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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
946a1734-e488-4568-8ddd-61fa1ea2d736
|
free-headgan-neural-talking-head-synthesis
|
2208.02210
| null |
https://arxiv.org/abs/2208.02210v1
|
https://arxiv.org/pdf/2208.02210v1.pdf
|
Free-HeadGAN: Neural Talking Head Synthesis with Explicit Gaze Control
|
We present Free-HeadGAN, a person-generic neural talking head synthesis system. We show that modeling faces with sparse 3D facial landmarks are sufficient for achieving state-of-the-art generative performance, without relying on strong statistical priors of the face, such as 3D Morphable Models. Apart from 3D pose and facial expressions, our method is capable of fully transferring the eye gaze, from a driving actor to a source identity. Our complete pipeline consists of three components: a canonical 3D key-point estimator that regresses 3D pose and expression-related deformations, a gaze estimation network and a generator that is built upon the architecture of HeadGAN. We further experiment with an extension of our generator to accommodate few-shot learning using an attention mechanism, in case more than one source images are available. Compared to the latest models for reenactment and motion transfer, our system achieves higher photo-realism combined with superior identity preservation, while offering explicit gaze control.
|
['Stefanos Zafeiriou', 'Viktoriia Sharmanska', 'Evangelos Ververas', 'Michail Christos Doukas']
|
2022-08-03
| null | null | null | null |
['gaze-estimation']
|
['computer-vision']
|
[ 1.01658534e-02 8.49557698e-01 2.80398101e-01 -5.98998189e-01
-7.66048670e-01 -4.20420438e-01 9.26327527e-01 -1.01255119e+00
-6.53211996e-02 5.01175821e-01 4.24417585e-01 3.39548886e-01
5.91509581e-01 -3.05891633e-01 -8.57153714e-01 -6.17538750e-01
2.43238404e-01 6.39999568e-01 -2.43770316e-01 -2.73879856e-01
-5.65566830e-02 5.53510845e-01 -1.90786111e+00 -9.14290845e-02
2.56204844e-01 1.01606631e+00 -3.04918438e-01 8.18188787e-01
2.48399049e-01 6.26521289e-01 -3.32686692e-01 -5.36610305e-01
3.39853644e-01 -8.50003779e-01 -6.82352066e-01 5.12538075e-01
9.64332461e-01 -5.52768707e-01 -1.39328435e-01 6.58259869e-01
7.50467658e-01 8.81776363e-02 6.35232389e-01 -1.50202799e+00
-7.83889413e-01 5.42908013e-02 -6.32586777e-01 -4.59636360e-01
7.57597089e-01 5.51685631e-01 7.32141495e-01 -1.09523475e+00
1.04438686e+00 1.54908514e+00 6.36092663e-01 1.22026169e+00
-1.41918695e+00 -6.74941659e-01 -2.85456218e-02 4.93288338e-02
-1.45439053e+00 -1.34323645e+00 9.01465476e-01 -3.91395152e-01
6.70220315e-01 1.07053411e-03 8.35474968e-01 1.52839994e+00
-2.01142859e-02 4.10721451e-01 1.03076851e+00 -4.43084061e-01
9.08805430e-02 1.84795529e-01 -6.03517711e-01 8.20570588e-01
-3.73072058e-01 6.49373680e-02 -8.86063516e-01 1.43133074e-01
8.23788822e-01 -4.88862723e-01 -3.46910715e-01 -6.92558110e-01
-8.11524332e-01 5.94401479e-01 4.74439085e-01 -1.23527430e-01
-4.04777706e-01 2.60559350e-01 -2.26408646e-01 1.02220312e-01
6.09454811e-01 2.62433618e-01 -2.81860549e-02 2.96856407e-02
-1.29979563e+00 3.77048999e-01 9.03182745e-01 1.25845933e+00
1.07791984e+00 2.74387717e-01 -2.38214120e-01 2.70119995e-01
4.99188304e-01 6.47288024e-01 2.32673764e-01 -1.51655614e+00
-2.02843994e-01 3.46658587e-01 -8.36487934e-02 -6.73540711e-01
-2.83907145e-01 -4.24484089e-02 -5.46941876e-01 5.89484572e-01
2.36349404e-01 -2.72719383e-01 -1.07361102e+00 2.28868246e+00
5.18490613e-01 3.78132284e-01 -1.65082499e-01 9.88358140e-01
7.81449974e-01 2.87429482e-01 -9.16192532e-02 -1.32715300e-01
1.38743138e+00 -7.28482723e-01 -6.19619727e-01 -3.03275973e-01
1.85259715e-01 -5.56376874e-01 1.02002704e+00 9.89429131e-02
-1.62315428e+00 -4.65619832e-01 -7.32851088e-01 -6.04341328e-01
-5.49455881e-02 -1.45096809e-01 3.77695203e-01 7.01577783e-01
-1.77995169e+00 3.85805666e-01 -7.31229842e-01 -6.64750278e-01
5.33391893e-01 5.90858400e-01 -7.27353215e-01 1.47190094e-01
-7.53275812e-01 1.21111023e+00 -2.57419258e-01 2.50752922e-02
-1.00389993e+00 -7.65063584e-01 -1.21372318e+00 2.06733821e-03
3.58750522e-02 -1.37299716e+00 1.24933171e+00 -1.23972154e+00
-2.18962860e+00 1.40211546e+00 -5.93770742e-01 -7.16631040e-02
6.93556309e-01 -9.76644754e-02 1.43332612e-02 7.75250271e-02
-2.30598941e-01 1.35316753e+00 1.31396711e+00 -1.34106231e+00
6.37258813e-02 -5.91682494e-01 -7.33722448e-02 3.83519351e-01
-1.04450382e-01 1.18812621e-01 -6.37593627e-01 -1.91727415e-01
-2.85478413e-01 -1.31218565e+00 1.85647264e-01 4.38000232e-01
-3.59886885e-01 4.19944758e-03 8.93032074e-01 -8.25298369e-01
4.53383982e-01 -1.97692704e+00 6.13826096e-01 1.52471170e-01
2.66937733e-01 -9.66658443e-02 -1.51229441e-01 7.80818388e-02
-2.41750062e-01 -1.38598666e-01 -1.57791495e-01 -1.24749386e+00
1.81017280e-01 -4.32631969e-02 -3.30371074e-02 7.45765209e-01
5.42017341e-01 1.21415436e+00 -4.11972880e-01 -3.98383915e-01
1.35698110e-01 1.02877092e+00 -9.00114119e-01 4.38768119e-01
-3.12859476e-01 9.39514756e-01 6.41807988e-02 4.61238682e-01
6.50087535e-01 -2.90508121e-01 -3.77908051e-02 -1.27575308e-01
9.51153636e-02 5.61627634e-02 -9.19284225e-01 2.24023151e+00
-5.18603683e-01 6.86654270e-01 5.30723572e-01 -1.61497757e-01
8.92923474e-01 4.52485055e-01 2.83325702e-01 -2.36060366e-01
5.13476074e-01 -1.71652853e-01 -2.28710324e-01 -3.36620241e-01
3.04862678e-01 -4.33485240e-01 1.08916737e-01 5.72419226e-01
6.56160891e-01 -4.91806656e-01 -2.62253255e-01 9.13722664e-02
6.48431003e-01 8.05908263e-01 -2.68651489e-02 -1.74834818e-01
4.90656972e-01 -6.45863831e-01 1.92092553e-01 -1.08253777e-01
1.03407182e-01 1.02972770e+00 5.60218751e-01 -2.59235930e-02
-1.27512383e+00 -8.54930341e-01 1.48349881e-01 1.05817807e+00
-2.29055092e-01 -3.41182262e-01 -1.23567784e+00 -2.44197249e-01
-1.93893507e-01 8.80487859e-01 -9.86653030e-01 -1.98529482e-01
-4.13293272e-01 -1.11026570e-01 6.09185219e-01 2.29924932e-01
2.23721415e-01 -1.00985253e+00 -5.89867234e-01 -3.69247884e-01
-1.84524478e-03 -1.04825103e+00 -7.36111641e-01 -4.94380772e-01
-3.37485284e-01 -8.35298300e-01 -9.92533445e-01 -4.44566160e-01
8.16320956e-01 -2.45344073e-01 1.09228456e+00 -1.54668048e-01
-1.58082977e-01 5.32211065e-01 1.25298038e-01 -4.39983875e-01
-4.75796610e-01 1.72001030e-02 2.16027513e-01 4.50337946e-01
1.13679692e-01 -9.09367561e-01 -6.59103453e-01 3.00041568e-02
-4.90974933e-01 1.79963693e-01 3.73592436e-01 5.28039277e-01
1.65754661e-01 -9.60993648e-01 1.31204560e-01 -6.81893766e-01
2.40215793e-01 -2.15802982e-01 -4.99715030e-01 5.00940494e-02
-4.78392035e-01 1.72420710e-01 1.01611957e-01 -3.03155899e-01
-1.39481616e+00 4.37318474e-01 -2.79797584e-01 -8.60016882e-01
-3.83792728e-01 -3.53711069e-01 -5.85632801e-01 -3.13455701e-01
8.30680728e-01 9.17846635e-02 5.89834154e-01 -4.19425160e-01
9.66928244e-01 4.74095076e-01 7.19162226e-01 -3.13112468e-01
9.28252637e-01 6.00604415e-01 2.82700896e-01 -7.74630368e-01
-5.70226789e-01 1.10910051e-02 -1.13461792e+00 -2.54280478e-01
1.05699456e+00 -1.16476429e+00 -1.10713112e+00 6.74800217e-01
-1.27337837e+00 -3.72312456e-01 -4.10732865e-01 1.31878853e-01
-1.05048835e+00 -5.26996441e-02 -4.40439880e-01 -6.69953108e-01
-4.47567075e-01 -1.04835737e+00 1.57876813e+00 2.22310409e-01
-5.96285045e-01 -8.61204982e-01 1.31518871e-01 2.89508611e-01
5.01185238e-01 4.68697250e-01 3.50817800e-01 -1.48088008e-01
-6.63408577e-01 -4.32150699e-02 1.43686816e-01 1.92831919e-01
-1.19142488e-01 1.74136966e-01 -1.54420602e+00 -4.14678484e-01
-1.68518443e-02 -4.91872728e-01 5.43038070e-01 3.02996963e-01
6.32520854e-01 -5.34792721e-01 -1.39227390e-01 1.24807966e+00
8.41472089e-01 -3.39457721e-01 7.74727821e-01 -2.02154666e-01
1.03144217e+00 1.01131296e+00 -2.78990477e-01 2.85104364e-01
7.63235867e-01 8.30589116e-01 5.10213315e-01 -2.14329645e-01
-4.65120673e-01 -4.84744340e-01 5.63391685e-01 3.52152884e-01
-2.93848723e-01 -1.42690958e-02 -6.69859946e-01 3.00870925e-01
-1.36633050e+00 -9.59108114e-01 3.62816691e-01 2.05289102e+00
9.56824660e-01 -3.88467669e-01 3.50752741e-01 -2.59901702e-01
4.40396398e-01 1.33218914e-01 -7.47841477e-01 -2.21809193e-01
-1.46987736e-01 4.07036811e-01 1.30167812e-01 6.80929840e-01
-6.76465988e-01 1.12612534e+00 6.29763126e+00 6.46519810e-02
-1.32755589e+00 1.25747263e-01 5.00068903e-01 -4.55376863e-01
-4.24740434e-01 1.54388202e-02 -8.35758328e-01 2.08761200e-01
1.04054773e+00 -6.93047494e-02 4.72196698e-01 7.32741416e-01
1.86484773e-02 2.95850694e-01 -1.39005148e+00 1.03616464e+00
7.29873896e-01 -1.28271616e+00 -2.28595473e-02 4.34891135e-01
6.09831214e-01 -9.68253762e-02 2.79374391e-01 -2.08736397e-02
2.49209106e-01 -1.22596717e+00 1.00656867e+00 9.33738768e-01
1.44464755e+00 -6.28845394e-01 1.85486332e-01 2.70573616e-01
-5.25998354e-01 2.37272710e-01 1.71294555e-01 2.32490584e-01
3.73134851e-01 -1.17538251e-01 -8.85837793e-01 3.08013380e-01
5.33306360e-01 7.04030275e-01 -5.18743455e-01 4.69621211e-01
-3.92645627e-01 5.71011305e-02 -3.84993285e-01 4.76924986e-01
-3.45155329e-01 3.54669802e-02 6.20598555e-01 8.94181132e-01
5.63608348e-01 1.17911294e-01 -3.11672658e-01 1.19072592e+00
-2.80262768e-01 -1.79729059e-01 -5.99806547e-01 4.91666585e-01
2.76008338e-01 1.35701489e+00 -9.44210738e-02 -9.85291675e-02
-1.80509612e-01 1.27868056e+00 2.62570202e-01 2.77411699e-01
-6.29327714e-01 1.01648323e-01 9.74598348e-01 5.70149362e-01
2.66709805e-01 -1.53283685e-01 -1.31576350e-02 -1.31581187e+00
-2.21877366e-01 -8.62148404e-01 -1.89418167e-01 -1.34398711e+00
-9.63327169e-01 7.51718938e-01 -2.34425329e-02 -8.07375431e-01
-9.28214312e-01 -3.72238547e-01 -7.57123768e-01 1.27168393e+00
-1.30663788e+00 -1.98418939e+00 -5.88817954e-01 1.00913882e+00
1.81862593e-01 -8.89574066e-02 1.00427449e+00 -7.92863294e-02
-4.85035032e-01 8.27658176e-01 -8.57030690e-01 -8.84279981e-02
1.05050385e+00 -1.01114333e+00 7.36369550e-01 7.18128443e-01
8.57281238e-02 6.91434979e-01 8.03346813e-01 -2.49384761e-01
-1.59668446e+00 -8.57653260e-01 7.87580311e-01 -9.02034104e-01
2.40093842e-01 -8.32460880e-01 -7.12104380e-01 1.10911584e+00
5.33107042e-01 4.55602519e-02 5.53795040e-01 4.32684533e-02
-4.08211976e-01 -8.66374299e-02 -1.15923119e+00 7.12925732e-01
1.31964374e+00 -6.58414900e-01 -3.83370340e-01 -4.75156121e-02
5.46163917e-01 -6.82085991e-01 -7.21897900e-01 1.79350916e-02
6.81823194e-01 -1.08042705e+00 7.75386035e-01 -4.56316680e-01
2.68869251e-01 -2.23569900e-01 2.30155334e-01 -1.35608304e+00
-3.40937406e-01 -1.37568581e+00 -2.61049539e-01 1.40281117e+00
1.88739032e-01 -3.38650584e-01 9.09853935e-01 1.01813877e+00
-4.17166874e-02 -1.98672652e-01 -9.12988961e-01 -2.36045882e-01
2.46771574e-02 -1.17613465e-01 7.76839435e-01 7.71650016e-01
-1.60360098e-01 8.10952783e-01 -7.29087174e-01 -2.66090855e-02
7.24463999e-01 -2.20705345e-01 1.34865212e+00 -1.20231032e+00
-2.98751801e-01 -3.69644552e-01 -3.99180233e-01 -8.99367154e-01
8.17375302e-01 -7.86584437e-01 6.30078241e-02 -1.05577159e+00
-2.23469473e-02 1.04026467e-01 3.57190132e-01 6.57067180e-01
1.47031531e-01 5.38536131e-01 4.05763149e-01 2.06430823e-01
-2.15967387e-01 7.51695037e-01 1.33301008e+00 3.90517801e-01
-2.07449719e-01 -1.44761696e-01 -8.12113702e-01 8.22611153e-01
2.73202032e-01 -8.85204747e-02 -4.89005387e-01 -3.70865315e-01
3.32012586e-02 1.57630071e-01 6.98177397e-01 -8.34739268e-01
3.54175150e-01 1.33713648e-01 6.44932747e-01 -9.18366686e-02
9.71947432e-01 -6.34760439e-01 5.53842366e-01 -1.52320564e-01
-2.29264110e-01 -2.78629139e-02 6.20634966e-02 2.87872374e-01
1.80383891e-01 3.92538816e-01 9.86817658e-01 -9.23645198e-02
-3.96124154e-01 5.95272899e-01 7.56375641e-02 -1.04095988e-01
9.34566617e-01 -3.02804857e-01 -2.01352954e-01 -9.15460706e-01
-9.87151921e-01 -5.89729100e-02 1.15205753e+00 5.09538651e-01
4.03285742e-01 -1.31298518e+00 -9.35612977e-01 7.28335440e-01
7.66773969e-02 2.57874783e-02 2.13983715e-01 7.44515717e-01
-2.47039661e-01 8.88994560e-02 -4.86485839e-01 -6.63157880e-01
-1.24840736e+00 5.24735212e-01 5.46149731e-01 3.97680014e-01
-5.77451408e-01 1.04744053e+00 3.96521568e-01 -2.97752231e-01
7.81317353e-02 1.16534039e-01 1.09099038e-01 8.52003843e-02
5.30234396e-01 1.04543328e-01 -2.18610037e-02 -1.38780022e+00
-4.27507669e-01 8.59756470e-01 3.73392999e-01 -5.88365316e-01
1.35242665e+00 -3.78246278e-01 -2.00516969e-01 3.45113158e-01
1.28264308e+00 3.68759558e-02 -1.86703932e+00 -8.66666660e-02
-5.95246077e-01 -3.19632977e-01 -9.46405083e-02 -5.92539787e-01
-1.07771134e+00 9.52810943e-01 3.71833771e-01 -4.79787767e-01
1.11110234e+00 2.57379323e-01 6.18772030e-01 -2.20936202e-02
3.15011114e-01 -5.75101256e-01 -2.29832847e-02 4.05658752e-01
1.17259049e+00 -1.05782902e+00 -1.66820601e-01 -1.99750155e-01
-7.54858494e-01 7.76481271e-01 5.48414350e-01 -1.99551761e-01
7.80399024e-01 1.58223033e-01 2.03009352e-01 -1.60896122e-01
-9.34232712e-01 -2.09287748e-01 5.99655330e-01 9.05156732e-01
3.61996144e-01 -2.72192508e-01 4.57558334e-01 2.11212009e-01
-7.93540597e-01 6.77116662e-02 3.66409212e-01 3.60840380e-01
1.77300833e-02 -9.58235681e-01 -2.25464627e-01 -2.83933699e-01
-2.77859598e-01 -3.37413549e-02 -5.04210830e-01 8.28845322e-01
1.16720036e-01 5.40728867e-01 4.20774221e-01 -2.85769761e-01
4.12036300e-01 5.10231912e-01 8.67637515e-01 -7.10196793e-01
-3.87232572e-01 2.00339183e-01 -5.39838374e-02 -8.95175934e-01
-5.35989940e-01 -8.69004071e-01 -8.55976343e-01 -6.29909635e-01
-6.70742467e-02 -4.17913437e-01 7.16791034e-01 7.51423180e-01
7.54060090e-01 1.62575036e-01 4.87741441e-01 -1.70565510e+00
-2.18193129e-01 -1.03422344e+00 -5.46558857e-01 4.99609560e-01
6.64026320e-01 -6.56024575e-01 -3.57770920e-01 6.26415551e-01]
|
[12.827779769897461, -0.29549139738082886]
|
61b10d7d-1a91-4ae2-995c-71e129055a17
|
pmct-patched-multi-condition-training-for
|
2207.04949
| null |
https://arxiv.org/abs/2207.04949v1
|
https://arxiv.org/pdf/2207.04949v1.pdf
|
pMCT: Patched Multi-Condition Training for Robust Speech Recognition
|
We propose a novel Patched Multi-Condition Training (pMCT) method for robust Automatic Speech Recognition (ASR). pMCT employs Multi-condition Audio Modification and Patching (MAMP) via mixing {\it patches} of the same utterance extracted from clean and distorted speech. Training using patch-modified signals improves robustness of models in noisy reverberant scenarios. Our proposed pMCT is evaluated on the LibriSpeech dataset showing improvement over using vanilla Multi-Condition Training (MCT). For analyses on robust ASR, we employed pMCT on the VOiCES dataset which is a noisy reverberant dataset created using utterances from LibriSpeech. In the analyses, pMCT achieves 23.1% relative WER reduction compared to the MCT.
|
['Mete Ozay', 'Karthikeyan Saravanan', 'Agnieszka Dobrowolska', 'Pablo Peso Parada']
|
2022-07-11
| null | null | null | null |
['robust-speech-recognition']
|
['speech']
|
[ 5.15461266e-01 -1.61402375e-01 7.21973300e-01 -7.90301263e-02
-1.62345409e+00 -4.79878128e-01 6.14886701e-01 -3.08174640e-01
-3.22109789e-01 4.13917303e-01 6.82184935e-01 -5.22799611e-01
-1.07303180e-01 8.04361627e-02 -7.26628482e-01 -7.85558105e-01
8.80686417e-02 -4.65617329e-03 -7.52466172e-02 -2.83142477e-01
-1.53220057e-01 2.92218894e-01 -1.43784857e+00 8.78386557e-01
5.71012318e-01 7.48983145e-01 4.89066571e-01 1.34707737e+00
4.64581847e-01 5.53641558e-01 -1.25740159e+00 -1.18436091e-01
2.91148335e-01 -4.24055576e-01 -4.94627535e-01 -7.90670589e-02
4.68398809e-01 1.44584060e-01 -4.06587601e-01 7.11249352e-01
1.10828757e+00 6.97382689e-01 4.95430082e-01 -7.00871408e-01
-1.04977012e-01 1.07625484e+00 -3.06003302e-01 5.17223120e-01
6.16218925e-01 -1.04153700e-01 7.22725391e-01 -1.14144135e+00
2.20022038e-01 1.41934347e+00 6.49515867e-01 5.42499661e-01
-1.20022523e+00 -6.75542235e-01 -1.57815754e-01 4.68080133e-01
-1.62165201e+00 -1.38757372e+00 6.88702166e-01 -6.06081188e-02
1.43926227e+00 1.03616941e+00 7.83936158e-02 1.35474575e+00
-2.68425077e-01 7.23537862e-01 1.21136045e+00 -7.92087495e-01
1.32005394e-01 -1.67980954e-01 -5.03586866e-02 -7.87772164e-02
-7.01928496e-01 4.72353935e-01 -7.80200720e-01 -9.67960507e-02
1.64763540e-01 -8.32034409e-01 -5.49995840e-01 7.58903146e-01
-1.27261496e+00 2.27024302e-01 -1.49188846e-01 6.02652550e-01
-4.36441988e-01 -1.08531654e-01 5.12434065e-01 7.10001707e-01
5.48945546e-01 3.92382622e-01 -5.20197392e-01 -4.53587592e-01
-1.36968565e+00 8.38403478e-02 5.21239817e-01 1.02175272e+00
-2.52742860e-02 8.35804999e-01 -5.01251876e-01 1.56864798e+00
3.54310840e-01 8.31854820e-01 6.58702016e-01 -6.15392089e-01
7.19542444e-01 -4.51299459e-01 -8.40778202e-02 -3.99186820e-01
-1.05059803e-01 -7.57271647e-01 -6.87907279e-01 -2.26994306e-01
2.48188227e-02 -3.07854474e-01 -1.29621112e+00 1.52713716e+00
2.48047560e-01 4.22723383e-01 4.67959195e-01 7.20779538e-01
9.06725287e-01 1.39738941e+00 -4.65161830e-01 -5.27902603e-01
9.22455609e-01 -1.07337630e+00 -1.11446726e+00 1.63223729e-01
3.94384921e-01 -1.38164294e+00 9.43612397e-01 9.06024098e-01
-1.12292182e+00 -8.10101688e-01 -1.10911489e+00 3.17500651e-01
-9.91967767e-02 3.30257744e-01 -6.23767495e-01 1.26662600e+00
-1.09293342e+00 3.41422290e-01 -4.59159553e-01 -7.13604363e-03
-1.92237020e-01 8.82707164e-02 -5.25848091e-01 -1.73272058e-01
-1.06153381e+00 8.21361005e-01 7.30483308e-02 1.50632992e-01
-1.36615539e+00 -7.27965891e-01 -9.55763698e-01 -5.22451326e-02
3.20741653e-01 2.04775676e-01 1.39613330e+00 -5.04194319e-01
-2.13332152e+00 2.21675649e-01 -5.21921098e-01 -7.07984984e-01
2.67265826e-01 -3.93053770e-01 -1.16479933e+00 1.16928533e-01
-4.34997946e-01 -3.07008680e-02 1.39585030e+00 -1.24869895e+00
-1.51282504e-01 2.43426397e-01 -7.59015620e-01 2.44451985e-02
2.15704203e-01 5.38415015e-01 -1.16251130e-02 -1.19625342e+00
2.63793170e-01 -6.96169555e-01 1.82146996e-01 -1.08562922e+00
-6.51422739e-01 9.25493091e-02 1.04184508e+00 -1.52457738e+00
1.38551164e+00 -2.53771234e+00 1.37793049e-01 3.19621176e-01
-5.50990105e-01 9.56396699e-01 -4.31710035e-01 4.51662779e-01
-3.89979392e-01 7.57123306e-02 -1.79929540e-01 -7.36019313e-01
1.28440619e-01 1.72657296e-01 -4.95538086e-01 3.80609781e-01
1.76196873e-01 3.71418566e-01 -4.39974815e-01 -1.55815840e-01
4.84589309e-01 7.37728596e-01 -4.76116955e-01 4.58187670e-01
6.45455644e-02 6.71824276e-01 3.89649659e-01 6.04179263e-01
6.74778283e-01 9.44650173e-01 -2.36312002e-01 -1.30226895e-01
-2.79229790e-01 7.43064940e-01 -1.33829820e+00 1.41808259e+00
-6.97841585e-01 7.34278500e-01 6.10806048e-01 -7.81457603e-01
1.06599617e+00 1.12026751e+00 -1.49384245e-01 -7.45902300e-01
1.57281488e-01 3.70786697e-01 3.10927421e-01 -4.15901124e-01
5.05526721e-01 -2.83130914e-01 2.32385993e-01 1.04564592e-01
3.63851070e-01 -5.69218636e-01 -2.30587035e-01 8.18002671e-02
1.37012565e+00 -1.96230814e-01 1.97698683e-01 -1.39189363e-01
8.45355451e-01 -7.10258365e-01 4.84505236e-01 7.45160460e-01
-2.82577008e-01 1.01854479e+00 -1.17883526e-01 4.37779486e-01
-1.04422414e+00 -1.25411391e+00 -2.75280207e-01 9.90833819e-01
-7.84455597e-01 -5.56338966e-01 -7.44362473e-01 -2.14130923e-01
-5.05458713e-01 1.15877867e+00 -2.22726718e-01 1.25174031e-01
-8.71291101e-01 -4.73131299e-01 1.07694757e+00 7.74893463e-02
2.92169243e-01 -8.43310177e-01 2.86793083e-01 2.93309271e-01
-3.93254876e-01 -1.35625124e+00 -7.18711138e-01 3.57552409e-01
-8.01668763e-02 -3.81189942e-01 -8.19629133e-01 -7.08538592e-01
1.26019672e-01 2.99868375e-01 5.87161243e-01 -3.01720828e-01
9.15048569e-02 3.51447672e-01 -8.97330523e-01 -4.29341912e-01
-1.14582849e+00 -3.34652752e-01 4.91692722e-01 3.97372991e-01
-3.73183280e-01 -7.37679541e-01 -2.47792620e-02 6.78344727e-01
-6.59104943e-01 -1.59837097e-01 3.98558587e-01 1.03640711e+00
5.13379037e-01 2.32271999e-01 1.00015163e+00 -3.08436275e-01
6.03182137e-01 -3.06467056e-01 -2.70383716e-01 1.60731882e-01
4.90439795e-02 -2.91796058e-01 6.37435019e-01 -7.91736603e-01
-1.46222293e+00 -3.26448381e-01 -8.54820311e-01 -5.58335900e-01
-3.82837594e-01 2.21308082e-01 -5.66419482e-01 5.93137443e-02
7.38165736e-01 3.47120523e-01 -4.12262648e-01 -8.73435318e-01
3.66394222e-01 1.37717211e+00 8.38399589e-01 -5.02611339e-01
8.01951647e-01 -3.08847427e-01 -4.95548457e-01 -1.64539266e+00
-2.90424019e-01 -6.49775386e-01 -3.37188959e-01 -2.36147672e-01
4.28348362e-01 -7.79399872e-01 -2.20517516e-01 4.82709438e-01
-1.10391033e+00 -5.27926028e-01 -1.57910228e-01 7.54511178e-01
-3.80260974e-01 2.67876446e-01 -5.69958687e-01 -1.33420122e+00
-4.92672235e-01 -1.18407989e+00 8.74980509e-01 -3.00997645e-01
-3.27725470e-01 -4.43013996e-01 5.90131134e-02 6.88376844e-01
7.06927419e-01 -1.20729186e-01 5.76915264e-01 -8.33907068e-01
9.28562060e-02 -1.96167454e-01 3.89603287e-01 1.07716966e+00
2.32490644e-01 -7.68305287e-02 -1.63962412e+00 -5.38950205e-01
3.27084243e-01 -2.35787537e-02 5.76846659e-01 4.05271709e-01
8.74392033e-01 -5.49245059e-01 1.29568174e-01 3.79298389e-01
7.46156633e-01 6.82432294e-01 8.61276150e-01 -2.80384094e-01
4.22412813e-01 2.07798675e-01 5.24656355e-01 1.66044027e-01
-2.90624261e-01 8.93323064e-01 -1.54709667e-01 1.34846315e-01
-6.97467566e-01 -7.01118708e-02 7.95113981e-01 1.85562706e+00
1.45916030e-01 -3.78708124e-01 -7.18106627e-01 7.46659815e-01
-1.07032847e+00 -1.01084280e+00 -1.38989374e-01 2.25905418e+00
9.60126638e-01 -1.39916372e-02 -3.40736620e-02 1.06248939e+00
8.54762733e-01 1.74731031e-01 4.52911817e-02 -7.83106208e-01
-4.59872246e-01 7.81835794e-01 6.39734417e-02 8.99089575e-01
-8.43138158e-01 7.08583295e-01 6.15171623e+00 1.40166605e+00
-1.09594095e+00 7.39601791e-01 2.70840108e-01 -3.50945652e-01
-5.99035770e-02 -3.09807986e-01 -5.15949965e-01 3.06246698e-01
1.63097322e+00 -2.95711271e-02 6.58647537e-01 3.26109767e-01
7.06386268e-01 -1.00650035e-01 -8.97657216e-01 1.07150388e+00
2.65336156e-01 -7.32300282e-01 -2.41140261e-01 -3.20394874e-01
8.11660826e-01 1.04324281e-01 3.23235273e-01 4.74289626e-01
-2.36225814e-01 -1.03942847e+00 1.00216424e+00 7.35190809e-02
8.63529921e-01 -7.14702487e-01 5.89872181e-01 2.73800522e-01
-1.15790153e+00 2.33524200e-02 -1.37189880e-01 2.43158236e-01
1.05150200e-01 5.33661962e-01 -1.40798283e+00 9.02897120e-01
4.66982335e-01 8.19894150e-02 -2.75783390e-01 1.13552380e+00
-1.47393972e-01 1.57501185e+00 -3.68251979e-01 2.58521974e-01
-5.28238341e-02 2.26564497e-01 1.35698676e+00 1.62973750e+00
4.69626635e-01 -8.50016326e-02 -4.64553922e-01 3.82967502e-01
4.78486232e-02 2.12504402e-01 -3.26115727e-01 8.55611339e-02
6.94955885e-01 9.24019098e-01 -9.02781039e-02 -3.17023844e-01
-3.21094580e-02 9.03949559e-01 -2.32047737e-01 5.38518250e-01
-6.70903385e-01 -4.27996725e-01 4.02080446e-01 -2.32324630e-01
5.75721860e-01 -2.71820575e-01 -3.53194773e-02 -7.20243335e-01
6.73043653e-02 -1.42495823e+00 1.35546103e-02 -8.16164851e-01
-9.02686119e-01 1.09579170e+00 8.54443386e-03 -1.12087739e+00
-2.04595387e-01 -2.83036351e-01 -6.68081582e-01 1.08898461e+00
-1.13676715e+00 -9.43882585e-01 2.98068255e-01 5.74975669e-01
9.63100314e-01 -2.94452310e-01 1.02328348e+00 6.54840410e-01
-6.98790312e-01 9.79582191e-01 2.04791471e-01 -2.56804109e-01
5.91482520e-01 -1.10048223e+00 6.74598217e-01 1.43814886e+00
4.00335401e-01 5.68833888e-01 9.34757650e-01 -2.96287626e-01
-1.27143645e+00 -1.36723244e+00 7.46443152e-01 -1.15911163e-01
2.08453253e-01 -6.29435241e-01 -1.02956450e+00 5.24065077e-01
4.62495118e-01 -3.42099309e-01 6.50294125e-01 -1.82363674e-01
-4.44259942e-01 -2.02823833e-01 -1.18520463e+00 3.98180008e-01
6.59052789e-01 -8.54473114e-01 -9.32826400e-01 5.89755997e-02
1.07974696e+00 -4.11018759e-01 -8.50169897e-01 5.03534555e-01
2.47800767e-01 -5.37034273e-01 7.30869830e-01 2.11373381e-02
-3.37617248e-01 -5.03765464e-01 -8.73310924e-01 -1.93952775e+00
1.20748557e-01 -1.52941918e+00 -3.55200432e-02 1.65753925e+00
7.03328967e-01 -5.40532887e-01 -1.76958025e-01 1.02492943e-02
-8.18223119e-01 -1.08450972e-01 -1.44899070e+00 -1.17285752e+00
-3.46920490e-02 -1.08013654e+00 4.79831338e-01 6.16824269e-01
-7.55045712e-02 3.20844650e-01 -7.11253822e-01 5.05974352e-01
3.60570759e-01 -7.76316762e-01 6.45260215e-01 -4.30403382e-01
-5.69037080e-01 1.71235893e-02 -5.92366010e-02 -7.66729414e-01
-3.25389057e-02 -7.79850185e-01 4.07871127e-01 -1.16487241e+00
-6.14321828e-01 -1.39642552e-01 -3.34249526e-01 2.44301945e-01
-8.38340670e-02 1.64458796e-01 3.29311937e-01 -4.19966221e-01
-1.10313281e-01 6.22678697e-01 9.80286062e-01 -2.68627435e-01
-3.23888183e-01 2.79692858e-01 -1.45964816e-01 2.50477761e-01
8.08404028e-01 -5.27849436e-01 -2.27744177e-01 -2.32727155e-01
-5.26421309e-01 2.83103764e-01 -3.67584787e-02 -1.41917372e+00
-3.26217897e-02 4.09545600e-01 7.10934103e-02 -8.25283885e-01
8.32933366e-01 -4.02723610e-01 5.16934395e-01 2.21723467e-01
-4.01997477e-01 -2.89568216e-01 6.94379270e-01 2.27647349e-01
-2.73379117e-01 -7.85003081e-02 1.01111960e+00 2.37924471e-01
-1.35818675e-01 -4.42533433e-01 -8.66304278e-01 -1.72865257e-01
4.38333720e-01 1.08464517e-01 -2.38843858e-01 -4.76936191e-01
-8.61008167e-01 -3.77445310e-01 -3.65289688e-01 3.81549835e-01
8.36770117e-01 -1.16576660e+00 -1.17817211e+00 4.86156344e-01
-2.58415282e-01 -1.99096367e-01 5.07217646e-01 8.28372896e-01
-4.23129601e-03 4.45590317e-01 2.94622183e-01 -3.75023216e-01
-1.93906760e+00 3.42912495e-01 4.64785516e-01 2.17717320e-01
-5.09145796e-01 9.95834172e-01 -1.39011383e-01 -4.55725640e-01
4.85270739e-01 -4.64419365e-01 -2.71216184e-02 -3.37792456e-01
7.50437558e-01 7.17504859e-01 7.79418230e-01 -1.05863786e+00
-2.70618051e-01 9.37698632e-02 -5.61348200e-02 -8.98828030e-01
1.08357823e+00 -1.32325232e-01 1.21833049e-01 6.21225655e-01
1.29772723e+00 5.49875200e-01 -5.75798154e-01 -2.70596117e-01
-1.03423692e-01 -4.48228836e-01 4.25243169e-01 -1.25681043e+00
-5.78523755e-01 7.70323634e-01 5.61648667e-01 1.71691209e-01
1.23679507e+00 -3.32917303e-01 7.60357022e-01 2.78330356e-01
1.30899459e-01 -1.04285848e+00 -5.01027703e-02 8.08654785e-01
1.59076893e+00 -5.95834196e-01 -3.63308430e-01 -2.38951311e-01
-5.24507284e-01 6.67570233e-01 1.67221889e-01 1.93330303e-01
7.12966502e-01 5.66594183e-01 4.17937934e-01 4.46363181e-01
-8.30360472e-01 -1.54565766e-01 3.67364168e-01 8.41031551e-01
1.89395830e-01 3.41786116e-01 -3.66277769e-02 6.34989321e-01
-7.01870382e-01 -9.18393850e-01 3.46441805e-01 7.52039671e-01
-2.39711583e-01 -1.14943826e+00 -1.17657554e+00 2.92441636e-01
-6.45189226e-01 -3.18300158e-01 -4.32993531e-01 1.43998176e-01
-1.52188063e-01 1.91893935e+00 -2.70980984e-01 -8.50794256e-01
5.87856770e-01 3.95248771e-01 3.83926302e-01 -4.89616394e-01
-1.00675046e+00 8.75415623e-01 5.88452578e-01 -2.11309329e-01
-1.28267884e-01 -8.21527898e-01 -8.37028623e-01 1.38054356e-01
-6.22258723e-01 3.74487251e-01 1.02770519e+00 1.13805008e+00
1.30675361e-01 1.21537888e+00 1.04108834e+00 -8.43629301e-01
-4.08280432e-01 -1.81266069e+00 -5.64087749e-01 6.78176731e-02
8.35410655e-01 -3.56562912e-01 -6.23170495e-01 7.11517185e-02]
|
[14.835609436035156, 6.068288326263428]
|
b3409b4f-d556-4306-9160-ca44128a8af4
|
edge-aware-regional-message-passing
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Li_Edge-Aware_Regional_Message_Passing_Controller_for_Image_Forgery_Localization_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Li_Edge-Aware_Regional_Message_Passing_Controller_for_Image_Forgery_Localization_CVPR_2023_paper.pdf
|
Edge-Aware Regional Message Passing Controller for Image Forgery Localization
|
Digital image authenticity has promoted research on image forgery localization. Although deep learning-based methods achieve remarkable progress, most of them usually suffer from severe feature coupling between the forged and authentic regions. In this work, we propose a two-step Edge-aware Regional Message Passing Controlling strategy to address the above issue. Specifically, the first step is to account for fully exploiting the edge information. It consists of two core designs: context-enhanced graph construction and threshold-adaptive differentiable binarization edge algorithm. The former assembles the global semantic information to distinguish the features between the forged and authentic regions, while the latter stands on the output of the former to provide the learnable edges. In the second step, guided by the learnable edges, a region message passing controller is devised to weaken the message passing between the forged and authentic regions. In this way, our ERMPC is capable of explicitly modeling the inconsistency between the forged and authentic regions and enabling it to perform well on refined forged images. Extensive experiments on several challenging benchmarks show that our method is superior to state-of-the-art image forgery localization methods qualitatively and quantitatively.
|
['Zheng-Jun Zha', 'Xueyang Fu', 'Jiawei Liu', 'Menglu Wang', 'Jiaying Zhu', 'Dong Li']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['graph-construction']
|
['graphs']
|
[ 6.87686875e-02 -1.25583231e-01 -5.62587641e-02 -1.12463437e-01
-7.42824852e-01 -2.94886619e-01 4.42288816e-01 5.34858443e-02
-7.46640041e-02 3.12802076e-01 -3.04912832e-02 -2.40252674e-01
9.78703126e-02 -7.33188391e-01 -7.36444235e-01 -1.06687987e+00
1.44486755e-01 -1.32029459e-01 2.84072518e-01 -3.07727098e-01
3.31764728e-01 3.34838748e-01 -9.90097046e-01 1.73992530e-01
1.00537097e+00 1.26430213e+00 1.64222255e-01 2.89806038e-01
1.69856343e-02 7.18350232e-01 -3.37088823e-01 -1.50648072e-01
2.92909354e-01 -5.93862772e-01 -4.57253784e-01 5.37655652e-01
2.45174989e-01 -4.40436810e-01 -6.13934755e-01 1.49193740e+00
2.65535057e-01 -1.32302260e-02 2.26857200e-01 -1.34055734e+00
-9.94314790e-01 4.74139959e-01 -8.41098249e-01 2.58805752e-01
5.08023649e-02 3.75724912e-01 8.34158838e-01 -1.10245550e+00
3.02859545e-01 9.43840146e-01 6.27033472e-01 2.35520571e-01
-8.79199922e-01 -6.63711429e-01 3.12272936e-01 4.17873085e-01
-1.57845938e+00 -3.54345977e-01 1.25014305e+00 -2.92095482e-01
2.77034909e-01 1.72786582e-02 5.19311786e-01 8.64182711e-01
3.37688506e-01 7.55741596e-01 8.99902225e-01 -3.16718340e-01
1.90909907e-01 -1.22642433e-02 -1.25857545e-02 1.04146338e+00
1.87313184e-01 2.54306406e-01 -4.05789882e-01 -2.89105680e-02
9.38489616e-01 1.72277287e-01 -7.88103521e-01 -4.29504454e-01
-9.52821672e-01 7.97679961e-01 7.81976819e-01 5.87378681e-01
-3.51541877e-01 1.83833808e-01 1.78157374e-01 2.03456163e-01
3.68324995e-01 1.14896938e-01 -1.30289733e-01 3.70646209e-01
-9.51608419e-01 -8.12496021e-02 3.55757892e-01 6.40596151e-01
1.04748964e+00 -8.60590208e-03 -1.61983564e-01 5.26974499e-01
2.90841430e-01 -8.77267867e-02 5.00264823e-01 -4.06034112e-01
4.61374015e-01 1.04566073e+00 1.32077083e-01 -1.57560742e+00
-1.93915114e-01 -6.90373540e-01 -1.06798255e+00 1.63500965e-01
5.96794248e-01 2.41025761e-02 -1.04808283e+00 1.53995013e+00
4.11965460e-01 5.10120690e-01 -2.39417300e-01 1.26129544e+00
4.39702511e-01 3.98334742e-01 -4.70826179e-02 1.70782611e-01
1.40378869e+00 -1.08022368e+00 -5.98621368e-01 -3.78606558e-01
7.05232084e-01 -5.73290527e-01 8.63489449e-01 1.65193722e-01
-1.04695475e+00 -4.56877500e-01 -1.17849374e+00 -2.62247529e-02
-4.62100923e-01 3.69215876e-01 5.21033406e-01 5.91416955e-01
-1.08301258e+00 5.14796972e-01 -6.39954209e-01 1.47560284e-01
5.72479546e-01 2.03962088e-01 -4.23558384e-01 -2.39210129e-01
-1.31211209e+00 5.85783124e-01 6.25923574e-01 5.32989442e-01
-9.39110279e-01 -5.42626202e-01 -9.78572249e-01 3.02945584e-01
5.48789442e-01 -5.66384256e-01 6.11879885e-01 -1.30270338e+00
-1.34316730e+00 6.82361543e-01 4.32862520e-01 -1.81436166e-01
7.09433734e-01 3.06012630e-01 -4.09127116e-01 2.86859155e-01
-8.09737742e-02 4.48900133e-01 1.20625222e+00 -1.50646520e+00
-8.69562209e-01 -6.08313978e-01 -8.52161199e-02 2.21573040e-01
-3.06772709e-01 -2.25850761e-01 -8.16957593e-01 -9.42812145e-01
4.28936601e-01 -6.02424622e-01 1.79058649e-02 2.68458903e-01
-4.28316832e-01 -2.07285602e-02 1.21478283e+00 -8.39770198e-01
1.08166230e+00 -2.31826520e+00 6.70411885e-02 3.32319170e-01
4.95645553e-01 2.39178434e-01 -1.79036558e-01 1.73391864e-01
-1.43682942e-01 -1.70404449e-01 -4.14784461e-01 -5.05007923e-01
-1.51770666e-01 6.18417896e-02 -1.62752956e-01 1.08313739e+00
2.83259511e-01 9.52714801e-01 -9.91352677e-01 -3.73714983e-01
3.39472830e-01 6.44203484e-01 -3.26726615e-01 9.56512168e-02
6.32597879e-02 5.28361857e-01 -4.82538074e-01 6.91614330e-01
1.05480802e+00 -2.08989486e-01 -7.25095859e-03 -4.07869101e-01
8.70723948e-02 -1.88923404e-01 -1.21850622e+00 1.66052568e+00
-2.49898806e-01 4.10625845e-01 6.08793557e-01 -1.06979895e+00
7.80806839e-01 1.02652423e-01 3.31385672e-01 -6.33938551e-01
4.24620867e-01 9.18615982e-02 -3.48097682e-01 -2.93471068e-01
6.03391111e-01 1.17982663e-02 -7.93232545e-02 1.93990186e-01
1.48546062e-02 1.36520028e-01 -3.72115403e-01 2.27359638e-01
9.18565392e-01 -7.09709898e-02 -8.20487067e-02 -2.96973348e-01
9.53168213e-01 -4.96617138e-01 7.04850137e-01 5.99573314e-01
-4.69323188e-01 6.12344325e-01 4.54498678e-01 -3.23256403e-01
-6.22643709e-01 -9.11693096e-01 2.56175548e-01 7.60286689e-01
9.38007653e-01 -1.49958253e-01 -1.01110756e+00 -1.03798985e+00
-3.32784168e-02 4.82056409e-01 -7.86575019e-01 -7.28912354e-01
-5.81706703e-01 -5.40694654e-01 4.87813801e-01 4.30135429e-01
8.95335674e-01 -7.52786696e-01 -2.13690832e-01 1.39109686e-01
-3.56010497e-01 -1.15964162e+00 -1.09910691e+00 3.02075967e-02
-4.63280976e-01 -1.13626862e+00 -5.08505344e-01 -1.11484206e+00
9.28920090e-01 5.68054736e-01 6.23745084e-01 7.58055866e-01
-3.95301908e-01 9.66018811e-02 -3.02510798e-01 2.98922896e-01
-3.86085629e-01 -1.86968371e-01 -4.21062112e-01 8.01252306e-01
-7.73922950e-02 -3.40837508e-01 -9.21221077e-01 5.43794334e-01
-1.24712479e+00 -3.53555977e-02 7.79681444e-01 9.97979164e-01
4.72279310e-01 4.85374391e-01 3.98441344e-01 -4.16853249e-01
4.72989500e-01 -5.16233981e-01 -7.04804361e-01 4.56284553e-01
-5.86547673e-01 1.30484700e-01 8.43251467e-01 -3.43698084e-01
-9.02740061e-01 1.29322872e-01 1.01416916e-01 -4.91328657e-01
-5.01225851e-02 2.12657765e-01 -5.99500835e-01 -5.55195153e-01
1.36805460e-01 5.88456035e-01 -2.25010365e-01 -2.00212508e-01
5.09937108e-01 6.14604354e-01 1.01534128e+00 -3.35909188e-01
9.63487804e-01 7.05523133e-01 -2.50984609e-01 -6.59736931e-01
-2.84038037e-01 -5.16849637e-01 -4.36678857e-01 -3.97060126e-01
7.27353632e-01 -6.42989933e-01 -5.05450189e-01 9.87885416e-01
-1.06792724e+00 -2.39906535e-01 -8.10064748e-02 1.22935273e-01
-3.67617667e-01 8.98941994e-01 -7.40958929e-01 -5.60074866e-01
-3.54440570e-01 -1.45714545e+00 1.35827994e+00 3.45470548e-01
4.98371571e-01 -1.16533136e+00 -3.15674245e-01 3.75014991e-01
2.78884888e-01 3.66133481e-01 9.10260201e-01 -3.39799047e-01
-7.96369076e-01 -4.06076193e-01 -6.06695652e-01 3.22483420e-01
3.22563231e-01 -3.66695791e-01 -6.55059814e-01 -4.99943107e-01
2.88664937e-01 -6.78402558e-02 1.06902301e+00 7.19160959e-02
1.16525567e+00 -2.77953595e-01 -3.86681437e-01 7.39721358e-01
1.42169905e+00 -3.82991470e-02 8.47043931e-01 1.82351947e-01
8.36064875e-01 3.13060075e-01 4.36085582e-01 2.21619383e-01
5.21177411e-01 6.79771662e-01 8.52606654e-01 -4.69829232e-01
-2.46191099e-01 -3.81706119e-01 3.12738270e-01 4.13066775e-01
5.11777759e-01 -3.06138247e-01 -4.68379974e-01 5.77310264e-01
-1.92054176e+00 -6.44756734e-01 1.67124178e-02 2.15908790e+00
5.54168284e-01 7.68800676e-02 -1.30095690e-01 2.04987228e-01
9.84866142e-01 3.66949141e-01 -4.38127100e-01 2.53131520e-02
-2.07451060e-01 -1.61121145e-01 5.59890151e-01 5.15221894e-01
-1.22414076e+00 1.08051610e+00 4.64218378e+00 1.09214902e+00
-1.10815227e+00 -5.58821075e-02 7.08283126e-01 5.47314167e-01
-2.08127752e-01 2.07450286e-01 -4.42481160e-01 6.75454080e-01
1.81419313e-01 2.46404424e-01 5.00505507e-01 5.94431400e-01
2.32484475e-01 -1.23606749e-01 -7.57812619e-01 9.40202236e-01
2.04818577e-01 -1.21842098e+00 5.88242002e-02 8.61313939e-02
6.09136820e-01 -4.74163413e-01 1.51588380e-01 -4.48825546e-02
4.46521444e-03 -7.84538150e-01 8.76994014e-01 3.68309647e-01
4.78665948e-01 -9.10805643e-01 7.15179682e-01 4.27034974e-01
-1.32287490e+00 -1.32359147e-01 -3.25033844e-01 3.57047915e-01
-3.29532810e-02 6.83502972e-01 -4.97478724e-01 7.19426036e-01
4.84036773e-01 5.69681287e-01 -4.83422577e-01 9.06015992e-01
-4.75148350e-01 1.86972827e-01 1.34336380e-02 5.58365405e-01
5.07021189e-01 -3.91955018e-01 5.68016827e-01 1.04943609e+00
9.02733058e-02 5.07286238e-03 2.74555922e-01 1.09861708e+00
-3.39301825e-01 -1.43471122e-01 -2.80787736e-01 2.78002203e-01
3.11011165e-01 1.42771137e+00 -9.85407531e-01 -1.08698793e-01
-3.59650820e-01 1.55065560e+00 4.29890752e-01 3.86231959e-01
-1.10167813e+00 -4.52374399e-01 6.71381950e-01 5.73449135e-02
4.19096589e-01 -3.34237128e-01 -2.06224799e-01 -1.22812152e+00
1.16496541e-01 -8.34227860e-01 4.08991665e-01 -4.84646738e-01
-1.12977600e+00 4.29615915e-01 -5.16577184e-01 -9.28054035e-01
4.65794981e-01 -4.10430849e-01 -8.15909028e-01 7.94491231e-01
-1.76067245e+00 -1.34721839e+00 -6.37097776e-01 9.13205385e-01
4.54726845e-01 2.67452598e-01 3.73905629e-01 4.31084394e-01
-9.09290493e-01 9.09693420e-01 -5.10210320e-02 4.33079928e-01
5.26929677e-01 -1.01171005e+00 4.38955784e-01 1.32850218e+00
7.67563581e-02 2.90478736e-01 3.61341447e-01 -7.52422929e-01
-1.42001104e+00 -1.24020970e+00 3.92877042e-01 9.08453390e-03
6.12300396e-01 -3.95201653e-01 -1.16457772e+00 4.40544546e-01
2.18245275e-02 2.92866379e-01 -5.09768352e-02 -8.01133752e-01
-4.87472683e-01 -5.29250316e-02 -1.22470391e+00 5.15168250e-01
8.34279358e-01 -6.15391314e-01 -1.89606816e-01 2.29714766e-01
4.11137789e-01 -4.41714823e-01 -5.62459648e-01 2.95296758e-01
1.44037351e-01 -1.01406431e+00 1.02180493e+00 -2.17349783e-01
2.07331434e-01 -4.67500716e-01 1.27570555e-01 -1.40507936e+00
-3.42087567e-01 -8.11984181e-01 -1.72478631e-01 1.12730038e+00
8.53213146e-02 -6.75900877e-01 8.58107746e-01 2.99100608e-01
-2.63949037e-01 -6.88946724e-01 -9.42188740e-01 -6.61406815e-01
-7.40272517e-04 -9.45890471e-02 6.14419580e-01 1.04386055e+00
-7.89779723e-02 -1.30247876e-01 -4.82555807e-01 5.18752158e-01
7.59283423e-01 8.11048225e-02 3.99089187e-01 -5.41033864e-01
-2.73050398e-01 -5.71674407e-01 -6.18284523e-01 -1.19073904e+00
-1.26704406e-02 -7.70054877e-01 2.12796390e-01 -1.26274967e+00
-6.16383068e-02 -4.32666898e-01 -4.12067920e-01 4.68773991e-01
-6.47506773e-01 2.81405121e-01 -5.80710731e-02 5.69970235e-02
-5.84112048e-01 5.86851954e-01 1.27646232e+00 -3.04686725e-01
-1.67032480e-01 -1.57780513e-01 -7.84203351e-01 6.77916825e-01
9.05432999e-01 -4.52380121e-01 -2.61909842e-01 -3.38507414e-01
-2.74685353e-01 -3.78858857e-02 7.86060989e-01 -9.31411624e-01
5.20112753e-01 5.00577241e-02 3.64301443e-01 -2.77839303e-01
3.83393802e-02 -9.43837821e-01 -1.35256305e-01 5.44559538e-01
-1.19530611e-01 -6.17192313e-02 -4.82074320e-02 9.59570348e-01
-2.14511484e-01 7.74993077e-02 1.14780104e+00 -1.66309532e-02
-8.55342269e-01 4.08976346e-01 -1.49047524e-01 4.95084897e-02
1.19062448e+00 -3.29794556e-01 -3.48622799e-01 -5.24649680e-01
-3.51446003e-01 2.31351480e-01 6.53035045e-01 5.66180408e-01
9.50615346e-01 -1.17970014e+00 -6.68132842e-01 6.33881807e-01
1.13758020e-01 -1.92613780e-01 4.09782350e-01 8.86972725e-01
-5.16098320e-01 -1.00321528e-02 9.60558802e-02 -5.03173649e-01
-1.04127491e+00 8.28926206e-01 7.19050407e-01 -4.33565110e-01
-7.47895718e-01 8.11991632e-01 4.10513610e-01 6.11785520e-03
2.89335757e-01 -2.52618432e-01 1.81935772e-01 -2.29820698e-01
5.39563179e-01 3.86637181e-01 1.83311105e-01 -8.15498590e-01
-3.45409393e-01 5.42944491e-01 -2.27681339e-01 2.88330227e-01
9.00330782e-01 -4.72817153e-01 -1.08646415e-01 -4.14348274e-01
1.48106718e+00 -1.10117637e-01 -1.56237555e+00 -2.85277337e-01
1.02959797e-01 -6.77312791e-01 4.75847602e-01 -5.71944773e-01
-1.79584050e+00 5.42757452e-01 6.01910889e-01 1.02760591e-01
1.53008902e+00 -5.86206168e-02 1.17449701e+00 -2.54675150e-01
4.37130302e-01 -1.01707661e+00 1.85853347e-01 3.98458168e-02
6.52257979e-01 -1.10305810e+00 -2.86645591e-01 -5.97741723e-01
-3.85789782e-01 1.00756013e+00 5.27626038e-01 -3.33450675e-01
4.88392919e-01 2.66249478e-01 9.68937855e-03 -3.50161880e-01
-8.96299407e-02 1.24114826e-01 3.37323278e-01 4.16449457e-01
-3.47668290e-01 -1.33585423e-01 -1.70472160e-01 5.49889386e-01
3.08687896e-01 -2.11704835e-01 3.19189370e-01 8.20781648e-01
-3.99451286e-01 -8.92877638e-01 -5.39890051e-01 3.08306627e-02
-3.38033140e-01 -6.80922195e-02 -5.96986294e-01 8.28290641e-01
3.25369120e-01 1.13948500e+00 -1.49127126e-01 -5.30049264e-01
1.87537402e-01 -2.79251963e-01 3.20128798e-01 -1.39098734e-01
-7.87271142e-01 2.03461740e-02 -4.25091833e-01 -7.43815660e-01
7.93668330e-02 -2.36279175e-01 -1.17675269e+00 -2.54780352e-01
-8.26391876e-01 1.44733876e-01 5.87632537e-01 9.14083719e-01
3.95778954e-01 4.16365087e-01 9.32418704e-01 -8.60162139e-01
-5.55658638e-01 -3.76448750e-01 -7.26255596e-01 4.55044150e-01
5.72478414e-01 -5.05340159e-01 -6.55009687e-01 -2.52152890e-01]
|
[12.365924835205078, 0.9017300009727478]
|
0dd2f335-100f-45df-aa5e-875cddf14dc8
|
medical-matting-a-new-perspective-on-medical
|
2106.09887
| null |
https://arxiv.org/abs/2106.09887v3
|
https://arxiv.org/pdf/2106.09887v3.pdf
|
Medical Matting: A New Perspective on Medical Segmentation with Uncertainty
|
It is difficult to accurately label ambiguous and complex shaped targets manually by binary masks. The weakness of binary mask under-expression is highlighted in medical image segmentation, where blurring is prevalent. In the case of multiple annotations, reaching a consensus for clinicians by binary masks is more challenging. Moreover, these uncertain areas are related to the lesions' structure and may contain anatomical information beneficial to diagnosis. However, current studies on uncertainty mainly focus on the uncertainty in model training and data labels. None of them investigate the influence of the ambiguous nature of the lesion itself.Inspired by image matting, this paper introduces alpha matte as a soft mask to represent uncertain areas in medical scenes and accordingly puts forward a new uncertainty quantification method to fill the gap of uncertainty research for lesion structure. In this work, we introduce a new architecture to generate binary masks and alpha mattes in a multitasking framework, which outperforms all state-of-the-art matting algorithms compared. The proposed uncertainty map is able to highlight the ambiguous regions and a novel multitasking loss weighting strategy we presented can improve performance further and demonstrate their concrete benefits. To fully-evaluate the effectiveness of our proposed method, we first labelled three medical datasets with alpha matte to address the shortage of available matting datasets in medical scenes and prove the alpha matte to be a more efficient labeling method than a binary mask from both qualitative and quantitative aspects.
|
['ZongYuan Ge', 'Xiufen Ye', 'Xin Zhao', 'Xuan Yao', 'Zhiwen Yang', 'Yelin Huang', 'Wanji He', 'Xin Wang', 'Donghao Zhang', 'Lie Ju', 'Lin Wang']
|
2021-06-18
| null | null | null | null |
['image-matting']
|
['computer-vision']
|
[ 5.05694926e-01 5.97502053e-01 -1.32394647e-02 -4.36646640e-01
-9.06510949e-01 -3.70146751e-01 3.53349835e-01 1.64433151e-01
-4.71645087e-01 1.12382913e+00 3.18479016e-02 -1.44755438e-01
-3.37578803e-01 -4.30875719e-01 -5.87889791e-01 -8.25908482e-01
2.66442835e-01 7.46467590e-01 3.58566374e-01 2.14311466e-01
-5.88359237e-02 1.72778755e-01 -1.30039120e+00 7.81007886e-01
1.28902662e+00 1.03201425e+00 5.61408460e-01 1.95051804e-01
-3.11634451e-01 4.34931576e-01 -7.21441686e-01 -6.01223588e-01
1.16780557e-01 -3.02068204e-01 -9.13622200e-01 2.30135053e-01
4.53744352e-01 6.08666353e-02 9.78247151e-02 1.26250875e+00
4.67007309e-01 -2.17377126e-01 9.07736540e-01 -1.01586962e+00
-2.39513174e-01 1.00248373e+00 -6.48712933e-01 2.87188470e-01
-1.78848863e-01 9.33691040e-02 5.34745812e-01 -7.52346516e-01
5.03572941e-01 1.07618332e+00 6.65560603e-01 4.10381913e-01
-1.06730950e+00 -3.68338406e-01 2.37872019e-01 3.46200347e-01
-1.29449975e+00 -1.07150428e-01 5.70491910e-01 -6.08397126e-01
2.27299988e-01 6.63357019e-01 4.07577991e-01 1.23251343e+00
5.19917727e-01 8.35431993e-01 1.72563386e+00 -2.61829287e-01
1.49158224e-01 4.29489136e-01 -7.50606358e-02 7.54796922e-01
3.39124322e-01 -5.77706099e-02 -2.69227266e-01 4.30882238e-02
4.99433547e-01 -2.40095705e-01 -5.18822432e-01 -3.41795504e-01
-1.43974924e+00 5.67459762e-01 6.59619033e-01 6.89528525e-01
-3.77313077e-01 1.03178680e-01 1.99269384e-01 -2.27006003e-01
7.29537666e-01 3.88652593e-01 -1.34357452e-01 1.08645763e-03
-1.43867958e+00 -1.69867650e-01 4.33323026e-01 5.22536039e-01
3.41768116e-01 -1.36992812e-01 -8.03965747e-01 7.03753889e-01
2.98126698e-01 3.06269407e-01 5.25367320e-01 -6.54565394e-01
1.34207577e-01 5.60152888e-01 -1.92675948e-01 -8.56089115e-01
-5.27362823e-01 -8.99888694e-01 -9.57858145e-01 4.72601682e-01
6.01469338e-01 -1.38891144e-02 -1.36241877e+00 1.41216481e+00
3.27472687e-01 2.23213881e-01 -2.12226063e-01 1.09452796e+00
8.88736606e-01 1.19058415e-01 1.67317607e-03 -3.46191704e-01
1.62222409e+00 -1.03896666e+00 -1.14369643e+00 -2.87242770e-01
4.10236686e-01 -8.65876198e-01 7.40630567e-01 4.87966001e-01
-1.08964241e+00 -3.24022532e-01 -9.96389627e-01 4.11604315e-01
-3.44858140e-01 3.61194104e-01 5.23305953e-01 8.40603352e-01
-9.29599941e-01 4.86659944e-01 -8.64721417e-01 1.15755044e-01
7.99031556e-01 2.55274802e-01 -2.02865735e-01 -5.83306700e-02
-1.20526767e+00 1.39713430e+00 6.42602205e-01 3.58625263e-01
-8.60501885e-01 -7.98566222e-01 -8.87318432e-01 -2.25514352e-01
6.89615369e-01 -7.63764620e-01 1.00062478e+00 -8.48904908e-01
-1.05293727e+00 9.65035021e-01 2.14576256e-02 -5.37218392e-01
1.13717818e+00 -4.86716479e-02 -2.63631910e-01 1.54668599e-01
1.94250152e-01 8.41106176e-01 1.13684022e+00 -1.61034083e+00
-4.59644645e-01 -2.62934476e-01 -1.81505173e-01 1.12674162e-01
1.57403145e-02 -2.14259312e-01 -3.12806100e-01 -9.37202394e-01
2.69970298e-01 -7.22330987e-01 -4.24957991e-01 8.65219831e-02
-5.32752812e-01 2.13773251e-01 6.60644293e-01 -7.57432461e-01
1.33651245e+00 -2.01372504e+00 2.59331167e-01 2.28035107e-01
4.28669363e-01 2.47313380e-01 3.46350282e-01 -3.97255093e-01
1.88690368e-02 2.35435292e-01 -9.94539380e-01 -4.48375493e-01
-1.68261990e-01 4.03427601e-01 1.96859073e-02 4.70177799e-01
2.80220926e-01 1.01734364e+00 -8.56154919e-01 -1.13485205e+00
4.46252227e-01 4.36275184e-01 -3.12000830e-02 -6.41267151e-02
-3.32267612e-01 6.63816452e-01 -1.97624996e-01 8.61005723e-01
9.49030161e-01 -2.45597869e-01 -1.86069205e-01 -5.55334568e-01
5.75603172e-03 -2.97087222e-01 -1.17644238e+00 1.79876339e+00
-4.10651386e-01 4.56419826e-01 3.37026954e-01 -8.09105992e-01
5.95100343e-01 4.32248175e-01 5.88344693e-01 -2.88815677e-01
2.38739222e-01 4.44096714e-01 2.68228859e-01 -5.69275200e-01
2.45987505e-01 -3.29704314e-01 1.26177713e-01 1.20968282e-01
-1.08010225e-01 -5.07764220e-01 4.20844667e-02 7.05157071e-02
9.15347159e-01 7.03044012e-02 2.13511184e-01 -3.90105188e-01
5.31396329e-01 -4.73563978e-03 5.17708123e-01 7.87214279e-01
-4.57125217e-01 1.01488566e+00 3.60856384e-01 -1.15562543e-01
-5.80999792e-01 -1.10919726e+00 -6.58662319e-01 4.51788962e-01
1.98700294e-01 -6.29970655e-02 -8.96297693e-01 -1.18126893e+00
-3.63757610e-02 7.26747453e-01 -9.47279215e-01 5.41301817e-02
-3.83109897e-01 -1.13830674e+00 4.96119410e-01 2.06932873e-01
6.82257771e-01 -9.20030236e-01 -7.59137452e-01 8.46089721e-02
-6.22437239e-01 -1.20726502e+00 -4.16140079e-01 1.36949003e-01
-8.67534220e-01 -1.10941625e+00 -1.05724633e+00 -4.32795227e-01
8.59228373e-01 -2.12490425e-01 1.03165936e+00 -5.46891727e-02
-5.02653003e-01 2.05540761e-01 -3.16659480e-01 -5.82189023e-01
-5.82610548e-01 1.23923339e-01 -3.06011319e-01 2.31249973e-01
-2.59481966e-01 -2.56768703e-01 -5.75039387e-01 5.15273213e-01
-1.13262272e+00 3.09539586e-01 1.06783593e+00 9.55342412e-01
6.20247245e-01 1.94736034e-01 2.74542212e-01 -1.13402891e+00
5.51155627e-01 -3.87014985e-01 -2.70479053e-01 5.41096270e-01
-7.45055199e-01 3.78398776e-01 -1.79492623e-01 -3.15451026e-01
-1.36954200e+00 5.99847808e-02 -4.79830764e-02 -4.44481343e-01
-1.49055600e-01 5.07857561e-01 2.59195328e-01 -2.73258716e-01
6.92759573e-01 -1.37993068e-01 2.21160978e-01 -2.87753075e-01
3.15494299e-01 4.97401804e-01 5.50973117e-01 -6.00907743e-01
5.63345671e-01 6.85203016e-01 1.40306145e-01 -3.44665080e-01
-1.00183725e+00 -3.35357755e-01 -5.75560927e-01 -5.73850334e-01
9.45083678e-01 -5.08775890e-01 -3.90772671e-01 2.92535841e-01
-1.20158362e+00 -2.18693297e-02 -4.22578961e-01 5.00735700e-01
-3.83562118e-01 5.79719007e-01 -3.77277970e-01 -7.99862862e-01
-3.18503380e-01 -1.66520298e+00 1.28223813e+00 1.13257147e-01
-1.20044984e-02 -9.36464787e-01 -2.15727493e-01 5.60494900e-01
6.39082730e-01 4.49932724e-01 7.04340398e-01 -3.84885907e-01
-5.78440964e-01 2.42541283e-01 -3.49776000e-01 3.43574762e-01
1.65045694e-01 -1.26141205e-01 -1.16397321e+00 -4.83680665e-02
2.53693759e-01 -1.00223690e-01 1.24113762e+00 8.34207833e-01
1.39938259e+00 1.14825435e-01 -5.70965707e-01 4.59725320e-01
1.20661545e+00 9.99381673e-03 6.17843449e-01 3.12346995e-01
5.79047561e-01 7.53271341e-01 8.68953228e-01 1.31309286e-01
7.06834570e-02 7.69864142e-01 8.75045061e-01 -4.85083967e-01
-4.66798604e-01 3.32771063e-01 1.37365898e-02 4.44769233e-01
-9.54375640e-02 -6.56319335e-02 -9.64167833e-01 5.21670759e-01
-2.00973105e+00 -5.53606927e-01 -3.51012051e-01 1.83689642e+00
1.02510941e+00 3.65815729e-01 -2.91411161e-01 8.60592052e-02
9.02056694e-01 2.56544519e-02 -2.19204903e-01 -1.28349066e-01
-3.04250509e-01 2.39395306e-01 6.22140646e-01 6.74444199e-01
-1.32066119e+00 7.12154984e-01 6.11390591e+00 1.31663036e+00
-1.00514436e+00 4.65192705e-01 9.00618792e-01 -1.71714916e-03
-4.69353706e-01 -1.86941519e-01 -4.56552714e-01 6.41497374e-01
4.10835683e-01 1.96310118e-01 -1.17848694e-01 6.04546010e-01
9.07521136e-03 -5.77148497e-01 -9.00978804e-01 1.02658427e+00
2.35536858e-01 -1.21099186e+00 -1.11700118e-01 5.45744896e-02
7.94919908e-01 -2.90060580e-01 3.61954719e-01 3.89698744e-02
-2.97729094e-02 -1.37004972e+00 7.65706122e-01 8.11834097e-01
7.73981690e-01 -2.21501604e-01 1.23641539e+00 1.20990135e-01
-7.53038704e-01 1.72228798e-01 -1.56229466e-01 4.02814358e-01
4.79619324e-01 1.20685613e+00 -1.26815724e+00 8.75518501e-01
6.28087044e-01 2.37837017e-01 -8.86537194e-01 1.41851532e+00
-2.10687235e-01 3.79361749e-01 -2.12643683e-01 2.82088816e-01
3.54817241e-01 -7.35240132e-02 7.30011284e-01 1.38631070e+00
3.30408007e-01 -1.38282657e-01 2.04770062e-02 1.21918690e+00
3.86475682e-01 -3.87613885e-02 -2.88338214e-01 2.45624095e-01
-1.25574604e-01 1.57033885e+00 -1.03001237e+00 -3.36859852e-01
8.09218511e-02 9.65004504e-01 -6.70898110e-02 1.20189749e-01
-1.10333657e+00 1.59831047e-01 1.49951860e-01 -3.55485156e-02
-6.75021857e-02 8.51712450e-02 -8.96940112e-01 -9.44748104e-01
1.78205371e-01 -7.81456292e-01 5.07524014e-01 -7.43986547e-01
-1.18219268e+00 9.47412908e-01 3.14808995e-01 -1.05889487e+00
2.68599782e-02 -6.51154995e-01 -2.56690472e-01 8.28329563e-01
-1.42396653e+00 -1.25472856e+00 -5.50566196e-01 3.41023743e-01
5.88590562e-01 1.06765188e-01 5.41978061e-01 3.47416878e-01
-3.75676602e-01 3.98665458e-01 -2.41287827e-01 -4.44700897e-01
8.92498672e-01 -1.61800337e+00 -2.27897778e-01 8.97352576e-01
8.46180990e-02 1.48877114e-01 8.02899659e-01 -8.43813539e-01
-3.81873995e-01 -9.21772778e-01 4.66778427e-01 -6.44978702e-01
3.78031790e-01 -1.25010386e-01 -9.78337288e-01 3.38750333e-01
2.16273353e-01 1.29229248e-01 3.10931861e-01 -3.04458648e-01
2.20286995e-01 3.62572409e-02 -1.52041829e+00 4.93560910e-01
8.30971301e-01 -1.09397419e-01 -6.43888474e-01 3.71090472e-01
7.04607069e-01 -8.02914679e-01 -8.73064339e-01 1.13828003e+00
2.34184772e-01 -1.01733720e+00 7.87458777e-01 2.82074627e-03
2.22177669e-01 -3.91247362e-01 1.40066236e-01 -1.45901287e+00
-7.18967915e-02 -2.22431093e-01 -3.38500887e-02 1.02115393e+00
6.62129462e-01 -4.70081121e-01 8.00668657e-01 5.65937042e-01
-6.35499239e-01 -1.01072133e+00 -1.26317406e+00 -3.64426672e-01
-1.66115999e-01 -5.02388000e-01 3.15376729e-01 8.60162556e-01
-1.59634590e-01 -2.98206389e-01 -2.56851524e-01 1.38544679e-01
5.94935894e-01 -1.51746660e-01 3.10441107e-02 -1.06687343e+00
-2.34773993e-01 -6.34613037e-01 -3.44089031e-01 -3.49915475e-01
-7.71237239e-02 -8.95115316e-01 2.04013035e-01 -1.75091720e+00
2.14491099e-01 -6.81848288e-01 -2.98693925e-01 3.80415201e-01
-3.75524700e-01 4.01820034e-01 2.26417743e-02 1.09567814e-01
-4.22510356e-01 3.99995148e-01 1.67778409e+00 -5.80899358e-01
2.28304043e-01 3.10715586e-01 -4.28431004e-01 7.59337544e-01
6.29318178e-01 -4.99583066e-01 -3.49306613e-01 -2.34283656e-01
-1.30148893e-02 -1.07904509e-01 4.48123127e-01 -1.10967457e+00
1.72208130e-01 7.09018018e-03 3.62804174e-01 -6.37112379e-01
4.30837870e-01 -1.06173992e+00 2.76156515e-01 6.09594882e-01
-1.30487829e-01 -1.68538064e-01 4.21278298e-01 4.26169991e-01
-3.17241222e-01 -5.51046908e-01 8.19630086e-01 -4.13556635e-01
-4.77578193e-01 9.88719389e-02 -1.91861153e-01 -1.83637645e-02
1.34499502e+00 -2.35361665e-01 -3.22658658e-01 -1.89768337e-02
-1.13202190e+00 2.63571799e-01 1.38883382e-01 2.94216961e-01
6.30698621e-01 -1.05608368e+00 -8.27753186e-01 -6.18031658e-02
3.23151946e-02 3.22502971e-01 5.21290362e-01 1.38493800e+00
-5.39597809e-01 4.82705474e-01 -2.26853058e-01 -1.04975510e+00
-1.32839942e+00 4.56537515e-01 6.22234285e-01 -5.11093080e-01
-4.14167672e-01 9.44076478e-01 3.55530500e-01 -1.15401410e-01
3.60286325e-01 -7.10463822e-01 -2.90134162e-01 1.31004989e-01
1.81743532e-01 3.13523650e-01 3.86838138e-01 -4.76250678e-01
-3.90193969e-01 4.38392729e-01 -1.24462493e-01 -3.77997398e-01
7.59180844e-01 -3.73209976e-02 -2.58045316e-01 4.35502738e-01
3.90768439e-01 -1.11362055e-01 -1.11642003e+00 -1.65564567e-01
9.30164754e-02 -5.05111396e-01 1.76508427e-01 -1.26249409e+00
-1.06746352e+00 8.18822622e-01 9.89017546e-01 -2.14684252e-02
9.96943176e-01 1.27463490e-01 3.39293301e-01 -3.09424132e-01
3.00949484e-01 -1.08412278e+00 1.37437731e-01 -1.02883250e-01
1.09545910e+00 -1.52021885e+00 -7.81710967e-02 -8.45366359e-01
-9.25400138e-01 9.03445721e-01 6.84076905e-01 5.33856392e-01
5.94063580e-01 4.19501424e-01 1.64520338e-01 -3.76862109e-01
-1.71975106e-01 -4.45698410e-01 7.19233215e-01 5.85256398e-01
3.36054355e-01 2.19985455e-01 -6.51584625e-01 4.69801575e-01
-2.14737114e-02 -1.36309519e-01 3.82256180e-01 6.49312675e-01
-4.17030483e-01 -1.09009171e+00 -7.49276221e-01 5.59256375e-01
-6.22536242e-01 -6.81833029e-02 -1.67927235e-01 6.27694845e-01
6.41641378e-01 7.60924637e-01 -2.56279409e-01 -1.43471569e-01
5.42351827e-02 9.58791673e-02 6.11487627e-01 -6.49047911e-01
-6.83118999e-01 1.55186011e-02 5.54428808e-02 -3.24937880e-01
-5.67736626e-01 -5.12947559e-01 -1.11203098e+00 2.37855062e-01
-5.43904364e-01 1.55199334e-01 8.71167004e-01 1.15792000e+00
-3.07685193e-02 1.12510467e+00 4.56342250e-02 -6.25825465e-01
-4.68320012e-01 -1.10426784e+00 -3.58733982e-01 3.56773257e-01
1.36666015e-01 -1.08193159e+00 -3.33696514e-01 -8.40388462e-02]
|
[14.397488594055176, -2.1147522926330566]
|
8ff4a0ba-51fc-4d16-a5d7-aec2d0d8a538
|
real-time-instance-segmentation-with
|
2106.12204
| null |
https://arxiv.org/abs/2106.12204v2
|
https://arxiv.org/pdf/2106.12204v2.pdf
|
Real-time Instance Segmentation with Discriminative Orientation Maps
|
Although instance segmentation has made considerable advancement over recent years, it's still a challenge to design high accuracy algorithms with real-time performance. In this paper, we propose a real-time instance segmentation framework termed OrienMask. Upon the one-stage object detector YOLOv3, a mask head is added to predict some discriminative orientation maps, which are explicitly defined as spatial offset vectors for both foreground and background pixels. Thanks to the discrimination ability of orientation maps, masks can be recovered without the need for extra foreground segmentation. All instances that match with the same anchor size share a common orientation map. This special sharing strategy reduces the amortized memory utilization for mask predictions but without loss of mask granularity. Given the surviving box predictions after NMS, instance masks can be concurrently constructed from the corresponding orientation maps with low complexity. Owing to the concise design for mask representation and its effective integration with the anchor-based object detector, our method is qualified under real-time conditions while maintaining competitive accuracy. Experiments on COCO benchmark show that OrienMask achieves 34.8 mask AP at the speed of 42.7 fps evaluated with a single RTX 2080 Ti. The code is available at https://github.com/duwt/OrienMask.
