, which generates and aggregates queries according to the features of the input frames. Extensive experimental results validate the effectiveness of our proposed methods: On the challenging ImageNet VID benchmark, when integrated with our proposed modules, the current state-of-the-art Transformer-based object detectors can be improved by more than 2.4% on mAP and 4.2% on AP50.
|
['Yiming Cui']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['video-object-detection']
|
['computer-vision']
|
[ 3.10378876e-02 -4.16378796e-01 -8.69765691e-03 -3.11564654e-01
-6.39213324e-01 -3.32011640e-01 4.86082613e-01 5.13212308e-02
-5.48518479e-01 2.80277610e-01 -1.72133282e-01 1.00951619e-01
9.86601710e-02 -7.75638878e-01 -6.87728405e-01 -6.16087019e-01
1.48063404e-02 1.45061165e-01 1.24584067e+00 -2.71481514e-01
1.04430191e-01 2.79687732e-01 -1.74924493e+00 5.68591535e-01
6.05129242e-01 1.39643979e+00 3.65885139e-01 5.58133364e-01
-1.11290924e-01 9.57998455e-01 -5.01857936e-01 -3.17209899e-01
3.94574523e-01 -2.68275112e-01 -5.19635022e-01 1.21419489e-01
5.30028164e-01 -7.93925107e-01 -6.55412376e-01 1.18377459e+00
4.20688808e-01 1.21274933e-01 1.62055850e-01 -1.35867774e+00
-5.40140986e-01 4.97023791e-01 -5.03534794e-01 6.56875670e-01
2.84721941e-01 1.72641799e-01 8.69128287e-01 -1.22785568e+00
3.59833449e-01 1.49131107e+00 4.15141195e-01 3.39275599e-01
-7.09682345e-01 -7.15881169e-01 5.92108250e-01 9.46864665e-01
-1.55727232e+00 -3.46307784e-01 3.27623576e-01 -3.59594911e-01
9.36209321e-01 2.52146304e-01 7.45677352e-01 6.72636092e-01
-1.14738710e-01 1.24088085e+00 7.06315160e-01 -5.23919985e-02
-1.10107437e-02 2.40659788e-01 2.23798200e-01 5.95897615e-01
2.96231687e-01 -6.43962771e-02 -5.48327267e-01 1.33737549e-01
4.30706650e-01 4.09955740e-01 -2.52185643e-01 -3.94787222e-01
-1.14605892e+00 7.34090149e-01 7.10039198e-01 4.17534053e-01
-4.20398831e-01 1.49684593e-01 4.17884290e-01 1.74052775e-01
3.45677406e-01 2.69146357e-02 -1.93987012e-01 1.79943785e-01
-1.16086292e+00 3.90296847e-01 4.36527312e-01 1.06754887e+00
6.86964869e-01 -6.53055906e-02 -7.87221372e-01 5.34693718e-01
3.98762614e-01 3.56058359e-01 3.07189524e-01 -6.55216336e-01
4.26421314e-01 7.19040692e-01 1.08264394e-01 -9.11425769e-01
-1.08128332e-01 -5.19450843e-01 -4.61772025e-01 1.40307054e-01
2.30744213e-01 1.83957189e-01 -9.56466973e-01 1.37268865e+00
3.95218015e-01 4.60479349e-01 -1.22016437e-01 1.16086149e+00
1.02320683e+00 6.70615137e-01 6.76855147e-02 -1.67408407e-01
1.62138891e+00 -1.11497355e+00 -6.49093449e-01 -3.89606170e-02
4.78828579e-01 -8.03162634e-01 5.88493526e-01 2.80798227e-01
-1.05536282e+00 -8.24881434e-01 -1.16112840e+00 6.13128878e-02
-4.04747725e-01 2.49827251e-01 2.63925612e-01 6.35855138e-01
-1.10298550e+00 3.71367425e-01 -8.90819192e-01 -3.78142655e-01
5.81030607e-01 3.35430622e-01 -1.31131053e-01 -1.92166314e-01
-1.04581094e+00 7.76253581e-01 6.28099382e-01 1.42790273e-01
-1.14587629e+00 -6.71720922e-01 -5.58990121e-01 2.74015069e-01
8.07201922e-01 -5.57636082e-01 1.24122524e+00 -8.08034480e-01
-1.02536607e+00 4.23626661e-01 -2.18723327e-01 -7.09767640e-01
6.52679563e-01 -4.19338256e-01 -1.84242710e-01 2.84173518e-01
1.71011388e-01 7.79965878e-01 7.69983172e-01 -8.36946785e-01
-1.29982388e+00 -2.22885072e-01 3.74836266e-01 7.85655826e-02
-5.81439316e-01 3.08861643e-01 -1.06435716e+00 -5.67224026e-01
-1.05529709e-03 -7.74979949e-01 -1.74762174e-01 3.57392550e-01
-8.61971453e-02 -5.34043550e-01 1.35606468e+00 -5.36937058e-01
1.23199630e+00 -2.28746653e+00 7.79469386e-02 -1.37229383e-01
4.51306880e-01 5.63076973e-01 -1.52111858e-01 4.45018709e-02
2.73122221e-01 -1.51709020e-01 -7.17309415e-02 -4.59556878e-01
-5.65552525e-02 -5.11578843e-02 -1.68256342e-01 2.57714093e-01
4.20410931e-01 8.52332830e-01 -8.96857679e-01 -7.52284348e-01
4.06999916e-01 4.05612111e-01 -7.38452971e-01 5.02023734e-02
-1.33653328e-01 9.11683589e-02 -5.87738812e-01 7.89262652e-01
7.98366070e-01 -9.29994732e-02 -9.17018801e-02 -5.30154586e-01
-2.01179311e-01 5.62889725e-02 -1.37327087e+00 1.44520116e+00
1.50625587e-01 7.62005091e-01 1.68360025e-02 -1.05301893e+00
6.47276163e-01 3.39260817e-01 8.50038290e-01 -7.98028290e-01
5.29354401e-02 2.82450825e-01 5.26089892e-02 -5.06702423e-01
5.00045836e-01 3.13151658e-01 1.82489008e-01 4.59756441e-02
2.83733040e-01 2.12385699e-01 6.02727354e-01 3.35761636e-01
1.02353764e+00 5.30766286e-02 1.85594872e-01 1.61593463e-02
7.64005840e-01 -3.45200375e-02 6.46077454e-01 8.87205422e-01
-3.30659330e-01 5.25112867e-01 1.72449157e-01 -5.55389524e-01
-8.23728979e-01 -1.05953765e+00 -4.91198264e-02 1.18968904e+00
4.51463073e-01 -6.67439103e-01 -7.02753901e-01 -8.37092161e-01
-1.78813934e-06 2.26390392e-01 -5.00221848e-01 -2.50081837e-01
-6.45316422e-01 -8.05885017e-01 5.14748454e-01 6.93719745e-01
9.61552143e-01 -8.10755193e-01 -8.06599498e-01 3.30116361e-01
-3.20669800e-01 -1.50324810e+00 -3.97919983e-01 -1.81782722e-01
-6.33462727e-01 -1.12523520e+00 -8.06967974e-01 -7.27508724e-01
2.18582243e-01 7.43278682e-01 8.00574481e-01 1.43362895e-01
-3.72939616e-01 4.08032745e-01 -5.98959208e-01 -5.74397266e-01
4.63975500e-03 -3.19370106e-02 5.94949699e-04 4.31810141e-01
3.89999837e-01 -2.37083465e-01 -7.43935227e-01 5.11273086e-01
-1.07004189e+00 -1.75360724e-01 7.34641194e-01 5.38147509e-01
4.20049787e-01 -3.04155536e-02 2.34104857e-01 -2.69127786e-01
1.71240047e-01 -4.40021187e-01 -6.56622112e-01 3.31609964e-01
-3.06052327e-01 -5.05198427e-02 2.31393337e-01 -5.45368671e-01
-9.12176013e-01 3.12187463e-01 -1.97167303e-02 -7.52081573e-01
1.95581883e-01 1.21681206e-01 -1.00959226e-01 -1.15627214e-01
3.83611292e-01 4.46874440e-01 -2.46977121e-01 -5.27347267e-01
2.31173173e-01 4.76714224e-01 4.07389641e-01 -2.82882720e-01
7.80038595e-01 6.04769468e-01 -1.86061904e-01 -6.46284103e-01
-6.44259870e-01 -7.57429063e-01 -3.18879724e-01 -3.77648681e-01
9.80394661e-01 -1.07108092e+00 -8.33289385e-01 4.49153483e-01
-1.20912313e+00 1.60839707e-01 -1.89708069e-01 6.46665871e-01
-3.81198168e-01 5.32653391e-01 -4.96141642e-01 -6.84246242e-01
-2.67135471e-01 -1.58841407e+00 1.23324943e+00 2.48719856e-01
3.34803402e-01 -3.41715336e-01 -2.80248851e-01 1.02601841e-01
5.56602538e-01 -1.82785258e-01 3.79825205e-01 -5.96595347e-01
-1.12245548e+00 -6.55712858e-02 -6.26398563e-01 4.36745018e-01
7.85463303e-03 -1.79796413e-01 -9.85906839e-01 -3.31335962e-01
9.91234779e-02 -3.96840833e-03 1.36900032e+00 2.54106104e-01
9.99195457e-01 -7.48717859e-02 -5.24910927e-01 5.17889738e-01
1.38608921e+00 2.29243875e-01 5.55236816e-01 3.39992732e-01
6.38814092e-01 3.77458155e-01 8.59339416e-01 5.39811850e-01
3.33500385e-01 1.16477072e+00 7.04318941e-01 2.57880371e-02
-2.51218945e-01 1.75291277e-03 6.35726273e-01 3.46171051e-01
-1.37756199e-01 -2.48076856e-01 -6.77942276e-01 6.27643585e-01
-2.16520429e+00 -1.06298304e+00 -1.55037120e-01 1.87427795e+00
3.28383237e-01 2.72067666e-01 3.18937451e-01 1.49744257e-01
7.33328462e-01 8.27766657e-02 -4.11687493e-01 2.68986255e-01
-1.66766167e-01 -3.47339083e-03 5.06494880e-01 1.71020851e-02
-1.39017582e+00 1.05922985e+00 5.81262064e+00 8.95838261e-01
-1.10897636e+00 4.23376441e-01 1.67055070e-01 -3.91964674e-01
2.90212929e-01 -1.06461821e-02 -9.76247013e-01 3.84352982e-01
4.53186154e-01 -1.99787810e-01 7.21750408e-02 9.54023778e-01
2.01865345e-01 -7.75486380e-02 -1.23569524e+00 1.17946529e+00
1.26872778e-01 -1.31561935e+00 2.15078101e-01 -1.66574195e-01
4.37862426e-01 7.79542252e-02 5.03771566e-02 4.88185108e-01
-4.97982949e-02 -5.16312778e-01 1.15027153e+00 4.31171536e-01
2.73486644e-01 -4.38486278e-01 9.54244494e-01 2.88312525e-01
-1.65244842e+00 -3.93231064e-01 -4.49355513e-01 -2.77227238e-02
2.66766846e-01 4.76293206e-01 -8.11442196e-01 6.63280547e-01
1.28901041e+00 8.77064228e-01 -8.76287878e-01 1.54467344e+00
1.71076283e-01 4.45125610e-01 -4.45750356e-01 -4.63697352e-02
3.13824326e-01 1.40854076e-01 6.70938909e-01 1.32904625e+00
4.52876508e-01 1.24576904e-01 4.45548117e-01 6.77877367e-01
3.17751430e-02 1.95893601e-01 -4.59658951e-01 2.56861180e-01
2.28745326e-01 1.28408015e+00 -7.18018055e-01 -7.79400408e-01
-6.74336910e-01 9.18198049e-01 2.18079612e-02 2.92283177e-01
-1.23854434e+00 -2.28705421e-01 7.75726318e-01 1.61340043e-01
7.46535301e-01 -1.37819901e-01 3.37966800e-01 -1.13151455e+00
4.64265972e-01 -8.28348935e-01 4.99955148e-01 -6.23998642e-01
-9.52942908e-01 6.22198999e-01 1.72938257e-01 -1.48423827e+00
-6.84262589e-02 -7.24510849e-01 -4.30074662e-01 3.85437846e-01
-1.45446479e+00 -1.00318491e+00 -5.59571207e-01 7.45262206e-01
8.40350330e-01 -6.89268261e-02 2.77108133e-01 8.07917476e-01
-5.63771546e-01 5.28430402e-01 -1.58708811e-01 3.92594963e-01
6.16093576e-01 -8.12203944e-01 3.71764958e-01 1.14918220e+00
3.67709160e-01 2.08758146e-01 5.15019298e-01 -3.41133803e-01
-1.33725905e+00 -1.25797820e+00 5.39736211e-01 -2.96417624e-01
4.21880513e-01 -2.97149360e-01 -9.18436110e-01 6.06569529e-01
1.17399491e-01 4.69367832e-01 2.45383009e-02 -4.23251808e-01
-3.93002510e-01 -4.50254232e-01 -9.66328800e-01 4.12691027e-01
1.28097153e+00 -3.22896034e-01 -5.27829170e-01 3.29758912e-01
7.80250907e-01 -2.09617525e-01 -5.52244842e-01 6.48038805e-01
5.13614774e-01 -1.06009865e+00 1.19639695e+00 -5.29068112e-01
-1.25712052e-01 -8.89334023e-01 -4.28326070e-01 -8.64489555e-01
-4.38104510e-01 -2.78937012e-01 -1.80180728e-01 1.09898031e+00
-7.66103417e-02 -4.49426115e-01 5.88599861e-01 4.52152267e-02
-2.30693594e-01 -4.35462236e-01 -1.09852493e+00 -9.93271470e-01
-5.66699564e-01 -6.15532339e-01 4.99949187e-01 4.03994501e-01
-2.08831012e-01 1.54833421e-01 -3.06105852e-01 2.15426296e-01
6.70118690e-01 -4.02348079e-02 5.94448924e-01 -1.11856627e+00
-2.68121779e-01 -5.94185829e-01 -9.29946721e-01 -1.35275209e+00
-3.20861667e-01 -6.44035995e-01 2.37562567e-01 -1.42994797e+00
4.26974624e-01 -1.52960822e-01 -5.41103661e-01 3.42305452e-01
-4.36047107e-01 5.65163255e-01 8.82070124e-01 1.52837843e-01
-1.19632256e+00 5.25804996e-01 9.13478196e-01 -5.13897479e-01
-4.86871004e-02 -6.23986498e-02 -3.89777929e-01 6.55288577e-01
4.62348402e-01 -5.29142797e-01 -2.92051524e-01 -6.05753899e-01
-1.95858702e-01 -2.48122081e-01 6.47763610e-01 -1.47844386e+00
6.21214390e-01 1.19656630e-01 3.85721266e-01 -8.45848978e-01
4.46524054e-01 -9.47644711e-01 6.96383882e-03 6.74669623e-01
4.88351360e-02 5.68707958e-02 2.88418323e-01 5.55788100e-01
-5.33738911e-01 3.11842356e-02 7.59835362e-01 -7.41670802e-02
-1.16141665e+00 4.89277780e-01 -3.99013966e-01 -1.45435318e-01
1.35677338e+00 -2.61662126e-01 -2.48869449e-01 -1.51880279e-01
-6.38004839e-01 4.34755653e-01 4.75529693e-02 7.42119491e-01
9.32989895e-01 -1.35738075e+00 -9.11970496e-01 3.79493944e-02
3.08107853e-01 -1.75155088e-01 3.04347157e-01 1.14880419e+00
-2.16982245e-01 6.41892374e-01 -3.31826627e-01 -1.02744901e+00
-1.46176660e+00 8.75812888e-01 4.71994698e-01 -1.09767638e-01
-4.53804433e-01 7.77714849e-01 6.10358298e-01 2.92113423e-01
2.83866495e-01 -5.98936260e-01 -3.37100178e-01 3.23532969e-01
8.24119866e-01 4.64385062e-01 2.10359842e-01 -6.26837790e-01
-6.63818300e-01 5.50620615e-01 -2.29606777e-01 4.30431999e-02
1.19851816e+00 2.17053629e-02 1.74081400e-01 1.00302897e-01
1.06661952e+00 -2.88722396e-01 -1.22158992e+00 -5.51670015e-01
-4.04564515e-02 -5.02068222e-01 2.57957689e-02 -2.17490122e-01
-1.31276262e+00 7.95951366e-01 9.79176164e-01 2.74586827e-01
1.37630284e+00 2.34996125e-01 5.73169410e-01 4.04245108e-01
4.02473986e-01 -8.58729422e-01 2.73702234e-01 3.91669899e-01
7.38154650e-01 -1.28279579e+00 -5.89636751e-02 -6.37652695e-01
-2.82999843e-01 1.03152573e+00 7.47497797e-01 -1.08394220e-01
6.32302642e-01 1.40205234e-01 -2.14291126e-01 -5.58150373e-02
-7.23065674e-01 -7.02769458e-01 4.46150959e-01 3.61806124e-01
7.98846856e-02 -1.94329455e-01 -2.87249863e-01 4.33037937e-01
3.97581458e-01 1.94883138e-01 2.59469897e-01 8.38163674e-01
-7.43919015e-01 -1.01573050e+00 -5.53808153e-01 3.24724793e-01
-3.69037479e-01 -1.06603198e-01 -1.51835471e-01 6.74351513e-01
4.25341994e-01 1.06014109e+00 1.50159776e-01 -5.34527302e-01
6.30931377e-01 -1.30236760e-01 3.23759437e-01 -5.07060170e-01
-4.77555394e-01 2.81508416e-02 -2.31563911e-01 -9.05832291e-01
-6.41382873e-01 -6.69540644e-01 -1.11709929e+00 -1.00547388e-01
-3.55498165e-01 -5.41845635e-02 4.57319945e-01 8.97993326e-01
3.14460546e-01 7.21389174e-01 2.80671924e-01 -9.35028434e-01
-5.88239729e-01 -9.57554519e-01 -1.90538809e-01 3.70047629e-01
3.64491761e-01 -9.26829576e-01 -4.99935113e-02 -2.27417033e-02]
|
[8.733757019042969, -0.17405490577220917]
|
575cec9a-bc92-421c-8f0c-c7e39bb14897
|
graph-neural-controlled-differential
|
2112.03558
| null |
https://arxiv.org/abs/2112.03558v1
|
https://arxiv.org/pdf/2112.03558v1.pdf
|
Graph Neural Controlled Differential Equations for Traffic Forecasting
|
Traffic forecasting is one of the most popular spatio-temporal tasks in the field of machine learning. A prevalent approach in the field is to combine graph convolutional networks and recurrent neural networks for the spatio-temporal processing. There has been fierce competition and many novel methods have been proposed. In this paper, we present the method of spatio-temporal graph neural controlled differential equation (STG-NCDE). Neural controlled differential equations (NCDEs) are a breakthrough concept for processing sequential data. We extend the concept and design two NCDEs: one for the temporal processing and the other for the spatial processing. After that, we combine them into a single framework. We conduct experiments with 6 benchmark datasets and 20 baselines. STG-NCDE shows the best accuracy in all cases, outperforming all those 20 baselines by non-trivial margins.
|
['Noseong Park', 'Jeehyun Hwang', 'Hwangyong Choi', 'Jeongwhan Choi']
|
2021-12-07
| null | null | null | null |
['spatio-temporal-forecasting']
|
['time-series']
|
[-9.53059494e-02 -5.90920448e-01 -4.66373004e-02 -1.11693991e-02
-1.45810768e-01 -8.25053453e-02 7.57948697e-01 -6.56855851e-02
-3.28132361e-01 4.65528011e-01 2.43169606e-01 -6.19997501e-01
-9.89525300e-03 -7.86341906e-01 -4.85385329e-01 -7.36567616e-01
-8.50986466e-02 1.51219564e-02 9.24399436e-01 -3.93603921e-01
3.11525822e-01 6.87175632e-01 -1.19909036e+00 2.14769408e-01
8.43641520e-01 1.19873583e+00 4.11286810e-03 6.06409669e-01
-5.10127246e-01 1.32811844e+00 -2.16811821e-01 -4.23671901e-01
1.04920477e-01 -4.05376524e-01 -4.47590590e-01 -2.69380987e-01
1.10422723e-01 1.70013145e-01 -7.96490610e-01 8.09890687e-01
4.42035764e-01 4.46055293e-01 6.35112762e-01 -1.35673785e+00
-7.06757009e-01 2.65044332e-01 -8.72581124e-01 7.88245976e-01
-8.06952715e-02 1.50077775e-01 5.60403526e-01 -7.02084064e-01
5.44018507e-01 1.34249663e+00 7.96449184e-01 3.17308038e-01
-8.90057862e-01 -7.22300828e-01 5.72013795e-01 4.97148305e-01
-1.39279926e+00 -2.87173569e-01 9.49877203e-01 -5.42956769e-01
9.99849916e-01 -7.73029253e-02 5.45230091e-01 6.64419472e-01
6.08067036e-01 7.93554246e-01 7.95497835e-01 -6.20844588e-02
1.76565319e-01 -4.00205582e-01 2.32751757e-01 6.94502413e-01
-7.02858046e-02 1.21688843e-01 -3.61374885e-01 1.61948204e-01
7.43157208e-01 1.93583772e-01 1.49630085e-01 4.56525385e-02
-1.05652320e+00 8.32225442e-01 8.01310003e-01 5.43026865e-01
-5.42730451e-01 4.03231770e-01 6.58945560e-01 2.24488795e-01
7.89381862e-01 -4.97248024e-02 7.94936121e-02 -8.56612772e-02
-8.14597189e-01 4.47313190e-01 2.97868341e-01 8.46061885e-01
4.62671489e-01 2.89810777e-01 -6.02728486e-01 5.70773542e-01
8.22695997e-03 3.19752723e-01 3.58791918e-01 -4.87756252e-01
7.24286616e-01 6.74939334e-01 -1.48053288e-01 -1.55484331e+00
-4.80237305e-01 -5.74966431e-01 -1.30444765e+00 -8.27625743e-04
2.10257336e-01 -3.80009413e-01 -9.81976628e-01 1.39555776e+00
3.06927979e-01 7.59885490e-01 -1.95260227e-01 7.16415226e-01
8.11602831e-01 1.26065493e+00 2.88597196e-01 -2.57360071e-01
9.88476753e-01 -1.10711908e+00 -8.93124878e-01 1.24904856e-01
7.29481995e-01 -6.15201831e-01 5.33246040e-01 1.57094061e-01
-9.47875798e-01 -6.58624470e-01 -8.39480758e-01 -1.04278788e-01
-6.65966332e-01 8.39551091e-02 6.52251005e-01 1.23537160e-01
-1.16212583e+00 4.56337959e-01 -8.61052930e-01 -3.00929904e-01
2.80661225e-01 1.88944712e-01 5.92006370e-03 3.24122638e-01
-1.38828957e+00 8.09711456e-01 1.93481401e-01 4.19131786e-01
-4.65171516e-01 -4.14571971e-01 -5.52487731e-01 -5.93851432e-02
4.46967065e-01 -3.34873945e-01 1.02691591e+00 -5.42957485e-01
-1.29739237e+00 4.98625457e-01 -4.32125539e-01 -8.13712895e-01
6.65294766e-01 3.92278023e-02 -9.61420476e-01 -2.06319615e-01
9.52890068e-02 3.53316098e-01 7.10655808e-01 -6.33195460e-01
-8.33694696e-01 -5.96226938e-02 -2.11232737e-01 -5.46052717e-02
-1.74133018e-01 2.60745734e-01 -7.00847208e-01 -8.83693516e-01
4.78721643e-03 -9.47168946e-01 -5.03152192e-01 -2.93542594e-01
-4.20793056e-01 -8.44755411e-01 1.01218259e+00 -6.47078753e-01
1.89791059e+00 -2.03103113e+00 -7.54816830e-02 6.32875192e-04
4.78869319e-01 6.89202130e-01 9.16600227e-02 4.86805409e-01
-9.36873928e-02 5.56213483e-02 -9.84696448e-02 -4.07575727e-01
-8.92523825e-02 -4.89458442e-02 -5.37059903e-01 2.22484365e-01
2.77844191e-01 1.24444914e+00 -7.44118989e-01 -5.58420420e-01
2.21609622e-01 4.66002524e-01 -1.82293758e-01 -2.32572835e-02
-3.23799789e-01 5.44475615e-01 -7.01975763e-01 1.30834818e-01
8.30941737e-01 -2.67832845e-01 -2.38564730e-01 -5.28665036e-02
-5.38176894e-01 3.11751693e-01 -1.08434594e+00 1.33736885e+00
-3.39853525e-01 8.62016559e-01 -4.40104812e-01 -1.00469458e+00
1.00153112e+00 1.40871003e-01 5.46337545e-01 -1.19232631e+00
3.66209373e-02 1.04679354e-01 -1.10023268e-01 -6.43231750e-01
6.67250335e-01 6.72098324e-02 5.58776483e-02 3.34894896e-01
-3.30697894e-01 4.16889727e-01 4.22356457e-01 2.40733147e-01
1.04036498e+00 -1.40164588e-02 -1.03264488e-01 -6.29983991e-02
7.83014894e-01 3.91682377e-03 6.37638509e-01 3.67870212e-01
-2.44114727e-01 4.79530603e-01 7.71982431e-01 -9.08835649e-01
-8.26480627e-01 -6.14575744e-01 3.41849297e-01 9.33380187e-01
1.64232254e-01 -4.29055780e-01 -5.13942778e-01 -6.01277530e-01
-6.69719875e-02 5.70887446e-01 -6.16065562e-01 -3.76121956e-04
-1.04080319e+00 -5.59435725e-01 5.74052215e-01 7.62587428e-01
1.00703895e+00 -1.04828501e+00 -1.85548246e-01 3.91733885e-01
-2.69975737e-02 -1.39259386e+00 -6.93085670e-01 -2.91502267e-01
-7.17371285e-01 -6.08531356e-01 -1.01864123e+00 -9.00227726e-01
2.91323364e-01 7.19875395e-01 9.80539620e-01 1.42272487e-01
1.07916355e-01 -3.80828679e-01 -2.88199961e-01 -3.62099618e-01
-1.11670874e-01 5.17482460e-01 -1.65505394e-01 5.32807887e-01
3.76634389e-01 -7.05688715e-01 -6.32734835e-01 3.63641083e-01
-7.49021828e-01 2.69857258e-01 5.36634684e-01 1.88298509e-01
4.62973475e-01 5.57188988e-02 5.29240012e-01 -9.61975455e-01
9.97213304e-01 -5.51476300e-01 -7.51086056e-01 1.92689806e-01
-5.63555360e-01 8.74960348e-02 8.47459853e-01 -3.73954415e-01
-9.90771651e-01 -6.48882538e-02 -1.24945484e-01 -6.15148425e-01
-9.45195258e-02 8.20428550e-01 3.58699083e-01 5.27405627e-02
5.92887521e-01 2.97218084e-01 -3.51993680e-01 -4.02878970e-01
2.83264965e-01 4.19148326e-01 4.66153443e-01 -1.97393999e-01
7.73646712e-01 3.68221492e-01 1.97289705e-01 -8.81480932e-01
-5.38169742e-01 -4.51977134e-01 -5.43244004e-01 -4.67709839e-01
1.23047662e+00 -6.20728135e-01 -7.49448299e-01 7.77611852e-01
-1.52319348e+00 -4.89311010e-01 1.75975785e-01 4.80505109e-01
-1.30120218e-01 1.49218872e-01 -6.39656067e-01 -9.29848790e-01
-1.92971647e-01 -9.23368692e-01 8.87820661e-01 3.08854431e-01
3.04895252e-01 -1.10275578e+00 2.09009707e-01 -1.46083012e-01
7.99518228e-01 5.14216363e-01 6.86142921e-01 -3.58247370e-01
-6.50440991e-01 -1.91804260e-01 -5.71408749e-01 3.79295833e-02
-1.76390335e-02 1.29903480e-01 -5.73085189e-01 1.62619263e-01
-1.76938877e-01 2.23456800e-01 1.25715232e+00 4.77271944e-01
1.17149782e+00 -6.41319826e-02 -4.55038697e-01 5.74030221e-01
1.27008188e+00 5.41512966e-01 8.13039362e-01 5.07961884e-02
9.75977361e-01 4.66025203e-01 3.42908084e-01 -6.55767322e-02
6.04703605e-01 7.01090753e-01 2.92288978e-03 -2.19440401e-01
-6.67125136e-02 -3.09325755e-01 3.54149073e-01 1.14793825e+00
-3.82226914e-01 -4.73688334e-01 -1.19068801e+00 4.48330462e-01
-2.32637382e+00 -1.13933253e+00 -7.69692719e-01 1.66746366e+00
1.23040475e-01 4.71168339e-01 4.77371484e-01 1.10763431e-01
1.02865899e+00 6.71010554e-01 -5.72353244e-01 -5.00210583e-01
-1.53185949e-02 1.24863379e-01 5.74848771e-01 3.57325584e-01
-1.22648895e+00 1.24161577e+00 5.75601721e+00 8.25225651e-01
-1.48079932e+00 1.71601865e-02 7.94455886e-01 2.68145144e-01
3.85134220e-02 -2.22673401e-01 -7.88981140e-01 5.34165919e-01
1.08543432e+00 -1.08477019e-01 2.34276488e-01 5.15742183e-01
6.57572210e-01 -5.50468899e-02 -5.25945723e-01 1.14601636e+00
-1.86876029e-01 -1.68335915e+00 2.87438501e-02 -8.51441175e-02
7.58620858e-01 1.32220775e-01 2.26803757e-02 3.49849731e-01
2.03403562e-01 -1.02321303e+00 4.26164836e-01 8.46101105e-01
6.52742982e-01 -7.69732833e-01 6.40325487e-01 2.88941711e-01
-1.86720610e+00 7.54488111e-02 -2.55857408e-01 -2.85093486e-01
5.28451979e-01 5.42835176e-01 -2.88665704e-02 6.81436658e-01
5.65999031e-01 1.20834303e+00 -5.83548546e-01 1.17738461e+00
-1.95121597e-02 7.28851378e-01 -1.42301977e-01 -4.45617944e-01
6.11106277e-01 -4.93800998e-01 4.29049343e-01 1.33279049e+00
1.64519295e-01 8.98225382e-02 7.58612454e-02 7.38043249e-01
1.81243233e-02 9.30225775e-02 -6.74205661e-01 -7.67304301e-02
1.96679577e-01 1.02062297e+00 -9.29028869e-01 -4.33240861e-01
-4.90785033e-01 1.02053034e+00 2.56784946e-01 5.10388017e-01
-1.20874321e+00 -6.02223933e-01 3.47792596e-01 3.30982469e-02
4.05722558e-01 -6.16191328e-01 -3.17722768e-01 -1.07280540e+00
1.47307307e-01 -4.25542355e-01 4.49869692e-01 -7.29793251e-01
-1.25148582e+00 7.75349438e-01 4.58137132e-02 -1.19789755e+00
-2.20976889e-01 -5.94665825e-01 -1.00841415e+00 1.02500260e+00
-1.73439240e+00 -1.11494339e+00 -4.32057142e-01 7.63522923e-01
6.67424738e-01 -9.30919200e-02 2.41675004e-01 6.90474689e-01
-9.07048345e-01 3.54022056e-01 -1.37006547e-02 3.45512956e-01
3.19354445e-01 -8.49856555e-01 1.14008069e+00 1.18478549e+00
-1.71757251e-01 3.24268728e-01 4.75610524e-01 -7.58443773e-01
-1.14005017e+00 -1.38789654e+00 1.26860154e+00 -7.19662309e-02
7.77534306e-01 -4.97270703e-01 -9.44961488e-01 6.82812512e-01
2.82023638e-01 7.79277608e-02 -1.14585102e-01 -2.38681167e-01
-1.49009809e-01 -3.09651852e-01 -5.83823681e-01 7.67078459e-01
1.30799758e+00 -4.23493326e-01 -1.33273631e-01 3.53911310e-01
9.14964795e-01 -4.76102859e-01 -3.66708338e-01 1.90744221e-01
1.73885792e-01 -9.72722948e-01 6.09339058e-01 -7.29754984e-01
3.98220181e-01 -5.56353748e-01 1.47654459e-01 -1.11637163e+00
-6.37542844e-01 -9.20797110e-01 -1.75478101e-01 1.27130902e+00
3.77199382e-01 -8.53314400e-01 7.65620768e-01 3.91357988e-01
-2.82367975e-01 -9.25497651e-01 -7.08075106e-01 -8.46903145e-01
1.98724553e-01 -6.88295424e-01 7.13827908e-01 7.23479688e-01
-5.54439127e-01 4.34839338e-01 -5.23439646e-01 -9.19497535e-02
2.22195476e-01 1.84172504e-02 8.32756042e-01 -1.21421170e+00
1.69906527e-01 -6.94475949e-01 -5.66718102e-01 -1.49632871e+00
5.25141098e-02 -8.36845636e-01 -1.82395741e-01 -1.78246891e+00
-1.87563226e-01 -2.60887682e-01 -3.86104584e-01 1.46316990e-01
-4.50736284e-02 -5.03115989e-02 1.54039696e-01 3.06599408e-01
-8.21230888e-01 6.01739049e-01 1.21776140e+00 -5.72420955e-02
-4.44092959e-01 2.44744897e-01 -4.18052286e-01 4.21593547e-01
9.88961875e-01 -2.78225690e-01 -4.56577033e-01 -4.81244832e-01
1.94448233e-01 -2.01460287e-01 1.98681504e-01 -1.26369917e+00
7.21147239e-01 -1.86132655e-01 5.21123558e-02 -9.47783113e-01
-1.37979612e-01 -4.92075592e-01 5.87713392e-03 2.97282964e-01
-2.66301930e-01 7.73680747e-01 1.93912923e-01 6.94797575e-01
-3.58753294e-01 5.23303390e-01 6.59892678e-01 -6.17863145e-04
-9.94865775e-01 7.71908939e-01 -6.58744991e-01 1.07926227e-01
9.81925249e-01 -8.65281746e-02 -3.10617238e-01 -5.00752628e-01
-5.15509248e-01 4.62743014e-01 -1.61855608e-01 5.69385350e-01
6.41686201e-01 -1.60627782e+00 -6.52501285e-01 4.75506298e-02
-1.42962739e-01 -8.16962197e-02 2.74918169e-01 1.26602888e+00
-7.79941738e-01 8.14653635e-01 4.96299975e-02 -5.76013148e-01
-7.69395769e-01 8.06357265e-01 5.97022653e-01 -5.90407908e-01
-7.43716240e-01 5.90122998e-01 1.53695747e-01 -7.91146681e-02
3.04612100e-01 -6.22376025e-01 -4.41903800e-01 -2.97394544e-02
4.44028735e-01 5.98286211e-01 1.27560601e-01 -8.73787105e-01
-2.88070768e-01 6.96261287e-01 3.23794112e-02 6.69743307e-03
1.42318535e+00 -7.62384310e-02 -1.10606201e-01 6.90615833e-01
1.23673403e+00 -3.78824770e-01 -1.03075433e+00 -4.45425034e-01
2.62373716e-01 1.35292094e-02 1.35390878e-01 -2.00958475e-01
-1.42633569e+00 1.17821944e+00 5.33116817e-01 7.06645370e-01
1.09981990e+00 -4.52363938e-01 1.30984640e+00 1.11723304e-01
3.05335280e-02 -1.09793258e+00 -1.89171061e-02 1.07129395e+00
8.29984248e-01 -8.38406861e-01 -2.89602965e-01 -5.08724749e-01
-6.34063303e-01 1.14252198e+00 4.37729597e-01 -4.89016324e-01
1.05769336e+00 1.75018981e-02 -5.85934632e-02 -4.20891434e-01
-7.23663986e-01 -4.68053192e-01 4.37466770e-01 3.14137191e-01
5.30551612e-01 -1.13518082e-01 -5.59374094e-01 2.49056131e-01
1.04185723e-01 2.48169348e-01 -4.21512574e-02 7.10886300e-01
-1.31982386e-01 -8.60884011e-01 5.11903912e-02 4.42306876e-01
-2.84480184e-01 5.67800105e-02 -3.72267693e-01 7.39733696e-01
1.06412068e-01 1.05836892e+00 2.49136731e-01 -7.46345520e-01
4.67732757e-01 -8.54369178e-02 -1.75683483e-01 -1.01894669e-01
-6.23421848e-01 -6.45704046e-02 -6.28256053e-02 -6.56055689e-01
-3.79065752e-01 -5.30841470e-01 -1.33128870e+00 -6.94350302e-01
-1.63985848e-01 1.15484916e-01 3.71270657e-01 9.82824802e-01
6.70897365e-01 8.04331124e-01 7.37813473e-01 -7.12705553e-01
1.65603682e-01 -8.03407550e-01 -3.24836910e-01 2.69789904e-01
3.91330719e-01 -6.29251599e-01 -1.45080671e-01 -2.05192298e-01]
|
[6.5148468017578125, 2.075610876083374]
|
370a03b6-e478-4b11-99e0-cf48d38db001
|
an-investigation-of-evaluation-metrics-for
|
2305.17364
| null |
https://arxiv.org/abs/2305.17364v1
|
https://arxiv.org/pdf/2305.17364v1.pdf
|
An Investigation of Evaluation Metrics for Automated Medical Note Generation
|
Recent studies on automatic note generation have shown that doctors can save significant amounts of time when using automatic clinical note generation (Knoll et al., 2022). Summarization models have been used for this task to generate clinical notes as summaries of doctor-patient conversations (Krishna et al., 2021; Cai et al., 2022). However, assessing which model would best serve clinicians in their daily practice is still a challenging task due to the large set of possible correct summaries, and the potential limitations of automatic evaluation metrics. In this paper, we study evaluation methods and metrics for the automatic generation of clinical notes from medical conversations. In particular, we propose new task-specific metrics and we compare them to SOTA evaluation metrics in text summarization and generation, including: (i) knowledge-graph embedding-based metrics, (ii) customized model-based metrics, (iii) domain-adapted/fine-tuned metrics, and (iv) ensemble metrics. To study the correlation between the automatic metrics and manual judgments, we evaluate automatic notes/summaries by comparing the system and reference facts and computing the factual correctness, and the hallucination and omission rates for critical medical facts. This study relied on seven datasets manually annotated by domain experts. Our experiments show that automatic evaluation metrics can have substantially different behaviors on different types of clinical notes datasets. However, the results highlight one stable subset of metrics as the most correlated with human judgments with a relevant aggregation of different evaluation criteria.
|
['Thomas Lin', 'George Michalopoulos', 'Wen-wai Yim', 'Asma Ben Abacha']
|
2023-05-27
| null | null | null | null |
['graph-embedding', 'knowledge-graph-embedding', 'text-summarization']
|
['graphs', 'graphs', 'natural-language-processing']
|
[ 2.63841838e-01 5.90096056e-01 -1.47250658e-02 -1.87419668e-01
-1.03403401e+00 -5.13874114e-01 6.98323011e-01 9.58361924e-01
-2.51663119e-01 1.00526190e+00 1.10254633e+00 -1.40085921e-01
-4.81058478e-01 -4.93091315e-01 3.06198716e-01 -4.29183215e-01
-1.97668392e-02 6.04091346e-01 -5.44324331e-02 -1.68153435e-01
5.89188159e-01 2.53924072e-01 -1.13331091e+00 5.93056560e-01
1.25041533e+00 5.24022996e-01 -1.06652379e-01 1.08281708e+00
-1.50784075e-01 1.20966077e+00 -1.07325172e+00 -7.58557260e-01
-3.04002941e-01 -9.48058963e-01 -9.16004598e-01 -7.25411251e-02
2.70750403e-01 -9.28676054e-02 -1.12729281e-01 8.89732718e-01
1.13424420e+00 -3.35852951e-02 9.07045901e-01 -1.02397323e+00
-7.99375415e-01 8.13511550e-01 1.13124065e-01 1.96744636e-01
7.88106680e-01 2.04401195e-01 9.18241203e-01 -6.04069829e-01
8.19178462e-01 8.65268528e-01 8.54333222e-01 7.86219656e-01
-9.66533303e-01 -2.69045740e-01 -3.52310121e-01 2.89424658e-01
-1.09967101e+00 -4.36485529e-01 6.24077082e-01 -7.03645825e-01
9.34835076e-01 6.64753675e-01 6.44070208e-01 1.07210803e+00
4.34785515e-01 5.34614027e-01 8.90483201e-01 -3.49822313e-01
3.30318093e-01 5.01534700e-01 4.08889711e-01 4.56322640e-01
7.12116241e-01 -3.52996886e-01 -3.62331808e-01 -6.38900638e-01
3.35776478e-01 -2.32217014e-01 -7.03935742e-01 2.61994928e-01
-1.47465181e+00 1.00156021e+00 1.00228205e-01 6.78694546e-01
-5.43836772e-01 -4.43782479e-01 5.60346007e-01 5.96774975e-03
3.76318485e-01 1.10636675e+00 -2.07315639e-01 -3.53152812e-01
-1.25023377e+00 3.77611965e-01 1.19566512e+00 7.33400881e-01
-2.08699554e-02 -8.01238120e-02 -1.02555919e+00 8.91198039e-01
-9.90132764e-02 3.07169139e-01 9.94996965e-01 -8.49396169e-01
4.59496498e-01 5.99859297e-01 1.92286208e-01 -1.37132192e+00
-6.89300716e-01 -4.86165732e-01 -1.08845770e+00 -3.98412377e-01
1.43911302e-01 -2.34390959e-01 -5.88599503e-01 1.35716748e+00
-8.09152424e-02 -2.63028830e-01 4.25981730e-01 5.68482637e-01
1.55932796e+00 4.04985726e-01 1.99645348e-02 -7.39504576e-01
1.30004323e+00 -1.10805213e+00 -1.28892350e+00 2.15053231e-01
8.08809221e-01 -9.53625381e-01 8.36180210e-01 3.19630802e-01
-1.39008546e+00 -3.50551128e-01 -8.20658088e-01 7.36692995e-02
-1.35163322e-01 2.80478865e-01 2.43935436e-01 5.68383098e-01
-1.19354367e+00 7.36119926e-01 -5.32433629e-01 -5.80767334e-01
1.38089061e-01 -2.25982852e-02 -6.75952584e-02 2.43367299e-01
-1.15428185e+00 1.11684692e+00 2.81889260e-01 -1.42870724e-01
-3.69725257e-01 -7.35642970e-01 -7.20804334e-01 1.63526326e-01
6.38305619e-02 -1.12072384e+00 1.15721107e+00 -3.77406150e-01
-1.29128551e+00 7.84366310e-01 -2.44082119e-02 -4.68064606e-01
6.59786582e-01 8.06904864e-03 -6.13742769e-01 4.48826730e-01
8.14567581e-02 3.95824224e-01 1.91559121e-01 -1.20990920e+00
-3.82390559e-01 -4.77258414e-02 -1.04884416e-01 1.86963513e-01
-4.81321841e-01 -2.19079331e-01 5.43680638e-02 -8.86996210e-01
-2.54650682e-01 -7.97290802e-01 -3.25616002e-01 -4.34445858e-01
-9.31740046e-01 -3.01415890e-01 9.41834599e-02 -9.42187607e-01
1.94748330e+00 -1.74835908e+00 -1.06578499e-01 -5.23050576e-02
5.14180362e-01 5.65266013e-01 -1.18435010e-01 9.90227699e-01
6.64927885e-02 3.90166670e-01 -5.06946921e-01 -1.92783058e-01
-1.81131124e-01 -6.15471900e-02 -1.97267026e-01 -1.05500281e-01
1.45066887e-01 8.68630111e-01 -1.21127748e+00 -9.64017570e-01
5.88768199e-02 4.89637196e-01 -5.53937137e-01 3.73742521e-01
1.50051415e-01 3.37272733e-01 -4.15450811e-01 3.86461556e-01
2.10591689e-01 -4.56450015e-01 3.26256365e-01 -4.44008172e-01
1.65728062e-01 4.43844467e-01 -7.40187049e-01 1.26573527e+00
-2.71391749e-01 5.65838933e-01 -6.70854807e-01 -5.79220831e-01
8.98547649e-01 7.57280767e-01 5.78509152e-01 -1.36342511e-01
8.58430564e-02 2.03192383e-01 1.67765290e-01 -9.71204221e-01
7.96788573e-01 -1.30840778e-01 2.98226085e-02 5.67211449e-01
-1.04699783e-01 -4.55525070e-01 4.94741946e-01 5.54510057e-01
1.41636610e+00 -4.98874784e-01 1.00993943e+00 -1.29676700e-01
5.25186837e-01 2.67374635e-01 4.68961209e-01 8.12174022e-01
-3.52093667e-01 1.01852334e+00 6.09416068e-01 -2.84857482e-01
-7.52804637e-01 -8.45507205e-01 -1.33545712e-01 2.26973653e-01
-2.45980322e-01 -8.16150367e-01 -8.86017859e-01 -7.24359632e-01
-1.08432695e-01 1.19636309e+00 -6.65756106e-01 -3.76468062e-01
-3.21560949e-01 -9.08367515e-01 8.40713322e-01 4.81580555e-01
1.67841420e-01 -1.25437510e+00 -7.98563242e-01 3.97926390e-01
-7.01675057e-01 -9.90581095e-01 -7.34228730e-01 -3.79806936e-01
-8.80983889e-01 -1.24122393e+00 -8.13486993e-01 -4.76542801e-01
5.85887730e-01 -1.08996909e-02 1.35698032e+00 2.53906190e-01
-2.93068022e-01 6.42363369e-01 -5.81675470e-01 -4.09450114e-01
-8.85340989e-01 -8.20748955e-02 -4.11972143e-02 -3.12431633e-01
1.37808040e-01 -3.11351746e-01 -8.59661162e-01 -2.08311453e-01
-1.10914242e+00 1.55423461e-02 6.68273091e-01 8.96022379e-01
3.52268696e-01 -4.00757462e-01 7.71400869e-01 -1.22083879e+00
1.59720993e+00 -2.98223823e-01 3.39899808e-01 4.64301050e-01
-9.25590754e-01 8.41146708e-02 6.20046437e-01 -2.72565991e-01
-8.22798371e-01 -5.28582752e-01 -4.65139411e-02 -3.19247805e-02
7.79517144e-02 6.75037324e-01 3.64759445e-01 3.81725580e-01
1.05413353e+00 2.74215609e-01 -9.80834737e-02 -1.87919736e-01
2.64063030e-01 8.70065033e-01 3.86623204e-01 -1.23159349e-01
3.76337856e-01 1.19134963e-01 -2.52479851e-01 -7.29552031e-01
-1.00344193e+00 -3.47689837e-01 -4.46132839e-01 -1.71221733e-01
9.84362066e-01 -4.26254779e-01 -4.55513239e-01 -1.07433915e-01
-1.38357282e+00 1.51067242e-01 -5.13886690e-01 6.76092207e-01
-5.26109815e-01 7.01598644e-01 -6.68826282e-01 -7.21648097e-01
-9.52065647e-01 -9.87627029e-01 7.90857613e-01 1.49142563e-01
-1.09080780e+00 -1.25474644e+00 6.05766118e-01 3.14127803e-01
4.10600632e-01 7.13634670e-01 1.10806501e+00 -1.07561934e+00
1.41144753e-01 -1.68173298e-01 -1.07822128e-01 3.94626856e-01
6.11406267e-01 1.76591694e-01 -7.97803283e-01 -1.25168934e-01
1.02916621e-01 -1.00703761e-01 7.60264635e-01 5.29061675e-01
1.00459981e+00 -8.51839781e-01 -3.17055821e-01 1.24846831e-01
1.21668339e+00 3.24893117e-01 5.58858097e-01 -8.37809965e-02
4.44334179e-01 7.50135720e-01 4.38587427e-01 5.68369329e-01
5.59454739e-01 3.28811049e-01 -1.89560294e-01 1.86689258e-01
-2.41415724e-01 -1.63709030e-01 1.75589725e-01 1.63775265e+00
-3.26223075e-01 -4.89591479e-01 -8.97473872e-01 7.34670281e-01
-1.88262475e+00 -1.17347157e+00 -2.90891528e-01 2.10816240e+00
1.14026618e+00 -7.52964169e-02 8.14491585e-02 2.13730335e-01
6.09672785e-01 4.73102890e-02 -1.63680077e-01 -6.41542733e-01
-2.68937610e-02 1.67103827e-01 -3.56280580e-02 4.00013059e-01
-6.88469887e-01 3.58619452e-01 6.69033337e+00 5.82502842e-01
-8.08746696e-01 -3.01081743e-02 5.15566766e-01 -3.45343538e-02
-5.69133937e-01 -3.52125525e-01 -4.50211972e-01 5.61662257e-01
1.07768059e+00 -8.17691445e-01 -1.80435002e-01 5.15360117e-01
3.14721316e-01 8.48133862e-02 -1.22870719e+00 1.00724566e+00
6.68253541e-01 -1.54600859e+00 4.25800651e-01 -7.00115561e-02
1.02091360e+00 -5.11805832e-01 -3.06622088e-01 3.81749161e-02
2.40245447e-01 -9.18242395e-01 1.97513461e-01 8.08410168e-01
6.71474934e-01 -3.56703103e-01 1.10812795e+00 2.48142462e-02
-6.64261878e-01 1.45161867e-01 -1.97066918e-01 2.98251897e-01
3.27237576e-01 9.65046167e-01 -1.24936938e+00 7.35669732e-01
2.78974175e-01 6.24057114e-01 -7.09210932e-01 1.22496533e+00
-1.94700643e-01 5.43149114e-01 1.70253143e-01 -3.52485567e-01
6.28671199e-02 -2.91070458e-03 6.05105460e-01 1.61062777e+00
5.73591113e-01 3.29276055e-01 -1.11177631e-01 7.23463833e-01
8.35357457e-02 4.87729311e-01 -7.10643113e-01 -3.25499713e-01
4.79527146e-01 1.39090943e+00 -7.18713939e-01 -7.50439227e-01
6.73836749e-03 7.21647143e-01 4.04674234e-03 8.98015127e-02
-6.79890811e-01 -5.76729059e-01 2.99401104e-01 -1.13214850e-02
-1.51657447e-01 2.66717911e-01 -5.39621592e-01 -1.01964617e+00
-6.68539405e-02 -1.00952983e+00 5.13247967e-01 -7.73259521e-01
-1.31541777e+00 1.08103549e+00 -4.68744487e-02 -1.67493260e+00
-5.05178213e-01 -1.85123146e-01 -7.73045480e-01 4.82290506e-01
-1.18861747e+00 -5.89285314e-01 -4.98445719e-01 2.62517303e-01
4.43780303e-01 -1.81544676e-01 1.16320765e+00 2.21990168e-01
-3.48871291e-01 6.65243268e-01 -8.39198902e-02 4.76244800e-02
8.56536865e-01 -1.40169048e+00 9.00084898e-02 3.71651113e-01
-1.44400978e-02 6.02530599e-01 8.65154564e-01 -6.90710068e-01
-6.33624434e-01 -1.16350973e+00 1.49141824e+00 -7.05522060e-01
3.80556703e-01 4.94766325e-01 -1.06803799e+00 3.60533267e-01
4.49349314e-01 -7.05064416e-01 1.18445754e+00 -2.88586289e-01
1.25968844e-01 1.56118110e-01 -1.19533527e+00 7.67211318e-01
9.40988004e-01 -2.09000379e-01 -9.11991060e-01 7.42475688e-01
7.75919855e-01 -1.94213524e-01 -1.19066072e+00 3.99515778e-01
3.78687173e-01 -1.12809706e+00 7.10821271e-01 -7.59627581e-01
9.11454141e-01 -7.94332922e-02 1.60224706e-01 -1.72954881e+00
-3.71906966e-01 -5.03356516e-01 2.33015865e-02 1.28835726e+00
6.22764170e-01 -5.66765547e-01 2.05902442e-01 6.64034903e-01
-2.82921284e-01 -9.72527683e-01 -5.27040780e-01 -4.34249312e-01
-2.50205398e-02 1.25473991e-01 4.80411351e-01 1.15727532e+00
5.66276908e-01 4.87802058e-01 -2.77511835e-01 -2.41212711e-01
3.09918851e-01 8.36848170e-02 5.43475330e-01 -1.24043679e+00
-1.45523220e-01 -6.39173031e-01 -4.11825687e-01 -2.95765191e-01
-2.66759276e-01 -1.03369725e+00 -5.59073947e-02 -2.30173373e+00
5.16885459e-01 -3.52695361e-02 -2.35293791e-01 3.36803228e-01
-6.12716198e-01 -4.57876213e-02 2.41487339e-01 3.46887678e-01
-4.56719905e-01 4.04583186e-01 1.34425855e+00 -1.82805389e-01
-3.36128086e-01 -1.52392581e-01 -1.08242309e+00 5.78090131e-01
8.42040181e-01 -5.02450883e-01 -4.10881460e-01 -1.62269443e-01
2.79853225e-01 3.35046202e-01 4.69552465e-02 -1.01595175e+00
2.70232201e-01 -9.29049999e-02 2.27081534e-02 -4.18593407e-01
-4.28388594e-03 -3.03538710e-01 8.76171887e-02 8.05785060e-01
-7.34315813e-01 3.35137695e-01 -1.68428589e-02 3.89629334e-01
-4.70632374e-01 -4.73252445e-01 4.36775863e-01 -2.30284780e-01
-1.52735049e-02 -7.78677762e-02 -4.05459136e-01 4.72717881e-01
7.53163338e-01 -1.38450578e-01 -6.26329601e-01 -8.18508506e-01
-8.10414195e-01 1.79269686e-01 1.45902202e-01 2.73085028e-01
7.91056514e-01 -1.27652514e+00 -1.15679169e+00 -4.87332374e-01
1.93915188e-01 -3.46698642e-01 2.30679139e-01 1.19022763e+00
-6.93445265e-01 5.92737675e-01 -9.02088135e-02 -3.44297349e-01
-1.33628082e+00 4.50716853e-01 3.11105866e-02 -8.52512002e-01
-5.44266641e-01 4.27672595e-01 2.65027676e-02 -2.89854437e-01
4.36079595e-03 -6.96446061e-01 -6.51336968e-01 3.43114167e-01
7.91852772e-01 6.40718400e-01 2.41535485e-01 -4.20805186e-01
-3.38460505e-01 5.26301563e-01 -1.96975786e-02 4.07341011e-02
1.15791225e+00 6.80514500e-02 -1.17873184e-01 3.99853945e-01
9.86656368e-01 1.81280047e-01 -2.43251115e-01 -6.60290495e-02
1.70967117e-01 -4.05636653e-02 -2.95052797e-01 -1.06628466e+00
-6.70921266e-01 7.97262967e-01 2.49544024e-01 6.11725867e-01
1.02566862e+00 -3.08513552e-01 8.50570798e-01 3.99391025e-01
-4.01557870e-02 -1.06135547e+00 2.68793285e-01 1.20518744e-01
1.23975503e+00 -1.17212474e+00 1.80942237e-01 -3.52274209e-01
-1.19213951e+00 1.20221674e+00 1.52424172e-01 2.37917662e-01
4.59817708e-01 -1.18751295e-01 3.67027044e-01 -2.60501504e-01
-9.31981146e-01 3.51746986e-03 4.88208294e-01 5.72671473e-01
8.44285786e-01 3.08327645e-01 -1.03460503e+00 8.14364970e-01
-6.53230906e-01 2.41106570e-01 1.05256963e+00 6.35739684e-01
-2.41145551e-01 -8.05731952e-01 -1.60209551e-01 8.77711594e-01
-5.03995895e-01 -2.60292798e-01 -7.22108543e-01 6.13011599e-01
-2.08125517e-01 1.37770605e+00 -3.15456331e-01 -4.60777074e-01
6.29449606e-01 2.16985181e-01 3.88116360e-01 -9.21999872e-01
-1.01904845e+00 -2.85207182e-01 5.38251936e-01 -2.83569157e-01
-6.00567520e-01 -5.69232881e-01 -1.12145877e+00 -1.27037570e-01
-2.30821624e-01 5.33954740e-01 1.96550786e-01 6.37657583e-01
6.24802828e-01 8.83888483e-01 4.29751277e-01 -3.63502622e-01
-7.56146312e-01 -1.33023775e+00 -2.57376462e-01 7.46265233e-01
2.71739155e-01 -2.05344066e-01 -5.21466374e-01 2.46419951e-01]
|
[12.263518333435059, 9.455862998962402]
|
1ab2d274-1ede-4e81-b63c-26253fceb963
|
adversarial-self-attention-for-language
|
2206.12608
| null |
https://arxiv.org/abs/2206.12608v3
|
https://arxiv.org/pdf/2206.12608v3.pdf
|
Adversarial Self-Attention for Language Understanding
|
Deep neural models (e.g. Transformer) naturally learn spurious features, which create a ``shortcut'' between the labels and inputs, thus impairing the generalization and robustness. This paper advances the self-attention mechanism to its robust variant for Transformer-based pre-trained language models (e.g. BERT). We propose \textit{Adversarial Self-Attention} mechanism (ASA), which adversarially biases the attentions to effectively suppress the model reliance on features (e.g. specific keywords) and encourage its exploration of broader semantics. We conduct a comprehensive evaluation across a wide range of tasks for both pre-training and fine-tuning stages. For pre-training, ASA unfolds remarkable performance gains compared to naive training for longer steps. For fine-tuning, ASA-empowered models outweigh naive models by a large margin considering both generalization and robustness.
|
['Min Zhang', 'Fei Huang', 'Pengjun Xie', 'Hai Zhao', 'Ruixue Ding', 'Hongqiu Wu']
|
2022-06-25
| null | null | null | null |
['paraphrase-identification', 'machine-reading-comprehension']
|
['natural-language-processing', 'natural-language-processing']
|
[ 2.31951430e-01 2.93627053e-01 -2.52341688e-01 -5.91297328e-01
-6.69815302e-01 -9.75861251e-01 9.45007920e-01 -5.31889349e-02
-5.40486217e-01 4.65400517e-01 2.29617804e-01 -5.69808543e-01
2.89690167e-01 -9.05301273e-01 -9.66993570e-01 -3.61908048e-01
1.18120566e-01 3.05538595e-01 3.25661778e-01 -5.37144721e-01
-1.14959404e-01 4.06443685e-01 -1.17322850e+00 4.24385905e-01
9.21930075e-01 9.81827557e-01 -1.17773868e-01 1.75003275e-01
-2.32186809e-01 8.86374295e-01 -6.45277679e-01 -8.08007121e-01
1.49170712e-01 1.36846915e-01 -7.20457971e-01 -2.40084693e-01
5.11426747e-01 -4.32618111e-01 -7.18265533e-01 9.94477093e-01
3.65169346e-01 2.21808329e-01 7.01602042e-01 -1.28917730e+00
-1.31172431e+00 1.11569750e+00 -3.12176257e-01 3.07282001e-01
-7.94958994e-02 5.56758285e-01 1.27538943e+00 -1.12644863e+00
1.96835890e-01 1.36902905e+00 8.34257066e-01 8.00917745e-01
-1.39341140e+00 -1.01637959e+00 5.75541019e-01 6.88755233e-03
-1.27401447e+00 -4.45927233e-01 6.98958337e-01 -3.98806363e-01
1.05640173e+00 3.51850510e-01 1.32149979e-01 1.76396334e+00
2.90739723e-03 9.81871426e-01 1.06098652e+00 -2.08634377e-01
1.55662701e-01 2.65143186e-01 2.19834745e-01 5.48046887e-01
8.37543458e-02 2.96984166e-01 -3.14900696e-01 -1.40959695e-01
6.41088247e-01 2.53432840e-01 2.56527681e-02 -2.34695494e-01
-1.02243805e+00 8.29829812e-01 9.12033081e-01 3.16883326e-01
-2.22302765e-01 1.94490209e-01 5.55931032e-01 4.02890414e-01
4.73972738e-01 9.26393092e-01 -7.21718252e-01 3.60459507e-01
-8.29380870e-01 4.30577770e-02 3.25795054e-01 1.11092651e+00
8.47748637e-01 3.38066965e-01 -8.00673425e-01 9.59206879e-01
-2.40877662e-02 4.10689980e-01 6.68880045e-01 -4.78496164e-01
3.64950836e-01 7.15175629e-01 -2.22244307e-01 -5.07082582e-01
-1.27217188e-01 -9.11156833e-01 -1.04472256e+00 -5.69401635e-03
2.53320307e-01 -1.58691689e-01 -1.33689117e+00 2.16949224e+00
-1.02212474e-01 1.04966290e-01 -5.27770221e-02 5.39205372e-01
7.78612375e-01 4.04023290e-01 6.07111990e-01 2.83485234e-01
1.29327738e+00 -1.27176952e+00 -4.03972507e-01 -6.26463175e-01
5.03092110e-01 -4.66776758e-01 1.69567800e+00 1.33991793e-01
-9.10662711e-01 -6.44916117e-01 -9.45467889e-01 -2.65531391e-01
-8.20205867e-01 -2.63134956e-01 7.11328268e-01 4.28439587e-01
-1.04625022e+00 8.25269461e-01 -6.65022135e-01 -2.80858874e-01
7.18933940e-01 5.29826343e-01 -3.99071664e-01 -3.05707771e-02
-1.65204549e+00 1.01715052e+00 3.23462248e-01 -3.81171852e-01
-1.22621870e+00 -9.14888322e-01 -7.99143493e-01 3.63500714e-01
5.47391653e-01 -7.02415228e-01 1.46093917e+00 -1.05918682e+00
-1.39705741e+00 9.21371281e-01 1.61596984e-01 -6.84770405e-01
5.62278986e-01 -5.84128082e-01 -3.26241046e-01 -2.84695327e-01
2.08177462e-01 8.37256908e-01 1.13804460e+00 -1.13441956e+00
-2.94741869e-01 -1.08823963e-01 4.89405751e-01 4.62833755e-02
-7.79922843e-01 -1.55274477e-02 -3.40087652e-01 -1.22519100e+00
-3.78140002e-01 -8.59039664e-01 -2.54924357e-01 -2.15254366e-01
-7.83704340e-01 -4.49822217e-01 6.64635181e-01 -3.11964124e-01
1.38033366e+00 -2.12582183e+00 -1.20438494e-01 -1.64555199e-02
4.53875840e-01 4.82099712e-01 -3.47556144e-01 2.79824585e-01
-3.63742799e-01 4.97119993e-01 -1.13989212e-01 -4.63676631e-01
1.34145066e-01 3.76271427e-01 -7.94627011e-01 2.23035052e-01
5.36106884e-01 1.31788945e+00 -8.59141052e-01 -1.92564502e-01
1.34184077e-01 2.48174727e-01 -7.76186049e-01 3.82123768e-01
-5.37405789e-01 2.74107814e-01 -5.41823983e-01 6.50682390e-01
4.49035227e-01 -3.83331299e-01 -3.22404176e-01 -1.52804732e-01
1.83352351e-01 5.18546224e-01 -4.66787577e-01 1.59132135e+00
-8.22395802e-01 1.72709823e-01 -2.28088751e-01 -8.33894134e-01
7.51032770e-01 1.35859072e-01 2.22726278e-02 -6.77048802e-01
1.48549810e-01 4.94384542e-02 -1.74137745e-02 -2.05900781e-02
3.38073283e-01 -2.78468430e-01 -3.34571898e-01 3.55138272e-01
3.84023756e-01 2.15894118e-01 -2.16498837e-01 4.58153903e-01
1.18695331e+00 7.65646622e-02 2.03920528e-01 -5.01618207e-01
4.42899048e-01 -1.41218513e-01 3.22431266e-01 1.18913913e+00
-1.47464350e-02 2.10421920e-01 2.85865933e-01 -4.00302231e-01
-9.35193002e-01 -1.17086184e+00 5.57198413e-02 1.94165540e+00
1.02056761e-03 -4.81646776e-01 -6.06270075e-01 -1.25954282e+00
3.57702494e-01 1.14561749e+00 -1.00647998e+00 -8.64476919e-01
-3.51182699e-01 -4.19628054e-01 8.23992193e-01 8.52182627e-01
4.01040107e-01 -1.13549328e+00 -1.15495868e-01 1.32026926e-01
2.11736664e-01 -8.83583426e-01 -7.27353632e-01 6.40569329e-01
-6.11020267e-01 -4.61846471e-01 -5.82127452e-01 -5.25703728e-01
5.68651021e-01 -2.07862668e-02 1.36356342e+00 1.07058339e-01
1.31569371e-01 6.11537732e-02 -2.17492431e-01 -4.47653800e-01
-3.69638860e-01 4.19064641e-01 1.15993507e-01 -9.46040154e-02
5.51550984e-01 -6.93368137e-01 -4.59394127e-01 3.41338187e-01
-9.28681850e-01 -1.60604417e-01 8.21032643e-01 1.00999916e+00
2.81983942e-01 -1.60713285e-01 7.48429477e-01 -1.20324123e+00
6.09133542e-01 -7.17482865e-01 -3.07976484e-01 3.28012377e-01
-8.06277871e-01 3.18764329e-01 1.15337420e+00 -7.29076207e-01
-9.26142156e-01 -3.29161763e-01 -8.91400427e-02 -7.24902093e-01
-1.35205105e-01 2.43889555e-01 -4.21782494e-01 9.02900919e-02
8.98269832e-01 2.31176496e-01 -3.08942169e-01 -7.48804867e-01
7.10625708e-01 5.59985340e-01 7.07533121e-01 -8.11053991e-01
1.16200840e+00 2.32306540e-01 -4.48915154e-01 -1.65067181e-01
-1.36455286e+00 -2.23572820e-01 -3.18874478e-01 3.45194787e-01
5.36404669e-01 -1.06018043e+00 -3.29019159e-01 2.24490017e-01
-9.53842163e-01 -3.96005929e-01 -5.24322093e-01 1.05842993e-01
-2.54703820e-01 -7.61028901e-02 -6.89833283e-01 -4.45063561e-01
-3.48457217e-01 -1.06503010e+00 1.07542574e+00 6.53159432e-03
-4.16173846e-01 -1.09263217e+00 -2.77698010e-01 1.00979969e-01
7.56407380e-01 -9.36609134e-02 1.13927591e+00 -1.46419227e+00
-4.25892383e-01 -1.63949966e-01 -2.66031235e-01 5.31241894e-01
1.21219076e-01 -2.88652927e-01 -1.47703159e+00 -3.91223371e-01
-1.29353136e-01 -5.31202316e-01 1.12931824e+00 -9.26676765e-02
1.48039341e+00 -6.66430414e-01 -4.10199940e-01 8.12930584e-01
1.14304483e+00 -4.13764417e-02 5.34687638e-01 4.75499034e-01
8.69011521e-01 3.19536209e-01 3.19697589e-01 -1.69882458e-02
1.75305650e-01 5.80081582e-01 4.79808807e-01 -3.88416708e-01
-1.21550061e-01 -7.67652154e-01 3.62737477e-01 4.32141304e-01
2.96978444e-01 -4.11147326e-01 -8.51675630e-01 5.76915145e-01
-1.50329471e+00 -6.43525839e-01 5.24384379e-01 2.03038931e+00
1.14929163e+00 5.70406556e-01 -5.29936068e-02 -1.35349473e-02
7.62466431e-01 1.24388747e-01 -8.92226160e-01 -4.73740965e-01
-4.70417179e-02 5.64077258e-01 5.56420207e-01 4.82286036e-01
-1.29050660e+00 1.45308685e+00 6.12642527e+00 1.27776802e+00
-1.41561723e+00 2.61601776e-01 7.29413152e-01 -4.92785126e-02
-7.41400421e-01 -6.04122989e-02 -6.76197529e-01 5.19481838e-01
8.79512966e-01 -3.33194554e-01 5.08916676e-01 1.18227077e+00
-3.35561424e-01 7.34734416e-01 -1.30552495e+00 3.92213732e-01
-2.79987961e-01 -1.25813937e+00 5.86457849e-01 -2.81722806e-02
6.82468593e-01 3.56478572e-01 4.26422536e-01 1.04215252e+00
7.11368501e-01 -1.29716861e+00 8.13746154e-01 1.94850132e-01
1.21214330e+00 -6.77344739e-01 4.96318817e-01 3.01238149e-01
-7.00721085e-01 -3.01065981e-01 -2.24249423e-01 4.56812903e-02
-1.92044348e-01 4.97401267e-01 -7.31570542e-01 3.53250593e-01
5.70951760e-01 5.75714946e-01 -7.97282338e-01 4.25446838e-01
-4.52690303e-01 8.15579474e-01 -3.05936605e-01 1.97678030e-01
5.72591960e-01 2.42934436e-01 4.37488765e-01 1.31599772e+00
-1.48775214e-02 -2.48469815e-01 2.59231210e-01 9.82909143e-01
-5.98795772e-01 -1.26337353e-02 -6.70373082e-01 -2.64739335e-01
6.54192209e-01 1.08537567e+00 -2.74429649e-01 -6.84010863e-01
-4.42324668e-01 9.02160466e-01 6.08574808e-01 4.76830781e-01
-9.53431189e-01 -4.33189809e-01 6.85010672e-01 3.05473626e-01
3.13291132e-01 1.96994200e-01 -4.85404253e-01 -1.17469180e+00
-1.95297971e-01 -9.22897756e-01 4.35499221e-01 -6.49980545e-01
-1.61930096e+00 7.95880854e-01 -9.90509391e-02 -8.80131185e-01
-2.76141137e-01 -5.01691341e-01 -7.48273790e-01 1.02310848e+00
-1.49208915e+00 -1.48054492e+00 4.99242917e-02 7.94508576e-01
5.08715570e-01 -4.01150733e-01 9.29899693e-01 2.23066241e-01
-5.44321060e-01 1.21952140e+00 1.97307058e-02 2.45341137e-01
7.36301780e-01 -1.39781260e+00 6.95355773e-01 8.67700815e-01
1.28300652e-01 1.19895947e+00 5.83807588e-01 -5.49778998e-01
-1.04635811e+00 -1.50281858e+00 8.57519150e-01 -7.73184597e-01
1.05789638e+00 -5.89626908e-01 -1.02688396e+00 1.06279373e+00
1.24184348e-01 2.79279426e-02 6.49374068e-01 2.77846158e-01
-1.05687726e+00 -4.24351636e-03 -1.06764483e+00 7.70571351e-01
1.26342452e+00 -9.08942103e-01 -8.83424938e-01 2.86389768e-01
1.24045646e+00 -1.88311581e-02 -8.11822474e-01 6.28898263e-01
2.84567654e-01 -7.07049966e-01 1.13644779e+00 -1.10316539e+00
4.40139949e-01 1.52755678e-01 -2.62755174e-02 -1.34047866e+00
-7.56951690e-01 -8.66512060e-01 -2.10151210e-01 1.39684010e+00
5.28469145e-01 -6.48268163e-01 5.21014690e-01 4.20356989e-01
-2.68269300e-01 -8.55234206e-01 -6.45696580e-01 -8.82852793e-01
5.72832465e-01 -3.88438523e-01 8.56618285e-01 1.17801523e+00
-2.11434692e-01 6.92971289e-01 -2.21069649e-01 -8.75033159e-03
3.91896546e-01 -1.05664775e-01 4.21378344e-01 -1.02282608e+00
-2.92716503e-01 -5.82491934e-01 -9.14338976e-02 -1.20637774e+00
3.63651842e-01 -1.24268067e+00 -1.46770075e-01 -9.75650549e-01
1.16464965e-01 -6.30151272e-01 -7.98423767e-01 9.77038920e-01
-5.76279044e-01 2.69529045e-01 1.30126905e-02 1.68298706e-01
-5.67469060e-01 6.03027284e-01 1.19713223e+00 -2.92566806e-01
1.30874142e-01 3.15155797e-02 -1.31158614e+00 7.38307118e-01
8.35690558e-01 -5.99320233e-01 -5.62799931e-01 -6.66629970e-01
-7.48650283e-02 -6.01797163e-01 3.90076935e-01 -6.85993433e-01
6.04125932e-02 -1.03130154e-01 2.87428230e-01 -6.55370280e-02
1.35024607e-01 -6.95457935e-01 -4.28306013e-01 2.17981234e-01
-8.57484043e-01 6.57355860e-02 4.19380009e-01 5.44795632e-01
-2.08953246e-02 9.63020921e-02 9.01280940e-01 -2.58557290e-01
-5.46692789e-01 5.28726816e-01 -1.25455648e-01 4.17526156e-01
6.44825161e-01 2.33626608e-02 -4.59380448e-01 -3.07525396e-01
-8.97729814e-01 3.33249778e-01 4.34710830e-01 6.84505284e-01
2.36844018e-01 -1.31760907e+00 -5.48936605e-01 4.68923658e-01
3.70668590e-01 6.73750043e-02 5.37757166e-02 4.64749664e-01
1.37989983e-01 5.55808663e-01 3.30694660e-04 -4.09086198e-01
-7.65950799e-01 1.01613390e+00 2.51231372e-01 -4.43427622e-01
-5.05479693e-01 1.24711728e+00 8.32358301e-01 -5.99531591e-01
4.09481257e-01 -3.44437212e-01 3.48257236e-02 -2.81005979e-01
4.25725907e-01 -1.00882882e-02 2.73236372e-02 -4.74597096e-01
-4.40801531e-01 6.12202622e-02 -5.97885609e-01 2.94761956e-01
1.22110689e+00 1.39184371e-01 1.34142235e-01 1.75208047e-01
1.22739100e+00 1.83985621e-01 -1.31619883e+00 -5.95318615e-01
8.11127424e-02 -1.78165525e-01 1.74108788e-01 -1.25588453e+00
-9.82083380e-01 1.07186365e+00 2.31674790e-01 2.62495220e-01
8.85049403e-01 1.63875714e-01 6.90517247e-01 4.21208054e-01
1.93821386e-01 -6.84677005e-01 2.77866900e-01 6.95194185e-01
1.00116432e+00 -9.91579413e-01 -4.34255600e-01 -2.22990528e-01
-7.19560564e-01 6.40552878e-01 9.52438712e-01 -3.09259027e-01
4.74480420e-01 2.94414490e-01 9.72022209e-03 3.58802751e-02
-9.80622828e-01 -1.63802907e-01 4.29682434e-01 5.49082935e-01
2.54430145e-01 -2.21203547e-02 2.50872701e-01 1.10245728e+00
-3.13934296e-01 -1.68553099e-01 1.36652714e-04 5.98122478e-01
-2.98947453e-01 -8.94096196e-01 -8.89761150e-02 5.34593403e-01
-6.38161302e-01 -6.23596966e-01 -5.17977893e-01 8.15481246e-01
3.00973922e-01 6.73675895e-01 -1.77101661e-02 -5.56444347e-01
5.52477956e-01 2.38607928e-01 1.76934615e-01 -8.91094565e-01
-9.55817819e-01 -1.35603532e-01 -1.51185328e-02 -4.63926733e-01
1.51397169e-01 -1.50119379e-01 -9.37932909e-01 -2.87118256e-01
-3.79553795e-01 1.78383306e-01 2.38184765e-01 8.37080598e-01
4.53009725e-01 7.35321045e-01 6.28058791e-01 -4.84767169e-01
-1.30582547e+00 -1.15938210e+00 -2.99387962e-01 7.64978647e-01
3.98325056e-01 -7.88586915e-01 -5.17457902e-01 -2.53573358e-01]
|
[10.541921615600586, 8.078655242919922]
|
50770d07-99fc-4263-b537-b30b9fa64503
|
less-is-more-learning-highlight-detection
|
1903.00859
| null |
http://arxiv.org/abs/1903.00859v1
|
http://arxiv.org/pdf/1903.00859v1.pdf
|
Less is More: Learning Highlight Detection from Video Duration
|
Highlight detection has the potential to significantly ease video browsing,
but existing methods often suffer from expensive supervision requirements,
where human viewers must manually identify highlights in training videos. We
propose a scalable unsupervised solution that exploits video duration as an
implicit supervision signal. Our key insight is that video segments from
shorter user-generated videos are more likely to be highlights than those from
longer videos, since users tend to be more selective about the content when
capturing shorter videos. Leveraging this insight, we introduce a novel ranking
framework that prefers segments from shorter videos, while properly accounting
for the inherent noise in the (unlabeled) training data. We use it to train a
highlight detector with 10M hashtagged Instagram videos. In experiments on two
challenging public video highlight detection benchmarks, our method
substantially improves the state-of-the-art for unsupervised highlight
detection.
|
['Kristen Grauman', 'Deepti Ghadiyaram', 'Yannis Kalantidis', 'Bo Xiong']
|
2019-03-03
|
less-is-more-learning-highlight-detection-1
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Xiong_Less_Is_More_Learning_Highlight_Detection_From_Video_Duration_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Xiong_Less_Is_More_Learning_Highlight_Detection_From_Video_Duration_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['highlight-detection']
|
['computer-vision']
|
[ 4.14771229e-01 -1.24301910e-01 -7.33789802e-01 -2.78714001e-01
-7.94022739e-01 -7.49515831e-01 2.72882253e-01 3.08912098e-01
-3.89452934e-01 2.98260748e-01 4.99488801e-01 -3.44301313e-02
2.81238407e-01 -4.13523257e-01 -7.93638706e-01 -5.29280305e-01
-5.45097649e-01 -3.49241436e-01 5.20115793e-01 2.64205247e-01
4.40737814e-01 -2.73027178e-02 -1.68038428e+00 4.00587171e-01
6.46203458e-01 1.02947342e+00 1.14943653e-01 5.00791371e-01
2.21562371e-01 1.25671279e+00 -5.05480826e-01 -2.29089260e-01
3.87276262e-01 -2.59347230e-01 -4.00671989e-01 4.07295942e-01
1.13806260e+00 -7.10417628e-01 -6.63943708e-01 9.91508424e-01
1.83901176e-01 4.14900929e-01 3.64668101e-01 -1.33572519e+00
-4.11490023e-01 7.74416089e-01 -9.24345315e-01 7.06189692e-01
4.24197644e-01 -2.84057073e-02 1.33211565e+00 -7.90730894e-01
9.17995453e-01 8.26720119e-01 6.12743199e-01 1.38729379e-01
-1.08214056e+00 -7.04126418e-01 5.70357144e-01 2.05247357e-01
-1.33951747e+00 -4.64537650e-01 8.84731352e-01 -4.70709473e-01
4.22845840e-01 3.61433566e-01 5.75799406e-01 9.65616763e-01
-3.54677916e-01 1.20562541e+00 8.06879580e-01 -1.50029585e-01
3.82812396e-02 1.94273159e-01 -5.05776256e-02 6.60547733e-01
9.81603265e-02 -2.57593215e-01 -1.20467913e+00 -2.11656719e-01
5.55145502e-01 3.25922102e-01 -5.53663731e-01 -6.65972590e-01
-1.07367373e+00 8.35701644e-01 2.45181262e-01 -4.34546098e-02
-2.79009014e-01 1.13535941e-01 5.19481659e-01 3.17717791e-01
6.21499121e-01 3.74129415e-01 -2.16616988e-01 -4.11016017e-01
-1.49152982e+00 3.23013395e-01 5.09205341e-01 1.08435786e+00
8.48991096e-01 -2.52346784e-01 -3.72472972e-01 5.79093158e-01
2.11770069e-02 3.43391985e-01 1.48879334e-01 -9.26715851e-01
4.26344842e-01 5.10235071e-01 1.90695941e-01 -1.28324080e+00
7.68705532e-02 -2.24123254e-01 -4.69467044e-02 -1.71504870e-01
3.59530360e-01 5.83962817e-03 -8.47718418e-01 1.68859684e+00
1.97114021e-01 3.18747580e-01 -4.95030731e-01 9.95787680e-01
4.57567096e-01 5.81714749e-01 1.02017596e-01 -2.55915284e-01
1.21826005e+00 -1.16372442e+00 -5.41157782e-01 9.21466127e-02
4.98822927e-01 -7.07570314e-01 1.18358541e+00 4.91613209e-01
-9.19047236e-01 -3.96991313e-01 -9.25866306e-01 -5.98836653e-02
-1.28426015e-01 2.84781586e-02 7.91080058e-01 5.83289802e-01
-8.49629164e-01 4.07443047e-01 -5.54239273e-01 -4.33127433e-01
5.01740634e-01 -2.22622883e-02 -1.38048559e-01 4.88065295e-02
-8.31149638e-01 2.09632114e-01 1.81921914e-01 -3.94602090e-01
-1.15854967e+00 -1.01243567e+00 -7.67400205e-01 1.49415910e-01
9.19729829e-01 -8.17061886e-02 1.23817885e+00 -1.36303234e+00
-1.08209229e+00 9.58261788e-01 -2.76251495e-01 -6.57388210e-01
7.27524579e-01 -7.56253421e-01 -2.73082614e-01 7.88820326e-01
1.63271815e-01 7.39422917e-01 1.17133904e+00 -1.17947304e+00
-1.06470942e+00 7.50904009e-02 2.71853149e-01 6.97203726e-02
-7.62512565e-01 1.91617146e-01 -8.02874267e-01 -9.63548064e-01
-3.40177715e-01 -9.02334094e-01 6.14641421e-02 1.64090768e-01
-6.42223597e-01 -3.43081541e-02 1.32329679e+00 -5.37288547e-01
1.67024136e+00 -2.35611653e+00 -6.61519468e-02 4.45040435e-01
5.47528386e-01 3.19494829e-05 3.61079462e-02 4.63406920e-01
1.63664460e-01 -5.22748418e-02 1.53530657e-01 -2.23703146e-01
-9.37478617e-02 -2.01532692e-01 -5.49160838e-01 3.51530194e-01
1.52296543e-01 3.93338799e-01 -1.29061496e+00 -7.21000135e-01
-7.77701885e-02 3.54491621e-01 -7.78601229e-01 2.74930209e-01
-3.30998033e-01 2.86895365e-01 -3.75402182e-01 9.57577348e-01
4.95053649e-01 -4.14075702e-01 2.82612413e-01 -2.66957998e-01
-4.46305960e-01 1.82320520e-01 -9.51093554e-01 1.56179404e+00
6.99920813e-03 1.25054717e+00 -1.28336519e-01 -6.91199839e-01
3.76580745e-01 2.07749352e-01 6.57307863e-01 -4.53140885e-01
-4.34151106e-02 -1.52441069e-01 -4.56882238e-01 -6.89575911e-01
8.37283850e-01 4.47694302e-01 9.35200155e-02 4.26312089e-01
-1.59042373e-01 5.09609759e-01 5.00568748e-01 7.10746706e-01
1.31537127e+00 7.34698623e-02 1.14615634e-01 -4.03596796e-02
8.20279047e-02 -1.34334698e-01 6.31894529e-01 8.99411917e-01
-4.38159466e-01 4.82381910e-01 7.89323032e-01 -2.60797679e-01
-7.88080752e-01 -1.14973581e+00 2.59020001e-01 1.74471819e+00
1.90880314e-01 -8.66037488e-01 -6.67884707e-01 -1.04587984e+00
1.27169117e-01 1.80570364e-01 -7.68099844e-01 4.91983406e-02
-4.51536745e-01 -2.99970107e-03 2.74683684e-01 5.04329503e-01
1.59610137e-01 -7.78705418e-01 -8.72368515e-01 -4.69236225e-02
-2.28370443e-01 -1.26940274e+00 -9.97132838e-01 -6.11809306e-02
-6.50630295e-01 -1.19744802e+00 -8.66735995e-01 -7.41054058e-01
8.20846736e-01 9.99093831e-01 1.18786037e+00 4.04864669e-01
-3.20596695e-01 6.01875007e-01 -5.37421942e-01 -2.38080025e-01
5.16749360e-02 5.68730868e-02 -2.80187093e-02 2.40564317e-01
4.81890947e-01 -3.69735152e-01 -8.86678576e-01 3.58705878e-01
-9.71477509e-01 9.91410296e-03 4.09059227e-01 6.47724152e-01
5.95474303e-01 1.22598968e-01 1.71284243e-01 -1.19811726e+00
4.31646071e-02 -5.55626154e-01 -4.29208130e-01 1.25039164e-02
-4.26628411e-01 -4.01759952e-01 4.48050290e-01 -6.27600133e-01
-8.36654246e-01 7.23453909e-02 7.10721791e-01 -7.50085831e-01
7.28270859e-02 4.46170509e-01 4.30139247e-03 -1.69793174e-01
2.81822830e-01 -7.84535427e-03 -3.30250800e-01 -4.65199798e-01
2.74745464e-01 3.94175440e-01 6.14893436e-01 -4.88566518e-01
9.10938799e-01 7.72055805e-01 -3.45996380e-01 -1.02846563e+00
-1.04685783e+00 -9.93857205e-01 -3.25510472e-01 -5.98982215e-01
5.21494746e-01 -1.17786300e+00 -4.72110689e-01 2.45213062e-01
-4.89859164e-01 -3.33131701e-01 4.54872660e-03 2.59676844e-01
-3.75804693e-01 6.03605747e-01 -7.02274203e-01 -8.34623098e-01
-9.47978348e-02 -8.34867775e-01 1.11555004e+00 4.10605997e-01
-2.99572110e-01 -7.19686151e-01 -1.42387062e-01 1.51843876e-01
2.23944038e-01 4.24282849e-01 5.66301286e-01 -4.61666793e-01
-9.50220108e-01 -3.70497078e-01 -4.41546619e-01 1.38015941e-01
1.84269071e-01 4.64451790e-01 -9.94418144e-01 -3.66463959e-01
-5.01067698e-01 -2.98168808e-01 1.21924019e+00 2.76002705e-01
1.34887052e+00 -4.65321422e-01 -4.84651446e-01 4.92687613e-01
1.32606208e+00 -2.57146150e-01 4.05591547e-01 5.01134753e-01
8.54401350e-01 6.55924201e-01 1.14369428e+00 8.05655837e-01
3.38166744e-01 5.81907034e-01 3.63066375e-01 -1.07052609e-01
1.13057479e-01 -6.18922234e-01 7.29980111e-01 3.83108675e-01
-1.81990132e-01 -2.73814410e-01 -4.93279576e-01 6.64661705e-01
-1.82537162e+00 -1.37423587e+00 2.12917268e-01 2.40249372e+00
7.44396448e-01 4.78557348e-01 7.96968699e-01 1.20762385e-01
8.17650139e-01 5.81014633e-01 -5.14424622e-01 2.19205603e-01
-1.28739830e-02 -2.82794908e-02 7.02570677e-01 -3.45589407e-03
-1.61107934e+00 8.04196358e-01 6.58913755e+00 5.33700407e-01
-1.19285762e+00 -1.13692492e-01 6.11707270e-01 -6.65799677e-01
-2.10754752e-01 5.58505021e-02 -5.46248138e-01 8.62612903e-01
7.96963513e-01 -2.25329399e-02 2.10455999e-01 9.75610733e-01
4.85865086e-01 -3.78662854e-01 -1.21175575e+00 9.56930578e-01
3.43333185e-01 -1.28079104e+00 5.61421104e-02 -3.28711756e-02
9.69000638e-01 1.20644942e-02 4.72537607e-01 2.02121750e-01
-1.31132379e-01 -6.18921340e-01 9.54182744e-01 2.69001693e-01
6.62165165e-01 -7.09828496e-01 1.64364398e-01 -1.62168741e-01
-1.37622035e+00 -1.95080429e-01 -1.25741974e-01 -1.05897561e-02
1.63196445e-01 6.49558127e-01 -6.01405978e-01 1.92990774e-04
9.02310193e-01 1.18293500e+00 -7.08876967e-01 1.29102552e+00
-2.09631920e-01 8.84517193e-01 -2.56060719e-01 1.47501141e-01
2.44366467e-01 6.78588822e-02 5.78272223e-01 1.60921943e+00
1.68972090e-01 -2.45750606e-01 4.96893138e-01 3.96843970e-01
-4.39241111e-01 2.52928525e-01 -5.60691178e-01 -3.62463683e-01
5.31278789e-01 1.35614955e+00 -9.78374183e-01 -5.50719678e-01
-7.70172298e-01 9.41969573e-01 2.37285703e-01 4.58545357e-01
-1.06176281e+00 -3.18734288e-01 6.33386612e-01 3.26043695e-01
6.87741339e-01 -1.37866572e-01 4.87387329e-01 -1.32456946e+00
1.78627521e-01 -1.00201893e+00 5.70941389e-01 -5.81917048e-01
-1.00231326e+00 1.31770417e-01 -2.57327974e-01 -1.49401414e+00
4.64507714e-02 -4.09844339e-01 -7.05919862e-01 -3.00731901e-02
-1.60055423e+00 -8.49282444e-01 -6.07338548e-01 4.19188201e-01
6.32665336e-01 2.81666994e-01 2.28456020e-01 6.99112475e-01
-5.18043101e-01 6.48738623e-01 4.55523022e-02 2.44768918e-01
1.20400095e+00 -1.25481594e+00 5.23471683e-02 8.84657562e-01
4.13798362e-01 5.83610415e-01 8.77131402e-01 -6.05716288e-01
-1.48437381e+00 -9.66297269e-01 5.20800531e-01 -4.93869662e-01
8.74561310e-01 -2.00908765e-01 -6.86634421e-01 7.27182746e-01
1.48019344e-01 3.37111987e-02 1.09563959e+00 2.56852537e-01
-6.75551176e-01 5.23534454e-02 -7.35579967e-01 5.41785657e-01
1.06669176e+00 -9.07306850e-01 -2.61027455e-01 2.93156534e-01
5.10343790e-01 -3.10769290e-01 -6.12334669e-01 1.42516494e-01
8.02738309e-01 -1.01207685e+00 8.23646069e-01 -5.46000779e-01
6.03772998e-01 -3.85944843e-01 -6.73329085e-02 -8.83742332e-01
2.00637672e-02 -9.92932498e-01 -6.68087840e-01 1.26886964e+00
1.94710091e-01 3.43697444e-02 1.03205204e+00 5.18613696e-01
1.07147381e-01 -6.02674544e-01 -4.54759568e-01 -7.63233364e-01
-8.53100181e-01 -1.14633568e-01 1.34806082e-01 1.09160066e+00
1.36391491e-01 -4.60541202e-03 -8.58827114e-01 1.97627619e-01
6.49257839e-01 1.80412561e-01 9.60749686e-01 -1.16604960e+00
-3.37378502e-01 -3.93980771e-01 -5.63786507e-01 -1.34844923e+00
-2.24461332e-01 -3.15425396e-01 1.00863107e-01 -9.56070125e-01
6.97171390e-01 -6.29803613e-02 -8.17539155e-01 3.29284489e-01
-3.18935901e-01 6.37447953e-01 2.96704888e-01 2.16684714e-01
-1.41117656e+00 1.14601672e-01 8.37476969e-01 -1.75438561e-02
-2.79613376e-01 -7.66496658e-02 -6.13155544e-01 8.37777197e-01
3.99682909e-01 -4.37775731e-01 -4.34192508e-01 -1.88191101e-01
3.51733685e-01 -4.09272164e-01 3.12923223e-01 -1.07177877e+00
-3.76271829e-03 -8.14025849e-02 3.69290918e-01 -7.16705322e-01
1.50051400e-01 -6.84205234e-01 -2.97807872e-01 -1.90081098e-03
-6.75282955e-01 9.56825092e-02 1.95594952e-02 9.04441059e-01
-3.11167955e-01 6.08754344e-03 5.28313100e-01 1.00834392e-01
-9.54817057e-01 3.96160007e-01 -4.36442673e-01 2.02019542e-01
9.68882084e-01 -1.72725543e-01 -3.68251175e-01 -5.84047377e-01
-3.00474793e-01 2.37229779e-01 9.51445401e-01 4.23252046e-01
4.04626846e-01 -1.07690132e+00 -2.36508265e-01 -4.33873497e-02
4.44637597e-01 -4.30337191e-01 2.47377545e-01 9.73111689e-01
-3.62799317e-01 9.36031863e-02 -1.84740331e-02 -4.73292291e-01
-1.42855155e+00 8.35523486e-01 -3.10000300e-01 8.16940218e-02
-7.21642494e-01 1.00007474e+00 2.39849329e-01 5.67232192e-01
7.26968050e-01 -3.60963255e-01 -3.49529460e-02 4.30655688e-01
8.52323055e-01 4.47229296e-01 -4.34882820e-01 -6.11082733e-01
-2.82897174e-01 3.28759789e-01 -5.67316890e-01 1.77088976e-01
1.16598499e+00 -2.73957253e-01 3.64360660e-01 4.13474053e-01
1.32343996e+00 6.70949638e-01 -1.66770244e+00 -3.77156436e-01
3.40191126e-01 -8.97393107e-01 4.56222109e-02 -2.68887520e-01
-1.23832738e+00 5.09418666e-01 4.09082234e-01 4.93774801e-01
1.21501780e+00 -1.07377559e-01 1.02619028e+00 3.74017894e-01
3.25601190e-01 -1.36027229e+00 4.13423717e-01 1.40534341e-02
2.13574886e-01 -1.42362571e+00 2.23256335e-01 -6.14701450e-01
-7.60321915e-01 1.06605446e+00 4.91633803e-01 -2.46947482e-01
3.79235744e-01 -9.24237259e-03 1.28453359e-01 -2.80000776e-01
-6.74976528e-01 -2.06971332e-01 3.53583395e-01 2.69295752e-01
5.42130291e-01 -3.61979418e-02 2.47995593e-02 1.83634132e-01
1.67055681e-01 -1.14847153e-01 4.98120576e-01 1.14097786e+00
-5.61515331e-01 -6.70465827e-01 -9.37776640e-02 6.40809655e-01
-8.73677552e-01 -1.57240599e-01 -4.09525990e-01 6.47334099e-01
-8.70998204e-02 8.00581932e-01 1.12324387e-01 -2.47117043e-01
6.89700469e-02 -1.92443043e-01 2.65739560e-01 -5.32188654e-01
-3.99951071e-01 3.28115016e-01 8.53099748e-02 -9.28532660e-01
-7.02331126e-01 -8.18986833e-01 -9.54137981e-01 -1.90253049e-01
-2.37782568e-01 2.58761883e-01 2.14739680e-01 5.22727907e-01
4.46119875e-01 3.70695978e-01 7.72307813e-01 -8.97438645e-01
-1.40843064e-01 -6.27915502e-01 -6.37191236e-01 7.67930865e-01
6.31720662e-01 -8.74933898e-01 -4.94335413e-01 4.68770087e-01]
|
[10.152966499328613, 0.4836825728416443]
|
c79c225d-7044-431a-a8ae-f415054ce021
|
human-image-generation-a-comprehensive-survey
|
2212.08896
| null |
https://arxiv.org/abs/2212.08896v1
|
https://arxiv.org/pdf/2212.08896v1.pdf
|
Human Image Generation: A Comprehensive Survey
|
Image and video synthesis has become a blooming topic in computer vision and machine learning communities along with the developments of deep generative models, due to its great academic and application value. Many researchers have been devoted to synthesizing high-fidelity human images as one of the most commonly seen object categories in daily lives, where a large number of studies are performed based on various deep generative models, task settings and applications. Thus, it is necessary to give a comprehensive overview on these variant methods on human image generation. In this paper, we divide human image generation techniques into three paradigms, i.e., data-driven methods, knowledge-guided methods and hybrid methods. For each route, the most representative models and the corresponding variants are presented, where the advantages and characteristics of different methods are summarized in terms of model architectures and input/output requirements. Besides, the main public human image datasets and evaluation metrics in the literature are also summarized. Furthermore, due to the wide application potentials, two typical downstream usages of synthesized human images are covered, i.e., data augmentation for person recognition tasks and virtual try-on for fashion customers. Finally, we discuss the challenges and potential directions of human image generation to shed light on future research.
|
['Tieniu Tan', 'Liang Wang', 'Zhang Zhang', 'Zhen Jia']
|
2022-12-17
| null | null | null | null |
['person-recognition', 'virtual-try-on']
|
['computer-vision', 'computer-vision']
|
[ 4.51729745e-01 3.12665962e-02 -9.39325765e-02 -3.27021599e-01
-3.13511491e-01 -1.60839781e-01 8.90512347e-01 -5.29260576e-01
-2.60902315e-01 7.18061328e-01 1.64858684e-01 2.24418879e-01
1.40294269e-01 -8.73789907e-01 -5.06067395e-01 -8.74665916e-01
4.46234822e-01 3.75539452e-01 -2.75694042e-01 -3.05404961e-01
-2.48297211e-02 2.85573214e-01 -2.03294063e+00 3.68989378e-01
8.02912474e-01 1.00028801e+00 1.32998407e-01 5.96679807e-01
7.03454688e-02 7.89081454e-01 -9.60281491e-01 -1.04358447e+00
8.54430422e-02 -9.93758321e-01 -4.27007705e-01 7.55231380e-01
3.84842634e-01 -2.97027320e-01 -7.59788752e-01 1.01352060e+00
8.98498416e-01 1.54938608e-01 6.77150011e-01 -1.71439421e+00
-1.19016612e+00 5.07682920e-01 -3.41294110e-01 1.90605536e-01
4.09673989e-01 4.33296651e-01 5.58734834e-01 -9.91371095e-01
6.93474591e-01 1.16671205e+00 3.62302154e-01 8.25441182e-01
-8.83612573e-01 -4.89363104e-01 2.51611292e-01 6.31356597e-01
-1.28100204e+00 -4.83825862e-01 9.28075790e-01 -5.71567774e-01
7.25794017e-01 3.53610754e-01 1.01735795e+00 1.52164650e+00
-1.22417416e-02 1.22162604e+00 1.00863600e+00 -5.27374268e-01
4.03390415e-02 3.99362117e-01 -7.35705793e-02 4.71250594e-01
3.68101984e-01 2.67624199e-01 -6.02631569e-01 3.05216342e-01
9.34079111e-01 -4.38350402e-02 -2.35923529e-01 -2.11669490e-01
-1.26254189e+00 8.60940337e-01 3.53173792e-01 2.14868113e-01
-4.30307657e-01 -1.21620901e-01 3.08531702e-01 -4.33078744e-02
2.81068206e-01 1.04071438e-01 3.95110726e-01 5.28018326e-02
-9.22968864e-01 7.95240104e-01 3.47832561e-01 1.35889781e+00
3.35570902e-01 5.87465823e-01 -4.89346027e-01 9.59303677e-01
2.18573302e-01 6.11535847e-01 6.36098325e-01 -5.88527858e-01
2.54565567e-01 5.06128550e-01 1.37926489e-02 -1.24522591e+00
-2.31781349e-01 -4.15381014e-01 -1.38108301e+00 -7.90993050e-02
1.26901671e-01 -7.52486661e-02 -1.13581014e+00 1.45150471e+00
1.12458877e-01 -3.96593800e-03 9.27731767e-02 1.13466477e+00
1.58428502e+00 7.53898799e-01 8.37586969e-02 -1.51562095e-01
1.57840741e+00 -1.19123316e+00 -9.95362043e-01 -2.96742320e-01
-6.16696961e-02 -7.49515176e-01 8.82669210e-01 4.31177258e-01
-1.24241281e+00 -1.06620550e+00 -9.33125615e-01 -1.12253822e-01
-3.84947628e-01 3.55768412e-01 7.32710242e-01 9.51440036e-01
-9.22046363e-01 2.15432376e-01 -6.04210556e-01 -5.04123926e-01
4.79069382e-01 -7.25713279e-03 -3.49099040e-02 -1.39080644e-01
-1.29617846e+00 5.42040944e-01 3.43972266e-01 2.49164447e-01
-8.68897438e-01 -4.04802561e-01 -9.28699195e-01 -2.90510714e-01
1.06230095e-01 -9.89500105e-01 1.26958072e+00 -9.20031905e-01
-1.38407755e+00 1.04046142e+00 -1.79049447e-02 -5.69644153e-01
8.49310994e-01 -5.24409227e-02 -6.99503183e-01 1.39472410e-02
-6.72152862e-02 8.55246723e-01 9.09349322e-01 -1.26534307e+00
-8.12738359e-01 -2.35150233e-01 -4.10289466e-02 3.18565875e-01
-4.09133971e-01 4.80026901e-02 -8.51331651e-01 -1.03316820e+00
-2.90576488e-01 -9.79437649e-01 -2.82068670e-01 -3.21592391e-01
-6.87474489e-01 -1.88164845e-01 6.73314512e-01 -6.19332910e-01
1.47854304e+00 -1.92690289e+00 2.68733382e-01 6.20722678e-03
2.43930668e-01 4.29371625e-01 3.52719091e-02 4.23934519e-01
4.73402217e-02 -5.54772429e-02 -1.22847058e-01 -3.91825438e-01
9.06677246e-02 5.22795655e-02 -3.99945587e-01 2.32901916e-01
1.36471763e-01 1.33786452e+00 -6.15668833e-01 -3.78080606e-01
5.94742596e-01 7.93771744e-01 -3.06519389e-01 2.92088419e-01
1.05625275e-03 6.21966362e-01 -2.40384400e-01 6.85345650e-01
5.26399374e-01 -3.16360772e-01 1.08294897e-02 -4.19209927e-01
1.12822197e-01 -1.34263352e-01 -1.15973508e+00 1.32688439e+00
-1.30228341e-01 6.36300683e-01 -3.78274977e-01 -8.95982683e-01
1.00446367e+00 3.70743215e-01 4.15633589e-01 -8.81677270e-01
2.20045537e-01 1.40935685e-02 -2.75793374e-02 -5.08186281e-01
8.02401364e-01 -1.51390629e-02 -1.15733638e-01 1.63585767e-01
2.03417968e-02 1.54431723e-02 7.12627351e-01 6.93540722e-02
4.83862430e-01 1.93265334e-01 3.63525271e-01 2.33486250e-01
5.65422535e-01 9.81903225e-02 4.10326213e-01 6.43723845e-01
-2.91466087e-01 9.85310495e-01 9.02555734e-02 -5.20165622e-01
-1.29563618e+00 -9.77408826e-01 3.56117859e-02 8.30038071e-01
3.06603760e-01 -3.38334054e-01 -1.09783196e+00 -2.84396380e-01
-2.46217296e-01 5.72291374e-01 -6.57917619e-01 -2.61858582e-01
-7.18688488e-01 -1.17424655e+00 4.94651139e-01 6.84970737e-01
1.07482028e+00 -1.69258869e+00 -6.19808257e-01 3.89191583e-02
-5.39135218e-01 -1.22733295e+00 -1.93614021e-01 -6.20270431e-01
-6.42118633e-01 -8.78631353e-01 -1.32395470e+00 -9.60524738e-01
6.51490510e-01 3.94300640e-01 1.24336481e+00 -2.78629810e-02
-4.93319869e-01 5.42100787e-01 -5.83929956e-01 -4.82006401e-01
-4.74206150e-01 -6.82212189e-02 1.21735167e-02 2.06920400e-01
4.70732510e-01 -1.22059233e-01 -8.03486407e-01 5.51685095e-01
-9.34700787e-01 5.75341523e-01 6.99938178e-01 8.89455914e-01
7.49063909e-01 9.35404934e-03 5.25045574e-01 -8.45007360e-01
6.63562298e-01 -2.17330143e-01 -3.34631205e-01 3.03971529e-01
-4.59696949e-01 -4.68122512e-01 4.34859723e-01 -3.98818135e-01
-1.36036706e+00 1.63117647e-02 -1.83690131e-01 -3.45639139e-01
-3.59801948e-01 3.14732313e-01 -4.51321572e-01 2.34845132e-01
8.10426056e-01 6.35884106e-01 -7.04084663e-03 -2.94363767e-01
5.35666466e-01 6.49631083e-01 7.86487758e-01 -2.89180577e-01
6.72311187e-01 3.38835746e-01 -4.48633313e-01 -1.02555645e+00
-6.69753134e-01 -9.28425938e-02 -4.56718862e-01 -5.83873093e-01
1.03493571e+00 -8.00746322e-01 -4.50558007e-01 1.01492667e+00
-1.16394317e+00 -1.68348730e-01 -3.70313287e-01 2.89874941e-01
-8.07020724e-01 2.29693040e-01 -7.39740431e-01 -7.92229652e-01
-4.23203856e-01 -1.48696887e+00 9.58847463e-01 6.34482563e-01
-2.43041217e-02 -7.75058031e-01 -2.71035284e-01 6.19551718e-01
3.01278710e-01 4.33714956e-01 5.37568569e-01 -2.55723983e-01
-7.05663085e-01 -3.14038783e-01 -1.85276568e-01 3.74925673e-01
1.88617930e-02 -1.34369820e-01 -1.00621676e+00 -2.36113548e-01
-2.31300041e-01 -2.59516448e-01 7.12948740e-01 6.24052823e-01
1.20028162e+00 -1.94394782e-01 -3.43599737e-01 6.60313964e-01
1.08343327e+00 5.07364988e-01 1.18091631e+00 3.13505977e-01
7.90694535e-01 6.46943331e-01 5.60409367e-01 5.10668337e-01
2.95941532e-01 8.96644473e-01 5.08722812e-02 -2.41764233e-01
-5.43053448e-01 -4.46965039e-01 1.51594624e-01 7.00468123e-01
-5.97735286e-01 -6.83783472e-01 -5.92759013e-01 2.64813095e-01
-1.87525678e+00 -1.24248791e+00 -2.19181255e-01 2.11899710e+00
4.88171935e-01 -6.07361048e-02 5.88119149e-01 2.91223854e-01
1.12919283e+00 1.53793558e-01 -5.51288426e-01 7.58206546e-02
-5.27459204e-01 1.48611488e-02 4.87754904e-02 -9.99279544e-02
-1.28188705e+00 8.84731948e-01 6.48811817e+00 8.53406906e-01
-8.39443445e-01 -5.32861054e-02 1.04411197e+00 8.34737569e-02
2.22522356e-02 -5.56938469e-01 -1.12043226e+00 6.33548260e-01
6.61182940e-01 -2.27224484e-01 3.47274095e-01 9.24237192e-01
1.02469884e-01 7.03075752e-02 -9.47906494e-01 1.55522656e+00
5.51049173e-01 -1.42455673e+00 3.96739334e-01 1.31418362e-01
8.99175107e-01 -5.89766920e-01 4.22980428e-01 3.30539882e-01
-1.32562146e-02 -1.06229436e+00 1.01422977e+00 5.30533969e-01
9.27632034e-01 -9.17876244e-01 6.41116202e-01 1.53155953e-01
-1.16446257e+00 -4.87483246e-03 -2.95648783e-01 6.54056445e-02
7.14645684e-01 4.94874507e-01 -2.27765888e-01 4.98727441e-01
8.34277928e-01 7.30192721e-01 -6.12568796e-01 1.10416198e+00
-3.16136241e-01 4.20646489e-01 1.14958353e-01 -3.42126414e-02
-1.12438679e-01 -2.31991589e-01 2.85132617e-01 1.21766043e+00
4.27613467e-01 1.94014519e-01 4.28969897e-02 1.04809320e+00
2.72405427e-02 -5.83786592e-02 -5.18919766e-01 -2.49708489e-01
2.51308590e-01 1.27502298e+00 -7.68044353e-01 -5.63140273e-01
-4.86317456e-01 1.17915714e+00 -2.17961162e-01 3.87118250e-01
-1.09393799e+00 -3.98483187e-01 5.26798487e-01 2.59226531e-01
-3.14723738e-02 -5.38371503e-02 -2.82189667e-01 -1.16890383e+00
-9.26454272e-03 -1.22032261e+00 3.97304147e-01 -7.85911679e-01
-1.26231611e+00 9.67625022e-01 2.75584549e-01 -1.43227911e+00
-5.68846464e-01 -6.03318930e-01 -3.01044077e-01 7.61215150e-01
-1.12112355e+00 -1.23240554e+00 -8.78901601e-01 5.41430414e-01
8.96204352e-01 -5.23990631e-01 5.61921954e-01 5.44502974e-01
-8.50867987e-01 8.15691471e-01 -2.97214746e-01 3.51536334e-01
4.17251468e-01 -7.67871797e-01 8.02209139e-01 9.20842350e-01
2.73740917e-01 4.29823965e-01 6.39411271e-01 -6.66127980e-01
-1.23808539e+00 -1.20975602e+00 8.36082041e-01 -2.49215990e-01
8.58065393e-03 -4.65010107e-01 -6.19020045e-01 6.58798635e-01
4.47808117e-01 -1.70078412e-01 5.37417233e-01 -4.43808228e-01
1.32918581e-01 -1.89631898e-02 -1.03245366e+00 9.17463183e-01
1.34203303e+00 2.92750876e-02 -5.90528771e-02 2.85331011e-01
3.23388547e-01 -6.33583009e-01 -5.63399732e-01 3.57601047e-01
4.43161100e-01 -1.31438887e+00 1.15642178e+00 -4.20124501e-01
6.17001355e-01 -2.36445338e-01 3.15896980e-02 -1.15919042e+00
-5.64523935e-01 -6.06629789e-01 -3.27162296e-01 1.28919637e+00
1.01110727e-01 -3.62925619e-01 9.92278397e-01 6.13968313e-01
-1.84569910e-01 -6.89000547e-01 -3.38502735e-01 -5.34797907e-01
-2.87741423e-01 -4.37264442e-01 5.92850089e-01 5.58317840e-01
-4.66649264e-01 4.84596252e-01 -9.93197799e-01 -3.01512569e-01
7.29088366e-01 5.61509244e-02 1.13773715e+00 -8.55187595e-01
-2.72958189e-01 -5.70800900e-01 -7.07762539e-01 -1.16617548e+00
-2.54805982e-01 -7.71565020e-01 -1.02000214e-01 -1.72955203e+00
4.29561079e-01 -8.07338357e-02 4.30047065e-02 1.37378111e-01
-2.90370226e-01 7.40908265e-01 4.38736975e-01 2.33210787e-01
-4.01086390e-01 5.64129770e-01 1.42165244e+00 -2.35908717e-01
-1.21298425e-01 1.92693546e-01 -8.68194580e-01 6.49463773e-01
7.65950263e-01 2.32218727e-01 -6.20726287e-01 -3.98636103e-01
3.86512019e-02 -1.52617157e-01 7.01069534e-01 -1.14242601e+00
1.42205670e-01 -1.52009696e-01 7.59287655e-01 -7.41941154e-01
4.89450097e-01 -4.73167926e-01 4.68346953e-01 4.30488020e-01
-2.95240730e-01 1.58794016e-01 -1.47328272e-01 4.50196892e-01
-4.46296245e-01 -3.47764045e-02 1.05934310e+00 -2.79488504e-01
-1.20290613e+00 5.11224687e-01 -2.45983705e-01 -1.29414443e-02
1.24655974e+00 -5.09436309e-01 -1.30722195e-01 -7.23539233e-01
-8.11022997e-01 -1.40201822e-01 6.36809319e-02 8.15044820e-01
9.73643124e-01 -1.77163076e+00 -9.56140876e-01 3.33100796e-01
2.87583202e-01 -2.37740815e-01 8.21484685e-01 5.82572639e-01
-4.34997976e-01 5.83151698e-01 -3.75851065e-01 -6.17168367e-01
-1.18121088e+00 7.80221939e-01 2.18602598e-01 -4.02498730e-02
-7.06840098e-01 7.94363618e-01 6.33665204e-01 -8.20453092e-02
2.64209926e-01 1.39866501e-01 -3.90461594e-01 -8.45796093e-02
8.31184208e-01 5.80425203e-01 1.07627146e-01 -1.00523198e+00
-4.16996963e-02 2.88357764e-01 -4.81598303e-02 3.30458246e-02
1.00561142e+00 -2.00971156e-01 3.63301337e-01 1.07037313e-01
7.39201069e-01 -6.13886356e-01 -1.15700102e+00 -1.88874558e-01
-4.90368158e-01 -5.47372699e-01 -2.86940515e-01 -7.88806260e-01
-1.46455598e+00 9.09242272e-01 6.94310427e-01 1.59445465e-01
1.27791321e+00 -3.06326267e-03 8.60850394e-01 -5.45445010e-02
5.34417391e-01 -1.15298724e+00 2.52840191e-01 1.15386412e-01
1.17526650e+00 -1.13392830e+00 -4.18274291e-02 -5.84794998e-01
-1.08628082e+00 7.12901294e-01 8.29026937e-01 7.32650794e-03
3.45611334e-01 1.45405647e-04 4.93729822e-02 4.98709381e-02
-2.37974048e-01 -2.85230964e-01 4.28477496e-01 1.04159820e+00
6.17095411e-01 5.78567609e-02 -3.61545295e-01 7.40244329e-01
-5.58396280e-01 1.38873458e-01 1.97319701e-01 6.16552472e-01
-1.99324653e-01 -1.10080969e+00 -5.61089098e-01 5.24467528e-01
-2.18946159e-01 8.65231678e-02 -1.84132010e-01 8.26402664e-01
2.90293813e-01 1.02005827e+00 3.24881896e-02 -4.97997671e-01
6.82732940e-01 -2.47777671e-01 7.33978748e-01 -4.27598894e-01
-3.12632859e-01 1.93133093e-02 -1.11890063e-01 -2.30660290e-01
-5.32359898e-01 -5.93833625e-01 -8.42776358e-01 -4.65446442e-01
-1.04574271e-01 -1.65436924e-01 4.51836199e-01 6.62982166e-01
3.67778689e-01 6.35747850e-01 3.32834691e-01 -1.14849782e+00
-1.60201922e-01 -8.62834275e-01 -5.24810970e-01 6.93391263e-01
-2.70541161e-01 -8.50846112e-01 9.88910943e-02 6.04211152e-01]
|
[12.015395164489746, -0.8162513971328735]
|
963408d3-00db-484a-8066-b43e791a7424
|
fewrel-20-towards-more-challenging-few-shot
|
1910.07124
| null |
https://arxiv.org/abs/1910.07124v1
|
https://arxiv.org/pdf/1910.07124v1.pdf
|
FewRel 2.0: Towards More Challenging Few-Shot Relation Classification
|
We present FewRel 2.0, a more challenging task to investigate two aspects of few-shot relation classification models: (1) Can they adapt to a new domain with only a handful of instances? (2) Can they detect none-of-the-above (NOTA) relations? To construct FewRel 2.0, we build upon the FewRel dataset (Han et al., 2018) by adding a new test set in a quite different domain, and a NOTA relation choice. With the new dataset and extensive experimental analysis, we found (1) that the state-of-the-art few-shot relation classification models struggle on these two aspects, and (2) that the commonly-used techniques for domain adaptation and NOTA detection still cannot handle the two challenges well. Our research calls for more attention and further efforts to these two real-world issues. All details and resources about the dataset and baselines are released at https: //github.com/thunlp/fewrel.
|
['Jie zhou', 'Tianyu Gao', 'Peng Li', 'Maosong Sun', 'Hao Zhu', 'Zhiyuan Liu', 'Xu Han']
|
2019-10-16
|
fewrel-20-towards-more-challenging-few-shot-1
|
https://aclanthology.org/D19-1649
|
https://aclanthology.org/D19-1649.pdf
|
ijcnlp-2019-11
|
['few-shot-relation-classification', 'few-shot-relation-classification']
|
['methodology', 'natural-language-processing']
|
[ 1.24669977e-01 2.46984974e-01 -6.67307317e-01 -3.15233260e-01
-6.74704671e-01 -4.75719154e-01 8.28246474e-01 2.27088600e-01
-1.35780409e-01 8.92646790e-01 6.33234531e-02 -2.63556600e-01
-1.43488988e-01 -8.02145302e-01 -3.93777132e-01 -2.32308462e-01
3.08536515e-02 9.01680708e-01 7.27432966e-01 -6.28359795e-01
-2.12072834e-01 9.03340802e-02 -1.61909914e+00 3.42128247e-01
7.50781238e-01 9.09384608e-01 -9.58430246e-02 4.10167307e-01
9.53083113e-02 9.51221704e-01 -3.34762782e-01 -6.97677612e-01
1.33544385e-01 -5.43199837e-01 -1.48485315e+00 -2.12628484e-01
2.12278992e-01 -2.24754781e-01 -6.19979203e-01 8.91882837e-01
3.56189400e-01 4.25738364e-01 6.90891206e-01 -1.39879775e+00
-9.84034836e-01 9.37587321e-01 -4.75287378e-01 6.62500262e-01
5.02528369e-01 9.63677317e-02 1.23278058e+00 -7.86513865e-01
1.04450572e+00 1.02706504e+00 7.32291877e-01 4.22859073e-01
-1.19840884e+00 -7.64656067e-01 8.38536769e-02 6.73490763e-01
-1.53538394e+00 -6.94729567e-01 5.57297111e-01 -5.11343479e-01
1.46050894e+00 2.92961955e-01 4.59564269e-01 1.51017678e+00
-2.10131392e-01 5.72256982e-01 8.43052864e-01 -5.57981730e-01
2.41151914e-01 6.40958473e-02 6.73420429e-01 2.79624552e-01
2.68389016e-01 2.82759033e-03 -4.39954787e-01 -2.21182182e-01
3.22595417e-01 -2.62087107e-01 -1.71939746e-01 -2.57713646e-01
-1.03760207e+00 8.00707519e-01 1.83633044e-01 6.76887989e-01
-7.44742006e-02 -4.67725992e-01 4.43704098e-01 5.34008682e-01
6.57351613e-01 7.23212421e-01 -7.44060755e-01 -3.58862638e-01
-4.52226758e-01 3.00240666e-01 1.04229903e+00 1.28266966e+00
7.74376214e-01 -4.19698685e-01 -3.03723097e-01 1.07271004e+00
-3.70175123e-01 -2.41013721e-01 5.01571238e-01 -7.04600692e-01
5.50154865e-01 5.19032598e-01 -4.06452604e-02 -7.73391068e-01
-4.84267682e-01 -1.01045921e-01 -5.39187789e-01 -2.45191172e-01
5.83658397e-01 4.10845987e-02 -7.55017698e-01 1.62721944e+00
3.88850749e-01 3.66741598e-01 8.63676071e-02 6.96842849e-01
1.30533600e+00 2.42304102e-01 8.18403512e-02 -3.46776724e-01
1.59141684e+00 -1.08873940e+00 -9.00527000e-01 -6.54079378e-01
9.21347082e-01 -8.00083637e-01 1.15773642e+00 -2.51466837e-02
-8.37309361e-01 -6.02043927e-01 -1.17221475e+00 -2.41365537e-01
-6.92859292e-01 -2.93081492e-01 1.00745273e+00 3.44226629e-01
-5.96862555e-01 6.19822025e-01 -6.12474561e-01 -1.08851457e+00
4.67252761e-01 -1.89802870e-01 -4.68046308e-01 -2.81173348e-01
-1.89050484e+00 1.33039832e+00 4.40701395e-01 -4.22013074e-01
-4.51215923e-01 -7.25276291e-01 -9.80650246e-01 -3.05045079e-02
1.18497109e+00 -4.21258301e-01 1.44134235e+00 -3.11596483e-01
-1.12510562e+00 1.24139619e+00 -4.81807552e-02 -3.56766909e-01
2.11061567e-01 -4.56458330e-01 -6.75324202e-01 -3.72625701e-02
3.48475158e-01 5.57524040e-02 3.50571096e-01 -8.21753800e-01
-5.71340144e-01 -1.62162453e-01 1.02964863e-01 -7.70419836e-02
-1.69663519e-01 3.58173013e-01 -3.05038482e-01 -6.52709484e-01
-1.91342413e-01 -7.40603030e-01 1.89142331e-01 -3.20189267e-01
-4.58870262e-01 -6.63393855e-01 6.82509124e-01 -3.18199664e-01
1.27386522e+00 -1.99533308e+00 -9.24776196e-02 -3.96491706e-01
2.73871392e-01 6.50736988e-01 -2.67634362e-01 6.08218551e-01
-5.34718812e-01 1.62747964e-01 -8.13728943e-02 -1.20289698e-01
-2.89087623e-01 1.90995336e-01 -2.57512510e-01 2.83164650e-01
4.41999912e-01 1.15820074e+00 -1.27520490e+00 -4.80152935e-01
1.05834268e-02 4.19886261e-02 -9.76884738e-02 3.57563585e-01
-1.33768722e-01 1.29453778e-01 -2.99732834e-01 8.66016388e-01
5.16284943e-01 -3.92555386e-01 3.01053494e-01 -2.34644875e-01
1.31481051e-01 7.00100005e-01 -9.14666533e-01 1.62185311e+00
6.92899451e-02 6.60592973e-01 -3.72522116e-01 -1.23504853e+00
8.88203919e-01 3.65387648e-01 4.99636948e-01 -3.96818846e-01
4.14758652e-01 7.78058264e-03 3.10987800e-01 -7.86798418e-01
3.86593223e-01 -1.35894671e-01 -2.28755012e-01 4.11144763e-01
4.94442731e-01 2.92376205e-02 4.51662630e-01 3.10100019e-01
1.68205118e+00 2.18046363e-02 9.59936738e-01 -9.11397487e-02
-1.24172447e-03 1.04418591e-01 9.81097400e-01 8.15450728e-01
-7.77319014e-01 6.08562529e-01 7.39660740e-01 -4.72615540e-01
-8.00688744e-01 -7.27413833e-01 -2.96047926e-01 1.34652019e+00
1.52700424e-01 -7.40108013e-01 -1.91268668e-01 -9.00110662e-01
1.05016530e-01 9.56097424e-01 -9.45936859e-01 -3.18505675e-01
-3.10334623e-01 -5.82075596e-01 5.95377386e-01 4.58583891e-01
3.65374982e-01 -9.82371271e-01 -3.61632586e-01 1.48629516e-01
-4.70351011e-01 -1.52453458e+00 -1.69174552e-01 3.85346621e-01
-3.30612242e-01 -1.36763835e+00 -3.38902891e-01 -6.22725964e-01
1.99234728e-02 4.98703688e-01 1.45126569e+00 4.91460860e-02
-3.26519310e-01 2.88729393e-03 -1.00073612e+00 -2.86632001e-01
-3.73859376e-01 3.18417758e-01 8.74996930e-03 -4.46910203e-01
1.02862906e+00 -8.93871963e-01 -6.30528331e-02 4.07706708e-01
-4.28536087e-01 -1.67949080e-01 3.34379762e-01 7.36133218e-01
2.53321290e-01 1.74006581e-01 7.27955878e-01 -1.38856661e+00
7.32323706e-01 -9.22743618e-01 -2.05018714e-01 4.27617490e-01
-5.68134069e-01 -3.06420207e-01 3.14353913e-01 -8.11418056e-01
-8.71567905e-01 -2.87649184e-01 3.03310268e-02 -4.29361284e-01
-3.33525181e-01 5.10450304e-01 -2.40438536e-01 3.00265104e-01
1.15980506e+00 -3.36649001e-01 -2.86934137e-01 -5.64583302e-01
5.17722666e-01 7.27266192e-01 4.01975960e-01 -5.54732025e-01
7.46051133e-01 1.72867790e-01 -2.20409751e-01 -7.25099862e-01
-1.51798642e+00 -7.16142416e-01 -1.14313626e+00 7.43257925e-02
6.82490170e-01 -8.33882451e-01 -2.22487077e-01 3.43417913e-01
-1.18380940e+00 -5.81516564e-01 -5.17092884e-01 2.43806332e-01
-3.52457970e-01 3.37778330e-01 -7.49737442e-01 -7.59549856e-01
-2.59011596e-01 -7.62708426e-01 7.20185161e-01 2.15143844e-01
-7.92868853e-01 -7.84433186e-01 3.03850830e-01 5.26487291e-01
1.88164353e-01 2.89293170e-01 8.22466731e-01 -1.24054146e+00
1.54596940e-01 -1.22553639e-01 -2.61266649e-01 -1.15195505e-01
3.11098486e-01 -1.21173158e-01 -1.07941580e+00 1.16681112e-02
1.78563073e-02 -9.19897616e-01 9.99617696e-01 -6.57126084e-02
8.09181213e-01 -5.08440025e-02 -5.69768548e-01 4.67161745e-01
7.73320615e-01 2.50909925e-01 5.17639220e-01 3.83206934e-01
5.19055426e-01 5.91232240e-01 1.08068061e+00 2.12807134e-01
8.01396251e-01 9.06934798e-01 5.30691631e-02 1.76726133e-01
-3.47278744e-01 -1.33543104e-01 -9.04067382e-02 6.42073750e-01
-2.12545499e-01 -1.63964942e-01 -1.21292675e+00 8.51112783e-01
-2.24433851e+00 -1.15808439e+00 -2.75296196e-02 1.79187095e+00
1.19896162e+00 2.60539949e-01 2.74583489e-01 1.36939213e-01
8.12204838e-01 2.85372764e-01 -6.37439787e-01 -2.75584549e-01
-2.65584052e-01 3.60210419e-01 -5.93833402e-02 2.42337167e-01
-1.41123474e+00 1.36626911e+00 6.13610363e+00 8.62552762e-01
-6.16186440e-01 3.81087542e-01 4.48386520e-01 3.38981077e-02
6.53201416e-02 4.02471006e-01 -7.94893861e-01 2.09121495e-01
8.24765742e-01 -2.46298268e-01 4.19634342e-01 9.25662041e-01
-6.85635388e-01 -1.11812375e-01 -1.38009131e+00 7.67758489e-01
1.19070657e-01 -1.07943451e+00 -3.13405275e-01 -2.53869951e-01
5.32472730e-01 9.59616154e-02 -4.36479449e-01 1.00655520e+00
5.27909040e-01 -9.78059709e-01 3.19979876e-01 2.59312153e-01
7.87222087e-01 -3.35194379e-01 9.25995708e-01 3.25876355e-01
-1.25627995e+00 1.10029429e-01 -4.71242666e-01 -5.15142322e-01
1.26352385e-01 5.55924296e-01 -6.16803348e-01 5.74223995e-01
8.84420455e-01 1.07513654e+00 -7.27423787e-01 6.51360691e-01
-4.58993673e-01 6.10688925e-01 -1.55453607e-01 7.05934837e-02
-2.60610819e-01 1.37922704e-01 6.29804790e-01 1.15238345e+00
4.98924665e-02 5.89176774e-01 1.52773872e-01 8.62285137e-01
-5.16522164e-03 -1.06565133e-01 -7.08751738e-01 -2.51294881e-01
8.69911492e-01 1.31110549e+00 -6.52385712e-01 -3.31199944e-01
-6.56736553e-01 7.38481700e-01 9.63956714e-01 2.85408288e-01
-6.69069350e-01 -4.45242643e-01 8.54951084e-01 1.93432897e-01
2.44652748e-01 1.12723120e-01 -1.72411546e-01 -1.46034300e+00
-9.76950079e-02 -8.18035066e-01 9.05191362e-01 -6.47189856e-01
-1.85528231e+00 6.23563826e-01 2.57054567e-01 -9.25312698e-01
-3.11666965e-01 -4.34173077e-01 -6.59790874e-01 5.26211679e-01
-1.39549446e+00 -1.25602925e+00 -2.67511129e-01 4.41354632e-01
5.49636781e-01 -2.81141639e-01 1.05309975e+00 2.83901364e-01
-8.46233606e-01 7.27745175e-01 -4.91371363e-01 3.36403102e-01
1.23166680e+00 -9.51189160e-01 7.42169797e-01 7.10435390e-01
1.43439978e-01 5.33184528e-01 9.02788341e-01 -7.25965738e-01
-1.08226991e+00 -8.76129746e-01 1.21596801e+00 -9.40509498e-01
1.17527437e+00 -5.34009933e-01 -1.33357954e+00 1.09603858e+00
3.58429521e-01 2.26093799e-01 9.56880808e-01 9.31612134e-01
-7.61065722e-01 2.41099626e-01 -1.03849185e+00 4.89386946e-01
1.49151385e+00 -4.22582567e-01 -1.04256368e+00 4.69528556e-01
8.86319160e-01 -3.07419688e-01 -9.95616138e-01 7.98556328e-01
4.35310930e-01 -9.47350979e-01 8.95520568e-01 -1.01140463e+00
5.67920208e-01 1.80651501e-01 -2.32939377e-01 -1.26179910e+00
-6.45288408e-01 -5.63986659e-01 -6.53017521e-01 1.63902903e+00
4.58437681e-01 -7.40147054e-01 3.25785846e-01 6.59131348e-01
1.96667388e-01 -9.29466248e-01 -8.93356621e-01 -1.11948657e+00
2.02533722e-01 -4.78103817e-01 5.15116751e-01 1.64987242e+00
6.16148114e-01 1.05055952e+00 -4.10898238e-01 -8.38110521e-02
2.89193243e-01 3.10788572e-01 8.33926380e-01 -1.44936275e+00
-3.49166304e-01 -3.07403147e-01 -4.50566947e-01 -5.49609184e-01
2.72058427e-01 -9.01571453e-01 -5.30921109e-02 -1.45031965e+00
6.24966919e-01 -3.38512987e-01 -4.09682155e-01 1.04113710e+00
-5.07273674e-01 3.82160209e-02 1.29944384e-01 4.04438764e-01
-7.74903297e-01 5.86179972e-01 9.26765561e-01 -9.11866501e-02
-1.67874679e-01 -1.25804199e-02 -9.96007621e-01 6.95538223e-01
9.28535044e-01 -4.70690608e-01 -4.98442084e-01 2.33746972e-02
2.04208210e-01 -4.01282683e-02 3.60927545e-02 -8.02926779e-01
8.33927765e-02 -3.71961296e-01 1.01964228e-01 -3.02534729e-01
5.36815166e-01 -4.58086133e-01 -1.82391867e-01 1.03355469e-02
-3.01071495e-01 -4.45780277e-01 2.52522707e-01 3.36589336e-01
-1.30184125e-02 -2.16183901e-01 7.03909993e-01 1.17675466e-02
-9.71981704e-01 1.26222238e-01 -2.02130005e-01 6.54872179e-01
1.29546106e+00 1.60380483e-01 -9.63643193e-01 -5.04573047e-01
-7.08423615e-01 2.63542384e-01 3.82657528e-01 8.18522990e-01
2.15223998e-01 -1.42071545e+00 -8.27064276e-01 -7.48985335e-02
6.95249081e-01 -3.04523278e-02 3.57018001e-02 7.65598297e-01
1.56518355e-01 2.37520486e-01 -1.46531373e-01 -8.94253403e-02
-1.28892171e+00 8.23067009e-01 5.45654781e-02 -4.82348233e-01
-7.61828721e-01 9.98339891e-01 -1.02942258e-01 -5.23606539e-01
-5.21529131e-02 4.55416813e-02 -4.29992348e-01 2.89253712e-01
5.52730858e-01 2.62409240e-01 -1.33617461e-01 -6.58119798e-01
-6.60918772e-01 9.23482180e-02 -4.12716210e-01 1.97476909e-01
1.37365282e+00 -1.53053710e-02 2.62687393e-02 8.00481200e-01
8.03590298e-01 -2.13547975e-01 -7.65997529e-01 -6.20506287e-01
2.44822666e-01 -4.91112292e-01 -2.79621452e-01 -8.59607220e-01
-6.11000776e-01 5.65552235e-01 1.52984131e-02 6.40385151e-01
8.21057200e-01 6.32725418e-01 8.13862920e-01 2.47963920e-01
3.93016636e-01 -1.09181464e+00 3.93856019e-02 9.54819322e-01
8.12027574e-01 -1.43666744e+00 2.31446609e-01 -7.63866901e-01
-8.42575014e-01 7.47376263e-01 9.85927463e-01 1.05158940e-01
8.16294134e-01 3.34846348e-01 2.77153756e-02 -3.76639903e-01
-1.17504966e+00 -7.37896085e-01 2.66332299e-01 8.95493388e-01
6.14397645e-01 5.19710854e-02 -3.43145728e-01 1.13936877e+00
-3.72098297e-01 5.72655117e-03 4.23852116e-01 9.91076291e-01
-2.52711296e-01 -1.07389665e+00 1.00243047e-01 5.96674621e-01
-1.23148628e-01 -1.22620076e-01 -8.28345239e-01 9.46213067e-01
1.34897783e-01 1.33065021e+00 -1.00610442e-01 -7.76412189e-01
5.96431196e-01 3.80614430e-01 3.21822315e-01 -1.08212984e+00
-2.09753498e-01 -3.39820921e-01 5.99612772e-01 -7.39593148e-01
-3.19434702e-01 -8.07914436e-01 -9.41023111e-01 -1.84918776e-01
-5.00012755e-01 -9.92359295e-02 -7.05023333e-02 1.18919992e+00
3.82773340e-01 6.70545340e-01 1.30540222e-01 -5.68207800e-01
-3.62633914e-01 -1.41887307e+00 -6.41127944e-01 5.82932770e-01
-9.89359245e-03 -1.04581082e+00 -4.29108053e-01 -1.60601497e-01]
|
[9.38143253326416, 8.709239959716797]
|
4311eabb-5355-4818-b6ae-ec5820977294
|
mdvit-multi-domain-vision-transformer-for
|
2307.02100
| null |
https://arxiv.org/abs/2307.02100v1
|
https://arxiv.org/pdf/2307.02100v1.pdf
|
MDViT: Multi-domain Vision Transformer for Small Medical Image Segmentation Datasets
|
Despite its clinical utility, medical image segmentation (MIS) remains a daunting task due to images' inherent complexity and variability. Vision transformers (ViTs) have recently emerged as a promising solution to improve MIS; however, they require larger training datasets than convolutional neural networks. To overcome this obstacle, data-efficient ViTs were proposed, but they are typically trained using a single source of data, which overlooks the valuable knowledge that could be leveraged from other available datasets. Naivly combining datasets from different domains can result in negative knowledge transfer (NKT), i.e., a decrease in model performance on some domains with non-negligible inter-domain heterogeneity. In this paper, we propose MDViT, the first multi-domain ViT that includes domain adapters to mitigate data-hunger and combat NKT by adaptively exploiting knowledge in multiple small data resources (domains). Further, to enhance representation learning across domains, we integrate a mutual knowledge distillation paradigm that transfers knowledge between a universal network (spanning all the domains) and auxiliary domain-specific branches. Experiments on 4 skin lesion segmentation datasets show that MDViT outperforms state-of-the-art algorithms, with superior segmentation performance and a fixed model size, at inference time, even as more domains are added. Our code is available at https://github.com/siyi-wind/MDViT.
|
['Rafeef Garbi', 'Ghassan Harmarneh', 'Nourhan Bayasi', 'Siyi Du']
|
2023-07-05
| null | null | null | null |
['skin-lesion-segmentation', 'medical-image-segmentation', 'lesion-segmentation', 'representation-learning', 'transfer-learning']
|
['medical', 'medical', 'medical', 'methodology', 'miscellaneous']
|
[ 4.43329841e-01 1.05871983e-01 -6.08659327e-01 -2.85733551e-01
-9.92113233e-01 -6.53381109e-01 2.04109475e-01 7.12567195e-02
-6.18505001e-01 8.53627264e-01 -7.74144055e-03 -2.48402908e-01
-6.25227019e-02 -6.78600013e-01 -7.13478565e-01 -7.91739106e-01
4.18748140e-01 5.18224239e-01 5.81068575e-01 -1.22842252e-01
-1.87182188e-01 2.09304035e-01 -1.13011527e+00 3.65736783e-01
1.36563563e+00 9.79753613e-01 4.47387874e-01 2.15839818e-01
-3.57033938e-01 7.75447249e-01 -5.45901954e-01 -5.22721827e-01
1.18681192e-01 -3.54093403e-01 -9.29723918e-01 -2.94465426e-04
2.80837417e-01 -2.67190814e-01 -3.37438971e-01 1.01342380e+00
6.21787608e-01 -1.92369491e-01 4.60924178e-01 -1.10267353e+00
-7.11401820e-01 4.76320237e-01 -6.33201301e-01 2.00642407e-01
-2.31535524e-01 4.64524984e-01 5.97344398e-01 -5.70819557e-01
7.74945021e-01 8.21464837e-01 8.08950067e-01 7.82870233e-01
-1.24985945e+00 -8.03962648e-01 1.31001621e-01 3.13308775e-01
-1.27975523e+00 -1.76708773e-01 7.40477085e-01 -3.59724641e-01
8.23102713e-01 1.70036122e-01 6.38589680e-01 1.25161672e+00
-9.58191082e-02 1.11866629e+00 1.10134947e+00 -1.89841524e-01
2.68754095e-01 2.83686668e-01 -7.78194889e-03 6.01570129e-01
1.52512893e-01 -1.73734367e-01 -5.32884300e-01 -5.58791086e-02
8.48049343e-01 -1.90638304e-02 -4.82731342e-01 -4.47970837e-01
-1.01882267e+00 8.04740310e-01 7.26306319e-01 4.06199634e-01
-4.79972094e-01 -2.85962105e-01 5.43558180e-01 1.84893295e-01
4.85365421e-01 4.10809100e-01 -7.90003181e-01 -9.37644020e-02
-7.79686570e-01 -1.42298222e-01 6.23429537e-01 8.30480337e-01
7.81611621e-01 -2.01401636e-01 -1.80038884e-01 1.11449099e+00
-2.54446119e-01 3.63259763e-01 7.42165387e-01 -6.98072493e-01
4.63167220e-01 9.74883616e-01 -4.53835666e-01 -6.09847784e-01
-3.48634839e-01 -4.76168185e-01 -1.11369658e+00 -1.41675264e-01
4.62223798e-01 -1.84532836e-01 -1.41459227e+00 1.87904382e+00
4.52989548e-01 4.35780704e-01 7.72020444e-02 8.78453553e-01
1.05198646e+00 1.02741078e-01 2.77090132e-01 3.17499824e-02
1.36893451e+00 -9.47490871e-01 -6.10607207e-01 -4.65329766e-01
7.58707047e-01 -4.88394231e-01 9.02524829e-01 3.77182126e-01
-9.34099734e-01 -3.02987814e-01 -7.03566670e-01 -2.01766863e-01
-4.70666200e-01 2.89495252e-02 6.80885613e-01 5.31789660e-01
-9.90705907e-01 2.81831801e-01 -8.15642655e-01 -3.28896761e-01
1.16376352e+00 4.07633901e-01 -3.47831219e-01 -4.46402580e-01
-1.23869026e+00 7.71178842e-01 5.37168503e-01 -7.62898922e-02
-7.39294648e-01 -1.17641950e+00 -7.23267555e-01 -2.17424020e-01
6.68493271e-01 -8.78567338e-01 1.07250452e+00 -1.26188493e+00
-1.33393300e+00 9.03137207e-01 -1.15664147e-01 -4.45606709e-01
6.97536826e-01 -1.30574748e-01 -2.25017726e-01 3.45753908e-01
1.50867000e-01 9.05902267e-01 6.93152666e-01 -1.13077211e+00
-4.43053335e-01 -4.58613545e-01 6.54224828e-02 2.40378872e-01
-6.20951295e-01 -4.02142912e-01 -8.29819024e-01 -5.88661432e-01
-8.39441046e-02 -9.45998311e-01 -4.25998956e-01 3.32601845e-01
-3.31147045e-01 -1.47018462e-01 6.26384735e-01 -7.97982991e-01
1.00731480e+00 -2.19395232e+00 2.34113768e-01 5.64291254e-02
4.73730117e-01 9.20524478e-01 -1.68607205e-01 4.54511158e-02
1.11756682e-01 1.60666760e-02 -7.49282002e-01 -1.04519874e-01
-4.90790993e-01 5.96929371e-01 3.63697074e-02 1.68323725e-01
5.32014012e-01 1.18426621e+00 -1.01223040e+00 -6.33774996e-01
3.10893059e-01 5.33766747e-01 -5.56967735e-01 -1.40354112e-02
-3.00583988e-01 7.02483654e-01 -4.84477520e-01 7.43512094e-01
8.89507115e-01 -7.72334039e-01 3.23517799e-01 -3.10042262e-01
2.69311994e-01 -2.25601476e-02 -8.96238565e-01 1.93957257e+00
-3.09901595e-01 3.69713157e-01 1.45433350e-02 -1.24262857e+00
6.69075191e-01 1.85095713e-01 6.20892882e-01 -9.24421072e-01
1.30325526e-01 2.88843572e-01 1.73061378e-02 -4.96422559e-01
3.28623615e-02 -2.13666603e-01 1.41666710e-01 3.04432791e-02
2.28966668e-01 4.14310358e-02 -6.00850210e-02 2.46601745e-01
1.17994642e+00 -2.41267562e-01 2.80008346e-01 4.28961366e-02
3.81538451e-01 3.77350658e-01 8.76255393e-01 6.58280730e-01
-4.31144446e-01 5.19213438e-01 4.01308149e-01 -1.96364418e-01
-7.12020159e-01 -1.12718093e+00 -2.79863536e-01 8.20705116e-01
4.03503120e-01 -2.83212420e-02 -7.34096348e-01 -9.22150612e-01
1.77035436e-01 3.91959816e-01 -7.39880085e-01 -4.31713820e-01
-4.35868025e-01 -6.84569478e-01 7.57214069e-01 7.14369714e-01
7.64878213e-01 -9.39373314e-01 -4.77173716e-01 1.46387890e-01
-3.53806704e-01 -1.29088783e+00 -3.26418906e-01 2.81320810e-01
-9.74940956e-01 -1.15343201e+00 -1.20123923e+00 -7.58608401e-01
7.23789275e-01 4.23515260e-01 1.14791894e+00 -1.19170256e-01
-5.67937315e-01 2.18603477e-01 -3.48320961e-01 -4.83188838e-01
-2.76134461e-01 2.72821605e-01 -2.98679829e-01 -1.96111664e-01
5.68187892e-01 -4.24074888e-01 -8.21217477e-01 3.00168008e-01
-1.06501293e+00 2.84398735e-01 9.13454831e-01 1.22275662e+00
8.30792367e-01 -1.30440429e-01 7.45072186e-01 -1.18328488e+00
4.44547027e-01 -6.81283712e-01 -2.87323684e-01 3.01753551e-01
-4.60552633e-01 -2.08496168e-01 5.79901576e-01 -7.45785594e-01
-1.21353662e+00 6.08717762e-02 5.58459982e-02 -7.29165316e-01
-1.97619408e-01 6.57013118e-01 -3.07763256e-02 -7.37440586e-02
7.62710273e-01 3.59532297e-01 3.28740090e-01 -4.00718629e-01
4.63637292e-01 6.77185893e-01 5.36315024e-01 -4.95525420e-01
4.76153910e-01 5.97139180e-01 -4.12273824e-01 -7.36478031e-01
-9.66310978e-01 -6.71726108e-01 -5.79658926e-01 1.16348285e-02
8.73911560e-01 -1.09618092e+00 -2.75364578e-01 7.22699642e-01
-8.98320496e-01 -5.65491378e-01 -4.23722267e-01 3.72798234e-01
-2.58636385e-01 3.47041637e-01 -4.82859284e-01 -2.63037115e-01
-3.36821973e-01 -1.18166363e+00 8.49602640e-01 4.85099256e-01
-1.14775104e-02 -1.02800786e+00 -1.50668651e-01 7.20096469e-01
4.04054224e-01 1.52309507e-01 8.76046956e-01 -7.01807261e-01
-4.69929546e-01 1.61161929e-01 -5.86728334e-01 7.01573789e-01
4.79425877e-01 -4.89040762e-01 -9.94748712e-01 -3.32507968e-01
-2.59208292e-01 -6.17470026e-01 1.03023064e+00 4.53101575e-01
1.46185029e+00 -9.73504707e-02 -5.83093345e-01 6.71657503e-01
1.36890924e+00 9.53576639e-02 5.83311319e-01 2.51380444e-01
8.54329586e-01 4.63033438e-01 5.13397694e-01 3.33491802e-01
5.43068767e-01 3.97009730e-01 2.89398879e-01 -6.41580999e-01
-4.48836803e-01 5.93411596e-03 3.09179593e-02 6.49286509e-01
1.34941280e-01 2.58013066e-02 -1.04257417e+00 9.10200536e-01
-1.82555854e+00 -4.74550694e-01 -8.62989854e-03 1.95895648e+00
1.35238695e+00 5.22587299e-02 1.11088581e-01 -2.11410865e-01
5.59911907e-01 -1.76197767e-01 -1.22692418e+00 -2.79236659e-02
-2.07599506e-01 3.95621836e-01 6.88277900e-01 3.41206044e-02
-9.44992602e-01 1.17113912e+00 5.50290585e+00 1.11237228e+00
-1.36628938e+00 2.48077601e-01 5.42327404e-01 -7.43974000e-02
-2.30279461e-01 -3.50429267e-01 -4.64112192e-01 5.56359470e-01
7.04693675e-01 -7.00397566e-02 6.80982172e-02 8.48422706e-01
-2.38253638e-01 -2.00208455e-01 -7.66677082e-01 9.27080929e-01
-1.59166679e-01 -1.54667723e+00 5.80329895e-02 1.14969671e-01
7.84076333e-01 3.75204116e-01 1.85051560e-01 3.65939677e-01
4.73273695e-01 -8.60883296e-01 5.90709038e-02 2.10847408e-01
9.13042068e-01 -6.10974371e-01 8.71752262e-01 3.31661105e-01
-9.00854826e-01 -7.04186596e-03 -3.07830602e-01 4.00246143e-01
-2.54070181e-02 8.14677119e-01 -1.06012690e+00 7.14542866e-01
8.12892616e-01 7.97821581e-01 -5.57490706e-01 9.87030447e-01
-1.10271156e-01 7.51811683e-01 -3.14783961e-01 3.07808071e-01
3.08021039e-01 -1.13508562e-02 2.34357536e-01 1.18814254e+00
3.20311226e-02 1.89376324e-01 1.90143958e-01 8.00199807e-01
-2.44585812e-01 -1.25244334e-01 -5.57038724e-01 6.22707838e-03
5.41550338e-01 1.06047022e+00 -4.87821400e-01 -4.61778998e-01
-5.28479695e-01 1.15630162e+00 4.82048899e-01 3.82723957e-01
-8.99790108e-01 -1.78033873e-01 9.35059905e-01 -9.21256617e-02
4.04806942e-01 1.66097850e-01 -4.99453753e-01 -1.06894064e+00
1.54358596e-01 -8.72167647e-01 6.93298101e-01 -3.76626939e-01
-1.62197602e+00 3.84776682e-01 -1.91467360e-01 -1.18250275e+00
1.33163944e-01 -5.15653253e-01 -7.32303932e-02 9.36974943e-01
-2.03161955e+00 -1.33910120e+00 -3.91959518e-01 9.78116989e-01
4.72207397e-01 -1.23824991e-01 7.17547357e-01 4.68239903e-01
-6.62227809e-01 9.98333812e-01 1.69291005e-01 2.04263583e-01
9.69275057e-01 -1.18793213e+00 8.97886083e-02 4.37573701e-01
-2.22426906e-01 5.17373204e-01 1.20832942e-01 -5.98626316e-01
-1.27765322e+00 -1.24097538e+00 3.69851053e-01 -3.55025023e-01
5.86841643e-01 -6.34033009e-02 -1.44141662e+00 5.74972928e-01
-7.46626779e-02 2.87769794e-01 1.03961575e+00 5.98343723e-02
-5.55120766e-01 -4.25960161e-02 -1.39655221e+00 5.07143557e-01
1.12800229e+00 -4.78623718e-01 -4.48226064e-01 2.73473620e-01
7.88430095e-01 -5.45465410e-01 -1.04200399e+00 5.88045239e-01
3.36702496e-01 -7.35625923e-01 1.04893494e+00 -4.78138506e-01
4.15783674e-01 -1.51002660e-01 8.37856755e-02 -1.32480681e+00
-1.93386853e-01 -2.58759081e-01 -2.27926727e-02 1.16551614e+00
3.32084060e-01 -8.36855710e-01 8.80943596e-01 6.78436697e-01
-9.95720476e-02 -8.44665229e-01 -1.12036264e+00 -8.28419745e-01
3.28044206e-01 -2.88557649e-01 4.72933173e-01 1.38187397e+00
-1.83784425e-01 2.16177434e-01 -1.83452576e-01 1.25224859e-01
4.50413376e-01 -1.20021790e-01 5.52509904e-01 -1.13887537e+00
-2.88562089e-01 -4.42655861e-01 -3.39319795e-01 -9.17030871e-01
3.61258015e-02 -1.12667823e+00 -3.03879697e-02 -1.66600025e+00
3.07474852e-01 -6.96744919e-01 -6.20437026e-01 8.90474141e-01
-4.24694270e-01 2.53844768e-01 9.56040099e-02 2.12007716e-01
-5.04479051e-01 4.20266867e-01 1.74691129e+00 -2.86612928e-01
-2.29062095e-01 -2.79480457e-01 -9.65254903e-01 7.22675383e-01
8.88500929e-01 -4.57280964e-01 -6.47221982e-01 -5.82595050e-01
-3.46518427e-01 -6.50507286e-02 3.48920077e-01 -8.64517093e-01
3.46764863e-01 -1.95377156e-01 4.83042985e-01 -4.43000644e-01
2.92659491e-01 -7.03812957e-01 -5.16248792e-02 5.24663508e-01
-2.37660661e-01 -5.66894293e-01 5.50827324e-01 6.29590631e-01
-3.01122248e-01 9.39016044e-02 9.22491133e-01 -1.95556566e-01
-1.02196300e+00 4.53092217e-01 -2.50350684e-02 2.77174890e-01
1.12990177e+00 -2.67868996e-01 -5.82279384e-01 1.19260356e-01
-5.71378946e-01 4.46849227e-01 3.96983981e-01 4.38226074e-01
6.61953270e-01 -1.07469392e+00 -6.41881526e-01 1.50711741e-02
3.81130069e-01 5.72560728e-01 7.59374380e-01 9.90569234e-01
-1.82365954e-01 1.90484032e-01 -2.79378116e-01 -8.90340924e-01
-1.18710947e+00 4.01594520e-01 2.60285974e-01 -4.66757327e-01
-7.06340611e-01 1.06697226e+00 4.69731867e-01 -6.77078366e-01
1.95507333e-01 -1.75201848e-01 1.22881234e-02 1.28922254e-01
5.06146669e-01 1.61225736e-01 1.67644382e-01 -2.07910806e-01
-5.22260308e-01 4.19059217e-01 -6.12404406e-01 2.73835897e-01
1.25627446e+00 6.63435906e-02 8.57980102e-02 6.55955970e-02
1.09090137e+00 -6.14056885e-01 -1.41235077e+00 -6.82873130e-01
-1.62920490e-01 -4.35010254e-01 1.63263828e-01 -1.31571460e+00
-1.32396436e+00 8.72353613e-01 7.39349306e-01 -3.00806433e-01
1.40543222e+00 2.17842534e-01 1.15527618e+00 6.15534037e-02
3.18298191e-01 -1.07247841e+00 2.29047373e-01 2.66310602e-01
5.32188535e-01 -1.51354170e+00 -2.11871266e-01 -5.90020835e-01
-1.08008111e+00 6.81896865e-01 9.15510714e-01 2.40635991e-01
6.24952912e-01 1.30406424e-01 2.47760266e-01 -1.24902308e-01
-5.46784699e-01 -5.09343326e-01 1.64954573e-01 1.02532816e+00
2.11851731e-01 2.12003604e-01 -2.11968452e-01 7.59453714e-01
2.42200524e-01 5.19557118e-01 1.91203535e-01 9.88622725e-01
-1.06871150e-01 -1.25380218e+00 1.21544879e-02 6.61026120e-01
-3.08932811e-01 -1.22393578e-01 -4.34533805e-01 7.50018597e-01
3.17149132e-01 7.60470450e-01 -1.83090553e-01 -3.49387974e-01
3.09699982e-01 -1.69098437e-01 4.15386081e-01 -5.12961030e-01
-6.44120276e-01 -6.76769838e-02 -3.96870412e-02 -6.22576535e-01
-4.39946711e-01 -4.49340224e-01 -1.31574130e+00 -1.39127389e-01
-1.76614463e-01 -2.24137932e-01 3.86346787e-01 8.52961540e-01
7.29241729e-01 8.08301568e-01 2.95059115e-01 -2.82293916e-01
-4.37369764e-01 -6.42183125e-01 -3.73358011e-01 4.63901669e-01
3.28125656e-01 -6.60625398e-01 -2.01861896e-02 6.44740388e-02]
|
[14.674539566040039, -2.075057029724121]
|
9385498c-254a-492e-b12b-1c571523a276
|
convolutional-neural-network-with
|
2202.06673
| null |
https://arxiv.org/abs/2202.06673v1
|
https://arxiv.org/pdf/2202.06673v1.pdf
|
Convolutional Neural Network with Convolutional Block Attention Module for Finger Vein Recognition
|
Convolutional neural networks have become a popular research in the field of finger vein recognition because of their powerful image feature representation. However, most researchers focus on improving the performance of the network by increasing the CNN depth and width, which often requires high computational effort. Moreover, we can notice that not only the importance of pixels in different channels is different, but also the importance of pixels in different positions of the same channel is different. To reduce the computational effort and to take into account the different importance of pixels, we propose a lightweight convolutional neural network with a convolutional block attention module (CBAM) for finger vein recognition, which can achieve a more accurate capture of visual structures through an attention mechanism. First, image sequences are fed into a lightweight convolutional neural network we designed to improve visual features. Afterwards, it learns to assign feature weights in an adaptive manner with the help of a convolutional block attention module. The experiments are carried out on two publicly available databases and the results demonstrate that the proposed method achieves a stable, highly accurate, and robust performance in multimodal finger recognition.
|
['Mingwen Wang', 'Zhongxia Zhang']
|
2022-02-14
| null | null | null | null |
['finger-vein-recognition']
|
['computer-vision']
|
[ 2.65163988e-01 -5.12676001e-01 7.53031299e-02 -4.15156841e-01
-5.37183173e-02 -1.45820603e-01 1.20837882e-01 -2.12162748e-01
-5.79729974e-01 4.05174583e-01 2.08462462e-01 6.51240945e-02
9.79016274e-02 -8.89638782e-01 -4.93775278e-01 -7.94654906e-01
4.89681512e-01 1.08917337e-02 3.48759979e-01 7.99195766e-02
3.86189312e-01 7.60321558e-01 -1.24717593e+00 1.61846101e-01
6.57370150e-01 1.15097678e+00 5.87149598e-02 3.63155752e-01
-3.44047725e-01 7.48931289e-01 -4.35294747e-01 -5.34654140e-01
1.64968386e-01 -3.48833710e-01 -5.65991879e-01 2.82844931e-01
4.58204806e-01 -5.50743580e-01 -7.38969266e-01 1.04329884e+00
8.39255452e-01 -2.14611039e-01 4.59066391e-01 -8.18312109e-01
-5.85926712e-01 2.94060230e-01 -7.32202530e-01 3.06983560e-01
-2.30515346e-01 3.59694779e-01 8.27771246e-01 -5.22589982e-01
2.12686062e-01 1.27676380e+00 3.78818423e-01 4.90845680e-01
-1.11015761e+00 -6.41471982e-01 2.59604156e-01 4.27901328e-01
-1.23171937e+00 -2.10403260e-02 9.20808196e-01 -2.11899340e-01
4.83478665e-01 2.08203778e-01 7.41980195e-01 9.48301792e-01
1.26370892e-01 8.76175582e-01 8.93644691e-01 -2.44154051e-01
-1.14373326e-01 -4.80056740e-02 2.26771742e-01 6.33004010e-01
2.44430289e-01 -9.62098241e-02 -1.93101987e-01 1.36563957e-01
1.38868630e+00 7.29909122e-01 -1.84626997e-01 -8.61202851e-02
-9.66636777e-01 6.60798252e-01 1.06678128e+00 4.68433142e-01
-7.62820542e-01 4.45280612e-01 2.61498988e-01 -8.20474848e-02
-1.96729720e-01 -7.44368555e-03 -1.74416229e-01 -4.35454212e-02
-4.47213501e-01 -6.67383224e-02 4.93511677e-01 6.07553244e-01
5.93801320e-01 -2.33039349e-01 -7.05958903e-01 7.14320958e-01
3.02576929e-01 3.75524998e-01 3.65281075e-01 -5.84910870e-01
4.24102783e-01 1.07180786e+00 -1.14826597e-01 -1.23774767e+00
-2.52010942e-01 -4.94149685e-01 -1.19627631e+00 2.80992359e-01
6.25992417e-01 -1.66375473e-01 -9.99520600e-01 1.30522501e+00
-9.90622044e-02 4.01354507e-02 -2.97247261e-01 1.13577986e+00
7.28360474e-01 3.38247389e-01 2.77463794e-01 3.03785741e-01
1.52112758e+00 -1.06618953e+00 -4.78052586e-01 -3.71249914e-02
-8.79957974e-02 -7.06033766e-01 9.05676067e-01 1.65248543e-01
-9.01729465e-01 -9.72303808e-01 -6.86286688e-01 -1.27240255e-01
-2.37900481e-01 3.44557583e-01 6.68756068e-01 4.92050320e-01
-4.52055514e-01 3.98596525e-01 -6.56552434e-01 -2.98279524e-01
7.44788706e-01 3.79744262e-01 -1.92093939e-01 -2.14281544e-01
-8.13971937e-01 5.52327633e-01 3.30677301e-01 6.74484611e-01
-4.75266367e-01 -2.90690452e-01 -5.84912300e-01 4.96254861e-01
3.04108337e-02 -3.46246302e-01 6.99501812e-01 -1.29509091e+00
-1.27414215e+00 3.92833114e-01 -3.19151223e-01 -6.24353215e-02
6.07846916e-01 -1.42599136e-01 -1.53479964e-01 2.39994138e-01
-3.16299140e-01 5.19842744e-01 9.11787927e-01 -8.54887903e-01
-5.70073009e-01 -5.10138273e-01 1.24139242e-01 -1.07628264e-01
-6.82491124e-01 4.28093560e-02 -8.28164041e-01 -6.61994338e-01
1.95160843e-02 -5.88232040e-01 -3.52726221e-01 1.77840799e-01
-4.15378809e-01 -1.91070199e-01 6.72085762e-01 -7.45444059e-01
1.25123858e+00 -2.07995462e+00 1.72856420e-01 5.72221637e-01
3.17981422e-01 5.61723590e-01 -1.74196228e-01 2.38905512e-02
3.84295881e-02 -1.87075526e-01 -9.23057273e-02 1.15361676e-01
-2.70759404e-01 1.57859191e-01 2.84666894e-03 2.58717462e-02
6.05795979e-01 1.05364120e+00 -5.91389656e-01 -5.45078099e-01
5.54848194e-01 7.32379615e-01 -4.84241039e-01 2.69732416e-01
4.62809466e-02 5.74014425e-01 -6.25060737e-01 6.11034691e-01
7.30302691e-01 -5.18995643e-01 9.40314606e-02 -6.52850628e-01
-1.87098548e-01 -3.23976696e-01 -1.22111094e+00 1.25223601e+00
-2.15070218e-01 5.41420400e-01 -3.73225302e-01 -7.33942449e-01
1.00294435e+00 7.67808855e-02 2.63433129e-01 -9.09567118e-01
4.23517108e-01 4.41341661e-03 3.07877153e-01 -5.91472447e-01
4.00119983e-02 1.59317300e-01 3.83680820e-01 2.86509693e-01
-2.56676406e-01 7.73448229e-01 1.05795392e-03 -2.35465318e-01
9.43013012e-01 -3.38167250e-01 -1.49717927e-01 2.12023169e-01
8.67735624e-01 -4.63896155e-01 3.46101999e-01 5.98750174e-01
-1.75052002e-01 6.41585171e-01 5.63074589e-01 -7.94654787e-01
-1.08630013e+00 -6.89895689e-01 -1.72165945e-01 7.45169461e-01
3.92552406e-01 -4.43074815e-02 -5.69007039e-01 -6.33620560e-01
2.62554616e-01 -2.42052317e-01 -8.17930281e-01 -7.37856477e-02
-7.32635558e-01 -6.43350303e-01 4.03272778e-01 1.01622057e+00
1.10204041e+00 -1.47028184e+00 -7.79571593e-01 2.58882999e-01
1.07318319e-01 -7.78618753e-01 -4.50060666e-01 -1.35268569e-01
-7.58945942e-01 -1.23723388e+00 -1.25847983e+00 -1.03698063e+00
1.06377387e+00 1.47015259e-01 4.34192210e-01 3.85213107e-01
-7.32965589e-01 6.25692680e-02 -2.80676037e-01 -1.79221153e-01
4.60776985e-01 2.16747344e-01 -6.13124788e-01 6.78068221e-01
5.98848999e-01 -2.37005964e-01 -9.12735105e-01 2.15698928e-01
-8.54515970e-01 -8.24967697e-02 1.03984833e+00 1.03349125e+00
4.66993093e-01 -7.39549771e-02 2.30669096e-01 -7.47855961e-01
4.79902416e-01 5.00239804e-02 -3.82803321e-01 2.86740571e-01
-2.08824068e-01 2.95015097e-01 4.39753473e-01 -3.36266279e-01
-9.02690589e-01 3.38284634e-02 -3.12158793e-01 -2.80476391e-01
-7.38336071e-02 2.74468035e-01 -2.00496674e-01 -3.52362543e-01
1.00571223e-01 4.28137779e-01 1.82981119e-01 -6.10131800e-01
6.16004579e-02 8.31626058e-01 2.56606489e-01 -2.26386666e-01
6.09546006e-01 3.54491830e-01 -1.39236748e-02 -6.72371387e-01
-3.74398202e-01 -1.30165368e-01 -6.16580248e-01 -2.04471245e-01
8.85637999e-01 -4.29649323e-01 -1.14499032e+00 9.22762573e-01
-1.12213969e+00 7.85229448e-03 1.73342600e-01 2.74244547e-01
9.63272005e-02 5.82369149e-01 -7.70158529e-01 -5.98634660e-01
-6.34182930e-01 -1.28564656e+00 7.04284310e-01 8.64597499e-01
8.41980278e-02 -7.92484462e-01 -4.32090700e-01 9.78909135e-02
8.12388837e-01 1.52598903e-01 1.07624781e+00 -2.59215415e-01
-7.91029871e-01 -3.80277604e-01 -1.02850068e+00 4.19733375e-01
3.99619222e-01 8.13073367e-02 -7.10035205e-01 -2.21200332e-01
-6.32329822e-01 3.66033688e-02 1.04778647e+00 4.81324255e-01
1.62189662e+00 -1.65962502e-01 -1.94021448e-01 5.49349904e-01
1.44594884e+00 1.96188182e-01 1.05284309e+00 3.66416007e-01
7.70156205e-01 5.82519293e-01 1.15362339e-01 4.62173402e-01
1.59028023e-01 5.78830123e-01 6.50217056e-01 -6.32966220e-01
-1.51458561e-01 1.35652348e-01 -1.88299760e-01 2.76521027e-01
-5.84596515e-01 2.27044299e-01 -4.43523198e-01 4.02866453e-01
-1.87898207e+00 -9.42662120e-01 -7.95949325e-02 2.12399983e+00
6.96721852e-01 2.03572549e-02 1.68880850e-01 4.15781587e-02
8.89869571e-01 1.26704976e-01 -6.44743443e-01 -1.19418241e-01
-4.22146823e-03 5.95403910e-01 4.44896370e-01 1.41664192e-01
-9.66519773e-01 8.08051586e-01 5.65379238e+00 4.30060923e-01
-1.32832611e+00 -3.25224429e-01 6.25948608e-01 2.03709938e-02
2.19024327e-02 -4.01284397e-01 -6.46016061e-01 9.04352307e-01
1.30986884e-01 3.18418682e-01 4.62637186e-01 6.65607512e-01
-7.41647929e-02 3.09909791e-01 -6.96894109e-01 1.24799252e+00
-5.95797002e-02 -1.23202062e+00 2.90692508e-01 -6.30474836e-02
3.78032953e-01 -3.04601133e-01 7.91687593e-02 -1.14626743e-01
-3.55346620e-01 -1.22264326e+00 2.92567432e-01 9.30382252e-01
4.68160003e-01 -9.79448617e-01 1.07837439e+00 -1.48799360e-01
-1.23003745e+00 -1.12240076e-01 -6.70925796e-01 -4.76657711e-02
-2.12246686e-01 3.26040924e-01 -9.05963406e-02 1.97042510e-01
8.56030107e-01 7.70010591e-01 -7.15353072e-01 1.29786134e+00
-3.81744862e-01 2.55474329e-01 8.05091113e-02 -4.05471325e-01
2.15517804e-01 -1.19926617e-01 2.65484434e-02 1.18273342e+00
1.97543297e-02 6.53639436e-03 -1.03584848e-01 9.70374584e-01
-1.27265319e-01 6.46551549e-02 -9.09717903e-02 -1.41666025e-01
1.92616671e-01 1.33396924e+00 -3.59766185e-01 -2.65756488e-01
-6.64629102e-01 1.19782674e+00 2.59648293e-01 4.56681311e-01
-6.32784724e-01 -8.67249489e-01 6.62218034e-01 -8.47391668e-04
6.91303968e-01 -5.11448719e-02 -3.72232348e-01 -9.83831942e-01
2.17130870e-01 -5.15238106e-01 2.67887533e-01 -4.71335977e-01
-1.41533291e+00 6.63047135e-01 -8.04632366e-01 -1.01621187e+00
2.63932824e-01 -9.18658376e-01 -7.51765013e-01 1.28376031e+00
-1.86254871e+00 -1.10247469e+00 -1.00946283e+00 8.61994088e-01
3.13001394e-01 -2.45651230e-01 6.09158576e-01 5.16913772e-01
-8.80081594e-01 7.59465039e-01 -2.22724095e-01 7.57717371e-01
7.62416959e-01 -9.25639212e-01 1.75260916e-01 6.49119616e-01
-3.98290575e-01 8.16190183e-01 2.86810920e-02 -3.90557587e-01
-1.24028254e+00 -1.13344955e+00 6.65971637e-01 -6.18620077e-04
1.37878910e-01 9.55310538e-02 -8.45386922e-01 6.02521598e-01
1.43112615e-01 1.50585428e-01 5.40345132e-01 -3.48269604e-02
-3.48951280e-01 -3.72649014e-01 -9.41574872e-01 6.03360772e-01
6.96301818e-01 -2.68789768e-01 -3.22929382e-01 -3.79641056e-02
1.06544033e-01 -2.20296577e-01 -8.82938385e-01 2.90329039e-01
1.03383183e+00 -8.82982194e-01 9.08939838e-01 -7.26879478e-01
6.15337074e-01 -4.14732695e-01 1.87086403e-01 -7.84015357e-01
-5.69141150e-01 3.60436961e-02 5.62674552e-02 1.15346384e+00
8.68946165e-02 -7.62649417e-01 9.92785513e-01 7.83444107e-01
3.80175948e-01 -7.86439717e-01 -6.62757993e-01 -2.65487134e-01
-2.30672643e-01 7.81219900e-02 6.06567860e-01 5.24682641e-01
-3.63369733e-01 1.17494375e-01 -4.30119932e-01 1.21212766e-01
7.17017829e-01 3.49614054e-01 6.58998609e-01 -1.25519025e+00
-2.38069937e-01 -5.30822396e-01 -7.67088652e-01 -1.10632765e+00
-4.15501207e-01 -5.30112922e-01 -2.75060028e-01 -1.53270912e+00
5.53175628e-01 -2.05529168e-01 -7.17154086e-01 6.66907787e-01
-4.92718875e-01 5.86703658e-01 2.91053712e-01 9.78033766e-02
-3.61381561e-01 3.57425123e-01 1.42698538e+00 -4.70748246e-01
-1.06150039e-01 5.51913343e-02 -7.90084958e-01 5.12390196e-01
6.21306300e-01 1.54581532e-01 2.44397908e-01 -6.40846908e-01
-2.71266878e-01 -4.27965701e-01 6.62579894e-01 -9.54007030e-01
3.72525275e-01 1.43144652e-01 1.08986866e+00 -3.51504952e-01
1.17170531e-02 -1.21512663e+00 -1.28721833e-01 9.72595215e-01
-3.72600675e-01 -1.88147336e-01 1.76663324e-01 3.08104038e-01
-2.98611075e-01 -1.57378003e-01 8.24833572e-01 -1.55571222e-01
-9.67418849e-01 6.10921443e-01 -1.30565777e-01 -5.37200451e-01
9.88295496e-01 -2.66780168e-01 -2.70193696e-01 -7.01334849e-02
-5.56819022e-01 1.77811980e-01 1.27706423e-01 4.56001073e-01
7.93886721e-01 -1.48495758e+00 -7.00008631e-01 5.73167384e-01
2.40069270e-01 -1.94467023e-01 5.06615281e-01 7.40761280e-01
-6.92234039e-01 3.51465434e-01 -6.79388642e-01 -5.26993990e-01
-1.36364090e+00 2.84719259e-01 4.72895920e-01 -5.83678856e-02
-6.24643445e-01 7.27213800e-01 3.58857177e-02 -3.15809399e-02
5.42748034e-01 -2.29612857e-01 -4.98014122e-01 -1.22198157e-01
8.15288126e-01 3.19264442e-01 -2.35361591e-01 -3.63485783e-01
-3.11529815e-01 1.02096355e+00 -2.77343065e-01 4.79227990e-01
1.27769566e+00 2.00216666e-01 -2.99146861e-01 -2.19566688e-01
1.02093136e+00 -2.14272082e-01 -1.49069369e+00 -4.31293547e-01
-2.58956313e-01 -7.47481465e-01 -5.99498004e-02 -7.99555123e-01
-1.52528143e+00 1.15645003e+00 8.93635094e-01 -2.99566872e-02
1.16527498e+00 -2.30224326e-01 9.55271244e-01 1.80170655e-01
1.06851630e-01 -8.38882148e-01 1.67957366e-01 3.81499529e-01
7.90435910e-01 -1.24249566e+00 -3.27827632e-01 -2.67819464e-01
-5.46119213e-01 1.61816168e+00 8.20036709e-01 -3.14414233e-01
4.38587248e-01 -4.37584892e-03 1.23619050e-01 -8.14909413e-02
-1.32692475e-02 -4.17986095e-01 3.75924975e-01 3.83870572e-01
4.93895769e-01 -1.51843935e-01 -3.82641077e-01 5.10451853e-01
3.96612465e-01 2.48785149e-02 -7.93762729e-02 7.76535273e-01
-4.55762476e-01 -1.29276884e+00 -6.26782998e-02 5.20820856e-01
-3.57828766e-01 -3.43073420e-02 -3.52172494e-01 5.56910813e-01
1.28356904e-01 6.52988911e-01 2.60208666e-01 -3.94761413e-01
5.31642795e-01 -1.75026864e-01 7.01239228e-01 -2.97073960e-01
-6.10108018e-01 6.62761182e-02 -4.82052863e-01 -4.96193528e-01
-3.78351361e-01 -2.28370309e-01 -1.16441655e+00 -3.16119760e-01
-1.20344780e-01 -9.67279300e-02 4.00441766e-01 8.61698151e-01
3.62722635e-01 9.67629373e-01 6.94319725e-01 -6.80895805e-01
-4.24786121e-01 -1.07528484e+00 -5.97344339e-01 5.77548146e-01
2.07963824e-01 -4.28317368e-01 7.21486285e-02 -1.84478611e-01]
|
[13.141435623168945, 0.9668226838111877]
|
cb1dee5c-d42a-471d-bf65-b4a547d53aa4
|
deep-spectral-methods-a-surprisingly-strong
|
2205.07839
| null |
https://arxiv.org/abs/2205.07839v1
|
https://arxiv.org/pdf/2205.07839v1.pdf
|
Deep Spectral Methods: A Surprisingly Strong Baseline for Unsupervised Semantic Segmentation and Localization
|
Unsupervised localization and segmentation are long-standing computer vision challenges that involve decomposing an image into semantically-meaningful segments without any labeled data. These tasks are particularly interesting in an unsupervised setting due to the difficulty and cost of obtaining dense image annotations, but existing unsupervised approaches struggle with complex scenes containing multiple objects. Differently from existing methods, which are purely based on deep learning, we take inspiration from traditional spectral segmentation methods by reframing image decomposition as a graph partitioning problem. Specifically, we examine the eigenvectors of the Laplacian of a feature affinity matrix from self-supervised networks. We find that these eigenvectors already decompose an image into meaningful segments, and can be readily used to localize objects in a scene. Furthermore, by clustering the features associated with these segments across a dataset, we can obtain well-delineated, nameable regions, i.e. semantic segmentations. Experiments on complex datasets (Pascal VOC, MS-COCO) demonstrate that our simple spectral method outperforms the state-of-the-art in unsupervised localization and segmentation by a significant margin. Furthermore, our method can be readily used for a variety of complex image editing tasks, such as background removal and compositing.
|
['Andrea Vedaldi', 'Iro Laina', 'Christian Rupprecht', 'Luke Melas-Kyriazi']
|
2022-05-16
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Melas-Kyriazi_Deep_Spectral_Methods_A_Surprisingly_Strong_Baseline_for_Unsupervised_Semantic_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Melas-Kyriazi_Deep_Spectral_Methods_A_Surprisingly_Strong_Baseline_for_Unsupervised_Semantic_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['unsupervised-semantic-segmentation', 'graph-partitioning']
|
['computer-vision', 'graphs']
|
[ 6.21579349e-01 9.54813436e-02 -2.24653572e-01 -3.26858401e-01
-6.79184794e-01 -8.81927311e-01 2.48127177e-01 1.96596414e-01
-4.36566681e-01 2.58399457e-01 -1.37602612e-01 -1.12580629e-02
-3.38468980e-03 -4.62717712e-01 -8.24722171e-01 -7.16067135e-01
2.02439949e-01 5.14852345e-01 4.27434057e-01 1.89622328e-01
1.25441134e-01 4.81132656e-01 -1.27535915e+00 -1.01887524e-01
1.09838665e+00 7.85164773e-01 2.16472343e-01 2.86156952e-01
-2.96568781e-01 4.40080017e-01 -4.43126470e-01 -1.04775555e-01
3.14681321e-01 -4.30692047e-01 -1.15724051e+00 7.90781915e-01
6.76411808e-01 4.15424891e-02 -1.46461502e-01 1.50684273e+00
-5.71347475e-02 2.54312128e-01 6.34143710e-01 -1.22071981e+00
-5.06987154e-01 3.99427325e-01 -8.76715064e-01 -1.68402791e-01
1.63303137e-01 -1.73164368e-01 1.02941346e+00 -8.86353374e-01
7.12716162e-01 9.91287708e-01 7.06934690e-01 3.01628411e-01
-1.90861225e+00 -3.02340806e-01 2.77818322e-01 -6.34917943e-03
-1.50626218e+00 -2.75146753e-01 8.15759599e-01 -7.86051571e-01
5.38021684e-01 3.09520483e-01 5.95781624e-01 5.14895439e-01
-3.68203431e-01 9.69007552e-01 1.00746667e+00 -5.36317945e-01
2.52314597e-01 -9.31600481e-02 1.93676710e-01 9.09709930e-01
1.50298089e-01 -4.89182383e-01 -2.47837543e-01 1.11336924e-01
8.05081487e-01 1.20184384e-01 -1.77814230e-01 -1.08313859e+00
-1.53698552e+00 6.83363914e-01 5.40856898e-01 9.22182947e-02
-1.89678535e-01 1.44407764e-01 2.56461740e-01 -1.79450616e-01
3.83914828e-01 4.92022872e-01 -3.99817497e-01 2.65603364e-01
-1.17736852e+00 -4.39333841e-02 7.63516307e-01 1.05632043e+00
1.34883296e+00 -4.50013950e-02 7.52089620e-02 9.76289392e-01
1.67287394e-01 3.19743752e-01 1.40233949e-01 -1.30643308e+00
6.64136335e-02 8.45149279e-01 -6.35502413e-02 -1.02323735e+00
-3.42166424e-01 -3.12764376e-01 -8.98497641e-01 5.76151861e-03
6.09161079e-01 -8.99200886e-02 -1.34036839e+00 1.65754616e+00
5.25572419e-01 1.99312598e-01 -1.17226489e-01 8.75983238e-01
5.95825553e-01 4.08301383e-01 -4.39942926e-02 -7.39417002e-02
1.25628793e+00 -1.37480891e+00 -5.18350422e-01 -5.86941719e-01
3.70956153e-01 -6.58392012e-01 1.09646571e+00 2.57969975e-01
-6.98569655e-01 -4.86108065e-01 -8.62865090e-01 -9.85981226e-02
-3.95899653e-01 1.45068362e-01 9.54444587e-01 4.79509205e-01
-1.06408763e+00 4.57410157e-01 -9.51127052e-01 -6.21113658e-01
6.27370000e-01 4.15235430e-01 -5.27066767e-01 -6.39534220e-02
-5.32638252e-01 3.50359917e-01 6.28079414e-01 1.36157185e-01
-8.68467927e-01 -4.88519341e-01 -1.03287232e+00 -1.21271402e-01
7.37443089e-01 -4.33261693e-01 1.00239325e+00 -1.37538350e+00
-1.16748321e+00 1.07972181e+00 -2.63246864e-01 -3.53870422e-01
3.12902927e-01 -1.14466943e-01 6.59187436e-02 4.10428345e-01
5.69758654e-01 9.78358269e-01 9.78064060e-01 -1.49747849e+00
-6.35289967e-01 -2.09631786e-01 9.21115950e-02 2.57526726e-01
-2.83441722e-01 -1.07713431e-01 -9.57429945e-01 -5.72702408e-01
6.16240740e-01 -1.07213342e+00 -6.49538577e-01 -1.41684460e-02
-8.92835319e-01 -2.49450374e-02 9.40355182e-01 -4.96492416e-01
7.97796130e-01 -2.18432808e+00 3.92147601e-01 3.87025863e-01
5.32110870e-01 1.62202702e-03 -1.06433876e-01 9.83530059e-02
-8.64476711e-02 7.23136142e-02 -7.43103683e-01 -2.44366005e-01
-4.30419855e-02 3.02219927e-01 5.20448312e-02 5.79407573e-01
2.02159971e-01 1.03000188e+00 -1.07410562e+00 -7.26101696e-01
1.81405336e-01 1.12004556e-01 -5.24030089e-01 4.14773151e-02
-2.94565588e-01 7.04741955e-01 -2.86743879e-01 7.60901153e-01
5.56073904e-01 -5.34814239e-01 1.24397323e-01 -3.05892378e-01
1.06099717e-01 -1.69514656e-01 -1.20525575e+00 2.06852388e+00
-3.18592638e-02 7.23421574e-01 2.77882129e-01 -1.46584606e+00
6.08705759e-01 -2.63584673e-01 8.64533901e-01 -2.46498272e-01
-3.46817859e-02 9.76041611e-03 -1.00999929e-01 -3.48520607e-01
4.63654399e-01 1.46530703e-01 -1.75628915e-01 4.71655875e-01
2.70051092e-01 -2.29532808e-01 6.40502870e-01 4.01722074e-01
1.01385438e+00 1.28542304e-01 5.60293160e-02 -3.58819664e-01
2.78890252e-01 2.37799019e-01 7.69390225e-01 7.95680285e-01
-2.57148057e-01 8.99531603e-01 5.40431499e-01 -2.04288974e-01
-8.29192281e-01 -1.25291348e+00 -3.94517556e-02 1.22142446e+00
4.43030685e-01 -3.39745283e-01 -1.27949488e+00 -8.41346860e-01
-4.29034643e-02 2.12151766e-01 -4.50250238e-01 1.74063385e-01
-2.33716056e-01 -6.82120919e-01 3.75010908e-01 5.66976547e-01
5.63889265e-01 -1.01078308e+00 -4.10764545e-01 1.79629028e-02
-3.88661414e-01 -1.42003787e+00 -7.68695593e-01 3.31938356e-01
-7.12579012e-01 -1.20517802e+00 -6.75054252e-01 -1.26181185e+00
1.26947880e+00 5.89766383e-01 1.04192209e+00 1.82214435e-02
-6.24351025e-01 8.21797311e-01 -2.07063600e-01 -7.21962601e-02
-1.52722210e-01 1.95012987e-01 -9.62832645e-02 4.22363698e-01
2.92844415e-01 -4.50328916e-01 -4.61798579e-01 3.31129491e-01
-1.10482907e+00 1.46845698e-01 5.51586688e-01 7.43332148e-01
1.16965461e+00 1.71735704e-01 2.81667292e-01 -1.34514320e+00
2.87988186e-01 -2.47332871e-01 -7.15441346e-01 3.01010817e-01
-4.01229054e-01 -5.72369806e-02 3.70998323e-01 -3.95075619e-01
-9.04849529e-01 7.94200122e-01 3.25069517e-01 -3.91026378e-01
-4.34429377e-01 4.71564710e-01 -2.49870211e-01 -2.83031732e-01
5.97613990e-01 3.19480747e-01 -7.26139247e-02 -3.68381709e-01
7.84711361e-01 3.98055315e-01 8.75244498e-01 -5.80781281e-01
1.02050924e+00 7.20694244e-01 -1.50691777e-01 -9.89025235e-01
-1.05567753e+00 -9.42063272e-01 -1.23096573e+00 -1.50051311e-01
1.19701636e+00 -7.77032793e-01 -2.73347706e-01 4.00260389e-01
-9.12791014e-01 -5.44963717e-01 -4.23556179e-01 2.02411339e-01
-5.90046942e-01 7.11440504e-01 -5.44802248e-01 -3.53794158e-01
1.34171709e-01 -1.14041746e+00 1.15625954e+00 3.89652908e-01
-2.10480377e-01 -1.06647897e+00 -1.60370812e-01 6.11299694e-01
-6.23593479e-02 3.92273486e-01 8.63015950e-01 -4.77061272e-01
-7.49082386e-01 -1.42803237e-01 -4.52628672e-01 5.25828242e-01
3.73385102e-01 1.21726431e-02 -8.25214267e-01 -2.37880453e-01
-3.82385969e-01 -3.64460737e-01 1.13271081e+00 3.49520653e-01
1.41931152e+00 -2.39034310e-01 -4.00561422e-01 8.91209960e-01
1.24117863e+00 -1.14762492e-01 3.86479169e-01 1.38467297e-01
1.29948258e+00 7.26393580e-01 4.86785740e-01 -5.34982085e-02
2.49270156e-01 3.77769828e-01 3.10406268e-01 -6.05233908e-01
-8.57167616e-02 -1.70411229e-01 2.28266511e-02 7.25413799e-01
1.74579993e-01 1.69388905e-01 -1.03889298e+00 7.84070015e-01
-2.07934451e+00 -5.73094785e-01 -1.16396278e-01 1.98994720e+00
8.79438698e-01 2.22106725e-02 1.20317183e-01 -8.70577022e-02
8.31553042e-01 2.01766610e-01 -8.10817778e-01 7.61904046e-02
-1.70246005e-01 1.65091723e-01 7.44287431e-01 3.40930194e-01
-1.56231391e+00 1.38636851e+00 6.58332396e+00 8.49205196e-01
-9.11597550e-01 -1.89717487e-02 7.37630486e-01 2.64786303e-01
-1.12471834e-01 2.28301838e-01 -4.26541865e-01 1.62947029e-01
3.29067677e-01 1.46373138e-01 6.19199932e-01 9.70164776e-01
-3.23770344e-02 -3.33354056e-01 -1.24383402e+00 1.00403714e+00
1.06229499e-01 -1.22316015e+00 -8.62042084e-02 -1.18939668e-01
1.19815171e+00 3.97137627e-02 -8.43766332e-03 -1.44462362e-02
4.09895331e-01 -1.07761383e+00 6.62323713e-01 2.87068993e-01
7.66116261e-01 -5.25151193e-01 3.07609111e-01 2.00171173e-01
-1.26335633e+00 2.51769871e-01 -3.89022589e-01 2.48310775e-01
-7.33205974e-02 6.60260439e-01 -4.73773628e-01 3.07523251e-01
7.59759068e-01 8.73329282e-01 -7.90942073e-01 1.12148499e+00
-4.26915675e-01 6.23603225e-01 -3.59293669e-01 3.02940249e-01
2.72918046e-01 -6.05689168e-01 2.87443101e-01 1.32136595e+00
-1.93142921e-01 -1.33298084e-01 6.92990839e-01 1.16429532e+00
-3.97695035e-01 1.77506939e-01 -5.16881168e-01 -2.10357383e-01
2.36011192e-01 1.68088901e+00 -1.57909513e+00 -3.09189558e-01
-3.91152799e-01 1.32654583e+00 4.34649348e-01 7.46514678e-01
-7.44665742e-01 -5.18697500e-01 3.94326836e-01 4.39148920e-04
4.02808726e-01 -4.46830273e-01 -3.17364067e-01 -1.21912527e+00
3.37427086e-03 -7.78052807e-01 1.36196762e-01 -5.87082624e-01
-1.18399239e+00 1.83484212e-01 -3.48422527e-02 -1.01332569e+00
9.08827782e-02 -4.34982479e-01 -4.70382184e-01 3.42517972e-01
-1.30019486e+00 -1.27260613e+00 -4.33414221e-01 6.11403704e-01
5.81173241e-01 8.17600191e-02 6.74706459e-01 2.54567087e-01
-7.65575349e-01 2.90251106e-01 3.35711718e-01 5.74802160e-01
5.97520888e-01 -1.52789128e+00 2.78296947e-01 1.23542070e+00
7.20258892e-01 6.90413713e-01 2.33207732e-01 -6.14230156e-01
-1.32652986e+00 -1.35812056e+00 2.57333100e-01 -3.57482970e-01
6.87667370e-01 -7.16429353e-01 -8.27440858e-01 7.36559153e-01
1.31015331e-01 3.10382903e-01 6.07469440e-01 9.15175080e-02
-4.20661390e-01 -2.94240713e-02 -8.23037624e-01 6.81547642e-01
1.16669118e+00 -5.51821351e-01 -3.17129791e-01 6.55836403e-01
6.54201567e-01 -4.17050719e-01 -5.29957473e-01 2.23503798e-01
2.09352911e-01 -7.89318562e-01 1.04013121e+00 -5.40997863e-01
1.94548726e-01 -5.35780489e-01 -3.46730184e-03 -1.13413215e+00
-2.61251569e-01 -7.68880069e-01 3.44130903e-01 1.27714038e+00
4.68712837e-01 -4.36820358e-01 9.92476463e-01 5.30460358e-01
-1.66483670e-02 -3.85715544e-01 -6.71494007e-01 -6.74846172e-01
-2.99952716e-01 -3.71215850e-01 9.09724981e-02 1.14575577e+00
-2.86288321e-01 4.44921285e-01 -8.38541612e-02 2.81161010e-01
9.89644647e-01 3.92862707e-01 9.36631620e-01 -1.30718982e+00
-2.24280953e-01 -4.56880212e-01 -3.61514777e-01 -1.29327583e+00
3.84222776e-01 -1.03787494e+00 6.19668067e-01 -1.72087002e+00
5.25395513e-01 -4.95136380e-01 -1.82146356e-01 7.03071713e-01
-2.05405220e-01 6.47006452e-01 6.70322776e-02 3.93132716e-01
-1.14967012e+00 2.87427634e-01 1.05541396e+00 -4.92849022e-01
-2.79604375e-01 -2.75810719e-01 -7.31805444e-01 1.17603683e+00
6.29096806e-01 -5.31314969e-01 -4.39068258e-01 -3.99220198e-01
-1.21515751e-01 -3.19893748e-01 4.12943333e-01 -8.38238835e-01
3.15116733e-01 -3.17178011e-01 3.22586566e-01 -4.73319292e-01
4.87949587e-02 -7.88574100e-01 -7.03569269e-03 7.38842636e-02
-2.07097113e-01 -3.82223934e-01 1.55773357e-01 7.76820719e-01
-3.23509306e-01 -2.65638918e-01 7.91034818e-01 -2.47091830e-01
-1.04436481e+00 3.43288749e-01 -3.62964690e-01 3.77577156e-01
1.15738618e+00 -4.40888882e-01 3.15464474e-02 -2.78831303e-01
-8.76812756e-01 2.45518267e-01 7.21627831e-01 2.18055040e-01
4.68267769e-01 -1.15374684e+00 -3.34717035e-01 2.45804384e-01
2.26570934e-01 6.52184188e-01 3.19248214e-02 1.04115283e+00
-8.30266118e-01 9.15165991e-02 -5.20967394e-02 -1.15493119e+00
-1.15784395e+00 4.27714616e-01 2.48294756e-01 6.35540951e-03
-5.38232148e-01 8.64844859e-01 6.93777204e-01 -5.36953568e-01
4.06966209e-01 -3.29768836e-01 1.07820421e-01 1.83162630e-01
3.77690531e-02 1.01793274e-01 -2.23324850e-01 -8.00470173e-01
-3.22502375e-01 8.14892173e-01 -6.11527590e-03 4.23107743e-02
1.28868937e+00 -1.97873309e-01 -5.59622288e-01 2.93944210e-01
1.19036710e+00 8.12179223e-02 -1.46141469e+00 -3.61860752e-01
3.34518611e-01 -3.46369624e-01 -1.57447353e-01 -4.54131752e-01
-1.33383858e+00 7.27767766e-01 2.94122964e-01 3.39659810e-01
1.14445221e+00 1.74299851e-01 5.94476283e-01 4.98690397e-01
2.32239902e-01 -1.23487735e+00 3.07479590e-01 4.30984616e-01
4.10066605e-01 -1.34935677e+00 3.14764008e-02 -7.76494563e-01
-5.15222371e-01 9.68320668e-01 4.85258341e-01 -2.02832833e-01
5.35301089e-01 1.36731341e-01 1.76171333e-01 -1.43638417e-01
1.56751990e-01 -5.33788204e-01 5.06845474e-01 6.83929443e-01
1.31626919e-01 5.30284047e-02 1.10071220e-01 3.53088528e-01
-1.79792056e-03 -4.93379951e-01 4.91845459e-01 8.43121231e-01
-4.92559969e-01 -9.83080864e-01 -3.25726390e-01 4.77318704e-01
-3.32228422e-01 -8.85012150e-02 -7.26060271e-01 6.69033468e-01
1.18860207e-01 8.13733399e-01 1.78713780e-02 -2.28532776e-01
1.21710770e-01 -7.10030831e-03 3.64943415e-01 -1.09627199e+00
-1.11585811e-01 4.54798430e-01 -2.59475112e-01 -7.35330582e-01
-7.12450683e-01 -6.87833488e-01 -1.43322778e+00 2.94509530e-01
-4.02851045e-01 3.71741727e-02 5.36174059e-01 9.59223211e-01
2.00643286e-01 5.59276044e-01 3.03444535e-01 -9.31093395e-01
-3.14513929e-02 -6.45219743e-01 -7.08931506e-01 7.57071435e-01
2.64878571e-01 -5.45760214e-01 -3.19249421e-01 6.61453843e-01]
|
[9.53364086151123, 0.8052259087562561]
|
5b4efda3-9752-48f8-bfa5-4be3bbbfb8a9
|
exploiting-spectral-augmentation-for-code
|
2010.07130
| null |
https://arxiv.org/abs/2010.07130v1
|
https://arxiv.org/pdf/2010.07130v1.pdf
|
Exploiting Spectral Augmentation for Code-Switched Spoken Language Identification
|
Spoken language Identification (LID) systems are needed to identify the language(s) present in a given audio sample, and typically could be the first step in many speech processing related tasks such as automatic speech recognition (ASR). Automatic identification of the languages present in a speech signal is not only scientifically interesting, but also of practical importance in a multilingual country such as India. In many of the Indian cities, when people interact with each other, as many as three languages may get mixed. These may include the official language of that province, Hindi and English (at times the languages of the neighboring provinces may also get mixed during these interactions). This makes the spoken LID task extremely challenging in Indian context. While quite a few LID systems in the context of Indian languages have been implemented, most such systems have used small scale speech data collected internally within an organization. In the current work, we perform spoken LID on three Indian languages (Gujarati, Telugu, and Tamil) code-mixed with English. This task was organized by the Microsoft research team as a spoken LID challenge. In our work, we modify the usual spectral augmentation approach and propose a language mask that discriminates the language ID pairs, which leads to a noise robust spoken LID system. The proposed method gives a relative improvement of approximately 3-5% in the LID accuracy over a baseline system proposed by Microsoft on the three language pairs for two shared tasks suggested in the challenge.
|
['Hemant Misra', 'Sundeep Teki', 'Pradeep Rangan']
|
2020-10-14
| null | null | null | null |
['spoken-language-identification']
|
['speech']
|
[-1.92877531e-01 -3.31169426e-01 1.58356637e-01 -1.63645476e-01
-1.28548861e+00 -6.58501029e-01 6.94657028e-01 -9.21343714e-02
-5.64094961e-01 6.95383787e-01 5.96671820e-01 -6.20926738e-01
3.68551016e-01 -2.29839593e-01 -2.01057151e-01 -6.34076059e-01
2.40315631e-01 6.44686341e-01 -6.18740823e-03 -3.89124334e-01
1.82928964e-01 5.01129985e-01 -1.48007047e+00 2.43777409e-01
7.63416171e-01 5.96463799e-01 3.58290583e-01 8.13062131e-01
-3.73252183e-01 6.16921842e-01 -7.15928316e-01 -1.90362353e-02
2.27508262e-01 -4.53735054e-01 -1.03145361e+00 2.92531043e-01
4.21804279e-01 2.25136932e-02 -2.01401398e-01 1.01987469e+00
8.03770602e-01 1.33889571e-01 5.10225654e-01 -8.08062315e-01
-2.67263472e-01 8.45434427e-01 -5.02068043e-01 2.14537993e-01
5.94218910e-01 7.58077158e-03 7.77631640e-01 -9.98817384e-01
2.38212243e-01 1.40875375e+00 4.97471482e-01 4.45338160e-01
-1.15041816e+00 -8.52632284e-01 -6.05442896e-02 2.55322177e-02
-1.78595638e+00 -9.47388232e-01 7.07420647e-01 -4.36773270e-01
9.17619526e-01 4.11258250e-01 1.52217627e-01 7.95522630e-01
-4.05011714e-01 7.04287529e-01 1.40893674e+00 -8.03166866e-01
2.30500042e-01 6.71539128e-01 2.08565816e-01 2.80393541e-01
-3.46959680e-01 -2.78752953e-01 -4.92397517e-01 -1.92992523e-01
1.75580800e-01 -5.93013763e-01 -2.83450305e-01 3.15374434e-01
-1.21945512e+00 9.22313631e-01 -1.35798201e-01 6.28468633e-01
-2.29884341e-01 -5.73552132e-01 5.33331633e-01 3.80621135e-01
6.24758840e-01 5.53806163e-02 -3.00721139e-01 -3.92017424e-01
-1.13024652e+00 -4.67373244e-02 9.33361590e-01 6.00112736e-01
5.37048936e-01 9.56097841e-02 2.30547369e-01 1.50604165e+00
5.82517922e-01 3.72645229e-01 6.52338386e-01 -5.16694367e-01
5.91452777e-01 2.81029761e-01 4.06042188e-02 -4.33776706e-01
-2.11539388e-01 -3.18597909e-03 -7.72435486e-01 8.83429945e-02
4.44859296e-01 -2.88337260e-01 -9.58548844e-01 1.42426884e+00
2.15331689e-01 -1.55723706e-01 4.29476589e-01 7.04301119e-01
8.89406145e-01 8.29586148e-01 -1.98899016e-01 -3.87076020e-01
1.54618037e+00 -7.42728949e-01 -7.62209237e-01 -3.69729310e-01
4.00845110e-01 -1.34949481e+00 1.22730505e+00 4.32520658e-01
-8.46081972e-01 -4.19350803e-01 -8.03426623e-01 2.12270439e-01
-3.59910786e-01 2.78015345e-01 8.53406042e-02 9.97829914e-01
-1.32955170e+00 -3.05320621e-01 -4.10701364e-01 -8.26904833e-01
-2.52320230e-01 4.70465302e-01 -5.88878632e-01 -1.36669464e-02
-1.16946805e+00 7.23589242e-01 1.08531125e-01 1.24052148e-02
-4.39234018e-01 -1.19984075e-01 -8.03343892e-01 -4.75530177e-01
1.25782743e-01 3.60468090e-01 1.30642462e+00 -7.71724343e-01
-1.56016052e+00 1.11465967e+00 -4.34449375e-01 -2.21000358e-01
4.48466063e-01 7.12773502e-02 -8.26144695e-01 -2.57681906e-01
2.14984387e-01 5.02584457e-01 5.41586339e-01 -1.09693170e+00
-7.40133762e-01 -3.61975104e-01 -3.34442228e-01 3.22602332e-01
-6.98689744e-02 8.30268443e-01 -5.53414762e-01 -6.47408724e-01
1.83623612e-01 -1.05600810e+00 1.90847330e-02 -6.86691880e-01
-4.36451852e-01 -2.01039746e-01 1.00542307e+00 -1.09052861e+00
1.19835365e+00 -2.46531963e+00 -2.92690307e-01 1.09435663e-01
-4.24283057e-01 4.21380579e-01 1.18254371e-01 5.63179553e-01
-3.97490300e-02 1.06856629e-01 -2.60082871e-01 -5.60143292e-01
-1.49815723e-01 2.33710647e-01 -1.77514389e-01 3.98832262e-01
-1.10930450e-01 1.39262512e-01 -4.56177831e-01 -2.97121733e-01
1.31659597e-01 4.96367782e-01 -1.53425500e-01 2.23994493e-01
2.11081997e-01 5.56276858e-01 9.72189680e-02 5.98751843e-01
5.95650136e-01 5.29019177e-01 -1.78638678e-02 1.66358918e-01
-6.14592731e-01 7.14833379e-01 -1.56417334e+00 1.10808098e+00
-5.61183214e-01 7.40448117e-01 5.84768653e-01 -9.32101369e-01
1.14816201e+00 7.99199700e-01 6.51489124e-02 -3.38288546e-01
-6.94009736e-02 6.60680473e-01 3.73347044e-01 -2.77474672e-01
6.11446917e-01 -1.19117074e-01 -1.80944249e-01 5.97898602e-01
-9.53322574e-02 -2.91113347e-01 1.36207968e-01 -1.70282591e-02
6.85391665e-01 -6.16431296e-01 3.16676080e-01 -3.02847505e-01
9.35691297e-01 -2.25036263e-01 4.18341070e-01 6.07555866e-01
-5.27219534e-01 7.94027567e-01 9.56239328e-02 -6.09923340e-02
-9.34989572e-01 -8.72754931e-01 -2.36271948e-01 1.06034577e+00
-3.50724459e-01 -1.70141771e-01 -6.96474373e-01 -2.68816590e-01
-5.23651600e-01 5.98595798e-01 1.33809783e-02 1.49730012e-01
-4.90002960e-01 -6.59936965e-01 9.84703124e-01 1.40573047e-02
7.32990921e-01 -1.29230523e+00 2.13807598e-01 4.17854130e-01
-4.56570953e-01 -1.31685770e+00 -1.05612791e+00 4.01461571e-01
-1.69312268e-01 -5.53165555e-01 -8.58156860e-01 -1.21979249e+00
2.48728037e-01 2.66252279e-01 6.79081202e-01 -5.12266636e-01
-2.07991898e-01 3.60052556e-01 -2.89816022e-01 -4.72480685e-01
-1.12696218e+00 2.19371468e-02 5.78339577e-01 3.33120763e-01
5.25140882e-01 -2.01844871e-01 2.85923462e-02 4.08970982e-01
-5.97164631e-01 -3.75706494e-01 2.58974224e-01 6.14356875e-01
2.46794045e-01 2.25608230e-01 6.85931265e-01 -4.80496585e-01
8.98279011e-01 -2.70910472e-01 -4.66987938e-01 1.46742925e-01
-1.07402995e-01 -1.69760451e-01 5.17264247e-01 -5.87341249e-01
-9.01242018e-01 3.41498166e-01 -5.96890807e-01 3.60826566e-03
-5.78729689e-01 4.85450983e-01 -4.89195645e-01 -4.69983108e-02
3.43055278e-01 5.48061967e-01 1.33617908e-01 -6.62928402e-01
-7.22945342e-03 1.75928724e+00 6.32226348e-01 -3.05431873e-01
5.64295888e-01 2.61783078e-02 -7.39802480e-01 -1.62994444e+00
-2.00033262e-01 -9.72843051e-01 -5.97500563e-01 -2.77968850e-02
8.37081254e-01 -1.19505394e+00 -3.54871809e-01 8.61809552e-01
-1.15999818e+00 -1.11017622e-01 1.64419562e-01 6.86196327e-01
-1.82313457e-01 4.10092175e-01 -5.17662048e-01 -1.14387393e+00
-2.52485573e-01 -1.54586220e+00 9.26608622e-01 1.15895882e-01
-6.07284904e-01 -8.96516562e-01 1.78496107e-01 5.84628999e-01
4.66194123e-01 -6.76674187e-01 7.11648285e-01 -9.28921580e-01
1.03170663e-01 -9.98262540e-02 3.43917347e-02 7.28129327e-01
7.09167361e-01 5.66463545e-02 -1.20899963e+00 -2.98528433e-01
-5.93707450e-02 -2.34392688e-01 5.67974448e-01 2.93868750e-01
3.38634163e-01 -2.76056677e-01 1.01665199e-01 5.73044531e-02
9.14252460e-01 6.47242427e-01 3.80723834e-01 1.31531492e-01
4.67542022e-01 7.23269045e-01 3.60501617e-01 3.25006992e-01
4.32582080e-01 8.31908643e-01 -2.01000407e-01 -1.83764130e-01
-1.14656091e-01 1.11945406e-01 9.34175551e-01 1.26787603e+00
2.95148045e-01 -5.04388697e-02 -1.18599355e+00 8.53820920e-01
-1.24264884e+00 -8.49824429e-01 -8.88666734e-02 2.49209356e+00
1.09022057e+00 -1.08117431e-01 5.07427692e-01 4.10980999e-01
8.85004580e-01 7.66345263e-02 -5.62216751e-02 -8.43474805e-01
-3.16794425e-01 1.79051414e-01 3.49648178e-01 9.53876257e-01
-1.21943545e+00 9.96238589e-01 6.13593102e+00 7.42806196e-01
-1.49392033e+00 6.15644157e-02 6.02579355e-01 1.77513227e-01
1.79472759e-01 -2.22925812e-01 -1.06216776e+00 4.10311669e-01
1.15917182e+00 -1.28392443e-01 6.29787564e-01 4.09325361e-01
7.19009161e-01 -3.00377578e-01 -8.63426089e-01 1.24243522e+00
2.21286580e-01 -7.40710318e-01 -1.44535244e-01 2.69314080e-01
5.06436527e-01 3.45371425e-01 8.64496231e-02 1.94464579e-01
2.87850261e-01 -1.14404285e+00 6.58634484e-01 -1.42851487e-01
7.62813210e-01 -9.75017250e-01 6.79345906e-01 5.40745080e-01
-1.25808358e+00 3.94696631e-02 -5.83559386e-02 1.02104031e-01
7.12919235e-02 1.83756545e-01 -1.18892670e+00 1.64126426e-01
8.13615620e-01 1.20131738e-01 -3.32673192e-01 9.12328303e-01
1.29167825e-01 1.02706683e+00 -5.01439273e-01 1.19118616e-01
3.57524872e-01 -2.59131312e-01 7.66508877e-01 1.37504220e+00
3.53046983e-01 -1.66446313e-01 3.14131021e-01 4.46402848e-01
7.16809705e-02 4.74107444e-01 -7.47212410e-01 -2.07805023e-01
4.84345376e-01 9.67818618e-01 -5.67857146e-01 -6.10194579e-02
-5.98916054e-01 1.05790174e+00 -1.86193019e-01 2.68016398e-01
-2.55644500e-01 -3.29969466e-01 1.01872313e+00 7.06587136e-02
4.09285509e-04 -3.03585172e-01 1.36146888e-01 -6.86698914e-01
8.10069516e-02 -1.41644502e+00 7.04593882e-02 -3.47654969e-01
-1.18749940e+00 9.01411653e-01 -3.21429491e-01 -1.22064018e+00
-5.04176974e-01 -3.53626043e-01 -6.17851317e-01 1.40920365e+00
-1.08494711e+00 -1.09888434e+00 1.48956060e-01 6.81523263e-01
9.63216305e-01 -7.55237520e-01 9.89559889e-01 5.78081846e-01
-5.65046072e-01 6.55857503e-01 2.53724605e-01 3.72188151e-01
8.97865415e-01 -1.05150330e+00 4.49584126e-01 8.62911880e-01
4.65856940e-01 4.95240301e-01 6.30182326e-01 -4.01511401e-01
-9.78811204e-01 -1.02305722e+00 1.34269142e+00 1.64676760e-03
7.57058203e-01 -6.00313127e-01 -9.22141075e-01 3.43871981e-01
4.24994528e-01 -2.07134262e-01 7.73415267e-01 -7.44061992e-02
-1.61590621e-01 -1.10639827e-02 -1.00765979e+00 5.36655724e-01
3.80584210e-01 -1.01306856e+00 -3.98789287e-01 3.15627366e-01
4.24855232e-01 -2.66875196e-02 -3.32780004e-01 -1.35047182e-01
2.78038979e-01 -8.40128660e-01 6.27199829e-01 -1.48491472e-01
-3.94614846e-01 -5.42769015e-01 -5.02012730e-01 -1.46104884e+00
2.23879069e-01 -8.69495332e-01 6.45861030e-01 1.84184110e+00
3.98728818e-01 -8.06645274e-01 4.49350804e-01 2.88264215e-01
9.49404761e-03 9.66565609e-02 -1.23515546e+00 -8.94579649e-01
7.11762607e-02 -6.11830056e-01 2.16337487e-01 8.01176548e-01
-1.66269541e-02 4.85881388e-01 -3.58346432e-01 3.52126867e-01
3.53537738e-01 -5.64401388e-01 6.87080503e-01 -1.13353360e+00
-1.12698466e-01 -3.86936456e-01 -3.87197345e-01 -7.38538027e-01
3.33671927e-01 -7.66946375e-01 2.76450992e-01 -1.39289415e+00
-7.04450980e-02 -4.28030908e-01 1.05494205e-02 5.12375772e-01
4.36683558e-02 2.41655052e-01 1.07514963e-01 1.85275823e-01
8.86254683e-02 1.78316787e-01 4.80899304e-01 -4.00152802e-01
-6.64994895e-01 3.69094878e-01 -6.20084405e-01 6.25842988e-01
7.99274921e-01 -1.87761828e-01 -1.63314283e-01 -9.87631455e-02
-4.36014354e-01 6.15611710e-02 4.00326448e-03 -1.02672744e+00
2.86739916e-01 1.41865820e-01 -3.36317986e-01 -5.99972606e-01
3.90990376e-01 -6.57774210e-01 1.97769791e-01 1.48967206e-01
-1.48053452e-01 3.77004631e-02 3.51265281e-01 -8.26908946e-02
-7.40930080e-01 -1.08880885e-01 1.07427931e+00 -1.08635493e-01
-6.59436643e-01 -8.65840465e-02 -9.69225645e-01 -7.43964612e-02
9.01654541e-01 -3.48048955e-01 4.72610891e-02 -7.61191487e-01
-5.74281454e-01 8.95068422e-02 2.67272741e-01 5.62977254e-01
3.68532509e-01 -1.02477837e+00 -1.11438990e+00 5.30931592e-01
1.88048199e-01 -3.46514314e-01 -7.64728040e-02 9.35546279e-01
-4.06953812e-01 5.08869469e-01 9.46690738e-02 -3.93631667e-01
-1.86064124e+00 -4.20750156e-02 4.58896846e-01 6.32417807e-03
-2.96280533e-01 8.06900859e-01 7.57988766e-02 -8.00111294e-01
5.59073746e-01 -5.57330251e-02 -4.19837475e-01 3.30149382e-01
7.34499216e-01 1.80560693e-01 3.30323756e-01 -1.40027988e+00
-5.61671555e-01 4.16686893e-01 -2.87031263e-01 -6.36605501e-01
9.55568671e-01 -3.83269399e-01 -3.02769989e-01 9.01819468e-01
1.21727848e+00 5.66975713e-01 -4.37858552e-01 -3.59230131e-01
1.48868889e-01 4.41418067e-02 1.23536631e-01 -7.48944938e-01
-6.65537775e-01 9.40911889e-01 7.44890332e-01 1.83051199e-01
1.10970628e+00 2.16023415e-01 5.70747137e-01 1.97136447e-01
3.04007530e-01 -1.18682194e+00 -5.11138082e-01 9.25323486e-01
8.86396170e-01 -1.38634038e+00 -6.22892380e-01 -2.64079183e-01
-8.18010032e-01 9.18638766e-01 2.36830935e-01 4.00940418e-01
6.45725191e-01 4.47714180e-01 6.46848202e-01 3.86800438e-01
-4.32353437e-01 -4.96053368e-01 2.39312544e-01 6.74093068e-01
8.08037400e-01 1.76539585e-01 -2.54834324e-01 3.85031492e-01
-3.22215438e-01 -5.23720980e-01 6.52298152e-01 7.11161137e-01
-3.88801277e-01 -1.39416122e+00 -9.72997427e-01 2.47725517e-01
-7.06319451e-01 -3.60320956e-01 -7.10508108e-01 5.32228649e-01
4.76595424e-02 1.39604962e+00 6.22498654e-02 -3.49246770e-01
2.14098305e-01 3.98561269e-01 -1.12069473e-01 -8.05088997e-01
-8.02528799e-01 5.31125069e-01 2.49340564e-01 5.40481880e-02
-3.11596990e-01 -9.42948282e-01 -1.22034168e+00 -1.02132939e-01
-1.08213373e-01 3.66336256e-01 7.99572468e-01 9.16996300e-01
-6.74494952e-02 -1.79010145e-02 8.05771649e-01 -6.91674590e-01
-3.05682123e-01 -1.25698841e+00 -9.42317724e-01 4.49285470e-02
5.64039648e-01 -3.09329540e-01 -6.07070267e-01 5.77124506e-02]
|
[14.178839683532715, 6.546756744384766]
|
e562e1f5-169c-402f-810a-5a060646515e
|
automated-surface-texture-analysis-via
|
2204.05968
| null |
https://arxiv.org/abs/2204.05968v1
|
https://arxiv.org/pdf/2204.05968v1.pdf
|
Automated Surface Texture Analysis via Discrete Cosine Transform and Discrete Wavelet Transform
|
Surface roughness and texture are critical to the functional performance of engineering components. The ability to analyze roughness and texture effectively and efficiently is much needed to ensure surface quality in many surface generation processes, such as machining, surface mechanical treatment, etc. Discrete Wavelet Transform (DWT) and Discrete Cosine Transform (DCT) are two commonly used signal decomposition tools for surface roughness and texture analysis. Both methods require selecting a threshold to decompose a given surface into its three main components: form, waviness, and roughness. However, although DWT and DCT are part of the ISO surface finish standards, there exists no systematic guidance on how to compute these thresholds, and they are often manually selected on case by case basis. This makes utilizing these methods for studying surfaces dependent on the user's judgment and limits their automation potential. Therefore, we present two automatic threshold selection algorithms based on information theory and signal energy. We use machine learning to validate the success of our algorithms both using simulated surfaces as well as digital microscopy images of machined surfaces. Specifically, we generate feature vectors for each surface area or profile and apply supervised classification. Comparing our results with the heuristic threshold selection approach shows good agreement with mean accuracies as high as 95\%. We also compare our results with Gaussian filtering (GF) and show that while GF results for areas can yield slightly higher accuracies, our results outperform GF for surface profiles. We further show that our automatic threshold selection has significant advantages in terms of computational time as evidenced by decreasing the number of mode computations by an order of magnitude compared to the heuristic thresholding for DCT.
|
['Yang Guo', 'Firas A. Khasawneh', 'Jisheng Chen', 'Melih C. Yesilli']
|
2022-04-12
| null | null | null | null |
['texture-classification']
|
['computer-vision']
|
[ 9.65264261e-01 -2.28533953e-01 2.74860978e-01 -1.29592076e-01
-8.45916927e-01 -2.15111062e-01 3.28414619e-01 5.09476840e-01
-3.32870066e-01 3.41256201e-01 -3.14847410e-01 -2.15987965e-01
-3.58251691e-01 -1.11190724e+00 -2.02490687e-01 -8.74787986e-01
1.35748684e-02 2.17087492e-01 4.83401418e-01 -2.83035040e-01
6.84470534e-01 6.93500757e-01 -2.00588465e+00 4.09138232e-01
8.31397891e-01 1.33863378e+00 2.48401925e-01 5.79878509e-01
-2.37349004e-01 -1.30400419e-01 -3.87240708e-01 -2.71129124e-02
1.37812793e-01 -2.67390460e-01 -4.86767113e-01 1.82045519e-01
1.81541339e-01 5.06349728e-02 6.55728698e-01 9.69770730e-01
3.59407663e-01 3.45137492e-02 1.07487035e+00 -3.94765496e-01
-1.46780044e-01 -1.68699637e-01 -6.44056737e-01 -1.01323590e-01
4.38864380e-01 -6.58924505e-02 9.10191178e-01 -8.77511859e-01
5.27024448e-01 9.11284149e-01 8.12983274e-01 3.00508514e-02
-1.27659595e+00 -3.52472663e-01 -4.41284567e-01 8.79685581e-02
-1.17615032e+00 -5.54297268e-01 8.93512368e-01 -5.69759667e-01
8.48696709e-01 3.89079839e-01 4.92671162e-01 2.59354085e-01
8.04543257e-01 1.30773827e-01 1.21864116e+00 -7.82740295e-01
6.04852080e-01 -1.42381579e-01 3.57806049e-02 4.75608349e-01
3.71504188e-01 -6.39748946e-02 -1.95080787e-01 -2.98061877e-01
8.07070374e-01 -3.73750687e-01 -1.95030183e-01 -9.51550081e-02
-9.16633844e-01 7.80308425e-01 -3.71794879e-01 5.44523716e-01
-6.52661622e-01 7.99153522e-02 3.92747909e-01 5.42114556e-01
6.61169112e-01 5.02259910e-01 -2.29623705e-01 -2.70128489e-01
-9.17174697e-01 2.84962468e-02 6.60708547e-01 2.84390241e-01
7.68391550e-01 -2.27011338e-01 2.38633826e-01 1.15345800e+00
2.06249148e-01 6.55315995e-01 2.50149459e-01 -1.08316076e+00
-2.54034817e-01 5.29381394e-01 1.58911601e-01 -1.41459310e+00
-3.08093786e-01 1.28118455e-01 -7.02938139e-01 5.82472444e-01
2.77319193e-01 2.74820536e-01 -9.25272405e-01 9.24045384e-01
1.78716585e-01 -4.05449152e-01 -1.38853475e-01 5.41945994e-01
3.57471049e-01 4.24222082e-01 -2.37264574e-01 -5.03857911e-01
1.38653052e+00 5.66029176e-02 -9.20679629e-01 8.13142583e-02
3.93038481e-01 -1.18057573e+00 1.20656085e+00 8.96855950e-01
-1.06852460e+00 -3.41147751e-01 -1.32159781e+00 2.87632883e-01
-1.65398940e-01 3.03139657e-01 6.34609401e-01 6.63950443e-01
-8.38453293e-01 9.62157428e-01 -1.12952125e+00 -2.68600553e-01
2.27018446e-01 4.39628065e-01 -3.97099435e-01 3.22885043e-03
-8.73949707e-01 8.82496595e-01 -5.89253306e-01 1.18023410e-01
9.40941349e-02 -2.97523916e-01 -7.19070554e-01 -2.18330845e-01
-1.23757802e-01 -1.91628113e-01 1.02269101e+00 -5.20721793e-01
-1.83750892e+00 8.53382826e-01 -4.03998852e-01 -1.31379133e-02
2.42264137e-01 -9.87583697e-02 -3.17205697e-01 4.61551249e-01
-2.65726889e-03 -4.72479500e-03 8.96756828e-01 -1.12674689e+00
-5.06392658e-01 -4.50551003e-01 -5.14252305e-01 -2.19627723e-01
-1.53425291e-01 -1.29030287e-01 -7.05682263e-02 -3.07135731e-01
6.09624028e-01 -5.52989662e-01 -3.12370658e-01 8.76193494e-03
2.16133948e-02 -1.11659475e-01 6.77419007e-01 -6.83459103e-01
1.14025664e+00 -2.22896862e+00 -3.96128714e-01 7.00787067e-01
2.34039314e-03 -1.69947997e-01 4.84666117e-02 5.90253890e-01
1.86011255e-01 1.43105581e-01 -4.75838274e-01 -4.75003049e-02
-2.22311720e-01 -4.73527424e-02 -4.01130132e-03 7.53346682e-01
2.12285072e-01 2.29005277e-01 -6.04143500e-01 -3.32248390e-01
4.18280840e-01 5.88356674e-01 -4.82642174e-01 -2.46727273e-01
1.37651160e-01 3.41816917e-02 -4.77924466e-01 7.14661896e-01
5.55066884e-01 8.02464038e-02 1.09410360e-01 -5.52299619e-01
-3.41431916e-01 1.37159988e-01 -1.20944774e+00 9.11887109e-01
-7.06644118e-01 5.92499614e-01 3.94648820e-01 -9.11799371e-01
1.50909376e+00 4.13798481e-01 7.58706212e-01 -6.92106724e-01
2.22622752e-01 8.01737845e-01 3.95705774e-02 -5.05720675e-01
4.00446057e-01 -5.67247093e-01 2.92635560e-01 1.48403749e-01
-4.98083055e-01 -8.28779697e-01 -2.50196494e-02 -4.05789256e-01
1.22021616e+00 -1.91980422e-01 2.34281600e-01 -6.25461996e-01
4.66021359e-01 1.24061540e-01 2.60610878e-01 2.68489897e-01
7.32429400e-02 7.75163531e-01 4.80348885e-01 -2.20736504e-01
-9.57181334e-01 -6.49368405e-01 -5.47864437e-01 4.23703402e-01
2.21757740e-01 -4.08940986e-02 -6.71743035e-01 7.84774050e-02
1.39217094e-01 5.88367283e-01 -6.28965735e-01 2.47369241e-02
-4.27541107e-01 -7.82315910e-01 -8.67083296e-02 1.17298856e-01
1.45341426e-01 -9.78310823e-01 -9.70347464e-01 3.55580539e-01
2.17056304e-01 -9.03697073e-01 -1.67761147e-01 2.08119333e-01
-1.06134784e+00 -1.20076036e+00 -4.02572095e-01 -5.28072596e-01
6.32995248e-01 2.80335099e-01 8.42918038e-01 1.83148265e-01
-5.55783451e-01 4.68927771e-01 -5.32834709e-01 -4.43295896e-01
-4.91057307e-01 -3.23011994e-01 2.93858089e-02 1.09554291e-01
3.77296299e-01 -7.43758321e-01 -7.60745287e-01 5.83140075e-01
-8.68583918e-01 -2.14547619e-01 6.63749099e-01 5.78949690e-01
9.80972111e-01 4.62399840e-01 6.67890131e-01 -7.94367194e-01
9.29682314e-01 -3.15150209e-02 -4.97646302e-01 -1.32816508e-01
-7.85789371e-01 -1.79721519e-01 3.68770391e-01 -1.74339026e-01
-7.19337463e-01 9.87221487e-03 -3.42475891e-01 -3.16891596e-02
-1.20785289e-01 5.52140892e-01 1.64629743e-01 -3.81885290e-01
7.19965041e-01 -8.99859611e-03 6.45240545e-01 -3.62862200e-01
-1.69379160e-01 8.88344526e-01 3.66142601e-01 -4.89517510e-01
4.31373179e-01 8.07192624e-01 1.27534986e-01 -1.50467551e+00
-2.13536873e-01 -7.14772284e-01 -4.32012916e-01 -4.54959899e-01
8.75514269e-01 -2.11878583e-01 -7.85138011e-01 3.48383188e-01
-9.17090178e-01 -2.97730774e-01 -1.86166570e-01 4.46162879e-01
-7.35532165e-01 5.84058762e-01 -4.50498849e-01 -1.01634192e+00
-4.48266119e-01 -1.16859114e+00 1.35598969e+00 -2.16359124e-01
-5.87525666e-01 -1.02147019e+00 -1.33346036e-01 2.58290112e-01
7.09524751e-01 6.69830799e-01 1.06405222e+00 -9.89618823e-02
1.09552756e-01 -6.38882339e-01 1.57588631e-01 5.08366704e-01
6.86890364e-01 3.97860378e-01 -8.17757845e-01 -1.86241925e-01
4.33745086e-01 3.03263403e-02 7.80263901e-01 6.47191107e-01
7.80262947e-01 2.46962443e-01 -4.26638275e-01 1.06683433e-01
1.51365924e+00 2.72530764e-01 9.06308711e-01 2.98300058e-01
7.46955574e-02 1.01903927e+00 9.22785163e-01 3.22776556e-01
-4.01632965e-01 6.83654666e-01 3.00882965e-01 -1.09405808e-01
-1.07966430e-01 3.99002761e-01 4.09919322e-01 8.67077887e-01
-3.78544778e-01 9.98130888e-02 -8.44958723e-01 5.72904885e-01
-1.06969118e+00 -7.96262443e-01 -4.09093976e-01 2.50375795e+00
6.66937530e-01 2.84423053e-01 -2.27699071e-01 7.81701624e-01
6.51656449e-01 -3.41894656e-01 -1.13526843e-01 -7.15723097e-01
8.72960240e-02 6.57197416e-01 5.47216296e-01 6.29602492e-01
-7.15779603e-01 4.73138958e-01 6.47649574e+00 9.69072044e-01
-1.39881670e+00 -3.42240423e-01 3.38939637e-01 3.43538880e-01
-4.11684632e-01 -2.88982362e-01 -4.21898663e-01 2.17868209e-01
6.76716745e-01 7.07170740e-02 1.45316109e-01 4.33919698e-01
5.48945665e-01 -5.53639829e-01 -7.69596100e-01 9.01427090e-01
-3.92155468e-01 -1.13689673e+00 -2.39992112e-01 2.40665123e-01
4.12173957e-01 -2.90632784e-01 -1.16294049e-01 -3.54556382e-01
-1.60695344e-01 -1.03105712e+00 4.74963963e-01 4.45321232e-01
9.56643939e-01 -5.94300747e-01 9.87924159e-01 -4.74870279e-02
-1.28495550e+00 3.50863248e-01 -3.64575326e-01 -9.04401988e-02
2.41895378e-01 1.31092715e+00 -6.27153277e-01 4.13041204e-01
5.76385200e-01 3.16404819e-01 -3.32192071e-02 7.98845887e-01
1.98061556e-01 6.91287279e-01 -6.68517411e-01 -1.52737915e-03
-8.62976685e-02 -5.36548257e-01 4.39097196e-01 8.25309157e-01
5.86284041e-01 2.28514761e-01 -1.96206227e-01 5.58954477e-01
6.85033739e-01 5.12106538e-01 -4.58400607e-01 8.57639089e-02
6.20983481e-01 1.12760186e+00 -1.19450152e+00 1.19367912e-01
-4.72455859e-01 6.84635401e-01 -3.81546736e-01 4.41078134e-02
-2.67159224e-01 -7.53032386e-01 6.20334387e-01 6.08964443e-01
3.11074495e-01 -5.17360866e-01 -9.02083397e-01 -3.95929009e-01
4.33892682e-02 -7.34316468e-01 -7.95881078e-02 -2.33636767e-01
-1.31610036e+00 3.33025038e-01 -8.25859532e-02 -1.23848891e+00
1.80198744e-01 -9.20490086e-01 -6.40161693e-01 7.55862832e-01
-1.17258823e+00 -6.32119060e-01 -3.00505251e-01 1.01802155e-01
3.22020322e-01 2.28525892e-01 8.13491464e-01 1.86814085e-01
-7.41755068e-02 7.32491910e-02 2.14486122e-01 -3.85888159e-01
4.21900004e-01 -1.03670371e+00 2.39669085e-01 3.13817650e-01
-4.34576601e-01 4.63166386e-01 8.84309590e-01 -7.61476815e-01
-1.50315142e+00 -4.64059204e-01 6.47547960e-01 -8.64310116e-02
4.92163211e-01 -1.38619423e-01 -1.05765593e+00 -1.35789245e-01
-2.34743610e-01 -4.18294221e-01 5.96441329e-01 -4.24523838e-03
4.17300522e-01 5.14710918e-02 -1.26235938e+00 4.47978765e-01
5.88544190e-01 -2.47990370e-01 -5.60451269e-01 8.41577500e-02
7.16879405e-03 -1.59359761e-02 -1.30961931e+00 8.31026971e-01
8.16105604e-01 -1.09751618e+00 5.99854469e-01 3.31614882e-01
3.45229626e-01 -3.09038520e-01 -4.05132651e-01 -9.81883109e-01
-1.99228495e-01 -4.07207370e-01 6.69115901e-01 9.55952406e-01
5.20378053e-01 -8.33671510e-01 8.31562936e-01 4.26016659e-01
-2.68092811e-01 -9.90430355e-01 -1.00618720e+00 -7.62071311e-01
-7.06950426e-02 -6.57690823e-01 2.03110948e-01 7.53267705e-01
1.01684764e-01 -2.54678521e-02 3.13802630e-01 4.32509854e-02
6.09330833e-01 1.58238396e-01 5.18494368e-01 -1.64329803e+00
1.39138162e-01 -6.09742463e-01 -6.29834950e-01 -4.51043993e-01
-3.62596482e-01 -4.40089136e-01 3.76944482e-01 -1.70084131e+00
-3.22754711e-01 -6.45104110e-01 5.28814942e-02 3.39162767e-01
1.34042680e-01 2.60701805e-01 -4.77841794e-01 3.57161671e-01
3.19473833e-01 4.01338637e-01 1.19204664e+00 4.25645970e-02
-4.56948131e-01 -2.55093426e-02 -3.22783589e-01 7.27203250e-01
7.82696605e-01 -1.44084215e-01 -1.58545747e-01 1.69422999e-02
2.80211747e-01 -8.81891251e-02 -6.66581746e-03 -1.07697856e+00
-1.32106826e-01 -1.52296916e-01 1.14900023e-01 -2.06375822e-01
2.08209842e-01 -8.34109247e-01 2.89546758e-01 5.13527036e-01
1.08235301e-02 -2.35877708e-01 3.07088494e-01 4.46327746e-01
-4.24273521e-01 -1.68148130e-01 1.12014401e+00 1.25429019e-01
-3.85577738e-01 -2.16416299e-01 -7.03996360e-01 -5.94180107e-01
8.13489795e-01 -8.88722181e-01 -2.05725990e-02 -4.59080815e-01
-6.82036698e-01 -2.40893140e-01 9.05111253e-01 -7.40178451e-02
7.56097734e-01 -9.14732277e-01 -5.34432530e-01 4.01962817e-01
-1.55666187e-01 -1.47229180e-01 1.43862098e-01 1.12467945e+00
-8.79146636e-01 5.63256480e-02 -4.60797064e-02 -7.48392344e-01
-1.22359502e+00 7.05809891e-02 1.22685388e-01 -1.37546333e-02
-6.05446696e-01 5.29669225e-01 -2.72549856e-02 9.19940397e-02
-4.00085986e-01 -5.70609808e-01 -2.95470804e-01 1.38703227e-01
2.95695215e-01 6.25656426e-01 6.30302012e-01 -3.48448008e-01
-3.85877490e-01 1.26169693e+00 2.27882534e-01 -1.44603938e-01
1.34864497e+00 3.90853509e-02 -3.41767639e-01 4.45262015e-01
1.12516236e+00 5.40324271e-01 -9.40986872e-01 4.45973516e-01
1.61507130e-01 -3.10257494e-01 3.48477781e-01 -2.08059058e-01
-8.08692217e-01 7.62209237e-01 6.44466162e-01 5.56571424e-01
1.32102692e+00 -1.46660864e-01 7.94405878e-01 1.44790247e-01
6.00396931e-01 -1.32892358e+00 -4.58886772e-02 1.39021948e-01
9.80829298e-01 -8.12229753e-01 3.19058925e-01 -1.02476966e+00
-2.02948704e-01 1.34074736e+00 -1.23432660e-02 -2.62914985e-01
8.43852282e-01 5.09629846e-01 3.08142930e-01 -5.64347923e-01
-6.01457894e-01 -1.77972227e-01 1.51694968e-01 3.50403160e-01
6.83058321e-01 -4.33134213e-02 -7.48221576e-01 3.46162021e-01
-3.07182699e-01 3.40188481e-02 4.72833872e-01 1.13926351e+00
-9.95015264e-01 -1.10538089e+00 -5.25589764e-01 8.38809967e-01
-6.55571699e-01 1.59795210e-01 -1.44575134e-01 7.80034363e-01
-9.68021676e-02 1.13544810e+00 8.41128603e-02 -4.81324196e-01
5.71184635e-01 -7.22480863e-02 4.49750930e-01 -6.30987465e-01
-1.90655544e-01 1.00587256e-01 2.72688150e-01 -5.15336633e-01
-4.31903094e-01 -8.16145480e-01 -1.30909562e+00 -7.14586228e-02
-6.23308480e-01 2.32817546e-01 1.00403738e+00 7.61522174e-01
2.45658875e-01 3.95249456e-01 5.84475040e-01 -1.00162530e+00
-3.96755666e-01 -8.40902746e-01 -8.98157299e-01 4.20828670e-01
5.39231934e-02 -1.00751686e+00 -8.07251930e-01 1.80044428e-01]
|
[13.001444816589355, -2.811244249343872]
|
c624e3ea-8337-42ea-9ea8-0330aedfe624
|
unsupervised-vision-and-vision-motion
|
2210.00413
| null |
https://arxiv.org/abs/2210.00413v2
|
https://arxiv.org/pdf/2210.00413v2.pdf
|
Unsupervised Visual Odometry and Action Integration for PointGoal Navigation in Indoor Environment
|
PointGoal navigation in indoor environment is a fundamental task for personal robots to navigate to a specified point. Recent studies solved this PointGoal navigation task with near-perfect success rate in photo-realistically simulated environments, under the assumptions with noiseless actuation and most importantly, perfect localization with GPS and compass sensors. However, accurate GPS signalis difficult to be obtained in real indoor environment. To improve the PointGoal navigation accuracy without GPS signal, we use visual odometry (VO) and propose a novel action integration module (AIM) trained in unsupervised manner. Sepecifically, unsupervised VO computes the relative pose of the agent from the re-projection error of two adjacent frames, and then replaces the accurate GPS signal with the path integration. The pseudo position estimated by VO is used to train action integration which assists agent to update their internal perception of location and helps improve the success rate of navigation. The training and inference process only use RGB, depth, collision as well as self-action information. The experiments show that the proposed system achieves satisfactory results and outperforms the partially supervised learning algorithms on the popular Gibson dataset.