|
['Tingming Bai', 'Yiman Chen', 'Chengyu Qiao', 'Shuya Chen', 'Zhiyu Xiang', 'Wentao Du']
|
2021-06-23
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Du_Real-Time_Instance_Segmentation_With_Discriminative_Orientation_Maps_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Du_Real-Time_Instance_Segmentation_With_Discriminative_Orientation_Maps_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['real-time-instance-segmentation', 'foreground-segmentation']
|
['computer-vision', 'computer-vision']
|
[ 3.24883550e-01 -4.21397611e-02 -2.96001881e-01 -3.85376871e-01
-8.43893647e-01 -4.97638404e-01 1.00775793e-01 4.99390550e-02
-4.35326666e-01 4.36000645e-01 -6.56175017e-01 -1.75265387e-01
3.02696407e-01 -7.02713847e-01 -7.70019889e-01 -7.71686256e-01
1.82010323e-01 4.37666297e-01 1.06837404e+00 4.12269145e-01
3.53253692e-01 5.49253404e-01 -1.47812831e+00 4.22820628e-01
8.74339283e-01 1.36475778e+00 5.13003826e-01 8.40018392e-01
-2.44825512e-01 4.08039629e-01 -7.29675174e-01 -2.24592954e-01
5.54837048e-01 -2.24585980e-02 -5.94592869e-01 3.35455656e-01
5.69510937e-01 -4.43866462e-01 -2.73852646e-01 7.72835135e-01
3.67915779e-01 -8.15552007e-03 2.40875304e-01 -1.26389003e+00
7.83152804e-02 4.64129925e-01 -9.93755877e-01 3.52341682e-01
-2.22080678e-01 2.84796238e-01 7.33607769e-01 -9.58580852e-01
5.28800607e-01 7.46193886e-01 3.73213261e-01 3.73814583e-01
-1.26900101e+00 -7.62153983e-01 4.03087854e-01 2.57452875e-01
-1.59214592e+00 -3.55123609e-01 3.75991821e-01 -1.19659796e-01
6.59742177e-01 5.63869298e-01 6.32419348e-01 3.16642672e-01
8.14271420e-02 9.86439049e-01 9.74310875e-01 -1.34981647e-01
1.21666647e-01 1.11343630e-01 1.70340553e-01 7.22553909e-01
2.46414244e-01 -2.71322012e-01 -4.88753915e-01 1.23107046e-01
9.09331203e-01 1.19631842e-01 -3.44950110e-01 -3.80609244e-01
-1.24780107e+00 4.47271138e-01 5.58706284e-01 1.14439569e-01
-1.68888673e-01 3.32945496e-01 2.17215627e-01 -2.74782926e-01
4.62658882e-01 1.22534037e-01 -5.14360905e-01 -5.30833437e-04
-1.24777114e+00 2.49477867e-02 5.84179699e-01 1.18601525e+00
9.60845768e-01 -7.15900064e-02 -1.91124991e-01 7.13148952e-01
2.77670715e-02 6.92788243e-01 2.85834700e-01 -9.60740805e-01
5.17176569e-01 7.72505164e-01 3.88512798e-02 -9.54348445e-01
-3.72537524e-01 -4.68492895e-01 -5.66789508e-01 2.01664388e-01
6.13005280e-01 -4.79793549e-03 -1.19814992e+00 1.03340077e+00
8.68797958e-01 6.16156697e-01 -2.41065279e-01 9.46809649e-01
5.65339506e-01 8.90527129e-01 -9.50910151e-02 -1.58417039e-02
1.62695956e+00 -1.11938858e+00 -3.41736823e-01 -3.54407549e-01
6.10937715e-01 -9.46485877e-01 7.76260912e-01 4.45606411e-01
-1.00519168e+00 -5.75247169e-01 -1.04170167e+00 -3.38633843e-02
-2.22520038e-01 4.34796751e-01 5.61994135e-01 6.05715573e-01
-7.56126583e-01 3.46175075e-01 -1.16686344e+00 1.15416534e-01
6.47806942e-01 7.04956293e-01 -1.97769385e-02 4.37865891e-02
-6.20530367e-01 4.42263544e-01 6.45895243e-01 3.54962289e-01
-5.86959720e-01 -7.95472503e-01 -5.67174852e-01 -5.10209464e-02
9.69485223e-01 -1.56435013e-01 1.35934341e+00 -9.77216542e-01
-1.34966683e+00 6.85985863e-01 -3.50017309e-01 -5.81725419e-01
6.37041390e-01 -1.75962836e-01 -2.55675763e-01 3.67973447e-01
2.92211264e-01 8.69434059e-01 1.03714371e+00 -1.17066014e+00
-1.27512038e+00 -2.26788893e-01 -5.06495982e-02 5.94503954e-02
-7.08263740e-02 -1.61723480e-01 -1.09047771e+00 -5.77949584e-01
4.45406169e-01 -1.00427628e+00 -5.31341791e-01 8.53535309e-02
-5.83101034e-01 5.56569137e-02 1.03212333e+00 -3.78832221e-01
1.31069589e+00 -2.24553823e+00 -2.65541762e-01 3.10554385e-01
2.50011027e-01 5.24374127e-01 2.85015941e-01 -1.75924584e-01
2.57876188e-01 -1.43015414e-01 -5.22905707e-01 -3.20370585e-01
-1.88226223e-01 1.96349323e-01 -3.25227588e-01 5.55691183e-01
2.94022769e-01 7.76514649e-01 -5.88055909e-01 -7.32564867e-01
5.93081474e-01 3.50154757e-01 -4.14787322e-01 1.35416165e-01
-2.42925107e-01 1.88755825e-01 -4.27080065e-01 8.10548186e-01
1.02524006e+00 -2.95080006e-01 2.04337060e-01 -2.29848549e-01
-3.66860121e-01 2.84623727e-02 -1.63591433e+00 1.36629009e+00
-2.86918640e-01 6.95753694e-01 1.25325695e-01 -7.92400777e-01
7.02741086e-01 3.53494585e-02 2.96077073e-01 -7.47150719e-01
7.72782937e-02 5.00538468e-01 -6.82341158e-02 -3.99901494e-02
6.89033508e-01 3.33126217e-01 -8.28118846e-02 1.84175909e-01
-4.11499083e-01 -1.23526687e-02 3.87064397e-01 1.95829257e-01
7.63253450e-01 1.81447595e-01 1.38181105e-01 -2.32656002e-01
5.54157794e-01 2.74373293e-01 9.11254048e-01 6.75873339e-01
-1.54408038e-01 7.86409676e-01 3.56850147e-01 -3.40206683e-01
-7.46541977e-01 -1.01690292e+00 -4.18944091e-01 1.00475693e+00
6.69628620e-01 -4.47997838e-01 -9.72059369e-01 -6.20245695e-01
8.44438840e-03 2.79256999e-01 -4.20110226e-01 3.81794542e-01
-1.04246867e+00 -7.31742620e-01 3.55604619e-01 7.65437543e-01
6.10811234e-01 -1.01835239e+00 -1.15800142e+00 5.21008074e-01
4.87730168e-02 -1.38751459e+00 -6.62576020e-01 2.36230060e-01
-9.83624876e-01 -1.01801169e+00 -7.59649098e-01 -7.85605907e-01
7.92804420e-01 5.25318444e-01 9.03190553e-01 2.80023724e-01
-7.22771287e-01 -9.32443142e-02 -2.32869655e-01 -1.78711459e-01
-7.37915486e-02 2.50794232e-01 -3.70614171e-01 1.71332583e-01
3.33151361e-03 -2.37135306e-01 -9.92742240e-01 6.92912161e-01
-9.85085964e-01 4.74945068e-01 5.29885769e-01 4.86112803e-01
1.22921276e+00 -8.12343210e-02 3.88297141e-01 -9.79124904e-01
-4.74406928e-01 -1.80072606e-01 -1.16868234e+00 5.83508983e-02
-3.39881003e-01 -2.01124161e-01 4.35285777e-01 -4.78867769e-01
-9.33759689e-01 4.70316082e-01 1.25994220e-01 -2.47738764e-01
-7.31050298e-02 -2.57156253e-01 -2.49238521e-01 -7.99180642e-02
1.14225619e-01 1.77255392e-01 -2.07770541e-01 -4.57577735e-01
3.91690820e-01 7.12185085e-01 5.88756621e-01 -4.11767185e-01
6.93711936e-01 7.43306220e-01 -1.43774450e-01 -8.45616639e-01
-6.12421453e-01 -6.56578183e-01 -5.15451968e-01 -2.61764020e-01
8.74199271e-01 -8.15805435e-01 -7.78716564e-01 5.78264296e-01
-9.00954425e-01 -6.01704657e-01 -1.53303936e-01 1.63188025e-01
-2.50041425e-01 2.94202209e-01 -5.93292534e-01 -6.33874655e-01
-4.10102904e-01 -1.44646513e+00 1.18197846e+00 5.31052351e-01
3.12672891e-02 -3.42804700e-01 -7.02177405e-01 4.60908502e-01
1.20624661e-01 1.66607037e-01 3.00112158e-01 -4.07908440e-01
-1.42072201e+00 -2.06074134e-01 -4.96696830e-01 1.94467723e-01
8.65379050e-02 2.14736462e-01 -9.07297015e-01 -2.29141768e-02
-1.87711656e-01 2.85959095e-01 8.21946621e-01 4.45020705e-01
1.53641593e+00 -1.75132137e-02 -6.70120835e-01 7.17553377e-01
1.50893736e+00 4.18625355e-01 4.61461335e-01 2.80893832e-01
9.78115141e-01 3.19534421e-01 1.05463493e+00 3.97396743e-01
9.52983275e-02 8.34486842e-01 3.02512348e-01 -2.12071136e-01
-2.63534129e-01 1.54893935e-01 1.10399164e-01 5.68568349e-01
1.82585344e-01 -2.84249395e-01 -9.94053423e-01 5.77431560e-01
-1.81621563e+00 -5.08957803e-01 -4.23466980e-01 2.36560059e+00
8.15053999e-01 5.01195133e-01 9.69180167e-02 8.75842422e-02
9.62045789e-01 1.93523884e-01 -6.63702846e-01 -1.62654832e-01
1.04100797e-02 2.89940536e-01 9.62522209e-01 6.31935120e-01
-1.19945753e+00 1.20199394e+00 4.76439476e+00 1.21706879e+00
-1.12111425e+00 1.97533541e-03 1.02182925e+00 -2.62264371e-01
2.63152897e-01 1.71957593e-02 -1.22316146e+00 7.19923854e-01
7.10793912e-01 1.31785765e-01 1.00416489e-01 9.61003125e-01
8.01834017e-02 -5.75465858e-01 -8.48199368e-01 8.44979465e-01
-2.27882877e-01 -1.53025746e+00 -2.39052296e-01 -3.69176641e-02
5.85978210e-01 -1.25144288e-01 -3.71235907e-02 6.54341206e-02
-3.32086951e-01 -7.28765488e-01 8.14730942e-01 2.43506040e-02
7.93886006e-01 -7.96286821e-01 4.34267730e-01 2.31740430e-01
-1.56507730e+00 1.60584167e-01 -4.01429623e-01 2.12596253e-01
2.16893300e-01 6.36678040e-01 -1.03885102e+00 4.22846884e-01
6.84727430e-01 3.64728957e-01 -4.81283635e-01 1.14644945e+00
-1.48110077e-01 6.33498967e-01 -6.12722933e-01 1.72205523e-01
2.79045194e-01 -1.02504246e-01 3.69569033e-01 1.40296090e+00
8.92607644e-02 2.39794448e-01 4.32187855e-01 4.83088553e-01
-3.67252738e-03 7.00185671e-02 7.27284998e-02 3.09615076e-01
6.09570980e-01 1.41041744e+00 -1.63354909e+00 -7.56116867e-01
-2.59283662e-01 1.16558719e+00 1.89154744e-02 1.09758303e-01
-1.38184440e+00 -2.98136503e-01 6.30990744e-01 2.69188881e-01
8.41125667e-01 -1.99779719e-01 -6.35062099e-01 -8.82009387e-01
3.11780840e-01 -7.42009819e-01 1.52725667e-01 -3.36121231e-01
-6.24497831e-01 6.63877487e-01 -9.22806710e-02 -1.27089775e+00
2.89849132e-01 -5.82325578e-01 -3.82946581e-01 6.06184244e-01
-1.46803474e+00 -7.61967421e-01 -3.97183746e-01 2.65722901e-01
8.30757499e-01 4.92493778e-01 3.99098963e-01 4.67452377e-01
-9.89058137e-01 5.63084960e-01 1.23972453e-01 1.30182907e-01
5.06815135e-01 -1.29657197e+00 5.40991664e-01 9.38245893e-01
2.00171843e-01 2.54943788e-01 5.16941428e-01 -6.02482855e-01
-1.19166422e+00 -1.20560515e+00 4.17982072e-01 -2.04265177e-01
3.27774584e-01 -4.54987586e-01 -1.09246349e+00 4.61148053e-01
-1.98488563e-01 4.55452502e-01 3.75521511e-01 -5.19129157e-01
-1.42751336e-01 -4.48307872e-01 -9.82590795e-01 6.54236376e-01
8.52158070e-01 -3.79684269e-02 -5.08883260e-02 3.76277238e-01
9.33441699e-01 -8.81994307e-01 -6.91225052e-01 3.61659825e-01
3.56505871e-01 -8.93430769e-01 9.36143637e-01 5.18686809e-02
8.53708163e-02 -7.90330291e-01 -5.74992178e-03 -4.38157231e-01
-6.59810938e-03 -6.34177923e-01 -1.93891108e-01 1.20697212e+00
4.46150959e-01 -7.22275496e-01 1.05216765e+00 7.73556113e-01
-1.27231687e-01 -1.15404916e+00 -1.05683279e+00 -7.96685338e-01
-4.46726054e-01 -5.75425029e-01 6.91099048e-01 4.24486548e-01
-6.37991071e-01 6.44005276e-03 -1.63335681e-01 6.11992359e-01
6.16721630e-01 6.61113679e-01 8.88882756e-01 -8.36809933e-01
-4.03121710e-01 -4.04326499e-01 -4.82958525e-01 -1.37204790e+00
-2.13023692e-01 -6.23792529e-01 2.71146446e-01 -1.32628345e+00
2.85576824e-02 -7.89841175e-01 -2.81122297e-01 5.42843878e-01
-4.99292344e-01 7.93637395e-01 4.17427242e-01 1.56262383e-01
-7.87240863e-01 1.62401989e-01 1.10147023e+00 1.29977882e-01
-3.35633308e-01 7.95655176e-02 -2.66287595e-01 9.09411490e-01
1.01291156e+00 -6.77418649e-01 -2.30220437e-01 -3.70586872e-01
-4.43143249e-01 1.29203856e-01 2.40470886e-01 -1.20800495e+00
1.49513543e-01 -6.19787648e-02 5.39069057e-01 -8.25715661e-01
5.22382319e-01 -8.37855577e-01 2.09075868e-01 5.57559431e-01
1.28079861e-01 -1.30171910e-01 3.94069582e-01 5.65282226e-01
6.88853040e-02 -1.51954204e-01 8.92326832e-01 2.39537492e-01
-1.04289854e+00 5.26173055e-01 -2.33272806e-01 -3.73543315e-02
1.42657173e+00 -6.87543333e-01 -1.55205265e-01 3.58704150e-01
-5.68190515e-01 3.19829851e-01 5.29271841e-01 1.18632637e-01
5.00425041e-01 -8.62474918e-01 -4.50284868e-01 1.56496704e-01
-1.27684906e-01 5.24479985e-01 5.87139785e-01 7.96812773e-01
-8.50569487e-01 3.16913605e-01 2.06925556e-01 -8.75384629e-01
-1.49212205e+00 3.79002452e-01 3.48978460e-01 -1.50338978e-01
-1.08459079e+00 9.68294263e-01 5.62120020e-01 1.99607193e-01
6.48341104e-02 -5.63369930e-01 2.39051953e-01 -5.40885292e-02
6.59600616e-01 4.50616032e-01 2.17843559e-02 -5.61003447e-01
-5.85437298e-01 5.59171379e-01 -2.93236941e-01 -4.70817313e-02
8.99981678e-01 -1.84962023e-02 1.25042155e-01 2.23943993e-01
9.37970817e-01 1.36784390e-01 -1.69738722e+00 -1.06091954e-01
2.22117171e-01 -8.37169945e-01 -1.87699758e-02 -5.69587827e-01
-1.42581487e+00 6.40917242e-01 6.64507985e-01 7.34975487e-02
1.20341396e+00 3.15411426e-02 1.14835024e+00 -8.20026025e-02
3.37808222e-01 -1.07191586e+00 -1.55941695e-01 7.32735395e-02
3.79169941e-01 -1.16365123e+00 7.85373673e-02 -9.83165622e-01
-5.35072327e-01 9.00609553e-01 9.06465352e-01 -2.81250119e-01
2.61970609e-01 5.50440252e-01 1.94903851e-01 -1.99403707e-03
-3.86728853e-01 -1.55564025e-01 1.77218392e-01 2.14770004e-01
1.85755983e-01 4.38836813e-01 -2.55677730e-01 3.81849259e-01
9.17971134e-02 -4.35328722e-01 4.84437793e-01 9.14655268e-01
-6.92261338e-01 -1.15401614e+00 -4.88542408e-01 4.57857549e-01
-6.14596903e-01 -2.04617493e-02 1.58264488e-01 8.88862610e-01
2.25094348e-01 8.49811852e-01 3.18807721e-01 -1.53624127e-02
2.36967564e-01 -1.89759493e-01 2.25468978e-01 -7.12768435e-01
-6.72348440e-01 3.55853468e-01 -1.17517777e-01 -8.91489208e-01
-1.34656549e-01 -4.86060739e-01 -1.80976117e+00 -7.24135861e-02
-5.26667416e-01 6.00408912e-02 6.41892195e-01 6.66006088e-01
5.01774311e-01 5.65155983e-01 3.55700582e-01 -9.64185536e-01
-1.40356924e-02 -4.93547589e-01 -4.56906646e-01 2.91658528e-02
1.73408046e-01 -4.88116384e-01 -1.87301692e-02 1.89998876e-02]
|
[9.409461975097656, -0.009846139699220657]
|
c91bc98d-035f-4978-a583-7b1615f47101
|
virtual-reality-to-study-the-gap-between
|
1912.09380
| null |
https://arxiv.org/abs/1912.09380v2
|
https://arxiv.org/pdf/1912.09380v2.pdf
|
A Transferable Adaptive Domain Adversarial Neural Network for Virtual Reality Augmented EMG-Based Gesture Recognition
|
Within the field of electromyography-based (EMG) gesture recognition, disparities exist between the offline accuracy reported in the literature and the real-time usability of a classifier. This gap mainly stems from two factors: 1) The absence of a controller, making the data collected dissimilar to actual control. 2) The difficulty of including the four main dynamic factors (gesture intensity, limb position, electrode shift, and transient changes in the signal), as including their permutations drastically increases the amount of data to be recorded. Contrarily, online datasets are limited to the exact EMG-based controller used to record them, necessitating the recording of a new dataset for each control method or variant to be tested. Consequently, this paper proposes a new type of dataset to serve as an intermediate between offline and online datasets, by recording the data using a real-time experimental protocol. The protocol, performed in virtual reality, includes the four main dynamic factors and uses an EMG-independent controller to guide movements. This EMG-independent feedback ensures that the user is in-the-loop during recording, while enabling the resulting dynamic dataset to be used as an EMG-based benchmark. The dataset is comprised of 20 able-bodied participants completing three to four sessions over a period of 14 to 21 days. The ability of the dynamic dataset to serve as a benchmark is leveraged to evaluate the impact of different recalibration techniques for long-term (across-day) gesture recognition, including a novel algorithm, named TADANN. TADANN consistently and significantly (p<0.05) outperforms using fine-tuning as the recalibration technique.
|
['François Laviolette', 'Erik Scheme', 'Gabriel Gagnon-Turcotte', 'Ulysse Côté-Allard', 'Kyrre Glette', 'Angkoon Phinyomark', 'Benoit Gosselin']
|
2019-12-16
| null | null | null | null |
['emg-gesture-recognition']
|
['medical']
|
[ 2.59791613e-01 -3.38672757e-01 -2.74785250e-01 7.88937509e-02
-6.09959424e-01 -6.83636189e-01 3.97102922e-01 -2.32004583e-01
-7.25732505e-01 4.34601486e-01 2.72489786e-01 -1.00659490e-01
-2.91350365e-01 -2.18008593e-01 -5.27928770e-01 -5.18220127e-01
-4.34250027e-01 7.34787583e-02 3.18822712e-01 -2.19581380e-01
3.95063698e-01 6.55776799e-01 -1.85754430e+00 1.23519920e-01
4.90857363e-01 7.92670310e-01 3.98209393e-01 4.89912599e-01
2.75712937e-01 -6.67630956e-02 -7.20778883e-01 2.65533239e-01
5.50405800e-01 -5.50022840e-01 -2.46332273e-01 7.70286843e-02
2.93266952e-01 -4.32673395e-01 -1.72759831e-01 3.11919957e-01
1.00378752e+00 3.80737990e-01 2.40027294e-01 -9.73268747e-01
8.76781903e-03 2.09170014e-01 -1.88196912e-01 1.00143179e-01
7.62648225e-01 5.34475148e-01 6.13909185e-01 -4.01545256e-01
9.94340956e-01 7.13233232e-01 6.31346166e-01 6.63116634e-01
-1.37160659e+00 -6.86458647e-01 -2.52086371e-02 2.68218249e-01
-1.36854649e+00 -2.79080570e-01 7.26168513e-01 -6.16125643e-01
9.77846086e-01 5.40272236e-01 1.19084239e+00 1.45601022e+00
2.45994344e-01 4.48754340e-01 1.26464045e+00 -4.53795463e-01
5.30212224e-01 -1.59196749e-01 -1.12512305e-01 -1.02151901e-01
2.60715783e-01 4.76040095e-01 -6.49872720e-01 1.03166461e-01
8.94171298e-01 -2.20057964e-02 -7.09855676e-01 -5.44456005e-01
-1.07968903e+00 2.97359675e-01 2.11135149e-01 5.68700433e-01
-6.72716856e-01 8.99767969e-03 4.47530687e-01 5.22533476e-01
-2.11140871e-01 7.55775273e-01 -4.62384999e-01 -1.07016158e+00
-7.75466323e-01 4.45868492e-01 8.91745746e-01 4.97107506e-01
2.22295113e-02 -6.36103749e-02 -1.61727950e-01 7.15193868e-01
-3.71753238e-02 2.62743443e-01 8.89092505e-01 -5.63777626e-01
3.91216069e-01 7.37638474e-01 6.67727739e-02 -9.31061924e-01
-6.27004266e-01 -2.81675816e-01 -1.67779773e-01 6.75355911e-01
7.37388015e-01 -1.41939417e-01 -8.51308107e-01 1.68888521e+00
3.73596996e-01 -1.74555957e-01 -3.79151344e-01 1.38846445e+00
2.41114289e-01 -2.53109857e-02 -1.04490086e-01 -2.92719811e-01
1.15151823e+00 -2.96244502e-01 -7.20440567e-01 -6.25377446e-02
4.81260002e-01 -5.28054774e-01 1.44850552e+00 7.67823577e-01
-6.13471806e-01 -4.08308327e-01 -1.19493628e+00 4.30977911e-01
-3.01875889e-01 4.51841541e-02 4.76075590e-01 6.79598093e-01
-5.26485920e-01 9.80975091e-01 -1.11759496e+00 -5.20937383e-01
-1.49386823e-01 7.28292942e-01 -6.30629063e-01 4.91625965e-01
-1.01909697e+00 1.12727153e+00 1.82156563e-01 4.14530456e-01
-3.89749438e-01 -6.34334028e-01 -4.15948093e-01 -4.26525921e-01
3.96758527e-01 -1.00552790e-01 8.11068058e-01 -8.25919628e-01
-2.00567794e+00 6.52200878e-01 3.20735097e-01 8.29752758e-02
9.81899023e-01 -4.54501748e-01 -3.22700381e-01 1.10723423e-02
-2.86747187e-01 -1.02464529e-02 7.36370385e-01 -9.68198776e-01
-2.04981521e-01 -5.02837658e-01 -1.12054303e-01 9.90165174e-02
-2.89675236e-01 -2.79397219e-01 -3.50517422e-01 -8.11652541e-01
1.16766155e-01 -1.26119304e+00 3.23942512e-01 -1.21174306e-01
-1.21840119e-01 2.20469266e-01 8.85844350e-01 -8.26538563e-01
1.37220883e+00 -2.20837760e+00 3.24785411e-01 4.23414379e-01
-1.33958310e-01 3.88789058e-01 -4.56851022e-03 7.02243209e-01
-2.40620911e-01 -8.27718601e-02 3.18228230e-02 2.03061432e-01
-2.86702514e-01 2.10317791e-01 2.06561178e-01 5.22698700e-01
-1.34601742e-01 6.49785340e-01 -6.82098925e-01 7.74049014e-02
5.12187302e-01 4.60817546e-01 -2.33689800e-01 2.11709931e-01
2.80858457e-01 8.63495350e-01 -2.36990377e-01 4.75165784e-01
1.52491122e-01 4.30731088e-01 3.68957192e-01 -3.99388254e-01
-2.48032004e-01 1.09286614e-01 -1.56919432e+00 1.80090129e+00
-4.78941947e-01 6.21544302e-01 1.38171539e-01 -6.31989419e-01
9.23471570e-01 5.13873219e-01 8.00078630e-01 -9.19443905e-01
5.11733711e-01 5.03675222e-01 3.87848765e-01 -8.24581563e-01
1.86618730e-01 5.86866178e-02 1.66559830e-01 4.11592185e-01
-4.27094065e-02 3.79894301e-02 2.92237490e-01 -4.40900087e-01
1.24141932e+00 5.84033847e-01 8.08588043e-02 -2.18712062e-01
8.30718428e-02 9.84818712e-02 3.26266110e-01 5.29640019e-01
-2.88154036e-01 5.86431563e-01 1.07870422e-01 2.83390749e-02
-7.36655533e-01 -8.88209343e-01 -7.54378513e-02 6.79022670e-01
1.15352020e-01 -2.26650208e-01 -6.38404548e-01 -1.25005573e-01
2.42096871e-01 4.20732349e-01 -6.47044301e-01 -2.63668716e-01
-6.23155594e-01 -3.05360258e-01 5.03575921e-01 6.09799504e-01
6.98598623e-02 -1.07457340e+00 -1.41334248e+00 3.90879095e-01
-4.30603102e-02 -7.43090332e-01 -3.18718284e-01 3.33666027e-01
-1.11164093e+00 -1.33307636e+00 -5.33824682e-01 -2.58077741e-01
2.35037923e-01 -1.80925019e-02 3.12419504e-01 -8.39785784e-02
-3.76833498e-01 5.29909253e-01 -7.40021169e-01 -2.86388248e-01
-2.08817244e-01 -1.35042788e-02 2.50236452e-01 -1.33719057e-01
1.52494222e-01 -8.20915043e-01 -6.31981969e-01 5.14210045e-01
-7.80928671e-01 -1.52477458e-01 6.80948794e-01 8.70171845e-01
4.76185441e-01 -4.14457440e-01 4.24300402e-01 -2.95459479e-01
1.03577530e+00 -5.08919396e-02 -3.29681903e-01 1.30039364e-01
-7.51176238e-01 -1.28361687e-01 4.24054474e-01 -1.09233987e+00
-6.00263953e-01 1.44572183e-01 8.39462802e-02 -4.96758372e-01
-1.32011471e-03 5.70693493e-01 -3.52121964e-02 -2.33749807e-01
7.74283409e-01 3.72263342e-02 6.21719778e-01 -5.80750585e-01
1.48627803e-01 9.39955771e-01 7.28807092e-01 -4.06825274e-01
5.78565359e-01 1.28862709e-01 -1.01632409e-01 -7.07682908e-01
1.43086433e-01 -4.58465517e-01 -9.11510110e-01 -7.24321187e-01
3.34782451e-01 -4.19889838e-01 -9.03945327e-01 6.02531791e-01
-6.35216177e-01 -7.32002556e-01 -2.17681885e-01 1.04744184e+00
-6.33729815e-01 3.10431011e-02 -2.87673265e-01 -9.68362391e-01
-3.68644536e-01 -1.14930177e+00 7.20144808e-01 1.27547905e-01
-8.19199681e-01 -3.82017702e-01 4.49561439e-02 2.53744870e-01
5.40234804e-01 7.84498334e-01 3.86977226e-01 -3.69664043e-01
5.61978966e-02 -7.82140732e-01 4.27120060e-01 1.09076127e-01
4.50411439e-01 -1.25900522e-01 -7.90494323e-01 -6.48904085e-01
1.61112383e-01 -1.39581785e-01 -2.11999356e-03 1.86651230e-01
6.82791650e-01 1.71842687e-02 -1.31636620e-01 1.78056762e-01
1.28827131e+00 3.96467507e-01 8.90649438e-01 6.07482731e-01
5.70846438e-01 6.05450690e-01 7.24443376e-01 1.12746008e-01
-2.12788105e-01 1.27013218e+00 1.58105031e-01 2.17966586e-01
-1.98603004e-01 -2.09721059e-01 3.34687024e-01 6.25233054e-01
-6.45808160e-01 1.94045052e-01 -7.39323497e-01 3.26346993e-01
-1.61193812e+00 -7.85491228e-01 -9.88838673e-02 2.79999256e+00
6.78079128e-01 -4.82140146e-02 3.63133579e-01 6.34982109e-01
4.88653868e-01 -3.44754487e-01 -7.56105244e-01 -3.63648236e-01
3.94814402e-01 4.53277022e-01 5.68434954e-01 4.96105179e-02
-5.61253130e-01 2.66672313e-01 6.00777292e+00 4.87691194e-01
-1.98591363e+00 -1.05683297e-01 -3.37126732e-01 -5.77679515e-01
3.05060446e-01 -1.38487384e-01 -2.16929689e-01 7.06532001e-01
8.71553242e-01 -8.76109153e-02 6.26272261e-01 6.87345207e-01
8.39253247e-01 -3.84642482e-01 -1.16773248e+00 8.82665634e-01
-9.63187870e-03 -8.09770286e-01 -4.12095070e-01 1.99453786e-01
2.32111484e-01 4.81745601e-02 -4.79438156e-01 7.64346495e-02
-6.46573544e-01 -7.19435573e-01 9.29331779e-01 5.48321605e-01
8.76293361e-01 -3.32819283e-01 4.49379623e-01 3.64214152e-01
-1.10868716e+00 -3.13726336e-01 5.45829594e-01 -2.67753601e-01
2.05485895e-01 7.33350143e-02 -4.64862198e-01 5.48463821e-01
5.31492889e-01 2.73738474e-01 -3.49699855e-01 1.01985931e+00
-2.63110578e-01 6.51051462e-01 -5.53139210e-01 -3.25957537e-01
-1.77052319e-01 -8.93084183e-02 9.11747575e-01 8.80102634e-01
2.19372064e-01 2.78632324e-02 -6.01732768e-02 7.14802980e-01
5.21118701e-01 2.14870632e-01 -4.70185190e-01 -1.90628916e-01
4.30337667e-01 1.04650199e+00 -6.20247900e-01 3.16906989e-01
-3.97229679e-02 9.52098191e-01 -1.24765910e-01 2.91902125e-01
-5.27934253e-01 -5.58491945e-01 4.74034786e-01 4.51153427e-01
9.04008374e-02 -5.63130200e-01 -4.79220062e-01 -8.39289725e-01
7.23975778e-01 -1.19310760e+00 1.27873242e-01 -5.39014399e-01
-9.27102089e-01 1.47779167e-01 1.42514125e-01 -1.67382038e+00
-5.63937843e-01 -6.41481340e-01 -5.25930882e-01 8.72356892e-01
-6.87723935e-01 -6.18013859e-01 -6.28021538e-01 3.48530471e-01
2.88945764e-01 2.50285178e-01 1.02643001e+00 5.65447509e-01
-5.37013829e-01 5.75370491e-01 -2.75958423e-02 -2.41718832e-02
6.79281414e-01 -7.55024314e-01 -1.19934380e-01 7.21183181e-01
-1.25654995e-01 9.86321032e-01 7.36299515e-01 -6.28346801e-01
-2.03771019e+00 -2.89450347e-01 2.50009239e-01 -3.71773124e-01
5.47595739e-01 -4.79777902e-01 -9.42898691e-01 2.11401090e-01
-5.45736134e-01 -2.53722906e-01 4.54306602e-01 -2.32929364e-02
5.03481030e-02 -2.02173591e-01 -1.11747992e+00 6.94134533e-01
1.13021362e+00 -4.09649074e-01 -5.27572691e-01 -2.00472206e-01
-6.87127784e-02 -7.24046767e-01 -1.08555210e+00 4.86220658e-01
1.56295836e+00 -6.00656509e-01 4.48786974e-01 -1.97142079e-01
1.51866838e-01 -2.62345970e-01 1.62923709e-01 -1.37394750e+00
1.19584143e-01 -6.32052124e-01 -1.36720479e-01 9.09589350e-01
2.53792256e-01 -8.68353248e-01 7.72071421e-01 7.05321133e-01
1.70126528e-01 -1.05265343e+00 -1.20215356e+00 -1.14293122e+00
-2.85746604e-01 -6.48565531e-01 3.24280173e-01 6.14324272e-01
4.88589644e-01 -1.36036873e-01 -2.59797156e-01 -2.91951001e-01
1.59308333e-02 -3.87047268e-02 1.16013205e+00 -1.09106541e+00
-4.90861177e-01 -5.12397528e-01 -9.01494622e-01 -7.16477334e-01
-5.56900561e-01 -5.85818172e-01 2.48753935e-01 -1.27463639e+00
-2.42905289e-01 -3.59789789e-01 -1.48120865e-01 5.33454061e-01
3.01259290e-02 1.31106883e-01 3.46683979e-01 3.78533989e-01
3.51666868e-01 1.59182042e-01 1.17145753e+00 1.66641414e-01
-8.80484939e-01 9.65793878e-02 -2.46228963e-01 1.63939670e-01
6.39817417e-01 -4.41846818e-01 -4.13156837e-01 -1.39342040e-01
-3.70032161e-01 1.26733780e-01 3.77058208e-01 -1.27221942e+00
-5.16893063e-03 -1.24050947e-02 2.37578556e-01 -9.02496874e-02
2.48945639e-01 -9.16054368e-01 7.38295197e-01 7.20615149e-01
-1.67803481e-01 -5.85263185e-02 3.47941935e-01 3.44161749e-01
-1.08240433e-01 8.73711333e-02 3.70291382e-01 3.16132933e-01
-7.00582743e-01 -1.52157545e-01 -4.20075476e-01 -2.43897825e-01
1.11840653e+00 -1.02821720e+00 1.40483215e-01 -2.61642814e-01
-7.96128035e-01 -3.23657580e-02 4.54478830e-01 8.42271984e-01
1.97428539e-01 -9.77084458e-01 -2.63866931e-01 3.77605081e-01
2.61466026e-01 -4.88071352e-01 2.24854201e-01 1.23929656e+00
-3.34849209e-01 1.66156381e-01 -6.27090693e-01 -6.88516557e-01
-1.40981162e+00 -1.15216024e-01 3.50658059e-01 2.17257743e-03
-1.03693914e+00 2.48512641e-01 -8.86170745e-01 -2.55125642e-01
3.88008982e-01 -1.81623504e-01 -6.96872026e-02 8.01154673e-02
2.25909188e-01 5.83577335e-01 4.29370850e-01 -2.38996387e-01
-3.80851805e-01 5.18099308e-01 2.86309928e-01 -4.52265710e-01
1.28806543e+00 1.02851197e-01 3.89812052e-01 9.36768115e-01
8.83586407e-01 -9.99414176e-02 -1.35461831e+00 4.18556601e-01
5.94656914e-02 -6.98063493e-01 -5.11623733e-02 -1.10132146e+00
-9.41754937e-01 4.95092839e-01 1.26247787e+00 -2.13031918e-01
1.11912310e+00 -6.30870938e-01 5.87755501e-01 9.63723660e-02
8.61189663e-01 -1.46217179e+00 -1.69284970e-01 3.08069140e-02
1.21343851e+00 -6.46555901e-01 9.87759382e-02 -2.30816111e-01
-4.35681194e-01 1.16346824e+00 4.64957118e-01 -2.46734113e-01
3.73662055e-01 4.39140290e-01 4.50102001e-01 -6.08563721e-02
-2.05371320e-01 -7.84381293e-03 3.30057621e-01 5.51683724e-01
4.35108423e-01 2.98425943e-01 -1.13896608e+00 5.85860670e-01
-2.81938195e-01 6.77198589e-01 2.79882163e-01 1.39232087e+00
1.16131991e-01 -1.29672408e+00 -3.81371707e-01 6.72090530e-01
-8.78488049e-02 6.28420532e-01 -3.71021688e-01 1.28003931e+00
1.13862604e-01 1.02421606e+00 -1.48386240e-01 -8.70989084e-01
9.58837867e-01 6.97264597e-02 7.22915709e-01 -3.95194739e-01
-1.00113535e+00 6.80823624e-03 3.12010534e-02 -9.82125342e-01
-2.50076354e-01 -8.39030564e-01 -1.38660371e+00 -1.71544299e-01
-6.54546738e-01 -1.48422211e-01 1.15629590e+00 8.91706824e-01
4.28113341e-01 5.63490748e-01 4.50954646e-01 -1.21537173e+00
-7.69215167e-01 -1.40709817e+00 -5.99979460e-01 8.83404255e-01
-1.23346783e-01 -1.17309356e+00 -4.28464413e-01 -2.05624998e-01]
|
[6.8321428298950195, 0.1981944888830185]
|
c484f571-61b6-40e2-95a1-127749e66f07
|
enhancing-chinese-multi-label-text
| null | null |
https://aclanthology.org/2022.rocling-1.4
|
https://aclanthology.org/2022.rocling-1.4.pdf
|
Enhancing Chinese Multi-Label Text Classification Performance with Response-based Knowledge Distillation
|
It’s difficult to optimize individual label performance of multi-label text classification, especially in those imbalanced data containing long-tailed labels. Therefore, this study proposes a response-based knowledge distillation mechanism comprising a teacher model that optimizes binary classifiers of the corresponding labels and a student model that is a standalone multi-label classifier learning from distilled knowledge passed by the teacher model. A total of 2,724 Chinese healthcare texts were collected and manually annotated across nine defined labels, resulting in 8731 labels, each containing an average of 3.2 labels. We used 5-fold cross-validation to compare the performance of several multi-label models, including TextRNN, TextCNN, HAN, and GRU-att. Experimental results indicate that using the proposed knowledge distillation mechanism effectively improved the performance no matter which model was used, about 2-3% of micro-F1, 4-6% of macro-F1, 3-4% of weighted-F1 and 1-2% of subset accuracy for performance enhancement.
|
['Kuo-Kai Shyu', 'Po-Lei Lee', 'Lung-Hao Lee', 'Po-Hsun Liao', 'Cheng-Fu Cao', 'Szu-Chi Huang']
| null | null | null | null |
rocling-2022-11
|
['multi-label-text-classification', 'multi-label-text-classification']
|
['methodology', 'natural-language-processing']
|
[ 6.02050386e-02 1.07062034e-01 -5.21770537e-01 -7.35147238e-01
-7.87784219e-01 -4.27038312e-01 -4.01567481e-02 6.88399434e-01
-4.65424240e-01 1.02276659e+00 -4.10023965e-02 -2.62980670e-01
-1.10250577e-01 -6.44342422e-01 -2.09004730e-01 -6.57345533e-01
5.78496516e-01 7.12004423e-01 1.49353966e-01 -1.05195992e-01
3.65810513e-01 -8.76087919e-02 -1.46893561e+00 6.05673373e-01
1.13368058e+00 1.10061753e+00 -3.10118586e-01 6.12336755e-01
-4.76788014e-01 1.03476226e+00 -8.85059416e-01 -5.11835694e-01
-2.32212126e-01 -2.72810519e-01 -9.24283981e-01 -2.42547184e-01
1.43299282e-01 2.23393290e-04 2.23542765e-01 9.62240636e-01
6.67126298e-01 -1.02825478e-01 9.45989788e-01 -1.29338932e+00
-7.12531269e-01 1.06018412e+00 -8.37254882e-01 -3.78710628e-02
-4.16027295e-04 -3.18245590e-01 1.08447540e+00 -5.70470870e-01
1.00342743e-01 1.23110151e+00 8.35519493e-01 4.61858630e-01
-9.98326302e-01 -1.19549954e+00 1.00701623e-01 2.24179417e-01
-1.45486784e+00 6.74306601e-02 4.34044570e-01 -5.75483859e-01
7.78241098e-01 2.20984608e-01 2.40116157e-02 5.93539834e-01
1.92150921e-01 7.85714805e-01 1.53446615e+00 -6.54933214e-01
-1.37312248e-01 6.97761476e-01 9.01736081e-01 4.78877783e-01
4.20714974e-01 -3.69054466e-01 -4.83182162e-01 -3.39091897e-01
-1.34057567e-01 -1.97963014e-01 1.40090203e-02 3.91066253e-01
-8.65769267e-01 9.37630057e-01 6.07839450e-02 3.95477712e-01
3.85647528e-02 -3.50125760e-01 5.77995062e-01 3.98622632e-01
8.59560907e-01 1.84104189e-01 -1.12082005e+00 -4.51090299e-02
-7.71018922e-01 5.14238067e-02 7.31667101e-01 8.49227846e-01
8.81660819e-01 -2.63707101e-01 -3.30722988e-01 1.05779338e+00
3.55388403e-01 7.09523737e-01 9.82608616e-01 -2.67807156e-01
5.10738790e-01 9.71176147e-01 -7.29445443e-02 -9.35105920e-01
-9.40794706e-01 -5.02972126e-01 -8.36344719e-01 -3.61387730e-01
2.43629768e-01 -3.92770648e-01 -7.78889418e-01 1.49612486e+00
5.13890147e-01 -2.59822667e-01 1.55676708e-01 3.98674279e-01
1.20561039e+00 6.63244545e-01 4.89014655e-01 -5.97241938e-01
1.42473471e+00 -1.09458899e+00 -9.29189026e-01 7.78699592e-02
1.21891999e+00 -8.49539101e-01 1.03560066e+00 4.09262210e-01
-4.96275276e-01 -5.50501049e-01 -8.33821833e-01 1.41502723e-01
-5.18778026e-01 4.26035702e-01 2.82975167e-01 1.05922329e+00
-5.89344263e-01 3.32800746e-01 -1.42707407e-01 8.91463012e-02
4.37164873e-01 2.89831966e-01 -2.18746424e-01 -7.99065307e-02
-1.36349630e+00 8.87237668e-01 8.36583853e-01 -2.67255545e-01
-4.95720685e-01 -7.99250722e-01 -3.74823153e-01 8.19492340e-02
2.90849566e-01 3.54694668e-03 1.29725623e+00 -8.58834386e-01
-1.37170660e+00 8.88878286e-01 -2.06741616e-02 2.19772682e-01
3.97250801e-01 -1.81439072e-01 -7.42405415e-01 -1.86088964e-01
7.29743168e-02 2.46579289e-01 2.60742605e-01 -1.20409369e+00
-9.93335664e-01 -4.79460746e-01 -3.50049704e-01 3.40868115e-01
-5.60448766e-01 1.35458902e-01 2.41132051e-01 -6.39340699e-01
3.55938599e-02 -8.43279898e-01 1.56687155e-01 -7.70520926e-01
-6.11375034e-01 -6.55540466e-01 9.14818168e-01 -7.51899362e-01
1.52002490e+00 -1.81567764e+00 -4.62138951e-01 2.44941011e-01
2.25597590e-01 3.17282230e-01 1.30574461e-02 1.43970281e-01
-2.64832705e-01 4.72774714e-01 8.49208832e-02 -2.60770526e-02
-5.88236675e-02 -2.07845401e-02 1.23390861e-01 3.76622498e-01
-1.16921924e-01 7.38106251e-01 -7.35388517e-01 -6.29439950e-01
-1.50737360e-01 -1.98076153e-03 -1.25513315e-01 2.88027227e-01
-1.77294001e-01 2.93238193e-01 -5.34188807e-01 7.79647470e-01
7.01086879e-01 -3.99690717e-01 2.09989637e-01 -2.14188159e-01
7.20386654e-02 -4.88231406e-02 -1.34388101e+00 1.00818038e+00
-3.29365939e-01 2.52781540e-01 -3.96882951e-01 -9.42661822e-01
1.25383687e+00 4.31186497e-01 4.98935252e-01 -7.19406664e-01
4.15861815e-01 2.78701335e-01 -6.15656935e-02 -7.47152627e-01
4.01668906e-01 -4.50787634e-01 -2.84273297e-01 8.57865214e-01
-2.00284421e-02 1.07586659e-01 1.38996527e-01 -7.91433901e-02
7.75308013e-01 -3.49523127e-01 2.57513195e-01 -2.87363470e-01
7.14490712e-01 -2.28975154e-02 7.82599330e-01 4.71156120e-01
-2.98964262e-01 2.63520002e-01 5.43114662e-01 -4.27411139e-01
-7.52861440e-01 -3.16753507e-01 -3.57773095e-01 1.79027390e+00
3.83006372e-02 -2.89367527e-01 -4.97768164e-01 -1.12332213e+00
8.81501287e-02 9.63741779e-01 -5.64066052e-01 -2.34887004e-01
-2.93778062e-01 -1.45701671e+00 8.33084345e-01 3.48209560e-01
6.97133780e-01 -8.96854103e-01 -8.23748261e-02 4.56205234e-02
-5.87060153e-01 -7.83341765e-01 -3.62571508e-01 5.63341498e-01
-7.84780920e-01 -1.23707187e+00 -4.12188500e-01 -7.05398142e-01
6.79393888e-01 1.54313622e-02 1.12136805e+00 8.29459950e-02
5.82716689e-02 -2.79481798e-01 -7.35824704e-01 -6.79327846e-01
-5.14629483e-01 3.90790850e-01 5.62096387e-02 6.86340928e-02
7.47952640e-01 -1.05662480e-01 -4.36984152e-02 5.47597647e-01
-6.57259107e-01 -5.47245238e-03 5.42672396e-01 9.71018076e-01
4.96587932e-01 3.78592879e-01 1.12781787e+00 -1.32686961e+00
7.04778492e-01 -5.35840034e-01 -3.17897350e-01 8.22575152e-01
-1.33199406e+00 2.20542606e-02 7.52601147e-01 -6.27160966e-01
-1.22328484e+00 -1.61927670e-01 -1.82679296e-02 2.23850533e-01
-1.49663284e-01 4.92754340e-01 -1.37003124e-01 4.77572717e-02
7.70549178e-01 -8.04921612e-02 -2.51709253e-01 -4.21640724e-01
3.74986917e-01 1.46891689e+00 1.04488075e-01 -6.15351617e-01
2.75392979e-01 -1.52220249e-01 -3.91519308e-01 -2.97619045e-01
-1.50036788e+00 -6.22259021e-01 -5.66192448e-01 -1.47708923e-01
7.74125695e-01 -8.86700034e-01 -1.02450573e+00 8.50814939e-01
-7.69114912e-01 -3.40225279e-01 1.75731909e-02 4.82753426e-01
-8.08875039e-02 1.31452784e-01 -6.74775839e-01 -8.16048086e-01
-7.37504542e-01 -9.68780637e-01 6.62067771e-01 6.62987649e-01
-1.29082292e-01 -8.91875923e-01 7.77709335e-02 8.89872134e-01
3.96964252e-01 -1.28385723e-01 1.35414958e+00 -1.32777143e+00
3.60026568e-01 -2.37216458e-01 -4.21253592e-01 4.70778495e-01
3.54025774e-02 -9.89073664e-02 -1.12332904e+00 -2.57198334e-01
-1.54727057e-01 -9.49484706e-01 5.24204373e-01 2.03357667e-01
1.22041059e+00 -2.50683278e-01 -3.84925604e-01 7.77164623e-02
1.46182835e+00 3.75560075e-01 2.96352863e-01 3.01301807e-01
8.35895479e-01 5.72565794e-01 6.80034757e-01 4.41530704e-01
7.75627971e-01 3.56421083e-01 1.88206241e-01 8.90313610e-02
7.85670727e-02 2.43797433e-02 3.04520968e-02 1.23000443e+00
3.67553502e-01 -3.36125195e-01 -1.07074833e+00 3.04682612e-01
-1.67814636e+00 -4.28377688e-01 -5.76842606e-01 1.88641632e+00
1.53556967e+00 4.03825119e-02 5.92240505e-02 4.73049581e-01
9.28993881e-01 -1.06117316e-01 -5.75316846e-01 -3.60409319e-01
-1.60137296e-01 6.71698079e-02 6.43351138e-01 3.90768915e-01
-1.15821970e+00 8.39951217e-01 6.45601845e+00 1.12198997e+00
-9.61110950e-01 3.18187684e-01 1.12206769e+00 1.37265384e-01
-1.86953023e-01 -2.48888209e-01 -1.34450161e+00 6.82455897e-01
1.26962924e+00 -1.57306150e-01 -4.25463021e-02 7.23014355e-01
-4.04731445e-02 -3.48558486e-01 -6.36297405e-01 8.28250825e-01
2.53817886e-01 -7.60984421e-01 -3.72020230e-02 -1.45088494e-01
1.15341246e+00 -2.87815213e-01 -1.85692117e-01 7.65165567e-01
4.83201325e-01 -1.16184115e+00 5.42396188e-01 3.69880468e-01
8.51328969e-01 -8.02865565e-01 1.07264781e+00 7.42247999e-01
-7.98158228e-01 -2.35196233e-01 -2.73625553e-01 1.21363379e-01
-1.96898982e-01 1.17941272e+00 -7.67606139e-01 7.52012610e-01
6.21215880e-01 3.88008744e-01 -6.95406139e-01 7.14825273e-01
-1.55755192e-01 9.61957693e-01 -1.37453973e-01 -4.65584785e-01
-3.97446938e-02 1.85104266e-01 -3.99474293e-01 1.31582451e+00
9.54265669e-02 2.90279955e-01 3.97990972e-01 3.86634134e-02
-3.08144599e-01 7.20463037e-01 1.90773204e-01 1.84988916e-01
6.35748744e-01 1.40046620e+00 -7.72037745e-01 -5.34250438e-01
-3.26789498e-01 3.30546707e-01 3.22676539e-01 7.47025758e-02
-9.90068853e-01 -7.62469888e-01 -6.50238618e-02 -4.02001262e-01
2.65761781e-02 4.93529111e-01 -7.70112753e-01 -1.01901031e+00
-2.37718850e-01 -9.96217906e-01 7.12986469e-01 -5.94117343e-01
-1.38175392e+00 3.55590671e-01 -7.99980238e-02 -8.34434152e-01
1.46452367e-01 -5.05145073e-01 -2.14366913e-01 9.46836650e-01
-1.51852334e+00 -1.08078527e+00 -6.11097336e-01 2.60687232e-01
2.28981853e-01 -2.74530202e-01 9.05235827e-01 6.03249073e-01
-8.71030629e-01 1.09519184e+00 5.47443867e-01 1.18982911e-01
1.17335176e+00 -1.18293691e+00 -4.16963011e-01 1.32014737e-01
-1.61877841e-01 -6.22504279e-02 3.47660869e-01 -7.92093039e-01
-6.04260206e-01 -1.18313515e+00 1.36497319e+00 -4.08814758e-01
3.41324478e-01 9.09836739e-02 -9.68033195e-01 5.80906332e-01
-1.05434649e-01 -2.29959060e-02 1.26256108e+00 3.02453816e-01
-4.75074619e-01 -3.95348936e-01 -1.40055525e+00 -2.05545072e-02
2.71142930e-01 -1.81168526e-01 -4.87166435e-01 6.15682781e-01
6.96779907e-01 -4.03114170e-01 -1.20492649e+00 5.58169305e-01
5.46924174e-01 -5.53576946e-01 4.32848126e-01 -7.78451085e-01
3.93878251e-01 -2.56903678e-01 -2.17489049e-01 -1.45997667e+00
-4.73410398e-01 3.10818672e-01 5.28545305e-02 1.61838484e+00
7.56280482e-01 -5.97641945e-01 6.64275825e-01 4.11453068e-01
2.16407198e-02 -9.32866395e-01 -6.67205453e-01 -4.53610837e-01
2.81049848e-01 -1.85545579e-01 5.77578962e-01 1.50982106e+00
-5.00186943e-02 5.26955307e-01 -3.09045494e-01 -9.47787091e-02
4.31304008e-01 -3.99638079e-02 4.63785231e-01 -1.48568916e+00
1.14038460e-01 -3.98780853e-01 4.94977273e-02 -6.47332728e-01
3.29985440e-01 -1.08437049e+00 2.47689039e-02 -1.27045381e+00
5.66508114e-01 -9.03161287e-01 -6.03778362e-01 9.80488896e-01
-6.55950367e-01 1.39412597e-01 -1.87508672e-01 1.54783934e-01
-6.83877349e-01 5.15714884e-01 1.18672562e+00 -2.60272115e-01
-1.07667856e-01 1.30751640e-01 -1.07260466e+00 4.66418356e-01
8.47354412e-01 -1.02151072e+00 -4.28866059e-01 -2.44499102e-01
2.84965187e-01 -1.90753639e-01 -5.11951923e-01 -6.89013720e-01
1.98022157e-01 -4.60635096e-01 4.21057045e-01 -6.21210217e-01
-2.77989209e-01 -6.10262752e-01 3.20958570e-02 4.98845279e-01
-8.69319260e-01 -2.60273386e-02 1.63579151e-01 4.31080431e-01
-1.22487107e-02 -6.43900096e-01 7.67822385e-01 5.37959002e-02
-3.61970037e-01 -5.69508597e-02 -1.65163502e-01 3.97314131e-01
1.26693308e+00 1.63448066e-01 -7.04502165e-01 1.59521941e-02
-4.27938581e-01 5.56921899e-01 -2.29435772e-01 2.85127252e-01
1.35591894e-01 -1.34574270e+00 -1.07189429e+00 1.01434641e-01
2.75627583e-01 1.88281387e-02 4.82071608e-01 6.62155926e-01
-4.05841053e-01 4.58759159e-01 -1.21824839e-03 -4.27630097e-01
-1.38343215e+00 2.40470678e-01 2.96852350e-01 -7.78739393e-01
2.31837835e-02 9.59746361e-01 -1.26565605e-01 -9.54313219e-01
2.50209272e-01 -1.42959371e-01 -5.19860148e-01 3.95948291e-01
5.43750703e-01 6.49963856e-01 2.90144563e-01 -6.27115071e-01
-4.00879145e-01 4.97284502e-01 -3.86033714e-01 1.48165271e-01
9.73430157e-01 -1.77490599e-02 -4.05451298e-01 8.03378701e-01
1.14863980e+00 -1.01498120e-01 -5.87773919e-01 -3.06057721e-01
2.02205285e-01 -1.51200324e-01 4.34213813e-04 -1.41295552e+00
-9.87188280e-01 4.79751766e-01 6.39283895e-01 1.57550618e-01
1.04085255e+00 -2.18649209e-01 6.90556824e-01 1.78384617e-01
7.47021735e-02 -1.41796732e+00 2.25456163e-01 7.44056165e-01
1.93644464e-01 -1.43260169e+00 4.28800359e-02 -2.34573022e-01
-6.43440902e-01 1.08710206e+00 9.13275361e-01 5.83776474e-01
8.42332840e-01 1.47976160e-01 4.80662107e-01 4.15196270e-02
-7.84670830e-01 1.79441124e-01 2.77881175e-01 2.62441754e-01
7.01836944e-01 3.43619972e-01 -6.24334157e-01 1.00930643e+00
-1.88844770e-01 -1.10616893e-01 2.93341249e-01 7.91105092e-01
-7.49591351e-01 -1.02403033e+00 -4.47188735e-01 9.28023577e-01
-8.00536990e-01 6.35084808e-02 -8.48482102e-02 5.16356528e-01
3.46765935e-01 1.36444962e+00 -2.42382661e-01 -8.40190351e-01
3.39059561e-01 5.03223240e-01 -4.14712466e-02 -6.21358275e-01
-9.34816420e-01 -1.62101507e-01 8.65008533e-02 4.37909141e-02
-5.25390625e-01 -4.29579020e-01 -1.34104276e+00 -3.19718152e-01
-9.59344268e-01 4.69040751e-01 8.43215227e-01 1.18273818e+00
4.45495583e-02 6.15809023e-01 8.37895155e-01 1.61785558e-01
-8.62578392e-01 -1.48803568e+00 -7.28661537e-01 2.93776065e-01
-8.67244881e-03 -5.68627298e-01 -4.85648245e-01 1.43594921e-01]
|
[9.528101921081543, 4.4637885093688965]
|
9598a41a-a816-4bbd-8c12-e4ae18aea8d0
|
a-dataset-for-telling-the-stories-of-social
| null | null |
https://aclanthology.org/D18-1117
|
https://aclanthology.org/D18-1117.pdf
|
A Dataset for Telling the Stories of Social Media Videos
|
Video content on social media platforms constitutes a major part of the communication between people, as it allows everyone to share their stories. However, if someone is unable to consume video, either due to a disability or network bandwidth, this severely limits their participation and communication. Automatically telling the stories using multi-sentence descriptions of videos would allow bridging this gap. To learn and evaluate such models, we introduce VideoStory a new large-scale dataset for video description as a new challenge for multi-sentence video description. Our VideoStory captions dataset is complementary to prior work and contains 20k videos posted publicly on a social media platform amounting to 396 hours of video with 123k sentences, temporally aligned to the video.
|
['ana', 'Sp Gella', 'Mike Lewis', 'Marcus Rohrbach']
|
2018-10-01
| null | null | null |
emnlp-2018-10
|
['video-description']
|
['computer-vision']
|
[ 5.15824072e-02 -8.52940083e-02 -6.22208059e-01 -5.05584419e-01
-8.71612072e-01 -5.89469135e-01 5.33054829e-01 4.70098183e-02
-2.70667404e-01 9.91324246e-01 9.65192080e-01 4.67063904e-01
1.61210135e-01 -4.07933563e-01 -7.33336151e-01 5.52605614e-02
3.14243175e-02 2.21103832e-01 2.82647640e-01 -1.04718357e-01
-5.63387107e-03 -1.59597993e-01 -1.50867438e+00 9.26466942e-01
8.12315196e-02 7.90469408e-01 1.88251466e-01 8.66368294e-01
-1.75445601e-01 1.21310341e+00 -5.23966372e-01 -7.80643940e-01
-1.71042770e-01 -3.65203112e-01 -8.23929548e-01 3.04546237e-01
8.61845672e-01 -9.15016949e-01 -1.03807402e+00 6.16068900e-01
2.07216978e-01 1.26501590e-01 3.61238778e-01 -1.60584724e+00
-7.22178936e-01 7.66120613e-01 -1.05835497e-01 2.32926443e-01
1.29647315e+00 -7.85161629e-02 9.39047575e-01 -4.22371477e-01
1.19431913e+00 1.04647791e+00 8.18874180e-01 6.36617541e-01
-8.47195029e-01 -5.26471257e-01 -7.83208832e-02 3.86090368e-01
-1.56740606e+00 -7.70955741e-01 5.09574652e-01 -4.65871662e-01
8.95528257e-01 6.02028370e-01 1.06527627e+00 1.65710139e+00
-3.22095186e-01 1.01566660e+00 2.44544297e-01 5.75713329e-02
-3.37428302e-01 2.29920536e-01 -9.83129665e-02 3.07960480e-01
-7.72651136e-02 -8.29252958e-01 -1.07837272e+00 -2.55262882e-01
4.65910614e-01 2.45612830e-01 -5.14455199e-01 1.26270562e-01
-1.49106252e+00 5.14910042e-01 9.31308046e-02 3.11096400e-01
-2.71215081e-01 3.92910361e-01 7.76035428e-01 3.58381182e-01
5.25215924e-01 1.23103857e-01 -6.03193231e-02 -8.60067248e-01
-1.04631805e+00 6.50436521e-01 8.30576479e-01 1.32384491e+00
5.01644254e-01 -4.13783252e-01 -4.31538194e-01 6.64923429e-01
-2.92563662e-02 4.24994588e-01 2.90093958e-01 -1.26830590e+00
8.33590984e-01 4.49360877e-01 2.73206502e-01 -1.16410136e+00
-3.01273882e-01 2.42262244e-01 -4.36150312e-01 -1.08121169e+00
4.58073109e-01 -2.17334330e-01 -1.12642854e-01 1.62195671e+00
-2.28430815e-02 5.40767312e-01 -2.01393068e-01 1.05075097e+00
1.22482967e+00 8.39355648e-01 -1.36351883e-01 -3.38856250e-01
1.35329020e+00 -8.26153040e-01 -9.52402890e-01 -5.01293456e-03
6.87434852e-01 -6.31076813e-01 8.02740455e-01 3.90927754e-02
-1.18485975e+00 -1.48744971e-01 -6.49376035e-01 -3.14215183e-01
-3.32090966e-02 -2.45263100e-01 4.00663406e-01 2.97663063e-01
-9.17637050e-01 3.81686896e-01 -3.65561157e-01 -5.94225764e-01
5.68629026e-01 -9.65382904e-02 -7.74611771e-01 -3.61656189e-01
-1.35955095e+00 3.76698524e-01 2.31513023e-01 -3.34426790e-01
-6.80126846e-01 -7.96114445e-01 -1.04186761e+00 -9.32059437e-02
3.47492337e-01 -4.90884721e-01 1.37572229e+00 -1.14389884e+00
-9.87923443e-01 9.05583858e-01 -9.78157669e-02 -6.03511631e-01
7.60165274e-01 -4.00488913e-01 -4.88500863e-01 8.07608962e-01
2.20575064e-01 1.05372858e+00 7.56201088e-01 -5.78241885e-01
-3.91388059e-01 1.35448435e-02 6.51264369e-01 2.01771170e-01
-8.98003280e-01 4.15512413e-01 -7.90131986e-01 -6.61889791e-01
-6.49727881e-01 -1.08108544e+00 4.64890152e-01 1.51868314e-01
-5.83375692e-01 1.69144887e-02 1.25641668e+00 -9.53602970e-01
1.27110660e+00 -2.35542083e+00 1.94816276e-01 -4.61056530e-01
1.99993730e-01 -1.28573537e-01 8.86827558e-02 9.99333441e-01
3.96974117e-01 3.69670838e-01 1.08475752e-01 -7.21474826e-01
-3.53407338e-02 5.56254350e-02 -2.41246000e-01 4.38465625e-01
-2.51347512e-01 6.57158196e-01 -9.60441649e-01 -6.70474708e-01
-1.13234393e-01 8.67003143e-01 -6.01104975e-01 1.04205430e-01
-3.66849661e-01 2.98164248e-01 -5.31010151e-01 4.83193338e-01
2.22923651e-01 -3.89217347e-01 1.11822903e-01 -7.45456666e-02
-1.00273155e-01 1.24133475e-01 -7.38658071e-01 2.06271458e+00
-2.31178865e-01 1.48382759e+00 -7.95167312e-02 -5.57555616e-01
1.44488335e-01 9.35567796e-01 6.99265420e-01 -2.16931924e-01
2.92388652e-03 -2.05143005e-01 -7.97279894e-01 -9.51289415e-01
6.47165418e-01 1.85258850e-01 -4.15232450e-01 5.40957630e-01
-1.71361387e-01 8.09191987e-02 5.71415424e-01 7.41578579e-01
1.22889483e+00 9.75689590e-02 1.52984113e-01 4.14430350e-01
1.59044012e-01 2.44150206e-01 3.50162715e-01 5.41529775e-01
-3.02955300e-01 1.00231659e+00 6.82178974e-01 -3.63881022e-01
-1.21079350e+00 -6.85453594e-01 7.11752623e-02 9.72025156e-01
2.50962172e-02 -1.01206362e+00 -7.94390857e-01 -3.08661520e-01
1.28038123e-01 2.50681162e-01 -2.09955961e-01 2.24738285e-01
-5.35842299e-01 1.10391779e-02 5.53060412e-01 3.41028363e-01
7.19440341e-01 -8.73868406e-01 -2.18585134e-01 1.44071318e-02
-9.23752904e-01 -1.67679846e+00 -1.09793878e+00 -1.11295557e+00
-2.63224304e-01 -1.21678603e+00 -9.06251669e-01 -7.28160620e-01
3.51527989e-01 8.82040203e-01 9.85170543e-01 3.37991446e-01
-1.57850206e-01 9.43724751e-01 -6.20917380e-01 1.70894831e-01
-1.40646711e-01 -8.32808167e-02 2.51173556e-01 1.36543334e-01
3.12497497e-01 -4.87267643e-01 -4.35843021e-01 3.02668035e-01
-7.09250808e-01 4.71216708e-01 -1.63100362e-01 2.78599679e-01
2.08127156e-01 -1.84711188e-01 6.42324805e-01 -5.49006522e-01
4.00452137e-01 -1.09187472e+00 9.59428698e-02 -5.30072115e-02
4.36377615e-01 -8.68902981e-01 4.65485156e-01 -4.72545922e-01
-6.62910998e-01 -2.00991929e-02 2.40921125e-01 -5.97214401e-01
1.75387203e-03 4.65708077e-01 1.77403569e-01 1.70711353e-01
1.03299029e-01 1.18932590e-01 1.83126740e-02 -3.22706342e-01
1.43810943e-01 1.06299162e+00 4.87831593e-01 1.08356280e-02
6.09312057e-01 6.17230952e-01 -4.86857861e-01 -1.37632990e+00
-9.75900292e-01 -6.12409174e-01 -4.00979817e-01 -7.52687871e-01
1.05127406e+00 -1.28337932e+00 -1.08950722e+00 2.83020705e-01
-1.25616038e+00 -1.33187935e-01 2.55720794e-01 6.23359442e-01
-6.74832344e-01 4.03427184e-01 -8.53407502e-01 -4.79896486e-01
-2.97300331e-02 -7.55159795e-01 8.92148256e-01 3.55155207e-02
-5.95431089e-01 -8.60910237e-01 -2.69206494e-01 1.12146974e+00
4.28020716e-01 2.37427592e-01 7.94976950e-02 -5.65279782e-01
-5.72909951e-01 -6.31014168e-01 -3.26837957e-01 1.32470444e-01
-4.76286933e-02 1.54123064e-02 -7.24650860e-01 -4.14827645e-01
-4.25357252e-01 -9.53311443e-01 7.06820071e-01 2.42783070e-01
1.23301136e+00 -7.13469505e-01 -3.67184877e-01 2.66705483e-01
9.80584800e-01 -4.50769871e-01 5.69312811e-01 2.78023332e-01
7.34288156e-01 5.25415719e-01 4.07753199e-01 7.99959958e-01
8.85479510e-01 9.32293057e-01 3.25006127e-01 5.91123581e-01
-2.63107091e-01 -5.12911022e-01 6.43370092e-01 7.21559703e-01
1.36099458e-02 -6.75481677e-01 -7.42906570e-01 5.49686313e-01
-1.87405860e+00 -1.64256763e+00 5.05040661e-02 1.87624574e+00
5.31524360e-01 5.99302836e-02 4.53168243e-01 -1.02589928e-01
9.88117099e-01 3.96742731e-01 -2.80744106e-01 1.66502729e-01
-2.78446019e-01 -9.58317399e-01 2.81312585e-01 2.85418481e-01
-9.88530636e-01 4.77545053e-01 6.70590925e+00 5.53721011e-01
-8.14967692e-01 3.00702602e-01 4.37811166e-01 -9.59647775e-01
-2.62267739e-01 -2.52254635e-01 -8.10907483e-01 7.36915648e-01
1.23476481e+00 -3.93538028e-01 5.20397842e-01 5.15556872e-01
5.69953322e-01 -2.00494137e-02 -1.21063280e+00 1.36010075e+00
7.69120038e-01 -1.96216011e+00 1.29562512e-01 -1.60067350e-01
6.95849061e-01 7.45276213e-02 -1.67485178e-01 2.43080452e-01
-6.05194330e-01 -8.31040144e-01 1.11653757e+00 4.82071131e-01
9.32383239e-01 -4.96370763e-01 5.75746238e-01 3.14928055e-01
-1.20232058e+00 2.35698074e-02 -1.22595109e-01 -4.56719220e-01
8.02712381e-01 2.71039397e-01 -5.82784712e-01 5.02825193e-02
7.49702334e-01 1.39831638e+00 -4.08489197e-01 1.00938702e+00
1.40328303e-01 4.34848398e-01 -1.60661846e-01 -1.21890008e-01
9.56381559e-02 1.23400211e-01 9.04506087e-01 1.31531608e+00
5.16517699e-01 1.88245922e-01 2.96849430e-01 3.78872812e-01
-6.94125593e-01 5.24414890e-02 -7.97174096e-01 -6.38909161e-01
7.29637802e-01 8.85317445e-01 -4.56006229e-01 -2.93484300e-01
-9.36020195e-01 1.12600577e+00 2.66789466e-01 2.78817087e-01
-9.40556288e-01 -9.29454640e-02 7.75585949e-01 6.16345763e-01
-1.30833447e-01 -3.08663458e-01 6.58374906e-01 -1.47222316e+00
4.37642872e-01 -5.96779644e-01 2.86897659e-01 -1.13169098e+00
-1.07115352e+00 3.12066972e-01 2.30734572e-01 -1.36934745e+00
-3.34940076e-01 -1.18129619e-03 -2.54585594e-01 -5.96021079e-02
-1.24883807e+00 -1.30262876e+00 -5.12838840e-01 5.89393437e-01
8.88291240e-01 -1.94201216e-01 5.41882217e-01 7.53683448e-01
-5.17844260e-01 3.54016006e-01 -2.41115168e-02 1.76652730e-01
1.15488291e+00 -5.36190927e-01 1.59187287e-01 3.24215919e-01
1.47650987e-01 2.41144240e-01 8.93282652e-01 -6.75193727e-01
-1.70263565e+00 -1.06920755e+00 1.16398692e+00 -7.15161264e-01
8.17768574e-01 -5.20437479e-01 -6.05317593e-01 1.07248151e+00
3.61174583e-01 -2.66651791e-02 1.13851821e+00 -2.40179658e-01
-1.82931811e-01 1.48851991e-01 -1.03885031e+00 6.10872090e-01
1.29330242e+00 -9.25435245e-01 -4.45129812e-01 9.50221956e-01
1.08009136e+00 -3.89003247e-01 -1.00749826e+00 -3.00777674e-01
8.47699523e-01 -8.37703407e-01 9.31862056e-01 -5.37863255e-01
8.14521134e-01 1.50717989e-01 -3.38180661e-01 -9.85765994e-01
1.57193944e-01 -1.06300259e+00 -1.71477437e-01 1.57174635e+00
2.84672797e-01 9.13823117e-03 8.08220148e-01 8.71260107e-01
-1.92052141e-01 -2.30837300e-01 -9.73195374e-01 -6.47101939e-01
-5.46142161e-01 -7.22926319e-01 4.83938813e-01 1.01179731e+00
4.18362141e-01 -1.50959557e-02 -9.81721461e-01 -2.84046769e-01
3.78059953e-01 -2.58331239e-01 7.97846198e-01 -9.53131974e-01
-8.40747058e-02 -1.39853448e-01 -7.46250153e-01 -1.06579554e+00
4.89546001e-01 -8.18322957e-01 -4.01325345e-01 -1.64294291e+00
9.54845250e-01 -3.61982058e-03 3.26428056e-01 3.56153250e-01
5.48502684e-01 8.38166654e-01 4.55900103e-01 4.65441853e-01
-1.34909105e+00 2.97568262e-01 1.06863368e+00 -2.35508889e-01
6.98605478e-02 -2.18493521e-01 -5.72892785e-01 6.16959691e-01
3.10264438e-01 -2.92750150e-01 -2.65022695e-01 -6.02873206e-01
7.83989549e-01 5.91598272e-01 4.71070439e-01 -9.79876161e-01
1.66012555e-01 -1.95973575e-01 8.03982615e-02 -4.91829932e-01
8.89006317e-01 -5.55945814e-01 3.49351913e-01 1.32580087e-01
-6.87057078e-01 -7.29363337e-02 -3.86551544e-02 6.84491158e-01
-4.30523932e-01 -1.31369410e-02 2.08441764e-01 -1.08598210e-01
-5.79329967e-01 3.87733132e-01 -5.91712892e-01 1.54337883e-01
1.25405169e+00 -3.23661447e-01 -6.69275641e-01 -1.16580427e+00
-8.80205691e-01 4.52789634e-01 7.82405257e-01 8.01094413e-01
7.50317514e-01 -1.68757045e+00 -9.51661944e-01 -2.31675878e-01
2.91894108e-01 -4.21163857e-01 6.90260768e-01 7.28948355e-01
-5.70753753e-01 5.94693184e-01 -2.43608519e-01 -3.91004205e-01
-1.47902310e+00 1.93347931e-01 -2.03427047e-01 5.92376471e-01
-1.02163517e+00 8.11278939e-01 -3.68423462e-01 3.78664255e-01
2.65312254e-01 8.84360000e-02 -3.79897386e-01 4.08906937e-01
1.23428428e+00 4.41101104e-01 -4.87337828e-01 -1.28639853e+00
-3.68327409e-01 2.99575686e-01 -3.19031961e-02 -7.82432209e-04
1.33994246e+00 -6.82291985e-01 8.07266533e-02 5.21237314e-01
1.57137132e+00 -1.28818899e-01 -1.21130323e+00 -8.21606368e-02
-3.80111694e-01 -8.00929606e-01 -3.89926955e-02 -6.67105988e-02
-9.52219248e-01 3.16512138e-01 -1.53186873e-01 1.88418582e-01
5.69670379e-01 2.46394083e-01 1.60059202e+00 4.26903784e-01
4.49958473e-01 -1.04157305e+00 4.88925010e-01 2.59386808e-01
1.04148352e+00 -1.41586196e+00 6.23139367e-02 -4.03742969e-01
-9.55274045e-01 1.13419509e+00 1.90754920e-01 9.52112228e-02
5.60715020e-01 -3.55795287e-02 -5.09765208e-01 -4.53627072e-02
-9.99519289e-01 2.77313471e-01 1.07448235e-01 3.27276975e-01
5.08736551e-01 -2.11227592e-02 -3.73668969e-02 7.01463878e-01
-2.33552992e-01 2.60693878e-01 1.10775173e+00 8.63429189e-01
-3.25931042e-01 -6.26313210e-01 -1.11712947e-01 4.07508492e-01
-7.45587051e-01 2.68270463e-01 -3.41168165e-01 5.20328045e-01
-1.98561847e-01 1.14316595e+00 4.04994994e-01 -2.80481070e-01
-2.66390406e-02 -1.26121834e-01 3.11878979e-01 -7.07764566e-01
-3.28512490e-01 -3.87027234e-01 7.49616504e-01 -6.40482962e-01
-6.59244657e-01 -1.05040658e+00 -1.15347803e+00 -7.48356283e-01
3.26070368e-01 8.25178176e-02 6.12989783e-01 9.97882485e-01
3.92453849e-01 1.01308532e-01 3.37747037e-01 -9.70266283e-01
2.80147105e-01 -8.10204148e-01 -4.23414618e-01 6.64768815e-01
4.37300593e-01 -4.63178337e-01 -5.05439281e-01 4.03835177e-01]
|
[10.478250503540039, 0.7812630534172058]
|
f0b552bd-3b06-4ccd-b522-ed7cb4b53315
|
the-pytorch-kaldi-speech-recognition-toolkit
|
1811.07453
| null |
http://arxiv.org/abs/1811.07453v2
|
http://arxiv.org/pdf/1811.07453v2.pdf
|
The PyTorch-Kaldi Speech Recognition Toolkit
|
The availability of open-source software is playing a remarkable role in the
popularization of speech recognition and deep learning. Kaldi, for instance, is
nowadays an established framework used to develop state-of-the-art speech
recognizers. PyTorch is used to build neural networks with the Python language
and has recently spawn tremendous interest within the machine learning
community thanks to its simplicity and flexibility.