|
['Chuan Lin', 'YongJie Li', 'Fuya Luo', 'Xianshi Zhang', 'Yijun Cao']
|
2022-10-02
| null | null | null | null |
['pointgoal-navigation']
|
['robots']
|
[-1.17726713e-01 1.23718888e-01 6.99901208e-02 -3.25598359e-01
-3.10793191e-01 -3.96480858e-01 5.35153091e-01 -6.30418807e-02
-8.37129712e-01 1.10450566e+00 -2.94674814e-01 2.61610240e-01
-8.41657724e-03 -8.45160484e-01 -8.98409724e-01 -8.54984760e-01
-1.12726584e-01 5.34382582e-01 5.54414749e-01 -5.49089372e-01
3.22045833e-01 2.63029337e-01 -1.54918134e+00 -7.35298276e-01
1.15281117e+00 8.54579389e-01 7.80922234e-01 6.58344209e-01
2.69129127e-01 6.52422607e-01 -4.82846946e-01 3.72609317e-01
1.54719815e-01 -1.04424894e-01 -2.45956793e-01 8.90392694e-05
-1.28159896e-01 -2.70426631e-01 -2.64884144e-01 1.08491910e+00
7.33061254e-01 5.59954107e-01 4.02400583e-01 -1.39843130e+00
-2.75414795e-01 6.46374449e-02 -1.29421309e-01 -2.25237250e-01
9.23674107e-01 3.75243664e-01 2.45646805e-01 -3.92200023e-01
7.64106214e-01 1.07717037e+00 8.89037549e-01 3.43338430e-01
-7.17535734e-01 -2.73419350e-01 1.24295652e-01 3.95967871e-01
-1.60243273e+00 -1.67390034e-01 6.85044825e-01 1.17801316e-02
1.01094413e+00 -7.36409947e-02 9.93430495e-01 9.93668079e-01
7.13717461e-01 4.46029156e-01 8.94354701e-01 -1.61419243e-01
6.60978198e-01 6.18348038e-03 -3.69734734e-01 8.12250555e-01
3.94898623e-01 1.60129324e-01 -4.33976144e-01 2.43971452e-01
1.00988674e+00 2.35795930e-01 -4.28084046e-01 -9.16934609e-01
-1.31004226e+00 3.85642081e-01 1.09078109e+00 9.66257751e-02
-8.18927705e-01 4.02856380e-01 -2.08425865e-01 1.46671504e-01
-4.06591028e-01 1.52862832e-01 -3.68541926e-01 -4.63774353e-01
-1.04704887e-01 4.26921993e-01 7.37073362e-01 1.29028165e+00
8.51372719e-01 1.85455590e-01 5.28262854e-01 4.31039155e-01
6.27724588e-01 9.37837064e-01 8.37827146e-01 -1.26471186e+00
4.28418040e-01 5.36153197e-01 7.98257649e-01 -1.22426975e+00
-9.04975176e-01 -2.00315714e-01 -6.08815908e-01 5.13864100e-01
2.90846735e-01 -4.32024866e-01 -8.99588645e-01 1.52283144e+00
7.34676361e-01 -2.46827267e-02 1.96744457e-01 1.24219596e+00
5.50458491e-01 5.01485944e-01 -5.07363454e-02 1.95241999e-02
1.02290893e+00 -7.82961130e-01 -1.03486502e+00 -6.92581594e-01
6.30258918e-01 -2.62449950e-01 7.91254997e-01 5.37462175e-01
-4.81868446e-01 -8.19603562e-01 -1.43127966e+00 -3.15466411e-02
-4.00722384e-01 8.33487585e-02 6.63952351e-01 2.34855548e-01
-9.74075973e-01 6.53192163e-01 -1.31925392e+00 -8.64135444e-01
-2.58681685e-01 5.68232954e-01 -5.96413851e-01 3.52408849e-02
-1.24223578e+00 1.16837561e+00 5.43999016e-01 2.17751741e-01
-9.18658972e-01 -5.05381916e-03 -1.02466404e+00 -6.99484468e-01
3.04950029e-01 -9.72863615e-01 1.21318042e+00 -3.52173597e-01
-2.11099291e+00 1.85753778e-01 -4.75973524e-02 -6.12064600e-01
4.03434783e-01 -4.32043344e-01 5.44746704e-02 -6.28628433e-02
4.74185735e-01 6.56519532e-01 4.50143307e-01 -1.40284789e+00
-8.95713449e-01 -6.51083291e-01 1.99022740e-01 1.09367716e+00
4.77898419e-01 -1.28505838e+00 -5.89860678e-01 4.42559347e-02
9.86890197e-01 -1.18010664e+00 -7.00492084e-01 1.66197345e-01
-1.65271342e-01 -6.55861273e-02 3.91026258e-01 -4.32067454e-01
4.29924399e-01 -1.88411987e+00 6.26204610e-02 1.80198886e-02
-3.60241592e-01 -3.25922310e-01 5.47761440e-01 3.01844597e-01
6.35783851e-01 -6.14857316e-01 3.48259807e-02 -3.14564019e-01
-3.61020677e-02 7.06443608e-01 5.43376245e-02 6.20233834e-01
-4.66397047e-01 5.69128036e-01 -1.26689899e+00 -4.87912625e-01
6.53844416e-01 6.75868750e-01 -3.16861659e-01 1.96982250e-01
-7.41886795e-02 9.21989858e-01 -7.55402684e-01 5.13403654e-01
3.65618020e-01 1.61813393e-01 1.18972406e-01 -3.42951156e-02
-3.62726361e-01 2.99884498e-01 -1.60919130e+00 2.46400762e+00
-5.03198743e-01 2.73024291e-02 8.39070454e-02 -7.27544069e-01
1.00718772e+00 1.54836461e-01 3.79047930e-01 -9.09651875e-01
3.73368770e-01 3.15095276e-01 -4.15083677e-01 -5.95745623e-01
4.55834359e-01 2.80785143e-01 -1.01900049e-01 -2.72790343e-01
-7.55858747e-03 -6.58754051e-01 -2.75287598e-01 -2.96821922e-01
9.91731584e-01 1.08104610e+00 6.51459694e-01 1.72376648e-01
6.72431588e-01 5.27071536e-01 5.16193271e-01 8.00699532e-01
-4.13339823e-01 3.48192930e-01 -4.40230638e-01 -2.74311781e-01
-7.56203651e-01 -1.23025417e+00 1.31045803e-01 5.29952049e-01
9.87339258e-01 -3.25008556e-02 -5.80809593e-01 -1.00453660e-01
1.01656377e-01 4.85205024e-01 -2.34574616e-01 -1.57758430e-01
-6.95031464e-01 -5.69446802e-01 1.27633825e-01 3.80812764e-01
1.17767465e+00 -7.53520250e-01 -1.13839829e+00 3.05895120e-01
-4.24768716e-01 -1.15651345e+00 2.86654115e-01 2.64277220e-01
-7.35754788e-01 -1.06301951e+00 -4.25425172e-01 -9.33819294e-01
7.07062423e-01 4.48638439e-01 2.84774095e-01 -2.81951457e-01
1.56866133e-01 4.94938046e-01 -4.58771646e-01 -2.51884639e-01
2.17616573e-01 -2.70218104e-01 4.65778083e-01 -5.15949547e-01
1.92611411e-01 -6.68279469e-01 -6.14377856e-01 4.29143012e-01
2.49681026e-02 -1.05686474e-03 4.60667431e-01 5.89603543e-01
9.05367494e-01 2.40578413e-01 3.17595676e-02 9.91994888e-02
3.22490782e-01 -1.71554103e-01 -6.95837259e-01 -3.39455158e-01
-4.06688780e-01 -7.96502158e-02 3.72341365e-01 -1.81872249e-01
-1.04946125e+00 5.14948845e-01 -3.33259404e-01 3.05635899e-01
-3.08223277e-01 2.67022192e-01 -2.65857786e-01 -3.81173015e-01
8.26264799e-01 3.94097686e-01 8.63590464e-02 -2.23810926e-01
3.67973417e-01 6.13889575e-01 7.41016567e-01 -2.04272315e-01
5.78384936e-01 7.95988798e-01 3.17765445e-01 -7.76076257e-01
-6.42273650e-02 -5.09081542e-01 -7.79071748e-01 -1.52598873e-01
8.76830935e-01 -1.10553193e+00 -1.31708884e+00 5.24676800e-01
-9.94851768e-01 -3.29858601e-01 -5.80962710e-02 8.57778192e-01
-1.00369430e+00 5.61733127e-01 -1.78524777e-01 -1.12595606e+00
-1.43479154e-01 -1.03254735e+00 9.48179245e-01 5.23009300e-01
-7.09316880e-02 -6.32180631e-01 2.26035163e-01 -2.07580864e-01
-3.87494676e-02 4.32690769e-01 -1.25420645e-01 -9.41703171e-02
-6.90216720e-01 -4.47059810e-01 2.55692691e-01 -3.31437677e-01
2.61855572e-01 -7.16378331e-01 -5.89039564e-01 -1.55222237e-01
2.63046235e-01 -6.37470633e-02 2.49171332e-01 4.22473699e-01
2.09935069e-01 1.32554188e-01 -7.89544880e-01 5.28059840e-01
1.61258030e+00 7.05216229e-01 5.24791658e-01 1.03686559e+00
7.04687655e-01 3.90240103e-01 1.25918865e+00 3.87463033e-01
1.10433376e+00 8.86484623e-01 9.87417400e-01 4.42784905e-01
2.33825475e-01 -4.39601988e-01 3.36760104e-01 5.29388607e-01
-3.63090485e-01 5.39018922e-02 -7.58565962e-01 4.04128134e-01
-2.29747772e+00 -6.30992174e-01 -1.89279646e-01 2.31422639e+00
5.72559297e-01 3.33558828e-01 -4.21669990e-01 2.43339062e-01
3.49412441e-01 -1.22422099e-01 -7.06165493e-01 9.19818431e-02
1.23382777e-01 -3.30472887e-01 8.91715169e-01 9.60989833e-01
-9.91861582e-01 1.07324123e+00 5.27156544e+00 1.78401053e-01
-1.00202727e+00 1.07944487e-02 -4.67922837e-01 3.12515289e-01
3.85886014e-01 5.20026572e-02 -9.13650692e-01 4.62908924e-01
6.65331900e-01 2.52473503e-01 4.30623651e-01 1.48724747e+00
2.27458671e-01 -8.22293639e-01 -7.15731919e-01 1.16992664e+00
1.04670919e-01 -6.67229652e-01 -6.50101066e-01 -2.73264319e-01
4.18296516e-01 1.54667377e-01 -3.26500118e-01 3.60117763e-01
6.15673780e-01 -3.32949072e-01 6.83670938e-01 7.36556113e-01
2.09035631e-02 -6.19377017e-01 8.85731757e-01 9.34773207e-01
-1.32057023e+00 -2.67361373e-01 -5.93751967e-01 -5.40029645e-01
5.72321892e-01 -4.15732414e-02 -1.08559132e+00 7.16285288e-01
8.43390286e-01 7.57463455e-01 -2.78794318e-01 1.21656120e+00
-7.98540056e-01 -3.43054324e-01 -8.20173860e-01 -4.85958695e-01
2.37884447e-01 -3.55698168e-01 6.12469494e-01 2.50568628e-01
5.59633195e-01 1.01575129e-01 4.54317898e-01 2.62342840e-01
7.97617853e-01 -2.36500293e-01 -6.96480334e-01 6.06229663e-01
4.32687104e-01 8.29392791e-01 -5.12541294e-01 -2.20800981e-01
1.96296737e-01 1.33487785e+00 1.00480668e-01 4.67942506e-01
-8.46573889e-01 -6.63261056e-01 5.68132997e-01 -1.31391466e-01
1.40547022e-01 -9.29397941e-01 -1.12423293e-01 -9.30317581e-01
1.16783731e-01 -2.16381088e-01 -6.08441536e-04 -1.25568736e+00
-4.59030151e-01 4.14366245e-01 -1.45600602e-01 -1.66756344e+00
-6.52641714e-01 -4.89192039e-01 -8.96751061e-02 7.14191258e-01
-1.49173760e+00 -9.00687516e-01 -9.06037271e-01 6.83081031e-01
4.62428898e-01 2.64655858e-01 9.18211281e-01 -7.24039823e-02
-1.64968465e-02 -1.00980863e-01 6.00036187e-03 -2.88370490e-01
6.18512750e-01 -1.39605844e+00 3.09734553e-01 5.34806132e-01
-3.24074447e-01 6.08973801e-01 1.08455038e+00 -1.01045632e+00
-1.65697455e+00 -5.93307316e-01 6.59367263e-01 -5.45819283e-01
1.81555092e-01 -5.52375726e-02 -2.96993315e-01 7.57015049e-01
-2.29132488e-01 -1.43587545e-01 -1.56213403e-01 -5.59027672e-01
5.59780955e-01 -2.41645128e-01 -1.32400668e+00 8.81605685e-01
1.32236278e+00 -1.63250379e-02 -8.10203731e-01 2.49009043e-01
8.47375631e-01 -1.09647167e+00 -5.11838675e-01 3.07296783e-01
6.50077164e-01 -8.04173529e-01 1.20680797e+00 2.23773330e-01
-5.70501626e-01 -8.84991050e-01 -3.71728808e-01 -1.22859669e+00
-2.05587491e-01 -3.55936885e-01 -1.97713766e-02 6.37685776e-01
-1.70050442e-01 -7.92871654e-01 1.01953590e+00 3.80492985e-01
-9.62667540e-02 -3.48989457e-01 -1.25761127e+00 -5.98359644e-01
-7.88915634e-01 -4.30498511e-01 5.75699806e-01 3.68650526e-01
2.04428807e-01 1.03413694e-01 -3.86000693e-01 7.91288197e-01
8.23877633e-01 -3.53884339e-01 1.30042183e+00 -1.06485748e+00
3.49209793e-02 2.86956638e-01 -9.38242197e-01 -1.72516835e+00
-2.38900349e-01 -2.27018014e-01 6.57684922e-01 -2.23735905e+00
-8.58582914e-01 -5.15825331e-01 1.58816189e-01 2.73888350e-01
1.31453767e-01 8.67679268e-02 -1.02007359e-01 2.78071404e-01
-6.98442638e-01 7.61873484e-01 1.28943849e+00 1.66739315e-01
-5.60005665e-01 1.87367588e-01 -1.54667916e-02 9.35751319e-01
6.68661714e-01 -2.64633179e-01 -7.29310691e-01 -4.49328899e-01
1.44007295e-01 1.76593557e-01 6.06972635e-01 -1.63987601e+00
7.80636013e-01 -1.18150733e-01 7.22254395e-01 -8.26720238e-01
9.50418651e-01 -1.13864648e+00 3.73502403e-01 1.01140690e+00
3.96159858e-01 5.80625571e-02 -1.81116119e-01 8.27299356e-01
1.07607879e-01 -1.28545493e-01 1.53064311e-01 -5.24582326e-01
-1.24963868e+00 8.78013670e-02 -4.40016806e-01 -6.15142703e-01
1.08840060e+00 -8.45710337e-01 -7.84673020e-02 -5.81474185e-01
-8.92359555e-01 4.63454932e-01 6.96636260e-01 3.72703850e-01
6.71240211e-01 -1.51230252e+00 2.78904915e-01 2.62419462e-01
-7.63797536e-02 5.24606824e-01 3.72224063e-01 1.05248523e+00
-8.72704804e-01 5.88725448e-01 -4.47630376e-01 -1.01721597e+00
-6.92326784e-01 3.38594764e-01 4.78657305e-01 1.99677885e-01
-6.58159018e-01 5.94339848e-01 -2.82427132e-01 -9.68090296e-01
3.46862465e-01 -4.60220605e-01 -5.26586413e-01 -5.30606151e-01
2.60165453e-01 4.27952945e-01 -1.31209940e-01 -7.85694599e-01
-5.71292341e-01 9.67718184e-01 6.07282162e-01 -4.91199791e-01
1.14013052e+00 -8.30153108e-01 3.84096116e-01 5.23760855e-01
6.83067203e-01 -3.16046417e-01 -1.54758465e+00 -2.80179549e-02
-1.83486745e-01 -4.48762506e-01 -2.82779545e-01 -6.02678239e-01
-6.94882646e-02 4.28321600e-01 1.08717692e+00 -2.15884596e-01
6.64592087e-01 -5.11128783e-01 5.28920949e-01 9.68273222e-01
1.41383982e+00 -1.28316700e+00 1.95915755e-02 5.52259386e-01
8.60244393e-01 -1.38511872e+00 2.15881184e-01 -3.26149285e-01
-6.09682739e-01 8.33927929e-01 8.52896333e-01 -4.91243750e-01
5.26682138e-01 -1.72488485e-02 1.92248464e-01 3.09053808e-01
-1.28205210e-01 -3.65566611e-01 -2.99054354e-01 1.28447044e+00
-2.75292844e-01 -1.15274593e-01 -3.06433111e-01 3.95296484e-01
-4.41092372e-01 6.47483766e-02 2.87664145e-01 1.64011312e+00
-1.07719970e+00 -7.61314571e-01 -5.57321012e-01 -4.42424059e-01
2.14381114e-01 4.70851779e-01 1.52764007e-01 1.07905984e+00
3.61029923e-01 9.74415123e-01 2.60278080e-02 -4.83664066e-01
5.98227143e-01 -2.96575189e-01 7.09472060e-01 -2.86542863e-01
-1.14592284e-01 3.42492200e-03 1.40300971e-02 -1.05529952e+00
-4.03244227e-01 -4.76458192e-01 -2.12849689e+00 1.17711581e-01
-1.97012603e-01 1.05650820e-01 1.19565809e+00 1.05655456e+00
1.83434635e-01 4.30654913e-01 2.02502951e-01 -1.34308565e+00
-3.97195131e-01 -8.57379198e-01 -2.69232690e-01 1.63572788e-01
4.67072219e-01 -1.02789235e+00 -2.10703120e-01 -1.84205294e-01]
|
[4.765561580657959, 0.58937007188797]
|
30f8b007-4742-470f-94f5-a6fd5835ac71
|
an-automatic-pipeline-for-atlas-based-fetal
|
2205.07575
| null |
https://arxiv.org/abs/2205.07575v1
|
https://arxiv.org/pdf/2205.07575v1.pdf
|
An automatic pipeline for atlas-based fetal and neonatal brain segmentation and analysis
|
The automatic segmentation of perinatal brain structures in magnetic resonance imaging (MRI) is of utmost importance for the study of brain growth and related complications. While different methods exist for adult and pediatric MRI data, there is a lack for automatic tools for the analysis of perinatal imaging. In this work, a new pipeline for fetal and neonatal segmentation has been developed. We also report the creation of two new fetal atlases, and their use within the pipeline for atlas-based segmentation, based on novel registration methods. The pipeline is also able to extract cortical and pial surfaces and compute features, such as curvature, thickness, sulcal depth, and local gyrification index. Results show that the introduction of the new templates together with our segmentation strategy leads to accurate results when compared to expert annotations, as well as better performances when compared to a reference pipeline (developing Human Connectome Project (dHCP)), for both early and late-onset fetal brains.
|
['Miguel A', 'González Ballester', 'Gemma', 'Piella', 'Fàtima', 'Crispi', 'Eduard', 'Gratacós', 'Elisenda', 'Eixarch', 'Nadine', 'Hahner', 'Valentin', 'Comte', 'Laura', 'Segales', 'Francesca', 'Crovetto', 'Oualid', 'Benkarim', 'Ayako', 'Nakaki', 'Andrea', 'Urru']
|
2022-05-16
| null | null | null | null |
['brain-segmentation']
|
['medical']
|
[-1.31958231e-01 4.11909491e-01 4.54190522e-01 -5.49399674e-01
-1.89995915e-01 -5.29982567e-01 4.61276472e-01 7.43338406e-01
-5.67243159e-01 3.88336331e-01 1.16138272e-01 -8.48646313e-02
-1.48995742e-01 -6.91041768e-01 -4.14668381e-01 -2.56867617e-01
-3.50930303e-01 1.07029641e+00 6.17587209e-01 4.43132520e-02
2.81417519e-01 8.95393550e-01 -1.15399933e+00 1.00715019e-01
9.10318017e-01 5.88261962e-01 3.45152944e-01 4.61245865e-01
-2.29820058e-01 -2.79940460e-02 -1.17828019e-01 -5.19475698e-01
1.01235777e-01 -4.22650248e-01 -1.00510097e+00 -3.02376002e-01
2.97722071e-01 -2.55649596e-01 2.97399074e-01 8.73787582e-01
7.32942998e-01 -2.35095322e-01 6.99509919e-01 -4.02471185e-01
7.46634752e-02 7.00117648e-01 -4.08427685e-01 5.07541895e-01
3.50012660e-01 -2.62563825e-01 3.83193269e-02 -7.91032076e-01
1.04533422e+00 6.95110142e-01 8.04236948e-01 5.74836671e-01
-1.02808988e+00 -5.28947234e-01 -1.88377559e-01 1.71757758e-01
-1.10934627e+00 -4.51464504e-01 8.90348777e-02 -1.06326914e+00
9.19221401e-01 -1.31354481e-02 1.30042112e+00 5.40355444e-01
1.36707872e-01 1.18363127e-01 1.30239582e+00 -3.75202209e-01
2.83964664e-01 -1.49640173e-01 5.89557849e-02 6.89787924e-01
3.98513913e-01 -2.13493273e-01 -2.53560208e-02 2.32278228e-01
9.46350336e-01 -2.92632759e-01 -8.61963034e-02 -4.53332484e-01
-1.19885027e+00 5.17931104e-01 -4.18354347e-02 1.09981883e+00
-5.45651674e-01 -2.37364799e-01 5.97164333e-01 -1.53654248e-01
6.52514935e-01 2.14181796e-01 -3.58704180e-01 -2.08283395e-01
-1.42438698e+00 1.37598380e-01 5.28680921e-01 5.29454648e-01
4.64658141e-01 -2.54523218e-01 -4.74290177e-03 9.05117750e-01
4.54790115e-01 1.35931775e-01 5.35405278e-01 -8.97671998e-01
3.52602065e-01 5.90135753e-01 -4.97199863e-01 -7.94268847e-01
-1.13468218e+00 -1.12659536e-01 -6.12033308e-01 3.01511467e-01
5.10970056e-01 -3.30757052e-01 -8.14648926e-01 1.41201448e+00
4.87793356e-01 -2.07922794e-02 -4.04550880e-01 7.87838519e-01
9.62961435e-01 -2.25667030e-01 3.97545882e-02 -3.68241161e-01
1.11144328e+00 -5.39194047e-01 -5.92106104e-01 1.46282136e-01
4.70549405e-01 -5.50642312e-01 3.78802329e-01 3.87100339e-01
-1.55742645e+00 -1.41029758e-02 -7.36590862e-01 1.09519333e-01
-3.86000454e-01 -9.77221355e-02 4.20495868e-01 9.98007238e-01
-1.46881616e+00 7.96925008e-01 -1.51834702e+00 -6.80363655e-01
5.88777125e-01 4.98536944e-01 -8.60544086e-01 3.15506250e-01
-6.44243956e-01 1.42369246e+00 2.64322013e-01 -1.81506529e-01
-5.05481601e-01 -8.25970709e-01 -7.02751815e-01 -8.84499475e-02
-2.35302597e-01 -4.63771373e-01 8.70874047e-01 -6.14172399e-01
-1.52879775e+00 1.40010142e+00 3.69525179e-02 -4.19685453e-01
8.50074887e-01 1.55490533e-01 -4.89218086e-02 6.97604060e-01
1.96186528e-01 7.02619135e-01 2.86567718e-01 -8.22688162e-01
-2.97776073e-01 -6.97880745e-01 -4.01633620e-01 -8.62018988e-02
5.80853857e-02 7.16895401e-01 -4.47219610e-01 -4.06428963e-01
4.58201230e-01 -5.32682180e-01 -3.68992299e-01 -2.22001806e-01
1.11944146e-01 1.45404324e-01 1.54165536e-01 -1.60034752e+00
8.89822423e-01 -1.59913361e+00 6.81236237e-02 5.42226315e-01
4.89588380e-01 3.85243386e-01 2.27138862e-01 1.02805234e-01
-4.10995126e-01 1.70621857e-01 -5.10940254e-01 -1.91641927e-01
-3.05261135e-01 -1.19554147e-01 6.49179697e-01 7.24477410e-01
-2.52336860e-01 7.40535736e-01 -8.82968903e-01 -4.86309350e-01
3.10996532e-01 4.10959065e-01 -3.99402916e-01 7.78677166e-02
4.23006773e-01 8.74931037e-01 -1.44611970e-01 1.59062013e-01
7.06128180e-01 3.30879420e-01 1.93716332e-01 2.63408363e-01
-5.42886853e-01 7.65709952e-02 -7.26676881e-01 1.99961364e+00
-1.77408412e-01 2.43663922e-01 2.87323028e-01 -9.45538163e-01
8.99740875e-01 8.18164825e-01 8.15760612e-01 -5.84696889e-01
5.47425687e-01 4.73763674e-01 3.55465561e-01 -6.42481029e-01
-3.82210702e-01 -1.72428951e-01 9.52134311e-01 3.67235124e-01
4.95310754e-01 -2.04673126e-01 7.13166714e-01 -1.81888357e-01
9.86799240e-01 3.47645789e-01 1.85161278e-01 -7.34872520e-01
6.79335773e-01 -3.95593137e-01 3.89701098e-01 -6.85303509e-02
-4.93254848e-02 9.77511585e-01 6.24550462e-01 -4.56503272e-01
-1.16519570e+00 -8.74224901e-01 -4.20552313e-01 3.51656795e-01
-4.37638938e-01 -1.57058850e-01 -1.51537371e+00 -6.51405036e-01
-3.64310771e-01 4.90355283e-01 -6.01066530e-01 4.86257493e-01
-6.31718874e-01 -8.86078894e-01 5.15212953e-01 2.25690633e-01
5.86239174e-02 -1.03581214e+00 -8.75031710e-01 2.83208966e-01
2.04543948e-01 -1.05477178e+00 -1.40874460e-01 -3.19682211e-01
-1.04053473e+00 -1.35125017e+00 -1.30521607e+00 -7.23871171e-01
7.85302222e-01 -7.03143835e-01 7.76445150e-01 2.14464724e-01
-5.05957544e-01 3.08634400e-01 -3.54958892e-01 -3.88148934e-01
-6.23154402e-01 1.18998639e-01 2.77632810e-02 -3.27644080e-01
-1.06612742e-01 -1.04314780e+00 -9.39516425e-01 5.42953350e-02
-7.20053792e-01 1.56940594e-01 3.52932006e-01 -5.42474650e-02
2.94788748e-01 -9.28229511e-01 5.72155833e-01 -9.08513963e-01
4.77503031e-01 -4.43544209e-01 -7.31470048e-01 2.43975297e-01
-9.42594349e-01 -4.69608873e-01 2.03192979e-01 8.93284827e-02
-7.96264589e-01 -1.74032643e-01 -9.30350125e-01 -3.65681611e-02
-6.18961453e-01 4.17381763e-01 1.84365153e-01 -3.96765828e-01
5.16601562e-01 -1.54675812e-01 2.06582114e-01 -7.70658493e-01
1.26200780e-01 1.55652285e-01 4.52749729e-01 -4.38862264e-01
1.79858103e-01 2.50926793e-01 2.10297108e-01 -7.07763493e-01
-5.77487517e-03 -4.80954319e-01 -1.35503769e+00 -3.80781233e-01
1.32335877e+00 -2.05649257e-01 -1.94749877e-01 3.63299906e-01
-1.35978007e+00 -4.58073229e-01 -1.04811177e-01 7.70841241e-01
-4.91715610e-01 4.82428759e-01 -4.81440723e-01 -3.91941726e-01
-8.40765834e-01 -1.44709408e+00 3.72597247e-01 2.63835192e-02
-2.53495395e-01 -1.27104485e+00 5.77382863e-01 2.44213030e-01
5.63049853e-01 6.90731645e-01 9.68445003e-01 -8.62133086e-01
1.01373002e-01 -1.52576387e-01 -1.11358143e-01 3.76372039e-01
-8.79190266e-02 1.33246571e-01 -6.66061580e-01 -1.78624280e-02
-7.23032355e-02 4.70651150e-01 4.51958030e-01 4.57060099e-01
7.36954927e-01 1.60235584e-01 -2.29275271e-01 6.62231803e-01
1.28074670e+00 3.42708170e-01 6.28867805e-01 2.30355263e-01
3.60517979e-01 1.03951478e+00 1.17661960e-01 2.14413807e-01
4.67398047e-01 4.76004303e-01 3.10599953e-01 -1.33884564e-01
-3.89996648e-01 4.92563069e-01 -2.08351657e-01 9.58229423e-01
-9.74286795e-01 4.73636478e-01 -1.65333819e+00 8.78256917e-01
-1.31434131e+00 -5.62499285e-01 -7.63703048e-01 2.42375898e+00
6.81338191e-01 -4.65837494e-02 2.62913555e-01 -1.66089907e-01
7.46492565e-01 -4.87562180e-01 3.57641764e-02 -6.50728047e-01
2.67210484e-01 8.95158350e-01 4.13600117e-01 5.23097873e-01
-6.82648003e-01 6.90352142e-01 7.27200127e+00 2.55765289e-01
-1.32489526e+00 7.01798379e-01 5.67084730e-01 1.57304376e-01
6.93153441e-02 -2.13406652e-01 -3.04465234e-01 4.07483488e-01
1.03247690e+00 6.91734441e-03 4.15022939e-01 4.67708498e-01
3.44172210e-01 -2.56888658e-01 -8.50324392e-01 5.22567391e-01
5.94740547e-02 -1.35486448e+00 -4.83075440e-01 -9.96606648e-02
6.43556416e-01 5.04619777e-01 -6.07990682e-01 -2.51990080e-01
-3.80886048e-01 -1.00173986e+00 7.32844651e-01 8.11001420e-01
1.17535627e+00 -5.67111850e-01 7.45615244e-01 1.19895443e-01
-8.38649273e-01 4.73368138e-01 5.89102395e-02 1.49204224e-01
2.83669710e-01 5.30738533e-01 -1.07345402e+00 5.11983871e-01
6.99236035e-01 3.69882911e-01 -6.14586055e-01 1.53736436e+00
-2.33246282e-01 5.02355516e-01 -2.87722498e-01 3.77666056e-01
-1.51940838e-01 -5.72466254e-01 5.71757853e-01 1.53138793e+00
4.33619082e-01 2.55255606e-02 -5.64017832e-01 9.41143572e-01
3.04301679e-01 8.08662951e-01 -2.21724838e-01 1.60943925e-01
1.55111970e-02 1.67280138e+00 -1.45721817e+00 -1.52796626e-01
-4.39257473e-01 6.53907359e-01 3.71007234e-01 1.42080374e-02
-3.88380587e-01 -1.28258422e-01 1.21930748e-01 6.81790173e-01
-2.30834801e-02 -3.08829457e-01 -5.01368225e-01 -7.74410367e-01
-6.16567023e-02 -4.40604538e-01 1.53196380e-01 -4.68149036e-01
-5.84502757e-01 5.66320837e-01 3.19845259e-01 -3.70783955e-01
-2.77019262e-01 -4.69002604e-01 -1.01565528e+00 9.84066010e-01
-1.05156803e+00 -1.00124991e+00 -1.49785399e-01 2.02403665e-01
1.29373580e-01 1.42256524e-02 9.51513410e-01 6.23818278e-01
-4.38638926e-01 2.45004043e-01 -2.45251954e-01 1.53250173e-01
2.84840763e-01 -1.45889997e+00 4.20707822e-01 7.45036364e-01
-3.40027988e-01 7.46895015e-01 5.10528147e-01 -7.75036097e-01
-5.62140942e-01 -8.27516139e-01 8.49458635e-01 -1.79913342e-01
6.47019148e-01 -8.74244049e-02 -8.64101231e-01 7.35829592e-01
3.10991049e-01 -1.11599371e-01 7.05118716e-01 -1.74642548e-01
2.29164809e-01 1.23271771e-01 -1.42638934e+00 1.99314654e-01
8.94364536e-01 2.88043380e-01 -6.91757441e-01 2.97559798e-01
6.10168502e-02 -5.31292081e-01 -1.38287508e+00 6.63450301e-01
6.92502975e-01 -1.29159558e+00 7.34712601e-01 -6.16662391e-02
3.09226125e-01 6.08099103e-02 7.48448253e-01 -1.10225439e+00
-2.93829688e-03 -4.11176503e-01 1.84668511e-01 1.24203122e+00
4.10138994e-01 -7.84077287e-01 5.63725114e-01 7.61257172e-01
-4.07236636e-01 -7.92585015e-01 -1.17140353e+00 -2.46209353e-01
3.54455918e-01 -2.65125483e-01 4.04118717e-01 8.52586031e-01
1.43774003e-01 -2.11621910e-01 5.93508899e-01 -9.19986963e-02
2.83060282e-01 -3.91612262e-01 2.73007482e-01 -1.45848656e+00
2.73844719e-01 -9.74832594e-01 -7.68277466e-01 -8.06616340e-03
-1.50764305e-02 -1.32239294e+00 -2.93081135e-01 -2.08343124e+00
8.82386789e-02 -4.60734308e-01 -9.52386931e-02 4.27020818e-01
1.72593206e-01 3.64663035e-01 -7.44598284e-02 -2.52230525e-01
1.94322392e-02 -1.50025487e-01 1.09660757e+00 6.42188907e-01
-1.85913984e-02 -3.71882180e-03 -1.30218700e-01 1.09644210e+00
8.43162477e-01 -4.74650919e-01 5.35411108e-03 -3.72525454e-01
2.30492204e-01 1.37716066e-04 1.01331964e-01 -1.14307737e+00
-9.60665569e-02 2.98642069e-01 3.78159493e-01 -4.32867050e-01
-1.88463047e-01 -5.45213282e-01 2.26171806e-01 6.20983660e-01
9.54126492e-02 2.41156667e-01 1.16367251e-01 -3.85717422e-01
1.22438684e-01 -6.92534864e-01 1.09933054e+00 -2.82812774e-01
2.42281333e-02 4.50135887e-01 -6.72377110e-01 -1.21538021e-01
1.04544663e+00 -3.34263802e-01 1.97045833e-01 3.88926044e-02
-1.23285770e+00 1.37659222e-01 7.04890013e-01 1.02025114e-01
3.81415069e-01 -6.94896579e-01 -9.70799446e-01 7.05973133e-02
-3.47481519e-01 1.03211544e-01 2.79123396e-01 1.71672785e+00
-1.48104584e+00 2.99469233e-01 -6.83079898e-01 -4.18050885e-01
-1.46162963e+00 4.16517615e-01 5.97292364e-01 -2.37616017e-01
-6.96483672e-01 6.10328078e-01 -1.09417818e-01 -4.99151409e-01
-4.02677804e-02 -6.49655461e-01 -9.08552408e-01 2.31038243e-01
3.97550613e-01 7.99799323e-01 4.65445876e-01 -9.80754256e-01
-3.41310382e-01 8.23174119e-01 4.07317251e-01 -1.80430204e-01
1.59554708e+00 -1.83328286e-01 -9.01055872e-01 2.07544863e-01
7.32866704e-01 3.18745404e-01 -6.12023056e-01 4.48862851e-01
4.34233010e-01 -4.59706858e-02 -1.08341947e-01 -8.09335113e-01
-1.31875455e+00 1.19411266e+00 9.61472094e-01 2.63927191e-01
8.92666161e-01 -7.05182329e-02 5.12152731e-01 -4.81164753e-01
4.41100240e-01 -8.39450359e-01 -7.25924492e-01 5.38218856e-01
1.08098912e+00 -6.04143262e-01 -1.41674608e-01 -6.48195565e-01
-1.15658648e-01 1.49566805e+00 4.12844747e-01 -9.27218273e-02
6.38640046e-01 4.05777603e-01 1.59902856e-01 -3.61384094e-01
-1.22505628e-01 -2.64970064e-01 5.82923710e-01 7.81195998e-01
8.83414149e-01 -5.40956929e-02 -1.31646299e+00 7.24580944e-01
-3.52185309e-01 -1.82477348e-02 3.08115512e-01 5.64534366e-01
-3.90305430e-01 -1.27389884e+00 -2.04276040e-01 5.97204626e-01
-1.00012219e+00 -1.16512470e-01 -2.60899395e-01 6.69041872e-01
7.09242880e-01 3.26847762e-01 -1.06009714e-01 1.93784431e-01
3.34089339e-01 2.67469823e-01 8.75170410e-01 -1.05617523e+00
-9.93445754e-01 2.24854555e-02 1.85092062e-01 -6.10757053e-01
-3.76125753e-01 -1.07038212e+00 -1.71633768e+00 3.02444220e-01
-1.17521614e-01 1.79587416e-02 1.33946240e+00 1.15221620e+00
4.48459312e-02 5.83308101e-01 1.27103776e-01 -1.03137147e+00
5.88865161e-01 -1.21831822e+00 -5.87013543e-01 2.01456636e-01
-1.27472684e-01 -6.50166512e-01 9.63265598e-02 1.45809337e-01]
|
[14.126568794250488, -2.431640386581421]
|
5c3d9b6a-f961-47d8-9325-da2c7dde9a68
|
automated-essay-scoring-system-for-nonnative
| null | null |
https://aclanthology.org/2020.lrec-1.157
|
https://aclanthology.org/2020.lrec-1.157.pdf
|
Automated Essay Scoring System for Nonnative Japanese Learners
|
In this study, we created an automated essay scoring (AES) system for nonnative Japanese learners using an essay dataset with annotations for a holistic score and multiple trait scores, including content, organization, and language scores. In particular, we developed AES systems using two different approaches: a feature-based approach and a neural-network-based approach. In the former approach, we used Japanese-specific linguistic features, including character-type features such as {``}kanji{''} and {``}hiragana.{''} In the latter approach, we used two models: a long short-term memory (LSTM) model (Hochreiter and Schmidhuber, 1997) and a bidirectional encoder representations from transformers (BERT) model (Devlin et al., 2019), which achieved the highest accuracy in various natural language processing tasks in 2018. Overall, the BERT model achieved the best root mean squared error and quadratic weighted kappa scores. In addition, we analyzed the robustness of the outputs of the BERT model. We have released and shared this system to facilitate further research on AES for Japanese as a second language learners.
|
['Satoru Katsumata', 'Hiroki Shimanaka', 'Mio Arai', 'Mamoru Komachi', 'Reo Hirao']
|
2020-05-01
| null | null | null |
lrec-2020-5
|
['automated-essay-scoring']
|
['natural-language-processing']
|
[-2.29520142e-01 -1.69265922e-02 8.81731957e-02 -5.15094101e-01
-8.68599057e-01 -6.46704197e-01 3.32907945e-01 4.09185410e-01
-8.09872270e-01 7.34209299e-01 3.06604028e-01 -4.43928063e-01
-1.54348120e-01 -6.80272996e-01 -3.71439815e-01 -3.46189797e-01
4.56682831e-01 1.16377376e-01 1.24751404e-01 -3.48731965e-01
7.00715661e-01 3.08556762e-02 -1.34070742e+00 3.26196760e-01
1.41421723e+00 1.05070949e+00 3.12167853e-01 6.69974029e-01
-1.80966958e-01 1.31884015e+00 -8.18975925e-01 -1.02157927e+00
-1.38564318e-01 -3.37874115e-01 -9.58170056e-01 -5.89393020e-01
5.91968715e-01 -4.71930563e-01 -3.23372394e-01 1.04698598e+00
2.59997487e-01 2.93027520e-01 8.31999123e-01 -6.72354460e-01
-1.39963293e+00 9.61093187e-01 -1.49965465e-01 1.01833910e-01
3.65284860e-01 1.17593639e-01 1.42425501e+00 -1.11487877e+00
3.07795972e-01 5.44787586e-01 7.81508386e-01 4.51068789e-01
-6.99651420e-01 -7.73025751e-01 1.21105999e-01 6.01373851e-01
-8.21837246e-01 -2.31547579e-01 4.90831077e-01 -6.29973233e-01
8.91265750e-01 1.59806814e-02 8.02585483e-01 9.92079198e-01
2.60165304e-01 1.08739519e+00 1.54364550e+00 -5.62745929e-01
-8.87821987e-02 2.02043042e-01 6.70688927e-01 9.69626009e-01
5.31041548e-02 -7.37146735e-02 -7.43640125e-01 1.81661054e-01
3.57772917e-01 -2.30134532e-01 -1.24890596e-01 4.65383559e-01
-1.37974918e+00 8.18702877e-01 3.56196135e-01 3.80548328e-01
-2.24464625e-01 -1.51901171e-01 4.20399547e-01 6.48300886e-01
3.66404593e-01 7.33970881e-01 -7.17830122e-01 -4.51281756e-01
-9.56511557e-01 1.07840173e-01 8.66620302e-01 8.60920310e-01
3.93600136e-01 1.44853756e-01 -5.62708497e-01 1.19212472e+00
3.34574521e-01 2.68224716e-01 1.00476742e+00 -7.89154708e-01
6.06795669e-01 4.40620840e-01 -1.44862786e-01 -8.56260121e-01
-4.65299845e-01 -5.21988153e-01 -5.18607438e-01 -1.34480104e-01
6.19356453e-01 -2.74852574e-01 -6.10169351e-01 1.92669582e+00
-3.35319042e-01 -3.84736270e-01 7.79294893e-02 9.03524280e-01
1.31150424e+00 6.45006239e-01 4.99219783e-02 -7.21937194e-02
1.47905183e+00 -1.45887983e+00 -7.77596474e-01 5.72431311e-02
9.75471675e-01 -7.65066266e-01 1.51081777e+00 6.03662312e-01
-1.31616557e+00 -6.52069986e-01 -1.19461823e+00 -3.04258376e-01
-4.29032654e-01 6.72872961e-01 3.53460431e-01 8.23711693e-01
-1.22453487e+00 4.86820936e-01 -4.33066368e-01 -2.27019802e-01
1.81428567e-02 1.15434512e-01 -3.60835403e-01 3.00769478e-01
-1.48096216e+00 1.15249646e+00 5.55419214e-02 2.24581838e-01
-4.92720664e-01 -4.78981167e-01 -7.50118971e-01 2.60074317e-01
-4.46029231e-02 -3.63263726e-01 1.59312057e+00 -9.54727173e-01
-2.12172413e+00 9.14075255e-01 2.05393247e-02 -7.74173811e-02
5.52074909e-02 -3.81217062e-01 -2.73368984e-01 -6.37290180e-02
1.25389341e-02 2.17823967e-01 2.33665213e-01 -5.18517435e-01
-5.53110838e-01 -3.91544819e-01 2.71294196e-03 3.40570778e-01
-1.01456332e+00 3.48383158e-01 1.43103004e-01 -7.96833038e-01
-8.90981779e-02 -8.03163588e-01 3.64854574e-01 -2.19751641e-01
-1.33132219e-01 -7.75468528e-01 1.51162613e-02 -1.34714639e+00
1.57355344e+00 -1.90011346e+00 1.07736424e-01 6.87170103e-02
2.75891900e-01 3.41324627e-01 -2.10865811e-01 3.44775945e-01
3.63802910e-01 2.27664292e-01 -9.41478163e-02 -1.86444759e-01
2.38186732e-01 -3.90571207e-01 1.76467746e-02 1.89665779e-02
-2.51874924e-02 9.80604470e-01 -1.01552606e+00 -3.37932050e-01
-3.90768588e-01 -4.60390486e-02 -2.96798050e-01 3.08835566e-01
5.32180518e-02 7.85878673e-02 -6.68164790e-02 6.84569657e-01
1.86705753e-01 -8.62138420e-02 -2.24547042e-03 2.74627060e-01
-4.96371448e-01 9.37693655e-01 -6.13660634e-01 1.54257905e+00
-5.47020376e-01 9.32837427e-01 -2.74168979e-02 -4.85274225e-01
1.29193497e+00 4.80585545e-01 3.81418467e-02 -7.22060680e-01
1.19525060e-01 5.73081076e-01 2.03359932e-01 -4.97358114e-01
7.33323216e-01 -2.52028517e-02 -2.85257518e-01 8.86436701e-01
2.32997924e-01 -3.12193096e-01 2.52464950e-01 3.12682018e-02
1.02972591e+00 1.60478324e-01 3.49523813e-01 -4.10737574e-01
6.02082849e-01 -4.23202105e-02 3.59032780e-01 6.19732320e-01
-5.91725230e-01 5.41700542e-01 5.16671538e-01 -2.55905300e-01
-7.76798189e-01 -9.29676056e-01 -8.30026437e-03 1.51355159e+00
-4.28482711e-01 -4.72798973e-01 -9.29085374e-01 -7.79034019e-01
-2.73267865e-01 1.00281334e+00 -5.07076204e-01 -3.19812745e-01
-4.94305760e-01 -3.88384402e-01 8.54453444e-01 5.88409781e-01
5.52401781e-01 -1.22496533e+00 -4.34428006e-01 1.96895361e-01
-4.55184430e-01 -6.74885273e-01 -7.96805382e-01 2.95200527e-01
-5.92410922e-01 -5.87753356e-01 -8.51915598e-01 -9.97683346e-01
2.53577292e-01 -6.57520965e-02 1.01399696e+00 9.80769619e-02
3.15011263e-01 1.98795661e-01 -4.91981626e-01 -6.24424815e-01
-1.99093059e-01 4.02699411e-01 1.35454908e-01 -3.06263417e-01
5.85645854e-01 3.13371867e-02 -1.87300161e-01 9.20699239e-02
-5.89726985e-01 7.58892391e-03 7.04246700e-01 1.03897095e+00
6.33056909e-02 -5.08653700e-01 6.99809790e-01 -5.88598192e-01
1.27458251e+00 -2.20456049e-01 -2.64866322e-01 5.43372035e-01
-7.58533001e-01 -9.98191983e-02 7.91524231e-01 -4.24745023e-01
-1.15172541e+00 -3.60217124e-01 -4.68558639e-01 2.00797588e-01
6.92547206e-03 9.91132677e-01 1.45312473e-01 -5.41195162e-02
4.92795676e-01 3.51705492e-01 -1.82171047e-01 -4.45059747e-01
-7.24588037e-02 1.04127288e+00 5.00101268e-01 -7.81175315e-01
4.13812906e-01 -5.53340197e-01 -5.83800435e-01 -5.61518490e-01
-1.25072992e+00 -6.44118665e-03 -8.95634890e-01 -4.74775046e-01
7.82572865e-01 -9.48500872e-01 -7.23838329e-01 9.13669229e-01
-1.13315237e+00 -5.29481113e-01 3.93795073e-02 8.18033755e-01
-3.16720277e-01 2.60757625e-01 -1.19690144e+00 -4.78947639e-01
-5.96419275e-01 -1.08023095e+00 4.47719395e-01 5.41997194e-01
-4.95407254e-01 -9.93043005e-01 3.10275525e-01 6.39005303e-01
5.28496325e-01 -2.55705953e-01 1.21910119e+00 -9.84517038e-01
8.29031914e-02 -1.11531079e-01 -1.83737963e-01 5.46596766e-01
-2.21850559e-01 3.06034982e-01 -9.31204259e-01 -1.55142248e-02
1.10393114e-01 -7.13988960e-01 8.05778682e-01 1.85825706e-01
1.13573563e+00 -4.52280939e-01 5.74016154e-01 3.64509284e-01
8.28900814e-01 7.03263581e-02 3.91274184e-01 6.66027725e-01
5.69038093e-01 7.52091289e-01 4.71898913e-01 2.14657038e-01
1.02089739e+00 5.93108833e-01 -2.38002554e-01 2.59151846e-01
-5.54899126e-03 -2.80116916e-01 9.22413528e-01 1.72379208e+00
-2.43552327e-01 -9.45181400e-02 -1.03486967e+00 4.39977080e-01
-1.68655121e+00 -8.34503531e-01 -5.54288626e-01 2.09390402e+00
1.21358073e+00 -4.77468129e-03 5.81636392e-02 -2.03527231e-02
4.98098046e-01 3.39225419e-02 -1.30684629e-01 -1.07136655e+00
-1.35901779e-01 2.27104053e-01 9.65387076e-02 3.07315797e-01
-8.76248360e-01 1.08499491e+00 5.88784981e+00 6.24101341e-01
-1.05377281e+00 2.06742167e-01 4.40478802e-01 9.60916653e-02
-3.42282653e-01 -3.32856268e-01 -7.13353693e-01 6.42563701e-01
1.27545607e+00 -1.03160068e-01 2.29512557e-01 6.50721014e-01
-1.72177657e-01 1.59696005e-02 -7.07171738e-01 4.42012757e-01
3.17071319e-01 -8.47894192e-01 -2.01645657e-01 -1.94843337e-01
8.10531437e-01 -5.21739684e-02 2.87466913e-01 8.42750013e-01
3.74069482e-01 -9.35689569e-01 1.24003637e+00 7.54023969e-01
7.79700041e-01 -5.72927535e-01 8.69248092e-01 4.57624793e-01
-8.26937675e-01 -1.98743403e-01 -3.25009078e-01 -5.35536468e-01
-1.48777500e-01 5.35926700e-01 -3.27918828e-01 2.25255579e-01
6.53977811e-01 5.86881101e-01 -8.69370580e-01 9.12014425e-01
-8.10478270e-01 1.10893238e+00 8.56105015e-02 -6.72241151e-01
1.60593346e-01 -2.17528507e-01 1.93193689e-01 1.29916835e+00
4.97942954e-01 5.89569286e-02 -1.88130423e-01 6.97259307e-01
-3.23301494e-01 4.29923058e-01 -2.49811232e-01 -1.41724765e-01
5.00447214e-01 1.42150819e+00 -4.14619029e-01 -2.44526789e-01
-6.69868052e-01 7.91184187e-01 8.25459003e-01 2.62226820e-01
-6.38998628e-01 -8.55633318e-01 2.37025470e-01 -4.26311433e-01
-4.17521447e-02 -4.13040698e-01 -9.18399036e-01 -1.30424321e+00
-4.88712713e-02 -8.53742361e-01 3.33810925e-01 -8.11431110e-01
-1.40967166e+00 6.24390185e-01 -3.95528793e-01 -8.90687168e-01
-1.94226205e-02 -9.57081378e-01 -6.66357875e-01 1.04802644e+00
-1.32526231e+00 -1.02479780e+00 -2.68833786e-01 2.90721059e-01
4.98973161e-01 -5.50229311e-01 9.56800461e-01 1.68750286e-01
-8.53176355e-01 9.86143053e-01 3.02585334e-01 4.08057392e-01
1.10766399e+00 -1.54592860e+00 1.40883014e-01 7.07376659e-01
1.59244061e-01 7.84100294e-01 6.55432791e-02 -5.09694576e-01
-9.48951662e-01 -7.21761823e-01 1.63274205e+00 -5.14730573e-01
8.70429814e-01 -2.29364514e-01 -9.62815166e-01 6.95486069e-01
5.11629820e-01 -6.27706409e-01 1.14369738e+00 2.97723502e-01
-4.15700346e-01 1.26040238e-03 -8.03399444e-01 5.60049176e-01
5.52605569e-01 -7.13612914e-01 -8.26620877e-01 -8.26273020e-03
5.05379796e-01 -2.95619905e-01 -1.20923924e+00 3.03641528e-01
1.02663112e+00 -1.00962007e+00 4.39125299e-01 -4.37269956e-01
1.02343559e+00 1.29732639e-01 1.74152911e-01 -1.68709075e+00
-7.91812062e-01 -1.16407916e-01 -8.54737684e-02 1.14158380e+00
7.11003423e-01 -4.91053939e-01 2.29352236e-01 6.64710045e-01
-7.14718461e-01 -1.02174711e+00 -7.62763619e-01 -5.98364472e-01
6.15870059e-01 -1.47765517e-01 4.80434597e-01 1.03259885e+00
4.15733099e-01 3.85338247e-01 -7.91324675e-02 -4.88323659e-01
1.05973676e-01 -1.42120004e-01 2.45606720e-01 -1.35024440e+00
4.00706306e-02 -1.08715868e+00 4.53912606e-03 -8.91233921e-01
4.64133084e-01 -1.09596241e+00 1.26621708e-01 -1.56514418e+00
3.78045976e-01 -1.67478725e-01 -6.76624715e-01 6.22511923e-01
-5.68814576e-01 2.40668967e-01 2.72118419e-01 1.73475817e-01
-4.72306758e-01 6.05949998e-01 1.14117253e+00 7.14486744e-03
-6.35513887e-02 1.54501230e-01 -9.77020860e-01 6.66854799e-01
1.00701332e+00 -3.32625121e-01 -2.83038151e-02 -8.13381851e-01
4.77162808e-01 -1.78133398e-01 -2.54676491e-02 -7.04527140e-01
4.62936163e-01 -3.27888936e-01 3.98230195e-01 -2.36751616e-01
8.19275677e-02 -1.42471150e-01 -5.56750596e-01 3.93296897e-01
-6.97873890e-01 4.54838574e-01 4.10382263e-02 -1.64628163e-01
-3.61657381e-01 -7.09028363e-01 6.02062345e-01 -2.19396412e-01
-4.39580679e-01 -1.25747845e-01 -7.34355688e-01 -1.04823731e-01
6.59299314e-01 -1.59711111e-02 -7.23911464e-01 -5.18245339e-01
-1.76071793e-01 2.71999031e-01 1.61002293e-01 4.96191978e-01
6.60651922e-01 -1.31953609e+00 -1.02792847e+00 4.06880528e-02
1.23458132e-01 -5.16224384e-01 1.00561827e-01 1.18668699e+00
-7.03724444e-01 6.16979182e-01 -4.28899139e-01 5.47209568e-02
-1.08561599e+00 -1.01992421e-01 1.09765448e-01 -4.78102952e-01
-1.36846602e-01 1.11405456e+00 -1.23950429e-01 -9.65391159e-01
2.01062843e-01 -2.02561438e-01 -7.83312798e-01 3.22148919e-01
5.34149945e-01 6.93617523e-01 2.41659746e-01 -6.22811675e-01
4.57980335e-02 2.67612278e-01 -1.95748508e-01 -3.79618764e-01
1.18486214e+00 2.99060941e-02 -3.87166589e-01 1.10453010e+00
8.63227367e-01 2.59878755e-01 -7.18755484e-01 -3.27560902e-01
3.26602846e-01 -2.09504113e-01 1.60736829e-01 -1.24486077e+00
-7.02204406e-01 1.05421734e+00 7.93560743e-02 2.28661507e-01
8.26087117e-01 -5.08190453e-01 9.06050801e-01 6.08952343e-01
1.64636284e-01 -1.54039061e+00 1.03255399e-01 1.30490255e+00
9.16742384e-01 -1.01209331e+00 -1.72646090e-01 1.31147102e-01
-9.09521759e-01 1.55943286e+00 1.08989942e+00 2.65407294e-01
4.03520107e-01 -1.08958572e-01 2.48107284e-01 7.69560412e-02
-9.20732260e-01 7.75589272e-02 7.56270885e-01 1.32802755e-01
1.09680152e+00 3.43103409e-01 -8.04657876e-01 1.40795183e+00
-7.69218981e-01 -1.65290639e-01 7.57369220e-01 6.69327736e-01
-5.27293324e-01 -8.28092575e-01 -3.15086663e-01 8.42134893e-01
-4.20650691e-01 -4.17396456e-01 -6.69211030e-01 4.15400565e-01
-1.30571932e-01 1.02518547e+00 -1.74559161e-01 -7.24702775e-01
2.62916297e-01 5.92796743e-01 4.75052595e-01 -7.81728864e-01
-1.28959358e+00 -4.76331890e-01 7.70997554e-02 3.11818700e-02
-1.11893535e-01 -8.56129348e-01 -1.07168770e+00 -2.73452371e-01
-5.32420993e-01 9.35434476e-02 6.92780614e-01 1.14411497e+00
-6.93094656e-02 5.09620428e-01 4.76098984e-01 -2.71816105e-01
-8.99845600e-01 -1.57195985e+00 -5.31814754e-01 -3.25709544e-02
2.92986240e-02 -3.71071577e-01 -3.07662070e-01 -9.43248346e-02]
|
[11.278922080993652, 9.389843940734863]
|
afd28a48-c95e-484e-bcaf-75eb79f453a8
|
cluster-labeling-by-word-embeddings-and
| null | null |
https://aclanthology.org/U18-1008
|
https://aclanthology.org/U18-1008.pdf
|
Cluster Labeling by Word Embeddings and WordNet's Hypernymy
|
Cluster labeling is the assignment of representative labels to clusters obtained from the organization of a document collection. Once assigned, the labels can play an important role in applications such as navigation, search and document classification. However, finding appropriately descriptive labels is still a challenging task. In this paper, we propose various approaches for assigning labels to word clusters by leveraging word embeddings and the synonymity and hypernymy relations in the WordNet lexical ontology. Experiments carried out using the WebAP document dataset have shown that one of the approaches stand out in the comparison and is capable of selecting labels that are reasonably aligned with those chosen by a pool of four human annotators.
|
['Massimo Piccardi', 'Hanieh Poostchi']
|
2018-12-01
|
cluster-labeling-by-word-embeddings-and-1
|
https://aclanthology.org/U18-1008
|
https://aclanthology.org/U18-1008.pdf
|
alta-2018-12
|
['learning-word-embeddings']
|
['methodology']
|
[-2.88616538e-01 -5.96894510e-02 -5.48084021e-01 -5.47000051e-01
-3.47821236e-01 -8.67119014e-01 8.37311089e-01 9.51853573e-01
-8.24653566e-01 5.11519194e-01 4.47038710e-01 -6.78352714e-02
-5.47102988e-01 -6.39135480e-01 2.23124668e-01 -5.31761706e-01
2.71845274e-02 9.87197995e-01 2.65082359e-01 -3.99971716e-02
4.88977879e-01 3.95145804e-01 -1.76267254e+00 2.45844215e-01
5.91092706e-01 9.84706223e-01 1.12465456e-01 1.42913952e-01
-8.07818294e-01 4.96007174e-01 -5.60422421e-01 -3.18077594e-01
7.62399584e-02 -1.36511385e-01 -1.19044876e+00 1.07300580e-01
1.80165544e-01 4.08170611e-01 1.88318580e-01 1.14843047e+00
1.68460667e-01 5.78750849e-01 8.89998496e-01 -1.22987652e+00
-4.49092209e-01 7.44092464e-01 -5.35321496e-02 2.88196392e-02
3.27829510e-01 -8.20648968e-01 1.64237165e+00 -8.51417124e-01
9.54721093e-01 1.13207579e+00 4.62410539e-01 4.22490925e-01
-1.17352021e+00 -4.83408988e-01 6.12640530e-02 3.11449319e-01
-1.71839595e+00 3.40897851e-02 5.42780101e-01 -6.27499700e-01
8.85184169e-01 2.28517607e-01 3.04940850e-01 6.31328344e-01
-4.88848805e-01 1.15286253e-01 8.53261471e-01 -9.28563118e-01
5.91654718e-01 5.53007185e-01 7.42596686e-01 3.10589552e-01
3.71827841e-01 -5.83778083e-01 -1.06766947e-01 -4.88018930e-01
1.57458588e-01 -1.19886674e-01 -2.15844631e-01 -7.40900457e-01
-1.11738253e+00 1.14681530e+00 4.48967576e-01 7.77063906e-01
-2.92118460e-01 -7.63558745e-02 5.75996697e-01 -1.09886512e-01
4.87637848e-01 1.14473391e+00 -2.67250240e-01 1.71637848e-01
-6.65141463e-01 -4.37464193e-02 7.68879831e-01 1.06145287e+00
8.72437596e-01 -6.44294024e-01 -3.90075855e-02 1.05285621e+00
3.93487781e-01 -2.43318081e-01 9.09076333e-01 -8.13241780e-01
6.75168559e-02 1.17227256e+00 4.20730799e-01 -1.14060926e+00
-6.82799935e-01 1.25756055e-01 -2.82879263e-01 -1.09188400e-01
1.27199680e-01 2.59193748e-01 -8.32381308e-01 1.42529571e+00
4.85354245e-01 -1.14495762e-01 9.24444422e-02 6.85093224e-01
8.89284492e-01 6.17441416e-01 4.47914273e-01 -1.00241698e-01
1.55183351e+00 -6.52098596e-01 -8.57408464e-01 -5.55534139e-02
8.47046018e-01 -8.61883700e-01 1.00780749e+00 1.57612115e-01
-2.18690604e-01 -4.66189563e-01 -9.17035460e-01 -4.51926403e-02
-9.27387238e-01 1.16137870e-01 5.52403510e-01 5.34715354e-01
-9.15058017e-01 3.00442278e-01 -3.54052097e-01 -9.62676883e-01
2.88195387e-02 4.29046541e-01 -5.95506847e-01 -9.96318385e-02
-1.25588214e+00 1.16108739e+00 1.09091449e+00 -4.72801208e-01
-2.37009123e-01 -6.93694651e-02 -7.94173837e-01 1.24373831e-01
2.08012775e-01 -3.36911753e-02 7.79727638e-01 -6.10923886e-01
-7.15870678e-01 1.10379469e+00 -7.54387155e-02 -1.31479174e-01
-3.05905402e-01 1.82826832e-01 -6.69066668e-01 1.34346604e-01
4.26780492e-01 7.53230155e-01 2.05277279e-01 -1.37452841e+00
-9.67696548e-01 -2.36973554e-01 1.83371261e-01 2.95559257e-01
-9.04687881e-01 3.17236215e-01 -5.58787167e-01 -4.15826499e-01
1.70842424e-01 -1.02962911e+00 -2.58246601e-01 -4.00769502e-01
-4.05162007e-01 -9.77960467e-01 5.42045653e-01 -4.13329244e-01
1.69898582e+00 -2.10181355e+00 1.27170637e-01 7.02277601e-01
3.13945502e-01 1.88487083e-01 1.92588806e-01 7.19543099e-01
-1.31409183e-01 5.08891642e-01 9.30137336e-02 -1.53426200e-01
2.99930423e-01 3.67419958e-01 -1.21870741e-01 2.81754673e-01
-3.34338158e-01 3.14246058e-01 -1.02406263e+00 -7.24086940e-01
9.60364640e-02 1.08250611e-01 -3.13939840e-01 1.40068874e-01
-1.64897889e-01 8.43205117e-03 -5.65130591e-01 3.34925324e-01
-3.21421982e-03 -2.39412040e-01 6.66632712e-01 -3.12707424e-01
-3.41604091e-02 3.65776688e-01 -1.35548210e+00 1.35581970e+00
-3.85279179e-01 6.30187452e-01 -5.31399310e-01 -1.06753016e+00
1.05123127e+00 4.50383157e-01 4.91877049e-01 -3.91612113e-01
2.57478356e-01 3.37418795e-01 3.17257717e-02 -5.44251919e-01
7.21938193e-01 -1.33996323e-01 -3.47026676e-01 7.44938612e-01
2.56082535e-01 1.46035045e-01 6.19028926e-01 1.70075208e-01
8.44158173e-01 -3.89700741e-01 7.00097203e-01 -5.31493783e-01
6.25998795e-01 4.40167099e-01 3.09319735e-01 3.76536667e-01
-8.56862888e-02 2.51792639e-01 2.70718873e-01 -5.80092490e-01
-1.15507817e+00 -6.89085543e-01 -3.91693860e-01 1.40076077e+00
5.73660396e-02 -1.01031291e+00 -5.64511955e-01 -7.62073696e-01
-8.70597810e-02 8.51760149e-01 -5.98287225e-01 -3.93479541e-02
-3.96362871e-01 -3.47708672e-01 3.71710271e-01 5.31278610e-01
-1.23342998e-01 -1.09445584e+00 -5.16938031e-01 1.39682710e-01
-3.06647450e-01 -1.01534510e+00 -3.13900560e-01 4.91805106e-01
-3.15825522e-01 -1.25188494e+00 -2.50907093e-01 -1.25166190e+00
8.07509720e-01 1.73849404e-01 1.14892268e+00 2.21488237e-01
-2.18865365e-01 3.43344718e-01 -8.84382188e-01 -2.11376503e-01
-2.81599104e-01 5.18828332e-01 3.40660483e-01 -1.70844376e-01
1.10165000e+00 -3.04601789e-02 -9.51663554e-02 4.45661098e-01
-1.14762247e+00 -5.10874450e-01 -1.94797918e-01 6.91540062e-01
5.57261288e-01 4.81900781e-01 4.86570090e-01 -1.20239365e+00
9.78128195e-01 -5.66371262e-01 -4.52875882e-01 4.45809126e-01
-9.29271042e-01 2.15629324e-01 4.93843377e-01 -3.86063755e-01
-6.31758034e-01 1.93622321e-01 8.83693174e-02 5.00643146e-05
-3.63599956e-01 7.33093023e-01 -1.91208079e-01 1.17258251e-01
7.13584185e-01 -3.88388097e-01 -3.92284542e-01 -4.54064667e-01
7.46459723e-01 1.07671428e+00 1.87147141e-01 -6.95154607e-01
5.40837348e-01 3.56557786e-01 -2.35072330e-01 -7.02226996e-01
-9.42018449e-01 -1.20911026e+00 -9.59150195e-01 -7.78537169e-02
1.10984051e+00 -5.36341906e-01 -3.05306315e-01 -5.61840057e-01
-1.01833439e+00 3.37022334e-01 -3.72377962e-01 6.33874357e-01
-1.43160671e-01 2.70412207e-01 -6.39160872e-02 -3.48553538e-01
-5.41183203e-02 -9.63043451e-01 6.76686823e-01 9.31267589e-02
-9.88083899e-01 -1.32280731e+00 2.59699762e-01 1.46526143e-01
1.68256223e-01 4.77143414e-02 1.60468292e+00 -1.53209054e+00
1.98773175e-01 -5.80251098e-01 -6.57781139e-02 3.32835704e-01
5.43153584e-01 -9.01203528e-02 -7.77810156e-01 -3.07601035e-01
-4.34115648e-01 -2.51150191e-01 5.78321159e-01 3.04910121e-03
8.91952038e-01 -2.49093190e-01 -6.86724544e-01 1.20796375e-01
1.52959442e+00 5.06716967e-01 2.94633687e-01 6.32161558e-01
7.21606612e-01 1.05376720e+00 6.42107606e-01 3.75500411e-01
1.19040474e-01 8.20714176e-01 2.61265308e-01 2.95952082e-01
9.33924690e-02 -1.27779931e-01 -2.92715847e-01 8.54169786e-01
2.29727611e-01 -3.92945021e-01 -1.34306109e+00 7.03906476e-01
-1.81783855e+00 -6.58723176e-01 -1.34738401e-01 2.21214128e+00
7.57552028e-01 2.15781368e-02 7.89150000e-02 2.11136028e-01
1.08863342e+00 -9.58972871e-02 -6.98099509e-02 -5.35479307e-01
1.62069216e-01 2.10873783e-01 4.35293913e-01 5.03214538e-01
-1.21036208e+00 1.08704162e+00 6.40907478e+00 7.59684682e-01
-6.97999597e-01 2.14247122e-01 2.17628881e-01 3.52010101e-01
-2.22494379e-01 2.18835935e-01 -9.94683087e-01 3.90607566e-01
8.23137999e-01 -4.03634220e-01 2.42580533e-01 8.68860722e-01
-2.57991910e-01 5.53140268e-02 -1.18990076e+00 7.79568791e-01
3.24970901e-01 -1.23742855e+00 2.39752382e-01 5.99586889e-02
7.13725984e-01 -2.52895653e-01 -3.26153159e-01 1.58604383e-01
6.27333760e-01 -1.10984731e+00 4.68428761e-01 1.12376049e-01
7.15091705e-01 -8.29801261e-01 9.31707978e-01 2.40269974e-01
-1.26057613e+00 -8.93717557e-02 -5.98939896e-01 9.02219638e-02
7.20946863e-03 3.82859081e-01 -1.04469371e+00 3.42498094e-01
6.62462711e-01 4.85689610e-01 -6.05139732e-01 1.11421883e+00
-4.54032958e-01 5.84025919e-01 -2.71259159e-01 -3.69508147e-01
5.22497654e-01 -2.20390141e-01 8.98715258e-02 1.34306121e+00
3.29043806e-01 -1.37336999e-01 5.94738841e-01 4.43700910e-01
-1.33978292e-01 5.82342923e-01 -4.73044723e-01 -1.97778031e-01
9.23326075e-01 1.35066640e+00 -1.21629739e+00 -3.44001889e-01
-2.10886508e-01 5.87811768e-01 4.54208970e-01 1.75572723e-01
-4.06948864e-01 -7.69226551e-01 6.93928659e-01 1.04587327e-03
1.99858442e-01 -1.50353476e-01 -5.11303656e-02 -6.85380995e-01
-1.47341058e-01 -4.06283200e-01 7.07030654e-01 -5.44178545e-01
-1.25387692e+00 9.41938519e-01 3.18255633e-01 -1.32393360e+00
-2.57571518e-01 -8.39200377e-01 -2.46785164e-01 6.03766561e-01
-1.09054017e+00 -7.47307360e-01 -3.44939083e-01 2.28060558e-01
1.74040109e-01 -2.76486397e-01 1.34603655e+00 3.69543850e-01
-3.77685905e-01 1.14743724e-01 4.08866435e-01 2.03807458e-01
8.19402039e-01 -1.42984807e+00 -1.04934466e-03 4.43315715e-01
6.92286670e-01 9.11539733e-01 6.89007163e-01 -4.14989740e-01
-3.33149552e-01 -9.75875378e-01 1.48900604e+00 -4.06913877e-01
7.64531493e-01 -2.90621638e-01 -1.00574315e+00 5.22405207e-01
4.40824747e-01 -5.03294989e-02 1.41515779e+00 2.69150317e-01
-4.89870518e-01 1.33705765e-01 -1.01245415e+00 3.48584592e-01
5.63972950e-01 -4.79973167e-01 -8.32161665e-01 6.81217074e-01
4.80317980e-01 1.56796843e-01 -9.21242535e-01 -1.76155761e-01
2.30750948e-01 -3.35963696e-01 8.94831717e-01 -9.84909654e-01
1.36832729e-01 -5.57871640e-01 -4.55973089e-01 -1.49121654e+00
-6.61529124e-01 -4.77003157e-02 2.89220780e-01 1.75559270e+00
3.61530632e-01 -3.91910136e-01 5.67968011e-01 5.38972378e-01
1.03232712e-01 -2.79199809e-01 -8.74599338e-01 -8.54008794e-01
-4.58265422e-03 -2.12329388e-01 6.64824665e-01 1.39214325e+00
5.39146364e-01 5.64654112e-01 4.37818887e-03 6.82860911e-02
3.67594719e-01 3.87950055e-02 3.50958228e-01 -1.91168141e+00
4.53304887e-01 -4.58187312e-01 -6.80060923e-01 -5.00082374e-01
6.85630500e-01 -1.39657402e+00 1.65983513e-01 -1.96550095e+00
-3.23013738e-02 -8.46227288e-01 -5.90799451e-01 5.08590579e-01
-3.31102102e-03 2.56293207e-01 3.79818231e-02 5.81432045e-01
-8.30145836e-01 2.92347390e-02 4.10435855e-01 -1.31175324e-01
3.54550891e-02 -4.65818852e-01 -6.74044728e-01 8.71157944e-01
9.25886810e-01 -8.88820767e-01 -4.05754447e-01 -4.04453546e-01
5.06210268e-01 -6.81737185e-01 -1.73672155e-01 -8.66627812e-01
5.00938833e-01 -2.08186522e-01 -1.95101835e-02 -1.42587021e-01
1.21569470e-01 -1.18033135e+00 2.41785683e-02 1.16303742e-01
-7.69132674e-01 2.20497638e-01 -1.09884493e-01 3.93442154e-01
-4.39107746e-01 -1.01352179e+00 5.61358392e-01 -1.38746500e-01
-1.26273251e+00 -2.64993072e-01 -5.87886512e-01 1.71195135e-01
1.15388775e+00 -1.81769580e-01 -9.06478465e-02 -1.15483649e-01
-7.20709145e-01 3.00188899e-01 5.47751129e-01 5.56529462e-01
2.77538776e-01 -1.63647020e+00 -3.64949197e-01 -2.33772650e-01
8.06230843e-01 -2.94081748e-01 -5.05459428e-01 1.03867851e-01
-6.03864372e-01 6.96874797e-01 -1.48844495e-01 -2.71833956e-01
-1.21689296e+00 7.25230217e-01 -1.70413867e-01 -2.63065070e-01
-3.85458171e-01 6.55213594e-01 -1.19481929e-01 -6.17346764e-01
3.26357782e-01 -1.72471896e-01 -1.05086315e+00 6.63979888e-01
4.23075885e-01 3.47595096e-01 1.45994589e-01 -1.08781397e+00
-4.81112808e-01 6.52780771e-01 -8.50469619e-02 7.79364258e-02
1.15934610e+00 -1.08821437e-01 -3.85001242e-01 6.65564835e-01
1.26707304e+00 -1.04077689e-01 -2.60599285e-01 -4.50812787e-01
9.14579391e-01 -3.22054237e-01 -9.90592688e-02 -4.91529316e-01
-6.42510593e-01 5.82491457e-01 6.19228184e-01 6.25089407e-01
7.40343690e-01 2.93774098e-01 4.17803705e-01 6.83060348e-01
3.67696524e-01 -1.46254826e+00 -6.67734221e-02 4.50326622e-01
3.79698455e-01 -1.14491642e+00 6.46296963e-02 -4.10654396e-01
-6.10485911e-01 1.11650717e+00 4.66858715e-01 2.16740090e-02
8.18083107e-01 -1.68319881e-01 2.97906876e-01 -4.50105786e-01
-3.61246556e-01 -4.86898899e-01 4.57483917e-01 6.15282416e-01
6.98347628e-01 1.87561721e-01 -8.28525484e-01 5.13949573e-01
-1.19811468e-01 -6.26025200e-01 3.54308635e-01 8.97957683e-01
-8.30213308e-01 -1.41981840e+00 -3.19680393e-01 5.57505012e-01
-2.65723795e-01 1.95660070e-02 -6.36295736e-01 7.26667464e-01
4.64992255e-01 1.06682777e+00 2.84828603e-01 -2.80719578e-01
1.83713034e-01 3.76341075e-01 -1.20522030e-01 -1.11326015e+00
-5.30156195e-01 -1.50979027e-01 1.68424398e-01 -6.10989593e-02
-7.30245709e-01 -3.36064488e-01 -1.49872673e+00 2.43029460e-01
-5.68052709e-01 8.85974526e-01 6.12761855e-01 9.69642699e-01
9.20445397e-02 2.34133452e-01 4.46154714e-01 -4.82594430e-01
-2.49861062e-01 -9.47418451e-01 -7.89493144e-01 1.04411125e+00
-3.63277584e-01 -9.16949272e-01 -2.42259696e-01 2.20484063e-01]
|
[10.178664207458496, 8.620189666748047]
|
2916c0d5-3b21-42a6-90b3-baa8910dfd44
|
lsoie-a-large-scale-dataset-for-supervised
|
2101.11177
| null |
https://arxiv.org/abs/2101.11177v1
|
https://arxiv.org/pdf/2101.11177v1.pdf
|
LSOIE: A Large-Scale Dataset for Supervised Open Information Extraction
|
Open Information Extraction (OIE) systems seek to compress the factual propositions of a sentence into a series of n-ary tuples. These tuples are useful for downstream tasks in natural language processing like knowledge base creation, textual entailment, and natural language understanding. However, current OIE datasets are limited in both size and diversity. We introduce a new dataset by converting the QA-SRL 2.0 dataset to a large-scale OIE dataset (LSOIE). Our LSOIE dataset is 20 times larger than the next largest human-annotated OIE dataset. We construct and evaluate several benchmark OIE models on LSOIE, providing baselines for future improvements on the task. Our LSOIE data, models, and code are made publicly available
|
['Stefan Larson', 'Jacob Solawetz']
|
2021-01-27
| null |
https://aclanthology.org/2021.eacl-main.222
|
https://aclanthology.org/2021.eacl-main.222.pdf
|
eacl-2021-2
|
['open-information-extraction']
|
['natural-language-processing']
|
[ 3.22782919e-02 6.30201101e-01 -6.92412555e-01 -5.76861262e-01
-1.16411221e+00 -7.12408006e-01 5.41547656e-01 5.68771780e-01
-4.06725347e-01 1.16491210e+00 7.98705935e-01 -6.39580429e-01
-3.10347583e-02 -1.07850146e+00 -9.70705867e-01 4.69057947e-01
-7.95675814e-02 8.12030196e-01 2.07595125e-01 -5.44248641e-01
1.39164805e-01 1.87070981e-01 -1.38708234e+00 1.02052283e+00
9.71584618e-01 9.14512873e-01 -3.01611543e-01 5.23364484e-01
-3.24336201e-01 1.11643147e+00 -4.19376552e-01 -9.86342728e-01
2.54907727e-01 -1.50906667e-01 -1.44186532e+00 -5.45216739e-01
4.71572876e-01 -2.41782367e-01 -4.37768340e-01 7.33967006e-01
1.56460330e-01 -1.13977149e-01 4.61051852e-01 -1.35117090e+00
-7.78484106e-01 1.19513798e+00 -1.98156219e-02 3.06094497e-01
8.38189781e-01 8.11480656e-02 1.68160081e+00 -1.23595333e+00
1.20770943e+00 1.34717047e+00 7.15751767e-01 4.10447866e-01
-7.99845994e-01 -3.37289512e-01 -2.89247066e-01 4.70768303e-01
-1.20918489e+00 -6.97563231e-01 9.67083126e-02 5.21618985e-02
1.76642430e+00 6.08059645e-01 3.18717360e-01 7.52238154e-01
3.34593087e-01 1.10393977e+00 8.49244893e-01 -3.64581257e-01
7.15490952e-02 -3.39245871e-02 5.11475027e-01 7.87340224e-01
5.69193661e-01 -3.20615441e-01 -7.92685628e-01 -3.25345188e-01
1.86290413e-01 -4.69949722e-01 1.23748317e-01 3.80104303e-01
-1.35519719e+00 7.11779296e-01 3.52365822e-01 -1.93935290e-01
-4.60239679e-01 -1.38564676e-01 5.80460966e-01 7.71188140e-01
2.59407640e-01 9.71007705e-01 -6.89435542e-01 -2.98952103e-01
-4.43723738e-01 6.71383679e-01 1.37716711e+00 1.17753792e+00
6.66979432e-01 -7.28976429e-01 -1.78058386e-01 6.76741838e-01
-3.54518630e-02 3.63506585e-01 2.38197759e-01 -1.13600302e+00
1.33407807e+00 8.35502863e-01 3.38319242e-01 -8.84487867e-01
-2.51715392e-01 1.12325050e-01 -3.19387883e-01 -6.60551250e-01
1.21313617e-01 -2.96654552e-01 -5.32759845e-01 1.38631368e+00
1.38869345e-01 -2.40805298e-01 6.55496180e-01 5.82095921e-01
1.38014579e+00 9.61219370e-01 4.17820849e-02 -1.17528997e-01
1.44060719e+00 -6.92776620e-01 -8.02697122e-01 -7.51672924e-01
9.60987449e-01 -6.07685149e-01 1.00904334e+00 3.86345387e-01
-1.18743920e+00 4.10392173e-02 -1.00377107e+00 -8.16081464e-01
-4.39437807e-01 -1.08217716e-01 9.95839298e-01 8.18152800e-02
-6.81621790e-01 2.37209544e-01 -4.48355645e-01 -3.32023054e-02
5.18729806e-01 9.49376225e-02 -6.17836654e-01 -4.46081460e-01
-1.86466289e+00 1.13152456e+00 9.47036922e-01 1.55387655e-01
-4.56979305e-01 -7.08078444e-01 -1.26409781e+00 1.85955659e-01
9.21060860e-01 -7.36485183e-01 1.42600107e+00 -2.63139009e-02
-8.45065773e-01 8.41092467e-01 -5.57001531e-01 -1.02950275e+00
7.75790438e-02 -4.27651167e-01 -6.58550680e-01 7.31540769e-02
4.51867372e-01 6.97571874e-01 -6.98911920e-02 -8.01080704e-01
-7.48968065e-01 -1.73200175e-01 3.84232879e-01 1.67099148e-01
-4.86878492e-02 3.04504305e-01 -3.50177437e-01 -1.82960078e-01
2.29839489e-01 -6.82945848e-01 -1.37844115e-01 -5.08533835e-01
-8.28741848e-01 -7.15643048e-01 3.90528202e-01 -6.51367843e-01
1.79903769e+00 -1.62303352e+00 -1.19529478e-01 -7.60189593e-02
4.40028578e-01 1.73434213e-01 -1.61271945e-01 7.14567304e-01
5.48133478e-02 2.34647468e-01 -3.14756691e-01 -1.51416942e-01
2.85411805e-01 5.07517040e-01 -5.93313158e-01 -2.61702240e-01
5.64408422e-01 1.49186301e+00 -9.24490571e-01 -8.64406705e-01
-4.71863389e-01 -6.95463359e-01 -6.85004830e-01 5.34079224e-02
-9.21647966e-01 -4.96487647e-01 -3.75900358e-01 7.87755311e-01
3.68201435e-01 -1.23969160e-01 7.84102604e-02 -2.51830757e-01
8.74184668e-02 1.33765411e+00 -9.29597497e-01 1.47338688e+00
-4.47123885e-01 6.96525455e-01 -4.45337951e-01 -5.41532278e-01
6.07371032e-01 2.75406808e-01 3.47981989e-01 -7.00276494e-01
-1.45714611e-01 3.17464501e-01 -1.06615372e-01 -7.56860971e-01
1.15591097e+00 -3.41257513e-01 -7.85222530e-01 6.54508531e-01
-1.03927720e-02 -3.00136894e-01 8.75926316e-01 8.25935304e-01
1.26392674e+00 -2.24212229e-01 6.25153601e-01 -4.47385982e-02
1.82524502e-01 5.78621805e-01 9.39364374e-01 6.37451291e-01
4.66896482e-02 5.23601808e-02 1.00506556e+00 -5.43595135e-01
-1.08246911e+00 -9.51161981e-01 -2.90890336e-01 7.55843639e-01
3.74487750e-02 -1.03898704e+00 -3.40818405e-01 -8.78220856e-01
2.30347499e-01 1.13208866e+00 -2.57192582e-01 -9.24171600e-03
-7.34901905e-01 -5.78492582e-01 1.13015354e+00 6.13552928e-01
3.06464285e-01 -1.22722685e+00 -3.64886582e-01 2.77658373e-01
-9.57618296e-01 -1.58282161e+00 -2.63293296e-01 8.84656236e-02
-6.18560612e-01 -1.13573432e+00 5.31102180e-01 -6.48262858e-01
2.31059328e-01 -1.43481478e-01 1.94682634e+00 -3.16131860e-01
-9.76531133e-02 -2.20233351e-01 -4.66781169e-01 -7.81710148e-01
-6.86903238e-01 4.90864515e-02 1.50047410e-02 -5.93027949e-01
1.11197746e+00 -4.20165360e-02 -8.28716606e-02 8.56170952e-02
-1.01761317e+00 1.98476613e-01 4.48040068e-01 7.76281953e-01
7.79820144e-01 3.25934850e-02 8.35356414e-01 -1.58515942e+00
8.33729446e-01 -9.00652051e-01 -4.03923541e-01 4.90357161e-01
-6.51555121e-01 3.78075033e-01 6.06770039e-01 3.17610532e-01
-1.35538054e+00 -3.10996294e-01 -4.33170199e-01 2.36398995e-01
-7.25605264e-02 1.36637604e+00 -2.89714366e-01 7.66052842e-01
7.51936316e-01 -2.83831060e-01 -4.18849081e-01 -1.82552561e-01
7.21756220e-01 9.32474136e-01 1.09220612e+00 -7.48089612e-01
4.01808798e-01 1.43992782e-01 -6.16161823e-02 -4.70541775e-01
-1.42687953e+00 -5.01376450e-01 -6.20814323e-01 4.80855167e-01
4.23794001e-01 -1.00236773e+00 -6.22272253e-01 -3.69495116e-02
-1.30738068e+00 -5.75794354e-02 -4.80913043e-01 1.62373364e-01
-3.75916123e-01 3.10279459e-01 -1.09847224e+00 -3.45956445e-01
-6.22701883e-01 -6.37836456e-01 8.74455571e-01 -1.50606006e-01
-5.91876686e-01 -7.56052554e-01 1.12471104e-01 5.51494062e-01
-2.64244914e-01 1.32525966e-01 1.14140415e+00 -7.75975287e-01
-5.68152130e-01 -2.21249476e-01 -2.76580721e-01 2.03028172e-01
-2.45739713e-01 -9.73867700e-02 -3.67287278e-01 2.45925859e-01
-1.67612255e-01 -1.05730784e+00 1.08468330e+00 -1.27781272e-01
9.33129489e-01 -8.66731763e-01 -1.23990282e-01 3.05052638e-01
1.47613895e+00 -1.40655488e-01 8.45217586e-01 5.05192220e-01
2.86581367e-01 6.64584219e-01 1.08675969e+00 4.47053999e-01
8.59569192e-01 1.11840680e-01 1.92614630e-01 3.47924113e-01
3.27443033e-02 -6.36964917e-01 3.60932797e-01 9.34036851e-01
2.48506382e-01 -4.66051906e-01 -1.10999334e+00 8.67465973e-01
-1.69267285e+00 -1.25100791e+00 -1.86699301e-01 1.64467585e+00
1.41469753e+00 4.56602007e-01 -2.65532285e-01 -6.62720948e-02
1.04697183e-01 3.65960971e-03 -3.94933403e-01 -7.81972706e-01
-4.66744602e-01 2.00554669e-01 4.39385235e-01 5.70304751e-01
-1.13866854e+00 1.18409121e+00 6.40787888e+00 5.90123475e-01
-4.43790078e-01 -5.75501397e-02 3.21083516e-01 -2.51971573e-01
-8.50290120e-01 2.09079087e-01 -1.19840837e+00 9.20346528e-02
1.34852624e+00 -4.56812680e-01 7.05787688e-02 6.29466772e-01
-2.24421218e-01 -3.60739738e-01 -1.53165913e+00 5.92842102e-01
-4.64645065e-02 -1.82107568e+00 6.93780258e-02 -1.88915521e-01
6.95437014e-01 3.70515972e-01 -3.60756397e-01 5.98074734e-01
5.36859035e-01 -1.20167458e+00 6.00596428e-01 4.05408144e-01
1.01824510e+00 -7.52082407e-01 9.95179772e-01 5.10780454e-01
-1.03227842e+00 -1.96986169e-01 -4.08379674e-01 -2.53047287e-01
3.56048644e-01 5.98415732e-01 -9.12016571e-01 7.18607545e-01
4.40974921e-01 8.26709747e-01 -7.47535110e-01 5.98539531e-01
-4.85105097e-01 6.10593259e-01 -5.46013534e-01 -1.54752821e-01
4.02466685e-01 -5.10075800e-02 6.25513434e-01 1.49061668e+00
-2.21341684e-01 5.69070995e-01 -6.14199713e-02 1.05657613e+00
-6.85002804e-01 -2.60193467e-01 -6.84461057e-01 -4.52797771e-01
8.87147546e-01 9.45173025e-01 -8.78071636e-02 -6.69054806e-01
-7.91544497e-01 7.15430498e-01 6.65949702e-01 1.75779555e-02
-6.03760362e-01 -7.69252002e-01 5.81578672e-01 -2.05722004e-01
1.21522523e-01 1.32827759e-02 -4.72929209e-01 -1.47897029e+00
4.35319006e-01 -1.19080853e+00 8.27825665e-01 -5.72635412e-01
-1.47036290e+00 5.66359699e-01 3.29726458e-01 -7.88076699e-01
-8.38476598e-01 -5.67532063e-01 -2.85753787e-01 6.79774046e-01
-1.36228979e+00 -7.35345721e-01 1.18489720e-01 3.31019551e-01
5.12831390e-01 2.64631864e-03 9.67761219e-01 3.31222892e-01
-5.91767669e-01 5.35918295e-01 -2.22188234e-01 3.98206621e-01
5.94355226e-01 -1.30523348e+00 1.07711053e+00 1.12940109e+00
3.13199550e-01 1.08767581e+00 6.62453890e-01 -9.12120044e-01
-1.69969320e+00 -1.03794312e+00 1.83445346e+00 -1.08984625e+00
8.32198143e-01 -2.85152107e-01 -1.00024915e+00 1.37749410e+00
-6.24386854e-02 3.21787479e-03 6.32987559e-01 6.55518770e-01
-5.69857359e-01 1.48052588e-01 -8.50371122e-01 5.68039596e-01
1.08532476e+00 -8.15325677e-01 -1.44128883e+00 3.07824343e-01
1.11611581e+00 -7.30612576e-01 -1.21383953e+00 4.34610933e-01
4.52801347e-01 -6.73668563e-01 8.88374686e-01 -1.25042856e+00
1.22527051e+00 -1.56191383e-02 -1.69054091e-01 -1.15164244e+00
1.56763300e-01 -4.85534757e-01 -5.24339318e-01 9.73976374e-01
1.17708421e+00 -3.78423035e-01 7.54884541e-01 1.05046928e+00
-3.16304207e-01 -1.28050542e+00 -8.57677102e-01 -6.53933942e-01
9.88396481e-02 -8.41834009e-01 8.66155326e-01 6.74080014e-01
6.79976046e-01 9.16854739e-01 -9.57087204e-02 6.06665164e-02
4.29292619e-01 6.84729993e-01 6.77329302e-01 -1.03059733e+00
-1.23957723e-01 -4.91305403e-02 -4.23902832e-03 -1.12666869e+00
3.65741938e-01 -1.14806664e+00 9.72499028e-02 -1.88216794e+00
3.82926345e-01 -6.91437781e-01 1.17564209e-01 7.60915399e-01
-4.03440148e-01 -9.71158743e-02 -3.27566154e-02 3.09312213e-02
-9.56393003e-01 4.60281968e-01 8.85523140e-01 -7.31850415e-02
-1.34783074e-01 -2.66518027e-01 -9.26629126e-01 6.32183850e-01
5.01622736e-01 -5.29380679e-01 -2.88628012e-01 -6.29415572e-01
8.19471598e-01 3.94740790e-01 1.88516542e-01 -3.77011538e-01
4.83596206e-01 -3.20545077e-01 -2.36116555e-02 -8.71359468e-01
8.92179608e-02 -3.39647710e-01 -2.33661607e-01 2.62609739e-02
-6.68247104e-01 2.16331869e-01 2.10881174e-01 2.49505594e-01
-6.55219615e-01 -5.08260250e-01 4.55174707e-02 -2.79437840e-01
-1.09793258e+00 2.69495726e-01 -1.80724874e-01 8.17792952e-01
5.21320879e-01 3.00301164e-01 -6.71522439e-01 -3.69148582e-01
-2.80165315e-01 5.28030813e-01 1.11509422e-02 2.87972450e-01
1.05347764e+00 -1.07679248e+00 -9.64523017e-01 -8.41303915e-02
5.34629107e-01 3.96742374e-01 -2.33540535e-01 4.66012836e-01
-5.97321749e-01 1.06909883e+00 -5.16077653e-02 2.95550883e-01
-1.04351938e+00 5.73388159e-01 2.87329871e-02 -5.42743742e-01
-5.18420935e-01 8.78547311e-01 -4.41656619e-01 -7.90349185e-01
7.13481903e-02 -6.35580003e-01 -5.41643836e-02 -2.17118159e-01
5.92783809e-01 3.56682628e-01 2.11939648e-01 -3.19319427e-01
-2.94283181e-01 -4.25589651e-01 -1.58683091e-01 -7.49506727e-02
1.31867230e+00 -2.11392403e-01 -8.11414957e-01 2.85531163e-01
1.07239854e+00 3.25884670e-01 -3.36730242e-01 -6.52866900e-01
6.53800547e-01 -3.69432390e-01 -1.87964037e-01 -7.45304763e-01
-1.35813281e-01 3.62986147e-01 -7.56790161e-01 -1.01684459e-01
1.02894354e+00 3.96196544e-01 1.48111272e+00 1.13758290e+00
4.53215092e-01 -1.04754269e+00 -2.39362001e-01 1.05689132e+00
9.84137356e-01 -1.38583994e+00 2.00344399e-01 -6.94812000e-01
-9.04965341e-01 8.61989021e-01 7.40842879e-01 3.48982930e-01
2.52935737e-01 5.54103553e-01 -2.31721699e-01 -5.78082442e-01
-1.50577021e+00 -9.99372080e-02 4.82067138e-01 1.35053471e-01
5.14540851e-01 1.56389892e-01 -4.68167156e-01 1.22449625e+00
-7.83551216e-01 1.32121608e-01 6.23134911e-01 8.77556741e-01
-3.55825335e-01 -1.07590330e+00 -3.13062072e-02 8.32393944e-01
-4.76957083e-01 -5.97974420e-01 -6.75079048e-01 6.90019548e-01
5.84779903e-02 1.00095117e+00 -8.92008543e-02 -2.75008023e-01
4.68504786e-01 1.25868991e-01 4.46532071e-01 -7.89283097e-01
-3.20385754e-01 -8.19200575e-01 1.13742936e+00 -7.52546966e-01
4.73635383e-02 -5.79738855e-01 -1.68170130e+00 -5.77283680e-01
-9.68745872e-02 3.82338166e-01 2.04282731e-01 1.21355808e+00
4.00422394e-01 1.38459295e-01 2.49085203e-01 1.87206164e-01
-6.14099205e-01 -8.28340888e-01 -2.51912504e-01 5.84450662e-01
1.23527087e-01 -8.08030292e-02 6.80196807e-02 -3.39975022e-02]
|
[9.738579750061035, 8.54740047454834]
|
8aa12a2c-2a86-4a58-bf95-9413980c0d23
|
no-fear-of-classifier-biases-neural-collapse
|
2303.10058
| null |
https://arxiv.org/abs/2303.10058v1
|
https://arxiv.org/pdf/2303.10058v1.pdf
|
No Fear of Classifier Biases: Neural Collapse Inspired Federated Learning with Synthetic and Fixed Classifier
|
Data heterogeneity is an inherent challenge that hinders the performance of federated learning (FL). Recent studies have identified the biased classifiers of local models as the key bottleneck. Previous attempts have used classifier calibration after FL training, but this approach falls short in improving the poor feature representations caused by training-time classifier biases. Resolving the classifier bias dilemma in FL requires a full understanding of the mechanisms behind the classifier. Recent advances in neural collapse have shown that the classifiers and feature prototypes under perfect training scenarios collapse into an optimal structure called simplex equiangular tight frame (ETF). Building on this neural collapse insight, we propose a solution to the FL's classifier bias problem by utilizing a synthetic and fixed ETF classifier during training. The optimal classifier structure enables all clients to learn unified and optimal feature representations even under extremely heterogeneous data. We devise several effective modules to better adapt the ETF structure in FL, achieving both high generalization and personalization. Extensive experiments demonstrate that our method achieves state-of-the-art performances on CIFAR-10, CIFAR-100, and Tiny-ImageNet.
|
['Chao Wu', 'Tao Lin', 'Rui He', 'Xinyi Shang', 'Zexi Li']
|
2023-03-17
| null | null | null | null |
['classifier-calibration', 'classifier-calibration']
|
['computer-vision', 'miscellaneous']
|
[-2.49103278e-01 -1.41237959e-01 -4.39395338e-01 -8.16412091e-01
-6.05705619e-01 -5.37078559e-01 2.86985606e-01 -3.46002936e-01
-3.03317755e-01 7.08920598e-01 1.09566003e-01 -2.48733163e-01
-2.67822272e-03 -5.69135666e-01 -8.87598455e-01 -7.79524565e-01
-2.89611295e-02 2.68527210e-01 1.00858606e-01 -2.12100506e-01
6.19589649e-02 4.75895315e-01 -2.02712393e+00 6.74472570e-01
8.70077968e-01 1.33592045e+00 -5.46790138e-02 4.19893265e-01
-3.00063848e-01 9.66276050e-01 -7.53071845e-01 -6.16240501e-01
4.35745150e-01 -9.13582668e-02 -8.18765879e-01 -3.77593189e-01
7.93911397e-01 -1.63870379e-01 -6.76803440e-02 1.13285410e+00
5.31225801e-01 -1.45272151e-01 2.80586451e-01 -1.61474121e+00
-6.32758439e-01 9.92173135e-01 -2.08044648e-01 2.99348354e-01
-1.33704364e-01 1.05139583e-01 8.76084149e-01 -9.44037199e-01
4.79589820e-01 1.32687366e+00 9.87931490e-01 8.91915917e-01
-1.16909373e+00 -1.08608222e+00 5.77542365e-01 2.43540213e-01
-1.40986407e+00 -6.12941742e-01 5.41798353e-01 -2.91397989e-01
8.35035801e-01 4.23904359e-01 4.89022493e-01 1.54164004e+00
1.18632369e-01 6.65620685e-01 1.00180972e+00 -4.09809172e-01
4.71607924e-01 4.23876077e-01 6.16040289e-01 6.14159763e-01
5.04226506e-01 8.77431333e-02 -8.78959417e-01 -5.13988554e-01
3.68905187e-01 -3.59667675e-03 -4.84625459e-01 -5.11631072e-01
-8.93397152e-01 9.29985881e-01 6.12566173e-01 2.82995194e-01
-1.30019814e-01 1.36532396e-01 4.86449808e-01 4.31074411e-01
4.11328793e-01 4.06401217e-01 -6.96390629e-01 6.77897409e-02
-6.83017015e-01 1.76479876e-01 9.56930339e-01 9.28421199e-01
8.94135177e-01 -1.14903413e-01 -7.80214295e-02 7.62682974e-01
2.56073549e-02 3.49005252e-01 7.30723083e-01 -9.17550862e-01
3.64988565e-01 8.18029761e-01 -1.91539302e-01 -7.78489351e-01
-1.82698667e-01 -1.09491253e+00 -8.47020209e-01 4.61001939e-04
3.47161472e-01 -1.38850346e-01 -5.73870659e-01 1.99865019e+00
3.99388433e-01 3.28043282e-01 1.44147739e-01 7.80105770e-01
6.36903346e-01 2.09308099e-02 1.93607770e-02 1.51274115e-01
1.05238509e+00 -1.08562124e+00 -4.70071554e-01 -1.87041864e-01
8.40977132e-01 -2.62112767e-01 1.14991713e+00 2.88953125e-01
-6.25119746e-01 -3.33514273e-01 -1.24517024e+00 2.53240377e-01
-4.83754098e-01 -2.25120842e-01 7.52267361e-01 9.66276705e-01
-8.24931443e-01 5.46285927e-01 -6.37189865e-01 -2.56312221e-01
8.82829785e-01 4.61189866e-01 -3.00769866e-01 -1.50594950e-01
-1.09393728e+00 7.86491752e-01 1.69624701e-01 -7.67749622e-02
-8.70607495e-01 -1.04267108e+00 -3.17819923e-01 1.57333970e-01
1.18094824e-01 -8.90252113e-01 1.48837268e+00 -1.38341105e+00
-1.28994668e+00 4.63706911e-01 -1.18147813e-01 -6.25915885e-01
5.71964443e-01 -5.54386824e-02 -3.59245777e-01 -2.89719224e-01
-2.21826524e-01 5.74272275e-01 9.26949322e-01 -1.43761182e+00
-9.47035849e-01 -5.03511131e-01 7.13746548e-02 6.06943443e-02
-9.00204718e-01 -7.43038282e-02 1.65041521e-01 -4.02191401e-01
5.98142482e-02 -7.06145942e-01 -5.37328199e-02 -1.35130137e-01
9.00330022e-02 -3.83587360e-01 9.28529739e-01 6.03333004e-02
1.35582221e+00 -1.99885404e+00 -2.64689982e-01 2.27812290e-01
4.26565558e-01 1.70134887e-01 -1.55595541e-01 -6.47647157e-02
4.93879104e-03 2.20242038e-01 1.25344962e-01 -3.06027621e-01
-8.34143609e-02 4.05441701e-01 -5.96848726e-01 4.36898470e-01
-1.08071186e-01 6.47027314e-01 -8.41651380e-01 -3.61187875e-01
-3.58826458e-01 4.82614160e-01 -8.38852286e-01 1.94656074e-01
-1.50884151e-01 3.06778014e-01 -4.27708387e-01 8.24768364e-01
8.01430285e-01 -4.15759832e-01 2.97495246e-01 -2.54649967e-01
3.92565243e-02 3.51114012e-02 -1.16133356e+00 1.60780883e+00
-4.06101942e-01 1.79754704e-01 2.35056460e-01 -8.51263583e-01
1.02188206e+00 1.00724861e-01 3.86117697e-01 -4.75875109e-01
2.51105189e-01 4.48070645e-01 -6.31931424e-02 -3.36348951e-01
2.17271239e-01 9.87122059e-02 1.32678092e-01 3.75703722e-01
3.61685753e-01 4.18266237e-01 -2.13190049e-01 5.41922078e-02
1.02946663e+00 -4.12430875e-02 -5.26468083e-02 -5.87769210e-01
4.28535968e-01 -1.42375454e-01 8.94892871e-01 1.09334278e+00
-3.99500549e-01 5.27428031e-01 2.51148194e-01 -9.94050860e-01
-6.48183584e-01 -6.66946054e-01 -4.13310349e-01 1.59404182e+00
-5.00176735e-02 -5.19290388e-01 -9.77141380e-01 -1.10733056e+00
2.36639306e-01 7.37841308e-01 -7.03489661e-01 -4.70707566e-01
-5.20231426e-01 -9.86067295e-01 7.29748249e-01 5.38955271e-01
7.27075517e-01 -4.19565082e-01 -7.62659073e-01 1.74159378e-01
-2.81158954e-01 -7.46336937e-01 -2.51136869e-01 5.01972735e-01
-8.34235430e-01 -1.14334857e+00 -2.49747247e-01 -5.22393048e-01
6.05173767e-01 3.96651268e-01 1.30411673e+00 2.66804010e-01
-1.30853027e-01 7.62997642e-02 -2.48622879e-01 -5.67950010e-01
-3.64553005e-01 4.72027659e-01 2.66033024e-01 1.47591516e-01
4.93814588e-01 -5.04318655e-01 -6.82509303e-01 7.10008025e-01
-5.83944619e-01 -1.95679069e-01 3.64997566e-01 1.09192121e+00
2.70778120e-01 -2.73027539e-01 7.46176839e-01 -8.93422842e-01
5.55379212e-01 -7.28022039e-01 -4.88603473e-01 5.65712512e-01
-9.99374449e-01 1.80245832e-01 8.03601086e-01 -6.11797035e-01
-1.08374333e+00 -4.44834679e-02 5.82383247e-03 -7.47117877e-01
4.17107195e-02 1.25009060e-01 -3.26411664e-01 -3.90009969e-01
1.27871609e+00 -2.21298382e-01 7.82525837e-02 -5.81226826e-01
2.15438947e-01 9.86677647e-01 4.50823575e-01 -1.12303317e+00
2.90702999e-01 5.30237556e-01 -2.92545378e-01 -1.36449948e-01
-1.12135386e+00 -1.95274860e-01 -5.14913857e-01 -1.45307645e-01
2.33246937e-01 -1.04804778e+00 -8.22579324e-01 3.68126005e-01
-1.01346755e+00 -1.07904315e-01 -3.96385014e-01 1.57299981e-01
-2.56663889e-01 -1.13527343e-01 -1.49149224e-01 -5.13847888e-01
-5.05207241e-01 -1.30937910e+00 6.28791213e-01 2.38816902e-01
-1.73509847e-02 -7.58271217e-01 1.26548514e-01 3.26062679e-01
9.49846983e-01 1.16123334e-02 8.74585986e-01 -8.27699840e-01
-4.15852934e-01 -1.50148034e-01 -4.65712622e-02 4.14799333e-01
-1.93592325e-01 -3.29700522e-02 -1.49396968e+00 -6.48648739e-01
9.92044210e-02 -4.39855456e-01 8.51110637e-01 -3.47284190e-02
1.52935719e+00 -4.14238483e-01 -4.94778246e-01 1.15518761e+00
1.41261220e+00 -2.09150985e-01 1.26285672e-01 5.24975836e-01
4.27604169e-01 4.04059172e-01 2.68962681e-01 4.95511025e-01
4.83372092e-01 4.76602644e-01 6.58390939e-01 2.40988165e-01
-2.55505383e-01 -1.98715180e-01 2.12937653e-01 6.66075289e-01
1.10529430e-01 5.31516597e-03 -1.08348763e+00 2.36132264e-01
-2.01092887e+00 -8.82663727e-01 2.22454295e-01 2.17409205e+00
6.37633383e-01 -1.35821804e-01 -1.03250831e-01 -2.00802106e-02
7.88394153e-01 -9.05647501e-02 -9.21334743e-01 -2.74598360e-01
-3.82143766e-01 -1.50374427e-01 6.18236065e-01 1.77261084e-01
-9.45143640e-01 7.59671509e-01 6.89017344e+00 6.33977950e-01
-1.37475204e+00 5.00325203e-01 7.21748292e-01 -3.14141959e-01
-1.77754685e-01 -6.42253458e-02 -1.20689034e+00 4.39791739e-01
1.18497729e+00 -1.98197693e-01 6.36129022e-01 1.30537403e+00
-5.20710409e-01 2.91163713e-01 -1.34611320e+00 1.10983014e+00
4.85157818e-02 -1.70484889e+00 2.15226442e-01 -9.99352038e-02
6.88635170e-01 4.89934355e-01 2.19432950e-01 5.45535326e-01
5.90067506e-01 -9.79979157e-01 1.00077116e+00 4.64048922e-01
8.27211082e-01 -7.29160964e-01 7.83421516e-01 4.77593124e-01
-9.01970625e-01 -7.96940923e-01 -5.72513819e-01 1.23987548e-01
-6.50591969e-01 5.33316374e-01 -7.85136461e-01 4.52798128e-01
1.21556723e+00 4.98901159e-01 -9.39082682e-01 8.76095772e-01
2.90283799e-01 5.50457656e-01 -3.83098632e-01 1.13733150e-01
-1.41557410e-01 2.19460309e-01 3.52353662e-01 1.07956660e+00
1.37154624e-01 -2.35283345e-01 1.48006082e-02 6.73737109e-01
-3.67348909e-01 -6.00742176e-02 -5.02679229e-01 3.87419164e-01
8.54416192e-01 1.17768002e+00 -2.21045390e-01 -1.96419656e-01
-2.93526739e-01 3.19636583e-01 7.92922556e-01 2.09500059e-01
-8.43336463e-01 7.79590011e-02 1.01360142e+00 -1.74510062e-01
1.67133108e-01 2.65419453e-01 -4.39527392e-01 -1.49681890e+00
-1.16457616e-03 -1.35948861e+00 7.70568311e-01 -1.57223716e-01
-1.56358182e+00 1.01275396e+00 -9.80352461e-02 -1.03177416e+00
-7.99558908e-02 -3.54856461e-01 -4.68091756e-01 5.94455481e-01
-1.47147882e+00 -1.20645320e+00 -7.11717665e-01 9.14685309e-01
3.72589082e-01 -5.69978654e-01 1.05136883e+00 2.86268592e-01
-9.29086328e-01 1.14394653e+00 3.72537434e-01 -1.76163703e-01
9.26442623e-01 -8.89750481e-01 1.05010271e-01 6.57727599e-01
-4.04538736e-02 7.43180871e-01 5.09373128e-01 -3.06755453e-01
-1.57849920e+00 -1.26238060e+00 5.02159715e-01 -6.74051404e-01
3.75196815e-01 -5.92614889e-01 -8.20929646e-01 6.95525527e-01
-1.85621418e-02 6.56088948e-01 8.26333225e-01 4.05263811e-01
-8.91106427e-01 -6.76189899e-01 -1.41580760e+00 3.14105839e-01
1.34931612e+00 -3.21675003e-01 -3.63964677e-01 4.40480977e-01
6.14079475e-01 -3.28224570e-01 -8.16273034e-01 4.42771673e-01
7.00252712e-01 -1.18698621e+00 7.66802430e-01 -8.28215599e-01
-6.12008236e-02 -7.20575973e-02 -6.18678153e-01 -1.34672046e+00
-4.38485473e-01 -7.44296014e-01 -1.79710895e-01 1.20890379e+00
2.60101050e-01 -1.08949542e+00 8.66549492e-01 7.73422718e-01
-8.18749145e-02 -9.28135455e-01 -1.14109540e+00 -7.63326347e-01
2.76100308e-01 -3.85675043e-01 1.39863050e+00 1.05002582e+00
-3.39770883e-01 1.41365066e-01 1.66264325e-01 1.22338377e-01
7.89941430e-01 9.70216170e-02 8.01052511e-01 -1.48258471e+00
-6.42217472e-02 -5.34707427e-01 -1.26332119e-01 -6.54627085e-01
4.34643239e-01 -1.15872645e+00 -2.54337400e-01 -6.72860861e-01
2.52761960e-01 -9.57550943e-01 -5.60974360e-01 6.74708545e-01
9.48097035e-02 -6.53877482e-02 3.14047337e-01 5.97468793e-01
-6.28636718e-01 4.29823935e-01 8.57981563e-01 -1.25754088e-01
8.63027722e-02 -1.63429290e-01 -9.75064039e-01 6.33807600e-01
7.96702325e-01 -6.10428870e-01 -2.63558537e-01 -7.23897040e-01
1.55048624e-01 -6.09219313e-01 3.15202177e-01 -1.13900125e+00
5.66780686e-01 -1.42476395e-01 5.04424095e-01 -2.11665213e-01
1.94770787e-02 -1.02065861e+00 2.89158642e-01 3.56437892e-01
-4.13535893e-01 1.45604163e-01 -7.93574005e-02 4.96926934e-01
-5.78723103e-02 -1.05951168e-01 1.02805638e+00 -2.55908698e-01
-4.50441360e-01 3.13540161e-01 5.79104610e-02 2.52404585e-02
1.04948366e+00 -9.30259898e-02 -9.30491388e-01 1.18335642e-01
-3.69633526e-01 3.09044719e-01 5.86450219e-01 5.20191491e-01
3.93311203e-01 -1.22424197e+00 -5.47939956e-01 6.86003804e-01
1.43988654e-01 7.63741285e-02 4.12609018e-02 7.53232062e-01
-3.08962494e-01 3.08405161e-01 -2.88141072e-01 -8.66298437e-01
-1.03039193e+00 5.20427167e-01 8.98622572e-01 -2.22064126e-02
-4.94583726e-01 1.17941189e+00 -8.34918488e-03 -6.65233374e-01
6.73140287e-01 -1.73013508e-01 1.63529754e-01 3.06218676e-02
5.48040569e-01 4.49455678e-01 5.25807202e-01 -4.52919781e-01
-5.04081547e-01 1.60155803e-01 -3.50228012e-01 3.87710631e-01
1.30212986e+00 -7.72491619e-02 -4.03559469e-02 2.04964668e-01
1.22179687e+00 -3.38935614e-01 -1.27152061e+00 -2.99582750e-01
-5.05117811e-02 -6.37028992e-01 2.25294665e-01 -8.92707109e-01
-1.22017860e+00 6.39789760e-01 9.52363253e-01 -1.91175807e-02
9.13195670e-01 -2.69791782e-01 5.81422925e-01 5.81080616e-01
7.67510116e-01 -1.04380989e+00 -1.28037930e-01 7.18766928e-01
7.98391700e-01 -1.09764111e+00 -2.22630948e-01 -1.00511715e-01
-4.34234351e-01 1.25815976e+00 8.75521779e-01 -3.12969089e-02
8.18994284e-01 3.85687023e-01 2.72463351e-01 -2.51041725e-03
-1.32079315e+00 3.51261437e-01 -1.41452804e-01 4.70553964e-01
1.49650738e-01 -7.02403784e-02 1.70900851e-01 8.77870142e-01
-3.31358045e-01 4.57767323e-02 1.84085488e-01 9.60488796e-01
-3.08965415e-01 -9.74965096e-01 -4.92872417e-01 3.58897895e-01
-4.24149990e-01 2.28206396e-01 -2.09918842e-01 5.69050848e-01
4.79980409e-01 8.91397834e-01 1.47086337e-01 -6.40200317e-01
2.48062402e-01 4.08000797e-01 4.33942139e-01 -2.94189990e-01
-1.20113277e+00 -4.15942669e-01 -1.25976026e-01 -8.24211180e-01
-1.30694509e-01 -3.84200811e-01 -9.31643903e-01 -4.25590813e-01
-5.10246933e-01 2.88190573e-01 9.70195949e-01 6.45042062e-01
8.55270445e-01 2.70722568e-01 8.49633038e-01 -6.79198384e-01
-1.38265622e+00 -8.06601703e-01 -2.01003134e-01 4.06877726e-01
4.19241160e-01 -7.81934142e-01 -6.50384784e-01 -1.97341442e-01]
|
[5.938273906707764, 6.280642986297607]
|
89e7c62b-e0d6-4c0c-9910-fef8194d1cb1
|
image-translation-for-medical-image
|
2010.02745
| null |
https://arxiv.org/abs/2010.02745v2
|
https://arxiv.org/pdf/2010.02745v2.pdf
|
Image Translation for Medical Image Generation -- Ischemic Stroke Lesions
|
Deep learning based disease detection and segmentation algorithms promise to improve many clinical processes. However, such algorithms require vast amounts of annotated training data, which are typically not available in the medical context due to data privacy, legal obstructions, and non-uniform data acquisition protocols. Synthetic databases with annotated pathologies could provide the required amounts of training data. We demonstrate with the example of ischemic stroke that an improvement in lesion segmentation is feasible using deep learning based augmentation. To this end, we train different image-to-image translation models to synthesize magnetic resonance images of brain volumes with and without stroke lesions from semantic segmentation maps. In addition, we train a generative adversarial network to generate synthetic lesion masks. Subsequently, we combine these two components to build a large database of synthetic stroke images. The performance of the various models is evaluated using a U-Net which is trained to segment stroke lesions on a clinical test set. We report a Dice score of $\mathbf{72.8}$% [$\mathbf{70.8\pm1.0}$%] for the model with the best performance, which outperforms the model trained on the clinical images alone $\mathbf{67.3}$% [$\mathbf{63.2\pm1.9}$%], and is close to the human inter-reader Dice score of $\mathbf{76.9}$%. Moreover, we show that for a small database of only 10 or 50 clinical cases, synthetic data augmentation yields significant improvement compared to a setting where no synthetic data is used. To the best of our knowledge, this presents the first comparative analysis of synthetic data augmentation based on image-to-image translation, and first application to ischemic stroke.
|
['Christian Federau', 'Jonathan Zopes', 'Moritz Platscher']
|
2020-10-05
| null | null | null | null |
['medical-image-generation']
|
['medical']
|
[ 6.03351116e-01 5.55310547e-01 9.98186693e-02 -3.73034060e-01
-1.15163207e+00 -4.94131774e-01 3.94086748e-01 1.68512329e-01
-7.03179359e-01 9.79977429e-01 -3.90762985e-02 -4.45333213e-01
-1.50157157e-02 -7.70815551e-01 -8.65031064e-01 -5.96735358e-01
-1.15710497e-01 7.93203533e-01 7.41032362e-02 2.23714441e-01
-1.23217009e-01 5.31662524e-01 -9.99533057e-01 2.97229648e-01
1.11906517e+00 9.38081861e-01 -4.29953821e-02 6.05452061e-01
1.64539427e-01 4.66429502e-01 -6.29292846e-01 -5.56956172e-01
5.75057328e-01 -7.63754547e-01 -7.67504454e-01 -6.92709722e-03
4.10357863e-01 -7.46137142e-01 -3.87237936e-01 1.04194736e+00
7.27998197e-01 -2.36653183e-02 8.74174774e-01 -1.07850420e+00
-4.83185142e-01 5.26040614e-01 -3.84764105e-01 1.91937625e-01
-4.33007739e-02 3.69153976e-01 3.49229306e-01 -4.79794621e-01
8.62965345e-01 5.82564652e-01 5.68746328e-01 6.60292625e-01
-1.29923630e+00 -7.24591076e-01 -3.81812245e-01 -2.12152004e-01
-1.11681032e+00 -1.44569010e-01 3.18394244e-01 -6.60113633e-01
5.37116945e-01 3.89099717e-01 6.03380024e-01 1.06105685e+00
9.50298086e-02 5.60439169e-01 1.43340552e+00 -2.03551635e-01
2.73997635e-01 -9.61687998e-04 3.71834412e-02 5.75959444e-01
3.37280542e-01 4.78484631e-02 1.18826307e-01 -6.14049807e-02
1.02503443e+00 4.82872017e-02 -3.77205580e-01 -2.24263787e-01
-1.33104146e+00 7.64410973e-01 5.26925266e-01 2.02395692e-01
-4.84551966e-01 8.69602561e-02 3.57187808e-01 7.51582012e-02
2.75217056e-01 5.03573895e-01 -1.25729442e-01 6.58859834e-02
-1.24629235e+00 2.92881161e-01 4.50214565e-01 7.00083494e-01
1.64134145e-01 6.51964471e-02 -3.08468878e-01 7.65725613e-01
-2.50511587e-01 6.88883424e-01 5.60928941e-01 -1.25020778e+00
4.12336737e-01 3.80674064e-01 1.70360476e-01 -6.53267860e-01
-5.96296251e-01 -5.97997248e-01 -1.01917398e+00 5.30990779e-01
8.81723404e-01 -4.06238824e-01 -1.25516570e+00 1.75964463e+00
-6.44555092e-02 -1.10662952e-01 3.35893556e-02 8.58336389e-01
6.86660230e-01 1.40337005e-01 1.82040036e-01 -2.01931193e-01
1.21980071e+00 -6.45314395e-01 -5.45026541e-01 -5.45770936e-02
7.35830128e-01 -6.09517992e-01 1.26182568e+00 1.94930524e-01
-1.48235977e+00 -1.99669570e-01 -8.98018539e-01 2.26694122e-01
-1.84735581e-01 7.04088435e-02 2.45040789e-01 8.91318321e-01
-9.78584647e-01 5.74217618e-01 -1.04680669e+00 -1.34370327e-01
1.07266271e+00 4.28342760e-01 -2.87961274e-01 -2.50033289e-01
-1.00903535e+00 9.63764489e-01 2.40567058e-01 -2.40339860e-01
-8.01885784e-01 -1.02581668e+00 -5.97929955e-01 -1.60493135e-01
1.51171461e-01 -6.83286846e-01 1.02421069e+00 -7.85465181e-01
-1.09778857e+00 1.09121931e+00 1.82305783e-01 -8.08926225e-01
1.24411881e+00 1.52719140e-01 -1.13725737e-01 4.95821148e-01
2.46152699e-01 9.38790619e-01 3.59490454e-01 -1.23777926e+00
-1.53840929e-01 -4.61360902e-01 -1.62632138e-01 1.79049131e-02
3.69087532e-02 1.21180505e-01 -1.40133813e-01 -8.63222599e-01
3.52166481e-02 -1.05547845e+00 -4.82047230e-01 1.37769714e-01
-3.90538990e-01 6.73035204e-01 3.39220792e-01 -1.08991313e+00
7.68698812e-01 -1.77603340e+00 -1.90388218e-01 5.31484008e-01
4.67417926e-01 4.74545240e-01 7.47065097e-02 -1.56211406e-01
-1.98142648e-01 2.88741320e-01 -9.30715203e-01 -7.15738460e-02
-2.29114398e-01 1.96737237e-02 8.37645605e-02 4.05667663e-01
7.58810416e-02 1.18834889e+00 -7.78716207e-01 -3.64778101e-01
1.73430666e-01 5.33805549e-01 -5.67917109e-01 2.27341782e-02
1.04351446e-01 8.65729332e-01 -4.21684176e-01 3.77849638e-01
6.36213481e-01 -2.14766666e-01 5.26220165e-02 -3.08773257e-02
3.85307580e-01 -2.67067075e-01 -8.70408058e-01 1.70466948e+00
-1.40741661e-01 4.03742194e-01 1.45248193e-02 -1.22962546e+00
6.98845029e-01 4.46442932e-01 9.21584725e-01 -7.46199965e-01
3.96352410e-01 4.16505277e-01 4.72822219e-01 -3.12740356e-01
-1.14718750e-01 -4.58800107e-01 5.53035885e-02 7.13250756e-01
-1.63126796e-01 -4.83229727e-01 1.93887591e-01 1.73841268e-01
1.45049846e+00 -1.46509916e-01 -1.81777358e-01 -1.60791159e-01
2.77794003e-01 1.48171604e-01 2.36826912e-01 8.42524290e-01
-5.25554180e-01 1.13285482e+00 6.69470608e-01 -2.23642960e-01
-1.36813712e+00 -1.35592258e+00 -4.11500782e-01 3.33897471e-01
-1.93368137e-01 1.32987857e-01 -1.35330284e+00 -6.69142842e-01
-2.89051265e-01 7.47327566e-01 -6.42517984e-01 -1.52651384e-01
-6.52738869e-01 -1.15407908e+00 8.84251535e-01 7.00040340e-01
3.96773070e-01 -1.23648679e+00 -9.36095834e-01 2.38380000e-01
-1.88212648e-01 -1.25429296e+00 -3.29031974e-01 -2.18391925e-01
-8.98433328e-01 -1.22588658e+00 -1.31793809e+00 -5.08813500e-01
9.15885866e-01 -3.49456459e-01 9.80948150e-01 7.37369582e-02
-6.12190545e-01 2.89931655e-01 -2.92848051e-01 -5.40470302e-01
-6.04923487e-01 -1.74897194e-01 -1.14563957e-01 -2.35755965e-01
3.16329114e-03 -6.70816898e-01 -1.13650715e+00 1.89245895e-01
-1.14377785e+00 1.59786001e-01 6.43545389e-01 9.40370262e-01
8.60178530e-01 -4.37422246e-01 6.74756765e-01 -1.01015174e+00
6.48440301e-01 -1.85399652e-01 -3.43811810e-01 9.11493152e-02
-6.18408203e-01 -1.38525814e-01 5.41694105e-01 -4.41248834e-01
-8.03362966e-01 7.31088072e-02 -8.31851512e-02 -4.05121654e-01
-3.23382020e-01 3.36934835e-01 1.71359658e-01 1.05718091e-01
1.03188801e+00 8.44171867e-02 2.94521242e-01 -2.75447816e-01
4.66085434e-01 5.12914121e-01 8.37175250e-01 -4.00626451e-01
4.96216923e-01 5.67763448e-01 5.46152070e-02 -2.66534418e-01
-5.20616353e-01 2.41606534e-01 -7.27833688e-01 -2.80927196e-02
1.09440207e+00 -5.26899517e-01 -2.41648242e-01 3.14938605e-01
-7.53196359e-01 -7.32833266e-01 -6.54110253e-01 7.68713772e-01
-8.80706012e-01 3.04445118e-01 -5.57051957e-01 -3.04467708e-01
-6.84045017e-01 -1.70755231e+00 7.08641052e-01 -8.51484984e-02
-2.09852546e-01 -6.97730958e-01 -3.06185305e-01 6.22338593e-01
4.26826268e-01 9.29802597e-01 9.34287906e-01 -7.96087682e-01
-3.25823456e-01 -5.21563649e-01 -5.10973632e-01 5.55597842e-01
2.09761053e-01 -4.32794064e-01 -6.80043161e-01 -2.25375727e-01
-1.43997639e-01 -3.60307366e-01 6.08370125e-01 7.53428221e-01
1.39176750e+00 -3.31694842e-04 -1.18152045e-01 4.56884801e-01
1.07823360e+00 5.33371091e-01 8.71975541e-01 8.64253491e-02
5.67367673e-01 5.09984791e-01 1.99979052e-01 2.46841326e-01
1.05325051e-01 4.03712869e-01 2.61735976e-01 -4.22975302e-01
-4.72324908e-01 1.68380514e-01 -1.92231670e-01 2.52277106e-01
-6.53890893e-02 -4.68047224e-02 -1.26062286e+00 5.80073714e-01
-1.42130268e+00 -7.30683625e-01 -2.15351552e-01 2.27977586e+00
9.56499755e-01 2.73370504e-01 1.82240278e-01 1.28359154e-01
7.11074591e-01 -3.01483959e-01 -5.74353039e-01 -2.48437002e-01
-9.50067863e-02 9.04631972e-01 6.13524735e-01 4.76997674e-01
-9.06933308e-01 7.58122981e-01 5.50918436e+00 5.96224964e-01
-1.14205194e+00 4.18568492e-01 1.06935132e+00 -3.37692291e-01
-1.28819764e-01 -3.07656437e-01 7.24200234e-02 6.35152936e-01
1.02754951e+00 -1.34939849e-01 3.33842069e-01 3.63253385e-01
3.41024697e-01 -1.37423590e-01 -9.93500769e-01 8.11015308e-01
-5.00951894e-02 -1.38224983e+00 -1.11961469e-01 8.72828662e-02
8.55300844e-01 1.99877411e-01 1.75905421e-01 -1.35066897e-01
2.80464143e-01 -1.45689094e+00 4.14006948e-01 5.46946168e-01
1.28220820e+00 -5.41649759e-01 7.81445384e-01 1.93924740e-01
-4.75957483e-01 3.28220993e-01 5.89618906e-02 4.15208966e-01
3.85397941e-01 6.05333149e-01 -9.76691306e-01 4.63590384e-01
6.09642684e-01 1.15007006e-01 -3.82761866e-01 1.10380769e+00
-2.30208077e-02 5.12962997e-01 -3.75731409e-01 4.61996078e-01
1.94307894e-01 -2.09287122e-01 3.47076774e-01 1.06038833e+00
3.96978825e-01 3.84813190e-01 -1.34887882e-02 1.09770703e+00
-2.75142908e-01 3.49864185e-01 -4.87671316e-01 7.19022155e-02
7.71663710e-02 9.74233449e-01 -9.51462567e-01 -5.73462009e-01
-2.83674121e-01 9.60211217e-01 3.54959853e-02 3.00084144e-01
-8.46742272e-01 -4.11586225e-01 2.35867873e-01 4.75456655e-01
-1.07863456e-01 5.74582741e-02 -7.48154223e-01 -1.03428459e+00
5.91498017e-02 -8.30264807e-01 3.39181006e-01 -8.13623250e-01
-1.08970356e+00 7.91654110e-01 1.25998348e-01 -1.09694946e+00
-3.03631902e-01 -5.39212942e-01 -4.56889182e-01 1.15137446e+00
-1.07194030e+00 -9.15029168e-01 -3.69003713e-01 5.05761445e-01
7.53076449e-02 -1.69947788e-01 9.05633330e-01 4.38120335e-01
-3.64326566e-01 8.04096520e-01 1.96169853e-01 4.06909168e-01
6.36362910e-01 -1.14515626e+00 2.15635926e-01 8.17142785e-01
-2.76925057e-01 3.81066740e-01 4.50145364e-01 -6.74865365e-01
-6.30977869e-01 -1.13630140e+00 4.50808942e-01 -5.26352763e-01
4.58193809e-01 5.81710115e-02 -9.92353201e-01 7.51700819e-01
-9.94135719e-03 3.24710220e-01 6.42717898e-01 -7.52587676e-01
2.06631329e-02 2.20221832e-01 -1.75132608e+00 7.28825212e-01
9.50492740e-01 -2.53641367e-01 -4.72361803e-01 6.07760310e-01
3.90886009e-01 -6.96131289e-01 -1.40843010e+00 6.74362123e-01
5.17543375e-01 -7.91347742e-01 1.04324067e+00 -8.35724115e-01
6.28267407e-01 -1.47022412e-03 -4.23874222e-02 -1.13446391e+00
2.81172544e-01 -3.59066725e-01 3.15278947e-01 7.39661098e-01
6.75062001e-01 -6.86564982e-01 1.02390254e+00 1.15973759e+00
-2.46121004e-01 -8.60525787e-01 -9.36949670e-01 -5.29038489e-01
7.89433300e-01 -4.61892188e-01 3.59447628e-01 9.55555439e-01
-2.13351905e-01 -3.30120176e-01 -1.04615048e-01 -1.61080390e-01
6.74289405e-01 -1.13849759e-01 4.38787848e-01 -8.97348940e-01
-9.64707136e-02 -5.63008308e-01 -3.87874842e-01 -2.67286897e-01
-1.65665448e-02 -1.19452465e+00 -2.28026018e-01 -1.71339869e+00
2.68542677e-01 -7.32503712e-01 -4.14068162e-01 5.41899562e-01
-7.18519092e-02 6.32370412e-01 2.06081972e-01 5.33868000e-02
1.15945324e-01 2.87244260e-01 1.53170693e+00 -1.44237921e-01
-2.42114231e-01 -9.91918147e-02 -7.10910618e-01 6.73501015e-01
1.19579971e+00 -4.27801639e-01 -4.86539006e-01 -3.82247448e-01
-3.97139400e-01 2.36936599e-01 5.74562550e-01 -9.50950325e-01
-1.67152390e-01 4.40420322e-02 5.32476783e-01 -2.59797454e-01
2.50338286e-01 -6.36852682e-01 2.26904139e-01 8.60140324e-01
-3.69449049e-01 -3.71929184e-02 1.63560957e-01 1.20308883e-01
3.22476067e-02 -1.50583982e-01 9.84568655e-01 -4.18950886e-01
2.21097711e-02 3.89103800e-01 -3.24709982e-01 4.23949778e-01
1.12710845e+00 -2.39594445e-01 -2.39493981e-01 -4.68378305e-01
-1.19785345e+00 -2.78129466e-02 3.58893961e-01 5.41972369e-02
4.43141818e-01 -1.14419603e+00 -8.70953798e-01 9.24156234e-02
-1.86727941e-01 8.91975164e-02 3.06095690e-01 1.25829494e+00
-8.98921669e-01 2.77830422e-01 -5.23791313e-01 -5.04164875e-01
-9.44068611e-01 3.33333284e-01 6.27557755e-01 -3.53181481e-01
-7.10491657e-01 5.09614885e-01 2.16998816e-01 -3.31309706e-01
1.14432290e-01 -4.45619136e-01 2.35267580e-01 -1.21187687e-01
3.02757382e-01 3.85277420e-01 2.30954051e-01 -6.45706415e-01
-1.76587790e-01 3.09911668e-01 -1.49709508e-02 -5.32293379e-01
1.23538971e+00 2.33831644e-01 6.01256192e-02 -2.92302400e-01
1.12695110e+00 -3.58369112e-01 -1.19425988e+00 -6.42286986e-02
-2.69358635e-01 -3.39051694e-01 -1.13950327e-01 -1.31743824e+00
-1.35475707e+00 9.72810209e-01 1.06805778e+00 -1.18994534e-01
1.09515262e+00 -1.65838823e-01 9.92151856e-01 1.21879550e-02
1.91772074e-01 -8.27913404e-01 -4.79822569e-02 -8.81444439e-02
9.06138480e-01 -1.19276202e+00 -1.71206638e-01 -3.40947211e-01
-7.82838166e-01 6.13307118e-01 3.73345405e-01 -1.77708283e-01
5.91273129e-01 3.41979265e-01 3.26664418e-01 -8.57873857e-02
8.31841975e-02 -3.05937063e-02 2.56117791e-01 6.90336645e-01
2.36362740e-01 2.29145795e-01 -7.75203228e-01 6.83821976e-01
-3.89123231e-01 2.40308404e-01 5.96643209e-01 9.63498414e-01
1.79221462e-02 -1.25712240e+00 -2.60674328e-01 9.11265612e-01
-8.42523158e-01 -2.13730589e-01 -2.82315463e-01 8.37072611e-01
1.79002777e-01 7.02643096e-01 -5.36954552e-02 2.13393867e-02
4.68430966e-01 2.55563349e-01 6.20363951e-01 -3.70911151e-01
-6.58292234e-01 -3.91910877e-03 -9.65501592e-02 -4.49479669e-01
-3.05781007e-01 -8.05736244e-01 -1.63288581e+00 -1.10107847e-01
2.58506656e-01 -7.68468231e-02 6.25059605e-01 7.54122257e-01
3.00423354e-01 6.91607118e-01 1.38974130e-01 -6.49545312e-01
-3.30793768e-01 -8.30247104e-01 -4.45683956e-01 7.22168982e-01
-2.54907017e-03 -5.24937153e-01 1.46772768e-02 1.99567080e-01]
|
[14.262007713317871, -2.0787737369537354]
|
37e46461-5140-4c7f-9d2d-f1af0165a373
|
harflow3d-a-latency-oriented-3d-cnn
|
2303.17218
| null |
https://arxiv.org/abs/2303.17218v6
|
https://arxiv.org/pdf/2303.17218v6.pdf
|
HARFLOW3D: A Latency-Oriented 3D-CNN Accelerator Toolflow for HAR on FPGA Devices
|
For Human Action Recognition tasks (HAR), 3D Convolutional Neural Networks have proven to be highly effective, achieving state-of-the-art results. This study introduces a novel streaming architecture based toolflow for mapping such models onto FPGAs considering the model's inherent characteristics and the features of the targeted FPGA device. The HARFLOW3D toolflow takes as input a 3D CNN in ONNX format and a description of the FPGA characteristics, generating a design that minimizes the latency of the computation. The toolflow is comprised of a number of parts, including i) a 3D CNN parser, ii) a performance and resource model, iii) a scheduling algorithm for executing 3D models on the generated hardware, iv) a resource-aware optimization engine tailored for 3D models, v) an automated mapping to synthesizable code for FPGAs. The ability of the toolflow to support a broad range of models and devices is shown through a number of experiments on various 3D CNN and FPGA system pairs. Furthermore, the toolflow has produced high-performing results for 3D CNN models that have not been mapped to FPGAs before, demonstrating the potential of FPGA-based systems in this space. Overall, HARFLOW3D has demonstrated its ability to deliver competitive latency compared to a range of state-of-the-art hand-tuned approaches being able to achieve up to 5$\times$ better performance compared to some of the existing works.
|
['Dimitrios Tzovaras', 'Christos-Savvas Bouganis', 'Alexander Montgomerie-Corcoran', 'Petros Toupas']
|
2023-03-30
| null | null | null | null |
['action-recognition-in-videos']
|
['computer-vision']
|
[ 1.75259963e-01 -1.90438870e-02 -1.47102535e-01 -4.11030799e-01
5.72128184e-02 -3.04491192e-01 2.64091074e-01 -1.24515602e-02
-4.73563194e-01 1.81979686e-01 -3.20426635e-02 -6.55482829e-01
-6.49302080e-02 -7.26575315e-01 -6.78584933e-01 3.28601338e-02
-4.57293481e-01 2.47192308e-01 4.31930959e-01 -1.27358973e-01
7.15903044e-02 1.08293414e+00 -1.95773554e+00 5.39962947e-01
-1.01944946e-01 1.40553451e+00 2.65074074e-02 1.02597320e+00
-1.42062038e-01 9.16720629e-01 -8.93037677e-01 -4.23340574e-02
7.34871745e-01 -2.75133342e-01 -6.03513479e-01 -9.37605202e-02
6.32959187e-01 -6.17886722e-01 -3.26086462e-01 6.13202333e-01
6.12190962e-01 -4.26434934e-01 1.46741301e-01 -1.25938559e+00
2.67963827e-01 6.69347644e-01 -1.47500811e-02 2.39668399e-01
2.06056476e-01 4.00574088e-01 4.57446963e-01 -4.23354626e-01
5.75575590e-01 1.31712234e+00 7.23330677e-01 4.65113848e-01
-8.18732440e-01 -6.47515297e-01 -4.31676179e-01 8.32982957e-02
-1.23827755e+00 -2.27610111e-01 3.61448437e-01 -4.24986601e-01
1.93583512e+00 7.92939663e-02 1.32003903e+00 9.39185500e-01
5.69908559e-01 4.30625886e-01 7.62426734e-01 -7.07597136e-01
6.82256997e-01 -2.14575917e-01 2.47052521e-01 7.43271470e-01
3.18178773e-01 2.31522411e-01 -5.64991653e-01 8.51876363e-02
9.68711078e-01 -5.01226902e-01 3.21802467e-01 -1.85779423e-01
-1.06830943e+00 5.48478067e-01 4.76606309e-01 2.07682267e-01
-5.06204784e-01 7.57579207e-01 9.62634921e-01 1.36697590e-01
-8.59077051e-02 2.58615702e-01 -7.10662544e-01 -4.69658852e-01
-9.53255951e-01 6.00716829e-01 9.64789927e-01 1.42721319e+00
4.01013762e-01 4.58803087e-01 -3.98926809e-03 1.07077494e-01
2.52075404e-01 1.08143404e-01 4.33673859e-01 -8.09826791e-01
4.32292998e-01 9.42084253e-01 -2.86561757e-01 -7.49113858e-01
-8.26866925e-01 -5.11519909e-01 -5.26990891e-01 6.79987967e-01
2.59732246e-01 -2.53536373e-01 -8.65882993e-01 1.28289187e+00
3.07321936e-01 -9.16283578e-02 2.78462559e-01 7.44466543e-01
7.42009759e-01 4.51478601e-01 -5.25645129e-02 4.12673801e-01
1.33799076e+00 -8.76504779e-01 -3.65476519e-01 -3.29732031e-01
8.45086336e-01 -7.18510926e-01 6.31735861e-01 4.03435439e-01
-1.37895155e+00 -1.03752923e+00 -1.77912438e+00 -9.23616905e-03
-3.70167792e-01 5.55149198e-01 6.37978554e-01 1.17089701e+00
-1.27293062e+00 6.45423472e-01 -1.21079624e+00 -5.58495879e-01
6.98941350e-01 7.58287430e-01 -1.34536892e-01 6.30992353e-02
-8.40944052e-01 9.66173887e-01 8.97778749e-01 1.07406460e-01
-9.86512661e-01 -9.90192354e-01 -7.37668335e-01 2.92293429e-01
-2.08447664e-03 -7.90033519e-01 1.52081048e+00 -1.14753735e+00
-1.73213220e+00 5.55955768e-01 5.36877036e-01 -1.10952115e+00
5.56555927e-01 -6.24132566e-02 -3.66465926e-01 2.09640160e-01
-2.25592121e-01 1.00768530e+00 6.93135738e-01 -1.55657362e-02
-9.62725103e-01 -7.63226999e-03 5.04747689e-01 -9.52366069e-02
-3.73827547e-01 3.73820215e-02 -4.41094548e-01 -4.29446757e-01
-4.69895303e-01 -8.83936167e-01 -3.23324740e-01 1.84317634e-01
-1.25845268e-01 2.65233755e-01 7.90911973e-01 -2.27829471e-01
1.00012100e+00 -2.18050122e+00 -3.04933190e-01 9.54556614e-02
-8.45133141e-02 7.99066186e-01 -5.01647890e-02 5.13886034e-01
-2.29180500e-01 -7.75859356e-02 3.37929189e-01 9.46763456e-02
2.45549351e-01 4.40017693e-02 5.14142327e-02 3.49658668e-01
7.01152682e-01 5.47056913e-01 -5.72329521e-01 -1.61446735e-01
5.85777283e-01 7.49531031e-01 -6.59984708e-01 5.39767332e-02
-2.46536985e-01 -3.73521030e-01 -4.41031337e-01 6.30832553e-01
4.96515453e-01 6.61965981e-02 3.83636057e-01 -3.82569402e-01
-4.61771756e-01 1.57869712e-01 -1.47768760e+00 1.60269535e+00
-3.79389763e-01 8.22326899e-01 -1.17878735e-01 -7.13159204e-01
1.30830801e+00 3.20129603e-01 1.12172335e-01 -6.40866101e-01
7.45337546e-01 4.53072190e-01 1.67598352e-01 -4.15399790e-01
4.30885255e-01 3.63748759e-01 -2.91912463e-02 2.76005179e-01
4.33984846e-01 -7.99051151e-02 4.61690813e-01 -3.11297446e-01
1.62546384e+00 4.76116896e-01 5.68186522e-01 -6.77056968e-01
5.49373329e-01 4.13863868e-01 1.85780257e-01 5.09990931e-01
-1.49557516e-01 1.53127715e-01 7.70067334e-01 -9.74905968e-01
-1.38229179e+00 -4.93167222e-01 1.10261224e-01 3.12045246e-01
-3.80426019e-01 -4.36710566e-01 -1.26758909e+00 -2.94443607e-01
-2.02077657e-01 4.56522167e-01 -5.51369548e-01 -1.03853323e-01
-6.15027845e-01 -2.53758579e-01 1.01862240e+00 8.69356096e-01
9.33256507e-01 -1.10107481e+00 -1.97600591e+00 6.55821204e-01
1.06176555e+00 -1.46124959e+00 -3.67553271e-02 6.54679954e-01
-1.17282009e+00 -9.48702514e-01 -8.13596919e-02 -7.36174643e-01
5.72093368e-01 -5.70859239e-02 1.01517475e+00 2.33609192e-02
-7.24436522e-01 1.16909064e-01 -5.47392011e-01 -9.24850643e-01
-5.18852830e-01 3.36776197e-01 -8.09979588e-02 -3.07977557e-01
3.52249831e-01 -4.42347020e-01 -3.08141947e-01 8.89321044e-02
-1.06521535e+00 2.44094640e-01 6.92488670e-01 5.48931301e-01
5.09696603e-01 3.26006591e-01 2.95134753e-01 -4.90511835e-01
3.27850133e-01 1.31209150e-01 -1.15374422e+00 -2.20924586e-01
-2.46582761e-01 -5.90897650e-02 8.89326513e-01 -3.46160173e-01
-5.92756152e-01 9.14364994e-01 -2.05844149e-01 -5.96264780e-01
-4.60685909e-01 4.25424606e-01 -3.29036474e-01 -1.75783575e-01
9.44367886e-01 -4.13927555e-01 3.27872574e-01 -1.58357263e-01
-4.48675044e-02 5.89805067e-01 3.14042598e-01 -1.61668062e-01
3.83758426e-01 3.11506063e-01 5.01369774e-01 -8.63202035e-01
-5.07484794e-01 1.20789241e-02 -8.70419025e-01 -4.34151858e-01
8.16879690e-01 -9.18236077e-01 -6.97440982e-01 6.33521616e-01
-1.20411754e+00 -6.80826247e-01 -5.18200815e-01 6.17046893e-01
-6.52756691e-01 -4.23198730e-01 -2.65056700e-01 -4.99459177e-01
-3.96742612e-01 -1.62930989e+00 6.59362435e-01 4.08472568e-01
-4.08808827e-01 -6.64638758e-01 -2.89037526e-01 -4.22779433e-02
5.02019346e-01 5.43948293e-01 9.34329867e-01 -7.68748045e-01
-7.45499909e-01 -5.81507981e-01 -1.28892541e-01 5.33307374e-01
-3.84170830e-01 4.09235954e-01 -9.52039897e-01 -5.81826605e-02
-2.68693209e-01 -1.19240120e-01 2.45947137e-01 5.02050281e-01
7.67169833e-01 1.52479559e-01 -1.97942674e-01 6.14639521e-01
1.61575973e+00 4.13640320e-01 7.20665216e-01 5.91902792e-01
2.75940567e-01 2.70302117e-01 7.13406861e-01 5.74585021e-01
-1.33048192e-01 7.48424768e-01 9.72124338e-01 -1.48075342e-01
-3.10976148e-01 1.17323808e-01 4.66532886e-01 3.15488070e-01
1.06687598e-01 -1.10927619e-01 -9.55076456e-01 3.62643927e-01
-1.62239575e+00 -7.03900039e-01 -2.05511004e-01 1.95903361e+00
1.89640209e-01 6.16636395e-01 3.23540479e-01 5.27881145e-01
6.02537692e-01 -1.20742202e-01 -4.23777193e-01 -1.17689562e+00
3.50333184e-01 9.18160975e-01 8.29426587e-01 -2.14239433e-02
-9.70703542e-01 8.95384610e-01 6.10449028e+00 6.18599296e-01
-1.40446126e+00 -2.02088282e-01 1.81536093e-01 -3.21316272e-01
6.23510003e-01 -5.49976043e-02 -1.26920569e+00 -4.08593789e-02
1.59717786e+00 -2.82250415e-03 2.02307284e-01 1.22884548e+00
2.47693107e-01 1.45323381e-01 -1.25155401e+00 7.16058254e-01
8.57897475e-02 -1.79128730e+00 -1.72311335e-03 6.54316545e-02
3.71714830e-01 -1.84086096e-02 -3.79051954e-01 1.26152128e-01
-4.98063676e-02 -1.13654542e+00 1.47890401e+00 1.00677907e-01
8.20237577e-01 -1.26328993e+00 1.02564871e+00 1.41170815e-01
-1.12522221e+00 -3.16737562e-01 -3.50997299e-01 -4.12557721e-01
-2.99178227e-03 4.05139208e-01 -1.23751271e+00 5.60737431e-01
8.92910421e-01 5.56797922e-01 -5.60519636e-01 1.21016324e+00
-1.41488671e-01 3.60313773e-01 -2.72927195e-01 -5.82109094e-01
5.79601407e-01 6.11034513e-01 9.60631445e-02 1.46602535e+00
4.73748058e-01 -6.21076748e-02 -3.38023394e-01 5.76200247e-01
2.33637094e-01 8.31361786e-02 -3.56762528e-01 -2.11748108e-01
3.78936470e-01 1.23954105e+00 -1.13228285e+00 -2.53139019e-01
-4.14491177e-01 4.93865669e-01 -2.10668556e-02 -3.91875803e-01
-1.02745879e+00 -6.85770452e-01 7.83112943e-01 1.70465305e-01
8.79967749e-01 -4.25451696e-01 -2.74064749e-01 -3.03501546e-01
-6.50290400e-03 -9.35015857e-01 8.65813568e-02 -8.86019647e-01
-3.52482319e-01 1.03370428e+00 6.63171187e-02 -1.49265659e+00
-3.04980636e-01 -1.07263684e+00 -2.27373466e-01 7.40889549e-01
-1.21116889e+00 -9.65507925e-01 -5.27740180e-01 2.85455912e-01
5.56308448e-01 -5.45927405e-01 9.90459085e-01 4.86417234e-01
-5.24478614e-01 4.87659842e-01 -4.73241866e-01 -1.25718251e-01
1.30587742e-01 -8.01701486e-01 8.87577593e-01 8.00616384e-01
-1.80557743e-01 8.73608291e-02 3.24357569e-01 -5.01460850e-01
-1.91439855e+00 -1.39430177e+00 5.83183885e-01 2.56695766e-02
4.51009661e-01 -5.22849500e-01 -1.52922213e-01 6.65192306e-01
1.87160328e-01 2.21227929e-01 4.30372477e-01 -6.39411390e-01
6.82399259e-04 -3.04751784e-01 -1.15053713e+00 5.01755357e-01
1.03926325e+00 1.92304123e-02 -7.39504918e-02 9.45521221e-02
4.88061845e-01 -1.16479874e+00 -9.27858710e-01 1.20328717e-01
7.76287913e-01 -1.21082473e+00 7.91906416e-01 -4.15540814e-01
7.26392746e-01 -3.38757515e-01 -1.88103423e-01 -7.87004471e-01
-1.91155262e-02 -4.76087630e-01 -1.30734578e-01 8.39179099e-01
3.47636789e-01 -2.11050749e-01 7.86545694e-01 1.55848175e-01
-7.04725504e-01 -7.46354520e-01 -9.14472342e-01 -7.95536995e-01
-4.77438092e-01 -8.53843451e-01 8.62852216e-01 3.16941947e-01
-2.72447586e-01 2.27775499e-01 1.26420259e-01 2.27840081e-01
2.19720662e-01 -3.51394325e-01 9.22668457e-01 -9.97865021e-01
-2.01586068e-01 -4.70116884e-01 -1.23982942e+00 -4.77195829e-01
-8.26474279e-02 -7.71690667e-01 -2.23378509e-01 -1.26427245e+00
-5.19033432e-01 -2.75158226e-01 3.21300536e-01 7.96204746e-01
7.84969151e-01 2.71547288e-01 3.19900423e-01 -3.76798242e-01
-1.39825344e-01 -1.82740331e-01 7.45940924e-01 8.82086810e-03
-3.59771520e-01 -2.37856507e-01 -4.30707246e-01 6.53698504e-01
7.62217343e-01 -4.11443949e-01 -4.50714827e-01 -5.60320556e-01
9.77103263e-02 -5.63226268e-02 4.43525553e-01 -1.89720047e+00
2.62799293e-01 3.62879813e-01 6.14080012e-01 -6.03917301e-01
2.24844903e-01 -1.13136590e+00 5.60775399e-01 9.17979717e-01
-2.87197411e-01 2.90353805e-01 8.54407549e-01 -4.54621762e-02
-6.44502640e-02 -3.90461147e-01 7.92164505e-01 2.85385773e-02
-9.88830447e-01 1.11093387e-01 -6.46060050e-01 -3.80410314e-01
1.31645358e+00 -6.55920804e-01 -1.24438293e-01 2.63125986e-01
-5.05352080e-01 -3.28903466e-01 2.67100513e-01 3.45752180e-01
4.86458987e-01 -1.18744946e+00 -4.88108397e-01 6.54962301e-01
-2.19170645e-01 -6.95898905e-02 2.89887041e-01 3.00233006e-01
-1.46024275e+00 8.37003469e-01 -1.15341341e+00 -6.67582989e-01
-1.42839909e+00 3.59334886e-01 5.39128602e-01 -1.30000919e-01
-6.79906428e-01 4.23102051e-01 -5.72626173e-01 -1.65308326e-01
3.32308233e-01 -8.06813180e-01 -5.15126698e-02 -3.85971777e-02
7.12677836e-01 3.99230063e-01 7.56723762e-01 -2.77543724e-01
-4.87596422e-01 3.00868332e-01 2.15793848e-01 -4.58021574e-02
1.48163974e+00 7.47925937e-01 2.54779607e-01 -2.11865202e-01
1.14221680e+00 -7.04524219e-01 -1.39371300e+00 5.37403584e-01
-1.69963226e-01 -2.01283917e-01 4.08389539e-01 -6.67180061e-01
-1.24746895e+00 8.07604551e-01 8.92413735e-01 7.77062625e-02
1.33286107e+00 -4.90383357e-01 5.78356743e-01 2.78453290e-01
4.95949268e-01 -9.73722696e-01 -1.65781170e-01 7.37635851e-01
5.85953891e-01 -2.37066910e-01 1.40173942e-01 -3.20006549e-01
-1.12689026e-01 1.90664709e+00 5.31947374e-01 -4.15669531e-01
7.29808271e-01 9.52848673e-01 1.46086533e-02 -2.80778170e-01
-7.44214892e-01 2.94010453e-02 -1.73326194e-01 8.32247198e-01
4.72430587e-02 -1.40065670e-01 -2.10351512e-01 5.94915271e-01
-4.77060556e-01 5.96529424e-01 7.44886816e-01 1.42198408e+00
-8.50695223e-02 -1.06100667e+00 -1.98396295e-01 2.56018490e-01
-2.88937449e-01 1.83826357e-01 -2.57765025e-01 1.37877452e+00
3.67555737e-01 4.39343274e-01 3.32722694e-01 -6.40759945e-01
9.35009539e-01 -2.20438346e-01 6.42670214e-01 -5.26090145e-01
-1.41009319e+00 -1.45118847e-01 4.51217592e-01 -8.53311956e-01
-3.93833905e-01 -6.28428340e-01 -1.27928710e+00 -1.89689919e-01
9.85347293e-03 -5.80717444e-01 1.25001609e+00 7.49563456e-01
7.08166063e-01 1.00475287e+00 1.50868863e-01 -1.28319037e+00
-3.90261114e-01 -4.52618301e-01 -5.84101140e-01 -5.20821631e-01
-8.93646404e-02 -4.58891064e-01 1.01322033e-01 2.34659404e-01]
|
[8.308221817016602, 2.739593029022217]
|
d5219285-9e77-48f9-9510-978198b618c9
|
multimodal-emotion-recognition-among-couples
|
2212.13917
| null |
https://arxiv.org/abs/2212.13917v1
|
https://arxiv.org/pdf/2212.13917v1.pdf
|
Multimodal Emotion Recognition among Couples from Lab Settings to Daily Life using Smartwatches
|
Couples generally manage chronic diseases together and the management takes an emotional toll on both patients and their romantic partners. Consequently, recognizing the emotions of each partner in daily life could provide an insight into their emotional well-being in chronic disease management. The emotions of partners are currently inferred in the lab and daily life using self-reports which are not practical for continuous emotion assessment or observer reports which are manual, time-intensive, and costly. Currently, there exists no comprehensive overview of works on emotion recognition among couples. Furthermore, approaches for emotion recognition among couples have (1) focused on English-speaking couples in the U.S., (2) used data collected from the lab, and (3) performed recognition using observer ratings rather than partner's self-reported / subjective emotions. In this body of work contained in this thesis (8 papers - 5 published and 3 currently under review in various journals), we fill the current literature gap on couples' emotion recognition, develop emotion recognition systems using 161 hours of data from a total of 1,051 individuals, and make contributions towards taking couples' emotion recognition from the lab which is the status quo, to daily life. This thesis contributes toward building automated emotion recognition systems that would eventually enable partners to monitor their emotions in daily life and enable the delivery of interventions to improve their emotional well-being.
|
['George Boateng']
|
2022-12-21
| null | null | null | null |
['multimodal-emotion-recognition', 'multimodal-emotion-recognition']
|
['computer-vision', 'speech']
|
[-1.32793322e-01 2.68895179e-01 -5.94974995e-01 -6.86843693e-01
-1.41715825e-01 -2.95007795e-01 -2.53459722e-01 5.04479647e-01
1.61041003e-02 8.53604138e-01 8.28900710e-02 3.38062167e-01
1.75216019e-01 -7.04300940e-01 5.58244050e-01 -4.70562398e-01
-2.70856827e-01 2.15081125e-01 -1.05765331e+00 -3.39884847e-01
1.56607449e-01 6.97031081e-01 -1.36831486e+00 1.16161339e-01
7.78121531e-01 9.69946325e-01 -6.25114501e-01 8.97026420e-01
4.78403196e-02 5.14680386e-01 -5.49075186e-01 -7.39716113e-01
-3.66828889e-01 -5.93659997e-01 -5.59454322e-01 4.21734154e-01
-8.71863663e-02 -2.67219812e-01 2.03096196e-01 5.52223206e-01
8.12569082e-01 -1.65610798e-02 2.94927716e-01 -1.48322093e+00
-4.54508573e-01 -7.96107426e-02 -1.40783653e-01 2.44369712e-02
1.02777958e+00 -1.64328851e-02 5.90961337e-01 -6.28148794e-01
6.63892150e-01 1.10531211e+00 8.44709754e-01 8.71206582e-01
-1.18937516e+00 -8.51558566e-01 -3.95216823e-01 2.64251176e-02
-1.09619164e+00 -9.39506233e-01 8.69119465e-01 -1.58360690e-01
1.42227936e+00 4.91537899e-01 1.47554493e+00 1.15068793e+00
2.72622913e-01 4.46343720e-01 1.27329087e+00 -2.05952555e-01
2.87673295e-01 7.81146765e-01 -2.64419541e-02 3.17082882e-01
2.32692182e-01 -1.30267352e-01 -6.75428510e-01 -5.23972690e-01
6.12029612e-01 -1.01858296e-01 -5.24737835e-02 7.45465085e-02
-8.81354213e-01 6.42263472e-01 -1.94519982e-01 8.23481441e-01
-6.70809567e-01 -3.02290469e-01 6.04651690e-01 6.44549966e-01
6.96904182e-01 2.59531528e-01 -3.65952015e-01 -1.10230672e+00
-6.07017219e-01 -1.84422702e-01 1.37782097e+00 4.64681417e-01
5.87811530e-01 -9.48632658e-02 1.13493718e-01 1.13455236e+00
3.76335800e-01 4.15609926e-01 3.46506298e-01 -1.10525584e+00
-8.19065347e-02 8.86536181e-01 2.14494467e-01 -1.59554255e+00
-5.37350774e-01 -9.42578837e-02 -1.01092732e+00 -6.02754056e-02
4.78129126e-02 -5.17128766e-01 -1.90372273e-01 1.46603632e+00
3.65079880e-01 -1.44097298e-01 6.00818634e-01 7.14686871e-01
1.08385777e+00 3.07038367e-01 1.97911814e-01 -1.05949032e+00
1.34466195e+00 -4.07932490e-01 -1.39101899e+00 -4.17287856e-01
7.54242718e-01 -6.61885858e-01 6.37227416e-01 6.22260571e-01
-1.48804843e+00 -2.37180784e-01 -9.02413249e-01 5.82810268e-02
-5.37019014e-01 8.81433859e-02 1.46296799e+00 1.45194328e+00
-1.30555511e+00 6.17514372e-01 -8.28217685e-01 -1.20789647e+00
3.20004642e-01 7.56484210e-01 -7.68616140e-01 2.74395853e-01
-1.05435109e+00 1.04989874e+00 -3.18185985e-01 3.35997820e-01
2.43611023e-01 -3.63945037e-01 -9.73084152e-01 -1.17317937e-01
-3.73463660e-01 -5.31542957e-01 8.97718668e-01 -1.30764520e+00
-1.82146788e+00 1.58506715e+00 -4.33634192e-01 4.99284387e-01
-6.00766651e-02 1.37871996e-01 -7.32081413e-01 2.57414490e-01
6.50726780e-02 6.30547643e-01 -3.57428491e-02 -7.84533799e-01
-2.44414836e-01 -6.80034995e-01 -3.79461646e-01 4.76092160e-01
-6.24346316e-01 4.48617548e-01 -9.96178314e-02 -3.65065366e-01
1.26820594e-01 -7.97591031e-01 -1.59937397e-01 1.43574402e-01
2.87214130e-01 -3.39524597e-01 3.56335491e-01 -6.36353552e-01
1.25614560e+00 -2.12375283e+00 -2.14190349e-01 1.69399157e-01
-2.09041372e-01 2.12903321e-01 1.39373913e-01 8.26874912e-01
-4.07144487e-01 2.89614826e-01 3.71313512e-01 -5.45549631e-01
1.05089940e-01 5.60346067e-01 2.90923864e-01 5.23500323e-01
1.76950082e-01 8.39510024e-01 -9.53828573e-01 -5.53262293e-01
2.81212479e-01 6.58462167e-01 -1.37282163e-01 6.13746822e-01
8.82501125e-01 2.34158427e-01 -1.73855364e-01 1.17357516e+00
6.56653643e-01 1.97018176e-01 5.76468289e-01 -6.97398782e-02
1.33412153e-01 4.84616980e-02 -1.02128696e+00 1.23004770e+00
-3.04537416e-01 4.07235026e-01 5.66192150e-01 -1.17148590e+00
1.48573124e+00 9.43187654e-01 8.31078827e-01 -5.24563193e-01
3.32407773e-01 2.12346733e-01 -2.65164196e-01 -8.75345290e-01
2.22155780e-01 -6.83899820e-01 -2.41161704e-01 3.69982660e-01
-1.32976666e-01 -1.75832406e-01 2.61504233e-01 -3.13760132e-01
8.97829950e-01 -4.41271037e-01 7.83095837e-01 1.47292316e-01
5.47224164e-01 -2.72710145e-01 9.32674944e-01 2.27760702e-01
-8.89880657e-01 1.28192365e-01 7.25298285e-01 -4.75533038e-01
-2.35655546e-01 -7.62054980e-01 -1.09895773e-01 8.52173388e-01
-2.78945297e-01 -2.23300308e-01 -1.62275523e-01 -1.98172107e-01
6.35461882e-03 4.39637959e-01 -4.88750041e-01 -2.99694270e-01
9.25684497e-02 -6.64600730e-01 5.56039155e-01 5.46874881e-01
2.62309194e-01 -1.33730161e+00 -4.64191347e-01 3.69608432e-01
-3.62390876e-01 -7.57472992e-01 1.01614304e-01 2.99947802e-02
-1.19525135e+00 -6.92643046e-01 -4.01807308e-01 -6.96401417e-01
6.60871804e-01 -2.60558039e-01 1.34639907e+00 7.66969472e-02
-4.77176815e-01 8.83484125e-01 -1.99330911e-01 -3.25927079e-01
9.57166404e-02 -3.48615646e-01 3.80015314e-01 -2.56253958e-01
8.23074937e-01 -1.09110415e+00 -5.86917162e-01 2.31265277e-01
-4.15178031e-01 -4.23497498e-01 3.31122398e-01 5.11922479e-01
8.83104652e-02 -7.37905726e-02 9.00691152e-01 -6.35473192e-01
7.82300174e-01 -7.15179443e-01 4.68110234e-01 1.40661553e-01
-6.78238451e-01 -9.93675649e-01 -7.44980872e-02 -4.65204418e-01
-1.07102621e+00 -2.79025167e-01 -7.73677379e-02 -4.16108333e-02
-5.65492749e-01 3.77556682e-01 -1.01956971e-01 -1.98838226e-02
1.46558985e-01 -5.20454586e-01 1.85775846e-01 -5.15971370e-02
-3.04113746e-01 1.25549495e+00 4.83497798e-01 -3.71482641e-01
-3.10723688e-02 3.15889955e-01 -4.25689295e-02 -7.76405632e-01
-3.09971452e-01 -6.10925615e-01 -4.65064764e-01 -6.40326440e-01
5.71931064e-01 -9.27857339e-01 -1.50910091e+00 5.27274370e-01
-8.50042045e-01 -7.03251585e-02 1.21888988e-01 6.80583119e-01
-4.81809616e-01 1.38978824e-01 -1.00478172e+00 -1.41422999e+00
-7.84213960e-01 -4.07254130e-01 8.98478627e-01 8.73892367e-01
-1.30970097e+00 -1.42880309e+00 2.12623268e-01 8.68658543e-01
2.64505535e-01 6.48716748e-01 3.13401878e-01 -2.00537682e-01
5.78960896e-01 -5.56266725e-01 -3.14087458e-02 8.11819509e-02
7.04187572e-01 1.79574400e-01 -8.33308518e-01 -2.76743978e-01
1.55643031e-01 -8.24193835e-01 -2.97059417e-01 1.05198264e-01
3.08631182e-01 -1.36099979e-02 -3.90075266e-01 1.81160659e-01
1.04884744e+00 6.21257007e-01 1.04238462e+00 -2.02063560e-01
-1.91556325e-03 1.13571370e+00 9.78655756e-01 8.36516380e-01
6.61494136e-01 2.14795515e-01 -1.07702557e-02 -2.84681231e-01
6.15572214e-01 4.79523927e-01 6.33174598e-01 8.16689014e-01
-2.48621702e-01 -1.66896090e-01 -6.11469626e-01 6.48916066e-01
-1.66571128e+00 -8.18742394e-01 -2.40130678e-01 1.88467026e+00
9.37199473e-01 -4.39101696e-01 4.74808812e-02 3.04570824e-01
5.87027311e-01 -3.79630655e-01 -4.45086360e-01 -1.48635995e+00
1.29501879e-01 4.08747107e-01 -3.29885572e-01 2.94990480e-01
-7.44462192e-01 4.96197671e-01 6.92185974e+00 5.85896103e-03
-1.19680274e+00 -3.70179027e-01 1.10035229e+00 -1.81699410e-01
1.07155360e-01 -9.72590744e-02 -2.33114347e-01 1.09606445e-01
1.02675736e+00 -1.46509513e-01 4.18024093e-01 6.87677443e-01
5.07818878e-01 -6.34369910e-01 -1.29075897e+00 1.56048179e+00
3.37461382e-01 -2.85837442e-01 -9.76885259e-01 8.19925517e-02
3.79562229e-01 -8.00677538e-01 -2.06144720e-01 4.18322235e-01
-3.23147118e-01 -1.17616642e+00 -4.27677363e-01 1.00349951e+00
5.67922354e-01 -9.33884442e-01 1.13039339e+00 -1.11826919e-02
-8.53147864e-01 2.50723928e-01 3.93699929e-02 -7.83076644e-01
4.51761298e-02 8.28190148e-01 -9.13324118e-01 2.78657913e-01
9.02538061e-01 9.90524948e-01 -1.01437755e-01 3.41227859e-01
1.93510011e-01 4.49916303e-01 -2.97640055e-01 -2.29191691e-01
-3.66062969e-01 -5.97847164e-01 1.78076029e-01 1.18334424e+00
4.44623023e-01 6.53054416e-01 -2.03151584e-01 5.59403896e-01
3.12984943e-01 4.44344640e-01 -6.06370270e-01 -8.85885954e-01
2.20920756e-01 1.76178801e+00 -8.66391242e-01 -2.78291821e-01
-3.25732023e-01 1.24734247e+00 1.69717416e-01 -6.67191595e-02
-1.27798140e-01 -3.63387138e-01 1.13025415e+00 -3.00340265e-01
-4.07292008e-01 2.85932332e-01 -5.74399412e-01 -9.49472666e-01
1.65443569e-02 -8.70553434e-01 5.39087236e-01 -9.54831958e-01
-1.62386322e+00 -2.11903471e-02 -4.29890364e-01 -8.85842443e-01
-4.62519705e-01 -2.77076930e-01 -7.71759152e-01 7.87377298e-01
-7.36381173e-01 -7.94405997e-01 -5.33876181e-01 1.66679516e-01
-4.04325761e-02 3.21272731e-01 1.42541277e+00 4.50075030e-01
-9.55488443e-01 5.62506855e-01 -5.36683977e-01 -1.11183986e-01
1.15598822e+00 -1.13739312e+00 -6.15997076e-01 -2.44132578e-01
-8.39700520e-01 7.26308346e-01 5.37586391e-01 -8.77745032e-01
-1.65165436e+00 -1.91387266e-01 1.68410468e+00 -7.17504770e-02
2.44949669e-01 -1.92717746e-01 -6.96086347e-01 4.25656617e-01
4.09490317e-01 -1.65080056e-01 1.64437175e+00 5.79494059e-01
5.21077037e-01 -5.26842237e-01 -1.89138532e+00 4.82183754e-01
8.48458707e-01 -4.67542171e-01 -2.97565609e-01 8.43639225e-02
-3.80606353e-01 -3.45139831e-01 -1.84087133e+00 4.59922254e-01
1.05605578e+00 -1.23214948e+00 7.53823042e-01 -4.16392237e-01
2.22115055e-01 3.63758236e-01 1.82961017e-01 -1.16799557e+00
-1.22953445e-01 -7.11340964e-01 1.95966676e-01 1.83071661e+00
-1.45780489e-01 -8.23771596e-01 8.95343661e-01 1.56221712e+00
4.26138818e-01 -1.03924012e+00 -7.25304008e-01 5.57674281e-02
-3.47908735e-01 -4.41414356e-01 2.90849924e-01 1.46016622e+00
1.20166337e+00 1.59426272e-01 -2.53212929e-01 -3.21688086e-01
1.26705661e-01 -4.81378920e-02 7.01309621e-01 -1.20913029e+00
2.34494671e-01 -2.75383264e-01 -8.53451610e-01 4.02046181e-02
-1.13516599e-02 -3.66325498e-01 -3.67931366e-01 -1.57274234e+00
1.54619426e-01 -2.58149475e-01 -5.67114837e-02 7.54491985e-01
-6.80409968e-02 3.88252348e-01 -3.51843327e-01 -4.00489777e-01
-3.42742085e-01 2.95379549e-01 9.10978317e-01 3.30033988e-01
-5.26478648e-01 -7.09525347e-02 -1.05586815e+00 5.09313703e-01
9.52035248e-01 -7.17120916e-02 -3.26072395e-01 4.84618247e-01
6.14521921e-01 7.59573519e-01 -4.33169976e-02 -7.27057338e-01
3.09974756e-02 -4.56298143e-01 6.98447704e-01 -5.18038690e-01
6.95967793e-01 -8.05237889e-01 6.08951986e-01 3.85784835e-01
1.57404080e-01 -1.11895621e-01 -3.14804614e-02 -1.96947172e-01
-1.81850418e-01 -4.61570360e-02 5.74330926e-01 -1.99625447e-01
-9.08876806e-02 -1.95087940e-01 -8.27826977e-01 -3.95842046e-01
1.26076424e+00 -6.32691503e-01 6.96798563e-02 -8.88443291e-01
-1.21821535e+00 5.06565213e-01 6.97524846e-01 1.99833080e-01
5.07587135e-01 -1.52826715e+00 -3.36028665e-01 2.14463472e-01
-5.47468513e-02 -5.64340472e-01 2.87519515e-01 1.26868188e+00
-2.83031702e-01 7.92103559e-02 -5.07153034e-01 -2.76114345e-01
-1.72377396e+00 1.34614065e-01 2.53866166e-01 -2.73607135e-01
7.14115202e-02 5.31678736e-01 -5.39214075e-01 -4.99315023e-01
2.00754493e-01 -6.59029782e-02 -5.56719601e-01 7.01474905e-01
4.12228703e-01 7.57717133e-01 -5.61439991e-02 -7.28209317e-01
-5.06868124e-01 4.07978922e-01 3.82898986e-01 -1.71036099e-03
1.41022015e+00 -5.56590855e-01 -7.92940617e-01 8.81528735e-01
1.24207056e+00 -4.93883826e-02 -8.24558362e-02 4.76089984e-01
-2.22679943e-01 -5.20365655e-01 -1.66076347e-01 -7.31073558e-01
-1.06491601e+00 4.98422951e-01 7.22930372e-01 3.05306554e-01
1.63236213e+00 -1.78136721e-01 8.86901438e-01 2.92925894e-01
8.79109949e-02 -1.57384431e+00 9.79949981e-02 4.28542122e-02
7.73225784e-01 -1.20115793e+00 4.60549183e-02 -6.93040550e-01
-9.24872816e-01 1.36665332e+00 7.02781022e-01 2.22797200e-01
6.14775717e-01 3.40839654e-01 5.03898025e-01 -2.05738321e-01
-9.80567634e-01 4.27430272e-02 -1.50903359e-01 7.99254119e-01
1.10375714e+00 2.32735813e-01 -7.62367845e-01 9.97664392e-01
-2.08660707e-01 5.63243330e-01 2.93585628e-01 1.08050346e+00
2.52445806e-02 -1.49366772e+00 -5.38393915e-01 6.68185353e-01
-5.15951097e-01 5.69440305e-01 -9.27504003e-01 2.21111804e-01
5.82929671e-01 1.65489924e+00 2.35259175e-01 -1.44400552e-01
4.08915192e-01 3.36778313e-01 4.01669025e-01 -3.43701094e-01
-9.07918870e-01 -2.02663913e-02 8.55490744e-01 -7.40150154e-01
-1.00281632e+00 -1.02273512e+00 -1.09595215e+00 -5.69863737e-01
1.26074135e-01 8.59512761e-02 5.40716469e-01 7.59176910e-01
4.29340512e-01 -4.90243733e-02 9.48292971e-01 -9.11395490e-01
4.84974980e-01 -1.07938468e+00 -1.03997886e+00 4.09720957e-01
-1.08489603e-01 -3.96674931e-01 -4.13584292e-01 -1.36079401e-01]
|
[13.514711380004883, 2.7606632709503174]
|
0164d459-cb59-4dae-9977-5980745c3d92
|
towards-zero-shot-code-switched-speech
|
2211.01458
| null |
https://arxiv.org/abs/2211.01458v2
|
https://arxiv.org/pdf/2211.01458v2.pdf
|
Towards Zero-Shot Code-Switched Speech Recognition
|
In this work, we seek to build effective code-switched (CS) automatic speech recognition systems (ASR) under the zero-shot setting where no transcribed CS speech data is available for training. Previously proposed frameworks which conditionally factorize the bilingual task into its constituent monolingual parts are a promising starting point for leveraging monolingual data efficiently. However, these methods require the monolingual modules to perform language segmentation. That is, each monolingual module has to simultaneously detect CS points and transcribe speech segments of one language while ignoring those of other languages -- not a trivial task. We propose to simplify each monolingual module by allowing them to transcribe all speech segments indiscriminately with a monolingual script (i.e. transliteration). This simple modification passes the responsibility of CS point detection to subsequent bilingual modules which determine the final output by considering multiple monolingual transliterations along with external language model information. We apply this transliteration-based approach in an end-to-end differentiable neural network and demonstrate its efficacy for zero-shot CS ASR on Mandarin-English SEAME test sets.
|
['Shinji Watanabe', 'Preethi Jyothi', 'Ondrej Klejch', 'Matthew Wiesner', 'Brian Yan']
|
2022-11-02
| null | null | null | null |
['transliteration']
|
['natural-language-processing']
|
[ 2.55269378e-01 4.27480005e-02 -1.19844615e-01 -4.04920757e-01
-1.34370279e+00 -8.89296710e-01 5.93028307e-01 -2.81041861e-01
-7.82346666e-01 3.11977088e-01 -2.69645780e-01 -1.02306736e+00
5.61881840e-01 -2.70869642e-01 -7.47665644e-01 -5.56251943e-01
4.17255372e-01 6.93265557e-01 -5.70645630e-02 -4.10683662e-01
-1.75824106e-01 5.00179768e-01 -8.74546468e-01 1.96162015e-01
1.25215340e+00 3.50776345e-01 5.61365962e-01 8.40025544e-01
-9.73898545e-02 6.94781780e-01 -5.65299928e-01 -4.18010354e-01
2.64107525e-01 -5.87311447e-01 -7.85839736e-01 3.40904325e-01
4.77284759e-01 -1.76705457e-02 -3.58018816e-01 1.13328803e+00
3.74030739e-01 2.07292587e-01 4.68620718e-01 -6.30041182e-01
-6.19083822e-01 1.12891901e+00 -3.45105797e-01 1.76275834e-01
1.01053707e-01 2.22108990e-01 9.98606980e-01 -1.45595467e+00
7.01576173e-01 1.11175811e+00 2.15458944e-01 7.21839845e-01
-1.30672324e+00 -5.32885969e-01 3.34682733e-01 2.24778000e-02
-1.41993010e+00 -1.15572619e+00 6.66184127e-01 -4.26644027e-01
1.28044677e+00 3.44616354e-01 3.09831411e-01 1.10660076e+00
-3.96226048e-01 1.00206125e+00 1.01155686e+00 -7.09805965e-01
1.65354475e-01 3.95626843e-01 3.20395708e-01 6.48648560e-01
-3.24158192e-01 -1.17849864e-01 -3.98209810e-01 3.73171002e-01
2.47615784e-01 -3.70995849e-01 -3.65213066e-01 -1.08832158e-01
-1.26288271e+00 6.63008809e-01 8.80381539e-02 6.56617939e-01
-1.46896809e-01 -3.54095399e-01 3.76341730e-01 5.88987410e-01
5.61966419e-01 1.79483563e-01 -4.34385270e-01 -1.75134279e-02
-1.46727884e+00 -4.99603301e-01 7.22565234e-01 1.25181127e+00
8.57656181e-01 5.06512940e-01 -1.88713476e-01 1.41992903e+00
8.89817476e-02 7.14370012e-01 6.90188885e-01 -3.53118628e-01
7.52839148e-01 2.76818246e-01 -3.71519804e-01 -1.30222008e-01
1.25701919e-01 -6.02888465e-01 -6.58117771e-01 -1.29960299e-01
1.85418487e-01 -3.28809857e-01 -1.23737752e+00 1.58097875e+00
1.64796293e-01 -4.17337306e-02 3.57834190e-01 8.13799381e-01
3.55818778e-01 9.48559463e-01 -4.50016931e-02 -1.78718567e-01
1.16295338e+00 -1.41136849e+00 -5.20601034e-01 -4.25621510e-01
7.50827312e-01 -9.61276650e-01 1.32129800e+00 2.40105078e-01
-1.21796286e+00 -4.94384766e-01 -1.01602185e+00 -2.22785920e-01
-3.99775386e-01 7.49980390e-01 -5.82620800e-02 5.14632404e-01
-1.18677843e+00 2.03620121e-01 -9.90325749e-01 -4.02673870e-01
2.32562963e-02 2.06267342e-01 -2.81397730e-01 -1.73389331e-01
-1.13977623e+00 9.63596404e-01 3.63778740e-01 1.71016291e-01
-1.19687974e+00 -2.88658619e-01 -9.36987460e-01 8.42725486e-02
5.90429425e-01 -1.29749328e-01 1.57636738e+00 -1.24598837e+00
-1.79263210e+00 1.08587420e+00 -4.77647245e-01 -3.98847491e-01
7.59707093e-01 1.02988034e-01 -3.92649442e-01 -5.03449002e-03
-3.92791815e-02 4.48585600e-01 9.39249575e-01 -1.08891165e+00
-7.49385417e-01 -3.76337707e-01 -1.82306901e-01 3.46863091e-01
-1.54038042e-01 4.43918020e-01 -8.19278359e-01 -7.51473606e-01
1.52339533e-01 -1.09088731e+00 -5.85314371e-02 -6.05577767e-01
-6.24639511e-01 -2.29647547e-01 6.57061160e-01 -1.02937901e+00
1.01574719e+00 -2.16035795e+00 5.13515115e-01 1.72066152e-01
-1.59797981e-01 6.06844783e-01 -2.42244095e-01 2.16968432e-01
4.46871184e-02 -1.57640621e-01 -5.53764105e-01 -8.76413882e-01
-9.63166058e-02 2.20894992e-01 -3.24783385e-01 5.55677354e-01
4.80236381e-01 8.84306908e-01 -8.16087365e-01 -3.24888557e-01
2.21072450e-01 4.75560874e-01 -4.37518567e-01 2.18303844e-01
-1.53866783e-01 6.53594196e-01 1.65576965e-01 6.67136014e-01
4.54620570e-01 2.17702553e-01 3.18779647e-01 2.77041912e-01
-4.33185667e-01 7.02215910e-01 -8.56668949e-01 1.78926539e+00
-8.82402182e-01 7.36983299e-01 5.52167892e-01 -1.06784022e+00
8.70567918e-01 5.41831255e-01 2.21469030e-02 -4.55584586e-01
8.81522819e-02 7.62984812e-01 1.90187208e-02 -1.57394379e-01
3.80358547e-01 -2.50402659e-01 -1.23685025e-01 3.38440895e-01
4.96037036e-01 -5.64502226e-03 2.04686597e-01 1.99951783e-01
8.10275197e-01 2.21731160e-02 1.12446390e-01 -2.33847812e-01
7.58303165e-01 1.66944191e-01 5.72564960e-01 6.71777248e-01
-2.16836140e-01 5.18567562e-01 -5.99020682e-02 1.21178091e-01
-1.15259743e+00 -1.16047549e+00 1.63542002e-01 1.44045293e+00
-2.74654657e-01 -6.44955859e-02 -1.11561799e+00 -7.30174899e-01
-4.51319277e-01 9.66002285e-01 -6.01587743e-02 1.95557512e-02
-9.61487234e-01 -3.34555924e-01 7.61033535e-01 8.38697031e-02
1.10217415e-01 -8.99947703e-01 -1.09826699e-01 3.39356869e-01
-3.27135891e-01 -1.21519423e+00 -8.98010433e-01 5.70582271e-01
-3.81967098e-01 -5.52248359e-01 -1.14550173e+00 -1.16956162e+00
6.86219335e-01 3.51889104e-01 7.74442077e-01 -2.29615778e-01
1.47989311e-03 1.02114983e-01 -3.00557464e-01 1.78437769e-01
-9.77584958e-01 4.87090677e-01 2.01939300e-01 4.24345195e-01
3.81128043e-01 -2.91474730e-01 -9.57431570e-02 1.69543460e-01
-7.22317517e-01 1.12684920e-01 5.91876328e-01 7.12056220e-01
6.12056673e-01 -5.08853197e-01 4.96700078e-01 -9.94137168e-01
3.97901267e-01 -4.42229748e-01 -8.31801355e-01 6.66654110e-01
-2.32114151e-01 7.00322166e-02 9.10848856e-01 -6.39501929e-01
-1.07135975e+00 3.76504391e-01 -3.12854081e-01 -7.10337877e-01
-2.29908779e-01 4.38881785e-01 -4.27747041e-01 1.19926870e-01
5.36668181e-01 7.10862994e-01 -3.18614006e-01 -6.72731042e-01
6.09691978e-01 1.30477619e+00 8.36968422e-01 -5.95541656e-01
7.12603629e-01 -2.60527022e-02 -8.65749598e-01 -1.25882983e+00
-4.14363682e-01 -6.86894536e-01 -9.94808316e-01 -1.81102946e-01
8.50241005e-01 -1.11484015e+00 -2.11454004e-01 4.84449595e-01
-1.56644237e+00 -3.98295939e-01 -4.83913757e-02 4.30934668e-01
-3.16999495e-01 2.34455869e-01 -6.78683341e-01 -8.74601007e-01
-2.74155468e-01 -1.68421066e+00 9.83471930e-01 -2.12145492e-01
5.70555544e-03 -6.86729074e-01 4.37359046e-03 4.29863334e-01
3.86511236e-01 -6.64904654e-01 7.45890856e-01 -9.37339962e-01
-5.44798255e-01 -3.80857438e-02 1.05048478e-01 6.87073529e-01
2.80394144e-02 8.69485289e-02 -1.23158014e+00 -5.71345985e-01
1.21630520e-01 -1.65415794e-01 7.61021137e-01 -1.50889519e-03
5.92945158e-01 -3.93123567e-01 3.62641178e-03 7.76620209e-01
1.02127683e+00 4.03325289e-01 1.43323680e-02 -1.37525126e-01
8.60932052e-01 7.27599561e-01 1.62575856e-01 -1.75629124e-01
3.86420220e-01 6.63583219e-01 -2.70960987e-01 -2.57934034e-01
-4.25073683e-01 -2.82865047e-01 9.36583936e-01 1.72540569e+00
4.13528919e-01 -2.00775012e-01 -1.22310364e+00 7.56466925e-01
-1.46867228e+00 -5.78286052e-01 3.89229283e-02 2.23350549e+00
9.69644845e-01 -8.92654285e-02 -3.77837308e-02 -1.34283647e-01
9.66003180e-01 9.05348435e-02 -5.03903985e-01 -4.56904829e-01
-2.95395434e-01 2.36707479e-01 4.69911307e-01 9.10542905e-01
-9.49970424e-01 1.66952491e+00 4.82953262e+00 9.73216534e-01
-1.52341127e+00 5.77779353e-01 5.44635415e-01 4.23744740e-03
-3.81122589e-01 2.75258511e-01 -9.60228980e-01 3.56078029e-01
1.26691771e+00 -5.44139929e-02 9.42937255e-01 7.60333002e-01
2.83715665e-01 1.17155321e-01 -1.26234984e+00 8.31401646e-01
9.18744951e-02 -9.34217751e-01 4.79094358e-03 -2.47627869e-01
7.46258676e-01 5.91992438e-01 -8.02841038e-02 5.95125079e-01
4.87324387e-01 -8.13633025e-01 1.02041733e+00 9.45194736e-02
1.16440487e+00 -6.15048587e-01 1.18594691e-01 5.99013329e-01
-1.13355279e+00 1.92605689e-01 -1.38082147e-01 4.12703425e-01
2.48824343e-01 3.17536443e-01 -1.11957419e+00 5.28347790e-01
1.82109401e-01 3.84805024e-01 -3.39767396e-01 6.08143389e-01
-3.59410286e-01 9.31463301e-01 -2.14684740e-01 2.29308680e-01
4.99858588e-01 -3.37134391e-01 8.00350547e-01 1.56132889e+00
4.01280284e-01 -4.00998145e-01 3.64462346e-01 8.26776922e-01
-2.73385942e-01 4.38556254e-01 -5.53685784e-01 -2.03088909e-01
5.07380486e-01 9.90788043e-01 -6.42353952e-01 -6.27688289e-01
-6.16476595e-01 1.52347434e+00 6.51551366e-01 6.61596119e-01
-4.12741363e-01 -5.60737133e-01 5.19195914e-01 -1.03035972e-01
1.86781257e-01 -6.58547282e-01 -6.29928559e-02 -1.56714129e+00
6.27942234e-02 -1.19115436e+00 -8.62064399e-03 -4.25960183e-01
-9.65841293e-01 1.10446274e+00 -4.11496639e-01 -1.05466604e+00
-4.07263994e-01 -5.12499630e-01 -3.88499886e-01 1.30813611e+00
-1.62971580e+00 -1.44044638e+00 3.72122139e-01 6.22608304e-01
1.24212790e+00 -5.00523388e-01 6.36855125e-01 6.53703213e-01
-8.79643977e-01 8.05592120e-01 2.34699413e-01 4.19427633e-01
6.62713468e-01 -1.12805533e+00 6.70907617e-01 1.44963586e+00
6.57515347e-01 7.46619344e-01 4.04613197e-01 -6.86822593e-01
-1.43712234e+00 -1.18027687e+00 1.34784615e+00 -9.50916260e-02
8.12823415e-01 -8.61318171e-01 -9.36083853e-01 9.56314743e-01
2.25504309e-01 -1.10866077e-01 3.46322030e-01 1.29851075e-02
-2.76556373e-01 -1.56485997e-02 -6.74925983e-01 8.22014570e-01
7.93480277e-01 -1.17570221e+00 -6.24454796e-01 1.93680301e-01
8.02485883e-01 -2.13374496e-01 -4.06212926e-01 3.60690989e-02
9.46575478e-02 -4.20366704e-01 4.87330973e-01 -4.09725398e-01
-6.76089972e-02 -3.05419981e-01 -2.53975809e-01 -1.53883433e+00
1.81657150e-01 -8.03443849e-01 3.09707850e-01 1.33297992e+00
7.68143952e-01 -6.46041095e-01 1.87461689e-01 2.35283986e-01
-5.73097050e-01 -2.49509767e-01 -1.21127605e+00 -9.66814458e-01
2.90942639e-01 -6.12703323e-01 3.20428461e-01 1.12703216e+00
-3.25250961e-02 6.65742040e-01 -2.48111039e-01 3.53723943e-01
4.32927668e-01 -1.07115597e-01 5.66895366e-01 -8.18475783e-01
-4.27514523e-01 -4.30728257e-01 2.39919841e-01 -1.20969534e+00
5.57566345e-01 -1.57999897e+00 5.83323598e-01 -1.02758217e+00
4.44055572e-02 -4.37178314e-01 -2.26197660e-01 4.95690733e-01
-5.53848669e-02 -7.76137114e-02 3.07171285e-01 3.74567837e-01
-3.65717888e-01 4.96110231e-01 8.93312216e-01 -3.76319945e-01
-2.68153638e-01 7.64702857e-02 -2.45507851e-01 3.99052948e-01
6.44155860e-01 -5.23153007e-01 -2.95017600e-01 -7.59473741e-01
-3.59553605e-01 2.72832841e-01 -3.83772552e-02 -8.99698675e-01
5.29200554e-01 -1.10894116e-02 -2.18661129e-01 -4.43238378e-01
1.84094056e-01 -6.30168736e-01 -1.14687242e-01 1.99203104e-01
-4.31431919e-01 -2.16465499e-02 2.30156034e-02 1.80731699e-01
-3.39977026e-01 -2.64991999e-01 1.01998389e+00 -2.27070689e-01
-5.25982559e-01 2.68177629e-01 -6.98566854e-01 -9.43388697e-03
8.57794702e-01 -8.78450554e-03 -8.94328812e-04 -3.75251696e-02
-8.50546539e-01 9.38033387e-02 4.38482612e-01 5.09796381e-01
3.31926614e-01 -9.32624400e-01 -8.60461950e-01 6.87219083e-01
7.73912147e-02 -1.31399095e-01 4.45013940e-02 8.89721274e-01
-2.74152219e-01 6.51006937e-01 2.52140582e-01 -6.90295041e-01
-1.18997169e+00 4.12229449e-01 5.43997228e-01 2.72500003e-03
-3.24174941e-01 1.04195333e+00 1.86091140e-01 -9.98026967e-01
3.77076179e-01 -3.65086377e-01 1.42639101e-01 1.30146109e-02
1.75491780e-01 6.19456880e-02 5.14925480e-01 -1.12180865e+00
-3.57419878e-01 3.01401079e-01 -3.18223774e-01 -6.62265837e-01
1.07838213e+00 -3.36959034e-01 -5.43773398e-02 7.78328300e-01
1.26099026e+00 2.80482382e-01 -1.07939243e+00 -5.35926163e-01
3.37898314e-01 3.36677954e-02 1.06064327e-01 -7.06874430e-01
-9.82897699e-01 1.30464840e+00 3.15899253e-01 -1.98984399e-01
7.56067812e-01 1.24649487e-01 8.25045645e-01 3.41736346e-01
4.67420429e-01 -1.44995284e+00 -4.06166345e-01 9.82802927e-01
6.55862808e-01 -1.36793554e+00 -8.99499536e-01 -7.31418803e-02
-7.68944860e-01 9.55663264e-01 3.96577060e-01 1.10519551e-01
4.45743471e-01 1.69262379e-01 2.90254533e-01 1.42644763e-01
-5.89299917e-01 -3.10254276e-01 3.52428406e-01 1.06714472e-01
5.01792073e-01 4.29513544e-01 -1.50349393e-01 4.81617630e-01
-2.52534121e-01 -3.64634633e-01 2.56354332e-01 8.56851935e-01
-4.43733603e-01 -1.20917189e+00 -3.15943211e-01 6.19877316e-03
-4.61816221e-01 -4.87335980e-01 -4.99789387e-01 2.74836302e-01
-3.19580808e-02 1.09312022e+00 9.80332419e-02 -2.74015933e-01
2.70779818e-01 5.29541373e-01 1.49345368e-01 -1.01936901e+00
-6.95884705e-01 4.87086415e-01 -1.51706403e-02 -2.84684807e-01
1.51078701e-01 -7.36670971e-01 -1.14329374e+00 -7.89867807e-03
-1.79162860e-01 4.11190540e-02 1.05550647e+00 1.09106445e+00
1.70217261e-01 5.71922123e-01 8.51765156e-01 -7.63617098e-01
-7.35514879e-01 -1.05525005e+00 -2.20571637e-01 4.81065214e-02
7.05362797e-01 -9.68763828e-02 -3.23112428e-01 2.35350892e-01]
|
[14.439079284667969, 7.063859939575195]
|
ad8f3d9d-07f2-482d-8cd7-ad258947209d
|
automatic-skin-lesion-segmentation-on
|
1808.06759
| null |
http://arxiv.org/abs/1808.06759v1
|
http://arxiv.org/pdf/1808.06759v1.pdf
|
Automatic skin lesion segmentation on dermoscopic images by the means of superpixel merging
|
We present a superpixel-based strategy for segmenting skin lesion on
dermoscopic images. The segmentation is carried out by over-segmenting the
original image using the SLIC algorithm, and then merge the resulting
superpixels into two regions: healthy skin and lesion. The mean RGB color of
each superpixel was used as merging criterion. The presented method is capable
of dealing with segmentation problems commonly found in dermoscopic images such
as hair removal, oil bubbles, changes in illumination, and reflections images
without any additional steps. The method was evaluated on the PH2 and ISIC 2017
dataset with results comparable to the state-of-art.
|
['Jonathan Avendaño', 'Diego Patiño', 'John Willian Branch']
|
2018-08-21
| null | null | null | null |
['skin-lesion-segmentation']
|
['medical']
|
[ 8.51551950e-01 1.97171614e-01 -6.01078011e-02 3.78377661e-02
-3.27971160e-01 -6.04580998e-01 1.95584953e-01 3.83912742e-01
-5.50560057e-01 5.41223526e-01 -4.61928725e-01 -1.59747422e-01
2.33622938e-01 -5.49922645e-01 -3.89894038e-01 -8.35441887e-01
1.18856877e-01 3.79936695e-02 8.89703512e-01 2.99716741e-01
4.21138495e-01 5.24945617e-01 -1.32472956e+00 3.20767969e-01
1.40683460e+00 8.31239402e-01 -1.06758513e-02 8.21353078e-01
-3.73469263e-01 3.82479697e-01 -5.53878367e-01 -4.63331521e-01
2.10484982e-01 -6.40692294e-01 -1.04780614e+00 8.02362502e-01
6.13903999e-01 -3.01468909e-01 4.01035219e-01 1.37009561e+00
8.33802894e-02 -3.55859041e-01 6.74533844e-01 -5.17268062e-01
-2.67158389e-01 -2.48670131e-02 -1.03382528e+00 -7.55703226e-02
3.96687776e-01 -5.46629960e-03 3.73814851e-01 -4.72232282e-01
1.08928549e+00 9.85430598e-01 3.99609864e-01 5.36996245e-01
-1.27959514e+00 -1.75491590e-02 -1.01662360e-01 3.92963856e-01
-1.13968301e+00 2.91317075e-01 3.52071136e-01 -4.59506601e-01
4.58353043e-01 7.95604885e-01 1.08884096e+00 4.69273001e-01
2.19905138e-01 7.88047016e-01 1.86682200e+00 -6.24476790e-01
3.78934652e-01 3.91964227e-01 2.64243484e-01 6.16757035e-01
4.71232653e-01 -2.85157681e-01 -4.39460315e-02 -8.00488219e-02
7.41097033e-01 -1.53409258e-01 -1.63501754e-01 8.77697244e-02
-4.77351934e-01 3.56255949e-01 3.70865911e-01 4.02993739e-01
-6.08006418e-01 -1.97085947e-01 1.55144736e-01 -1.14548452e-01
6.38579130e-01 2.47323290e-01 3.52240026e-01 1.54854894e-01
-1.39193749e+00 -1.00478046e-01 7.23431826e-01 3.79285812e-01
4.67924476e-01 -5.18927336e-01 -9.99152511e-02 6.55824959e-01
5.20557351e-02 4.83958662e-01 1.21946946e-01 -1.07304716e+00
-7.77329877e-02 8.11273754e-01 7.83016682e-02 -6.23235762e-01
-1.19340792e-01 1.59661844e-01 -4.15662497e-01 4.73683864e-01
6.91329122e-01 9.07947123e-02 -1.75732183e+00 6.14166141e-01
7.62422800e-01 7.78273540e-03 -2.85272926e-01 8.21224749e-01
6.12571299e-01 2.83001393e-01 1.90599576e-01 -4.37316895e-01
1.24443614e+00 -8.58837545e-01 -9.66215730e-01 1.87927753e-01
5.39663211e-02 -1.07337677e+00 5.15825033e-01 9.34761345e-01
-1.41804552e+00 -2.95074105e-01 -9.75924015e-01 1.66613430e-01
-2.66589671e-01 5.82071356e-02 2.06515729e-01 9.93642390e-01
-1.10561192e+00 6.55890584e-01 -9.11194921e-01 -6.53904855e-01
3.16240579e-01 2.14981601e-01 -2.23572314e-01 -5.48914680e-03
-7.51320779e-01 8.87184083e-01 2.84747422e-01 2.67337412e-01
-4.28156704e-01 -5.63336194e-01 -3.09096038e-01 -5.34966171e-01
2.01806173e-01 -1.12227961e-01 7.93889821e-01 -1.23177040e+00
-1.66408896e+00 1.27084661e+00 -3.92650455e-01 -4.56083328e-01
7.28236020e-01 8.17560107e-02 -4.14880157e-01 8.61191630e-01
-4.10521448e-01 3.72906268e-01 1.01439035e+00 -1.50348103e+00
-7.97200739e-01 -3.99276972e-01 -3.36774476e-02 1.71573445e-01
1.25936747e-01 1.31527141e-01 -7.91664541e-01 2.51622368e-02
1.91352144e-01 -1.07675576e+00 -3.50233346e-01 1.82170495e-01
-8.76029372e-01 1.63093209e-01 6.56264365e-01 -1.15144145e+00
1.27237201e+00 -1.91368604e+00 -5.62487170e-02 6.57915175e-01
3.45384002e-01 7.39790678e-01 -2.83205342e-02 1.51044324e-01
4.70586913e-03 3.44787419e-01 -4.05977935e-01 -1.18264191e-01
-6.02348983e-01 3.30940448e-02 3.23996961e-01 6.36314154e-01
-8.53185579e-02 4.60934401e-01 -7.84278274e-01 -1.06591153e+00
5.32066762e-01 6.03911042e-01 1.40195549e-01 -1.55775562e-01
-2.09734857e-01 2.82336265e-01 -1.15065254e-01 9.64214027e-01
1.38573325e+00 1.70880899e-01 3.13900739e-01 -2.03969613e-01
-3.61937582e-01 -2.54576921e-01 -1.31453943e+00 1.27100229e+00
-1.15807466e-01 2.89705604e-01 4.69975859e-01 -3.10678333e-01
6.64599180e-01 2.78401464e-01 5.96464515e-01 -5.66106260e-01
9.28847771e-03 5.01259267e-01 -1.64621025e-01 -9.63412881e-01
3.86945397e-01 -2.23184466e-01 6.53199077e-01 1.68743551e-01
-1.32912740e-01 -3.96141231e-01 6.76708043e-01 1.11642540e-01
6.06759906e-01 8.18012953e-02 1.43407553e-01 -2.15079442e-01
7.43439794e-01 2.77796835e-01 2.20124707e-01 2.17677593e-01
-4.49313492e-01 5.97607136e-01 8.16822946e-01 6.26759455e-02
-9.01219249e-01 -1.21383560e+00 -5.68009138e-01 3.55546832e-01
6.07093215e-01 3.11949309e-02 -1.51277030e+00 -6.18725359e-01
2.64052190e-02 4.42343086e-01 -9.59971130e-01 4.50446725e-01
-2.83680975e-01 -7.86923587e-01 3.05664927e-01 1.14921041e-01
5.17646492e-01 -7.39251316e-01 -7.11820006e-01 -4.16712798e-02
-9.88103598e-02 -8.40322912e-01 -2.65696317e-01 -2.83275217e-01
-8.91930282e-01 -1.44410419e+00 -1.11747932e+00 -6.53839707e-01
1.07300699e+00 2.89533347e-01 6.85422659e-01 2.93640077e-01
-1.14328980e+00 1.85534015e-01 -3.62507671e-01 -2.39722252e-01
-5.80788791e-01 -3.32719684e-01 -5.95044255e-01 3.19443107e-01
1.92095846e-01 -8.56203586e-03 -9.59213495e-01 1.30693838e-01
-1.24211419e+00 -7.69826546e-02 5.14869273e-01 4.61967438e-01
1.00735629e+00 -4.07472858e-03 -1.50368869e-01 -1.50868428e+00
4.87887412e-01 -2.02106610e-01 -5.16478062e-01 5.74582338e-01
-4.73449528e-01 -4.03553665e-01 2.55975753e-01 -1.78441659e-01
-1.38428080e+00 1.78959548e-01 1.01123177e-01 2.63728737e-03
-4.95904326e-01 1.39060065e-01 4.19001848e-01 -5.98848462e-01
6.90879226e-01 4.42034863e-02 3.29025805e-01 -4.48593140e-01
3.13187420e-01 8.61778080e-01 5.02320051e-01 1.94882363e-01
5.11031985e-01 8.52088273e-01 1.82220325e-01 -1.16136765e+00
-4.53112543e-01 -8.51160586e-01 -8.73191059e-01 -5.69578886e-01
1.17546928e+00 -1.40558690e-01 -3.79495621e-01 1.02336097e+00
-1.04643428e+00 -1.67046398e-01 -3.17051977e-01 1.71298712e-01
-7.60427490e-02 9.29007769e-01 -9.10328209e-01 -1.04128194e+00
-3.39256883e-01 -1.05677307e+00 6.06011450e-01 8.78691316e-01
-2.26746430e-03 -1.34214020e+00 2.63968945e-01 6.25708044e-01
1.34037852e-01 7.16860592e-01 7.20687509e-01 -2.61736780e-01
-3.05627286e-01 -3.87739956e-01 -1.78306103e-01 6.93679452e-01
1.50004908e-01 7.79081345e-01 -1.06766403e+00 -9.41679776e-02
-1.81052864e-01 -9.03180763e-02 1.01322055e+00 4.28985476e-01
1.09385443e+00 3.54690328e-02 -3.39303970e-01 5.28247356e-01
1.91507292e+00 3.64240497e-01 9.26797688e-01 -3.34855802e-02
4.23812121e-01 1.01906979e+00 9.25945044e-01 -3.69620062e-02
-9.76262093e-02 2.38894030e-01 6.58974230e-01 -8.98700595e-01
-4.59782600e-01 1.12738349e-01 -1.00744709e-01 2.23085329e-01
-5.20530701e-01 -1.43528163e-01 -5.89794874e-01 7.58353293e-01
-1.31832445e+00 -6.03223979e-01 -7.98014581e-01 2.24669647e+00
8.92254472e-01 -1.28881261e-01 3.50263864e-01 1.39841020e-01
1.00106716e+00 -1.17286526e-01 -5.63422859e-01 -8.85950625e-01
-8.10975581e-02 5.93273103e-01 8.82408917e-01 7.43469000e-01
-1.05808878e+00 8.24690580e-01 7.32663631e+00 9.52879488e-01
-1.41559100e+00 -2.75081042e-02 4.48653162e-01 -8.72570723e-02
-3.90347153e-01 -1.08862594e-01 -3.94701630e-01 4.55062360e-01
5.29547274e-01 1.52643144e-01 3.32898825e-01 2.17452180e-02
8.01405981e-02 -1.17187524e+00 -4.26514030e-01 6.15911782e-01
1.09810777e-01 -9.10643578e-01 -1.45761281e-01 1.45807371e-01
1.02108514e+00 -3.68160754e-01 2.83614457e-01 -7.60710835e-01
-1.02665849e-01 -1.09564972e+00 3.36853713e-01 6.31564081e-01
1.05143464e+00 -4.89755392e-01 7.78411508e-01 -2.89202720e-01
-8.70642781e-01 3.93900663e-01 -1.09391384e-01 5.33734679e-01
8.24942589e-02 7.20809102e-01 -9.72004235e-01 6.29973888e-01
4.31774557e-01 3.39786947e-01 -9.12687361e-01 1.59991050e+00
-3.14439416e-01 6.26843333e-01 -4.34051096e-01 -2.80302614e-02
3.29516858e-01 -6.72469914e-01 5.07701099e-01 1.24503922e+00
-2.38901153e-02 -2.51213521e-01 -3.41790915e-01 9.93168414e-01
4.87243414e-01 4.00218904e-01 -2.68539344e-03 -7.34740794e-02
2.67749399e-01 1.56255269e+00 -1.05241597e+00 -3.44748944e-01
-1.88383788e-01 1.39942145e+00 -3.87470871e-01 6.91672146e-01
-6.27659500e-01 -5.44456840e-01 -8.79480876e-03 4.15839523e-01
1.33882746e-01 3.07585329e-01 -5.74315965e-01 -5.26813328e-01
7.69687891e-02 -5.18284380e-01 4.49298173e-01 -4.43017751e-01
-6.74161434e-01 3.62442791e-01 -4.04521264e-02 -1.02772057e+00
1.20808296e-01 -6.92836523e-01 -9.08625901e-01 1.06757283e+00
-1.58288503e+00 -1.21489882e+00 -5.38626254e-01 4.23559010e-01
1.04896650e-01 6.18938684e-01 6.39795244e-01 -6.87453523e-02
-5.89065313e-01 2.37122774e-01 3.30279619e-01 -4.74008203e-01
7.44602978e-01 -1.80813336e+00 -1.06307760e-01 1.00698948e+00
-5.47321618e-01 3.51659566e-01 7.03931570e-01 -8.93559158e-01
-8.05330634e-01 -7.54316449e-01 6.77450955e-01 1.06741525e-01
5.00616312e-01 8.83754566e-02 -9.18380618e-01 -1.13566953e-03
5.83894610e-01 -1.76725045e-01 8.03797126e-01 -2.96715915e-01
-2.43358687e-02 -1.73193738e-01 -1.73350239e+00 3.57623011e-01
1.96631715e-01 -2.80427694e-01 -7.91405365e-02 4.21144694e-01
1.97034255e-02 -6.34492874e-01 -1.10730529e+00 1.43175244e-01
5.87813377e-01 -1.08702683e+00 7.54066527e-01 -7.73326531e-02
3.94223720e-01 -1.21516660e-01 5.89937806e-01 -1.11164916e+00
9.51200277e-02 -7.02064574e-01 2.99795061e-01 9.37851965e-01
4.32893187e-01 -5.42728484e-01 8.97979736e-01 5.94629884e-01
3.38875175e-01 -7.39923179e-01 -7.63675690e-01 -4.09381241e-01
-1.08694946e-02 3.93616021e-01 -1.55008147e-02 5.10395408e-01
1.36278570e-01 -5.97233593e-01 9.00902748e-02 -1.50947228e-01
1.14876473e+00 1.38119251e-01 2.00269535e-01 -1.07453573e+00
-4.72779609e-02 -3.38405788e-01 -2.59782881e-01 -3.12853247e-01
-5.84853053e-01 -4.87710983e-01 1.07982084e-01 -1.75568759e+00
2.54831254e-01 -1.51560649e-01 -1.80587798e-01 2.89587975e-01
-5.07891774e-01 6.10552311e-01 2.25776941e-01 1.46747798e-01
-3.87542903e-01 -5.52430332e-01 1.75212765e+00 -1.10381730e-01
-3.53881121e-01 -8.66791699e-03 -2.90119112e-01 6.39171898e-01
6.90897763e-01 4.99623679e-02 -1.17890947e-01 1.88693076e-01
-1.66885197e-01 -1.41399384e-01 4.41040158e-01 -8.43496978e-01
2.73146659e-01 -2.74984092e-01 1.44110456e-01 -6.15920424e-01
5.11689596e-02 -6.97439849e-01 2.28256673e-01 9.29232776e-01
-9.36609879e-02 -7.95825481e-01 6.21967465e-02 4.75788087e-01
-2.39441872e-01 -7.18761384e-01 1.37322295e+00 -3.23411435e-01
-5.82678497e-01 -1.53698295e-01 -4.79541391e-01 -3.67461056e-01
1.61349118e+00 -6.60322905e-01 -4.78325963e-01 1.27506629e-02
-1.04542267e+00 9.92020741e-02 9.34730470e-01 -2.46415809e-01
5.00549138e-01 -7.13276744e-01 -5.70852935e-01 9.95122045e-02
-1.55423164e-01 -6.29582256e-03 7.45774388e-01 1.43783319e+00
-1.47560358e+00 2.29579017e-01 -4.32527184e-01 -6.31976008e-01
-1.82445979e+00 3.23504955e-01 6.02221072e-01 -3.24505478e-01
-2.68809766e-01 8.19867849e-01 -2.35340089e-01 2.59319007e-01
1.50619954e-01 -4.33642417e-01 -2.95913965e-01 5.89033365e-02
4.45177644e-01 8.97042930e-01 1.18741259e-01 -4.10399795e-01
-2.79347032e-01 9.01814461e-01 -1.94459364e-01 -1.34420335e-01
7.45381415e-01 -2.95070171e-01 -5.27339637e-01 3.71345222e-01
8.75261307e-01 1.02494217e-01 -1.05994999e+00 2.37965316e-01
-3.66121531e-01 -5.94090641e-01 1.14500239e-01 -1.08018804e+00
-1.14543545e+00 1.00547874e+00 8.65459204e-01 6.09874725e-01
1.31313372e+00 -3.27256531e-01 9.84213173e-01 -4.65193659e-01
4.48576622e-02 -1.66488028e+00 -2.04136133e-01 -2.63462692e-01
5.20334661e-01 -8.13024700e-01 2.33855903e-01 -1.16540539e+00
-8.10740948e-01 1.29313338e+00 4.28189188e-01 -2.90519655e-01
5.10698557e-01 1.44449547e-01 3.19519401e-01 -5.80228753e-02
-1.15220979e-01 -5.10984898e-01 5.12267590e-01 7.34581232e-01
2.64179677e-01 2.22163901e-01 -1.08398438e+00 -1.22350916e-01
6.30104601e-01 1.95818007e-01 7.76457250e-01 8.33861232e-01
-4.76151437e-01 -1.13748431e+00 -5.45599520e-01 5.05205035e-01
-6.72407746e-01 2.70947605e-01 -7.83597827e-01 8.17691207e-01
6.02314651e-01 9.29861069e-01 2.56849807e-02 -1.97685897e-01
1.50893167e-01 -1.51027709e-01 8.63415718e-01 -4.42413956e-01
-8.75253439e-01 5.22294164e-01 5.61395586e-02 -6.07318163e-01
-7.20339417e-01 -7.58023322e-01 -1.21382225e+00 -1.58809323e-03
-2.67288029e-01 -1.73086748e-01 9.32224333e-01 6.00584805e-01
-2.99649507e-01 1.93860531e-01 7.08080232e-01 -3.31298649e-01
-2.84625322e-01 -7.73097336e-01 -1.20993364e+00 8.14935684e-01
2.90430397e-01 -2.13355362e-01 -4.98382092e-01 4.22838569e-01]
|
[15.607157707214355, -3.0402674674987793]
|
e56f252f-002e-40cd-9d47-423a18e09f92
|
stock-price-prediction-using-machine-learning
|
2009.10819
| null |
https://arxiv.org/abs/2009.10819v1
|
https://arxiv.org/pdf/2009.10819v1.pdf
|
Stock Price Prediction Using Machine Learning and LSTM-Based Deep Learning Models
|
Prediction of stock prices has been an important area of research for a long time. While supporters of the efficient market hypothesis believe that it is impossible to predict stock prices accurately, there are formal propositions demonstrating that accurate modeling and designing of appropriate variables may lead to models using which stock prices and stock price movement patterns can be very accurately predicted. In this work, we propose an approach of hybrid modeling for stock price prediction building different machine learning and deep learning-based models. For the purpose of our study, we have used NIFTY 50 index values of the National Stock Exchange (NSE) of India, during the period December 29, 2014 till July 31, 2020. We have built eight regression models using the training data that consisted of NIFTY 50 index records during December 29, 2014 till December 28, 2018. Using these regression models, we predicted the open values of NIFTY 50 for the period December 31, 2018 till July 31, 2020. We, then, augment the predictive power of our forecasting framework by building four deep learning-based regression models using long-and short-term memory (LSTM) networks with a novel approach of walk-forward validation. We exploit the power of LSTM regression models in forecasting the future NIFTY 50 open values using four different models that differ in their architecture and in the structure of their input data. Extensive results are presented on various metrics for the all the regression models. The results clearly indicate that the LSTM-based univariate model that uses one-week prior data as input for predicting the next week open value of the NIFTY 50 time series is the most accurate model.
|
['Jaydip Sen', 'Sidra Mehtab', 'Abhishek Dutta']
|
2020-09-20
| null | null | null | null |
['stock-price-prediction']
|
['time-series']
|
[-6.30892754e-01 -2.87770003e-01 -4.07904088e-01 -2.97250330e-01
-1.43047035e-01 -4.93134022e-01 7.97265410e-01 -1.22735091e-01
-4.74268436e-01 1.03360724e+00 1.62506342e-01 -9.28374112e-01
-2.43316278e-01 -1.30285072e+00 -6.39062464e-01 -3.98171276e-01
-4.80805427e-01 3.28945696e-01 -1.16462752e-01 -6.58854723e-01
6.22427046e-01 4.73492593e-01 -1.25258660e+00 8.30013677e-03
3.22667658e-01 1.68741846e+00 -7.42583498e-02 4.39359426e-01
-3.19545060e-01 1.24018443e+00 -5.10225177e-01 -2.92341381e-01
9.23238456e-01 -1.45646438e-01 -2.93991983e-01 -7.63257921e-01
-1.61980867e-01 -7.05256402e-01 -1.98584691e-01 7.49016345e-01
1.26390532e-01 -1.70745790e-01 2.89330751e-01 -1.24552059e+00
-7.43825138e-01 1.11222839e+00 -5.22886574e-01 7.09683478e-01
-3.92220199e-01 6.98841810e-02 1.21939993e+00 -7.74540663e-01
2.28541777e-01 7.29800105e-01 9.62964892e-01 -4.92256805e-02
-9.79969680e-01 -1.15729046e+00 6.77112043e-02 1.96242660e-01
-9.51230705e-01 -9.17770639e-02 7.65736699e-01 -7.01787829e-01
1.25215828e+00 1.25010699e-01 1.05242980e+00 6.18127882e-01
7.15976894e-01 3.13082367e-01 1.36220789e+00 -3.30202192e-01
1.15125217e-01 1.47657812e-01 1.29578769e-01 7.48301074e-02
4.25771624e-01 6.29279315e-01 -3.09890062e-01 -1.92175224e-01
8.26547027e-01 3.11449468e-01 1.98216468e-01 4.75462556e-01
-1.17218697e+00 1.12967730e+00 4.49242294e-01 7.19558954e-01
-8.99029255e-01 3.34522784e-01 3.30289423e-01 8.90184462e-01
8.37110937e-01 4.79775816e-01 -1.10727215e+00 -1.53517753e-01
-1.49726284e+00 4.75851744e-01 9.95244384e-01 2.75391638e-01
4.58663702e-01 8.75303149e-01 8.94299373e-02 3.95153850e-01
3.26225787e-01 5.49349368e-01 1.07453799e+00 -4.51557994e-01
3.39587659e-01 3.60588402e-01 2.96909183e-01 -1.05652213e+00
-5.27509153e-01 -7.71794319e-01 -6.76558554e-01 4.60665256e-01
2.54788190e-01 -4.98476624e-01 -7.94187665e-01 1.42379236e+00
-4.87229705e-01 4.27757263e-01 3.47064704e-01 3.75999123e-01
3.07112724e-01 1.04627204e+00 -3.43454890e-02 -3.84572178e-01
8.24073255e-01 -6.78776145e-01 -4.84222680e-01 9.27256048e-02
4.90721226e-01 -3.80784810e-01 3.43019098e-01 2.19515875e-01
-9.89254653e-01 -5.25326133e-01 -1.07544947e+00 5.67018688e-01
-9.62628543e-01 -2.68370181e-01 5.88054359e-01 2.67618895e-01
-1.06631672e+00 1.11575389e+00 -6.36081815e-01 3.25890183e-01
-2.08924767e-02 3.87852937e-01 2.67257124e-01 9.28723752e-01
-1.74136186e+00 1.24526846e+00 5.77401698e-01 2.53959149e-01
-2.62594283e-01 -9.35264051e-01 -4.22226429e-01 2.38358751e-01
-3.34245324e-01 -2.95128435e-01 1.26807976e+00 -1.16131318e+00
-1.26322210e+00 3.04168552e-01 4.12846208e-01 -1.37817490e+00
6.31943405e-01 5.89897111e-02 -6.98868155e-01 -4.59421486e-01
-2.08559446e-02 2.49913886e-01 5.19273460e-01 -6.31967664e-01
-9.41663623e-01 -2.89129496e-01 -2.39176124e-01 -3.37301075e-01
-2.04621822e-01 9.36602429e-02 6.68257713e-01 -9.91321266e-01
3.59796104e-03 -7.34815598e-01 -1.72043353e-01 -7.37604022e-01
2.93396208e-02 -1.15761898e-01 5.83847463e-01 -1.05633187e+00
1.48532975e+00 -1.59914267e+00 -6.44677401e-01 5.44883072e-01
-1.99833617e-01 6.34815767e-02 3.03742856e-01 5.11466324e-01
-5.16825616e-01 3.70015085e-01 -1.90501869e-01 1.83885638e-02
1.68304756e-01 3.04000795e-01 -1.19238865e+00 1.57760695e-01
9.82344523e-02 1.17579520e+00 -3.19189578e-01 2.02846780e-01
2.47006536e-01 1.86995521e-01 9.03421119e-02 -5.66291101e-02
-3.43539685e-01 7.61831105e-02 -2.27436841e-01 4.87378567e-01
5.99295139e-01 -1.46425083e-01 -2.36294657e-01 2.46796787e-01
-8.07925344e-01 3.88995767e-01 -6.80803359e-01 8.19769323e-01
-4.08099204e-01 7.89511681e-01 -5.83247483e-01 -9.73694503e-01
1.28343999e+00 4.04752225e-01 6.88511431e-01 -1.13634586e+00
1.03981458e-01 8.64786267e-01 3.51238102e-01 -6.11947440e-02
5.05243421e-01 -6.17462277e-01 7.34488666e-02 6.03933632e-01
-4.18064952e-01 1.80595577e-01 2.03657761e-01 -5.30863762e-01
6.49832487e-01 2.40760177e-01 1.59643903e-01 -3.31882119e-01
2.55394340e-01 4.40035015e-02 5.40581465e-01 4.09722716e-01
-2.41563153e-02 5.70442975e-02 4.71311331e-01 -1.11034667e+00
-1.30082595e+00 -5.76838791e-01 -2.94789582e-01 9.02656734e-01
-8.09247732e-01 3.49152416e-01 -3.38775665e-02 1.11660617e-03
5.26321113e-01 1.21846652e+00 -9.38551247e-01 3.13646853e-01
-6.10895455e-01 -1.14866769e+00 5.68063796e-01 4.47313070e-01
6.30106866e-01 -1.36725533e+00 -1.09734607e+00 5.69420516e-01
3.83007884e-01 -5.52296400e-01 3.29419494e-01 5.31729698e-01
-1.15424836e+00 -6.69629097e-01 -9.88337100e-01 -3.48792523e-01
-7.74535164e-02 -3.51883501e-01 1.25045884e+00 -2.11336650e-02
4.02215719e-01 -2.37614602e-01 -7.41720051e-02 -1.09742224e+00
-2.27582008e-01 4.11421582e-02 -1.07114501e-01 -9.92652848e-02
5.61434388e-01 -7.38407671e-01 -4.44589108e-01 -7.64815137e-02
-7.58893847e-01 -4.85895341e-03 6.13288820e-01 4.94200408e-01
3.47349912e-01 8.02469403e-02 1.13233256e+00 -4.84564841e-01
6.68000281e-01 -9.77126718e-01 -1.11245525e+00 5.41728921e-02
-1.33998871e+00 4.19956911e-03 5.64104319e-01 -1.85250059e-01
-6.77592933e-01 -4.39016432e-01 -2.65249938e-01 -4.77834284e-01
4.73823220e-01 1.10894692e+00 8.80851030e-01 1.95278212e-01
1.06007099e-01 6.05301261e-01 2.95269135e-02 -5.97648323e-01
-1.98707834e-01 3.80223006e-01 4.41986978e-01 -1.53444335e-01
8.69047940e-01 2.32802048e-01 2.79477797e-02 -4.93683010e-01
-5.62469363e-01 2.08395254e-02 -6.60643101e-01 -2.24182427e-01
4.31400239e-01 -9.75641787e-01 -5.46792269e-01 8.55977476e-01
-7.47383475e-01 -4.05019820e-01 -4.23344582e-01 5.04424274e-01
-3.34837705e-01 -4.13537472e-01 -7.40108132e-01 -1.13997412e+00
-7.08632588e-01 -7.73003340e-01 1.66169822e-01 1.21104121e-01
-2.38407642e-01 -1.37509716e+00 3.85154456e-01 -3.63571346e-01
1.12723577e+00 7.88634896e-01 9.41389859e-01 -1.17871511e+00
-3.13220739e-01 -3.94579828e-01 -1.55129373e-01 5.31506956e-01
1.12507828e-02 2.22645253e-01 -6.34016752e-01 -7.90245086e-03
2.98077166e-01 -4.93537374e-02 9.01701093e-01 7.93049157e-01
5.99551201e-01 -6.67244077e-01 3.97637427e-01 6.51342094e-01
1.77704394e+00 7.24373460e-01 7.09856331e-01 1.32716644e+00
1.85028985e-01 4.35290813e-01 3.42050225e-01 5.97143292e-01
4.06764150e-01 1.48509771e-01 3.78926635e-01 -1.51425442e-02
5.85074008e-01 -2.80275613e-01 5.20502687e-01 8.79493058e-01
-2.99522132e-01 3.02059442e-01 -1.00225461e+00 4.29191411e-01
-1.55138850e+00 -1.29490995e+00 -1.16556235e-01 1.96124530e+00
6.67232215e-01 5.74948609e-01 2.79034913e-01 3.36641192e-01
1.57501474e-01 5.07116199e-01 -7.20904648e-01 -5.45310020e-01
-2.50906318e-01 2.98032075e-01 1.26734567e+00 1.39756694e-01
-9.55868900e-01 6.26658022e-01 6.46878433e+00 2.69175977e-01
-1.58406305e+00 -2.07939729e-01 1.02079713e+00 -6.05672672e-02
-4.49582458e-01 -6.30360320e-02 -8.19644392e-01 8.40186894e-01
1.75893009e+00 -6.37499571e-01 1.58353463e-01 9.34557557e-01
7.08223283e-01 1.18018836e-02 -5.95602751e-01 5.91937244e-01
-3.38547230e-01 -1.66965556e+00 3.67503278e-02 3.35262030e-01
7.87075341e-01 3.86683762e-01 3.54514003e-01 5.27849913e-01
3.14848036e-01 -9.56957102e-01 1.08429575e+00 1.09407473e+00
2.94847548e-01 -9.40016866e-01 1.05950737e+00 4.40882236e-01
-9.76037204e-01 -5.13722658e-01 -3.83198053e-01 -6.86523378e-01
2.13355552e-02 5.29212832e-01 -4.16308463e-01 3.40174198e-01
8.30764592e-01 7.82683492e-01 -2.75521606e-01 6.15188241e-01
2.39134863e-01 6.94995582e-01 -3.11352670e-01 -4.15536352e-02
7.38127291e-01 -4.09278780e-01 1.16287068e-01 8.88146579e-01
8.26132596e-01 6.31214827e-02 -3.34126621e-01 8.93606722e-01
1.26796901e-01 9.45999995e-02 -6.76576138e-01 -1.57991961e-01
2.53681332e-01 6.14246070e-01 -4.53275114e-01 -4.20649797e-01
-7.92073667e-01 3.87012959e-02 -2.42853776e-01 2.86584347e-01
-6.98265612e-01 -2.44051427e-01 4.98260498e-01 2.26400241e-01
3.02754253e-01 -2.25752011e-01 -8.39402914e-01 -1.03567898e+00
-1.79920435e-01 -6.56019032e-01 2.13419288e-01 -7.58490145e-01
-1.18480480e+00 6.41406536e-01 1.34128645e-01 -1.13381195e+00
-8.11613500e-01 -6.15493059e-01 -1.02135456e+00 1.27027237e+00
-2.00377440e+00 -7.80365646e-01 3.93195689e-01 2.70113349e-01
4.94977057e-01 -7.60916114e-01 5.78438938e-01 -1.26509145e-01
-4.32801843e-01 -5.67355491e-02 7.09754944e-01 4.63273913e-01
-5.67725860e-02 -1.14457214e+00 9.07575071e-01 6.67024672e-01
-3.52315558e-03 4.94515479e-01 7.06961215e-01 -9.24603522e-01
-8.34320247e-01 -9.52858448e-01 1.13154233e+00 7.75671899e-02
1.34247494e+00 3.35019171e-01 -9.57916558e-01 1.09903252e+00
2.64729410e-01 -4.06186074e-01 5.61023355e-01 -4.28125024e-01
-9.80689824e-02 -2.68808395e-01 -1.03806269e+00 1.45293206e-01
-9.91900712e-02 -1.96182087e-01 -8.93195927e-01 -9.88219306e-02
4.77156341e-01 5.35694733e-02 -1.21170783e+00 4.56042349e-01
9.81391728e-01 -1.20927668e+00 8.44907463e-01 -5.23777843e-01
3.57983649e-01 1.66405708e-01 -2.40279570e-01 -1.38032341e+00
-2.38409653e-01 -2.99665093e-01 -1.47687733e-01 8.89601469e-01
8.64710808e-01 -1.30258572e+00 7.12584674e-01 9.91506875e-01
8.27560276e-02 -9.58475351e-01 -1.00582325e+00 -7.69749641e-01
7.11286306e-01 -4.96039271e-01 1.11853695e+00 1.02898645e+00
-3.46370578e-01 -2.65400857e-01 -4.93496090e-01 -1.29868388e-01
3.26194078e-01 3.99766773e-01 4.92022157e-01 -1.52659619e+00
-1.41328067e-01 -8.16872954e-01 -1.55788511e-01 -3.40548635e-01
2.06791162e-01 -4.90497023e-01 -6.70448780e-01 -1.46832645e+00
-2.68913418e-01 -4.30268764e-01 -8.65920067e-01 6.19912803e-01
4.34272856e-01 -7.87609220e-02 3.76792222e-01 7.70712733e-01
5.97234249e-01 3.22521806e-01 7.42623568e-01 -6.18962310e-02
-2.24112019e-01 3.49525690e-01 -4.06226009e-01 7.75097966e-01
1.18331552e+00 -1.84847414e-01 -1.23984545e-01 -1.13807656e-01
6.60512805e-01 1.70246989e-01 3.27051222e-01 -8.77455950e-01
-1.08057298e-01 -3.54246885e-01 8.60800564e-01 -9.96732652e-01
1.07378796e-01 -7.92373836e-01 5.61876595e-01 9.61912036e-01
-3.21379513e-01 8.74994040e-01 4.14841920e-01 4.14888188e-02
-4.83593315e-01 -2.22479105e-01 4.91661906e-01 -4.68115658e-01
-7.76517749e-01 2.93813497e-01 -3.00527334e-01 -3.47641736e-01
1.07671428e+00 -3.29447687e-01 -1.67967211e-02 -5.15728951e-01
-6.62881613e-01 2.53299356e-01 9.46245342e-02 4.49432850e-01
4.43946004e-01 -1.28418493e+00 -9.07760441e-01 3.40203971e-01
-5.73968828e-01 -6.71291173e-01 -2.38241896e-01 6.06198728e-01
-7.63875425e-01 8.53669524e-01 -5.17036140e-01 9.19838622e-02
-4.16339964e-01 4.38670009e-01 5.95737040e-01 -5.78258514e-01
-5.24140716e-01 4.68495071e-01 -3.87031227e-01 1.13187589e-01
-7.06645176e-02 -8.87331069e-01 -5.26281059e-01 7.31079280e-01
5.59924781e-01 4.11999494e-01 5.80344312e-02 -8.18053663e-01
4.72799465e-02 5.02649724e-01 2.04082668e-01 -3.27254653e-01
2.02067256e+00 5.44262379e-02 -1.60185680e-01 1.16771805e+00
1.11713445e+00 -4.63234037e-01 -1.04096663e+00 -5.96778393e-02
6.44678235e-01 -1.48854017e-01 2.63531923e-01 -8.26141059e-01
-1.45158291e+00 6.08956456e-01 6.02720976e-01 6.54413939e-01
1.00047278e+00 -8.13050985e-01 1.09593689e+00 2.13040322e-01
2.83941120e-01 -1.16511607e+00 -7.21098900e-01 6.65128767e-01
9.90754664e-01 -9.94451225e-01 -1.28526717e-01 7.06057668e-01
-4.23958421e-01 1.40515196e+00 -1.81775335e-02 -5.02146423e-01
1.24811447e+00 2.66900539e-01 1.57969102e-01 -3.38383652e-02
-1.09089780e+00 6.06813207e-02 4.01328951e-02 -5.61533421e-02
3.75783592e-01 1.87357292e-01 -2.43057117e-01 6.96472883e-01
-7.68157840e-01 3.06316525e-01 5.99798501e-01 8.60536695e-01
-5.40477037e-01 -7.58326948e-01 -5.51165879e-01 9.63163614e-01
-9.28226650e-01 -3.68349314e-01 -1.00317232e-01 1.12652493e+00
-2.56250769e-01 5.47469974e-01 5.61487794e-01 -3.31099331e-01
1.82430893e-01 4.99190718e-01 -3.37125093e-01 -2.02884842e-02
-9.66629744e-01 -1.55230775e-03 -1.16516195e-01 -5.84563874e-02
-2.42811337e-01 -7.93426156e-01 -1.17460930e+00 -8.09522510e-01
3.66727114e-02 1.56680509e-01 8.47438514e-01 1.07262862e+00
1.47827892e-02 3.30156982e-01 9.71829295e-01 -9.65396523e-01
-1.13506901e+00 -1.21616614e+00 -9.75583136e-01 -6.25184253e-02
5.80921531e-01 -4.54044819e-01 -5.70997775e-01 -1.18236005e-01]
|
[4.515870094299316, 4.189015865325928]
|
28bed18e-b443-4dfc-8f86-7c34da5735cf
|
feature-augmented-machine-reading
|
2211.09438
| null |
https://arxiv.org/abs/2211.09438v1
|
https://arxiv.org/pdf/2211.09438v1.pdf
|
Feature-augmented Machine Reading Comprehension with Auxiliary Tasks
|
While most successful approaches for machine reading comprehension rely on single training objective, it is assumed that the encoder layer can learn great representation through the loss function we define in the predict layer, which is cross entropy in most of time, in the case that we first use neural networks to encode the question and paragraph, then directly fuse the encoding result of them. However, due to the distantly loss backpropagating in reading comprehension, the encoder layer cannot learn effectively and be directly supervised. Thus, the encoder layer can not learn the representation well at any time. Base on this, we propose to inject multi granularity information to the encoding layer. Experiments demonstrate the effect of adding multi granularity information to the encoding layer can boost the performance of machine reading comprehension system. Finally, empirical study shows that our approach can be applied to many existing MRC models.
|
['Yifeng Xie']
|
2022-11-17
| null | null | null | null |
['machine-reading-comprehension']
|
['natural-language-processing']
|
[ 6.70011878e-01 8.02282572e-01 -1.57329395e-01 -4.81441468e-01
-7.20694363e-01 -3.05210531e-01 3.71468484e-01 5.58360100e-01
-4.47766036e-01 7.33899474e-01 4.55995649e-01 -4.53568578e-01
-1.11327164e-01 -1.18873596e+00 -1.04241419e+00 -2.84271270e-01
3.06587189e-01 2.91946858e-01 6.60027936e-02 -2.83170342e-01
2.01910317e-01 -2.55610347e-01 -1.42846966e+00 4.87496197e-01
1.28597283e+00 1.02466476e+00 4.98511642e-01 8.09496343e-01
-1.53439581e-01 1.29036045e+00 -4.58178818e-01 -5.90835690e-01
-6.83700293e-02 -7.50960410e-01 -1.37790942e+00 -4.99730587e-01
3.27863872e-01 -6.29897773e-01 -4.82433379e-01 1.04438543e+00
2.45129555e-01 1.69726133e-01 8.34809780e-01 -6.95165396e-01
-8.57059658e-01 9.65798736e-01 -2.75168389e-01 1.98437795e-01
7.61211574e-01 -2.33724132e-01 1.27886713e+00 -4.37137395e-01
3.16185981e-01 1.07223094e+00 5.03351808e-01 4.38448668e-01
-8.49175632e-01 -2.69677490e-01 3.37104619e-01 4.77662563e-01
-7.71572411e-01 -2.72055626e-01 8.02092791e-01 -7.75182620e-02
1.00369430e+00 1.51449502e-01 4.80768949e-01 1.07412863e+00
4.14984882e-01 1.38870072e+00 1.07435143e+00 -6.53148174e-01
-1.55991490e-03 7.97443762e-02 7.01730967e-01 8.51741076e-01
-1.62458628e-01 5.90238757e-02 -3.17167431e-01 3.09558064e-01
2.32783601e-01 -5.93598671e-02 -6.39819801e-01 7.56874159e-02
-7.99113989e-01 8.44728470e-01 5.93234003e-01 2.97806077e-02
-2.56846398e-01 -4.68704440e-02 2.89377689e-01 7.53866136e-01
3.78283530e-01 5.06488144e-01 -6.07451856e-01 -2.30430394e-01
-9.94203925e-01 6.40372885e-03 1.09027076e+00 8.19644570e-01
5.76127231e-01 -4.61817950e-01 -4.10630524e-01 6.39480233e-01
2.69424021e-01 1.56425774e-01 6.09823525e-01 -7.46563673e-01
7.85932899e-01 6.10080719e-01 -5.05439699e-01 -7.54442692e-01
-2.94537902e-01 -5.99949598e-01 -9.30063486e-01 -1.97349727e-01
2.72975802e-01 -2.47992069e-01 -7.50355124e-01 1.77350831e+00
-3.31575871e-01 7.32771382e-02 3.20193917e-01 5.81726372e-01
7.78071404e-01 7.07372785e-01 -3.27628106e-02 -1.77156106e-01
1.18610179e+00 -1.26351345e+00 -9.48980570e-01 -3.25943351e-01
6.00513995e-01 -3.74957055e-01 1.09699821e+00 3.88162553e-01
-1.52269673e+00 -6.02228403e-01 -1.50010979e+00 -5.63730180e-01
-5.26306987e-01 -2.34961919e-02 4.69458550e-01 2.76441723e-01
-8.37914109e-01 5.48338592e-01 -6.23597622e-01 -3.45272943e-02
3.75163049e-01 1.96869463e-01 -1.37376159e-01 -3.13782215e-01
-1.63745618e+00 1.19671524e+00 8.26150596e-01 -1.25578046e-01
-7.55480647e-01 -5.16671002e-01 -8.43494833e-01 6.37016654e-01
1.33314013e-01 -9.55097795e-01 1.36472321e+00 -1.08019459e+00
-1.60536981e+00 4.01073694e-01 -2.08675310e-01 -6.59502625e-01
3.57505262e-01 -5.19674540e-01 1.62881892e-02 1.73506647e-01
-2.70176917e-01 6.63953424e-01 5.67820668e-01 -1.14930892e+00
-6.79801583e-01 -2.87255764e-01 5.46002567e-01 5.10873556e-01
-3.25425208e-01 -5.34361362e-01 -4.40879390e-02 -3.37513655e-01
2.37694681e-02 -3.26817811e-01 2.24966690e-01 -4.46235240e-01
-3.89120936e-01 -4.44388568e-01 5.72423875e-01 -1.26943684e+00
1.40129411e+00 -1.91612566e+00 3.71136248e-01 2.13627405e-02
1.73795924e-01 2.32604757e-01 -1.96956158e-01 4.50808465e-01
-6.52094632e-02 1.60081521e-01 -3.31018031e-01 -3.09452593e-01
3.76278423e-02 2.83816040e-01 -4.86653715e-01 -3.08369659e-02
2.96769410e-01 1.12113655e+00 -1.00692677e+00 -2.46467620e-01
-1.19298985e-02 2.93628603e-01 -5.56762815e-01 6.66307807e-01
-5.00117421e-01 -3.66736203e-02 -4.93306011e-01 1.30155012e-01
5.75527847e-01 -3.41449708e-01 -1.18506856e-01 8.15799534e-02
3.38099092e-01 8.95199478e-01 -5.78354537e-01 2.00561786e+00
-6.84471190e-01 7.11395085e-01 -3.15523922e-01 -1.21522212e+00
6.20815098e-01 3.50572497e-01 -1.62793651e-01 -9.18197930e-01
1.68326318e-01 7.80466769e-04 8.95471871e-02 -5.58455765e-01
4.38478738e-01 1.11460932e-01 1.66752070e-01 5.46727538e-01
2.47059345e-01 1.30300358e-01 -7.89991319e-02 2.56615818e-01
1.05868483e+00 1.59838349e-01 3.44343603e-01 1.98941335e-01
6.65787399e-01 -2.68812329e-01 2.98789777e-02 8.89737844e-01
1.26182631e-01 5.28793156e-01 6.47329926e-01 7.26082847e-02
-9.32564437e-01 -9.63564873e-01 -1.24982111e-01 1.22775650e+00
1.27478883e-01 -4.02217627e-01 -1.13148057e+00 -1.00323009e+00
-2.97671020e-01 1.00487769e+00 -6.74044013e-01 -7.47773826e-01
-6.43902957e-01 -4.28097785e-01 6.43891811e-01 6.69263422e-01
9.89848673e-01 -8.15411747e-01 -7.66497433e-01 2.54119813e-01
-4.20489877e-01 -8.00943494e-01 -6.78044558e-02 4.74165916e-01
-9.45252359e-01 -8.48984063e-01 -4.36178714e-01 -8.67671430e-01
5.23031771e-01 -9.49964598e-02 1.41282117e+00 4.20104206e-01
1.44879043e-01 2.80804157e-01 -6.02011025e-01 -5.11522293e-01
-4.92873073e-01 6.12388253e-01 -6.35992348e-01 -3.65871817e-01
5.27743638e-01 -3.79839033e-01 -4.71100926e-01 -2.20692337e-01
-8.35613191e-01 3.87254119e-01 6.51424050e-01 1.08111429e+00
2.38032773e-01 2.44980991e-01 8.57337892e-01 -8.44040811e-01
1.07451367e+00 -6.07829273e-01 -1.16010882e-01 7.76649594e-01
-6.72803342e-01 4.08867121e-01 7.08139122e-01 -1.11761265e-01
-1.20934439e+00 -3.60739022e-01 -4.35215801e-01 2.03387477e-02
-1.60652831e-01 8.35313499e-01 -3.19832325e-01 3.36524516e-01
2.59479821e-01 3.80611628e-01 -7.81213865e-02 -5.73122203e-01
3.87598962e-01 7.67943501e-01 4.30120081e-01 -4.25877184e-01
4.15537536e-01 -2.16079623e-01 -3.19641024e-01 -4.17780399e-01
-1.32715547e+00 -2.09021084e-02 -5.78630149e-01 9.39967558e-02
1.14728737e+00 -8.13416064e-01 -5.78631818e-01 3.02182913e-01
-1.33714294e+00 -1.38741970e-01 -3.23159665e-01 4.35631394e-01
-6.06646657e-01 2.44264722e-01 -5.68956017e-01 -7.41042912e-01
-2.67668098e-01 -1.03199589e+00 8.80741835e-01 3.95625174e-01
-1.13107443e-01 -1.42375052e+00 -9.58864316e-02 5.18666625e-01
5.69097579e-01 -5.89243881e-02 1.57481503e+00 -8.49753857e-01
-6.55668676e-01 -1.93488076e-01 -1.88285023e-01 6.88325346e-01
-1.59787491e-01 -5.02167106e-01 -1.32239294e+00 -2.09152669e-01
4.39375103e-01 -8.14475715e-01 1.24450541e+00 1.85625434e-01
1.58951581e+00 -4.43945169e-01 -3.86721715e-02 5.47577977e-01
1.33080101e+00 -9.88209620e-03 9.69716072e-01 2.41025895e-01
5.47493398e-01 7.88477600e-01 3.50662857e-01 -2.34906487e-02
8.53133440e-01 2.12271512e-01 4.78372961e-01 3.77190784e-02
-2.45238826e-01 -7.37152636e-01 4.98371929e-01 1.25874782e+00
-9.23607945e-02 -5.43051064e-01 -7.24527001e-01 1.57695696e-01
-1.77910280e+00 -8.88088584e-01 2.56303728e-01 2.03173232e+00
1.09564745e+00 1.75393835e-01 -5.27065158e-01 2.28687674e-01
1.47749633e-01 1.38124719e-01 -5.02287030e-01 -7.88037658e-01
-1.00779124e-01 3.88434559e-01 2.84340233e-01 7.89852858e-01
-9.40594435e-01 6.86803699e-01 7.01832914e+00 7.63181806e-01
-8.20029974e-01 6.25406206e-02 7.00046360e-01 2.89234847e-01
-4.61865842e-01 -1.32941986e-02 -6.57892942e-01 2.90384769e-01
1.10249054e+00 -1.76581368e-01 4.10412252e-01 4.37201232e-01
-2.82839626e-01 -2.03388661e-01 -1.53010297e+00 5.04642546e-01
2.85162717e-01 -9.27218080e-01 2.53095001e-01 -6.60803989e-02
4.89978641e-01 -2.47198611e-01 1.04410902e-01 7.08100975e-01
9.65136662e-02 -1.52785313e+00 2.80141830e-01 9.02110815e-01
3.42584819e-01 -7.96527743e-01 1.05274665e+00 9.83868659e-01
-6.87222064e-01 -1.85259432e-01 -6.76471591e-01 -3.73001724e-01
9.27923247e-02 4.28125411e-01 -7.85597682e-01 8.01701546e-01
2.26948619e-01 6.39721394e-01 -6.19592905e-01 8.64029586e-01
-6.36074245e-01 8.26784492e-01 1.05098411e-02 -8.28301385e-02
1.77704006e-01 -6.57468736e-02 1.68413058e-01 1.06796479e+00
2.29572043e-01 1.50244176e-01 1.01443395e-01 6.92758858e-01
-4.65082288e-01 3.70603949e-02 -5.56220889e-01 5.49744815e-02
2.30644897e-01 5.83434761e-01 1.06946006e-01 -3.11839759e-01
-5.03510714e-01 1.09905350e+00 7.84070015e-01 3.81881297e-01
-7.61517704e-01 -5.09891450e-01 7.98836444e-03 -4.25454043e-02
1.04423806e-01 5.99756166e-02 -5.07538021e-01 -1.36739540e+00
1.47953436e-01 -9.35624719e-01 2.96366483e-01 -7.45348096e-01
-1.19517076e+00 3.87478381e-01 8.15044157e-03 -7.42036641e-01
-4.04656321e-01 -4.76377457e-01 -7.64834046e-01 1.01973319e+00
-1.95722342e+00 -9.78168011e-01 -1.17162704e-01 3.95085901e-01
6.63038671e-01 -1.33617699e-01 9.95065153e-01 9.34932157e-02
-4.34917837e-01 9.35003340e-01 1.68073788e-01 1.30218759e-01
4.08977360e-01 -1.63051593e+00 3.03624347e-02 6.44403160e-01
5.00319898e-02 6.88817441e-01 4.42807257e-01 -3.77539247e-01
-1.26799428e+00 -7.03052640e-01 1.28342104e+00 -3.57403815e-01
3.74913722e-01 -3.90324980e-01 -1.32549238e+00 7.94966102e-01
9.30683374e-01 -7.11429298e-01 7.83077419e-01 3.00532669e-01
-3.83369684e-01 1.26725703e-01 -1.00587201e+00 2.90366888e-01
6.94024801e-01 -7.20526576e-01 -1.37818885e+00 3.65068167e-01
9.96247709e-01 -4.07245845e-01 -1.16151345e+00 4.19677883e-01
3.56529206e-01 -9.00045455e-01 8.01528096e-01 -8.48562956e-01
1.01554334e+00 1.90306947e-01 -1.60607100e-01 -1.44866323e+00
-1.52621642e-01 1.24941412e-02 -5.21327019e-01 1.15225697e+00
7.32456028e-01 -5.41158259e-01 3.97843480e-01 5.59222758e-01
-1.36365026e-01 -1.11354887e+00 -8.72293591e-01 -4.29049969e-01
7.13038266e-01 -1.51673108e-01 7.58235395e-01 7.59234548e-01
3.39413285e-01 8.50569308e-01 -1.98290825e-01 6.51836023e-02
3.97931069e-01 -6.95803622e-03 2.65637100e-01 -1.16248512e+00
-6.88073218e-01 -4.20354456e-01 -4.43459116e-02 -1.76055801e+00
4.83893096e-01 -1.09529400e+00 6.08366504e-02 -1.74623156e+00
4.61476803e-01 -1.49736509e-01 -4.69420940e-01 3.07532370e-01
-6.04996502e-01 -5.94341099e-01 2.08768040e-01 -5.76592833e-02
-4.69250560e-01 7.26080060e-01 1.57414162e+00 -2.71937460e-01
2.38518029e-01 -1.72138866e-02 -7.59075820e-01 6.64111733e-01
7.98972964e-01 -2.11904183e-01 -8.14857900e-01 -8.20752501e-01
4.15267706e-01 1.86675027e-01 2.29842514e-01 -9.04744864e-01
3.59528720e-01 2.83981413e-01 5.02419114e-01 -5.50277233e-01
2.55282342e-01 -7.72696078e-01 -5.77743351e-01 3.06676388e-01
-1.01142442e+00 -1.27467841e-01 -1.31260917e-01 4.90525156e-01
-6.43542945e-01 -7.48588383e-01 5.27990103e-01 -9.54273045e-02
-3.17451358e-01 1.52638778e-02 -9.52249318e-02 2.75803685e-01
5.86295784e-01 3.26958671e-02 -5.13245583e-01 -5.89105010e-01
-5.97215116e-01 5.40148377e-01 1.96390092e-01 3.73079985e-01
6.25815332e-01 -9.73570049e-01 -7.46644437e-01 2.70729214e-01
-1.61730945e-01 2.08309740e-01 5.20792929e-03 5.87603390e-01
-1.98231325e-01 5.32590747e-01 -1.50682777e-01 -3.80583912e-01
-8.49850059e-01 4.36831087e-01 4.30748671e-01 -7.47723937e-01
-4.81198281e-01 7.59000003e-01 2.12252825e-01 -4.53204513e-01
5.59828401e-01 -4.09385830e-01 -6.77843511e-01 -3.65303271e-02
6.76153302e-01 2.95461535e-01 1.10917622e-02 -1.42408341e-01
1.47876129e-01 3.42253685e-01 -4.75675106e-01 7.14629935e-03
1.11596012e+00 -2.59919971e-01 -6.39652759e-02 4.96370435e-01
1.55152380e+00 -4.65910137e-01 -1.05541980e+00 -1.24679677e-01
5.89917712e-02 3.11773233e-02 1.24787912e-01 -1.10331368e+00
-7.81782150e-01 1.45533061e+00 4.27009910e-01 1.81768358e-01
1.34780824e+00 -1.51089683e-01 9.49530244e-01 7.59230733e-01
7.89297745e-02 -1.12944567e+00 -4.44767922e-02 1.00887156e+00
7.54276335e-01 -1.27117789e+00 -2.22801521e-01 -2.64202803e-01
-4.21599627e-01 1.17713571e+00 6.88728988e-01 -1.01624615e-01
5.12884021e-01 1.51554182e-01 -2.44473293e-01 -2.67865253e-03
-9.65786338e-01 1.10263623e-01 3.75513345e-01 4.30366665e-01
8.21298838e-01 4.11651768e-02 -4.08883989e-01 7.41432428e-01
-4.74558234e-01 -4.36956547e-02 5.25601804e-01 8.38283718e-01
-7.15426862e-01 -1.16554308e+00 1.32013708e-01 6.66440725e-01
-4.85968590e-01 -3.54683012e-01 -3.65857780e-01 5.35685718e-01
2.79688872e-02 9.26149249e-01 2.17337292e-02 -4.03598934e-01
3.28964233e-01 4.69873011e-01 5.30174911e-01 -4.63133812e-01
-7.41537213e-01 -5.19741237e-01 1.97505027e-01 -3.42648447e-01
-2.61465222e-01 -3.24734330e-01 -1.19560349e+00 -5.01852483e-02
-4.83215511e-01 3.12733263e-01 4.99951959e-01 1.19070458e+00
7.65318573e-02 8.55773807e-01 5.78469872e-01 5.37750497e-02
-9.30793345e-01 -1.12419951e+00 -2.78391182e-01 3.39098781e-01
5.84005117e-01 -2.79051006e-01 -4.13567901e-01 4.55554314e-02]
|
[11.24478530883789, 8.139999389648438]
|
98021d72-7e7a-41b1-b218-c702789f3d15
|
strubert-structure-aware-bert-for-table
|
2203.14278
| null |
https://arxiv.org/abs/2203.14278v1
|
https://arxiv.org/pdf/2203.14278v1.pdf
|
StruBERT: Structure-aware BERT for Table Search and Matching
|
A large amount of information is stored in data tables. Users can search for data tables using a keyword-based query. A table is composed primarily of data values that are organized in rows and columns providing implicit structural information. A table is usually accompanied by secondary information such as the caption, page title, etc., that form the textual information. Understanding the connection between the textual and structural information is an important yet neglected aspect in table retrieval as previous methods treat each source of information independently. In addition, users can search for data tables that are similar to an existing table, and this setting can be seen as a content-based table retrieval. In this paper, we propose StruBERT, a structure-aware BERT model that fuses the textual and structural information of a data table to produce context-aware representations for both textual and tabular content of a data table. StruBERT features are integrated in a new end-to-end neural ranking model to solve three table-related downstream tasks: keyword- and content-based table retrieval, and table similarity. We evaluate our approach using three datasets, and we demonstrate substantial improvements in terms of retrieval and classification metrics over state-of-the-art methods.