The PyTorch-Kaldi project aims to bridge the gap between these popular
toolkits, trying to inherit the efficiency of Kaldi and the flexibility of
PyTorch. PyTorch-Kaldi is not only a simple interface between these software,
but it embeds several useful features for developing modern speech recognizers.
For instance, the code is specifically designed to naturally plug-in
user-defined acoustic models. As an alternative, users can exploit several
pre-implemented neural networks that can be customized using intuitive
configuration files. PyTorch-Kaldi supports multiple feature and label streams
as well as combinations of neural networks, enabling the use of complex neural
architectures. The toolkit is publicly-released along with a rich documentation
and is designed to properly work locally or on HPC clusters.
Experiments, that are conducted on several datasets and tasks, show that
PyTorch-Kaldi can effectively be used to develop modern state-of-the-art speech
recognizers.
|
['Titouan Parcollet', 'Mirco Ravanelli', 'Yoshua Bengio']
|
2018-11-19
| null | null | null | null |
['noisy-speech-recognition', 'distant-speech-recognition']
|
['speech', 'speech']
|
[-3.80695939e-01 -2.97016412e-01 2.57056653e-01 -4.13797259e-01
-4.09678280e-01 -3.44021380e-01 5.07123053e-01 -2.02711254e-01
-5.93304694e-01 9.52263325e-02 -4.67225499e-02 -5.91836989e-01
1.03135094e-01 -6.64317250e-01 -2.97552466e-01 -8.22066009e-01
-7.96570107e-02 6.32870555e-01 3.96242589e-01 -4.05973077e-01
-1.96931571e-01 6.72513306e-01 -2.11432505e+00 3.16769511e-01
3.51683915e-01 6.84759617e-01 4.94734675e-01 9.34296668e-01
-4.97181982e-01 4.01487619e-01 -6.78447962e-01 -3.19809288e-01
-8.77807289e-02 2.55183205e-02 -4.89759207e-01 -5.47801077e-01
1.65048942e-01 2.65075136e-02 -2.39513516e-01 6.69866264e-01
8.23575795e-01 4.60319072e-02 1.71846561e-02 -9.44222212e-01
-2.61347890e-01 8.41681957e-01 2.43094042e-01 2.00850397e-01
5.93802035e-02 2.42517084e-01 7.91150451e-01 -8.21853280e-01
1.65322617e-01 1.18137681e+00 6.32802904e-01 6.10743284e-01
-1.04259360e+00 -6.67943776e-01 -2.55037159e-01 2.26610154e-01
-1.36902690e+00 -8.43220949e-01 4.70854491e-01 -3.88027102e-01
1.50751460e+00 5.50037742e-01 5.25418758e-01 1.25916231e+00
-1.89759612e-01 7.37353444e-01 8.18546891e-01 -5.59899628e-01
3.22413296e-01 2.72553921e-01 3.34656149e-01 5.70724010e-01
-3.81863594e-01 2.39719823e-01 -5.87493896e-01 -2.00606868e-01
6.48233712e-01 -1.89901665e-01 -1.63492382e-01 1.10963903e-01
-1.04929638e+00 6.79911673e-01 1.83041673e-02 7.84252942e-01
-1.68724418e-01 -1.42197579e-01 8.14755082e-01 4.33157474e-01
4.03871000e-01 1.50650159e-01 -4.60097641e-01 -8.65034401e-01
-9.31058228e-01 1.29892245e-01 1.13149083e+00 5.10151982e-01
7.01041102e-01 5.20709813e-01 3.81645769e-01 1.39176571e+00
3.50277066e-01 3.75866354e-01 1.03052402e+00 -5.95448434e-01
8.24477710e-03 2.54383683e-01 -5.92915893e-01 -4.49710786e-01
-3.48388016e-01 -4.80988860e-01 -7.78213978e-01 5.54398417e-01
3.67267221e-01 -3.88037823e-02 -8.68904173e-01 1.32281840e+00
2.83195466e-01 1.31002188e-01 1.35328323e-01 6.34249449e-01
1.00269818e+00 8.55451643e-01 -1.04732681e-02 1.87747791e-01
1.11622024e+00 -1.00584304e+00 -3.48871052e-01 -1.69604838e-01
7.08514154e-01 -8.18211615e-01 1.16815507e+00 5.54155231e-01
-1.01343846e+00 -6.59778595e-01 -8.99433374e-01 2.04720229e-01
-8.29302013e-01 -4.36857268e-02 6.28917873e-01 1.02868199e+00
-1.56547093e+00 6.66392148e-01 -1.05423784e+00 -4.77773607e-01
3.27179916e-02 3.46967369e-01 -4.38194573e-01 1.79592580e-01
-1.12603915e+00 8.50934207e-01 5.29683530e-01 -9.63287875e-02
-5.01308680e-01 -6.17954135e-01 -8.32507730e-01 1.61632493e-01
1.20396227e-01 -4.90298450e-01 1.41956949e+00 -1.06676006e+00
-2.15400171e+00 9.50073659e-01 1.21090800e-01 -4.45605606e-01
2.09255427e-01 4.05910574e-02 -6.34037197e-01 -1.62413463e-01
-4.36567873e-01 3.82093489e-01 7.52886295e-01 -4.83956456e-01
-4.09492910e-01 -1.09699555e-01 -3.49326640e-01 2.96103349e-03
-6.50835395e-01 6.03351176e-01 -6.46662951e-01 -4.94857997e-01
-4.21821356e-01 -6.40122116e-01 6.47199387e-03 -2.77066380e-01
-1.93452969e-01 -4.17690128e-01 7.89644659e-01 -5.35610020e-01
1.02451515e+00 -2.39480066e+00 -5.36437407e-02 1.47523835e-01
-1.99078377e-02 9.87893760e-01 -2.96850652e-02 5.72065771e-01
-1.66314259e-01 -1.41694963e-01 -2.13219658e-01 -6.34334087e-01
2.35964835e-01 5.08058846e-01 -9.31772068e-02 1.40729010e-01
2.16835663e-02 4.45977479e-01 -5.80510378e-01 7.06225261e-02
6.47452772e-01 8.52249980e-01 -2.54035980e-01 2.75007129e-01
-1.19838469e-01 1.63599521e-01 -5.67178279e-02 2.99928486e-01
6.48366570e-01 5.37422523e-02 8.01993534e-02 4.27230358e-01
-6.18934155e-01 7.42545247e-01 -1.14632452e+00 1.55841017e+00
-6.13327622e-01 6.56155705e-01 3.32394749e-01 -1.09065127e+00
1.03995264e+00 6.04224801e-01 1.80722371e-01 -3.31214547e-01
9.79034305e-02 4.69385356e-01 6.72125164e-03 -3.91538292e-01
3.04234058e-01 -4.47632000e-02 2.27595091e-01 4.94269788e-01
3.93141061e-01 1.15729671e-03 8.31222311e-02 -1.26183167e-01
1.07585037e+00 4.25919183e-02 1.55745029e-01 -9.92416218e-02
5.99756598e-01 -3.00349832e-01 2.77278334e-01 6.19590044e-01
8.01539496e-02 4.51148808e-01 1.13208465e-01 -4.89830881e-01
-1.07980561e+00 -1.13264823e+00 -4.12556052e-01 1.59676015e+00
-9.18180525e-01 -5.13251901e-01 -7.99738944e-01 -2.18260840e-01
-1.16132855e-01 5.28846800e-01 -1.47024691e-01 3.45961422e-01
-4.14682448e-01 -5.08236468e-01 1.12465191e+00 3.47336292e-01
3.50313783e-01 -1.55028450e+00 -4.69771892e-01 3.67514551e-01
3.55297208e-01 -8.74104738e-01 -9.70035717e-02 5.02318919e-01
-4.79988277e-01 -5.46899259e-01 -8.33204865e-01 -8.47046554e-01
8.84296820e-02 1.19858891e-01 1.23344231e+00 1.86446041e-01
-2.90617973e-01 4.93804157e-01 -5.14387310e-01 -4.18784529e-01
-8.93055975e-01 4.49984789e-01 1.80670276e-01 -4.88884524e-02
4.24153298e-01 -8.94645274e-01 1.86126769e-01 1.76498771e-01
-1.07376504e+00 2.48108298e-01 2.77209938e-01 6.00401282e-01
1.88823611e-01 -2.03313246e-01 6.97243035e-01 -6.59652114e-01
4.92649227e-01 -4.16979343e-01 -7.99876750e-01 -5.99920154e-02
-3.77393752e-01 -9.62515473e-02 7.96930969e-01 -2.83728927e-01
-9.11374152e-01 6.31159395e-02 -1.23606455e+00 -4.17850912e-01
-7.59776413e-01 7.42639422e-01 -2.01751947e-01 8.19629058e-03
4.20474082e-01 4.33772445e-01 2.70240217e-01 -9.19405282e-01
3.03162217e-01 1.25934803e+00 4.75626707e-01 -5.05346775e-01
2.91467041e-01 4.69888151e-02 -7.62236893e-01 -1.53807735e+00
-4.37460542e-02 -6.37402594e-01 -3.14596087e-01 -2.29007658e-02
6.28915846e-01 -6.57117188e-01 -6.73770010e-01 1.10133183e+00
-1.15039849e+00 -6.30903363e-01 -2.08831623e-01 3.50167930e-01
-2.30889365e-01 1.92295521e-01 -6.91139638e-01 -8.10512841e-01
-5.95799983e-01 -1.23978615e+00 5.70829451e-01 2.70876110e-01
-1.38776675e-01 -1.09817207e+00 2.95303851e-01 2.54975911e-02
8.58973920e-01 -3.94443810e-01 6.30575657e-01 -9.91721690e-01
-1.90376729e-01 -1.81786895e-01 1.26864791e-01 9.05149281e-01
-6.35913238e-02 4.29685533e-01 -1.56085324e+00 -1.85820848e-01
-5.07755615e-02 -1.73156157e-01 7.10705757e-01 1.41004622e-01
1.19353080e+00 -1.79048568e-01 -6.56185523e-02 7.08303928e-01
8.79947126e-01 2.29208395e-01 7.64147162e-01 4.10795301e-01
4.56941068e-01 6.16508782e-01 -1.50639340e-01 3.08812320e-01
2.20897287e-01 9.47231233e-01 2.35199392e-01 -1.13678515e-01
4.97645661e-02 1.39005661e-01 6.91326797e-01 1.50940859e+00
1.39590025e-01 5.46430349e-02 -1.12530184e+00 2.79421806e-01
-1.68025661e+00 -1.01169384e+00 -3.16329062e-01 2.41256881e+00
9.43921030e-01 1.68341026e-02 3.80543917e-01 3.53763670e-01
5.06286442e-01 1.81270227e-01 -1.61533564e-01 -8.93486142e-01
-1.44860342e-01 7.36666620e-01 -1.06407136e-01 4.96563584e-01
-9.65576947e-01 1.02872980e+00 6.21325970e+00 1.00680065e+00
-1.69217515e+00 3.18372369e-01 2.01276734e-01 -1.66282728e-01
7.87273794e-02 -3.64444792e-01 -1.02434981e+00 5.68516910e-01
1.68711102e+00 -2.10435167e-01 6.03115022e-01 1.17193484e+00
1.87272131e-01 1.83104828e-01 -9.37769949e-01 9.78536546e-01
-1.37268379e-01 -1.59438300e+00 -2.76633352e-01 5.66127663e-03
-2.80363038e-02 8.74184430e-01 -8.17828104e-02 4.06883597e-01
3.16228777e-01 -9.53357697e-01 7.15338290e-01 1.20897613e-01
7.70705819e-01 -8.22952628e-01 6.34878755e-01 4.06744838e-01
-1.01943910e+00 3.24267507e-01 -4.09656018e-01 -2.33210877e-01
8.55282098e-02 8.00203204e-01 -9.78658378e-01 3.80126715e-01
9.09384608e-01 6.01402402e-01 -2.94582278e-01 1.19726813e+00
-1.09335341e-01 9.14669991e-01 -6.47923708e-01 -7.22411275e-02
2.35103562e-01 -7.60976523e-02 4.29602951e-01 1.85880303e+00
1.96267530e-01 -4.48191345e-01 7.75109837e-03 5.89336395e-01
3.26309860e-01 2.91617125e-01 -6.76994681e-01 -1.88555375e-01
4.66274887e-01 1.36247039e+00 -3.76009315e-01 -3.74274582e-01
-4.18994755e-01 7.48882353e-01 2.97353238e-01 6.12458736e-02
-5.97901046e-01 -6.58365786e-01 1.18801427e+00 4.55432646e-02
3.80642623e-01 -6.26448512e-01 1.35667222e-02 -8.57404113e-01
-2.62710989e-01 -1.10698485e+00 1.97076246e-01 -7.43720055e-01
-1.15154028e+00 1.07546198e+00 -2.74522305e-01 -5.63687742e-01
-5.15396118e-01 -8.81980062e-01 -8.22119653e-01 1.13693881e+00
-1.22117484e+00 -9.36062276e-01 -1.52553052e-01 5.57054579e-01
5.09304583e-01 -4.92104471e-01 1.41428983e+00 6.88599706e-01
-8.04858506e-01 4.74479496e-01 2.91065991e-01 2.37758100e-01
5.37316203e-01 -1.13754332e+00 8.39653254e-01 5.62085629e-01
4.80912745e-01 6.64258838e-01 4.81854260e-01 -9.27373990e-02
-1.22630012e+00 -7.38340497e-01 8.90445948e-01 -9.73558128e-02
8.30141425e-01 -7.38241732e-01 -1.31405652e+00 5.14507234e-01
3.77100080e-01 -2.19973296e-01 9.56644118e-01 2.53971845e-01
-5.19990325e-01 -3.10645878e-01 -6.73518896e-01 3.97896677e-01
5.55674851e-01 -7.01817155e-01 -4.33685094e-01 2.19275966e-01
5.78897417e-01 -3.87107700e-01 -7.28496671e-01 -1.40588850e-01
6.33047223e-01 -1.51633728e+00 8.45281482e-01 -8.01764801e-02
-9.57626924e-02 -3.25592346e-02 7.13223740e-02 -1.34209085e+00
-1.28687490e-02 -8.63107979e-01 1.38170868e-01 1.53989697e+00
4.60796386e-01 -1.20693612e+00 5.36266983e-01 2.32137874e-01
-5.57731271e-01 -4.74438816e-01 -1.24960160e+00 -8.61697197e-01
6.91498742e-02 -1.05560970e+00 7.16839194e-01 7.34308124e-01
1.37620598e-01 1.07875630e-01 -7.05782995e-02 6.54197335e-02
1.78795904e-01 -4.88870561e-01 9.31500256e-01 -1.25931835e+00
-8.85332406e-01 -7.44117022e-01 -4.68242645e-01 -8.91102910e-01
2.50838578e-01 -1.05004203e+00 6.63360730e-02 -1.12997913e+00
-3.89338404e-01 -7.53576159e-01 -2.71504492e-01 9.52731907e-01
2.66371369e-01 1.77044570e-01 2.55438149e-01 2.18392059e-01
-6.78166002e-02 2.51277119e-01 5.87797761e-01 7.29821548e-02
-4.25563723e-01 1.99295536e-01 -3.61998141e-01 6.59170151e-01
8.96720886e-01 -5.30495644e-01 -1.69505998e-01 -4.56169516e-01
1.09958099e-02 -4.34867650e-01 3.71443987e-01 -1.49419880e+00
2.38890231e-01 3.47748429e-01 -1.69673145e-01 -3.17411244e-01
6.28369033e-01 -4.56198692e-01 3.19525570e-01 2.03998521e-01
-4.68430519e-02 -8.89554899e-03 6.16838872e-01 -1.85157686e-01
-1.54838145e-01 -3.72997344e-01 8.67271662e-01 -1.01016626e-01
-7.42081523e-01 4.27969582e-02 -7.99794197e-01 -1.31515950e-01
6.16587043e-01 -1.16274148e-01 -2.40852803e-01 -2.69092977e-01
-5.75683117e-01 -2.51195043e-01 3.34802568e-01 6.28797948e-01
4.46459711e-01 -8.83257031e-01 -6.75886869e-01 8.71660590e-01
7.62569010e-02 -2.35167578e-01 1.71589777e-01 7.11035550e-01
-6.38782084e-01 4.82706130e-01 -9.08704698e-02 -7.28407443e-01
-1.41169190e+00 2.10746169e-01 4.83708024e-01 -1.32244974e-01
-9.26807940e-01 8.93480539e-01 -9.34242830e-02 -8.63213956e-01
3.99202645e-01 -2.06846595e-01 3.70782129e-02 -1.82072833e-01
9.83413756e-01 1.17011793e-01 5.50786436e-01 -3.62777829e-01
-4.10028905e-01 1.85776073e-02 -4.63822223e-02 -3.70394826e-01
1.63614273e+00 3.22278321e-01 -1.77400947e-01 8.44378769e-01
1.07085335e+00 -1.59635752e-01 -8.82357061e-01 -5.49218990e-02
-2.96663810e-02 -9.21858568e-03 2.75924444e-01 -8.79877627e-01
-1.06284595e+00 1.20331872e+00 5.56044221e-01 4.45706278e-01
8.82925093e-01 -3.42237204e-02 7.90859461e-01 2.71735013e-01
4.52926219e-01 -8.21597636e-01 -3.66139978e-01 1.03650546e+00
7.14309335e-01 -7.50485003e-01 -5.89794993e-01 -1.50966262e-02
-4.31909174e-01 1.40526617e+00 1.98118523e-01 1.13997586e-01
6.88766718e-01 8.25358093e-01 4.94197577e-01 7.88337216e-02
-7.61767924e-01 -3.69031012e-01 -6.89092278e-02 7.36591399e-01
7.29940951e-01 1.17652841e-01 9.65179428e-02 5.01661479e-01
-4.10775512e-01 9.37518254e-02 2.33017743e-01 7.67787397e-01
-4.87920672e-01 -1.59934640e+00 -5.69420934e-01 2.33828023e-01
-3.39506179e-01 -3.98582995e-01 -1.02574058e-01 5.07722318e-01
3.50433327e-02 8.25834632e-01 2.54316539e-01 -4.79763001e-01
3.97939570e-02 4.74542111e-01 1.27382234e-01 -7.82089949e-01
-1.20159721e+00 6.88274503e-02 2.16810554e-01 -3.32946986e-01
5.99070676e-02 -3.88945997e-01 -1.19972956e+00 -5.23317635e-01
-7.51405880e-02 3.72235388e-01 1.33101785e+00 9.29982126e-01
5.91410637e-01 4.85667825e-01 2.27283254e-01 -1.19383311e+00
-3.86176378e-01 -9.52473879e-01 -6.84673131e-01 -2.13341817e-01
1.22420244e-01 -2.85591304e-01 -4.63320732e-01 -1.81483492e-01]
|
[14.349201202392578, 6.40721321105957]
|
33c39291-f8f6-4381-8894-3bcf2aff32c0
|
a-network-community-detection-method-with
|
2305.13012
| null |
https://arxiv.org/abs/2305.13012v1
|
https://arxiv.org/pdf/2305.13012v1.pdf
|
A network community detection method with integration of data from multiple layers and node attributes
|
Multilayer networks are in the focus of the current complex network study. In such networks multiple types of links may exist as well as many attributes for nodes. To fully use multilayer -- and other types of complex networks in applications, the merging of various data with topological information renders a powerful analysis. First, we suggest a simple way of representing network data in a data matrix where rows correspond to the nodes, and columns correspond to the data items. The number of columns is allowed to be arbitrary, so that the data matrix can be easily expanded by adding columns. The data matrix can be chosen according to targets of the analysis, and may vary a lot from case to case. Next, we partition the rows of the data matrix into communities using a method which allows maximal compression of the data matrix. For compressing a data matrix, we suggest to extend so called regular decomposition method for non-square matrices. We illustrate our method for several types of data matrices, in particular, distance matrices, and matrices obtained by augmenting a distance matrix by a column of node degrees, or by concatenating several distances matrices corresponding to layers of a multilayer network. We illustrate our method with synthetic power-law graphs and two real networks: an Internet autonomous systems graph and a world airline graph. We compare the outputs of different community recovery methods on these graphs, and discuss how incorporating node degrees as a separate column to the data matrix leads our method to identify community structures well-aligned with tiered hierarchical structures commonly encountered in complex scale-free networks.
|
['Tomi Räty', 'Lasse Leskelä', 'Hannu Reittu']
|
2023-05-22
| null | null | null | null |
['community-detection']
|
['graphs']
|
[ 1.25953943e-01 2.01767743e-01 7.62658268e-02 3.02386340e-02
4.04670656e-01 -8.38216603e-01 5.12597024e-01 4.80693221e-01
-1.23824179e-01 6.67921364e-01 1.12000354e-01 -4.40755874e-01
-5.38863361e-01 -1.29686773e+00 -3.81547987e-01 -5.30801535e-01
-1.04012108e+00 7.58375585e-01 6.02953255e-01 -5.30699909e-01
1.28734916e-01 7.07988322e-01 -1.14410806e+00 2.46590063e-01
6.77591562e-01 4.98305678e-01 -1.97023638e-02 7.14041114e-01
-2.65611589e-01 4.89169627e-01 -5.98054767e-01 -3.18580747e-01
4.67784196e-01 -1.61980733e-01 -8.55232239e-01 5.35476029e-01
3.39845754e-02 2.42816970e-01 -4.32886183e-01 1.04217148e+00
1.07085906e-01 -2.99747199e-01 7.78880060e-01 -1.44805479e+00
3.40899229e-02 9.52041924e-01 -7.84921825e-01 8.03152174e-02
5.03852367e-01 -2.86128014e-01 1.10103452e+00 -6.30066514e-01
1.04533625e+00 1.33235323e+00 7.64043212e-01 -8.84974226e-02
-1.66636240e+00 -4.68830645e-01 6.23214208e-02 -6.54220721e-03
-1.50500178e+00 -2.38570452e-01 7.97274530e-01 -7.68942237e-01
5.46012223e-01 4.35870171e-01 1.00328279e+00 4.59808439e-01
1.17935605e-01 9.98135135e-02 9.32329476e-01 -3.83334875e-01
2.12689802e-01 -4.07582000e-02 2.39703760e-01 7.34415650e-01
7.83584237e-01 -3.64456892e-01 3.91472653e-02 -6.41544223e-01
6.97042584e-01 1.66996479e-01 -1.40605927e-01 -8.37110043e-01
-1.39608479e+00 8.38881433e-01 6.44399166e-01 5.17037809e-01
-2.59045601e-01 -9.16508287e-02 2.72510290e-01 7.80932546e-01
2.33342752e-01 2.11688250e-01 -5.10129556e-02 6.18104339e-01
-7.44926870e-01 8.84253085e-02 1.22189486e+00 8.92292857e-01
1.12843227e+00 -3.69334519e-01 4.82927352e-01 7.15600193e-01
2.29714572e-01 1.85557887e-01 -1.76380411e-01 -6.58373475e-01
6.55243337e-01 1.06006849e+00 -3.11659932e-01 -1.57684350e+00
-6.43615603e-01 -3.54772300e-01 -1.59274137e+00 3.16258110e-02
5.55664301e-01 -5.45601211e-02 -4.09880072e-01 1.60406995e+00
2.45370641e-01 -2.86525011e-01 -8.37525353e-03 4.15916234e-01
3.53978395e-01 4.31696415e-01 -6.23740196e-01 -4.40355957e-01
1.25199437e+00 -5.08679092e-01 -3.98145556e-01 3.08381431e-02
6.94195807e-01 -4.23185557e-01 4.97754574e-01 2.91147709e-01
-9.28708553e-01 -1.15632154e-01 -1.04797602e+00 5.53632438e-01
-5.14003992e-01 -3.89433056e-01 4.13944989e-01 4.03571427e-01
-1.27244484e+00 7.92123079e-01 -4.61068094e-01 -5.79506516e-01
-7.41367936e-02 5.05082369e-01 -8.21171463e-01 -1.44858196e-01
-1.19625473e+00 4.95150328e-01 4.15525019e-01 7.80277848e-02
-3.03551435e-01 -1.55148402e-01 -6.68784082e-01 1.23788372e-01
4.47429866e-01 -6.60934448e-01 1.88769937e-01 -7.33800530e-01
-5.69045842e-01 6.89075172e-01 1.74645036e-01 -3.12257230e-01
2.51071781e-01 8.41333687e-01 -5.41763783e-01 2.71999687e-01
1.80356473e-01 1.71127766e-01 6.88941300e-01 -1.45491445e+00
-1.74334347e-01 -2.36276731e-01 3.42507780e-01 -1.55594394e-01
-4.51635689e-01 -6.83433861e-02 -1.45131022e-01 -5.27352691e-01
4.99098152e-01 -1.07221365e+00 -6.44899964e-01 3.93463373e-02
-8.57933342e-01 9.55914930e-02 7.13000536e-01 -3.77998501e-01
1.57541704e+00 -2.01082945e+00 5.63940585e-01 1.20130002e+00
9.97435629e-01 -2.23043218e-01 -2.68789113e-01 1.10860240e+00
-3.52654457e-01 5.57398438e-01 -5.20413101e-01 -3.15534584e-02
-2.26485506e-01 4.75010216e-01 2.21592888e-01 5.59606552e-01
6.13369308e-02 2.91546673e-01 -7.86497056e-01 -5.58132768e-01
-3.11474621e-01 1.07369557e-01 -7.05381811e-01 -1.38273686e-01
2.75443196e-01 5.97967282e-02 -2.89964288e-01 4.23557043e-01
7.16238141e-01 -6.15284860e-01 8.88364911e-01 -2.89986789e-01
-1.12462789e-01 8.13741833e-02 -1.76254964e+00 9.71586049e-01
1.13161758e-01 4.68822837e-01 5.25703371e-01 -1.00187004e+00
8.52778673e-01 1.58345610e-01 6.65773034e-01 -2.75313146e-02
-1.30162552e-01 5.40968291e-02 6.44119561e-01 -1.25767529e-01
3.35863978e-01 5.07500768e-02 -4.97940481e-02 6.44075572e-01
-2.69613922e-01 -6.83286935e-02 8.65092993e-01 7.68747628e-01
1.71535063e+00 -1.00949883e+00 3.44599873e-01 -6.15062773e-01
6.07009411e-01 -9.12549570e-02 4.61958855e-01 3.87006551e-01
1.54189333e-01 3.32790971e-01 1.25349605e+00 -3.15599918e-01
-1.26931143e+00 -9.29907680e-01 -6.43846169e-02 5.03009140e-01
-1.08696222e-02 -8.07990253e-01 -5.51257193e-01 -4.42047954e-01
2.74945587e-01 -3.64522278e-01 -6.63568199e-01 2.21937209e-01
-5.06835818e-01 -8.21399450e-01 3.87892872e-01 -1.22858472e-01
1.55955911e-01 -7.40738213e-01 2.37002060e-01 2.06365019e-01
-2.05038190e-01 -9.75284338e-01 -3.45053881e-01 2.07133844e-01
-1.06915450e+00 -1.38341379e+00 -3.55162382e-01 -9.15452719e-01
1.15959382e+00 4.50469077e-01 1.17864037e+00 6.80245757e-01
-8.31471011e-02 9.24498439e-02 -2.94040918e-01 4.23342317e-01
-6.93789780e-01 2.14656651e-01 4.40342993e-01 2.08616287e-01
-3.97990882e-01 -1.17583358e+00 -3.40969235e-01 4.26912099e-01
-1.13858402e+00 5.08443639e-02 4.67432082e-01 6.59810245e-01
1.63503870e-01 4.89134341e-01 3.46099615e-01 -9.56639051e-01
1.07262111e+00 -9.09788966e-01 -4.69297439e-01 2.42845416e-01
-4.96855795e-01 1.26809418e-01 5.57744324e-01 -3.93375307e-01
1.62828825e-02 -1.20806796e-02 3.72964352e-01 -1.51897788e-01
3.64927322e-01 9.57307220e-01 -1.36252180e-01 -4.04097676e-01
5.96681297e-01 -1.26919240e-01 4.02790129e-01 -3.47405642e-01
3.39842528e-01 5.06315470e-01 3.45273130e-02 -5.13003111e-01
1.08380461e+00 4.06067550e-01 4.86879081e-01 -9.28190053e-01
5.14573567e-02 -3.59249473e-01 -1.10359848e+00 -2.28818819e-01
2.79119998e-01 -5.53660214e-01 -7.43245304e-01 2.22578764e-01
-1.01050925e+00 -9.72390696e-02 -1.38873070e-01 1.09435774e-01
-6.92569166e-02 7.92194009e-01 -9.19820607e-01 -4.65263277e-01
1.48988023e-01 -8.36586773e-01 2.57123291e-01 -4.69261974e-01
-1.18150823e-01 -1.29802632e+00 3.53936225e-01 -1.55350521e-01
3.90750945e-01 4.36383098e-01 1.27886939e+00 -6.44604564e-01
-6.56133115e-01 -2.21738219e-01 -2.73393154e-01 1.01950012e-01
2.58908898e-01 4.56119895e-01 -7.72103146e-02 -5.68872809e-01
-4.24866110e-01 2.06171513e-01 6.68752074e-01 -5.85024850e-03
6.98864996e-01 -7.45332301e-01 -6.40423536e-01 3.79687250e-01
1.35106885e+00 -2.04260737e-01 6.86369359e-01 1.76916316e-01
8.28308523e-01 9.87615108e-01 -1.63552672e-01 4.32191908e-01
4.66816992e-01 4.97850746e-01 6.18957818e-01 -1.35688409e-01
1.99455336e-01 8.13715979e-02 1.66199863e-01 1.49390972e+00
-2.28113353e-01 -1.74929887e-01 -1.10118961e+00 4.24142331e-01
-1.75671721e+00 -1.12833214e+00 -5.42951882e-01 2.16510296e+00
7.73089945e-01 3.30590874e-01 5.23788035e-01 5.00580907e-01
1.07595849e+00 1.60478339e-01 -1.86880857e-01 -8.35671201e-02
-3.76370013e-01 -2.32921004e-01 4.96477991e-01 7.59068191e-01
-7.15956926e-01 3.15133154e-01 7.03872776e+00 4.91064876e-01
-6.85352623e-01 -2.65102744e-01 2.15863660e-01 2.15288386e-01
-5.69652021e-01 4.08834398e-01 -2.43651897e-01 2.60215551e-01
8.64937901e-01 -3.89739037e-01 6.71095490e-01 3.37068349e-01
-1.04124574e-02 -1.52482302e-03 -1.08293319e+00 6.04537845e-01
-3.51362467e-01 -1.36164653e+00 3.61605883e-02 7.21632421e-01
6.78623259e-01 -8.54028240e-02 -4.13234979e-01 -1.30536184e-01
5.62218487e-01 -8.21873903e-01 2.99515128e-01 3.66229326e-01
6.43022716e-01 -5.13490260e-01 4.78301257e-01 4.39128876e-01
-1.65901875e+00 -7.71498829e-02 -4.06051725e-01 -2.69769043e-01
1.34523083e-02 1.00941849e+00 -7.68100858e-01 6.98234737e-01
4.74311292e-01 8.90616298e-01 -6.62132859e-01 1.15913367e+00
4.76301700e-01 3.03062260e-01 -6.38309896e-01 3.00563611e-02
4.16846164e-02 -7.46006310e-01 7.46328354e-01 1.14290178e+00
3.10945928e-01 -8.16499442e-02 3.07326347e-01 6.47496700e-01
-2.03579143e-01 1.60679489e-01 -1.07074130e+00 -1.21856093e-01
6.37626529e-01 1.45170784e+00 -1.06493688e+00 -3.12841207e-01
-4.43941057e-01 5.42028725e-01 4.02184635e-01 4.21935290e-01
-1.87070310e-01 -5.65030873e-01 6.36920214e-01 7.64652908e-01
1.32044166e-01 -6.09260499e-01 8.47990364e-02 -1.15083170e+00
-1.45501550e-02 -1.01303875e+00 2.90477663e-01 -4.54903394e-01
-1.52051353e+00 8.05231869e-01 1.70136794e-01 -1.46208966e+00
-2.10838050e-01 -4.07476455e-01 -5.20510554e-01 6.23799026e-01
-9.70704496e-01 -6.63101375e-01 -1.39241353e-01 8.38494241e-01
-3.75515878e-01 -1.85356230e-01 6.32263184e-01 4.50041741e-01
-6.08085513e-01 9.64069217e-02 2.29882225e-01 3.94476950e-01
3.64530414e-01 -1.10940039e+00 5.44016480e-01 7.10119247e-01
2.10758038e-02 8.30276370e-01 6.99587762e-01 -7.62292802e-01
-1.23722291e+00 -7.40817130e-01 9.19606209e-01 -8.39614421e-02
1.28189528e+00 -7.52595007e-01 -1.02617455e+00 7.66589463e-01
1.19488828e-01 1.08039558e-01 4.16193217e-01 1.76676974e-01
-3.82489353e-01 -1.40468225e-01 -1.19612062e+00 7.36326277e-01
1.28238654e+00 -4.76476997e-01 -1.70833677e-01 4.57181871e-01
5.84721565e-01 2.24727139e-01 -1.21002221e+00 2.72452563e-01
3.64758641e-01 -1.11533666e+00 1.00400531e+00 -4.78028774e-01
1.18306570e-01 -6.38647854e-01 -1.13369115e-01 -1.39060819e+00
-7.92558610e-01 -5.72542191e-01 -3.82780023e-02 1.04571319e+00
4.96072263e-01 -8.89056921e-01 7.68424630e-01 1.60279453e-01
4.05169904e-01 -5.44336975e-01 -8.63946378e-01 -6.55233681e-01
-1.89088225e-01 1.58928737e-01 6.21568382e-01 1.28157139e+00
4.02543664e-01 5.66719711e-01 -2.36096814e-01 1.54364109e-01
8.63057613e-01 -3.16191502e-02 7.85488427e-01 -1.92144692e+00
-2.06144139e-01 -6.19235694e-01 -6.18082404e-01 -7.51825929e-01
-7.58552551e-02 -1.07148468e+00 -5.38237214e-01 -1.53474450e+00
3.30328256e-01 -1.03620136e+00 2.16909334e-01 4.26648438e-01
3.21113646e-01 1.51072472e-01 2.99394727e-01 5.70437729e-01
-3.00593823e-01 1.74683943e-01 1.29866934e+00 -3.61317933e-01
-1.90450475e-01 3.75977084e-02 -5.64392805e-01 6.44182682e-01
5.54215789e-01 -5.83377004e-01 -4.04314399e-01 1.90382600e-02
6.91579223e-01 5.48897684e-01 9.10994634e-02 -9.33062851e-01
4.41164374e-01 -1.96593255e-01 -6.70973882e-02 -4.82503325e-01
2.16227517e-01 -1.08241045e+00 7.89747059e-01 7.25641131e-01
-5.89255914e-02 6.10696495e-01 -2.51084268e-01 6.01234436e-01
-2.63897747e-01 -1.26946166e-01 4.64599222e-01 -1.04653150e-01
-1.50645882e-01 2.90987343e-01 -4.90770817e-01 -2.14874670e-01
7.99974740e-01 -4.89665240e-01 -5.32901466e-01 -5.43954134e-01
-1.14698052e+00 3.72831672e-01 8.17938328e-01 9.42942724e-02
5.04185259e-01 -1.51218796e+00 -8.25743616e-01 2.93080062e-01
2.25868914e-02 -1.62655026e-01 -1.16857328e-01 1.02788413e+00
-6.61025167e-01 -8.76254216e-02 -4.45932567e-01 -6.33926213e-01
-1.46531951e+00 7.29206026e-01 2.79961765e-01 -6.70004666e-01
-4.69482422e-01 1.80136323e-01 1.15782455e-01 -5.39589345e-01
6.95795577e-04 -2.74615794e-01 -4.20580983e-01 3.27124268e-01
2.09662110e-01 2.99430162e-01 -1.22923359e-01 -6.77389443e-01
-4.88393456e-01 6.67731643e-01 1.44360945e-01 -1.34416930e-02
1.35883427e+00 -4.45795029e-01 -1.03748322e+00 4.31590587e-01
1.08281577e+00 4.10684407e-01 -5.50212622e-01 -3.40459317e-01
1.16973445e-01 -1.47907078e-01 -6.18857443e-01 -8.38985294e-02
-1.26053417e+00 5.09632230e-01 -1.52981296e-01 1.16431141e+00
1.04202187e+00 -1.56735610e-02 5.54387420e-02 5.82069278e-01
5.56577146e-01 -4.60319221e-01 1.69090346e-01 6.72707617e-01
8.74821246e-01 -7.94647634e-01 3.23622614e-01 -7.92068899e-01
-1.41943067e-01 1.17825294e+00 1.04653046e-01 -4.23606187e-01
1.09442639e+00 4.40458030e-01 -3.43764812e-01 -3.13997626e-01
-8.81753802e-01 -2.33145580e-01 9.61093009e-02 5.38817942e-01
1.17258638e-01 2.14070648e-01 -3.72383058e-01 1.45725250e-01
-2.68362701e-01 -6.21309876e-01 1.07116747e+00 7.41699517e-01
-6.44293427e-01 -1.49202740e+00 -5.48942149e-01 6.56496227e-01
4.81017046e-02 -1.45775318e-01 -7.55801916e-01 8.99004936e-01
-1.04293473e-01 9.33888137e-01 2.08838940e-01 -8.05173755e-01
2.70253599e-01 -3.12467277e-01 3.02847981e-01 -7.20928848e-01
-4.92775649e-01 -7.75415674e-02 4.37459081e-01 -1.37642577e-01
-4.27970290e-01 -5.45516253e-01 -1.03408635e+00 -1.05608261e+00
-3.55542183e-01 3.40974152e-01 3.74251962e-01 5.64966559e-01
2.97066212e-01 4.37569618e-01 8.53383660e-01 -8.84133577e-01
-1.20384665e-02 -8.76484573e-01 -1.09677720e+00 3.66850108e-01
4.07020628e-01 -5.69332778e-01 -6.61627471e-01 -1.79289728e-01]
|
[6.952853679656982, 5.283052444458008]
|
4bfc942d-c9bb-4ff6-9768-c5a31cf144b9
|
principled-analysis-of-energy-discourse
| null | null |
https://aclanthology.org/2021.alta-1.11
|
https://aclanthology.org/2021.alta-1.11.pdf
|
Principled Analysis of Energy Discourse across Domains with Thesaurus-based Automatic Topic Labeling
|
With the increasing impact of Natural Language Processing tools like topic models in social science research, the experimental rigor and comparability of models and datasets has come under scrutiny. Especially when contributing to research on topics with worldwide impacts like energy policy, objective analyses and reliable datasets are necessary. We contribute toward this goal in two ways: first, we release two diachronic corpora covering 23 years of energy discussions in the U.S. Energy Information Administration. Secondly, we propose a simple and theoretically sound method for automatic topic labelling drawing on political thesauri. We empirically evaluate the quality of our labels, and apply our labelling to topics induced by diachronic topic models on our energy corpora, and present a detailed analysis.
|
['Lea Frermann', 'Alfonso Martinez Arranz', 'Thomas Scelsi']
| null | null | null | null |
alta-2021-12
|
['topic-models']
|
['natural-language-processing']
|
[ 1.68699086e-01 5.62698662e-01 -5.31751096e-01 -4.41750258e-01
-1.03660738e+00 -9.87215042e-01 1.30263424e+00 7.38508224e-01
-4.39708292e-01 6.24486744e-01 1.03778934e+00 -7.36910880e-01
-1.95851266e-01 -9.43861961e-01 -4.14141059e-01 -4.73460793e-01
2.55250335e-01 5.55552781e-01 4.65096273e-02 1.71322394e-02
4.95464593e-01 8.05551335e-02 -1.21418226e+00 7.64570087e-02
9.79509771e-01 5.28749406e-01 -1.54047664e-02 3.46519828e-01
-5.08684635e-01 5.22376060e-01 -4.18586880e-01 -5.97117484e-01
-7.95200840e-03 -3.54422450e-01 -1.33840895e+00 -7.27506876e-02
1.07446223e-01 2.63936877e-01 -1.65305406e-01 1.23140001e+00
2.51953900e-01 1.82709917e-01 8.22378099e-01 -1.22040939e+00
-1.63793206e-01 1.12739348e+00 -4.26886559e-01 3.42517287e-01
2.47823998e-01 -2.22284392e-01 1.30536878e+00 -5.10868132e-01
1.00188291e+00 1.20699441e+00 5.18334746e-01 7.76003078e-02
-1.15436733e+00 -6.99518263e-01 1.58721656e-01 8.78876075e-02
-1.02852976e+00 -3.24559212e-01 9.60409284e-01 -8.49274218e-01
6.71568692e-01 3.67878199e-01 6.56290472e-01 1.09390378e+00
2.34893531e-01 3.35661650e-01 1.57622349e+00 -6.07240975e-01
3.96495342e-01 3.10658664e-01 4.32331771e-01 7.37788528e-02
5.00525355e-01 -2.71279186e-01 -4.48272437e-01 -5.63552916e-01
2.07730383e-01 -4.31739926e-01 6.51613027e-02 -2.19580173e-01
-1.07888544e+00 1.11915922e+00 1.30015204e-03 7.77627110e-01
-4.76295263e-01 -7.93799385e-02 5.01424313e-01 3.09047792e-02
1.11317003e+00 3.44240904e-01 -3.84849697e-01 -4.85204101e-01
-9.84581113e-01 3.09269071e-01 1.00180697e+00 6.66802406e-01
5.42683840e-01 -5.93484342e-01 1.63317308e-01 6.14299178e-01
6.47530735e-01 4.96431679e-01 1.66385382e-01 -9.37984109e-01
5.45164049e-01 5.34113526e-01 2.80948400e-01 -1.18365598e+00
-6.31117761e-01 -5.43414317e-02 -5.79725623e-01 -1.72768325e-01
3.86605144e-01 -1.29410028e-01 -5.81133127e-01 1.53221273e+00
4.55460757e-01 -6.02688551e-01 1.53987288e-01 5.08972049e-01
7.54856527e-01 8.74372065e-01 8.85056198e-01 -4.32391554e-01
1.85673547e+00 -3.07097644e-01 -9.44039106e-01 -3.61464359e-03
6.01106226e-01 -9.14532602e-01 8.05761218e-01 1.88023403e-01
-9.79667544e-01 5.02505116e-02 -5.24470866e-01 -1.99414551e-01
-4.39909518e-01 -3.03180546e-01 8.41563880e-01 7.75180757e-01
-8.42328191e-01 3.85227174e-01 -9.63382959e-01 -9.03245509e-01
3.25434864e-01 -2.54142489e-02 -1.65323690e-02 3.49661618e-01
-1.35253716e+00 1.03717256e+00 4.88269866e-01 -1.19006105e-01
-3.50707740e-01 -3.92370135e-01 -7.54866779e-01 4.54954952e-02
5.47980845e-01 -3.30757290e-01 1.48254621e+00 -5.44051766e-01
-1.13254678e+00 9.20647383e-01 -1.97172344e-01 -4.35538352e-01
3.95062685e-01 -1.47407837e-02 -3.57819229e-01 2.95962412e-02
2.57038057e-01 4.59349543e-01 1.01389118e-01 -1.14568710e+00
-7.98453450e-01 -2.76242912e-01 1.68378904e-01 1.81248233e-01
-4.25057918e-01 7.55282700e-01 5.47626242e-02 -6.56934321e-01
1.55633137e-01 -9.48964357e-01 -3.30101669e-01 -6.27182722e-01
-3.59961987e-01 -8.12133372e-01 5.77703893e-01 -9.56816375e-01
1.28227365e+00 -1.76978338e+00 6.29382953e-02 1.43678501e-01
3.23610991e-01 -3.07531267e-01 6.99564397e-01 6.64112449e-01
2.91802753e-02 7.35884249e-01 -1.91614822e-01 -3.94493900e-02
4.31577712e-01 7.22198561e-02 -6.11186802e-01 4.74510819e-01
-1.23283729e-01 6.95533454e-01 -9.78098333e-01 -6.17871761e-01
1.68359369e-01 1.81471139e-01 -4.86540437e-01 -1.80030033e-01
-4.83406484e-01 1.38104305e-01 -6.83847606e-01 4.22870785e-01
1.30393088e-01 -2.93328941e-01 5.08420348e-01 -1.50293633e-01
-6.85729980e-01 1.09648073e+00 -8.20514083e-01 1.59001064e+00
-3.37916434e-01 8.43239486e-01 3.58759277e-02 -6.87835932e-01
5.39153934e-01 4.75078851e-01 7.78163671e-01 -5.87460220e-01
3.63804191e-01 -1.44140506e-02 -1.60459146e-01 -2.54771382e-01
8.72711897e-01 -6.48970008e-01 -4.65665221e-01 8.82913888e-01
1.73604619e-02 -5.26992142e-01 2.12719202e-01 3.90769541e-01
6.78534627e-01 2.61846986e-02 4.31794316e-01 -9.19428766e-01
5.22564091e-02 4.63517070e-01 4.60329980e-01 3.14153165e-01
-1.08400844e-01 2.15651363e-01 6.77780986e-01 -4.36602026e-01
-1.36583316e+00 -5.82337976e-01 -6.91019475e-01 9.98219550e-01
-1.79015934e-01 -7.75664091e-01 -7.44845748e-01 -7.87501633e-01
-3.22429508e-01 1.13501084e+00 -4.70107824e-01 4.15314108e-01
-3.80949080e-01 -9.85662878e-01 1.19199961e-01 1.86416253e-01
3.22940081e-01 -8.31458271e-01 -5.34924507e-01 1.39841780e-01
-5.27187586e-01 -1.01626205e+00 1.33602664e-01 1.84354886e-01
-6.20210290e-01 -1.13681722e+00 -4.48153257e-01 -3.32249492e-01
2.87423223e-01 4.20983844e-02 1.27373815e+00 -4.58345175e-01
2.36134201e-01 3.04125607e-01 -3.05228025e-01 -7.89445043e-01
-7.41650522e-01 4.54464853e-01 -3.96173522e-02 -6.69131100e-01
7.36673415e-01 -4.65619653e-01 -4.46472377e-01 1.34028271e-01
-9.07794893e-01 2.61221051e-01 1.36154771e-01 1.72197238e-01
9.86219123e-02 3.92039478e-01 4.59096670e-01 -1.16145825e+00
6.52610481e-01 -8.59502017e-01 -7.35326350e-01 9.94680375e-02
-7.89945066e-01 -8.15677494e-02 -6.50792047e-02 1.92803502e-01
-1.27540052e+00 -4.96477634e-01 -2.09286705e-01 3.99745315e-01
-3.01541001e-01 7.02859044e-01 -9.37635452e-03 5.41933477e-01
6.79973483e-01 -4.65718150e-01 -4.46266055e-01 -4.99698848e-01
6.57447278e-01 9.54905868e-01 2.20592365e-01 -8.23405623e-01
9.57794070e-01 5.87316215e-01 -4.54645038e-01 -8.82489741e-01
-9.70042169e-01 -6.22901499e-01 -5.21026850e-01 -3.58936191e-01
1.04600394e+00 -9.71743047e-01 -3.72904122e-01 2.58168697e-01
-1.18703556e+00 -1.62640408e-01 -3.56909394e-01 6.53999746e-01
-3.96616131e-01 2.15110093e-01 -3.10277134e-01 -8.28979433e-01
-1.33206949e-01 -8.90508175e-01 9.13765252e-01 2.49318212e-01
-8.30930531e-01 -1.36587834e+00 4.24438894e-01 4.51830149e-01
2.94002801e-01 3.44319642e-01 1.06590879e+00 -8.05470049e-01
-2.90493965e-01 1.64604962e-01 -1.83843225e-01 -1.53942347e-01
3.20745975e-01 9.43694562e-02 -1.11311305e+00 -5.27401678e-02
1.61940321e-01 -2.79518127e-01 7.22924411e-01 2.98252106e-01
6.47810400e-01 -5.02443969e-01 -5.22386611e-01 -2.64715016e-01
1.28153634e+00 1.68014243e-01 2.97977626e-01 6.12802267e-01
3.08999956e-01 1.09665072e+00 4.73377705e-01 3.00581902e-01
7.70664454e-01 5.80547988e-01 -1.81737319e-01 4.97498829e-03
2.95905862e-02 -2.64388710e-01 7.73989409e-02 1.21345282e+00
-7.06230327e-02 -3.02921772e-01 -1.36111999e+00 9.21588838e-01
-1.66545951e+00 -1.01944864e+00 -2.59702355e-01 1.78515244e+00
9.95669007e-01 3.40426534e-01 1.80118725e-01 -1.45688094e-02
7.07174420e-01 4.63667840e-01 2.71346569e-02 -3.76612335e-01
-6.93947226e-02 -2.13756296e-03 4.15742487e-01 4.95000124e-01
-1.05033708e+00 7.62832522e-01 6.73573208e+00 7.19677448e-01
-5.56875587e-01 4.31177765e-01 7.44725287e-01 1.52987942e-01
-9.21051860e-01 3.08500171e-01 -7.40223169e-01 4.32410598e-01
1.21347463e+00 -7.50828207e-01 -1.74527854e-01 6.45188630e-01
4.81641859e-01 -3.77009213e-01 -6.92666471e-01 2.31579378e-01
-2.12853312e-01 -1.38026023e+00 -3.43969434e-01 4.23280716e-01
9.23425436e-01 9.50213522e-02 -2.47167140e-01 1.46537140e-01
1.00938165e+00 -7.20688105e-01 1.01775575e+00 9.90657806e-02
4.99318063e-01 -5.46115935e-01 7.29897976e-01 3.10845435e-01
-9.17414188e-01 1.23763554e-01 -2.29542851e-01 -1.46807522e-01
4.61728245e-01 9.09177482e-01 -6.74551785e-01 7.24467814e-01
6.54183567e-01 3.37828428e-01 -1.35985956e-01 7.55920529e-01
-3.50487322e-01 1.24086022e+00 -5.14095485e-01 -4.14730944e-02
3.37345272e-01 -3.76091748e-01 4.75172848e-01 1.13662946e+00
8.00670609e-02 4.02679592e-01 1.54986098e-01 6.60192430e-01
-2.45210826e-01 3.34780544e-01 -6.27151310e-01 -4.81315613e-01
4.00097191e-01 1.31295729e+00 -1.26056850e+00 -3.71109098e-01
-5.23597836e-01 2.06644144e-02 3.97865102e-02 6.16064519e-02
-5.96852422e-01 -1.53286224e-02 5.29492319e-01 2.51844943e-01
-3.23626280e-01 -3.83160681e-01 -5.43791413e-01 -9.61974859e-01
-1.90075934e-01 -6.99589014e-01 4.89330173e-01 -4.22872603e-01
-1.28944063e+00 3.96102160e-01 6.15277290e-01 -5.15071213e-01
-3.25998902e-01 -1.58378974e-01 -5.66015244e-01 6.94255769e-01
-1.35362375e+00 -9.90256965e-01 5.39901666e-02 -3.17170501e-01
6.47208929e-01 3.48407209e-01 6.28877699e-01 6.95492998e-02
-3.39090735e-01 -2.71718264e-01 1.25043839e-01 -1.05079480e-01
6.18206024e-01 -1.40683365e+00 8.98608923e-01 5.48271656e-01
1.96045116e-01 5.51772892e-01 1.12132847e+00 -9.53067303e-01
-9.75711584e-01 -8.42912734e-01 1.44782937e+00 -7.08205938e-01
1.30016124e+00 -6.28024697e-01 -8.13836336e-01 7.68588781e-01
7.26280928e-01 -9.17824507e-01 9.40409839e-01 7.43786633e-01
-9.95269641e-02 5.04963398e-01 -9.32419479e-01 4.37049478e-01
7.90879607e-01 -6.07536495e-01 -1.11386740e+00 7.15979576e-01
7.76991427e-01 -1.69413999e-01 -1.08660161e+00 2.90205032e-01
4.63197678e-01 -4.36168820e-01 5.57056069e-01 -4.76899147e-01
5.19655943e-01 -6.67158887e-03 -1.74489632e-01 -1.21405458e+00
-2.37153977e-01 -8.31981301e-01 4.78678674e-01 1.68561161e+00
4.94514942e-01 -5.44536412e-01 8.11328650e-01 1.00749063e+00
2.07241736e-02 -2.50916570e-01 -8.95212471e-01 -4.46554840e-01
7.15276837e-01 -7.36811399e-01 5.12057245e-01 1.41096354e+00
4.11660612e-01 2.80894816e-01 1.61645159e-01 -7.37593323e-02
6.23994648e-01 2.48927549e-01 5.27485549e-01 -1.46635997e+00
2.36497283e-01 -4.89969641e-01 2.59823911e-02 -4.98018503e-01
1.61448285e-01 -8.48256528e-01 -4.40302864e-02 -1.88231730e+00
6.90794706e-01 -5.91605902e-01 -4.73935567e-02 3.84009957e-01
-5.50762191e-03 7.87744522e-02 2.15963736e-01 4.00417686e-01
-5.36520541e-01 5.19623995e-01 5.21295965e-01 -1.26310065e-01
-1.21607326e-01 -2.47140080e-01 -9.50283706e-01 1.10604215e+00
9.39150989e-01 -7.24763155e-01 -8.16883072e-02 -1.09637171e-01
7.47693598e-01 -2.25315437e-01 1.10892139e-01 -6.51685953e-01
1.27862647e-01 -3.85716319e-01 -3.39223593e-02 -6.36330187e-01
-2.09262297e-01 -7.01318741e-01 3.92967641e-01 1.73435047e-01
-4.71143335e-01 -7.28183798e-03 1.93599567e-01 4.28603560e-01
-8.21309164e-02 -2.67344028e-01 3.60416144e-01 -1.57533050e-01
-3.32629532e-01 -8.82685333e-02 -6.54318094e-01 2.52324224e-01
9.82401609e-01 3.26022953e-01 -4.73835260e-01 -2.77670026e-01
-6.61776066e-01 3.45202863e-01 7.45440602e-01 4.07180399e-01
-3.26873451e-01 -1.26420367e+00 -7.18142569e-01 -4.25608963e-01
-4.94776201e-03 8.53021443e-03 4.05103229e-02 5.69983363e-01
-2.28571624e-01 8.88888836e-01 3.01979333e-01 -3.19250673e-01
-1.01554120e+00 3.17647487e-01 -1.68038204e-01 -4.41712379e-01
-5.68154931e-01 2.99674004e-01 3.50906909e-01 -2.75536954e-01
6.89803716e-03 -4.17203933e-01 -4.19610649e-01 6.04042590e-01
2.65506834e-01 4.91938025e-01 -3.63690243e-03 -7.73871839e-01
-1.75251842e-01 2.13433847e-01 4.00752872e-02 -4.69106346e-01
1.49432802e+00 -3.83259833e-01 -2.04378873e-01 8.23753357e-01
7.33876467e-01 2.69560337e-01 -8.32741320e-01 -1.34208158e-01
5.77532053e-01 -1.08149000e-01 3.49868149e-01 -8.32704842e-01
-2.78460413e-01 4.87514228e-01 5.76966032e-02 1.12336075e+00
7.02432632e-01 4.76083308e-01 5.84503233e-01 -1.73243100e-03
2.37982690e-01 -1.20702648e+00 -5.09060264e-01 2.44082615e-01
7.90302634e-01 -1.16775286e+00 2.42807254e-01 -5.94328761e-01
-3.66007984e-01 8.02611232e-01 -1.37479171e-01 3.83671790e-01
7.91595161e-01 2.21100017e-01 6.09774776e-02 -7.78394282e-01
-6.92946017e-01 6.01800019e-03 2.99695879e-01 1.44753367e-01
6.21420860e-01 1.33957505e-01 -8.89284968e-01 6.31391108e-01
-6.18199289e-01 -2.36071736e-01 6.15784228e-01 8.95094991e-01
-7.06820011e-01 -1.18294275e+00 -4.65924710e-01 2.73260653e-01
-8.50369275e-01 -1.24732189e-01 -5.93372703e-01 8.90557885e-01
-1.79770052e-01 1.06830728e+00 5.52148968e-02 1.09139467e-02
4.83410666e-03 4.44765389e-01 3.87178250e-02 -6.44521892e-01
-6.10638857e-01 2.12334678e-01 5.56179345e-01 -1.66177750e-01
-9.02801752e-01 -1.23353207e+00 -1.01644123e+00 -4.61318225e-01
-5.25654614e-01 8.37777376e-01 1.27089036e+00 1.07647848e+00
1.10303134e-01 3.23157012e-01 6.03747427e-01 -5.53692997e-01
-3.11387360e-01 -1.16824734e+00 -4.59282964e-01 1.92719862e-01
-2.69721508e-01 -6.32950127e-01 -4.80729401e-01 1.12998612e-01]
|
[9.150323867797852, 9.666997909545898]
|
79e03c34-8eea-4c5d-9031-f545f573ab79
|
compositional-generalization-in-a-deep
|
1904.09708
| null |
https://arxiv.org/abs/1904.09708v3
|
https://arxiv.org/pdf/1904.09708v3.pdf
|
Compositional generalization in a deep seq2seq model by separating syntax and semantics
|
Standard methods in deep learning for natural language processing fail to capture the compositional structure of human language that allows for systematic generalization outside of the training distribution. However, human learners readily generalize in this way, e.g. by applying known grammatical rules to novel words. Inspired by work in neuroscience suggesting separate brain systems for syntactic and semantic processing, we implement a modification to standard approaches in neural machine translation, imposing an analogous separation. The novel model, which we call Syntactic Attention, substantially outperforms standard methods in deep learning on the SCAN dataset, a compositional generalization task, without any hand-engineered features or additional supervision. Our work suggests that separating syntactic from semantic learning may be a useful heuristic for capturing compositional structure.
|
["Randall C. O'Reilly", 'Jason Jo', 'Yoshua Bengio', 'Jake Russin']
|
2019-04-22
| null | null | null | null |
['systematic-generalization']
|
['reasoning']
|
[ 5.06071091e-01 5.22369683e-01 -1.44405171e-01 -9.07517672e-01
-3.78629327e-01 -7.91349769e-01 9.28272724e-01 2.51804620e-01
-5.80342352e-01 6.60199106e-01 4.84937847e-01 -6.58630967e-01
8.75383466e-02 -7.93518007e-01 -1.13684821e+00 -3.80001634e-01
1.34119928e-01 7.07473099e-01 7.80708045e-02 -4.63185221e-01
3.44826162e-01 2.64876187e-01 -1.04077864e+00 6.83741629e-01
7.97619700e-01 5.23722708e-01 3.68528783e-01 2.87051827e-01
-5.83612502e-01 3.63629907e-01 -2.94674218e-01 -4.68177915e-01
3.06000143e-01 -9.35448647e-01 -1.26056325e+00 -2.54769951e-01
7.48361945e-01 -4.69163097e-02 1.20248441e-02 1.07844329e+00
1.61665261e-01 1.87052473e-01 5.72434962e-01 -6.31461740e-01
-1.55233276e+00 1.03258991e+00 -2.79295444e-02 3.78503531e-01
4.57684577e-01 3.27532403e-02 1.21819425e+00 -9.54743743e-01
7.26062953e-01 1.58470452e+00 8.13604653e-01 1.10143018e+00
-1.59646010e+00 -4.13786620e-01 2.93954432e-01 -1.27772123e-01
-7.91950703e-01 -3.81722093e-01 5.30302644e-01 -3.99045914e-01
1.28597903e+00 -6.94462284e-02 4.64425147e-01 1.40909946e+00
3.54470551e-01 7.46003568e-01 1.35601962e+00 -6.34739399e-01
2.25049898e-01 1.43867079e-02 2.78625339e-01 7.09255278e-01
3.19690555e-01 3.53486329e-01 -4.77013767e-01 -5.09748086e-02
5.33015907e-01 -7.87608996e-02 -3.27473581e-02 -2.79054224e-01
-1.22502327e+00 9.05252814e-01 5.37086070e-01 4.32629079e-01
-3.96854468e-02 3.46341819e-01 6.39967501e-01 6.79243684e-01
2.89970070e-01 9.02289927e-01 -8.47730577e-01 3.59017640e-01
-7.10533917e-01 3.10977936e-01 6.82278275e-01 1.01998210e+00
7.90793061e-01 1.00257002e-01 -1.28373830e-02 6.69705987e-01
1.38864443e-01 2.74665475e-01 1.00332892e+00 -9.98740196e-01
1.58835530e-01 3.90349776e-01 -5.82574666e-01 -3.33801240e-01
-3.98606837e-01 -2.35876694e-01 -3.70220900e-01 -9.03860033e-02
4.66161817e-01 -1.33839920e-01 -9.59737301e-01 2.17340136e+00
-1.21323116e-01 -2.22013146e-01 1.31505340e-01 7.16573656e-01
6.49216890e-01 4.46105540e-01 4.59610522e-01 2.58542430e-02
1.24602604e+00 -7.73961246e-01 -2.04682961e-01 -5.30629873e-01
7.61690080e-01 -4.36143786e-01 1.60874307e+00 2.17181697e-01
-1.29947078e+00 -4.12903696e-01 -8.17397237e-01 -4.57383275e-01
-6.47650301e-01 -5.12902260e-01 9.91251647e-01 5.86696386e-01
-1.34719682e+00 8.06940615e-01 -5.75264752e-01 -8.28499615e-01
6.15433037e-01 4.43285227e-01 -4.15114164e-01 3.78759280e-02
-1.23395264e+00 1.14275825e+00 5.61148405e-01 -3.00368249e-01
-7.32365906e-01 -8.10780942e-01 -1.05416894e+00 1.28925458e-01
1.04611717e-01 -1.27311301e+00 1.60938406e+00 -1.78042793e+00
-1.51238763e+00 1.39082181e+00 -2.88394779e-01 -7.28816152e-01
8.79302621e-03 -1.83051139e-01 7.24810511e-02 -2.83191949e-02
6.47912398e-02 1.03229403e+00 9.66041446e-01 -1.04501069e+00
-1.51910931e-01 -5.33668876e-01 1.93742692e-01 1.72066346e-01
-7.86452964e-02 2.98688322e-01 5.45726180e-01 -9.33160067e-01
1.10105783e-01 -8.82779717e-01 -2.07587600e-01 -1.36796832e-01
3.68269496e-02 -5.68627059e-01 3.13025087e-01 -5.17944276e-01
6.59604371e-01 -1.77254868e+00 3.11385900e-01 2.83332281e-02
7.40386397e-02 1.70818806e-01 -3.69043499e-01 3.43622744e-01
-3.38529944e-01 4.09426212e-01 -6.33705974e-01 -1.65186405e-01
2.63443530e-01 6.02027535e-01 -6.15636885e-01 2.39504695e-01
4.67371672e-01 1.39467931e+00 -1.15072966e+00 -2.40833517e-02
-3.30631942e-01 -1.00493670e-01 -8.51872325e-01 -1.41824141e-01
-5.98087907e-01 3.35544646e-01 -1.53589457e-01 3.39334577e-01
1.93588570e-01 2.92675067e-02 2.84073949e-01 4.20221210e-01
1.19735748e-01 1.07714069e+00 -3.28965724e-01 2.11416030e+00
-5.48896134e-01 5.69863200e-01 -1.10813528e-01 -1.69764817e+00
7.83189774e-01 3.88688534e-01 -1.95460245e-01 -7.10960686e-01
1.54653028e-01 4.13032949e-01 4.72646803e-01 -3.50364566e-01
4.79856469e-02 -8.26211870e-01 -2.91854441e-01 8.69875491e-01
5.12550235e-01 -2.72798032e-01 1.71180591e-02 2.56411023e-02
1.08378279e+00 3.37166488e-01 3.65415335e-01 -8.23687911e-01
3.96355093e-01 -1.96439624e-02 4.40880209e-01 7.64886558e-01
-1.18543290e-01 4.14453954e-01 3.14374685e-01 -6.52628601e-01
-1.18213713e+00 -1.32287228e+00 -8.06830153e-02 1.62547004e+00
-4.21152443e-01 -2.78057545e-01 -9.39384699e-01 -9.07606840e-01
-2.53019594e-02 9.04903889e-01 -6.31117284e-01 -4.80067819e-01
-9.09589112e-01 -6.13337874e-01 6.47522390e-01 8.29674482e-01
1.52129382e-01 -1.58341098e+00 -5.20323336e-01 9.38660651e-02
4.19773787e-01 -8.87311935e-01 -4.25244391e-01 5.25055826e-01
-1.23648822e+00 -5.81809342e-01 -2.67312586e-01 -1.21962774e+00
7.17562973e-01 1.67788323e-02 1.31965375e+00 1.98409483e-01
-8.43361169e-02 4.62024957e-01 -2.55009294e-01 -4.40428346e-01
-7.59511590e-01 1.97955459e-01 1.71260536e-01 -5.59079885e-01
7.35429406e-01 -7.58958995e-01 -2.09426031e-01 -2.90934622e-01
-9.16825175e-01 -1.61032438e-01 5.86190522e-01 1.13983846e+00
1.86561018e-01 -6.53671801e-01 1.09330368e+00 -1.19024897e+00
9.40976560e-01 -6.85799956e-01 -2.13034555e-01 5.41194007e-02
-6.14786148e-01 5.98040640e-01 1.12200737e+00 -4.19548303e-01
-1.04299653e+00 4.74606976e-02 -2.71158159e-01 9.20217484e-02
-6.12822115e-01 4.94791269e-01 -1.11706018e-01 1.31759420e-03
8.16728592e-01 3.99143368e-01 -1.10862087e-02 -4.88361508e-01
7.32523620e-01 1.62913248e-01 5.33003986e-01 -1.19362545e+00
7.05806434e-01 2.67758429e-01 -3.15721176e-04 -7.16449261e-01
-9.84091520e-01 1.15304496e-02 -1.01612794e+00 7.10217834e-01
1.01827180e+00 -5.02266765e-01 -2.87820667e-01 -6.34040982e-02
-1.41941381e+00 -4.78309095e-01 -6.70764387e-01 4.21205878e-01
-9.54874158e-01 3.20508420e-01 -7.03583717e-01 -1.20451346e-01
-3.01585495e-01 -7.04644740e-01 1.13261139e+00 -5.15911691e-02
-7.71068454e-01 -1.46706557e+00 1.25373662e-01 2.82984450e-02
6.03616774e-01 -1.21576779e-01 1.43989432e+00 -1.13533986e+00
-3.15882564e-01 2.53826708e-01 -6.47922158e-02 5.68282902e-01
1.11632146e-01 -4.67972606e-01 -8.15238178e-01 -7.56550133e-02
3.38295460e-01 -5.43977559e-01 1.18675697e+00 2.07514420e-01
1.11639261e+00 -3.81496489e-01 -1.33741349e-01 7.66111016e-01
1.00592864e+00 -5.58671467e-02 2.61777639e-01 2.96664238e-01
4.59813178e-01 8.41470718e-01 -2.68789440e-01 -2.63372600e-01
2.62859344e-01 1.91730440e-01 -1.21259488e-01 8.60437080e-02
-2.42781818e-01 -4.38579023e-01 6.72058165e-01 6.80031359e-01
9.05986950e-02 2.33190209e-01 -9.40406680e-01 5.54062307e-01
-1.49894905e+00 -1.01135957e+00 2.12848470e-01 1.87787247e+00
1.25687170e+00 1.61734611e-01 -1.49686858e-01 -2.75315374e-01
3.82034242e-01 -6.69190064e-02 -4.16192919e-01 -1.15716612e+00
-1.46113634e-01 1.12154579e+00 3.68851006e-01 5.83872557e-01
-7.71622300e-01 1.39641345e+00 7.62086344e+00 3.75191987e-01
-9.76731658e-01 3.85926038e-01 2.58094907e-01 1.33702725e-01
-6.57335162e-01 2.21498594e-01 -6.59559309e-01 1.38818190e-01
1.25477445e+00 -2.43970037e-01 7.40870774e-01 4.59189564e-01
-3.00485659e-02 3.28709781e-01 -1.83163583e+00 4.21741366e-01
3.46108615e-01 -1.37427270e+00 4.83082682e-01 -2.00282559e-01
7.06519246e-01 1.40570059e-01 7.04267099e-02 4.55069005e-01
6.21095419e-01 -1.33500242e+00 6.78438842e-01 2.28794098e-01
4.84675348e-01 -2.47724608e-01 2.99516201e-01 4.77187455e-01
-5.79063714e-01 -1.64482206e-01 -5.47631800e-01 -4.54762995e-01
1.45780705e-02 4.65967245e-02 -6.44734979e-01 4.04879339e-02
2.82513589e-01 5.90094030e-01 -7.60971606e-01 4.33758795e-01
-6.49735987e-01 7.08182752e-01 -1.45823032e-01 -6.32375330e-02
5.01739919e-01 -9.39099118e-02 5.08940279e-01 1.42608988e+00
1.73547179e-01 4.56468351e-02 8.59889612e-02 1.27057564e+00
-3.90325993e-01 2.24517509e-01 -9.23460007e-01 -9.18939337e-02
1.72006980e-01 6.90241933e-01 -5.10243773e-01 -4.23626542e-01
-6.25213325e-01 1.05197406e+00 5.72884440e-01 4.20694053e-01
-3.92147988e-01 1.17549775e-02 4.52845871e-01 1.02411509e-01
2.67061085e-01 -4.55322921e-01 -6.65272593e-01 -1.32212377e+00
2.32923795e-02 -9.42611694e-01 3.08680534e-01 -5.82860649e-01
-1.57545948e+00 3.22312891e-01 1.33081183e-01 -3.94194812e-01
-3.88055831e-01 -1.10509062e+00 -8.10310483e-01 1.01906466e+00
-1.20090032e+00 -1.23043990e+00 4.21873838e-01 5.06324053e-01
9.48351443e-01 -3.55076194e-01 1.00601709e+00 -1.20171532e-01
-5.12612797e-02 7.41526127e-01 -1.60545692e-01 1.10258544e-02
6.78037524e-01 -1.57616174e+00 9.06599641e-01 7.38610148e-01
5.13893187e-01 1.25752664e+00 6.31738484e-01 -4.13453281e-01
-1.32729673e+00 -9.65745211e-01 1.28463411e+00 -7.32359111e-01
8.08962941e-01 -7.38151133e-01 -1.14283645e+00 1.14540040e+00
6.65210128e-01 -1.39890432e-01 8.05845559e-01 2.91719884e-01
-8.07422996e-01 2.51475513e-01 -9.23866510e-01 6.39734089e-01
1.56347334e+00 -8.04199934e-01 -1.62956583e+00 4.64754343e-01
1.10169220e+00 9.24606323e-02 -4.11335111e-01 2.39380479e-01
4.46010560e-01 -5.41650891e-01 7.76851416e-01 -1.42260861e+00
6.51706636e-01 3.66981477e-02 -2.75041908e-01 -1.65924561e+00
-4.64271665e-01 -5.18942058e-01 1.18555717e-01 8.64576459e-01
5.70701599e-01 -8.35379243e-01 5.76079249e-01 5.37635922e-01
-6.82078660e-01 -6.16780043e-01 -7.64805973e-01 -1.02833295e+00
1.11727440e+00 -4.06732976e-01 5.28826833e-01 1.18068528e+00
2.87686497e-01 7.82508194e-01 3.78786653e-01 -3.82064670e-01
3.81015718e-01 1.10692024e-01 4.50022161e-01 -1.22589314e+00
-5.22418916e-01 -1.01611936e+00 -2.30563983e-01 -1.19803739e+00
9.94695663e-01 -1.73124349e+00 -2.67751445e-03 -1.37107515e+00
1.99834093e-01 -3.19874175e-02 -1.54748186e-01 5.14245987e-01
-1.02451574e-02 3.27321514e-02 1.44335572e-02 1.64272252e-03
-2.64228940e-01 5.10966897e-01 1.06425095e+00 -1.57126546e-01
1.72761783e-01 -3.32351208e-01 -1.18518007e+00 9.51438606e-01
8.93418193e-01 -4.79163468e-01 -4.17053878e-01 -8.45916331e-01
4.26939011e-01 -6.38981104e-01 3.77351403e-01 -3.42727721e-01
6.99118199e-03 -2.41816610e-01 2.99538136e-01 1.24256268e-01
-2.80526936e-01 -6.39467239e-01 -5.02713680e-01 5.07724583e-01
-8.14252555e-01 2.94851989e-01 2.04631031e-01 3.43053252e-01
-1.00610912e-01 -5.73521852e-01 7.24340022e-01 -6.65863335e-01
-7.32814729e-01 -4.91620740e-03 -6.08391404e-01 3.63692045e-01
6.39886618e-01 -2.00047761e-01 -4.80050951e-01 1.61504913e-02
-1.06775177e+00 1.83119404e-03 5.84965765e-01 5.23728371e-01
4.76395190e-01 -1.27571023e+00 -7.36878693e-01 2.05657840e-01
5.36037982e-03 -1.43809885e-01 -3.81659865e-01 6.97635472e-01
-4.23215628e-01 4.42190617e-01 -2.47365743e-01 -4.18822736e-01
-6.54121697e-01 8.57217252e-01 4.32672083e-01 1.31920636e-01
-6.51904643e-01 8.91781688e-01 6.42232358e-01 -8.47661734e-01
-1.73459649e-01 -8.14237595e-01 1.86326101e-01 -3.71171802e-01
2.86254734e-01 -2.89075911e-01 5.75090982e-02 -4.90668058e-01
-2.29998112e-01 4.00396198e-01 -1.80561170e-01 -5.47624305e-02
1.29178035e+00 -1.47617459e-01 -5.69793522e-01 7.00088441e-01
1.17068958e+00 -8.80980939e-02 -6.82285428e-01 -4.26114410e-01
6.80508256e-01 -3.68993878e-02 -4.66928393e-01 -7.42401600e-01
-3.74712884e-01 1.21023035e+00 6.40790686e-02 1.24624424e-01
8.45024347e-01 4.13528770e-01 1.00993574e+00 8.40220571e-01
2.31039882e-01 -9.60673273e-01 -7.07289055e-02 1.09647655e+00
9.40132439e-01 -1.18770230e+00 -2.88568079e-01 -1.30801350e-02
-3.08214247e-01 1.35992241e+00 6.11001790e-01 -5.89125633e-01
5.24828434e-01 1.79137483e-01 -1.76854745e-01 -2.83338308e-01
-1.02330339e+00 -3.41935679e-02 2.74488211e-01 6.33613348e-01
9.46611226e-01 8.35829303e-02 -6.48192286e-01 7.95293152e-01
-5.72450578e-01 -3.18588972e-01 4.91736829e-01 8.82080615e-01
-6.54007077e-01 -1.26007235e+00 -1.19406365e-01 3.29639494e-01
-5.03475904e-01 -6.20916545e-01 -8.07962537e-01 5.12047648e-01
2.71626443e-01 3.08529317e-01 1.17840968e-01 8.11711401e-02
2.63655633e-01 8.53252649e-01 9.84367907e-01 -1.43108082e+00
-8.13133299e-01 -4.13422316e-01 -3.75516643e-03 -4.97804731e-01
-4.85855162e-01 -8.66030097e-01 -1.52464569e+00 -3.34335156e-02
2.30723560e-01 -5.25501892e-02 4.53557611e-01 1.28012276e+00
2.16608360e-01 2.32624993e-01 1.04208365e-01 -6.83240235e-01
-8.49995971e-01 -9.43458676e-01 -7.36773834e-02 7.76332378e-01
2.98185378e-01 -3.64025801e-01 -1.97178602e-01 4.78217870e-01]
|
[10.703245162963867, 9.100137710571289]
|
e76aa31c-7557-4674-84dd-cb90550f990a
|
gadbench-revisiting-and-benchmarking
|
2306.12251
| null |
https://arxiv.org/abs/2306.12251v1
|
https://arxiv.org/pdf/2306.12251v1.pdf
|
GADBench: Revisiting and Benchmarking Supervised Graph Anomaly Detection
|
With a long history of traditional Graph Anomaly Detection (GAD) algorithms and recently popular Graph Neural Networks (GNNs), it is still not clear (1) how they perform under a standard comprehensive setting, (2) whether GNNs outperform traditional algorithms such as tree ensembles, and (3) their efficiency on large-scale graphs. In response, we present GADBench -- a comprehensive benchmark for supervised anomalous node detection on static graphs. GADBench provides a thorough comparison across 23 distinct models on ten real-world GAD datasets ranging from thousands to millions of nodes ($\sim$6M). Our main finding is that tree ensembles with simple neighborhood aggregation outperform all other baselines, including the latest GNNs tailored for the GAD task. By making GADBench available as an open-source tool, we offer pivotal insights into the current advancements of GAD and establish a solid foundation for future research. Our code is available at https://github.com/squareRoot3/GADBench.