|
['Jeff Heflin', 'Brian D. Davison', 'Shuo Zhang', 'Zhiyu Chen', 'Mohamed Trabelsi']
|
2022-03-27
| null | null | null | null |
['table-retrieval', 'table-search']
|
['natural-language-processing', 'natural-language-processing']
|
[ 2.82013342e-02 -1.34405911e-01 -6.53642118e-01 -5.02518058e-01
-1.50785041e+00 -8.83063734e-01 4.67882365e-01 1.24628913e+00
-2.07980916e-01 5.45532465e-01 6.67283595e-01 -9.80332196e-02
-2.68999636e-01 -9.66804445e-01 -9.07694578e-01 -4.82601970e-02
2.27633074e-01 8.45453560e-01 2.56666124e-01 -5.28248906e-01
4.09814566e-01 2.21707270e-01 -1.71013415e+00 1.01119125e+00
8.45187962e-01 1.60489118e+00 1.59397811e-01 2.87878811e-01
-8.40563953e-01 1.07591999e+00 -6.69396520e-01 -5.79720020e-01
4.51086462e-02 -1.75574094e-01 -9.37012732e-01 -2.01637775e-01
7.44654894e-01 -1.97226346e-01 -4.73079890e-01 8.08856010e-01
2.72997439e-01 -3.02671511e-02 4.95953172e-01 -9.59551156e-01
-7.98689783e-01 9.74610448e-01 -5.71741700e-01 1.67242587e-01
5.33773661e-01 -5.18038809e-01 1.62458229e+00 -1.07239425e+00
8.27750742e-01 1.37098420e+00 2.90807933e-01 5.72115853e-02
-1.08068180e+00 -2.43091553e-01 2.68271983e-01 1.69600502e-01
-1.39231050e+00 -5.31994224e-01 7.14018583e-01 -2.93924302e-01
9.63368893e-01 5.32331526e-01 3.29720527e-01 5.98210752e-01
2.85314351e-01 1.03268659e+00 2.16903776e-01 -3.57063234e-01
1.19780511e-01 4.21856612e-01 3.06998581e-01 4.22788620e-01
5.00732481e-01 -6.41947865e-01 -7.43775666e-01 -4.07496206e-02
1.32020935e-01 1.32205009e-01 -3.90521213e-02 -6.37999892e-01
-1.19553995e+00 6.92660570e-01 7.76947200e-01 8.87585655e-02
-3.95406753e-01 -6.19812869e-03 7.94751883e-01 1.72277510e-01
3.75252038e-01 6.19607031e-01 -5.82078755e-01 5.89815415e-02
-6.97212815e-01 5.58664620e-01 9.95898426e-01 1.36741912e+00
8.32141340e-01 -6.00864470e-01 -6.95214450e-01 1.00969207e+00
2.92075515e-01 4.00396615e-01 3.96900356e-01 -5.29218316e-01
1.23340619e+00 1.14072692e+00 1.77159429e-01 -1.25108409e+00
-3.09968799e-01 -3.83987278e-01 -6.36784375e-01 -7.39880025e-01
1.58672914e-01 6.03804410e-01 -7.71464169e-01 1.47833931e+00
2.66916156e-01 -7.83115387e-01 1.04190297e-01 7.06993699e-01
1.26496863e+00 7.75624394e-01 -1.02124326e-01 -8.68963078e-02
1.67414153e+00 -8.51600051e-01 -1.12884247e+00 -3.06883454e-01
7.30516016e-01 -7.61736393e-01 1.08421183e+00 1.49236247e-01
-1.15023685e+00 -3.24134290e-01 -1.08234870e+00 -8.68603826e-01
-1.03697765e+00 3.70887995e-01 3.98818105e-01 2.56348521e-01
-8.27163696e-01 4.78140146e-01 -4.34791535e-01 -3.67702067e-01
2.45363340e-01 9.14915502e-02 -3.06212187e-01 -2.05628350e-01
-1.32765627e+00 6.57340229e-01 4.14606929e-01 1.61151275e-01
-2.63175189e-01 -9.26031411e-01 -1.10012853e+00 3.74624848e-01
1.01992965e+00 -5.79515219e-01 1.18629038e+00 -3.00447848e-02
-5.49003899e-01 9.28670824e-01 -5.28832734e-01 -2.90679634e-01
9.63192526e-03 -2.74679482e-01 -4.26437289e-01 1.35130674e-01
4.89059895e-01 3.64679724e-01 3.46292228e-01 -1.33780718e+00
-5.95633388e-01 -7.17311144e-01 9.30867624e-03 2.95318544e-01
-2.12445438e-01 4.51392345e-02 -1.12121129e+00 -8.18604410e-01
2.98670411e-01 -4.19521391e-01 2.56973326e-01 5.42731844e-02
-7.76565433e-01 -3.27696413e-01 4.50167239e-01 -8.91126215e-01
1.97459483e+00 -1.95844352e+00 -9.18634795e-03 3.71279925e-01
4.28137422e-01 -2.39062056e-01 5.15681412e-03 1.02623320e+00
5.59667908e-02 4.51543689e-01 -2.27349237e-01 -6.35950044e-02
2.67964929e-01 -5.94167002e-02 -8.04543376e-01 -1.54472694e-01
2.77793258e-01 1.19555867e+00 -6.90981269e-01 -5.94467938e-01
-3.25583279e-01 2.59765029e-01 -4.34678793e-01 1.01036109e-01
-4.19768214e-01 -3.46623898e-01 -5.55199921e-01 9.97414887e-01
5.23267210e-01 -5.74087560e-01 1.87632352e-01 -5.08009434e-01
1.15051970e-01 1.07086027e+00 -1.16165173e+00 1.58495426e+00
-3.91277432e-01 2.89921463e-01 -2.75746524e-01 -6.53875053e-01
9.36253965e-01 -1.58792902e-02 4.30877417e-01 -1.11276793e+00
-4.77517582e-02 3.23309779e-01 -4.79240716e-01 -1.54728368e-01
8.94883335e-01 6.14039540e-01 -5.04445434e-01 4.17573780e-01
-2.27039561e-01 -1.11772351e-01 8.13462019e-01 7.18488991e-01
1.02763617e+00 -6.90664873e-02 4.54071611e-01 -2.55130708e-01
7.23673761e-01 1.89752355e-01 1.71913922e-01 8.98775697e-01
4.22098637e-01 6.34345651e-01 7.35041678e-01 -4.50006127e-01
-9.10186946e-01 -9.66114223e-01 -1.28438860e-01 1.17975330e+00
3.70264262e-01 -1.21434474e+00 -5.00460505e-01 -6.18315160e-01
6.85534000e-01 3.75991523e-01 -9.40967023e-01 -4.10549998e-01
-6.45031929e-01 -2.18998075e-01 7.33000189e-02 6.82493746e-01
1.23189539e-01 -9.75237429e-01 -2.50706434e-01 4.09591287e-01
-3.93195480e-01 -1.03038180e+00 -9.04601932e-01 5.52029729e-01
-7.17801929e-01 -1.00141430e+00 -1.56515181e-01 -6.45329475e-01
3.92547190e-01 3.33534092e-01 1.85011661e+00 1.78480119e-01
-2.90623784e-01 2.18270451e-01 -3.98697138e-01 -4.35505033e-01
-2.24695712e-01 3.66853356e-01 -3.03021729e-01 -1.02923691e-01
4.61827070e-01 -1.14466831e-01 -4.26647484e-01 2.66563833e-01
-1.11373949e+00 -2.60290414e-01 6.54257357e-01 6.95410788e-01
9.58557248e-01 -1.15230151e-01 3.07839572e-01 -1.15573466e+00
7.91804314e-01 -4.86884713e-01 -7.43372679e-01 8.07652593e-01
-8.62244010e-01 6.43796444e-01 6.44131243e-01 -8.45794603e-02
-5.56042135e-01 -9.84710604e-02 1.29619196e-01 -2.50004590e-01
3.31300884e-01 1.24702859e+00 -5.42381346e-01 5.51702142e-01
4.83906597e-01 2.54674286e-01 -2.34440789e-01 -9.35643196e-01
3.52312654e-01 5.41744292e-01 5.28262913e-01 -7.31524587e-01
8.25266123e-01 2.58912534e-01 -7.09000081e-02 -2.35960007e-01
-1.31286287e+00 -7.34269559e-01 -7.01512098e-01 8.12198743e-02
5.17682254e-01 -1.04557490e+00 -7.30754673e-01 -8.57401490e-02
-9.55750763e-01 4.09839571e-01 -3.66897583e-01 -8.37796926e-02
-2.70921499e-01 5.52482642e-02 -4.26769733e-01 -5.89960575e-01
-3.25776577e-01 -1.04988754e+00 1.49825859e+00 -1.13922998e-01
-3.04726452e-01 -7.21779168e-01 -1.45340398e-01 4.04516995e-01
3.07264835e-01 -1.55879647e-01 1.57445788e+00 -1.05303657e+00
-9.35470283e-01 -4.54798996e-01 -4.31159347e-01 -2.81898320e-01
2.05152199e-01 -1.51865780e-01 -7.45738029e-01 -8.09286907e-02
-4.66237485e-01 -5.18359244e-01 1.14074039e+00 -2.60574538e-02
1.53912032e+00 -6.86161578e-01 -4.31925535e-01 4.40763652e-01
1.45659840e+00 4.98126268e-01 5.87516606e-01 5.38707316e-01
8.61284435e-01 8.82838905e-01 7.51776338e-01 5.35857141e-01
7.42160618e-01 8.37831616e-01 3.23346645e-01 1.82998627e-01
5.22127934e-02 -8.11109483e-01 -1.47234201e-01 6.72345161e-01
9.33677733e-01 -6.87746942e-01 -9.57799315e-01 4.57015187e-01
-1.80315137e+00 -8.19299340e-01 -1.68803129e-02 2.31084895e+00
1.17666304e+00 2.44599313e-01 6.82621449e-02 3.25991333e-01
4.94273335e-01 2.56580770e-01 -6.95900738e-01 -9.74714234e-02
-2.20964611e-01 -2.17091486e-01 4.03053969e-01 1.02672398e-01
-1.28887463e+00 7.01569796e-01 5.83898020e+00 7.82252848e-01
-7.79784143e-01 -6.51340425e-01 8.14854860e-01 -6.12633117e-02
-6.61571860e-01 -2.22610250e-01 -1.20917451e+00 3.62075597e-01
8.35337162e-01 -6.82404578e-01 3.69482517e-01 7.68195748e-01
-2.21733093e-01 -2.28343815e-01 -1.66990507e+00 1.04457819e+00
2.09164292e-01 -1.58300209e+00 6.95465267e-01 -3.75709310e-02
2.07642600e-01 -5.56750715e-01 2.64503330e-01 4.95936632e-01
-1.04696676e-01 -8.97553921e-01 9.02566731e-01 3.48581910e-01
7.17040122e-01 -7.94430375e-01 6.95719957e-01 -1.77000791e-01
-1.67176247e+00 -4.63158898e-02 -2.27139771e-01 5.09504855e-01
-4.70863730e-01 4.35587406e-01 -5.24169445e-01 6.65817380e-01
1.01881135e+00 9.62130189e-01 -9.89116311e-01 8.80569100e-01
1.90294385e-01 3.71983610e-02 -1.73125863e-01 -2.54691809e-01
1.77668333e-01 -1.06411748e-01 1.77574247e-01 9.74435270e-01
-1.78114306e-02 -1.66032478e-01 2.13701531e-01 1.02901483e+00
-7.09086835e-01 4.30688620e-01 -6.45397186e-01 -3.87380719e-01
6.30552828e-01 1.10291135e+00 -7.61030853e-01 -4.60090578e-01
-3.03403199e-01 4.97707933e-01 5.16540289e-01 2.29659036e-01
-3.49561781e-01 -7.49288738e-01 5.32508016e-01 1.70415014e-01
5.24622381e-01 2.78600246e-01 -4.98352349e-01 -1.18196678e+00
9.28751945e-01 -1.08940160e+00 8.47074032e-01 -7.87788868e-01
-1.24674058e+00 5.60094893e-01 8.29995871e-02 -1.34732962e+00
-4.98839140e-01 -4.76292849e-01 1.48349166e-01 9.91931438e-01
-1.48475218e+00 -8.18457067e-01 -3.18673730e-01 2.90335923e-01
5.11229932e-01 -1.34595186e-01 6.29980028e-01 4.75711673e-01
-4.88591999e-01 8.30562949e-01 4.10834998e-01 5.15438795e-01
9.02360022e-01 -1.49908245e+00 5.65774262e-01 6.86580837e-01
2.54008681e-01 1.10321295e+00 4.27198827e-01 -8.17358792e-01
-2.00077844e+00 -1.06035876e+00 1.38400197e+00 -7.39827156e-01
6.76255584e-01 -8.48885715e-01 -1.30531991e+00 4.18007314e-01
2.92219147e-02 -1.14070393e-01 5.73891044e-01 3.77838075e-01
-9.34029222e-01 -7.65813291e-01 -8.94013464e-01 5.38495839e-01
8.00942361e-01 -8.31729054e-01 -5.68397403e-01 2.74319082e-01
1.05657768e+00 -5.38320720e-01 -7.50628710e-01 2.32893795e-01
6.23059392e-01 -4.16050494e-01 1.06879723e+00 -6.97727621e-01
7.60983407e-01 -1.95281729e-01 -4.78148043e-01 -1.01572847e+00
-9.84850377e-02 -4.35642123e-01 -4.64536637e-01 1.33372688e+00
8.74668300e-01 -1.89933777e-01 7.17752695e-01 8.00112426e-01
4.56824824e-02 -8.59606683e-01 -5.58911324e-01 -6.19848728e-01
-6.62136972e-02 -1.22473776e-01 9.23427641e-01 6.25514507e-01
2.75890142e-01 7.11187243e-01 -3.42484228e-02 -1.72949761e-01
3.38888675e-01 4.91394907e-01 5.82231641e-01 -1.62095714e+00
3.71269852e-01 -5.68628371e-01 -1.04219921e-01 -1.14165092e+00
-9.08930451e-02 -1.19692624e+00 7.19604939e-02 -1.86910582e+00
4.68694121e-01 -3.65502268e-01 -4.29307222e-01 5.00465274e-01
-3.37667286e-01 -2.06034750e-01 2.44961217e-01 3.89602482e-01
-8.95387590e-01 4.37580734e-01 1.03832936e+00 -5.20164967e-01
-7.54599124e-02 -3.24183345e-01 -1.24350476e+00 -8.31855461e-02
2.87248254e-01 -5.94349444e-01 -5.04096866e-01 -4.07659620e-01
8.25070441e-01 4.80221629e-01 -2.04576954e-01 -7.98320711e-01
5.53330719e-01 -3.14028934e-02 5.72133958e-01 -1.28038430e+00
2.70759463e-01 -9.70777929e-01 -4.00469005e-01 -3.39816511e-02
-1.01788139e+00 6.16314888e-01 3.91181797e-01 5.92571735e-01
-6.66148186e-01 1.12599060e-02 2.63993979e-01 -1.41293138e-01
-4.44264382e-01 4.52236295e-01 -1.85941592e-01 4.71865982e-01
2.51776338e-01 1.60723120e-01 -4.67237234e-01 -5.09311497e-01
-2.09537357e-01 5.13294518e-01 2.69050151e-01 7.94554889e-01
8.15807283e-01 -1.45263970e+00 -5.42762399e-01 2.23783717e-01
8.20942819e-01 6.76342770e-02 -2.04600915e-01 3.85934263e-01
-3.43976498e-01 9.02749419e-01 5.95809035e-02 -3.77566427e-01
-1.03776765e+00 9.74942207e-01 9.84683353e-03 -6.83205962e-01
-4.43538547e-01 6.27064705e-01 2.09193543e-01 -5.22285283e-01
6.81058109e-01 -7.36551464e-01 -4.18338805e-01 5.26664495e-01
8.63834679e-01 -1.37830809e-01 7.18178272e-01 -2.52606928e-01
-4.53401715e-01 4.04586077e-01 -6.40975714e-01 1.54693238e-02
1.10576272e+00 -2.33605951e-01 -3.58849287e-01 7.95630515e-01
1.54500186e+00 1.81558549e-01 -3.65269423e-01 -6.32026196e-01
6.44048989e-01 -5.28799891e-01 -8.56756940e-02 -1.27200091e+00
-8.96728396e-01 6.90989912e-01 1.01270780e-01 1.57288998e-01
1.15153873e+00 2.25327194e-01 7.85135865e-01 1.11974895e+00
1.94770500e-01 -1.00759602e+00 6.97389171e-02 6.73546612e-01
1.10939896e+00 -1.45377839e+00 7.11368993e-02 -6.25121534e-01
-5.99219501e-01 1.13395011e+00 5.76811373e-01 4.54852939e-01
4.66565192e-01 3.64293337e-01 1.07099459e-01 -4.44917887e-01
-1.22305989e+00 -3.27632219e-01 9.47040737e-01 2.45002195e-01
7.24160016e-01 -4.85490590e-01 -3.64925265e-02 6.83257937e-01
-1.74333975e-01 -3.71500909e-01 2.89472491e-01 1.24396431e+00
-3.80576283e-01 -1.05104208e+00 -2.28288561e-01 9.10946727e-01
-4.49344575e-01 -3.43516141e-01 -8.54394853e-01 6.49628222e-01
-5.91723204e-01 9.55048859e-01 2.93190241e-01 -3.37670833e-01
5.42118371e-01 1.93764731e-01 -3.42197791e-02 -6.54464960e-01
-8.03861380e-01 -6.02742322e-02 7.44550824e-02 -7.18267202e-01
2.70585179e-01 -3.94731164e-01 -1.08947396e+00 -2.13151276e-01
1.46934599e-01 5.69322169e-01 6.35247350e-01 5.96708357e-01
5.05209744e-01 5.48092723e-01 6.16998494e-01 -1.69287652e-01
-5.02447605e-01 -7.78202415e-01 -3.90142977e-01 4.94468778e-01
5.69082141e-01 -5.55228055e-01 -6.93746880e-02 -1.17063351e-01]
|
[9.714163780212402, 7.8971099853515625]
|
e0ee4231-9dc1-4de5-ac74-7bb7c348c9b6
|
rcp-recurrent-closest-point-for-point-cloud
| null | null |
http://openaccess.thecvf.com//content/CVPR2022/html/Gu_RCP_Recurrent_Closest_Point_for_Point_Cloud_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Gu_RCP_Recurrent_Closest_Point_for_Point_Cloud_CVPR_2022_paper.pdf
|
RCP: Recurrent Closest Point for Point Cloud
|
3D motion estimation including scene flow and point cloud registration has drawn increasing interest. Inspired by 2D flow estimation, recent methods employ deep neural networks to construct the cost volume for estimating accurate 3D flow. However, these methods are limited by the fact that it is difficult to define a search window on point clouds because of the irregular data structure. In this paper, we avoid this irregularity by a simple yet effective method. We decompose the problem into two interlaced stages, where the 3D flows are optimized point-wisely at the first stage and then globally regularized in a recurrent network at the second stage. Therefore, the recurrent network only receives the regular point-wise information as the input. In the experiments, we evaluate the proposed method on both the 3D scene flow estimation and the point cloud registration task. For 3D scene flow estimation, we make comparisons on the widely used FlyingThings3D and KITTI datasets. For point cloud registration, we follow previous works and evaluate the data pairs with large pose and partially overlapping from ModelNet40. The results show that our method outperforms the previous method and achieves a new state-of-the-art performance on both 3D scene flow estimation and point cloud registration, which demonstrates the superiority of the proposed zero-order method on irregular point cloud data. Our source code is available at https://github.com/gxd1994/RCP.
|
['Ping Tan', 'Siyu Zhu', 'Zuozhuo Dai', 'Weihao Yuan', 'Chengzhou Tang', 'Xiaodong Gu']
|
2022-01-01
| null | null | null |
cvpr-2022-1
|
['point-cloud-registration', 'scene-flow-estimation']
|
['computer-vision', 'computer-vision']
|
[-4.41176325e-01 -7.33192563e-01 -1.74739920e-02 -1.53791577e-01
-2.41436064e-01 -4.38918322e-01 3.85791093e-01 -1.22994736e-01
-5.15054703e-01 3.69154990e-01 -2.96941716e-02 -2.08414674e-01
-8.15085880e-03 -8.43840718e-01 -5.72137833e-01 -5.48477054e-01
-2.41346434e-01 3.83748263e-01 6.40858471e-01 -2.17172220e-01
4.23774093e-01 8.67161214e-01 -1.58720958e+00 -3.50573510e-01
1.04395509e+00 1.08325374e+00 2.28250012e-01 5.80039442e-01
-4.45111930e-01 6.47845209e-01 -4.53335375e-01 -3.41156987e-03
6.57948017e-01 -1.52668685e-01 -6.82177484e-01 -4.19999212e-02
6.51956201e-01 -5.49321115e-01 -6.10718429e-01 8.44856381e-01
5.44142842e-01 4.97309387e-01 2.78499275e-01 -1.26045859e+00
-8.34222734e-02 -1.76442474e-01 -8.85983288e-01 3.84932190e-01
1.07757345e-01 3.47630441e-01 5.84585071e-01 -9.18395340e-01
5.55087984e-01 1.23333502e+00 6.84135973e-01 2.58995831e-01
-7.60758758e-01 -8.40681195e-01 2.55205065e-01 1.78142428e-01
-1.45104122e+00 -2.72943467e-01 9.61800456e-01 -6.22266293e-01
1.03554881e+00 -6.77490085e-02 9.91786599e-01 3.96092415e-01
1.06809691e-01 5.34329951e-01 6.07317567e-01 2.06441492e-01
5.01330346e-02 -4.70407158e-01 2.77054086e-02 8.15670490e-01
1.50575548e-01 3.18325818e-01 -3.55775833e-01 5.94097637e-02
1.28580940e+00 2.03602672e-01 -4.38151360e-01 -3.95053029e-01
-1.49951148e+00 7.89231777e-01 9.07910407e-01 1.98227823e-01
-3.31056982e-01 3.89107645e-01 4.19373304e-01 7.03927083e-03
7.53934801e-01 1.54459730e-01 -2.96108037e-01 -2.71195412e-01
-9.40037668e-01 4.07123059e-01 7.26934552e-01 1.04714262e+00
9.99587476e-01 1.74912915e-01 1.26023069e-01 5.62755764e-01
5.22060215e-01 6.59277320e-01 3.06497604e-01 -1.21064532e+00
5.85243881e-01 6.35934711e-01 8.97945464e-02 -1.46784425e+00
-3.90673816e-01 -4.22802031e-01 -1.02978110e+00 3.11533988e-01
5.00029027e-01 -7.32748061e-02 -7.41945267e-01 1.34842002e+00
5.86959124e-01 8.26937318e-01 -1.60995007e-01 1.32802081e+00
1.04122448e+00 9.53459620e-01 -1.86954767e-01 -4.96667475e-02
9.92190242e-01 -1.07750320e+00 -4.82885480e-01 -9.02900770e-02
6.33581400e-01 -7.86143482e-01 7.09339023e-01 -8.65285546e-02
-1.05502927e+00 -6.96342349e-01 -9.67922270e-01 -2.63587832e-01
-6.80895820e-02 4.22715321e-02 7.39924073e-01 7.96526447e-02
-1.09803164e+00 6.61502898e-01 -1.12825990e+00 -2.70757437e-01
5.41416347e-01 2.73593545e-01 -3.00552666e-01 6.43183440e-02
-1.04048550e+00 6.12113416e-01 2.05126569e-01 5.10487199e-01
-6.15131199e-01 -9.31190372e-01 -8.70437205e-01 -1.40837416e-01
1.29353955e-01 -9.99755025e-01 9.68497157e-01 -3.65186512e-01
-1.55464041e+00 9.00898039e-01 -3.03000838e-01 -3.59152645e-01
6.24442220e-01 -2.08513469e-01 1.16878003e-01 1.40769288e-01
2.02861920e-01 7.94586599e-01 5.68499029e-01 -9.56074536e-01
-7.67614901e-01 -3.36869597e-01 1.19820178e-01 4.07936096e-01
1.94067836e-01 -2.75520265e-01 -6.93800092e-01 -3.41186702e-01
3.64133447e-01 -9.37240839e-01 -3.56543034e-01 3.49924982e-01
-2.22730502e-01 -2.84396499e-01 9.84426916e-01 -4.08144832e-01
1.03068244e+00 -2.05343175e+00 1.11488551e-01 -1.12966627e-01
3.84021223e-01 3.46376330e-01 -2.00158767e-02 5.91942370e-02
3.60595696e-02 1.01870187e-01 -5.02809227e-01 -6.14475012e-01
-2.25835145e-01 6.03638627e-02 -3.56628478e-01 8.31539214e-01
2.33729944e-01 7.99549460e-01 -9.11203265e-01 -5.26323378e-01
7.52812743e-01 6.36017144e-01 -7.05707312e-01 3.20290774e-01
-1.39769102e-02 7.44550347e-01 -7.35992730e-01 5.20717859e-01
1.08802211e+00 -2.90851057e-01 -5.40606976e-01 -2.84261703e-01
-7.93052733e-01 3.52112442e-01 -1.34996700e+00 2.18004704e+00
-4.41302150e-01 7.16604471e-01 -5.34776449e-02 -8.69612694e-01
1.05001009e+00 3.05125192e-02 8.32410395e-01 -4.00365174e-01
2.69216776e-01 3.18049908e-01 -1.80664226e-01 -4.15730000e-01
6.32052481e-01 5.22476472e-02 1.92434371e-01 1.89429715e-01
-1.01252884e-01 -5.28563917e-01 1.52192652e-01 -1.59240775e-02
9.04295087e-01 4.61403936e-01 1.30061746e-01 -2.26225078e-01
8.07357430e-01 3.66009593e-01 8.24887931e-01 4.08501804e-01
-4.70665514e-01 8.38515639e-01 2.41896048e-01 -7.62759387e-01
-8.82980764e-01 -8.07410955e-01 -2.00322345e-01 2.07915947e-01
8.50224853e-01 -3.08559358e-01 -3.60175997e-01 -5.07698596e-01
4.45735939e-02 1.50196955e-01 -2.75211602e-01 1.93797082e-01
-8.90094519e-01 -6.28903925e-01 3.90340447e-01 2.73026496e-01
1.17615211e+00 -9.27633107e-01 -9.27231371e-01 2.22995579e-01
-4.41737175e-01 -1.25746834e+00 -6.37470305e-01 -3.98024976e-01
-1.19891298e+00 -9.98100281e-01 -7.60403454e-01 -7.45580018e-01
5.38529932e-01 7.80839801e-01 1.12475348e+00 4.43563640e-01
-6.09114133e-02 4.96003963e-02 -1.69330463e-01 -4.88370545e-02
7.49797970e-02 2.54087687e-01 -3.33289914e-02 -1.17714331e-01
2.74137884e-01 -7.45951414e-01 -9.22971129e-01 4.96791303e-01
-6.95951462e-01 1.13907844e-01 2.34129995e-01 4.27714080e-01
7.12483644e-01 -1.14589967e-01 2.83455197e-02 -2.05628753e-01
1.99370086e-01 -3.28672320e-01 -1.05013990e+00 -2.64571041e-01
-1.14064254e-01 -1.44079641e-01 4.53805596e-01 -1.30100667e-01
-9.49927568e-01 2.41202652e-01 -3.68591845e-01 -8.99641275e-01
-1.86892226e-01 2.18088374e-01 1.31240189e-01 -4.91642267e-01
4.00874346e-01 1.12750523e-01 2.46109301e-03 -2.69403785e-01
1.46558821e-01 2.88941383e-01 4.23099667e-01 -2.86657453e-01
1.13653326e+00 8.29305172e-01 2.43656605e-01 -8.79820645e-01
-6.06715143e-01 -6.72701657e-01 -8.60379696e-01 -4.31686789e-01
1.02749896e+00 -1.05066979e+00 -1.14119267e+00 7.23338187e-01
-1.55437171e+00 -2.88534462e-01 -1.87299162e-01 8.27310681e-01
-5.51672339e-01 4.93387550e-01 -5.55622876e-01 -5.61994910e-01
-4.04515922e-01 -1.37694037e+00 1.16266012e+00 3.44237536e-01
1.93699047e-01 -1.06296384e+00 2.71979660e-01 4.13293019e-02
3.98176730e-01 4.64514941e-01 1.82037026e-01 6.59763142e-02
-1.19814336e+00 1.15313306e-01 -3.70556206e-01 4.13728729e-02
1.67248979e-01 1.24686830e-01 -8.57628167e-01 -9.69238952e-02
3.01959157e-01 -2.61383094e-02 8.00563157e-01 6.55808389e-01
9.95984316e-01 1.47258416e-01 -1.44860640e-01 1.43021274e+00
1.48554778e+00 1.63672730e-01 5.69790542e-01 4.18493390e-01
1.07888603e+00 6.00060821e-01 7.07964420e-01 3.71856153e-01
5.12786865e-01 7.07237601e-01 6.69661939e-01 -1.72505319e-01
-1.21733770e-01 -2.97780424e-01 4.50337641e-02 9.93531644e-01
-3.86429012e-01 -1.09466329e-01 -9.78407979e-01 4.48388875e-01
-1.90294492e+00 -9.19039369e-01 -4.79573250e-01 2.22082472e+00
3.70706856e-01 -6.61666691e-02 -8.02794322e-02 -1.83504149e-01
6.84076965e-01 5.68381965e-01 -6.85141265e-01 9.62749347e-02
-6.72109649e-02 -9.64864269e-02 6.34688675e-01 7.08460033e-01
-1.18232262e+00 1.19484675e+00 4.92450619e+00 6.56337082e-01
-1.33817077e+00 4.16119630e-03 4.31178302e-01 -1.70064047e-01
-1.67844556e-02 1.10726289e-01 -9.07610118e-01 5.12217939e-01
3.91086936e-01 -1.86485767e-01 3.81997287e-01 5.20620286e-01
4.68041718e-01 -7.37013519e-02 -7.54589319e-01 1.33410788e+00
-1.54141456e-01 -1.57884634e+00 -9.90703469e-04 -3.34043913e-02
5.92015386e-01 5.15079916e-01 -2.53079623e-01 -2.89936550e-02
7.14620948e-02 -6.63343728e-01 5.31873226e-01 4.37859207e-01
6.10117376e-01 -6.82614267e-01 5.40359139e-01 3.72533441e-01
-1.64383519e+00 3.50348443e-01 -5.45349240e-01 -2.46044159e-01
5.73009551e-01 5.91604352e-01 -3.46601814e-01 7.22005427e-01
9.49638903e-01 1.45713031e+00 -2.62583703e-01 1.46417618e+00
-7.59788975e-02 1.51760533e-01 -6.16927683e-01 1.38826028e-01
4.56364363e-01 -7.31422126e-01 8.71872723e-01 8.40035737e-01
5.88336527e-01 2.04924718e-01 1.70008466e-01 1.20088065e+00
1.15503186e-04 -6.46742061e-02 -8.38116705e-01 5.06467223e-01
3.75587970e-01 1.29990315e+00 -8.49531174e-01 -2.20662579e-01
-3.49141747e-01 5.82810462e-01 1.62259132e-01 3.30078661e-01
-8.40327024e-01 -4.58263457e-01 1.00199199e+00 -1.12444407e-03
5.64654768e-02 -6.58510208e-01 -1.79238290e-01 -1.46145904e+00
8.30770284e-02 -1.76974595e-01 9.70303640e-02 -7.02995181e-01
-1.11924696e+00 8.02904069e-01 1.33526891e-01 -1.77455115e+00
-1.11252092e-01 -3.55722487e-01 -7.59249568e-01 1.02233195e+00
-1.94289446e+00 -7.18316615e-01 -8.37561727e-01 6.03108823e-01
5.45106590e-01 2.23328605e-01 1.82438999e-01 5.18385768e-01
-5.65740049e-01 -2.72539966e-02 -3.32786083e-01 1.77750856e-01
4.53805894e-01 -8.84197414e-01 8.83472621e-01 1.03012276e+00
-1.34424478e-01 3.81682158e-01 3.38004589e-01 -7.13229537e-01
-1.31099153e+00 -1.16110659e+00 7.97738194e-01 -3.59322965e-01
5.36004722e-01 -2.42029771e-01 -1.11431837e+00 5.25951207e-01
-9.77423042e-02 3.42295617e-01 1.30353402e-03 -4.63756889e-01
1.96990356e-01 -1.98615327e-01 -9.95958507e-01 4.14432526e-01
1.46852696e+00 -1.57872260e-01 -3.48435223e-01 2.24015459e-01
9.54933167e-01 -8.55079412e-01 -8.70318353e-01 6.19639337e-01
2.76224166e-01 -1.12528098e+00 1.10535586e+00 8.21124837e-02
3.61302942e-01 -7.48170793e-01 1.86792418e-01 -1.25188601e+00
-1.84367344e-01 -7.35040426e-01 6.18152171e-02 8.30473185e-01
2.67715063e-02 -8.94115388e-01 9.07427847e-01 3.20064604e-01
-4.07839417e-01 -5.48704028e-01 -1.04590225e+00 -6.73208296e-01
-5.79565763e-02 -4.07588273e-01 7.84873545e-01 9.55764055e-01
-6.34946823e-01 2.21849859e-01 -1.84189081e-01 2.80924141e-01
7.57450700e-01 4.32010382e-01 1.04826617e+00 -1.25711524e+00
2.11713612e-01 -5.69523931e-01 -5.49881637e-01 -1.82100606e+00
2.10146949e-01 -8.37677836e-01 1.03180192e-01 -1.47853303e+00
-4.47543532e-01 -7.87875056e-01 1.70077622e-01 8.00186172e-02
-1.02752388e-01 1.00425251e-01 4.02489245e-01 6.04171872e-01
-3.96618158e-01 8.95938218e-01 1.64144814e+00 6.02548420e-02
-6.30267859e-01 6.34799004e-02 -9.57245007e-02 7.29009986e-01
7.96972096e-01 -3.39506447e-01 -4.66766030e-01 -9.72165763e-01
-1.94910690e-02 2.05600366e-01 5.61396062e-01 -1.24353027e+00
4.43653464e-01 -9.61901471e-02 2.03884572e-01 -1.05576479e+00
4.74941194e-01 -8.24642420e-01 4.89340611e-02 5.13055921e-01
1.05367348e-01 3.91112328e-01 3.95765066e-01 4.50453252e-01
-3.98312718e-01 8.74811709e-02 6.95812285e-01 -2.37138361e-01
-9.18562710e-01 1.16915870e+00 1.01744771e-01 1.80462822e-01
8.19290280e-01 -2.98965514e-01 -3.41472954e-01 -2.12675750e-01
-1.70083314e-01 4.34877902e-01 6.35506988e-01 5.09832501e-01
8.47117007e-01 -1.36342752e+00 -7.16751039e-01 4.38524991e-01
-1.04940087e-01 9.17627335e-01 4.28738952e-01 9.79352117e-01
-1.19191480e+00 3.49448949e-01 -2.54245490e-01 -1.13082135e+00
-7.89945543e-01 2.54492521e-01 6.67951167e-01 -4.91338819e-02
-8.22263539e-01 7.04045355e-01 4.22448099e-01 -6.18800163e-01
5.92137687e-03 -6.15617931e-01 -2.92460442e-01 -2.32157722e-01
2.93097675e-01 4.99527812e-01 -1.27539694e-01 -9.38781500e-01
-4.18654531e-01 1.28126228e+00 3.69951248e-01 1.62325770e-01
1.24855709e+00 -1.50276870e-01 -2.16181532e-01 3.61050844e-01
1.52222431e+00 -3.33190769e-01 -1.55903971e+00 -8.00973848e-02
-4.01480913e-01 -9.44646358e-01 2.33032703e-01 1.39432386e-01
-1.57893670e+00 1.15905035e+00 5.64452350e-01 4.72090533e-03
1.11136496e+00 -2.80387759e-01 1.05147684e+00 1.90003410e-01
3.73700798e-01 -4.81804222e-01 -3.21350008e-01 8.46067369e-01
7.58181751e-01 -1.31016171e+00 1.87290795e-02 -5.81391096e-01
-2.78733462e-01 1.10403001e+00 8.18032265e-01 -5.95557630e-01
7.37847984e-01 -6.01441227e-02 5.56506515e-02 -2.86382139e-01
-4.74459141e-01 -1.85788095e-01 1.72644973e-01 3.85609418e-01
1.63107768e-01 -3.03160578e-01 -1.23431072e-01 -2.54991651e-01
-2.31184736e-01 2.12555617e-01 3.61614883e-01 7.84246147e-01
-2.37616241e-01 -8.02125096e-01 -2.96263576e-01 1.16888322e-01
-1.08174726e-01 6.99847788e-02 1.02966920e-01 8.86241734e-01
3.25168110e-02 7.63298631e-01 5.31234384e-01 -3.83705407e-01
5.79486251e-01 -5.66929042e-01 2.77430683e-01 -2.49367833e-01
-5.36135137e-01 -7.58721307e-02 -3.03626329e-01 -1.00347853e+00
-8.41801226e-01 -5.80151618e-01 -1.35229015e+00 -6.09828770e-01
-1.18362568e-01 1.19736291e-01 6.24236584e-01 7.01250017e-01
5.73344111e-01 3.85404885e-01 8.52422416e-01 -1.37807310e+00
1.74711831e-02 -6.82224512e-01 -2.46791244e-01 3.49040210e-01
5.33611715e-01 -8.36297631e-01 -6.71887636e-01 -9.15542692e-02]
|
[8.5460786819458, -2.0629398822784424]
|
387b5218-f90c-46f5-b843-8908e7810423
|
unsupervised-boundary-aware-language-model
|
2210.15231
| null |
https://arxiv.org/abs/2210.15231v1
|
https://arxiv.org/pdf/2210.15231v1.pdf
|
Unsupervised Boundary-Aware Language Model Pretraining for Chinese Sequence Labeling
|
Boundary information is critical for various Chinese language processing tasks, such as word segmentation, part-of-speech tagging, and named entity recognition. Previous studies usually resorted to the use of a high-quality external lexicon, where lexicon items can offer explicit boundary information. However, to ensure the quality of the lexicon, great human effort is always necessary, which has been generally ignored. In this work, we suggest unsupervised statistical boundary information instead, and propose an architecture to encode the information directly into pre-trained language models, resulting in Boundary-Aware BERT (BABERT). We apply BABERT for feature induction of Chinese sequence labeling tasks. Experimental results on ten benchmarks of Chinese sequence labeling demonstrate that BABERT can provide consistent improvements on all datasets. In addition, our method can complement previous supervised lexicon exploration, where further improvements can be achieved when integrated with external lexicon information.
|
['Min Zhang', 'Meishan Zhang', 'Pengjun Xie', 'Yanzhao Zhang', 'Dingkun Long', 'Peijie Jiang']
|
2022-10-27
| null | null | null | null |
['chinese-named-entity-recognition', 'chinese-word-segmentation']
|
['natural-language-processing', 'natural-language-processing']
|
[ 5.46972007e-02 -1.29566684e-01 -3.35571140e-01 -5.11418104e-01
-8.50220680e-01 -7.38050520e-01 1.96022287e-01 5.29069565e-02
-6.43513262e-01 6.54888570e-01 2.60478377e-01 -5.93403399e-01
5.36562145e-01 -7.56804883e-01 -4.29723233e-01 -3.89939576e-01
3.57712865e-01 3.39316875e-01 3.46189708e-01 -1.37005001e-01
2.54337341e-01 2.52128653e-02 -8.48733187e-01 1.79383159e-01
1.15536106e+00 8.04501772e-01 5.83947599e-01 4.38056476e-02
-8.39347064e-01 5.07872641e-01 -5.55005968e-01 -3.42745513e-01
-8.93163159e-02 -5.07718384e-01 -9.31934118e-01 2.71163285e-01
-4.65087891e-01 4.28150631e-02 3.42453003e-01 1.19923341e+00
1.06308587e-01 1.60870090e-01 4.02867526e-01 -5.84786773e-01
-5.98800182e-01 9.94048059e-01 -4.03820217e-01 -2.40448490e-02
3.02798569e-01 9.75236204e-03 1.23953009e+00 -9.68259037e-01
8.18529487e-01 9.14353132e-01 5.37242651e-01 5.53518057e-01
-1.06440055e+00 -6.73299670e-01 7.18372405e-01 2.29146108e-02
-1.55254698e+00 -1.96557030e-01 7.14279115e-01 -3.46086681e-01
1.10157788e+00 -5.31360246e-02 5.76461077e-01 5.66503108e-01
-2.00570732e-01 1.26676381e+00 7.97505438e-01 -6.99913979e-01
1.57028332e-01 6.89708516e-02 4.52982306e-01 6.57847703e-01
2.00093567e-01 -4.87275571e-01 -2.36145929e-01 1.39041245e-01
5.55083156e-01 -2.23942697e-01 -1.69077545e-01 -3.51651050e-02
-1.07984054e+00 7.02523172e-01 5.21374345e-02 5.06586254e-01
-2.44465604e-01 -2.36416563e-01 4.74169642e-01 4.93400805e-02
5.44232309e-01 6.13622904e-01 -7.48106539e-01 -3.21776420e-01
-7.02845812e-01 -2.37629503e-01 7.43257761e-01 1.38987124e+00
9.89255369e-01 -5.08307442e-02 -4.77287136e-02 8.78799975e-01
3.36467147e-01 2.66211718e-01 5.70545375e-01 -5.89053333e-01
6.60244942e-01 9.11605299e-01 1.30099922e-01 -4.43309218e-01
-3.74971360e-01 -1.71135738e-01 -4.16609585e-01 -4.91574019e-01
3.30117315e-01 -3.57220560e-01 -1.11456609e+00 1.58912897e+00
2.20029935e-01 9.23945978e-02 8.28853771e-02 7.55163133e-01
6.51332557e-01 5.66135466e-01 2.32790083e-01 -2.71006942e-01
1.50969172e+00 -9.65583563e-01 -9.71539915e-01 -4.47454900e-01
1.04631281e+00 -7.97204792e-01 1.37480080e+00 4.16430384e-01
-6.99718714e-01 -4.18808490e-01 -7.65762866e-01 4.18106206e-02
-3.56988966e-01 2.59631008e-01 7.74008393e-01 8.19228351e-01
-8.12781394e-01 2.41910636e-01 -1.04447854e+00 -1.69852316e-01
3.70358020e-01 2.02514425e-01 -2.58859336e-01 -1.91604555e-01
-1.37394464e+00 5.46403825e-01 8.30793321e-01 2.76286602e-01
-5.44462264e-01 -2.01384321e-01 -1.12630916e+00 -2.76543340e-03
7.36786783e-01 -5.74401841e-02 1.36094511e+00 -8.28298748e-01
-1.62646687e+00 6.99946165e-01 -3.88291955e-01 -1.59849226e-01
1.60560068e-02 -3.94105047e-01 -4.00144011e-01 -5.83058037e-02
1.55376866e-01 7.05060005e-01 2.04125687e-01 -1.08175921e+00
-6.26442194e-01 5.24680577e-02 -1.19955249e-01 1.45160213e-01
-5.29670000e-01 5.43960631e-01 -1.17060232e+00 -6.63710535e-01
2.44258121e-01 -8.12408984e-01 -6.33677185e-01 -7.35732853e-01
-4.16200548e-01 -4.73218650e-01 4.26030040e-01 -7.55381167e-01
1.66425431e+00 -2.09140253e+00 -2.46269047e-01 1.41640127e-01
-2.06892312e-01 4.61450398e-01 -1.44413427e-01 2.17374101e-01
3.04884821e-01 5.97289681e-01 -3.99811357e-01 -3.62732768e-01
-7.72729218e-02 3.32549989e-01 -1.37483731e-01 8.92135203e-02
5.87156832e-01 1.19257343e+00 -9.32179689e-01 -7.19242990e-01
-1.31631687e-01 4.64031473e-02 -5.74787915e-01 1.63170561e-01
-5.72585344e-01 5.99054873e-01 -8.47898662e-01 8.12799811e-01
5.69385290e-01 -3.27908546e-01 4.85830963e-01 2.74417698e-01
-1.90215826e-01 8.46908152e-01 -1.00622356e+00 1.72987890e+00
-4.90214616e-01 2.22601190e-01 -6.79860786e-02 -8.71705294e-01
1.01610637e+00 3.96727532e-01 1.43195733e-01 -5.94359040e-01
2.26833016e-01 2.57095069e-01 1.19131930e-01 -2.41192430e-01
6.11907721e-01 -6.68143779e-02 -5.29717267e-01 4.94269818e-01
-1.58891305e-01 -8.36395174e-02 4.21165824e-01 1.27389193e-01
8.81430030e-01 4.38124269e-01 5.19404292e-01 -1.69355363e-01
5.94372749e-01 3.98815334e-01 1.15401161e+00 5.20830333e-01
-2.75039315e-01 6.02267206e-01 5.06796122e-01 -9.57776830e-02
-7.70670533e-01 -5.85169256e-01 -5.60668446e-02 1.15285122e+00
1.96933880e-01 -6.52149200e-01 -1.00769472e+00 -1.09776962e+00
-4.14677143e-01 7.03894079e-01 -2.69747704e-01 1.45576701e-01
-1.00361967e+00 -7.74190307e-01 7.07803071e-01 9.07048583e-01
3.59091312e-01 -1.35072601e+00 -7.15562776e-02 5.02304554e-01
-3.31847459e-01 -1.35607111e+00 -8.35809886e-01 3.71190250e-01
-9.08913374e-01 -7.37202585e-01 -4.36048210e-01 -1.01495707e+00
7.09518969e-01 1.39451027e-01 9.95065331e-01 3.08370888e-01
9.39118192e-02 -1.29540965e-01 -8.30276489e-01 -2.38999069e-01
-3.50346029e-01 3.76755625e-01 -1.67917177e-01 -1.73400834e-01
7.94108987e-01 -9.87777188e-02 -2.41189480e-01 4.93073970e-01
-7.71381319e-01 5.01086712e-02 5.32470107e-01 9.30019200e-01
6.73238993e-01 -1.97037891e-01 6.53318048e-01 -1.26768529e+00
4.74769145e-01 -1.87069163e-01 -7.38932073e-01 4.41841036e-01
-5.96568108e-01 1.75760522e-01 6.99444950e-01 -4.58385289e-01
-1.60688984e+00 2.61019230e-01 -4.78037000e-01 1.06326103e-01
-3.73564452e-01 1.01508236e+00 -7.66982138e-01 2.02963695e-01
1.39100090e-01 3.27550918e-01 -3.79412740e-01 -7.89051831e-01
3.88740450e-01 8.33822012e-01 2.25709096e-01 -7.50049412e-01
3.39602232e-01 1.21045835e-01 -6.35388017e-01 -6.54886782e-01
-9.38679755e-01 -7.70482481e-01 -1.01331758e+00 2.38938674e-01
9.01347697e-01 -9.06433105e-01 -1.70011804e-01 5.11105418e-01
-1.05447769e+00 -5.21554708e-01 2.17703387e-01 5.16356051e-01
-1.19504191e-01 6.80393755e-01 -8.19288969e-01 -7.42894232e-01
-9.61887613e-02 -1.23636973e+00 8.91213357e-01 2.59432375e-01
-2.88763613e-01 -1.14993739e+00 -1.43416479e-01 3.55978996e-01
1.49424732e-01 -3.51613730e-01 8.21620405e-01 -1.00925326e+00
-6.79842234e-01 -1.61915720e-01 -1.16639920e-01 2.78678149e-01
3.46107930e-01 -1.98136955e-01 -7.87345529e-01 -1.55690815e-02
-2.86440164e-01 -3.75866443e-01 8.60808134e-01 9.93334800e-02
1.13618648e+00 7.47729912e-02 -5.38143337e-01 5.62644958e-01
1.20493293e+00 5.46986401e-01 5.41287720e-01 4.24245805e-01
8.16873729e-01 5.53237975e-01 1.03028762e+00 3.09691936e-01
5.95273376e-01 4.61441219e-01 -9.76352021e-02 -7.61505440e-02
7.17613995e-02 -2.81960398e-01 3.54932815e-01 1.38490403e+00
2.09336370e-01 -3.25632662e-01 -1.26455986e+00 7.41313338e-01
-1.63649476e+00 -2.83363819e-01 7.05535337e-02 1.85416448e+00
1.37851024e+00 3.26142192e-01 -2.64209718e-01 -6.41676262e-02
7.72340059e-01 -1.12360381e-01 -3.55244547e-01 -1.69126660e-01
-1.11618951e-01 1.47027045e-01 3.02639216e-01 4.31274980e-01
-1.19970560e+00 1.74613726e+00 5.92709923e+00 9.89234746e-01
-1.11691427e+00 1.34655103e-01 5.88351965e-01 4.99927968e-01
-5.54647684e-01 3.32861036e-01 -1.34788585e+00 2.63854295e-01
7.10510612e-01 4.04880494e-02 7.96792209e-02 8.86010110e-01
8.73396322e-02 -2.22757861e-01 -7.70817816e-01 5.04472315e-01
-1.28483281e-01 -9.50044453e-01 -9.75938961e-02 -3.31421494e-02
7.75112331e-01 -2.43388931e-03 -4.39259022e-01 5.88378966e-01
6.35443151e-01 -7.18960345e-01 5.03184319e-01 7.28097185e-02
8.11906040e-01 -7.30013251e-01 8.86240721e-01 5.47940433e-01
-1.41972685e+00 3.38776082e-01 -3.58439505e-01 -6.23241030e-02
4.96012509e-01 4.64232355e-01 -1.05993187e+00 5.91989100e-01
3.48239690e-01 6.13615274e-01 -4.50266361e-01 1.03145707e+00
-7.11198449e-01 1.25550246e+00 -2.23326355e-01 -3.86367291e-01
4.47571665e-01 -2.72477895e-01 1.49205819e-01 1.51998091e+00
-3.79239731e-02 2.72100717e-01 6.84862077e-01 5.95228076e-01
-8.13979134e-02 5.54565191e-01 -3.12770367e-01 -4.28735286e-01
5.58769763e-01 1.21012938e+00 -1.38816535e+00 -4.89271522e-01
-7.37114131e-01 8.29437792e-01 5.20640314e-01 3.25114727e-01
-5.81298470e-01 -5.20579159e-01 6.25222981e-01 -3.51156890e-01
6.76919043e-01 -4.28009063e-01 -5.63513458e-01 -1.36484361e+00
-1.28247485e-01 -7.87198484e-01 3.97530824e-01 -2.39252239e-01
-1.07045853e+00 7.18978345e-01 -2.25873277e-01 -1.07595599e+00
-2.29429677e-01 -7.34302104e-01 -5.39921045e-01 9.63124514e-01
-1.53764653e+00 -9.63132143e-01 1.68045908e-01 1.59480006e-01
6.85871422e-01 -1.54192541e-02 7.89702237e-01 2.41286278e-01
-8.86090338e-01 7.56160915e-01 -5.46061173e-02 7.63487875e-01
6.15780294e-01 -1.24142635e+00 7.86109209e-01 1.15777719e+00
4.20874417e-01 9.90854681e-01 2.65319258e-01 -1.08662689e+00
-1.01780081e+00 -1.19660783e+00 1.17307568e+00 -3.08217555e-01
6.97441161e-01 -6.00586832e-01 -1.26214981e+00 7.98326671e-01
1.50119677e-01 -6.84212595e-02 1.05433977e+00 4.35410470e-01
-1.02587081e-01 2.83081740e-01 -4.31470573e-01 6.03953421e-01
1.07253599e+00 -4.84615058e-01 -7.23628104e-01 -1.29932463e-01
1.04612541e+00 -4.03292567e-01 -6.51422739e-01 2.89900750e-01
2.32522056e-01 -2.48372599e-01 4.73410726e-01 -6.97262049e-01
-5.97992986e-02 -4.25789237e-01 3.27892862e-02 -1.26230252e+00
-1.45938069e-01 -5.12120008e-01 3.40994328e-01 1.60986280e+00
9.03473258e-01 -4.29530919e-01 8.01944911e-01 6.60403430e-01
-4.44125623e-01 -6.57727361e-01 -4.84793782e-01 -8.44964445e-01
1.61188394e-01 -8.35494816e-01 8.72160733e-01 1.00052154e+00
2.80738056e-01 3.28633487e-01 -1.78444326e-01 6.24850430e-02
9.81572736e-03 2.71247119e-01 4.72936273e-01 -1.06839657e+00
-5.76429330e-02 -5.20684481e-01 -2.03397647e-02 -1.52325618e+00
4.93392020e-01 -9.05609369e-01 6.64088249e-01 -1.57430613e+00
1.43205464e-01 -9.10657823e-01 -4.51069057e-01 8.01376343e-01
-7.30553269e-01 1.49077833e-01 2.40738448e-02 -3.99176404e-02
-1.06130278e+00 5.54265916e-01 1.26050627e+00 1.64090544e-01
-3.83554965e-01 -4.43973653e-02 -8.14488590e-01 8.47193182e-01
1.00644767e+00 -5.66634417e-01 -1.06836624e-01 -5.18047750e-01
2.46446222e-01 -1.00368425e-01 -5.22253215e-01 -4.86258477e-01
2.33711496e-01 -4.40161824e-01 9.81053635e-02 -6.53003871e-01
-8.62511098e-02 -4.72991198e-01 -3.53017479e-01 8.59199017e-02
-1.13007233e-01 -9.83551964e-02 1.99449867e-01 3.08160871e-01
-5.01993418e-01 -6.11187339e-01 3.28560382e-01 -4.00301665e-01
-1.12975764e+00 1.91407904e-01 -4.66879398e-01 3.70667964e-01
7.99062192e-01 -6.18216246e-02 -4.12594453e-02 -3.16213742e-02
-5.79464376e-01 4.59255517e-01 4.89992052e-01 3.21204215e-01
4.31077600e-01 -1.11815429e+00 -4.73167062e-01 2.40945518e-01
2.49594152e-01 2.58137703e-01 -2.47467339e-01 6.40079737e-01
-3.74788493e-01 6.63382173e-01 2.23566711e-01 -4.46818620e-01
-9.97007906e-01 6.15146637e-01 -1.08590439e-01 -5.25092363e-01
-4.35263067e-01 8.86212885e-01 3.04591537e-01 -5.64501226e-01
1.77766904e-01 -5.44725239e-01 -5.01708865e-01 8.51415694e-02
4.79355425e-01 -3.08064461e-01 1.03425212e-01 -5.97215533e-01
-5.15474737e-01 4.72022325e-01 -3.18758607e-01 -1.60309657e-01
1.02534986e+00 -3.05153251e-01 -1.33728519e-01 3.94564122e-01
6.86422706e-01 2.94949263e-01 -1.17015982e+00 -4.46956843e-01
9.16190326e-01 -1.93317115e-01 -9.91690382e-02 -7.05600798e-01
-8.48641336e-01 9.71855521e-01 -2.08933890e-01 -1.11726113e-02
1.15978801e+00 -2.20161844e-02 9.90681231e-01 6.22995555e-01
6.79976165e-01 -1.22117007e+00 -1.01848431e-01 1.02553487e+00
1.72860846e-01 -1.37303150e+00 -3.92156124e-01 -7.45199740e-01
-9.76106822e-01 7.81801999e-01 8.74842405e-01 3.17855358e-01
6.44557059e-01 4.39198405e-01 5.07700324e-01 2.45493710e-01
-7.89377749e-01 -7.33878374e-01 2.81526685e-01 4.22259301e-01
8.37139904e-01 3.73344272e-02 -6.61072552e-01 1.13247764e+00
9.91619527e-02 -1.65069610e-01 2.68296510e-01 1.07477427e+00
-5.50232708e-01 -1.72876513e+00 -1.41425073e-01 2.47588247e-01
-7.59504318e-01 -6.18153334e-01 -3.72681648e-01 7.73591101e-01
1.05595991e-01 9.35149252e-01 -1.91019978e-02 -1.95746288e-01
1.64994836e-01 3.99269015e-01 5.09429686e-02 -1.13427258e+00
-6.47843122e-01 6.57976091e-01 4.28068846e-01 -2.13970289e-01
-1.87905535e-01 -8.69299471e-01 -1.74802136e+00 4.20401752e-01
-8.55226159e-01 5.25437891e-01 4.42760170e-01 1.14682853e+00
1.08342603e-01 3.69257331e-01 3.27177435e-01 -2.66591728e-01
-2.52459049e-01 -1.13950586e+00 -3.03775221e-01 2.70313948e-01
-1.68225393e-01 -4.87681687e-01 -1.17991129e-02 1.98030606e-01]
|
[9.997269630432129, 10.106274604797363]
|
0129612c-b7f7-4214-a6e2-d92164d84e0f
|
learning-to-fuse-2d-and-3d-image-cues-for
|
1611.05708
| null |
http://arxiv.org/abs/1611.05708v3
|
http://arxiv.org/pdf/1611.05708v3.pdf
|
Learning to Fuse 2D and 3D Image Cues for Monocular Body Pose Estimation
|
Most recent approaches to monocular 3D human pose estimation rely on Deep
Learning. They typically involve regressing from an image to either 3D joint
coordinates directly or 2D joint locations from which 3D coordinates are
inferred. Both approaches have their strengths and weaknesses and we therefore
propose a novel architecture designed to deliver the best of both worlds by
performing both simultaneously and fusing the information along the way. At the
heart of our framework is a trainable fusion scheme that learns how to fuse the
information optimally instead of being hand-designed. This yields significant
improvements upon the state-of-the-art on standard 3D human pose estimation
benchmarks.
|
['Pablo Márquez-Neila', 'Pascal Fua', 'Mathieu Salzmann', 'Bugra Tekin']
|
2016-11-17
|
learning-to-fuse-2d-and-3d-image-cues-for-1
|
http://openaccess.thecvf.com/content_iccv_2017/html/Tekin_Learning_to_Fuse_ICCV_2017_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2017/papers/Tekin_Learning_to_Fuse_ICCV_2017_paper.pdf
|
iccv-2017-10
|
['monocular-3d-human-pose-estimation']
|
['computer-vision']
|
[-1.89088538e-01 1.04585923e-02 -3.20352428e-02 -4.68147576e-01
-8.05823326e-01 -4.43005323e-01 7.38053143e-01 -3.57546397e-02
-7.31114209e-01 6.30651474e-01 3.49252999e-01 4.20995727e-02
1.28505707e-01 -3.84414613e-01 -8.96997213e-01 -3.28362018e-01
-2.90573929e-02 8.36494684e-01 2.67505050e-02 -3.62316519e-01
4.24072184e-02 5.62141061e-01 -1.31663764e+00 3.33219767e-03
1.81648433e-01 1.06500483e+00 -3.11875433e-01 8.48414302e-01
3.60968262e-01 6.05984211e-01 -4.98795658e-01 -3.85023117e-01
6.45799041e-01 -1.64634347e-01 -6.47330165e-01 9.49579030e-02
7.28170514e-01 -7.04078794e-01 -6.77429676e-01 5.84979534e-01
8.03914428e-01 -3.73951979e-02 6.07596278e-01 -1.16273999e+00
-2.34160081e-01 6.86870068e-02 -5.81939340e-01 5.71849830e-02
9.77721453e-01 3.67091149e-01 8.54499221e-01 -9.32015538e-01
6.78934038e-01 1.37298036e+00 9.03716803e-01 3.09772909e-01
-1.15607774e+00 -3.34451318e-01 2.97956318e-01 -8.70572031e-02
-1.33063352e+00 -5.02828658e-01 6.58763945e-01 -6.38025343e-01
1.44743204e+00 -1.90902650e-01 7.20331192e-01 1.07204735e+00
3.74690294e-01 9.84537005e-01 8.93173516e-01 -4.66248333e-01
-1.47218466e-01 -3.18696856e-01 -2.98592716e-01 8.62458706e-01
1.36602551e-01 2.27637902e-01 -6.85750902e-01 -6.19992912e-02
1.00848293e+00 -8.02729686e-04 -1.12148859e-01 -1.06799102e+00
-1.34457004e+00 5.16042709e-01 7.55601823e-01 -1.99097916e-01
-3.54523301e-01 4.41929758e-01 1.07809506e-01 3.85602303e-02
3.21643949e-01 3.48111033e-01 -6.35645747e-01 -4.99616489e-02
-7.52544224e-01 9.98757482e-01 5.89238226e-01 8.48238468e-01
5.70887148e-01 -3.44062567e-01 -4.97375429e-02 2.30339527e-01
4.50814754e-01 4.11101073e-01 1.64880350e-01 -1.01202703e+00
6.84423566e-01 6.41522050e-01 4.96148050e-01 -9.27658558e-01
-7.43223667e-01 -3.95226985e-01 -1.38946861e-01 5.25104761e-01
7.01807022e-01 -3.12401712e-01 -1.23875546e+00 1.84051609e+00
5.17162263e-01 -1.53313413e-01 -2.85076678e-01 9.63914514e-01
6.02859974e-01 3.09875160e-01 8.17803815e-02 5.44388413e-01
1.11326456e+00 -1.02476454e+00 -3.49266589e-01 -5.80380559e-01
5.18215477e-01 -7.20737517e-01 5.52625656e-01 2.75487036e-01
-1.32290506e+00 -6.70010448e-01 -1.19101918e+00 -4.21446472e-01
-3.14911127e-01 2.00628087e-01 4.68430668e-01 4.62999851e-01
-1.05683076e+00 6.03505552e-01 -1.06134713e+00 -1.63460225e-01
2.13919327e-01 6.17240727e-01 -8.78288925e-01 8.03418756e-02
-9.90515828e-01 1.41134810e+00 2.66509026e-01 2.50759125e-01
-5.34053326e-01 -5.13143003e-01 -1.11277771e+00 -2.75213242e-01
5.67585051e-01 -1.33441341e+00 1.46489418e+00 -2.24449709e-01
-1.45289290e+00 1.12167728e+00 -3.69809680e-02 -5.42687297e-01
1.00085497e+00 -1.12597847e+00 2.83911377e-01 -3.03480756e-02
1.63291872e-01 1.14509976e+00 6.42252147e-01 -1.25936425e+00
-5.32857239e-01 -6.51627183e-01 1.70264855e-01 5.01792908e-01
1.92624733e-01 -2.57197738e-01 -7.08659410e-01 -5.57661891e-01
2.55399525e-01 -1.27319729e+00 -2.80206114e-01 3.49589050e-01
-4.88212496e-01 -2.53197283e-01 4.45638567e-01 -7.75905192e-01
7.91971624e-01 -1.57652020e+00 7.69437671e-01 7.39468634e-02
4.23956603e-01 1.88599691e-01 1.05531842e-01 2.98449010e-01
-1.36649422e-02 -2.88225472e-01 1.53898820e-01 -7.24096894e-01
3.30002964e-01 -4.56743576e-02 1.38983220e-01 7.28741050e-01
3.43765438e-01 1.24672973e+00 -8.34603071e-01 -3.13544989e-01
5.64309478e-01 7.30875075e-01 -7.46261954e-01 3.22691709e-01
-1.86258957e-01 6.22249663e-01 -1.80938721e-01 5.11230767e-01
3.55394095e-01 -1.62164032e-01 1.83989376e-01 -3.28845322e-01
2.09552944e-01 4.13008124e-01 -1.26191807e+00 2.23770046e+00
-2.38181621e-01 4.50622320e-01 -8.03942457e-02 -7.55080223e-01
8.02144468e-01 3.46071750e-01 7.03130245e-01 -3.09706062e-01
3.68294477e-01 1.66552752e-01 -1.83703825e-01 -2.05841988e-01
3.93944561e-01 -1.28841147e-01 -3.56782794e-01 3.30940336e-01
3.45930517e-01 -1.93606958e-01 -2.62608469e-01 -1.83631089e-02
9.51746523e-01 8.66642356e-01 5.93111813e-01 2.24149987e-01
3.75971913e-01 -2.35136449e-01 2.65106797e-01 5.32459795e-01
-3.60547066e-01 7.34482527e-01 5.62636852e-01 -7.77001143e-01
-1.20078659e+00 -1.41284251e+00 3.70612621e-01 7.68123209e-01
-1.41032832e-02 -4.34357673e-01 -5.10646522e-01 -9.09122348e-01
3.95905375e-01 2.42511645e-01 -8.17147553e-01 -6.59948066e-02
-8.45114112e-01 -9.28429421e-03 4.75138247e-01 9.97476876e-01
3.50284308e-01 -6.44491851e-01 -9.89428222e-01 5.02475910e-02
-2.59777725e-01 -1.35174251e+00 -4.83312517e-01 3.13433290e-01
-5.43089271e-01 -1.01912045e+00 -9.79750574e-01 -5.55317223e-01
4.33476925e-01 1.24403276e-01 1.36070490e+00 -2.06662908e-01
-2.12526560e-01 3.65807176e-01 -6.40362967e-04 -5.05251825e-01
2.04611734e-01 2.41963550e-01 2.06051782e-01 -3.52059782e-01
4.93538558e-01 -3.61165613e-01 -5.53479850e-01 2.04022691e-01
-2.59765327e-01 -6.82183355e-02 5.87075353e-01 7.50407457e-01
4.11134094e-01 -3.57908994e-01 -6.61846297e-03 -6.04059756e-01
1.05051011e-01 -1.31144643e-01 -4.39250827e-01 4.47672382e-02
-4.56609242e-02 2.26923957e-01 1.40010193e-01 -8.85587409e-02
-8.53331745e-01 7.87007928e-01 -2.63197511e-01 -5.54516494e-01
-4.00289655e-01 1.32749841e-01 -2.28659615e-01 -1.46939605e-01
6.43865228e-01 -1.98014468e-01 3.83717939e-02 -5.82061887e-01
5.42763829e-01 2.51879483e-01 9.32594836e-01 -5.14902115e-01
8.29348087e-01 4.73356009e-01 1.66605040e-01 -3.14614505e-01
-1.01360631e+00 -3.48606646e-01 -1.43027747e+00 -1.42559543e-01
1.11171782e+00 -1.19481969e+00 -7.13160157e-01 6.46962225e-01
-1.11922729e+00 -1.44404560e-01 -3.91195156e-02 5.15660405e-01
-9.13369119e-01 2.94234276e-01 -4.91561502e-01 -5.99210262e-01
2.55165552e-03 -1.21359742e+00 1.66548574e+00 -7.81945139e-02
-6.47439718e-01 -7.76025355e-01 2.30855212e-01 2.86909819e-01
-4.59553645e-04 6.76639855e-01 5.24251759e-01 -3.08379024e-01
-4.67579663e-01 -7.42608786e-01 -1.12523980e-01 1.86375659e-02
-7.39502534e-03 -4.71247852e-01 -8.45635653e-01 -4.14510280e-01
-2.53587514e-01 -4.86491978e-01 7.96467125e-01 4.81858134e-01
6.37480140e-01 -3.59752625e-02 -4.60637093e-01 7.21629739e-01
1.09588242e+00 -2.70226657e-01 3.93834859e-01 4.68763560e-01
8.74874353e-01 5.89280367e-01 4.48196501e-01 4.70231146e-01
7.36566067e-01 1.26138937e+00 5.26900887e-01 -9.10131335e-02
-2.48535015e-02 -5.69730103e-01 -9.74669121e-03 1.64675429e-01
-2.17064515e-01 -6.26157671e-02 -9.27650392e-01 1.62520438e-01
-1.92084932e+00 -6.63957894e-01 4.96444702e-01 2.06066871e+00
6.14765108e-01 5.66629350e-01 5.16198933e-01 1.67740718e-01
2.66630948e-01 2.38196105e-01 -6.11748695e-01 -4.11401875e-02
2.90171862e-01 1.42231569e-01 4.36013401e-01 6.21045411e-01
-1.57510483e+00 7.99390614e-01 7.23754787e+00 4.34034653e-02
-8.65174770e-01 -3.06982100e-01 1.90750331e-01 -3.89462858e-01
1.77903593e-01 -1.06134064e-01 -9.66006696e-01 1.29029676e-01
4.59559113e-01 4.18256551e-01 9.90220606e-02 6.49646699e-01
-1.63715497e-01 -1.57001734e-01 -1.46590900e+00 1.20470870e+00
2.33600557e-01 -9.49449003e-01 -9.95139871e-03 2.35749319e-01
6.28321290e-01 -5.44511192e-02 5.50406836e-02 1.05608694e-01
4.32003200e-01 -1.19697142e+00 1.09161437e+00 7.16688693e-01
4.69112813e-01 -6.87146008e-01 8.01295280e-01 4.39707756e-01
-1.21844041e+00 5.80001101e-02 1.42508298e-01 -4.43064272e-01
3.49378467e-01 7.97467604e-02 -7.90297627e-01 4.91648674e-01
5.91287494e-01 8.25576186e-01 -5.92067897e-01 9.29822028e-01
-4.76401120e-01 -3.43778461e-01 -4.80837554e-01 1.86581627e-01
2.34933242e-01 2.62211025e-01 4.82207149e-01 9.55545008e-01
8.90055150e-02 -1.00056037e-01 5.53031564e-01 5.98897636e-01
1.03012249e-01 -4.30133909e-01 -6.76921010e-01 3.19060624e-01
3.10627043e-01 8.21725905e-01 -6.09050572e-01 -1.93617955e-01
-3.91969889e-01 9.23220932e-01 5.67278147e-01 5.73223177e-03
-8.08967113e-01 -2.60734767e-01 8.68515968e-01 1.02472201e-01
5.77422082e-01 -7.46467471e-01 -4.21426386e-01 -1.09792650e+00
1.71701416e-01 -7.83340454e-01 3.88245255e-01 -8.07113469e-01
-1.06892776e+00 3.63331169e-01 1.48720637e-01 -1.17861402e+00
-6.37841582e-01 -8.93653750e-01 -1.70188218e-01 9.08339977e-01
-1.09765363e+00 -1.33190310e+00 -2.97593743e-01 5.56539893e-01
2.77582109e-01 6.15596846e-02 6.48351073e-01 1.04715131e-01
-1.10847548e-01 7.02801108e-01 -6.23046875e-01 2.12003499e-01
9.03127432e-01 -1.51959324e+00 1.00132418e+00 6.63898587e-01
2.37997085e-01 5.28676212e-01 7.73135126e-01 -4.66341764e-01
-1.34116697e+00 -4.23321098e-01 1.18576360e+00 -1.22368658e+00
1.98994473e-01 -5.43579996e-01 -3.02351862e-01 9.64146614e-01
-1.19324848e-01 3.99281271e-02 2.56110370e-01 2.42278874e-01
-4.85031962e-01 7.42839351e-02 -9.93445516e-01 5.30853868e-01
1.13790989e+00 -6.03481114e-01 -8.52055848e-01 1.37955099e-01
4.84194398e-01 -9.91136730e-01 -8.52375448e-01 4.32078660e-01
8.97901416e-01 -1.09548092e+00 1.48616254e+00 -7.40572870e-01
4.81076807e-01 -3.84866208e-01 -2.33535886e-01 -1.22886348e+00
-2.75258422e-01 -5.22828102e-01 -2.94552147e-01 5.27822196e-01
1.27951756e-01 -1.40689492e-01 1.17076647e+00 6.17411792e-01
4.78838198e-02 -9.22466695e-01 -8.96031559e-01 -4.17958885e-01
4.51473966e-02 -2.22290516e-01 5.32483160e-01 2.67972142e-01
-8.53954628e-02 5.85264325e-01 -6.13903463e-01 -3.30972783e-02
7.12561667e-01 -2.11666971e-01 1.34986246e+00 -9.57954347e-01
-3.91461194e-01 -4.23184723e-01 -8.40023398e-01 -1.41785657e+00
2.50588283e-02 -4.56239730e-01 1.69781581e-01 -1.48893535e+00
-4.39239852e-02 2.05613729e-02 -7.98177645e-02 5.63020706e-01
-3.59298408e-01 4.34246689e-01 3.91173422e-01 4.09688912e-02
-6.78788781e-01 3.67927015e-01 1.19811189e+00 8.00473988e-02
-1.68442428e-01 1.28788874e-02 -5.66180766e-01 9.46121871e-01
4.04262215e-01 -1.41347334e-01 -2.50001878e-01 -7.06806719e-01
1.72893569e-01 1.44130528e-01 8.02751064e-01 -1.28129447e+00
2.98984736e-01 2.45016545e-01 1.09819198e+00 -1.05824888e+00
8.20722163e-01 -6.94728136e-01 1.41403582e-02 5.32174170e-01
-3.67509574e-01 3.17417055e-01 7.40458295e-02 5.64559937e-01
-3.30812647e-03 3.83069903e-01 7.21439540e-01 -5.17701805e-01
-6.83092892e-01 2.99872577e-01 5.25045656e-02 8.72084573e-02
9.88812506e-01 -2.56096125e-01 8.51610601e-02 -5.90806246e-01
-9.12554681e-01 9.36548933e-02 5.39067566e-01 4.81591731e-01
4.73419249e-01 -1.48268402e+00 -5.80279589e-01 3.58265579e-01
-5.03708236e-03 5.03261313e-02 -7.13168532e-02 6.48580194e-01
-6.09777927e-01 8.09710443e-01 -4.86586124e-01 -8.68640006e-01
-9.95545626e-01 5.10438859e-01 5.69287837e-01 -4.34328675e-01
-6.11620843e-01 1.04069054e+00 -6.19923361e-02 -7.75444090e-01
5.76450884e-01 -1.58159569e-01 4.99240346e-02 -1.41033292e-01
2.39335507e-01 3.03082913e-01 1.97609305e-01 -9.38294530e-01
-5.65342426e-01 8.17126632e-01 7.51678869e-02 -3.19481999e-01
1.34983730e+00 -2.09282473e-01 2.69182146e-01 3.43733788e-01
1.32615066e+00 -1.66760311e-01 -1.77548909e+00 -3.75045329e-01
-9.13634077e-02 -7.23985434e-01 -1.33735910e-01 -9.04215276e-01
-8.02301466e-01 8.23323309e-01 4.14472252e-01 -4.57863092e-01
7.26911485e-01 5.01805767e-02 9.62000310e-01 4.44972813e-01
3.91502321e-01 -9.81911659e-01 3.15748006e-01 5.59143424e-01
7.60843694e-01 -1.27255642e+00 4.82469231e-01 -2.23031387e-01
-4.75393921e-01 1.05130947e+00 6.77843511e-01 -6.83206856e-01
8.20676327e-01 3.42985183e-01 1.77598953e-01 -1.85997471e-01
-6.45664036e-01 -2.63749629e-01 7.54604459e-01 6.81229532e-01
6.37095809e-01 -5.21592274e-02 7.29107931e-02 3.56274784e-01
-3.89476955e-01 -6.61657751e-02 -1.30707175e-01 1.25651789e+00
-3.11491936e-01 -1.03935063e+00 -4.47943032e-01 1.38287976e-01
-3.31953406e-01 3.54006320e-01 -7.10875094e-01 1.01791394e+00
2.07484230e-01 6.60494983e-01 -1.06112935e-01 -6.12353981e-01
7.79574573e-01 9.52486172e-02 1.06617808e+00 -5.00347078e-01
-4.81050998e-01 1.95104510e-01 -3.56904976e-02 -9.05291259e-01
-4.41041261e-01 -7.90340006e-01 -7.88217902e-01 -3.03893149e-01
-1.45019278e-01 -4.31828976e-01 5.46523213e-01 1.19038427e+00
1.27569407e-01 3.09211642e-01 1.91055432e-01 -1.74858463e+00
-7.39871800e-01 -6.94507539e-01 -2.15105936e-01 3.51008147e-01
5.94332874e-01 -1.18678403e+00 5.76071963e-02 -2.21168593e-01]
|
[6.933110237121582, -0.9832189679145813]
|
a1db8fa4-22f7-4bad-af7c-0cad349b2e6e
|
fsar-federated-skeleton-based-action
|
2306.11046
| null |
https://arxiv.org/abs/2306.11046v1
|
https://arxiv.org/pdf/2306.11046v1.pdf
|
FSAR: Federated Skeleton-based Action Recognition with Adaptive Topology Structure and Knowledge Distillation
|
Existing skeleton-based action recognition methods typically follow a centralized learning paradigm, which can pose privacy concerns when exposing human-related videos. Federated Learning (FL) has attracted much attention due to its outstanding advantages in privacy-preserving. However, directly applying FL approaches to skeleton videos suffers from unstable training. In this paper, we investigate and discover that the heterogeneous human topology graph structure is the crucial factor hindering training stability. To address this limitation, we pioneer a novel Federated Skeleton-based Action Recognition (FSAR) paradigm, which enables the construction of a globally generalized model without accessing local sensitive data. Specifically, we introduce an Adaptive Topology Structure (ATS), separating generalization and personalization by learning a domain-invariant topology shared across clients and a domain-specific topology decoupled from global model aggregation.Furthermore, we explore Multi-grain Knowledge Distillation (MKD) to mitigate the discrepancy between clients and server caused by distinct updating patterns through aligning shallow block-wise motion features. Extensive experiments on multiple datasets demonstrate that FSAR outperforms state-of-the-art FL-based methods while inherently protecting privacy.
|
['Chenyang Si', 'Min Zhang', 'Tianyu Guo', 'Shitong Sun', 'Hong Liu', 'Jingwen Guo']
|
2023-06-19
| null | null | null | null |
['skeleton-based-action-recognition', 'action-recognition-in-videos']
|
['computer-vision', 'computer-vision']
|
[ 1.80280849e-01 -1.25801221e-01 -5.25647879e-01 -3.73387814e-01
-7.62032688e-01 -6.77368879e-01 3.68129790e-01 -2.13417962e-01
-2.51871467e-01 6.19525909e-01 3.82414430e-01 -4.93284985e-02
-2.97876388e-01 -5.65979838e-01 -8.77147615e-01 -1.02728236e+00
-2.65655935e-01 -1.21627683e-02 3.75139862e-01 1.37698859e-01
1.03116609e-01 6.59274876e-01 -1.31104362e+00 4.56967831e-01
6.49619460e-01 1.14692950e+00 -4.60661888e-01 1.16140038e-01
8.10456574e-02 9.25610900e-01 -4.30309236e-01 -6.75478339e-01
6.04814112e-01 -3.86637270e-01 -9.16400909e-01 2.28968233e-01
7.19427586e-01 -5.96896648e-01 -7.41383612e-01 9.49708164e-01
3.58082831e-01 7.75435492e-02 1.41019315e-01 -1.66979742e+00
-4.31563526e-01 3.41501117e-01 -5.21728814e-01 3.87722328e-02
3.86444658e-01 4.32428777e-01 9.33198571e-01 -2.35928982e-01
7.91569650e-01 9.05729711e-01 5.12548566e-01 6.84675872e-01
-1.20133436e+00 -6.51822686e-01 3.87280256e-01 6.40141666e-01
-1.44869196e+00 -4.33373570e-01 8.63373101e-01 -2.42334038e-01
5.43867230e-01 5.80165148e-01 5.23792446e-01 1.37530196e+00
7.71612227e-02 9.22196031e-01 1.05471420e+00 7.71413520e-02
4.14689332e-01 -1.84577301e-01 -2.97681957e-01 8.50309908e-01
3.76305670e-01 -8.98738950e-02 -1.04310644e+00 -6.41681135e-01
7.98860788e-01 1.73161313e-01 -5.79997480e-01 -1.21887767e+00
-1.15726137e+00 5.56496501e-01 2.21202463e-01 -4.57044281e-02
-9.99031737e-02 7.91256055e-02 7.19262362e-01 4.00348604e-01
-8.64882022e-02 1.21672079e-02 -4.70952004e-01 9.60387290e-03
-6.94661736e-01 2.67979037e-02 8.72134328e-01 8.02138388e-01
8.25294018e-01 -1.67963892e-01 -1.63619474e-01 1.25391722e-01
1.35813847e-01 1.26498878e-01 5.72826982e-01 -1.06931520e+00
5.24344444e-01 7.31907666e-01 -1.72658011e-01 -1.25533366e+00
8.62549096e-02 -8.17307830e-02 -8.84717762e-01 -7.58310035e-02
4.33553249e-01 6.14228239e-03 -4.24073935e-01 1.86470795e+00
8.09656918e-01 3.43185961e-01 2.79592704e-02 9.89843965e-01
1.36364922e-01 2.82797664e-02 1.05067812e-01 -1.27234176e-01
1.13910854e+00 -9.59784269e-01 -3.06181341e-01 2.92617261e-01
8.51769388e-01 -7.09009692e-02 7.71692216e-01 3.62610072e-01
-6.78653777e-01 -3.85439619e-02 -9.32659864e-01 7.19921663e-02
-3.16021293e-01 -2.44999886e-01 6.00824416e-01 9.53992486e-01
-7.77047575e-01 6.18735373e-01 -1.08943117e+00 -5.58497488e-01
1.02747810e+00 4.68548566e-01 -9.17183042e-01 -6.81286156e-02
-1.02842307e+00 2.39077970e-01 1.51375890e-01 -1.99736536e-01
-9.64706957e-01 -5.88903487e-01 -6.79055870e-01 -8.84203054e-03
8.61185610e-01 -8.39672089e-01 8.36708128e-01 -7.56852090e-01
-1.58054435e+00 7.47600675e-01 2.74834365e-01 -7.48059988e-01
8.75401914e-01 -8.30718204e-02 -4.23691630e-01 6.59868777e-01
-1.03311516e-01 4.63804416e-02 1.22794533e+00 -1.01709712e+00
-6.10852063e-01 -7.81194210e-01 1.25072390e-01 1.78903475e-01
-8.70844722e-01 -5.08141965e-02 -2.58504957e-01 -7.40699410e-01
-1.10999495e-02 -6.73006833e-01 -1.88003272e-01 5.96537471e-01
-3.35614443e-01 9.92718786e-02 1.23391151e+00 -6.13159776e-01
1.29036176e+00 -2.29829192e+00 1.34998813e-01 5.01537740e-01
3.10618460e-01 2.15669438e-01 -1.25573844e-01 4.71891284e-01
2.87229180e-01 6.40819687e-03 -4.55731899e-02 -1.31734952e-01
2.64942255e-02 3.83190542e-01 -3.68229866e-01 8.58422339e-01
-2.50126868e-01 8.42643976e-01 -7.38298595e-01 -8.73142302e-01
3.23641044e-03 3.56076688e-01 -5.35910785e-01 -1.39062256e-02
-1.87114969e-01 7.55720139e-01 -9.62029219e-01 9.76391792e-01
6.42685592e-01 -3.93258035e-01 7.33717501e-01 -3.09821010e-01
1.00374073e-01 -2.11101964e-01 -1.19145572e+00 2.06557250e+00
1.11386418e-01 -6.34905100e-02 4.45817620e-01 -1.18029571e+00
8.26296091e-01 3.51790845e-01 9.51072931e-01 -3.40834290e-01
-5.85497022e-02 5.91124073e-02 -6.76956892e-01 -4.74665076e-01
1.21970642e-02 2.52173096e-01 6.25876039e-02 6.70933366e-01
-2.21352577e-01 7.74208546e-01 -4.02078301e-01 1.68596745e-01
1.48759937e+00 2.91186124e-01 1.09096728e-01 -6.66747510e-05
7.71782279e-01 -1.81414902e-01 9.79052842e-01 4.98678833e-01
-7.08262563e-01 3.40440780e-01 5.36493540e-01 -6.85931504e-01
-4.08067971e-01 -1.05992079e+00 1.90136313e-01 1.08351922e+00
4.01557952e-01 -6.18503094e-01 -7.12972760e-01 -1.38222730e+00
3.21442515e-01 1.08097143e-01 -4.84518111e-01 -3.89022410e-01
-7.16760337e-01 -2.98756272e-01 8.45498502e-01 4.73461896e-01
8.66048932e-01 -5.53717613e-01 -8.17497432e-01 -4.62835617e-02
-4.26932156e-01 -9.13713455e-01 -7.53659666e-01 -3.40220988e-01
-8.68963003e-01 -1.29222369e+00 -4.44715887e-01 -3.52945000e-01
6.48621798e-01 3.92191440e-01 5.14178097e-01 -2.30187047e-02
-4.78347301e-01 8.03864717e-01 -2.45329320e-01 2.66822219e-01
5.21509610e-02 1.62002221e-01 1.62213579e-01 7.92985439e-01
3.93884927e-01 -7.92092860e-01 -7.88853049e-01 6.80131614e-01
-1.10403895e+00 -2.65700996e-01 5.10925114e-01 6.33503139e-01
6.42739594e-01 7.51725435e-02 2.31820837e-01 -7.43138373e-01
1.05165936e-01 -4.47630405e-01 -4.06559736e-01 6.67685509e-01
-5.66707969e-01 -5.84587976e-02 6.42102540e-01 -5.04597604e-01
-1.12669361e+00 4.90846008e-01 5.90532959e-01 -8.68189454e-01
-3.40570182e-01 3.38767879e-02 -7.62833834e-01 -5.55467904e-01
4.32591558e-01 4.93329823e-01 3.98550808e-01 -6.52430892e-01
3.03675950e-01 4.03711200e-01 5.44051528e-01 -8.10888469e-01
7.99659014e-01 1.05301535e+00 2.36988872e-01 -5.43027520e-01
-3.72978032e-01 -4.04567152e-01 -6.17640555e-01 -2.23630488e-01
7.35567868e-01 -8.05377424e-01 -8.95405948e-01 6.43089116e-01
-6.63602829e-01 -6.85617886e-03 -5.05680561e-01 2.85596430e-01
-8.36594045e-01 9.30492401e-01 -5.30382574e-01 -4.36446667e-01
-2.18389466e-01 -8.43765616e-01 7.67729104e-01 -1.17265083e-01
1.02377154e-01 -5.50893664e-01 1.82999432e-01 8.44486177e-01
3.50646138e-01 5.60877383e-01 6.31376266e-01 -7.12655246e-01
-1.00353932e+00 -2.43701845e-01 -4.72073257e-02 1.82365835e-01
2.58517265e-01 -3.94913882e-01 -7.25929022e-01 -6.57605946e-01
6.73671952e-03 -5.19054770e-01 6.53953850e-01 -2.13950157e-01
1.44852555e+00 -7.72381961e-01 -4.10605133e-01 9.09076750e-01
1.28103971e+00 -4.86649573e-01 6.12977445e-01 4.97919232e-01
8.51036429e-01 5.54917395e-01 4.51605916e-01 8.84361744e-01
3.91340882e-01 6.17677450e-01 5.39937854e-01 3.74167979e-01
1.51442021e-01 -5.95294595e-01 5.52276909e-01 1.89263344e-01
1.17108664e-02 2.10533783e-01 -4.65090722e-01 3.09510559e-01
-2.15938115e+00 -1.03895402e+00 3.54281753e-01 2.32867837e+00
7.26063490e-01 -3.74591440e-01 3.82449597e-01 -1.39939696e-01
6.01665258e-01 5.58079600e-01 -9.05456245e-01 5.59996665e-02
-2.33679757e-01 -1.59119695e-01 8.07112098e-01 -2.54643619e-01
-1.20667970e+00 7.33978093e-01 5.10789394e+00 9.63443577e-01
-1.00302994e+00 1.85328484e-01 2.72395134e-01 -2.75206268e-01
-1.26789510e-01 1.06553823e-01 -4.91352022e-01 4.00789529e-01
5.96070409e-01 -2.82037258e-01 4.58274484e-01 1.05897462e+00
-1.14191368e-01 3.33653063e-01 -1.05354512e+00 9.35312808e-01
-9.90660768e-03 -1.39638710e+00 4.20686156e-01 4.80513394e-01
5.49181700e-01 -2.82369316e-01 3.95612828e-02 -9.08431262e-02
1.84879810e-01 -6.70539081e-01 5.11389554e-01 5.83693326e-01
8.62174690e-01 -7.03886569e-01 1.64201558e-01 2.25899354e-01
-1.35597932e+00 -3.97274971e-01 -2.61230558e-01 4.12444919e-01
-6.43077567e-02 6.24702461e-02 -2.28920162e-01 1.00495732e+00
9.14211929e-01 6.73644900e-01 -4.95251358e-01 7.81041086e-01
1.20757282e-01 5.02354801e-01 -4.09915686e-01 4.28236753e-01
-9.83630121e-02 -2.41545122e-03 6.47954404e-01 7.30094075e-01
-6.16925210e-03 9.33436602e-02 3.50697786e-01 5.09057224e-01
-7.09526017e-02 2.52422422e-01 -6.44289315e-01 -6.11530095e-02
5.11719584e-01 9.75621939e-01 -3.06727499e-01 8.19865018e-02
-6.00851774e-01 1.32985604e+00 4.50045079e-01 3.22072744e-01
-7.38369524e-01 -5.42373657e-02 1.08724582e+00 1.60891518e-01
7.08753467e-01 -2.53487647e-01 3.61645557e-02 -1.47973609e+00
3.59888077e-01 -1.09054112e+00 1.09495461e+00 1.42539501e-01
-1.38474369e+00 1.35527119e-01 -2.45213777e-01 -1.42057788e+00
1.65498972e-01 -1.52136147e-01 -4.56294864e-01 2.39064664e-01
-1.37361503e+00 -1.53025651e+00 -1.85879946e-01 1.37510240e+00
-3.34325843e-02 -2.58161247e-01 9.10725713e-01 1.90843612e-01
-8.14769030e-01 1.12316358e+00 1.35421544e-01 1.44157693e-01
8.51091444e-01 -6.24374986e-01 -1.64630711e-02 8.66918564e-01
1.78884417e-01 6.31678402e-01 1.87711954e-01 -7.63945341e-01
-2.00671959e+00 -1.26306856e+00 5.00967681e-01 -4.13041830e-01
6.60257280e-01 -1.91447780e-01 -9.63754535e-01 7.85367191e-01
-3.55619997e-01 4.99606848e-01 8.58045816e-01 -3.89017820e-01
-9.75357234e-01 -4.66621697e-01 -1.59478700e+00 4.78589058e-01
1.39642119e+00 -6.31152153e-01 -3.94663289e-02 2.38481447e-01
6.14609957e-01 -1.56281199e-02 -1.03469348e+00 3.08974326e-01
6.78938150e-01 -1.31760657e+00 8.78301501e-01 -9.49134707e-01
-1.55845731e-01 -3.97266150e-01 -3.74010861e-01 -5.00294745e-01
-3.40155303e-01 -1.19940150e+00 -8.17982256e-01 1.15124404e+00
-2.33657628e-01 -9.17512715e-01 1.42292845e+00 9.24517095e-01
4.58606899e-01 -6.44516528e-01 -1.43872094e+00 -1.14679980e+00
-2.58520037e-01 2.12208927e-02 1.06178248e+00 1.15846622e+00
1.07690774e-01 -4.28019643e-01 -5.92846572e-01 3.78274262e-01
1.08750606e+00 2.22394228e-01 1.09873915e+00 -9.16514099e-01
-4.46238518e-01 -1.37034535e-01 -8.17818820e-01 -8.99209738e-01
2.15633616e-01 -6.75844848e-01 -4.70430762e-01 -8.58014762e-01
1.34858191e-01 -2.40900755e-01 -6.14909530e-01 8.03262293e-01
3.83750379e-01 -6.80474332e-03 1.91349819e-01 5.13806939e-01
-9.99513328e-01 7.38983154e-01 1.04986668e+00 5.28012253e-02
3.56811769e-02 -4.20613848e-02 -5.07002354e-01 3.59122217e-01
8.54507804e-01 -2.71834701e-01 -5.50580919e-01 -3.82903576e-01
-2.97352195e-01 -8.97861198e-02 6.26404643e-01 -1.01776004e+00
5.28956175e-01 -4.93892521e-01 1.04322948e-01 -1.97910234e-01
1.69003323e-01 -1.23851764e+00 4.16108280e-01 4.93766814e-01
-2.11323336e-01 -4.33913022e-01 -2.18568131e-01 1.28872800e+00
3.12443487e-02 4.97503072e-01 5.63550711e-01 -9.71466303e-02
-7.22516179e-01 7.57295668e-01 -5.33428304e-02 -8.00841749e-02
1.36618352e+00 -4.85210180e-01 -3.13716590e-01 -1.87669247e-01
-5.04035056e-01 3.22267711e-01 1.00903118e+00 3.41783285e-01
6.32838607e-01 -1.28964555e+00 -4.36742425e-01 5.35986364e-01
3.07939768e-01 -2.25421414e-01 5.89897037e-01 9.80267823e-01
-1.66184530e-01 2.34170169e-01 -3.57735723e-01 -3.48432839e-01
-1.48147810e+00 7.78059661e-01 2.00250089e-01 -2.11143836e-01
-9.97359335e-01 6.54769957e-01 1.35543823e-01 -2.48505950e-01
6.47697091e-01 3.67673039e-01 4.09723103e-01 -2.83868045e-01
5.23969352e-01 8.34509015e-01 -1.23129919e-01 -7.23582804e-01
-5.31223774e-01 3.47501189e-01 -2.36112267e-01 2.13653192e-01
1.11457193e+00 -3.36546689e-01 -7.04278648e-02 -1.78857774e-01
1.15187955e+00 -2.08936512e-01 -1.59853578e+00 -4.60760802e-01
5.66570973e-03 -1.01044726e+00 -2.04168215e-01 -4.59586114e-01
-1.34489620e+00 2.93560207e-01 5.04185796e-01 -1.79775611e-01
1.14900434e+00 -1.10967614e-01 1.33425987e+00 3.94508183e-01
9.86328542e-01 -1.14622343e+00 1.21273510e-01 -1.31821901e-01
5.06374180e-01 -9.71048951e-01 9.57583785e-02 -3.27192724e-01
-6.68110251e-01 8.52008820e-01 7.22440004e-01 2.94358462e-01
6.19063377e-01 -1.74612582e-01 -8.02754164e-02 -1.91856146e-01
-6.63441181e-01 2.15079293e-01 -5.41025735e-02 7.76878595e-01
-4.94764626e-01 -2.41391528e-02 -1.26042977e-01 7.43394375e-01
2.97950983e-01 1.53698102e-01 5.63324317e-02 1.57207716e+00
-1.23581454e-01 -1.33691299e+00 -3.64832669e-01 2.48065531e-01
-3.98215264e-01 5.17937601e-01 -6.63959920e-01 4.66566980e-01
-7.71165565e-02 6.24272168e-01 -3.52243096e-01 -5.59265971e-01
9.42700282e-02 1.96858793e-01 4.27531779e-01 -2.03082487e-01
-4.66105133e-01 -2.41783351e-01 -1.75152838e-01 -1.30076027e+00
-3.20536405e-01 -9.46124792e-01 -1.08720779e+00 -4.61135387e-01
-3.46435681e-02 -1.02340449e-02 2.97131777e-01 6.16804302e-01
9.26155806e-01 -3.49337727e-01 8.77409875e-01 -4.23988163e-01
-1.12661040e+00 -1.17624074e-01 -8.98884058e-01 5.45309126e-01
3.73751342e-01 -5.04683793e-01 -4.61692184e-01 2.22333863e-01]
|
[5.837756156921387, 6.675413131713867]
|
7aaac243-2c72-4f49-83b5-9261e7b9fdee
|
oriented-reppoints-for-aerial-object
|
2105.11111
| null |
https://arxiv.org/abs/2105.11111v4
|
https://arxiv.org/pdf/2105.11111v4.pdf
|
Oriented RepPoints for Aerial Object Detection
|
In contrast to the generic object, aerial targets are often non-axis aligned with arbitrary orientations having the cluttered surroundings. Unlike the mainstreamed approaches regressing the bounding box orientations, this paper proposes an effective adaptive points learning approach to aerial object detection by taking advantage of the adaptive points representation, which is able to capture the geometric information of the arbitrary-oriented instances. To this end, three oriented conversion functions are presented to facilitate the classification and localization with accurate orientation. Moreover, we propose an effective quality assessment and sample assignment scheme for adaptive points learning toward choosing the representative oriented reppoints samples during training, which is able to capture the non-axis aligned features from adjacent objects or background noises. A spatial constraint is introduced to penalize the outlier points for roust adaptive learning. Experimental results on four challenging aerial datasets including DOTA, HRSC2016, UCAS-AOD and DIOR-R, demonstrate the efficacy of our proposed approach. The source code is availabel at: https://github.com/LiWentomng/OrientedRepPoints.
|
['Jianke Zhu', 'Kaixuan Hu', 'Yijie Chen', 'Wentong Li']
|
2021-05-24
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Li_Oriented_RepPoints_for_Aerial_Object_Detection_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Li_Oriented_RepPoints_for_Aerial_Object_Detection_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['object-detection-in-aerial-images']
|
['computer-vision']
|
[-1.65984184e-01 -2.53162146e-01 -8.92630592e-02 -2.87928820e-01
-6.23796225e-01 -7.21141517e-01 4.05155808e-01 7.61991250e-04
-1.91965938e-01 3.88298512e-01 -3.36137414e-01 8.29300806e-02
-6.76359236e-01 -8.17309618e-01 -6.96336746e-01 -8.60640943e-01
-3.43678385e-01 2.73212314e-01 2.36146718e-01 -2.01536164e-01
3.43816131e-01 9.44961131e-01 -1.73804772e+00 -4.03903499e-02
9.86207366e-01 1.16419995e+00 1.40865430e-01 5.84385216e-01
2.35817641e-01 1.87845230e-01 -4.75478321e-01 -1.06325913e-02
8.00232649e-01 9.15771425e-02 -1.84880868e-01 3.17485690e-01
8.18838298e-01 -4.43897277e-01 -3.51336636e-02 1.16707337e+00
4.03947264e-01 2.06203878e-01 8.91912341e-01 -1.43009734e+00
-5.36200345e-01 2.59976923e-01 -7.41890311e-01 2.46655717e-01
2.64087766e-01 2.91748315e-01 1.00538242e+00 -1.24039292e+00
2.63408244e-01 1.08619630e+00 8.53927672e-01 -3.53661291e-02
-7.37817466e-01 -6.29496276e-01 6.02404833e-01 3.90175909e-01
-1.81253958e+00 -1.14435181e-01 8.05464149e-01 -5.44358850e-01
2.21184969e-01 4.80605662e-01 6.78783894e-01 9.23946619e-01
-1.86782375e-01 7.21777737e-01 8.56524408e-01 -2.83849895e-01
2.43755616e-02 9.82960016e-02 2.26119533e-01 5.49039304e-01
4.25069541e-01 1.04219720e-01 2.07845084e-02 -1.47228211e-01
5.43465137e-01 4.66436207e-01 -5.40480316e-01 -5.79058826e-01
-1.10645878e+00 4.58363026e-01 9.59174871e-01 1.23188034e-01
-4.31459755e-01 -3.47130626e-01 4.59736213e-02 -2.17630100e-02
4.74116266e-01 3.09863597e-01 -5.81697464e-01 4.92155731e-01
-6.86216533e-01 3.01973999e-01 2.37030208e-01 1.47700500e+00
7.35422552e-01 6.44756034e-02 -2.71615624e-01 5.88874698e-01
4.34214622e-01 1.08732390e+00 3.57092887e-01 -4.74624783e-01
4.86361653e-01 9.40467179e-01 5.11158884e-01 -1.45539200e+00
-4.74564582e-01 -5.49178958e-01 -8.05858910e-01 3.14245343e-01
3.92830044e-01 -3.69106159e-02 -9.29889381e-01 9.51602876e-01
8.05340767e-01 2.34651357e-01 4.83291335e-02 1.08486152e+00
8.50555837e-01 7.78107047e-01 -2.53886521e-01 4.90222797e-02
1.15617585e+00 -7.22806513e-01 -3.72994721e-01 -9.83001515e-02
3.71987015e-01 -7.25671053e-01 1.18344176e+00 4.70385432e-01
-6.13040745e-01 -8.41750979e-01 -1.08567011e+00 2.93665618e-01
-5.13036728e-01 9.05298352e-01 2.14633763e-01 2.50963569e-01
-5.04914224e-01 4.34227169e-01 -5.47323167e-01 -3.21722567e-01
5.87489963e-01 2.87095875e-01 -2.24267393e-01 6.20606132e-02
-6.16538525e-01 4.57976550e-01 5.56877673e-01 5.32382846e-01
-7.31109619e-01 -6.00187540e-01 -5.79851210e-01 -1.76852852e-01
5.54298818e-01 -2.96003252e-01 8.01690400e-01 -1.04281092e+00
-1.09085727e+00 3.80243212e-01 1.30117074e-01 -3.28721702e-01
6.67841792e-01 -5.86219788e-01 -2.92640865e-01 8.84078890e-02
1.70015916e-01 3.40802401e-01 1.18314803e+00 -1.27324259e+00
-8.89527738e-01 -6.45376861e-01 4.68746163e-02 5.00078440e-01
-4.55335855e-01 -3.04326534e-01 -1.70470461e-01 -6.02567613e-01
4.93714571e-01 -9.13745582e-01 -2.06057087e-01 3.64105850e-01
-5.12954950e-01 -1.65935904e-01 9.12800729e-01 -5.10172486e-01
9.36081290e-01 -2.24802542e+00 2.49189585e-01 4.14602160e-01
-1.91365741e-02 1.57153994e-01 -1.81360934e-02 1.99274197e-01
-7.96392187e-02 -1.08753696e-01 -1.86128244e-01 1.92833915e-01
-2.00294331e-01 -3.29321891e-01 -3.35866302e-01 7.74544001e-01
1.93297565e-01 2.25144103e-01 -8.51140976e-01 -3.41598958e-01
4.64442998e-01 3.04887623e-01 -2.34512642e-01 1.83221936e-01
-1.13702007e-01 2.34875619e-01 -7.34450936e-01 1.07100952e+00
1.11187148e+00 6.63888529e-02 -3.66992503e-01 -5.37709951e-01
-2.61613220e-01 -4.26311374e-01 -1.71865249e+00 1.03960776e+00
8.98028258e-03 1.13428898e-01 1.67758611e-03 -5.80547035e-01
1.20779836e+00 1.44715253e-02 3.76282334e-01 -2.21140087e-01
1.80167362e-01 1.48487866e-01 -2.51569092e-01 -3.97504061e-01
5.34675598e-01 4.52706724e-01 2.49790087e-01 -5.49613297e-01
-1.33803740e-01 -2.99869418e-01 1.37784984e-02 -1.56755060e-01
4.57743824e-01 3.88428092e-01 6.90307498e-01 -3.09523314e-01
7.19338655e-01 2.55956888e-01 5.61545849e-01 4.86541092e-01
-1.42727926e-01 5.99345207e-01 -7.74426088e-02 -6.32835805e-01
-7.83355415e-01 -8.40213656e-01 -3.90818745e-01 7.08085239e-01
4.11367476e-01 -3.10824811e-01 -5.44945180e-01 -8.15679848e-01
-7.07917884e-02 5.30493796e-01 -5.36647856e-01 1.07305489e-01
-3.38606507e-01 -7.43020415e-01 1.75219700e-01 4.66475695e-01
5.60522616e-01 -6.54910445e-01 -5.83390415e-01 -1.76707417e-01
-1.62444443e-01 -8.55260372e-01 -3.57887059e-01 1.72798671e-02
-9.21854258e-01 -1.31951654e+00 -8.22843850e-01 -4.15016711e-01
8.84995818e-01 7.18597949e-01 6.59315348e-01 9.67385471e-02
-5.06475508e-01 4.90960598e-01 -6.33025289e-01 -6.98984683e-01
2.25320518e-01 8.19998831e-02 1.93002507e-01 1.50159702e-01
3.55946153e-01 -2.25666121e-01 -7.61423707e-01 6.86582625e-01
-6.66013956e-01 -2.55321145e-01 6.34438753e-01 6.88687086e-01
9.61529851e-01 6.49584830e-02 1.98435336e-01 -4.34526175e-01
-9.82330590e-02 -5.94338119e-01 -1.10774064e+00 1.73737690e-01
-3.82682621e-01 -5.00698924e-01 8.26797187e-01 -4.10079688e-01
-5.28287947e-01 3.37743640e-01 1.40382707e-01 -7.04025686e-01
-5.70819616e-01 3.22310738e-02 -3.04180354e-01 -3.46796155e-01
9.54170763e-01 2.37036735e-01 -3.59480679e-01 -2.03902259e-01
2.47406974e-01 6.63844824e-01 1.14869900e-01 -3.95967036e-01
1.47546411e+00 5.41022301e-01 1.62248328e-01 -1.17985833e+00
-7.99058199e-01 -7.43691266e-01 -1.01796889e+00 -5.17623186e-01
6.61077499e-01 -1.05046880e+00 -2.59853184e-01 3.76057893e-01
-9.29460704e-01 -6.16585650e-03 -4.70508039e-01 5.04893601e-01
-3.82637233e-01 3.11764210e-01 4.57209758e-02 -1.04992509e+00
-3.49457383e-01 -1.00247025e+00 1.25622702e+00 5.36524296e-01
3.27138215e-01 -5.37939847e-01 -1.87624797e-01 -6.97905757e-03
-4.35695387e-02 3.99037212e-01 3.54032427e-01 -7.09917903e-01
-9.89293277e-01 -4.52295721e-01 -1.47384167e-01 2.82841086e-01
6.18502870e-02 4.83835757e-01 -8.14954102e-01 -4.72735882e-01
-1.97229400e-01 -1.61596984e-01 3.79779965e-01 5.38547814e-01
1.15057957e+00 -3.00680310e-01 -5.22556663e-01 9.19048071e-01
1.54669762e+00 2.97959924e-01 3.60925317e-01 5.59735239e-01
6.81172788e-01 4.49289501e-01 1.50159502e+00 7.00236320e-01
8.11795667e-02 6.61344409e-01 1.02203095e+00 -1.40009999e-01
2.51275152e-01 -9.26981121e-02 1.95820123e-01 2.46302575e-01
-4.97490048e-01 -1.93964243e-01 -7.66374052e-01 4.84587789e-01
-1.74321473e+00 -9.00042892e-01 -6.01215601e-01 2.33276463e+00
1.04675908e-02 -1.56614065e-01 1.26703486e-01 1.96977675e-01
6.46109760e-01 1.33522332e-01 -5.82005918e-01 5.95233083e-01
-1.94666117e-01 -2.91548550e-01 8.85931730e-01 2.37738982e-01
-1.61954188e+00 9.01954353e-01 4.89973450e+00 8.97589743e-01
-8.79358828e-01 -2.66042978e-01 3.40567619e-01 1.87457725e-01
3.22330594e-01 -2.56063044e-01 -1.22580004e+00 2.98457325e-01
3.26776475e-01 5.94993383e-02 2.57059425e-01 1.44918191e+00
1.74321234e-01 2.06954718e-01 -5.22694290e-01 9.81836081e-01
1.36349931e-01 -8.00603688e-01 1.58149824e-02 -9.33132097e-02
4.92424786e-01 1.67396571e-02 2.84496337e-01 1.25965133e-01
4.82379682e-02 -6.70258760e-01 8.78178298e-01 7.46151447e-01
6.41616404e-01 -7.69103110e-01 7.18219519e-01 3.74781072e-01
-1.42975390e+00 -4.76676524e-01 -8.66620839e-01 2.51176238e-01
-3.31629664e-01 2.77940333e-01 -9.59744394e-01 1.06993723e+00
1.09804904e+00 1.07520652e+00 -1.02160168e+00 1.48541415e+00
-2.22065464e-01 3.78566414e-01 -4.97249216e-01 -1.18242323e-01
2.81839013e-01 -4.76483256e-01 9.74899292e-01 8.36626410e-01
8.23005795e-01 2.36412019e-01 4.59049046e-01 5.69838524e-01
3.77807766e-01 5.95002294e-01 -9.03445721e-01 4.47451741e-01
4.76459175e-01 1.60739207e+00 -7.57870436e-01 -8.93657431e-02
-1.67946041e-01 7.15856731e-01 5.69945797e-02 2.52997398e-01
-1.00886762e+00 -4.05678064e-01 3.40222806e-01 3.23690385e-01
5.15132308e-01 -3.01760882e-01 9.12989974e-02 -1.03020728e+00
2.64853746e-01 -9.60138261e-01 4.58592296e-01 -8.75501573e-01
-1.24688172e+00 7.90914059e-01 3.36442798e-01 -2.06754375e+00
2.78701663e-01 -9.88477170e-01 -6.83891594e-01 6.95936799e-01
-1.49842846e+00 -1.43114960e+00 -9.60589468e-01 6.07094169e-01
7.88009524e-01 -3.46530110e-01 5.48826098e-01 1.49407372e-01
-5.63888073e-01 2.09632665e-01 2.97008455e-01 -2.42255013e-02
6.81198657e-01 -1.18322694e+00 -6.69775382e-02 1.03172672e+00
1.48640946e-01 3.27970952e-01 6.86145067e-01 -5.80512524e-01
-1.20270228e+00 -1.44740248e+00 -6.27203211e-02 -4.84844387e-01
5.25827646e-01 -2.19057411e-01 -7.76263177e-01 7.75973141e-01
-1.77481890e-01 2.60446459e-01 4.85047430e-01 -8.19849297e-02
-2.16786396e-02 -3.77093107e-01 -1.14390445e+00 5.83089948e-01
9.59985137e-01 2.16243580e-01 -4.27895725e-01 8.17131579e-01
4.28807735e-01 -3.93611461e-01 -7.70409107e-01 7.67245889e-01
2.71623284e-01 -9.58120406e-01 1.27438271e+00 -3.33183557e-01
-6.40759571e-03 -8.44876766e-01 -3.94640535e-01 -1.22445929e+00
-5.18918991e-01 -1.63695067e-01 1.42068535e-01 1.22020006e+00
1.60008878e-01 -5.61960816e-01 6.55536890e-01 -4.28390168e-02
-3.50519210e-01 -5.82330287e-01 -7.31898010e-01 -8.60441029e-01
-2.66163021e-01 -8.70959610e-02 6.91436231e-01 6.70462728e-01
-6.14444971e-01 9.56689566e-03 -3.07121754e-01 1.13216674e+00
7.51048326e-01 1.77235946e-01 1.13605964e+00 -1.38988769e+00
-6.83994144e-02 -1.28128842e-01 -6.03447437e-01 -1.11072922e+00
-1.15274593e-01 -7.38325059e-01 1.40313506e-01 -1.31036890e+00
-4.07705933e-01 -7.23374724e-01 1.23447925e-02 2.74683148e-01
-2.62814105e-01 1.43285081e-01 1.22986399e-01 4.56242412e-01
-5.50308645e-01 8.18076134e-01 9.70581710e-01 -2.58707672e-01
-2.79628724e-01 5.51814497e-01 -3.27800632e-01 1.04932678e+00
9.47679162e-01 -4.32839572e-01 -4.44927245e-01 -3.41264695e-01
-2.33273759e-01 -3.44634086e-01 6.54744983e-01 -1.55819118e+00
1.31360471e-01 -2.70072341e-01 7.96171486e-01 -1.17340970e+00
4.88202721e-01 -1.45145953e+00 -2.65033974e-04 3.02225858e-01
9.80793312e-02 5.15762977e-02 2.60655105e-01 7.68928766e-01
2.19094083e-02 -4.86571431e-01 8.19702029e-01 -1.43842280e-01
-6.53644383e-01 6.69913352e-01 7.75070488e-02 -1.67477876e-01
1.53389299e+00 -4.47004050e-01 -1.46941200e-01 9.80533566e-03
-5.22601843e-01 3.21028650e-01 5.45163035e-01 2.46592879e-01
8.32355142e-01 -1.13876498e+00 -6.71857059e-01 3.46761823e-01
4.64250684e-01 4.05287683e-01 4.04788017e-01 7.99249470e-01
-6.61751270e-01 1.92428693e-01 -1.63477242e-01 -8.85556102e-01
-1.49319398e+00 7.10921824e-01 4.23424542e-01 3.13004524e-01
-6.22610867e-01 5.87143242e-01 6.49478287e-02 -5.18151999e-01
3.41117680e-01 -6.30466580e-01 -6.00405633e-01 1.13014311e-01
5.76834440e-01 5.23804486e-01 -8.48473161e-02 -9.05465245e-01
-3.23038369e-01 1.01674199e+00 7.02733397e-02 4.78082180e-01
1.20083809e+00 -7.46136755e-02 1.61110312e-01 4.76951391e-01
5.72994828e-01 2.06430554e-01 -1.46700633e+00 -9.22673866e-02
1.99976210e-02 -9.29058552e-01 -1.96295992e-01 -4.38557833e-01
-7.80513525e-01 6.87866449e-01 1.04605949e+00 2.49601036e-01
1.07110918e+00 -3.09728354e-01 9.34269875e-02 6.51535690e-01
4.54182744e-01 -7.37143219e-01 -7.64294565e-02 1.92522436e-01
1.18427098e+00 -1.34274423e+00 2.68622428e-01 -5.85151374e-01
-6.02924824e-01 1.37179828e+00 8.55649471e-01 -4.25565958e-01
5.30134678e-01 -1.42490072e-02 -2.61222310e-02 -1.22450642e-01
1.84840001e-02 -2.31283903e-01 5.10431349e-01 7.83120096e-01
-2.61286069e-02 -7.02901185e-03 7.69503340e-02 4.08815563e-01
2.08177716e-02 -2.55892366e-01 3.57559264e-01 6.75379336e-01
-6.27152622e-01 -5.43712735e-01 -9.35814798e-01 3.25951308e-01
-1.60273835e-01 -1.29149295e-02 -2.85630882e-01 1.14387250e+00
2.82411397e-01 6.01949871e-01 -1.26321614e-01 -2.09099650e-01
8.20798218e-01 -1.28773212e-01 1.11333378e-01 -4.89528060e-01
-1.84166357e-01 2.55410522e-02 -3.59727621e-01 -3.72531950e-01
-3.14825624e-01 -7.69775391e-01 -9.41681862e-01 2.73625165e-01
-6.75798178e-01 1.41199827e-01 4.73770976e-01 3.16066146e-01
3.35622817e-01 4.65569884e-01 7.45522320e-01 -1.20702994e+00
-7.68419743e-01 -8.94990623e-01 -5.78530312e-01 8.44669938e-02
3.59342933e-01 -8.92272592e-01 -7.28043020e-01 -1.28068581e-01]
|
[8.692573547363281, -0.77676922082901]
|
27bf116c-1c9e-49b4-b44e-070291194fa1
|
unsupervised-mutual-transformer-learning-for
|
2305.02032
| null |
https://arxiv.org/abs/2305.02032v1
|
https://arxiv.org/pdf/2305.02032v1.pdf
|
Unsupervised Mutual Transformer Learning for Multi-Gigapixel Whole Slide Image Classification
|
Classification of gigapixel Whole Slide Images (WSIs) is an important prediction task in the emerging area of computational pathology. There has been a surge of research in deep learning models for WSI classification with clinical applications such as cancer detection or prediction of molecular mutations from WSIs. Most methods require expensive and labor-intensive manual annotations by expert pathologists. Weakly supervised Multiple Instance Learning (MIL) methods have recently demonstrated excellent performance; however, they still require large slide-level labeled training datasets that need a careful inspection of each slide by an expert pathologist. In this work, we propose a fully unsupervised WSI classification algorithm based on mutual transformer learning. Instances from gigapixel WSI (i.e., image patches) are transformed into a latent space and then inverse-transformed to the original space. Using the transformation loss, pseudo-labels are generated and cleaned using a transformer label-cleaner. The proposed transformer-based pseudo-label generation and cleaning modules mutually train each other iteratively in an unsupervised manner. A discriminative learning mechanism is introduced to improve normal versus cancerous instance labeling. In addition to unsupervised classification, we demonstrate the effectiveness of the proposed framework for weak supervision for cancer subtype classification as downstream analysis. Extensive experiments on four publicly available datasets show excellent performance compared to the state-of-the-art methods. We intend to make the source code of our algorithm publicly available soon.