|
['Jia Li', 'Peilin Zhao', 'Ziqi Gao', 'Fengrui Hua', 'Jianheng Tang']
|
2023-06-21
| null | null | null | null |
['graph-anomaly-detection', 'anomaly-detection', 'benchmarking', 'benchmarking']
|
['graphs', 'methodology', 'miscellaneous', 'robots']
|
[ 1.19019665e-01 2.21823901e-01 -1.92111686e-01 -4.87984857e-03
-3.87343258e-01 -4.72272784e-01 3.47130984e-01 6.55769289e-01
-1.20297574e-01 5.73249102e-01 3.10015511e-02 -8.97687972e-01
-1.22671865e-01 -1.13212240e+00 -5.91109395e-01 -3.60622436e-01
-9.10814881e-01 5.37309945e-01 4.99875486e-01 -2.74890125e-01
5.41743413e-02 5.97914398e-01 -1.03140604e+00 -1.30060971e-01
8.60967577e-01 9.56347823e-01 -6.96083784e-01 9.62022781e-01
1.61086485e-01 8.39802861e-01 -3.36615711e-01 -7.87965953e-01
1.77377865e-01 -1.43429890e-01 -8.89012456e-01 -3.35519403e-01
7.09369540e-01 -3.15830588e-01 -8.57087791e-01 1.13254213e+00
4.27767873e-01 -3.57497483e-02 4.52004075e-01 -1.54898798e+00
-6.87291265e-01 1.02472067e+00 -5.87899804e-01 7.93828547e-01
1.60253122e-01 2.93997198e-01 1.44585145e+00 -4.18112844e-01
5.48120797e-01 1.05213344e+00 1.26190686e+00 3.73575211e-01
-1.34025538e+00 -4.58480269e-01 4.64777738e-01 9.11674500e-02
-1.15783739e+00 -4.54672396e-01 6.59136355e-01 -1.63387135e-01
1.30417669e+00 4.34943736e-01 5.36890090e-01 1.37989175e+00
2.39303991e-01 6.86465323e-01 5.79638958e-01 -1.33852392e-01
4.83657569e-02 -5.69522023e-01 5.80543041e-01 9.90879238e-01
9.61550832e-01 -2.20918525e-02 -2.75438190e-01 -5.69705844e-01
4.95108634e-01 -8.50679725e-03 -2.14731559e-01 -2.88348496e-01
-9.31704938e-01 9.31574523e-01 5.83421886e-01 4.64034706e-01
-3.14395428e-01 4.99168873e-01 7.91050911e-01 5.20398438e-01
6.63260937e-01 5.94816089e-01 -4.78821248e-01 -1.45953506e-01
-4.90093172e-01 6.15929402e-02 1.15216112e+00 6.20854318e-01
6.22993946e-01 4.25668180e-01 7.45451376e-02 4.25301611e-01
1.96743384e-02 -8.72997846e-03 5.71745522e-02 -5.78485787e-01
3.35509449e-01 8.64882290e-01 -3.77168834e-01 -1.41761470e+00
-6.39797032e-01 -7.94064999e-01 -1.20218122e+00 -1.68564767e-01
6.89014435e-01 -8.97562727e-02 -7.87338018e-01 1.65838516e+00
2.13628411e-01 4.28926855e-01 -3.75029892e-01 2.41137654e-01
1.06517577e+00 1.62423447e-01 3.72979580e-03 1.07186332e-01
8.16070437e-01 -9.06058848e-01 -2.48485997e-01 -4.39195186e-01
1.27012551e+00 -2.25889623e-01 9.61084306e-01 3.55091721e-01
-7.97228098e-01 3.46165858e-02 -1.01881301e+00 -3.51860523e-02
-6.49208009e-01 -6.13179743e-01 1.03732800e+00 8.14217567e-01
-1.56486523e+00 9.28876162e-01 -1.13411927e+00 -7.64392376e-01
4.75284040e-01 2.57603019e-01 -3.67654502e-01 8.35702792e-02
-1.00243878e+00 5.30011356e-01 4.22635883e-01 -8.35175216e-02
-9.32923853e-01 -6.55999780e-01 -9.61534917e-01 4.66466993e-02
7.40982413e-01 -6.00668252e-01 1.06187022e+00 -7.06302464e-01
-5.78628838e-01 8.80257428e-01 7.42170289e-02 -9.19191360e-01
2.27460161e-01 -2.57230867e-02 -6.66064203e-01 -1.01780161e-01
7.19595775e-02 1.78500917e-02 3.72253865e-01 -1.04747081e+00
-4.05113488e-01 -4.70374584e-01 -1.90093100e-01 -3.28613579e-01
-4.86219674e-01 -2.34137792e-02 -2.73989290e-01 -7.02307105e-01
1.06213182e-01 -6.45297587e-01 -3.77098352e-01 -3.47409099e-01
-1.02462840e+00 -3.68726999e-01 7.14433312e-01 -6.19916797e-01
1.68914688e+00 -1.88631105e+00 -2.24261925e-01 5.97313344e-01
9.39012170e-01 3.30471456e-01 -2.97102869e-01 7.77966857e-01
-3.79783630e-01 3.35711747e-01 -4.08489317e-01 -2.84065545e-01
8.44039302e-03 1.77879080e-01 -1.36880368e-01 4.95922297e-01
8.52272585e-02 1.23674810e+00 -9.60676551e-01 -1.96110964e-01
-9.06241089e-02 9.53657851e-02 -3.17951918e-01 -3.13796878e-01
-3.10958982e-01 3.13872583e-02 -5.01833797e-01 1.15201306e+00
5.22447586e-01 -8.56544614e-01 3.19524437e-01 2.62946516e-01
3.71231794e-01 3.39373708e-01 -8.04101408e-01 1.26873803e+00
3.28743547e-01 5.22334874e-01 7.36333430e-02 -1.07243454e+00
8.86012852e-01 -1.43581718e-01 4.40502584e-01 -5.99449992e-01
2.90162209e-02 2.82411635e-01 3.35371226e-01 -2.15846207e-03
4.33609903e-01 6.50759935e-01 -6.39031231e-02 7.62656510e-01
-1.20575670e-02 4.80020523e-01 3.80751759e-01 8.05329084e-01
2.34439421e+00 -3.57730746e-01 3.95409346e-01 -2.52010167e-01
2.18589410e-01 -4.68310304e-02 3.97016138e-01 1.12680852e+00
-4.90966052e-01 2.64648288e-01 9.85688865e-01 -1.11900198e+00
-9.24759328e-01 -1.04770911e+00 2.32889637e-01 1.17012334e+00
-2.05474675e-01 -9.45240200e-01 -5.28558254e-01 -1.21971393e+00
3.88285786e-01 4.68672633e-01 -6.74496531e-01 -1.11826174e-01
-6.57790720e-01 -1.20844924e+00 8.63992631e-01 5.81529438e-01
2.77356565e-01 -1.15393746e+00 2.10149288e-01 1.69315591e-01
-4.79524583e-02 -1.10472775e+00 -3.06505919e-01 1.53969482e-01
-1.05242252e+00 -1.47457778e+00 1.38340309e-01 -5.03045857e-01
6.58513904e-01 3.57626408e-01 1.88540959e+00 8.75995338e-01
-3.88232380e-01 4.31544900e-01 -3.16796988e-01 -2.06636563e-01
-4.96477008e-01 5.42185366e-01 1.83348030e-01 -3.24673623e-01
5.86103857e-01 -1.32637250e+00 -5.69368124e-01 8.39394554e-02
-5.74966073e-01 -5.69641113e-01 4.34003294e-01 4.45738703e-01
2.37150148e-01 2.33131871e-02 6.85554922e-01 -1.27378023e+00
8.26985359e-01 -8.12642694e-01 -6.19685233e-01 -1.83946639e-02
-1.08873534e+00 -1.31159782e-01 6.36165857e-01 8.65582973e-02
-1.98526993e-01 -5.81110835e-01 -4.55707520e-01 -3.65651511e-02
-1.21229760e-01 5.32832086e-01 2.69318998e-01 -2.34603941e-01
9.57722008e-01 1.87812909e-01 -7.62422159e-02 -4.20933574e-01
8.63725618e-02 2.71434281e-02 7.18183994e-01 -6.05823934e-01
1.01658750e+00 3.36645126e-01 1.71929404e-01 -5.88234961e-01
-5.69490254e-01 -3.21296871e-01 -5.28044701e-01 1.30651087e-01
4.45604473e-01 -6.23590708e-01 -5.87080002e-01 6.25828683e-01
-7.66683280e-01 -7.16837049e-01 7.15972707e-02 -3.26582611e-01
-3.16465013e-02 9.58371341e-01 -9.71808374e-01 -7.07281709e-01
-8.09573412e-01 -5.22536039e-01 6.72769368e-01 -9.78832170e-02
-3.92579436e-01 -1.35902810e+00 7.32343346e-02 9.03185308e-02
4.33744937e-01 7.14086592e-01 1.11971557e+00 -1.27438962e+00
-7.03407407e-01 -2.84447759e-01 -4.39610153e-01 8.76895413e-02
4.47630361e-02 3.48238766e-01 -6.11895561e-01 -5.90895295e-01
-7.53816247e-01 -2.05069229e-01 1.21161950e+00 2.84565568e-01
1.39253962e+00 -3.60488147e-01 -6.61020696e-01 8.70368183e-01
1.34505260e+00 -1.70947269e-01 6.07797325e-01 3.86719942e-01
1.00453699e+00 5.27835116e-02 6.61835298e-02 1.18667953e-01
6.19116902e-01 1.35928959e-01 9.48673010e-01 -5.50444424e-02
-1.22361034e-02 -3.68347108e-01 2.85969913e-01 7.16285348e-01
-1.60747975e-01 -6.61005378e-01 -1.33944643e+00 6.43349409e-01
-1.88981676e+00 -8.40608001e-01 -4.96068060e-01 2.09542203e+00
1.71333969e-01 4.96192813e-01 6.59249425e-01 1.05232403e-01
7.31593907e-01 4.78051186e-01 -7.00859427e-01 -4.22017992e-01
-1.12587266e-01 2.44624510e-01 5.05819201e-01 2.65366197e-01
-1.42462182e+00 8.24944258e-01 6.51289320e+00 6.94993615e-01
-7.82082021e-01 6.29372075e-02 9.12410259e-01 1.28324181e-01
-2.39478067e-01 -1.09699182e-02 -4.76231784e-01 3.94516736e-01
1.18536425e+00 -2.31574476e-01 4.11904037e-01 9.34123993e-01
-3.14104408e-01 2.22798083e-02 -1.17408216e+00 5.70767343e-01
-6.40545487e-02 -1.32287574e+00 -2.75969803e-01 3.77258778e-01
6.89532101e-01 8.75241339e-01 -1.55758664e-01 4.94718820e-01
9.61452603e-01 -1.32589960e+00 -2.08997294e-01 1.88745737e-01
4.96510506e-01 -5.19520044e-01 7.03444302e-01 1.99889660e-01
-1.25210202e+00 -6.86884373e-02 -2.38423154e-01 -3.00110932e-02
-2.21991956e-01 7.14989126e-01 -8.10545444e-01 8.07428122e-01
9.57424402e-01 9.25370991e-01 -1.10505319e+00 9.04785156e-01
2.75632516e-02 1.03211784e+00 -5.83938360e-01 1.19358152e-01
2.32242972e-01 -2.73394845e-02 9.57092702e-01 1.06878257e+00
2.55235821e-01 -3.68965596e-01 1.62614852e-01 6.99881375e-01
-3.63804430e-01 2.42647203e-03 -9.47803617e-01 -4.04649675e-01
6.07533276e-01 1.50129271e+00 -8.60857487e-01 -4.07135785e-02
-5.06105483e-01 6.47049487e-01 7.74851561e-01 2.59541482e-01
-5.45341849e-01 -4.63781446e-01 6.83092535e-01 2.32952878e-01
1.81803405e-01 -1.63320556e-01 -1.09650880e-01 -1.17167187e+00
1.30823150e-01 -1.10677791e+00 1.15675640e+00 -4.55351263e-01
-1.61256278e+00 7.98131108e-01 -4.85274374e-01 -7.07960486e-01
-8.77419785e-02 -8.32515419e-01 -1.13368011e+00 3.17351252e-01
-1.09390402e+00 -1.04329479e+00 -5.35004616e-01 4.88926649e-01
-6.91095293e-02 -2.25146189e-01 9.81214881e-01 2.01197967e-01
-1.07211125e+00 8.06931496e-01 1.02928214e-01 5.60123801e-01
4.71851468e-01 -1.50957417e+00 1.38893330e+00 1.26754320e+00
3.70906293e-01 3.19343358e-01 7.06138432e-01 -8.98764312e-01
-1.29433131e+00 -1.24107754e+00 4.92854059e-01 -9.12383139e-01
1.15526414e+00 -4.01521772e-01 -1.26898503e+00 1.35898781e+00
-8.84481072e-02 5.01998842e-01 3.67077231e-01 7.11376667e-01
-5.40591776e-01 -1.87746853e-01 -1.06059730e+00 5.66623390e-01
1.65016913e+00 -1.29053622e-01 1.15633637e-01 4.70516771e-01
7.20225990e-01 -2.72831500e-01 -1.04485333e+00 6.61216795e-01
2.78369337e-01 -1.42581701e+00 8.75499129e-01 -8.71071994e-01
1.41772717e-01 9.31925178e-02 9.65820476e-02 -1.29722977e+00
-4.84933645e-01 -9.08444881e-01 -8.04745436e-01 9.57175970e-01
6.30352199e-01 -1.34516597e+00 1.07769012e+00 2.02699170e-01
-1.99856102e-01 -1.03637826e+00 -8.41955721e-01 -7.32368946e-01
6.03016503e-02 -3.54751796e-01 6.59300745e-01 1.12333250e+00
-2.23310038e-01 1.51000187e-01 -1.68493137e-01 3.49680841e-01
7.05101848e-01 -1.15955457e-01 9.35478568e-01 -1.72531354e+00
-2.36813053e-01 -5.98507404e-01 -7.75135040e-01 -5.94911993e-01
1.15768962e-01 -1.20271671e+00 -6.95152938e-01 -1.51934314e+00
2.22264379e-01 -4.46055084e-01 -5.86490512e-01 8.05232704e-01
-2.90165871e-01 4.24602836e-01 -3.30252886e-01 1.80154797e-02
-9.37647820e-01 2.79847264e-01 7.82835960e-01 3.67479138e-02
4.88639751e-04 1.63089812e-01 -1.08917797e+00 7.96202362e-01
1.06921780e+00 -4.31438237e-01 -1.93598270e-01 -2.23129451e-01
5.55676162e-01 -3.18784505e-01 5.92560232e-01 -1.17007875e+00
1.61281377e-01 2.62227714e-01 3.05348456e-01 -5.82314849e-01
-7.55039081e-02 -2.99828142e-01 -1.02470011e-01 5.85811496e-01
1.53390050e-01 7.41372406e-01 1.70408964e-01 8.24550152e-01
1.32815883e-01 9.92780626e-02 3.78814846e-01 -7.68070146e-02
-8.20599377e-01 8.20526958e-01 -4.98102531e-02 4.07226443e-01
8.97619188e-01 -2.08468780e-01 -1.03302073e+00 -6.72586858e-01
-7.63882518e-01 4.46124196e-01 6.24729514e-01 3.26312780e-01
2.88370907e-01 -1.08283496e+00 -7.95105696e-01 1.64314702e-01
2.52122968e-01 -4.85667475e-02 1.64064467e-01 1.09515011e+00
-5.34266472e-01 9.09563377e-02 9.83753279e-02 -4.02909577e-01
-1.35412180e+00 5.30829787e-01 6.38264954e-01 -6.57834232e-01
-9.02412593e-01 9.69033957e-01 -1.63645357e-01 -5.86913407e-01
2.22474545e-01 1.52504332e-02 3.33128035e-01 -4.89076257e-01
1.88575894e-01 5.44693172e-01 1.86970785e-01 -3.55867326e-01
-4.41384554e-01 -6.58111051e-02 -3.79129618e-01 7.52465427e-01
1.41986763e+00 9.18173194e-02 -5.13703406e-01 1.71212897e-01
7.02176332e-01 1.01622708e-01 -7.52951503e-01 -3.54059100e-01
4.22547042e-01 -1.92935884e-01 -1.97149977e-01 -6.62754416e-01
-1.17640591e+00 4.49061334e-01 9.06655490e-02 8.68349314e-01
1.16054296e+00 1.33097038e-01 8.66610467e-01 5.40361106e-01
3.33059698e-01 -6.35002375e-01 2.24273756e-01 6.14470422e-01
6.36895001e-01 -1.25620949e+00 1.65561527e-01 -5.45505345e-01
-2.18603373e-01 9.44082558e-01 1.02196729e+00 -2.52023607e-01
5.44106781e-01 2.68370211e-01 -1.46809533e-01 -5.56682706e-01
-1.03503668e+00 -1.52555078e-01 7.54745537e-03 7.10716128e-01
3.78716290e-01 2.88294926e-02 2.33678535e-01 4.04235691e-01
-1.99830428e-01 -5.27956784e-01 5.64323008e-01 6.67732477e-01
-3.80723238e-01 -1.15126848e+00 1.04002275e-01 1.13719618e+00
-6.59285665e-01 -2.21306577e-01 -7.98327625e-01 1.03628302e+00
-3.27546120e-01 8.47376585e-01 8.57783929e-02 -5.17384648e-01
1.64499342e-01 1.21710040e-01 2.36573681e-01 -4.84378338e-01
-5.48349321e-01 -5.72542787e-01 6.57348216e-01 -9.15354669e-01
3.05846512e-01 -5.03157198e-01 -7.52350867e-01 -1.09293985e+00
-5.89222685e-02 3.02876122e-02 5.39878383e-02 5.21110356e-01
6.81243181e-01 5.14725745e-01 2.10341811e-01 -3.97740483e-01
-4.52118576e-01 -1.14131069e+00 -5.89068055e-01 2.07752436e-01
2.74187624e-01 -3.15651149e-01 -6.72049940e-01 -8.57638896e-01]
|
[6.721827983856201, 5.897677898406982]
|
01d8ab35-f60a-4eb7-8e86-dbb4ca0b7496
|
self-supervised-training-for-blind-multi
|
2004.06957
| null |
https://arxiv.org/abs/2004.06957v4
|
https://arxiv.org/pdf/2004.06957v4.pdf
|
Self-Supervised training for blind multi-frame video denoising
|
We propose a self-supervised approach for training multi-frame video denoising networks. These networks predict frame t from a window of frames around t. Our self-supervised approach benefits from the video temporal consistency by penalizing a loss between the predicted frame t and a neighboring target frame, which are aligned using an optical flow. We use the proposed strategy for online internal learning, where a pre-trained network is fine-tuned to denoise a new unknown noise type from a single video. After a few frames, the proposed fine-tuning reaches and sometimes surpasses the performance of a state-of-the-art network trained with supervision. In addition, for a wide range of noise types, it can be applied blindly without knowing the noise distribution. We demonstrate this by showing results on blind denoising of different synthetic and realistic noises.
|
['Jérémy Anger', 'Valéry Dewil', 'Gabriele Facciolo', 'Thibaud Ehret', 'Pablo Arias', 'Axel Davy']
|
2020-04-15
| null | null | null | null |
['video-denoising', 'video-temporal-consistency']
|
['computer-vision', 'computer-vision']
|
[ 2.41081104e-01 -2.60641545e-01 1.02183565e-01 -3.27123195e-01
-7.08284974e-01 -3.75508219e-01 4.78237748e-01 -2.41385028e-01
-6.84428751e-01 8.20095718e-01 1.92643106e-01 1.80798829e-01
8.65608081e-02 -4.06564415e-01 -1.04348779e+00 -8.33258033e-01
-6.77092448e-02 -6.82936534e-02 3.48395586e-01 3.16494703e-02
1.23440996e-02 2.43862689e-01 -1.57815874e+00 3.36967915e-01
5.48634708e-01 1.18120515e+00 3.59999448e-01 8.85559618e-01
2.27004781e-01 1.03668284e+00 -5.94635308e-01 -4.26214427e-01
5.67236960e-01 -6.49999022e-01 -5.30175984e-01 3.69294435e-01
9.38090861e-01 -6.37571692e-01 -6.34963989e-01 1.23891175e+00
3.16133738e-01 3.63105178e-01 1.74777821e-01 -7.41452575e-01
-2.56318271e-01 3.19588244e-01 -3.14724147e-01 4.62467998e-01
3.42650294e-01 3.43533546e-01 6.22276843e-01 -5.36205411e-01
7.72072911e-01 1.23739290e+00 9.19582784e-01 8.39955389e-01
-1.23800313e+00 -3.87573332e-01 2.34591141e-01 3.65533590e-01
-9.43057358e-01 -8.36190522e-01 6.76820219e-01 -4.06265110e-01
6.71373725e-01 -2.18951821e-01 6.33644819e-01 1.24470460e+00
1.25639975e-01 3.91709864e-01 1.01024258e+00 -2.06006080e-01
4.45677131e-01 -3.93932402e-01 -3.44767630e-01 5.28244138e-01
-2.65378840e-02 4.18758899e-01 -6.89754367e-01 8.88543725e-02
8.68113160e-01 -8.61179233e-02 -6.28309906e-01 -3.41812849e-01
-1.21613085e+00 4.87206310e-01 3.56286138e-01 3.27882528e-01
-5.42408645e-01 3.74984711e-01 5.06791890e-01 5.16183674e-01
6.98116541e-01 2.96054315e-03 -4.23990041e-01 -2.56247044e-01
-1.39369166e+00 -2.70627029e-02 6.69421375e-01 3.94383281e-01
8.50819945e-01 4.82547611e-01 -4.33893986e-02 6.00291252e-01
2.18966275e-01 3.68054390e-01 5.70995808e-01 -1.51393771e+00
2.90460825e-01 -1.99002609e-01 3.74912620e-01 -8.10477138e-01
-3.14576328e-02 -3.74177247e-01 -1.18065774e+00 6.70507729e-01
7.57679582e-01 -4.61336613e-01 -1.06647849e+00 1.76939046e+00
2.11313680e-01 8.22934687e-01 6.34465590e-02 1.17038298e+00
4.22557294e-01 6.28047645e-01 -8.71853009e-02 -7.46621609e-01
6.50879204e-01 -1.03516746e+00 -9.00109410e-01 -3.57785106e-01
5.71914017e-03 -7.89580762e-01 4.72230405e-01 7.80776441e-01
-1.33597100e+00 -8.66701722e-01 -8.22128296e-01 2.96334833e-01
1.34793893e-01 -7.47635663e-02 -2.54421011e-02 5.42973340e-01
-1.38014984e+00 1.20522547e+00 -1.12486398e+00 -3.56381238e-01
4.82858509e-01 2.76630461e-01 -5.36704361e-01 -1.27225205e-01
-9.87280965e-01 6.84858680e-01 1.98437363e-01 3.79341125e-01
-1.51869833e+00 -6.25384748e-01 -7.77462959e-01 -1.53758228e-01
3.49918664e-01 -9.75743651e-01 1.11963367e+00 -1.59730113e+00
-1.76951885e+00 6.12714350e-01 -5.99293411e-01 -9.61567402e-01
8.05784702e-01 -6.05946422e-01 -4.41183776e-01 5.65675259e-01
6.32416755e-02 3.83353829e-01 1.90453744e+00 -1.46934307e+00
-5.75582266e-01 -8.98424685e-02 1.49778202e-01 1.16822183e-01
-3.05587441e-01 -9.48473513e-02 -3.87084156e-01 -9.96779025e-01
6.72412440e-02 -5.81023932e-01 -2.30756938e-01 2.51328409e-01
-1.20782338e-01 4.26257759e-01 9.65573311e-01 -8.65056634e-01
9.48683321e-01 -2.22886348e+00 4.32845265e-01 -1.87275782e-02
2.28394896e-01 4.74553734e-01 -2.42167979e-01 1.32969972e-02
-3.14465374e-01 -1.41087428e-01 -3.78914207e-01 -7.19593287e-01
-5.33126175e-01 2.90298790e-01 -2.32218057e-01 6.76698983e-01
-9.87363141e-03 3.49950135e-01 -1.06786239e+00 -1.24755800e-01
4.09457624e-01 6.87407613e-01 -5.18113911e-01 4.95184332e-01
-2.38196716e-01 9.16540444e-01 1.31388500e-01 1.20021351e-01
6.98731542e-01 -3.75883989e-02 1.41313121e-01 -5.14453650e-01
7.24305063e-02 -4.09373529e-02 -1.44376576e+00 1.83391511e+00
-4.45102692e-01 6.65969491e-01 5.01301408e-01 -1.14709091e+00
6.47158504e-01 5.22680879e-01 4.60252553e-01 -2.68234193e-01
3.86811979e-02 1.24250278e-01 -1.70974597e-01 -6.58131778e-01
1.54622579e-02 -2.27831513e-01 6.84285223e-01 3.26178819e-01
3.81040335e-01 2.19306186e-01 3.83794487e-01 1.50242662e-02
1.32842600e+00 4.29957926e-01 4.94952314e-02 8.80865529e-02
8.23084056e-01 -4.21738744e-01 8.65810871e-01 9.75604951e-01
-4.31415230e-01 9.17438626e-01 2.35649198e-01 -6.58284962e-01
-1.10640657e+00 -9.61175263e-01 1.79209232e-01 8.41306150e-01
2.19098642e-01 -2.52337366e-01 -9.58377182e-01 -6.54166162e-01
-1.48784474e-01 1.50688753e-01 -5.34016788e-01 1.18413158e-02
-7.47969389e-01 -4.47049260e-01 1.45192489e-01 2.83908844e-01
7.07083583e-01 -8.89229834e-01 -4.52720463e-01 2.74455696e-01
-3.48667473e-01 -1.38093317e+00 -5.32845020e-01 8.81637335e-02
-1.04823208e+00 -1.08800638e+00 -8.11093569e-01 -8.75361919e-01
6.57899916e-01 2.59921640e-01 1.14542627e+00 1.19689941e-01
2.09606975e-01 4.48291451e-01 -1.29228905e-01 4.13339496e-01
-6.69477880e-01 -4.15743560e-01 3.56540918e-01 5.59076011e-01
-2.01073095e-01 -7.92055368e-01 -7.16029644e-01 3.03015500e-01
-9.81011629e-01 -2.73199201e-01 1.11208543e-01 1.05186176e+00
5.59880257e-01 2.69400388e-01 3.42408121e-01 -4.79769319e-01
2.66486585e-01 -1.40040919e-01 -7.22559869e-01 9.80924293e-02
-1.23315282e-01 8.22044685e-02 9.30683613e-01 -6.16070628e-01
-1.23321223e+00 3.12568009e-01 2.18994301e-02 -9.40496266e-01
-3.15772444e-01 7.54588097e-02 3.21505331e-02 -3.74388725e-01
6.66426241e-01 7.76344836e-02 2.38179415e-01 -4.27427053e-01
4.00753438e-01 2.84016818e-01 9.98084188e-01 -3.21155012e-01
1.06474209e+00 7.95113087e-01 -4.97442931e-02 -7.18611121e-01
-9.10039186e-01 -3.97210687e-01 -6.77501619e-01 -4.56221372e-01
8.02739024e-01 -1.22129190e+00 -6.92607760e-01 9.21661198e-01
-1.35911047e+00 -4.56409901e-01 -3.99590433e-01 5.76613367e-01
-6.25929773e-01 7.66234338e-01 -8.54797244e-01 -6.66035116e-01
-1.27686840e-02 -1.05358803e+00 8.04709017e-01 1.27321213e-01
1.12512589e-01 -1.19250107e+00 5.49594164e-02 2.06729442e-01
4.04296428e-01 4.24202941e-02 1.91225752e-01 -2.32380629e-01
-6.38389528e-01 1.08306393e-01 4.67321686e-02 9.57774460e-01
2.96952814e-01 -8.34767371e-02 -1.13945174e+00 -5.28169632e-01
5.56391001e-01 -2.14065239e-01 1.22133720e+00 7.57497549e-01
1.04586053e+00 -3.26628953e-01 1.79966763e-01 9.55435336e-01
1.54361713e+00 5.17077334e-02 5.79237044e-01 5.31854391e-01
6.52307868e-01 2.51921952e-01 2.42068186e-01 3.94763231e-01
2.22576261e-02 4.02397692e-01 7.99908817e-01 6.71060085e-02
-2.96525508e-01 -1.98904574e-02 5.30465424e-01 3.77648979e-01
-3.31511855e-01 -1.58026606e-01 -5.18397689e-01 6.83198869e-01
-1.93699193e+00 -1.15890598e+00 2.29355335e-01 2.41710210e+00
8.29256952e-01 2.43320122e-01 6.99325204e-02 1.99657619e-01
9.74426806e-01 4.68749821e-01 -5.66916227e-01 1.00855209e-01
-4.49654698e-01 8.20174813e-02 4.76541191e-01 9.16374862e-01
-1.41484249e+00 9.23333406e-01 6.47159624e+00 5.08845448e-01
-1.23899281e+00 2.83109188e-01 6.68817163e-01 -5.34907021e-02
1.26540273e-01 -2.15131752e-02 -3.32650900e-01 4.69827265e-01
8.55255842e-01 -3.64138111e-02 9.06068683e-01 4.03315187e-01
6.48516953e-01 -2.79937238e-01 -1.10816824e+00 1.15298104e+00
1.89711183e-01 -1.46803415e+00 -7.72482157e-02 -4.37750548e-01
9.70188797e-01 -2.38103811e-02 -1.32682309e-01 -2.89774090e-01
1.38941497e-01 -7.07531989e-01 6.33084595e-01 7.61027277e-01
4.59357679e-01 -4.54207748e-01 6.72554672e-01 1.50360584e-01
-9.22002792e-01 -2.75343984e-01 -2.93536276e-01 -1.21840306e-01
4.74506527e-01 8.45133066e-01 -1.22926287e-01 4.09257472e-01
1.03835595e+00 1.16223490e+00 -2.96425611e-01 1.30754733e+00
-4.85772580e-01 6.50914431e-01 -1.93246618e-01 7.60386944e-01
1.12288170e-01 -3.12878728e-01 8.62778962e-01 9.70719993e-01
3.66296113e-01 -1.47233576e-01 -4.14356701e-02 3.67893845e-01
-1.21940546e-01 -4.39078867e-01 -4.10197943e-01 4.29131776e-01
1.36266991e-01 1.07161093e+00 -3.60699624e-01 -5.40401280e-01
-4.33013916e-01 1.36120641e+00 7.08163008e-02 8.23151648e-01
-6.48118556e-01 3.95140871e-02 7.78740466e-01 6.03910908e-03
7.35950947e-01 -1.59573093e-01 -9.28890333e-02 -1.46790159e+00
2.19495237e-01 -8.90978277e-01 1.81127653e-01 -1.04886031e+00
-1.39200211e+00 8.23794127e-01 -4.23491389e-01 -1.47565806e+00
-4.93911922e-01 -4.87935513e-01 -6.60896420e-01 6.55399799e-01
-1.58199966e+00 -6.57206059e-01 -4.12528545e-01 8.88799250e-01
4.07019496e-01 -3.53862494e-01 5.30664563e-01 2.53605217e-01
-4.70119238e-01 2.36660093e-01 4.15547401e-01 1.99491471e-01
9.92807388e-01 -1.17136574e+00 2.87032515e-01 1.39293230e+00
2.89117694e-02 2.68915415e-01 1.03413701e+00 -6.38711214e-01
-1.11749756e+00 -1.26654470e+00 6.39041007e-01 3.97471413e-02
8.40307713e-01 1.04363710e-02 -1.04158151e+00 6.20507777e-01
4.17598486e-01 6.19202197e-01 -5.21268621e-02 -3.76611054e-01
-3.91053766e-01 -6.19611025e-01 -1.23409212e+00 3.31206590e-01
1.09110975e+00 -6.74859822e-01 -4.66856033e-01 3.63502443e-01
4.09742236e-01 -5.85720360e-01 -7.60383725e-01 1.09440096e-01
2.87844747e-01 -1.41861558e+00 9.93629336e-01 -4.34205890e-01
4.91838336e-01 -4.47991192e-01 3.21078859e-02 -1.60817659e+00
-9.85340551e-02 -1.25004745e+00 -4.13799047e-01 9.52074587e-01
-1.09904349e-01 -3.44575316e-01 8.38136733e-01 1.59980521e-01
1.01427183e-01 -1.23825297e-01 -1.16718578e+00 -7.98708916e-01
-2.27142110e-01 -4.28992778e-01 8.02885741e-03 6.67921901e-01
-4.78846639e-01 -4.05021757e-03 -9.08902347e-01 4.19977784e-01
1.17491746e+00 -3.02889198e-01 5.65428615e-01 -8.91548991e-01
-3.99195492e-01 4.19986658e-02 -6.65315866e-01 -1.18236482e+00
3.28878313e-01 -2.21810058e-01 2.30582803e-01 -1.23893583e+00
-2.12616622e-01 2.91810930e-01 -2.24980831e-01 2.55813092e-01
-1.05152130e-01 5.09502470e-01 2.18817055e-01 2.44640931e-01
-6.40279770e-01 5.04269063e-01 1.13518572e+00 -1.52031645e-01
-5.05256355e-02 5.69566153e-02 -2.68644094e-01 9.76183712e-01
5.91083109e-01 -4.97778445e-01 -1.61982968e-01 -7.62954175e-01
-1.33805335e-01 2.91475892e-01 6.78817630e-01 -1.25083482e+00
4.26263124e-01 1.85384870e-01 4.48961020e-01 -7.07152933e-02
3.80382210e-01 -9.60494101e-01 1.66736811e-01 4.44395840e-01
-2.60306686e-01 -3.18337642e-02 8.21766481e-02 8.66657197e-01
-5.55184186e-01 -2.06596106e-01 1.00131500e+00 -2.20348224e-01
-8.56135488e-01 3.09405774e-01 -3.93892854e-01 1.05748422e-01
6.56262815e-01 -2.32592061e-01 -2.31122360e-01 -7.99011528e-01
-1.25356984e+00 -1.40740618e-01 4.73641247e-01 2.65122235e-01
6.26265228e-01 -1.17645466e+00 -5.91711164e-01 3.46960366e-01
-4.35900241e-01 -1.31340623e-01 3.72433424e-01 6.43531263e-01
-4.42876160e-01 -2.82747298e-01 -2.89316803e-01 -8.77606094e-01
-1.12225509e+00 4.48815912e-01 8.48744035e-01 -1.06414070e-03
-7.05286682e-01 6.81195498e-01 -2.44197808e-02 1.08355559e-01
4.98558760e-01 -1.17610864e-01 -4.89754751e-02 -2.01903403e-01
7.12100685e-01 3.74342918e-01 2.05609780e-02 -8.08399320e-01
-1.82999000e-01 7.22742856e-01 2.71492541e-01 -1.02089196e-01
1.48485100e+00 -4.53944087e-01 -2.51515657e-01 3.97564173e-01
1.14238942e+00 -1.72516659e-01 -2.09286380e+00 -3.53006035e-01
-2.53369540e-01 -6.20544732e-01 2.64655977e-01 -5.92035770e-01
-1.56820536e+00 5.76594949e-01 9.83412445e-01 1.11559533e-01
1.47726154e+00 -3.66841197e-01 7.28068292e-01 3.50971729e-01
1.07752532e-01 -9.41624880e-01 1.93572372e-01 5.06620765e-01
6.15680635e-01 -1.33987737e+00 -1.58195361e-01 -2.28740916e-01
-3.55289847e-01 1.31119406e+00 3.72472167e-01 -3.78722727e-01
7.13761806e-01 3.51446569e-01 3.98516476e-01 3.69503260e-01
-6.13462448e-01 -1.85274437e-01 1.11456051e-01 8.08850467e-01
1.62700117e-01 -5.84839702e-01 1.53876647e-01 4.52286266e-02
1.71143383e-01 3.32927763e-01 6.88363314e-01 6.04031742e-01
-3.43730956e-01 -9.82632816e-01 -5.79816639e-01 1.73188761e-01
-5.81393898e-01 -1.42708570e-01 2.35381097e-01 3.14096153e-01
2.32073978e-01 1.17034554e+00 4.05837074e-02 -1.44877449e-01
1.69407412e-01 -1.42539978e-01 5.28190911e-01 -1.81579590e-01
-5.60280561e-01 3.98695499e-01 -1.00660682e-01 -9.59167778e-01
-1.21475565e+00 -7.53683448e-01 -7.30493367e-01 -1.97582766e-01
1.40367329e-01 -1.31643182e-02 2.89321363e-01 1.17047048e+00
8.79304111e-02 4.07833248e-01 7.75592208e-01 -1.21007216e+00
-4.74302232e-01 -7.21089244e-01 -4.09442753e-01 6.38584018e-01
1.07958210e+00 -2.97527999e-01 -8.70733976e-01 6.88927352e-01]
|
[11.36500358581543, -2.1828155517578125]
|
3602c97c-63ac-46b8-b7ef-c07b7fc01003
|
image-augmentation-for-multitask-few-shot
|
2102.12295
| null |
https://arxiv.org/abs/2102.12295v1
|
https://arxiv.org/pdf/2102.12295v1.pdf
|
Image Augmentation for Multitask Few-Shot Learning: Agricultural Domain Use-Case
|
Large datasets' availability is catalyzing a rapid expansion of deep learning in general and computer vision in particular. At the same time, in many domains, a sufficient amount of training data is lacking, which may become an obstacle to the practical application of computer vision techniques. This paper challenges small and imbalanced datasets based on the example of a plant phenomics domain. We introduce an image augmentation framework, which enables us to extremely enlarge the number of training samples while providing the data for such tasks as object detection, semantic segmentation, instance segmentation, object counting, image denoising, and classification. We prove that our augmentation method increases model performance when only a few training samples are available. In our experiment, we use the DeepLabV3 model on semantic segmentation tasks with Arabidopsis and Nicotiana tabacum image dataset. The obtained result shows a 9% relative increase in model performance compared to the basic image augmentation techniques.
|
['Mariia Pukalchik', 'Dmitrii Shadrin', 'Sergey Nesteruk']
|
2021-02-24
| null | null | null | null |
['object-counting']
|
['computer-vision']
|
[ 5.76239526e-01 -1.12358641e-04 -4.41889539e-02 -2.76310384e-01
-3.07839870e-01 -4.38089609e-01 2.17221841e-01 4.48559940e-01
-4.75280464e-01 6.15626514e-01 -6.38584137e-01 -3.69830757e-01
2.60839075e-01 -9.95360970e-01 -7.60305643e-01 -6.25384748e-01
3.20702881e-01 6.45502865e-01 1.30804315e-01 -4.11359556e-02
1.31181329e-01 5.70892811e-01 -1.44987524e+00 1.66345879e-01
1.00852394e+00 1.25445306e+00 3.70191813e-01 4.39116448e-01
-5.12307346e-01 2.71124333e-01 -6.04367852e-01 -7.17521310e-02
3.60223681e-01 -1.91070646e-01 -7.34706700e-01 5.93881667e-01
3.98604363e-01 -2.33984783e-01 6.01255260e-02 1.14384687e+00
3.30239773e-01 -1.21514432e-01 3.06351155e-01 -1.43693638e+00
-4.73491907e-01 4.39645201e-01 -1.00626087e+00 1.10342689e-01
-4.01208639e-01 1.94716021e-01 7.44863451e-01 -4.28878605e-01
5.85723877e-01 1.10138392e+00 4.96093750e-01 3.47274035e-01
-1.24373722e+00 -4.25577134e-01 1.44370168e-01 2.03886196e-01
-1.04671860e+00 -2.07298547e-02 6.98628247e-01 -3.14468682e-01
3.97397757e-01 3.12129527e-01 6.28706574e-01 6.70904279e-01
-3.01326036e-01 9.26560760e-01 8.87429774e-01 -4.17946637e-01
4.37916964e-01 -2.73750667e-02 3.64619136e-01 5.30050516e-01
3.81802171e-01 -2.14859158e-01 3.55487615e-02 2.73970723e-01
8.17629099e-01 8.76839980e-02 -6.48166835e-02 -3.51829827e-01
-9.51430798e-01 6.86410248e-01 5.99870741e-01 3.69350612e-01
-4.21905726e-01 -9.37817693e-02 5.22364676e-01 2.08691105e-01
5.34211516e-01 4.34070021e-01 -6.26264036e-01 2.92539060e-01
-8.70424032e-01 1.90784201e-01 6.41745746e-01 6.90011799e-01
6.45040870e-01 2.09958330e-01 1.32835433e-01 9.79866028e-01
-1.65389240e-01 3.28144670e-01 4.03367430e-01 -8.26740444e-01
2.90901303e-01 1.03934526e+00 7.37185031e-02 -8.59124124e-01
-3.50779563e-01 -3.90440494e-01 -1.22190738e+00 3.08480501e-01
7.43720531e-01 5.35672419e-02 -1.17361188e+00 1.75521755e+00
5.00384033e-01 1.64817005e-01 -1.86167598e-01 6.55019343e-01
5.79044402e-01 6.17968619e-01 1.06868036e-01 -2.36441776e-01
1.27857494e+00 -8.35701585e-01 -6.78821087e-01 -3.16318333e-01
5.88927567e-01 -5.59325278e-01 1.27698207e+00 7.00513899e-01
-8.05230916e-01 -8.06793451e-01 -8.82157683e-01 -2.26352680e-02
-6.68367743e-01 3.75048846e-01 9.66685414e-01 6.85080171e-01
-5.66808522e-01 5.63145757e-01 -8.11671734e-01 -4.27870244e-01
1.14970255e+00 4.53013718e-01 -4.68636811e-01 -2.85915673e-01
-4.60011512e-01 5.46048343e-01 7.20267475e-01 3.10915619e-01
-7.70612776e-01 -7.55903244e-01 -5.32487035e-01 3.21572244e-01
4.74872351e-01 -3.04503858e-01 9.29180801e-01 -1.13461483e+00
-9.81598318e-01 9.81005788e-01 2.19854727e-01 -6.52403891e-01
5.11720181e-01 -1.33633345e-01 8.70026425e-02 -1.47678331e-01
-5.77212088e-02 9.24680471e-01 7.53297031e-01 -1.17027628e+00
-6.93226099e-01 -9.29542601e-01 -5.40352687e-02 -4.97048460e-02
-6.15769565e-01 -1.76534876e-01 -3.44284266e-01 -3.55286777e-01
1.69716701e-01 -8.08922708e-01 -5.36502779e-01 3.07521403e-01
-4.70650494e-01 6.84919432e-02 1.21199179e+00 -6.35332227e-01
5.07476747e-01 -2.06630707e+00 9.37136859e-02 7.26947049e-03
2.36116290e-01 6.59980714e-01 -3.66521955e-01 -1.74026117e-01
-2.63262510e-01 2.95240372e-01 -5.89679897e-01 -1.75443545e-01
-2.38775000e-01 4.38643754e-01 -3.76078300e-02 2.97721773e-01
3.71347100e-01 7.47877300e-01 -5.56274593e-01 -2.84309477e-01
4.19713467e-01 2.62069792e-01 -4.49535191e-01 1.46745458e-01
-4.98899639e-01 4.87279147e-01 -3.85897189e-01 8.30508709e-01
1.09530795e+00 -3.08924913e-01 1.31532475e-01 -2.12215424e-01
-2.67029321e-03 -3.55921030e-01 -1.21274257e+00 1.57413483e+00
-2.71721750e-01 5.01654685e-01 2.37509474e-01 -1.65840447e+00
8.94144833e-01 -2.91183665e-02 6.59184575e-01 -7.69079566e-01
2.20653042e-01 1.19025502e-02 2.59819835e-01 -2.34604791e-01
4.41752933e-02 1.51228577e-01 2.96149492e-01 1.33814588e-01
1.01229213e-02 -1.76668659e-01 6.74989343e-01 7.90277123e-02
9.68712211e-01 -3.03031467e-02 8.14249963e-02 -4.60125469e-02
4.95230883e-01 2.22943678e-01 6.02061868e-01 4.93903250e-01
-2.76825249e-01 4.93783027e-01 7.99253464e-01 -6.54834688e-01
-1.28138769e+00 -6.90096319e-01 -1.91009358e-01 8.20663452e-01
-6.94780052e-02 -8.39344226e-03 -1.10147822e+00 -6.78809345e-01
3.15359347e-02 3.83107215e-01 -6.40674770e-01 -1.06544659e-01
-2.89927930e-01 -1.51481926e+00 3.72690827e-01 5.11847973e-01
9.19463873e-01 -1.29077733e+00 -6.33287787e-01 5.96703812e-02
-7.62447193e-02 -1.42769992e+00 2.35104650e-01 3.09139878e-01
-1.16391969e+00 -1.13493299e+00 -4.55191642e-01 -7.75529802e-01
8.54842901e-01 2.37138957e-01 1.08976209e+00 2.24845037e-01
-8.75886321e-01 -2.33576715e-01 -3.42319667e-01 -8.10148716e-01
-4.23805207e-01 1.57534361e-01 -3.08674574e-01 -1.37959808e-01
2.33526602e-01 -5.63734472e-01 -4.07877147e-01 -4.26238403e-02
-1.16702199e+00 8.44622478e-02 5.07332623e-01 9.82316375e-01
6.66780949e-01 1.57514349e-01 7.23533809e-01 -1.15481949e+00
2.62224376e-01 -1.13829449e-01 -1.05007184e+00 1.03056848e-01
-5.85171521e-01 -3.61412883e-01 7.84540653e-01 -3.10864687e-01
-7.04133630e-01 3.21251214e-01 -2.89261997e-01 -1.14062555e-01
-6.32768989e-01 5.06758392e-01 -4.58312809e-01 -1.13909610e-01
5.59194624e-01 -4.71516803e-04 2.50840753e-01 -5.77791810e-01
2.21746907e-01 5.53361535e-01 3.98791850e-01 -2.74099916e-01
5.44040918e-01 5.15476823e-01 2.11840883e-01 -9.94822621e-01
-7.72268713e-01 -3.43817890e-01 -8.54954779e-01 7.61432797e-02
6.50663316e-01 -4.99549419e-01 -6.68872654e-01 8.03567469e-01
-9.72850144e-01 -2.71000147e-01 -4.30382013e-01 3.32325175e-02
-3.50027651e-01 4.37224627e-01 -4.59545434e-01 -4.84884083e-01
-5.78654170e-01 -1.20944142e+00 9.71487522e-01 2.88937896e-01
3.74786586e-01 -6.45054281e-01 -2.77106196e-01 6.11320138e-01
1.81850016e-01 3.92173588e-01 9.99243855e-01 -8.90917063e-01
-6.51639163e-01 -6.26334727e-01 -5.65087140e-01 7.76037276e-01
2.18431890e-01 1.15362518e-01 -1.10431242e+00 -1.46685794e-01
-1.72708603e-03 -5.34297228e-01 9.03278112e-01 5.74970067e-01
1.78249049e+00 2.67359853e-01 -2.20993832e-01 5.69740176e-01
1.47207987e+00 2.31213093e-01 7.26835608e-01 1.60595879e-01
9.67716694e-01 5.62361240e-01 7.27929831e-01 3.78366441e-01
-1.50955245e-02 3.67933780e-01 8.78408074e-01 -6.26217246e-01
1.53221920e-01 3.43822241e-01 -3.87034535e-01 3.98022264e-01
5.70958629e-02 -3.75492841e-01 -9.86408889e-01 6.00805759e-01
-1.68935359e+00 -6.50327861e-01 -1.40250489e-01 2.13507986e+00
6.40822709e-01 5.10824397e-02 -6.66294023e-02 6.16523623e-01
5.98085046e-01 -1.79321989e-01 -8.59659433e-01 -1.40289843e-01
-1.04386546e-01 2.59792566e-01 4.31599230e-01 6.49376437e-02
-1.32397938e+00 1.08111382e+00 5.19195509e+00 7.07352340e-01
-1.14594686e+00 -1.96019653e-02 1.26926315e+00 3.52461934e-01
2.60759920e-01 -2.33725131e-01 -4.89221752e-01 4.11584765e-01
6.04088128e-01 2.22958103e-01 2.55402654e-01 9.20777500e-01
1.53812900e-01 -2.04070300e-01 -8.45401585e-01 9.87936497e-01
-1.99425817e-01 -1.28064001e+00 1.62006944e-01 1.27876461e-01
6.18878722e-01 -1.17256477e-01 -7.00447634e-02 1.76811680e-01
-5.57509903e-03 -9.83709753e-01 -6.07124269e-02 -1.65689960e-01
3.21675420e-01 -8.59060347e-01 8.59218240e-01 6.03505373e-01
-6.45475447e-01 -1.89201847e-01 -6.99812174e-01 1.54912770e-01
-1.31907910e-01 9.15177524e-01 -9.43417907e-01 3.88909310e-01
5.82262814e-01 3.95857036e-01 -6.43322885e-01 1.16015303e+00
2.55718529e-01 7.21655130e-01 -5.24968624e-01 2.58240312e-01
1.35228783e-01 -5.27381897e-01 8.73675793e-02 6.43346012e-01
1.67849764e-01 -8.80563557e-02 3.80681753e-01 8.38979244e-01
-3.73341173e-01 2.41685972e-01 -5.19227922e-01 -2.90157229e-01
1.51175544e-01 1.56702816e+00 -1.31423378e+00 -3.22460294e-01
-2.87554920e-01 1.01975334e+00 1.58835202e-01 6.82753772e-02
-7.43110836e-01 -1.45484447e-01 2.35144466e-01 -6.15421571e-02
2.38812029e-01 -1.07236467e-01 -6.10488415e-01 -9.10317779e-01
-1.47568658e-01 -9.82815802e-01 1.78848356e-01 -5.08841097e-01
-8.91610324e-01 2.26149827e-01 -4.34622973e-01 -6.84468746e-01
2.44608685e-01 -9.01738763e-01 -4.62181479e-01 6.90711200e-01
-1.39074862e+00 -1.32315063e+00 -8.14050615e-01 2.71341026e-01
6.72919452e-01 -1.59547105e-01 8.39666426e-01 5.97843349e-01
-9.02338207e-01 2.09401667e-01 2.23543853e-01 2.31240064e-01
3.30647498e-01 -1.16249454e+00 5.24214268e-01 9.08817768e-01
2.47791827e-01 8.88548493e-02 4.88572210e-01 -4.09627497e-01
-1.22736287e+00 -1.11060083e+00 2.98749208e-01 -1.61116704e-01
3.23855549e-01 -3.03441316e-01 -9.94645834e-01 5.39956570e-01
-9.25809741e-02 3.53884995e-01 5.17213523e-01 -4.59698066e-02
1.23869721e-02 -4.18450385e-01 -1.49616206e+00 4.72487867e-01
6.34402633e-01 5.72541803e-02 1.12155333e-01 7.34031022e-01
5.48729718e-01 -2.42615715e-01 -6.07386172e-01 6.28380716e-01
3.00388694e-01 -7.71107852e-01 1.05214989e+00 -8.33090901e-01
4.18477774e-01 -2.35811487e-01 -1.53076455e-01 -1.16347587e+00
1.95367020e-02 -9.01254565e-02 5.67739233e-02 1.27163184e+00
1.49059176e-01 -2.02990279e-01 1.02578390e+00 3.99891138e-01
-2.00024061e-02 -4.84197795e-01 -6.47233605e-01 -5.01058161e-01
1.18036434e-01 -2.62825131e-01 4.72450286e-01 9.67060566e-01
-6.33449495e-01 2.15647683e-01 -6.15289919e-02 3.32809649e-02
5.86486340e-01 3.06137204e-01 7.74592876e-01 -1.62758005e+00
6.61328137e-02 -2.28410155e-01 -5.46141982e-01 -4.72237885e-01
8.85364041e-02 -6.79981709e-01 -5.38835004e-02 -1.44454920e+00
3.05678189e-01 -5.54762781e-01 -2.64949024e-01 5.57907164e-01
-2.48615488e-01 4.22223419e-01 2.06211463e-01 -2.99530208e-01
-2.87930161e-01 3.37940931e-01 1.41827619e+00 -4.22579318e-01
-1.24490261e-01 2.35215396e-01 -6.79145575e-01 8.16415370e-01
1.19751966e+00 -2.36115023e-01 -4.90662456e-01 -3.60238522e-01
-1.81813329e-01 -3.39485109e-01 4.82733011e-01 -1.03013158e+00
-2.20777139e-01 -9.64120999e-02 6.77123666e-01 -5.13034880e-01
2.02166244e-01 -9.73891199e-01 -2.27424204e-01 6.38454497e-01
-2.41632238e-01 -1.46149099e-01 5.41511476e-01 3.48164856e-01
-1.35656208e-01 -3.31946701e-01 1.06054366e+00 -3.83529335e-01
-8.02915335e-01 4.60908979e-01 -5.54602146e-02 -4.54415232e-02
1.04627013e+00 -2.50770181e-01 -2.34793946e-01 -9.59631801e-03
-7.86500335e-01 1.40847906e-01 2.31843621e-01 3.03169042e-01
4.49887067e-01 -9.16208208e-01 -5.27339518e-01 3.63493145e-01
-8.28640983e-02 3.97457153e-01 2.06516057e-01 7.43584573e-01
-7.69516170e-01 2.84027401e-02 -7.44679332e-01 -7.43731618e-01
-1.39407182e+00 7.42860317e-01 2.78565940e-02 -2.65363187e-01
-4.13622320e-01 7.94001400e-01 2.95877844e-01 -6.00195706e-01
3.11485738e-01 -6.04626238e-01 -3.15918416e-01 1.72989368e-01
4.46438015e-01 2.70961642e-01 3.26954573e-01 -1.59063488e-01
-7.19014779e-02 2.04360485e-01 -2.27052107e-01 4.06221211e-01
1.50468028e+00 1.52189583e-01 -3.47166568e-01 1.98811442e-01
7.85846353e-01 -5.53044438e-01 -9.78929162e-01 1.42824695e-01
1.04206301e-01 -3.98831666e-01 2.46562466e-01 -8.82612586e-01
-1.42317462e+00 1.21542394e+00 1.07260442e+00 4.45134342e-01
1.36818993e+00 -3.44384819e-01 6.72349274e-01 6.20893121e-01
1.90032497e-01 -1.20776498e+00 -4.47536707e-02 3.24410528e-01
5.44122577e-01 -1.58518982e+00 -1.14198096e-01 -5.71607232e-01
-4.96927291e-01 8.80906165e-01 9.04944301e-01 1.80865675e-02
4.84811783e-01 5.02506733e-01 1.21496264e-02 -1.37349730e-02
-4.97847050e-01 -3.37522894e-01 -1.66260764e-01 6.98524237e-01
4.08314675e-01 -7.56706763e-03 -2.28703007e-01 4.36477214e-01
2.45759398e-01 1.84776992e-01 5.55722058e-01 7.68995643e-01
-5.68953454e-01 -1.22438419e+00 -3.03169906e-01 6.43221498e-01
-5.82691550e-01 -7.88231567e-02 -4.45407540e-01 7.97238529e-01
3.58411074e-01 6.08169734e-01 1.83333725e-01 1.37449250e-01
2.03811899e-01 2.78271250e-02 5.27688324e-01 -5.28708756e-01
-3.69204968e-01 8.64100084e-02 -2.16886416e-01 -3.54436219e-01
-3.35515589e-01 -4.28669214e-01 -1.17323363e+00 -1.84111342e-01
-4.40283984e-01 -1.89164013e-01 9.76061106e-01 8.54173064e-01
2.14108348e-01 8.08034003e-01 3.39839906e-01 -6.72297001e-01
-5.18852711e-01 -9.91043031e-01 -5.94843328e-01 4.64929998e-01
5.26471660e-02 -4.57323641e-01 4.96839844e-02 3.92390013e-01]
|
[9.100342750549316, -1.4821122884750366]
|
74c69c71-1b44-4a41-b639-9eb8590809c8
|
online-multi-object-tracking-framework-with
|
1907.13347
| null |
https://arxiv.org/abs/1907.13347v1
|
https://arxiv.org/pdf/1907.13347v1.pdf
|
Online Multi-Object Tracking Framework with the GMPHD Filter and Occlusion Group Management
|
In this paper, we propose an efficient online multi-object tracking framework based on the GMPHD filter and occlusion group management scheme where the GMPHD filter utilizes hierarchical data association to reduce the false negatives caused by miss detection. The hierarchical data association consists of two steps: detection-to-track and track-to-track associations, which can recover the lost tracks and their switched IDs. In addition, the proposed framework is equipped with an object grouping management scheme which handles occlusion problems with two main parts. The first part is "track merging" which can merge the false positive tracks caused by false positive detections from occlusions, where the false positive tracks are usually occluded with a measure. The measure is the occlusion ratio between visual objects, sum-of-intersection-over-area (SIOA) we defined instead of the IOU metric. The second part is "occlusion group energy minimization (OGEM)" which prevents the occluded true positive tracks from false "track merging". We define each group of the occluded objects as an energy function and find an optimal hypothesis which makes the energy minimal. We evaluate the proposed tracker in benchmark datasets such as MOT15 and MOT17 which are built for multi-person tracking. An ablation study in training dataset shows that not only "track merging" and "OGEM" complement each other but also the proposed tracking method has more robust performance and less sensitive to parameters than baseline methods. Also, SIOA works better than IOU for various sizes of false positives. Experimental results show that the proposed tracker efficiently handles occlusion situations and achieves competitive performance compared to the state-of-the-art methods. Especially, our method shows the best multi-object tracking accuracy among the online and real-time executable methods.
|
['Kin-Choong Yow', 'Young-chul Yoon', 'Young-min Song', 'Kwangjin Yoon', 'Moongu Jeon']
|
2019-07-31
| null | null | null | null |
['online-multi-object-tracking', 'real-time-multi-object-tracking']
|
['computer-vision', 'computer-vision']
|
[-3.52646798e-01 -4.22808766e-01 -1.50528073e-01 3.45920138e-02
-4.82531488e-01 -4.03970003e-01 2.51937956e-01 1.88483953e-01
-4.06033754e-01 8.23432207e-01 -8.69820565e-02 8.54171291e-02
-1.40093207e-01 -7.39437819e-01 -8.30337286e-01 -7.34824181e-01
-1.77950580e-02 6.56143010e-01 1.01888764e+00 2.25667417e-01
3.90578955e-02 4.65473562e-01 -1.89862454e+00 -4.96973842e-02
7.81696975e-01 9.39554811e-01 2.54330218e-01 3.33869249e-01
-1.63999662e-01 3.54667425e-01 -7.08237588e-01 -2.27585703e-01
4.77993459e-01 6.93431348e-02 -2.57430941e-01 -8.03328212e-03
7.32779860e-01 -3.56439769e-01 -2.64783353e-01 1.13110888e+00
7.24339664e-01 -3.63386958e-03 3.97469014e-01 -1.64455271e+00
-1.42731622e-01 1.48791268e-01 -9.84899223e-01 6.44067585e-01
1.85867965e-01 1.50344074e-01 3.94713491e-01 -7.41970181e-01
5.78786075e-01 1.56566358e+00 1.01744676e+00 2.96629637e-01
-1.07092988e+00 -1.12672472e+00 2.26853505e-01 2.98031300e-01
-1.68438482e+00 -2.27914020e-01 1.16336130e-01 -6.00146472e-01
4.95044321e-01 5.40567517e-01 5.47785282e-01 3.48481387e-01
4.36863542e-01 7.11999238e-01 7.84229755e-01 -2.30843708e-01
-8.18571970e-02 2.18477488e-01 5.78349650e-01 6.40390098e-01
8.63457978e-01 4.12524134e-01 -2.93147832e-01 -2.41002724e-01
5.37946045e-01 1.71364099e-01 -9.92640406e-02 -3.99439394e-01
-1.12415969e+00 5.73005021e-01 3.95883828e-01 1.37168512e-01
-2.66466171e-01 1.09372117e-01 3.64506364e-01 -1.36575317e-02
1.94088712e-01 -4.62899715e-01 -1.99664384e-01 1.69085056e-01
-8.42247903e-01 4.00505394e-01 3.07788700e-01 1.21898639e+00
6.64160669e-01 -5.86892515e-02 -6.61645114e-01 5.26616395e-01
6.29164457e-01 8.58860433e-01 1.80590108e-01 -4.61787373e-01
4.05320525e-01 8.79627943e-01 4.65568513e-01 -1.09738922e+00
-5.95396757e-01 -7.13603616e-01 -7.19252050e-01 3.35900575e-01
4.12416875e-01 -1.24548554e-01 -8.83359730e-01 1.71966350e+00
9.96569037e-01 5.08333743e-01 -2.26358041e-01 8.60041261e-01
9.14318144e-01 4.85755771e-01 2.87357777e-01 -5.94705701e-01
1.72386515e+00 -8.23670447e-01 -1.18669617e+00 -2.14405032e-03
3.50315809e-01 -1.03327072e+00 2.71870136e-01 7.04086646e-02
-9.57947493e-01 -1.04010403e+00 -1.06839955e+00 4.10207987e-01
-4.36975807e-01 3.48000407e-01 3.71727526e-01 7.52812028e-01
-8.21779251e-01 2.78466076e-01 -7.75267243e-01 -5.70663154e-01
4.28685188e-01 6.04489863e-01 -9.46634859e-02 1.88085541e-01
-8.02742660e-01 8.38274479e-01 6.71394348e-01 1.10100277e-01
-6.80618346e-01 -7.53763258e-01 -5.04068494e-01 -9.97552872e-02
5.16526103e-01 -5.87759674e-01 7.53003538e-01 -4.17384326e-01
-7.61913180e-01 5.76226294e-01 -3.61567348e-01 -4.55108345e-01
6.14636838e-01 -3.15453112e-01 -8.21595252e-01 -2.99293369e-01
4.25448269e-01 5.67448616e-01 5.27612031e-01 -1.24744022e+00
-1.25666356e+00 -2.74642438e-01 -3.89669329e-01 7.67919943e-02
-1.74747363e-01 2.88712293e-01 -8.14163327e-01 -5.58132231e-01
1.68916538e-01 -9.78533268e-01 1.28732875e-01 1.92953438e-01
-3.53086621e-01 -4.98937190e-01 1.45818353e+00 -3.78971785e-01
1.57918906e+00 -2.00504160e+00 -2.95983076e-01 1.93870872e-01
1.61402822e-01 5.18063843e-01 2.32706860e-01 1.10606074e-01
1.76475033e-01 -3.08708400e-01 4.76948172e-01 -3.88928026e-01
1.07039865e-02 5.71997203e-02 -1.23371996e-01 6.35394812e-01
-3.19334567e-01 5.14979661e-01 -7.71083593e-01 -9.78072286e-01
5.56040645e-01 3.97606105e-01 -1.67559892e-01 3.08172312e-02
3.73859107e-02 3.96483004e-01 -5.04747510e-01 8.40941846e-01
1.25233269e+00 -1.62341893e-01 -1.00576296e-01 -5.86794138e-01
-6.46393359e-01 -3.90521348e-01 -2.05811405e+00 1.22297359e+00
3.97345126e-01 1.61920995e-01 -3.90466377e-02 -3.48524511e-01
8.48364294e-01 1.88094839e-01 7.26885736e-01 -4.86914605e-01
2.89408207e-01 -2.64288578e-02 -2.24472225e-01 -2.36548275e-01
6.00843489e-01 2.86866933e-01 1.01020813e-01 4.87800799e-02
-2.79267013e-01 1.07477009e+00 4.33280736e-01 3.11754137e-01
1.06443942e+00 2.30244666e-01 1.53378561e-01 -4.48090971e-01
7.67513275e-01 2.00965211e-01 1.18336844e+00 8.98458242e-01
-5.84676504e-01 2.90573630e-02 -2.68110424e-01 -4.75264847e-01
-5.15514255e-01 -1.32638884e+00 -3.21388006e-01 7.57458031e-01
9.86552060e-01 -4.27411914e-01 -4.20593351e-01 -5.86943686e-01
3.64729017e-01 2.22030774e-01 -3.98218781e-01 -3.56351957e-02
-5.72846413e-01 -9.21665788e-01 4.48603898e-01 5.36297619e-01
6.26919985e-01 -9.52219844e-01 -6.69766426e-01 3.17256600e-01
-2.32377693e-01 -1.06612754e+00 -7.02822030e-01 -3.18926901e-01
-7.06398308e-01 -1.22268546e+00 -2.49761954e-01 -6.66487753e-01
6.64792299e-01 5.96063793e-01 7.92904615e-01 4.51079816e-01
-5.00878453e-01 1.19408295e-01 -2.35530168e-01 -6.36119962e-01
-5.52783459e-02 -2.93298542e-01 3.80947292e-01 1.53352365e-01
4.05853659e-01 -9.23749581e-02 -7.05133915e-01 9.05922174e-01
-4.36426520e-01 -4.29808162e-02 6.01834714e-01 3.50097239e-01
1.03519166e+00 3.07531506e-01 5.22927761e-01 -4.14405197e-01
-8.32384378e-02 -3.81605446e-01 -1.00353074e+00 4.05056357e-01
-4.48899001e-01 -2.25082055e-01 2.79824495e-01 -6.71506047e-01
-9.52322185e-01 1.72012299e-01 2.67305642e-01 -6.04674160e-01
3.91836986e-02 -3.69967788e-01 -4.02537405e-01 -3.34019244e-01
2.37294227e-01 4.98877354e-02 -2.60069966e-01 -6.85380220e-01
2.41536126e-02 3.71549189e-01 7.93641150e-01 -3.88020456e-01
1.16153848e+00 7.96105206e-01 1.67361796e-01 -5.27482748e-01
-5.33979952e-01 -8.43234062e-01 -4.34832692e-01 -6.56039894e-01
9.34599161e-01 -1.05497730e+00 -1.30942178e+00 4.01446730e-01
-1.05826366e+00 4.78173524e-01 -8.87867659e-02 6.88866377e-01
-4.54500876e-02 5.11118531e-01 -3.70065570e-01 -1.16599536e+00
-4.89116043e-01 -1.06148291e+00 1.10676563e+00 8.20704043e-01
3.23143989e-01 -4.91512567e-01 8.44331458e-02 1.07847326e-01
2.28226513e-01 5.36770701e-01 1.82776287e-01 -4.34250265e-01
-1.04394460e+00 -2.65777946e-01 -3.52129698e-01 -3.14575613e-01
1.44349113e-02 -1.80130020e-01 -6.05709076e-01 -8.17115009e-01
-3.20297897e-01 1.99589103e-01 6.18588209e-01 6.12468660e-01
7.17143774e-01 5.99833718e-03 -1.29119444e+00 4.39855218e-01
1.69652057e+00 4.34880763e-01 6.61939621e-01 5.83270133e-01
7.84609556e-01 8.12060758e-02 1.26318502e+00 4.37989265e-01
4.69268352e-01 1.12703133e+00 5.22714317e-01 -1.65594295e-01
-3.88181865e-01 -1.00985117e-01 2.44967014e-01 5.19837141e-01
7.08506703e-02 -3.18068147e-01 -4.15378571e-01 4.82816964e-01
-2.28445220e+00 -1.20729887e+00 -7.64336526e-01 2.54264593e+00
3.27689916e-01 4.13904428e-01 4.95369196e-01 -3.83532122e-02
1.22535336e+00 -2.40724295e-01 -3.93412173e-01 3.81122321e-01
-9.27027538e-02 -1.34269059e-01 8.53163660e-01 2.89240420e-01
-1.52376878e+00 7.40553737e-01 5.34430933e+00 9.19000447e-01
-5.06376326e-01 4.14713830e-01 -2.30623111e-01 -2.07447130e-02
5.27450740e-01 5.03054895e-02 -1.79864764e+00 9.32476640e-01
5.13709366e-01 1.72938421e-01 -1.69113338e-01 6.09436870e-01
2.75711954e-01 -3.37445438e-01 -7.30677962e-01 1.01268554e+00
-1.99845806e-01 -1.06549037e+00 -3.15891989e-02 1.42405495e-01
4.05834198e-01 -2.36834258e-01 -3.21586251e-01 3.41121048e-01
2.91368127e-01 -2.80060887e-01 9.26803887e-01 6.85941637e-01
3.84864956e-01 -6.80649817e-01 6.49461031e-01 3.39440882e-01
-2.23169303e+00 -9.51972529e-02 -5.96901536e-01 3.17575812e-01
4.83716637e-01 4.95681942e-01 -3.67511183e-01 1.00173545e+00
1.01975679e+00 3.95225734e-01 -6.42501414e-01 1.76909137e+00
4.03577864e-01 1.04149126e-01 -6.52719378e-01 2.11789817e-01
-2.14354992e-01 2.85866521e-02 8.64603102e-01 1.11230516e+00
2.40904704e-01 7.60585591e-02 8.59824181e-01 5.59130073e-01
3.81139308e-01 9.10649151e-02 -1.77758604e-01 7.31759489e-01
9.17001069e-01 1.35867107e+00 -1.01518130e+00 -5.74927807e-01
-4.03980464e-01 3.89316797e-01 4.34002206e-02 -1.11973189e-01
-1.45875990e+00 -7.44258463e-02 5.30328989e-01 2.78944194e-01
5.87084532e-01 1.86299697e-01 7.59784058e-02 -8.13387692e-01
2.00556725e-01 -5.07618845e-01 8.73827636e-01 -4.41205114e-01
-9.44024444e-01 4.21281219e-01 2.23302364e-01 -1.69325042e+00
4.03626621e-01 -1.95129484e-01 -5.41182756e-01 6.53213561e-01
-1.19255292e+00 -1.32635796e+00 -6.39773965e-01 6.93101466e-01
5.35652995e-01 6.95100520e-03 2.26763278e-01 1.12670088e+00
-8.51770580e-01 8.99787009e-01 2.09191367e-02 -5.74784540e-02
7.40847349e-01 -8.43987763e-01 -6.00101054e-02 1.18518400e+00
-3.61787736e-01 3.85456890e-01 8.37048769e-01 -1.31145787e+00
-1.29239297e+00 -1.38952780e+00 6.07623219e-01 -3.70186090e-01
2.21835792e-01 -2.03241229e-01 -7.63193607e-01 8.43193531e-01
-1.43397853e-01 1.49817333e-01 2.39780590e-01 -1.20244615e-01
1.00834183e-01 -3.85461390e-01 -1.28957832e+00 1.72128245e-01
1.27183878e+00 4.26248610e-01 -4.98065948e-01 4.80913341e-01
4.72136319e-01 -5.41393220e-01 -9.00693715e-01 7.12702274e-01
6.44408464e-01 -7.36151099e-01 1.24032533e+00 -6.25757724e-02
-8.67497802e-01 -1.37381256e+00 8.53931084e-02 -6.14215553e-01
-6.99548244e-01 -4.11901116e-01 -4.23324317e-01 1.77177393e+00
-1.04548909e-01 -4.84211087e-01 6.68588340e-01 1.56892642e-01
-2.28132039e-01 -2.93569207e-01 -1.00523174e+00 -1.21475804e+00
-7.38554418e-01 1.08824462e-01 5.91687858e-01 7.24992633e-01
-6.33835495e-01 4.24790457e-02 -6.37525976e-01 6.61125839e-01
1.25368607e+00 8.82759839e-02 1.10891867e+00 -1.58625758e+00
-8.70684534e-02 -6.95885420e-02 -5.15847325e-01 -8.47607315e-01
-4.04272377e-01 -6.11979008e-01 6.54804483e-02 -1.42587805e+00
4.82382357e-01 -6.73206687e-01 -4.05722469e-01 4.95287240e-01
-3.65866959e-01 1.86963290e-01 3.60986531e-01 4.95550841e-01
-1.17531574e+00 2.69305587e-01 8.76687527e-01 -1.16703928e-01
-3.82306278e-01 1.25605017e-01 -2.60520279e-01 5.90701044e-01
3.20372760e-01 -9.18593943e-01 -1.16230370e-02 -1.80211663e-01
-3.82277399e-01 -7.16620013e-02 4.84911174e-01 -1.52123439e+00
6.03771627e-01 4.22478206e-02 6.85103178e-01 -1.62930465e+00
2.63270617e-01 -1.07722259e+00 1.00037205e+00 1.00059927e+00
4.65880036e-01 3.34293008e-01 5.30259609e-01 7.49240875e-01
2.34175250e-01 1.43965855e-01 9.16428149e-01 2.34100297e-01
-8.28385711e-01 3.64924312e-01 6.72990382e-02 -3.12893778e-01
1.54436016e+00 -5.36638975e-01 -7.19622850e-01 2.54438549e-01
-5.86434662e-01 8.10294926e-01 2.75233030e-01 6.96677506e-01
2.23140270e-01 -1.81602728e+00 -7.09660709e-01 6.88101873e-02
7.80736655e-02 -2.69800037e-01 4.34122413e-01 1.09470940e+00
-2.52591431e-01 2.85166711e-01 -2.35087708e-01 -1.03390610e+00
-1.86244559e+00 7.74336576e-01 2.20851049e-01 -5.55512607e-01
-7.51668155e-01 3.87688309e-01 3.55380774e-01 1.10000208e-01
5.58719516e-01 -9.78480950e-02 -2.33063936e-01 -3.44676860e-02
7.74434090e-01 7.86012650e-01 -2.04800382e-01 -1.05325580e+00
-8.23417246e-01 8.82495761e-01 -1.76092535e-01 3.93993646e-01
7.03844547e-01 -2.96323925e-01 1.11648224e-01 -2.67465562e-02
5.66623092e-01 -6.59148097e-02 -1.17713356e+00 -1.78074643e-01
-2.14228649e-02 -6.56497598e-01 -2.42409438e-01 -7.06819773e-01
-1.06714475e+00 1.24097928e-01 1.47831607e+00 1.23004623e-01
9.37027633e-01 -1.72966182e-01 9.60827529e-01 -1.79630920e-01
5.61368227e-01 -1.03432107e+00 -2.57250547e-01 1.61087766e-01
3.07934582e-01 -1.05047524e+00 3.23199630e-01 -6.50811791e-01
-1.28475264e-01 5.57633758e-01 1.06133771e+00 -1.08250096e-01
5.96896708e-01 4.77692246e-01 -2.15485424e-01 -2.90006250e-01
-3.48983407e-01 -5.68386436e-01 3.83257091e-01 4.46309566e-01
-2.93454558e-01 -1.44989723e-02 -7.72862613e-01 4.14499372e-01
1.23537600e-01 1.28509536e-01 -5.26383705e-02 1.02113199e+00
-8.78175199e-01 -9.50262725e-01 -1.14633262e+00 4.43584919e-01
-4.13472950e-01 5.38192809e-01 1.48490995e-01 1.12051296e+00
8.37711573e-01 1.06994283e+00 2.21361518e-01 -4.82666612e-01
6.90444648e-01 -2.35964999e-01 3.64084125e-01 -1.92019522e-01
-8.02794397e-01 3.56607407e-01 -5.61556406e-02 -4.88771528e-01
-5.83243251e-01 -7.12158620e-01 -1.41410780e+00 -4.79112744e-01
-8.92702639e-01 1.48602232e-01 4.11071002e-01 9.22977805e-01
3.95624042e-01 6.88027799e-01 3.21927428e-01 -6.09978199e-01
-1.50648519e-01 -7.91663468e-01 -3.95578802e-01 6.05063200e-01
1.25701129e-01 -1.43966401e+00 -1.35601813e-03 -9.03111398e-02]
|
[6.505015850067139, -1.9859169721603394]
|
b2af89d6-7040-426d-b2f5-ec976d79908e
|
multi-label-ecg-classification-using
| null | null |
https://ieeexplore.ieee.org/abstract/document/9662750
|
https://www.cinc.org/archives/2021/pdf/CinC2021-075.pdf
|
Multi-Label ECG Classification Using Convolutional Neural Networks in a Classifier Chain
|
Over the last decade, AI has shown its feasibility in classifying heart-related diagnoses from ECGs. Earlier studies have mainly focused on 12 and 2-lead ECGs, but we aim to classify 26 different diagnoses based on 12, 6, 4, 3, and 2-lead ECGs in this study. We trained a supervised model on a dataset containing 88 253 ECGs with 26 different diagnoses used as ground truth. The training and classification steps can be separated into three parts. (1) Pan Tompkins algorithm was used to find peaks and calculate the average heart rate. (2) The average heart rate and the Fourier transformed ECG signal was used to train convolutional neural networks (CNN) system that classified the ECGs with regular or irregular rhythms. 9 out of 26 classes were classified in this step. (3) Finally, CNN models in a classifier chain were trained to classify the remaining 17 diagnoses. The classification results from step 2 and the raw ECG signal were used as input to the classifier chain in step 3. Our team, CardiOUS, achieved a PhysioNet Challenge score of −0.63 for all sets of leads on the hidden test set. Based on the test score, our team placed at 38th out of 39 teams in the official ranking.