|
['Nasir Rajpoot', 'Naoufel Werghi', 'Talha Qaiser', 'Arif Mahmood', 'Sajid Javed']
|
2023-05-03
| null | null | null | null |
['whole-slide-images', 'multiple-instance-learning', 'pseudo-label']
|
['computer-vision', 'methodology', 'miscellaneous']
|
[ 7.52018213e-01 2.60600924e-01 -5.37214160e-01 -5.26997328e-01
-1.33215773e+00 -3.97197932e-01 2.78376907e-01 5.04993081e-01
-4.25008088e-01 6.96483672e-01 -1.57724991e-01 -2.80381113e-01
-1.71868116e-01 -6.92066491e-01 -7.06455171e-01 -1.40731132e+00
4.09062564e-01 5.65353811e-01 1.33279011e-01 2.24351674e-01
3.31665911e-02 3.49941581e-01 -1.23400235e+00 4.89912838e-01
8.59316587e-01 8.88212144e-01 1.53598294e-01 4.30489033e-01
-3.41760308e-01 7.03132987e-01 -3.43965590e-01 -1.94704279e-01
9.89753008e-02 -3.86118054e-01 -8.14488947e-01 1.32187277e-01
2.83439159e-01 3.14120799e-01 2.19696276e-02 1.07919872e+00
3.60428542e-01 -2.35821158e-01 7.04902411e-01 -1.09323454e+00
-1.72737762e-01 4.76765186e-01 -7.00209796e-01 1.95596740e-02
-3.12068135e-01 2.36960594e-02 1.11026967e+00 -6.96327567e-01
8.79854977e-01 7.09360480e-01 6.72219992e-01 5.88444114e-01
-1.34828937e+00 -8.43369782e-01 -1.06850632e-01 2.33404681e-01
-1.50024402e+00 -3.29120547e-01 6.56631708e-01 -3.65293801e-01
6.87764227e-01 5.20846665e-01 4.29101020e-01 7.52648652e-01
2.32565507e-01 9.14601624e-01 1.17985249e+00 -4.44587827e-01
8.33927989e-02 4.04550254e-01 3.81902158e-01 9.32495773e-01
6.77433908e-02 -3.32037956e-01 -4.49449062e-01 -8.27983320e-02
2.99683303e-01 4.67286438e-01 -1.98016480e-01 -2.33317897e-01
-1.32766187e+00 6.41407192e-01 4.44269359e-01 5.87496221e-01
1.26444608e-01 -1.09228015e-01 4.55106616e-01 2.64873236e-01
8.03355575e-01 2.21074939e-01 -3.37975174e-01 3.06252360e-01
-1.01641965e+00 -2.67205745e-01 4.18335855e-01 7.47731924e-01
7.44949639e-01 -7.53139198e-01 -2.26653129e-01 9.94481981e-01
2.24143684e-01 2.11438179e-01 6.33755326e-01 -1.94216236e-01
1.91634670e-01 1.02829456e+00 -3.47093582e-01 -8.18031967e-01
-5.45038223e-01 -7.38936484e-01 -1.18123281e+00 -6.63783997e-02
2.92944700e-01 3.85505289e-01 -9.60113406e-01 1.44837570e+00
4.78560716e-01 6.25149906e-01 5.81552200e-02 6.18743181e-01
7.87066400e-01 4.04687136e-01 2.16417059e-01 -2.77369380e-01
1.30960298e+00 -9.90143120e-01 -7.44185925e-01 7.95109197e-02
1.01815832e+00 -5.43494999e-01 8.45832229e-01 2.59097844e-01
-7.38452971e-01 -1.46904811e-01 -9.56727266e-01 -1.59203172e-01
-4.27659422e-01 6.01832032e-01 4.82892722e-01 3.20114434e-01
-7.86300898e-01 3.97748202e-01 -1.24361444e+00 -4.41052467e-01
8.46981883e-01 5.48383176e-01 -7.38206267e-01 5.87325804e-02
-6.92958951e-01 5.62707484e-01 2.44708866e-01 1.77337989e-01
-8.82437170e-01 -8.66717696e-01 -9.43437636e-01 2.79698446e-02
2.92085439e-01 -3.18808883e-01 8.97149622e-01 -8.72040153e-01
-1.47725511e+00 1.44830370e+00 -4.87882674e-01 -2.60691136e-01
3.08457792e-01 3.87920797e-01 -2.26292893e-01 3.32827568e-02
2.65737563e-01 1.68801337e-01 6.13502502e-01 -1.09233642e+00
-8.37031484e-01 -5.95969081e-01 -3.36635917e-01 -7.45805055e-02
-6.31137252e-01 -3.94765913e-01 -4.44965035e-01 -5.97845316e-01
3.29015315e-01 -1.06009579e+00 -2.02709749e-01 7.11405426e-02
-8.22503328e-01 -3.15027207e-01 7.90395916e-01 -5.49455106e-01
1.04784584e+00 -2.34690857e+00 2.46649891e-01 4.36355740e-01
3.37269574e-01 3.62513959e-01 -1.45348027e-01 8.87200609e-02
-2.26281777e-01 1.88331544e-01 -3.68189216e-01 -6.96126640e-01
-1.78238720e-01 7.55394846e-02 9.66663063e-02 8.38651717e-01
2.35522494e-01 8.50913703e-01 -1.01267827e+00 -6.74852431e-01
7.58275911e-02 2.90041894e-01 -2.42412388e-01 2.84190685e-01
9.88551229e-03 7.24892199e-01 -1.80605754e-01 7.85746038e-01
4.85983878e-01 -7.56511867e-01 2.12407306e-01 -4.60577071e-01
8.33357424e-02 1.37876883e-01 -8.63043368e-01 1.75218832e+00
-4.29387152e-01 4.38764155e-01 3.07177007e-02 -1.43595254e+00
6.44837558e-01 3.67109835e-01 4.20003593e-01 -2.82436818e-01
6.02045096e-02 4.12209094e-01 -9.37349573e-02 -5.85482359e-01
-4.05200005e-01 -4.50409353e-01 1.07739024e-01 2.90948808e-01
1.66604847e-01 1.52909890e-01 2.64072537e-01 1.52337655e-01
1.33847630e+00 -2.94618070e-01 3.95232558e-01 -2.53297806e-01
9.73966300e-01 1.09920479e-01 9.33501005e-01 4.25049603e-01
-1.48387685e-01 3.35282505e-01 3.90380561e-01 -3.77291650e-01
-8.39337528e-01 -8.21448267e-01 -5.76955855e-01 8.28846157e-01
1.02247179e-01 -2.09120169e-01 -6.15490317e-01 -1.07566249e+00
-5.68832383e-02 1.79433212e-01 -9.29289162e-01 -7.65211210e-02
-4.23637748e-01 -1.04865324e+00 5.52116871e-01 3.50794822e-01
1.69409916e-01 -7.10172415e-01 -7.96812400e-03 1.16557077e-01
-2.07820043e-01 -7.43479669e-01 -3.00706089e-01 3.70974094e-01
-6.01161122e-01 -1.19506323e+00 -5.32263517e-01 -1.22976696e+00
1.27443945e+00 3.20338964e-01 6.58636212e-01 3.39566618e-01
-9.09604251e-01 -1.18489467e-01 -2.14224368e-01 -3.72944921e-01
-5.68432689e-01 2.42675200e-01 -1.91772133e-01 4.40016061e-01
4.30939406e-01 -3.04656118e-01 -7.57367611e-01 2.80188739e-01
-1.15879619e+00 3.18890035e-01 8.86965632e-01 1.24239993e+00
1.19622040e+00 8.79904479e-02 5.51868796e-01 -1.42301309e+00
1.19602531e-02 -5.33427835e-01 -5.27557909e-01 4.55001563e-01
-2.67219752e-01 1.40861720e-01 5.83092749e-01 -2.24778414e-01
-1.14628077e+00 1.96126685e-01 -2.64127016e-01 -6.31578565e-02
-2.73449928e-01 6.91119611e-01 -1.81637466e-01 -2.43198201e-01
3.88455868e-01 2.40725100e-01 1.64508462e-01 -2.11039469e-01
-1.03183173e-01 8.55225027e-01 2.77943552e-01 5.93420025e-03
8.00952971e-01 8.75250101e-01 1.23204157e-01 -6.40476108e-01
-1.27500319e+00 -8.76831949e-01 -5.09456277e-01 7.94801116e-02
8.92094016e-01 -6.38816297e-01 -7.36333847e-01 6.47957683e-01
-5.35081387e-01 -4.41169709e-01 -7.47919455e-02 2.86496103e-01
-2.95199215e-01 2.19225779e-01 -8.91002297e-01 -2.87358761e-01
-4.57877249e-01 -1.22387338e+00 1.43526852e+00 2.40202531e-01
-2.46197321e-02 -1.08469069e+00 3.61082882e-01 6.30932450e-01
1.93149343e-01 2.07272783e-01 1.18401313e+00 -8.91566336e-01
-5.30909061e-01 -5.29941082e-01 -1.70506984e-01 1.81396902e-01
6.31760120e-01 -4.91159223e-02 -1.06546521e+00 -4.26682800e-01
-2.09462389e-01 -3.94927233e-01 9.98785257e-01 2.61679500e-01
1.40751362e+00 -1.43289015e-01 -1.01688564e+00 8.31509888e-01
1.60446858e+00 1.96398050e-02 2.49312162e-01 2.72845745e-01
7.37127423e-01 6.01288557e-01 7.67942071e-01 2.13745862e-01
1.95368201e-01 3.96616131e-01 2.08081841e-01 -4.97569054e-01
-6.06293082e-02 4.79022563e-02 -1.26461253e-01 8.18824410e-01
2.32746825e-01 -2.49489576e-01 -9.38155353e-01 5.61233044e-01
-1.69959497e+00 -7.67538488e-01 -6.53858110e-02 2.14875937e+00
1.16481662e+00 -6.71043471e-02 -4.99432653e-01 3.77284884e-01
6.00482047e-01 -1.42765030e-01 -6.23868346e-01 3.11731964e-01
2.41573453e-02 1.89891592e-01 3.66975665e-01 4.96917248e-01
-1.30379128e+00 7.75950551e-01 4.46176863e+00 1.07981801e+00
-1.21472943e+00 2.55906314e-01 1.20553434e+00 -4.84321602e-02
-1.42286748e-01 -2.65637130e-01 -9.61985290e-01 3.77751410e-01
5.85341692e-01 4.98079807e-02 -1.12682007e-01 5.51927388e-01
1.37536764e-01 -2.34502684e-02 -1.11713278e+00 9.62142885e-01
1.00388080e-01 -1.62043643e+00 -1.50585234e-01 1.24163069e-01
7.97293007e-01 1.51057588e-02 6.52849972e-02 1.13449335e-01
-8.92214244e-04 -8.43825638e-01 5.82668297e-02 4.96601611e-01
9.95170772e-01 -7.12487757e-01 1.20254040e+00 3.82152021e-01
-1.10960686e+00 -4.05285545e-02 -2.26547480e-01 5.32889426e-01
-1.77264467e-01 9.85824108e-01 -1.05049622e+00 4.63017732e-01
5.69418907e-01 9.34480548e-01 -5.82974434e-01 8.79282773e-01
-1.11953653e-01 7.38394618e-01 -2.02466592e-01 7.55193532e-02
8.30844119e-02 -1.11324243e-01 1.87356293e-01 1.29731309e+00
2.97637582e-01 1.21871337e-01 1.61340296e-01 4.71059620e-01
-1.39754727e-01 2.41652057e-01 -3.27647626e-01 -9.24832523e-02
2.02373669e-01 1.79868519e+00 -1.03406131e+00 -5.11510551e-01
-3.22576642e-01 9.18576360e-01 5.04095614e-01 1.41101360e-01
-6.63349509e-01 -2.49540195e-01 5.19111574e-01 2.36204602e-02
-6.12752810e-02 4.72042829e-01 -2.62567699e-01 -1.27263212e+00
-1.26126837e-02 -6.21642530e-01 5.54314137e-01 -4.05597687e-02
-1.52445555e+00 4.36089337e-01 -4.00416940e-01 -1.39755511e+00
9.78610441e-02 -6.42882526e-01 -6.41393602e-01 4.57403630e-01
-1.89346623e+00 -1.28172553e+00 -5.81829190e-01 4.30207849e-01
3.59791934e-01 -1.26263201e-01 1.03285646e+00 3.46196562e-01
-8.64302576e-01 9.17731404e-01 4.30160910e-01 2.47645780e-01
1.00866866e+00 -1.31204045e+00 -1.51117176e-01 5.09904563e-01
-2.29470693e-02 4.96606171e-01 3.40700895e-01 -5.91693521e-01
-1.23152947e+00 -1.52777040e+00 9.24269915e-01 -1.35570914e-01
7.08325028e-01 -4.90192592e-01 -9.28279042e-01 7.47014165e-01
-6.71011303e-03 4.46624637e-01 1.49545920e+00 -1.54748768e-01
-1.21805839e-01 -3.88049096e-01 -1.27192140e+00 3.77532244e-01
8.33137751e-01 -3.71336490e-01 3.27534950e-03 7.84952998e-01
3.29873264e-01 -3.74721795e-01 -1.00306034e+00 5.05186617e-01
2.63411850e-01 -5.54191351e-01 7.29709029e-01 -3.24815661e-01
2.26496994e-01 -5.50414920e-01 4.02680039e-01 -1.21246111e+00
-3.13123494e-01 -1.83042780e-01 3.63464326e-01 1.30207193e+00
5.39960444e-01 -7.34685242e-01 1.04799998e+00 2.97680795e-01
-2.56220698e-01 -1.11626709e+00 -9.85089123e-01 -4.34353828e-01
-2.22858399e-01 -7.32438192e-02 2.29850560e-01 1.05263996e+00
3.87178630e-01 1.62819147e-01 8.42248574e-02 3.01788062e-01
8.15260053e-01 2.25211695e-01 5.49563646e-01 -1.17397928e+00
-3.54875177e-01 -3.35523814e-01 -6.81159794e-01 -3.27172995e-01
3.48115623e-01 -1.40634310e+00 2.38025725e-01 -1.38878560e+00
7.18098044e-01 -7.22843826e-01 -6.92958474e-01 8.39261174e-01
-4.50487673e-01 6.03107214e-01 -2.95755327e-01 3.62229943e-01
-7.76714981e-01 3.22799057e-01 9.15736198e-01 -6.46007776e-01
1.31437043e-02 -1.09558646e-02 -5.29372931e-01 6.72654688e-01
7.83976018e-01 -6.62570953e-01 -2.49974057e-01 -1.34386972e-01
1.22481491e-02 -1.11307472e-01 1.95885241e-01 -9.31793749e-01
2.96758562e-01 -4.08943668e-02 2.63600588e-01 -4.72849578e-01
1.41587287e-01 -8.01722527e-01 1.58301249e-01 5.71914613e-01
-6.55674636e-01 -5.79376042e-01 -4.84924577e-02 6.44562125e-01
-4.05254036e-01 -1.67670250e-01 1.00835133e+00 1.28334993e-03
-2.11899415e-01 6.32185161e-01 -2.67187148e-01 -3.67004871e-01
1.35050249e+00 -7.61903683e-03 -2.84916103e-01 2.52799571e-01
-8.88616800e-01 2.74854511e-01 3.56361181e-01 -1.94131245e-03
4.32254583e-01 -1.07460105e+00 -8.28904927e-01 3.08906049e-01
5.33486843e-01 3.32965404e-01 4.91855592e-01 1.16538942e+00
-3.64538223e-01 4.66289818e-01 1.71449915e-01 -8.53190482e-01
-1.53108656e+00 3.81346762e-01 2.28806838e-01 -9.44371641e-01
-4.87018436e-01 1.23058212e+00 5.33169866e-01 -6.14390373e-01
2.51655668e-01 -3.24054182e-01 -1.83414802e-01 -8.67870227e-02
7.15822816e-01 9.90585191e-04 3.66119176e-01 -4.05884117e-01
-4.07320589e-01 4.52022582e-01 -5.07404685e-01 2.73172259e-01
1.51185870e+00 1.25469029e-01 -5.11464357e-01 4.95406836e-01
1.57718241e+00 -8.42971280e-02 -9.27215874e-01 -3.59984070e-01
1.69851869e-01 -2.63484836e-01 -9.27262530e-02 -6.49540007e-01
-1.17647088e+00 7.96528876e-01 8.12927842e-01 -1.03755273e-01
1.19510281e+00 2.05256790e-01 7.38398492e-01 2.97907889e-01
1.92386940e-01 -8.58703196e-01 -9.67106670e-02 -3.55220363e-02
3.96416575e-01 -1.47619021e+00 -1.64060026e-01 -6.36839449e-01
-2.41174579e-01 9.68354881e-01 4.62730765e-01 2.42054611e-02
6.58403337e-01 5.64257622e-01 3.32422882e-01 -1.79763243e-01
-7.61026919e-01 1.10121360e-02 1.35395512e-01 2.57115394e-01
5.95193684e-01 1.13307863e-01 -3.28269750e-01 6.41114712e-01
2.80785173e-01 1.36884034e-01 1.04650423e-01 9.50613856e-01
-1.45810246e-01 -1.31542158e+00 -1.45880535e-01 8.17637324e-01
-5.73228359e-01 -8.34914949e-03 -9.65359583e-02 4.96388227e-01
1.71300918e-01 7.07764089e-01 1.23865576e-02 -1.16066843e-01
-4.35635559e-02 7.78199881e-02 1.76936343e-01 -8.47889781e-01
-5.13184726e-01 1.21308833e-01 -3.19834650e-01 -4.23526406e-01
-5.32009840e-01 -6.33157670e-01 -1.50960696e+00 2.86389887e-01
-5.28031111e-01 2.60629326e-01 5.77449441e-01 1.16981602e+00
2.98182249e-01 5.26183367e-01 6.59122527e-01 -5.65382659e-01
-2.09131271e-01 -9.30880249e-01 -5.20167112e-01 6.81941688e-01
4.70607311e-01 -5.39644301e-01 -5.07857323e-01 3.45269024e-01]
|
[15.067697525024414, -2.7858808040618896]
|
d65fc99d-fb5b-495c-8a93-c1f89c1a09a5
|
semantic-code-search-for-smart-contracts
|
2111.14139
| null |
https://arxiv.org/abs/2111.14139v1
|
https://arxiv.org/pdf/2111.14139v1.pdf
|
Semantic Code Search for Smart Contracts
|
Semantic code search technology allows searching for existing code snippets through natural language, which can greatly improve programming efficiency. Smart contracts, programs that run on the blockchain, have a code reuse rate of more than 90%, which means developers have a great demand for semantic code search tools. However, the existing code search models still have a semantic gap between code and query, and perform poorly on specialized queries of smart contracts. In this paper, we propose a Multi-Modal Smart contract Code Search (MM-SCS) model. Specifically, we construct a Contract Elements Dependency Graph (CEDG) for MM-SCS as an additional modality to capture the data-flow and control-flow information of the code. To make the model more focused on the key contextual information, we use a multi-head attention network to generate embeddings for code features. In addition, we use a fine-tuned pretrained model to ensure the model's effectiveness when the training data is small. We compared MM-SCS with four state-of-the-art models on a dataset with 470K (code, docstring) pairs collected from Github and Etherscan. Experimental results show that MM-SCS achieves an MRR (Mean Reciprocal Rank) of 0.572, outperforming four state-of-the-art models UNIF, DeepCS, CARLCS-CNN, and TAB-CS by 34.2%, 59.3%, 36.8%, and 14.1%, respectively. Additionally, the search speed of MM-SCS is second only to UNIF, reaching 0.34s/query.
|
['Longxiang Gao', 'Jiangshan Yu', 'Yong Xiang', 'Chaochen Shi']
|
2021-11-28
| null | null | null | null |
['code-search', 'code-search']
|
['computer-code', 'computer-vision']
|
[-5.26923597e-01 -3.07546526e-01 -7.52864242e-01 -8.25082138e-02
-8.00219238e-01 -7.88991094e-01 5.27068257e-01 -1.19339131e-01
-1.04176611e-01 1.21212468e-01 3.83265227e-01 -8.05685818e-01
1.59444213e-01 -6.30628467e-01 -7.44311631e-01 -3.71671438e-01
1.28641620e-01 2.06665322e-01 3.05144280e-01 -2.89435863e-01
5.52252591e-01 -2.01906055e-01 -1.01619434e+00 3.35871875e-01
1.10813296e+00 8.99470866e-01 4.63690102e-01 2.92479694e-01
-4.19036716e-01 9.35836673e-01 -3.42379332e-01 -7.80512393e-01
-1.20453630e-02 -5.06741442e-02 -8.63968551e-01 -1.06132996e+00
-6.79147691e-02 -4.44804102e-01 -6.27434731e-01 1.40138388e+00
2.49496341e-01 -4.41607952e-01 1.73710242e-01 -1.22497284e+00
-1.15857911e+00 1.06321192e+00 -4.71429825e-01 1.29707739e-01
3.19699138e-01 2.67816842e-01 1.39202726e+00 -8.98049355e-01
7.25390434e-01 9.83299494e-01 6.22685850e-01 6.39868021e-01
-9.23087358e-01 -1.13671768e+00 -6.93388879e-02 2.57895947e-01
-1.15307593e+00 -1.77791059e-01 6.58273637e-01 -6.02828443e-01
1.37963772e+00 1.78357624e-02 5.20201087e-01 1.08044505e+00
3.26959521e-01 9.09666479e-01 5.66132188e-01 -6.29281923e-02
-4.10711691e-02 -8.12592804e-02 -1.18390612e-01 8.25132430e-01
2.08661348e-01 2.25530595e-01 -2.43779778e-01 -4.56405044e-01
5.14299393e-01 2.97745258e-01 -2.76332140e-01 -2.68955350e-01
-1.13083363e+00 1.10257769e+00 6.75139666e-01 7.06978202e-01
1.25026271e-01 8.17871630e-01 8.93003345e-01 3.49056721e-01
1.04053266e-01 6.09932005e-01 -7.36076474e-01 -6.68317318e-01
-7.52026439e-01 1.64106473e-01 9.32845592e-01 1.44463480e+00
6.77147686e-01 -2.17231929e-01 -1.39485776e-01 6.50004804e-01
5.59692502e-01 5.54223537e-01 6.61668003e-01 -8.39740157e-01
9.49851096e-01 1.05749440e+00 -1.63950488e-01 -9.28786397e-01
7.61516020e-03 -6.54869735e-01 -4.29703027e-01 -3.61423761e-01
-6.47757128e-02 2.96766847e-01 -5.72282135e-01 1.60746777e+00
-1.98475830e-02 -1.21062018e-01 -7.11423904e-02 8.17175686e-01
7.12777734e-01 5.98644674e-01 -1.86389878e-01 2.70888090e-01
1.24619031e+00 -1.50871217e+00 -2.79353619e-01 -2.42585018e-01
1.13177109e+00 -7.41247714e-01 1.33448637e+00 -6.58499673e-02
-6.14479125e-01 -2.18466327e-01 -8.14291120e-01 -2.00593248e-01
-3.99890512e-01 6.56462926e-03 7.87867606e-01 4.47131544e-01
-9.11110222e-01 4.02667969e-01 -7.58059025e-01 -9.08679739e-02
4.98041928e-01 1.18708029e-01 -1.61025360e-01 -1.57032296e-01
-1.27179050e+00 5.87542892e-01 2.83820778e-01 -3.74708503e-01
-1.17438102e+00 -8.87153685e-01 -7.40167856e-01 5.50974786e-01
3.31727266e-01 -2.25844711e-01 1.21640551e+00 -6.31163836e-01
-1.14222407e+00 7.22530782e-01 2.46087089e-03 -1.15572333e-01
2.55619913e-01 -7.09240213e-02 -5.42016149e-01 -5.22590503e-02
3.96496624e-01 1.48399323e-01 3.44039142e-01 -1.00298464e+00
-4.56699640e-01 -9.96880755e-02 4.50811863e-01 -4.50335503e-01
-2.73014456e-01 4.11017120e-01 -9.56761479e-01 -7.79057801e-01
-3.49699736e-01 -1.00988960e+00 9.76507962e-02 -2.98341423e-01
-1.38035148e-01 -5.44484496e-01 7.01176107e-01 -7.00044453e-01
1.82524002e+00 -2.47611547e+00 7.89318904e-02 1.01673752e-01
2.33724952e-01 3.61962467e-01 -4.04722631e-01 7.23133802e-01
9.77942124e-02 5.10173023e-01 -2.96962589e-01 -2.85614133e-02
3.67883444e-01 1.62299536e-02 -2.92974174e-01 2.88452476e-01
-1.15517251e-01 1.44064367e+00 -1.08845794e+00 -3.76009643e-01
-3.67225885e-01 1.08156621e-01 -1.06733000e+00 1.29962578e-01
-5.59544623e-01 6.52830675e-02 -8.64532769e-01 9.67445433e-01
3.62029076e-01 -6.57966733e-01 2.58163244e-01 2.04643279e-01
-1.91106871e-01 5.35385370e-01 -3.29834789e-01 2.27520013e+00
-8.65293682e-01 5.58082521e-01 -3.06564756e-02 -7.96717584e-01
7.56550014e-01 2.02953458e-01 4.17005539e-01 -1.21587002e+00
-2.42949724e-02 6.68619335e-01 -1.01392597e-01 -6.67692065e-01
2.33552232e-01 4.75237846e-01 -5.52054822e-01 7.19940484e-01
-2.41239831e-01 3.37653846e-01 2.06858758e-02 3.82112890e-01
1.44419611e+00 6.41990006e-02 -8.07935968e-02 -5.29596627e-01
5.47516167e-01 -6.85489178e-02 7.33886719e-01 6.67177022e-01
-8.06533545e-02 2.09613770e-01 9.36465204e-01 -5.80729842e-01
-8.66095960e-01 -5.65804064e-01 6.72791600e-02 1.23026240e+00
3.92189801e-01 -8.55452299e-01 -7.90615201e-01 -1.09454858e+00
2.91254491e-01 5.20716608e-01 -6.37772143e-01 -3.87854218e-01
-8.19881737e-01 -2.75910914e-01 6.18566036e-01 5.94868064e-01
7.09378302e-01 -9.61936414e-01 -6.14456296e-01 3.20370167e-01
-3.98130566e-01 -6.94486976e-01 -1.15597701e+00 -2.15892866e-01
-5.30360341e-01 -1.48334575e+00 -4.45724219e-01 -9.26925361e-01
5.83713174e-01 1.15757264e-01 1.23344767e+00 8.23940933e-01
9.90461558e-04 -8.73400047e-02 -6.43092394e-01 4.15510982e-01
-4.47947800e-01 4.15903211e-01 -4.56314564e-01 -5.73430300e-01
4.27357525e-01 -3.79526973e-01 -9.63420868e-01 4.77305084e-01
-8.67946565e-01 -1.40829548e-01 6.08552694e-01 1.17855132e+00
2.01753840e-01 -3.81912082e-01 4.65265900e-01 -8.22357357e-01
6.30189478e-01 -9.14898574e-01 -1.00568283e+00 4.67555612e-01
-1.33568478e+00 2.35823706e-01 8.55175316e-01 -2.64218718e-01
-8.23464990e-01 -2.12596044e-01 -1.38703108e-01 -5.69978952e-01
6.78919077e-01 9.23553288e-01 1.52261863e-02 -2.83051372e-01
4.47943866e-01 3.43406767e-01 -1.47967383e-01 -7.34035611e-01
3.58577728e-01 9.28387702e-01 4.02987868e-01 -7.49746263e-01
6.23685360e-01 7.88374096e-02 -5.05301714e-01 3.67049307e-01
-2.01566249e-01 -4.95501906e-01 1.31202444e-01 3.72627944e-01
6.41577065e-01 -8.06448936e-01 -8.80875170e-01 3.54170918e-01
-1.29903495e+00 -4.30855721e-01 3.48554313e-01 2.65332460e-01
-3.85133445e-01 4.89745170e-01 -7.90423512e-01 -1.85868368e-01
-7.21388757e-01 -1.76923072e+00 9.68333483e-01 5.37095964e-02
5.89938425e-02 -7.88747549e-01 1.77315190e-01 4.95451778e-01
8.69893193e-01 -1.32409051e-01 1.35483599e+00 -7.31186390e-01
-9.78459418e-01 -1.54670879e-01 -5.94998837e-01 1.50943011e-01
-9.85281635e-03 -1.83337241e-01 -4.68487769e-01 -5.27947664e-01
-2.05373228e-01 -1.82586104e-01 7.99153805e-01 -1.66692957e-01
1.40026748e+00 -6.77522480e-01 -5.91275275e-01 9.11061585e-01
1.72579157e+00 5.78438580e-01 5.26335537e-01 5.52851796e-01
6.72018528e-01 -3.41248363e-02 3.68794292e-01 3.36766839e-01
7.51681268e-01 8.00623119e-01 8.55399370e-01 4.63042468e-01
1.28120482e-02 -6.28871918e-01 3.28705251e-01 1.10185194e+00
3.06075454e-01 8.08543786e-02 -1.34514844e+00 8.40362370e-01
-1.90766120e+00 -7.83963680e-01 -4.03707672e-04 1.82568061e+00
1.06038713e+00 -5.74717969e-02 -2.30047554e-01 -4.86430556e-01
5.15472114e-01 1.49331972e-01 -6.29149079e-01 -3.17607164e-01
3.08442861e-01 1.10346422e-01 6.73335314e-01 2.74580717e-01
-8.52214992e-01 9.58805263e-01 5.20742321e+00 9.92576063e-01
-1.10508001e+00 4.21310157e-01 2.12020576e-01 7.52512962e-02
-8.57759535e-01 5.35097599e-01 -3.98232371e-01 1.29003215e+00
7.15104103e-01 -4.39246625e-01 1.00631416e+00 1.31861591e+00
-3.87738466e-01 3.35385472e-01 -1.04883218e+00 1.06089675e+00
-2.30746001e-01 -1.70414591e+00 -3.52240533e-01 5.00016883e-02
1.08963251e+00 6.51144862e-01 -1.02078892e-01 7.04595089e-01
6.66626453e-01 -1.01776266e+00 7.43006825e-01 1.99211255e-01
1.10652244e+00 -4.30441946e-01 8.03402781e-01 1.76092759e-01
-1.38371849e+00 -5.70039988e-01 -9.48696360e-02 3.90149653e-01
-3.34386647e-01 2.39129588e-01 -2.87529558e-01 5.38826942e-01
8.04514647e-01 8.65020573e-01 -5.72402120e-01 1.00357938e+00
-2.14892074e-01 4.39285874e-01 -5.38009889e-02 -4.44116116e-01
5.09047806e-01 1.15573756e-01 9.96913686e-02 1.39738488e+00
5.64344347e-01 -2.67564893e-01 -3.62212360e-02 1.33920455e+00
-5.19544244e-01 -1.49247617e-01 -3.57109308e-01 -4.13260430e-01
9.40043211e-01 9.04445589e-01 -2.10026085e-01 -3.36221218e-01
-8.89970243e-01 7.01611280e-01 3.10214132e-01 3.94267052e-01
-9.92001235e-01 -8.57454598e-01 7.53021955e-01 -2.54304469e-01
5.35728097e-01 4.69017699e-02 4.09702547e-02 -1.34616697e+00
3.65452170e-01 -1.18989348e+00 4.23521340e-01 -5.03490090e-01
-9.85144973e-01 4.16141570e-01 -2.78497994e-01 -1.04200554e+00
-1.84306130e-01 -2.29696512e-01 -7.26531625e-01 8.39181781e-01
-1.58179915e+00 -1.09003842e+00 -8.84536505e-02 3.21417242e-01
3.34039271e-01 -4.91544515e-01 6.11400902e-01 6.46714985e-01
-3.65073740e-01 9.03973579e-01 5.53313017e-01 5.16900480e-01
3.23620498e-01 -9.85462189e-01 8.48454118e-01 6.51126027e-01
-6.77830204e-02 1.16871774e+00 1.07364491e-01 -6.20711327e-01
-2.01219773e+00 -9.05645847e-01 1.02425504e+00 -3.83371979e-01
1.09312773e+00 -4.52172250e-01 -8.69342387e-01 5.54068148e-01
1.71182841e-01 1.88060552e-01 3.56103569e-01 -5.83630130e-02
-1.01475334e+00 -2.92208225e-01 -8.49107385e-01 2.99707115e-01
1.27032518e+00 -1.11993766e+00 -3.85275543e-01 1.63858026e-01
1.11862826e+00 -4.21840400e-01 -1.08258700e+00 2.01611206e-01
6.97019279e-01 -9.27759886e-01 6.76811516e-01 -5.60113072e-01
8.69799972e-01 -1.56767160e-01 -1.89750373e-01 -1.18701279e+00
-3.30025524e-01 -7.49715745e-01 -2.53841847e-01 1.16725850e+00
4.54504013e-01 -8.13409626e-01 6.35319173e-01 4.51508760e-01
-3.44743937e-01 -1.03037298e+00 -1.07166326e+00 -9.21539307e-01
3.18994462e-01 -2.38401189e-01 1.29511058e+00 1.27446306e+00
5.06376147e-01 -1.36677518e-01 2.17160378e-02 -3.18754405e-01
2.21080139e-01 8.26112330e-01 5.85566521e-01 -8.35370600e-01
-4.40013677e-01 -9.73698080e-01 -7.17003420e-02 -1.25036454e+00
3.95012885e-01 -1.25449085e+00 -2.91780174e-01 -1.44987965e+00
5.20459414e-01 -7.23248780e-01 -5.03239095e-01 7.37996697e-01
-3.93816680e-02 -5.13725162e-01 8.95712599e-02 6.07927799e-01
-5.42592943e-01 6.54701352e-01 1.13513625e+00 -5.11867285e-01
2.96242177e-01 -3.32792580e-01 -8.75794291e-01 1.14795707e-01
6.78177893e-01 -6.95715368e-01 -3.13186318e-01 -1.07742882e+00
7.30585754e-01 2.77654737e-01 1.70017481e-01 -3.58904392e-01
5.21070242e-01 -1.71617582e-01 -5.32157421e-01 -2.62048215e-01
-4.73103195e-01 -8.55309129e-01 3.66122067e-01 8.58258545e-01
-2.82071114e-01 2.59364128e-01 -1.61501929e-01 5.76793611e-01
-4.33306247e-01 -4.75254834e-01 3.74645621e-01 -2.82093585e-01
-6.35467172e-01 4.43889737e-01 2.18666628e-01 3.47813129e-01
6.94677591e-01 3.10318649e-01 -9.88072634e-01 4.23615724e-02
1.57955214e-01 6.58241272e-01 8.03856611e-01 9.64696229e-01
3.66945237e-01 -1.54246318e+00 -3.16404432e-01 1.98426381e-01
6.02676868e-01 -1.58095896e-01 -1.81611460e-02 8.14587057e-01
-7.46647894e-01 7.84284532e-01 1.58193991e-01 -1.25816628e-01
-7.31624842e-01 7.44550049e-01 1.62682667e-01 -3.60363990e-01
-5.04113674e-01 7.59110987e-01 1.39357001e-01 -5.05873263e-01
1.07434466e-01 -4.90348339e-01 2.13553801e-01 -3.59105885e-01
2.72539139e-01 1.82785764e-01 -1.59253225e-01 -3.89183104e-01
-7.13204384e-01 6.86830044e-01 -7.07116351e-02 5.12618482e-01
1.30713892e+00 3.54836345e-01 -5.68911493e-01 -3.71907145e-01
1.70519316e+00 2.05115512e-01 -1.03934872e+00 -3.22350025e-01
2.92526484e-01 -7.43697584e-01 -3.13851750e-03 -9.47040260e-01
-1.61788666e+00 8.05284560e-01 2.27807254e-01 2.67063797e-01
7.79976785e-01 3.81126076e-01 1.00203478e+00 2.87093848e-01
7.32666314e-01 -9.69163656e-01 1.11089811e-01 6.47699237e-01
8.05058539e-01 -1.09452355e+00 -4.58572716e-01 4.11505885e-02
-2.21307904e-01 9.74856853e-01 5.57959557e-01 1.18510880e-01
5.56148410e-01 1.60109177e-01 -2.29748681e-01 -5.03859460e-01
-9.09225583e-01 3.21245909e-01 -1.19453622e-02 2.06771150e-01
2.19892010e-01 -5.92350066e-02 -3.91454011e-01 9.59301829e-01
1.31274879e-01 1.14713162e-01 1.91772178e-01 9.15589094e-01
-1.40726745e-01 -1.18545830e+00 3.13843459e-01 4.51506883e-01
-7.32929230e-01 -5.49845755e-01 -1.30349830e-01 6.48542643e-01
-1.58781588e-01 7.83050358e-01 -2.90622562e-01 -8.28772902e-01
1.84188798e-01 -1.28103942e-01 -1.13525070e-01 -4.73667771e-01
-8.72462034e-01 -3.01444948e-01 -6.07078895e-02 -9.50515032e-01
1.50393341e-02 -3.47672015e-01 -1.30203807e+00 -4.84495491e-01
-4.82662141e-01 4.01702881e-01 8.05364430e-01 5.55550039e-01
7.90504754e-01 2.08700985e-01 8.14373970e-01 3.58596407e-02
-7.24048316e-01 -7.47479498e-01 -8.01815614e-02 3.75697792e-01
3.09154570e-01 -6.28517151e-01 -4.50837851e-01 -2.30893910e-01]
|
[7.487397193908691, 8.067889213562012]
|
ba9debeb-41c1-437b-8f25-55c08898d6f0
|
metacorrection-domain-aware-meta-loss
|
2103.05254
| null |
https://arxiv.org/abs/2103.05254v1
|
https://arxiv.org/pdf/2103.05254v1.pdf
|
MetaCorrection: Domain-aware Meta Loss Correction for Unsupervised Domain Adaptation in Semantic Segmentation
|
Unsupervised domain adaptation (UDA) aims to transfer the knowledge from the labeled source domain to the unlabeled target domain. Existing self-training based UDA approaches assign pseudo labels for target data and treat them as ground truth labels to fully leverage unlabeled target data for model adaptation. However, the generated pseudo labels from the model optimized on the source domain inevitably contain noise due to the domain gap. To tackle this issue, we advance a MetaCorrection framework, where a Domain-aware Meta-learning strategy is devised to benefit Loss Correction (DMLC) for UDA semantic segmentation. In particular, we model the noise distribution of pseudo labels in target domain by introducing a noise transition matrix (NTM) and construct meta data set with domain-invariant source data to guide the estimation of NTM. Through the risk minimization on the meta data set, the optimized NTM thus can correct the noisy issues in pseudo labels and enhance the generalization ability of the model on the target data. Considering the capacity gap between shallow and deep features, we further employ the proposed DMLC strategy to provide matched and compatible supervision signals for different level features, thereby ensuring deep adaptation. Extensive experimental results highlight the effectiveness of our method against existing state-of-the-art methods on three benchmarks.
|
['Yixuan Yuan', 'Baopu Li', 'Chen Yang', 'Xiaoqing Guo']
|
2021-03-09
| null |
http://openaccess.thecvf.com//content/CVPR2021/html/Guo_MetaCorrection_Domain-Aware_Meta_Loss_Correction_for_Unsupervised_Domain_Adaptation_in_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Guo_MetaCorrection_Domain-Aware_Meta_Loss_Correction_for_Unsupervised_Domain_Adaptation_in_CVPR_2021_paper.pdf
|
cvpr-2021-1
|
['synthetic-to-real-translation']
|
['computer-vision']
|
[ 4.08677995e-01 2.21457586e-01 -3.78957421e-01 -6.49713337e-01
-9.24749672e-01 -4.33389217e-01 3.26420814e-01 -1.32814854e-01
-2.74977982e-01 5.86578608e-01 4.24060319e-03 3.74828838e-02
-5.05903810e-02 -8.61056685e-01 -8.02461147e-01 -8.62391174e-01
6.41229689e-01 6.02542520e-01 1.25240535e-01 6.82719722e-02
-1.22935779e-01 -2.05754302e-02 -1.12881339e+00 1.30930051e-01
1.46417165e+00 1.27530313e+00 4.06820208e-01 -6.14295565e-02
-4.04270202e-01 5.28000951e-01 -5.81863642e-01 -5.00229061e-01
4.00341094e-01 -4.85380143e-01 -9.15822446e-01 4.52259570e-01
7.60647357e-02 -1.80277005e-01 -2.87980586e-01 1.33199930e+00
4.40642506e-01 2.65142024e-01 8.77584040e-01 -1.08718061e+00
-6.93318665e-01 5.93174636e-01 -6.12634778e-01 2.10282374e-02
-3.06696802e-01 1.83757901e-01 8.06853652e-01 -8.98762822e-01
6.72529042e-01 1.18718493e+00 6.45842433e-01 8.30603063e-01
-1.26699567e+00 -8.42532039e-01 4.34326947e-01 1.83310315e-01
-1.19358969e+00 -3.54174674e-01 9.76714253e-01 -3.81383151e-01
2.55499125e-01 -2.34851420e-01 1.54901370e-01 1.26460981e+00
-3.15289587e-01 1.08275402e+00 1.11108744e+00 -4.78732616e-01
4.58021343e-01 3.80626798e-01 2.20311448e-01 4.08795625e-01
2.37099491e-02 4.03379202e-02 -3.62821251e-01 3.83076258e-02
5.78363657e-01 -1.79824561e-01 -3.15570040e-03 -6.79473698e-01
-1.05126047e+00 7.57023036e-01 3.26495618e-01 3.16396095e-02
-4.14812118e-01 -5.29322207e-01 6.19330585e-01 1.72141641e-01
4.90781695e-01 2.04079449e-01 -7.39560843e-01 2.90273190e-01
-5.45181990e-01 -4.45583761e-02 3.73706222e-01 1.19579709e+00
9.75779474e-01 1.61365382e-02 -4.04144108e-01 1.29247034e+00
4.28004622e-01 4.53943729e-01 7.42481112e-01 -9.84025478e-01
7.70317674e-01 8.29299331e-01 -2.88531706e-02 -6.50025725e-01
-1.48459882e-01 -8.18066537e-01 -8.18504035e-01 -4.11091447e-02
6.17430687e-01 -2.59493053e-01 -1.22016025e+00 1.90702128e+00
7.39880502e-01 4.26109284e-01 4.03726548e-01 8.70712340e-01
5.90367734e-01 6.18004143e-01 3.05330217e-01 -1.78090379e-01
9.75136280e-01 -1.00729871e+00 -6.71749234e-01 -5.64624667e-01
7.53061950e-01 -4.86295819e-01 1.13879061e+00 2.25360885e-01
-6.55635118e-01 -9.04300153e-01 -1.10138035e+00 1.77392915e-01
-7.93684721e-02 3.25246692e-01 1.00645855e-01 5.37253737e-01
-3.64012122e-01 5.12252450e-01 -6.18566990e-01 -2.01882288e-01
8.83263707e-01 2.47088015e-01 -1.85254797e-01 -3.62214267e-01
-1.42443848e+00 7.16918588e-01 7.99861372e-01 8.25045165e-03
-1.15489912e+00 -7.32042313e-01 -9.05619204e-01 -2.66289413e-01
5.13198614e-01 -5.28476357e-01 1.30206466e+00 -1.15372407e+00
-1.62348926e+00 1.07587719e+00 3.96270268e-02 -5.89914441e-01
4.82290566e-01 -4.99247424e-02 -4.56251025e-01 -1.23770788e-01
4.67700571e-01 6.46197319e-01 9.53029275e-01 -1.48803163e+00
-9.93561387e-01 -5.25869727e-01 -3.20703804e-01 4.88541812e-01
-5.17960191e-01 -5.27810276e-01 -3.60285193e-01 -7.66598046e-01
3.65457982e-01 -6.92481518e-01 -2.36246109e-01 -3.82653743e-01
-4.74737585e-01 -1.56341940e-01 7.46523023e-01 -7.35424578e-01
1.09694231e+00 -2.31429744e+00 1.92873463e-01 3.59007835e-01
7.51577318e-02 5.30130327e-01 -2.28496864e-01 -2.95392334e-01
-2.28986144e-02 -3.46171498e-01 -6.98309481e-01 -3.48269612e-01
-7.62193874e-02 4.38922971e-01 -2.85190493e-01 3.12740654e-01
3.01802725e-01 6.70285225e-01 -9.41491902e-01 -6.54708385e-01
1.22386254e-01 1.13224544e-01 -3.71526480e-01 4.71531123e-01
-4.34751183e-01 8.68540108e-01 -8.81203294e-01 5.27745962e-01
9.75014985e-01 -2.33220175e-01 3.29840444e-02 -1.92562491e-01
3.62573445e-01 1.45713225e-01 -1.18057394e+00 1.94307435e+00
-5.50781548e-01 4.55990508e-02 6.03500605e-02 -1.37475419e+00
1.36439359e+00 1.36235906e-02 5.49046695e-01 -7.42967844e-01
3.74807209e-01 3.47523779e-01 -1.23610295e-01 -3.67993474e-01
2.20618490e-02 -3.23727608e-01 -1.38690144e-01 1.05584726e-01
3.78601283e-01 1.60639420e-01 -3.43578964e-01 -9.27091017e-02
7.57110357e-01 3.12796235e-01 4.33706678e-02 -1.56305239e-01
7.33858168e-01 2.65027314e-01 1.13808751e+00 5.17747939e-01
-5.10767877e-01 5.52410066e-01 4.73262817e-02 -1.90378532e-01
-9.99266505e-01 -1.08726633e+00 -3.22303414e-01 1.00436902e+00
3.58084828e-01 2.94625968e-01 -1.06728256e+00 -1.21592093e+00
-6.08740672e-02 9.08654749e-01 -5.71431160e-01 -6.68285906e-01
-3.60543609e-01 -8.98889363e-01 3.02720845e-01 6.25131965e-01
9.42067564e-01 -9.38374341e-01 2.14468464e-02 4.79995131e-01
-3.49546552e-01 -1.20391190e+00 -6.03685498e-01 3.54357600e-01
-9.88470376e-01 -8.22112739e-01 -7.54471421e-01 -1.02865720e+00
7.63570786e-01 4.63902391e-02 8.27247202e-01 -6.04902744e-01
1.33329034e-01 2.07025602e-01 -3.23751003e-01 -2.52400100e-01
-7.96462655e-01 1.28707618e-01 7.92812407e-02 4.42368597e-01
6.48341954e-01 -4.75553572e-01 -4.37850416e-01 5.19360542e-01
-7.66964674e-01 -1.98602118e-02 6.19153619e-01 1.09670877e+00
9.70777929e-01 2.90441811e-01 9.91869330e-01 -1.24659050e+00
3.29465717e-01 -7.98865139e-01 -5.99382222e-01 2.03953475e-01
-8.14724088e-01 3.91256586e-02 7.93180645e-01 -6.70509696e-01
-1.59552419e+00 3.65096271e-01 -9.24452208e-03 -6.35124981e-01
-3.78991246e-01 3.05026144e-01 -9.03210163e-01 2.05235705e-01
7.84241319e-01 3.39093238e-01 -5.49467802e-02 -5.89200497e-01
4.80862856e-01 9.02539372e-01 7.72281170e-01 -7.79572964e-01
7.48427033e-01 4.31568414e-01 -2.04446003e-01 -2.50149816e-01
-1.29135573e+00 -5.78064680e-01 -8.52349699e-01 -6.10433705e-02
7.03770816e-01 -1.03176105e+00 4.91262153e-02 7.60969460e-01
-8.76117587e-01 -3.42210352e-01 -5.94299912e-01 4.13665146e-01
-6.40651762e-01 2.68872172e-01 -3.12556505e-01 -5.83315432e-01
-1.91327229e-01 -1.12121296e+00 8.87554109e-01 3.86069596e-01
1.31749243e-01 -1.10732675e+00 -1.14553012e-01 5.80072105e-01
-1.63119547e-02 1.75038829e-01 9.98475611e-01 -9.92551744e-01
-3.04546446e-01 -1.13644995e-01 -3.22929859e-01 9.57910419e-01
3.55304241e-01 -6.99734569e-01 -1.19522035e+00 -1.41645193e-01
3.78168613e-01 -4.38617319e-01 8.34767759e-01 4.98654097e-01
1.06267297e+00 -1.30576476e-01 -4.02935326e-01 6.01136148e-01
1.21646667e+00 2.10983604e-01 2.38340884e-01 5.28592944e-01
9.31755245e-01 7.01088965e-01 1.15502918e+00 3.67411226e-01
4.94532406e-01 5.21099150e-01 2.71527499e-01 4.77255136e-02
-2.80337214e-01 -6.04837954e-01 2.27028310e-01 8.90076280e-01
5.93450487e-01 -8.96646380e-02 -9.70393479e-01 8.06469083e-01
-1.76610851e+00 -2.11828366e-01 5.16992100e-02 2.25188541e+00
1.00807691e+00 3.43943506e-01 5.87709099e-02 8.34932998e-02
1.10764694e+00 -1.73330233e-01 -1.20519173e+00 8.74364525e-02
-1.48189425e-01 -1.56006366e-01 5.87575853e-01 2.58173496e-01
-1.27936101e+00 9.47529197e-01 5.21337414e+00 1.35008526e+00
-8.34827006e-01 3.93022954e-01 8.22688937e-01 3.64360243e-01
-2.55731970e-01 -1.28746152e-01 -9.13782299e-01 6.31653607e-01
7.67561436e-01 -1.33974642e-01 1.88043758e-01 1.12555015e+00
5.77731244e-02 2.20866233e-01 -1.03159356e+00 7.57054150e-01
-2.43606210e-01 -8.86214972e-01 8.57776850e-02 -3.91382240e-02
1.05930948e+00 -1.16410986e-01 2.08862469e-01 5.74448764e-01
5.02378345e-01 -5.21318257e-01 5.60638070e-01 1.88297242e-01
8.91467333e-01 -7.87244201e-01 7.13012099e-01 5.60832441e-01
-8.74366581e-01 -2.81502515e-01 -6.25946343e-01 3.27613801e-01
-1.90458551e-01 6.44731641e-01 -9.18848991e-01 6.28853500e-01
5.12425184e-01 9.67209578e-01 -4.02707070e-01 7.33317256e-01
-9.74014997e-02 8.98761332e-01 -1.50536001e-01 4.94145662e-01
2.13846982e-01 -2.63154715e-01 6.31883264e-01 7.43713617e-01
1.36582166e-01 -5.90241216e-02 2.43628085e-01 9.98137355e-01
-3.09100538e-01 2.25177288e-01 -2.91107923e-01 2.47693002e-01
7.60445833e-01 9.19834256e-01 -5.32814562e-01 -2.21743390e-01
-3.70044619e-01 1.11253452e+00 3.90386492e-01 3.95699561e-01
-7.25343347e-01 -8.86209309e-02 6.09114766e-01 -8.08148235e-02
1.74898371e-01 3.44233811e-01 -6.44356549e-01 -1.12509680e+00
1.66361973e-01 -8.38334918e-01 6.80966854e-01 -3.96550745e-01
-1.69794214e+00 3.85084599e-01 -8.12803134e-02 -1.58473098e+00
-7.90513586e-03 -3.25310618e-01 -4.43419397e-01 9.43852067e-01
-1.59203839e+00 -1.15772998e+00 -1.29535794e-01 5.94311237e-01
7.04910994e-01 -4.63851511e-01 6.23272121e-01 3.74535531e-01
-7.67604768e-01 8.57931793e-01 5.77886105e-01 2.12276846e-01
7.78659403e-01 -1.08133757e+00 4.16843951e-01 7.41641700e-01
-2.48771518e-01 2.19120532e-01 3.17580104e-01 -7.02175617e-01
-8.57011557e-01 -1.56342459e+00 2.59411246e-01 -4.31245387e-01
3.67414266e-01 -2.14647189e-01 -1.28157246e+00 6.25619471e-01
-3.54650021e-01 1.11243963e-01 5.37042081e-01 -4.07361649e-02
-3.36072713e-01 -3.24790895e-01 -1.48773682e+00 2.59703815e-01
9.92344320e-01 -3.43693763e-01 -7.18995094e-01 2.09219530e-01
1.00415170e+00 -5.05555868e-01 -8.97660673e-01 6.86680257e-01
1.02066442e-01 -5.98892570e-01 8.51425648e-01 -6.04833424e-01
2.87883818e-01 -2.65837133e-01 -1.65530890e-02 -1.51764143e+00
-3.03460956e-01 -1.23972341e-01 2.15536058e-02 1.82066953e+00
3.14173967e-01 -5.68554699e-01 9.79137301e-01 6.07103944e-01
-2.99876481e-01 -5.02915502e-01 -1.00046420e+00 -9.24788833e-01
3.61109287e-01 -3.82666528e-01 7.41281092e-01 1.23165393e+00
-4.15023565e-01 2.61000991e-01 -7.30177909e-02 4.42462683e-01
8.05884659e-01 2.44065765e-02 5.28950989e-01 -1.19799089e+00
-1.00830629e-01 -1.42226279e-01 -1.82660952e-01 -1.07762778e+00
5.24541497e-01 -1.05884421e+00 2.99507320e-01 -1.27802062e+00
6.48874715e-02 -8.37366462e-01 -6.47924721e-01 4.23845500e-01
-3.97498548e-01 -1.44012511e-01 -1.59781843e-01 4.08859432e-01
-5.97233355e-01 8.48947823e-01 1.45696223e+00 -2.65884310e-01
-3.37382674e-01 1.84744075e-01 -7.38034248e-01 7.59687901e-01
6.57807708e-01 -7.23569989e-01 -6.30996823e-01 -4.71108794e-01
-4.58545476e-01 -1.21066213e-01 8.86606127e-02 -9.81525004e-01
-3.77426818e-02 -1.96279511e-01 3.42412084e-01 -2.08829299e-01
1.55198164e-02 -9.74825859e-01 -2.66625881e-01 1.10816114e-01
-4.92067784e-01 -9.29608107e-01 1.77527413e-01 8.55217457e-01
-3.23922157e-01 -3.38793039e-01 1.13784146e+00 5.95303550e-02
-9.50850368e-01 3.70622873e-01 4.55423705e-02 4.87817049e-01
1.07473528e+00 -2.54783750e-01 -2.32722610e-02 -2.06214990e-02
-8.41035008e-01 6.06700540e-01 3.73348415e-01 4.12744582e-01
3.61886203e-01 -1.52314866e+00 -6.91854000e-01 2.93085486e-01
3.66060644e-01 6.76649690e-01 5.57272494e-01 3.29407245e-01
1.11823618e-01 7.01975599e-02 -2.74932623e-01 -6.87358439e-01
-6.14275336e-01 5.89845657e-01 4.80713308e-01 -2.32023761e-01
-4.37034905e-01 9.89416659e-01 5.17382443e-01 -9.23027813e-01
2.83031791e-01 -6.64635301e-02 -1.82109535e-01 -2.51486413e-02
2.64845461e-01 3.63545150e-01 5.16092554e-02 -6.65372312e-01
-1.66709542e-01 5.25337279e-01 -2.43290573e-01 1.71821833e-01
1.08102989e+00 -6.28480017e-01 3.86468410e-01 4.03162152e-01
1.13333249e+00 -3.54285419e-01 -1.78122568e+00 -8.19212437e-01
1.01597823e-01 -2.68495500e-01 7.38978982e-02 -9.54951763e-01
-1.10169590e+00 8.87369394e-01 8.68966043e-01 -3.07654977e-01
1.40576482e+00 -6.88489963e-05 1.00043225e+00 8.62580538e-02
1.49438336e-01 -1.44665241e+00 1.28448904e-01 2.90100813e-01
3.24958920e-01 -1.50867617e+00 -5.02269328e-01 -5.44234037e-01
-1.01329863e+00 6.34019554e-01 1.04348779e+00 -2.75123268e-02
5.10568976e-01 -1.28178880e-01 3.59668404e-01 2.39738464e-01
-2.35797107e-01 -1.65687501e-01 1.64954662e-01 1.04511762e+00
-2.88589329e-01 4.69541699e-02 4.51830775e-02 1.33259284e+00
2.07340628e-01 5.37225232e-02 1.26421511e-01 6.12023652e-01
-3.47091734e-01 -1.25552654e+00 -4.49831724e-01 4.39427733e-01
-1.55471146e-01 1.23505510e-01 -1.09024651e-01 4.52267557e-01
4.19089973e-01 8.72277796e-01 -2.60593325e-01 -5.15374005e-01
6.36531532e-01 3.48094016e-01 1.55261546e-01 -7.58480549e-01
-1.30984083e-01 6.38733804e-02 -2.36750171e-01 -1.93426192e-01
-2.65308887e-01 -7.29930222e-01 -1.48324728e+00 2.69805849e-01
-4.20956552e-01 9.54752490e-02 4.25349414e-01 1.18606114e+00
3.48807335e-01 5.99460959e-01 9.51601207e-01 -2.80943513e-01
-8.88944268e-01 -1.03952980e+00 -6.96208656e-01 6.43644035e-01
2.48519972e-01 -7.86376774e-01 -1.94595695e-01 1.67568535e-01]
|
[9.766409873962402, 1.5468709468841553]
|
41512081-d66c-4431-8681-7c79a5d5d0ad
|
adaptive-and-implicit-regularization-for
|
2208.05640
| null |
https://arxiv.org/abs/2208.05640v1
|
https://arxiv.org/pdf/2208.05640v1.pdf
|
Adaptive and Implicit Regularization for Matrix Completion
|
The explicit low-rank regularization, e.g., nuclear norm regularization, has been widely used in imaging sciences. However, it has been found that implicit regularization outperforms explicit ones in various image processing tasks. Another issue is that the fixed explicit regularization limits the applicability to broad images since different images favor different features captured by different explicit regularizations. As such, this paper proposes a new adaptive and implicit low-rank regularization that captures the low-rank prior dynamically from the training data. The core of our new adaptive and implicit low-rank regularization is parameterizing the Laplacian matrix in the Dirichlet energy-based regularization, which we call the regularization AIR. Theoretically, we show that the adaptive regularization of \ReTwo{AIR} enhances the implicit regularization and vanishes at the end of training. We validate AIR's effectiveness on various benchmark tasks, indicating that the AIR is particularly favorable for the scenarios when the missing entries are non-uniform. The code can be found at https://github.com/lizhemin15/AIR-Net.
|
['Bao Wang', 'Hongxia Wang', 'Tao Sun', 'Zhemin Li']
|
2022-08-11
| null | null | null | null |
['matrix-completion']
|
['methodology']
|
[ 1.77574277e-01 -9.61226225e-02 -1.99160099e-01 -3.35040867e-01
-5.68051696e-01 -3.06224227e-01 3.37036997e-01 -2.03624085e-01
-5.13717473e-01 4.85897899e-01 3.47810984e-01 3.38104963e-02
-2.65315354e-01 -2.26299360e-01 -5.65532982e-01 -1.04150081e+00
1.46712407e-01 -1.68229931e-03 1.78851753e-01 3.25244702e-02
2.70303160e-01 2.36581713e-01 -1.22299111e+00 2.38223821e-01
9.90783215e-01 6.38885796e-01 4.29354519e-01 7.67602995e-02
5.80539741e-02 4.05698895e-01 1.14121726e-02 -4.81135473e-02
3.10152143e-01 -2.46203914e-01 -5.75519860e-01 6.68263137e-02
1.78366944e-01 4.69436236e-02 -2.25077063e-01 1.29015684e+00
4.47369605e-01 2.79113352e-01 8.04415703e-01 -9.04362500e-01
-6.48195922e-01 2.72378296e-01 -1.01495624e+00 3.48775566e-01
-5.11336289e-02 3.18738483e-02 1.03561759e+00 -1.48622692e+00
5.58115780e-01 1.15894926e+00 6.24277294e-01 3.45810920e-01
-1.36054766e+00 -5.12894332e-01 3.44145596e-01 2.00584784e-01
-1.57564580e+00 -3.83776456e-01 7.95447946e-01 -6.67645514e-01
3.02492172e-01 2.24077314e-01 1.74546137e-01 7.27294981e-01
1.22311413e-01 7.12254822e-01 1.27757335e+00 -3.69261980e-01
-3.18785140e-04 9.66902375e-02 5.11995971e-01 7.24867344e-01
4.30624276e-01 -4.14204150e-02 -3.34084541e-01 -2.72541612e-01
7.57012784e-01 6.22377619e-02 -5.20087361e-01 -3.63294512e-01
-9.38191831e-01 7.85957277e-01 4.32639867e-01 3.94559205e-01
-2.59545267e-01 8.28979611e-02 3.83006722e-01 -3.00206821e-02
5.89868009e-01 8.11150074e-02 -1.88705623e-01 2.58804888e-01
-7.89530218e-01 -2.26039603e-01 3.59038442e-01 5.81015825e-01
1.02613676e+00 -8.51738304e-02 -3.54197770e-01 1.24022102e+00
5.45367658e-01 3.82127017e-01 3.34882230e-01 -9.89373744e-01
9.71219987e-02 3.46652925e-01 -3.04708462e-02 -1.15076852e+00
-3.79447818e-01 -6.85757518e-01 -1.24653625e+00 7.34133348e-02
4.50172842e-01 -6.16580136e-02 -7.66389668e-01 1.83967900e+00
2.12693289e-01 5.84148467e-01 -3.59718800e-01 1.22179437e+00
8.38193238e-01 5.41119635e-01 1.44285783e-01 -4.17107731e-01
1.23874569e+00 -8.52369130e-01 -9.91124153e-01 -2.56501049e-01
5.27445614e-01 -8.49775314e-01 1.45399666e+00 2.46654257e-01
-8.35367918e-01 -1.79938793e-01 -7.84951925e-01 -1.76819235e-01
-2.82048732e-02 5.38799167e-01 4.99152660e-01 3.82172406e-01
-7.70429790e-01 2.11707786e-01 -7.02390730e-01 -2.07054421e-01
1.78365618e-01 2.85351664e-01 -2.71388263e-01 -2.59612948e-01
-1.20141542e+00 6.34090364e-01 -3.37065780e-03 4.50972289e-01
-4.53880161e-01 -6.52971089e-01 -5.23202121e-01 -1.02883123e-01
4.84518349e-01 -4.73484188e-01 7.08852828e-01 -1.10471714e+00
-1.38246572e+00 8.72794032e-01 -3.15985322e-01 1.36320284e-02
4.45664316e-01 -4.20255780e-01 1.09716453e-01 1.25730157e-01
1.92361604e-02 3.07924956e-01 1.13279164e+00 -1.40985477e+00
1.64431885e-01 -2.04098910e-01 -4.43370081e-02 2.41393253e-01
-4.14773196e-01 -1.22035980e-01 -8.59253407e-01 -9.48617935e-01
3.29101682e-01 -1.15946162e+00 -4.84834969e-01 -2.28547920e-02
-2.80978471e-01 1.93244480e-02 3.61621380e-01 -5.24395645e-01
1.28361809e+00 -2.46723151e+00 1.17197596e-01 4.77007568e-01
1.74855158e-01 1.81657910e-01 -1.81832686e-01 1.05266944e-01
-3.77245396e-01 9.17000994e-02 -6.15020633e-01 -1.07434995e-01
-1.71683654e-01 3.28345537e-01 -4.04872075e-02 7.96390951e-01
-4.97495420e-02 5.85381925e-01 -8.57681274e-01 -3.81991029e-01
5.64269116e-03 7.19854176e-01 -6.73369050e-01 -2.46791691e-02
1.99884832e-01 7.68081784e-01 -5.94453871e-01 2.51816303e-01
9.66511965e-01 -6.55257881e-01 3.17746192e-01 -5.88392675e-01
-2.24763870e-01 -3.06662153e-02 -1.32110476e+00 1.52245581e+00
-9.47844312e-02 4.46410596e-01 5.49728632e-01 -1.15841901e+00
4.49118912e-01 1.90054551e-01 7.61259079e-01 -5.58590293e-01
1.69588357e-01 3.09477836e-01 -4.66645397e-02 -4.16408867e-01
2.29819253e-01 -9.75982696e-02 2.77616799e-01 2.93799758e-01
-3.19158405e-01 3.12455654e-01 2.94887066e-01 2.98732638e-01
1.04770672e+00 -1.33148253e-01 3.00604790e-01 -7.29720414e-01
8.64672601e-01 -4.24875408e-01 9.13679183e-01 7.29179323e-01
-6.95981458e-02 7.47341931e-01 3.99159551e-01 -9.21980217e-02
-5.99082947e-01 -1.05336499e+00 -4.15577114e-01 1.14903426e+00
1.94941312e-01 -3.72447759e-01 -6.11005902e-01 -4.88457471e-01
-1.66758120e-01 3.25767875e-01 -5.01200676e-01 -1.24046966e-01
-4.93161708e-01 -1.14876258e+00 1.79109558e-01 4.46253568e-02
4.57065463e-01 -6.28310859e-01 -8.78757238e-02 -7.88123608e-02
-4.50783849e-01 -1.08737075e+00 -7.82646239e-01 1.09983094e-01
-9.88657892e-01 -1.08057559e+00 -8.38389456e-01 -5.75905383e-01
1.21657813e+00 4.50868189e-01 9.11623061e-01 2.72275895e-01
-3.38758856e-01 6.82899237e-01 -4.11803693e-01 8.14818591e-02
7.36203268e-02 -8.31137300e-02 4.17916253e-02 3.07200700e-01
8.17127228e-02 -5.28611660e-01 -7.87732899e-01 5.19082069e-01
-1.07882476e+00 -3.68500082e-03 7.60245800e-01 1.07531750e+00
9.29463029e-01 -1.31801322e-01 4.79310870e-01 -1.26030815e+00
6.01348042e-01 -4.37321186e-01 -5.55414379e-01 1.65183961e-01
-7.59922624e-01 1.78330302e-01 1.72895953e-01 -5.36321044e-01
-1.16404891e+00 1.81091845e-01 -3.58239785e-02 -3.76702219e-01
2.84959644e-01 7.85767794e-01 -1.39001794e-02 -2.30758831e-01
5.67845225e-01 -2.29667109e-02 -1.78414151e-01 -5.98199725e-01
3.95187497e-01 1.78430825e-01 1.62752062e-01 -7.17774153e-01
9.53070521e-01 5.60293317e-01 1.18034579e-01 -1.09914291e+00
-1.20033336e+00 -7.09535003e-01 -5.44593751e-01 -2.75400132e-01
8.47479582e-01 -9.04949009e-01 -3.81260097e-01 4.05243635e-01
-9.18162286e-01 -3.01146030e-01 -2.82927662e-01 7.97621667e-01
-3.44973117e-01 7.45458186e-01 -6.67668164e-01 -5.67409575e-01
-9.09717754e-02 -1.25133681e+00 5.35783589e-01 2.84883915e-03
3.78481187e-02 -9.83919382e-01 1.33920208e-01 2.54682779e-01
2.34015152e-01 -4.37685475e-02 1.01791716e+00 -2.02413037e-01
-5.89826226e-01 1.40747711e-01 -5.05327463e-01 5.08384824e-01
1.82477191e-01 -2.31030896e-01 -7.77417183e-01 -3.59977901e-01
1.84938446e-01 -1.04217134e-01 1.01427400e+00 7.45942593e-01
1.32148361e+00 -1.80854008e-01 -1.08546121e-02 6.56364024e-01
1.41068113e+00 -4.93362546e-01 7.10429549e-01 5.71430065e-02
6.32412136e-01 4.21888918e-01 6.05598927e-01 4.60440546e-01
1.29902009e-02 6.70401871e-01 3.21394563e-01 -3.64801705e-01
-9.23492983e-02 3.82981330e-01 5.45977712e-01 1.33119535e+00
-3.08986545e-01 7.72505999e-02 -7.94029355e-01 3.25301021e-01
-2.07149816e+00 -7.77328253e-01 -5.07778227e-01 2.32739711e+00
1.01327538e+00 -1.30898699e-01 -4.02380735e-01 -1.65645242e-01
8.36826086e-01 2.12152213e-01 -5.76148331e-01 -8.80070105e-02
-1.75019696e-01 1.01763844e-01 6.88148737e-01 6.98936939e-01
-1.10100722e+00 7.71437287e-01 5.93002415e+00 9.59507406e-01
-8.99576187e-01 4.46122199e-01 5.87440789e-01 -2.85857376e-02
-3.16474408e-01 3.17360857e-03 -4.95375305e-01 3.64889205e-01
5.04125297e-01 -1.26935571e-01 4.37022865e-01 4.37627494e-01
6.80314958e-01 -2.64799684e-01 -8.28163147e-01 1.11631584e+00
2.43990570e-02 -1.08430612e+00 5.31551950e-02 7.47381374e-02
8.35956991e-01 1.53589949e-01 2.99796164e-01 1.04103692e-01
1.80098042e-02 -8.07663441e-01 2.93286443e-01 7.35970259e-01
8.34773123e-01 -3.92991692e-01 6.00712597e-01 2.60260850e-01
-1.07639849e+00 1.47690162e-01 -4.72949058e-01 2.27450758e-01
9.80199054e-02 1.30341136e+00 -2.59797126e-01 4.01021749e-01
5.60374260e-01 8.71137798e-01 -4.41061348e-01 1.13174260e+00
-4.30538654e-01 7.08229899e-01 -2.95374185e-01 6.53604031e-01
1.14465855e-01 -7.80702591e-01 7.26730227e-01 1.37492311e+00
2.41306335e-01 4.69931573e-01 4.10708457e-01 6.92687392e-01
-1.55130222e-01 4.36054915e-01 -3.76658648e-01 1.25560522e-01
-1.02271140e-01 1.38893139e+00 -7.51895607e-01 4.46103476e-02
-5.94686568e-01 7.63824344e-01 2.28026554e-01 7.76006103e-01
-8.04352999e-01 1.54100377e-02 4.85567123e-01 2.57207662e-01
4.56698462e-02 -4.76522177e-01 -3.39148134e-01 -1.40966964e+00
-5.92797883e-02 -6.09640837e-01 4.05391663e-01 -5.78163683e-01
-1.50413871e+00 2.91588217e-01 7.99378529e-02 -1.19291878e+00
2.96289951e-01 -6.03341579e-01 -3.55729014e-01 7.86852241e-01
-1.63280761e+00 -6.79974437e-01 -4.48580176e-01 6.41043842e-01
4.22512650e-01 1.31974638e-01 4.26092565e-01 8.30240130e-01
-7.18363702e-01 5.72702527e-01 2.96674997e-01 5.16410135e-02
9.93073940e-01 -8.12397778e-01 -4.66297120e-01 7.72245109e-01
-1.71689019e-01 8.65562558e-01 7.28264928e-01 -7.02818871e-01
-1.25105846e+00 -9.63052273e-01 3.12367946e-01 4.97270562e-02
7.34432697e-01 -1.91268280e-01 -1.19142902e+00 4.40104663e-01
1.67690098e-01 3.74394357e-01 7.79117346e-01 4.99304160e-02
-3.12509984e-01 -1.50457267e-02 -9.13311422e-01 6.40594363e-01
9.35122073e-01 -5.08094490e-01 -2.34633222e-01 5.57233036e-01
2.75576264e-01 -1.04461715e-01 -6.11882150e-01 5.97849369e-01
3.63569736e-01 -7.25442827e-01 9.86055255e-01 -2.56916136e-01
2.78273761e-01 -4.94434595e-01 -1.88567474e-01 -9.72165823e-01
-7.01987505e-01 -3.70184213e-01 -8.69661197e-02 9.53589976e-01
3.66753399e-01 -8.36952448e-01 5.81088006e-01 5.67567289e-01
-6.23800568e-02 -7.96501100e-01 -8.59710634e-01 -7.72287548e-01
4.40917276e-02 -2.50009149e-01 -2.64539331e-01 1.08848345e+00
-4.54551995e-01 2.56407917e-01 -5.23229003e-01 3.04043114e-01
8.16818595e-01 -1.18770115e-01 3.64776075e-01 -1.22220111e+00
-1.18760191e-01 -2.76437730e-01 9.31262225e-03 -1.26909542e+00
3.81658256e-01 -1.16905475e+00 1.91122845e-01 -1.29861569e+00
5.28514206e-01 -6.02181852e-01 -6.18487537e-01 4.56892312e-01
-2.54320264e-01 3.71861547e-01 4.81199734e-02 5.85692286e-01
-5.23225009e-01 7.58528292e-01 1.30578566e+00 -9.23430175e-02
-1.72640905e-01 -1.34803921e-01 -6.26861691e-01 1.16724277e+00
8.84533465e-01 -6.97146893e-01 -5.00501573e-01 -4.25565988e-01
3.84723663e-01 -4.09863561e-01 1.78336099e-01 -6.96580648e-01
2.08992258e-01 -3.09246749e-01 6.80953488e-02 -3.46676528e-01
3.05707902e-01 -8.65662575e-01 9.27585140e-02 2.89552391e-01
-3.54513377e-01 -2.28311434e-01 -1.64757203e-02 6.99013233e-01
-2.04738364e-01 -5.72525680e-01 1.06190979e+00 -1.80091888e-01
-5.63826799e-01 4.49045479e-01 -4.31636125e-01 3.94214302e-01
6.38372540e-01 -8.51534680e-03 -1.47324592e-01 -2.67251581e-01
-9.88890827e-01 9.76119563e-02 4.11635071e-01 -1.60197634e-02
6.80991650e-01 -1.31526709e+00 -6.84628904e-01 -3.21729369e-02
-5.06815091e-02 -2.58558035e-01 4.26244080e-01 1.53650689e+00
-1.98097900e-01 7.49340430e-02 1.50235757e-01 -7.36974955e-01
-1.37716413e+00 2.52342314e-01 2.91731566e-01 -4.73128915e-01
-7.74960756e-01 6.14337742e-01 8.79004240e-01 -3.32168370e-01
1.04672760e-01 -1.29058912e-01 -3.02486837e-01 4.27175462e-02
2.82236665e-01 3.31569463e-01 -1.46871328e-01 -7.81447291e-01
-4.88335818e-01 8.96665633e-01 -7.36472532e-02 -6.27736151e-02
1.45531845e+00 -3.54232639e-01 -5.49073458e-01 5.81775367e-01
1.15198398e+00 3.21738690e-01 -1.15168178e+00 -4.64536160e-01
1.41917884e-01 -5.54137468e-01 2.62378067e-01 -4.07316178e-01
-1.32198644e+00 6.96208835e-01 7.30622053e-01 -1.91945970e-01
1.26577222e+00 -2.10577130e-01 4.05663967e-01 4.38714892e-01
1.74319550e-01 -1.23616219e+00 2.48026237e-01 8.16523135e-01
1.08280087e+00 -1.14783156e+00 5.31379580e-01 -8.85539591e-01
-5.97077429e-01 7.60164320e-01 3.72170568e-01 -2.46891275e-01
9.77601230e-01 4.68950905e-02 -1.00248272e-03 -2.80222893e-01
-5.03935397e-01 -1.81319311e-01 5.60193241e-01 7.24922270e-02
6.74942613e-01 -8.65703449e-02 -7.68226862e-01 4.80755955e-01
4.79837626e-01 9.03288182e-03 5.28128386e-01 6.66899800e-01
-2.87107229e-01 -1.21148849e+00 -3.52369219e-01 4.95325774e-01
-5.81967592e-01 -1.84266940e-01 -1.24377899e-01 4.20320183e-01
1.55744180e-01 7.78978407e-01 -3.66314471e-01 1.45869195e-01
6.22808337e-02 -5.32166250e-02 3.40743721e-01 -7.24320829e-01
-4.18757141e-01 3.47975910e-01 -1.74265102e-01 -6.14082992e-01
-7.11059630e-01 -6.89411342e-01 -1.47652423e+00 5.09412475e-02
-2.81680256e-01 3.68628144e-01 4.39343929e-01 7.56917357e-01
3.26481760e-01 4.55358982e-01 5.07949829e-01 -6.83044016e-01
-3.95446569e-01 -6.89805210e-01 -6.83054686e-01 5.33635736e-01
2.58144408e-01 -9.96832013e-01 -7.76842475e-01 1.21378474e-01]
|
[7.625667572021484, 4.2894697189331055]
|
52f20bec-ce7c-4500-a759-c460d63292ba
|
dilbert-customized-pre-training-for-domain
|
2109.00571
| null |
https://arxiv.org/abs/2109.00571v1
|
https://arxiv.org/pdf/2109.00571v1.pdf
|
DILBERT: Customized Pre-Training for Domain Adaptation withCategory Shift, with an Application to Aspect Extraction
|
The rise of pre-trained language models has yielded substantial progress in the vast majority of Natural Language Processing (NLP) tasks. However, a generic approach towards the pre-training procedure can naturally be sub-optimal in some cases. Particularly, fine-tuning a pre-trained language model on a source domain and then applying it to a different target domain, results in a sharp performance decline of the eventual classifier for many source-target domain pairs. Moreover, in some NLP tasks, the output categories substantially differ between domains, making adaptation even more challenging. This, for example, happens in the task of aspect extraction, where the aspects of interest of reviews of, e.g., restaurants or electronic devices may be very different. This paper presents a new fine-tuning scheme for BERT, which aims to address the above challenges. We name this scheme DILBERT: Domain Invariant Learning with BERT, and customize it for aspect extraction in the unsupervised domain adaptation setting. DILBERT harnesses the categorical information of both the source and the target domains to guide the pre-training process towards a more domain and category invariant representation, thus closing the gap between the domains. We show that DILBERT yields substantial improvements over state-of-the-art baselines while using a fraction of the unlabeled data, particularly in more challenging domain adaptation setups.
|
['Roi Reichart', 'Yftah Ziser', 'Entony Lekhtman']
|
2021-09-01
| null | null | null | null |
['aspect-extraction']
|
['natural-language-processing']
|
[ 2.07575321e-01 -1.89114660e-02 -3.39041740e-01 -6.31058991e-01
-1.03608477e+00 -1.18930507e+00 9.22245681e-01 4.84617025e-01
-3.57212245e-01 7.12343574e-01 3.09645593e-01 -2.57834524e-01
1.60721183e-01 -6.78365707e-01 -4.08596277e-01 -6.44948125e-01
4.44157094e-01 9.00952220e-01 2.34610230e-01 -4.58793789e-01
5.99550530e-02 3.81472319e-01 -1.21192992e+00 3.48310798e-01
6.20053887e-01 8.97284627e-01 9.48657049e-04 3.67546320e-01
-5.73172808e-01 2.19776943e-01 -6.38248980e-01 -5.65143704e-01
2.62377113e-01 -3.12065482e-01 -8.87256861e-01 2.41534978e-01
2.41857454e-01 1.90584362e-01 5.02716303e-02 9.00361776e-01
3.60967219e-01 2.33042762e-01 1.20564437e+00 -1.04484642e+00
-5.78561723e-01 5.44280648e-01 -5.19906223e-01 1.17973283e-01
1.35689899e-01 -1.22784935e-01 1.14011478e+00 -8.34351659e-01
5.77129722e-01 1.23276258e+00 5.62138021e-01 4.18409467e-01
-1.48079681e+00 -4.27909464e-01 4.98585790e-01 -9.47335958e-02
-1.04939485e+00 -3.61310512e-01 6.91381633e-01 -5.63006282e-01
9.00153637e-01 -2.22201005e-01 6.92886636e-02 1.20369172e+00
-1.23746015e-01 7.45907307e-01 1.16883862e+00 -5.80788434e-01
3.89089406e-01 6.17891312e-01 3.78926754e-01 1.56367701e-02
2.98107386e-01 -1.97279364e-01 -2.48800650e-01 -2.31139600e-01
4.57307041e-01 -1.58616260e-01 7.68358260e-02 -8.11945200e-01
-1.03404677e+00 9.39604282e-01 1.03786334e-01 5.53041995e-01
-2.38991484e-01 -5.95172346e-01 6.74317956e-01 5.84995031e-01
7.52711296e-01 6.98286593e-01 -1.05133438e+00 -8.64676833e-02
-8.10893297e-01 2.41818264e-01 1.05486560e+00 1.04703748e+00
8.03627074e-01 -2.65593082e-01 -1.57170206e-01 1.18654692e+00
1.37350271e-02 3.00424874e-01 5.38854718e-01 -4.30113614e-01
7.18741357e-01 7.30210841e-01 1.18433833e-01 -5.83044469e-01
-2.50312060e-01 -3.39377046e-01 -7.97444224e-01 -8.40398425e-04
7.50717402e-01 -2.46290237e-01 -1.14755821e+00 1.73554635e+00
4.47066545e-01 -3.30564797e-01 4.02180612e-01 4.54030901e-01
5.85157931e-01 5.67814231e-01 3.35333407e-01 -2.47226488e-02
1.55828404e+00 -9.09330308e-01 -4.83505756e-01 -7.08961844e-01
4.69380707e-01 -8.12771797e-01 1.20992649e+00 3.42699587e-01
-6.63424551e-01 -4.79349256e-01 -9.61474717e-01 -6.30045384e-02
-8.78848374e-01 1.89454347e-01 4.76699054e-01 4.99010235e-01
-5.07642031e-01 3.84894669e-01 -5.70559800e-01 -7.26585507e-01
2.73358852e-01 3.15858275e-01 -4.58022833e-01 -3.43669683e-01
-1.14733374e+00 8.17461014e-01 4.68400836e-01 -5.67193568e-01
-5.96174717e-01 -7.70581543e-01 -9.37851667e-01 8.50077998e-03
4.69450891e-01 -5.08143246e-01 1.45754743e+00 -1.20206881e+00
-1.35537446e+00 1.23952794e+00 -1.66661337e-01 -3.71665478e-01
3.07482839e-01 -2.80872375e-01 -5.81304610e-01 -2.03372747e-01
3.56542677e-01 4.80507106e-01 1.11355329e+00 -1.12951648e+00
-7.93603063e-01 -3.38919431e-01 1.74038038e-01 2.74685532e-01
-4.04868662e-01 1.48844719e-01 -3.85423005e-01 -7.71123648e-01
-2.89380014e-01 -7.96296239e-01 -2.47762322e-01 -2.14446038e-01
-1.22415170e-01 -5.03589332e-01 7.25881934e-01 -3.66692394e-01
1.02540243e+00 -2.42465091e+00 7.24548474e-02 2.79666893e-02
-3.15662883e-02 5.28491020e-01 -3.30552280e-01 4.47630554e-01
-1.75584033e-01 -6.08750358e-02 -4.98325050e-01 -3.83112341e-01
3.72432321e-02 4.54404771e-01 -4.56662267e-01 2.01931462e-01
5.17594755e-01 5.65529048e-01 -9.76729751e-01 -2.89016753e-01
1.94528457e-02 2.35796496e-01 -3.85359645e-01 2.96739906e-01
-3.00649464e-01 5.24562538e-01 -3.54364306e-01 3.93950522e-01
6.19717121e-01 -1.53942093e-01 1.76940799e-01 5.01578562e-02
2.00207997e-02 8.43850553e-01 -1.18403423e+00 1.69824648e+00
-7.88420916e-01 6.42496943e-01 1.03739291e-01 -1.38418591e+00
1.05425572e+00 3.17313731e-01 4.67508435e-01 -4.16504323e-01
1.82304450e-03 3.08895916e-01 -1.02432445e-01 -1.67666763e-01
3.71415257e-01 -5.71777344e-01 -4.46658522e-01 4.62092370e-01
5.28826118e-01 -4.57059592e-01 4.17886406e-01 3.56025994e-02
9.95329559e-01 8.16453770e-02 9.11583424e-01 -2.39876851e-01
6.05853319e-01 3.25006217e-01 4.67936128e-01 6.01998091e-01
-3.82741749e-01 8.36672604e-01 6.19353235e-01 -1.93749234e-01
-1.00797844e+00 -1.26075566e+00 -3.08441907e-01 1.44030094e+00
-9.78048444e-02 -2.37069488e-01 -4.90219057e-01 -1.29480457e+00
1.35083079e-01 7.24095583e-01 -4.99998719e-01 -3.32569003e-01
-4.30861413e-01 -7.24762857e-01 2.79670179e-01 4.46059555e-01
4.61782604e-01 -8.52098286e-01 -1.34241360e-03 3.70175898e-01
-1.17696479e-01 -1.33358991e+00 -3.36233467e-01 6.79811180e-01
-7.47002542e-01 -8.55521202e-01 -7.51597703e-01 -1.06315041e+00
5.50632477e-01 1.64209664e-01 1.62013829e+00 -6.29730523e-01
2.59988874e-01 3.69107246e-01 -5.63894391e-01 -5.72731078e-01
-6.16107285e-01 5.12364805e-01 7.32618794e-02 6.16374724e-02
1.10181236e+00 -5.84522665e-01 -2.38120556e-01 2.55825669e-01
-9.39629078e-01 -4.86364573e-01 6.64817214e-01 7.80649483e-01
5.35766602e-01 2.31126383e-01 6.73976421e-01 -1.24783647e+00
6.56765223e-01 -7.68063962e-01 -2.83575416e-01 1.27701268e-01
-4.66028214e-01 2.30411664e-01 9.15994167e-01 -7.65492320e-01
-1.11490619e+00 1.19791873e-01 -1.34980813e-01 9.47224647e-02
-6.32516623e-01 4.36383158e-01 -4.87615913e-01 3.15412492e-01
9.17554677e-01 -2.27225795e-02 -2.89397895e-01 -7.71446168e-01
4.18256938e-01 9.20767426e-01 5.47497213e-01 -5.98993599e-01
9.50074792e-01 2.82848179e-01 -3.74232709e-01 -8.26282740e-01
-1.12065196e+00 -1.01979327e+00 -1.05134630e+00 5.13377666e-01
4.72061396e-01 -1.03850889e+00 3.87097865e-01 2.84201354e-01
-1.07317972e+00 -2.56013632e-01 -6.06668472e-01 2.54182041e-01
-4.24983889e-01 1.55783564e-01 -2.07388967e-01 -3.77884179e-01
-3.39383073e-02 -9.40756977e-01 1.10872531e+00 5.52880727e-02
-5.45875907e-01 -1.26537192e+00 3.55980963e-01 -3.80943231e-02
3.55416328e-01 2.24729460e-02 1.23699617e+00 -1.47139633e+00
5.60908690e-02 -3.45619112e-01 -2.14432344e-01 7.30949104e-01
5.39857328e-01 -3.49037081e-01 -1.01841319e+00 -9.67960209e-02
9.93239135e-02 -2.39053741e-01 8.11236739e-01 1.20373040e-01
5.33071041e-01 -1.46313846e-01 -3.58628929e-01 3.11545640e-01
1.30740261e+00 3.56720723e-02 2.98841864e-01 5.40255725e-01
5.12804866e-01 8.04816723e-01 7.93440342e-01 2.57807970e-01
4.35313672e-01 7.50946701e-01 -2.10476950e-01 -1.87534332e-01
-2.05404416e-01 -2.38255590e-01 5.28451979e-01 6.02718711e-01
3.60182047e-01 -1.66972309e-01 -9.74028289e-01 1.04471111e+00
-1.54908037e+00 -5.37806392e-01 2.81937540e-01 2.22477341e+00
1.24096584e+00 2.41754726e-01 3.89058203e-01 7.62750283e-02
5.50690711e-01 9.70450640e-02 -6.41687512e-01 -6.63253188e-01
-1.10595115e-01 3.45708221e-01 3.04252803e-01 2.82230735e-01
-1.51499724e+00 1.02714872e+00 5.74021578e+00 8.42473507e-01
-1.11458588e+00 1.76042184e-01 4.57712442e-01 2.39455223e-01
-1.50751114e-01 -3.57698351e-02 -1.02206433e+00 2.88949966e-01
1.00735736e+00 -3.77171278e-01 1.66768178e-01 1.09409142e+00
-8.85588229e-02 9.41808671e-02 -1.39865506e+00 9.04946864e-01
5.61424978e-02 -7.09015369e-01 8.08194503e-02 -1.17117912e-01
7.95275450e-01 2.46099651e-01 -1.86026990e-02 7.62964725e-01
6.62747741e-01 -7.62186706e-01 2.56969690e-01 -1.86197996e-01
7.17381716e-01 -6.67255640e-01 7.03357279e-01 4.02948499e-01
-1.02032447e+00 -2.31944937e-02 -5.08591175e-01 6.83593825e-02
1.93567510e-04 8.54243159e-01 -9.45339322e-01 3.42238426e-01
5.34749210e-01 8.47839534e-01 -5.28484643e-01 8.45605314e-01
-3.57622087e-01 7.03079045e-01 -3.52673590e-01 1.99750662e-01
3.26458991e-01 -1.06299438e-01 6.93169117e-01 1.44824064e+00
3.12848808e-03 -1.47144258e-01 1.83746979e-01 5.89098454e-01
-2.12640733e-01 2.19956756e-01 -9.38270867e-01 -1.66587412e-01
2.62251586e-01 1.31543088e+00 -5.80039442e-01 -5.27569056e-01
-6.69188142e-01 9.43967581e-01 4.00284350e-01 4.63006556e-01
-4.73112166e-01 -4.90779936e-01 1.02024710e+00 2.01983958e-01
4.88861412e-01 -2.57582366e-01 -3.95692796e-01 -1.42888677e+00
1.22363053e-01 -1.05421805e+00 4.77401108e-01 -1.42081007e-01
-1.67499030e+00 6.80861056e-01 2.13185742e-01 -1.33674240e+00
-5.47543347e-01 -8.37486982e-01 -4.15507615e-01 8.00929725e-01
-1.79510427e+00 -1.07087088e+00 1.13069892e-01 5.60564101e-01
9.01227534e-01 -3.18844795e-01 9.01869595e-01 3.05897593e-01
-1.69066936e-01 6.67689264e-01 4.37641859e-01 1.85864508e-01
1.28342950e+00 -1.54294145e+00 6.96656704e-01 8.19391131e-01
3.93240482e-01 6.24715924e-01 6.97921276e-01 -3.11482370e-01
-8.24320436e-01 -1.31765985e+00 1.17691076e+00 -8.36052775e-01
7.89141893e-01 -6.93620741e-01 -1.04500461e+00 7.11246908e-01
1.14463240e-01 -1.63633630e-01 8.32602680e-01 4.97058451e-01
-7.62476861e-01 -6.21588863e-02 -1.18950272e+00 5.60872614e-01
7.78816044e-01 -6.97458923e-01 -1.00312626e+00 4.30156231e-01
5.60821533e-01 -1.77676275e-01 -7.13591218e-01 4.06803906e-01
2.27218270e-01 -6.16894841e-01 1.03215909e+00 -7.44370878e-01
3.93940568e-01 -9.77167934e-02 -1.72902897e-01 -1.75677991e+00
-3.29357058e-01 -5.16613841e-01 -4.85750847e-03 1.85526443e+00
5.65447271e-01 -7.19584882e-01 5.37090242e-01 5.58213115e-01
9.11746472e-02 -3.87383819e-01 -7.21441150e-01 -8.68444204e-01
5.83188474e-01 -3.37852776e-01 5.25758147e-01 9.13729608e-01
-1.38226477e-02 8.69565725e-01 4.75885719e-02 6.31310744e-03
2.32178569e-01 1.39752895e-01 7.91141808e-01 -1.44900763e+00
-3.70276749e-01 -4.50906157e-01 -3.20631832e-01 -1.25386918e+00
2.41134480e-01 -9.32835281e-01 1.71998128e-01 -1.35186291e+00
1.12208966e-02 -5.29809058e-01 -3.39674622e-01 5.89048862e-01
-2.52389342e-01 7.92160854e-02 4.52094860e-02 1.43950716e-01
-4.99822587e-01 3.38233382e-01 8.70875895e-01 -3.94188225e-01
-4.24241126e-01 3.93693715e-01 -1.09613335e+00 7.97060609e-01
9.31611300e-01 -5.97155929e-01 -3.59070063e-01 -4.07801241e-01
3.31060477e-02 -6.47096753e-01 -2.66876779e-02 -8.64884138e-01
-9.67617631e-02 -1.31153196e-01 1.72769323e-01 -2.59511471e-01
1.54674932e-01 -1.02386737e+00 -5.83052337e-01 -1.12406172e-01
-3.73386204e-01 -1.36025503e-01 3.96422714e-01 5.49260616e-01
-5.22530079e-01 -2.70735860e-01 9.97062504e-01 -1.46237522e-01
-7.95924008e-01 8.39708522e-02 -4.08825517e-01 5.24305642e-01
7.49076962e-01 -6.74798191e-02 -6.16426282e-02 -1.83622196e-01
-5.92373371e-01 2.99298745e-02 4.55761194e-01 6.37174904e-01
-1.04545422e-01 -1.10249710e+00 -6.28182232e-01 3.33635658e-01
4.92979228e-01 2.67367095e-01 -1.38535783e-01 5.46142161e-01
1.37385637e-01 4.56342578e-01 -4.10169996e-02 -5.01066029e-01
-1.05063272e+00 6.42587543e-01 9.30102989e-02 -8.34683836e-01
-3.43108505e-01 6.50643170e-01 6.03236914e-01 -8.52821648e-01
5.98849840e-02 -4.51883912e-01 -4.61353958e-01 4.63783950e-01
5.04032254e-01 3.00332550e-02 3.69953960e-01 -6.13538444e-01
-3.81137043e-01 5.99777222e-01 -4.93221074e-01 -2.06785411e-01
1.32482266e+00 -2.82870173e-01 1.65448427e-01 5.48183918e-01
1.29242516e+00 1.08045883e-01 -1.13974631e+00 -7.13630319e-01
2.69656688e-01 -5.93290776e-02 -2.98838377e-01 -1.08896470e+00
-6.92394674e-01 1.00180852e+00 2.19107866e-01 2.77705312e-01
1.22956371e+00 2.43480861e-01 6.73617721e-01 3.80644351e-01
3.13974410e-01 -1.16232157e+00 -1.76147372e-01 9.67732728e-01
6.76963925e-01 -1.49208319e+00 -1.21223308e-01 -3.23740423e-01
-8.66905034e-01 9.60630536e-01 3.57680827e-01 -2.40276963e-01
7.51636446e-01 1.59469366e-01 3.24748009e-01 3.41777839e-02
-5.93830287e-01 -3.69238615e-01 3.86248112e-01 1.07239819e+00
5.39189339e-01 7.57008120e-02 -1.10146910e-01 8.15847814e-01
-1.36157528e-01 -2.01185271e-01 2.59497434e-01 7.78566360e-01
-2.67768115e-01 -1.59231353e+00 -2.78168321e-01 4.20119017e-01
-6.50048256e-01 -2.36617476e-01 -6.61989331e-01 8.35464060e-01
2.83793770e-02 9.66676950e-01 -7.89528191e-02 6.32928386e-02
6.39496744e-01 3.73412490e-01 1.05852485e-01 -1.12170970e+00
-6.37903452e-01 4.33021821e-02 1.99125767e-01 -1.59915835e-01
-3.70178163e-01 -8.32004964e-01 -9.77032781e-01 1.75046846e-01
4.84994315e-02 2.49593064e-01 5.68108737e-01 1.16543853e+00
4.69347268e-01 2.57093281e-01 5.56988597e-01 -7.52158165e-01
-7.33759046e-01 -9.91692424e-01 -5.51095486e-01 7.13943958e-01
5.27936578e-01 -8.81786585e-01 -3.77884269e-01 1.93869412e-01]
|
[10.76356029510498, 7.9528303146362305]
|
298b82be-b189-4cd8-84fa-def2c7786563
|
remote-sensing-change-detection-with
|
2304.06710
| null |
https://arxiv.org/abs/2304.06710v1
|
https://arxiv.org/pdf/2304.06710v1.pdf
|
Remote Sensing Change Detection With Transformers Trained from Scratch
|
Current transformer-based change detection (CD) approaches either employ a pre-trained model trained on large-scale image classification ImageNet dataset or rely on first pre-training on another CD dataset and then fine-tuning on the target benchmark. This current strategy is driven by the fact that transformers typically require a large amount of training data to learn inductive biases, which is insufficient in standard CD datasets due to their small size. We develop an end-to-end CD approach with transformers that is trained from scratch and yet achieves state-of-the-art performance on four public benchmarks. Instead of using conventional self-attention that struggles to capture inductive biases when trained from scratch, our architecture utilizes a shuffled sparse-attention operation that focuses on selected sparse informative regions to capture the inherent characteristics of the CD data. Moreover, we introduce a change-enhanced feature fusion (CEFF) module to fuse the features from input image pairs by performing a per-channel re-weighting. Our CEFF module aids in enhancing the relevant semantic changes while suppressing the noisy ones. Extensive experiments on four CD datasets reveal the merits of the proposed contributions, achieving gains as high as 14.27\% in intersection-over-union (IoU) score, compared to the best-published results in the literature. Code is available at \url{https://github.com/mustansarfiaz/ScratchFormer}.
|
['Fahad Shahbaz Khan', 'Salman Khan', 'Rao Muhammad Anwer', 'Sanath Narayan', 'Hisham Cholakkal', 'Mustansar Fiaz', 'Mubashir Noman']
|
2023-04-13
| null | null | null | null |
['change-detection']
|
['computer-vision']
|
[ 5.37369251e-01 -2.30527505e-01 2.35464554e-02 -4.55561429e-01
-1.03020692e+00 -2.03127712e-01 7.99238205e-01 1.44972011e-01
-6.22070670e-01 3.83581191e-01 3.55608106e-01 1.35584921e-01
1.20678239e-01 -6.01531267e-01 -8.97953033e-01 -7.96670914e-01
1.36987373e-01 1.49809137e-01 3.74445498e-01 -3.62572342e-01
1.34797469e-01 1.23077564e-01 -1.67278695e+00 3.70406181e-01
9.54569280e-01 1.35677981e+00 3.99033278e-01 3.94121587e-01
-5.53523824e-02 6.14484489e-01 -2.59372205e-01 -3.22623134e-01
4.33629096e-01 -4.45133269e-01 -5.92432976e-01 -1.16491526e-01
6.04019165e-01 -1.01229019e-01 -4.03501362e-01 1.16566324e+00
6.73458278e-01 6.89458549e-02 2.15721756e-01 -8.88821721e-01
-7.39913225e-01 3.57671797e-01 -6.08531654e-01 6.23626709e-01
6.29238188e-02 3.57574672e-01 1.19866693e+00 -1.09035182e+00
6.11592054e-01 8.55398834e-01 7.29111314e-01 3.51828635e-01
-1.51548898e+00 -8.05096090e-01 3.91506314e-01 4.78198051e-01
-1.34430122e+00 -6.62421107e-01 9.63333368e-01 -4.07836765e-01
1.03148746e+00 1.91587284e-01 5.71182370e-01 1.05539703e+00
1.66036021e-02 8.36000502e-01 1.06527770e+00 -3.63093197e-01
1.23454519e-01 -5.84648326e-02 7.51825422e-02 4.87228721e-01
4.03001830e-02 1.29233883e-03 -5.54138184e-01 2.12764621e-01
5.15109181e-01 2.74861455e-01 -4.12161678e-01 -5.53491771e-01
-1.23688948e+00 7.73547590e-01 8.97485495e-01 4.87333775e-01
-6.64970160e-01 1.25470340e-01 5.34433067e-01 2.42203534e-01
5.95169485e-01 4.40588027e-01 -5.15825629e-01 -6.18726090e-02
-1.02526045e+00 1.96515873e-01 1.23549700e-01 5.68881035e-01
9.37556148e-01 -1.08507663e-01 -4.06909198e-01 1.00703514e+00
-4.54417616e-02 3.59753817e-01 7.41384327e-01 -5.19993782e-01
3.83222401e-01 6.92296803e-01 -2.24575758e-01 -8.86876881e-01
-2.02157035e-01 -9.76976156e-01 -8.94973993e-01 5.67895584e-02
4.51195240e-02 2.01903284e-01 -1.13058448e+00 1.93663514e+00
2.40777791e-01 3.32074434e-01 -2.53317595e-01 8.91659737e-01
6.85689330e-01 4.74332273e-01 5.36756665e-02 1.35136098e-01
1.15534925e+00 -1.12699342e+00 -4.67447281e-01 -3.78761142e-01
3.89899403e-01 -5.10680914e-01 1.44675553e+00 2.35400617e-01
-8.71746540e-01 -6.16414070e-01 -1.10183609e+00 -1.74535915e-01
-3.89246792e-01 1.87131599e-01 3.18865329e-01 1.60691977e-01
-1.03126276e+00 3.86832505e-01 -8.64395797e-01 -3.83683294e-01
8.47757578e-01 3.45795989e-01 -3.60006481e-01 -3.62088889e-01
-1.10076189e+00 6.64606929e-01 2.30021611e-01 1.26243398e-01
-9.63374376e-01 -9.55316126e-01 -7.90574610e-01 9.66354162e-02
4.40862477e-01 -6.22794807e-01 1.06686068e+00 -1.46589422e+00
-1.33292353e+00 7.68866956e-01 -1.77181542e-01 -6.40855134e-01
3.69979173e-01 -4.99605864e-01 -3.23456049e-01 1.86344996e-01
2.27925405e-01 8.04687321e-01 9.17355001e-01 -1.22132576e+00
-6.49608374e-01 -3.60305607e-01 4.05078679e-02 8.25709403e-02
-4.38823551e-01 -2.10204899e-01 -8.43786716e-01 -9.69733119e-01
9.32729896e-03 -9.00752366e-01 -3.11522335e-01 -1.14709325e-01
-1.95496753e-01 7.00287521e-02 8.11452031e-01 -6.52246296e-01
1.38086462e+00 -2.30194759e+00 1.24867119e-01 1.48236066e-01
1.35864288e-01 5.62427521e-01 -4.71071869e-01 2.50337332e-01
-2.54622817e-01 -1.28077254e-01 -5.04129112e-01 -3.69567007e-01
-3.54356505e-02 -1.01242118e-01 -7.20323026e-02 4.51154828e-01
4.68055308e-01 8.73603523e-01 -8.80235255e-01 -1.17208749e-01
3.34315181e-01 5.23935318e-01 -8.84050608e-01 6.96013644e-02
-9.86652300e-02 4.04927671e-01 -1.92497641e-01 4.90515977e-01
6.58115149e-01 -3.46548557e-01 -5.34788184e-02 -4.61984217e-01
-5.89154102e-02 4.01337922e-01 -1.01858819e+00 2.09645200e+00
-6.22673869e-01 5.19366682e-01 -1.67002678e-01 -1.19323730e+00
6.81611300e-01 -4.42215391e-02 6.35000646e-01 -1.07042527e+00
1.32956013e-01 2.48441935e-01 1.05961431e-02 -2.39656001e-01
1.62119418e-01 -8.56371038e-03 -9.66607630e-02 1.32989883e-01
2.60775357e-01 4.08210270e-02 1.96136415e-01 2.65877664e-01
1.34829116e+00 1.39571935e-01 2.45156422e-01 -2.86726415e-01
7.46296525e-01 -1.08645387e-01 7.58973420e-01 4.92411494e-01
-2.20612347e-01 7.04638541e-01 2.03389898e-01 -3.66713136e-01
-8.32731962e-01 -9.25457656e-01 -8.92306417e-02 1.11144698e+00
2.06261501e-01 -3.69444221e-01 -7.04074979e-01 -8.07144582e-01
2.68855859e-02 6.49051368e-01 -8.66870344e-01 -4.50552583e-01
-4.49733049e-01 -7.42953718e-01 2.67668366e-01 5.23181379e-01
7.92238593e-01 -8.86995554e-01 -5.31715810e-01 2.72320002e-01
-2.08210111e-01 -1.07901359e+00 -6.31726265e-01 2.56181359e-01
-5.69647133e-01 -9.23245311e-01 -5.60326815e-01 -6.44117594e-01
6.65723503e-01 4.35557395e-01 1.00826037e+00 -1.13606483e-01
-3.28360677e-01 2.91014910e-01 -3.37152898e-01 -2.90271640e-01
9.18979943e-02 2.04126090e-01 -2.78321147e-01 3.24177921e-01
4.26769704e-01 -6.17438734e-01 -9.07118380e-01 1.18153505e-01
-9.29202080e-01 -2.53233477e-03 7.99113512e-01 1.12759173e+00
7.55797386e-01 -1.40111148e-01 7.79008329e-01 -7.98404932e-01
2.37803191e-01 -5.27013421e-01 -4.03624326e-01 2.01856121e-02
-7.56122053e-01 6.58532605e-02 5.52277863e-01 -2.93858469e-01
-1.13710725e+00 1.70584366e-01 -2.30129927e-01 -5.82371294e-01
-5.48137091e-02 3.52832198e-01 -1.57822877e-01 1.07085146e-02
5.40241480e-01 4.24694180e-01 -8.39221925e-02 -6.02826834e-01
2.93831140e-01 4.29022044e-01 6.86615467e-01 -2.31251508e-01
7.37673104e-01 5.84131479e-01 -3.79967570e-01 -3.81152779e-01
-9.57147360e-01 -7.01667547e-01 -5.83324969e-01 -2.28844732e-02
6.65599108e-01 -1.30579591e+00 -1.43618762e-01 6.62917316e-01
-6.20414019e-01 -3.10720444e-01 -4.68676507e-01 3.46762985e-01
-3.06909204e-01 9.95855406e-02 -3.85050178e-01 -2.09363088e-01
-5.58752179e-01 -1.20556331e+00 1.08470213e+00 2.57788729e-02
-1.35647943e-02 -7.45706439e-01 1.67447582e-01 4.08304006e-01
7.79460490e-01 1.53910413e-01 7.75445402e-01 -6.34900510e-01
-5.73939562e-01 -1.95197284e-01 -3.22501868e-01 6.39983356e-01
2.13728830e-01 -3.93334210e-01 -1.07394457e+00 -4.11314607e-01
-2.58902431e-01 -3.31490189e-01 1.30794001e+00 3.44729543e-01
1.33248198e+00 -8.60844404e-02 -2.30518416e-01 6.12631559e-01
1.74288583e+00 -9.50313136e-02 5.43483675e-01 4.26509053e-01
7.20193446e-01 2.17724353e-01 5.05394042e-01 4.27931726e-01
5.85886240e-01 8.81873548e-01 5.16062021e-01 -3.12821597e-01
-5.47015667e-01 -1.98537618e-01 3.89387548e-01 5.85073233e-01
1.51450127e-01 3.41944210e-02 -8.17922831e-01 8.95357549e-01
-1.67592669e+00 -8.57362628e-01 1.30098507e-01 2.11809969e+00
1.01987517e+00 2.49046594e-01 1.14041113e-03 1.25164464e-01
6.20172381e-01 2.68566847e-01 -7.87434042e-01 -1.58556877e-03
-1.86087236e-01 5.42614818e-01 4.67894197e-01 3.12306494e-01
-1.30094278e+00 9.79970396e-01 4.95713997e+00 8.72952580e-01
-1.55764902e+00 5.07443309e-01 5.59692144e-01 -5.02584636e-01
-3.00294727e-01 -5.00735901e-02 -5.28785169e-01 5.13828516e-01
6.86240911e-01 -3.85479140e-03 3.87056559e-01 5.73958695e-01
1.82016790e-01 -2.49381468e-01 -8.86179209e-01 9.47176218e-01
2.58192897e-01 -1.28166461e+00 6.72328696e-02 -2.76704669e-01
9.24572647e-01 5.54807127e-01 1.36118934e-01 4.12749320e-01
2.67233700e-01 -6.31119788e-01 8.15943658e-01 4.51215327e-01
7.44776666e-01 -5.15161455e-01 7.94932544e-01 1.15473242e-02
-1.13955688e+00 -2.98057944e-01 -4.60199527e-02 8.43635052e-02
-1.53683826e-01 9.38757181e-01 -5.60048521e-01 6.06039941e-01
1.04892874e+00 1.01884675e+00 -8.59704554e-01 1.06466043e+00
-4.91139628e-02 7.49282539e-01 -3.19734037e-01 4.09918040e-01
3.86652529e-01 4.01240028e-02 5.62629580e-01 1.37981927e+00
2.28803262e-01 -1.77868068e-01 3.48991752e-02 6.97763503e-01
-2.51354754e-01 1.35587886e-01 -2.75938511e-01 2.20195085e-01
3.36848199e-01 1.27026010e+00 -4.36745286e-01 -4.03258234e-01
-6.17892683e-01 1.19684565e+00 3.37007433e-01 2.70526618e-01
-8.73916626e-01 -4.15223181e-01 8.52582574e-01 1.40900016e-01
9.10080791e-01 7.34339803e-02 -3.07098746e-01 -1.23733056e+00
2.16622651e-01 -1.00434828e+00 3.35079461e-01 -4.75618780e-01
-1.14465034e+00 6.87813044e-01 -2.12661922e-01 -1.19461560e+00
6.81562722e-02 -3.45708467e-02 -5.86230278e-01 7.17909932e-01
-1.88841164e+00 -1.29629219e+00 -5.00937164e-01 7.70292044e-01
7.55428076e-01 -1.08279251e-02 5.33066094e-01 7.29868174e-01
-8.36639524e-01 7.37565577e-01 1.44505203e-01 -5.06543182e-02
8.85183990e-01 -1.16171265e+00 3.59667748e-01 1.14106178e+00
4.52905819e-02 4.53387141e-01 4.93992269e-01 -3.37522656e-01
-1.18173492e+00 -1.33941209e+00 7.61663973e-01 -1.39255017e-01
5.45331299e-01 -4.13513780e-01 -1.06340766e+00 4.94221210e-01
3.18283409e-01 4.91283298e-01 4.78698194e-01 1.27105974e-02
-5.99441826e-01 -5.40119290e-01 -9.52345729e-01 3.58018190e-01
1.29844940e+00 -4.89277363e-01 -4.52678353e-01 1.56720709e-02
5.83993971e-01 -1.57858104e-01 -7.46011972e-01 4.98436183e-01
3.82039219e-01 -1.00131559e+00 9.56105769e-01 -3.19538236e-01
4.21976030e-01 -4.54951078e-01 -3.24669391e-01 -1.51372099e+00
-5.96160889e-01 -3.68455857e-01 2.63403922e-01 1.35629439e+00
2.71294177e-01 -6.78936660e-01 3.92718434e-01 1.24472350e-01
-3.16436350e-01 -9.29858744e-01 -8.31191719e-01 -5.97737849e-01
-1.61292613e-01 -4.02033269e-01 6.51711762e-01 9.49559927e-01
-3.64121944e-01 4.12945479e-01 -1.40834168e-01 6.81060702e-02
4.89546299e-01 1.20551877e-01 6.51747167e-01 -1.02005601e+00
-1.65526584e-01 -5.00509501e-01 -4.61518079e-01 -7.85905600e-01
9.27193463e-02 -1.06953561e+00 1.41891688e-01 -1.42678666e+00
2.93749154e-01 -5.36512911e-01 -8.00303519e-01 7.53593206e-01
-4.39478517e-01 5.67377806e-01 2.30131760e-01 2.16620445e-01
-7.27203131e-01 8.90303731e-01 1.01211321e+00 -2.88560420e-01
-1.35478392e-01 -3.51613432e-01 -9.47304845e-01 5.70630014e-01
6.43826723e-01 -3.69657069e-01 -3.90271515e-01 -5.31248629e-01
-1.62574947e-01 -6.51379764e-01 5.73894382e-01 -1.28233397e+00
1.15386419e-01 2.13525698e-01 3.35457861e-01 -4.15976584e-01
1.95921987e-01 -7.82966077e-01 9.56096947e-02 5.50182045e-01
-3.60993743e-01 -7.45072514e-02 2.28008807e-01 6.24118865e-01
-4.84489024e-01 1.71384797e-01 1.04787171e+00 -2.60153582e-04
-1.01384389e+00 3.60460401e-01 -6.89745545e-02 9.36260074e-02
1.02252579e+00 2.75370141e-04 -2.14697659e-01 -1.67333633e-01
-4.62510467e-01 9.55228508e-02 4.68036175e-01 5.33735037e-01
4.37974513e-01 -1.40115988e+00 -6.83864236e-01 5.00781298e-01
4.61686879e-01 -1.01487666e-01 4.43259567e-01 1.03793120e+00
-5.84201887e-02 2.90838212e-01 -2.40907356e-01 -6.93920791e-01
-1.11162615e+00 4.69897807e-01 3.41093689e-01 -4.26245213e-01
-7.03807235e-01 9.39257324e-01 3.88153076e-01 -3.06496590e-01
9.19835716e-02 -4.76332307e-01 4.36905585e-02 8.07673559e-02
4.41498011e-01 1.48309201e-01 5.35197914e-01 -6.49718761e-01
-5.81969678e-01 5.89915395e-01 -2.74379849e-01 1.25231341e-01
1.64196646e+00 -2.18577489e-01 6.80421963e-02 1.18569084e-01
1.45909357e+00 -7.71994963e-02 -1.43171334e+00 -8.00512433e-01
-1.64515093e-01 -5.21639764e-01 4.53804642e-01 -9.72388387e-01
-1.36447585e+00 7.75771439e-01 9.09992754e-01 -3.77798975e-01
1.70658553e+00 -4.47186008e-02 9.38718677e-01 2.17519045e-01
8.37673321e-02 -1.06392694e+00 2.02332705e-01 4.11677480e-01
1.15021026e+00 -1.33428943e+00 -1.48349926e-01 -2.13103950e-01
-7.15744734e-01 5.53799868e-01 5.86876929e-01 -3.81282598e-01
7.88366675e-01 1.31897390e-01 -8.26970339e-02 -1.38455704e-01
-8.98489714e-01 -5.49041450e-01 3.52622837e-01 2.85154223e-01
1.91528887e-01 -1.69330895e-01 -2.72979259e-01 6.22723401e-01
1.48173690e-01 2.34950960e-01 7.53597170e-02 9.68085587e-01
-2.70262361e-01 -1.01257467e+00 -3.64341885e-02 6.39548481e-01
-4.22831416e-01 -3.13065946e-01 -2.16391131e-01 5.09962618e-01
4.04748648e-01 6.65196836e-01 8.98540765e-02 -4.93634373e-01
5.96723676e-01 -2.10552383e-02 3.11270267e-01 -4.95203763e-01
-8.88968170e-01 6.92060441e-02 -1.28051057e-01 -9.34050679e-01
-5.42937517e-01 -8.89798164e-01 -1.04070175e+00 1.18301600e-01
-1.59431085e-01 -1.89916074e-01 4.78138804e-01 7.27290452e-01
8.04663360e-01 7.06277072e-01 7.06241906e-01 -8.81411791e-01
-3.63315433e-01 -1.07802689e+00 -2.01638848e-01 7.35455155e-01
5.18101752e-01 -8.67211401e-01 -2.53246665e-01 1.12093262e-01]
|
[9.83020305633545, 0.1941520869731903]
|
34883ebc-eb7b-48a5-8848-aef20b68a0d5
|
taylorimnet-for-fast-3d-shape-reconstruction
|
2201.06845
| null |
https://arxiv.org/abs/2201.06845v1
|
https://arxiv.org/pdf/2201.06845v1.pdf
|
TaylorImNet for Fast 3D Shape Reconstruction Based on Implicit Surface Function
|
Benefiting from the contiguous representation ability, deep implicit functions can extract the iso-surface of a shape at arbitrary resolution. However, utilizing the neural network with a large number of parameters as the implicit function prevents the generation speed of high-resolution topology because it needs to forward a large number of query points into the network. In this work, we propose TaylorImNet inspired by the Taylor series for implicit 3D shape representation. TaylorImNet exploits a set of discrete expansion points and corresponding Taylor series to model a contiguous implicit shape field. After the expansion points and corresponding coefficients are obtained, our model only needs to calculate the Taylor series to evaluate each point and the number of expansion points is independent of the generating resolution. Based on this representation, our TaylorImNet can achieve a significantly faster generation speed than other baselines. We evaluate our approach on reconstruction tasks from various types of input, and the experimental results demonstrate that our approach can get slightly better performance than existing state-of-the-art baselines while improving the inference speed with a large margin.