|
['Pål Haugar Brekke', 'Eraraya Morenzo Muten', 'Bjørn-Jostein Singstad']
|
2022-01-10
| null | null | null |
computing-in-cardiology-2022-1
|
['ecg-classification']
|
['medical']
|
[ 1.19745329e-01 2.53455900e-03 2.25090623e-01 -3.22374851e-01
-8.35491598e-01 -4.52309728e-01 -2.88707286e-01 3.59482467e-01
-2.12632418e-01 8.41643572e-01 -1.63468599e-01 -5.19069493e-01
-3.93611789e-01 -6.83878422e-01 -1.46477312e-01 -6.14809096e-01
-4.94301379e-01 5.19867539e-01 -1.56389307e-02 -2.84716487e-03
8.06759000e-02 4.61082399e-01 -1.03672922e+00 4.82656658e-01
8.03741634e-01 1.38065886e+00 -5.34898758e-01 1.05074751e+00
4.27917480e-01 8.35166276e-01 -1.03298926e+00 -1.33285761e-01
1.62516013e-01 -1.07597613e+00 -7.65247881e-01 -2.18503103e-01
8.05140212e-02 -2.22097933e-01 -1.55762151e-01 3.65150034e-01
1.08960128e+00 -4.83054608e-01 5.26717603e-01 -1.02265167e+00
-6.83575645e-02 9.03521776e-01 -1.04508758e-01 4.98281777e-01
2.62780011e-01 1.00209415e-01 6.91996574e-01 -6.52415752e-01
3.14497977e-01 3.75700891e-01 1.29577112e+00 2.08429903e-01
-1.05361295e+00 -7.04330206e-01 -6.30678296e-01 3.71019751e-01
-1.64932168e+00 -6.15836978e-02 7.30056465e-01 -5.31005144e-01
7.71001637e-01 4.02846605e-01 1.31046546e+00 7.30275333e-01
4.05906916e-01 1.03176340e-01 1.08731592e+00 -3.59302968e-01
1.27827227e-01 -9.17019248e-02 1.87904149e-01 4.69077289e-01
3.14256489e-01 3.70340496e-02 -2.04567164e-01 -2.81567037e-01
6.70181155e-01 -1.55077636e-01 -2.69712150e-01 3.57726067e-01
-1.42838478e+00 6.35027409e-01 3.19242358e-01 4.09998924e-01
-6.75277054e-01 -1.96241885e-01 5.47860146e-01 6.56256616e-01
2.89326668e-01 7.52893806e-01 -6.26299024e-01 -1.99909627e-01
-1.07627475e+00 1.70128971e-01 9.31023479e-01 2.62995213e-01
2.69985855e-01 1.96158290e-01 -4.26287323e-01 7.34067440e-01
1.11632198e-02 4.81669724e-01 4.39926803e-01 -8.45846176e-01
1.70926318e-01 5.98022819e-01 -1.32328466e-01 -1.00064015e+00
-9.48396325e-01 -9.34685707e-01 -1.11025858e+00 6.43280745e-02
5.08572042e-01 -5.45475245e-01 -9.05140996e-01 1.05724406e+00
-8.00511539e-02 -4.20494284e-03 3.12250108e-02 9.98492002e-01
1.14104486e+00 1.97900414e-01 -1.55834571e-01 -2.25563273e-01
1.27262652e+00 -2.30407983e-01 -5.94867289e-01 4.25334603e-01
6.49397254e-01 -6.71326637e-01 4.54452693e-01 8.49501014e-01
-1.20042586e+00 -8.30290675e-01 -1.18341303e+00 4.06941056e-01
2.56578960e-02 4.91262347e-01 2.89147317e-01 6.47587061e-01
-1.17857432e+00 1.02569759e+00 -7.26368308e-01 -2.78141975e-01
4.88891512e-01 3.45464706e-01 -4.38220575e-02 3.07906210e-01
-1.55121887e+00 8.95840168e-01 3.03659976e-01 1.95723191e-01
-8.48745465e-01 -6.87601447e-01 -5.16989946e-01 6.71462044e-02
-2.53867298e-01 -7.32848763e-01 8.30661893e-01 -8.71928930e-01
-1.14696097e+00 1.04181504e+00 1.95663646e-01 -5.99183202e-01
7.13616073e-01 2.24712640e-01 -6.28380299e-01 2.51688957e-01
1.63526922e-01 1.59377560e-01 6.50054634e-01 -8.24764073e-01
-7.58257627e-01 -3.03892910e-01 -1.19106367e-01 -3.50066759e-02
1.42068624e-01 -7.97198266e-02 -4.41882014e-03 -5.63103199e-01
5.06400943e-01 -8.28998148e-01 1.36022151e-01 -5.15066564e-01
-6.01513565e-01 -1.24150053e-01 2.18433097e-01 -8.56312037e-01
1.42246938e+00 -2.09072208e+00 -2.16403455e-02 4.92306471e-01
7.52966940e-01 2.07711637e-01 3.13117087e-01 3.09262604e-01
-5.92876434e-01 1.53122231e-01 -1.03687234e-01 4.17156741e-02
-3.44878972e-01 -9.94810835e-02 2.76452489e-02 3.87648463e-01
3.26734602e-01 6.45888090e-01 -7.16620803e-01 -3.44908476e-01
1.08800903e-01 4.64440197e-01 -5.02612963e-02 1.38533652e-01
5.99704146e-01 7.75228381e-01 -3.93154379e-03 5.50162733e-01
3.79396081e-01 -2.24875763e-01 2.62357116e-01 -4.25047100e-01
9.81097370e-02 4.23959851e-01 -8.52033257e-01 1.21131992e+00
7.08945245e-02 7.37179637e-01 -5.09418249e-01 -1.10715175e+00
1.16769850e+00 8.98221612e-01 9.07141566e-01 -3.76959622e-01
1.89546347e-01 3.75596166e-01 8.70789945e-01 -6.38670206e-01
-3.31133574e-01 -1.01014167e-01 3.16019319e-02 4.54416156e-01
-7.88296834e-02 9.11418572e-02 1.78889066e-01 -1.34620994e-01
1.26795912e+00 -1.56876937e-01 2.30390579e-01 -2.26272896e-01
2.51695782e-01 8.09013769e-02 6.43861294e-01 8.87970924e-01
-3.75531524e-01 1.11189651e+00 8.01912725e-01 -1.23005021e+00
-7.78938174e-01 -1.08377755e+00 -5.10682821e-01 3.42057019e-01
-4.19678181e-01 -4.95407045e-01 -6.51224494e-01 -4.59248334e-01
-1.89524904e-01 2.72618473e-01 -6.60328269e-01 -2.28297681e-01
-4.90559161e-01 -9.53897297e-01 1.17587841e+00 6.49873734e-01
4.88978058e-01 -1.36479425e+00 -1.08147252e+00 3.70622218e-01
-5.28031588e-01 -4.64672953e-01 1.86336353e-01 5.10505497e-01
-8.50747764e-01 -1.38819218e+00 -8.67448926e-01 -6.11312747e-01
3.30840051e-01 -3.86588871e-01 1.28792775e+00 3.24521273e-01
-6.21104836e-01 -8.07196833e-04 -3.97953719e-01 -8.93769741e-01
-3.66691440e-01 1.88949868e-01 4.40290719e-02 1.55581906e-01
1.68316826e-01 -6.76070094e-01 -7.84562886e-01 1.08423501e-01
-2.84051090e-01 -1.77705511e-01 4.59026098e-01 7.21799910e-01
4.33096588e-01 -9.01030228e-02 8.86078835e-01 -7.20833302e-01
7.29252875e-01 -3.92577082e-01 -1.97596643e-02 -8.77553225e-02
-7.93678880e-01 -6.55075669e-01 5.72596192e-01 -2.62273222e-01
-1.47492886e-01 1.77672170e-02 -3.66374433e-01 -3.93437475e-01
-3.38467479e-01 7.57838845e-01 3.98828238e-01 3.66205543e-01
9.47820306e-01 1.03286475e-01 -5.56452796e-02 -1.37293441e-02
-5.36228657e-01 7.45928049e-01 5.23608088e-01 -2.80242682e-01
4.78934526e-01 1.61424890e-01 1.05987683e-01 -5.56433916e-01
-8.37415218e-01 -2.19883397e-01 -8.30857754e-01 -3.63151431e-01
1.11512518e+00 -7.48449802e-01 -6.88407362e-01 6.43790126e-01
-1.06467032e+00 -2.53092647e-01 -3.77479732e-01 6.60933375e-01
-1.99916273e-01 1.52469069e-01 -6.73682809e-01 -8.36770713e-01
-8.33424449e-01 -7.38779366e-01 7.62339234e-01 4.74342257e-02
-8.51153076e-01 -8.13794374e-01 6.00914434e-02 1.15343891e-01
4.42474693e-01 8.28386545e-01 9.48316932e-01 -8.03846836e-01
1.37726948e-01 -5.59242189e-01 6.54275529e-03 3.83960843e-01
2.79800683e-01 -4.12064306e-02 -1.13538122e+00 -1.33575976e-01
3.25094461e-02 -1.09171905e-01 5.79191089e-01 4.99311388e-01
1.04380107e+00 2.39804551e-01 -7.02315643e-02 7.00202703e-01
9.69463229e-01 6.38673067e-01 8.84178638e-01 6.45459816e-02
6.57746196e-01 1.84943751e-01 1.97578758e-01 3.60278249e-01
2.34211445e-01 3.21502090e-01 1.45313472e-01 -6.31060123e-01
-8.06789473e-03 2.42946550e-01 -2.70924214e-02 9.00454938e-01
-6.44133627e-01 2.33952731e-01 -1.34794950e+00 5.06169617e-01
-1.50108588e+00 -8.51136327e-01 -7.06902385e-01 2.19810867e+00
7.25578964e-01 4.70148921e-01 2.29280263e-01 9.47369099e-01
6.15100503e-01 -4.25100684e-01 -5.10926247e-01 -4.03220236e-01
6.68681115e-02 6.97555006e-01 6.13986589e-02 -1.57453030e-01
-1.13487875e+00 7.56751746e-02 6.92381811e+00 4.89670485e-02
-1.60115170e+00 -9.67053398e-02 9.01782751e-01 7.36565143e-02
4.83896613e-01 -2.92277992e-01 -2.35501066e-01 4.61466908e-01
1.20722461e+00 1.99147314e-02 1.92452207e-01 4.11384910e-01
2.84984469e-01 1.11134693e-01 -1.16844082e+00 1.17979741e+00
5.87154515e-02 -1.09420156e+00 -3.19660842e-01 -8.66490975e-02
4.65548337e-01 1.41663387e-01 -3.36636245e-01 2.56720781e-01
-4.22636300e-01 -1.28314281e+00 5.72557271e-01 7.80865431e-01
1.29937458e+00 -6.75297678e-01 1.23865199e+00 1.98156685e-01
-9.63123322e-01 -1.37497574e-01 -5.39617836e-02 -3.23976338e-01
-2.42505953e-01 6.76307559e-01 -9.63349640e-01 8.43674958e-01
9.01761830e-01 8.56030166e-01 -5.08720040e-01 1.12100387e+00
-1.92331582e-01 1.20598972e+00 -2.00081885e-01 4.07933652e-01
-2.99355030e-01 1.52121753e-01 3.89671713e-01 1.06685567e+00
2.77347565e-01 1.70965418e-01 2.13532388e-01 7.05127120e-01
1.63844272e-01 -1.27260402e-01 -2.37094998e-01 2.86672711e-01
1.77696690e-01 1.32241058e+00 -1.01116967e+00 -6.31878376e-01
5.04466929e-02 5.64961076e-01 -2.11794242e-01 2.20517024e-01
-1.00038850e+00 -9.50438976e-01 3.96846831e-02 3.91454935e-01
-1.60258368e-01 3.08808386e-01 -8.78194213e-01 -9.17747974e-01
-1.88426804e-02 -1.01364064e+00 5.56079865e-01 -7.36551344e-01
-1.30930495e+00 9.19101000e-01 -3.17773968e-01 -1.35325432e+00
-2.30461508e-01 -3.09363663e-01 -8.82264137e-01 1.29402804e+00
-9.13465500e-01 -4.92576301e-01 -6.77819967e-01 4.94608581e-01
1.73728064e-01 -1.15877643e-01 1.38756108e+00 7.19870389e-01
-3.95807743e-01 5.30205309e-01 -4.88152862e-01 7.64458716e-01
7.22855806e-01 -1.44468307e+00 1.27733678e-01 3.60260069e-01
-9.21642035e-02 5.21025658e-01 2.17060745e-01 -4.18466508e-01
-7.28242695e-01 -1.14697695e+00 1.29373932e+00 -5.00850201e-01
1.55012503e-01 2.81462282e-01 -7.68870115e-01 3.36037815e-01
3.93328741e-02 1.17942244e-01 8.44273567e-01 5.59815168e-02
1.15310371e-01 -3.61836851e-01 -9.97552872e-01 2.78268382e-03
5.96398950e-01 -4.34948564e-01 -5.32322764e-01 2.45785471e-02
-1.01231799e-01 -7.72439182e-01 -1.41501653e+00 5.93247712e-01
1.07579565e+00 -8.47892225e-01 7.16106594e-01 -5.69314599e-01
6.78968728e-01 -2.28691712e-01 4.24995601e-01 -1.33811545e+00
-4.87109631e-01 -5.82506359e-01 2.80030549e-01 5.88185132e-01
5.84036589e-01 -7.96689391e-01 3.58163267e-01 1.69818662e-02
-2.75997460e-01 -1.10041118e+00 -6.86098218e-01 -2.51069754e-01
-4.53962050e-02 -4.67444956e-01 2.86957383e-01 1.11505270e+00
5.38969133e-03 4.24663574e-01 -1.89465523e-01 -1.99771762e-01
3.93351644e-01 7.89680779e-02 4.17235374e-01 -1.62283111e+00
-3.15967083e-01 -3.11060369e-01 -5.96217155e-01 -1.81473866e-02
-6.13760710e-01 -1.12941051e+00 -1.51942328e-01 -1.75505400e+00
4.93787080e-02 -5.99259496e-01 -8.04232597e-01 8.00058424e-01
-1.76025063e-01 8.52797031e-01 1.33117527e-01 2.69413650e-01
-2.18925774e-01 -3.68073285e-01 9.96308327e-01 -1.01475231e-01
-3.37308705e-01 3.88325155e-01 -6.59995079e-01 6.31727338e-01
1.07748842e+00 -5.65869451e-01 -3.26510191e-01 -1.07348673e-01
3.17799211e-01 3.49951386e-01 4.17155087e-01 -1.55636251e+00
-2.75571585e-01 4.71386075e-01 1.10581517e+00 -6.62232876e-01
7.39976764e-02 -5.47035098e-01 5.29757679e-01 8.82228017e-01
-3.46888334e-01 1.29750162e-01 1.98296160e-02 -1.63279325e-01
-1.65371343e-01 6.41664788e-02 7.06097841e-01 -3.28567065e-02
1.93571657e-01 1.80369452e-01 -5.67775607e-01 8.71839672e-02
8.36305976e-01 -3.25588226e-01 3.10225654e-02 -3.77907693e-01
-1.48479223e+00 -4.36053276e-02 -2.57804632e-01 1.64725482e-01
6.81622207e-01 -1.23540115e+00 -1.26763475e+00 2.44391322e-01
2.35368498e-02 -9.44136977e-02 1.30031943e-01 1.46095133e+00
-1.03547335e+00 2.52995610e-01 -5.11502922e-01 -1.02549911e+00
-1.21214354e+00 -9.01445374e-02 8.57781827e-01 -3.34748924e-01
-8.32576811e-01 4.85248715e-01 -4.59587365e-01 6.99726343e-02
2.76031256e-01 -4.51536179e-01 -4.24880803e-01 3.29637796e-01
4.71181512e-01 4.52711433e-01 4.86085385e-01 -2.96121180e-01
-5.24372458e-01 5.03296256e-01 3.33915889e-01 1.09147623e-01
1.22290218e+00 2.85971940e-01 -2.47522563e-01 7.31278062e-01
9.76361036e-01 -3.27831507e-01 -5.28397620e-01 3.20385635e-01
-1.44560322e-01 4.20831218e-02 -1.58079848e-01 -1.12683105e+00
-1.23953593e+00 1.08288598e+00 1.01560104e+00 7.36209035e-01
1.32051635e+00 -3.56035858e-01 7.19292760e-01 1.42555922e-01
2.72894621e-01 -8.31381798e-01 -3.65354806e-01 3.83418262e-01
7.15177059e-01 -6.08232975e-01 -1.42188549e-01 -4.63182479e-02
-6.54452384e-01 1.49559331e+00 1.81571007e-01 -2.74823368e-01
7.65441597e-01 1.99770182e-01 5.49245000e-01 -3.62099409e-01
-5.68781137e-01 -2.03125421e-02 2.98528731e-01 7.03501940e-01
8.02404761e-01 2.76033700e-01 -6.45062208e-01 8.48912954e-01
-3.63691568e-01 4.02006358e-01 5.45875847e-01 6.71801686e-01
-1.49398297e-01 -7.62656152e-01 -3.01474631e-01 8.41659009e-01
-9.11234617e-01 5.45053855e-02 -4.15657580e-01 4.42568451e-01
7.92686701e-01 1.20659137e+00 2.69841738e-02 -7.06435442e-01
4.52355713e-01 4.26068187e-01 3.59622478e-01 -5.04554749e-01
-1.24882495e+00 7.58638084e-02 2.64112711e-01 -2.56335646e-01
-7.83473253e-02 -5.23314118e-01 -1.26724362e+00 3.80037259e-03
1.68750510e-02 2.67505735e-01 4.91540790e-01 8.11087608e-01
2.62289703e-01 1.14461446e+00 6.47868872e-01 -5.60569167e-01
-4.89755481e-01 -1.34276199e+00 -7.22544551e-01 4.01478767e-01
4.34536994e-01 -2.43379384e-01 -3.25010329e-01 4.63147342e-01]
|
[14.333642959594727, 3.296204090118408]
|
d9624b0c-e9e1-4dd2-af1e-379684808c71
|
beyond-one-model-fits-all-a-survey-of-domain
|
2305.18703
| null |
https://arxiv.org/abs/2305.18703v3
|
https://arxiv.org/pdf/2305.18703v3.pdf
|
Large Language Models, Natural Language Processing, Domain Specialization
|
Large language models (LLMs) have significantly advanced the field of natural language processing (NLP), providing a highly useful, task-agnostic foundation for a wide range of applications. However, directly applying LLMs to solve sophisticated problems in specific domains meets many hurdles, caused by the heterogeneity of domain data, the sophistication of domain knowledge, the uniqueness of domain objectives, and the diversity of the constraints (e.g., various social norms, cultural conformity, religious beliefs, and ethical standards in the domain applications). Domain specification techniques are key to make large language models disruptive in many applications. Specifically, to solve these hurdles, there has been a notable increase in research and practices conducted in recent years on the domain specialization of LLMs. This emerging field of study, with its substantial potential for impact, necessitates a comprehensive and systematic review to better summarize and guide ongoing work in this area. In this article, we present a comprehensive survey on domain specification techniques for large language models, an emerging direction critical for large language model applications. First, we propose a systematic taxonomy that categorizes the LLM domain-specialization techniques based on the accessibility to LLMs and summarizes the framework for all the subcategories as well as their relations and differences to each other. Second, we present an extensive taxonomy of critical application domains that can benefit dramatically from specialized LLMs, discussing their practical significance and open challenges. Last, we offer our insights into the current research status and future trends in this area.
|
['Liang Zhao', 'Jian Pei', 'Quanquan Gu', 'Xuchao Zhang', 'Carl Yang', 'Chris White', 'Haifeng Chen', 'Zhengzhang Chen', 'Yanchi Liu', 'Haoyu Wang', 'Wei Cheng', 'Amit Panalkar', 'Tianjiao Zhao', 'Hejie Cui', 'Yun Li', 'Tanmoy Chowdhury', 'Junxiang Wang', 'Can Zheng', 'Chengyuan Deng', 'Jiaying Lu', 'Xujiang Zhao', 'Chen Ling']
|
2023-05-30
| null | null | null | null |
['chatbot', 'chatbot']
|
['methodology', 'natural-language-processing']
|
[ 9.72543135e-02 1.11708730e-01 -8.24300706e-01 -3.88431340e-01
-4.52453524e-01 -9.57071304e-01 6.60340667e-01 7.40970448e-02
-2.78063655e-01 7.28079498e-01 2.55824089e-01 -3.77042949e-01
-3.85146916e-01 -5.38879275e-01 -1.75914809e-01 -5.52363582e-02
1.15894884e-01 6.22579694e-01 4.42745350e-02 -3.74299198e-01
3.68002057e-01 4.68810946e-01 -1.50803030e+00 3.26862454e-01
1.03477561e+00 6.95420027e-01 4.52928752e-01 4.49554697e-02
-7.43093610e-01 7.13603318e-01 -6.74853861e-01 -2.69240469e-01
1.05773270e-01 -2.11397544e-01 -9.44951653e-01 2.01307923e-01
1.86681710e-02 -3.75106335e-02 7.29227215e-02 9.76054788e-01
1.01917945e-01 1.83394536e-01 7.12378979e-01 -1.50916028e+00
-7.01399982e-01 5.22130191e-01 -3.00429642e-01 -7.32568055e-02
5.76225221e-01 -1.56313986e-01 9.92593527e-01 -6.88428998e-01
7.10741222e-01 1.40445304e+00 4.15275633e-01 6.45923257e-01
-1.08784580e+00 -7.11142778e-01 6.33712769e-01 8.11980739e-02
-1.49654329e+00 -3.66121382e-01 6.42542839e-01 -7.44889319e-01
1.26961088e+00 2.09965155e-01 2.61165529e-01 9.15714622e-01
1.28864899e-01 7.02455878e-01 9.94156897e-01 -7.30567932e-01
3.99427474e-01 5.50031364e-01 3.45628411e-01 1.30548835e-01
4.37484652e-01 -2.36928418e-01 -3.58419061e-01 -5.53066671e-01
6.86509013e-01 -2.25522578e-01 7.94423446e-02 -6.07275009e-01
-9.40838099e-01 8.90927672e-01 -3.17350507e-01 5.85791528e-01
-1.88306451e-01 -5.22428215e-01 4.08141106e-01 3.05496007e-01
3.69755238e-01 6.83059335e-01 -7.68963099e-01 -1.61905438e-01
-9.15875018e-01 5.82673132e-01 1.10434711e+00 1.12605453e+00
7.10577130e-01 3.22170779e-02 3.44609886e-01 1.18542874e+00
4.60196704e-01 2.67952770e-01 5.83225131e-01 -9.40800250e-01
5.38825870e-01 8.46091628e-01 2.82747835e-01 -9.86973226e-01
-4.05096620e-01 -1.97150096e-01 -5.17889798e-01 -2.45591104e-01
2.92782038e-01 -8.62145051e-02 -4.98655587e-01 1.97522724e+00
2.35447556e-01 -2.29373634e-01 2.87687927e-01 5.58000267e-01
6.27857327e-01 7.05162764e-01 5.99771678e-01 -3.91472459e-01
1.42056334e+00 -6.52718723e-01 -6.21920526e-01 -9.74862039e-01
8.15617919e-01 -7.40136266e-01 1.27985346e+00 2.72637606e-01
-8.93286169e-01 -2.43983701e-01 -6.83532774e-01 -1.62565783e-01
-6.25365973e-01 -1.28637835e-01 8.15083981e-01 6.43984079e-01
-8.00895035e-01 -3.97879444e-02 -5.32808304e-01 -8.36656272e-01
-1.45427018e-01 3.54433328e-01 -1.71823949e-01 -2.42804140e-01
-1.56594634e+00 1.11263478e+00 6.29803121e-01 -2.92440891e-01
-2.08491862e-01 -7.99978614e-01 -9.74102855e-01 -6.90981895e-02
5.44362426e-01 -5.24050653e-01 1.42388773e+00 -8.68996739e-01
-1.28905344e+00 9.83593643e-01 -3.33002150e-01 -3.38255346e-01
1.98301435e-01 2.27166433e-02 -8.22168052e-01 -2.52480060e-01
3.37305218e-01 4.44440037e-01 2.43698761e-01 -1.11030591e+00
-9.01797593e-01 -1.56044036e-01 3.16702247e-01 3.67957503e-01
-6.51442766e-01 7.07116187e-01 -4.86144781e-01 -6.39371037e-01
-2.39132330e-01 -6.55625284e-01 -4.66217428e-01 -3.26538146e-01
1.83813095e-01 -2.98902512e-01 6.29762650e-01 -3.08132648e-01
1.92653847e+00 -2.16332269e+00 1.28477588e-01 1.08977668e-01
8.34662765e-02 2.27197170e-01 -2.80188978e-01 9.40611422e-01
1.15815915e-01 4.16772723e-01 -3.03380191e-01 2.98323222e-02
2.27078721e-01 4.24080431e-01 -5.89243829e-01 1.17022738e-01
3.82686779e-02 7.56028831e-01 -8.59907150e-01 -5.50351143e-01
1.57610804e-01 1.18308455e-01 -4.75559890e-01 3.20238322e-02
-6.14323199e-01 1.47369921e-01 -5.33055186e-01 6.64268196e-01
6.26864672e-01 -2.39454985e-01 6.30178630e-01 3.08239192e-01
-3.49225670e-01 4.92517591e-01 -1.41663897e+00 1.45710373e+00
-5.47763348e-01 3.13650191e-01 5.11713684e-01 -9.81007755e-01
9.15302038e-01 4.27122235e-01 6.06532633e-01 -4.96363461e-01
-3.02464962e-01 4.34897840e-01 1.76680274e-02 -5.03605306e-01
5.63703656e-01 -3.35211396e-01 -5.27598023e-01 5.97205698e-01
-1.71479702e-01 -3.78303051e-01 5.13197243e-01 1.78709373e-01
6.58640504e-01 -2.03700364e-01 9.64539766e-01 -4.70814288e-01
6.00738525e-01 3.93225729e-01 9.35857892e-01 6.04899764e-01
-2.88226277e-01 2.20192075e-02 5.14461100e-01 -3.65930974e-01
-7.64766812e-01 -7.98238695e-01 -1.57789126e-01 1.23293376e+00
4.05407287e-02 -6.26824915e-01 -4.49093908e-01 -5.68439007e-01
3.28888893e-02 8.35910738e-01 -1.81421578e-01 1.36352917e-02
-5.81695139e-01 -6.83851719e-01 5.96379936e-01 3.98429811e-01
2.78949827e-01 -9.38171804e-01 -3.65097046e-01 3.17410171e-01
-4.55830455e-01 -1.38301539e+00 -2.35575046e-02 -6.25236630e-02
-9.18095589e-01 -8.47192764e-01 -4.25546974e-01 -1.10759735e+00
4.08881187e-01 3.73823434e-01 1.33014190e+00 -2.71129668e-01
2.30435610e-01 4.82421130e-01 -3.13176304e-01 -6.10711515e-01
-5.36461055e-01 1.92058191e-01 3.21648329e-01 -3.97766113e-01
9.92374778e-01 -3.56961131e-01 5.80698997e-02 4.49075818e-01
-1.13794148e+00 -1.00140944e-01 4.12611693e-01 3.95727396e-01
3.00039828e-01 3.80898863e-01 8.63168299e-01 -1.01833570e+00
1.34887254e+00 -7.73863554e-01 -6.02720857e-01 5.49033403e-01
-6.29777312e-01 -2.88318276e-01 7.22370863e-01 -5.42569816e-01
-1.14045930e+00 -2.77315974e-01 1.17117248e-01 3.32465738e-01
-4.84427810e-01 1.03885186e+00 -4.82461005e-01 6.31640032e-02
8.07700932e-01 1.17035113e-01 -6.62382916e-02 -4.67090696e-01
2.46994272e-01 9.42950070e-01 3.79583091e-02 -1.01927221e+00
5.99830925e-01 1.03263535e-01 -3.88116241e-01 -1.08452106e+00
-6.75066888e-01 -6.34841800e-01 -6.08539939e-01 6.36469722e-02
3.14553231e-01 -8.72986913e-01 -2.64244503e-03 3.14142704e-01
-1.15356445e+00 -3.70960027e-01 7.23095983e-03 4.05642629e-01
-3.40316445e-01 7.16987729e-01 -4.03328180e-01 -8.33899736e-01
-7.61967376e-02 -1.19425011e+00 6.37970090e-01 9.77236405e-02
-8.60995412e-01 -1.37968028e+00 -1.26488134e-01 4.99205440e-01
4.23501760e-01 -3.43582071e-02 1.52183282e+00 -8.59092236e-01
-2.09311724e-01 -1.18101567e-01 -1.65459160e-02 2.97754645e-01
3.37761581e-01 -1.85921580e-01 -5.80770314e-01 -9.52887386e-02
9.04281139e-02 -2.27564976e-01 -4.92630294e-03 3.32746297e-01
7.24375069e-01 -3.72848600e-01 -5.32571852e-01 9.85844955e-02
1.30682945e+00 4.86783326e-01 1.99326292e-01 6.16286516e-01
1.61248744e-01 1.03711629e+00 1.04466033e+00 5.05121291e-01
5.87561071e-01 5.57499230e-01 -2.47416466e-01 4.84555289e-02
2.14237839e-01 -4.15171944e-02 4.29365188e-01 7.83878505e-01
2.38941684e-01 -2.12374136e-01 -1.41228342e+00 6.31160617e-01
-1.91410923e+00 -7.53452420e-01 8.93253759e-02 2.11396813e+00
9.39123988e-01 1.50546255e-02 1.66000023e-01 -1.76174954e-01
6.27271175e-01 1.47779226e-01 -5.54103971e-01 -6.39991403e-01
-2.13342190e-01 -8.92876610e-02 9.24192369e-02 6.44836307e-01
-8.57337475e-01 1.34764242e+00 7.44257450e+00 7.79800177e-01
-1.12918127e+00 -1.59099981e-01 1.02338076e-01 5.27593447e-03
-5.09492874e-01 1.40915558e-01 -1.04801261e+00 2.63177305e-01
8.65018189e-01 -7.36676693e-01 4.99087155e-01 1.03521991e+00
4.71933156e-01 -8.84951837e-03 -1.18861377e+00 8.38006556e-01
5.28148934e-02 -1.27802646e+00 2.21787542e-01 1.87190145e-01
6.06910408e-01 1.24265708e-01 -2.14116387e-02 4.37575161e-01
2.96903074e-01 -7.73989201e-01 5.86688280e-01 -2.30659738e-01
7.03546822e-01 -5.59291840e-01 3.99688840e-01 8.41403842e-01
-1.06945419e+00 -3.23640764e-01 -4.12944049e-01 -6.09495103e-01
2.92180359e-01 5.20180643e-01 -7.26089656e-01 5.32102883e-01
4.89373744e-01 6.29362822e-01 -1.31577998e-02 6.82235003e-01
-1.09767411e-02 4.45832133e-01 -3.98566723e-01 1.41074732e-02
2.32658461e-01 -2.41943017e-01 4.52172816e-01 1.47133470e+00
1.31121762e-02 8.59796703e-02 4.72064316e-01 7.36875534e-01
2.43640631e-01 3.45940441e-01 -8.20727885e-01 -4.30494249e-01
8.50873888e-01 9.17017162e-01 -4.66140836e-01 -3.52062434e-01
-9.68563795e-01 2.65127122e-01 1.42816380e-01 6.23236358e-01
-5.21053970e-01 -3.18512499e-01 1.14984691e+00 3.95287156e-01
-3.28462780e-01 -6.27064884e-01 -6.22139394e-01 -1.42980242e+00
-1.01572936e-02 -1.52497983e+00 6.16491795e-01 -3.67307842e-01
-1.31641066e+00 4.58363771e-01 6.67014062e-01 -1.16515028e+00
-5.58013439e-01 -6.16621733e-01 -9.77638885e-02 9.38079059e-01
-1.53508353e+00 -1.09393048e+00 1.72599822e-01 4.21258658e-01
7.18172729e-01 -3.12597960e-01 1.06458151e+00 3.12467456e-01
-4.69697237e-01 3.09663802e-01 1.09756321e-01 -1.00524150e-01
8.67827356e-01 -6.07118666e-01 3.19790870e-01 8.41551244e-01
-2.22180903e-01 1.32180870e+00 5.90761244e-01 -8.24737728e-01
-1.24592769e+00 -9.66845214e-01 1.54957080e+00 -5.89184165e-01
9.09771979e-01 -7.50317812e-01 -9.36246395e-01 9.94162023e-01
-1.49903828e-02 -6.98145390e-01 1.13499022e+00 3.81969601e-01
-3.63992184e-01 1.72829942e-03 -1.33644855e+00 7.82535255e-01
7.86046922e-01 -5.05821705e-01 -8.55066478e-01 3.02521050e-01
5.48763931e-01 -1.33514434e-01 -8.89838874e-01 3.94659609e-01
4.85654712e-01 -5.43062210e-01 9.25219476e-01 -7.59102166e-01
1.11759201e-01 -2.70857811e-01 -4.06849146e-01 -9.44768250e-01
-5.31047165e-01 -4.18386668e-01 -1.03850383e-02 1.44810188e+00
4.19956416e-01 -8.96677792e-01 6.23569906e-01 1.48772991e+00
4.42725420e-02 -5.86005151e-01 -6.16951346e-01 -8.92027020e-01
6.08930111e-01 -8.07835162e-01 7.25202501e-01 1.18144655e+00
4.42559332e-01 4.04154330e-01 -2.47371972e-01 7.57762641e-02
3.96637172e-01 2.74609715e-01 7.20080912e-01 -1.46253729e+00
-6.06391579e-02 -5.39785922e-01 -1.46868154e-01 -1.31324983e+00
4.87181962e-01 -8.52656305e-01 -4.37473446e-01 -1.84181511e+00
1.26763478e-01 -4.53821361e-01 -9.71321762e-02 5.05891204e-01
2.75605232e-01 -4.15979654e-01 2.30517745e-01 3.24878573e-01
-4.43870038e-01 -6.51430041e-02 7.46142149e-01 -6.46329001e-02
-6.08220994e-01 -7.34173432e-02 -1.34341085e+00 9.68594730e-01
8.67517769e-01 -3.07966650e-01 -7.87371755e-01 -7.36314654e-01
3.97439778e-01 -2.29545549e-01 -2.07635805e-01 -5.83811820e-01
2.98119634e-01 -1.09615302e+00 -1.40033200e-01 -2.15148821e-01
2.11015478e-01 -9.61046278e-01 1.49457529e-02 2.17907503e-01
-3.41412365e-01 7.25815957e-03 4.59361076e-01 2.18376160e-01
-5.89329004e-01 -3.01952958e-01 7.04573572e-01 -3.78103793e-01
-1.41508341e+00 -2.20413860e-02 -8.92245293e-01 3.55561554e-01
1.17643166e+00 -4.35371995e-01 -3.68414037e-02 -3.01396310e-01
-3.29625636e-01 4.92297620e-01 5.86396456e-01 9.08866167e-01
3.29487979e-01 -1.01848269e+00 -4.08346444e-01 2.26942018e-01
3.90553415e-01 -1.30008850e-02 1.34450078e-01 3.49965841e-01
-7.61322677e-02 8.81009579e-01 -1.40734583e-01 -2.34435737e-01
-1.12121427e+00 6.05545759e-01 1.07401066e-01 -4.48113829e-01
-1.35703892e-01 3.45785201e-01 5.97101271e-01 -8.08308601e-01
2.40507990e-01 -4.17919606e-01 -2.74000883e-01 -1.55260218e-02
4.98737872e-01 2.17404768e-01 -1.17356017e-01 -5.66556871e-01
-7.74205863e-01 5.66205204e-01 -1.61769688e-01 -1.12172402e-01
1.27959943e+00 -3.28881294e-01 -3.74550372e-01 4.70296681e-01
7.06834733e-01 7.35143423e-02 -6.07447267e-01 -4.17591780e-01
4.64619935e-01 -2.74547249e-01 -2.50439525e-01 -9.86864567e-01
-3.60608935e-01 6.44910336e-01 -1.40582651e-01 1.93256959e-01
1.10020018e+00 2.08866537e-01 5.57951212e-01 3.05492252e-01
6.31486058e-01 -1.51117992e+00 -3.07549477e-01 1.01773906e+00
9.15688455e-01 -9.83090758e-01 -7.07858801e-02 -5.75489759e-01
-8.60156357e-01 8.68435562e-01 7.78766751e-01 4.54268605e-01
8.55309844e-01 4.81446803e-01 3.66011739e-01 1.67834818e-01
-7.16507912e-01 1.33882999e-01 2.40084603e-02 9.00569379e-01
7.62908757e-01 2.53604781e-02 -7.18418062e-01 1.01643395e+00
-1.47292957e-01 2.38297597e-01 3.07919860e-01 1.12232494e+00
-5.38820267e-01 -1.81102049e+00 -6.07514620e-01 2.63382971e-01
-3.95877272e-01 -1.06428333e-01 -7.50470757e-01 9.79374290e-01
1.12613007e-01 1.27686906e+00 -1.96270689e-01 -1.13012411e-01
4.05315667e-01 1.54622987e-01 9.64204594e-02 -1.04552770e+00
-4.78514135e-01 5.01894429e-02 3.12481433e-01 -1.22267812e-01
-2.99090862e-01 -6.95064664e-01 -1.27516925e+00 -3.22623998e-01
8.74291062e-02 3.59520316e-01 5.19122362e-01 9.29289758e-01
6.00046754e-01 -1.95971742e-01 2.53357217e-02 -2.47830108e-01
-6.40351236e-01 -7.05132127e-01 -6.59556627e-01 2.81471074e-01
9.01853517e-02 -6.79746926e-01 -1.36944473e-01 5.56611083e-02]
|
[10.775446891784668, 8.78807544708252]
|
1ce24a5c-f5b5-4612-894d-2f8b54eaf5dd
|
playing-codenames-with-language-graphs-and
|
2105.05885
| null |
https://arxiv.org/abs/2105.05885v1
|
https://arxiv.org/pdf/2105.05885v1.pdf
|
Playing Codenames with Language Graphs and Word Embeddings
|
Although board games and video games have been studied for decades in artificial intelligence research, challenging word games remain relatively unexplored. Word games are not as constrained as games like chess or poker. Instead, word game strategy is defined by the players' understanding of the way words relate to each other. The word game Codenames provides a unique opportunity to investigate common sense understanding of relationships between words, an important open challenge. We propose an algorithm that can generate Codenames clues from the language graph BabelNet or from any of several embedding methods - word2vec, GloVe, fastText or BERT. We introduce a new scoring function that measures the quality of clues, and we propose a weighting term called DETECT that incorporates dictionary-based word representations and document frequency to improve clue selection. We develop BabelNet-Word Selection Framework (BabelNet-WSF) to improve BabelNet clue quality and overcome the computational barriers that previously prevented leveraging language graphs for Codenames. Extensive experiments with human evaluators demonstrate that our proposed innovations yield state-of-the-art performance, with up to 102.8% improvement in precision@2 in some cases. Overall, this work advances the formal study of word games and approaches for common sense language understanding.
|
['Cynthia Rudin', 'Rachel Lea Draelos', 'Anna Sun', 'Divya Koyyalagunta']
|
2021-05-12
| null | null | null | null |
['board-games']
|
['playing-games']
|
[-2.85519004e-01 -1.12586662e-01 -2.58502245e-01 1.51957020e-01
-7.33396590e-01 -9.51206684e-01 5.70354521e-01 4.27642077e-01
-7.73944318e-01 3.86390418e-01 5.58762193e-01 -6.45523965e-01
-3.04011643e-01 -1.11356878e+00 -2.42569372e-01 -1.92453459e-01
1.26518518e-01 4.78988647e-01 4.77746069e-01 -9.70343411e-01
6.66985750e-01 -3.56925666e-01 -1.36158836e+00 1.45677999e-01
9.50415015e-01 4.55008298e-01 4.07430619e-01 8.42935264e-01
-5.25336623e-01 7.78693497e-01 -5.94073951e-01 -1.04463565e+00
1.41813889e-01 -3.16575468e-01 -7.22251832e-01 -7.04704344e-01
5.65698966e-02 1.31217062e-01 -5.86780965e-01 1.13470531e+00
5.59476316e-01 3.10658485e-01 5.67584813e-01 -1.02535784e+00
-1.05518675e+00 1.14649689e+00 -5.28531909e-01 4.72806990e-01
7.15474546e-01 -2.08073538e-02 1.89988899e+00 -1.01416874e+00
4.87364411e-01 1.06353688e+00 6.79104090e-01 4.81404245e-01
-8.60524356e-01 -8.08415651e-01 1.15840271e-01 5.27894020e-01
-1.53019679e+00 1.94718078e-01 4.47948992e-01 -6.66104198e-01
1.37014914e+00 1.92384541e-01 8.87233257e-01 9.09747422e-01
1.91078469e-01 6.45623922e-01 6.27298415e-01 -6.96407020e-01
-6.76040873e-02 1.98938288e-02 5.90522051e-01 8.66568148e-01
5.58733821e-01 1.83090400e-02 -9.11703527e-01 -2.70520270e-01
6.70704544e-01 -2.16305792e-01 -3.35578114e-01 -1.54006228e-01
-9.72733855e-01 1.39500296e+00 1.16325900e-01 3.69580567e-01
8.82112086e-02 2.52077937e-01 6.91344738e-01 3.69317949e-01
3.74233812e-01 1.06942701e+00 -3.68225247e-01 -5.92055559e-01
-4.75776047e-01 3.40420723e-01 8.68787944e-01 7.19215035e-01
5.56707382e-01 3.08912136e-02 -2.57866830e-01 1.10529077e+00
2.50083834e-01 3.24590534e-01 8.34074378e-01 6.39730170e-02
4.10657018e-01 6.88466728e-01 -3.61869186e-01 -1.61409187e+00
-2.33556494e-01 -5.19035697e-01 -3.48227471e-01 -1.75311968e-01
1.94391638e-01 7.02060238e-02 -6.01818502e-01 1.61496961e+00
-1.67604297e-01 4.98602957e-01 -1.73260476e-02 8.36984515e-01
1.25730598e+00 4.79971945e-01 1.72879487e-01 5.24737775e-01
1.70281315e+00 -1.00239420e+00 -6.40815020e-01 -6.76912785e-01
9.26945210e-01 -7.97862649e-01 1.50878990e+00 5.07715285e-01
-6.32780433e-01 -5.98223507e-01 -1.29816484e+00 -3.99621087e-04
-6.47588015e-01 -1.04349159e-01 9.87953484e-01 9.85116243e-01
-7.97719002e-01 2.77282864e-01 -4.12562579e-01 -4.30664331e-01
-2.41444334e-02 2.18076423e-01 -2.69823313e-01 -8.65686610e-02
-1.82589662e+00 1.12462175e+00 7.33153582e-01 -6.16998136e-01
-5.25887787e-01 -8.85093510e-01 -1.39231205e+00 7.22302794e-02
6.61469460e-01 -6.84991717e-01 7.84982204e-01 -2.43060991e-01
-1.10517669e+00 9.93480325e-01 1.81743145e-01 -3.97997588e-01
-2.26422980e-01 -3.56628060e-01 -7.68703640e-01 -2.37681851e-01
2.62866974e-01 1.96600363e-01 4.01122808e-01 -8.33917558e-01
-6.27830088e-01 3.40952203e-02 5.15058458e-01 4.22025621e-01
-6.78495884e-01 5.99691123e-02 -9.03498173e-01 -9.72748637e-01
-7.87178650e-02 -6.77279115e-01 -3.21995765e-01 -5.37709177e-01
-3.12458068e-01 -3.22267205e-01 -1.82735939e-02 -4.51093256e-01
2.04127383e+00 -1.94757152e+00 -5.93624497e-03 3.77239048e-01
8.48841906e-01 3.27063113e-01 -3.91296595e-01 7.93974221e-01
-9.35791060e-02 3.41527700e-01 2.69236356e-01 1.02268286e-01
2.64687657e-01 2.90903181e-01 -3.10114235e-01 -4.44979221e-03
-1.96639702e-01 1.22093308e+00 -1.29180229e+00 -3.88253517e-02
1.68566272e-01 1.76615030e-01 -8.79536927e-01 -8.78114104e-02
4.83957268e-02 -5.40587664e-01 -4.67499137e-01 2.09821895e-01
2.89680988e-01 -2.76724190e-01 1.86723709e-01 9.31136981e-02
2.94449151e-01 4.98785108e-01 -1.23853743e+00 1.83962655e+00
-5.38640976e-01 6.08724952e-01 -8.07393253e-01 -6.99698985e-01
1.06126022e+00 -8.85780901e-02 -7.22599998e-02 -6.25846386e-01
1.28298074e-01 9.23686549e-02 2.01502278e-01 -5.92143118e-01
1.05119991e+00 -8.37675035e-02 -5.38651586e-01 4.27801818e-01
3.51259768e-01 -3.38995397e-01 4.12906587e-01 7.23284304e-01
1.38201201e+00 -3.38491768e-01 7.71291137e-01 -3.59673530e-01
3.93875062e-01 1.48623720e-01 2.47516647e-01 9.29616928e-01
1.37436330e-01 4.44343686e-01 5.08423984e-01 -3.80210191e-01
-6.59604311e-01 -1.02104783e+00 4.81743485e-01 1.67736530e+00
5.03601909e-01 -1.41207397e+00 -4.25093472e-01 -5.19900441e-01
1.28735080e-01 7.78054655e-01 -7.72493958e-01 -5.21327496e-01
-2.22563028e-01 -6.10919297e-01 9.45963502e-01 6.39244020e-01
-5.66380396e-02 -7.86442995e-01 -3.19812559e-02 2.89903462e-01
-8.77498314e-02 -1.06715202e+00 -8.34399045e-01 4.84356619e-02
-1.23229198e-01 -1.38669848e+00 -2.15019226e-01 -8.05913448e-01
1.44748032e-01 5.07180095e-01 1.82437503e+00 4.72325534e-01
-3.48577917e-01 5.69678426e-01 -9.66726720e-01 -4.09977049e-01
-7.00565428e-02 9.43678319e-02 2.16860279e-01 -4.26977187e-01
1.15668797e+00 -3.51425886e-01 -4.67817128e-01 1.42150491e-01
-9.56863403e-01 -2.39273846e-01 2.76119828e-01 1.05822110e+00
2.55371630e-01 -1.13879256e-01 2.57587373e-01 -9.79112148e-01
1.61347270e+00 -6.28364980e-01 -2.06112564e-01 3.23371679e-01
-6.31023467e-01 8.80452171e-02 4.00383770e-01 -5.42174697e-01
-2.15551093e-01 -5.02545953e-01 -2.48954207e-01 -1.12296306e-01
4.17928487e-01 1.16430426e+00 1.67908296e-01 -3.09614599e-01
1.02673340e+00 3.28632057e-01 -2.84034133e-01 -1.54420957e-01
8.48053098e-01 3.91549855e-01 3.59060049e-01 -7.71379113e-01
8.72618973e-01 4.45409641e-02 -6.89095438e-01 -8.52207541e-01
-6.43857718e-01 -9.97392714e-01 -2.94190586e-01 -1.15842164e-01
8.29626977e-01 -1.01270723e+00 -7.56742060e-01 -1.68626755e-01
-1.21151602e+00 2.77891994e-01 -1.66035578e-01 5.51174760e-01
-1.11875921e-01 4.69656944e-01 -3.25857460e-01 -5.11577845e-01
-3.80363911e-01 -9.85517442e-01 5.96021950e-01 2.27254912e-01
-7.63121426e-01 -1.22037828e+00 4.69539672e-01 4.12651360e-01
2.52060801e-01 -1.83706701e-01 1.08586574e+00 -1.22858298e+00
-2.58770168e-01 -1.35478958e-01 -9.16941613e-02 4.32556272e-02
4.07898007e-03 -3.34654987e-01 -5.53681016e-01 -2.02417597e-01
-5.76402247e-01 -4.02494222e-01 1.07230604e+00 2.03342006e-01
7.56534874e-01 -7.46581051e-03 -2.84973532e-01 4.64677632e-01
1.59471750e+00 2.60828644e-01 6.27357781e-01 6.23480260e-01
7.86823392e-01 1.84830800e-01 5.23758829e-01 6.65354550e-01
6.17916465e-01 5.91651201e-01 2.05256030e-01 1.53775543e-01
-4.81034592e-02 -7.78999388e-01 3.05442363e-01 1.47476411e+00
-1.98385477e-01 -4.41567719e-01 -1.22707164e+00 9.07946885e-01
-1.72827899e+00 -8.21982086e-01 -1.60114557e-01 1.90684390e+00
6.33846939e-01 3.45718980e-01 4.76395562e-02 1.29918858e-01
5.17151713e-01 2.91329175e-01 -1.04113169e-01 -5.09118915e-01
-2.35832855e-01 8.30810189e-01 4.42663938e-01 8.26521218e-01
-7.76129901e-01 1.73668671e+00 5.85352468e+00 1.67527986e+00
-5.00591338e-01 1.88174427e-01 6.87870234e-02 2.58947790e-01
-7.62518466e-01 -7.39037097e-02 -7.22143650e-01 1.32417351e-01
2.56404817e-01 -7.50386417e-01 3.77222776e-01 9.04964685e-01
-2.82449812e-01 1.41471505e-01 -6.94851041e-01 1.32809520e+00
4.92221415e-01 -1.50144744e+00 3.62067312e-01 -2.12354586e-01
5.22336364e-01 -5.03460020e-02 1.51932672e-01 7.01652825e-01
1.04424679e+00 -1.32037640e+00 3.48779649e-01 1.58181980e-01
8.92782152e-01 -9.97840643e-01 6.86676681e-01 -3.14209610e-03
-1.67722082e+00 8.97604302e-02 -7.21605837e-01 -3.57573628e-01
-1.11257263e-01 2.40056351e-01 -9.07127380e-01 5.95624745e-01
4.29319412e-01 7.18237758e-01 -7.07094729e-01 9.35272694e-01
-6.72438800e-01 7.56839156e-01 7.93423504e-02 -7.23063171e-01
5.69915414e-01 -1.48095056e-01 5.60797870e-01 1.28994477e+00
4.29980874e-01 4.28155512e-01 3.61955583e-01 7.03499973e-01
7.20171332e-02 5.82373857e-01 -7.32328415e-01 -4.13469732e-01
7.43675709e-01 1.08466017e+00 -7.08825231e-01 -1.30143315e-01
-5.59641957e-01 8.40198934e-01 3.62948895e-01 1.48636311e-01
-6.84520721e-01 -7.03286231e-01 1.25921309e+00 -1.15125650e-03
2.19071984e-01 -3.40383738e-01 -3.38209152e-01 -1.27575457e+00
-2.71165192e-01 -1.22115469e+00 7.21088350e-01 -6.10282719e-01
-1.50856304e+00 8.75585973e-01 -2.69381791e-01 -1.25110078e+00
-1.50345623e-01 -8.28135967e-01 -7.31142402e-01 6.83942974e-01
-1.08956611e+00 -9.33854640e-01 -5.87536544e-02 6.12354040e-01
8.28450203e-01 -8.64055097e-01 1.03126037e+00 6.45929053e-02
-1.51717722e-01 8.95012975e-01 -2.91015208e-02 4.12096083e-01
5.64115584e-01 -1.34340847e+00 9.46669996e-01 8.38539124e-01
1.06808960e+00 1.05615306e+00 7.27898002e-01 -7.28596032e-01
-1.12852454e+00 -6.29748225e-01 8.70689869e-01 -6.42310977e-01
1.30445242e+00 -5.00144660e-01 -7.62459278e-01 2.43988156e-01
3.72601867e-01 -2.96388268e-01 1.30019748e+00 5.60008526e-01
-7.69790471e-01 3.50827873e-01 -4.15944070e-01 9.94774282e-01
1.18687308e+00 -8.91808689e-01 -9.95699048e-01 7.06697330e-02
1.05312455e+00 -3.67398113e-01 -4.95668381e-01 -3.07620447e-02
3.04247707e-01 -6.25995576e-01 1.26615465e+00 -1.10355246e+00
5.38376987e-01 -2.76084036e-01 -2.67228961e-01 -1.85896838e+00
-6.07938290e-01 -7.67568290e-01 1.09994352e-01 8.87871027e-01
6.56606495e-01 -3.91402185e-01 7.60038078e-01 2.86531031e-01
-8.99131410e-03 -7.00092018e-01 -6.27365291e-01 -8.08586240e-01
1.55764922e-01 -1.08871150e+00 6.17399335e-01 1.31952405e+00
6.61965489e-01 5.63568354e-01 -4.64362621e-01 -4.89778183e-02
1.64290518e-01 -2.89830476e-01 6.19492829e-01 -1.12711632e+00
-5.81158698e-01 -7.24282622e-01 -1.04756582e+00 -1.33087099e+00
2.27140710e-01 -1.22485948e+00 -1.94571644e-01 -1.39303410e+00
3.18248749e-01 -1.50099635e-01 -6.17662549e-01 3.66650581e-01
-7.42311239e-01 4.02929932e-01 1.83974236e-01 -3.96959633e-01
-6.41694307e-01 4.13073123e-01 1.06252670e+00 -3.90576363e-01
1.07131131e-01 -4.05220151e-01 -1.29902685e+00 7.61194587e-01
5.57718873e-01 -5.05653381e-01 -8.58191609e-01 -5.96649945e-01
1.01529455e+00 -3.87117505e-01 -4.82304720e-03 -7.13430941e-01
5.78260243e-01 -3.61626372e-02 -2.60847628e-01 -5.62943444e-02
2.64629185e-01 -2.87491411e-01 -3.80642682e-01 3.23375940e-01
-3.32724661e-01 5.20394683e-01 4.21444893e-01 8.22543859e-01
-4.16624904e-01 -3.80699396e-01 1.53247252e-01 -1.36767551e-01
-1.45146739e+00 1.63693726e-01 -4.68486786e-01 7.06614137e-01
8.24503720e-01 -3.74288768e-01 -3.57076943e-01 -6.46990538e-01
-5.40975869e-01 2.00445443e-01 1.02168908e-02 9.54036534e-01
1.16472054e+00 -1.47925377e+00 -9.06186938e-01 4.60300222e-02
7.00962365e-01 -7.51314938e-01 8.66674706e-02 1.07728124e-01
-6.99683607e-01 3.39884102e-01 6.36931658e-02 -6.76859990e-02
-1.49036694e+00 3.71740788e-01 2.96090413e-02 -5.51021516e-01
-3.91263813e-01 1.58826315e+00 2.26338729e-01 -5.27406335e-01
6.93298504e-02 -2.05743730e-01 -7.80589759e-01 1.91013008e-01
7.72002816e-01 6.63944557e-02 8.88729617e-02 -5.09014606e-01
-3.20022851e-01 7.19637871e-01 -2.13526100e-01 -9.10305530e-02
1.21544564e+00 1.67501539e-01 -8.60755965e-02 2.60224462e-01
7.87825406e-01 4.31004941e-01 -1.54142439e-01 -4.04593408e-01
3.20324481e-01 -7.48302042e-01 4.35701795e-02 -7.83936024e-01
-9.47544456e-01 6.31320000e-01 1.96523890e-01 2.14405745e-01
7.68809915e-01 -3.79492976e-02 8.81798029e-01 4.51015323e-01
5.94158113e-01 -9.83346045e-01 3.59445035e-01 1.17019999e+00
6.43077314e-01 -1.04696524e+00 -1.95815206e-01 -4.92618799e-01
-1.06091022e+00 1.04857945e+00 7.92801917e-01 -2.26517498e-01
6.51894927e-01 2.89200157e-01 1.92986101e-01 -4.87018317e-01
-8.58906448e-01 -7.99806595e-01 8.04370224e-01 6.69482529e-01
5.76093316e-01 1.35435179e-01 -7.38977313e-01 1.39224279e+00
-6.73376143e-01 -5.44008851e-01 4.79297042e-01 3.94572914e-01
-5.65010548e-01 -1.20839107e+00 -8.23195055e-02 4.84655946e-01
-3.18392873e-01 -9.73994851e-01 -6.41763508e-01 8.10894728e-01
2.19092727e-01 1.09766531e+00 -1.91650406e-01 -1.19064486e+00
3.25885236e-01 -9.58098993e-02 1.46320507e-01 -1.10791898e+00
-8.15367341e-01 -3.73815358e-01 2.46654734e-01 -4.49564010e-01
1.13463789e-01 3.36097553e-02 -1.08053434e+00 -5.40991783e-01
-6.67412043e-01 5.56494832e-01 2.19539091e-01 9.36309934e-01
1.18806057e-01 6.23097003e-01 1.78522598e-02 -5.88234477e-02
2.06762343e-03 -8.60391915e-01 -7.38077939e-01 4.46909934e-01
-2.64762670e-01 -8.30138326e-01 -9.69652385e-02 -3.57580990e-01]
|
[10.474672317504883, 8.764177322387695]
|
8edd4d18-311e-41b8-85c7-2e7c58edfd4c
|
audio-retrieval-with-wavtext5k-and-clap
|
2209.14275
| null |
https://arxiv.org/abs/2209.14275v1
|
https://arxiv.org/pdf/2209.14275v1.pdf
|
Audio Retrieval with WavText5K and CLAP Training
|
Audio-Text retrieval takes a natural language query to retrieve relevant audio files in a database. Conversely, Text-Audio retrieval takes an audio file as a query to retrieve relevant natural language descriptions. Most of the literature train retrieval systems with one audio captioning dataset, but evaluating the benefit of training with multiple datasets is underexplored. Moreover, retrieval systems have to learn the alignment between elaborated sentences describing audio content of variable length ranging from a few seconds to several minutes. In this work, we propose a new collection of web audio-text pairs and a new framework for retrieval. First, we provide a new collection of about five thousand web audio-text pairs that we refer to as WavText5K. When used to train our retrieval system, WavText5K improved performance more than other audio captioning datasets. Second, our framework learns to connect language and audio content by using a text encoder, two audio encoders, and a contrastive learning objective. Combining both audio encoders helps to process variable length audio. The two contributions beat state of the art performance for AudioCaps and Clotho on Text-Audio retrieval by a relative 2% and 16%, and Audio-Text retrieval by 6% and 23%.
|
['Huaming Wang', 'Benjamin Elizalde', 'Soham Deshmukh']
|
2022-09-28
| null | null | null | null |
['audio-captioning']
|
['audio']
|
[ 2.70366132e-01 -1.57082126e-01 -1.05055846e-01 -2.07919404e-01
-2.05404043e+00 -7.26513088e-01 5.28124511e-01 3.97423387e-01
-3.99074197e-01 5.06796122e-01 5.24501383e-01 2.60120749e-01
-2.58326322e-01 -3.89261782e-01 -8.92461181e-01 -2.07146540e-01
-2.30864182e-01 6.33480310e-01 3.85177314e-01 -3.25434655e-01
1.87218040e-01 -7.16768727e-02 -1.80636752e+00 7.17448115e-01
1.67099893e-01 1.42998731e+00 1.52356163e-01 1.14480472e+00
-2.91280538e-01 7.30995536e-01 -8.30494940e-01 -2.20176965e-01
6.91741779e-02 -1.67533621e-01 -1.18470764e+00 -5.01364648e-01
8.48276854e-01 -5.87243974e-01 -4.85702574e-01 4.07296568e-01
1.00216973e+00 1.00137502e-01 5.67837417e-01 -1.38535905e+00
-5.13109744e-01 1.18018186e+00 -8.15079287e-02 1.81993902e-01
1.04376900e+00 -1.70213833e-01 1.51531124e+00 -8.03409278e-01
5.15367806e-01 1.21237373e+00 5.17731071e-01 3.84696126e-01
-1.08033299e+00 -9.26416039e-01 -2.82352656e-01 3.44762594e-01
-1.89049363e+00 -8.23164046e-01 7.24915266e-01 -4.02662188e-01
1.06778276e+00 3.57951999e-01 7.68469870e-01 1.21552420e+00
-2.10464727e-02 9.23227489e-01 2.01720908e-01 -4.23641026e-01
-1.51757151e-02 9.82566830e-03 -5.94716966e-02 1.15913726e-01
-4.70397949e-01 -1.93145305e-01 -1.24301684e+00 -3.13118339e-01
2.29947716e-01 -3.68550003e-01 -1.34391189e-01 2.44782850e-01
-1.16689992e+00 7.43819594e-01 -7.43899867e-02 2.21671849e-01
-2.17909813e-01 5.49138248e-01 1.00551236e+00 8.19072485e-01
2.43818954e-01 6.33032799e-01 -2.21959174e-01 -4.53481466e-01
-1.25570774e+00 4.95234132e-01 6.91069126e-01 1.37316215e+00
6.64005935e-01 -3.76565419e-02 -3.58553022e-01 1.06028795e+00
3.34292889e-01 8.89596343e-01 7.34721959e-01 -1.00635302e+00
6.56463623e-01 -5.42680323e-02 -4.33740392e-02 -8.37988973e-01
-9.05042365e-02 8.29850435e-02 -4.15433615e-01 -7.20637619e-01
-4.29413631e-04 8.57593864e-03 -4.11297530e-01 1.46925676e+00
-1.14824280e-01 9.33001265e-02 1.05961695e-01 7.48324037e-01
1.06442428e+00 1.15358090e+00 -1.07470728e-01 6.11355202e-03
1.40616679e+00 -8.66296291e-01 -8.53942990e-01 -4.81877550e-02
5.43862820e-01 -1.21553040e+00 1.31438649e+00 4.26696479e-01
-1.39243770e+00 -5.76387346e-01 -9.23736095e-01 -2.71785378e-01
-3.04244339e-01 -3.52551304e-02 2.80338526e-01 1.81765229e-01
-1.28288627e+00 2.32031360e-01 -4.55304921e-01 -3.64358574e-01
-1.03573695e-01 2.99481541e-01 -3.31657350e-01 1.88687563e-01
-1.60245037e+00 1.11404911e-01 5.30222833e-01 -4.21981573e-01
-1.22817659e+00 -1.01202273e+00 -7.40565419e-01 2.36840442e-01
3.17706198e-01 -6.09736860e-01 1.71524262e+00 -6.22126877e-01
-1.47109985e+00 6.24524176e-01 7.89803788e-02 -6.68321311e-01
9.97146741e-02 -5.97137332e-01 -6.59634233e-01 9.93513942e-01
2.44818375e-01 1.09003603e+00 1.13292754e+00 -7.10283220e-01
-5.71135640e-01 1.62173375e-01 -6.53635040e-02 1.06290251e-01
-6.97363615e-01 4.24311996e-01 -8.08748662e-01 -7.91080475e-01
-3.77236903e-01 -7.86463559e-01 5.34003735e-01 -1.90525018e-02
-3.83459479e-01 -3.47910911e-01 7.37413466e-01 -5.86751640e-01
1.66051626e+00 -2.34054565e+00 8.17751214e-02 -6.39907923e-03
-4.23187315e-02 -5.23444042e-02 -5.08123517e-01 1.10006940e+00
-1.27873076e-02 2.86856294e-01 1.83706552e-01 -2.78229535e-01
3.65990222e-01 -1.81148410e-01 -9.23338413e-01 -1.27649337e-01
2.46137261e-01 7.43299901e-01 -9.23804998e-01 -9.42037940e-01
-6.05749413e-02 7.06882298e-01 -7.30152249e-01 3.19888800e-01
-1.23207405e-01 -1.13493256e-01 -3.41729462e-01 6.59504533e-01
-1.08593151e-01 -5.09892143e-02 -4.34303164e-01 -5.80176748e-02
1.55635804e-01 5.63385606e-01 -1.14765799e+00 2.07897329e+00
-5.95823765e-01 1.18441331e+00 -4.01572958e-02 -5.03457606e-01
9.12344217e-01 1.09631932e+00 8.19496810e-01 -8.11961591e-01
-7.04369619e-02 4.02148813e-01 -6.78432107e-01 -6.95519149e-01
8.77541482e-01 -2.73571201e-02 -4.60074067e-01 6.83046877e-01
4.24412042e-01 -7.05818176e-01 2.94841290e-01 4.86299098e-01
1.18586707e+00 -3.42376828e-01 -2.48955905e-01 2.99761593e-01
3.96346152e-01 -2.34933943e-01 -2.14229837e-01 8.90934527e-01
7.65231550e-02 9.83401477e-01 1.74312830e-01 -4.71116044e-02
-1.06375289e+00 -9.73436058e-01 3.54204625e-02 1.72100055e+00
-1.16161786e-01 -1.11265433e+00 -5.10947645e-01 -1.13285489e-01
-1.41261920e-01 2.86831468e-01 -1.74314797e-01 -4.07653749e-01
-5.49656153e-01 2.00592563e-01 1.26170349e+00 4.41943496e-01
8.26056674e-02 -1.25017321e+00 -2.83539355e-01 2.94434696e-01
-6.82998240e-01 -1.14290464e+00 -9.88504887e-01 1.04391173e-01
-5.32979071e-01 -6.98588431e-01 -9.12960172e-01 -9.16412592e-01
-3.38138998e-01 2.14126468e-01 1.40148425e+00 -1.29123583e-01
-3.23742330e-01 1.00820386e+00 -8.16527605e-01 -5.72486043e-01
-4.03210849e-01 6.51400983e-01 1.30181998e-01 -1.66130707e-01
2.34728217e-01 -6.96607947e-01 -3.34960103e-01 1.23342194e-01
-1.34539986e+00 -2.87899822e-01 5.01387358e-01 6.16195440e-01
5.37049770e-01 -2.61689365e-01 8.35617721e-01 -1.15172952e-01
8.94856930e-01 -4.79635030e-01 -2.51010925e-01 1.22793183e-01
-2.22092539e-01 1.22131266e-01 4.75315064e-01 -7.35192776e-01
-4.52383965e-01 1.09228313e-01 -2.21623018e-01 -7.16006696e-01
8.78269002e-02 6.87901676e-01 3.30074936e-01 4.27419573e-01
6.89141929e-01 2.95686364e-01 -2.51632899e-01 -5.77921450e-01
3.68748248e-01 1.12403202e+00 6.54647470e-01 -8.11279535e-01
8.29308808e-01 7.59844109e-02 -5.53760350e-01 -1.07752776e+00
-6.53345168e-01 -9.67472792e-01 -4.67808425e-01 -3.91593456e-01
6.62159264e-01 -1.13775945e+00 -5.11254907e-01 6.17931634e-02
-1.12431419e+00 -2.32776776e-01 -4.92635697e-01 6.04904413e-01
-8.63046706e-01 1.60901278e-01 -7.25106239e-01 -7.06429780e-01
-8.54771078e-01 -9.24017608e-01 2.02461553e+00 -3.52007926e-01
-4.71295297e-01 -5.31972706e-01 1.90801889e-01 2.71263987e-01
4.12419409e-01 -3.15262049e-01 5.30266583e-01 -7.97258258e-01
-4.73824620e-01 -4.50716943e-01 -4.94924970e-02 7.78504163e-02
-9.83650312e-02 4.73067574e-02 -1.18316329e+00 -3.74192268e-01
-6.66788876e-01 -1.07327557e+00 7.41102993e-01 1.28548115e-01
1.24919474e+00 -4.20033902e-01 -5.70586324e-02 2.17164457e-01
1.07330012e+00 2.27677971e-01 5.12382329e-01 2.10956693e-01
2.38072082e-01 5.21073580e-01 7.14923739e-01 6.12272024e-01
1.64835155e-01 1.04415560e+00 1.35296449e-01 4.24795032e-01
-1.83650419e-01 -5.23368418e-01 6.48133934e-01 1.65511727e+00
4.87756640e-01 -1.96112782e-01 -9.31483269e-01 8.01064551e-01
-1.44875860e+00 -1.05044365e+00 4.87553716e-01 1.91785860e+00
1.32918060e+00 4.03173789e-02 2.97224700e-01 5.01026630e-01
5.06449640e-01 4.31699529e-02 -2.27429658e-01 -1.05682127e-01
1.44462943e-01 3.49781126e-01 6.34277314e-02 5.41183949e-01
-1.01094019e+00 6.88095570e-01 7.07238865e+00 1.11832058e+00
-1.27908742e+00 -8.97484720e-02 -1.18338510e-01 -5.47748625e-01
-2.90677756e-01 -2.42242485e-01 -7.84995914e-01 3.90882164e-01
1.56067812e+00 -5.87525606e-01 4.95695323e-01 6.60919368e-01
1.56253397e-01 3.46040308e-01 -1.51368749e+00 1.35602617e+00
2.30286375e-01 -1.13576543e+00 4.99843597e-01 -3.86140943e-01
2.00215712e-01 3.48481536e-02 7.95881599e-02 6.36664152e-01
-2.63721049e-01 -9.68679547e-01 1.16613972e+00 6.53163612e-01
1.13005996e+00 -5.27908444e-01 6.25642657e-01 -1.05017237e-01
-1.68174231e+00 6.54606670e-02 -1.42908871e-01 2.49759436e-01
2.97117651e-01 1.44564614e-01 -1.07882452e+00 5.74625909e-01
1.03282940e+00 7.07993865e-01 -6.68036342e-01 1.25546074e+00
2.49882743e-01 7.62359262e-01 -4.84058082e-01 -6.47898093e-02
2.11582139e-01 5.09968102e-01 6.54329181e-01 1.45854497e+00
4.54965532e-01 -2.88353860e-01 2.25313574e-01 2.86538154e-01
-2.81655401e-01 4.57069457e-01 -7.23227859e-01 -4.98077571e-01
9.73845541e-01 8.31493855e-01 -2.86800325e-01 -4.36867326e-01
-1.38317391e-01 7.43143559e-01 -1.97742134e-01 2.85681754e-01
-7.41922915e-01 -1.06009710e+00 3.35091949e-01 1.06377058e-01
4.76671383e-02 3.47530730e-02 5.95965445e-01 -8.86451840e-01
2.74543583e-01 -9.35673356e-01 4.97283995e-01 -1.31214976e+00
-1.21775210e+00 5.99038601e-01 3.27758789e-01 -1.57753456e+00
-7.94722855e-01 -1.73670687e-02 6.83135390e-02 5.47221005e-01
-1.49778962e+00 -7.90029764e-01 -2.48379648e-01 6.79550171e-01
8.29071820e-01 -3.28689635e-01 1.07016480e+00 9.95644748e-01
6.68175593e-02 7.41502941e-01 1.62692405e-02 2.61100620e-01
1.39670646e+00 -1.07406592e+00 1.41096637e-01 -3.19621079e-02
4.06578362e-01 5.49782336e-01 6.56186283e-01 -1.29900113e-01
-1.61645460e+00 -1.07548296e+00 1.08281994e+00 -4.50531363e-01
9.27176118e-01 -5.11522472e-01 -8.60713542e-01 6.55677199e-01
4.18439358e-01 -8.95017460e-02 8.61968398e-01 -1.08540855e-01
-6.30883098e-01 -5.16137064e-01 -4.86278832e-01 4.13170546e-01
5.74934840e-01 -1.36639214e+00 -8.27896595e-01 4.17575061e-01
1.59758258e+00 -2.86061406e-01 -1.24557328e+00 -1.16182476e-01
9.10928905e-01 -2.44076267e-01 1.23757088e+00 -4.70386863e-01
4.79025215e-01 -1.31224200e-01 -4.86794621e-01 -9.59550500e-01
2.27287546e-01 -1.06488967e+00 1.73404487e-03 1.56114435e+00
4.19960260e-01 -2.98314150e-02 3.38934243e-01 -1.39065087e-01
-3.69059682e-01 -2.54973352e-01 -9.42257941e-01 -8.54277730e-01
-1.49254665e-01 -9.56244111e-01 5.96415460e-01 6.83540463e-01
2.70007581e-01 6.99114501e-01 -6.07000530e-01 -7.50604272e-02
1.00282423e-01 -1.57072946e-01 8.20890665e-01 -1.34616995e+00
-2.56570935e-01 -3.46940964e-01 -3.87195796e-01 -1.14055479e+00
1.37164012e-01 -9.99388754e-01 2.39748999e-01 -1.19196880e+00
6.17405996e-02 -1.29503384e-01 -9.94672328e-02 5.57871401e-01
3.76817733e-01 5.83689868e-01 2.66423613e-01 2.31540143e-01
-1.09610534e+00 6.21335387e-01 8.58737528e-01 -7.43138492e-01
-4.00076777e-01 -1.91717371e-01 -4.27772135e-01 1.63311645e-01
4.62346226e-01 -6.55087590e-01 -5.74051380e-01 -6.25134051e-01
6.21680021e-01 4.98684824e-01 1.94961458e-01 -1.21921599e+00
4.40129429e-01 2.03878403e-01 -1.59571186e-01 -8.93525362e-01
8.62982512e-01 -8.51546288e-01 5.37077561e-02 -1.34114310e-01
-1.03419971e+00 1.19180769e-01 3.87743145e-01 4.87336397e-01
-8.70855212e-01 -3.96922648e-01 1.85350478e-01 1.25721753e-01
-3.29131544e-01 2.19449610e-01 -7.20931470e-01 4.99463528e-01
3.69777113e-01 1.73493028e-01 2.14170087e-02 -1.02597237e+00
-6.81864560e-01 3.43257338e-01 -2.12063968e-01 6.83549821e-01
8.20477307e-01 -1.69057095e+00 -7.71221459e-01 1.78766400e-02
6.16781592e-01 -1.35032400e-01 -1.10723951e-03 4.60445702e-01
-2.81748712e-01 9.71729934e-01 1.23958969e-02 -8.25460911e-01
-1.42772758e+00 1.60495684e-01 -4.88068834e-02 3.74893621e-02
-5.05109429e-01 7.25949824e-01 -3.77115846e-01 -1.06658347e-01
8.69736850e-01 -4.41299647e-01 -1.54016674e-01 5.76599360e-01
9.01416719e-01 1.40346408e-01 3.12883645e-01 -5.72004259e-01
-7.51218274e-02 6.91657126e-01 -6.39626756e-02 -6.37201369e-01
1.22393298e+00 -2.95785576e-01 1.25886694e-01 9.10540879e-01
1.60723126e+00 4.72206995e-02 -5.70067108e-01 -4.80390966e-01
1.56405821e-01 -1.04808494e-01 5.06314337e-02 -4.41027164e-01
-6.56196117e-01 9.72681463e-01 4.72961068e-01 4.98554617e-01
1.30819809e+00 2.61243910e-01 1.16832888e+00 9.58950579e-01
2.74459213e-01 -1.24544156e+00 6.95308566e-01 7.39357889e-01
1.14224136e+00 -9.66218472e-01 -3.26955408e-01 8.23903177e-03
-3.94980848e-01 1.20142686e+00 1.44476771e-01 1.77569181e-01
5.45545518e-01 1.36109576e-01 1.48816764e-01 -2.49470428e-01
-1.04795980e+00 -1.27547413e-01 5.72143018e-01 4.14929211e-01
5.44432163e-01 -3.09051037e-01 2.39265651e-01 6.79279029e-01
-7.50594735e-01 -6.39544129e-02 2.79101610e-01 9.35137928e-01
-4.62294430e-01 -9.93349969e-01 -4.88726050e-01 3.13476175e-01
-7.09021568e-01 -2.51941711e-01 -5.50047874e-01 6.66958690e-01
-6.51395977e-01 1.07393193e+00 1.88841775e-01 -5.33614039e-01
3.81410182e-01 5.25915742e-01 5.73718734e-02 -7.28874147e-01
-7.92035401e-01 4.47065681e-01 -4.70276689e-03 -6.18902326e-01
-4.66142327e-01 -4.53836650e-01 -1.11805797e+00 4.29201163e-02
-3.38260889e-01 6.93440616e-01 5.65372407e-01 7.05802798e-01
2.92330325e-01 6.94292665e-01 5.88212013e-01 -8.91401947e-01
-3.09015602e-01 -1.11343896e+00 -6.38608515e-01 2.08189055e-01
7.26686537e-01 -2.31252298e-01 -3.45261455e-01 4.32078779e-01]
|
[15.322793960571289, 4.950413703918457]
|
e3ffcf32-8983-4c48-b022-2cbdc97d76bc
|
uot-uwf-partai-at-semeval-2021-task-5-self
|
2104.13164
| null |
https://arxiv.org/abs/2104.13164v1
|
https://arxiv.org/pdf/2104.13164v1.pdf
|
UoT-UWF-PartAI at SemEval-2021 Task 5: Self Attention Based Bi-GRU with Multi-Embedding Representation for Toxicity Highlighter
|
Toxic Spans Detection(TSD) task is defined as highlighting spans that make a text toxic. Many works have been done to classify a given comment or document as toxic or non-toxic. However, none of those proposed models work at the token level. In this paper, we propose a self-attention-based bidirectional gated recurrent unit(BiGRU) with a multi-embedding representation of the tokens. Our proposed model enriches the representation by a combination of GPT-2, GloVe, and RoBERTa embeddings, which led to promising results. Experimental results show that our proposed approach is very effective in detecting span tokens.
|
['Jafar Razmara', 'Mostafa Rahgouy', 'Taher Rahgooy', 'Hamed Babaei Giglou']
|
2021-04-27
| null |
https://aclanthology.org/2021.semeval-1.129
|
https://aclanthology.org/2021.semeval-1.129.pdf
|
semeval-2021
|
['toxic-spans-detection']
|
['natural-language-processing']
|
[ 2.04207506e-02 -1.15324974e-01 -2.38636747e-01 3.38273495e-02
-6.34701610e-01 -2.29803517e-01 6.44812226e-01 5.33145189e-01
-1.95918590e-01 4.93207902e-01 8.95013571e-01 -2.48044223e-01
2.12853968e-01 -7.68276215e-01 -2.21777171e-01 -5.73676229e-01
9.36140418e-02 -3.14068615e-01 2.40927905e-01 -1.91312909e-01
7.79498160e-01 2.46862084e-01 -1.05119658e+00 3.03588986e-01
8.06782067e-01 7.03825355e-01 2.85574757e-02 5.66379428e-01
-2.43788689e-01 1.04852867e+00 -8.62118602e-01 -4.04228210e-01
-9.87518281e-02 -6.10203505e-01 -7.44284391e-01 -1.37021825e-01
9.69037414e-02 -1.11676976e-02 -5.48075199e-01 8.29667628e-01
7.93436527e-01 1.01206966e-01 8.65638137e-01 -1.09119511e+00
-1.13773394e+00 1.13191497e+00 -7.77217388e-01 5.27889371e-01
3.56291533e-01 -8.14924836e-02 1.20928836e+00 -1.02296388e+00
3.40002298e-01 1.38744867e+00 6.62813306e-01 3.03257763e-01
-6.76955819e-01 -5.82208037e-01 3.29256415e-01 4.11013961e-01
-1.36681604e+00 -3.70434858e-02 1.12552619e+00 -3.19269300e-01
1.29042268e+00 9.88892019e-02 6.39378965e-01 1.27838731e+00
4.53418732e-01 8.41605306e-01 9.75965738e-01 -4.87220228e-01
-1.38371602e-01 1.27238572e-01 4.39223945e-01 4.22636509e-01
3.03575754e-01 -3.50594640e-01 -6.12312734e-01 -1.69197559e-01
9.56813097e-02 2.22232491e-01 -2.59790719e-01 3.70844245e-01
-9.09391642e-01 9.78297472e-01 5.06502151e-01 6.36346757e-01
-4.75707740e-01 2.77976155e-01 1.11439145e+00 -9.67528671e-03
8.98075104e-01 2.94768274e-01 6.56537637e-02 -2.45645285e-01
-8.09482336e-01 7.38179684e-02 3.93898338e-01 7.51019716e-01
3.08644205e-01 8.28073248e-02 -9.06696022e-01 7.41299689e-01
3.03031594e-01 1.63851485e-01 7.07478642e-01 1.56221762e-01
6.67021513e-01 9.54968631e-01 6.48658350e-02 -1.22079849e+00
-4.96169180e-01 -4.18200612e-01 -6.09046102e-01 -1.42572850e-01
-3.74324262e-01 -1.89028755e-01 -8.06548238e-01 1.39184642e+00
6.44374564e-02 1.42953649e-01 1.00057989e-01 7.00449109e-01
8.90085280e-01 9.30100441e-01 3.40276420e-01 -1.40622914e-01
1.33538651e+00 -9.91422832e-01 -9.32227671e-01 1.66203171e-01
8.02167892e-01 -8.88618231e-01 1.24038959e+00 2.64707386e-01
-8.69646490e-01 -4.18237954e-01 -1.12643552e+00 -2.55948633e-01
-7.45440722e-01 7.49547407e-03 1.59866244e-01 7.10039020e-01
-7.14434505e-01 5.39094448e-01 -2.94607550e-01 -3.79607052e-01
2.94764102e-01 -1.20300956e-01 -5.59013970e-02 1.55016482e-01
-1.47961557e+00 9.83026385e-01 5.02874017e-01 7.32840300e-02
-8.93284321e-01 -2.54879713e-01 -7.92991042e-01 1.75122097e-01
1.49379313e-01 -3.53174031e-01 9.09078538e-01 -3.50500226e-01
-1.14529634e+00 5.40023565e-01 -2.14157760e-01 -5.32342792e-01
2.29131252e-01 -3.80206406e-01 -5.65969408e-01 9.53549072e-02
-1.09259650e-01 2.14151666e-01 6.58414900e-01 -8.69957387e-01
-3.17665339e-01 -4.24984813e-01 -4.74364534e-02 3.29118520e-01
-1.05439222e+00 4.60628450e-01 -2.40293160e-01 -1.07631540e+00
-3.98238420e-01 -5.46461523e-01 1.65652279e-02 -6.75892115e-01
-8.74737680e-01 -9.17405725e-01 1.00435042e+00 -8.02080512e-01
2.04797125e+00 -2.09061623e+00 -2.36382969e-02 -8.18909239e-03
8.19354281e-02 3.70446354e-01 -1.07389972e-01 1.11194503e+00
9.04236883e-02 6.41573131e-01 -4.61839847e-02 -4.89145935e-01
1.91386864e-01 -3.95366997e-01 -4.80833679e-01 4.92687464e-01
1.24229670e-01 7.84362674e-01 -7.95532942e-01 -6.27493918e-01
-4.45318036e-02 5.98121047e-01 -2.55982801e-02 1.66277528e-01
-9.71976444e-02 -6.73826784e-02 -5.34254253e-01 5.32054126e-01
6.16358399e-01 2.23400787e-01 -2.78316945e-01 2.60738786e-02
-2.92256802e-01 5.18094063e-01 -6.75155640e-01 1.43777132e+00
-5.06075621e-01 5.46226501e-01 -7.07775950e-01 -5.36700606e-01
1.21888661e+00 4.03943241e-01 7.21665993e-02 -5.02962232e-01
3.62480164e-01 -2.88750175e-02 -2.30585515e-01 -7.39138424e-01
9.73784208e-01 -9.42725912e-02 -2.10898221e-01 5.96887946e-01
-2.42887795e-01 3.54295045e-01 3.72650564e-01 3.35837096e-01
1.42093742e+00 -9.58799571e-02 1.75979346e-01 -9.96474177e-04
5.84827781e-01 -3.21065485e-01 5.38577199e-01 4.18610662e-01
-1.52352810e-01 6.93960607e-01 7.81096697e-01 -1.24188326e-01
-1.08307147e+00 -4.34839845e-01 8.87670889e-02 1.05302739e+00
6.10252917e-02 -1.00373399e+00 -7.31861353e-01 -8.86959195e-01
8.10072571e-02 8.26888204e-01 -9.96700823e-01 -3.64410162e-01
-3.94465625e-01 -4.83984530e-01 9.50178862e-01 7.17442274e-01
4.30893004e-01 -1.41451645e+00 -6.59308136e-01 3.64788830e-01
-4.20052372e-02 -6.52102232e-01 -8.57339084e-01 1.75593495e-01
-7.27452159e-01 -8.87851536e-01 -8.55170608e-01 -8.06532443e-01
4.31122661e-01 3.86981010e-01 5.77493608e-01 2.57335939e-02
-2.04665899e-01 -1.56966418e-01 -9.66754198e-01 -4.66488808e-01
-4.22580261e-03 1.33551300e-01 -1.99266866e-01 1.29874200e-01
4.91867036e-01 -3.80895495e-01 -5.44173241e-01 8.41977596e-02
-9.96377826e-01 -2.87921399e-01 6.25803530e-01 5.88679254e-01
2.42866844e-01 -1.84840262e-01 9.93732274e-01 -1.07790363e+00
1.47841883e+00 -7.55667925e-01 1.35835201e-01 2.20668897e-01
-7.17995286e-01 -2.69247368e-02 8.89585555e-01 -3.88271987e-01
-9.62554336e-01 -4.92596507e-01 -2.40668058e-01 -5.61083734e-01
1.10448949e-01 6.77563071e-01 -1.08055949e-01 4.09019589e-01
4.88211602e-01 2.17104882e-01 -5.79658985e-01 -7.42577195e-01
3.67376804e-01 1.21610487e+00 1.32008851e-01 -1.34460390e-01
5.11712015e-01 5.71193919e-02 -2.90744692e-01 -5.30440927e-01
-6.54564440e-01 -7.16112852e-01 -3.78992617e-01 -1.39406025e-01
8.10759664e-01 -6.95621192e-01 -4.88136858e-01 3.68487418e-01
-1.22750711e+00 1.78027228e-01 -1.21007130e-01 3.65043998e-01
-8.20097178e-02 6.62745714e-01 -6.83678627e-01 -1.08890808e+00
-9.98723626e-01 -7.96977282e-01 7.80071795e-01 2.37935275e-01
-3.35017592e-01 -8.96292210e-01 4.15729642e-01 9.56924930e-02
4.63682473e-01 4.95436907e-01 9.65902328e-01 -7.54778862e-01
1.51685536e-01 -4.46266383e-01 -3.18718910e-01 2.62617737e-01
1.04941279e-01 6.07893914e-02 -9.73495781e-01 -1.75137371e-01
-2.49114871e-01 -5.16966701e-01 1.29905653e+00 -3.96952033e-04
1.25491762e+00 -5.57236910e-01 -3.63001645e-01 1.94047093e-01
1.35919905e+00 1.84906721e-01 6.36077225e-01 3.39442074e-01
7.78617978e-01 2.35575378e-01 7.12229967e-01 8.94102097e-01
1.32139698e-01 3.52710783e-01 6.67769790e-01 1.04242057e-01
-1.64273053e-01 -5.84480405e-01 7.90299654e-01 7.94481754e-01
6.17618486e-02 -8.25498700e-01 -7.87610173e-01 8.65214229e-01
-1.85591936e+00 -1.08169210e+00 -3.60808402e-01 1.66655576e+00
7.06063032e-01 4.48037177e-01 2.58792728e-01 5.29271245e-01
8.76324415e-01 6.36211038e-01 -4.76168245e-01 -1.16611600e+00
1.64549053e-02 -1.16689324e-01 3.25394690e-01 1.28349420e-02
-9.70667124e-01 8.89147758e-01 6.08968163e+00 1.03127897e+00
-1.03690362e+00 2.32298806e-01 3.78942728e-01 8.53055269e-02
-6.57856226e-01 -2.55986094e-01 -7.95321345e-01 7.25686014e-01
1.04220653e+00 -3.79915088e-01 -2.10480124e-01 7.88536251e-01
5.59750497e-01 7.41031468e-02 -7.47169375e-01 7.51310468e-01
6.40893936e-01 -9.41426158e-01 3.67137611e-01 -7.92091489e-02
4.47635531e-01 -3.26760232e-01 2.55286813e-01 3.46584052e-01
1.03889182e-01 -9.95417833e-01 7.59162188e-01 4.26581085e-01
6.17913604e-01 -1.07293773e+00 9.78697181e-01 1.61138713e-01
-1.17432559e+00 -2.25887969e-01 -5.65374017e-01 -1.44809820e-02
2.54926253e-02 7.33025432e-01 -1.03126180e+00 5.55088401e-01
4.68068451e-01 1.12380254e+00 -7.21136808e-01 1.25806737e+00
-5.13884723e-01 8.92381310e-01 1.47817761e-01 -6.87187254e-01
4.87826347e-01 4.56080884e-02 5.55018067e-01 1.63752317e+00
3.33649904e-01 -2.09843904e-01 -6.18837364e-02 7.87907958e-01
-4.24728781e-01 6.04798555e-01 -7.96412051e-01 -2.30873764e-01
4.32915509e-01 1.38765955e+00 -6.78804278e-01 -3.85728061e-01
-1.81423753e-01 9.76110816e-01 3.96962345e-01 7.94921443e-02
-9.23959076e-01 -1.01406145e+00 2.50426292e-01 -5.23585752e-02
5.29767632e-01 5.09715006e-02 -3.80480677e-01 -8.76997828e-01
9.64892954e-02 -3.47675741e-01 4.47611213e-01 -7.41145730e-01
-1.28183246e+00 6.96482301e-01 -1.89295277e-01 -1.28210926e+00
2.03826904e-01 -1.16468236e-01 -1.27057946e+00 8.44022393e-01
-1.46380401e+00 -1.32135868e+00 -1.84217185e-01 3.87798667e-01
7.83618569e-01 1.11487024e-01 6.02033675e-01 2.04115421e-01
-9.48684096e-01 7.34023809e-01 3.22125629e-02 2.16873422e-01
7.88362026e-01 -1.18688011e+00 3.21989685e-01 9.72790837e-01
-2.34503597e-02 7.80849338e-01 7.08783865e-01 -8.19625497e-01
-9.93596673e-01 -1.36815095e+00 1.20422339e+00 -1.37991846e-01
6.21091723e-01 -2.16345817e-01 -8.85764837e-01 5.38293004e-01
7.91039467e-01 -2.42226973e-01 8.00883412e-01 1.33502364e-01
-4.70897615e-01 9.49642360e-02 -8.95908713e-01 3.86518836e-01
8.93411756e-01 -4.22048479e-01 -8.32012296e-01 3.11839163e-01
8.56380224e-01 1.72116160e-02 -6.89411759e-01 1.07619174e-01
1.20172851e-01 -5.82795441e-01 4.14725363e-01 -6.54021502e-01
6.81668878e-01 -1.94074363e-01 2.34753340e-01 -1.41465783e+00
-5.11833131e-01 -3.89693141e-01 -3.28236997e-01 1.80990505e+00
3.13111126e-01 -2.62542963e-01 2.86857873e-01 7.45953918e-02
-6.38462126e-01 -1.03016627e+00 -7.30387568e-01 -7.63434291e-01
9.06161666e-02 -2.25706711e-01 6.00045860e-01 7.88619757e-01
5.75368047e-01 5.58809280e-01 -6.28289223e-01 -1.61833018e-01
1.17331699e-01 8.10723156e-02 3.10267031e-01 -8.23518395e-01
3.03490460e-01 -6.30380273e-01 -4.06110287e-01 -7.16550350e-01
1.02621466e-01 -1.00489247e+00 -8.71373191e-02 -1.79354191e+00
5.21260381e-01 -8.36307108e-02 -7.86338449e-01 5.85313082e-01
-4.79806155e-01 2.98693657e-01 1.44742802e-01 7.47718364e-02
-8.41622055e-01 7.46676683e-01 8.11280012e-01 -3.36479604e-01
-1.25695646e-01 -2.89603710e-01 -8.22592020e-01 4.22913313e-01
1.17200196e+00 -6.06107533e-01 -2.08743304e-01 -2.65944302e-01
2.88775206e-01 -4.51049209e-01 -1.49060592e-01 -8.79832447e-01
1.13377512e-01 2.11014915e-02 2.08467126e-01 -1.02878475e+00
8.70947316e-02 -3.94631147e-01 -3.61523688e-01 4.69985098e-01
-6.14047647e-01 3.69231284e-01 7.94517249e-02 6.63433909e-01
-2.72804618e-01 -5.39805114e-01 4.55862105e-01 3.85480225e-02
-4.96107399e-01 1.06634922e-01 -4.99613374e-01 1.20471045e-02
1.18258095e+00 -5.34750819e-02 -3.98503423e-01 -1.62192822e-01
-2.59550214e-01 2.61653811e-01 3.01061898e-01 6.05697393e-01
9.75691557e-01 -1.32453418e+00 -8.56133223e-01 -3.75714004e-01
4.32830304e-01 -4.01319623e-01 1.46442860e-01 6.73152268e-01
-3.20610106e-01 4.91123736e-01 -1.33781791e-01 6.40520006e-02
-1.39225554e+00 8.37995589e-01 -1.35476157e-01 -5.92834532e-01
-7.83553004e-01 6.83881998e-01 -3.40523273e-01 1.60905749e-01
3.54539156e-01 -3.07264864e-01 -8.80395532e-01 5.81812918e-01
6.40743494e-01 6.27527595e-01 1.07111521e-02 -7.15355635e-01
-3.66941631e-01 6.48135245e-01 -2.88183987e-01 1.04140773e-01
1.25379586e+00 -2.80186310e-02 3.58114019e-02 7.09527671e-01
1.30722713e+00 2.20069826e-01 -7.07064986e-01 -1.58320844e-01
1.08230777e-01 -3.73257369e-01 1.08682327e-01 -6.01535439e-01
-8.54950190e-01 1.27198362e+00 3.14596772e-01 3.83149385e-01
9.11620557e-01 -2.64586717e-01 1.16845739e+00 -5.99938817e-02
-1.00216985e-01 -1.17605948e+00 3.53358150e-01 5.90384007e-01
1.08698046e+00 -8.98277998e-01 -5.46648577e-02 9.91310365e-03
-7.48584807e-01 1.17956245e+00 6.74265027e-01 -2.06151068e-01
3.78247023e-01 -1.49429351e-01 -1.87790602e-01 -2.30032027e-01
-8.53466094e-01 -2.90196121e-01 1.85669735e-02 2.18601093e-01
6.17229581e-01 -2.00005412e-01 -1.11474204e+00 6.46451354e-01
9.10833776e-02 -1.14375949e-01 7.43276238e-01 1.01587617e+00
-6.66753054e-01 -8.79677296e-01 -2.53110051e-01 5.05778193e-01
-6.09036684e-01 -1.63365036e-01 -6.62302196e-01 4.33248222e-01
1.78404942e-01 1.11033821e+00 -1.24174654e-01 -8.21291327e-01
2.63653129e-01 3.32990885e-01 -1.08198822e-01 -7.97600985e-01
-1.09381557e+00 -1.46429427e-03 3.98939103e-02 -7.25467950e-02
-2.40412682e-01 -4.12054062e-01 -1.34882736e+00 -2.54262984e-01
-5.66789389e-01 5.08551359e-01 5.68226457e-01 5.09369969e-01
4.56327260e-01 7.57125795e-01 9.84464705e-01 -4.64300096e-01
-5.09468496e-01 -1.62597585e+00 -7.34057188e-01 5.42795658e-01
1.62292659e-01 -4.97453243e-01 -4.18770641e-01 -2.11449981e-01]
|
[8.966055870056152, 10.574906349182129]
|
f99dde8b-6699-444c-803b-2d41124928f6
|
explainable-hierarchical-imitation-learning
|
2105.07348
| null |
https://arxiv.org/abs/2105.07348v1
|
https://arxiv.org/pdf/2105.07348v1.pdf
|
Explainable Hierarchical Imitation Learning for Robotic Drink Pouring
|
To accurately pour drinks into various containers is an essential skill for service robots. However, drink pouring is a dynamic process and difficult to model. Traditional deep imitation learning techniques for implementing autonomous robotic pouring have an inherent black-box effect and require a large amount of demonstration data for model training. To address these issues, an Explainable Hierarchical Imitation Learning (EHIL) method is proposed in this paper such that a robot can learn high-level general knowledge and execute low-level actions across multiple drink pouring scenarios. Moreover, with EHIL, a logical graph can be constructed for task execution, through which the decision-making process for action generation can be made explainable to users and the causes of failure can be traced out. Based on the logical graph, the framework is manipulable to achieve different targets while the adaptability to unseen scenarios can be achieved in an explainable manner. A series of experiments have been conducted to verify the effectiveness of the proposed method. Results indicate that EHIL outperforms the traditional behavior cloning method in terms of success rate, adaptability, manipulability and explainability.
|
['Zhengyou Zhang', 'Dongsheng Zhang', 'Lei Wei', 'Qiang Li', 'Yu Zheng', 'Dandan Zhang']
|
2021-05-16
| null | null | null | null |
['action-generation']
|
['computer-vision']
|
[-2.78962016e-01 2.57480294e-01 -9.65511203e-02 -3.67163241e-01
1.12692453e-01 -3.94716620e-01 2.88078129e-01 -9.84471068e-02
1.15212716e-01 5.33161938e-01 -2.53934622e-01 -3.04597206e-02
-2.57475019e-01 -6.41676009e-01 -8.88817668e-01 -5.58393776e-01
-1.42456487e-01 4.85022724e-01 7.02682212e-02 -3.23306441e-01
2.88169205e-01 4.88252610e-01 -1.54427159e+00 3.68241109e-02
1.13399243e+00 4.83832598e-01 9.74101961e-01 5.77224672e-01
1.44176617e-01 9.11905348e-01 -3.89170885e-01 1.89085379e-01
1.55973732e-01 -4.56862301e-01 -5.44409037e-01 3.60502690e-01
-3.73099029e-01 -8.15170884e-01 -5.03542066e-01 8.49056363e-01
3.92657444e-02 1.78230181e-01 6.58463001e-01 -1.85426247e+00
-7.89393067e-01 8.02172482e-01 5.33929020e-02 -6.34408176e-01
5.10009646e-01 4.27161038e-01 6.46162510e-01 -4.58911389e-01
3.81043941e-01 1.49236429e+00 2.11215943e-01 7.59984374e-01
-9.15032208e-01 -6.56507134e-01 3.36900473e-01 5.40832162e-01
-1.06849098e+00 2.55364388e-01 5.86584806e-01 -2.87778586e-01
8.35216522e-01 9.89345536e-02 9.11829770e-01 8.87649715e-01
6.26475215e-01 1.07081616e+00 9.02548671e-01 -1.50955677e-01
4.47191894e-01 1.80493996e-01 -1.88014284e-01 8.08033884e-01
3.65691662e-01 1.30337268e-01 -1.43811852e-01 3.58326048e-01
1.19788027e+00 4.70292598e-01 -1.87103137e-01 -6.30567729e-01
-1.12355256e+00 5.54949343e-01 8.47807169e-01 3.24394703e-01
-5.37587047e-01 4.38382417e-01 2.11874992e-01 2.86844820e-01
-3.99846584e-01 7.72457659e-01 -4.31812137e-01 -2.82862157e-01
-1.20039731e-01 4.28892732e-01 1.02803326e+00 1.34540999e+00
6.75759852e-01 2.63997614e-01 6.94852918e-02 3.87214363e-01
4.75670725e-01 5.11014581e-01 5.37654161e-01 -1.01791930e+00
2.69518048e-01 1.02518559e+00 4.01083112e-01 -9.85341012e-01
-6.48013830e-01 3.04235309e-01 -5.77858269e-01 4.26384449e-01
-1.03010029e-01 -2.01329246e-01 -9.50222552e-01 1.46065259e+00
3.48337561e-01 1.43398475e-02 2.28112429e-01 1.07700288e+00
6.19436741e-01 8.95894289e-01 2.10996792e-01 1.21871710e-01
1.00149918e+00 -1.15920818e+00 -8.89167011e-01 -2.14408740e-01
5.67569315e-01 -1.93949923e-01 1.21416712e+00 3.14513147e-01
-6.45242035e-01 -7.11514056e-01 -1.25385153e+00 5.91755696e-02
-4.90094572e-02 2.57904589e-01 9.10846591e-01 1.89321429e-01
-6.91675842e-01 8.17348301e-01 -1.10006166e+00 -5.14400482e-01
-3.54554737e-03 6.86868370e-01 -3.20011556e-01 -1.72638282e-01
-9.51212108e-01 1.08027411e+00 7.45016277e-01 3.52572411e-01
-1.44307530e+00 -1.10061079e-01 -9.88576055e-01 1.58869356e-01
4.06427830e-01 -3.02236646e-01 1.38615680e+00 -7.12017119e-01
-1.79305470e+00 6.50318936e-02 1.87738448e-01 -9.71528813e-02
4.55309182e-01 -1.77181467e-01 4.31054309e-02 -1.74482947e-03
1.06233366e-01 6.45258963e-01 6.80865705e-01 -1.40169072e+00
-6.27915621e-01 -2.40431845e-01 4.06036913e-01 5.79062760e-01
-1.04535989e-01 -4.27735746e-01 -2.84252197e-01 -1.79731920e-01
2.97862232e-01 -1.39411271e+00 -2.23867238e-01 1.49813011e-01
-3.30767363e-01 -4.58845466e-01 9.00090694e-01 -5.71932018e-01
6.69749081e-01 -2.06493783e+00 7.01434910e-01 -2.10213605e-02
8.56115818e-02 -2.85492726e-02 -1.01065099e-01 6.96933568e-01
3.43047112e-01 -6.95896521e-02 -1.46891205e-02 2.71631815e-02
3.60960901e-01 6.90263331e-01 1.18077658e-01 1.41646028e-01
1.88087299e-03 8.42912674e-01 -1.01384377e+00 -2.01991007e-01
5.62476754e-01 5.99198006e-02 -4.89156097e-01 7.59562373e-01
-3.90257478e-01 8.03736031e-01 -9.82694626e-01 7.07931340e-01
3.27098757e-01 8.84288631e-04 4.20878053e-01 1.35227978e-01
-2.86274254e-02 -1.90562174e-01 -1.06914997e+00 1.44382989e+00
-6.75725281e-01 3.27330798e-01 1.47521868e-01 -7.84333587e-01
1.15334427e+00 2.98708320e-01 2.80794650e-01 -4.23973382e-01
3.11432868e-01 2.48019710e-01 2.43136600e-01 -1.12470651e+00
4.36904371e-01 -2.68078386e-03 -2.44916379e-01 4.21757609e-01
-2.52506614e-01 -6.92266762e-01 -8.18106383e-02 -4.68693972e-02
8.31463158e-01 6.55997813e-01 1.51549906e-01 -4.80339043e-02
2.90123492e-01 2.30299532e-01 4.00447816e-01 7.39067018e-01
-3.08210105e-01 -7.15462714e-02 2.40749314e-01 -4.58947182e-01
-1.20770514e+00 -7.35354424e-01 4.15112883e-01 8.90461206e-01
8.67554665e-01 1.60840936e-02 -8.42992902e-01 -3.93843353e-01
1.34508491e-01 1.11133993e+00 -4.91377443e-01 -5.57736874e-01
-5.68954885e-01 -4.76200804e-02 9.13830400e-02 6.08616650e-01
7.65376329e-01 -1.75407624e+00 -8.65696311e-01 3.66617590e-01
-2.02438667e-01 -1.03414440e+00 -1.38259009e-01 -4.29799967e-02
-8.79716456e-01 -1.12587559e+00 -3.64073873e-01 -1.22696924e+00
9.73258436e-01 4.96278733e-01 3.04443836e-01 4.26780760e-01
-2.33455777e-01 4.16405231e-01 -6.24305367e-01 -4.66021925e-01
-8.03573608e-01 -1.82356209e-01 3.53396982e-01 -2.30749324e-01
1.57251433e-01 -4.88770664e-01 -6.32664025e-01 6.36093438e-01
-8.25992346e-01 4.49283332e-01 7.48118520e-01 7.65647173e-01
3.60800922e-01 3.84096146e-01 7.02040732e-01 -1.57345399e-01
1.03713083e+00 -3.75258327e-01 -5.34822524e-01 3.45207393e-01
-5.07254362e-01 6.04511388e-02 8.65751922e-01 -7.16351688e-01
-1.11808729e+00 1.51651517e-01 1.06361791e-01 -2.82458633e-01
-9.37506557e-02 2.66584247e-01 -1.56768441e-01 1.10994196e-02
3.55432302e-01 5.99637985e-01 3.51814777e-01 -3.21940750e-01
4.71844733e-01 8.57647419e-01 2.30108216e-01 -6.66810572e-01
6.86720550e-01 5.31152189e-02 -8.05566385e-02 -4.16004121e-01
-1.62693530e-01 5.61659224e-02 -6.77176595e-01 -5.14999807e-01
8.91320884e-01 -6.45901501e-01 -1.41654682e+00 6.08343124e-01
-1.09244776e+00 -6.56205952e-01 1.61157131e-01 5.94258130e-01
-8.51239920e-01 1.64249063e-01 -9.14092600e-01 -1.01522887e+00
-2.10747570e-02 -1.55439317e+00 8.36969495e-01 5.22701740e-01
-2.63845831e-01 -5.31643569e-01 -6.38789415e-01 1.37180254e-01
3.57686311e-01 2.71958470e-01 1.11025417e+00 -4.76402074e-01
-9.33512449e-01 -3.81295323e-01 -8.57385546e-02 1.42895281e-01
3.35439086e-01 -1.06368400e-01 -1.39297336e-01 -3.71690005e-01
1.69112146e-01 -6.10537946e-01 -8.49902034e-02 1.83650777e-01
1.09096360e+00 -4.87893909e-01 -5.08324444e-01 6.49559032e-03
1.13723898e+00 7.95112669e-01 6.99253082e-01 6.67693436e-01
7.92346895e-01 6.38849080e-01 1.13842428e+00 3.49574655e-01
7.22566366e-01 6.08123004e-01 9.11965013e-01 2.51526147e-01
2.49011680e-01 -6.06304228e-01 3.93659115e-01 6.74409449e-01
-1.81145608e-01 2.71977317e-02 -4.37566072e-01 2.75583357e-01
-2.21467948e+00 -8.12305510e-01 -1.10381834e-01 1.82104278e+00
4.17786598e-01 -8.24215636e-02 -1.78410590e-01 2.45716609e-02
7.15785921e-01 -4.92744744e-01 -1.00715232e+00 -6.37170196e-01
6.98012114e-01 -5.71963310e-01 1.81367740e-01 3.21524441e-01
-5.46536803e-01 1.03050554e+00 6.02672529e+00 3.93638104e-01
-1.05975032e+00 -2.75189817e-01 2.19467536e-01 3.38022947e-01
-2.31458247e-03 -4.07082140e-02 -4.69127446e-01 4.64475274e-01
3.18927228e-01 -1.57235965e-01 8.42388749e-01 1.32442629e+00
6.17727697e-01 -1.64243922e-01 -1.38931704e+00 7.35329330e-01
-2.26995319e-01 -7.80664146e-01 -1.49823073e-02 -3.66222948e-01
4.22487795e-01 -4.23222214e-01 -1.23087958e-01 8.56370986e-01
3.07778031e-01 -9.84850705e-01 8.37205350e-01 3.61571521e-01
3.05322170e-01 -6.02567077e-01 4.64116216e-01 9.32207048e-01
-1.04063630e+00 -6.24555349e-01 -5.77057004e-01 -4.18776572e-01
2.18921706e-01 -4.40254718e-01 -1.17234635e+00 4.78438109e-01
4.46978778e-01 5.25168478e-01 -3.90506275e-02 9.37769711e-01
-8.41121495e-01 1.86611608e-01 -1.78555086e-01 -9.99614775e-01
3.97304446e-01 -4.01134044e-01 5.15657902e-01 5.44426382e-01
4.97248411e-01 2.74255067e-01 3.08008879e-01 1.05599833e+00
2.05530614e-01 -2.91141924e-02 -6.41257048e-01 -4.30815527e-03
4.38956529e-01 1.07007039e+00 -5.41463375e-01 -2.85399675e-01
5.48169725e-02 1.21131408e+00 3.47091854e-01 2.81131297e-01
-1.10922301e+00 -4.26070333e-01 2.98007011e-01 -3.13718796e-01
2.22545207e-01 -5.46966136e-01 -4.37075235e-02 -9.82514620e-01
1.68853924e-02 -9.20274973e-01 -2.35775545e-01 -1.20623505e+00
-8.05117488e-01 3.93653691e-01 7.67415613e-02 -1.29775763e+00
-2.18756840e-01 -5.37042439e-01 -5.62019646e-01 5.21626949e-01
-1.23701954e+00 -1.06818104e+00 -6.51914239e-01 4.51349020e-01
9.41423416e-01 -1.20312728e-01 8.09728265e-01 -1.42318577e-01
-5.89797437e-01 2.18552396e-01 -9.92201921e-03 -2.82743901e-01
8.53100151e-04 -1.04274595e+00 2.35098358e-02 3.64094645e-01
-3.26498181e-01 5.49405515e-01 9.72778261e-01 -7.37807751e-01
-1.91191912e+00 -9.98876870e-01 3.33035648e-01 -1.87073320e-01
5.71811736e-01 -3.16851467e-01 -9.46001112e-01 8.13075304e-01
-1.23359688e-01 -5.72095096e-01 1.15265526e-01 -1.40230402e-01
1.85458362e-01 -1.38657028e-02 -1.25747919e+00 1.00869417e+00
8.11676085e-01 -6.54673064e-03 -8.19347680e-01 6.74952626e-01
9.70698178e-01 -5.64944804e-01 -7.61150360e-01 2.03208342e-01
6.47321939e-01 -6.32723689e-01 4.53547597e-01 -6.73589587e-01
6.63076878e-01 -4.26161200e-01 1.70254439e-01 -1.36187255e+00
-4.31824505e-01 -5.30063808e-01 1.69009715e-02 7.80998588e-01
2.56915241e-01 -6.21135414e-01 5.76259732e-01 1.20823038e+00
-3.66667241e-01 -7.68899024e-01 -6.31573021e-01 -1.02409804e+00
-2.98816770e-01 -1.74081668e-01 8.69726360e-01 6.34182036e-01
6.50776803e-01 1.99315920e-01 -7.10582197e-01 3.80990863e-01
3.48112851e-01 3.21773857e-01 1.10459220e+00 -7.54364431e-01
-4.41522568e-01 -1.20382920e-01 -3.91515583e-01 -1.37322748e+00
1.08279079e-01 -7.68956959e-01 6.65189505e-01 -2.08904290e+00
1.83209822e-01 -6.06773078e-01 1.44016221e-02 8.21908176e-01
-1.41251892e-01 -5.09787440e-01 5.02311110e-01 5.58664680e-01
-5.02661824e-01 7.82680154e-01 1.96901751e+00 -1.07253753e-01
-5.78642130e-01 7.81818628e-02 -2.76292294e-01 6.37390196e-01
1.03234339e+00 -5.09514093e-01 -6.03748381e-01 -4.87966597e-01
-2.97641695e-01 5.32226026e-01 1.51370883e-01 -9.13114488e-01
1.16590865e-01 -5.36136210e-01 1.12890832e-01 -9.00625288e-02
4.89418864e-01 -1.13545620e+00 2.04158932e-01 1.12483418e+00
-2.58086413e-01 7.79873952e-02 2.49540821e-01 7.95495391e-01
1.45659983e-01 -3.53488058e-01 3.02114099e-01 -1.73954099e-01
-1.04754162e+00 1.31923899e-01 -7.07048357e-01 -8.94528627e-01
1.65349472e+00 -5.06782234e-01 1.25953674e-01 -5.97740710e-01
-7.00232923e-01 6.57186985e-01 3.16513479e-01 8.57171714e-01
7.68133998e-01 -1.32815397e+00 -4.76388961e-01 -5.32786734e-03
1.43409565e-01 -1.48605304e-02 3.48122269e-01 4.10188407e-01
-8.50298226e-01 2.27889776e-01 -5.23683429e-01 -5.98019838e-01
-1.03101420e+00 5.68529248e-01 2.98491925e-01 3.45320702e-02
-9.26971316e-01 4.26587433e-01 2.16746211e-01 -8.11056256e-01
1.59527123e-01 -4.16015208e-01 -2.57145047e-01 -7.80802906e-01
1.84672296e-01 1.82328165e-01 -5.91823697e-01 -2.51679599e-01
-6.50245696e-02 2.70996541e-01 2.59900298e-02 1.33186460e-01
1.44345105e+00 -1.71871915e-01 -1.16798550e-01 3.10409695e-01
6.52638435e-01 -7.59195328e-01 -1.68683171e+00 3.50786090e-01
-1.97712272e-01 -5.09361565e-01 -5.12312353e-01 -7.54432440e-01
-6.64486587e-01 6.84951663e-01 2.97419190e-01 3.27571034e-01
7.30691910e-01 -2.68819243e-01 6.68424726e-01 7.79519439e-01
7.72744656e-01 -1.15402150e+00 5.22782564e-01 1.56404465e-01
1.29699874e+00 -1.25556123e+00 -2.00868413e-01 -4.35004503e-01
-9.25947785e-01 1.16627431e+00 1.10938275e+00 -2.48180255e-01
1.54653683e-01 -3.64525430e-02 -8.87088254e-02 -1.45977467e-01
-3.31114769e-01 3.46995413e-01 -1.08437628e-01 7.31881797e-01
-1.10299453e-01 3.15654665e-01 -3.88232738e-01 5.12655675e-01
-1.19836815e-01 1.34784192e-01 8.10468137e-01 1.15815604e+00
-7.75233746e-01 -7.97872961e-01 -1.85531616e-01 1.31053761e-01
2.95428127e-01 5.81508160e-01 -9.97226387e-02 1.14155591e+00
-1.49119273e-01 1.11922860e+00 -2.90073723e-01 -4.44316924e-01
4.69496012e-01 -2.82655865e-01 5.39709270e-01 -6.21453583e-01
-1.57490492e-01 -4.56860602e-01 -7.87628442e-02 -6.81479275e-01
-1.14770561e-01 -3.01570117e-01 -1.88054037e+00 -1.65050045e-01
-3.15432161e-01 1.12921789e-01 9.55661178e-01 1.13128507e+00
3.00378293e-01 6.74136341e-01 8.39297235e-01 -1.12514818e+00
-9.19959486e-01 -9.37228501e-01 -6.51987791e-01 6.25371397e-01
-7.31056277e-03 -8.26327503e-01 -2.40445331e-01 -1.05194934e-01]
|
[4.589056491851807, 0.9433034062385559]
|
e88a56f0-3f4d-4c39-ae9b-bb2b19ebb16b
|
exploring-transfer-learning-for-low-resource
|
1901.04276
| null |
http://arxiv.org/abs/1901.04276v1
|
http://arxiv.org/pdf/1901.04276v1.pdf
|
Exploring Transfer Learning for Low Resource Emotional TTS
|
During the last few years, spoken language technologies have known a big
improvement thanks to Deep Learning. However Deep Learning-based algorithms
require amounts of data that are often difficult and costly to gather.
Particularly, modeling the variability in speech of different speakers,
different styles or different emotions with few data remains challenging. In
this paper, we investigate how to leverage fine-tuning on a pre-trained Deep
Learning-based TTS model to synthesize speech with a small dataset of another
speaker. Then we investigate the possibility to adapt this model to have
emotional TTS by fine-tuning the neutral TTS model with a small emotional
dataset.
|
['Kevin El Haddad', 'Noé Tits', 'Thierry Dutoit']
|
2019-01-14
|
exploring-transfer-learning-for-low-resource-1
| null | null |
advances-in-intelligent-systems-and-computing
|
['emotional-speech-synthesis', 'expressive-speech-synthesis']
|
['speech', 'speech']
|
[-2.91610152e-01 3.10913205e-01 7.58316666e-02 -8.97018790e-01
-6.43394947e-01 -5.27645171e-01 5.44362068e-01 -1.53257445e-01
-1.90270200e-01 7.19372571e-01 2.39781186e-01 -2.98646856e-02
4.25030798e-01 -3.28673691e-01 -5.45005679e-01 -4.14462119e-01
8.05796534e-02 6.13521278e-01 -2.90781319e-01 -5.12383699e-01
-3.61360699e-01 4.94139403e-01 -1.61118019e+00 2.88724393e-01
6.66460276e-01 1.11223841e+00 -6.19050078e-02 8.46396089e-01
-5.62697053e-01 5.86608291e-01 -1.10855651e+00 -4.05112028e-01
1.19318597e-01 -6.25060380e-01 -6.28326595e-01 -9.45908658e-04
2.69758195e-01 -1.67248383e-01 -1.46508031e-02 7.93379724e-01
8.83286417e-01 3.72080654e-01 4.73135829e-01 -1.21588624e+00
-6.13383293e-01 9.54903841e-01 -1.92247424e-02 -1.23915218e-01
2.72018075e-01 1.50941253e-01 6.79302633e-01 -5.39179325e-01
4.76612568e-01 1.50813746e+00 5.47474146e-01 1.03557396e+00
-1.35355806e+00 -6.04706943e-01 2.96287667e-02 1.06482908e-01
-1.07345653e+00 -7.75690794e-01 1.07539737e+00 -1.51920676e-01
8.58670950e-01 2.90386438e-01 6.38511896e-01 1.92159045e+00
-1.60227880e-01 6.46904528e-01 1.36424649e+00 -4.12197530e-01
4.91523921e-01 6.46319568e-01 -2.57378727e-01 1.60133719e-01
-6.08086944e-01 6.11510426e-02 -5.02573371e-01 8.95252153e-02
3.97317372e-02 -5.18072426e-01 -1.88359991e-01 -1.28071266e-03
-9.05618250e-01 1.03088546e+00 1.03658400e-01 5.48614979e-01
-2.60850161e-01 -7.04800636e-02 6.53986752e-01 8.46144795e-01
5.85782588e-01 6.74103856e-01 -7.28741705e-01 -4.81904238e-01
-1.14227247e+00 1.20332457e-01 1.33395195e+00 6.47545159e-01
6.28347635e-01 7.26056278e-01 -1.23817764e-01 1.26877797e+00
-1.68425888e-02 4.35384780e-01 8.85308743e-01 -8.70292664e-01
3.36232297e-02 -4.78966944e-02 -1.02816718e-02 -6.58749104e-01
-4.66009885e-01 -4.90875542e-01 -8.91304016e-01 3.63588691e-01
4.75934774e-01 -7.01047003e-01 -8.71328473e-01 1.99110866e+00
2.23279491e-01 -1.91237982e-02 3.40453565e-01 7.80141294e-01
7.26972640e-01 9.63739276e-01 -1.29788712e-01 -2.32134074e-01
9.12644744e-01 -1.00931656e+00 -8.72965395e-01 -3.72997820e-01
2.69040555e-01 -8.49455714e-01 1.68204951e+00 6.62471235e-01
-1.16587317e+00 -7.45234907e-01 -9.20613945e-01 -5.16505502e-02
-5.92568934e-01 -7.73073286e-02 3.80652159e-01 8.89838576e-01
-1.24729693e+00 5.18093288e-01 -4.11719620e-01 -3.73891354e-01
-3.53811309e-02 3.19340467e-01 -3.52148339e-02 4.75810945e-01
-1.52590168e+00 9.86736119e-01 1.63400903e-01 -6.66943416e-02
-7.57164359e-01 -5.62882543e-01 -6.34851098e-01 2.01077387e-01
2.51343548e-01 -5.66176534e-01 1.48897123e+00 -1.67852998e+00
-2.51509261e+00 7.92316675e-01 -1.16480887e-01 -5.37148535e-01
6.25073433e-01 -3.35907340e-02 -6.63936913e-01 -2.25303352e-01
-5.58441460e-01 8.01521957e-01 1.22628307e+00 -1.30269575e+00
-1.58783421e-02 5.51696718e-02 -1.46454141e-01 -8.92431810e-02
-4.78973746e-01 1.42819479e-01 1.16704248e-01 -8.01495314e-01
-4.93513584e-01 -1.00785708e+00 -4.70545739e-02 -3.56112212e-01
-3.00102979e-01 -2.56667525e-01 6.45180941e-01 -4.71085221e-01
8.32366109e-01 -2.17143178e+00 2.94924974e-01 1.31872326e-01
-2.24794075e-01 3.72912824e-01 -3.58885258e-01 3.27835202e-01
-1.12706088e-01 2.76698709e-01 3.43990177e-02 -6.52021945e-01
4.63791519e-01 4.15358692e-01 -4.99801993e-01 -1.63080901e-01
3.14818084e-01 6.06660008e-01 -6.05274379e-01 -2.37230018e-01
1.14432998e-01 6.49226427e-01 -4.35697109e-01 6.31054103e-01
-4.78215158e-01 7.44363606e-01 -3.01523767e-02 2.18447551e-01
4.50497091e-01 3.31039459e-01 -3.52981016e-02 9.61783007e-02
-1.97365448e-01 3.25778067e-01 -9.20706689e-01 1.68472648e+00
-9.40375209e-01 7.57884860e-01 2.06228584e-01 -1.04922497e+00
1.36795497e+00 4.54495370e-01 2.68650174e-01 -6.40180647e-01
6.50515139e-01 3.01620543e-01 1.90212429e-01 -5.91737390e-01
3.43358666e-01 -9.13648546e-01 -3.05648476e-01 4.29755926e-01
3.51545960e-01 -7.01428115e-01 -2.86662757e-01 -3.49374205e-01
6.46964431e-01 -3.21502447e-01 -4.61002626e-02 1.58289168e-02
5.72940052e-01 -6.32216036e-01 4.06300783e-01 5.33027470e-01
-3.63859773e-01 6.61173642e-01 5.36913455e-01 -3.59683275e-01
-9.80713487e-01 -7.78244793e-01 1.66078180e-01 1.25897646e+00
-5.46288133e-01 -3.90444696e-02 -1.07856655e+00 -2.69031495e-01
-3.15793902e-01 1.08144939e+00 -7.39159107e-01 -4.48367655e-01
-2.59558976e-01 -4.41473395e-01 9.19949114e-01 2.51062930e-01
2.06970945e-01 -1.21510148e+00 -2.75231123e-01 2.70366549e-01
-3.69710401e-02 -1.06266332e+00 -2.77829617e-01 5.72421432e-01
-4.66990143e-01 -2.15402991e-02 -9.82003272e-01 -6.99232459e-01
-1.01949945e-01 -5.25922596e-01 1.27957547e+00 -3.38857591e-01
1.17214799e-01 3.30816209e-02 -3.63555402e-01 -7.59140968e-01
-9.47839200e-01 3.80896091e-01 3.93094033e-01 3.68478686e-01
3.14047635e-01 -7.67992914e-01 -4.06962894e-02 6.92762509e-02
-8.34983826e-01 -2.18510568e-01 3.17671686e-01 7.53248334e-01
1.27621934e-01 1.10067697e-02 9.42117751e-01 -5.45048654e-01
8.84476781e-01 -4.71007824e-01 -1.97430745e-01 2.02545926e-01
-2.45969325e-01 3.88645798e-01 1.05503678e+00 -9.48070765e-01
-1.23865950e+00 -1.83766618e-01 -6.15790009e-01 -6.16661251e-01
-5.22036314e-01 4.25955385e-01 -4.74028766e-01 2.23320797e-01
5.86606920e-01 -2.45527867e-02 -1.24218129e-01 -2.61950701e-01
5.81414998e-01 9.30689633e-01 4.40803856e-01 -6.72653556e-01
7.18221486e-01 5.41206487e-02 -4.01509732e-01 -9.46034491e-01
-8.93875480e-01 2.36595407e-01 -5.27229190e-01 -3.34467858e-01
7.69094825e-01 -6.40179038e-01 -6.45284355e-01 5.21674991e-01
-1.11225128e+00 -8.65813136e-01 -5.42569578e-01 4.26035255e-01
-5.92294753e-01 3.58084515e-02 -5.40477872e-01 -7.72002995e-01
-2.84631997e-01 -9.71213818e-01 8.54556203e-01 3.39306116e-01
-6.22034490e-01 -1.05782759e+00 4.05472338e-01 1.72010183e-01
8.90876830e-01 1.40883133e-01 6.99995160e-01 -8.04252148e-01
1.18422203e-01 3.67420055e-02 3.18673462e-01 7.74157703e-01
2.70039231e-01 4.21060532e-01 -1.40081120e+00 1.57961268e-02
3.92929435e-01 -8.63100350e-01 5.03445804e-01 3.32734704e-01
1.14653420e+00 -2.52373546e-01 4.92179275e-01 7.60363996e-01
6.67416394e-01 2.41804570e-01 3.31926584e-01 9.59972218e-02
5.11937141e-01 9.48895097e-01 2.32397735e-01 3.68798584e-01
2.49915868e-01 6.87034965e-01 3.41827981e-02 -1.05811097e-01
-8.03989992e-02 2.86229495e-02 6.24614418e-01 1.31888640e+00
1.20198980e-01 -4.14270133e-01 -7.93498337e-01 2.69006521e-01
-1.38478327e+00 -9.77568686e-01 4.06944394e-01 1.81847966e+00
1.20149648e+00 2.26873264e-01 2.23539963e-01 1.40436620e-01
4.75185782e-01 3.07154238e-01 -8.31856847e-01 -1.30240750e+00
-5.06831467e-01 5.08879721e-01 -1.69943899e-01 6.18266881e-01
-7.45136201e-01 1.04554021e+00 6.36466360e+00 7.29745567e-01
-1.89369345e+00 2.66883262e-02 7.26243377e-01 -4.16019201e-01
-5.31642973e-01 -3.65642458e-01 -5.42626858e-01 5.08967936e-01
1.79791117e+00 -3.90102744e-01 6.60422385e-01 7.59210825e-01
4.77180004e-01 3.76790166e-01 -1.25721622e+00 1.01225352e+00
2.46562302e-01 -7.23556995e-01 1.16198407e-02 -1.49566576e-01
6.77271008e-01 3.53203043e-02 4.48720932e-01 8.84912133e-01
3.23774964e-01 -1.16566169e+00 8.02296460e-01 5.44652879e-01
5.20173311e-01 -9.08738673e-01 6.29444897e-01 3.34618419e-01
-5.47447920e-01 1.28460348e-01 -2.14080229e-01 -6.52840659e-02
1.08875953e-01 5.40037215e-01 -9.33501840e-01 1.01903565e-01
6.72483325e-01 1.86508596e-01 -2.48261139e-01 4.82267261e-01
-6.09311974e-03 8.75607908e-01 -3.49049509e-01 -1.98584184e-01
3.99419844e-01 -1.37175992e-01 4.10816044e-01 1.43391907e+00
5.80821693e-01 -2.48845965e-01 6.87905997e-02 8.34707022e-01
-3.53816837e-01 1.65601552e-01 -4.99601990e-01 -3.25905412e-01
2.23901004e-01 1.33061993e+00 -3.29576701e-01 -4.43926603e-01
-2.89254397e-01 1.26139784e+00 2.50840694e-01 3.83133292e-01
-9.10122097e-01 -2.09102601e-01 7.65268922e-01 -3.11126262e-01
2.13012442e-01 -1.71638444e-01 -2.20703796e-01 -1.23083448e+00
-3.14809710e-01 -1.29042542e+00 -1.84506476e-01 -9.34595048e-01
-1.40826595e+00 1.09125745e+00 -3.94909501e-01 -8.69406402e-01
-7.78729677e-01 -4.20547336e-01 -6.32391751e-01 9.60543215e-01
-1.39096284e+00 -8.64195228e-01 -1.61600605e-01 5.78568220e-01
7.71185458e-01 -3.47114265e-01 9.45098877e-01 1.32315785e-01
-5.70185959e-01 7.49078393e-01 8.28869492e-02 -2.88954768e-02
1.09328032e+00 -1.38425457e+00 3.19460601e-01 2.67998844e-01
1.53925791e-01 3.56029332e-01 1.05698109e+00 1.06809929e-01
-1.05514967e+00 -8.70541573e-01 8.40149403e-01 -9.40080658e-02
8.00661922e-01 -7.31466711e-01 -1.30322099e+00 5.17709494e-01
8.33151460e-01 -2.33324945e-01 9.80910718e-01 2.80043095e-01
-4.14080590e-01 -2.65015483e-01 -1.02688515e+00 7.51143098e-01
4.39547718e-01 -7.83173680e-01 -7.04737604e-01 -9.52058211e-02
7.81586766e-01 -2.32854083e-01 -9.92137015e-01 7.29725957e-02
5.58890998e-01 -1.09300423e+00 7.22198069e-01 -7.43803084e-01
1.45690128e-01 1.72660455e-01 -2.75849402e-01 -2.06936646e+00
7.44697452e-02 -9.39788640e-01 1.39737815e-01 1.63603508e+00
3.61295372e-01 -5.21855295e-01 5.62559962e-01 7.86749363e-01
-1.64528713e-01 -4.23995525e-01 -8.22241545e-01 -9.66600239e-01
5.57877541e-01 -4.96218741e-01 8.57707083e-01 1.15509570e+00
3.87400053e-02 5.58797717e-01 -6.09847486e-01 -1.13528930e-01
9.20495093e-02 -5.04787341e-02 9.48224247e-01 -1.28405523e+00
-4.16906744e-01 -6.24975562e-01 1.60068721e-02 -5.74628055e-01
8.92077565e-01 -6.47220612e-01 2.96300441e-01 -9.52175856e-01
-3.47961187e-01 -1.98415920e-01 -3.22751909e-01 3.34196627e-01
-1.49229262e-02 -3.95139344e-02 2.71748185e-01 -3.42683613e-01
-1.67819142e-01 1.11865675e+00 8.87896359e-01 -2.28349045e-01
-3.53063673e-01 -1.05878636e-02 -4.58965421e-01 8.10619295e-01
1.26272047e+00 -3.73774618e-01 -4.81624961e-01 -3.48662794e-01
2.24210516e-01 7.52321035e-02 4.91271392e-02 -1.10022843e+00
-9.21303108e-02 -1.16230331e-01 2.44126320e-01 -1.15070738e-01
7.97640204e-01 -6.98662400e-01 7.03603402e-02 -2.13450398e-02
-7.92196333e-01 -2.83459306e-01 4.85132992e-01 1.12824559e-01
-5.68740427e-01 -2.23810315e-01 9.02171075e-01 -5.31075597e-02
-3.06417793e-01 4.77579571e-02 -7.17790484e-01 1.76967129e-01
5.76244533e-01 1.12564899e-01 1.32397905e-01 -9.27081943e-01
-9.19604123e-01 -1.34053782e-01 3.05847675e-01 6.89379275e-01
2.68653661e-01 -1.26400685e+00 -6.76938713e-01 2.16238573e-01
-1.58131003e-01 -4.46897805e-01 3.32639903e-01 3.91810894e-01
-1.98591128e-03 1.53816968e-01 -4.48797703e-01 -3.61679345e-01
-1.29964173e+00 5.31487584e-01 6.28109336e-01 -1.23136872e-02
-1.47661120e-01 9.34530556e-01 -6.94393367e-02 -7.84812212e-01
3.78259242e-01 -5.92484951e-01 -6.02279864e-02 4.31365132e-01
3.01214963e-01 1.37616083e-01 1.08664380e-02 -6.29771113e-01
-2.18181252e-01 3.86657029e-01 1.44422844e-01 -6.98982477e-01
1.23772204e+00 -3.33997637e-01 2.89363861e-01 1.21092737e+00
1.30409849e+00 1.87136769e-01 -1.37616396e+00 -6.53942972e-02
6.47636829e-03 2.50410996e-02 1.50678873e-01 -9.21709359e-01
-1.18772686e+00 1.34810770e+00 4.51804817e-01 4.49031323e-01
1.05700421e+00 -1.60649374e-01 9.79508579e-01 6.13674283e-01
8.88926461e-02 -1.41537321e+00 2.00115725e-01 7.07667053e-01
1.07552493e+00 -1.33668876e+00 -7.01117158e-01 3.00851673e-01
-9.90702748e-01 1.32522535e+00 3.20852548e-01 1.53319106e-01
7.84938037e-01 4.23415959e-01 7.26613700e-01 2.64722914e-01
-1.12337303e+00 -4.49050888e-02 2.47437194e-01 5.45113325e-01
6.31127536e-01 1.46097317e-01 1.81279823e-01 6.75361156e-01
-8.94529819e-01 6.38167262e-02 6.22650445e-01 2.77267337e-01
-3.08603793e-01 -1.31692088e+00 -3.34856153e-01 -2.48436965e-02
-5.42632341e-01 2.66786832e-02 -8.25001597e-01 5.14164925e-01
2.70313378e-02 1.29818749e+00 -2.40233727e-02 -4.83295143e-01
3.43362957e-01 6.38910353e-01 3.31769794e-01 -5.53189754e-01
-7.99413383e-01 1.63472757e-01 2.48765677e-01 -3.20693761e-01
-3.87925655e-01 -7.48403013e-01 -1.15410650e+00 -2.70261407e-01
2.19273329e-01 2.13743910e-01 9.08651829e-01 9.82062221e-01
2.37626672e-01 5.80983877e-01 9.90689039e-01 -1.03241765e+00
-6.78179562e-01 -1.12932587e+00 -6.28239751e-01 4.13078725e-01
4.92358685e-01 -3.52578640e-01 -6.36628151e-01 6.91491663e-02]
|
[14.275425910949707, 6.282937049865723]
|
eaaf06bb-0dc5-4ba2-ae8b-fc32e613a009
|
factorized-attention-self-attention-with
|
1812.01243
| null |
https://arxiv.org/abs/1812.01243v9
|
https://arxiv.org/pdf/1812.01243v9.pdf
|
Efficient Attention: Attention with Linear Complexities
|
Dot-product attention has wide applications in computer vision and natural language processing. However, its memory and computational costs grow quadratically with the input size. Such growth prohibits its application on high-resolution inputs. To remedy this drawback, this paper proposes a novel efficient attention mechanism equivalent to dot-product attention but with substantially less memory and computational costs. Its resource efficiency allows more widespread and flexible integration of attention modules into a network, which leads to better accuracies. Empirical evaluations demonstrated the effectiveness of its advantages. Efficient attention modules brought significant performance boosts to object detectors and instance segmenters on MS-COCO 2017. Further, the resource efficiency democratizes attention to complex models, where high costs prohibit the use of dot-product attention. As an exemplar, a model with efficient attention achieved state-of-the-art accuracies for stereo depth estimation on the Scene Flow dataset. Code is available at https://github.com/cmsflash/efficient-attention.