|
['Shenghua Gao', 'Jiale Xu', 'Yuting Xiao']
|
2022-01-18
| null | null | null | null |
['3d-shape-representation']
|
['computer-vision']
|
[-2.87438631e-01 1.41856775e-01 -5.69206290e-02 -1.98197886e-01
-9.07616973e-01 -5.45839846e-01 5.56878209e-01 -1.95954204e-01
-1.74280956e-01 6.21840358e-01 1.33266956e-01 -1.58520639e-01
5.83591089e-02 -1.27791643e+00 -1.17311120e+00 -2.85034120e-01
2.97976118e-02 8.09687078e-01 5.66439509e-01 -2.46290103e-01
2.68926680e-01 7.55597055e-01 -1.11349070e+00 2.84254432e-01
8.97563756e-01 1.30189562e+00 3.53157222e-01 3.82729590e-01
-5.20200968e-01 2.61593878e-01 -5.25875211e-01 -2.74132520e-01
2.96264440e-01 9.64015871e-02 -7.14652002e-01 -5.18929064e-01
7.64326811e-01 -8.99444878e-01 -2.44227409e-01 8.66753399e-01
6.99490190e-01 1.34815559e-01 7.02246606e-01 -6.17618442e-01
-7.73626506e-01 3.98206264e-01 -5.43961525e-01 3.15930173e-02
1.30310297e-01 -1.00414835e-01 8.74320447e-01 -1.46528471e+00
7.93359160e-01 1.41020942e+00 1.00208652e+00 1.85737222e-01
-1.31789398e+00 -7.93743670e-01 1.19751535e-01 -7.73056522e-02
-1.58197808e+00 -4.39432770e-01 8.20775688e-01 6.26412919e-03
1.03137577e+00 4.75888513e-02 7.63049066e-01 6.42954290e-01
-3.13901454e-01 6.93754077e-01 5.64801514e-01 -3.48057784e-02
1.44300526e-02 -2.82636136e-01 -1.50800899e-01 8.98910224e-01
-4.86011431e-02 1.58619583e-01 -3.81088197e-01 -2.02235267e-01
1.86805665e+00 -1.25494733e-01 -3.37911695e-01 -2.91083962e-01
-1.19057488e+00 9.01602507e-01 9.45505619e-01 1.62413865e-01
-3.16559017e-01 6.07449234e-01 2.85720646e-01 1.24289624e-01
5.13074160e-01 4.58088905e-01 -4.45679039e-01 -8.06248561e-02
-1.11069870e+00 3.19306195e-01 7.14646518e-01 1.19843924e+00
9.00105953e-01 1.26702949e-01 -2.13212848e-01 9.18376744e-01
8.05086046e-02 5.80339074e-01 -1.52475657e-02 -1.22447073e+00
6.16846263e-01 4.74131346e-01 1.46604121e-01 -1.28273141e+00
-1.58520207e-01 -5.46326518e-01 -1.14274848e+00 4.14235666e-02
3.72718573e-01 1.06954493e-01 -7.85328448e-01 1.51156807e+00
3.86905700e-01 3.86103064e-01 -3.59981626e-01 1.00220311e+00
9.30721104e-01 9.55777526e-01 -4.37392712e-01 1.57001130e-02
1.24851513e+00 -9.95940030e-01 -3.93513411e-01 3.37799132e-01
3.29913348e-01 -7.59631634e-01 9.54733968e-01 1.73939541e-01
-1.58398128e+00 -6.07017577e-01 -9.03900564e-01 -5.50976813e-01
-5.05226478e-02 2.52051711e-01 5.55397511e-01 -2.23476216e-01
-1.22027671e+00 7.91871071e-01 -7.27459490e-01 2.24565983e-01
5.45292735e-01 1.29701152e-01 -1.48705930e-01 -2.07151901e-02
-1.12313569e+00 6.23768151e-01 1.23087078e-01 1.08752593e-01
-6.73803747e-01 -1.13422811e+00 -8.59826148e-01 3.16263765e-01
3.09051156e-01 -1.08171976e+00 1.17936862e+00 -4.84473079e-01
-1.35979962e+00 4.39309120e-01 -4.73744184e-01 -4.59738225e-01
6.85648441e-01 -2.56026298e-01 1.00115322e-01 4.81937438e-01
1.44261852e-01 8.92786920e-01 8.37612629e-01 -1.12937689e+00
-2.62398362e-01 -1.10870667e-01 1.08384848e-01 1.09001823e-01
-1.83141809e-02 -3.86657417e-01 -7.19573259e-01 -8.93589079e-01
4.24632788e-01 -5.69971919e-01 -1.67947173e-01 4.98836190e-01
-2.25063130e-01 -4.68744844e-01 7.86111176e-01 -5.74181020e-01
1.12592971e+00 -1.95530927e+00 6.61396459e-02 3.10577244e-01
3.34016889e-01 2.03943282e-01 -1.53169289e-01 2.64563143e-01
2.73179919e-01 3.32160294e-01 -2.21228749e-01 -2.67299801e-01
-2.23690365e-02 1.62185192e-01 -6.89438641e-01 1.31652638e-01
2.02601939e-01 1.19706821e+00 -8.49542916e-01 -5.25009871e-01
2.10826043e-02 7.43342280e-01 -8.60855579e-01 8.01972598e-02
-4.52878207e-01 2.53431797e-01 -7.61156321e-01 3.89779449e-01
9.21802521e-01 -5.53571284e-01 -1.50763482e-01 -5.42077482e-01
-1.17591582e-01 7.80876696e-01 -1.07693243e+00 2.10392141e+00
-5.95789850e-01 3.86958480e-01 4.02789051e-03 -8.34403455e-01
1.15667653e+00 2.67037630e-01 2.03910232e-01 -6.87807739e-01
-2.42128402e-01 4.11031455e-01 -3.83947968e-01 4.84222546e-03
4.44917262e-01 1.93463102e-01 1.74899355e-01 5.20110130e-01
-2.67386526e-01 -3.08995694e-01 -4.10354510e-02 1.74561009e-01
7.44848669e-01 3.55749786e-01 -1.03726402e-01 -2.64367968e-01
3.99010032e-01 -1.64750308e-01 5.86135805e-01 5.65203369e-01
4.22155559e-01 7.62144566e-01 3.76008213e-01 -8.86780858e-01
-1.35851312e+00 -1.20050669e+00 -2.67373115e-01 6.33651733e-01
2.48450696e-01 -5.03747880e-01 -6.82505190e-01 -4.26424980e-01
1.12018980e-01 4.19000894e-01 -3.54000688e-01 2.74878770e-01
-1.28378856e+00 -3.43084931e-02 5.99970400e-01 9.06356692e-01
8.67998779e-01 -9.64186311e-01 -4.36984986e-01 3.04141045e-01
-3.30007851e-01 -9.90522802e-01 -7.88698435e-01 -4.11978543e-01
-1.26454127e+00 -7.72116125e-01 -9.73152578e-01 -9.39465880e-01
9.23881292e-01 1.73292950e-01 1.37333882e+00 4.37881887e-01
1.62832558e-01 -1.06302612e-01 6.22855797e-02 -3.89817520e-03
-1.18795171e-01 2.84459114e-01 -4.75061625e-01 -4.60758001e-01
-1.76811919e-01 -1.05899024e+00 -9.08834696e-01 2.94742972e-01
-6.88815415e-01 4.00843948e-01 5.29261649e-01 8.35793734e-01
1.05653512e+00 -1.24388663e-02 6.81326628e-01 -5.63705742e-01
4.30922806e-01 -2.27933273e-01 -7.31887460e-01 9.05104876e-02
-2.23885626e-01 4.45169121e-01 8.01407278e-01 -4.20084268e-01
-8.76942873e-01 -2.17481554e-01 -3.56324315e-01 -7.69493401e-01
2.62632221e-01 5.43340147e-01 1.33461639e-01 6.50519365e-03
5.10592163e-01 3.69648188e-01 -7.92328268e-03 -9.44535077e-01
6.20059609e-01 2.00910121e-02 5.97424209e-01 -9.37107444e-01
8.82598460e-01 5.53857505e-01 1.38240367e-01 -5.71731985e-01
-8.21666598e-01 -1.06367677e-01 -5.18401384e-01 1.34710550e-01
4.41385150e-01 -8.40431333e-01 -6.86944902e-01 2.95720369e-01
-1.70093203e+00 -2.57194012e-01 -3.36507708e-01 1.15122534e-01
-4.67765480e-01 3.37757885e-01 -9.27067995e-01 -5.10477960e-01
-7.54755139e-01 -1.14967859e+00 1.33522284e+00 -1.06265105e-01
-2.02479567e-02 -7.83837855e-01 -1.67332113e-01 -1.31829292e-01
7.01577008e-01 9.96785983e-02 1.03848398e+00 -2.29107589e-02
-1.24443018e+00 7.81846717e-02 -5.65719843e-01 -5.35964221e-02
-1.69421420e-01 -1.78538442e-01 -6.57679975e-01 -1.73455805e-01
-6.32207170e-02 -1.76979214e-01 9.59967256e-01 3.30169678e-01
1.64661050e+00 -6.45287097e-01 -2.63935119e-01 1.21546304e+00
1.34854102e+00 -2.77984589e-01 7.19055772e-01 2.11091340e-01
6.71835005e-01 1.54357523e-01 3.30110312e-01 2.19107747e-01
5.49147367e-01 7.11688459e-01 4.89285022e-01 -2.20293075e-01
-2.74902403e-01 -7.74296463e-01 7.00951144e-02 9.37167287e-01
-2.97693014e-01 2.20427468e-01 -5.77833533e-01 4.90722805e-01
-1.69192600e+00 -9.35047626e-01 4.89166304e-02 2.18967295e+00
1.04947484e+00 2.30656639e-02 5.37631661e-03 -2.30950058e-01
6.58346951e-01 1.56993851e-01 -6.51663601e-01 -3.17801237e-01
-1.04496293e-02 4.80328709e-01 3.54517758e-01 6.19862378e-01
-7.18527496e-01 9.65033889e-01 6.56452608e+00 1.18264258e+00
-1.25496006e+00 -9.97081995e-02 5.42370915e-01 -1.18752897e-01
-7.29770541e-01 -3.13482992e-02 -8.72926831e-01 3.32953185e-01
3.93276900e-01 4.53760801e-03 6.29340708e-01 7.85070121e-01
8.09940845e-02 3.44800442e-01 -1.03922868e+00 9.92463708e-01
-2.37825468e-01 -1.85046625e+00 4.84863639e-01 6.14850856e-02
7.38577545e-01 1.43609658e-01 -7.64551293e-03 1.19037539e-01
1.23643935e-01 -1.29145694e+00 7.79924870e-01 5.94942033e-01
1.28936768e+00 -7.53665030e-01 2.99157739e-01 6.16623700e-01
-1.58858275e+00 2.89935350e-01 -7.93846190e-01 6.79401979e-02
3.67236584e-01 6.81195676e-01 -6.86360240e-01 5.39938331e-01
5.26561379e-01 6.98960006e-01 -7.49605298e-02 8.40318143e-01
-2.70472348e-01 4.07851398e-01 -7.27849782e-01 1.73680276e-01
1.89889356e-01 -3.88092428e-01 5.45693994e-01 9.94869888e-01
7.08252072e-01 2.15497479e-01 8.18400607e-02 1.46930528e+00
-3.70056927e-01 1.84246469e-02 -5.27803719e-01 2.88837999e-01
1.02476895e+00 1.12430024e+00 -4.19685960e-01 -5.00735641e-01
-2.41076082e-01 7.29106486e-01 8.04694653e-01 4.67954457e-01
-8.04292321e-01 -2.66584277e-01 3.17005843e-01 4.39176708e-01
7.50907481e-01 -4.67301458e-01 -4.56043929e-01 -9.71638083e-01
2.54513413e-01 -4.97639269e-01 1.18822105e-01 -1.02222812e+00
-1.16483045e+00 6.86167181e-01 -1.04379784e-02 -1.17021906e+00
-2.92313933e-01 -2.95501024e-01 -5.82802951e-01 1.26106918e+00
-1.74920487e+00 -1.27036285e+00 -1.88430429e-01 4.81135905e-01
3.48108292e-01 1.65320635e-01 8.28337133e-01 2.57552743e-01
8.44913721e-02 6.62921309e-01 -2.47059569e-01 4.22705829e-01
4.41821814e-01 -1.10680354e+00 8.82337987e-01 4.34659362e-01
8.36726800e-02 9.48703408e-01 4.34851572e-02 -6.41581655e-01
-1.30336022e+00 -1.02238011e+00 7.55626023e-01 -2.25537583e-01
1.42287731e-01 -1.85477927e-01 -1.16568851e+00 6.70676351e-01
-8.63191858e-02 3.01634878e-01 7.20136017e-02 -4.93167825e-02
-5.89030445e-01 -1.92561552e-01 -1.05855811e+00 6.09676957e-01
1.42403531e+00 -3.99396449e-01 -5.58863640e-01 1.33969784e-02
1.01523483e+00 -7.77571857e-01 -1.16685379e+00 5.92027187e-01
5.78086793e-01 -8.54175568e-01 1.42219710e+00 -2.25226611e-01
7.31077433e-01 -4.05070335e-01 7.83774406e-02 -1.10981691e+00
-5.46338975e-01 -6.70966506e-01 -5.35503745e-01 9.78238940e-01
2.90349275e-01 -7.80654073e-01 6.52138591e-01 3.66000980e-01
-1.98258281e-01 -1.30265415e+00 -9.65782344e-01 -5.87864280e-01
3.93834859e-01 -2.59178549e-01 1.13681304e+00 6.85453475e-01
-5.95775127e-01 4.12937790e-01 -1.88017920e-01 1.10241450e-01
5.94899654e-01 5.93410969e-01 9.15796340e-01 -1.18627131e+00
-3.08261096e-01 -4.81455117e-01 2.26586927e-02 -2.21194005e+00
-7.17421547e-02 -1.05697083e+00 -3.41190100e-01 -1.69062972e+00
-5.36709726e-02 -1.00866711e+00 2.13085134e-02 3.43662888e-01
-1.19894221e-02 1.76509798e-01 2.49276072e-01 4.17584777e-01
-2.97581464e-01 7.64839351e-01 2.05888867e+00 -5.21909148e-02
-8.84641930e-02 -2.88111985e-01 -6.19132459e-01 9.57339168e-01
5.02125442e-01 -1.72160327e-01 -5.15244007e-01 -1.03218627e+00
2.76570410e-01 3.47499937e-01 4.29356098e-01 -6.06606483e-01
3.20842475e-01 -3.57928290e-03 7.74051428e-01 -1.16668856e+00
5.88393629e-01 -6.74115598e-01 2.63526678e-01 1.79652542e-01
-2.85739422e-01 2.41323978e-01 1.64806247e-01 3.87772322e-01
-5.87262437e-02 -2.09472120e-01 5.84655464e-01 -3.53084505e-01
-2.80913621e-01 8.08016777e-01 2.68632472e-01 2.27600023e-01
3.49072546e-01 -2.37194479e-01 -4.16755825e-01 -3.12542975e-01
-5.46946704e-01 1.72619671e-01 6.88783646e-01 1.73761472e-01
9.12315428e-01 -1.76240444e+00 -6.52764618e-01 3.21359992e-01
-5.37313282e-01 1.09418273e+00 1.15497537e-01 6.63143456e-01
-7.42613733e-01 3.19014788e-01 2.07817823e-01 -6.97198987e-01
-6.68859184e-01 3.70385885e-01 5.27996659e-01 -4.21692312e-01
-1.15545821e+00 7.59093106e-01 5.52653015e-01 -5.07221758e-01
1.64193019e-01 -5.26326716e-01 3.90901826e-02 -3.38217229e-01
4.81657475e-01 3.54459882e-01 -2.10078627e-01 -3.81911159e-01
1.78692527e-02 1.03059864e+00 -2.19509080e-02 -8.73822495e-02
1.34855211e+00 3.15102518e-01 -2.05228239e-01 8.06552395e-02
1.33700359e+00 1.99704453e-01 -1.40476024e+00 -3.89595449e-01
-5.27528942e-01 -6.01029396e-01 -1.17795793e-02 -6.99665844e-01
-1.22968781e+00 1.00122416e+00 -1.01381399e-01 4.78064492e-02
9.09773111e-01 -1.75144345e-01 1.40332079e+00 3.61207455e-01
6.49159610e-01 -8.33410144e-01 3.05296071e-02 7.51133740e-01
1.31880736e+00 -7.86235273e-01 -7.24412650e-02 -7.73701966e-01
-5.96465878e-02 1.23333788e+00 5.41271448e-01 -5.24533510e-01
6.81236446e-01 3.53375703e-01 -2.24652573e-01 -2.76284456e-01
-6.92031264e-01 2.04561919e-01 4.78328079e-01 4.14493471e-01
3.95482957e-01 -8.49312767e-02 -2.01745331e-01 5.00699282e-01
-4.80724603e-01 -1.47597548e-02 -3.20122540e-02 4.64933723e-01
-5.07225811e-01 -1.03362405e+00 -2.20564663e-01 4.26054180e-01
-3.14790249e-01 -3.48420203e-01 -6.21467866e-02 6.25640571e-01
-1.63627982e-01 2.54491448e-01 3.91866624e-01 1.41941577e-01
4.02678311e-01 -2.66250074e-01 6.52465165e-01 -4.26287740e-01
-2.01056749e-01 2.53974319e-01 1.76371932e-02 -7.38619804e-01
-1.06117174e-01 -2.29614496e-01 -1.71399164e+00 -5.41179776e-01
-2.54449844e-01 8.12468454e-02 4.93631423e-01 5.43259978e-01
7.64201283e-01 3.45830113e-01 5.13377309e-01 -1.00440586e+00
-7.56485581e-01 -9.16425943e-01 -1.08320177e-01 4.22554851e-01
2.30696991e-01 -7.04150558e-01 -8.50291103e-02 -1.79983735e-01]
|
[8.730974197387695, -3.6092123985290527]
|
03c7f183-142f-414d-984d-0ff38cc7d870
|
dnn-based-mask-estimation-for-distributed
|
2011.01714
| null |
https://arxiv.org/abs/2011.01714v1
|
https://arxiv.org/pdf/2011.01714v1.pdf
|
DNN-based mask estimation for distributed speech enhancement in spatially unconstrained microphone arrays
|
Deep neural network (DNN)-based speech enhancement algorithms in microphone arrays have now proven to be efficient solutions to speech understanding and speech recognition in noisy environments. However, in the context of ad-hoc microphone arrays, many challenges remain and raise the need for distributed processing. In this paper, we propose to extend a previously introduced distributed DNN-based time-frequency mask estimation scheme that can efficiently use spatial information in form of so-called compressed signals which are pre-filtered target estimations. We study the performance of this algorithm under realistic acoustic conditions and investigate practical aspects of its optimal application. We show that the nodes in the microphone array cooperate by taking profit of their spatial coverage in the room. We also propose to use the compressed signals not only to convey the target estimation but also the noise estimation in order to exploit the acoustic diversity recorded throughout the microphone array.
|
['Slim Essid', 'Irina Illina', 'Romain Serizel', 'Nicolas Furnon']
|
2020-11-03
| null | null | null | null |
['noise-estimation']
|
['medical']
|
[ 4.03649420e-01 1.21874258e-01 6.39544070e-01 -1.84525520e-01
-8.32393527e-01 -3.39538038e-01 2.84595907e-01 -1.40676558e-01
-6.82157755e-01 5.45201421e-01 5.76625824e-01 -1.76959783e-01
-5.93766630e-01 -4.98026133e-01 -5.12452006e-01 -1.13574719e+00
-3.96612853e-01 -6.70162439e-02 -5.09356894e-02 3.80189046e-02
-1.42517552e-01 5.55266440e-01 -1.59810412e+00 -1.83954835e-02
5.60509324e-01 1.09335923e+00 8.29755962e-01 1.13354182e+00
4.65040952e-02 5.72750151e-01 -1.20312142e+00 6.55363873e-02
3.13396990e-01 -3.75911266e-01 1.93995431e-01 1.42098978e-01
2.26596985e-02 -4.28179622e-01 -5.23118019e-01 1.10852432e+00
1.39361942e+00 4.65898097e-01 1.77105948e-01 -6.25099540e-01
2.04219490e-01 7.46185839e-01 -2.35409766e-01 3.81498754e-01
-1.73879161e-01 -1.67376667e-01 4.43461746e-01 -9.55919206e-01
7.74214417e-02 9.80093658e-01 6.59226596e-01 5.72254181e-01
-4.19698387e-01 -6.93269372e-01 -5.43071143e-02 2.51384020e-01
-1.28810287e+00 -1.16372180e+00 9.10491943e-01 1.45088688e-01
7.75919557e-01 4.51895058e-01 4.00105059e-01 6.22930050e-01
-3.36134225e-01 6.59137428e-01 9.13799167e-01 -7.58623660e-01
6.25318170e-01 -5.00538424e-02 -3.75180483e-01 2.14130297e-01
3.51522490e-02 3.35017860e-01 -1.16548812e+00 -1.67084172e-01
5.48436701e-01 -3.08948368e-01 -5.85142493e-01 1.90959215e-01
-8.93095016e-01 5.46002150e-01 3.39620918e-01 8.14963877e-01
-8.37198675e-01 1.99829817e-01 2.14235540e-02 3.22914362e-01
7.81435907e-01 2.59212345e-01 -3.44334036e-01 -1.88193709e-01
-1.06972468e+00 -7.91485384e-02 8.73465240e-01 4.35939431e-01
5.06725013e-01 4.47202682e-01 4.64253016e-02 1.19022739e+00
3.40672314e-01 6.31266356e-01 1.51268721e-01 -1.17422950e+00
4.59115505e-01 -3.99249345e-01 -2.54231747e-02 -9.82170165e-01
-2.27433860e-01 -1.15818357e+00 -1.12283373e+00 2.12771431e-01
1.22610003e-01 -7.68080354e-01 -5.56447744e-01 1.79179060e+00
5.84566057e-01 6.27793133e-01 3.63339216e-01 8.95986915e-01
2.66472548e-01 7.80166864e-01 -4.76950377e-01 -4.73546803e-01
6.95136786e-01 -9.50279832e-01 -1.08493233e+00 -2.90631503e-01
2.68657565e-01 -7.20050514e-01 2.35369485e-02 8.51137340e-01
-1.41514087e+00 -3.93013686e-01 -1.05685544e+00 5.72158277e-01
-1.13197438e-01 -4.20779437e-02 9.93020162e-02 1.16013527e+00
-1.60580337e+00 2.61915624e-01 -8.09941411e-01 3.77778187e-02
3.59025508e-01 5.93151748e-01 -1.08081445e-01 -3.63412291e-01
-8.73990417e-01 3.83263856e-01 -1.34938359e-01 5.45817256e-01
-1.04345751e+00 -5.86153388e-01 -6.80195808e-01 1.74310192e-01
2.16115683e-01 -3.82940024e-01 1.37510860e+00 -7.76742280e-01
-1.53709650e+00 7.55017772e-02 -5.81783831e-01 -4.53683913e-01
1.66593328e-01 -3.88961099e-02 -3.76760423e-01 3.33916932e-01
-3.03241044e-01 3.77130210e-01 6.62468135e-01 -1.24404132e+00
-8.09946477e-01 -4.25678968e-01 -1.87478706e-01 3.97324711e-01
-1.04574049e+00 4.09553349e-02 -2.70456314e-01 -7.89335966e-01
1.69976071e-01 -4.33962494e-01 -8.82336438e-01 -5.46753407e-04
-3.10314447e-01 9.28472281e-02 7.30760694e-01 -9.79146719e-01
1.17008662e+00 -2.57899904e+00 1.12082608e-01 4.36614454e-01
4.52434987e-01 5.07027090e-01 -2.50848830e-01 5.54263234e-01
3.14382792e-01 -4.38364208e-01 -2.72643924e-01 -1.02909672e+00
-1.82282135e-01 3.64627093e-01 -4.19874080e-02 6.14286780e-01
-3.15375388e-01 9.07810628e-02 -6.76724315e-01 -1.25261992e-01
4.37164098e-01 7.94686437e-01 -5.71023405e-01 3.97208810e-01
1.26663834e-01 7.05747485e-01 -3.70938271e-01 1.94508642e-01
1.10681462e+00 5.27695119e-01 1.39740363e-01 1.52808562e-01
-1.63254306e-01 2.03606322e-01 -1.36345255e+00 1.68079555e+00
-9.46874917e-01 8.69176269e-01 1.38157439e+00 -1.30525458e+00
6.60234094e-01 7.78093278e-01 3.67818594e-01 -5.16361833e-01
7.20182657e-02 5.75839639e-01 5.41268848e-02 -3.58935714e-01
2.81271398e-01 -7.14161247e-02 3.86180848e-01 2.98473567e-01
3.14228386e-02 1.81581467e-01 -3.33601475e-01 -9.48807970e-02
1.44156647e+00 -9.06528473e-01 -2.01809756e-03 -4.96316254e-02
5.77741086e-01 -7.07967341e-01 1.79508865e-01 1.06957114e+00
-6.65988848e-02 3.17654848e-01 -2.74402499e-01 2.16194138e-01
-8.11152518e-01 -8.80359590e-01 2.00648502e-01 1.04896712e+00
-1.53163537e-01 -4.75071482e-02 -9.76696610e-01 -9.08296928e-02
-5.58291078e-01 5.18628776e-01 -1.37587339e-01 2.24306285e-01
-6.74752176e-01 -4.08832580e-01 5.63969135e-01 3.57302427e-01
5.98643541e-01 -7.33332694e-01 -5.68094671e-01 5.61248541e-01
-1.12004355e-01 -1.23392713e+00 -3.09489876e-01 5.94258070e-01
-5.67998052e-01 -3.04951966e-01 -1.11308730e+00 -8.11613619e-01
6.21451318e-01 5.90201616e-01 6.42024994e-01 -4.73115640e-03
4.34622504e-02 7.49974370e-01 -3.05526406e-01 -5.61477423e-01
-4.99984860e-01 -2.73848206e-01 1.68075383e-01 2.50149876e-01
-1.74818393e-02 -1.08251202e+00 -6.85502708e-01 2.12596655e-01
-8.93091857e-01 -3.87348205e-01 5.39048195e-01 6.99814141e-01
1.51059315e-01 6.97216809e-01 8.60412419e-01 -5.68052121e-02
8.11825752e-01 -3.45587075e-01 -5.49035370e-01 -6.45931661e-02
6.41115010e-02 -4.37877357e-01 5.09380698e-01 -3.58771980e-01
-1.30632222e+00 1.14548102e-01 -6.74594462e-01 -3.31621081e-01
-1.70308113e-01 5.85845470e-01 -4.73165095e-01 -3.30410659e-01
3.62648338e-01 2.75481611e-01 -3.74061316e-02 -7.00122535e-01
2.83822864e-01 1.17425644e+00 6.14703655e-01 -8.31888020e-02
6.64855540e-01 5.27845562e-01 1.14385210e-01 -1.33295691e+00
-2.91117132e-01 -6.82572126e-01 -1.24777354e-01 -4.15875107e-01
5.02235770e-01 -9.64663267e-01 -5.34395158e-01 6.34123266e-01
-1.40723002e+00 -2.71975368e-01 -2.91791916e-01 7.72078156e-01
-2.03148633e-01 3.78706068e-01 -4.74148482e-01 -1.69550323e+00
-2.21554071e-01 -8.61447930e-01 9.13940787e-01 1.05974652e-01
3.89933616e-01 -9.95687783e-01 -7.02994317e-02 6.04073592e-02
9.63876247e-01 -2.32585475e-01 1.64997622e-01 -5.78438699e-01
-6.67660892e-01 -3.99842635e-02 2.10765079e-01 5.48224032e-01
2.08398044e-01 -8.72982502e-01 -1.49631143e+00 -2.73776472e-01
7.06577599e-01 1.96704492e-01 8.95070374e-01 8.63391399e-01
1.03288937e+00 -4.02462691e-01 -3.14793378e-01 5.03758430e-01
1.23741424e+00 3.08993995e-01 4.49994504e-01 -2.26605177e-01
2.51981497e-01 8.86889398e-01 1.60574421e-01 6.88475668e-01
-1.84002861e-01 5.79133928e-01 5.21494925e-01 -1.71748728e-01
-4.11025077e-01 1.72201470e-01 3.37500513e-01 1.21276939e+00
3.50930661e-01 -8.98718894e-01 -4.31111097e-01 8.77981722e-01
-1.51284707e+00 -6.42753363e-01 1.88174009e-01 2.07671595e+00
5.62754393e-01 -2.70178437e-01 -2.00198829e-01 5.28553486e-01
7.97914505e-01 2.60443717e-01 -3.00084502e-01 -1.85484365e-01
-3.88117403e-01 7.50304282e-01 5.81305563e-01 7.42067516e-01
-6.80486977e-01 2.29311869e-01 6.04240990e+00 1.22768855e+00
-9.23237979e-01 5.67318022e-01 4.76150423e-01 -2.47587383e-01
-2.53062516e-01 -7.01039910e-01 -4.46852595e-01 1.34035945e-01
1.24726820e+00 6.67778701e-02 6.40664220e-01 4.86681044e-01
5.10517299e-01 -2.16086641e-01 -7.35636055e-01 9.42337990e-01
3.55787342e-03 -1.18774629e+00 -6.43441498e-01 3.97887498e-01
5.33060491e-01 1.42336905e-01 2.10446596e-01 -3.32910448e-01
2.13537037e-01 -9.80182052e-01 5.28210878e-01 1.96440205e-01
5.05845845e-01 -7.47039974e-01 6.38373435e-01 7.50106037e-01
-1.11050951e+00 -4.87114787e-01 -2.98631340e-01 -3.68327886e-01
5.56711137e-01 1.24706137e+00 -9.65327322e-01 5.57510316e-01
7.01417208e-01 2.37870917e-01 2.61603296e-01 1.38335466e+00
-1.61614865e-01 8.17048967e-01 -8.56202900e-01 -1.52293831e-01
2.16057718e-01 1.43521994e-01 1.00130463e+00 1.07859457e+00
1.06478643e+00 -1.15039265e-02 -1.98824599e-01 4.10222352e-01
-8.36763903e-02 -2.15049312e-01 -7.12919056e-01 3.64407033e-01
6.88082337e-01 9.86451685e-01 -3.63273829e-01 -1.32158458e-01
-2.23172650e-01 6.70833468e-01 -1.64354429e-01 5.51882863e-01
-2.11196989e-01 -3.37531775e-01 8.02510023e-01 -1.78353086e-01
7.97336400e-01 -6.22514784e-01 -2.50608921e-01 -4.36141789e-01
1.71266094e-01 -7.66415834e-01 -1.23383895e-01 -6.79728448e-01
-9.78576601e-01 7.46537030e-01 -2.57389426e-01 -8.58548939e-01
-2.63107091e-01 -5.08491814e-01 -5.82612574e-01 1.06734335e+00
-1.51167202e+00 -7.37852275e-01 -7.12451413e-02 6.28543198e-01
7.34831214e-01 -1.90863624e-01 7.78803825e-01 8.39323878e-01
-2.37787932e-01 5.33809543e-01 6.62007749e-01 -1.06232436e-02
1.54072344e-01 -8.82129014e-01 8.10807049e-02 1.20242321e+00
2.00969800e-01 2.99687803e-01 9.48667109e-01 -2.99846858e-01
-1.20241725e+00 -9.52482760e-01 8.02140176e-01 2.55310833e-01
2.09322721e-01 -6.28961384e-01 -5.34431219e-01 5.19544259e-02
7.28665471e-01 -1.07546344e-01 7.13924110e-01 -3.23623091e-01
3.09456974e-01 -3.70712489e-01 -1.08888459e+00 3.11634094e-01
1.06139016e+00 -2.81987667e-01 -1.68628972e-02 4.07607943e-01
6.48444951e-01 -2.38252714e-01 -3.19734275e-01 1.13409050e-01
1.46149486e-01 -9.79080021e-01 1.03988826e+00 2.58988202e-01
-1.61880314e-01 -1.96801707e-01 -6.64568722e-01 -1.59572077e+00
2.73633420e-01 -1.08625782e+00 -7.62119219e-02 1.23142576e+00
2.87231177e-01 -8.06766331e-01 9.41732883e-01 6.52089566e-02
-4.41683829e-01 -4.31713313e-01 -1.77619982e+00 -6.62595034e-01
-2.02934027e-01 -9.14314747e-01 4.23801452e-01 2.86374420e-01
-2.38687575e-01 -8.00583735e-02 -5.73288679e-01 6.74184680e-01
7.88844585e-01 -7.80991375e-01 5.33693314e-01 -7.79369354e-01
-6.17633700e-01 -2.33449623e-01 -2.55523801e-01 -1.73049212e+00
1.18951380e-01 -2.73870856e-01 5.56080580e-01 -1.46985126e+00
-4.77869034e-01 -5.04971206e-01 -3.14495623e-01 -2.20473498e-01
1.10208936e-01 1.60026118e-01 1.93270177e-01 -3.05766135e-01
-3.57782304e-01 5.33596277e-01 8.91790688e-01 -7.41689280e-02
-5.05315512e-02 4.83680993e-01 -6.97222769e-01 4.43993181e-01
6.04973197e-01 -4.98423189e-01 -4.48758721e-01 -8.20298195e-01
-4.28336160e-03 3.84344250e-01 2.87826329e-01 -1.30473042e+00
8.90959918e-01 3.89061004e-01 9.63905230e-02 -6.52248800e-01
9.55833554e-01 -1.14020181e+00 6.87576085e-02 5.16449392e-01
-4.33127671e-01 -3.32205504e-01 4.08367328e-02 8.31951737e-01
-4.49825972e-01 -3.01114947e-01 4.85125571e-01 1.01251330e-03
-2.67826945e-01 -9.04504061e-02 -9.02727067e-01 -5.87381303e-01
6.69224083e-01 -8.47796872e-02 5.56815788e-02 -1.11751258e+00
-7.28402078e-01 3.34495045e-02 -3.50151777e-01 -1.47775993e-01
7.57172346e-01 -7.53660619e-01 -8.03108156e-01 2.07311049e-01
-6.39439523e-01 4.97003496e-02 7.27917373e-01 6.90628529e-01
-1.84071381e-02 4.50237036e-01 4.55329776e-01 -5.68139851e-01
-1.26653838e+00 2.62802243e-01 6.03710413e-01 -9.49975252e-02
-2.97010183e-01 1.37131953e+00 3.06470066e-01 -2.43291840e-01
6.87279522e-01 -7.81923831e-02 -1.51657656e-01 -4.33200330e-01
7.85704195e-01 4.39097553e-01 3.39848965e-01 -3.18565160e-01
-2.08330780e-01 3.94006342e-01 6.33434355e-01 -8.03044260e-01
1.47797978e+00 -6.81643486e-01 -2.04052389e-01 -1.22115523e-01
1.21505654e+00 6.58951998e-01 -1.08734584e+00 -4.94058579e-01
-3.03901464e-01 -5.57309806e-01 4.41917300e-01 -7.29241550e-01
-1.34820569e+00 1.10133541e+00 9.00440693e-01 3.37153524e-01
1.72600412e+00 4.77551557e-02 9.33203280e-01 3.52765471e-01
3.78710270e-01 -7.73992240e-01 1.20185474e-02 3.48312467e-01
5.77215552e-01 -5.27321935e-01 -4.96499896e-01 -3.71785223e-01
1.70951962e-01 9.96160626e-01 6.64902255e-02 2.82269895e-01
9.39863801e-01 8.03558528e-01 2.75573194e-01 1.46707833e-01
-6.48848355e-01 -1.86132088e-01 -2.87179172e-01 1.20089161e+00
-1.80187654e-02 -1.35882292e-02 2.71417052e-02 4.96906728e-01
-3.01514715e-01 -3.59065473e-01 2.86624402e-01 8.50591362e-01
-7.71502852e-01 -1.24724996e+00 -8.53237987e-01 2.81294823e-01
-5.71074545e-01 -2.64618367e-01 1.33820123e-03 -8.10093060e-02
1.09649986e-01 1.78039289e+00 -8.64683241e-02 -5.07946074e-01
2.07784355e-01 -3.97109121e-01 1.32188216e-01 -3.40881377e-01
-6.13392532e-01 6.31859064e-01 3.34507585e-01 -1.19971819e-02
-5.86987853e-01 -4.66840595e-01 -7.47630239e-01 -1.27313420e-01
-5.21577895e-01 4.89475757e-01 1.28150475e+00 9.78563666e-01
4.43782002e-01 9.65199769e-01 1.04756176e+00 -1.06022727e+00
-6.22366488e-01 -9.40736413e-01 -7.43063450e-01 -5.01686752e-01
8.59927535e-01 -9.99870077e-02 -8.42997551e-01 -2.91781574e-01]
|
[14.996808052062988, 5.900113582611084]
|
01e217b9-c919-4d85-a2b8-0f00dd05ac41
|
a-term-based-methodology-for-query
|
1601.04615
| null |
http://arxiv.org/abs/1601.04615v2
|
http://arxiv.org/pdf/1601.04615v2.pdf
|
A Term-Based Methodology for Query Reformulation Understanding
|
Key to any research involving session search is the understanding of how a
user's queries evolve throughout the session. When a user creates a query
reformulation, he or she is consciously retaining terms from their original
query, removing others and adding new terms. By measuring the similarity
between queries we can make inferences on the user's information need and how
successful their new query is likely to be. By identifying the origins of added
terms we can infer the user's motivations and gain an understanding of their
interactions.
In this paper we present a novel term-based methodology for understanding and
interpreting query reformulation actions. We use TREC Session Track data to
demonstrate how our technique is able to learn from query logs and we make use
of click data to test user interaction behavior when reformulating queries. We
identify and evaluate a range of term-based query reformulation strategies and
show that our methods provide valuable insight into understanding query
reformulation in session search.
|
['Wang Jun', 'Yang Hui', 'Sloan Marc']
|
2016-01-19
| null | null | null | null |
['session-search']
|
['natural-language-processing']
|
[ 4.26341116e-01 -1.07099950e-01 -4.60051447e-01 -5.50459564e-01
-8.44038069e-01 -1.08879340e+00 8.27452004e-01 6.94033027e-01
-7.31094897e-01 2.15664148e-01 4.03682560e-01 -8.45690906e-01
-3.76208067e-01 -4.87248033e-01 -4.19994414e-01 1.64432928e-01
-7.38422945e-02 4.99437094e-01 4.73294646e-01 -4.56343919e-01
8.05419743e-01 4.08020824e-01 -1.67286813e+00 4.28512841e-01
6.86248124e-01 4.19202596e-01 3.24707299e-01 1.07205033e+00
-2.31288970e-01 6.72604561e-01 -5.97579539e-01 -1.83918476e-01
6.97676539e-02 -3.86631906e-01 -1.28373170e+00 -2.53237009e-01
2.61496931e-01 -4.14079249e-01 -3.41677040e-01 6.53895378e-01
1.22348063e-01 3.18389028e-01 2.63811767e-01 -1.10625136e+00
-2.43818671e-01 4.84800726e-01 1.28920108e-01 7.74294078e-01
1.05562139e+00 -1.94090202e-01 1.27737498e+00 -4.66451287e-01
6.71239138e-01 9.38652694e-01 1.51306584e-01 3.56021911e-01
-1.49788952e+00 -3.95407468e-01 2.45317027e-01 1.83559582e-01
-1.17655063e+00 -7.08018541e-01 5.15378475e-01 -4.50873256e-01
8.99899304e-01 9.80733991e-01 4.43381190e-01 6.77190244e-01
-1.71161771e-01 8.23070884e-01 7.67127216e-01 -6.78854048e-01
1.79127485e-01 5.11626542e-01 6.36055291e-01 2.81365126e-01
3.29464749e-02 6.79852441e-02 -5.99071622e-01 -7.72973895e-01
2.47160047e-01 -2.16954462e-02 -1.33440062e-01 -3.46524060e-01
-7.86108434e-01 6.45830154e-01 4.29384649e-01 7.19743550e-01
-4.30736601e-01 2.12376807e-02 2.55553219e-02 5.05922616e-01
-5.02797961e-02 9.99139905e-01 -3.33023906e-01 -6.22488201e-01
-6.87429488e-01 7.70264804e-01 1.27474427e+00 8.90316486e-01
7.99185872e-01 -8.61809373e-01 -4.79599923e-01 7.40711570e-01
3.15337390e-01 2.66804844e-01 4.94862407e-01 -1.07027364e+00
1.27410457e-01 7.06137061e-01 6.05645835e-01 -7.67462432e-01
9.05185491e-02 -4.46417928e-01 4.37566161e-01 -3.61387670e-01
2.02509493e-01 1.99009314e-01 -8.96680117e-01 1.63609588e+00
-2.58961052e-01 -4.41669792e-01 -2.03501284e-01 2.98442155e-01
2.42230520e-01 1.22304089e-01 2.02324674e-01 -3.31892490e-01
1.49356925e+00 -2.05094546e-01 -6.10769153e-01 -3.44097048e-01
1.14270639e+00 -9.67436731e-01 1.16334331e+00 9.77451168e-03
-1.04158878e+00 -3.24138373e-01 -6.16832435e-01 -1.44693911e-01
-4.77909744e-01 -5.48702002e-01 7.13531256e-01 7.18258858e-01
-9.56604481e-01 4.84824568e-01 -7.33334661e-01 -8.79183233e-01
-1.34145841e-01 3.79989296e-01 3.79724830e-01 3.11282184e-02
-1.10885215e+00 6.22820318e-01 9.40740407e-02 -3.40118080e-01
-5.75925410e-01 -6.63533688e-01 -3.11239064e-01 4.93040234e-01
6.15750194e-01 -4.89458561e-01 2.15758824e+00 -2.02876404e-01
-6.82064772e-01 7.65377522e-01 -1.04110992e+00 -2.85294026e-01
1.51027525e-02 -2.35730521e-02 -3.41016650e-01 -2.33019143e-01
3.27419221e-01 1.53683782e-01 4.09260780e-01 -1.24151480e+00
-1.14986253e+00 -6.42047584e-01 4.57662106e-01 6.23707414e-01
-2.25166008e-01 1.28334522e-01 -7.20132232e-01 -6.79791942e-02
2.22787991e-01 -1.08275008e+00 4.28061672e-02 -5.55895329e-01
-2.37215906e-01 -4.55048859e-01 7.78320312e-01 -3.95438403e-01
2.07096791e+00 -1.93894053e+00 -1.50840566e-01 6.67138338e-01
4.72609848e-01 1.85209289e-01 4.98192981e-02 1.12005019e+00
2.64847815e-01 8.26530576e-01 3.01989913e-01 -4.89310138e-02
7.85872489e-02 1.55146763e-01 -2.96099752e-01 -4.04623777e-01
-6.66386724e-01 1.03991389e+00 -7.63874471e-01 -4.44415599e-01
-1.44189462e-01 5.79518676e-02 -7.66997635e-01 3.20383549e-01
-3.31891418e-01 9.36893225e-02 -8.64039421e-01 5.39567888e-01
9.42712948e-02 -4.24909234e-01 9.39598009e-02 1.50653362e-01
-2.77934402e-01 7.12049127e-01 -5.32547295e-01 1.36389291e+00
-4.45852190e-01 6.45117402e-01 9.75970551e-02 -4.73620683e-01
1.51087254e-01 2.02878386e-01 3.29667896e-01 -9.93461430e-01
-1.29055023e-01 1.12183459e-01 -4.82558459e-03 -7.68538237e-01
7.01774895e-01 1.82102263e-01 -1.17767595e-01 1.03804386e+00
-5.33804774e-01 3.76179665e-02 5.25729239e-01 7.80492663e-01
1.41684055e+00 -6.90298975e-01 1.44975096e-01 -3.39508094e-02
4.77148950e-01 1.08698249e-01 -8.83616507e-02 1.67273271e+00
-2.71650881e-01 -1.87336907e-01 3.09561431e-01 -7.60235218e-03
-6.39417648e-01 -1.04724240e+00 9.20066833e-02 1.81178927e+00
2.59742290e-01 -7.39314377e-01 -4.59885806e-01 -5.44716954e-01
1.53611273e-01 1.16122663e+00 -5.26432872e-01 -3.45122248e-01
-2.86194563e-01 -7.31089264e-02 3.13755512e-01 1.82307065e-01
2.76111037e-01 -8.38000298e-01 -5.02953947e-01 7.05196187e-02
-5.72403371e-01 -6.51947200e-01 -7.66722202e-01 -1.01813506e-02
-8.06599081e-01 -1.08127034e+00 -1.08030722e-01 -5.67483366e-01
4.48688060e-01 7.78061867e-01 1.06898201e+00 4.94480819e-01
-3.83721650e-01 1.03813541e+00 -3.23685884e-01 -4.22226757e-01
-3.34171206e-01 3.41348380e-01 -4.47700381e-01 -4.01576102e-01
8.85016024e-01 -3.91647428e-01 -8.10542583e-01 5.64017951e-01
-1.02930844e+00 -4.18282300e-01 4.42074746e-01 3.97793710e-01
-6.93441480e-02 1.32043615e-01 8.58317167e-02 -1.22956693e+00
1.67205071e+00 -5.59214234e-01 -3.85918587e-01 6.88587308e-01
-1.22851503e+00 3.45536143e-01 -2.46354043e-01 -2.04549447e-01
-1.29020000e+00 -2.57714450e-01 1.48257136e-01 3.60942572e-01
-5.97477406e-02 7.87308156e-01 4.87206995e-01 -1.15293555e-01
9.38153267e-01 4.37045604e-01 -2.16054544e-01 -6.01249039e-01
3.52991402e-01 7.75656819e-01 3.10794920e-01 -4.94177580e-01
7.54334450e-01 1.66902766e-01 -7.12339401e-01 -9.44598138e-01
-6.49653912e-01 -1.34739935e+00 -3.24734986e-01 -2.06977800e-01
6.58158362e-01 -3.25910598e-01 -1.39790034e+00 -1.74950920e-02
-6.54718995e-01 -4.22905833e-02 -6.87939534e-03 2.60269433e-01
-2.64186203e-01 2.47421518e-01 -3.26845616e-01 -9.16978776e-01
1.47734269e-01 -8.37087274e-01 9.10385072e-01 4.66956168e-01
-8.36747229e-01 -1.12966394e+00 3.50362718e-01 7.66937733e-01
7.51046717e-01 -7.80251622e-01 1.24785888e+00 -1.18055975e+00
-8.56728435e-01 -5.74045300e-01 -4.55170125e-02 -3.71332377e-01
3.95312458e-01 -3.46196353e-01 -6.71492100e-01 -3.44797105e-01
1.15827210e-01 -1.33643821e-02 4.39575672e-01 1.60044432e-01
1.00287950e+00 -3.48873317e-01 -7.34307885e-01 8.59377533e-02
1.16022205e+00 6.61091685e-01 4.58640844e-01 2.72397995e-01
5.03563955e-02 5.32846808e-01 3.79294515e-01 3.15527350e-01
3.88201982e-01 7.74617434e-01 -5.16756438e-03 2.98733264e-01
6.03029907e-01 -6.41645551e-01 -1.15685768e-01 4.32955176e-02
9.52060968e-02 -5.62038779e-01 -7.21152365e-01 3.27441543e-01
-1.72307050e+00 -1.16989660e+00 3.54228228e-01 2.55912781e+00
7.57888317e-01 4.02691811e-01 9.57235768e-02 -1.11451507e-01
4.52411681e-01 -3.85440253e-02 -6.84708357e-01 -3.09647113e-01
7.23437548e-01 3.42258722e-01 6.32569134e-01 1.07349813e+00
-5.57160735e-01 9.32774246e-01 7.32192516e+00 1.96347505e-01
-6.47924721e-01 -4.03493106e-01 1.51973933e-01 -1.18188947e-01
-6.37937427e-01 4.79868233e-01 -9.12457466e-01 1.80471852e-01
9.34865415e-01 -9.36227500e-01 8.72725785e-01 6.20733082e-01
3.06996614e-01 -6.05296731e-01 -1.53706276e+00 7.09446132e-01
5.42018637e-02 -1.05919945e+00 -2.04900987e-02 4.01127279e-01
7.42824525e-02 -1.95079923e-01 -3.23665068e-02 7.10288644e-01
4.93639886e-01 -8.47729087e-01 -1.10507384e-01 6.12629414e-01
-5.26814200e-02 -3.52647960e-01 2.18561545e-01 6.88885033e-01
-8.77412558e-01 -2.66160667e-01 2.41176888e-01 -2.07466587e-01
7.52506703e-02 -1.94469288e-01 -1.58405590e+00 6.22764602e-02
3.50976884e-01 4.12961617e-02 -1.15063822e+00 9.90301430e-01
2.21397564e-01 6.59104705e-01 -4.60270107e-01 -3.74861419e-01
-7.96948187e-03 1.45050019e-01 5.80708504e-01 7.70005167e-01
1.08985133e-01 1.84567779e-01 2.73382425e-01 7.68327236e-01
8.05825647e-03 8.26517940e-02 -6.00232005e-01 -6.38431609e-01
7.28468835e-01 8.63842845e-01 -4.92870390e-01 -5.58671653e-01
-3.37050021e-01 1.01240516e+00 -2.89657544e-02 7.64200568e-01
-2.48302817e-01 -5.62447488e-01 7.11549938e-01 4.33403939e-01
-3.25023979e-02 -2.15579122e-01 1.21729478e-01 -1.07832658e+00
2.27315530e-01 -8.63089204e-01 4.85627145e-01 -8.92902017e-01
-6.38527453e-01 1.80669472e-01 4.27241862e-01 -3.55809987e-01
-7.81783283e-01 -1.05060615e-01 -6.43344820e-01 1.07236052e+00
-1.12802398e+00 -4.09638524e-01 -3.52897763e-01 4.37299848e-01
5.16320825e-01 2.11405471e-01 7.78785169e-01 2.23790169e-01
-1.11933261e-01 5.40797591e-01 -8.09203833e-02 -9.43773910e-02
7.07974553e-01 -1.11210430e+00 2.83949316e-01 4.54232424e-01
2.42150664e-01 1.51438606e+00 1.00206339e+00 -8.12584043e-01
-1.50518274e+00 -1.53992996e-01 1.43466234e+00 -7.61099160e-01
5.61355889e-01 -3.38452011e-01 -1.11978400e+00 9.43166256e-01
9.48260874e-02 -7.58359969e-01 9.34460402e-01 7.53957212e-01
-1.72513694e-01 8.79374593e-02 -7.89636075e-01 8.93844664e-01
1.17017651e+00 -1.06999874e+00 -8.49808455e-01 5.07634103e-01
5.40655792e-01 1.07166141e-01 -4.48335588e-01 1.66729644e-01
8.94362509e-01 -1.13147712e+00 1.02090847e+00 -8.04503381e-01
-3.99337560e-01 3.54216732e-02 -1.28393337e-01 -9.68234956e-01
-3.98033708e-01 -7.73762345e-01 1.45651549e-01 8.34757566e-01
6.19372010e-01 -6.34755909e-01 1.01748276e+00 1.49546134e+00
4.43526179e-01 -3.11290890e-01 -3.27189773e-01 -4.55446780e-01
-3.73271644e-01 -2.98895717e-01 4.56939608e-01 4.43031698e-01
3.12209815e-01 4.88058031e-01 1.25379354e-01 -3.95153463e-02
4.25287127e-01 -2.79773287e-02 8.05760860e-01 -1.41383088e+00
-3.36958408e-01 -4.87947285e-01 5.86248860e-02 -1.60642612e+00
-4.67583030e-01 -8.47836018e-01 -1.66249126e-01 -1.37235439e+00
3.04239810e-01 -2.78381735e-01 -3.81357312e-01 2.04448327e-01
-6.03942536e-02 -4.52699751e-01 -5.52593591e-03 5.71160734e-01
-6.86725855e-01 -5.95637001e-02 7.05461740e-01 1.68218836e-01
-7.80704975e-01 6.60685658e-01 -1.10015905e+00 4.03630853e-01
6.50606275e-01 -4.13324326e-01 -9.55703318e-01 -9.80242789e-02
5.67539573e-01 2.12578699e-01 2.20889747e-01 -4.82200652e-01
7.77875960e-01 -3.27722549e-01 1.69416121e-03 -6.27244890e-01
2.48712465e-01 -6.97564542e-01 -5.60274571e-02 6.09989107e-01
-1.23287618e+00 1.87584385e-01 -1.02357455e-01 7.83754766e-01
-1.58231989e-01 -4.92927164e-01 9.74766687e-02 -3.35828483e-01
-7.11196423e-01 6.96619675e-02 -7.78586566e-01 2.15079620e-01
8.54164898e-01 -2.76501507e-01 2.62632370e-02 -9.67265189e-01
-9.68684196e-01 7.02723384e-01 4.57676828e-01 5.99993646e-01
3.68707657e-01 -9.68764842e-01 -1.32205009e-01 4.46952105e-01
5.55751920e-01 -7.98341095e-01 -2.06910253e-01 2.67762691e-01
-2.22811356e-01 1.14969969e+00 4.18814898e-01 -5.51730335e-01
-1.81464946e+00 1.74698830e-01 1.50577903e-01 -3.07594627e-01
1.90221861e-01 7.51438618e-01 1.12463869e-01 -1.15153894e-01
5.78216791e-01 -2.20040366e-01 -1.24523461e-01 -2.45333873e-02
4.29588050e-01 8.62592161e-02 3.07299085e-02 -3.85608375e-02
-2.84768164e-01 1.95122302e-01 -7.41600454e-01 -6.72529399e-01
7.11468935e-01 -5.79035342e-01 6.65726811e-02 4.93382961e-01
1.48237121e+00 1.98323429e-01 -3.94152373e-01 -5.43901682e-01
4.54863697e-01 -9.92767096e-01 -1.23724915e-01 -1.01059341e+00
-7.44012743e-02 4.27182168e-01 5.00201821e-01 9.17814553e-01
9.46359515e-01 3.07493240e-01 6.34055555e-01 1.18387628e+00
3.01573396e-01 -9.59970295e-01 2.24703491e-01 4.27482605e-01
6.22779191e-01 -1.15900493e+00 7.48035125e-03 -2.48795554e-01
-3.39730561e-01 7.44788349e-01 3.23679030e-01 5.47710896e-01
7.86929309e-01 -4.10062224e-01 9.11770537e-02 -7.03479767e-01
-1.03141284e+00 -2.82134056e-01 2.82791227e-01 6.67713657e-02
7.04265356e-01 -2.29686603e-01 -6.42815650e-01 -1.45245239e-01
-3.87394458e-01 1.05777346e-02 2.98303127e-01 1.45766807e+00
-8.25513422e-01 -1.55453813e+00 8.52887034e-02 8.06272030e-01
-4.93044347e-01 -1.90214202e-01 -8.49162638e-01 5.93257427e-01
-5.83015323e-01 1.27128112e+00 -1.72661602e-01 -4.89520341e-01
4.71632719e-01 7.01458812e-01 2.53399819e-01 -9.86606777e-01
-5.09775758e-01 5.45487739e-02 3.03934813e-01 -7.02085912e-01
-1.38610944e-01 -7.33885288e-01 -8.96785080e-01 -2.06958145e-01
-3.72404575e-01 8.47073615e-01 8.49075019e-01 8.07055235e-01
7.58252025e-01 1.32548183e-01 7.41794944e-01 -6.11548778e-03
-6.61291540e-01 -8.49509358e-01 -1.16224669e-01 8.97010684e-01
2.82304049e-01 -4.20959741e-01 -6.56277239e-01 6.04586937e-02]
|
[12.011751174926758, 7.709576606750488]
|
a749a870-cbba-45c1-9dd1-5e999c828064
|
reinforced-co-training
|
1804.06035
| null |
http://arxiv.org/abs/1804.06035v1
|
http://arxiv.org/pdf/1804.06035v1.pdf
|
Reinforced Co-Training
|
Co-training is a popular semi-supervised learning framework to utilize a
large amount of unlabeled data in addition to a small labeled set. Co-training
methods exploit predicted labels on the unlabeled data and select samples based
on prediction confidence to augment the training. However, the selection of
samples in existing co-training methods is based on a predetermined policy,
which ignores the sampling bias between the unlabeled and the labeled subsets,
and fails to explore the data space. In this paper, we propose a novel method,
Reinforced Co-Training, to select high-quality unlabeled samples to better
co-train on. More specifically, our approach uses Q-learning to learn a data
selection policy with a small labeled dataset, and then exploits this policy to
train the co-training classifiers automatically. Experimental results on
clickbait detection and generic text classification tasks demonstrate that our
proposed method can obtain more accurate text classification results.
|
['Lei LI', 'William Yang Wang', 'Jiawei Wu']
|
2018-04-17
|
reinforced-co-training-1
|
https://aclanthology.org/N18-1113
|
https://aclanthology.org/N18-1113.pdf
|
naacl-2018-6
|
['clickbait-detection']
|
['natural-language-processing']
|
[ 2.67130286e-01 1.98400207e-03 -9.72475886e-01 -6.68706298e-01
-8.14365089e-01 -4.23797816e-01 3.35441738e-01 2.40915492e-01
-4.84836608e-01 9.94882584e-01 -1.08800188e-01 -4.30305600e-01
1.09090053e-01 -7.87755430e-01 -4.79300022e-01 -5.79288721e-01
3.85032415e-01 5.81619620e-01 2.70471841e-01 3.50260854e-01
3.03852886e-01 -1.37392610e-01 -1.61233759e+00 3.98255318e-01
1.53855515e+00 8.40977788e-01 2.24301144e-01 3.13477933e-01
-3.74052078e-01 7.47662723e-01 -4.51094538e-01 -6.37992173e-02
2.79809803e-01 -7.09178448e-01 -5.77601731e-01 5.08164525e-01
2.43492544e-01 -3.25992793e-01 8.43059719e-02 1.00756204e+00
2.64795899e-01 8.74498487e-02 5.38290739e-01 -1.11038077e+00
-5.21901608e-01 8.10282052e-01 -6.10736966e-01 -8.21783300e-03
1.37369558e-01 -1.60148125e-02 1.07362998e+00 -8.89232516e-01
3.77829254e-01 9.26378191e-01 4.81614918e-01 5.73758364e-01
-1.09306633e+00 -1.09042311e+00 3.70099038e-01 -1.53771371e-01
-1.38163877e+00 -1.23358987e-01 9.69069421e-01 -2.23383889e-01
1.67350903e-01 1.02654226e-01 6.92643642e-01 8.30451190e-01
-1.91669568e-01 1.31994593e+00 1.62131882e+00 -8.55751932e-01
5.62962353e-01 8.36777568e-01 5.06745994e-01 6.83483064e-01
2.86997497e-01 3.04415911e-01 -3.90472084e-01 -5.37592828e-01
4.40557152e-01 3.03369641e-01 -1.79755427e-02 -5.04626334e-01
-8.95680547e-01 1.06382668e+00 1.44776881e-01 2.13391691e-01
-1.94182128e-01 -4.21760440e-01 3.26185524e-01 3.11664253e-01
8.10877383e-01 2.62484401e-01 -6.51445687e-01 1.17461652e-01
-1.20270765e+00 -8.41756165e-02 7.76703358e-01 9.13246632e-01
1.11765110e+00 -1.69395849e-01 -1.90080240e-01 9.67951357e-01
5.96521676e-01 4.19807941e-01 9.67890143e-01 -3.73219430e-01
7.27200687e-01 9.81356800e-01 1.95945650e-01 -3.20409507e-01
-2.09814087e-02 -5.07589757e-01 -3.69416535e-01 1.35549635e-01
3.17323148e-01 -4.91863877e-01 -9.42117393e-01 1.36997819e+00
5.93669891e-01 1.16915211e-01 3.10745873e-02 7.23769546e-01
2.82806367e-01 3.53772849e-01 2.75542617e-01 -5.67812800e-01
8.75704110e-01 -1.27864397e+00 -7.85878718e-01 -9.31842476e-02
9.83709633e-01 -5.58354795e-01 1.06083667e+00 4.89070445e-01
-5.93728900e-01 -7.15620697e-01 -1.12405777e+00 5.47965944e-01
-1.67441100e-01 4.23682421e-01 6.09543502e-01 9.27076876e-01
-3.30442578e-01 6.61409199e-01 -5.46646059e-01 -1.44435436e-01
7.02296555e-01 3.64129841e-01 7.99365118e-02 -3.50813210e-01
-1.14877975e+00 3.78243953e-01 9.04379308e-01 -3.25710297e-01
-8.21400642e-01 -2.78786302e-01 -4.23112303e-01 2.56844424e-02
6.34096265e-01 -1.54683843e-01 1.22493649e+00 -1.54131901e+00
-1.44972599e+00 5.73994577e-01 -1.52923897e-01 -3.93507570e-01
7.09533870e-01 -3.19709122e-01 -6.15619838e-01 3.04174796e-02
1.73870921e-01 5.90630949e-01 1.11661124e+00 -1.35267389e+00
-1.01169205e+00 -4.68767613e-01 -4.12694812e-01 4.14977938e-01
-6.86458886e-01 -2.29230374e-01 -2.46446878e-01 -5.22607565e-01
1.00580886e-01 -8.96886826e-01 -3.81811291e-01 -2.53775328e-01
-5.11633158e-01 -5.13223588e-01 1.07715762e+00 -6.73008412e-02
1.35311675e+00 -1.92541492e+00 -6.89549029e-01 5.07679105e-01
2.03492224e-01 5.52030027e-01 -8.86545777e-02 2.33754888e-01
-2.88505349e-02 3.18549573e-01 -1.51706291e-02 -2.07234144e-01
-3.28453988e-01 5.51252253e-02 -1.44449919e-01 2.31188685e-01
4.35857475e-02 6.07249260e-01 -1.22171593e+00 -9.02338207e-01
1.00604400e-01 -1.84837803e-01 -4.53208506e-01 4.03878093e-01
-4.90115196e-01 5.16743898e-01 -1.05308414e+00 6.19960845e-01
6.65466487e-01 -5.88107347e-01 4.12032694e-01 8.41558278e-02
6.12561144e-02 1.96100339e-01 -1.30218387e+00 1.35441768e+00
-3.02918613e-01 7.38361105e-02 -4.86257792e-01 -1.14083314e+00
1.14769089e+00 2.68150955e-01 5.25222540e-01 -3.64597440e-01
3.01493436e-01 4.34767038e-01 -3.44552919e-02 -4.63267386e-01
1.53837755e-01 -5.67247393e-03 2.98986048e-01 9.97087181e-01
6.31969273e-02 3.28410357e-01 2.80117869e-01 1.92976981e-01
5.57832658e-01 2.14084432e-01 4.25428540e-01 -1.37721866e-01
5.19535422e-01 1.53836861e-01 6.04791880e-01 9.26837981e-01
-3.92988443e-01 1.75770432e-01 8.85620713e-02 -1.62323192e-01
-9.43695664e-01 -5.11017919e-01 -2.94408172e-01 1.24717045e+00
2.01745540e-01 -3.23453188e-01 -5.44284761e-01 -1.71334469e+00
1.38410470e-02 7.56985664e-01 -5.68771362e-01 -3.33123416e-01
-2.20209777e-01 -6.58599615e-01 1.28412619e-01 4.68634933e-01
6.31901979e-01 -9.05988693e-01 -2.24031806e-01 1.81160659e-01
2.61269957e-02 -6.98124826e-01 -7.15341866e-01 5.13765335e-01
-1.26116049e+00 -1.22634590e+00 -4.74627495e-01 -9.88824010e-01
1.12542140e+00 4.52266574e-01 7.86592066e-01 1.81348026e-01
2.34014034e-01 4.84398417e-02 -6.99084997e-01 -4.56371754e-01
-5.61870217e-01 4.99235913e-02 -4.52403091e-02 1.78330854e-01
8.74677539e-01 -1.68980822e-01 -4.88235295e-01 6.30373418e-01
-7.47591853e-01 -4.53584269e-03 6.35405600e-01 1.40257335e+00
6.42618060e-01 2.55965024e-01 9.95400071e-01 -1.53172374e+00
5.96919775e-01 -6.62611663e-01 -4.17068720e-01 4.98377085e-01
-1.39292431e+00 1.81696773e-01 7.45778680e-01 -9.16050434e-01
-1.17304254e+00 2.82456607e-01 3.28665316e-01 -4.26883996e-01
-2.01678216e-01 6.89392328e-01 -1.08408228e-01 8.80274326e-02
7.53680050e-01 1.81731209e-01 8.25955346e-02 -4.81126130e-01
2.97348410e-01 1.05157197e+00 -1.43551379e-01 -5.00278771e-01
7.28140891e-01 3.69061887e-01 -5.74478209e-01 -1.08483322e-01
-1.35442793e+00 -8.09215307e-01 -7.52793550e-01 -7.80151412e-02
3.51924866e-01 -9.06768620e-01 -2.51878142e-01 -2.27228254e-02
-4.50393587e-01 -2.13361233e-01 -4.47555870e-01 9.84147191e-01
-2.37190545e-01 4.07740742e-01 -1.78819716e-01 -1.09853017e+00
-3.47550392e-01 -9.24659789e-01 7.69656181e-01 5.37951708e-01
9.21065174e-03 -9.25371945e-01 1.34331271e-01 4.24513429e-01
-4.82380763e-03 -3.27255011e-01 5.57414353e-01 -1.52036715e+00
-3.73481333e-01 -5.50199509e-01 -9.14972723e-02 4.11060780e-01
4.24288154e-01 -1.16230197e-01 -8.37937891e-01 -5.19558847e-01
-1.03009917e-01 -1.08985078e+00 7.65899181e-01 1.77976981e-01
1.40713787e+00 -2.60259688e-01 -5.86003780e-01 6.23324141e-03
1.34034514e+00 3.79634142e-01 1.48791537e-01 2.20326409e-01
5.18908501e-01 6.37221694e-01 1.19713032e+00 6.23044968e-01
5.44129461e-02 1.75336853e-01 1.82964697e-01 -9.69097987e-02
2.48260185e-01 -7.38311410e-01 5.52033111e-02 7.27892518e-01
3.73903871e-01 -1.25112981e-01 -5.59116304e-01 2.57887691e-01
-1.82681108e+00 -7.05070555e-01 1.53823286e-01 2.45666170e+00
1.28131056e+00 2.25014240e-01 3.82470429e-01 3.42817843e-01
1.11595809e+00 -2.34907404e-01 -9.23145235e-01 3.14324588e-01
1.28700435e-01 1.64046526e-01 5.43555379e-01 1.79931641e-01
-1.24952650e+00 8.39578986e-01 6.52820539e+00 1.10611057e+00
-1.08323491e+00 6.25058711e-02 7.83471227e-01 4.53322604e-02
-3.61542374e-01 1.80783957e-01 -1.06335402e+00 6.17830992e-01
9.01454985e-01 -5.46686053e-02 5.06413803e-02 1.20747983e+00
2.46958926e-01 -2.03964233e-01 -9.95445192e-01 7.00928450e-01
2.19464879e-02 -1.08825672e+00 -4.50295489e-03 -8.99111181e-02
1.09148884e+00 -1.47055075e-01 -8.55317339e-02 6.86428905e-01
7.81420469e-01 -7.32211411e-01 3.93613964e-01 1.45229712e-01
6.40165091e-01 -6.99124813e-01 6.89853311e-01 7.76209712e-01
-8.81919801e-01 -3.11776847e-01 -3.63143086e-01 3.59429151e-01
-1.80234656e-01 6.41691446e-01 -9.84052896e-01 3.66310716e-01
3.79814267e-01 9.38724160e-01 -6.95136547e-01 1.11227977e+00
-2.36674383e-01 1.21539783e+00 -1.02134317e-01 -5.42165458e-01
1.35684341e-01 -2.56529391e-01 3.11481301e-02 8.49407971e-01
-8.37032795e-02 -1.43451199e-01 1.03196132e+00 7.24930286e-01
-5.08971475e-02 5.70811450e-01 -2.04947203e-01 -1.99127644e-01
7.47466862e-01 1.09561539e+00 -7.00958490e-01 -7.05982089e-01
-5.42122662e-01 6.91619039e-01 4.43924040e-01 3.68210971e-01
-6.74183011e-01 -3.17099184e-01 -4.66718942e-01 -6.84690922e-02
2.91489750e-01 2.35297859e-01 -2.74848044e-01 -1.30623889e+00
-8.53048712e-02 -9.08947945e-01 6.20578527e-01 -3.93771440e-01
-1.58299637e+00 3.21785659e-01 -1.72768105e-02 -1.71910453e+00
-2.05081716e-01 -1.50872007e-01 -5.33033669e-01 9.11494017e-01
-1.72209609e+00 -1.03515661e+00 -2.93770164e-01 5.84621370e-01
6.28329456e-01 -4.20400679e-01 5.54441571e-01 1.31900191e-01
-5.95114291e-01 7.29093432e-01 5.69459677e-01 1.81385085e-01
1.00602770e+00 -1.19715750e+00 -3.40588868e-01 2.86859274e-01
1.84935704e-01 7.24000812e-01 3.19630295e-01 -9.65848625e-01
-8.48824143e-01 -1.15791571e+00 4.99241918e-01 -8.21056962e-02
4.60034728e-01 -1.20480664e-01 -1.09448922e+00 6.21973157e-01
6.38073832e-02 3.02429795e-01 1.29492748e+00 1.73344523e-01
-4.30244625e-01 -6.20772243e-02 -1.23783278e+00 2.58467555e-01
5.78641713e-01 -3.95524830e-01 -5.75297713e-01 6.24940932e-01
5.96280038e-01 -5.77612221e-02 -7.04558671e-01 2.66469419e-01
5.53090215e-01 -6.97131813e-01 3.48164707e-01 -9.32486713e-01
2.22379163e-01 -2.31098399e-01 1.97591871e-01 -1.42804420e+00
-3.48500013e-02 -2.59293884e-01 -1.29200429e-01 1.28891134e+00
5.42491853e-01 -6.77495122e-01 1.38789618e+00 4.95357126e-01
3.15938622e-01 -7.79455721e-01 -4.01290208e-01 -7.78554916e-01
-6.66664168e-02 -6.24894686e-02 5.43594122e-01 1.25696266e+00
2.83991337e-01 4.81238902e-01 -4.39159572e-01 -1.11686110e-01
6.66238487e-01 5.20507157e-01 7.61911273e-01 -1.47130907e+00
-4.42347944e-01 -2.83069182e-02 2.64173329e-01 -1.40553486e+00
1.76777810e-01 -9.41871822e-01 1.99711725e-01 -9.46178555e-01
4.57017004e-01 -1.08084345e+00 -5.91544032e-01 4.94145334e-01
-8.36125135e-01 -1.39854312e-01 -1.88827828e-01 7.13323057e-01
-9.80874538e-01 7.56178319e-01 1.40998256e+00 -1.52704820e-01
-5.82926571e-01 4.38709438e-01 -7.26909518e-01 5.01273274e-01
8.56887460e-01 -7.35185087e-01 -6.65479064e-01 2.35135093e-01
-2.96632022e-01 1.21090543e-02 -2.37873420e-01 -7.06954956e-01
2.80081648e-02 -4.31850404e-01 5.98904431e-01 -8.02686930e-01
-2.64470279e-01 -9.54624414e-01 -4.47092414e-01 5.05966425e-01
-1.07756793e+00 -5.67274511e-01 -2.14110911e-01 9.88018274e-01
-1.88862458e-01 -7.90022790e-01 8.66423547e-01 -1.66160569e-01
-3.88658643e-01 4.34368938e-01 -2.60298550e-01 1.03390053e-01
1.04358411e+00 -1.30891830e-01 -1.60518095e-01 -1.62653953e-01
-6.30643010e-01 6.05420351e-01 1.88805610e-01 2.51327366e-01
3.82381767e-01 -1.52198040e+00 -5.08002758e-01 2.25333050e-01
4.43148345e-01 -8.26460272e-02 -1.32273734e-01 4.54286963e-01
1.11787483e-01 3.35788906e-01 -7.65198004e-03 -6.53154910e-01
-1.15315986e+00 8.36656749e-01 5.06122708e-02 -6.79856777e-01
-1.28133968e-01 5.43932498e-01 -2.00711831e-01 -6.54367745e-01
2.87866145e-01 1.31762326e-01 -5.84571362e-01 -9.69903320e-02
5.10617614e-01 1.19870514e-01 -1.55234620e-01 -1.67660341e-02
1.34003922e-01 1.10687844e-01 -5.12760937e-01 -6.12667613e-02
8.14395785e-01 -2.28226066e-01 2.98006356e-01 6.81017220e-01
1.27158225e+00 2.18648955e-01 -1.28327048e+00 -8.08406472e-01
2.23061904e-01 -7.22902060e-01 7.34674335e-02 -8.43355119e-01
-9.86588717e-01 6.05028927e-01 6.03511333e-01 2.40611210e-01
8.64254355e-01 -3.05942208e-01 5.65820813e-01 4.51988220e-01
3.02200735e-01 -1.54824650e+00 5.97701013e-01 6.86383992e-02
1.70218766e-01 -1.73524833e+00 1.28809765e-01 -5.03616869e-01
-8.40072274e-01 1.10689497e+00 1.29763865e+00 -8.15002248e-02
8.42519879e-01 -2.03991458e-01 1.22612648e-01 1.92620263e-01
-7.71952033e-01 -2.43060306e-01 2.02308387e-01 3.33503932e-01
5.71239233e-01 -1.92378182e-02 -6.95460439e-01 4.75984097e-01
3.77597690e-01 3.34516883e-01 1.70509532e-01 1.27240860e+00
-6.75871730e-01 -1.44913960e+00 -4.49260205e-01 1.07718718e+00
-2.16909871e-01 -8.31959248e-02 -2.56516606e-01 4.89266127e-01
1.97031632e-01 1.09739411e+00 -9.24672037e-02 -5.95880508e-01
-9.82135907e-02 2.95915961e-01 1.10918000e-01 -9.82093871e-01
-5.87692797e-01 5.03943443e-01 3.34905391e-03 -1.13993697e-01
-7.24786043e-01 -5.52794397e-01 -1.10220063e+00 3.20701391e-01
-1.28341997e+00 7.88066506e-01 3.92846078e-01 1.10878611e+00
2.29175568e-01 3.40097100e-01 1.33151698e+00 -2.21226037e-01
-1.17325330e+00 -1.19957948e+00 -7.78229177e-01 4.53367800e-01
8.46345872e-02 -6.09816730e-01 -4.12102073e-01 3.51184234e-02]
|
[9.483733177185059, 3.9742233753204346]
|
8f196365-9a9a-44bd-b235-7488f9a80dcd
|
egolocate-real-time-motion-capture
|
2305.01599
| null |
https://arxiv.org/abs/2305.01599v1
|
https://arxiv.org/pdf/2305.01599v1.pdf
|
EgoLocate: Real-time Motion Capture, Localization, and Mapping with Sparse Body-mounted Sensors
|
Human and environment sensing are two important topics in Computer Vision and Graphics. Human motion is often captured by inertial sensors, while the environment is mostly reconstructed using cameras. We integrate the two techniques together in EgoLocate, a system that simultaneously performs human motion capture (mocap), localization, and mapping in real time from sparse body-mounted sensors, including 6 inertial measurement units (IMUs) and a monocular phone camera. On one hand, inertial mocap suffers from large translation drift due to the lack of the global positioning signal. EgoLocate leverages image-based simultaneous localization and mapping (SLAM) techniques to locate the human in the reconstructed scene. On the other hand, SLAM often fails when the visual feature is poor. EgoLocate involves inertial mocap to provide a strong prior for the camera motion. Experiments show that localization, a key challenge for both two fields, is largely improved by our technique, compared with the state of the art of the two fields. Our codes are available for research at https://xinyu-yi.github.io/EgoLocate/.
|
['Feng Xu', 'Christian Theobalt', 'Shaohua Pan', 'Vladislav Golyanik', 'Marc Habermann', 'Yuxiao Zhou', 'Xinyu Yi']
|
2023-05-02
| null | null | null | null |
['simultaneous-localization-and-mapping']
|
['computer-vision']
|
[-1.97741494e-01 -4.93154585e-01 -2.27362394e-01 4.84650396e-02
-5.23963511e-01 -5.78485608e-01 5.51676810e-01 -3.84226650e-01
-4.86974716e-01 5.33279836e-01 2.42343187e-01 2.50453442e-01
4.66746062e-01 -4.13822442e-01 -7.12086618e-01 -5.41853487e-01
3.23273450e-01 1.93025902e-01 1.34639710e-01 9.87481400e-02
7.76564702e-02 1.39124840e-01 -1.17787516e+00 -4.79967594e-01
5.38882077e-01 1.01908791e+00 3.67442757e-01 8.31794381e-01
3.70977223e-01 7.81585395e-01 -7.04991221e-02 2.02678174e-01
4.41472858e-01 -2.33521864e-01 -2.20705718e-01 3.49642895e-02
6.67825222e-01 -4.19961780e-01 -6.96099579e-01 1.14814067e+00
7.01852918e-01 4.86555994e-02 -9.65626314e-02 -1.30567479e+00
-2.30791226e-01 -1.17501870e-01 -6.85268521e-01 -3.98510508e-02
1.01342261e+00 2.20536381e-01 3.81989628e-01 -1.14367998e+00
6.46989405e-01 9.26356912e-01 8.86121809e-01 3.29418600e-01
-6.83495641e-01 -4.34382617e-01 -1.73659414e-01 -2.37973928e-02
-1.75521624e+00 -7.94898331e-01 6.75206304e-01 -5.17076612e-01
5.30441165e-01 1.66555271e-01 7.18396962e-01 1.18091369e+00
2.52847165e-01 5.07946372e-01 6.92336202e-01 -2.55609781e-01
1.97986037e-01 -4.36845124e-02 -2.10506275e-01 8.49888384e-01
4.89572793e-01 3.67999598e-02 -1.06952894e+00 4.84494902e-02
1.18188775e+00 5.21971762e-01 -6.34297788e-01 -7.79425740e-01
-1.87136376e+00 2.59399205e-01 6.07612610e-01 6.70449659e-02
-5.12990415e-01 4.65762794e-01 -8.65661055e-02 -9.41391811e-02
1.91943049e-01 1.56025201e-01 -9.42505747e-02 -6.10724568e-01
-8.26788008e-01 -6.57543615e-02 5.48017681e-01 1.38117790e+00
9.45014834e-01 -1.88978221e-02 3.50926250e-01 2.25581303e-01
4.57884461e-01 1.12796986e+00 6.44705892e-01 -1.12205076e+00
6.90854192e-01 3.80317241e-01 6.40979409e-01 -1.35483384e+00
-4.86719131e-01 -2.56065518e-01 -6.88883603e-01 -7.19373375e-02
4.19565231e-01 -5.39444625e-01 -5.73828578e-01 1.66467929e+00
6.93710208e-01 7.49391675e-01 -2.47073606e-01 1.41626143e+00
5.40788651e-01 2.89051116e-01 -4.48767662e-01 6.70127422e-02
1.27929902e+00 -1.11831367e+00 -6.98347926e-01 -9.03212428e-01
5.83701015e-01 -6.12135887e-01 8.01556706e-01 4.89839204e-02
-7.84484029e-01 -6.09763801e-01 -1.19850481e+00 -1.13208689e-01
-3.41848843e-02 1.79026246e-01 5.04046738e-01 4.97057974e-01
-1.11299169e+00 3.49511802e-02 -1.42741132e+00 -7.68773377e-01
-1.60983130e-01 1.57943293e-01 -6.60358071e-01 -3.85133207e-01
-9.86755073e-01 7.82772541e-01 -5.84285855e-02 2.17248201e-01
-4.59066689e-01 -4.06802267e-01 -1.19967151e+00 -4.25513476e-01
4.02101308e-01 -1.16233742e+00 1.15926826e+00 -5.26451409e-01
-1.52241647e+00 6.00640655e-01 -5.12477100e-01 -3.08967918e-01
7.07430422e-01 -8.13541710e-01 -1.71917692e-01 2.02211410e-01
3.87299597e-01 3.84357423e-01 6.89989924e-01 -9.64763463e-01
-5.90535581e-01 -7.02395558e-01 -2.71275818e-01 5.68041146e-01
4.52422164e-02 -4.49492306e-01 -1.07524514e+00 -2.88702905e-01
7.67365336e-01 -1.35446608e+00 -2.70081192e-01 -2.65908148e-02
-3.64930630e-01 6.61822855e-01 6.13454700e-01 -5.05753636e-01
1.05026615e+00 -2.15110087e+00 9.54078883e-02 3.42412405e-02
2.38677412e-01 -2.14233801e-01 3.76654357e-01 2.32742101e-01
3.88208538e-01 -4.19317663e-01 1.13023639e-01 -8.37152064e-01
-1.25926539e-01 7.09766671e-02 -1.52275681e-01 1.11857140e+00
-6.70487344e-01 1.03612995e+00 -1.16167772e+00 -3.24017942e-01
5.67058444e-01 6.46212280e-01 -2.22242787e-01 5.31550460e-02
4.34339374e-01 1.03200221e+00 -4.99066561e-01 8.93576086e-01
6.11081719e-01 -3.43800217e-01 -2.35104728e-02 1.27373850e-02
-2.33117506e-01 2.74272531e-01 -1.43793023e+00 2.57607985e+00
-3.43341023e-01 5.67989647e-01 3.43174368e-01 -2.40069360e-01
5.28733552e-01 3.61373186e-01 5.57787240e-01 -6.65317833e-01
1.61268130e-01 1.96286663e-01 -6.49097025e-01 -2.08202496e-01
6.49242938e-01 2.27093086e-01 -1.68982178e-01 1.28370032e-01
-2.90040702e-01 1.15191117e-01 -4.07214701e-01 1.09915711e-01
1.31769502e+00 5.03649712e-01 6.18424833e-01 1.80835590e-01
3.32836092e-01 8.05515125e-02 1.01932919e+00 9.32017744e-01
-4.46402878e-01 9.07796025e-01 -3.49324644e-01 -3.43328953e-01
-7.33971477e-01 -1.14060533e+00 2.35991508e-01 5.33083498e-01
8.24155152e-01 -6.60936773e-01 -6.78855598e-01 -1.87893152e-01
6.74061896e-03 4.79409806e-02 -3.23463291e-01 1.38747677e-01
-6.79479003e-01 -3.13562840e-01 2.68970430e-01 5.71239531e-01
7.68917620e-01 -3.62555087e-01 -1.11654055e+00 6.79987594e-02
-5.11383593e-01 -1.45833480e+00 -6.96490288e-01 -4.28555161e-01
-7.63451219e-01 -1.03333485e+00 -8.40768278e-01 -2.97837585e-01
7.20799506e-01 1.09569561e+00 7.44565308e-01 -7.62091801e-02
-1.04125896e-02 8.38783562e-01 -2.73958772e-01 -1.33525237e-01
5.43550372e-01 -1.51563033e-01 7.21611023e-01 2.35980406e-01
2.79668182e-01 -6.90232813e-01 -9.60700691e-01 5.32659888e-01
-1.77861080e-01 2.99197882e-01 3.98574442e-01 4.74848688e-01
7.41384983e-01 -3.38019103e-01 -2.72812337e-01 -5.15006721e-01
-2.37734213e-01 -5.43207824e-01 -7.39848137e-01 -2.86476523e-01
-3.98437530e-01 -5.40087938e-01 1.32229820e-01 -3.14251482e-01
-6.51427150e-01 6.71965897e-01 3.43213886e-01 -7.41751373e-01
-2.31696293e-01 3.78499359e-01 -2.82364577e-01 -2.47660041e-01
6.62169337e-01 2.32011989e-01 -1.97130479e-02 -4.15549964e-01
3.21891010e-01 5.08487582e-01 1.14543509e+00 -1.11537702e-01
9.38820362e-01 1.02129126e+00 -3.55807580e-02 -8.95814598e-01
-7.18307078e-01 -9.20228422e-01 -8.16686749e-01 -2.37619877e-01
8.48990142e-01 -1.57568729e+00 -5.11527956e-01 6.04684055e-01
-9.69932556e-01 -2.28234187e-01 1.19427079e-02 1.02903616e+00
-5.59562325e-01 5.42495370e-01 -3.67246628e-01 -7.59788156e-01
-1.34195924e-01 -1.03613067e+00 1.33090401e+00 3.96937102e-01
-1.67971566e-01 -7.97649384e-01 3.64768922e-01 4.03287679e-01
6.03785776e-02 5.31159699e-01 -5.07587492e-01 1.82783768e-01
-7.41010129e-01 -6.79447472e-01 1.28930911e-01 -3.59717846e-01
2.40045071e-01 -5.79479992e-01 -9.63554740e-01 -4.75387245e-01
2.72373945e-01 2.24428058e-01 5.19515932e-01 5.46020508e-01
1.78774744e-01 -7.15031177e-02 -6.96235001e-01 1.12952149e+00
1.30872238e+00 -9.35858488e-02 4.65116292e-01 6.17570758e-01
1.30881011e+00 1.28436044e-01 7.64954686e-01 5.28298795e-01
9.89242673e-01 1.11687064e+00 4.27164376e-01 -2.78134141e-02
-4.18271683e-02 -5.87788165e-01 5.90187550e-01 8.66217315e-01
-1.12674423e-01 6.41402006e-02 -1.20064044e+00 4.10560369e-01
-2.25875473e+00 -7.79295623e-01 -2.38662362e-01 2.52127504e+00
3.11921418e-01 -1.28904670e-01 2.05442458e-02 -1.83229998e-01
7.17122734e-01 2.60027409e-01 -5.33371508e-01 5.79523265e-01
-5.42659983e-02 -5.79386055e-01 1.00451255e+00 8.35702956e-01
-1.11638927e+00 1.00851655e+00 5.14286470e+00 3.07346433e-02
-1.17329669e+00 3.89172465e-01 -2.47942299e-01 -4.56732959e-01
3.19002628e-01 3.47189188e-01 -8.26512277e-01 7.40449607e-01
7.69196212e-01 4.48655598e-02 3.97930294e-01 1.17045677e+00
1.50465831e-01 -5.19209743e-01 -9.50654864e-01 1.64384234e+00
2.86841720e-01 -1.06595790e+00 -8.29680324e-01 3.63416523e-01
7.20813274e-01 5.73209226e-01 -3.60454135e-02 -7.02027082e-02
5.80033325e-02 -5.31023502e-01 9.23512399e-01 6.84500158e-01
7.05759466e-01 -4.91408288e-01 6.70718729e-01 6.80071831e-01
-1.46409142e+00 2.11257145e-01 -3.25160354e-01 -5.00048339e-01
5.52863121e-01 7.46468425e-01 -6.70612693e-01 5.54416299e-01
7.52032399e-01 9.19841349e-01 -5.90701222e-01 1.00700080e+00
-4.81942326e-01 2.67261535e-01 -6.08381093e-01 4.58931983e-01
-8.92148092e-02 -3.81111920e-01 8.45372856e-01 8.32104802e-01
5.80789685e-01 9.25493911e-02 6.55825555e-01 4.73459542e-01
1.48518965e-01 -4.01066720e-01 -9.82174993e-01 4.69869196e-01
7.11996496e-01 1.11347544e+00 -4.73296374e-01 -3.29621732e-01
-4.28187937e-01 1.35733974e+00 -7.66801462e-02 3.74205530e-01
-8.09144497e-01 -2.14847282e-01 8.73777747e-01 2.88229644e-01
5.44418097e-02 -9.81530070e-01 -3.48049700e-01 -1.86318326e+00
3.19575101e-01 -4.29314941e-01 1.07537210e-01 -9.49033082e-01
-5.31853497e-01 3.31650585e-01 -3.74098301e-01 -1.74109077e+00
-5.49572587e-01 -1.52643695e-01 -3.27735692e-01 8.57713163e-01
-1.29937959e+00 -1.16614616e+00 -9.51785088e-01 8.51228058e-01
2.48649403e-01 1.33812025e-01 5.26972651e-01 3.29145819e-01
-5.84193647e-01 2.37686858e-01 6.94325119e-02 2.13658065e-01
9.84596372e-01 -1.17450368e+00 7.52231300e-01 1.40342462e+00
3.55405241e-01 8.42505038e-01 6.48640335e-01 -8.78621757e-01
-2.08797312e+00 -8.80665600e-01 8.79681230e-01 -1.05220377e+00
3.82490844e-01 -6.02273941e-01 -4.03349549e-01 1.01907992e+00
-2.53282815e-01 4.23961461e-01 3.44869465e-01 -1.80687875e-01
-3.57128009e-02 8.56200084e-02 -7.60485113e-01 5.84001422e-01
1.14092433e+00 -7.05516279e-01 -1.97179213e-01 2.19179615e-01
5.27714550e-01 -1.18319154e+00 -4.77753550e-01 6.82867691e-02
7.90397525e-01 -1.09106791e+00 1.12275422e+00 2.25493163e-01
-2.70515263e-01 -8.31102490e-01 -4.72810179e-01 -1.08463824e+00
-3.61656427e-01 -8.19219589e-01 -5.95699370e-01 1.01806831e+00
-2.68981397e-01 -7.93556988e-01 9.19000208e-01 6.48559928e-01
2.35685796e-01 -1.06123701e-01 -1.08045065e+00 -7.67442584e-01
-7.96928704e-01 -6.21510327e-01 5.05559981e-01 8.59738052e-01
-3.35878655e-02 4.32795793e-01 -8.97555053e-01 6.81393921e-01
8.24344456e-01 -1.59756355e-02 1.38241446e+00 -8.96629453e-01
-1.65535554e-01 1.95627183e-01 -8.65306795e-01 -1.54943788e+00
-2.34726742e-01 -3.57247502e-01 2.58737415e-01 -1.38798201e+00
-3.07940185e-01 -2.22850487e-01 -3.14192362e-02 9.37962681e-02
-7.56421238e-02 4.66957420e-01 3.05897176e-01 7.32669294e-01
-8.29825580e-01 4.98628080e-01 7.40038037e-01 3.06700081e-01
-2.84119517e-01 -2.81737298e-02 -5.20922542e-01 1.11489022e+00
5.90360224e-01 -3.89592677e-01 -2.56693155e-01 -9.24141109e-01
3.65620285e-01 3.21310282e-01 6.82028353e-01 -1.60949075e+00
1.01837420e+00 -1.42399326e-01 6.50175273e-01 -5.57784617e-01
6.84710860e-01 -8.79414678e-01 4.83520031e-01 4.14159626e-01
3.88997793e-01 4.78331149e-01 -1.87092170e-01 8.67074430e-01
-1.13264151e-01 3.94121975e-01 4.18449134e-01 -2.52914906e-01
-1.00835764e+00 5.08347034e-01 4.67280811e-03 -4.20229621e-02
7.11864114e-01 -4.04216468e-01 -2.43395939e-01 -7.55991459e-01
-3.66178602e-01 2.62543947e-01 1.18278861e+00 4.79098618e-01
5.05158782e-01 -1.65769219e+00 -2.96668261e-01 2.44146585e-01
2.20472470e-01 3.05437148e-01 2.36249253e-01 1.42181039e+00
-6.91794813e-01 5.06944001e-01 1.62358321e-02 -1.01097357e+00
-9.27264214e-01 3.55535269e-01 2.11638853e-01 2.39454851e-01
-9.12417293e-01 5.99453449e-01 2.98723638e-01 -3.16581249e-01
1.01279564e-01 -1.09344229e-01 2.47214034e-01 -3.58344764e-01
8.64227891e-01 5.75669825e-01 -1.44196898e-01 -1.21621251e+00
-7.33032346e-01 9.88230169e-01 5.21610439e-01 -5.02080500e-01
8.44501078e-01 -1.08552492e+00 1.53106049e-01 5.81131756e-01
9.71102834e-01 5.11458158e-01 -1.56826532e+00 -4.57240582e-01
-1.11785889e-01 -8.90861034e-01 1.72465950e-01 -3.20940226e-01
-7.79982507e-01 6.50893390e-01 6.64046347e-01 -3.97745460e-01
9.16227996e-01 -3.88059825e-01 9.38002288e-01 3.49310338e-01
1.07155383e+00 -9.28903699e-01 -2.72315949e-01 5.70306718e-01
5.23690522e-01 -1.46597779e+00 1.77834600e-01 -3.44446331e-01
-6.77922845e-01 5.48583329e-01 5.71084082e-01 -9.44746062e-02
4.57506388e-01 2.07977206e-01 3.15734655e-01 -3.11912913e-02
-2.98225135e-01 -3.02160412e-01 -2.01459485e-03 6.99881852e-01
3.61796916e-02 8.83938279e-03 3.18033367e-01 4.29062217e-01
-2.84161240e-01 -3.67206824e-03 4.36765403e-01 1.34579933e+00
-2.80250996e-01 -6.10833824e-01 -7.89301336e-01 -3.21793884e-01
-1.50719613e-01 8.69423300e-02 -2.78761178e-01 6.04094207e-01
2.86890566e-03 9.79876101e-01 -9.89831835e-02 -6.62571490e-01
2.87402183e-01 -2.47215942e-01 2.60244608e-01 -5.01780331e-01
-1.69626340e-01 2.50155926e-01 -1.77272409e-01 -1.39084423e+00
-1.76007882e-01 -7.29714692e-01 -1.09568465e+00 -3.51148099e-01
-1.96551651e-01 -1.36413842e-01 9.78475273e-01 7.21210480e-01
7.53209651e-01 1.36532277e-01 5.22203028e-01 -1.39612877e+00
-4.36317958e-02 -6.32184982e-01 -6.38597429e-01 1.68342099e-01
7.96958029e-01 -8.41699600e-01 -1.93404153e-01 1.03278227e-01]
|
[7.55694055557251, -2.081800699234009]
|
8215d4af-f709-439f-8af5-6cbbd6ecde56
|
road-detection-via-a-dual-task-network-based
|
2208.08116
| null |
https://arxiv.org/abs/2208.08116v1
|
https://arxiv.org/pdf/2208.08116v1.pdf
|
Road detection via a dual-task network based on cross-layer graph fusion modules
|
Road detection based on remote sensing images is of great significance to intelligent traffic management. The performances of the mainstream road detection methods are mainly determined by their extracted features, whose richness and robustness can be enhanced by fusing features of different types and cross-layer connections. However, the features in the existing mainstream model frameworks are often similar in the same layer by the single-task training, and the traditional cross-layer fusion ways are too simple to obtain an efficient effect, so more complex fusion ways besides concatenation and addition deserve to be explored. Aiming at the above defects, we propose a dual-task network (DTnet) for road detection and cross-layer graph fusion module (CGM): the DTnet consists of two parallel branches for road area and edge detection, respectively, while enhancing the feature diversity by fusing features between two branches through our designed feature bridge modules (FBM). The CGM improves the cross-layer fusion effect by a complex feature stream graph, and four graph patterns are evaluated. Experimental results on three public datasets demonstrate that our method effectively improves the final detection result.
|
['Xueyun Chen', 'Hongkun Liu', 'Wurui Shi', 'Zican Hu']
|
2022-08-17
| null | null | null | null |
['edge-detection']
|
['computer-vision']
|
[ 8.01672265e-02 -2.64076054e-01 -4.70348522e-02 -2.96715975e-01
-1.25157356e-01 1.14046670e-01 8.37655723e-01 -1.17660061e-01
-3.17846566e-01 5.15368879e-01 1.99411381e-02 -3.04035813e-01
-3.65045786e-01 -1.45408297e+00 -4.09966141e-01 -8.69549453e-01
-7.83023983e-02 -2.53033668e-01 8.20981383e-01 -5.64367473e-01
8.69475752e-02 4.65088218e-01 -1.85733294e+00 2.84186080e-02
1.34419858e+00 1.13525677e+00 2.20311567e-01 2.46681586e-01
-5.75995266e-01 4.61720139e-01 -1.87524378e-01 -3.84941965e-01
2.78916329e-01 -5.47422096e-03 -2.03805164e-01 -1.51130219e-03
1.77485198e-01 -2.48911530e-01 -6.67575717e-01 1.14700937e+00
7.18956113e-01 -7.22863078e-02 4.60681558e-01 -1.38395941e+00
-4.24602360e-01 6.91585779e-01 -7.63773024e-01 2.60861404e-02
-1.29100829e-01 3.08697879e-01 8.51800799e-01 -9.16099370e-01
9.09982249e-02 1.45212269e+00 8.07347298e-01 -2.01529060e-02
-7.31129408e-01 -9.99840617e-01 6.09736145e-01 4.47201133e-01
-1.69942188e+00 -3.39383811e-01 9.63412166e-01 -3.41543347e-01
4.85680610e-01 2.76771605e-01 8.07428837e-01 5.18476367e-01
-6.22432940e-02 8.92315328e-01 8.76564145e-01 1.76828325e-01
-3.92918468e-01 1.34362116e-01 6.25455379e-02 8.46161723e-01
5.36468804e-01 1.20672313e-02 -1.03527471e-01 3.93308401e-01
7.87136495e-01 3.42011273e-01 -2.27897346e-01 -1.40282258e-01
-1.13948500e+00 7.06744015e-01 1.12122738e+00 3.12819004e-01
-2.89463401e-01 -4.65414673e-02 3.99907619e-01 1.69523373e-01
5.29403329e-01 -1.79218709e-01 -1.64171025e-01 4.12608057e-01
-7.50245512e-01 1.09879449e-01 3.24882597e-01 8.34950328e-01
1.28472328e+00 3.19960080e-02 -4.16568667e-01 9.39619660e-01
4.13127720e-01 6.24622226e-01 1.11753635e-01 -3.31424743e-01
6.95104480e-01 1.23215628e+00 -4.16091263e-01 -1.52114952e+00
-6.53062463e-01 -7.93770671e-01 -1.29137611e+00 1.77352667e-01
-9.89770633e-04 -3.88788909e-01 -8.14621329e-01 1.32074952e+00
4.24960792e-01 4.23583776e-01 -1.11912467e-01 8.74829233e-01
1.37461543e+00 6.53708398e-01 1.53385773e-01 7.26880180e-03
1.35810590e+00 -9.41120923e-01 -6.19901359e-01 -2.76320428e-01
5.57236075e-01 -6.37677372e-01 9.09630477e-01 8.32867101e-02
-5.34477592e-01 -8.76652896e-01 -1.20819438e+00 -1.69334169e-02
-7.92347729e-01 3.71593416e-01 8.05313587e-01 4.21017915e-01
-6.75416946e-01 3.23018223e-01 -1.41219586e-01 -2.57245481e-01
7.17816770e-01 1.88717961e-01 -3.27507198e-01 -1.46182597e-01
-1.38053691e+00 5.16524613e-01 5.33577740e-01 8.75091612e-01
-3.73708785e-01 -5.27518809e-01 -7.76567936e-01 1.56573549e-01
5.77472091e-01 -6.04047537e-01 5.61421156e-01 -4.04801309e-01
-1.12427294e+00 3.59738946e-01 1.15348682e-01 -3.39225456e-02
7.43400514e-01 -1.96469828e-01 -9.51743901e-01 -4.48691659e-02
1.08235046e-01 4.61317301e-01 8.04579258e-01 -1.05240738e+00
-1.21894479e+00 -1.75635532e-01 3.36621255e-02 2.15662360e-01
-4.97801304e-01 -2.55361080e-01 -4.72546697e-01 -4.26544040e-01
1.86374202e-01 -3.38180572e-01 -3.70266676e-01 -5.79689294e-02
-5.23228228e-01 -3.23378533e-01 1.15111327e+00 -4.43921715e-01
1.72428906e+00 -2.06119084e+00 -2.75506943e-01 4.38696355e-01
7.18416810e-01 7.35120833e-01 -1.52142107e-01 3.65863532e-01
8.29036385e-02 1.78838447e-01 -3.23204756e-01 -1.07964699e-03
-1.17930859e-01 7.67157003e-02 -3.70491855e-02 2.95206189e-01
5.77825963e-01 1.04999638e+00 -9.18716550e-01 -9.55790758e-01
5.48104227e-01 5.46410501e-01 -1.85112849e-01 4.28443542e-03
1.32901564e-01 2.59080440e-01 -9.51680660e-01 6.26961052e-01
1.05272090e+00 -9.94349942e-02 -2.94235677e-01 -4.70964432e-01
-4.61901635e-01 3.35302055e-02 -1.55459905e+00 1.21638751e+00
-3.32560956e-01 2.27135196e-01 4.70089400e-03 -9.88487422e-01
1.38387454e+00 -7.95848519e-02 4.11085367e-01 -9.13602889e-01
1.26991242e-01 2.66806394e-01 7.94162527e-02 -7.31047392e-01
3.48127842e-01 1.53847978e-01 5.82285738e-03 -1.43497080e-01
-3.74339312e-01 2.27740064e-01 1.92411035e-01 1.92157134e-01
7.32327700e-01 -2.74234172e-03 -9.99539793e-02 -1.41697362e-01
1.07955718e+00 -1.80112392e-01 7.63301313e-01 5.55227637e-01
-1.15966357e-01 3.79457265e-01 4.03608799e-01 -5.30761838e-01
-5.68271399e-01 -8.57321799e-01 -2.64309675e-01 7.50142038e-01
5.16194046e-01 -3.43208700e-01 -3.73872906e-01 -7.86913216e-01
1.39386088e-01 9.93944034e-02 -4.73325372e-01 -3.43283147e-01
-6.49774075e-01 -9.70322728e-01 6.09529257e-01 5.19927561e-01
1.37580395e+00 -9.85535502e-01 -1.61196694e-01 4.12419200e-01
-1.17990538e-01 -1.06389844e+00 -1.03163764e-01 -3.80275905e-01
-6.66335106e-01 -1.05061126e+00 -5.68689167e-01 -6.21206880e-01
4.46727455e-01 1.03582215e+00 7.79975057e-01 6.02920532e-01
-1.78121939e-01 -3.00925344e-01 -3.85325402e-01 -4.12457645e-01
9.05413181e-02 3.70516390e-01 -4.19643521e-01 6.00511789e-01
2.65834630e-01 -7.71150231e-01 -8.30392838e-01 4.45551783e-01
-7.66027451e-01 3.29853088e-01 1.03208601e+00 5.06773412e-01
1.63908541e-01 3.63698572e-01 8.24397802e-01 -8.35189223e-01
5.37084341e-01 -5.99934757e-01 -4.73972768e-01 2.36645088e-01
-6.78662360e-01 -2.65763789e-01 5.68048716e-01 -6.07419536e-02
-1.32791865e+00 -4.21088189e-02 -3.55447322e-01 -2.14180917e-01
-1.67468175e-01 6.10513806e-01 -6.12328410e-01 -1.98582053e-01
3.10928285e-01 3.56357932e-01 1.80060789e-02 -5.20779252e-01
5.78159094e-01 8.46705258e-01 4.21901256e-01 -2.13542834e-01
1.12938380e+00 4.32906568e-01 2.04067662e-01 -7.73573637e-01
-5.92090189e-01 -4.45927858e-01 -6.18589282e-01 -5.51235735e-01
7.83768475e-01 -1.06698966e+00 -8.33845496e-01 8.81390095e-01
-8.48034501e-01 1.86038315e-01 3.24070454e-02 4.66550022e-01
3.84517461e-02 4.07966644e-01 -1.93035156e-01 -8.46405327e-01
-2.63493687e-01 -1.06747925e+00 1.19449258e+00 6.72514915e-01
7.41483927e-01 -7.06868589e-01 -3.50131392e-01 1.53920576e-01
6.18488252e-01 2.47198507e-01 7.97958314e-01 -3.68313223e-01
-8.28486562e-01 -1.80139437e-01 -1.03613782e+00 3.70467633e-01
2.14817315e-01 2.31767669e-01 -1.04492366e+00 2.27719378e-02
-4.88551438e-01 1.93821222e-01 1.48059690e+00 2.31069207e-01
1.14023030e+00 -7.29330257e-02 -6.71079278e-01 6.48361087e-01
1.43949115e+00 -1.30474225e-01 7.58930087e-01 3.78159314e-01
1.28577268e+00 7.84206092e-01 6.00642979e-01 1.78618774e-01
7.36275077e-01 3.36646885e-01 6.72742903e-01 -6.09755278e-01
-4.22689825e-01 -2.13500738e-01 2.12146431e-01 8.03782642e-01
-4.09042209e-01 2.23228638e-03 -7.26917624e-01 4.52188730e-01
-2.05523729e+00 -1.12198198e+00 -7.45673358e-01 2.04025245e+00
3.16656321e-01 3.82638842e-01 3.58729541e-01 3.94249231e-01
9.66774523e-01 4.41719353e-01 -4.67953533e-01 1.48458630e-01
-3.83588821e-01 -1.96323901e-01 6.48094118e-01 1.55218944e-01
-1.24986589e+00 8.12814713e-01 5.05506754e+00 1.28068244e+00
-1.18779850e+00 -1.63078196e-02 2.71432221e-01 3.72695655e-01
-3.46718490e-01 9.46303234e-02 -1.05721247e+00 4.53740537e-01
2.24288493e-01 2.99244560e-02 1.08154044e-01 5.30118227e-01
2.05987275e-01 1.22501917e-01 -2.81857282e-01 9.78209078e-01
-3.41032416e-01 -1.33724368e+00 2.02366650e-01 1.85501024e-01
4.67119366e-01 1.88087836e-01 -1.29921675e-01 6.18882000e-01
1.80043891e-01 -8.56366992e-01 4.17720646e-01 7.94602633e-01
6.81428015e-01 -7.56224930e-01 8.17550838e-01 3.95222336e-01
-2.08648133e+00 -3.68502349e-01 -5.12309551e-01 -1.07906446e-01
2.70476073e-01 1.00698674e+00 -3.86180013e-01 1.33920097e+00
6.73746943e-01 1.22408581e+00 -9.21779931e-01 1.30857956e+00
-1.69371501e-01 4.81757432e-01 -3.31846088e-01 -3.56517211e-02
4.40112054e-01 -5.46042860e-01 5.67713618e-01 1.35460901e+00
1.54090896e-01 -1.47988245e-01 4.79502857e-01 7.22193003e-01
6.56845123e-02 2.74859667e-01 -7.58610308e-01 4.04169053e-01
5.09260654e-01 1.84676731e+00 -6.26030743e-01 -3.93259615e-01
-7.99851060e-01 2.45611668e-01 2.56096184e-01 3.13960463e-01
-9.81078029e-01 -9.16503131e-01 5.96231520e-01 3.16936374e-01
3.65681201e-01 -1.36190906e-01 -2.44845033e-01 -1.11532271e+00
2.18125895e-01 -4.73463148e-01 4.47316259e-01 -4.81306672e-01
-1.40503681e+00 4.85819906e-01 -1.16581433e-01 -1.51596034e+00
5.61520755e-01 -3.86806279e-01 -9.75679219e-01 9.97778356e-01
-2.34421134e+00 -1.59005857e+00 -8.51414800e-01 7.32581317e-01
1.56802326e-01 -5.21313362e-02 6.22581244e-02 8.04497540e-01
-1.07695019e+00 4.75706100e-01 -3.31964642e-01 3.26822311e-01
3.27870250e-01 -8.80185664e-01 4.35707241e-01 1.02357829e+00
-3.00280482e-01 3.13110173e-01 -1.54561829e-02 -5.92821896e-01
-1.24825025e+00 -1.57000852e+00 6.70149744e-01 1.32389411e-01
5.48391819e-01 -1.20661527e-01 -1.11392701e+00 2.56428421e-01
-1.85878590e-01 1.40270159e-01 1.52615637e-01 9.83270928e-02
-3.07028055e-01 -7.10948288e-01 -9.12524283e-01 5.91789961e-01
1.48839319e+00 -2.94383615e-01 -3.37896645e-01 2.69866079e-01
8.73114526e-01 1.57689735e-01 -8.01499605e-01 7.99734592e-01
5.19136965e-01 -1.04998672e+00 1.05310965e+00 -3.41287047e-01
2.72158712e-01 -9.78981555e-01 7.69360438e-02 -9.70030844e-01
-5.77285945e-01 -2.92735159e-01 2.10278288e-01 1.60539734e+00
2.58557737e-01 -1.02475405e+00 4.80349451e-01 -1.59218997e-01
-4.99013782e-01 -7.83334792e-01 -8.02891552e-01 -7.84612000e-01
-1.91331729e-01 -4.81851339e-01 1.24476957e+00 7.92463660e-01
-4.89178658e-01 7.25454450e-01 -2.91785270e-01 2.91874707e-01
4.83890593e-01 1.93467304e-01 1.07517147e+00 -1.91564953e+00
2.78656572e-01 -1.07226253e+00 -4.63020712e-01 -1.09457481e+00
-3.01826507e-01 -9.35947835e-01 -5.14336303e-02 -1.88431180e+00
-7.54453689e-02 -7.96177208e-01 -2.89386332e-01 5.28841913e-01
-5.55349469e-01 6.68133721e-02 1.01915471e-01 2.51741558e-01
-4.13098186e-01 8.85590672e-01 1.62026250e+00 -2.10498944e-01
-8.21673796e-02 3.60154323e-02 -8.95573318e-01 7.87638366e-01
7.63381481e-01 -1.91329658e-01 -4.43306684e-01 -4.54454333e-01
3.03306401e-01 -3.22809756e-01 5.76845050e-01 -1.32628441e+00
5.22422731e-01 -7.31496960e-02 3.22325140e-01 -9.14787948e-01
-3.02168122e-03 -8.49896073e-01 -2.44502500e-02 4.30824518e-01
3.82191360e-01 -2.47144327e-01 9.64921489e-02 6.51575327e-01
-3.26519787e-01 3.17710847e-01 5.18506408e-01 6.06568418e-02
-1.15961683e+00 6.67508960e-01 1.36516290e-02 -2.51779288e-01
9.95065391e-01 -5.84008157e-01 -5.16161323e-01 1.90653466e-02
-3.79963905e-01 7.26855338e-01 -4.64550182e-02 6.23163521e-01
5.37163615e-01 -1.52402186e+00 -1.03742063e+00 5.26130974e-01
3.29546750e-01 3.48726541e-01 5.15257120e-01 1.01476991e+00
-2.33852431e-01 2.37326939e-02 -2.66659826e-01 -7.59678066e-01
-7.93457210e-01 3.42522979e-01 3.55924934e-01 -3.10682118e-01
-7.06658185e-01 6.15181267e-01 2.81503052e-01 -5.85388601e-01
-1.52568251e-01 -6.18114062e-02 -8.60624433e-01 4.71965373e-01
4.30455059e-01 7.04489827e-01 1.83011994e-01 -7.75255740e-01
-3.30570638e-01 9.75232303e-01 5.71590550e-02 4.10960257e-01
1.33019531e+00 -3.12182248e-01 -8.67747813e-02 1.26995474e-01
9.31476116e-01 1.51869643e-03 -1.16936624e+00 -6.38533115e-01
-4.56371844e-01 -4.32252765e-01 2.72061050e-01 -3.96925420e-01
-1.59555864e+00 1.09478045e+00 4.78395939e-01 4.94094998e-01
1.36408424e+00 -2.99550772e-01 1.04827440e+00 2.69105166e-01
1.80177659e-01 -8.35753143e-01 -2.64681399e-01 3.54519427e-01
5.73775291e-01 -1.18850589e+00 1.17322686e-03 -9.07413363e-01
-2.87319630e-01 1.06950557e+00 8.11141789e-01 -1.83398545e-01
9.92771327e-01 4.15416434e-02 -1.13390602e-01 -4.79418963e-01
-2.59477198e-01 -8.66965830e-01 4.56746936e-01 5.27877152e-01
2.50835456e-02 2.03520238e-01 -5.48378825e-01 5.56936681e-01
4.80234288e-02 -5.85699826e-02 9.40725356e-02 6.74915731e-01
-7.84003139e-01 -9.10693109e-01 -1.02273524e-01 6.76078260e-01
1.13116823e-01 -1.50849774e-01 -6.08448274e-02 8.82648468e-01
6.29084289e-01 1.10490716e+00 8.18677992e-02 -9.49577749e-01
6.61078513e-01 -3.74891788e-01 -1.26445040e-01 -2.79229522e-01
-5.79382598e-01 -1.25833541e-01 1.42890394e-01 -5.01324654e-01
-5.76731026e-01 -4.92195159e-01 -1.09142709e+00 -4.73723412e-01
-6.10143304e-01 6.71050139e-03 3.64855230e-01 1.02152014e+00
4.93507564e-01 7.82696366e-01 1.03657103e+00 -6.09965801e-01
-1.43725574e-01 -9.73408639e-01 -6.02059007e-01 9.27144755e-03
2.50585914e-01 -1.01148880e+00 -1.57008007e-01 -3.58476818e-01]
|
[9.270235061645508, -0.862943172454834]
|
3ea754d8-fec4-44be-af7e-e2619f2ea731
|
multi-channel-and-multi-microphone-acoustic
|
2103.02552
| null |
https://arxiv.org/abs/2103.02552v1
|
https://arxiv.org/pdf/2103.02552v1.pdf
|
Multi-Channel and Multi-Microphone Acoustic Echo Cancellation Using A Deep Learning Based Approach
|
Building on the deep learning based acoustic echo cancellation (AEC) in the single-loudspeaker (single-channel) and single-microphone setup, this paper investigates multi-channel AEC (MCAEC) and multi-microphone AEC (MMAEC). We train a deep neural network (DNN) to predict the near-end speech from microphone signals with far-end signals used as additional information. We find that the deep learning approach avoids the non-uniqueness problem in traditional MCAEC algorithms. For the AEC setup with multiple microphones, rather than employing AEC for each microphone, a single DNN is trained to achieve echo removal for all microphones. Also, combining deep learning based AEC with deep learning based beamforming further improves the system performance. Experimental results show the effectiveness of both bidirectional long short-term memory (BLSTM) and convolutional recurrent network (CRN) based methods for MCAEC and MMAEC. Furthermore, deep learning based methods are capable of removing echo and noise simultaneously and work well in the presence of nonlinear distortions.
|
['DeLiang Wang', 'Hao Zhang']
|
2021-03-03
| null | null | null | null |
['acoustic-echo-cancellation', 'acoustic-echo-cancellation']
|
['medical', 'speech']
|
[ 6.57550320e-02 -4.12920266e-01 9.48502541e-01 -1.43537238e-01
-1.23420084e+00 -4.81476486e-01 3.22572052e-01 -5.12134492e-01
-5.44430733e-01 1.77661657e-01 7.35833764e-01 -5.67365587e-01
-1.42036565e-02 -2.13088214e-01 -7.72473872e-01 -8.24030161e-01
-1.26401454e-01 -3.38540852e-01 -5.71824573e-02 -4.29560281e-02
-1.21986195e-01 5.18209040e-01 -1.27056968e+00 5.74444890e-01
3.76323074e-01 9.69503403e-01 3.70307773e-01 1.41939914e+00
-9.47624668e-02 8.25836122e-01 -7.92763531e-01 -3.56239438e-01
2.56402433e-01 -4.15236503e-01 -8.46462771e-02 -5.14512956e-01
3.57122362e-01 -4.01684463e-01 -3.59739542e-01 7.34246373e-01
1.17153358e+00 2.59429693e-01 4.48045820e-01 -6.24193966e-01
-4.39854741e-01 6.15272164e-01 -3.42812210e-01 9.58507285e-02
9.99612361e-02 -9.56698880e-02 7.10115969e-01 -1.62567234e+00
-2.73519665e-01 1.24287593e+00 1.20559847e+00 6.94918990e-01
-5.46168268e-01 -9.18726802e-01 -4.69919760e-03 7.92027190e-02
-1.16502213e+00 -1.18391931e+00 8.67593884e-01 -1.09861657e-01
1.23478270e+00 3.77087086e-01 3.74445260e-01 1.27029610e+00
1.49141625e-01 7.67080069e-01 7.98518538e-01 -6.47686064e-01
1.79555833e-01 -2.19511464e-01 -2.37780824e-01 2.97734112e-01
-3.28027904e-01 4.57783490e-01 -5.72554767e-01 -1.34185259e-03
6.45235658e-01 8.97738710e-02 -4.60459232e-01 5.97301364e-01
-9.17242229e-01 5.74903309e-01 1.18354641e-01 4.27114367e-01
-5.65056026e-01 5.29195905e-01 3.85210514e-01 5.52234173e-01
3.47192943e-01 3.44626904e-01 -4.22698408e-01 -2.68872142e-01
-1.04480684e+00 -3.59468386e-02 1.03331816e+00 7.24919021e-01
1.44316912e-01 9.16024446e-01 8.74274373e-02 1.31400919e+00
4.49384362e-01 9.20107245e-01 7.08547175e-01 -1.12359786e+00
5.64217508e-01 -5.40131271e-01 -6.00995943e-02 -9.66644704e-01
-2.92551398e-01 -1.05880880e+00 -1.30188549e+00 2.43596315e-01
8.88098851e-02 -7.01563537e-01 -9.32835519e-01 1.65077460e+00
-1.48327693e-01 5.37268460e-01 3.49857867e-01 1.00904512e+00
7.89974570e-01 9.52279687e-01 -2.89649934e-01 -1.94071293e-01
6.09716058e-01 -1.33219898e+00 -1.09267175e+00 -3.08062583e-01
3.39501262e-01 -1.23191941e+00 4.81568307e-01 4.48926598e-01
-1.11415017e+00 -5.64871609e-01 -9.52803075e-01 4.47890796e-02
-3.65060300e-01 -1.68627381e-01 3.17676440e-02 7.94531882e-01
-1.38249552e+00 1.45539373e-01 -7.51217842e-01 4.45375770e-01
-2.47376770e-01 4.62616801e-01 -1.91126168e-01 -1.31268948e-01
-1.25335348e+00 7.00668097e-01 -4.54720587e-01 9.61203754e-01
-1.13316607e+00 -6.65497959e-01 -6.93770528e-01 3.83587271e-01
-8.30648243e-02 -3.29280823e-01 1.38279939e+00 -1.08348536e+00
-2.04546189e+00 -9.00320411e-02 -4.12358522e-01 -5.11521518e-01
1.36131600e-01 -6.86341703e-01 -9.22191262e-01 8.95129424e-03
-3.35312814e-01 4.99198079e-01 1.15306985e+00 -1.41036093e+00
-5.89389205e-01 4.22341600e-02 -3.25209945e-01 1.76536486e-01
-2.76216060e-01 1.05565868e-01 -1.74388826e-01 -8.65734279e-01
1.80357158e-01 -5.71446896e-01 -2.47708172e-01 -4.55081910e-01
-2.89616853e-01 2.11755782e-01 8.03687274e-01 -1.15242672e+00
1.33673096e+00 -2.25083494e+00 -1.39914146e-02 2.35646039e-01
1.02892704e-01 8.13953519e-01 -7.10376263e-01 6.16131842e-01
-1.69343159e-01 -1.11202098e-01 -2.15320706e-01 -8.32624376e-01
-8.55100602e-02 -2.03182120e-02 -2.99050719e-01 2.35878229e-01
4.15312424e-02 7.98917353e-01 -3.88778001e-01 2.20021158e-02
2.92541027e-01 8.42917740e-01 -5.51095963e-01 3.26275319e-01
5.01486063e-01 4.64250058e-01 2.44509637e-01 3.99094999e-01
9.33606327e-01 3.94221872e-01 -1.30074292e-01 1.82602867e-01
-3.65512609e-01 3.39340210e-01 -1.28323364e+00 1.48573518e+00
-1.25937712e+00 8.58447313e-01 1.03743625e+00 -8.11412275e-01
8.44527006e-01 9.72644210e-01 8.96778777e-02 -1.00658786e+00
-1.32418245e-01 7.95273781e-01 3.58021319e-01 -3.61058563e-01
1.51455045e-01 -3.91177595e-01 3.34796071e-01 4.39396083e-01
1.77129135e-01 2.66450465e-01 -5.33482075e-01 -1.83287844e-01
1.09572256e+00 -5.71264625e-01 -2.69240975e-01 -8.11038911e-02
6.47841752e-01 -9.26499486e-01 6.03544235e-01 9.47622955e-01
-1.23500220e-01 1.00084233e+00 -4.42951083e-01 -3.59079614e-02
-8.49746168e-01 -8.51143599e-01 3.11846524e-01 1.11062050e+00
-3.41223627e-01 -1.95385054e-01 -7.54220188e-01 1.06028900e-01
-1.97550163e-01 5.78081250e-01 5.73832057e-02 1.04307802e-03
-1.04014456e+00 -1.74043849e-01 8.65596831e-01 5.86654365e-01
3.82403791e-01 -9.38599586e-01 -2.14838490e-01 7.44757712e-01
-3.05287421e-01 -9.64347184e-01 -7.45862305e-01 3.77487332e-01
-5.56591511e-01 -4.62741226e-01 -1.24459052e+00 -1.09563792e+00
2.25321613e-02 4.20360118e-01 7.73016989e-01 -1.81845739e-01
1.08029321e-01 6.34883523e-01 -1.64658055e-01 -4.47479397e-01
-4.47254151e-01 -1.97432131e-01 2.72164673e-01 2.47857571e-01
1.67635933e-01 -8.66024911e-01 -8.33782017e-01 3.07490323e-02
-7.31108069e-01 -2.36617297e-01 7.56868303e-01 7.88756549e-01
1.37356296e-01 1.45422921e-01 9.44290578e-01 -2.63811350e-01
8.87312770e-01 -3.93454313e-01 -4.14481521e-01 -7.01668719e-03
-2.77334899e-01 -3.64473879e-01 8.62147808e-01 -5.33395529e-01
-1.29399967e+00 -1.40812263e-01 -1.03053200e+00 -5.66297948e-01
-1.17026687e-01 5.28156936e-01 -5.59688881e-02 3.79440933e-02
1.41300216e-01 2.89879262e-01 -2.52788305e-01 -1.02926064e+00
1.80624306e-01 1.12756157e+00 7.13210285e-01 -2.65425891e-02
5.82940698e-01 1.47029897e-03 -2.98661768e-01 -9.68342483e-01
-5.05840838e-01 -6.69919908e-01 -3.10292214e-01 -3.63179684e-01
7.93917239e-01 -1.28899109e+00 -7.83579409e-01 1.07206023e+00
-1.57356071e+00 -3.58636439e-01 1.44840524e-01 9.55496967e-01
3.70666906e-02 4.75520380e-02 -9.76304293e-01 -1.34812117e+00
-5.87288260e-01 -1.01314211e+00 9.60681617e-01 7.69665837e-02
1.26945287e-01 -1.29596114e+00 5.98030351e-02 1.33580938e-01
9.69580710e-01 -3.04911554e-01 5.77898562e-01 -5.77458203e-01
-4.73141491e-01 -3.73894572e-01 1.86801270e-01 9.83447969e-01
-7.37464726e-02 -4.90798324e-01 -1.40868890e+00 -3.99244010e-01
5.02185881e-01 1.20594382e-01 1.02003658e+00 7.51308918e-01
8.61533225e-01 -3.96431446e-01 2.27275178e-01 6.53076589e-01
1.35794938e+00 4.74311173e-01 5.34155488e-01 -1.20415524e-01
7.32025146e-01 4.06553119e-01 -3.56490046e-01 9.09812227e-02
2.21747130e-01 2.94518590e-01 3.50636125e-01 -4.58129108e-01
-5.59499085e-01 -1.63566381e-01 7.10342407e-01 1.98663342e+00
2.14678884e-01 -7.38338351e-01 -7.45050251e-01 7.68177867e-01
-1.43604386e+00 -7.36172855e-01 -3.34167480e-01 1.99551892e+00
5.44534087e-01 -2.02749789e-01 -4.23407584e-01 3.56992066e-01
7.53177881e-01 1.08849108e-01 -3.38612586e-01 -8.18602622e-01
-3.69018495e-01 6.50293589e-01 4.92294192e-01 1.13230383e+00
-8.84538710e-01 5.10007143e-01 6.32182455e+00 7.52971113e-01
-1.25126135e+00 4.73480284e-01 3.85922283e-01 -1.58008158e-01
-3.66900802e-01 -5.60049474e-01 -6.18328154e-01 3.00703704e-01
1.55091524e+00 6.64949596e-01 7.89201200e-01 2.78223574e-01
3.44400495e-01 1.72475874e-01 -9.44721699e-01 1.21927214e+00
5.49043640e-02 -1.02900445e+00 -2.98535079e-01 -1.26240805e-01
6.37521029e-01 3.26937973e-01 3.46983671e-01 4.02875632e-01
1.86292678e-01 -1.11779118e+00 7.87283242e-01 7.35958934e-01
7.68367946e-01 -6.75021410e-01 9.30620134e-01 5.02672851e-01
-1.12237239e+00 -4.09408331e-01 -2.32637390e-01 -2.42392197e-01
2.86830127e-01 7.12042153e-01 -3.23624790e-01 3.70254099e-01
7.96219230e-01 3.64890635e-01 2.23234501e-02 1.04395878e+00
-3.83581631e-02 1.08735466e+00 -5.21197379e-01 1.21272258e-01
3.79906058e-01 -1.24652348e-02 7.53747880e-01 1.74074388e+00
6.14556134e-01 1.08639300e-01 -3.67530137e-01 5.29460967e-01
-1.79574430e-01 -2.55416662e-01 -3.04736763e-01 1.55979156e-01
5.31593561e-01 9.97605026e-01 -3.62764858e-02 1.80394724e-02
-5.37224054e-01 1.09722269e+00 -2.11850479e-02 1.01151025e+00
-4.30975586e-01 -8.20516825e-01 7.18891025e-01 -3.77711684e-01
5.94639122e-01 -4.21926439e-01 -2.24337101e-01 -8.76545906e-01
1.32317424e-01 -8.99043083e-01 -2.85477281e-01 -6.36735678e-01
-1.33481812e+00 6.27280712e-01 -7.47107983e-01 -1.00422764e+00
-4.17160630e-01 -6.19238615e-01 -1.03229415e+00 1.32923007e+00
-1.96181774e+00 -7.57991195e-01 3.09528083e-01 7.27951050e-01
6.81751788e-01 -1.98679656e-01 1.02119994e+00 9.09761667e-01
-5.05793035e-01 6.92473233e-01 7.52516747e-01 3.82599235e-01
6.61641061e-01 -1.10737467e+00 4.81084108e-01 1.08415878e+00
7.25015923e-02 8.56342614e-01 4.63755697e-01 -1.70972422e-01
-1.47290683e+00 -1.16463244e+00 1.08231008e+00 -3.14174369e-02
3.17242712e-01 -8.30628395e-01 -8.86810005e-01 3.45705301e-01
5.26171565e-01 -5.82916662e-02 8.92769098e-01 -1.83277447e-02
-3.43551785e-01 -4.06477630e-01 -4.84438211e-01 3.80239785e-01
7.40287542e-01 -9.70409989e-01 -3.73467952e-01 4.16437276e-02
7.81690836e-01 -2.97677755e-01 -5.59167683e-01 2.59033442e-01
7.07878053e-01 -9.71005559e-01 1.01334965e+00 -2.36098304e-01
2.40560651e-01 -1.16987951e-01 -6.14711165e-01 -1.48820198e+00
-2.86303043e-01 -8.54483724e-01 -2.36947715e-01 1.37169480e+00
6.57194853e-01 -8.11704993e-01 2.28839800e-01 2.72771508e-01
-8.16312432e-01 -4.12579507e-01 -1.07900202e+00 -6.67624056e-01
4.38185871e-01 -8.06717336e-01 4.33769554e-01 3.92768711e-01
-5.87726653e-01 2.67895311e-01 -8.33824456e-01 5.42478025e-01
3.89919519e-01 -2.02335954e-01 3.40963423e-01 -8.56299818e-01
-4.47289497e-01 -3.03754687e-01 2.91914731e-01 -1.56800175e+00
1.30957723e-01 -5.46881914e-01 5.46583533e-01 -1.47602046e+00
-4.73499686e-01 -1.55442789e-01 -5.91074765e-01 3.86778787e-02
-3.84148061e-02 2.33730987e-01 3.33052635e-01 -7.77942836e-02
-3.70863229e-01 6.73385918e-01 1.00754738e+00 1.21564316e-02
-8.74425620e-02 3.74792725e-01 -4.26837862e-01 7.01412082e-01
5.19503653e-01 -5.18742204e-01 -4.48352434e-02 -1.09565282e+00
-2.17422172e-02 3.86243284e-01 3.51641804e-01 -1.12187564e+00
7.95835853e-01 6.70656025e-01 5.08775294e-01 -5.89534163e-01
6.72013521e-01 -1.01878417e+00 -3.26744579e-02 3.01149666e-01
-5.32096446e-01 -4.72679846e-02 2.62051553e-01 7.16216922e-01
-6.79948568e-01 -1.09605230e-01 5.76826096e-01 -1.43739805e-01
-2.64220923e-01 -1.19980440e-01 -1.04573333e+00 -2.17607558e-01
-3.47000360e-02 4.74765375e-02 6.48483336e-02 -9.94698644e-01
-7.57664680e-01 5.68965413e-02 -5.91220140e-01 4.18825477e-01
8.95324707e-01 -1.30788672e+00 -8.25037837e-01 5.69128633e-01
-6.27685010e-01 -1.96642265e-01 4.50110346e-01 7.60286927e-01
-2.32609317e-01 6.81950271e-01 4.29246455e-01 -3.99147213e-01
-1.02813876e+00 9.55294222e-02 8.07796240e-01 5.04645221e-02
-4.27960932e-01 1.28369355e+00 2.79685795e-01 -8.47023666e-01
8.16536307e-01 -2.04007924e-01 -5.51407486e-02 -4.27235872e-01
5.27030587e-01 5.68550408e-01 3.35232437e-01 -5.77418506e-01
-2.43464559e-01 5.82388520e-01 1.87632680e-01 -6.08975887e-01
1.40672493e+00 -4.68008697e-01 -9.74614471e-02 4.38878447e-01
1.60986257e+00 4.88964140e-01 -1.08538234e+00 -3.12725931e-01
-1.30578533e-01 2.94692628e-02 5.09165406e-01 -1.06148982e+00
-1.27505481e+00 1.50740910e+00 8.69511247e-01 -1.28959134e-01
1.41902757e+00 -6.02028012e-01 1.25943565e+00 4.30058748e-01
-2.78086126e-01 -1.02079141e+00 5.16115837e-02 9.22761083e-01
1.02928627e+00 -9.69989419e-01 -8.59395444e-01 1.15825109e-01
-2.45971441e-01 1.35680199e+00 2.90733188e-01 -7.79595673e-02
1.19402373e+00 8.23108435e-01 6.54610217e-01 1.42520681e-01
-7.66349733e-01 6.05639406e-02 1.90112069e-01 5.32803535e-01
4.92090642e-01 -8.21343064e-03 3.19560885e-01 8.50900888e-01
-1.70730323e-01 -5.58110535e-01 3.27885419e-01 6.80721641e-01
-3.51873934e-01 -1.01764679e+00 -5.20792961e-01 9.89325643e-02
-7.92834640e-01 -6.08611584e-01 -2.73619741e-01 2.61563480e-01
-7.17777535e-02 1.66273546e+00 9.22766775e-02 -4.29394037e-01
2.96985865e-01 2.15319619e-01 -1.50878593e-01 -2.21056283e-01
-1.00249350e+00 8.00861537e-01 -8.25851038e-02 -3.05018693e-01
-4.22551006e-01 -3.36176395e-01 -1.06276846e+00 -1.72768354e-01
-6.80539966e-01 3.16104561e-01 1.02472913e+00 8.48161399e-01
5.36658704e-01 8.81418407e-01 8.19731832e-01 -7.55243540e-01
-4.92447466e-01 -1.18078995e+00 -6.35388315e-01 -8.70418083e-03
1.27415645e+00 1.30585097e-02 -7.39079356e-01 -6.32518679e-02]
|
[14.966389656066895, 6.008331775665283]
|
67b808ed-68f2-455a-8d7a-13c3880501ed
|
puzzle-solving-without-search-or-human
|
2109.02797
| null |
https://arxiv.org/abs/2109.02797v1
|
https://arxiv.org/pdf/2109.02797v1.pdf
|
Puzzle Solving without Search or Human Knowledge: An Unnatural Language Approach
|
The application of Generative Pre-trained Transformer (GPT-2) to learn text-archived game notation provides a model environment for exploring sparse reward gameplay. The transformer architecture proves amenable to training on solved text archives describing mazes, Rubik's Cube, and Sudoku solvers. The method benefits from fine-tuning the transformer architecture to visualize plausible strategies derived outside any guidance from human heuristics or domain expertise. The large search space ($>10^{19}$) for the games provides a puzzle environment in which the solution has few intermediate rewards and a final move that solves the challenge.