|
['Shuai Yi', 'Haiyu Zhao', 'Hongsheng Li', 'Zhuoran Shen', 'Mingyuan Zhang']
|
2018-12-04
| null | null | null | null |
['stereo-depth-estimation', 'extractive-document-summarization']
|
['computer-vision', 'natural-language-processing']
|
[-1.26210049e-01 -6.43771365e-02 -2.50224978e-01 -2.11934999e-01
-5.11290014e-01 -1.37224153e-01 6.29409790e-01 -1.34459168e-01
-6.94787800e-01 2.89441943e-01 8.35975260e-02 -3.05390984e-01
2.34751597e-01 -7.20477343e-01 -6.63483024e-01 -3.66784513e-01
1.28620520e-01 1.58038765e-01 3.52273464e-01 1.21339764e-02
3.70427549e-01 4.57973957e-01 -1.59553182e+00 1.75200224e-01
9.93966877e-01 1.00781655e+00 5.36056042e-01 5.71852148e-01
-1.14860557e-01 9.45432603e-01 -2.94231981e-01 -5.73940277e-01
2.80631423e-01 -1.87789202e-02 -8.53450418e-01 -9.89617482e-02
8.77239883e-01 -6.42713428e-01 -7.13021517e-01 1.03148901e+00
4.84381944e-01 2.61152416e-01 4.47259098e-01 -1.16676426e+00
-9.48789001e-01 3.08869451e-01 -8.05975795e-01 6.24080300e-01
7.74610147e-04 4.03497607e-01 1.24758971e+00 -1.13643265e+00
3.21480483e-01 1.29159188e+00 6.06369615e-01 3.49640638e-01
-1.02033472e+00 -6.12948298e-01 3.27341050e-01 5.11895239e-01
-1.32404053e+00 -5.28319716e-01 4.79466200e-01 -4.59665745e-01
1.65035403e+00 -1.44309103e-02 6.55753851e-01 9.78121340e-01
8.02509934e-02 1.14722598e+00 5.37356555e-01 -7.52437860e-02
-2.74201445e-02 -1.06765442e-01 1.45382941e-01 6.83892488e-01
4.28970337e-01 1.06687978e-01 -3.35481465e-01 5.09961128e-01
9.53320861e-01 8.70016664e-02 -3.26681674e-01 -1.70912713e-01
-1.04248762e+00 8.36562514e-01 9.75171328e-01 2.73780912e-01
-3.61660957e-01 4.08275306e-01 4.94403929e-01 -1.98268369e-01
5.58928132e-01 5.38956463e-01 -3.11779827e-01 -2.32722312e-01
-9.19179261e-01 -4.27159742e-02 1.84523195e-01 1.01193464e+00
6.74485326e-01 2.34356001e-01 -1.31810099e-01 9.17282999e-01
1.97454587e-01 5.20077705e-01 2.70549625e-01 -1.12486875e+00
6.48939669e-01 5.70377529e-01 -1.45151272e-01 -9.19520438e-01
-4.45078284e-01 -6.27013922e-01 -9.37430501e-01 2.05429867e-01
4.94309157e-01 6.19993955e-02 -9.20893610e-01 1.61658597e+00
7.18740076e-02 1.65035099e-01 -3.13788205e-01 1.31860471e+00
8.47452760e-01 6.11331820e-01 2.06609368e-01 1.99646145e-01
1.38494122e+00 -1.41328418e+00 -5.87206841e-01 -6.21678293e-01
5.48043132e-01 -7.24243343e-01 1.15208173e+00 2.49633521e-01
-1.12747002e+00 -7.86849082e-01 -9.47791874e-01 -6.55448675e-01
-1.15655482e-01 7.63849169e-02 1.06731462e+00 3.62200230e-01
-1.03230894e+00 5.62055826e-01 -9.85273302e-01 -3.30054790e-01
1.06209576e+00 3.64000678e-01 -1.50806367e-01 -5.90511560e-02
-8.58002484e-01 8.79506052e-01 1.81534380e-01 1.13136530e-01
-7.08316326e-01 -9.46441174e-01 -1.03576648e+00 4.17322516e-01
1.49006292e-01 -8.63570929e-01 1.45449471e+00 -1.00505221e+00
-1.48648131e+00 7.91194677e-01 -1.92703769e-01 -6.45265877e-01
5.70995092e-01 -6.44894958e-01 -4.51174192e-03 1.41861379e-01
1.97424516e-01 1.20463395e+00 8.56930137e-01 -5.57367980e-01
-7.75005400e-01 -1.40283078e-01 2.72916436e-01 1.92348674e-01
-3.95031095e-01 -9.64489579e-02 -7.41348088e-01 -5.11663020e-01
-3.19948681e-02 -7.37234294e-01 -1.49521306e-01 1.95937410e-01
-1.15627334e-01 -2.53528446e-01 6.86566651e-01 -4.58198309e-01
1.10471392e+00 -2.30954194e+00 1.01549961e-01 -5.14317870e-01
3.18772316e-01 5.32626808e-01 -2.10441262e-01 -5.71412295e-02
-4.67131771e-02 1.08742997e-01 -2.00246245e-01 -5.19947469e-01
-2.02983782e-01 7.55594671e-02 -3.26744080e-01 5.00506163e-01
6.21068895e-01 1.08965933e+00 -8.63905787e-01 -4.48674351e-01
4.68580037e-01 5.27904272e-01 -9.94517684e-01 5.43351918e-02
6.73211040e-03 1.81565672e-01 -1.80087134e-01 6.28618121e-01
6.81513488e-01 -6.09486401e-01 -2.14174822e-01 -3.34940493e-01
-2.48313531e-01 4.29132760e-01 -8.13097179e-01 1.77340889e+00
-5.64025283e-01 1.23356533e+00 -1.06327660e-01 -8.71138632e-01
6.38204038e-01 -1.66873413e-03 2.68696010e-01 -1.05842507e+00
3.21663141e-01 1.08545281e-01 2.18693882e-01 -5.58802605e-01
6.84157550e-01 1.44700214e-01 3.25829566e-01 3.45252790e-02
2.22417027e-01 -5.51973097e-02 2.53194839e-01 1.73504233e-01
7.80282676e-01 1.41915590e-01 3.11170161e-01 -4.20708090e-01
4.76530820e-01 -3.40778939e-02 4.91336942e-01 6.36043608e-01
-4.94364887e-01 6.59217477e-01 2.42640585e-01 -5.74996173e-01
-9.96037483e-01 -8.61156344e-01 -2.20678568e-01 1.19287825e+00
2.09378779e-01 -4.52644348e-01 -5.12666464e-01 -5.23221433e-01
9.40022245e-02 3.86909217e-01 -6.28924549e-01 1.16721600e-01
-6.06082022e-01 -4.31974858e-01 4.43222582e-01 9.13643241e-01
7.89477289e-01 -1.20693171e+00 -1.02715087e+00 4.22468670e-02
-3.97563487e-01 -1.17867541e+00 -5.59915841e-01 -1.71926796e-01
-9.28483665e-01 -1.01350868e+00 -9.46200132e-01 -6.34398520e-01
5.85538149e-01 5.83974183e-01 1.11631954e+00 5.46116978e-02
-5.40471077e-01 1.58420518e-01 -9.21831951e-02 -4.83771890e-01
2.39833593e-01 3.51571947e-01 -2.46537387e-01 -7.73204789e-02
3.13201874e-01 -2.94268996e-01 -8.98548305e-01 8.51047784e-02
-7.56838381e-01 2.41511747e-01 5.45025766e-01 9.17990804e-01
3.33064079e-01 -5.81500769e-01 3.62687051e-01 -6.22818828e-01
2.25525588e-01 -2.64163524e-01 -8.10891509e-01 -2.51125574e-01
-4.24188495e-01 9.51144919e-02 4.79371101e-01 -3.10446322e-01
-1.14317203e+00 2.57119667e-02 -1.55531168e-01 -5.82143188e-01
6.37566969e-02 2.83518404e-01 2.01939642e-01 5.02313636e-02
6.01369917e-01 -3.24831158e-02 -5.42687960e-02 -3.88661951e-01
3.03007305e-01 4.69898045e-01 5.45467615e-01 -2.84440964e-01
3.61572444e-01 5.03696799e-01 -1.16191044e-01 -9.30442274e-01
-1.05272353e+00 -4.08418745e-01 -3.91727179e-01 -1.11645684e-02
8.95128429e-01 -1.20162082e+00 -7.60135293e-01 5.72296977e-01
-1.28401983e+00 -4.66766030e-01 -1.63619474e-01 6.57299519e-01
-4.76667106e-01 2.07094103e-01 -7.35792398e-01 -5.87289453e-01
-3.50257039e-01 -1.10898244e+00 9.09011006e-01 3.47635061e-01
-3.02652538e-01 -7.95989931e-01 -2.43626013e-01 5.39036751e-01
5.27502775e-01 -2.50718385e-01 4.84945357e-01 -4.22379598e-02
-9.45764542e-01 1.72996223e-01 -8.12418520e-01 3.05285990e-01
-3.99425514e-02 1.82567328e-01 -1.32874060e+00 -2.73699611e-01
-3.83557379e-01 -2.50699818e-01 1.21528924e+00 6.90499961e-01
1.27095163e+00 -7.25415796e-02 -1.80443391e-01 1.11112499e+00
1.31971228e+00 3.17430533e-02 6.90658510e-01 3.88239503e-01
9.48967040e-01 4.56938565e-01 5.83100855e-01 3.86611193e-01
4.21615690e-01 6.74559355e-01 5.32900274e-01 -3.31362277e-01
-4.49950576e-01 -1.42356858e-01 1.47196457e-01 6.26887202e-01
-2.77518004e-01 -1.33786500e-01 -1.09960771e+00 9.33004260e-01
-1.80024254e+00 -1.00312984e+00 -1.61158606e-01 1.87806928e+00
4.67541903e-01 2.56473273e-01 -6.22052997e-02 -7.69083435e-03
5.35189688e-01 2.87475318e-01 -6.04346991e-01 -5.35499811e-01
-6.35017380e-02 1.64539605e-01 4.53661770e-01 5.51900923e-01
-1.20301414e+00 1.18334639e+00 5.73234081e+00 7.88565278e-01
-1.31995571e+00 1.40572309e-01 7.92919278e-01 -3.50250155e-01
1.14378892e-01 -3.64060044e-01 -8.43449295e-01 4.06351060e-01
7.78934717e-01 -2.19795570e-01 3.76463473e-01 1.02892566e+00
9.47851986e-02 -1.83260962e-01 -1.12286866e+00 1.32745755e+00
-9.71311778e-02 -1.63064444e+00 -2.47836765e-02 9.01445225e-02
8.25334132e-01 6.09474957e-01 2.09976882e-01 4.59676445e-01
-9.89828408e-02 -9.04947102e-01 7.85116971e-01 1.65261745e-01
8.59267116e-01 -6.96836114e-01 8.06712210e-01 1.09557785e-01
-1.23075354e+00 -2.54835904e-01 -4.95744526e-01 -5.42796195e-01
1.62925199e-01 2.92642266e-01 -5.89695215e-01 2.51872003e-01
9.34490860e-01 9.24567640e-01 -5.63495040e-01 1.17749429e+00
-2.30163723e-01 3.58734280e-01 -2.53065348e-01 1.37981176e-01
4.40388978e-01 -6.68551773e-02 3.87064338e-01 1.51779044e+00
2.25139499e-01 3.69620137e-02 -1.55764624e-01 8.78781140e-01
-2.46058509e-01 -2.77031772e-02 -6.39134407e-01 5.35397306e-02
3.41669887e-01 1.24004018e+00 -7.02009618e-01 -5.60315073e-01
-7.09519744e-01 8.17868292e-01 7.75150895e-01 3.01875889e-01
-1.11436427e+00 -3.67996544e-01 1.03406644e+00 3.58717203e-01
4.38696980e-01 -3.05047452e-01 -6.89599812e-01 -1.16352808e+00
-5.88325448e-02 -5.80829740e-01 3.51971239e-01 -6.61311626e-01
-1.10389495e+00 7.35935152e-01 -1.94243759e-01 -1.20853829e+00
2.75349207e-02 -7.10441589e-01 -4.43496466e-01 8.26553583e-01
-1.62795794e+00 -8.97971570e-01 -6.36560261e-01 4.95966464e-01
8.35709453e-01 5.55289052e-02 4.64278013e-01 6.19953513e-01
-8.55224133e-01 6.58752084e-01 -3.45057070e-01 2.62201786e-01
6.40586674e-01 -9.88031983e-01 7.88754225e-01 1.03399754e+00
1.71774179e-01 4.15802300e-01 2.67572284e-01 -2.53860950e-01
-1.12029958e+00 -1.15466547e+00 9.39968467e-01 -3.80500197e-01
7.15807617e-01 -2.07991302e-01 -9.77406263e-01 7.15529323e-01
3.00457567e-01 2.88451761e-01 1.65013403e-01 2.19464898e-01
-4.87559021e-01 -5.70470914e-02 -7.61136711e-01 7.59614527e-01
1.24746943e+00 -5.06907821e-01 -4.93466944e-01 7.59570748e-02
6.85653150e-01 -4.85395133e-01 -4.70835984e-01 5.13605297e-01
4.73310649e-01 -1.16968608e+00 1.00782871e+00 -2.86429495e-01
7.01301515e-01 -2.90546149e-01 -5.06656952e-02 -9.20076251e-01
-7.32168496e-01 -4.21967387e-01 -3.37018579e-01 8.67865980e-01
2.89330512e-01 -7.04264283e-01 5.54938853e-01 6.36447132e-01
-3.60998839e-01 -8.73030841e-01 -8.89060140e-01 -7.66140759e-01
4.82535884e-02 -5.66376686e-01 2.74475753e-01 9.18825328e-01
-7.54552633e-02 5.25955498e-01 -1.62298486e-01 1.18893370e-01
5.20243526e-01 6.99543133e-02 6.71653390e-01 -9.77371991e-01
-3.89599860e-01 -8.36450577e-01 -5.57061315e-01 -1.50689793e+00
6.62346855e-02 -8.24471951e-01 -1.01268068e-01 -1.48070681e+00
1.99293360e-01 -2.74420381e-01 -1.71085149e-01 4.94548559e-01
-3.96848232e-01 5.79194725e-01 5.48970580e-01 1.99241772e-01
-6.63533270e-01 7.24623024e-01 1.34502566e+00 -2.34431162e-01
-2.13185638e-01 -3.00348878e-01 -6.03327632e-01 7.42686272e-01
1.00701129e+00 -2.20625997e-01 -3.44511718e-01 -1.01275396e+00
-6.83406070e-02 -2.16079339e-01 4.34868544e-01 -1.22323096e+00
2.33892113e-01 3.23992595e-02 3.51698577e-01 -5.70161521e-01
4.06961232e-01 -5.27514338e-01 -1.52804643e-01 5.36953330e-01
-8.84025022e-02 9.55123678e-02 6.18278861e-01 2.45660618e-01
-4.32785749e-01 -4.61680517e-02 1.08783650e+00 3.26206386e-02
-9.73941445e-01 5.33279121e-01 -1.92090943e-01 1.72101632e-01
9.34041023e-01 -3.42463583e-01 -5.61079741e-01 -2.46544868e-01
-4.36979473e-01 1.18500546e-01 3.29155803e-01 7.00638175e-01
5.96758008e-01 -1.13145423e+00 -6.15343690e-01 3.37419957e-01
2.96928026e-02 1.90147907e-01 5.41051090e-01 9.72452164e-01
-7.70924330e-01 8.52116108e-01 -4.26982760e-01 -8.78590226e-01
-1.09993529e+00 6.18769884e-01 3.67921442e-01 3.75742353e-02
-7.87987649e-01 1.28431237e+00 6.89837635e-01 7.42665306e-02
1.21486515e-01 -5.90663254e-01 -2.32211370e-02 -4.11833916e-03
6.02096915e-01 4.49049860e-01 -5.27759455e-03 -5.36157668e-01
-4.47874159e-01 6.92591131e-01 -1.46338940e-01 1.63590088e-01
1.20882440e+00 -6.71517029e-02 1.54111817e-01 1.15280889e-01
1.10244167e+00 -3.46651882e-01 -1.80895662e+00 -1.84119657e-01
-1.59311488e-01 -7.36785948e-01 3.47421587e-01 -4.86155361e-01
-1.38768864e+00 1.22772956e+00 3.95719230e-01 -5.29742390e-02
1.15613902e+00 -3.76092829e-02 6.42951071e-01 2.57577032e-01
1.88988879e-01 -8.64885092e-01 5.36921620e-02 7.73539841e-01
9.39579129e-01 -1.57963336e+00 -1.23543315e-01 -4.29132938e-01
-6.36566818e-01 9.14498508e-01 1.05926645e+00 -2.83701330e-01
4.30890560e-01 2.93547928e-01 -5.05918041e-02 -8.31125826e-02
-7.79102921e-01 -3.54888499e-01 3.95796597e-01 5.29226184e-01
5.96706331e-01 6.82647228e-02 -5.33575483e-04 2.94094533e-01
-1.69834197e-01 1.98264495e-02 3.42926323e-01 4.82875109e-01
-3.15889806e-01 -5.68088591e-01 2.75500701e-03 3.47197264e-01
-4.44247663e-01 -3.93457919e-01 1.15605719e-01 9.02822316e-01
1.25346988e-01 8.53122056e-01 5.61377645e-01 -1.02825940e-01
3.27802867e-01 -2.24155545e-01 4.89772618e-01 -4.78802890e-01
-5.72424173e-01 -2.93355770e-02 2.19121352e-02 -9.36329603e-01
-2.61604369e-01 -4.53994513e-01 -1.14468551e+00 -5.49022496e-01
-1.65649295e-01 -3.75277519e-01 4.17765409e-01 6.09345853e-01
7.07870126e-01 7.73182511e-01 2.30717540e-01 -1.06583202e+00
-3.59769881e-01 -1.10121441e+00 -7.36339763e-02 3.89079660e-01
4.00880158e-01 -7.36142278e-01 -1.83308825e-01 -3.94397154e-02]
|
[9.554512023925781, 0.7786177396774292]
|
b4262131-1a39-46aa-bbbd-090152a440c9
|
instrument-independent-dastgah-recognition-of
|
1812.07017
| null |
http://arxiv.org/abs/1812.07017v3
|
http://arxiv.org/pdf/1812.07017v3.pdf
|
Instrument-Independent Dastgah Recognition of Iranian Classical Music Using AzarNet
|
In this paper, AzarNet, a deep neural network (DNN), is proposed to
recognizing seven different Dastgahs of Iranian classical music in Maryam
Iranian classical music (MICM) dataset. Over the last years, there has been
remarkable interest in employing feature learning and DNNs which lead to
decreasing the required engineering effort. DNNs have shown better performance
in many classification tasks such as audio signal classification compares to
shallow processing architectures. Despite image data, audio data need some
preprocessing steps to extract spectra and temporal features. Some
transformations like Short-Time Fourier Transform (STFT) have been used in the
state of art researches to transform audio signals from time-domain to
time-frequency domain to extract both temporal and spectra features. In this
research, the STFT output results which are extracted features are given to
AzarNet for learning and classification processes. It is worth noting that, the
mentioned dataset contains music tracks composed with two instruments (violin
and straw). The overall f1 score of AzarNet on test set, for average of all
seven classes was 86.21% which is the best result ever reported in Dastgah
classification according to our best knowledge.
|
['Ali Ahmadi', 'Shahla RezezadehAzar', 'Saber Malekzadeh', 'Maryam Samami']
|
2018-12-17
| null | null | null | null |
['recognizing-seven-different-dastgahs-of']
|
['music']
|
[ 3.10986429e-01 -6.49288476e-01 2.81920195e-01 2.95777433e-03
-4.02509212e-01 -6.96783245e-01 4.06003535e-01 -1.68165594e-01
-3.60990793e-01 6.03998780e-01 1.36693031e-01 1.55155957e-01
-8.01677704e-01 -6.98700428e-01 -2.46293560e-01 -7.80609190e-01
-3.12893391e-01 1.41043961e-01 -1.59366384e-01 -2.42303252e-01
6.19303763e-01 7.08532870e-01 -1.80329442e+00 4.34896976e-01
4.39043403e-01 1.27535546e+00 -1.60328224e-01 7.96857059e-01
4.01427336e-02 5.80716252e-01 -8.98816764e-01 -1.50124937e-01
3.72558922e-01 -6.86863542e-01 -7.48979330e-01 -2.98802644e-01
3.99941891e-01 9.18172374e-02 -3.06742281e-01 9.06213164e-01
8.70050490e-01 5.31937599e-01 9.39761996e-01 -9.31101561e-01
-6.24449909e-01 1.06544352e+00 -4.69976723e-01 5.10045648e-01
1.51931539e-01 -1.37894094e-01 9.18152750e-01 -7.64773369e-01
4.08647031e-01 9.31731343e-01 7.53430068e-01 2.80901134e-01
-6.99496806e-01 -9.79899883e-01 -5.57410479e-01 7.84440339e-01
-1.26169026e+00 -2.13875353e-01 1.18491280e+00 -4.78551805e-01
9.05337751e-01 1.86794639e-01 9.76596117e-01 9.02694643e-01
4.25043732e-01 6.28696442e-01 1.10635388e+00 -5.17661691e-01
-5.90601005e-02 -2.59264946e-01 9.51193497e-02 6.55886754e-02
-1.54300705e-01 2.97283530e-01 -8.93855512e-01 2.26020783e-01
6.35722458e-01 -2.19859183e-01 -9.44889188e-02 3.20340157e-01
-1.16635215e+00 7.12786317e-01 2.58354336e-01 1.04181409e+00
-6.98248863e-01 3.66902538e-02 8.49114716e-01 6.51923180e-01
-3.69872190e-02 5.02102077e-01 -5.29334068e-01 -6.19628072e-01
-1.08057451e+00 1.83891520e-01 5.50372720e-01 3.38651240e-01
-5.72973788e-02 9.40630615e-01 2.24479511e-01 8.47013950e-01
-2.16981217e-01 4.34955746e-01 1.08361113e+00 -7.89832413e-01
6.81604296e-02 4.58010972e-01 -4.93636757e-01 -1.24825287e+00
-4.22753304e-01 -9.06992912e-01 -1.05964470e+00 2.80352771e-01
3.74645054e-01 -2.23634288e-01 -7.19158053e-01 1.41651726e+00
4.61632051e-02 3.46064568e-01 4.03046221e-01 1.03497481e+00
9.89605010e-01 9.27790761e-01 -2.89487273e-01 -5.80132306e-02
1.34348345e+00 -4.67403889e-01 -6.14658952e-01 4.33317184e-01
1.52335048e-01 -1.42184758e+00 9.46152806e-01 1.28906500e+00
-8.11218977e-01 -9.89370704e-01 -1.28203833e+00 2.93613881e-01
-4.20500785e-01 3.14829260e-01 6.21886671e-01 4.99840140e-01
-5.09702742e-01 9.29703534e-01 -3.46928298e-01 -4.46735650e-01
2.33370334e-01 6.04823232e-01 -2.29776993e-01 4.47913408e-01
-1.25250936e+00 5.70867240e-01 8.80317867e-01 5.04058935e-02
-8.71330619e-01 -4.05676365e-01 -1.95398837e-01 5.31471930e-02
6.34060577e-02 -2.25555882e-01 1.26085961e+00 -1.36981857e+00
-1.67706275e+00 5.77244759e-01 4.04071361e-01 -7.73155749e-01
2.28665143e-01 -3.40357959e-01 -1.01293004e+00 1.53404027e-01
-2.38153651e-01 3.56549412e-01 9.19701874e-01 -5.51380932e-01
-6.09795153e-01 -2.83771098e-01 -2.86630809e-01 1.09078579e-01
-4.21469659e-01 2.37217583e-02 4.04134423e-01 -9.26042497e-01
1.79688364e-01 -8.79209995e-01 5.83679259e-01 -7.39892006e-01
-3.17025512e-01 -3.42188120e-01 1.08789289e+00 -8.46915305e-01
1.19070220e+00 -2.33402395e+00 1.08666137e-01 2.75627106e-01
-4.67567444e-01 5.30848265e-01 -2.46517226e-01 6.26796246e-01
-3.43334258e-01 -3.75331849e-01 3.64810228e-02 5.32261133e-01
-6.03269190e-02 6.75786883e-02 -3.95843565e-01 3.81772995e-01
-7.93310031e-02 4.04446840e-01 -3.11931282e-01 -3.05794179e-01
4.33844209e-01 6.69572353e-01 -2.77727008e-01 -3.50203604e-01
-4.42421390e-03 7.24168360e-01 -1.69119522e-01 5.94959915e-01
6.00442469e-01 6.26539588e-01 -2.26293936e-01 -4.83838260e-01
-3.98325980e-01 1.44950181e-01 -1.33461785e+00 1.75409889e+00
-1.99607432e-01 9.10438657e-01 -3.10910225e-01 -1.33877778e+00
1.18761480e+00 8.38703036e-01 8.39947939e-01 -7.38577366e-01
6.97028399e-01 4.30083305e-01 6.56571567e-01 -5.06436229e-01
3.54642630e-01 -3.51387680e-01 3.10984552e-02 2.17306942e-01
3.51122022e-01 1.71178970e-02 2.20504478e-01 -5.42196631e-01
4.72866684e-01 1.86893120e-01 3.52245241e-01 -1.70908365e-02
8.33790064e-01 5.56024164e-02 5.22704482e-01 3.17645997e-01
-6.70737773e-02 3.55083555e-01 1.27722219e-01 -5.74657798e-01
-1.02400589e+00 -8.53768408e-01 -1.18695103e-01 1.07000160e+00
-5.05635440e-01 7.25358259e-03 -5.29023767e-01 1.09291859e-01
-2.21907660e-01 5.30407369e-01 -3.69641811e-01 -3.10303658e-01
-6.62333786e-01 -3.01896483e-01 1.10426533e+00 4.29254889e-01
9.13959146e-01 -1.64390182e+00 -7.98112392e-01 5.51348627e-01
1.37504518e-01 -6.68224454e-01 -2.65258789e-01 3.87499243e-01
-1.12977934e+00 -1.09200859e+00 -7.75570571e-01 -9.45501626e-01
-3.43801677e-01 -2.39620194e-01 7.58742869e-01 -5.99046052e-01
-6.77376330e-01 -9.65147838e-03 -5.46397746e-01 -9.30422544e-01
-9.63061675e-02 1.28345668e-01 2.88570702e-01 8.12476352e-02
5.94477952e-01 -1.14401019e+00 -5.26563585e-01 -1.05564050e-01
-8.48805428e-01 -3.99226129e-01 7.92009175e-01 7.59222627e-01
5.32486916e-01 5.28519750e-01 9.21792209e-01 -4.88044590e-01
6.87944949e-01 -1.76328838e-01 -4.49943334e-01 -2.26068944e-01
-2.01199979e-01 -3.34496260e-01 9.21526134e-01 -6.88561201e-01
-7.54513204e-01 -4.96475771e-03 -1.52607977e-01 -3.32457244e-01
-3.07020396e-01 6.95044696e-01 1.08347543e-01 1.50841298e-02
8.07269335e-01 5.18393755e-01 -3.21397424e-01 -6.32669330e-01
6.59868345e-02 9.49285328e-01 9.54215705e-01 -5.48317015e-01
6.28004611e-01 1.03497803e-01 2.06741348e-01 -9.93303955e-01
-7.74816394e-01 -5.32178104e-01 -7.43302166e-01 -4.47705418e-01
9.84458268e-01 -5.30701816e-01 -1.10357141e+00 6.11190498e-01
-9.55048680e-01 3.68969887e-01 -2.24205628e-01 1.10091650e+00
-5.31273484e-01 3.05620097e-02 -4.22542423e-01 -1.09561419e+00
-7.33044147e-01 -5.86092949e-01 5.44625700e-01 4.96538818e-01
-4.39117998e-01 -5.87207913e-01 4.79513891e-02 1.71371743e-01
3.81123841e-01 4.61780757e-01 1.07407045e+00 -7.79501975e-01
-9.28956419e-02 -2.20285103e-01 1.89711452e-01 7.02382505e-01
1.81145042e-01 7.54048377e-02 -1.25185943e+00 -8.97005871e-02
2.93560714e-01 -2.43793130e-01 7.91179240e-01 6.74812019e-01
1.02251899e+00 3.35568301e-02 5.05181909e-01 5.70402443e-01
1.43473804e+00 1.12234032e+00 5.39642811e-01 6.54613972e-01
3.18469524e-01 1.01839967e-01 6.09235108e-01 7.58675516e-01
-4.34872299e-01 3.48191857e-01 2.41321385e-01 2.23788753e-01
-1.92849353e-01 -2.89218463e-02 6.02423191e-01 1.23053217e+00
-4.81179208e-01 -1.48405246e-02 -6.49849355e-01 4.05752867e-01
-1.36725020e+00 -1.31284988e+00 -1.85382977e-01 1.96615410e+00
5.15194595e-01 3.55363376e-02 3.27969909e-01 1.41615701e+00
5.65215170e-01 -7.76670799e-02 -5.65028906e-01 -7.11459458e-01
-2.89650679e-01 9.44334209e-01 6.24354091e-03 -1.05973057e-01
-1.23342335e+00 7.27709234e-01 4.77346754e+00 8.46825778e-01
-1.66041088e+00 -1.23861134e-01 -1.20581999e-01 -2.45828465e-01
4.15655673e-01 -2.25549325e-01 -1.00986198e-01 1.52381733e-01
1.22535717e+00 -1.88967958e-01 6.34297192e-01 5.49784601e-01
3.09956223e-01 1.46340236e-01 -8.18223238e-01 1.46005893e+00
6.17991015e-02 -9.26307499e-01 2.20768467e-01 -1.08890854e-01
6.33122325e-01 -2.16829494e-01 3.76626939e-01 4.37855691e-01
-3.18477064e-01 -1.11160898e+00 5.83769321e-01 5.81985533e-01
5.47053516e-01 -1.53120148e+00 9.55353141e-01 1.16569899e-01
-1.36891735e+00 -3.35868567e-01 -3.50166172e-01 -3.09129447e-01
-1.76142380e-01 3.82083625e-01 -7.18780994e-01 8.45950305e-01
6.96739674e-01 9.44017708e-01 -1.32816002e-01 1.25401092e+00
1.46103993e-01 1.00196183e+00 -1.78824812e-01 9.73106176e-02
4.60094869e-01 -3.21205288e-01 6.28095627e-01 1.04592395e+00
8.32181692e-01 -6.24034442e-02 -2.33865395e-01 4.31579828e-01
8.25544223e-02 4.57560956e-01 -5.70682347e-01 -4.94255275e-01
3.00909728e-01 1.12726808e+00 -7.66156256e-01 -2.51169175e-01
6.55643234e-04 6.78591609e-01 -6.03258669e-01 9.87702832e-02
-7.44425118e-01 -8.52121115e-01 3.04796934e-01 -2.46407717e-01
2.60826498e-01 -1.60309106e-01 -2.37920776e-01 -6.36830926e-01
-2.09341332e-01 -1.08795512e+00 5.91057599e-01 -7.18446851e-01
-1.17792177e+00 7.52973020e-01 -2.16838792e-01 -1.71280396e+00
-2.71298498e-01 -7.72332191e-01 -7.03562438e-01 8.35301757e-01
-1.05221868e+00 -9.44401205e-01 -1.41724825e-01 8.69556963e-01
7.09560275e-01 -7.90192127e-01 1.04145861e+00 6.63354814e-01
-2.19171882e-01 3.25523585e-01 2.70144612e-01 3.71660769e-01
7.27542460e-01 -1.06271124e+00 -2.40982682e-01 4.79227185e-01
7.18449235e-01 3.71882051e-01 6.66599810e-01 -1.91503435e-01
-1.41205406e+00 -7.86758602e-01 5.51228225e-01 2.93465018e-01
4.52202260e-01 1.94401219e-01 -6.09586895e-01 4.06661928e-01
6.58938587e-01 -5.44335544e-01 9.92355406e-01 -3.18300538e-02
-2.13601127e-01 -4.23509002e-01 -8.74790967e-01 2.64263421e-01
4.82291549e-01 -5.28014362e-01 -8.35725248e-01 -6.32063895e-02
-5.09114899e-02 -1.47835568e-01 -9.94613945e-01 3.12534720e-01
9.10766244e-01 -1.11020613e+00 7.17039466e-01 -5.48225582e-01
2.71108925e-01 -4.46873963e-01 -3.72981340e-01 -1.32660592e+00
-3.87790501e-02 -4.34792370e-01 1.70151606e-01 1.28500402e+00
1.02280611e-02 -3.22946787e-01 6.15845144e-01 -4.93091464e-01
-2.12689713e-01 -2.71572024e-01 -9.12598789e-01 -8.52169991e-01
-1.00552939e-01 -7.11746514e-01 4.14996088e-01 1.10856044e+00
-2.71278858e-01 4.87307668e-01 -4.94019240e-01 -1.46824881e-01
4.02096599e-01 3.62288028e-01 5.57173491e-01 -1.89717257e+00
-2.96203613e-01 -4.86550301e-01 -8.76229227e-01 -3.17933381e-01
-1.15705155e-01 -9.68276143e-01 -5.22260427e-01 -1.29647744e+00
-9.38917100e-02 1.77160576e-01 -6.77647352e-01 2.21552223e-01
7.75018990e-01 5.86760163e-01 3.01196963e-01 -1.62174515e-02
1.47405401e-01 3.00552487e-01 1.39595711e+00 -2.75693178e-01
-3.03807765e-01 1.93390742e-01 -3.54082763e-01 8.50846529e-01
1.03465092e+00 -4.67880338e-01 -6.02008343e-01 -2.16615453e-01
4.77775633e-02 2.04573363e-01 1.56281367e-01 -1.84175122e+00
3.96883152e-02 8.95102993e-02 8.55167985e-01 -1.08000290e+00
5.86584628e-01 -8.46923053e-01 5.18225193e-01 5.95181704e-01
-3.88000548e-01 1.06359646e-01 3.54645848e-01 1.59527138e-01
-8.18691015e-01 -2.82664359e-01 8.25224698e-01 -7.76462480e-02
-1.01221204e+00 -2.11657416e-02 -4.25971925e-01 -1.97511852e-01
7.87218511e-01 -5.67763984e-01 2.08111718e-01 -4.06803071e-01
-1.09112799e+00 -6.11050844e-01 -4.28034425e-01 3.45969498e-01
5.35447776e-01 -1.63942993e+00 -8.38346720e-01 7.34459534e-02
-2.56896347e-01 -4.56257820e-01 6.35803640e-01 8.70113015e-01
-6.16809726e-01 5.13085186e-01 -7.66970217e-01 -5.07816315e-01
-1.54978526e+00 2.20120117e-01 1.73311368e-01 1.18692748e-01
-7.25048184e-01 6.83544159e-01 -4.32459176e-01 -1.01464264e-01
3.13620865e-01 -3.46244067e-01 -7.47132480e-01 5.22704542e-01
2.30168030e-01 5.03765047e-01 1.49179399e-01 -7.90573061e-01
-3.06530803e-01 1.02243233e+00 1.44962579e-01 -2.73506939e-01
1.51841247e+00 4.49516654e-01 -2.99299508e-02 8.66742671e-01
1.08259106e+00 4.07290868e-02 -4.40855235e-01 5.97370602e-02
1.69278339e-01 -2.42333934e-01 6.17090352e-02 -1.13742864e+00
-1.26183438e+00 1.16995275e+00 1.08891547e+00 1.48439571e-01
1.55641854e+00 -6.71368122e-01 9.40303445e-01 6.30516171e-01
1.02756657e-01 -1.35042632e+00 4.29367535e-02 8.50611627e-01
9.14406598e-01 -5.42890966e-01 -3.08893919e-01 4.24729139e-01
-5.41190445e-01 1.73164988e+00 3.01429600e-01 -5.07066488e-01
6.81126833e-01 1.11823134e-01 2.48277456e-01 -1.26842394e-01
-5.24581313e-01 -5.43002486e-02 4.47777539e-01 5.20612061e-01
9.65979576e-01 -7.66440555e-02 -5.08495033e-01 6.99822783e-01
-9.77745950e-01 2.61955082e-01 3.99359137e-01 7.54419863e-01
-5.69221973e-01 -1.03068185e+00 -7.20667005e-01 4.32841033e-01
-8.74712527e-01 8.04005042e-02 -5.56037962e-01 1.05734766e+00
5.11734068e-01 9.58762050e-01 -2.29012668e-01 -6.89593732e-01
4.35269803e-01 4.19898123e-01 6.93034172e-01 -1.77323148e-01
-9.66127038e-01 3.40757102e-01 -1.52204126e-01 1.27886524e-02
-6.46400213e-01 -4.77801591e-01 -1.40915561e+00 -2.15326846e-01
-1.41449973e-01 4.53721792e-01 8.36562872e-01 8.31346035e-01
-1.39582917e-01 1.04640746e+00 5.23316145e-01 -8.68409693e-01
-3.90610904e-01 -1.45552897e+00 -9.84386563e-01 4.16010588e-01
1.66583598e-01 -5.40561795e-01 -6.59789890e-03 2.07627848e-01]
|
[15.793532371520996, 5.173585891723633]
|
26ccda84-b8f5-4a36-83ad-a56961e9ab7c
|
reasoning-with-language-model-prompting-a
|
2212.09597
| null |
https://arxiv.org/abs/2212.09597v5
|
https://arxiv.org/pdf/2212.09597v5.pdf
|
Reasoning with Language Model Prompting: A Survey
|
Reasoning, as an essential ability for complex problem-solving, can provide back-end support for various real-world applications, such as medical diagnosis, negotiation, etc. This paper provides a comprehensive survey of cutting-edge research on reasoning with language model prompting. We introduce research works with comparisons and summaries and provide systematic resources to help beginners. We also discuss the potential reasons for emerging such reasoning abilities and highlight future research directions. Resources are available at https://github.com/zjunlp/Prompt4ReasoningPapers (updated periodically).
|
['Huajun Chen', 'Fei Huang', 'Chuanqi Tan', 'Shumin Deng', 'Yunzhi Yao', 'Xiang Chen', 'Ningyu Zhang', 'Yixin Ou', 'Shuofei Qiao']
|
2022-12-19
| null | null | null | null |
['mathematical-reasoning', 'arithmetic-reasoning', 'logical-reasoning', 'common-sense-reasoning']
|
['natural-language-processing', 'reasoning', 'reasoning', 'reasoning']
|
[-2.91626394e-01 8.92326653e-01 -9.45424438e-01 -5.25286734e-01
-5.02612770e-01 -6.13470316e-01 5.67764342e-01 4.36289608e-01
-1.27778992e-01 7.88135171e-01 6.81592941e-01 -9.71195757e-01
-3.23529482e-01 -6.63612068e-01 -1.85208656e-02 -1.00510143e-01
2.44328484e-01 7.11889446e-01 -9.15894192e-03 -5.60606480e-01
3.21804136e-01 2.27123380e-01 -9.51509595e-01 7.42564738e-01
8.20509732e-01 3.46816927e-01 -1.80740375e-02 5.90390444e-01
-6.60873771e-01 1.62445974e+00 -5.96141636e-01 -4.53806818e-01
-2.97748059e-01 -1.10642053e-03 -1.42826939e+00 -4.70251888e-01
-2.36021295e-01 -1.90559745e-01 -3.38210106e-01 1.14981925e+00
4.84040588e-01 -3.96318361e-02 9.51672196e-02 -1.61738968e+00
-4.75357085e-01 1.23414314e+00 -2.09184825e-01 2.91175693e-01
1.09802556e+00 8.45362246e-02 7.13598669e-01 -4.01616961e-01
7.95865476e-01 1.41980100e+00 4.14584845e-01 1.06648123e+00
-4.53962833e-01 -8.23529899e-01 4.11154121e-01 4.91275102e-01
-1.15491641e+00 -6.32066131e-01 6.37751877e-01 -1.70204520e-01
1.25881183e+00 8.57461512e-01 3.99005711e-01 8.97735536e-01
3.70384127e-01 9.63747680e-01 1.07866001e+00 -5.20865202e-01
1.16001323e-01 2.62714446e-01 6.51489973e-01 7.87741542e-01
1.55000895e-01 -1.69671729e-01 -5.59408545e-01 -5.65793335e-01
6.73825204e-01 -2.16300994e-01 -8.13829899e-02 3.33080113e-01
-1.21599054e+00 6.48071945e-01 1.94032684e-01 4.51617658e-01
-4.63150680e-01 -5.01616895e-02 4.23379451e-01 6.19620085e-01
9.05416608e-02 6.00020826e-01 -6.12531126e-01 -3.62482578e-01
-3.82341206e-01 6.11484647e-01 1.37840891e+00 1.03567457e+00
5.68244653e-03 -1.80183247e-01 -4.45918053e-01 6.75989091e-01
6.49941802e-01 2.81630188e-01 1.43042549e-01 -1.28389096e+00
2.73714989e-01 6.33994579e-01 3.94290626e-01 -7.72165060e-01
-8.00716639e-01 -3.60477209e-01 -7.19501734e-01 -1.70183271e-01
1.09755442e-01 -4.14630473e-01 2.53942027e-03 1.57632804e+00
7.02681720e-01 -9.74035487e-02 4.70276296e-01 8.02827477e-01
1.60552311e+00 5.75251400e-01 4.37495887e-01 -5.05663157e-01
1.85646403e+00 -1.40308404e+00 -1.14891267e+00 -3.42875063e-01
7.80196369e-01 -1.03554296e+00 9.09886539e-01 4.68961239e-01
-1.44881356e+00 1.19263723e-01 -5.53481519e-01 -2.17480049e-01
-2.03283608e-01 -3.24418932e-01 1.34614038e+00 2.10692838e-01
-1.09674346e+00 -6.77448884e-02 -7.65793085e-01 -4.79919076e-01
4.65709604e-02 -3.78811024e-02 2.57071137e-01 -2.20727295e-01
-1.62230968e+00 1.08418000e+00 3.48267317e-01 -5.87524287e-02
-1.56413108e-01 -7.34890163e-01 -7.46485114e-01 -2.51734227e-01
8.73803318e-01 -1.02438045e+00 2.14238167e+00 -3.38843912e-01
-1.68845928e+00 9.32261348e-01 -3.64727259e-01 -3.70882601e-01
6.90159321e-01 -1.22661814e-01 -7.38728225e-01 1.35649147e-03
7.47187287e-02 3.44352126e-01 -2.13361144e-01 -8.12421679e-01
-5.46663046e-01 7.99972713e-02 8.75999272e-01 3.76633972e-01
1.11490712e-01 8.26007962e-01 -4.27786827e-01 -3.91672462e-01
3.85284275e-02 -4.15042549e-01 -5.38851082e-01 7.15233311e-02
-4.19549018e-01 -6.71573043e-01 2.57743448e-01 -5.06706536e-01
1.76624668e+00 -1.79949820e+00 -1.96394980e-01 1.00775644e-01
3.80986989e-01 2.19536528e-01 -1.93385378e-01 1.25654840e+00
-1.44416451e-01 3.79368186e-01 2.35786289e-01 3.38351160e-01
2.75132179e-01 8.15713778e-03 -4.70146030e-01 -7.21074734e-03
4.64194305e-02 1.17567420e+00 -1.15985262e+00 -6.99212253e-01
1.68744177e-01 1.51583090e-01 -3.11382443e-01 2.67467588e-01
-8.43187153e-01 4.08580124e-01 -7.97662318e-01 7.35044718e-01
5.33587635e-01 -6.99333668e-01 5.22699237e-01 3.92672062e-01
-1.81766033e-01 7.03563929e-01 -1.01532423e+00 1.57235527e+00
-5.25102735e-01 1.10104240e-01 3.66839230e-01 -6.56732559e-01
5.49388409e-01 7.79508173e-01 1.38412923e-01 -6.98828816e-01
-1.87866110e-02 3.54330130e-02 2.26095140e-01 -7.51748562e-01
2.81728804e-01 -1.07008912e-01 -2.83144295e-01 7.41220057e-01
-7.54781961e-01 -3.16087157e-01 4.29915100e-01 5.96364141e-01
9.97003496e-01 -3.27079058e-01 9.23716307e-01 -2.73526460e-01
8.19404721e-01 3.73185992e-01 5.75834751e-01 8.18914831e-01
-7.21444339e-02 -3.70991796e-01 6.77754998e-01 -7.02906489e-01
-3.00886691e-01 -4.55398232e-01 3.37604098e-02 9.95331049e-01
7.27162212e-02 -1.12529612e+00 -4.63802755e-01 -4.39280272e-01
-2.03891560e-01 1.09740651e+00 -2.48005942e-01 5.24736084e-02
-4.21581149e-01 -2.26734340e-01 6.90620065e-01 5.87634146e-01
5.63973606e-01 -1.37870336e+00 -6.10967696e-01 1.05425477e-01
-7.60015070e-01 -1.15828156e+00 1.19006790e-01 -3.92475069e-01
-7.58362889e-01 -1.18532872e+00 6.08196389e-03 -5.53013027e-01
4.45708007e-01 2.75955319e-01 1.19011045e+00 1.04555130e+00
1.37586415e-01 6.23877347e-01 -3.22305888e-01 -6.15917146e-01
-5.89072287e-01 -1.06217287e-01 -1.34766638e-01 -1.08094800e+00
3.10843855e-01 -5.70272990e-02 -2.82260656e-01 2.63807744e-01
-8.13413084e-01 6.49794638e-01 2.10821003e-01 6.08296514e-01
6.73219860e-02 8.87733623e-02 4.90031958e-01 -1.35472381e+00
1.26421416e+00 -6.97265267e-01 -4.76635277e-01 5.98736703e-01
-6.81068361e-01 -4.94836606e-02 4.30721670e-01 -1.93598922e-02
-1.27399182e+00 -1.02139425e+00 -4.55220491e-01 5.45161784e-01
-2.15486348e-01 1.17177212e+00 2.35780969e-01 2.54381627e-01
5.76818109e-01 -3.21601301e-01 -1.47463113e-01 -9.72270817e-02
2.87840277e-01 7.47368693e-01 1.28110945e-01 -1.10166478e+00
3.70928586e-01 2.08791941e-01 -2.78336078e-01 -5.68769395e-01
-8.70828688e-01 -2.90466547e-01 2.75305659e-01 -2.60807842e-01
2.69841433e-01 -7.14014411e-01 -1.11089373e+00 1.28283380e-02
-1.58052635e+00 -8.50276649e-01 -1.72197423e-03 3.26761961e-01
-4.44283187e-01 2.65640110e-01 -1.07232189e+00 -9.71808791e-01
-6.57781243e-01 -7.96743870e-01 5.29096782e-01 5.86414099e-01
-9.28109407e-01 -1.25762963e+00 -8.61019269e-02 8.05514038e-01
5.23685098e-01 -1.52923226e-01 1.18632877e+00 -8.25166643e-01
-2.24897400e-01 8.86916146e-02 -1.64117247e-01 -4.92853820e-01
-1.47673428e-01 1.52431682e-01 -5.94142735e-01 7.37836212e-02
1.55471012e-01 -3.57498437e-01 -5.24306744e-02 2.59043783e-01
1.08746493e+00 -7.00865805e-01 -6.28060102e-01 5.83624467e-02
7.84098268e-01 3.72071773e-01 4.31776226e-01 4.64600027e-01
3.20240371e-02 8.32123578e-01 1.16871190e+00 6.38152421e-01
8.87137234e-01 5.04893243e-01 -2.64062695e-02 9.25916806e-02
-4.05661091e-02 -3.50423413e-03 1.62440781e-02 9.09560621e-01
-2.44289592e-01 -1.98393583e-01 -1.65311563e+00 -2.77796704e-02
-2.31042075e+00 -9.33864832e-01 -3.95906568e-02 1.54103541e+00
1.04477847e+00 1.15812697e-01 -1.78308412e-01 -1.10057369e-01
4.87338245e-01 7.13303089e-02 -4.66650367e-01 -8.48438382e-01
1.21695913e-01 -3.93776558e-02 -1.84951127e-01 1.03192043e+00
-4.84617203e-01 1.27332830e+00 6.67800951e+00 4.62724507e-01
-1.11460543e+00 1.31247997e-01 2.17777878e-01 1.14264823e-01
-6.35913432e-01 3.19921225e-02 -4.24038917e-01 -1.31051928e-01
1.01941097e+00 -8.23534966e-01 5.67206264e-01 6.47042990e-01
5.48639655e-01 -1.78394899e-01 -8.62396777e-01 7.89622307e-01
-2.52590746e-01 -1.59770393e+00 -5.19402623e-02 -4.32884991e-01
3.18526119e-01 -1.88949838e-01 -1.59540445e-01 4.04755592e-01
6.70143723e-01 -8.70893061e-01 6.72085702e-01 4.73253995e-01
3.24232012e-01 -3.93144995e-01 8.53603840e-01 8.04159343e-01
-8.74733806e-01 -2.56293863e-01 1.11163214e-01 -7.61641026e-01
3.24855834e-01 4.16650862e-01 -7.72976637e-01 9.54690397e-01
4.67708290e-01 6.28431797e-01 -9.06558707e-02 8.83341670e-01
-6.38259768e-01 3.36006671e-01 -1.31210685e-01 -2.86129147e-01
-3.05948462e-02 3.99727151e-02 4.62260902e-01 1.25110352e+00
-3.49224061e-01 8.41322303e-01 3.95033121e-01 5.42286217e-01
2.92588383e-01 1.13343149e-01 -2.94453472e-01 -3.75148624e-01
1.00812852e+00 9.70932543e-01 -8.22798967e-01 -7.11204350e-01
-5.05584717e-01 2.86049217e-01 2.53469110e-01 4.33009803e-01
-8.15905273e-01 2.02990696e-02 6.37815893e-01 -7.88991079e-02
-7.26481318e-01 -1.91320062e-01 -3.68029088e-01 -1.25528502e+00
-2.58846972e-02 -1.56209183e+00 8.78658533e-01 -1.05534494e+00
-9.73316371e-01 5.93632936e-01 3.95470768e-01 -7.74848044e-01
-4.07851458e-01 -3.94371152e-01 -5.51139891e-01 5.32412708e-01
-1.58479583e+00 -1.08849299e+00 -3.07630509e-01 3.90874088e-01
6.67334855e-01 2.04975709e-01 1.34485400e+00 1.86269879e-01
-5.22880077e-01 1.18726842e-01 -8.39822352e-01 6.95245564e-02
6.09580278e-01 -8.11098754e-01 4.32954699e-01 6.59574449e-01
-2.41561592e-01 1.25495064e+00 8.34311783e-01 -7.00084925e-01
-1.65321815e+00 -6.30422652e-01 1.35444582e+00 -3.12432230e-01
7.18795598e-01 8.98175910e-02 -7.54837751e-01 9.07086790e-01
5.68712652e-01 -5.03362775e-01 8.35605681e-01 3.06233823e-01
-2.44415611e-01 2.06250250e-01 -1.17523324e+00 1.05077267e+00
1.10396612e+00 -4.61266547e-01 -7.78035522e-01 7.78348505e-01
7.83342123e-01 -1.00785828e+00 -7.46692121e-01 2.17293650e-01
6.51811242e-01 -7.36137509e-01 8.42473269e-01 -9.27455664e-01
4.18619066e-01 -1.93173543e-01 2.00439245e-01 -7.71064043e-01
-3.79591554e-01 -9.61272359e-01 -3.41701090e-01 8.79954100e-01
4.60026324e-01 -1.20768678e+00 4.44458574e-01 1.32350862e+00
-6.18785545e-02 -9.80501473e-01 -3.37601125e-01 -3.85182947e-01
6.63138330e-02 -7.43015766e-01 7.49074221e-01 1.21844947e+00
1.04039204e+00 2.89028287e-01 1.40856281e-01 2.18178272e-01
2.31173024e-01 2.47271657e-01 5.19700229e-01 -1.01492274e+00
-2.45915592e-01 -6.27955198e-01 2.81898528e-01 -1.01234508e+00
3.77182543e-01 -9.98821080e-01 -4.74089712e-01 -2.31019616e+00
9.81306061e-02 -5.71789265e-01 -1.24202743e-01 1.02154946e+00
6.06156513e-02 -4.66729432e-01 1.42446607e-01 1.19510256e-02
-8.22349906e-01 -2.03579590e-01 1.40784335e+00 -1.14979871e-01
-6.81460723e-02 1.68934315e-01 -1.29281104e+00 8.78830492e-01
1.36846054e+00 -3.56395364e-01 -6.32558644e-01 -5.23721516e-01
8.17359090e-01 6.92329705e-01 -6.76821396e-02 -4.71806824e-01
7.21782088e-01 -1.19299364e+00 -3.96960467e-01 -1.59555346e-01
1.15110926e-01 -7.58887410e-01 3.06716263e-01 9.44591939e-01
-7.32474446e-01 4.09113556e-01 5.28556108e-01 -2.06964895e-01
-1.59741372e-01 -3.25344443e-01 2.22002849e-01 -2.21086979e-01
-8.05550456e-01 -1.70957536e-01 -7.56314695e-01 1.81463897e-01
1.05835497e+00 3.52536231e-01 -7.50861824e-01 -7.22168148e-01
-3.73240501e-01 9.01144743e-01 2.04200193e-01 6.10728025e-01
4.98729795e-01 -8.94898236e-01 -6.95574999e-01 -1.82025626e-01
1.65335938e-01 -6.43740892e-02 4.19757903e-01 8.94937694e-01
-6.32502735e-01 7.87884355e-01 -9.83838066e-02 -6.53946251e-02
-1.43730795e+00 5.74996650e-01 1.81106374e-01 -3.72522563e-01
-5.53419113e-01 7.84961224e-01 -2.24791169e-01 -5.81878126e-01
3.71594995e-01 -4.24826622e-01 -2.71417528e-01 -5.17006159e-01
9.08665657e-01 1.78857833e-01 -1.57178476e-01 8.54844674e-02
-7.69078255e-01 1.01883061e-01 -2.33751014e-01 -2.54712492e-01
1.07086766e+00 -1.39533952e-01 -5.96455038e-01 4.21486318e-01
6.84522316e-02 -2.29177717e-02 -9.33619291e-02 -3.69797260e-01
3.56876969e-01 -1.67478189e-01 -2.56073892e-01 -1.22329140e+00
-5.88566005e-01 3.99929553e-01 -3.09440941e-01 3.67749095e-01
8.24002683e-01 2.58638859e-01 2.85401702e-01 7.08928466e-01
7.11441875e-01 -8.47320497e-01 -2.47486085e-01 6.24777198e-01
1.30884767e+00 -9.85889494e-01 2.02194989e-01 -9.64225709e-01
-7.82645524e-01 1.19404590e+00 7.02152848e-01 3.56402457e-01
6.43418252e-01 8.64880383e-01 7.70339429e-01 -4.31814432e-01
-1.42940152e+00 3.20700437e-01 -2.35871390e-01 4.76776451e-01
9.05529261e-01 2.39152834e-01 -7.95274258e-01 8.77158403e-01
-4.04392958e-01 4.11383986e-01 5.69585800e-01 1.34005845e+00
1.86236408e-02 -1.58310938e+00 -4.72088009e-01 1.98239625e-01
-3.38862032e-01 -2.39681527e-01 -5.69426119e-01 6.26862049e-01
-5.36038935e-01 1.51088405e+00 -2.51756996e-01 -4.03070860e-02
2.99923360e-01 6.89297169e-02 3.79521728e-01 -7.16764569e-01
-9.11182821e-01 -3.56866777e-01 8.60374928e-01 -6.92816973e-01
-5.86124599e-01 -3.75592172e-01 -1.66351426e+00 -7.67292798e-01
8.94276053e-02 5.55259645e-01 2.94584453e-01 8.11810553e-01
2.79839158e-01 6.36821330e-01 -1.39599741e-01 9.88115743e-02
-3.89293760e-01 -7.48927295e-01 4.52812426e-02 -2.40282997e-01
6.52503222e-02 -4.07804102e-01 6.27359599e-02 -2.93663859e-01]
|
[9.50910758972168, 7.4019389152526855]
|
b5738911-95bf-4de7-8f38-fb19f766218d
|
adaptive-multi-view-ica-estimation-of-noise
|
2102.10964
| null |
https://arxiv.org/abs/2102.10964v1
|
https://arxiv.org/pdf/2102.10964v1.pdf
|
Adaptive Multi-View ICA: Estimation of noise levels for optimal inference
|
We consider a multi-view learning problem known as group independent component analysis (group ICA), where the goal is to recover shared independent sources from many views. The statistical modeling of this problem requires to take noise into account. When the model includes additive noise on the observations, the likelihood is intractable. By contrast, we propose Adaptive multiView ICA (AVICA), a noisy ICA model where each view is a linear mixture of shared independent sources with additive noise on the sources. In this setting, the likelihood has a tractable expression, which enables either direct optimization of the log-likelihood using a quasi-Newton method, or generalized EM. Importantly, we consider that the noise levels are also parameters that are learned from the data. This enables sources estimation with a closed-form Minimum Mean Squared Error (MMSE) estimator which weights each view according to its relative noise level. On synthetic data, AVICA yields better sources estimates than other group ICA methods thanks to its explicit MMSE estimator. On real magnetoencephalograpy (MEG) data, we provide evidence that the decomposition is less sensitive to sampling noise and that the noise variance estimates are biologically plausible. Lastly, on functional magnetic resonance imaging (fMRI) data, AVICA exhibits best performance in transferring information across views.
|
['Bertrand Thirion', 'Alexandre Gramfort', 'Aapo Hyvärinen', 'Pierre Ablin', 'Hugo Richard']
|
2021-02-22
| null | null | null | null |
['multi-view-learning']
|
['computer-vision']
|
[ 9.09548774e-02 6.14663959e-02 9.96588320e-02 -3.70161325e-01
-1.04970300e+00 -5.42125881e-01 5.51941693e-01 -5.48672915e-01
-2.42331237e-01 6.51834428e-01 5.99324882e-01 2.06405431e-01
-2.94767022e-01 -3.17118078e-01 -9.10639107e-01 -9.69265521e-01
-2.18607321e-01 2.60439456e-01 -5.56884646e-01 3.48270088e-01
-1.30565092e-01 -3.89590152e-02 -1.26994872e+00 1.97828300e-02
8.21293592e-01 8.63814056e-01 3.22974533e-01 5.50874114e-01
3.87234747e-01 7.79021621e-01 -5.55903077e-01 -1.99273497e-01
2.53905773e-01 -7.72580266e-01 -2.64670163e-01 2.81236053e-01
3.00910264e-01 -2.10175365e-01 -9.43230540e-02 1.21968865e+00
4.39858228e-01 2.47604623e-02 8.84383559e-01 -1.37762880e+00
-6.77339077e-01 5.92959642e-01 -7.30535328e-01 2.18152508e-01
1.95970729e-01 -3.63381833e-01 7.79023349e-01 -1.14633429e+00
6.20312333e-01 1.28284216e+00 5.03229916e-01 2.06407040e-01
-1.51580143e+00 -6.69358552e-01 4.66574311e-01 2.48952150e-01
-1.36353135e+00 -7.72068560e-01 7.98312604e-01 -5.58256388e-01
5.75905621e-01 -6.16324553e-03 4.63208497e-01 1.44790995e+00
5.10322094e-01 6.64746583e-01 1.37608528e+00 -2.73171335e-01
7.37975955e-01 1.23118937e-01 3.19063663e-02 3.62155259e-01
3.19722086e-01 -9.94972810e-02 -8.98792624e-01 -4.75756347e-01
7.10319877e-01 -9.00262147e-02 -2.91035414e-01 -6.07262135e-01
-1.41359913e+00 7.96957433e-01 4.05195877e-02 2.36034229e-01
-7.46652067e-01 2.03568727e-01 6.11563176e-02 2.60354012e-01
6.68739200e-01 1.12615740e-02 -3.35290968e-01 3.60563695e-01
-8.52143109e-01 -1.77663714e-01 8.54285836e-01 7.97117233e-01
4.61434454e-01 4.25896138e-01 4.71806139e-01 8.14205527e-01
5.98030508e-01 1.08531284e+00 5.11360705e-01 -1.34026015e+00
3.69651139e-01 -5.48102073e-02 1.24128442e-02 -9.91691709e-01
-4.41475660e-01 -6.22567952e-01 -1.10505974e+00 2.93017209e-01
4.54068661e-01 -4.44644213e-01 -7.56637812e-01 2.38846421e+00
5.38060293e-02 3.16004843e-01 1.16191611e-01 8.48339260e-01
3.64913523e-01 3.95552188e-01 -8.00669119e-02 -8.23223770e-01
1.21797204e+00 -4.91264462e-01 -1.15816748e+00 -6.45518601e-01
6.04864322e-02 -3.59900147e-01 3.24588001e-01 7.85243869e-01
-1.35536563e+00 -3.21239531e-01 -1.07781410e+00 3.34985286e-01
-6.60456764e-03 1.92876335e-03 5.53968966e-01 7.45410621e-01
-1.05739653e+00 8.56710523e-02 -1.06749547e+00 -1.47599518e-01
3.27275068e-01 1.36472240e-01 -6.97901547e-01 -3.42559218e-02
-6.77832961e-01 8.49098980e-01 -1.89200819e-01 5.48092909e-02
-1.08022726e+00 -6.06919348e-01 -9.69358027e-01 1.32324500e-02
3.23521405e-01 -9.91949081e-01 6.70315385e-01 -1.33341444e+00
-1.35190117e+00 5.03612816e-01 -6.19564593e-01 -1.72603384e-01
2.02370137e-01 -1.07974015e-01 -4.50715899e-01 4.52495426e-01
3.44733864e-01 3.21964890e-01 1.47130275e+00 -1.50698411e+00
1.53182089e-01 -7.94213593e-01 -4.30976182e-01 2.13972121e-01
2.05623582e-02 6.01463392e-02 -1.50986046e-01 -7.16876149e-01
7.84850001e-01 -9.18602705e-01 -1.55658171e-01 -3.72631997e-02
-2.08023459e-01 4.37567681e-01 2.35188439e-01 -1.03785133e+00
6.48619115e-01 -2.03650618e+00 7.24777579e-01 3.96393120e-01
3.72858167e-01 -6.37902260e-01 -1.59783900e-01 2.16918916e-01
-3.52309406e-01 -6.17516376e-02 -3.83049339e-01 -5.12954533e-01
-1.67806074e-01 -1.03419693e-02 -1.86275601e-01 9.68826473e-01
-2.01302469e-01 7.89138496e-01 -5.22932470e-01 -1.42646000e-01
-2.95422040e-02 4.21345204e-01 -6.05951130e-01 3.03910468e-02
3.76236677e-01 8.17126691e-01 -1.52151614e-01 3.68626207e-01
6.85238242e-01 -5.89380026e-01 4.18084770e-01 -3.19648296e-01
2.45037615e-01 -2.00991407e-01 -1.49040318e+00 1.82188189e+00
-4.82546747e-01 5.22166729e-01 7.19814062e-01 -1.11122012e+00
3.35434735e-01 6.36682332e-01 6.06422961e-01 -4.40208405e-01
1.57317072e-01 1.24531731e-01 2.13032201e-01 -2.80422449e-01
-1.79438829e-01 -3.83549601e-01 2.75928173e-02 7.90341258e-01
7.44602919e-01 8.57733861e-02 -3.31830829e-01 5.99306405e-01
9.18286502e-01 -1.61107793e-01 5.03346741e-01 -3.48108530e-01
7.91958794e-02 -6.02217793e-01 5.20432591e-01 9.65770006e-01
-4.62803282e-02 8.49313915e-01 2.95978189e-01 2.22336352e-01
-8.27685893e-01 -1.39915264e+00 -1.58636764e-01 6.66032195e-01
-2.45993793e-01 -1.33088201e-01 -7.59690166e-01 -2.89462268e-01
-3.29390854e-01 6.52157426e-01 -6.53026521e-01 -4.19618785e-02
-2.23643765e-01 -1.19603443e+00 1.61899194e-01 5.34082353e-01
2.26031706e-01 -4.40887630e-01 -2.81113088e-01 2.85479605e-01
-5.61152101e-01 -1.14802372e+00 -2.97759295e-01 3.19201350e-01
-9.14997756e-01 -8.40830386e-01 -7.77325571e-01 -2.11332470e-01
7.28943050e-01 3.73392135e-01 8.27512980e-01 -5.08386850e-01
4.13979813e-02 1.06814647e+00 1.90269724e-02 -5.16668677e-01
-9.03502032e-02 -6.03062391e-01 4.92496789e-01 3.76822323e-01
3.71397361e-02 -1.11798537e+00 -3.80312860e-01 2.18944699e-01
-6.99281394e-01 -9.39137116e-02 3.49017471e-01 7.24404573e-01
6.27395630e-01 -2.66867727e-01 1.00196767e+00 -4.87667382e-01
5.16257763e-01 -8.62535059e-01 -4.15052652e-01 2.40639433e-01
-1.93705842e-01 -7.72393346e-02 4.02096152e-01 -5.18073440e-01
-1.28720891e+00 2.97024869e-03 2.71989703e-01 -6.12108648e-01
-2.04926789e-01 5.64954221e-01 -5.01715124e-01 7.51102949e-03
6.54155910e-01 3.11176181e-01 1.48020402e-01 -4.85500872e-01
5.73362827e-01 2.37444744e-01 4.57012862e-01 -2.74614513e-01
3.62355232e-01 8.50980639e-01 -7.85453767e-02 -9.48726237e-01
-7.40837395e-01 -1.85699016e-01 -6.69067383e-01 -2.41032586e-01
8.30774069e-01 -1.46441555e+00 -4.98667508e-01 5.42147994e-01
-1.01253581e+00 -3.18034142e-02 3.24646235e-02 1.04121625e+00
-9.23369288e-01 2.50318110e-01 -5.12859821e-01 -9.09610212e-01
2.80389767e-02 -9.96260285e-01 7.99609125e-01 -1.58710212e-01
-1.88201874e-01 -1.21657956e+00 -8.51900578e-02 3.43161672e-01
2.72682637e-01 7.58279786e-02 7.26896286e-01 -7.08150566e-01
-3.84838253e-01 -9.13563184e-03 1.33823767e-01 3.90809357e-01
1.65642664e-01 -5.35229027e-01 -1.33188736e+00 -1.63537726e-01
9.46035504e-01 -8.03131238e-02 8.42740417e-01 1.13251877e+00
9.36732411e-01 -2.48011932e-01 -2.39851326e-01 6.79061055e-01
1.29932988e+00 1.65708601e-01 2.66108602e-01 -2.21643925e-01
6.17653668e-01 7.08319187e-01 -1.42743304e-01 4.95454609e-01
5.31755388e-01 3.73411149e-01 2.13725001e-01 2.20177710e-01
2.51764935e-02 -1.10488132e-01 4.85789984e-01 1.29491794e+00
4.19665277e-02 -3.56415391e-01 -7.16365457e-01 3.80204916e-01
-1.80114937e+00 -1.08555996e+00 -2.12505743e-01 2.31266022e+00
2.64900565e-01 -3.31527978e-01 -1.39779195e-01 -2.02679887e-01
5.07920206e-01 3.65869403e-02 -7.89898932e-01 1.64474934e-01
-5.52970231e-01 -9.82872620e-02 3.83667111e-01 6.15685701e-01
-1.00743878e+00 2.43777797e-01 7.53011179e+00 4.64648813e-01
-6.35835350e-01 6.63726091e-01 6.08952522e-01 -3.89316291e-01
-3.61527264e-01 -1.38119817e-01 -2.80877709e-01 5.04137695e-01
1.15543759e+00 -3.66264910e-01 7.57490218e-01 4.73770618e-01
3.26322496e-01 -6.33104146e-01 -1.05264008e+00 1.12655306e+00
6.18584156e-01 -9.04136360e-01 -2.87919462e-01 2.80956864e-01
7.72347033e-01 1.66773766e-01 8.24730843e-02 -1.36337563e-01
4.16148812e-01 -7.46434271e-01 9.03645754e-01 8.94518495e-01
3.75452280e-01 -6.30670726e-01 3.59481394e-01 6.88617766e-01
-7.84929514e-01 -1.37773126e-01 -4.31842119e-01 -5.59938252e-02
5.97051859e-01 9.38739598e-01 1.31247947e-02 4.06188667e-01
5.66947877e-01 7.61488140e-01 -3.28242153e-01 7.85725832e-01
-2.51628131e-01 6.51378095e-01 -4.36350971e-01 8.16590369e-01
-3.80788296e-01 -6.43911779e-01 8.72398615e-01 7.38323748e-01
5.12606978e-01 3.34781140e-01 -1.80376545e-02 1.12647486e+00
6.44410551e-02 -5.20661734e-02 -6.92706704e-01 3.27224672e-01
3.65902722e-01 1.19309461e+00 -8.19620728e-01 -4.38530713e-01
-5.44037819e-01 9.69730914e-01 2.62806028e-01 8.77398789e-01
-5.35352170e-01 3.00879389e-01 4.99472976e-01 -4.78588521e-01
2.53807813e-01 -2.72671163e-01 -3.97538364e-01 -1.73057127e+00
9.44625512e-02 -9.31703210e-01 9.38714221e-02 -1.05400336e+00
-1.69561338e+00 3.49636674e-01 3.69104713e-01 -1.15414929e+00
-4.77341831e-01 -3.13928813e-01 -4.87531066e-01 1.00879562e+00
-8.86327147e-01 -9.92289960e-01 1.41562298e-01 7.26411104e-01
3.70151281e-01 -1.46589935e-01 6.67793572e-01 1.60896033e-01
-5.27780890e-01 1.79598987e-01 3.20683628e-01 -2.70670265e-01
7.71621287e-01 -1.00684190e+00 4.62648943e-02 9.06898141e-01
2.83803374e-01 8.56932878e-01 7.17506051e-01 -7.07933724e-01
-1.55647087e+00 -6.59912348e-01 3.98023278e-01 -4.97763693e-01
7.43560374e-01 -6.38274074e-01 -9.98378992e-01 1.09375322e+00
3.72619867e-01 2.93495595e-01 1.08819199e+00 1.23761572e-01
-4.04302508e-01 3.35969180e-01 -1.09767318e+00 1.06171340e-01
9.97650862e-01 -5.15302598e-01 -5.35081208e-01 2.72747427e-01
2.84495384e-01 3.37716378e-02 -8.40787590e-01 -7.42381439e-02
6.46081805e-01 -1.05078220e+00 9.96077120e-01 -5.58633268e-01
1.55646160e-01 -1.43402815e-01 -5.84064662e-01 -1.86142719e+00
-5.12781620e-01 -1.41100600e-01 -2.70265877e-01 1.09907997e+00
3.14930767e-01 -8.42324853e-01 1.95277974e-01 7.75025725e-01
1.43553674e-01 -9.52772126e-02 -1.01293004e+00 -9.50603426e-01
5.55886589e-02 -7.44870245e-01 5.62390350e-02 1.00906444e+00
1.34040404e-03 2.90234327e-01 -6.51969731e-01 4.95288014e-01
1.23945749e+00 -1.57554209e-01 2.65945196e-01 -1.14173269e+00
-5.28945923e-01 8.94129872e-02 -1.47468060e-01 -7.79922485e-01
5.36849201e-01 -8.69383931e-01 2.30286177e-02 -1.33766556e+00
4.91696149e-01 1.07383072e-01 -7.16268122e-02 1.25881702e-01
-1.22854210e-01 6.70658574e-02 4.01471734e-01 2.65086144e-01
-5.40018916e-01 6.05119586e-01 7.23585427e-01 8.74053761e-02
-7.49646053e-02 -8.89299735e-02 -8.12069893e-01 1.22722793e+00
5.43019235e-01 -7.33165801e-01 -4.60093141e-01 -4.59330618e-01
3.76722366e-01 5.37326694e-01 5.62531352e-01 -8.18218768e-01
4.04604614e-01 1.19682521e-01 7.47304022e-01 -3.21351588e-01
6.33661985e-01 -7.72360981e-01 6.34220660e-01 -3.18655781e-02
-1.23399191e-01 -1.45363901e-02 7.75753707e-02 9.47998464e-01
-2.07551852e-01 -1.86176941e-01 5.68078518e-01 -3.55641395e-01
-1.88033938e-01 -2.93263979e-02 -9.70824480e-01 1.27543077e-01
6.29243612e-01 3.04788388e-02 -8.78382102e-02 -9.58537579e-01
-1.20132315e+00 -8.53354633e-02 2.10995391e-01 1.58022061e-01
5.71694434e-01 -1.39864671e+00 -7.39196479e-01 2.96860993e-01
-1.58417493e-01 -6.39976084e-01 6.07168734e-01 1.33361912e+00
4.67190415e-01 5.20867147e-02 -9.62254405e-02 -6.19289637e-01
-9.27836478e-01 6.18532002e-01 3.81833732e-01 2.10495144e-01
-3.79951566e-01 7.51132011e-01 7.19918072e-01 -3.18606853e-01
-1.17219813e-01 9.37104598e-02 -1.20284602e-01 5.65758646e-01
9.01618838e-01 6.53587461e-01 -1.22449724e-02 -9.07233000e-01
-4.30019081e-01 5.11937022e-01 3.47108483e-01 -7.16408193e-01
1.41850901e+00 -6.91619992e-01 -3.26989233e-01 9.02672052e-01
1.20961368e+00 1.71949372e-01 -1.26849508e+00 -3.26005638e-01
-2.55600840e-01 -2.19223246e-01 3.31824243e-01 -7.32507825e-01
-1.28166783e+00 7.90015459e-01 5.87020218e-01 -1.85841337e-01
1.14220083e+00 6.40953481e-02 -1.70617774e-01 2.34405175e-01
6.90013826e-01 -7.91436076e-01 -1.02911212e-01 2.56276935e-01
1.11668718e+00 -1.28677070e+00 7.92265162e-02 -2.87458003e-01
-7.51575291e-01 8.31949711e-01 1.54607102e-01 -2.80129403e-01
1.07641649e+00 4.11511779e-01 -1.29114911e-01 -2.09190279e-01
-9.15097117e-01 7.48384818e-02 3.28245759e-01 6.55041873e-01
5.42924181e-02 1.23289272e-01 2.20002662e-02 1.14827335e+00
3.78171764e-02 -2.35711455e-01 7.35323787e-01 5.87098360e-01
-2.48778895e-01 -6.13721550e-01 -7.87648916e-01 6.34808004e-01
-5.63065588e-01 -1.87610313e-01 -1.32783771e-01 3.81651342e-01
-1.71598747e-01 1.31597114e+00 1.83158256e-02 1.16739221e-01
9.83103812e-02 3.51488888e-01 6.41647518e-01 -5.26892900e-01
-1.81911394e-01 7.24565446e-01 -2.29589447e-01 -7.36237288e-01
-8.11592638e-01 -1.01574814e+00 -8.08711052e-01 1.26674965e-01
-4.15010870e-01 -1.05479270e-01 9.22921658e-01 1.07500291e+00
4.23527867e-01 4.61689264e-01 4.67753679e-01 -1.05505466e+00
-3.36514145e-01 -9.20208514e-01 -1.01309192e+00 1.85469374e-01
3.32890779e-01 -7.71523058e-01 -8.69279563e-01 3.01291198e-01]
|
[7.757248401641846, 4.403214931488037]
|
8069a57f-e413-4c86-a954-23bb8fcd1d0c
|
learning-bone-suppression-from-dual-energy
|
1811.02628
| null |
http://arxiv.org/abs/1811.02628v1
|
http://arxiv.org/pdf/1811.02628v1.pdf
|
Learning Bone Suppression from Dual Energy Chest X-rays using Adversarial Networks
|
Suppressing bones on chest X-rays such as ribs and clavicle is often expected
to improve pathologies classification. These bones can interfere with a broad
range of diagnostic tasks on pulmonary disease except for musculoskeletal
system. Current conventional method for acquisition of bone suppressed X-rays
is dual energy imaging, which captures two radiographs at a very short interval
with different energy levels; however, the patient is exposed to radiation
twice and the artifacts arise due to heartbeats between two shots. In this
paper, we introduce a deep generative model trained to predict bone suppressed
images on single energy chest X-rays, analyzing a finite set of previously
acquired dual energy chest X-rays. Since the relatively small amount of data is
available, such approach relies on the methodology maximizing the data
utilization. Here we integrate the following two approaches. First, we use a
conditional generative adversarial network that complements the traditional
regression method minimizing the pairwise image difference. Second, we use Haar
2D wavelet decomposition to offer a perceptual guideline in frequency details
to allow the model to converge quickly and efficiently. As a result, we achieve
state-of-the-art performance on bone suppression as compared to the existing
approaches with dual energy chest X-rays.