|
['Ryerson Burdick', 'David Noever']
|
2021-09-07
| null | null | null | null |
['rubik-s-cube']
|
['graphs']
|
[-1.25720531e-01 4.75296021e-01 2.00260013e-01 2.14112118e-01
-1.01430011e+00 -9.58316505e-01 5.67675054e-01 -2.85833031e-01
-2.09569857e-01 1.10112703e+00 4.57006782e-01 -9.51524198e-01
-7.02876806e-01 -9.37279046e-01 -2.99167693e-01 -3.99480462e-01
-4.21171218e-01 8.77774000e-01 1.32965883e-02 -9.28119600e-01
3.35537612e-01 -1.88995123e-01 -1.20405543e+00 3.89244020e-01
5.91429591e-01 8.14827681e-01 4.49778348e-01 8.33455741e-01
-2.11891323e-01 1.30370820e+00 -9.54536736e-01 -6.39324605e-01
5.88640273e-01 -4.96994883e-01 -9.77512836e-01 -4.05592114e-01
-3.24480981e-01 -1.87470123e-01 -4.74083602e-01 7.78746724e-01
2.57671475e-01 4.20778841e-01 4.38099712e-01 -1.15701640e+00
-8.83277714e-01 7.89408624e-01 -9.81640592e-02 4.77405936e-01
7.39587307e-01 7.81373560e-01 1.19048131e+00 -3.85063261e-01
1.04051554e+00 9.25646067e-01 5.08870065e-01 6.06110573e-01
-1.16757393e+00 -4.36327338e-01 -4.88552153e-02 1.45666853e-01
-1.02679777e+00 4.25668694e-02 5.03134847e-01 -2.11683542e-01
1.67306280e+00 6.62443578e-01 1.02752650e+00 1.48086023e+00
1.27932310e-01 5.33755362e-01 9.56517577e-01 -4.73172516e-02
4.50118899e-01 -3.88250679e-01 -4.59127188e-01 7.11984634e-01
-2.81888768e-02 4.87341493e-01 -6.21896029e-01 -5.23429692e-01
1.24441719e+00 -4.87498969e-01 -2.07532346e-02 -4.48605746e-01
-9.82158959e-01 9.46862817e-01 4.09335643e-01 -2.09198877e-01
-2.91933745e-01 4.03619200e-01 4.23700422e-01 4.98734444e-01
2.90919811e-01 1.20882940e+00 -4.47165221e-01 -1.10801470e+00
-7.98088133e-01 7.83995688e-01 7.21482277e-01 1.18428814e+00
2.41962433e-01 6.32083833e-01 -1.50776044e-01 4.74980712e-01
4.14088294e-02 2.72193947e-03 5.12165785e-01 -1.25649679e+00
8.09286356e-01 6.03579223e-01 3.31765205e-01 -7.41225421e-01
-3.08583170e-01 -7.18488097e-01 -1.61887079e-01 5.95201850e-01
5.28908670e-01 -1.10684723e-01 -8.03677380e-01 1.47576582e+00
-9.78152454e-02 -7.12353215e-02 3.74670118e-01 8.97714019e-01
9.23607767e-01 6.36455238e-01 3.22671793e-03 9.01816040e-02
1.09570324e+00 -9.59834218e-01 -1.71245664e-01 -6.83238029e-01
8.14145029e-01 -2.74114996e-01 1.53498697e+00 7.19336271e-01
-1.71984112e+00 2.35910952e-01 -8.98519874e-01 -1.60293028e-01
-3.19172740e-01 -3.80825162e-01 1.02658713e+00 6.09126925e-01
-1.33142805e+00 6.46716297e-01 -6.07824147e-01 -4.08664383e-02
5.38921893e-01 2.22217351e-01 -2.23633245e-01 7.80691626e-03
-9.82259631e-01 9.35680151e-01 6.60524666e-01 -1.46782950e-01
-1.12197435e+00 -4.26845074e-01 -8.56753230e-01 3.61090958e-01
7.56445229e-01 -8.19115043e-01 1.53716218e+00 -3.63755375e-01
-1.72993898e+00 7.34413087e-01 4.81825978e-01 -3.77021551e-01
6.34420097e-01 1.87124416e-01 1.56875879e-01 -1.23622425e-01
8.71041119e-02 3.35986823e-01 4.11717951e-01 -7.18667030e-01
-3.86038363e-01 9.12123993e-02 6.32276237e-01 6.19725645e-01
1.97253600e-01 1.67870641e-01 -2.40313947e-01 -6.22055054e-01
1.60590723e-01 -5.44914544e-01 -6.06187284e-01 -7.43334651e-01
-2.53561586e-01 -1.74094409e-01 2.67666280e-02 -5.89234948e-01
1.15355766e+00 -1.81381691e+00 4.76116151e-01 2.78919607e-01
3.33527148e-01 -3.42492372e-01 -3.39470834e-01 6.15860701e-01
-1.12792313e-01 2.83014387e-01 2.67444462e-01 6.78939894e-02
4.46518183e-01 2.69725472e-01 -4.25039798e-01 -9.23555419e-02
-5.33692613e-02 1.23738635e+00 -1.21552420e+00 -3.37450244e-02
-1.05173178e-01 -1.60356358e-01 -1.03884447e+00 1.03004396e-01
-7.98111439e-01 1.89918458e-01 -5.34214377e-01 7.26483762e-01
8.39099586e-02 -4.68418121e-01 4.29911137e-01 8.15651238e-01
1.09056924e-02 9.67205107e-01 -1.13136458e+00 2.03645945e+00
-1.20072424e-01 3.49329293e-01 -1.15587540e-01 -4.76956725e-01
8.19949329e-01 8.56531337e-02 1.04110152e-01 -1.18431473e+00
1.50590092e-01 2.12979734e-01 8.37840214e-02 -9.23657864e-02
1.08817971e+00 2.60542542e-01 -5.05992830e-01 9.39106345e-01
1.20669380e-01 -5.44870257e-01 5.68863690e-01 4.63163525e-01
1.67851317e+00 6.20394289e-01 9.50655043e-02 -2.75414407e-01
-2.77796805e-01 4.50573772e-01 4.69409436e-01 1.06961572e+00
2.94059575e-01 3.92015636e-01 1.09448802e+00 -7.20215142e-01
-1.22160566e+00 -1.28274441e+00 5.89799166e-01 1.35218775e+00
-2.57709652e-01 -1.23033965e+00 -4.47177231e-01 -4.06604856e-01
-4.89758909e-01 9.00386274e-01 -6.84447765e-01 -1.05371945e-01
-6.17647350e-01 -7.02708006e-01 5.92548192e-01 5.28429389e-01
2.94223666e-01 -1.35474873e+00 -1.03241038e+00 3.88479382e-01
-4.21574831e-01 -3.67739379e-01 -8.69120806e-02 6.96363747e-01
-9.35867131e-01 -1.25329411e+00 -2.93388844e-01 -4.17729318e-01
3.14118534e-01 -1.17806979e-01 1.76214695e+00 2.03762963e-01
-4.78698313e-01 2.98350275e-01 -2.02688158e-01 -2.08702713e-01
-3.56435627e-01 1.57782719e-01 -2.99087554e-01 -1.43568134e+00
3.68090197e-02 -8.67067516e-01 -4.97743279e-01 4.59810272e-02
-3.30315500e-01 3.70519787e-01 1.09731808e-01 1.10887229e+00
3.36380750e-01 -2.69103311e-02 5.69164455e-02 -6.29444659e-01
1.32816207e+00 -7.57345796e-01 -8.12462270e-01 1.55784152e-02
-4.47516531e-01 2.04034507e-01 5.51724732e-01 -2.23934844e-01
-7.67664433e-01 -5.37293315e-01 6.03638291e-02 -1.38595149e-01
3.95622253e-01 6.70739055e-01 1.46054015e-01 2.84244835e-01
1.22281003e+00 4.50786263e-01 -5.40433824e-01 -1.92847624e-01
5.68854928e-01 -1.64122373e-01 7.93397844e-01 -1.25588357e+00
6.40292585e-01 -2.13961422e-01 -3.38777721e-01 1.56495441e-02
-1.07220121e-01 4.29164886e-01 1.81237519e-01 -1.97797883e-02
4.04131591e-01 -7.11538672e-01 -1.18310583e+00 -1.26597166e-01
-7.88172066e-01 -1.02908349e+00 -7.91526139e-01 1.83141138e-03
-1.13640428e+00 -1.40019834e-01 -7.00676978e-01 -8.35638821e-01
-2.37360150e-01 -1.06935215e+00 6.13453805e-01 4.28980768e-01
-5.69691777e-01 -7.74517536e-01 1.47138119e-01 3.52829814e-01
5.80403686e-01 2.27295414e-01 1.19898617e+00 -5.77730358e-01
-9.80740428e-01 4.13831890e-01 4.66472805e-02 -5.52141964e-01
-1.94479957e-01 -2.38069773e-01 -4.33478087e-01 -1.75468028e-01
-6.15280978e-02 -6.13994360e-01 2.19080225e-01 1.93926588e-01
1.17644334e+00 -5.60023427e-01 -2.62653418e-02 8.25884402e-01
1.04808414e+00 3.99320722e-01 9.97892559e-01 1.19144773e+00
1.21326700e-01 6.45860657e-03 4.64912891e-01 7.86290109e-01
3.84796262e-01 4.27428246e-01 8.75932097e-01 3.52358878e-01
3.90029132e-01 -3.76428664e-01 1.23190641e-01 1.96547598e-01
-5.23817420e-01 -2.75808096e-01 -1.08254683e+00 4.72232223e-01
-1.87726712e+00 -1.16110241e+00 1.92873925e-01 1.83805859e+00
9.43462551e-01 5.74820638e-01 2.66841710e-01 -1.17461041e-01
-8.60221982e-02 2.74432003e-01 -5.65270782e-01 -8.36890638e-01
-1.30119368e-01 9.08551753e-01 2.84294993e-01 4.19566035e-01
-4.59017634e-01 1.24622595e+00 8.11788750e+00 9.89287376e-01
-4.72252101e-01 -1.80472791e-01 4.91339207e-01 -7.45753825e-01
-6.80036724e-01 1.05213054e-01 -1.08933285e-01 3.74260187e-01
9.61696267e-01 -6.13591671e-01 1.32419527e+00 9.59599018e-01
-2.96660252e-02 -2.24788800e-01 -9.77846801e-01 1.00584877e+00
-4.10079896e-01 -1.78661585e+00 -2.78957397e-01 3.06935459e-01
3.78056377e-01 -9.15096924e-02 6.94373310e-01 8.71933341e-01
1.57330537e+00 -1.59013236e+00 9.21087861e-01 5.25007173e-02
8.64284277e-01 -9.17542040e-01 1.02580957e-01 4.83250916e-01
-8.36061180e-01 -3.84969801e-01 -5.77755988e-01 -7.02478170e-01
-2.40903988e-01 -3.90138507e-01 -1.17060006e+00 4.91205066e-01
8.65461886e-01 2.59796649e-01 -5.26262581e-01 7.63657987e-01
-5.00531554e-01 3.28437656e-01 -3.18260491e-01 -8.49332735e-02
7.26077378e-01 -4.04841065e-01 6.22280419e-01 6.51483715e-01
6.19930565e-01 5.02149522e-01 -1.33633390e-01 1.28530622e+00
8.92330036e-02 -8.06479827e-02 -7.57718861e-01 -3.22413325e-01
4.55174983e-01 1.03421581e+00 -9.14460182e-01 4.14385311e-02
2.73969799e-01 8.24473798e-01 5.80704808e-01 4.48242426e-01
-8.02013636e-01 -1.68385310e-03 8.84993970e-01 4.45623789e-03
2.91163296e-01 -4.75917906e-01 -5.38614392e-01 -9.30826306e-01
-1.98402002e-01 -1.27846360e+00 6.23098075e-01 -1.55500138e+00
-8.08903098e-01 8.11574042e-01 2.01144684e-02 -9.05958474e-01
-1.04127681e+00 -5.52403331e-01 -7.95728624e-01 1.22413671e+00
-7.17829823e-01 -7.38264799e-01 1.34547073e-02 5.39941669e-01
5.08267581e-01 -6.41236544e-01 1.17377758e+00 -4.34775501e-01
-3.21563840e-01 4.89927590e-01 1.92316905e-01 -1.95291251e-01
-9.50004756e-02 -1.62147331e+00 1.12129152e+00 5.78889251e-01
1.90700978e-01 6.17742062e-01 1.01198161e+00 -6.53973281e-01
-1.28759813e+00 -3.57827514e-01 1.75628573e-01 -7.90918469e-01
8.22528005e-01 -5.46263933e-01 -5.01207471e-01 8.82890165e-01
3.83989006e-01 -3.97953480e-01 6.76349461e-01 2.94566512e-01
-4.07427490e-01 6.14820540e-01 -1.16202176e+00 1.10663879e+00
1.44439888e+00 -4.58604723e-01 -7.02868581e-01 2.62328416e-01
4.61199760e-01 -1.45655298e+00 -4.25138742e-01 -4.00139600e-01
2.74358183e-01 -1.11172295e+00 1.08252513e+00 -1.11972439e+00
9.58487689e-01 1.39469607e-02 -6.70754090e-02 -1.71570575e+00
-4.26617563e-01 -1.23871219e+00 -1.91923186e-01 3.85569066e-01
4.17742670e-01 -3.63614380e-01 1.29957664e+00 1.08722413e+00
-2.76826978e-01 -8.13244045e-01 -9.33440208e-01 -6.52158320e-01
3.10120881e-01 -5.20313501e-01 1.04487765e+00 6.99198008e-01
5.48375964e-01 -1.38544329e-02 -2.21036777e-01 -3.95130634e-01
4.34833199e-01 -5.12008034e-02 8.15349996e-01 -9.83586848e-01
-8.20952952e-01 -6.24443173e-01 -3.23075801e-02 -1.00534105e+00
-2.29569793e-01 -1.02745378e+00 -2.28581160e-01 -1.62723434e+00
2.25215964e-02 -5.50133407e-01 -1.26537293e-01 7.61900008e-01
1.08308151e-01 5.91546185e-02 3.61306518e-01 -3.13105993e-02
-8.69684637e-01 4.84842002e-01 1.17119479e+00 -4.74662594e-02
-2.76444823e-01 -5.08751452e-01 -1.39422584e+00 5.70818603e-01
8.62282217e-01 -5.47499001e-01 -8.08676898e-01 -5.27254879e-01
1.17393851e+00 4.69721794e-01 2.23805249e-01 -6.80345833e-01
1.72397017e-01 -5.86611271e-01 3.14110488e-01 -2.42927030e-01
4.56044286e-01 -2.54848421e-01 4.12249833e-01 3.01690906e-01
-2.65435249e-01 5.67728043e-01 5.72212160e-01 3.23641717e-01
3.12723577e-01 -1.17366351e-01 8.64014179e-02 -7.28866816e-01
-6.56000733e-01 -1.65991828e-01 -7.11143196e-01 5.30279040e-01
8.22521865e-01 -7.17647493e-01 -8.04320157e-01 -7.40620494e-01
-9.53678489e-01 4.70988303e-01 6.38492882e-01 3.57662857e-01
7.01066494e-01 -1.21182466e+00 -6.14068508e-01 5.49742840e-02
-1.20874204e-01 2.16691777e-01 2.01662228e-01 1.76280681e-02
-8.98075402e-01 2.40174994e-01 -7.25592673e-01 -7.99398199e-02
-9.47981715e-01 2.90338457e-01 5.03746629e-01 -9.47193086e-01
-8.56510043e-01 1.11945772e+00 -6.16214648e-02 -3.54076535e-01
6.13559363e-03 -3.69135857e-01 -9.21754865e-04 -4.99293417e-01
4.36941624e-01 3.08652550e-01 6.33398369e-02 2.14740753e-01
-1.47166640e-01 -3.43758494e-01 -1.17271103e-01 -4.42660034e-01
1.72868538e+00 2.37582475e-01 1.27584547e-01 -7.94692338e-03
8.98058489e-02 -2.97538489e-01 -1.41380787e+00 8.62435699e-02
4.32797819e-02 -6.14514887e-01 -2.91715086e-01 -1.30814159e+00
-6.83799684e-01 4.89072174e-01 1.99414752e-02 2.22827852e-01
8.23972046e-01 -2.67035455e-01 4.01339412e-01 7.02519357e-01
8.36880267e-01 -1.11891651e+00 4.94255126e-01 1.10997832e+00
9.52729762e-01 -6.10923350e-01 -1.18523955e-01 9.60337669e-02
-7.09097266e-01 1.09715474e+00 7.93958068e-01 -3.13131392e-01
-1.93131819e-01 4.05072957e-01 -2.14502618e-01 -7.12335885e-01
-1.30316722e+00 1.39870971e-01 -1.98921636e-01 8.18605602e-01
1.82668880e-01 -8.85072425e-02 -1.74617127e-01 1.24263024e+00
-1.15827024e+00 -1.96591035e-01 9.81137097e-01 1.07532525e+00
-1.52491301e-01 -1.02475047e+00 -3.83446485e-01 4.92167115e-01
-2.86543548e-01 -5.62308609e-01 -4.09703314e-01 8.81930053e-01
-3.05321306e-01 5.97732306e-01 -5.22715179e-03 -3.19248855e-01
1.88185960e-01 1.25961617e-01 8.49280953e-01 -7.89414942e-01
-1.17522216e+00 -9.51570552e-03 2.97081679e-01 -9.12493169e-01
6.33374512e-01 -4.21454012e-01 -1.18795991e+00 -7.59392262e-01
3.46488744e-01 3.96777838e-01 2.36396268e-01 6.48885131e-01
2.46128634e-01 7.52571285e-01 -8.64991397e-02 -7.26625979e-01
-4.93713349e-01 -6.42175257e-01 -7.72309065e-01 -3.27010192e-02
-2.57814527e-01 -6.85043812e-01 -1.31969098e-02 -4.27913010e-01]
|
[3.7462754249572754, 1.4301806688308716]
|
f7b1709b-7af9-464c-af71-ea096586962f
|
provably-efficient-generalized-lagrangian
|
2306.00212
| null |
https://arxiv.org/abs/2306.00212v1
|
https://arxiv.org/pdf/2306.00212v1.pdf
|
Provably Efficient Generalized Lagrangian Policy Optimization for Safe Multi-Agent Reinforcement Learning
|
We examine online safe multi-agent reinforcement learning using constrained Markov games in which agents compete by maximizing their expected total rewards under a constraint on expected total utilities. Our focus is confined to an episodic two-player zero-sum constrained Markov game with independent transition functions that are unknown to agents, adversarial reward functions, and stochastic utility functions. For such a Markov game, we employ an approach based on the occupancy measure to formulate it as an online constrained saddle-point problem with an explicit constraint. We extend the Lagrange multiplier method in constrained optimization to handle the constraint by creating a generalized Lagrangian with minimax decision primal variables and a dual variable. Next, we develop an upper confidence reinforcement learning algorithm to solve this Lagrangian problem while balancing exploration and exploitation. Our algorithm updates the minimax decision primal variables via online mirror descent and the dual variable via projected gradient step and we prove that it enjoys sublinear rate $ O((|X|+|Y|) L \sqrt{T(|A|+|B|)}))$ for both regret and constraint violation after playing $T$ episodes of the game. Here, $L$ is the horizon of each episode, $(|X|,|A|)$ and $(|Y|,|B|)$ are the state/action space sizes of the min-player and the max-player, respectively. To the best of our knowledge, we provide the first provably efficient online safe reinforcement learning algorithm in constrained Markov games.
|
['Mihailo R. Jovanović', 'Zhaoran Wang', 'Zhuoran Yang', 'Xiaohan Wei', 'Dongsheng Ding']
|
2023-05-31
| null | null | null | null |
['multi-agent-reinforcement-learning']
|
['methodology']
|
[-1.15275718e-01 3.78566027e-01 -3.82523417e-01 1.08228676e-01
-1.16655695e+00 -6.50439858e-01 2.77733225e-02 1.97008207e-01
-1.25418890e+00 1.30343020e+00 -2.57454157e-01 -2.97075182e-01
-4.93642509e-01 -8.10688555e-01 -7.64011919e-01 -9.19932008e-01
-6.36508226e-01 6.11393809e-01 -9.87973660e-02 -2.97068506e-01
3.04761380e-01 4.08685878e-02 -1.03724670e+00 -6.53078556e-01
8.70706439e-01 1.30275810e+00 1.68247655e-01 7.72734106e-01
2.14228556e-01 9.95275438e-01 -5.24360001e-01 -3.17556471e-01
6.05941415e-01 -5.74150860e-01 -7.12042391e-01 2.17613548e-01
-5.57450473e-01 -5.72505176e-01 -5.09436548e-01 1.17953336e+00
3.88813287e-01 6.48649633e-01 2.67240614e-01 -1.38456333e+00
-1.09445877e-01 6.17435038e-01 -8.38626206e-01 1.09451391e-01
1.49989367e-01 4.79557902e-01 1.16815162e+00 -1.32462597e-02
6.02524519e-01 1.03181982e+00 2.38042492e-02 6.74120903e-01
-1.06797290e+00 -4.85404193e-01 6.44306481e-01 -8.04994702e-02
-1.01898277e+00 4.95443866e-02 3.17183495e-01 -1.72306776e-01
9.54690814e-01 1.72498330e-01 9.05154526e-01 6.40135050e-01
2.38124486e-02 8.09557378e-01 1.13256025e+00 -3.06891084e-01
9.73668575e-01 -2.09164798e-01 -2.64579087e-01 9.37642753e-01
-1.58082411e-01 4.98221219e-01 -4.15631175e-01 -1.43940061e-01
6.38497412e-01 4.69510518e-02 5.11142090e-02 -5.46883643e-01
-6.29320443e-01 1.41936183e+00 2.01768484e-02 -4.32325870e-01
-5.52957773e-01 5.82634807e-01 3.53657037e-01 4.85932082e-01
1.83025002e-01 3.99710923e-01 -4.23189342e-01 -5.50958991e-01
-6.46474063e-01 5.52451074e-01 8.98503482e-01 8.66983235e-01
6.86940491e-01 2.58045703e-01 -1.50552064e-01 4.18886155e-01
2.02380374e-01 7.35869586e-01 8.40613022e-02 -1.43069530e+00
7.87087858e-01 7.68065900e-02 7.59034574e-01 -3.46372515e-01
-2.67523348e-01 -3.54886234e-01 -3.26358974e-01 6.95289254e-01
6.10245824e-01 -8.61649632e-01 -5.48590720e-01 2.21456623e+00
2.86666930e-01 -2.43205979e-01 1.89535856e-01 9.30026352e-01
-3.49960655e-01 6.24689579e-01 -2.07541555e-01 -9.51881409e-01
8.83774579e-01 -9.24544334e-01 -5.88314474e-01 -3.84351522e-01
4.60662484e-01 -2.56290555e-01 8.29425395e-01 4.57296222e-01
-1.68398654e+00 3.17944467e-01 -1.03703070e+00 5.21206439e-01
-8.11460912e-02 -4.43391740e-01 6.24914885e-01 7.42887139e-01
-6.65414870e-01 7.57696390e-01 -9.07993376e-01 3.04494858e-01
2.99505711e-01 6.11307204e-01 5.62214106e-02 2.53485352e-01
-1.13043261e+00 8.22029769e-01 3.60249430e-01 1.72457956e-02
-1.49838471e+00 -1.69997230e-01 -8.01486433e-01 7.15088099e-02
1.28618157e+00 -2.55677789e-01 1.45766664e+00 -6.75198674e-01
-1.94910562e+00 4.98985827e-01 3.53623688e-01 -7.25929737e-01
7.95475543e-01 2.79004853e-02 3.75227749e-01 1.12764589e-01
3.41913790e-01 1.79699466e-01 6.63682699e-01 -7.92773962e-01
-8.39967310e-01 -4.53871340e-01 6.69485807e-01 7.39732087e-01
-1.16199434e-01 -1.20138213e-01 -7.92294741e-02 -2.71225214e-01
-2.79785812e-01 -1.12056386e+00 -7.56029546e-01 -4.43635471e-02
-1.28606945e-01 1.49210155e-01 9.98131409e-02 -4.05247748e-01
1.09343255e+00 -1.90957737e+00 3.56177866e-01 3.49391490e-01
3.21117006e-02 -1.32286370e-01 6.17405847e-02 4.44262773e-01
5.70810914e-01 -8.48103017e-02 -5.01381218e-01 -4.40193206e-01
3.70835930e-01 5.08351684e-01 -2.92430967e-01 6.53299570e-01
-5.44007838e-01 4.69664991e-01 -9.94488955e-01 -2.13844419e-01
-7.90880248e-03 -2.37050533e-01 -7.97066450e-01 2.83360600e-01
-4.44073260e-01 1.71867073e-01 -7.43256390e-01 3.73845756e-01
5.25551438e-01 7.37114474e-02 4.64946300e-01 8.24282348e-01
-3.71402174e-01 -1.12162091e-01 -1.58641911e+00 1.65142977e+00
-5.21950126e-01 1.12650521e-01 6.26653850e-01 -1.06656778e+00
4.99600291e-01 5.20197302e-02 8.60867202e-01 -8.66572559e-01
2.09069446e-01 1.67191789e-01 -1.99278191e-01 -2.07796365e-01
5.73186874e-01 -5.82294106e-01 -5.67140043e-01 7.86643088e-01
-8.30392540e-02 -6.03687353e-02 3.59000921e-01 3.06355745e-01
1.10419428e+00 1.85127139e-01 2.13746160e-01 -8.97499248e-02
1.27452731e-01 -2.25530937e-02 9.25194561e-01 1.07800436e+00
-4.75215167e-01 -1.33647576e-01 1.20982015e+00 -1.66275069e-01
-9.15705204e-01 -1.02380133e+00 3.24584275e-01 1.14154327e+00
5.03033161e-01 -1.71939164e-01 -6.28984272e-01 -6.59220934e-01
1.53302297e-01 6.70172155e-01 -7.91652739e-01 -4.19251844e-02
-4.15094405e-01 -9.37853515e-01 2.63835639e-01 3.48134995e-01
5.41700602e-01 -1.03798711e+00 -1.14639950e+00 4.52733427e-01
8.06223880e-03 -6.67413294e-01 -8.53120387e-01 5.89305460e-01
-4.72670972e-01 -1.00472200e+00 -6.38598382e-01 -2.71370798e-01
6.77671671e-01 -2.55375654e-01 6.56655371e-01 -4.53331560e-01
-3.11424017e-01 5.03175974e-01 -1.05690844e-01 -2.98506469e-01
1.01312641e-02 -2.04041854e-01 2.29674652e-01 -2.21939459e-01
-3.42771977e-01 -2.81250447e-01 -8.17015827e-01 7.88782313e-02
-7.23079622e-01 -1.96301252e-01 2.40727127e-01 9.27707314e-01
7.48945594e-01 5.25003970e-02 3.10176492e-01 -5.67194760e-01
7.74488330e-01 -4.08784270e-01 -1.38686728e+00 4.91536446e-02
-6.63304508e-01 4.03395355e-01 7.58036137e-01 -3.92300934e-01
-9.25893605e-01 9.08963606e-02 2.70271719e-01 -2.94260174e-01
6.00724101e-01 3.96732807e-01 4.51984964e-02 1.22471880e-02
3.42335492e-01 3.37409437e-01 1.63792565e-01 -8.56101662e-02
3.68284672e-01 1.48290485e-01 2.36681417e-01 -9.01347280e-01
5.21120012e-01 4.01356995e-01 2.05408171e-01 -9.95174944e-02
-6.52915299e-01 4.22950275e-02 -1.22392904e-02 -3.01088601e-01
6.55947089e-01 -7.37959981e-01 -1.75236213e+00 2.51230150e-01
-5.04756153e-01 -8.36571097e-01 -7.37563014e-01 5.16961336e-01
-1.29784596e+00 2.63497084e-01 -5.21146774e-01 -1.58441579e+00
-1.81943297e-01 -1.11781442e+00 4.90453720e-01 3.81479204e-01
4.46212977e-01 -8.13297510e-01 2.80721992e-01 2.83383042e-01
2.91035771e-01 3.95905584e-01 3.47162813e-01 -8.75443071e-02
-6.63827419e-01 3.46903689e-02 2.70927668e-01 1.71321630e-01
-3.03760976e-01 -5.26425064e-01 -2.05725402e-01 -7.18218207e-01
1.06995724e-01 -6.77900314e-01 6.38730347e-01 3.96397829e-01
8.13499570e-01 -9.20284033e-01 1.87743269e-02 4.69810098e-01
1.73266864e+00 5.24847806e-01 2.21313372e-01 5.38226485e-01
2.81879157e-02 1.14669949e-01 1.04681849e+00 1.34150040e+00
3.42137992e-01 4.51044291e-01 9.06811833e-01 3.99220467e-01
9.02677536e-01 -1.68468356e-01 7.25544870e-01 -3.86788100e-02
-3.36088575e-02 -2.23063707e-01 -5.29437363e-01 5.11328459e-01
-2.24677300e+00 -1.02759707e+00 6.73770964e-01 2.82583714e+00
1.02853680e+00 4.29553002e-01 4.64835048e-01 -2.79855490e-01
5.35911560e-01 5.94050176e-02 -1.11068738e+00 -7.78701186e-01
7.09249750e-02 3.46640527e-01 1.08013725e+00 9.49935973e-01
-8.20237577e-01 8.93982947e-01 5.18344116e+00 8.70229721e-01
-7.35870123e-01 7.80437887e-02 6.08315766e-01 -9.82465327e-01
-1.11761205e-02 1.97142631e-01 -6.38079405e-01 8.50479662e-01
7.92983353e-01 -3.75914782e-01 1.23865438e+00 9.29129422e-01
4.24507976e-01 -6.80413365e-01 -8.62379253e-01 7.35256970e-01
-4.35080916e-01 -1.15269649e+00 -8.03890884e-01 5.07852077e-01
8.61763120e-01 -1.47135973e-01 2.31498763e-01 5.26089549e-01
9.36884284e-01 -8.47532213e-01 8.74571383e-01 2.87073284e-01
6.99040890e-01 -1.36805248e+00 3.04200649e-01 5.17450273e-01
-1.09384120e+00 -5.46954155e-01 -1.48304150e-01 -4.72259134e-01
3.88376355e-01 -1.13799935e-02 -1.00901030e-01 1.58874050e-01
4.06793267e-01 -1.87584996e-01 5.54198205e-01 7.90801525e-01
-2.74142802e-01 3.87730673e-02 -5.36426842e-01 -2.65197963e-01
8.12322557e-01 -7.70245194e-01 5.61681926e-01 5.65070450e-01
5.79263903e-02 4.55199808e-01 8.02800596e-01 8.12670588e-01
2.16901042e-02 4.91928831e-02 -2.02300370e-01 -5.29777035e-02
3.58175516e-01 9.88951325e-01 -8.03827822e-01 -1.49640635e-01
-9.45249796e-02 1.04004717e+00 4.73935515e-01 3.43616009e-01
-1.15095699e+00 -4.94352758e-01 8.95963013e-01 -3.14909011e-01
3.86439234e-01 -3.59720677e-01 -5.33165317e-03 -9.62612092e-01
1.91236049e-01 -5.17887235e-01 6.62793159e-01 -7.79466778e-02
-8.24711800e-01 1.49348944e-01 -1.61835790e-01 -8.42952847e-01
-6.25307143e-01 -3.15263718e-01 -5.11277676e-01 7.75086522e-01
-1.35606062e+00 -4.60420698e-01 3.69892895e-01 8.08232009e-01
3.03140134e-01 -2.57739365e-01 6.85218334e-01 -1.82049364e-01
-7.31403589e-01 5.09542584e-01 6.05892241e-01 -1.14843763e-01
-6.88130385e-04 -1.49514318e+00 -2.31148779e-01 6.74769938e-01
-5.48148334e-01 1.29284173e-01 8.12548757e-01 -4.20432270e-01
-1.53262424e+00 -6.81912184e-01 1.49871975e-01 2.85549253e-01
7.98970401e-01 -3.11605692e-01 -8.67332593e-02 6.80435061e-01
5.90957850e-02 8.06889236e-02 5.32238007e-01 -9.03181508e-02
7.67270029e-02 6.20868383e-03 -1.43001378e+00 8.08620870e-01
7.99537063e-01 -2.67562777e-01 6.99931011e-02 4.19754684e-01
4.40081269e-01 -8.26764166e-01 -8.23084831e-01 -2.13302195e-01
4.12649930e-01 -6.44330740e-01 5.13222933e-01 -6.60302103e-01
1.62400141e-01 7.87273124e-02 -2.22350731e-01 -1.30294669e+00
7.47269467e-02 -1.32396626e+00 -3.85537803e-01 4.88406092e-01
4.96436298e-01 -5.66194892e-01 1.13599253e+00 9.52381253e-01
5.77625930e-02 -1.08040071e+00 -1.39052856e+00 -1.00789022e+00
4.54786509e-01 -2.03915581e-01 1.26029983e-01 4.72775340e-01
5.93767524e-01 -1.73278391e-01 -7.88140595e-01 1.06214574e-02
9.65763450e-01 2.44412109e-01 2.95801789e-01 -3.54058176e-01
-9.94811237e-01 -3.02687407e-01 3.25993985e-01 -9.57214236e-01
1.67046964e-01 -5.07795095e-01 2.09070280e-01 -1.16595638e+00
4.66731757e-01 -5.60884714e-01 -3.11039567e-01 5.45648098e-01
5.16629517e-02 -3.59175742e-01 6.30377114e-01 -1.58365190e-01
-1.16681480e+00 9.15838778e-01 1.18848276e+00 -5.28879389e-02
-5.14973819e-01 1.88110799e-01 -5.53077996e-01 4.17380959e-01
9.42179918e-01 -5.13124943e-01 -4.53593105e-01 -2.88431466e-01
4.28762227e-01 1.03496683e+00 -1.47124201e-01 -5.28980553e-01
1.77104861e-01 -9.36485887e-01 -2.52038956e-01 -6.09387681e-02
6.13893390e-01 -4.31978464e-01 -2.38015562e-01 9.47383702e-01
-5.41179538e-01 2.82973312e-02 -1.08976930e-01 7.26605773e-01
1.71270415e-01 -5.01425087e-01 1.04664183e+00 -5.32959044e-01
-3.89896542e-01 5.79399109e-01 -7.32252777e-01 3.63963187e-01
1.34872949e+00 5.12706004e-02 2.74121165e-02 -8.59763801e-01
-9.66434062e-01 9.88208294e-01 2.54872262e-01 -1.76635936e-01
5.40284753e-01 -9.73336399e-01 -4.39443976e-01 -3.13735723e-01
-4.53738570e-01 -1.06334649e-02 3.89784366e-01 6.08925700e-01
-1.89215228e-01 1.82189196e-01 -3.22075129e-01 4.85861264e-02
-7.96772480e-01 6.67718947e-01 6.89791083e-01 -8.81392896e-01
-1.86753705e-01 7.80501246e-01 -2.58200169e-01 -7.63585344e-02
4.67519999e-01 1.72439158e-01 4.45144117e-01 1.34469301e-01
3.07115138e-01 6.88491106e-01 -4.54546988e-01 -9.74461436e-02
-1.86508596e-01 -7.23372996e-02 -2.24694788e-01 -1.01007342e+00
1.32771361e+00 -3.13819647e-01 3.12430203e-01 1.71252750e-02
9.81944919e-01 -2.50187874e-01 -1.98854768e+00 -2.05673605e-01
-2.68855631e-01 -3.76355052e-01 -1.24008060e-02 -7.79744029e-01
-1.09400833e+00 4.74603623e-01 6.69148743e-01 2.26213530e-01
8.31512034e-01 -2.61732608e-01 6.87434971e-01 3.25985402e-01
7.75626898e-01 -1.85430729e+00 4.42321539e-01 8.17351103e-01
5.04719317e-01 -1.06848681e+00 -6.63943440e-02 3.00076485e-01
-8.86426985e-01 7.79031515e-01 5.57189465e-01 -4.07436520e-01
2.67091006e-01 3.50270599e-01 -3.03322881e-01 1.51686803e-01
-8.57367516e-01 -5.19388318e-01 -6.98829770e-01 1.84517875e-01
-1.76127210e-01 4.13189918e-01 -6.82800472e-01 5.31655312e-01
-9.28075463e-02 -1.57958388e-01 8.07547510e-01 1.44343138e+00
-6.89018428e-01 -1.17266345e+00 -2.24685937e-01 2.96396077e-01
-6.71569526e-01 1.00727543e-01 1.59699276e-01 5.91993213e-01
-1.21595822e-01 9.41714287e-01 8.12837258e-02 2.84318298e-01
1.25856593e-01 -2.47464761e-01 6.95130289e-01 -4.01566893e-01
-4.60189670e-01 1.75733253e-01 -7.10193738e-02 -7.11088419e-01
8.66297334e-02 -8.09491694e-01 -1.63018298e+00 -3.85394692e-01
-2.05780506e-01 4.26472306e-01 6.17706180e-01 9.77495909e-01
-5.92230745e-02 3.34752977e-01 1.13224745e+00 -5.99571288e-01
-1.23457456e+00 -5.42367339e-01 -1.01065993e+00 4.34113927e-02
3.03912371e-01 -6.07225180e-01 -4.14756000e-01 -6.87196136e-01]
|
[4.362606525421143, 2.8576531410217285]
|
8016cb61-c2b5-40b5-a754-913522e60d9b
|
bidirectional-hierarchical-attention-networks
| null | null |
https://aclanthology.org/2021.findings-emnlp.51
|
https://aclanthology.org/2021.findings-emnlp.51.pdf
|
Bidirectional Hierarchical Attention Networks based on Document-level Context for Emotion Cause Extraction
|
Emotion cause extraction (ECE) aims to extract the causes behind the certain emotion in text. Some works related to the ECE task have been published and attracted lots of attention in recent years. However, these methods neglect two major issues: 1) pay few attentions to the effect of document-level context information on ECE, and 2) lack of sufficient exploration for how to effectively use the annotated emotion clause. For the first issue, we propose a bidirectional hierarchical attention network (BHA) corresponding to the specified candidate cause clause to capture the document-level context in a structured and dynamic manner. For the second issue, we design an emotional filtering module (EF) for each layer of the graph attention network, which calculates a gate score based on the emotion clause to filter the irrelevant information. Combining the BHA and EF, the EF-BHA can dynamically aggregate the contextual information from two directions and filters irrelevant information. The experimental results demonstrate that EF-BHA achieves the competitive performances on two public datasets in different languages (Chinese and English). Moreover, we quantify the effect of context on emotion cause extraction and provide the visualization of the interactions between candidate cause clauses and contexts.
|
['Yi Zhao', 'Guangming Lu', 'Guimin Hu']
| null | null | null | null |
findings-emnlp-2021-11
|
['emotion-cause-extraction']
|
['natural-language-processing']
|
[ 6.34949356e-02 -1.70481443e-01 3.27835754e-02 -3.94705713e-01
-2.94495702e-01 -3.53653282e-01 3.04607272e-01 2.85905868e-01
-4.58036482e-01 3.80563170e-01 6.44629776e-01 -7.39836246e-02
-2.33485922e-01 -8.79537523e-01 -2.20187098e-01 -4.12606925e-01
6.63521588e-02 -2.22855434e-01 1.90347627e-01 -2.98978180e-01
2.72123218e-01 1.44940123e-01 -1.51996720e+00 4.75803375e-01
1.25953746e+00 1.12359488e+00 2.48033941e-01 2.11740971e-01
-6.75202906e-01 8.91046286e-01 -7.23467588e-01 -3.14384043e-01
-2.98444629e-01 -6.76794827e-01 -6.28662288e-01 -1.00429900e-01
-4.01712984e-01 1.64761379e-01 1.16569929e-01 1.28850436e+00
5.66295803e-01 3.42190787e-02 4.24680084e-01 -1.35071003e+00
-5.95571458e-01 8.86009812e-01 -6.23081565e-01 2.40872920e-01
3.32250655e-01 -4.61191051e-02 1.17224920e+00 -9.97816026e-01
6.13789141e-01 1.39078319e+00 2.71346778e-01 3.67519885e-01
-6.56298101e-01 -8.01041305e-01 8.31443548e-01 4.74450290e-01
-1.27395630e+00 7.63108730e-02 1.21508539e+00 -3.09097230e-01
1.01645815e+00 4.19239014e-01 9.05346274e-01 9.77013171e-01
1.70664504e-01 9.24466729e-01 1.02281594e+00 -2.37951010e-01
2.04269007e-01 7.20217377e-02 5.50363362e-01 5.51748276e-01
-3.40119526e-02 -2.99103409e-01 -3.60384166e-01 6.76981583e-02
1.16859280e-01 -1.75515786e-01 -5.46552539e-01 1.96918577e-01
-8.79626453e-01 8.70652497e-01 4.22612458e-01 5.76706350e-01
-3.61162871e-01 -5.13143167e-02 7.98024118e-01 1.70336857e-01
4.89725024e-01 4.97249663e-01 -6.07389987e-01 -5.01702465e-02
-3.98705482e-01 4.88410192e-03 6.67168856e-01 9.74914134e-01
5.48726797e-01 -1.08627342e-01 -4.66956884e-01 7.35759616e-01
1.41320899e-01 1.55126885e-01 3.59783113e-01 -2.03118220e-01
5.99209070e-01 1.18636119e+00 -1.10204808e-01 -1.48947930e+00
-5.56738436e-01 -5.76199234e-01 -8.54693234e-01 -2.48607621e-01
-3.03891957e-01 -6.09620392e-01 -6.36990905e-01 1.95042562e+00
4.12695944e-01 1.19352965e-02 -1.16525292e-02 1.11136603e+00
1.44035566e+00 8.90575171e-01 3.87860030e-01 -5.51967621e-01
1.58492076e+00 -8.91902030e-01 -1.35373235e+00 -4.14406002e-01
5.20471156e-01 -6.29885972e-01 1.40236318e+00 3.24316889e-01
-6.96076572e-01 -3.08521003e-01 -1.08140516e+00 -1.16035111e-01
-6.06496394e-01 3.71528566e-01 6.55637205e-01 2.41607204e-01
-5.75998127e-01 1.99134856e-01 -4.21748310e-01 -1.57625958e-01
1.93527177e-01 3.87240022e-01 4.29995880e-02 1.47537738e-01
-1.85694480e+00 6.11532450e-01 5.83921194e-01 4.76984441e-01
-1.45115063e-01 -4.85317171e-01 -7.58230209e-01 5.04569709e-01
8.02960992e-01 -3.48572642e-01 6.17578208e-01 -1.33414340e+00
-1.15168893e+00 3.44117492e-01 -2.15164885e-01 8.14660415e-02
1.38752177e-01 -4.30111557e-01 -8.01866651e-01 1.19974315e-02
-1.41794845e-01 2.63100207e-01 4.38339531e-01 -1.16451776e+00
-6.42071128e-01 -3.77478004e-01 -5.03208302e-02 4.18114275e-01
-7.08075821e-01 2.51958072e-01 -9.77669775e-01 -7.83149838e-01
-1.01931445e-01 -5.04303813e-01 -1.48203373e-01 -5.74675560e-01
-5.70781291e-01 -5.33186078e-01 1.01835108e+00 -6.47554040e-01
2.02041554e+00 -2.28441358e+00 3.26401532e-01 2.19146088e-01
3.13198030e-01 9.17087495e-02 -9.30077583e-02 3.84737402e-01
-1.15784295e-01 5.26263595e-01 -8.63928571e-02 5.70966229e-02
-2.59784814e-02 7.19674081e-02 -2.89333791e-01 -1.75302371e-01
3.70184392e-01 9.02539909e-01 -9.21950221e-01 -7.34312952e-01
-1.01293184e-01 5.21306932e-01 -5.61019361e-01 3.45435560e-01
-2.68466622e-01 8.81705135e-02 -8.17494988e-01 3.90954196e-01
5.97080350e-01 -1.39545456e-01 2.95391113e-01 -5.37657022e-01
-1.45840511e-01 2.47503251e-01 -1.23424149e+00 1.15748382e+00
-3.22441041e-01 3.63168269e-01 2.63262242e-01 -5.34428537e-01
1.08588636e+00 1.15210645e-01 3.43025148e-01 -7.50991285e-01
5.33403814e-01 -2.10126471e-02 1.48048729e-01 -8.21133435e-01
4.18233424e-01 -9.46589932e-02 -4.59350705e-01 1.77254587e-01
-1.40628204e-01 2.95495659e-01 2.56515473e-01 3.79063576e-01
1.02204359e+00 -1.33928344e-01 1.79592326e-01 -2.88892746e-01
7.42421091e-01 -1.90197542e-01 1.11805272e+00 3.07661623e-01
-2.56006181e-01 1.54173151e-01 8.69266331e-01 -3.68124306e-01
-3.00265819e-01 -2.60369331e-01 3.94675195e-01 1.12832618e+00
5.63688338e-01 -9.05334592e-01 -6.84671164e-01 -9.41572189e-01
-3.68325680e-01 7.44095683e-01 -8.25474739e-01 -4.52778131e-01
-5.37263334e-01 -9.01931703e-01 9.02246088e-02 5.69223046e-01
7.54353344e-01 -1.45093179e+00 -6.07432544e-01 2.76371628e-01
-5.59191644e-01 -9.67610180e-01 -5.62006116e-01 2.47815564e-01
-3.40347320e-01 -8.89167011e-01 -9.57790539e-02 -7.30037689e-01
5.63876748e-01 -5.70064858e-02 1.08580005e+00 4.17649925e-01
-4.73251604e-02 1.21940486e-01 -7.11201072e-01 -5.40710688e-01
1.22958034e-01 2.68703878e-01 -4.60197091e-01 2.04403624e-01
7.58477211e-01 -3.77481014e-01 -5.51468253e-01 2.45762989e-01
-8.57085943e-01 3.12847048e-01 5.71913362e-01 6.65972531e-01
6.15667105e-01 3.39586139e-01 6.94440484e-01 -1.02277792e+00
1.28787744e+00 -5.26341379e-01 -4.11083251e-01 2.87062377e-01
-7.52885282e-01 3.37162660e-03 7.78502405e-01 -3.30240935e-01
-1.25957692e+00 -1.11713432e-01 -2.21479803e-01 -2.25589082e-01
1.56604685e-02 9.48898256e-01 -8.32400322e-01 5.52553594e-01
5.10944007e-03 2.40022112e-02 -7.45721698e-01 -3.26024175e-01
2.80122727e-01 5.90356112e-01 2.68895268e-01 -5.43342888e-01
3.27254295e-01 2.10407414e-02 -2.59638011e-01 -3.06630194e-01
-9.81545806e-01 -2.59305984e-01 -2.46877223e-01 -3.72021496e-01
1.16743028e+00 -6.27256036e-01 -7.57844329e-01 1.32518068e-01
-1.25732386e+00 -8.76154564e-03 7.24171102e-02 2.86452562e-01
1.11678123e-01 2.00094223e-01 -5.82188368e-01 -9.11123633e-01
-7.40143836e-01 -1.12006652e+00 8.85094821e-01 4.75898117e-01
-1.90197781e-01 -6.34559512e-01 -2.47384220e-01 -1.77753821e-01
2.14302540e-01 4.49480712e-01 1.34317040e+00 -5.45476913e-01
-2.89353698e-01 -4.58128378e-02 -3.65507662e-01 -3.94252688e-02
3.19209099e-02 2.21858397e-01 -7.72154987e-01 1.42422661e-01
8.11020061e-02 -8.21910799e-02 8.94494474e-01 1.16993912e-01
1.36214852e+00 -6.21258676e-01 -3.68368894e-01 4.10238385e-01
1.46322250e+00 5.96774220e-01 5.46543241e-01 1.62909254e-01
8.68019462e-01 7.17297733e-01 7.81917989e-01 4.17337984e-01
3.79147261e-01 3.73373002e-01 5.79450309e-01 -3.84273440e-01
2.44421348e-01 -3.38839263e-01 1.98980942e-01 1.13563740e+00
-1.18558081e-02 -4.40666199e-01 -6.27925336e-01 4.83180493e-01
-1.89612389e+00 -7.32091904e-01 -3.91780198e-01 1.65467513e+00
7.22934067e-01 2.77821034e-01 -1.97803184e-01 1.37687385e-01
8.60604942e-01 2.23462448e-01 -5.46208560e-01 -6.04776263e-01
-2.62362987e-01 -2.63508618e-01 -9.75511894e-02 3.70255172e-01
-1.03207242e+00 9.28372443e-01 5.02985144e+00 9.45212305e-01
-1.15072608e+00 -1.09008737e-01 6.58962131e-01 -3.85056734e-02
-7.33504057e-01 2.62647923e-02 -5.89380562e-01 6.48006320e-01
3.55466187e-01 -7.54633695e-02 3.09425026e-01 7.56702960e-01
2.81844825e-01 6.92988858e-02 -6.77481592e-01 8.68354380e-01
-4.39199731e-02 -8.26413453e-01 1.13918737e-01 -1.04485862e-01
4.61146057e-01 -3.30399513e-01 -2.49706656e-01 6.09624505e-01
1.60799511e-02 -7.65300214e-01 7.09956884e-01 5.04747808e-01
5.40688813e-01 -1.23769593e+00 9.37387645e-01 1.91798717e-01
-1.56997705e+00 -2.76305765e-01 -2.30182588e-01 2.61010956e-02
1.63138025e-02 8.02723587e-01 -2.42047027e-01 6.25181556e-01
9.03394163e-01 6.11385345e-01 -6.98643804e-01 6.56933188e-01
-7.65476227e-01 7.46001482e-01 -1.42468840e-01 -5.12823224e-01
3.26241523e-01 -2.29455382e-01 5.55276990e-01 1.55370641e+00
2.50820547e-01 4.54610080e-01 5.14677055e-02 1.05097711e+00
-1.43440112e-01 7.04151392e-01 -9.23545286e-02 -1.30865723e-01
5.45521915e-01 1.58538496e+00 -9.34899509e-01 -3.58397812e-01
-3.45890403e-01 8.51778567e-01 4.70380187e-01 4.43325669e-01
-8.80256534e-01 -8.65687430e-01 3.15590501e-01 -2.58455962e-01
3.45292956e-01 1.86600998e-01 -3.00719649e-01 -1.16685617e+00
1.15843870e-01 -8.17965329e-01 7.42461026e-01 -9.35663104e-01
-1.26828325e+00 9.35592294e-01 -2.19547063e-01 -9.05196726e-01
2.01270610e-01 -2.87692547e-01 -9.51766193e-01 8.47912669e-01
-1.35349846e+00 -8.72313678e-01 -4.96033818e-01 4.99380231e-01
4.12688643e-01 2.46297032e-01 6.27455950e-01 4.22009975e-01
-9.83148634e-01 4.14766431e-01 -5.64539075e-01 2.31770113e-01
5.02261341e-01 -1.29209304e+00 -6.75579011e-02 1.00069630e+00
-2.42494181e-01 7.91081190e-01 6.27687514e-01 -8.64613593e-01
-1.29332781e+00 -1.01055443e+00 1.17019427e+00 -9.54550803e-02
3.64946455e-01 -5.82257569e-01 -1.04606390e+00 3.58512282e-01
4.69764739e-01 -1.04528837e-01 6.26679063e-01 4.19982374e-01
-2.33257115e-01 -2.27133781e-01 -8.39872062e-01 7.28847563e-01
9.26089525e-01 -1.49360880e-01 -6.55059576e-01 -1.72900870e-01
1.15103579e+00 -1.62115142e-01 -5.30727983e-01 5.50579965e-01
2.63957322e-01 -8.77275765e-01 4.95285153e-01 -5.10437071e-01
8.43347311e-01 -5.06337345e-01 1.86834842e-01 -1.25794780e+00
-4.95817631e-01 -4.67116624e-01 -7.89457187e-02 1.75185275e+00
4.59884137e-01 -2.90498585e-01 2.38888398e-01 7.40994871e-01
-1.81706518e-01 -1.28544104e+00 -5.60608268e-01 -3.00054818e-01
-2.11945534e-01 -5.13821304e-01 1.06962633e+00 1.05101192e+00
1.46971196e-01 8.95967901e-01 -3.64543349e-01 1.54075082e-02
-1.37852162e-01 7.17642546e-01 2.76945233e-01 -1.18496859e+00
-5.96766323e-02 -6.05791628e-01 1.13544315e-01 -7.46338546e-01
1.55721724e-01 -8.78194690e-01 5.91863915e-02 -1.85583842e+00
3.51566821e-01 -8.87559801e-02 -4.99710143e-01 6.15970492e-01
-8.48174095e-01 -3.95242423e-01 1.99037924e-01 -1.12858787e-01
-8.60093534e-01 7.38286316e-01 1.32176387e+00 -1.19729660e-01
-3.63239169e-01 -6.18219435e-01 -1.10350871e+00 8.14794004e-01
6.54245496e-01 -4.36924607e-01 -5.31875074e-01 -3.29571784e-01
7.60305643e-01 -2.41226643e-01 7.35562146e-02 -4.96602178e-01
1.52690917e-01 -4.10605252e-01 4.07202750e-01 -7.32522070e-01
-2.66806245e-01 -9.60231006e-01 -1.40222535e-01 1.82991654e-01
-3.32531095e-01 1.94806769e-01 2.68574983e-01 5.65131128e-01
-3.53757709e-01 -1.88097712e-02 3.82770032e-01 2.13669147e-03
-9.48808670e-01 1.94561303e-01 -3.42207760e-01 3.48413289e-01
7.74463654e-01 1.74461499e-01 -2.39640325e-01 -3.82799327e-01
-4.73977238e-01 6.27413929e-01 -1.01570776e-02 6.69476449e-01
5.92367828e-01 -1.43574739e+00 -4.80864942e-01 3.16289514e-02
2.82082349e-01 -1.19771414e-01 4.10858601e-01 6.73645258e-01
6.17753714e-02 8.62729475e-02 8.89049694e-02 -6.87702149e-02
-1.33967483e+00 9.48914766e-01 2.34021932e-01 -6.48307621e-01
-4.42289978e-01 7.43494868e-01 3.33074898e-01 -1.54370740e-01
1.37028888e-01 -3.84042799e-01 -8.95758331e-01 3.85068059e-01
3.82573187e-01 7.16733187e-02 -1.87838189e-02 -5.16450286e-01
-5.71232855e-01 4.76155609e-01 1.54959038e-01 8.12809318e-02
1.38244319e+00 -2.88775712e-01 -3.28769684e-01 2.16291755e-01
1.07303238e+00 2.60070652e-01 -8.46214354e-01 1.57458428e-02
8.34605321e-02 -1.21961765e-01 2.60515243e-01 -1.09003675e+00
-1.40257716e+00 9.20468092e-01 3.69027168e-01 3.26881856e-01
1.60843098e+00 -1.26495004e-01 8.39915216e-01 2.51706634e-02
-1.41043633e-01 -1.38207352e+00 -1.13851935e-01 6.45232379e-01
1.12465847e+00 -8.56621504e-01 -5.33890761e-02 -7.92261362e-01
-7.30976701e-01 1.01595378e+00 1.00932157e+00 1.02345929e-01
5.54600298e-01 5.23950517e-01 6.69084936e-02 -5.77832162e-01
-7.99467325e-01 -3.82843196e-01 6.20489478e-01 1.68570742e-01
5.23274660e-01 -1.03180729e-01 -1.02346420e+00 1.71873343e+00
-5.45627512e-02 -1.28685191e-01 1.18830748e-01 7.65303791e-01
-3.07767928e-01 -8.70755494e-01 -1.69005543e-01 3.87637645e-01
-6.11234128e-01 -2.96113372e-01 -9.98281360e-01 7.22950637e-01
5.44788241e-01 1.02712202e+00 -1.78780541e-01 -6.24522626e-01
4.85659987e-01 7.98066407e-02 -1.71480477e-01 -3.93412262e-01
-9.77987289e-01 5.40347934e-01 1.86763763e-01 -5.76996207e-01
-3.68681490e-01 -2.58943826e-01 -1.63627803e+00 2.85110414e-01
-6.23943150e-01 4.15513247e-01 3.82405579e-01 9.56791997e-01
5.84150016e-01 1.07456899e+00 5.57532966e-01 -1.42302275e-01
1.98737621e-01 -8.59336078e-01 -5.41969299e-01 6.21694446e-01
-5.76654151e-02 -6.04357660e-01 -4.59634393e-01 -2.78480202e-01]
|
[12.625174522399902, 6.212848663330078]
|
dc5878ba-3e0c-457e-a19f-8bc33e12e801
|
configurable-spatial-temporal-hierarchical
|
2305.07328
| null |
https://arxiv.org/abs/2305.07328v1
|
https://arxiv.org/pdf/2305.07328v1.pdf
|
Configurable Spatial-Temporal Hierarchical Analysis for Flexible Video Anomaly Detection
|
Video anomaly detection (VAD) is a vital task with great practical applications in industrial surveillance, security system, and traffic control. Unlike previous unsupervised VAD methods that adopt a fixed structure to learn normality without considering different detection demands, we design a spatial-temporal hierarchical architecture (STHA) as a configurable architecture to flexibly detect different degrees of anomaly. The comprehensive structure of the STHA is delineated into a tripartite hierarchy, encompassing the following tiers: the stream level, the stack level, and the block level. Specifically, we design several auto-encoder-based blocks that possess varying capacities for extracting normal patterns. Then, we stack blocks according to the complexity degrees with both intra-stack and inter-stack residual links to learn hierarchical normality gradually. Considering the multisource knowledge of videos, we also model the spatial normality of video frames and temporal normality of RGB difference by designing two parallel streams consisting of stacks. Thus, STHA can provide various representation learning abilities by expanding or contracting hierarchically to detect anomalies of different degrees. Since the anomaly set is complicated and unbounded, our STHA can adjust its detection ability to adapt to the human detection demands and the complexity degree of anomaly that happened in the history of a scene. We conduct experiments on three benchmarks and perform extensive analysis, and the results demonstrate that our method performs comparablely to the state-of-the-art methods. In addition, we design a toy dataset to prove that our model can better balance the learning ability to adapt to different detection demands.
|
['Jing Liu', 'Zhaoyang Xia', 'Jing Teng', 'Chengxin Pang', 'Tian Wang', 'Yang Liu', 'Xinhua Zeng', 'Kai Cheng']
|
2023-05-12
| null | null | null | null |
['video-anomaly-detection', 'human-detection']
|
['computer-vision', 'computer-vision']
|
[-9.42832828e-02 -4.96980786e-01 1.92651585e-01 -1.42263040e-01
-6.66410252e-02 -4.12804514e-01 1.25293821e-01 -2.38381371e-01
1.42033324e-01 -2.01351449e-01 -1.39843538e-01 -4.06816602e-01
-9.80125368e-02 -8.22052419e-01 -5.53208351e-01 -7.15311408e-01
-4.11736339e-01 -6.09515123e-02 8.46824944e-01 -2.43493557e-01
1.20582856e-01 5.75145960e-01 -1.69074142e+00 4.81328100e-01
7.44184196e-01 1.29508996e+00 2.22510491e-02 5.26749969e-01
-3.20396960e-01 8.89754355e-01 -7.38344550e-01 -1.60159953e-02
5.32484770e-01 -3.47140521e-01 -2.41619512e-01 7.91649163e-01
8.89116526e-02 -4.67296243e-01 -6.74092770e-01 1.08324027e+00
2.66308278e-01 -1.31210580e-01 3.90672356e-01 -1.60433328e+00
-3.87391031e-01 2.86275893e-01 -8.46934438e-01 6.08656704e-01
3.20329010e-01 6.02372408e-01 5.79467356e-01 -6.95992947e-01
1.44781664e-01 1.24626577e+00 4.93664682e-01 4.16537583e-01
-6.97526038e-01 -6.30319357e-01 8.13038409e-01 6.30635917e-01
-1.05788517e+00 5.59579879e-02 1.06094468e+00 -6.51577115e-01
5.86765051e-01 2.38067925e-01 8.56660604e-01 9.57359195e-01
1.52548105e-01 9.20265317e-01 6.11034095e-01 6.08819798e-02
2.47955173e-01 -4.83258337e-01 -2.59639844e-02 7.03178585e-01
2.65618563e-01 -2.44510248e-01 -3.92777592e-01 1.57584324e-02
9.69725490e-01 2.22271264e-01 -3.05760682e-01 -7.12238073e-01
-1.26860094e+00 5.26970267e-01 2.39882141e-01 1.09449185e-01
-3.57049286e-01 -2.93332309e-01 8.92584264e-01 6.77174032e-01
-5.24686165e-02 2.16587916e-01 -5.94706655e-01 5.80767356e-02
-4.46564019e-01 1.22724036e-02 4.17576790e-01 1.16722524e+00
4.06146824e-01 5.90724051e-01 -3.40565681e-01 6.37130141e-01
2.42739886e-01 1.83629721e-01 6.64118350e-01 -8.09758186e-01
5.39388955e-01 8.77459049e-01 -4.04133976e-01 -1.23989987e+00
-3.16734254e-01 -3.71791482e-01 -1.27604163e+00 3.85738164e-01
2.58514047e-01 -4.48961090e-03 -1.12226999e+00 1.50689077e+00
5.04361212e-01 5.44066966e-01 -9.61144045e-02 7.88034976e-01
4.91005629e-01 8.10884774e-01 -2.60806173e-01 -5.23532510e-01
1.23599279e+00 -8.61380219e-01 -8.47430646e-01 6.47308230e-02
5.82362711e-01 -5.66751659e-01 1.22061372e+00 7.55957246e-01
-1.03448033e+00 -6.99784338e-01 -1.29949474e+00 4.11701888e-01
-1.75506815e-01 -1.31191239e-01 3.56814861e-01 4.06084001e-01
-6.86099589e-01 1.96724191e-01 -9.55588400e-01 -1.64721236e-02
3.33793372e-01 1.64058656e-01 -3.80714461e-02 -1.40642107e-01
-1.12588048e+00 1.36343062e-01 4.24577892e-01 4.22777385e-01
-1.12943292e+00 -5.32814801e-01 -1.00223112e+00 -1.42347068e-02
5.81696868e-01 -6.19410455e-01 9.96912599e-01 -9.46466804e-01
-1.36252272e+00 4.62931246e-01 1.59785822e-01 -2.15446964e-01
6.30068719e-01 8.00987110e-02 -7.86762416e-01 1.08667918e-01
-1.58315480e-01 8.20473805e-02 1.19535565e+00 -9.45318341e-01
-9.33721542e-01 -3.29492360e-01 3.19919795e-01 1.83535516e-01
-5.36370695e-01 -7.45221693e-03 -7.31088400e-01 -9.64356244e-01
5.25898457e-01 -5.89322090e-01 -2.21768990e-01 8.67056996e-02
-3.45605701e-01 -1.41860425e-01 1.36525190e+00 -3.69843125e-01
1.61874247e+00 -2.67978787e+00 3.39881390e-01 4.46965903e-01
2.58462876e-01 2.51026213e-01 -7.75921047e-02 -4.40767705e-02
-1.88856080e-01 -1.49825245e-01 -2.92682290e-01 1.89485103e-01
-7.71052465e-02 4.18501318e-01 -2.35182524e-01 2.18541652e-01
3.32664728e-01 3.33420038e-01 -9.36711371e-01 -5.34364581e-01
3.07444274e-01 3.10580228e-02 -6.37352049e-01 4.63006467e-01
-1.66091844e-01 3.76173526e-01 -7.27350235e-01 9.78315473e-01
7.51915991e-01 -2.26481348e-01 5.01175188e-02 -3.57834429e-01
6.75689504e-02 -1.77701607e-01 -1.53659272e+00 1.36665261e+00
3.07610687e-02 2.37620860e-01 2.69476265e-01 -1.14327788e+00
8.75251293e-01 1.20575078e-01 7.22962320e-01 -7.44987667e-01
-1.40894815e-01 4.97994311e-02 1.82020068e-01 -7.10254788e-01
1.42248794e-01 3.56221974e-01 -7.15727508e-02 3.49742956e-02
-1.17341980e-01 3.07896674e-01 2.26440102e-01 9.79275107e-02
1.47946858e+00 -1.91123694e-01 7.66826943e-02 -9.07314494e-02
1.01784527e+00 -2.77998030e-01 1.00649130e+00 4.92473900e-01
-4.36059535e-01 5.27314782e-01 9.41371977e-01 -8.06719422e-01
-9.82546806e-01 -1.22916090e+00 -1.50894560e-02 9.28671479e-01
4.70407218e-01 -4.26221639e-01 -5.41355550e-01 -8.40583503e-01
-1.89688206e-01 1.47157639e-01 -5.02355874e-01 -3.94380242e-01
-7.96467602e-01 -7.96298504e-01 2.54319668e-01 5.98013163e-01
8.67653191e-01 -1.09704792e+00 -5.22083998e-01 -5.93729690e-02
1.87640041e-02 -1.20572376e+00 -4.51363742e-01 -2.52312105e-02
-7.39733696e-01 -1.15724635e+00 -3.97461981e-01 -1.03005660e+00
6.95202172e-01 2.34144554e-01 9.63082850e-01 1.59854114e-01
-2.33772084e-01 6.05102539e-01 -4.76234645e-01 -1.89121604e-01
-3.54296088e-01 -1.07122645e-01 1.95219472e-01 1.54466167e-01
1.77196786e-01 -6.87282979e-01 -6.18728518e-01 6.98990583e-01
-1.42135394e+00 -1.60498485e-01 5.44834018e-01 6.58052742e-01
7.15765297e-01 5.82023919e-01 2.38464400e-01 -5.83338559e-01
4.92874742e-01 -6.13824308e-01 -7.81863153e-01 4.48819622e-02
-1.99566811e-01 -1.98335022e-01 7.99829185e-01 -6.15289688e-01
-7.93949604e-01 1.16029315e-01 -5.23710884e-02 -1.07796848e+00
-2.65304685e-01 2.49864846e-01 -6.75601482e-01 2.20620841e-01
2.71190405e-01 3.04323047e-01 1.90553721e-02 -1.62338465e-01
-4.88690883e-02 3.52340251e-01 6.62190795e-01 -5.51154315e-01
1.02572954e+00 1.75043494e-01 1.62365660e-01 -6.59308374e-01
-4.89874691e-01 -3.21654230e-01 -6.75172985e-01 -4.78720516e-01
8.27916443e-01 -8.52617443e-01 -6.48631990e-01 1.00531840e+00
-8.28198969e-01 -2.58013397e-01 -2.60398835e-01 1.75915673e-01
-3.96105587e-01 7.21575439e-01 -8.71112585e-01 -4.08976376e-01
6.87222630e-02 -1.27823341e+00 8.11757386e-01 -1.71501130e-01
3.20536941e-01 -6.10920608e-01 -3.17667335e-01 -1.77273989e-01
2.55374938e-01 5.25128543e-01 1.18597507e+00 -5.08166075e-01
-8.45958650e-01 2.56147217e-02 3.63632254e-02 4.31088924e-01
2.66782105e-01 1.98431045e-01 -6.06864214e-01 -3.29973400e-01
2.67036289e-01 1.55040370e-02 6.65672719e-01 1.92459762e-01
1.91231227e+00 -3.16228360e-01 -4.30479161e-02 8.42640758e-01
9.71616387e-01 4.37753022e-01 9.39738214e-01 6.90575719e-01
9.75867927e-01 4.96462643e-01 7.56422281e-01 5.03343463e-01
1.61844179e-01 5.44684470e-01 8.97683084e-01 1.06152698e-01
2.64171690e-01 9.91573334e-02 7.07568407e-01 1.12457585e+00
1.70522146e-02 -2.15244636e-01 -6.58528984e-01 2.91704237e-01
-1.95494616e+00 -1.10200381e+00 -7.28200302e-02 2.11575079e+00
3.55648398e-01 6.00071132e-01 4.44030553e-01 5.91085553e-01
8.92740011e-01 3.70861560e-01 -9.10234511e-01 -3.29169363e-01
-2.59003434e-02 -4.97146010e-01 1.48543432e-01 -5.47743328e-02
-1.18460500e+00 3.40558887e-01 6.18084002e+00 8.41580868e-01
-1.02035439e+00 -3.13945353e-01 6.81385040e-01 -2.67796665e-02
-1.46311507e-01 -5.76104522e-01 -4.09935355e-01 9.19650793e-01
3.36726785e-01 1.01671010e-01 1.98786423e-01 8.18371594e-01
-1.74154311e-01 5.48699796e-01 -1.03355908e+00 1.07934654e+00
-3.00243646e-02 -9.82508659e-01 4.30913806e-01 -1.03968263e-01
5.25710344e-01 -2.76752949e-01 2.64463514e-01 3.84185970e-01
-1.11942012e-02 -5.68597078e-01 5.97467422e-01 3.41747463e-01
4.78098989e-01 -6.83268666e-01 6.89537346e-01 2.62535304e-01
-1.61240315e+00 -6.21747971e-01 -2.86127448e-01 1.15139738e-01
1.04596250e-01 5.57012856e-01 -2.39772201e-01 6.35448039e-01
1.07576370e+00 8.46109271e-01 -5.04660666e-01 1.01278114e+00
-3.66554037e-02 4.52820420e-01 -1.81264907e-01 3.99493158e-01
2.30790764e-01 -1.97444141e-01 8.50522518e-01 8.25722337e-01
5.75161755e-01 1.01746917e-01 3.84923935e-01 4.40827698e-01
2.03607559e-01 -1.77315965e-01 -5.77524245e-01 4.13472056e-01
5.43617308e-01 9.39449251e-01 -4.92274791e-01 -2.23204434e-01
-7.46583402e-01 9.85387921e-01 -7.32972994e-02 3.26559991e-01
-1.05005276e+00 -3.21604431e-01 8.27622116e-01 2.95709819e-01
6.86376274e-01 -3.40917587e-01 2.28062458e-02 -1.15831017e+00
4.90397304e-01 -1.19995236e+00 9.58852768e-01 -6.17968500e-01
-1.46634758e+00 7.53477275e-01 -4.17218730e-02 -1.91688502e+00
-4.50865030e-02 -9.14888442e-01 -8.49675357e-01 1.48674712e-01
-1.28037405e+00 -8.43627512e-01 -7.24495411e-01 9.40448344e-01
8.41931045e-01 -4.42323864e-01 3.65052104e-01 4.97197241e-01
-1.08238304e+00 6.50180101e-01 -2.28730559e-01 2.77825505e-01
4.09628123e-01 -1.13008034e+00 3.01468104e-01 1.21270287e+00
-3.35068077e-01 1.08199134e-01 3.47393692e-01 -5.24719417e-01
-1.31583357e+00 -1.35291851e+00 -2.80551612e-01 -2.05785573e-01
7.67802179e-01 -3.16565305e-01 -1.34791565e+00 7.50655651e-01
-6.10351861e-02 3.11620682e-01 3.20290804e-01 -1.65145531e-01
-4.57331657e-01 -5.20454705e-01 -9.66179609e-01 6.07937336e-01
1.37583435e+00 -1.39068857e-01 -4.17546570e-01 1.72815204e-01
1.05634832e+00 -7.12045968e-01 -7.10222661e-01 9.07186925e-01
2.08533525e-01 -1.23147857e+00 1.08630705e+00 -4.89402562e-01
4.74767745e-01 -8.31668973e-01 -1.57494649e-01 -9.88234103e-01
-7.37380087e-01 -4.73233253e-01 -6.69687271e-01 1.20059180e+00
5.63852824e-02 -5.63299298e-01 6.50126159e-01 1.78740248e-01
-6.07398748e-01 -1.18720257e+00 -7.39064872e-01 -7.89984643e-01
-3.28905612e-01 -4.66251016e-01 9.97724235e-01 9.66034055e-01
-2.06701458e-01 -2.10935287e-02 -3.31270248e-01 7.28001177e-01
3.87345403e-01 -5.46599068e-02 7.62418747e-01 -1.05441201e+00
-2.83281207e-01 -5.72360098e-01 -1.07290912e+00 -1.17796886e+00
-2.39376560e-01 -4.08642173e-01 -1.29824996e-01 -9.88547385e-01
-7.99549296e-02 -3.48062873e-01 -7.56746471e-01 3.20503503e-01
-1.90909252e-01 -7.56035671e-02 -9.90515277e-02 2.88982511e-01
-8.85548770e-01 5.49084783e-01 1.34869587e+00 -2.84421057e-01
-2.98119605e-01 -9.44164582e-03 -3.42738867e-01 9.77237463e-01
7.30230570e-01 7.50270579e-03 -7.19585538e-01 -6.27807498e-01
1.59838319e-01 -1.65347248e-01 2.22856149e-01 -1.32623816e+00
2.90042937e-01 -1.82783827e-01 5.70446968e-01 -7.95769095e-01
-6.27582101e-03 -1.05003273e+00 -9.29235891e-02 4.45325196e-01
1.09978449e-02 6.87056124e-01 2.35675067e-01 7.53899157e-01
-3.73816937e-01 1.73593715e-01 4.64373112e-01 -7.53068104e-02
-1.12454534e+00 8.15260768e-01 -5.99967360e-01 1.65692121e-02
1.42596686e+00 -4.83781099e-01 -1.07660368e-01 -2.45907605e-01
-6.46390915e-01 6.46139205e-01 4.46399629e-01 5.62411427e-01
8.06874573e-01 -1.59176183e+00 -3.87412786e-01 7.95612216e-01
2.66249239e-01 6.86233222e-01 4.99203622e-01 7.32647955e-01
-6.17178500e-01 -1.00887626e-01 -4.17067647e-01 -1.06541121e+00
-1.02104127e+00 1.04135668e+00 4.52315480e-01 -3.29039872e-01
-7.72009075e-01 4.68423724e-01 6.03482604e-01 -2.70713568e-01
5.42441070e-01 -4.50518757e-01 -2.68077284e-01 -2.65554845e-01
5.65968812e-01 4.78812069e-01 -4.51494008e-02 -4.13313746e-01
-1.51374638e-01 7.01353490e-01 -7.77387321e-02 6.32098198e-01
9.49637890e-01 -1.56779334e-01 -9.55960378e-02 3.94101828e-01
8.20522726e-01 -1.67236134e-01 -1.40498972e+00 -1.73170403e-01
-1.29842728e-01 -6.36337876e-01 -2.89638340e-01 -1.21055178e-01
-1.51598144e+00 6.36210740e-01 6.71871781e-01 6.33076966e-01
1.71034992e+00 -1.62176758e-01 8.49325180e-01 3.63603920e-01
1.19112536e-01 -8.46200526e-01 6.73349082e-01 5.87267995e-01
6.55251861e-01 -9.28589046e-01 -1.80466756e-01 -6.59610450e-01
-6.79802120e-01 1.11173332e+00 1.34480107e+00 -7.89587200e-02
6.93548441e-01 3.85854661e-01 5.00359647e-02 -3.56679171e-01
-5.99379778e-01 9.17271376e-02 2.53399402e-01 6.96259081e-01
-1.10875800e-01 -2.95027256e-01 2.70123869e-01 5.28678417e-01
1.72820598e-01 -4.54908580e-01 3.72364074e-01 8.37387204e-01
-4.25287962e-01 -9.59958673e-01 -4.62357223e-01 5.90114713e-01
-2.41217434e-01 4.03198779e-01 8.99084657e-02 7.15529382e-01
4.59405959e-01 6.51245236e-01 4.73200649e-01 -6.92997575e-01
6.59076631e-01 -2.01477289e-01 1.99015737e-01 -3.29495460e-01
-2.65790801e-02 -3.94802913e-03 -4.00848985e-01 -7.84305573e-01
-2.99026281e-01 -6.37023509e-01 -1.22658420e+00 -1.33627653e-01
-2.33539604e-02 -1.06428914e-01 -1.17070407e-01 8.04395080e-01
3.69963378e-01 1.08889592e+00 1.12989604e+00 -5.91951251e-01
-3.39184880e-01 -5.71900666e-01 -6.07376516e-01 5.14064372e-01
6.12874329e-01 -7.10472107e-01 -4.14267838e-01 -1.43967066e-02]
|
[7.897280216217041, 1.5824859142303467]
|
1c224742-0665-4578-a202-7ccfb6f00af1
|
sigmoidally-preconditioned-off-policy
|
2205.10047
| null |
https://arxiv.org/abs/2205.10047v6
|
https://arxiv.org/pdf/2205.10047v6.pdf
|
The Sufficiency of Off-Policyness and Soft Clipping: PPO is still Insufficient according to an Off-Policy Measure
|
The popular Proximal Policy Optimization (PPO) algorithm approximates the solution in a clipped policy space. Does there exist better policies outside of this space? By using a novel surrogate objective that employs the sigmoid function (which provides an interesting way of exploration), we found that the answer is ``YES'', and the better policies are in fact located very far from the clipped space. We show that PPO is insufficient in ``off-policyness'', according to an off-policy metric called DEON. Our algorithm explores in a much larger policy space than PPO, and it maximizes the Conservative Policy Iteration (CPI) objective better than PPO during training. To the best of our knowledge, all current PPO methods have the clipping operation and optimize in the clipped policy space. Our method is the first of this kind, which advances the understanding of CPI optimization and policy gradient methods. Code is available at https://github.com/raincchio/P3O.
|
['Yi Chang', 'Bei Jiang', 'Randy Goebel', 'Zhiwei Yang', 'Zhixiao Sun', 'Haiyin Piao', 'Hengshuai Yao', 'Hechang Chen', 'Dongcui Diao', 'Xing Chen']
|
2022-05-20
| null | null | null | null |
['policy-gradient-methods']
|
['methodology']
|
[-3.07018220e-01 1.20553285e-01 -9.08711553e-01 1.32148787e-01
-5.81600189e-01 -8.37551713e-01 5.87218523e-01 -5.57290465e-02
-7.30417550e-01 1.32328963e+00 2.84570694e-01 -7.31283247e-01
-2.06681132e-01 -3.65286618e-01 -8.88350010e-01 -7.29107141e-01
-1.81956485e-01 3.66960824e-01 2.49277622e-01 -4.22743946e-01
5.04183352e-01 2.85502851e-01 -1.24769163e+00 -6.66169301e-02
1.13201022e+00 9.03677166e-01 3.46035719e-01 5.57463348e-01
6.60675615e-02 4.66404796e-01 -5.16950846e-01 1.35102823e-01
5.15278697e-01 -2.97649711e-01 -7.52292871e-01 -6.10166848e-01
1.24552347e-01 -1.54292911e-01 -1.24070877e-02 1.29310977e+00
4.85284656e-01 5.85663795e-01 1.45395949e-01 -1.16411054e+00
-6.05274260e-01 5.29369593e-01 -3.05136472e-01 3.48555505e-01
2.39557698e-01 2.52234131e-01 7.62732327e-01 -7.08275676e-01
7.40568221e-01 1.28490353e+00 5.00712276e-01 6.30875945e-01
-1.09876657e+00 -3.10114652e-01 5.47147632e-01 -9.82333347e-03
-9.83283937e-01 -2.25778684e-01 6.71115100e-01 -3.06918442e-01
8.82211745e-01 4.84256804e-01 8.95868838e-01 1.15016580e+00
2.40196690e-01 1.17402518e+00 1.62334073e+00 -4.77712065e-01
6.77659929e-01 4.69087720e-01 -4.51888531e-01 5.32527030e-01
1.51035720e-02 1.00937569e+00 -3.70294422e-01 -4.04057980e-01
8.43767107e-01 -2.10724160e-01 -5.99777937e-01 -5.98079026e-01
-1.07110012e+00 9.41132665e-01 3.64578754e-01 4.37283576e-01
-4.39150929e-01 3.04629236e-01 1.39930412e-01 4.53625530e-01
5.00028133e-01 1.07024097e+00 -6.26307189e-01 -7.15876281e-01
-9.32464182e-01 7.27996588e-01 1.09565473e+00 4.92556095e-01
3.92868906e-01 6.77651688e-02 -2.49599949e-01 6.06638908e-01
1.40613630e-01 4.57649738e-01 5.66947758e-01 -1.53043580e+00
4.16755646e-01 1.10316202e-02 9.33025479e-01 -5.20525932e-01
-2.27356941e-01 -6.20272934e-01 -6.57830834e-02 8.78747463e-01
9.19089258e-01 -7.27641702e-01 -6.17665708e-01 1.86911380e+00
4.71366197e-01 -2.49930650e-01 -1.95421681e-01 1.17225814e+00
-1.64456040e-01 6.29223168e-01 -1.15134373e-01 -3.93481642e-01
9.60943699e-01 -1.05332649e+00 -6.74555123e-01 -4.41033483e-01
4.64471787e-01 -6.95460975e-01 1.44289732e+00 4.72639918e-01
-1.20199955e+00 -1.10796571e-01 -8.89291644e-01 4.08624023e-01
-3.44307333e-01 -6.35730401e-02 8.12146902e-01 4.42615628e-01
-1.20319915e+00 1.13498354e+00 -7.84244955e-01 -3.80018234e-01
3.54607373e-01 2.27337405e-01 9.20400992e-02 3.94029230e-01
-1.37835920e+00 1.42820489e+00 5.88944793e-01 -2.09340766e-01
-9.03608084e-01 -7.40974545e-01 -3.41207176e-01 -4.95804213e-02
9.18030858e-01 -3.61602366e-01 1.63033271e+00 -1.00181830e+00
-2.09591627e+00 2.75945485e-01 -1.44668683e-01 -6.03523731e-01
7.69931734e-01 -4.85387862e-01 -1.05602890e-02 -2.25759387e-01
-2.02416003e-01 6.09063566e-01 9.54238892e-01 -1.24299276e+00
-6.46528542e-01 -2.79984176e-01 1.19583488e-01 6.15186274e-01
-1.42641887e-01 1.19548328e-01 1.30636886e-01 -7.06718862e-01
-3.72270614e-01 -1.06393600e+00 -4.31451321e-01 8.99583548e-02
-3.25262509e-02 -4.27038014e-01 6.50979936e-01 -4.90141541e-01
1.53882742e+00 -1.73855269e+00 8.71262029e-02 2.16610298e-01
1.55947625e-03 5.73177993e-01 1.47241876e-01 4.33362216e-01
1.20234326e-01 3.29338253e-01 -2.35321879e-01 1.36829289e-02
2.12697804e-01 4.59650666e-01 -6.83184206e-01 5.32063842e-01
-4.23084766e-01 9.20613408e-01 -1.44304442e+00 -2.29980633e-01
1.22069061e-01 4.93480302e-02 -6.81195915e-01 -1.46866918e-01
-5.92338443e-01 7.11370409e-01 -6.95273221e-01 5.18876433e-01
2.42468491e-01 3.62408236e-02 9.23765674e-02 6.24802351e-01
-7.97316194e-01 1.23700827e-01 -9.27316904e-01 1.67048919e+00
-2.35722855e-01 5.60702324e-01 2.74141699e-01 -1.15597999e+00
8.01172435e-01 4.11239207e-01 7.54083037e-01 -5.52479506e-01
1.94839925e-01 4.59480941e-01 -9.58077088e-02 -2.01709479e-01
2.99655259e-01 -5.95996194e-02 3.33096951e-01 3.45876813e-01
-1.97869644e-01 -1.56444818e-01 2.72082806e-01 -3.22488397e-01
8.20845842e-01 6.65760040e-01 5.44296980e-01 -8.13384295e-01
2.72073716e-01 1.56085029e-01 7.09403157e-01 1.14734387e+00
-8.26299548e-01 2.61451025e-02 7.39487410e-01 -3.04478705e-01
-9.61401463e-01 -9.35040951e-01 -1.77500919e-01 1.40434253e+00
1.40873827e-02 -1.61384478e-01 -6.46902859e-01 -6.68622434e-01
3.98729384e-01 9.43090200e-01 -7.37929225e-01 1.06069177e-01
-7.82081366e-01 -5.22929013e-01 1.39720023e-01 3.84385735e-01
2.32441828e-01 -1.08501279e+00 -7.42195487e-01 3.81738186e-01
5.69784194e-02 -5.26556849e-01 -6.84789419e-01 2.51466006e-01
-1.01590514e+00 -7.82721877e-01 -1.07540059e+00 -4.00561422e-01
3.90224844e-01 -2.75488913e-01 9.99011755e-01 -4.17782128e-01
2.85990179e-01 2.25298777e-01 -2.74618506e-01 -8.60491872e-01
-2.74760038e-01 -1.79056060e-02 2.61877120e-01 -4.12906170e-01
1.07133783e-01 -6.61911607e-01 -8.93247843e-01 4.47013795e-01
-3.14473778e-01 -2.11910173e-01 2.51862317e-01 7.86828220e-01
5.56003392e-01 -3.79866451e-01 4.07935411e-01 -4.48638499e-01
1.19985199e+00 -5.46063900e-01 -1.14337730e+00 4.27513532e-02
-1.03982508e+00 4.83926654e-01 8.69138598e-01 -7.13037968e-01
-9.60707903e-01 -1.98581204e-01 9.67459660e-03 -6.64178431e-01
1.17499821e-01 8.17944482e-02 5.04567385e-01 -1.93419009e-01
1.08416307e+00 9.06927213e-02 1.25706077e-01 -7.78921187e-01
6.01590395e-01 1.08000189e-01 3.09091598e-01 -1.10771394e+00
5.02002716e-01 5.14646947e-01 -2.52789527e-01 -2.23963618e-01
-8.25541079e-01 -2.58525014e-01 1.30242873e-02 -2.35787228e-01
5.75377166e-01 -2.48280719e-01 -9.15101171e-01 -1.69142604e-01
-7.27762282e-01 -9.18604314e-01 -8.37787509e-01 7.44034827e-01
-8.59235644e-01 1.61868080e-01 -2.40301341e-01 -1.04198813e+00
-2.95460492e-01 -1.22614884e+00 5.81716776e-01 4.76602763e-01
-2.13974059e-01 -1.07486928e+00 4.99585539e-01 -3.44194174e-01
6.55375659e-01 2.04645216e-01 3.30068618e-01 -4.81209338e-01
-3.00045218e-03 1.73557773e-01 2.01900154e-01 3.04876566e-01
-1.46910220e-01 -2.20743507e-01 -6.35115623e-01 -7.34505534e-01
2.82464057e-01 -1.85903832e-01 8.07080090e-01 8.98390770e-01
1.23744524e+00 -7.31288195e-01 -3.64577383e-01 6.92394197e-01
1.38139880e+00 5.80947936e-01 2.27329940e-01 8.42821538e-01
8.88256431e-02 2.48095602e-01 8.12463999e-01 6.08191311e-01
-1.25244139e-02 7.44752109e-01 5.09175301e-01 8.65454972e-02
1.78909317e-01 -5.21625936e-01 6.46774411e-01 2.14141220e-01
-2.60098130e-01 -1.11157097e-01 -8.57877254e-01 5.11591613e-01
-2.24005675e+00 -9.95411634e-01 5.41667104e-01 2.29770827e+00
1.03970468e+00 1.66126400e-01 3.06066275e-01 -3.86164963e-01
5.75769365e-01 2.85700858e-01 -9.19435203e-01 -1.06112170e+00
1.51275992e-01 2.62949795e-01 9.60820735e-01 1.00571597e+00
-1.08444023e+00 1.05262291e+00 6.88617992e+00 9.13034976e-01
-1.21872902e+00 2.23421872e-01 3.61246288e-01 -5.11142731e-01
-7.59280249e-02 4.17498618e-01 -7.39213109e-01 8.80622387e-01
7.32703924e-01 -3.13039303e-01 1.00805330e+00 1.28317750e+00
5.82443535e-01 -2.72771209e-01 -6.53441250e-01 6.01170719e-01
-4.68305260e-01 -1.23951149e+00 -6.14175618e-01 1.78243458e-01
1.00260210e+00 1.81006119e-01 2.97900677e-01 4.95472133e-01
6.94350600e-01 -8.34053040e-01 7.79493034e-01 5.14087260e-01
3.71226549e-01 -7.86078930e-01 1.80709332e-01 5.91842592e-01
-7.54718423e-01 -5.04329145e-01 -3.87941331e-01 -7.95523524e-02
9.44571868e-02 3.73410970e-01 -7.78041363e-01 4.60755825e-03
6.28678620e-01 3.84580791e-01 -5.78722320e-02 1.15465498e+00
-5.79175174e-01 5.62063098e-01 -6.41416430e-01 -3.06444854e-01
7.48297632e-01 -4.13331002e-01 1.21100605e+00 7.23245859e-01
2.98485309e-01 -1.91991046e-01 2.93015927e-01 9.20780122e-01
1.97555944e-01 4.12642434e-02 -5.51235855e-01 -3.68927419e-02
5.53245664e-01 7.41259277e-01 -2.74635911e-01 -2.64731973e-01
1.15122972e-02 6.16989434e-01 3.48909497e-01 7.57610440e-01
-8.25262964e-01 -2.67289907e-01 8.69481623e-01 -8.04334506e-02
2.75294632e-01 -2.87381738e-01 -2.66376436e-01 -1.11840761e+00
1.66401654e-01 -8.26078892e-01 3.03062439e-01 -4.04357612e-01
-7.76788831e-01 1.15814440e-01 3.19970340e-01 -1.01994300e+00
-3.33808035e-01 -5.76458454e-01 -6.48230076e-01 7.56005049e-01
-1.37096417e+00 -1.63812280e-01 4.11090195e-01 2.96140730e-01
4.26883876e-01 -5.33389673e-02 5.07669926e-01 -3.53527740e-02
-3.66109878e-01 4.98088479e-01 9.16694701e-01 -4.95624393e-01
6.29418850e-01 -1.53256702e+00 9.00316685e-02 7.37326443e-01
-3.66715342e-01 5.97212672e-01 1.06515539e+00 -4.82193619e-01
-1.17581570e+00 -5.89833200e-01 5.34981847e-01 -4.62024838e-01
7.71811426e-01 1.02918521e-01 -7.67611206e-01 6.29233718e-01
1.52317405e-01 -1.29085332e-01 1.41558051e-01 7.23003373e-02
2.27258325e-01 1.64486870e-01 -1.12951946e+00 9.35143530e-01
1.05555916e+00 -1.51390284e-01 -5.58939934e-01 6.22698069e-01
5.87435544e-01 -5.77219665e-01 -9.89993691e-01 4.22447205e-01
5.91998398e-01 -8.51508975e-01 1.03150797e+00 -9.29943860e-01
-8.01988225e-03 -1.40461490e-01 5.04317284e-02 -1.71886289e+00
-1.77369952e-01 -1.30694866e+00 -6.55543327e-01 5.61620057e-01
3.93107504e-01 -1.12594676e+00 8.33919048e-01 3.43538254e-01
-1.47496283e-01 -1.61248231e+00 -1.02808750e+00 -1.23786807e+00
6.38884008e-01 -2.59592056e-01 6.28855944e-01 6.94658875e-01
3.91689837e-01 -3.30493897e-01 -3.56907308e-01 -3.20740640e-01
5.40444374e-01 1.34801105e-01 2.38026783e-01 -6.45714223e-01
-5.72776198e-01 -1.02086556e+00 4.58922118e-01 -1.38351643e+00
4.66800183e-02 -7.51200676e-01 5.31264693e-02 -1.33296812e+00
-4.98583555e-01 -6.78844929e-01 -4.48951721e-01 4.82477069e-01
-1.97037891e-01 -4.17622894e-01 4.18005109e-01 3.51290077e-01
-3.62275690e-01 7.27179885e-01 1.81116366e+00 2.94823319e-01
-6.21923327e-01 7.53115565e-02 -7.33293951e-01 9.50869262e-01
1.53994477e+00 -5.05091310e-01 -4.64386910e-01 -3.52255404e-01
1.28864303e-01 1.97652608e-01 5.22729754e-02 -7.14612722e-01
2.18926882e-03 -7.13956416e-01 2.51805067e-01 -1.26650572e-01
3.15599680e-01 -2.63908386e-01 -3.61571252e-01 7.48222947e-01
-4.59725142e-01 9.06180814e-02 1.44426718e-01 3.92508656e-01
2.09333315e-01 -2.96134830e-01 9.60677505e-01 -4.36120838e-01
-7.10157692e-01 2.00753585e-01 -4.96157736e-01 4.18484658e-01
9.18404043e-01 -1.36067510e-01 -4.30623680e-01 -4.48765635e-01
-7.36034989e-01 5.20367324e-01 6.48114920e-01 1.62754133e-01
2.28675157e-01 -1.28608131e+00 -3.91418487e-01 -2.77423024e-01
-4.35880452e-01 -4.25382733e-01 -4.82969910e-01 8.57719719e-01
-3.13158393e-01 7.77699292e-01 -8.50902200e-02 -2.92622805e-01
-6.00464225e-01 6.42371476e-01 6.35145366e-01 -5.21376073e-01
-5.75641692e-01 9.23815250e-01 -1.29889429e-01 -4.74263996e-01
2.51677632e-01 -4.31861073e-01 -1.62134506e-02 -2.43628085e-01
3.04019392e-01 6.23296797e-01 -4.59642798e-01 3.37011218e-02
-3.69964659e-01 2.99515277e-01 2.09306136e-01 -5.61129570e-01
9.82370615e-01 2.26213902e-01 1.39595151e-01 5.71378320e-02
1.05641019e+00 -1.19656526e-01 -1.77171993e+00 -3.38792503e-02
8.95960107e-02 -8.56913269e-01 1.31050304e-01 -9.53843892e-01
-6.45929992e-01 4.82768446e-01 7.07753479e-01 1.02358028e-01
8.44533265e-01 -2.55743861e-01 6.48502588e-01 4.03655440e-01
3.96602660e-01 -1.64350224e+00 -1.46260992e-01 6.95151985e-01
1.26209939e+00 -1.06164157e+00 9.85477790e-02 3.63532096e-01
-7.02146590e-01 7.55985618e-01 5.96534252e-01 -4.24089044e-01
6.46890879e-01 -2.39046116e-04 -9.19998065e-02 1.45033926e-01
-8.44872355e-01 -2.07813576e-01 -1.39159663e-02 4.24210608e-01
1.37238175e-01 2.55481511e-01 -9.12836134e-01 -4.69374359e-02
-4.46033686e-01 1.24766245e-01 3.52889821e-02 1.26132369e+00
-8.42712700e-01 -1.26062417e+00 -5.91405749e-01 2.92020530e-01
-4.31699961e-01 6.99819550e-02 -2.43392870e-01 7.32378125e-01
-3.86704989e-02 8.43091726e-01 -3.00217152e-01 2.33200882e-02
2.11245418e-01 7.80827478e-02 6.17734432e-01 -1.21685408e-01
-5.13594687e-01 -7.29646087e-02 1.72930658e-02 -1.09926867e+00
8.87411460e-02 -4.91831452e-01 -9.81129050e-01 -4.50643063e-01
-3.03668454e-02 5.40653586e-01 7.69568086e-01 7.26929724e-01
2.87349373e-01 1.67483777e-01 7.16074765e-01 -1.07548511e+00
-1.15907085e+00 -6.25784516e-01 -3.07514995e-01 2.23781690e-01
4.99251097e-01 -9.45441902e-01 -3.01121056e-01 -7.40973949e-01]
|
[4.108235836029053, 2.2584352493286133]
|
f0c5eb27-91ef-46ea-ac6a-432b21489b1d
|
acoustic-echo-cancellation-by-combining
|
2005.09237
| null |
https://arxiv.org/abs/2005.09237v1
|
https://arxiv.org/pdf/2005.09237v1.pdf
|
Acoustic Echo Cancellation by Combining Adaptive Digital Filter and Recurrent Neural Network
|
Acoustic Echo Cancellation (AEC) plays a key role in voice interaction. Due to the explicit mathematical principle and intelligent nature to accommodate conditions, adaptive filters with different types of implementations are always used for AEC, giving considerable performance. However, there would be some kinds of residual echo in the results, including linear residue introduced by mismatching between estimation and the reality and non-linear residue mostly caused by non-linear components on the audio devices. The linear residue can be reduced with elaborate structure and methods, leaving the non-linear residue intractable for suppression. Though, some non-linear processing methods have already be raised, they are complicated and inefficient for suppression, and would bring damage to the speech audio. In this paper, a fusion scheme by combining adaptive filter and neural network is proposed for AEC. The echo could be reduced in a large scale by adaptive filtering, resulting in little residual echo. Though it is much smaller than speech audio, it could also be perceived by human ear and would make communication annoy. The neural network is elaborately designed and trained for suppressing such residual echo. Experiments compared with prevailing methods are conducted, validating the effectiveness and superiority of the proposed combination scheme.
|
['Pei Zhao', 'Tengrong Su', 'Hua Huang', 'Lu Ma']
|
2020-05-19
| null | null | null | null |
['acoustic-echo-cancellation', 'acoustic-echo-cancellation']
|
['medical', 'speech']
|
[ 1.59256216e-02 -2.63810039e-01 7.00940132e-01 8.41451064e-02
-2.32901484e-01 -2.48532131e-01 4.48564067e-02 -1.87611341e-01
-3.41051400e-01 4.89436537e-01 5.46954453e-01 -1.09721914e-01
-6.97852746e-02 -4.70891207e-01 -2.00757906e-01 -9.16749597e-01
-2.02412549e-02 -5.35837293e-01 4.98432338e-01 -4.10158277e-01
5.26765920e-02 1.96838841e-01 -1.74440241e+00 1.63749725e-01
1.11202621e+00 9.10719156e-01 4.14106041e-01 2.88442791e-01
-2.96848118e-01 3.32280248e-01 -1.10719848e+00 -2.78897155e-02
1.34791151e-01 -5.76242983e-01 1.69202968e-01 -9.88545865e-02
-1.69779539e-01 -3.82083237e-01 -3.42347831e-01 1.42971098e+00
1.06077695e+00 1.91555843e-01 6.15135491e-01 -7.06035852e-01
-2.47053087e-01 4.87574130e-01 -4.52888489e-01 -2.11838316e-02
3.37032497e-01 3.48835960e-02 2.67833650e-01 -9.65867937e-01
-2.14562956e-02 1.40038049e+00 9.38573122e-01 4.49999422e-01
-8.00214350e-01 -1.15175736e+00 -1.62290707e-01 1.90346897e-01
-1.43027699e+00 -6.12681150e-01 1.04296291e+00 -2.53451597e-02
6.84959292e-01 6.02170587e-01 8.16670358e-01 6.83487535e-01
1.71913967e-01 4.30622637e-01 1.22944105e+00 -6.14954591e-01
-1.06110973e-02 3.74511987e-01 6.32975698e-02 2.80166835e-01
-2.61842739e-02 4.52393144e-01 -8.54588822e-02 -7.65361711e-02
5.21062613e-01 -3.30526638e-03 -1.00429940e+00 3.13750207e-01
-7.00716853e-01 2.54095942e-01 4.41007227e-01 6.25274420e-01
-4.12348628e-01 -4.03363168e-01 4.40884262e-01 2.59319723e-01
2.84804791e-01 2.58223861e-01 -2.44199947e-01 -4.42352742e-02
-7.91825056e-01 2.23462403e-01 7.65024960e-01 5.47763109e-01
2.25152805e-01 6.34728372e-01 3.18636969e-02 1.22031391e+00
4.18394864e-01 7.56553650e-01 6.97712302e-01 -5.72893679e-01
3.66878062e-01 1.11875169e-01 1.28300637e-01 -1.37836456e+00
-5.45883536e-01 -6.95326090e-01 -1.13787401e+00 4.36646640e-01
2.46192738e-01 -4.13962692e-01 -7.46945441e-01 1.54475200e+00
3.13000232e-01 1.88703910e-01 1.85966138e-02 1.23460245e+00
9.18074012e-01 1.22287238e+00 -1.72861591e-01 -7.94781983e-01
1.10491705e+00 -8.41330111e-01 -1.52422440e+00 3.35516147e-02
9.68356133e-02 -1.37357318e+00 9.96991992e-01 6.06318653e-01
-1.02070570e+00 -8.50173175e-01 -1.21438825e+00 3.59891266e-01
-7.98460841e-03 -2.03662559e-01 1.39567688e-01 8.41717660e-01
-8.51317227e-01 4.11429346e-01 -1.74807012e-01 2.58925050e-01
-4.71898347e-01 1.84153780e-01 -7.01364577e-02 3.94567460e-01
-1.43730748e+00 8.25714111e-01 2.48205349e-01 9.15025651e-01
-1.60692245e-01 -4.14061040e-01 -3.38840127e-01 1.44295067e-01
3.60939682e-01 -5.15676811e-02 1.31486404e+00 -9.32894528e-01
-1.63666666e+00 -2.63651550e-01 -9.19477940e-02 8.74685273e-02
5.00090837e-01 -3.90483856e-01 -1.26495993e+00 -1.86937094e-01
-3.04675490e-01 6.65221661e-02 1.11301148e+00 -1.18525946e+00
-5.92056274e-01 -6.14903755e-02 -2.70173967e-01 4.09111351e-01
-3.90716612e-01 -2.62422133e-02 -1.48688152e-01 -8.29585254e-01
4.81131554e-01 -6.55132830e-01 -1.92049891e-01 -2.42020518e-01
-1.76665246e-01 5.31517975e-02 1.07832766e+00 -1.12003732e+00
1.64734709e+00 -2.41017294e+00 -3.52882504e-01 1.92476779e-01
-4.69658002e-02 9.79774594e-01 -3.75181660e-02 4.54721332e-01
-5.86902164e-02 -4.87596802e-02 -1.72271982e-01 1.50745153e-01
-3.97384256e-01 -2.44058073e-01 -2.27996632e-01 3.11551690e-01
-8.77320245e-02 1.68262005e-01 -4.94857311e-01 -2.71105915e-01
2.03461125e-01 7.10092247e-01 -5.43453574e-01 3.53888154e-01
2.78022498e-01 3.39101642e-01 -2.67159641e-01 6.67472720e-01
1.16607285e+00 8.10175180e-01 -2.75207520e-01 -4.20351923e-01
-3.52446973e-01 1.88501969e-01 -1.78763831e+00 9.61203873e-01
-5.55624247e-01 3.44288141e-01 9.02159095e-01 -7.35457301e-01
1.18753767e+00 6.78124368e-01 3.80432680e-02 -8.52444351e-01
1.32564947e-01 5.95079780e-01 4.40320045e-01 -8.60169530e-01
3.54147911e-01 -4.44171876e-01 3.99677634e-01 4.46462594e-02
-4.26129043e-01 -2.36047730e-01 -3.21059346e-01 -3.19495350e-01
6.73448503e-01 -9.20995027e-02 7.25246221e-02 -3.84797454e-01
9.95936811e-01 -5.10172248e-01 9.71357346e-01 3.02691370e-01
-2.78335571e-01 4.02827770e-01 -2.24858329e-01 -1.45541370e-01
-8.50958526e-01 -6.58420920e-01 -1.61035463e-01 6.75763309e-01
4.87947434e-01 -2.71089256e-01 -8.19323361e-01 4.22422262e-03
-2.44790599e-01 5.08094847e-01 2.67440230e-01 -3.95825773e-01
-7.72235513e-01 -4.25792217e-01 5.13447642e-01 1.38969421e-01
5.80344200e-01 -1.41180241e+00 -1.41840398e-01 7.07266569e-01
-3.66761148e-01 -6.75837934e-01 -7.51410306e-01 -3.53389196e-02
-8.39181066e-01 -6.49746835e-01 -9.34361041e-01 -8.59113276e-01
4.10574734e-01 7.19857275e-01 4.35567737e-01 3.53947759e-01
-3.52025665e-02 -1.48108661e-01 -2.70721942e-01 -8.06955993e-01
-4.91536468e-01 -6.53361440e-01 3.97433609e-01 7.51932934e-02
7.08140619e-03 -6.67570114e-01 -7.84365714e-01 3.66687655e-01
-8.61332059e-01 -6.93117008e-02 5.81544995e-01 8.36254120e-01
6.20838404e-02 8.87991190e-01 7.33568788e-01 -4.10089999e-01
1.20280719e+00 -5.25235645e-02 -5.31528711e-01 -2.23930582e-01
-4.99112636e-01 -3.76025200e-01 1.07091391e+00 -7.47289658e-01
-1.56931579e+00 -4.47551131e-01 -6.34530067e-01 -2.44119197e-01
-1.59882486e-01 3.71765554e-01 -4.87140328e-01 -6.74764216e-02
3.66632462e-01 2.35483646e-01 1.45471290e-01 -7.37530053e-01
8.02228227e-03 1.26639450e+00 4.61875439e-01 -3.64546329e-02
8.91431272e-01 -2.01763019e-01 -1.51766032e-01 -1.21638727e+00
-3.77427369e-01 -3.30963641e-01 4.75637428e-02 -4.72272068e-01
4.87873346e-01 -8.20922315e-01 -6.14565849e-01 8.57189894e-01
-1.34930420e+00 3.14024359e-01 2.14207709e-01 1.03067136e+00
2.05573305e-01 4.63343352e-01 -7.96619713e-01 -1.38100982e+00
-5.34481406e-01 -1.15021515e+00 2.78342515e-01 6.87404692e-01
-4.13368829e-02 -6.16652369e-01 -2.24522904e-01 9.19842049e-02
9.22893167e-01 -3.73683810e-01 7.75013506e-01 -2.08816096e-01
-2.89861172e-01 -3.22911769e-01 8.46648663e-02 5.85726738e-01
2.10603103e-01 -4.15119492e-02 -1.21606779e+00 -2.01360524e-01
7.83828735e-01 7.46629089e-02 5.83614111e-01 3.43286484e-01
8.83863628e-01 -5.02526939e-01 -8.24872777e-02 4.40005362e-01
1.26715887e+00 9.92092967e-01 8.63361120e-01 4.14373018e-02
3.38209987e-01 8.00854862e-01 6.45791888e-01 4.80071688e-03
-3.82285416e-01 5.03987491e-01 2.60118783e-01 -4.64326680e-01
-2.27386415e-01 -1.58394530e-01 4.42791551e-01 1.70864248e+00
-1.98800251e-01 -3.34192336e-01 -2.28368595e-01 1.79045647e-01
-1.38463092e+00 -9.37310934e-01 -6.58836305e-01 2.06917691e+00
8.70336115e-01 1.84802905e-01 -1.97036073e-01 6.42720759e-01
1.00076890e+00 1.80727705e-01 -2.79390991e-01 -7.47894526e-01
-2.50663757e-01 2.15777203e-01 2.33807027e-01 6.08391166e-01
-7.33336568e-01 3.58585745e-01 6.02749634e+00 1.25417948e+00
-1.40341055e+00 1.58475265e-02 1.87333778e-01 1.71319321e-01
-3.76398027e-01 -1.91565931e-01 -7.11714089e-01 5.88737905e-01
7.30651200e-01 3.78393792e-02 4.34591115e-01 5.23988783e-01
4.58762050e-01 -6.13254234e-02 -2.73090422e-01 1.03742564e+00
-2.02542767e-01 -5.52275717e-01 -2.38755003e-01 -1.77961066e-01
3.68078440e-01 -6.58371150e-01 -5.04678711e-02 4.42869008e-01
-5.41086018e-01 -7.22105026e-01 6.95300221e-01 5.32616258e-01
4.68498796e-01 -8.73147070e-01 9.75882113e-01 7.27080643e-01
-1.24800134e+00 -1.33997381e-01 -6.29892230e-01 -3.29805374e-01
3.25249195e-01 9.56981421e-01 -3.71654540e-01 5.62112927e-01
6.40244603e-01 6.21204935e-02 -1.37416888e-02 1.33401489e+00
-2.18290284e-01 7.98114240e-01 -5.14653444e-01 -3.08986276e-01
-3.11886780e-02 -5.53069770e-01 9.74153161e-01 1.27060497e+00
6.97311282e-01 4.14784253e-01 -1.40278250e-01 6.93530798e-01
2.62214124e-01 5.77786505e-01 -6.58737719e-01 1.26154333e-01
5.73716342e-01 1.20484209e+00 -2.79012620e-01 -2.34695613e-01
-5.26332259e-01 7.89901078e-01 -3.84460121e-01 4.89975929e-01
-8.08312714e-01 -9.70774472e-01 2.48415694e-01 2.02534348e-01
-4.77509312e-02 5.82336038e-02 -6.97536068e-03 -7.41727829e-01
3.02111685e-01 -9.73131835e-01 -1.68327481e-01 -6.83925509e-01
-1.06246519e+00 7.55233586e-01 -3.78604531e-01 -1.55718863e+00
-8.46079662e-02 -4.28918719e-01 -9.89069819e-01 1.23157811e+00
-1.09888256e+00 -3.15761447e-01 -3.50560024e-02 4.24371094e-01
6.41601682e-01 -1.64273679e-01 7.91695416e-01 7.44669974e-01
-4.32179958e-01 6.45767033e-01 3.40659112e-01 -4.24148500e-01
8.91229212e-01 -7.90720344e-01 -3.49942267e-01 8.79571140e-01
-3.01623523e-01 6.65378273e-01 9.27710533e-01 -6.24772131e-01
-1.06312680e+00 -6.12103522e-01 8.25698018e-01 5.91757417e-01
3.25921625e-01 -3.06296080e-01 -1.36086810e+00 -2.31533349e-01
3.93421501e-01 -4.09137905e-01 3.36042255e-01 -9.02874023e-02
7.15923980e-02 -4.69731271e-01 -1.05242717e+00 7.71561444e-01
7.55465746e-01 -4.38008875e-01 -7.62456656e-01 -8.95658806e-02
8.61032128e-01 -2.81511307e-01 -5.42254627e-01 7.09561944e-01
6.12195194e-01 -1.31660044e+00 8.14348757e-01 1.07270174e-01
1.15987398e-01 -7.29852915e-01 1.52238846e-01 -1.42441440e+00
-5.31622112e-01 -7.57040381e-01 2.81054497e-01 1.73882747e+00
3.99780184e-01 -8.97425354e-01 2.20405772e-01 3.59962791e-01
-4.73754376e-01 -4.18332487e-01 -7.22651780e-01 -8.41129899e-01
-2.94316202e-01 -2.79539526e-01 4.13269848e-01 5.70357084e-01
8.07055384e-02 5.37301421e-01 -7.52136648e-01 2.97524571e-01
3.86331886e-01 -2.28702545e-01 4.41799670e-01 -1.04538703e+00
-2.92275429e-01 -4.28064793e-01 -1.19825108e-02 -1.19561410e+00
-3.35314363e-01 -3.21588993e-01 4.77443695e-01 -1.22626698e+00
-3.39297295e-01 -4.33726430e-01 -3.88138473e-01 -1.03706151e-01
-4.45447057e-01 3.01508624e-02 1.80754796e-01 2.07168162e-01
2.36382440e-01 8.04385245e-01 1.60713875e+00 -5.29812090e-02
-4.65050131e-01 3.19083542e-01 -4.19782668e-01 1.03747022e+00
6.61302805e-01 -3.20356935e-01 -4.35339808e-01 -1.96904436e-01
-3.20400566e-01 5.97760439e-01 -3.44959497e-02 -1.11933088e+00
4.32427853e-01 2.41307095e-01 3.42138469e-01 -7.77988672e-01
5.42185068e-01 -1.23144472e+00 4.45414424e-01 6.41512930e-01
3.06504257e-02 -1.92449808e-01 3.42015386e-01 4.18062508e-01
-6.61132574e-01 -4.30512846e-01 8.50780666e-01 -3.99713516e-02
-3.49975318e-01 -2.01427832e-01 -5.77069581e-01 -3.96395624e-01
4.94772702e-01 -4.21321720e-01 -3.40434052e-02 -7.44869351e-01
-5.36760628e-01 -3.51400152e-02 -1.59416273e-01 2.63610959e-01
6.36366129e-01 -1.38374507e+00 -5.57992339e-01 4.76284415e-01
-4.54183042e-01 -1.51210651e-01 7.66674817e-01 8.20701897e-01
-3.84276211e-01 3.99594486e-01 -2.13710777e-02 -2.34124109e-01
-1.32401907e+00 4.93418902e-01 3.53756428e-01 7.59759843e-02
-7.55390048e-01 6.74344659e-01 4.85684007e-01 -3.54929090e-01
5.38364947e-01 -8.48979950e-02 -5.19858301e-01 -9.35774520e-02
8.38074028e-01 4.92717564e-01 -2.21014749e-02 -6.23542666e-01
-1.16589345e-01 7.73602068e-01 1.49685457e-01 -1.69810578e-01
1.01814163e+00 -3.62956822e-01 -1.81053251e-01 3.45935762e-01
1.00824893e+00 8.56380761e-01 -7.72758126e-01 -9.96540412e-02
-4.15283471e-01 -4.81084943e-01 1.52065694e-01 -8.81867826e-01
-9.78592336e-01 9.63439345e-01 9.04713869e-01 4.19449180e-01
1.54852545e+00 -7.94363678e-01 1.08778000e+00 4.15515080e-02
2.90819108e-01 -1.19206238e+00 -1.48064420e-01 3.10591340e-01
1.06280744e+00 -8.28697562e-01 -2.16742381e-01 -7.15951681e-01
-2.99818337e-01 1.22677314e+00 6.93310142e-01 -1.26665294e-01
7.18826294e-01 5.24297357e-01 4.75155711e-01 4.03122962e-01
-2.90051073e-01 1.03300087e-01 2.63047695e-01 6.22853577e-01
5.87687671e-01 -1.55002370e-01 -8.74688983e-01 9.56052125e-01
-3.41837257e-01 -1.80573761e-01 4.41084206e-01 6.37663960e-01
-7.81538367e-01 -1.06218207e+00 -9.65232968e-01 2.16351479e-01
-7.82125294e-01 -4.66912389e-02 1.65067334e-02 6.35384023e-01
3.57984334e-01 1.50016284e+00 -3.16126794e-02 -6.44125640e-01
7.50681281e-01 -5.85314296e-02 9.87043679e-02 1.84650943e-02
-5.87036133e-01 1.02524197e+00 8.88643712e-02 -2.37271979e-01
-1.54839501e-01 -2.24981651e-01 -1.34845173e+00 -1.83539346e-01
-9.17349517e-01 4.80127215e-01 5.48424661e-01 6.46031976e-01
-3.72293741e-02 8.24558139e-01 7.55115271e-01 -6.90262675e-01
-7.18394339e-01 -1.33084321e+00 -1.01413429e+00 3.78083974e-01
4.21758324e-01 -4.78661716e-01 -7.67989218e-01 -3.39248836e-01]
|
[15.10736083984375, 5.949859619140625]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.