|
['Dong Yul Oh', 'Il Dong Yun']
|
2018-11-05
| null | null | null | null |
['bone-suppression-from-dual-energy-chest-x']
|
['medical']
|
[ 5.31338036e-01 6.35117590e-02 -2.17391048e-02 5.50167561e-02
-1.09927928e+00 -1.94826052e-01 3.27181458e-01 -2.19574451e-01
-2.50200659e-01 7.30178535e-01 2.13972166e-01 -9.58540887e-02
-2.38016903e-01 -9.79428411e-01 -6.49732769e-01 -1.00136900e+00
3.44286919e-01 4.00496006e-01 3.45602393e-01 -7.24753663e-02
-2.09434077e-01 4.20285970e-01 -1.15442359e+00 3.92837793e-01
4.82882082e-01 6.72511160e-01 2.24315315e-01 8.88623476e-01
2.62286872e-01 7.92940617e-01 -5.34424186e-01 -3.16242069e-01
3.24211091e-01 -8.89864504e-01 -5.57585657e-01 5.21993116e-02
-4.78841960e-02 -5.20285428e-01 -5.00294268e-01 8.64165127e-01
7.38100827e-01 2.32201785e-01 6.43926263e-01 -8.09751809e-01
-6.20100200e-01 4.71988171e-01 -1.15956163e+00 5.32745659e-01
2.64839202e-01 1.76755264e-01 3.43645900e-01 -7.50631630e-01
3.99219066e-01 7.51321971e-01 8.82488012e-01 8.37470174e-01
-1.09892631e+00 -6.35465026e-01 -5.28293371e-01 4.81280945e-02
-1.12161887e+00 -4.40998785e-02 9.78301704e-01 -1.96324661e-01
5.05764246e-01 6.46534801e-01 8.45624924e-01 1.31964839e+00
7.80053437e-01 3.97378504e-01 1.25743186e+00 -5.68532646e-01
-4.66799922e-02 -3.56418878e-01 -4.89961430e-02 8.26951981e-01
1.63411219e-02 2.91723311e-01 -6.27599597e-01 -2.87084579e-01
8.32840085e-01 5.85486352e-01 -5.94760954e-01 -9.21987966e-02
-1.11562634e+00 7.19196439e-01 2.90922135e-01 4.37738538e-01
-6.78482115e-01 3.51748168e-01 2.06538439e-01 3.30195129e-02
3.67729455e-01 -5.57596609e-02 2.29187503e-01 2.49173313e-01
-1.06787884e+00 1.88936144e-01 2.29373544e-01 2.19216451e-01
4.87038903e-02 3.45847942e-02 -4.51260448e-01 8.43968332e-01
3.65444005e-01 4.72718894e-01 7.19623923e-01 -7.06987441e-01
1.93874195e-01 3.96984350e-03 -3.72927606e-01 -5.30314267e-01
-1.99440569e-01 -6.02553904e-01 -1.02088034e+00 5.20273864e-01
2.89289057e-01 1.23128839e-01 -1.36356175e+00 1.72906101e+00
5.37972569e-01 3.49619478e-01 -1.32089108e-01 1.05098855e+00
7.98937798e-01 4.75167215e-01 1.07202090e-01 -6.04068160e-01
1.56987774e+00 -8.57941806e-01 -1.04557252e+00 -8.96436721e-02
-1.68002516e-01 -1.03337359e+00 1.08425832e+00 4.87663746e-01
-1.51126158e+00 -5.26003957e-01 -1.15757239e+00 5.02640493e-02
2.20702976e-01 -3.08076203e-01 2.13872358e-01 7.35579789e-01
-7.88635910e-01 6.68799877e-01 -1.35854542e+00 6.17973357e-02
3.00067008e-01 3.99002314e-01 -1.07059486e-01 -1.46796867e-01
-1.23421681e+00 8.51600766e-01 -2.04539135e-01 -7.77421892e-02
-8.42765570e-01 -1.05869818e+00 -5.67848384e-01 -1.50178954e-01
4.38015938e-01 -1.13091099e+00 1.15311408e+00 -5.22000849e-01
-1.53563786e+00 8.94185722e-01 1.30621135e-01 -3.10682952e-01
8.22906375e-01 -3.00992757e-01 -2.94690758e-01 5.55078268e-01
2.63134181e-01 3.44386816e-01 1.07064879e+00 -1.23743820e+00
-1.30190447e-01 -5.51377594e-01 -3.83030772e-01 2.81294972e-01
1.49900131e-02 -1.01638436e-01 -5.11059284e-01 -1.15310121e+00
4.35653806e-01 -1.10346973e+00 -2.59614676e-01 -5.37089035e-02
-4.26570535e-01 3.45943063e-01 7.97306120e-01 -9.65385616e-01
1.08649349e+00 -1.96869409e+00 2.71836400e-01 1.00511521e-01
4.63850558e-01 -2.27415577e-01 4.16881323e-01 1.31844729e-01
-3.48735899e-01 4.02622074e-02 -5.74755371e-01 -3.64568412e-01
-4.72766638e-01 1.75317198e-01 -2.65934050e-01 6.89586461e-01
-2.50870645e-01 7.92286873e-01 -5.89357436e-01 -6.09667838e-01
3.11867237e-01 7.51188695e-01 -3.52631986e-01 2.26383224e-01
2.73133874e-01 8.41587663e-01 -3.43998253e-01 6.26725078e-01
6.92571759e-01 -1.22779518e-01 -1.39047951e-01 -2.34019756e-01
2.46256009e-01 -2.11542502e-01 -7.83276379e-01 1.96371222e+00
-5.14114082e-01 5.15190586e-02 5.84160537e-02 -5.30143142e-01
3.49665046e-01 6.97308719e-01 1.00848699e+00 -5.85957766e-01
6.62490353e-02 1.33691043e-01 -5.42610958e-02 -4.13528651e-01
-8.11168924e-02 -1.06119263e+00 1.20132372e-01 7.14800596e-01
-2.92479277e-01 -5.85510850e-01 -1.55190200e-01 3.31173651e-02
1.37896609e+00 -4.77414280e-02 2.37434030e-01 9.34894457e-02
3.55373770e-01 -3.23879331e-01 2.68374413e-01 6.69877887e-01
-7.18210265e-02 1.24934733e+00 3.68051901e-02 -2.43893445e-01
-8.84210587e-01 -1.54272640e+00 -3.00344408e-01 6.64350271e-01
1.05507962e-01 1.35091156e-01 -7.55815625e-01 -5.98417699e-01
-3.77971917e-01 7.04654515e-01 -7.45422244e-01 -3.99072766e-01
-9.40580130e-01 -1.13418639e+00 6.14904821e-01 6.57752752e-01
3.54023099e-01 -1.01769674e+00 -1.11293316e+00 2.11460039e-01
-3.02479833e-01 -6.67831838e-01 -6.15117967e-01 3.52033198e-01
-8.91451120e-01 -1.00131190e+00 -1.22890556e+00 -3.52552652e-01
3.73546958e-01 2.22557545e-01 1.19239473e+00 2.13770032e-01
-6.63054883e-01 5.56458056e-01 -2.79536456e-01 -3.79484951e-01
-5.69685638e-01 -4.34466600e-01 -9.15598646e-02 -1.59685344e-01
-6.91395476e-02 -9.10443246e-01 -1.00298834e+00 1.87413581e-02
-1.17395818e+00 8.93395469e-02 7.53817558e-01 1.13002133e+00
1.20809960e+00 1.27846912e-01 3.18941236e-01 -9.67843294e-01
6.35508239e-01 -6.10542774e-01 4.76028547e-02 2.56444991e-01
-5.05104423e-01 -1.62286073e-01 6.49225354e-01 -3.84378433e-01
-1.19939876e+00 -1.27158910e-01 -3.35388273e-01 -8.34568501e-01
-6.86371059e-04 7.59674534e-02 3.50489646e-01 -1.78630680e-01
6.04450941e-01 3.99611264e-01 1.04611181e-02 -1.61448881e-01
6.24491721e-02 1.05574518e-01 9.18653905e-01 -2.04623386e-01
6.84010506e-01 7.53196180e-01 3.44346553e-01 -5.31305432e-01
-8.75008821e-01 -2.50764698e-01 -3.44601303e-01 -4.71680194e-01
1.31903887e+00 -4.77612674e-01 -4.79897350e-01 2.05251709e-01
-8.04565668e-01 2.65056267e-02 -6.14964426e-01 8.70581448e-01
-6.88104153e-01 6.16253316e-01 -9.74447310e-01 -5.25882423e-01
-6.70916140e-01 -1.25162327e+00 1.14080870e+00 6.31711483e-02
-1.80170476e-01 -7.85509825e-01 4.61403430e-01 5.85247040e-01
2.23857567e-01 6.95512056e-01 9.91948724e-01 -1.88225463e-01
-4.60108221e-01 -5.30849136e-02 3.56604010e-01 2.27015689e-01
2.73256153e-01 -3.35845232e-01 -7.87761688e-01 -3.09028357e-01
1.03375447e+00 -3.47281694e-01 8.27863038e-01 9.58478034e-01
1.46424663e+00 1.99356139e-01 -1.90708667e-01 9.38171029e-01
1.47209811e+00 4.34293330e-01 6.15004838e-01 -2.50900108e-02
4.99495536e-01 2.94090301e-01 5.40113688e-01 2.93696105e-01
-2.46016890e-01 4.73470211e-01 3.93555075e-01 -3.98456156e-01
-5.39396703e-01 -7.43848905e-02 -1.93732738e-01 8.36163342e-01
-3.11871946e-01 -3.20646882e-01 -9.83778834e-01 3.72191608e-01
-1.28780544e+00 -1.01002121e+00 -3.33326757e-01 2.06791782e+00
8.36592197e-01 7.94984773e-02 -2.07024649e-01 3.57622713e-01
7.55266905e-01 1.64385796e-01 -4.77084965e-01 1.95191875e-01
2.43019983e-01 9.32570934e-01 4.64646161e-01 4.65617836e-01
-8.34555626e-01 -4.91760746e-02 6.82102776e+00 8.84035051e-01
-1.17282414e+00 6.57011271e-01 6.56998277e-01 -5.11286259e-01
-4.65359539e-01 -3.04557085e-01 -1.51327282e-01 6.31626427e-01
6.89624608e-01 8.76688585e-02 1.67544872e-01 4.15008307e-01
9.86368284e-02 -4.00931060e-01 -8.39096129e-01 9.94841516e-01
1.32407263e-01 -1.12907767e+00 -1.76312789e-01 -3.68430861e-03
5.47564447e-01 -2.48821363e-01 3.66819620e-01 -5.04368357e-02
-2.39933524e-02 -1.12830484e+00 5.61277151e-01 6.47218287e-01
1.00074112e+00 -9.04260814e-01 5.41311204e-01 3.88377160e-01
-8.13561499e-01 2.67859876e-01 -4.11056168e-02 4.46262181e-01
5.15603364e-01 7.03054249e-01 -4.81383026e-01 7.21689999e-01
6.79688990e-01 -6.11108840e-02 -1.72750741e-01 5.38205564e-01
-2.02711210e-01 5.81485212e-01 -3.01996469e-01 5.90536714e-01
-1.26409903e-01 -2.95293778e-01 6.76037133e-01 7.90076375e-01
4.81036514e-01 6.40000820e-01 -3.37379090e-02 7.45929956e-01
-6.99663088e-02 -1.37089893e-01 -6.14934802e-01 7.13017344e-01
4.62462790e-02 9.51131403e-01 -1.07649577e+00 -2.61143208e-01
-4.91420656e-01 1.10756493e+00 -1.76940948e-01 1.21618606e-01
-1.29688048e+00 1.51024744e-01 -2.68312216e-01 4.90813911e-01
-1.65816948e-01 1.50386825e-01 -5.81591904e-01 -7.83600211e-01
1.03503250e-01 -8.15805852e-01 5.40229023e-01 -9.49657917e-01
-1.02879357e+00 6.06377184e-01 1.27143547e-01 -1.17789924e+00
-3.85340899e-01 2.82100998e-02 -7.71875978e-01 8.77138972e-01
-1.28768516e+00 -1.22096574e+00 -5.08546412e-01 6.93641484e-01
7.55641162e-01 1.37275070e-01 8.29803050e-01 4.66951996e-01
-2.38060966e-01 3.19311768e-01 -2.86316071e-02 -2.47090340e-01
7.63880372e-01 -1.31947768e+00 -8.08037296e-02 9.25728798e-01
1.44617617e-01 3.70011985e-01 8.55866730e-01 -8.82966340e-01
-1.16978061e+00 -8.04612458e-01 1.37623206e-01 -4.08659756e-01
1.31551161e-01 1.90639943e-01 -1.07670927e+00 6.13079309e-01
3.81858110e-01 2.99795449e-01 1.09969592e+00 -5.66262364e-01
3.69278938e-01 6.80627748e-02 -1.31670797e+00 4.02996510e-01
7.95432270e-01 -3.12624514e-01 -8.31270695e-01 3.60673606e-01
4.58516091e-01 -7.58560479e-01 -9.61846650e-01 6.53997302e-01
5.25808275e-01 -1.17955458e+00 1.35699856e+00 -2.25136817e-01
7.92541564e-01 -8.08131509e-03 3.55176032e-02 -1.13238549e+00
-2.61074841e-01 -4.19561118e-01 -1.57868788e-01 6.10891640e-01
-3.61008756e-02 -2.88362175e-01 9.77862298e-01 2.50295818e-01
-2.87524134e-01 -1.03695869e+00 -1.05220330e+00 -3.77733856e-01
-5.23248315e-02 -3.21704149e-01 7.81732127e-02 7.73168564e-01
-6.48754656e-01 1.92015618e-01 -4.26148713e-01 1.24910437e-01
9.18122709e-01 1.91512525e-01 3.13388258e-01 -7.68326461e-01
-8.13127935e-01 -7.37735704e-02 8.64920318e-02 -5.64062595e-01
-2.03931645e-01 -7.50472188e-01 -1.22261189e-01 -1.30614519e+00
4.90625709e-01 -1.80008963e-01 -4.02706504e-01 1.88576341e-01
-5.19580960e-01 7.50922441e-01 5.14470413e-02 4.04654115e-01
1.75643146e-01 4.71153527e-01 1.70226872e+00 5.34853786e-02
-2.34109536e-02 3.73073429e-01 -4.35578734e-01 9.47840691e-01
4.62797791e-01 -9.23892796e-01 -6.45994365e-01 -2.54214585e-01
-8.41547251e-02 9.13445592e-01 4.88618761e-01 -1.02185917e+00
1.68831095e-01 -1.36966363e-01 6.94273412e-01 -8.83102477e-01
5.85806012e-01 -8.03820789e-01 7.08670318e-01 1.00070572e+00
-6.31560087e-02 1.76307276e-01 5.52966446e-02 7.00430274e-01
-4.38840866e-01 -4.84239191e-01 1.29638469e+00 -5.89136362e-01
7.44154826e-02 2.21594617e-01 -3.27993274e-01 -1.03426889e-01
1.30827260e+00 -1.71182036e-01 2.57391483e-01 -2.53247917e-01
-1.11739850e+00 -2.96277344e-01 2.63457090e-01 2.36350242e-02
8.58434677e-01 -1.28833902e+00 -7.90726662e-01 3.04927170e-01
-4.84504163e-01 1.55149713e-01 6.91354156e-01 1.14895415e+00
-8.83612275e-01 -5.28632943e-03 -3.44638675e-01 -7.90412545e-01
-1.44025874e+00 5.71541607e-01 4.16645408e-01 -9.05376434e-01
-9.76323903e-01 8.54073167e-01 4.88673419e-01 1.28383934e-01
-3.20858777e-01 -1.26818255e-01 5.47764152e-02 -2.61561066e-01
2.91608065e-01 3.73058856e-01 3.54995072e-01 -3.06057066e-01
-1.55628219e-01 7.95033395e-01 -6.35305559e-03 -4.02468801e-01
1.41744554e+00 -1.19014196e-01 4.07130346e-02 3.25428754e-01
8.00301731e-01 4.88367140e-01 -1.08229637e+00 1.97465762e-01
-7.46677458e-01 -5.56696951e-01 2.64637858e-01 -6.53972566e-01
-1.13207150e+00 8.58269751e-01 9.38528240e-01 1.61903635e-01
1.68294799e+00 2.18429212e-02 1.05211926e+00 -3.53632629e-01
-3.37761790e-02 -7.59106338e-01 5.10966778e-01 -3.68976355e-01
9.93923008e-01 -8.44652474e-01 3.38143080e-01 -6.01389110e-01
-5.39896071e-01 1.03592527e+00 4.13841695e-01 -1.44379497e-01
5.26158690e-01 6.93828166e-01 1.43321440e-01 -4.01593208e-01
-5.01350045e-01 1.90395668e-01 2.11018562e-01 4.41555291e-01
4.75464433e-01 -4.87255566e-02 -5.48722804e-01 4.62996095e-01
-1.08603500e-01 -1.56583831e-01 4.61268336e-01 1.03163958e+00
-1.57186359e-01 -1.06806815e+00 -9.67607379e-01 4.55962151e-01
-1.28105557e+00 -7.34199807e-02 2.25227624e-02 8.69852841e-01
8.25068131e-02 4.40841645e-01 -2.15464354e-01 3.19686309e-02
3.09055120e-01 5.25429007e-03 8.06011915e-01 -7.06058919e-01
-6.23573422e-01 5.10175943e-01 -5.69626927e-01 -3.84289235e-01
-5.52300513e-01 -5.65868556e-01 -1.34630966e+00 -9.17424932e-02
-3.31129462e-01 -2.84053206e-01 4.24106508e-01 7.71848202e-01
-3.80494326e-01 1.34846485e+00 5.94271779e-01 -5.83972454e-01
-7.07666755e-01 -8.14135849e-01 -6.77895665e-01 5.79310656e-01
3.64870816e-01 -6.75511360e-01 -3.05980384e-01 2.19936877e-01]
|
[13.515238761901855, -2.51350736618042]
|
bcfcbf4e-2d5e-4ef7-9833-1b0175994b41
|
a-context-integrated-relational-spatio
|
2009.12469
| null |
https://arxiv.org/abs/2009.12469v1
|
https://arxiv.org/pdf/2009.12469v1.pdf
|
A Context Integrated Relational Spatio-Temporal Model for Demand and Supply Forecasting
|
Traditional methods for demand forecasting only focus on modeling the temporal dependency. However, forecasting on spatio-temporal data requires modeling of complex nonlinear relational and spatial dependencies. In addition, dynamic contextual information can have a significant impact on the demand values, and therefore needs to be captured. For example, in a bike-sharing system, bike usage can be impacted by weather. Existing methods assume the contextual impact is fixed. However, we note that the contextual impact evolves over time. We propose a novel context integrated relational model, Context Integrated Graph Neural Network (CIGNN), which leverages the temporal, relational, spatial, and dynamic contextual dependencies for multi-step ahead demand forecasting. Our approach considers the demand network over various geographical locations and represents the network as a graph. We define a demand graph, where nodes represent demand time-series, and context graphs (one for each type of context), where nodes represent contextual time-series. Assuming that various contexts evolve and have a dynamic impact on the fluctuation of demand, our proposed CIGNN model employs a fusion mechanism that jointly learns from all available types of contextual information. To the best of our knowledge, this is the first approach that integrates dynamic contexts with graph neural networks for spatio-temporal demand forecasting, thereby increasing prediction accuracy. We present empirical results on two real-world datasets, demonstrating that CIGNN consistently outperforms state-of-the-art baselines, in both periodic and irregular time-series networks.
|
['Hongjie Chen', 'Ryan A. Rossi', 'Hoda Eldardiry', 'Kanak Mahadik']
|
2020-09-25
| null | null | null | null |
['irregular-time-series']
|
['time-series']
|
[-2.68766046e-01 -4.70933199e-01 -5.03842354e-01 -4.90177572e-01
3.97331044e-02 -6.78342879e-01 7.46801019e-01 3.72236729e-01
3.86783004e-01 5.70566773e-01 6.30880356e-01 -5.92317939e-01
-4.72471029e-01 -1.24972773e+00 -6.26732111e-01 -4.73819137e-01
-7.49231756e-01 3.52455795e-01 2.04259366e-01 -6.08658016e-01
-2.08202451e-01 5.14433146e-01 -1.23423767e+00 1.06133752e-01
6.99830890e-01 1.25205743e+00 3.51558961e-02 5.42916656e-01
3.92558193e-03 1.11514199e+00 -5.39536297e-01 1.40220122e-02
1.65394753e-01 -1.59503043e-01 -2.93861121e-01 -2.06206352e-01
-1.13226597e-04 -2.32367054e-01 -7.76038706e-01 4.90215540e-01
2.99026012e-01 3.43124658e-01 6.23964965e-01 -1.39100885e+00
-1.06202543e+00 9.38700438e-01 -3.58756214e-01 6.75724030e-01
3.77847664e-02 2.93154176e-02 1.04968214e+00 -6.12001061e-01
2.79038310e-01 9.99949753e-01 6.94434702e-01 -1.49744242e-01
-1.28942871e+00 -5.16498864e-01 1.05981600e+00 4.17293549e-01
-1.45015526e+00 1.01202920e-01 1.32845747e+00 -4.40060586e-01
1.41096032e+00 1.69362575e-01 7.93157458e-01 9.93420601e-01
2.96239465e-01 6.23382330e-01 6.38178408e-01 -1.28582399e-02
2.21316114e-01 -4.95209873e-01 2.64168173e-01 -9.19558480e-02
-8.19747001e-02 1.78970292e-01 -2.96418697e-01 -1.03879407e-01
2.48868629e-01 5.91911495e-01 -1.60578564e-01 1.98875904e-01
-8.25327098e-01 6.17573678e-01 7.33857036e-01 5.36456108e-01
-9.02791381e-01 2.93963462e-01 3.11751336e-01 2.59651482e-01
1.08235323e+00 -2.64066786e-01 -7.70896077e-01 6.71278834e-02
-8.82116497e-01 2.10326046e-01 9.87751663e-01 8.39851916e-01
6.28680825e-01 5.78950107e-01 -1.34516910e-01 5.96684754e-01
1.65621936e-01 8.96048009e-01 9.29120332e-02 -1.91409454e-01
8.03351104e-01 6.78750873e-01 7.77875483e-02 -1.78273344e+00
-8.81467879e-01 -5.21073461e-01 -1.21801019e+00 -5.07394254e-01
1.35353327e-01 -5.50070345e-01 -9.13005352e-01 1.90309358e+00
3.55097532e-01 9.52839911e-01 -2.39136517e-01 6.27020657e-01
7.01614499e-01 1.03685462e+00 1.42454505e-01 -4.29387510e-01
7.19383419e-01 -6.38789415e-01 -9.76800978e-01 6.07674532e-02
6.18099332e-01 -2.28934392e-01 6.10858142e-01 1.53465346e-01
-7.05943108e-01 -3.89955372e-01 -5.82252324e-01 5.17011106e-01
-8.59330177e-01 -2.42466092e-01 4.37133700e-01 1.39991626e-01
-1.06134844e+00 2.77986318e-01 -5.50664961e-01 -1.63406745e-01
-1.41833454e-01 1.91045597e-01 1.52049378e-01 -4.51515475e-03
-1.80707073e+00 7.10368335e-01 1.72806606e-01 3.91041636e-01
-5.96067548e-01 -8.72085512e-01 -7.68263400e-01 2.50688344e-01
6.22595727e-01 -2.82591611e-01 1.01522815e+00 -6.57632172e-01
-9.15491164e-01 -2.75507987e-01 -8.33947882e-02 -5.29810369e-01
-6.03273548e-02 2.53893226e-01 -1.42980540e+00 -4.38148618e-01
-3.19469661e-01 -1.69270024e-01 7.01002359e-01 -1.10281682e+00
-6.57238364e-01 -3.64467621e-01 2.06348658e-01 5.01394309e-02
-1.59658894e-01 -1.95974648e-01 -2.55371273e-01 -1.00749874e+00
-3.03654194e-01 -9.54505742e-01 -1.43413201e-01 -8.75483632e-01
-3.05233687e-01 -5.81235707e-01 1.29008508e+00 -7.86907434e-01
2.15661049e+00 -2.07330942e+00 -8.43621790e-02 5.48201025e-01
2.01056674e-01 8.05440471e-02 -1.25338972e-01 1.09634411e+00
-3.07268985e-02 1.99772537e-01 -2.05389887e-01 -1.60416499e-01
1.59601167e-01 7.42359817e-01 -6.69942975e-01 1.94249228e-01
-1.67510360e-01 1.23518050e+00 -8.86276841e-01 2.13449821e-01
1.53240696e-01 8.23245049e-01 -1.63714364e-01 -9.63327810e-02
-5.59094846e-01 2.47097313e-01 -3.94516498e-01 5.18521667e-01
6.77627146e-01 -4.90198791e-01 5.41056693e-01 -9.59210247e-02
-5.14115207e-02 3.22564483e-01 -1.03611755e+00 9.23531651e-01
-6.37406111e-01 5.56464851e-01 -2.17902496e-01 -1.17430973e+00
1.00407720e+00 4.54353631e-01 7.04841316e-01 -1.03502417e+00
-2.88516611e-01 4.42220122e-02 -1.44532084e-01 -2.60154575e-01
6.11457646e-01 2.82850236e-01 -3.11909348e-01 6.14162028e-01
-4.33363765e-01 1.75647080e-01 8.49110410e-02 1.44346818e-01
9.12352085e-01 -2.02019989e-01 3.74922231e-02 -1.01311930e-01
2.67415076e-01 -1.51745170e-01 7.33647466e-01 3.28092396e-01
-2.75418554e-02 2.17719957e-01 5.67723036e-01 -1.03986025e+00
-6.24117494e-01 -5.79758346e-01 3.85744989e-01 1.29515505e+00
1.17303347e-02 -2.56597757e-01 -9.00183152e-03 -6.72554135e-01
4.55791622e-01 9.73978639e-01 -8.44974756e-01 1.96954817e-01
-9.67040896e-01 -8.00075173e-01 1.92260593e-01 7.01650858e-01
5.86208664e-02 -9.72132504e-01 -3.31324935e-02 5.25012314e-01
-1.10477082e-01 -9.83352482e-01 -7.02334464e-01 -3.43557633e-02
-5.12372196e-01 -8.27246010e-01 -2.34830275e-01 -3.18258733e-01
2.51376152e-01 5.18288910e-01 1.49157584e+00 3.42710428e-02
2.83297688e-01 6.67101502e-01 -3.83095801e-01 -3.59231323e-01
-3.13211083e-02 2.84543842e-01 7.11527169e-02 3.18116695e-01
3.08860689e-01 -1.26959956e+00 -9.94357884e-01 4.23826516e-01
-1.13415515e+00 -2.37659112e-01 -9.37706530e-02 3.91234189e-01
7.03766942e-01 4.24257129e-01 1.15009916e+00 -6.95906222e-01
8.46453428e-01 -1.30358827e+00 -5.86819708e-01 2.90239275e-01
-9.27219391e-01 -4.86834317e-01 9.00448263e-01 -5.58440089e-01
-9.26282167e-01 -1.94134682e-01 2.15702862e-01 -4.22844440e-01
-1.80207223e-01 1.36754072e+00 1.19935580e-01 4.66965199e-01
4.34230536e-01 2.21667215e-01 -6.71937823e-01 -3.86633813e-01
5.72268546e-01 3.63028705e-01 2.89520353e-01 -2.77722150e-01
6.63988352e-01 2.37080351e-01 1.94811165e-01 -6.04831815e-01
-4.90320653e-01 -3.78067911e-01 -5.62957585e-01 -3.38770300e-01
4.36878324e-01 -9.56211507e-01 -4.88691986e-01 4.50250149e-01
-8.86185586e-01 -9.16842520e-01 -2.15456098e-01 4.03129429e-01
-1.34519294e-01 -9.74460393e-02 -4.63158756e-01 -9.79042590e-01
-1.07482776e-01 -5.40367842e-01 6.72766685e-01 -1.87063336e-01
-1.03991285e-01 -1.84494627e+00 2.56414175e-01 -4.71054912e-01
8.75252843e-01 6.76394463e-01 9.57199454e-01 -5.89763343e-01
-4.16070968e-01 -2.63535827e-01 -2.52329588e-01 -8.70889425e-02
6.10731125e-01 4.71127331e-02 -5.93933702e-01 -6.81870282e-02
-4.55345899e-01 3.94288242e-01 6.28024161e-01 6.61252320e-01
1.10386503e+00 -8.90569150e-01 -4.69535500e-01 3.71245623e-01
1.31726038e+00 5.63725650e-01 2.32621610e-01 -2.79933602e-01
1.16314363e+00 6.63296819e-01 1.92511454e-01 6.80463195e-01
1.34117424e+00 7.00703979e-01 7.45256424e-01 -1.05403319e-01
1.25321662e-02 -2.32061505e-01 6.14504553e-02 9.77199316e-01
-1.74351603e-01 -9.20037210e-01 -1.30744612e+00 9.72626686e-01
-2.27198267e+00 -1.15818799e+00 -3.73907804e-01 1.78712964e+00
3.50664467e-01 -1.27134055e-01 5.94478786e-01 -3.16073708e-02
4.96426791e-01 7.72619784e-01 -9.65932012e-01 -4.31202203e-01
-2.77059227e-01 -3.98068637e-01 5.80175698e-01 5.52915156e-01
-9.94126737e-01 5.64767480e-01 6.28023720e+00 4.56129372e-01
-1.34269977e+00 6.62127882e-02 7.60357440e-01 -1.46016851e-01
-7.30488598e-01 -4.11457568e-01 -5.08149981e-01 7.43727624e-01
1.30604768e+00 -2.94476777e-01 7.93353140e-01 6.52654767e-01
5.54525256e-01 3.17954034e-01 -8.70050371e-01 6.38088346e-01
-3.59825976e-02 -1.28304648e+00 1.47485156e-02 2.08455727e-01
1.07879722e+00 5.60717583e-01 2.29272276e-01 3.25541049e-01
7.25966394e-01 -8.92198682e-01 3.29989165e-01 1.00144720e+00
4.59291518e-01 -9.43280518e-01 4.66332793e-01 4.99785960e-01
-1.96155691e+00 -3.18948269e-01 3.22118282e-01 -2.68604100e-01
6.31360173e-01 8.64200234e-01 -3.93392205e-01 8.85164142e-01
5.23428679e-01 1.20654237e+00 -1.82302415e-01 5.13438821e-01
1.77657120e-02 1.08778656e+00 -4.67676044e-01 1.63711548e-01
3.05274218e-01 -1.33908212e-01 3.61359119e-01 1.22498512e+00
5.98406672e-01 3.07747483e-01 5.25199294e-01 4.05158222e-01
-1.40560403e-01 -1.94300450e-02 -9.31779563e-01 6.71155751e-02
5.21251798e-01 8.97873700e-01 -3.73459548e-01 -4.47262317e-01
-7.43872702e-01 3.54920685e-01 2.64087822e-02 9.78357553e-01
-8.52206349e-01 5.72027676e-02 7.67558277e-01 2.57659853e-01
6.02912486e-01 -6.49212241e-01 -1.25629202e-01 -1.03194809e+00
1.41032621e-01 -3.88861656e-01 7.43298352e-01 -5.13287365e-01
-1.75959933e+00 5.93598664e-01 2.94243664e-01 -1.22618198e+00
-7.01627195e-01 -1.29874989e-01 -8.78281832e-01 8.63091171e-01
-1.82756639e+00 -1.43705440e+00 -1.99932247e-01 9.39814448e-01
4.20173883e-01 1.91413984e-02 4.66843635e-01 3.07439625e-01
-5.04256308e-01 1.77538589e-01 1.21116482e-01 -2.67179552e-02
1.53095201e-01 -1.13297379e+00 8.93154025e-01 8.24405551e-01
9.95585546e-02 1.61756620e-01 3.68835449e-01 -9.83512223e-01
-1.42825377e+00 -1.63561618e+00 1.25558507e+00 -3.70183080e-01
1.12244403e+00 -1.50983915e-01 -1.06505108e+00 8.83485436e-01
1.80946887e-01 4.54718739e-01 6.90956295e-01 2.38430947e-01
-5.78098834e-01 -4.36647892e-01 -8.50771606e-01 3.91821951e-01
1.16415811e+00 -6.65034771e-01 -1.72519043e-01 4.58002955e-01
1.16464901e+00 -1.74875110e-01 -1.06332719e+00 5.23719728e-01
4.84282970e-01 -6.38290226e-01 8.26882064e-01 -3.43672574e-01
-2.68706195e-02 -3.36813152e-01 -4.85469639e-01 -1.85064924e+00
-6.70626342e-01 -5.48074663e-01 -7.95215726e-01 1.05583942e+00
6.35830283e-01 -1.09206080e+00 1.42287314e-01 9.45726931e-01
-1.41290873e-01 -7.21594751e-01 -8.40417981e-01 -8.36704969e-01
3.73590402e-02 -9.54336584e-01 1.65913939e+00 1.36246336e+00
-1.67124439e-02 6.69970587e-02 -8.35980356e-01 3.83229673e-01
8.72257352e-02 3.83356720e-01 4.96740192e-01 -1.25684190e+00
-1.26413137e-01 -6.76789343e-01 -9.17592496e-02 -9.48387027e-01
8.35448578e-02 -6.20412827e-01 -4.76606786e-01 -1.63671231e+00
-6.08912051e-01 -5.65539062e-01 -7.45959461e-01 4.60399359e-01
4.64179106e-02 2.88320277e-02 1.48898184e-01 2.51543581e-01
-4.07890797e-01 4.54136938e-01 9.29427445e-01 -3.11077118e-01
-7.34723628e-01 1.36749640e-01 -3.58985037e-01 2.94924766e-01
1.18297470e+00 -1.38136983e-01 -9.54941273e-01 -5.55105925e-01
8.84084105e-01 2.88757384e-01 9.67450961e-02 -5.37060261e-01
3.47754657e-01 -7.94046462e-01 -2.97004450e-02 -1.14716780e+00
8.66723880e-02 -1.13584793e+00 6.37448490e-01 -2.90661044e-02
-2.07127035e-02 7.07318306e-01 3.01460236e-01 9.50626671e-01
-3.13394696e-01 9.05527174e-01 -3.81491214e-01 2.45849952e-01
-6.40452147e-01 1.03352666e+00 -4.41156328e-01 -2.32172146e-01
8.87011826e-01 2.99922358e-02 -4.86926734e-01 -9.23778355e-01
-9.33378994e-01 5.48456013e-01 -1.02131411e-01 6.78894699e-01
2.84442514e-01 -1.65884590e+00 -6.00629270e-01 8.20439160e-02
3.95197868e-02 -1.61123529e-01 6.16606176e-01 7.44251609e-01
1.19517505e-01 5.16285598e-01 5.20149052e-01 -4.09108728e-01
-6.94469035e-01 1.02789640e+00 2.07422286e-01 -5.51122427e-01
-4.63914514e-01 2.08186954e-01 1.31057709e-01 -5.11861086e-01
9.28986520e-02 -9.08313394e-01 -5.46361446e-01 4.89731818e-01
4.04966027e-01 5.86685658e-01 6.18567914e-02 -1.10800099e+00
-3.13651353e-01 5.62522769e-01 5.77670634e-01 2.24498913e-01
1.63220227e+00 -6.14586771e-01 -2.90596075e-02 7.88816273e-01
8.60556006e-01 -2.36731574e-01 -1.08860445e+00 -7.65657067e-01
-1.66975129e-02 -2.23621219e-01 1.91084698e-01 -1.03189182e+00
-1.51123250e+00 5.74506104e-01 2.51472592e-01 1.40388489e+00
1.53104734e+00 -9.22009721e-02 1.13943744e+00 1.90775007e-01
3.00859600e-01 -1.06767142e+00 -3.54905516e-01 7.71344483e-01
1.03113282e+00 -1.04511404e+00 -2.51648754e-01 -9.62968245e-02
-6.03807449e-01 8.10482085e-01 2.80865520e-01 -2.40429744e-01
1.54311144e+00 1.63508922e-01 -5.66101186e-02 -3.13249052e-01
-1.07635045e+00 -3.94393444e-01 7.92494655e-01 5.19589305e-01
1.45992294e-01 8.46613050e-01 7.22994432e-02 4.71231341e-01
-2.05534119e-02 -4.58589494e-02 1.23166382e-01 6.51025057e-01
9.71009061e-02 -8.57274175e-01 -2.95079239e-02 4.16128784e-01
-1.97691157e-01 -6.49266541e-02 -1.17747359e-01 7.50159323e-01
1.73780501e-01 1.23371673e+00 3.85228574e-01 -6.51121914e-01
5.39764225e-01 7.38223344e-02 -3.04214627e-01 -2.94295341e-01
-5.57751179e-01 3.65785100e-02 1.35601252e-01 -6.31022215e-01
-4.75294620e-01 -6.10708177e-01 -9.55540836e-01 -6.15515888e-01
-3.22401933e-02 -2.41920918e-01 6.49674356e-01 9.01915669e-01
8.76209915e-01 6.03705764e-01 1.08160520e+00 -9.08814907e-01
3.01727682e-01 -9.40058410e-01 -6.79798424e-01 1.91000059e-01
8.14074695e-01 -6.64080322e-01 -2.79426277e-01 -1.87166065e-01]
|
[6.6675004959106445, 2.5791690349578857]
|
0ff3b05f-4014-4bf9-a66c-70bcad9e381c
|
parameterizing-the-cost-function-of-dynamic
|
2301.10350
| null |
https://arxiv.org/abs/2301.10350v2
|
https://arxiv.org/pdf/2301.10350v2.pdf
|
Parameterizing the cost function of Dynamic Time Warping with application to time series classification
|
Dynamic Time Warping (DTW) is a popular time series distance measure that aligns the points in two series with one another. These alignments support warping of the time dimension to allow for processes that unfold at differing rates. The distance is the minimum sum of costs of the resulting alignments over any allowable warping of the time dimension. The cost of an alignment of two points is a function of the difference in the values of those points. The original cost function was the absolute value of this difference. Other cost functions have been proposed. A popular alternative is the square of the difference. However, to our knowledge, this is the first investigation of both the relative impacts of using different cost functions and the potential to tune cost functions to different tasks. We do so in this paper by using a tunable cost function {\lambda}{\gamma} with parameter {\gamma}. We show that higher values of {\gamma} place greater weight on larger pairwise differences, while lower values place greater weight on smaller pairwise differences. We demonstrate that training {\gamma} significantly improves the accuracy of both the DTW nearest neighbor and Proximity Forest classifiers.
|
['Geoffrey I. Webb', 'Chang Wei Tan', 'Matthieu Herrmann']
|
2023-01-24
| null | null | null | null |
['dynamic-time-warping']
|
['time-series']
|
[ 3.23886484e-01 -3.28576982e-01 -2.38362372e-01 -4.37518448e-01
-4.08350348e-01 -8.35039020e-01 7.05644727e-01 5.93819499e-01
-7.26216912e-01 5.18888593e-01 2.02340260e-01 -4.00292307e-01
-3.97209376e-01 -7.62157023e-01 -3.66354644e-01 -7.22812414e-01
-6.61228716e-01 2.82872945e-01 6.31650090e-01 -2.79729724e-01
5.48753977e-01 6.59816325e-01 -1.25586414e+00 6.42305389e-02
5.60609460e-01 7.79078960e-01 -1.09426409e-01 4.77339059e-01
2.97988176e-01 -1.38555661e-01 -7.52752244e-01 -8.39762688e-02
6.16578162e-01 -4.60530311e-01 -5.73585272e-01 -3.44952673e-01
1.33910611e-01 -1.74377754e-03 -3.53890210e-02 8.26366246e-01
3.82411629e-01 5.57443619e-01 8.25534582e-01 -1.31093907e+00
-2.97447920e-01 2.96640605e-01 -6.49358690e-01 7.48117507e-01
4.60918099e-01 9.24962386e-03 1.07457697e+00 -4.92853612e-01
7.06598878e-01 7.04932988e-01 7.91676998e-01 -5.63319512e-02
-1.59447443e+00 -5.88738620e-01 2.52388977e-02 4.25163917e-02
-1.25702929e+00 -8.62618312e-02 7.05327213e-01 -4.85293061e-01
1.13399220e+00 3.69871914e-01 6.74059093e-01 6.85476482e-01
4.37150180e-01 -8.34037811e-02 1.19379449e+00 -6.35877013e-01
2.44835883e-01 -3.84676367e-01 -1.99161042e-02 1.81271777e-01
2.63008863e-01 9.28066224e-02 -4.32173312e-01 -4.19076920e-01
7.13387609e-01 1.65332586e-01 -1.99349895e-01 -2.50399679e-01
-1.43185270e+00 7.68016398e-01 1.85059339e-01 5.65058529e-01
-3.26120183e-02 3.02271005e-02 1.90974340e-01 4.93381381e-01
4.66034263e-01 7.82021284e-01 -3.47952932e-01 -4.15375680e-01
-8.22652876e-01 3.42151314e-01 6.54625475e-01 4.49001998e-01
6.05261266e-01 -4.40548211e-01 -2.59356424e-02 5.75102270e-01
-1.83982909e-01 5.19305430e-02 5.67964613e-01 -9.66992974e-01
4.92718458e-01 4.24340636e-01 6.09669350e-02 -1.15149271e+00
-5.60965836e-01 5.46784550e-02 -2.43704185e-01 2.74601877e-01
9.81079757e-01 -1.28344625e-01 -6.08777940e-01 1.90767646e+00
2.60612935e-01 1.66106448e-01 -3.98951024e-01 6.92053199e-01
-2.01608270e-01 6.32465899e-01 3.30773718e-03 -4.20120686e-01
1.09234190e+00 -3.26836497e-01 -3.41924667e-01 -1.98938072e-01
6.28760099e-01 -1.04752052e+00 1.13022339e+00 1.63136311e-02
-8.97098780e-01 -7.69183561e-02 -1.27943826e+00 1.76789880e-01
-3.21735770e-01 -6.56977355e-01 4.03341472e-01 4.10749704e-01
-8.45826268e-01 1.22884238e+00 -1.09402061e+00 -4.32371587e-01
-2.40311533e-01 3.25465828e-01 -3.68713975e-01 3.82153958e-01
-1.11796606e+00 1.13438916e+00 8.46807733e-02 -2.66222179e-01
2.87556611e-02 -5.69168150e-01 -3.89583379e-01 -8.63172188e-02
4.32589315e-02 -1.04303740e-01 8.95188451e-01 -7.43412554e-01
-1.02775598e+00 7.53638387e-01 -1.72337368e-02 -4.83566523e-01
5.07645905e-01 1.43212125e-01 -5.36946893e-01 -1.95688710e-01
-5.47828190e-02 5.57259798e-01 6.09163880e-01 -6.17800951e-01
-1.63233027e-01 -5.31180322e-01 -4.94993217e-02 1.65412277e-01
-2.60827184e-01 1.07188784e-01 5.80655411e-03 -8.95091593e-01
4.13416833e-01 -1.41444468e+00 -8.36729184e-02 1.28226936e-01
5.54774962e-02 -6.84893429e-02 6.75612748e-01 -4.79200333e-01
1.33238721e+00 -2.15385771e+00 1.37932017e-01 4.73924369e-01
-6.11462258e-02 -4.55662310e-02 -7.26329014e-02 7.84380853e-01
-3.11452985e-01 1.69499099e-01 -3.20356458e-01 2.98094064e-01
-2.49952003e-01 2.41830453e-01 -2.33100474e-01 6.48556530e-01
3.88598591e-02 2.70769924e-01 -8.77576113e-01 -1.68668121e-01
-1.13784857e-01 3.77988547e-01 -4.41525102e-01 -2.20718399e-01
1.23475619e-01 1.65558130e-01 -1.27098173e-01 -1.41789904e-02
3.64692479e-01 8.47945884e-02 3.07643682e-01 -1.36928752e-01
-5.69437265e-01 3.58018249e-01 -1.16808105e+00 1.29827154e+00
-3.90439518e-02 7.29768336e-01 -5.52578747e-01 -8.55878890e-01
1.13064861e+00 1.92767471e-01 7.33046830e-01 -5.67622602e-01
4.20573652e-02 3.71155977e-01 5.70394456e-01 -2.29580030e-01
4.00565386e-01 -4.82232273e-01 -6.51194975e-02 7.23764002e-01
-4.76089150e-01 -3.67828041e-01 2.78560698e-01 -1.84026405e-01
1.21577191e+00 -7.13204220e-02 4.76657420e-01 -3.95450383e-01
-6.70681521e-03 -2.35304072e-01 5.07089019e-01 3.13030809e-01
-4.48676705e-01 6.41578734e-01 7.29352534e-01 -6.92835569e-01
-1.43194091e+00 -9.41103101e-01 -2.79541701e-01 9.75691974e-01
1.29650593e-01 -4.45805669e-01 -5.90223730e-01 -3.50257754e-01
2.17789695e-01 6.43644691e-01 -6.68514788e-01 -2.33264163e-01
-8.85511339e-01 -7.31161058e-01 5.44615448e-01 7.23349750e-01
1.41816169e-01 -7.94848204e-01 -1.11524916e+00 -2.14811489e-02
-1.78241119e-01 -7.56513178e-01 -9.61110055e-01 3.88608992e-01
-1.22772634e+00 -8.74420643e-01 -5.09641469e-01 -5.15213192e-01
6.83136582e-01 2.33359143e-01 6.84648156e-01 -9.06111598e-02
-2.81848580e-01 9.66919810e-02 -4.24700767e-01 -3.99747640e-01
-6.27088398e-02 3.73104005e-03 1.98045477e-01 -4.19422030e-01
2.93768227e-01 -8.96956086e-01 -6.25434101e-01 6.76006198e-01
-1.01380050e+00 -2.21725687e-01 -4.61486876e-02 6.33378804e-01
7.60661304e-01 -2.69927904e-02 1.56976402e-01 -4.87776518e-01
7.52567828e-01 -3.57824594e-01 -3.61313760e-01 9.35630128e-02
-6.78166032e-01 4.60905671e-01 3.35673571e-01 -9.19085443e-01
-3.36690009e-01 -1.48669630e-01 3.14556062e-01 -3.01870316e-01
2.42556840e-01 4.78589416e-01 4.45882350e-01 -6.28299033e-03
6.93841457e-01 -1.89649940e-01 8.03860351e-02 -3.23115110e-01
2.02584594e-01 2.44865581e-01 4.12247956e-01 -6.48301005e-01
4.96946454e-01 5.29032171e-01 2.90679008e-01 -4.49736625e-01
-3.14127356e-01 -7.47678205e-02 -7.52289653e-01 -1.77559957e-01
6.90110028e-01 -1.61459863e-01 -4.94072944e-01 1.16167925e-01
-7.97739506e-01 -2.86489278e-01 -3.71386409e-01 6.16798937e-01
-5.07960916e-01 2.91592419e-01 -8.36932883e-02 -4.34223503e-01
-6.58866540e-02 -9.23384309e-01 6.56740606e-01 8.66783559e-02
-8.94822896e-01 -8.50398600e-01 4.02971327e-01 -1.43529952e-01
3.06264281e-01 7.04537868e-01 1.09277606e+00 -8.48299861e-01
1.43199891e-01 -4.75079507e-01 2.58376956e-01 7.79350698e-02
3.92491072e-01 3.89343470e-01 -3.76467377e-01 -4.90422934e-01
7.62256160e-02 4.42739576e-01 5.57720423e-01 3.67301196e-01
8.37832510e-01 -4.07704055e-01 -3.59582841e-01 3.61486107e-01
1.16335797e+00 6.73740625e-01 4.64520335e-01 7.75447369e-01
2.88870752e-01 7.54282057e-01 5.01901865e-01 3.56196105e-01
4.80158739e-02 1.10895431e+00 1.34673893e-01 9.24862102e-02
2.12053955e-01 -2.94937473e-02 2.83374071e-01 5.72211504e-01
-3.87229949e-01 -5.21308966e-02 -1.21649861e+00 4.86463010e-01
-1.63605952e+00 -1.14388013e+00 1.55039772e-01 2.76120257e+00
8.13102722e-01 5.40975451e-01 3.39379221e-01 3.91563833e-01
7.80648172e-01 4.16303575e-01 -5.12783706e-01 -6.50884449e-01
1.98754877e-01 2.90927678e-01 6.76777303e-01 6.30774856e-01
-8.50364447e-01 3.52182716e-01 6.72745180e+00 4.23829406e-01
-1.21621943e+00 -1.81524277e-01 3.66752505e-01 -6.38055325e-01
-2.35676989e-01 4.12048727e-01 -3.38969648e-01 7.57485867e-01
9.27291512e-01 -4.95416135e-01 4.91229683e-01 5.23649812e-01
2.11586267e-01 -1.61452174e-01 -1.22341752e+00 3.96823823e-01
-2.05277875e-01 -9.36840594e-01 -3.15266252e-01 2.65380681e-01
6.02161527e-01 2.98622046e-02 5.22043481e-02 -3.02317113e-01
1.57096118e-01 -8.24388444e-01 8.17061722e-01 3.71121824e-01
5.20422101e-01 -7.02317655e-01 4.45331365e-01 1.00038446e-01
-1.16662252e+00 -2.46615894e-02 6.45666122e-02 -3.60193908e-01
2.96847254e-01 5.97800970e-01 -7.87462831e-01 -3.48679610e-02
5.51175952e-01 1.86398983e-01 -2.63850540e-01 8.00942779e-01
1.84059471e-01 3.91919166e-01 -6.09436810e-01 3.54223251e-02
2.11688317e-02 -4.33066994e-01 7.32909977e-01 9.77651417e-01
5.18753827e-01 3.38393562e-02 2.04232246e-01 2.87457258e-01
4.41323787e-01 -9.05356109e-02 -6.94392681e-01 -7.96124265e-02
1.02118278e+00 7.85668194e-01 -1.02939558e+00 1.03394523e-01
-3.35519135e-01 6.00158095e-01 1.26477674e-01 1.28736019e-01
-8.63324404e-01 -6.06523335e-01 9.45986032e-01 6.15108073e-01
2.11839154e-01 -8.02409351e-01 -4.62446392e-01 -6.66563869e-01
2.20011219e-01 -6.06466234e-01 7.31559336e-01 -5.16225278e-01
-1.11672223e+00 4.38261420e-01 4.61088121e-01 -1.43540084e+00
-3.42191368e-01 -2.69140244e-01 -5.79754531e-01 9.21291709e-01
-7.65062213e-01 -3.78639221e-01 -8.63918737e-02 3.60912651e-01
2.29838267e-01 2.17317611e-01 8.03809702e-01 2.05549821e-02
-4.01506841e-01 3.72077554e-01 7.21664876e-02 -9.09678042e-02
8.83000195e-01 -9.42512035e-01 5.59091926e-01 7.72862196e-01
3.65094431e-02 8.77451003e-01 1.22430265e+00 -6.58267856e-01
-8.43135595e-01 -5.27980804e-01 1.00742006e+00 -4.19532776e-01
7.92934954e-01 -1.53100699e-01 -9.84580755e-01 8.20716977e-01
-9.89287496e-02 -8.25634450e-02 8.18518460e-01 2.92940848e-02
-7.09600806e-01 -1.25889868e-01 -1.36807537e+00 8.49000454e-01
9.22048748e-01 -4.85989273e-01 -4.38050479e-01 2.14020796e-02
5.57865560e-01 -2.28069276e-01 -1.36215127e+00 4.87803966e-01
9.33751464e-01 -9.73199189e-01 1.05168486e+00 -4.72682536e-01
2.20972359e-01 -5.00134885e-01 -3.55727017e-01 -1.30994260e+00
-4.39299673e-01 -4.65005606e-01 1.65386215e-01 9.39035594e-01
5.09934247e-01 -8.74683917e-01 4.29045618e-01 8.49740386e-01
1.80420443e-01 -9.85181987e-01 -1.27546990e+00 -1.00848329e+00
2.20632121e-01 -2.03751758e-01 6.61429703e-01 1.17080224e+00
4.65835601e-01 -1.82640195e-01 -5.86377876e-03 -9.09941941e-02
1.66015327e-01 9.11586732e-02 2.05672085e-01 -1.22840130e+00
-2.43127093e-01 -6.29893839e-01 -6.78638458e-01 -3.49224091e-01
-3.16119492e-01 -7.58172393e-01 -2.45428026e-01 -1.03286040e+00
-5.11071784e-03 -4.35869813e-01 -1.49564460e-01 6.69261694e-01
-1.28448918e-01 4.81714448e-03 2.35939190e-01 6.43728852e-01
2.62758285e-01 7.79356807e-02 9.21269894e-01 2.35378206e-01
-5.36364377e-01 -2.84918547e-01 -4.09166873e-01 5.21906614e-01
9.35211301e-01 -6.74672902e-01 -4.71905679e-01 -4.81740177e-01
1.22251086e-01 2.27427818e-02 -1.32331010e-02 -8.88316929e-01
1.48387283e-01 -4.72340167e-01 2.51549155e-01 -2.36596495e-01
3.31087828e-01 -6.20143533e-01 5.38245380e-01 5.47894180e-01
-4.84164178e-01 8.86418700e-01 2.15839610e-01 5.56914032e-01
5.17174117e-02 -9.00409073e-02 8.68204057e-01 6.71379194e-02
-2.56480396e-01 -2.32667357e-01 -2.60375917e-01 -1.40068069e-01
1.33150256e+00 -5.29324055e-01 -1.80410832e-01 -3.13093483e-01
-4.15434420e-01 -1.04020759e-01 7.25127339e-01 5.21584570e-01
5.68393394e-02 -1.27507281e+00 -4.21589226e-01 1.76096901e-01
-2.82109063e-02 -4.06481653e-01 -2.69369513e-01 1.04001665e+00
-4.75647926e-01 2.06969917e-01 -5.06605685e-01 -3.54572743e-01
-1.57151830e+00 4.78952080e-01 3.45282614e-01 -1.61017552e-01
-6.31604433e-01 5.93725741e-01 -2.77516127e-01 -7.25070313e-02
-6.72147423e-02 -4.28525567e-01 7.86861256e-02 3.75509590e-01
3.54424804e-01 7.70337462e-01 2.05110073e-01 -5.57477593e-01
-5.28007090e-01 7.61054397e-01 -2.10799262e-01 -6.53010249e-01
1.33992076e+00 1.29277974e-01 -9.24493968e-02 6.31685674e-01
1.10765028e+00 -1.03347793e-01 -1.35701120e+00 1.14010841e-01
1.50446534e-01 -6.99814677e-01 -2.49063656e-01 -4.52512175e-01
-6.13894641e-01 4.10943329e-01 6.95769608e-01 5.17944336e-01
1.09264159e+00 -1.03947394e-01 6.29643798e-01 1.64470188e-02
4.79813993e-01 -1.06027520e+00 1.25430226e-01 4.23247516e-01
7.08171308e-01 -5.99364817e-01 2.45365873e-01 -2.00905859e-01
-5.96080720e-01 1.03301394e+00 2.98833251e-01 -2.69806623e-01
4.12453413e-01 3.85875255e-01 -1.91111695e-02 -1.84307862e-02
-4.54166412e-01 1.94299504e-01 7.26868138e-02 3.82798433e-01
7.80622840e-01 1.64184034e-01 -1.06723630e+00 2.40433868e-02
-4.89991099e-01 -1.25082165e-01 3.26418221e-01 1.30690241e+00
-3.14687788e-01 -1.30605316e+00 -4.66066182e-01 4.13269430e-01
-1.33599564e-01 1.72828361e-02 -4.66964066e-01 7.75853038e-01
5.92592992e-02 7.65981674e-01 5.84103942e-01 -4.83406723e-01
4.76575196e-01 3.14279616e-01 6.16092920e-01 -2.79675931e-01
-4.84149992e-01 -1.43925279e-01 6.91488087e-02 -3.26718837e-01
-2.75558650e-01 -1.04555857e+00 -1.37344491e+00 -5.50272524e-01
-3.22415382e-01 -1.95024192e-01 6.69061601e-01 8.61534774e-01
2.34189227e-01 5.35714142e-02 6.30458713e-01 -8.20798695e-01
-6.45926356e-01 -7.15596020e-01 -5.28070152e-01 6.85179412e-01
1.74385920e-01 -7.73544312e-01 -4.32783574e-01 -2.50962432e-02]
|
[7.306468963623047, 3.4372661113739014]
|
8bea1121-7dd7-4a9c-bd2a-250e1791482e
|
retrievalfuse-neural-3d-scene-reconstruction
|
2104.00024
| null |
https://arxiv.org/abs/2104.00024v2
|
https://arxiv.org/pdf/2104.00024v2.pdf
|
RetrievalFuse: Neural 3D Scene Reconstruction with a Database
|
3D reconstruction of large scenes is a challenging problem due to the high-complexity nature of the solution space, in particular for generative neural networks. In contrast to traditional generative learned models which encode the full generative process into a neural network and can struggle with maintaining local details at the scene level, we introduce a new method that directly leverages scene geometry from the training database. First, we learn to synthesize an initial estimate for a 3D scene, constructed by retrieving a top-k set of volumetric chunks from the scene database. These candidates are then refined to a final scene generation with an attention-based refinement that can effectively select the most consistent set of geometry from the candidates and combine them together to create an output scene, facilitating transfer of coherent structures and local detail from train scene geometry. We demonstrate our neural scene reconstruction with a database for the tasks of 3D super resolution and surface reconstruction from sparse point clouds, showing that our approach enables generation of more coherent, accurate 3D scenes, improving on average by over 8% in IoU over state-of-the-art scene reconstruction.
|
['Angela Dai', 'Matthias Nießner', 'Qi Shan', 'Fangchang Ma', 'Justus Thies', 'Yawar Siddiqui']
|
2021-03-31
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Siddiqui_RetrievalFuse_Neural_3D_Scene_Reconstruction_With_a_Database_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Siddiqui_RetrievalFuse_Neural_3D_Scene_Reconstruction_With_a_Database_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['3d-scene-reconstruction', 'scene-generation']
|
['computer-vision', 'computer-vision']
|
[ 3.94416094e-01 2.40770161e-01 4.83572811e-01 -4.35559809e-01
-1.17305017e+00 -5.43882191e-01 6.03982449e-01 -2.70621292e-02
4.80117723e-02 4.71521556e-01 3.82206857e-01 1.90380931e-01
1.75468046e-02 -1.12559831e+00 -1.17451203e+00 -5.27430952e-01
9.21063200e-02 1.05887175e+00 1.61130562e-01 -3.86688411e-02
2.57399797e-01 9.15533423e-01 -1.73773479e+00 3.70897770e-01
8.66000950e-01 9.54857886e-01 6.61416650e-01 5.97639680e-01
-4.81673300e-01 7.49701321e-01 -2.94114798e-01 2.28415187e-02
3.87680680e-01 -3.49432707e-01 -8.72837007e-01 3.71612370e-01
8.69523764e-01 -7.31500268e-01 -3.38312596e-01 7.89435565e-01
3.80772054e-01 8.77721608e-02 7.01235712e-01 -4.99982238e-01
-4.65912849e-01 2.05458686e-01 -3.31648499e-01 -9.71894041e-02
3.97642136e-01 8.02082792e-02 6.94719076e-01 -1.29022944e+00
9.79080200e-01 1.24390328e+00 6.04606748e-01 3.44514430e-01
-1.57264042e+00 -4.71432447e-01 1.77976206e-01 -3.36300254e-01
-1.47147119e+00 -5.33877552e-01 9.05638278e-01 -4.89211947e-01
1.33576024e+00 1.75733179e-01 8.74520063e-01 5.19851089e-01
-2.81815492e-02 4.53516871e-01 7.63645113e-01 -7.01134801e-02
3.39181095e-01 -1.91194668e-01 -2.87854910e-01 7.13755548e-01
7.09891459e-03 -1.51529595e-01 -7.28686750e-01 -1.94767281e-01
1.53140950e+00 3.83532699e-03 -2.37344071e-01 -6.93053663e-01
-1.30302370e+00 7.64814317e-01 8.05519581e-01 -1.47275642e-01
-7.64435828e-01 2.54975617e-01 -1.81533173e-01 -8.45720619e-02
8.16649556e-01 7.26918995e-01 -2.64736563e-01 2.31838569e-01
-1.12344921e+00 6.57139838e-01 6.98943675e-01 1.03205073e+00
1.16571069e+00 2.40664631e-01 2.58897319e-02 6.68335974e-01
1.64344162e-01 5.78569710e-01 -2.52556920e-01 -1.27107430e+00
3.04818749e-01 6.63564086e-01 1.98140144e-02 -7.76231110e-01
-1.15915053e-01 -5.09982467e-01 -9.67335463e-01 2.93987274e-01
-8.09494480e-02 2.49028563e-01 -1.24803686e+00 1.49603021e+00
7.02729285e-01 4.76573527e-01 -1.85971603e-01 8.65834594e-01
1.10972536e+00 9.12525654e-01 -1.50198013e-01 4.60874215e-02
8.76329899e-01 -6.45553052e-01 7.47735724e-02 -3.61014754e-01
6.05623238e-02 -6.57861829e-01 6.29200399e-01 1.85543686e-01
-1.40557778e+00 -6.02331400e-01 -7.11561978e-01 -4.72453624e-01
1.27393633e-01 -4.04806584e-01 6.74324632e-01 -1.31784424e-01
-1.38503003e+00 6.61886871e-01 -8.68667901e-01 -8.97247717e-02
8.41720998e-01 9.26682279e-02 -4.32172686e-01 -4.75252151e-01
-4.81941521e-01 5.94569147e-01 3.64807308e-01 -1.79143354e-01
-1.29794133e+00 -1.35103488e+00 -1.06591880e+00 2.27281228e-02
9.74338725e-02 -1.36852205e+00 1.20432138e+00 -5.85003436e-01
-1.24779952e+00 1.05418444e+00 -3.16991150e-01 -1.10520609e-01
1.98494494e-01 -3.16571891e-01 3.92882377e-01 1.78072974e-01
2.94603258e-01 8.35739136e-01 9.28740323e-01 -1.70082617e+00
-3.59246612e-01 -2.97794610e-01 -5.30916303e-02 5.40152192e-01
6.35363281e-01 -4.33533490e-01 -6.02104604e-01 -3.53009164e-01
7.79133439e-01 -6.40069723e-01 -5.97960889e-01 -3.02460156e-02
-4.60164368e-01 1.42183349e-01 6.01405203e-01 -6.55124843e-01
4.70339209e-01 -1.88697195e+00 7.04562664e-01 2.39863411e-01
5.67990124e-01 -3.59508663e-01 -8.01295489e-02 2.10433260e-01
-6.45009354e-02 -1.69696733e-01 -5.19199431e-01 -8.38928282e-01
-2.13943377e-01 1.63539499e-01 -6.31815255e-01 2.82883584e-01
4.34846163e-01 9.93424714e-01 -9.60914910e-01 -3.46668512e-01
6.43090665e-01 7.76530385e-01 -1.07308078e+00 6.20041072e-01
-6.85797274e-01 9.22461331e-01 -6.08731925e-01 4.74980652e-01
8.35666299e-01 -6.43652916e-01 -1.04818188e-01 -3.01052243e-01
-1.55442044e-01 5.78025520e-01 -1.24154019e+00 2.26849222e+00
-5.09621799e-01 3.49550962e-01 1.30485296e-01 -8.53591740e-01
8.40720654e-01 4.91136871e-02 6.98077381e-01 -4.25006866e-01
-9.95655134e-02 3.57845649e-02 -6.41590953e-01 -7.73341730e-02
8.24713290e-01 -3.52784097e-01 -4.36822474e-02 4.49261963e-01
3.80397528e-01 -1.20384812e+00 -3.92022580e-01 4.36715811e-01
1.00220692e+00 5.30317843e-01 -2.55532693e-02 -7.51000941e-02
-4.49414812e-02 3.04432482e-01 1.08081505e-01 7.52496362e-01
6.78213656e-01 1.13404119e+00 3.30306478e-02 -6.18599057e-01
-1.56500161e+00 -1.40287244e+00 -7.79442042e-02 5.45688987e-01
9.08371657e-02 -4.24895346e-01 -6.26400828e-01 -1.23282276e-01
2.06297394e-02 8.03685546e-01 -4.07479525e-01 1.98555831e-02
-6.82109237e-01 -4.88970876e-01 -8.98618028e-02 3.06901097e-01
4.44313139e-01 -1.13536477e+00 -6.60608172e-01 3.58616352e-01
-1.25163838e-01 -1.13684523e+00 -9.03180093e-02 1.39972582e-01
-1.16946101e+00 -8.15167367e-01 -4.65581089e-01 -5.84065855e-01
8.95278573e-01 4.13650930e-01 1.80518854e+00 7.08090663e-02
-2.60031581e-01 3.46473008e-01 7.14493766e-02 -1.21837758e-01
-5.93477130e-01 -1.91076938e-02 -3.04828882e-01 -1.85355738e-01
-2.06805214e-01 -1.02270758e+00 -5.27091920e-01 -2.41184980e-01
-7.20714211e-01 7.83264458e-01 5.56960762e-01 4.71897334e-01
1.27640963e+00 -5.53360544e-02 5.40542044e-02 -9.87641156e-01
9.58128124e-02 -6.40783250e-01 -6.80247545e-01 -1.82854027e-01
-9.15437341e-02 2.80451894e-01 4.43233758e-01 -8.03308859e-02
-1.24170649e+00 4.08816665e-01 -2.99014151e-01 -8.92722309e-01
-3.56650531e-01 3.79086643e-01 -3.33852582e-02 -1.40079353e-02
6.41480088e-01 5.42231739e-01 -2.55822867e-01 -5.81054091e-01
6.18678093e-01 -5.49874343e-02 7.73039997e-01 -6.65412843e-01
1.03245652e+00 5.35046220e-01 2.29586825e-01 -8.28925252e-01
-1.08878338e+00 -2.02121481e-01 -9.78848100e-01 -6.25011995e-02
8.89451027e-01 -1.35515559e+00 -1.89567879e-01 4.72863644e-01
-1.32085216e+00 -6.56638265e-01 -5.97504199e-01 1.81375086e-01
-8.07624400e-01 -4.89173308e-02 -5.92712700e-01 -5.15579820e-01
-4.15165722e-01 -1.06326997e+00 1.77952850e+00 -1.45301253e-01
-1.31763533e-01 -5.51976442e-01 1.60447940e-01 1.85848251e-01
3.59984577e-01 6.59347534e-01 1.03668487e+00 2.61059374e-01
-1.53691828e+00 1.34730428e-01 -2.29018554e-01 1.52990492e-02
1.54828057e-01 -2.20480487e-01 -1.11243892e+00 -9.30150524e-02
1.88181892e-01 -2.41316706e-01 8.58985782e-01 6.81446135e-01
1.24492562e+00 -8.77903700e-02 -3.39633167e-01 1.15907955e+00
1.70134819e+00 -3.15571040e-01 7.03679442e-01 -1.22788861e-01
1.19395673e+00 3.49812746e-01 1.70409709e-01 4.56492752e-01
4.26946938e-01 4.89982814e-01 5.92052579e-01 -1.67071223e-01
-5.14542758e-01 -6.42729819e-01 -1.98060647e-01 8.09457660e-01
-1.98087305e-01 3.07346620e-02 -7.71863341e-01 4.57185239e-01
-1.52985311e+00 -9.39646542e-01 6.50469661e-02 2.19577074e+00
8.32235098e-01 -8.62426963e-03 -2.04893708e-01 -2.22736403e-01
2.49485195e-01 3.38332742e-01 -8.71488214e-01 4.98814397e-02
-4.50217491e-03 5.72485864e-01 3.22720647e-01 7.29054272e-01
-8.89758706e-01 1.25606883e+00 6.67718220e+00 5.76839805e-01
-1.12596571e+00 -4.83159488e-03 7.78260648e-01 -3.31710845e-01
-8.17385018e-01 -3.80378403e-02 -6.56818092e-01 7.73220807e-02
6.78666532e-01 7.93155655e-02 6.36272490e-01 8.81393433e-01
1.45974162e-03 -8.01208168e-02 -1.20535636e+00 1.18539822e+00
9.73894373e-02 -1.95787477e+00 5.37228703e-01 -2.04433147e-02
1.11269355e+00 5.31686246e-01 -2.52355129e-01 -1.76186506e-02
7.03793526e-01 -1.28991592e+00 1.03715444e+00 8.36408913e-01
1.05546045e+00 -6.97064877e-01 4.01824787e-02 5.57360351e-01
-1.16503930e+00 5.02850831e-01 -5.55945635e-01 3.26561183e-02
4.38185573e-01 8.69590282e-01 -9.55366015e-01 5.91512918e-01
7.59056866e-01 7.96304524e-01 -2.94567376e-01 8.95876944e-01
-6.67138323e-02 3.29244375e-01 -5.99307716e-01 4.81607407e-01
-5.02811233e-03 -2.28735834e-01 6.61478937e-01 7.95742095e-01
6.10482693e-01 4.48625743e-01 1.52430639e-01 1.65613985e+00
-2.47072741e-01 -2.46424109e-01 -8.33486259e-01 2.59045094e-01
5.11517048e-01 1.21072471e+00 -7.04401672e-01 -4.73517090e-01
-9.23432410e-02 1.05572200e+00 6.99272573e-01 2.80799806e-01
-5.19295633e-01 3.41444850e-01 6.46412253e-01 4.65206653e-01
4.64425653e-01 -5.86902738e-01 -5.76954901e-01 -1.17375433e+00
-9.35458466e-02 -5.47328591e-01 -1.23923928e-01 -1.19389403e+00
-1.03543532e+00 7.99136817e-01 6.92641810e-02 -1.15134394e+00
-3.44297796e-01 -1.52139112e-01 -4.41068560e-01 1.25858235e+00
-1.37171948e+00 -1.12955737e+00 -8.04370522e-01 5.46967983e-01
7.32308388e-01 2.45365873e-01 8.08685124e-01 -9.41554755e-02
1.32495239e-01 -2.47635171e-01 -1.51374146e-01 -2.64385790e-01
1.35029748e-01 -1.25909698e+00 9.78063822e-01 7.46495724e-01
2.93069899e-01 3.97520930e-01 5.18154562e-01 -7.64111161e-01
-1.55440640e+00 -1.30762911e+00 4.45089787e-01 -8.20772767e-01
-2.09839698e-02 -5.67840099e-01 -1.03962469e+00 6.67262018e-01
-1.87192395e-01 1.13296784e-01 2.02501535e-01 -1.47740155e-01
-3.77067149e-01 2.65215546e-01 -1.14534128e+00 5.00275195e-01
1.46724582e+00 -5.60421109e-01 -4.30779099e-01 3.51349056e-01
1.00290859e+00 -9.86327946e-01 -8.40229809e-01 3.33191186e-01
2.26758361e-01 -1.04179287e+00 1.39252734e+00 -2.96061844e-01
8.94046783e-01 -4.51938182e-01 -2.98678964e-01 -1.37311101e+00
-5.71356714e-01 -5.01431286e-01 -1.15102194e-01 7.93375254e-01
1.77239537e-01 -1.46271080e-01 9.15765703e-01 6.54303551e-01
-5.78893304e-01 -6.63160920e-01 -7.57397890e-01 -1.30480617e-01
4.58591059e-02 -5.85916817e-01 8.85596633e-01 7.46720195e-01
-9.18335795e-01 5.09531081e-01 -2.32912302e-01 4.82340723e-01
1.06354535e+00 7.00256169e-01 1.11751664e+00 -1.31031048e+00
-3.85480225e-01 -1.53496996e-01 -1.57638133e-01 -1.50859010e+00
6.17124550e-02 -1.08517611e+00 3.64193946e-01 -2.01561093e+00
1.93904832e-01 -7.62946248e-01 2.51300097e-01 2.31173813e-01
-8.60880017e-02 4.11428690e-01 4.63670082e-02 4.09153402e-01
-4.46417451e-01 7.01714993e-01 1.50125742e+00 1.27649456e-01
-4.09255892e-01 -3.32455367e-01 -7.50380754e-01 6.22984171e-01
2.85054892e-01 -3.61182481e-01 -4.12624508e-01 -8.47105145e-01
1.93263203e-01 2.87899464e-01 7.00103045e-01 -9.62927639e-01
4.20017764e-02 -2.97407359e-01 7.98033714e-01 -1.01816523e+00
8.21609795e-01 -5.46666741e-01 7.13658333e-01 -8.67466349e-03
-4.41643447e-02 -3.39217365e-01 3.74271244e-01 3.61087590e-01
4.57784310e-02 2.38139823e-01 7.25460172e-01 -6.56644642e-01
-7.20761538e-01 7.71125853e-01 1.19250543e-01 -1.18765049e-01
4.77786541e-01 -3.40035468e-01 -3.99653465e-02 -3.87816608e-01
-6.61513984e-01 -6.50129318e-02 9.19305027e-01 1.32473752e-01
8.82952869e-01 -1.26536846e+00 -8.56001675e-01 5.46669781e-01
-1.37823388e-01 1.31155765e+00 5.52935243e-01 1.61878988e-01
-7.43608534e-01 2.42546812e-01 -7.50392750e-02 -9.41429198e-01
-9.24143851e-01 3.98724914e-01 4.97857541e-01 -2.12085657e-02
-1.16974044e+00 1.09084129e+00 8.11579883e-01 -5.16972721e-01
-2.86375642e-01 -5.74494720e-01 3.91376942e-01 -5.57427049e-01
2.31688514e-01 -1.90757513e-01 7.58957267e-02 -8.34121823e-01
-2.13992342e-01 7.91025341e-01 3.06021750e-01 -2.62261927e-02
1.92149699e+00 6.13816343e-02 -2.89122820e-01 3.19964141e-01
1.06004012e+00 -2.86503900e-02 -1.68241298e+00 -4.23251122e-01
-5.98570108e-01 -7.06142008e-01 2.65648425e-01 -4.90453660e-01
-1.02695990e+00 6.15529120e-01 1.33908898e-01 -4.43984978e-02
9.40160871e-01 6.60215735e-01 7.48074472e-01 1.91106290e-01
7.01041758e-01 -4.06997532e-01 6.73518702e-02 5.83745062e-01
1.26835382e+00 -9.73501503e-01 2.21747741e-01 -7.00056195e-01
-2.67801821e-01 8.81596088e-01 4.39685255e-01 -6.70704424e-01
7.25533485e-01 3.25103551e-01 -4.42325503e-01 -7.29201734e-01
-7.11321473e-01 2.44405754e-02 4.39207315e-01 5.45335770e-01
1.66515216e-01 -4.45618518e-02 7.63709605e-01 5.21197096e-02
-6.71862662e-01 -1.40431792e-01 2.77064383e-01 6.91701353e-01
-6.28246665e-01 -7.39966571e-01 -2.55349815e-01 6.54840648e-01
1.68955550e-01 -2.10045278e-01 -1.98342130e-01 2.74769902e-01
1.07550301e-01 2.97525764e-01 5.62574744e-01 -1.93070427e-01
4.34201837e-01 -2.79821958e-02 8.27494442e-01 -1.13735056e+00
-1.01723239e-01 1.87675446e-01 -6.81030378e-02 -9.42730129e-01
-3.51237774e-01 -6.97173238e-01 -1.09156740e+00 -2.72962660e-01
7.67270029e-02 -1.99602365e-01 5.73738575e-01 8.47091258e-01
6.84910655e-01 6.06318593e-01 4.52343196e-01 -1.71510601e+00
1.66563392e-01 -7.56488502e-01 -3.98469448e-01 4.74723160e-01
4.33076829e-01 -6.11673415e-01 -1.56987146e-01 2.77581573e-01]
|
[8.957077980041504, -3.327646017074585]
|
6891c9bb-798a-43be-b982-6a33fd524921
|
synthetic-pre-training-tasks-for-neural
|
2212.09864
| null |
https://arxiv.org/abs/2212.09864v2
|
https://arxiv.org/pdf/2212.09864v2.pdf
|
Synthetic Pre-Training Tasks for Neural Machine Translation
|
Pre-training models with large crawled corpora can lead to issues such as toxicity and bias, as well as copyright and privacy concerns. A promising way of alleviating such concerns is to conduct pre-training with synthetic tasks and data, since no real-world information is ingested by the model. Our goal in this paper is to understand the factors that contribute to the effectiveness of pre-training models when using synthetic resources, particularly in the context of neural machine translation. We propose several novel approaches to pre-training translation models that involve different levels of lexical and structural knowledge, including: 1) generating obfuscated data from a large parallel corpus 2) concatenating phrase pairs extracted from a small word-aligned corpus, and 3) generating synthetic parallel data without real human language corpora. Our experiments on multiple language pairs reveal that pre-training benefits can be realized even with high levels of obfuscation or purely synthetic parallel data. We hope the findings from our comprehensive empirical analysis will shed light on understanding what matters for NMT pre-training, as well as pave the way for the development of more efficient and less toxic models.
|
['Rogerio Feris', 'Julian McAuley', 'Rameswar Panda', 'Graeme Blackwood', 'Zexue He']
|
2022-12-19
| null | null | null | null |
['nmt']
|
['computer-code']
|
[ 5.24750113e-01 1.19568102e-01 -1.50356725e-01 -5.94662540e-02
-1.13791251e+00 -9.56151307e-01 8.08689356e-01 6.71563968e-02
-7.59421349e-01 1.05184686e+00 2.75237679e-01 -8.63554180e-01
4.08008754e-01 -7.00187624e-01 -1.05407739e+00 -3.32614481e-01
2.80529886e-01 4.98019964e-01 -2.63849169e-01 -2.94598192e-01
2.04523131e-01 2.49319211e-01 -9.65483189e-01 4.98557121e-01
1.08553433e+00 1.92836508e-01 3.72120887e-02 3.36383641e-01
-9.14870575e-02 2.06863433e-01 -9.86976802e-01 -1.19776750e+00
5.76324999e-01 -5.25106728e-01 -7.59066224e-01 -5.92404306e-02
2.21214399e-01 -3.33427697e-01 9.25261974e-02 1.05905998e+00
6.32299721e-01 -3.04234177e-01 4.46619064e-01 -9.63967741e-01
-8.82291913e-01 9.43858147e-01 -3.25662673e-01 1.38229251e-01
1.81928024e-01 7.12529063e-01 9.06861424e-01 -7.03600526e-01
7.14476466e-01 1.15613079e+00 5.99125087e-01 7.33772874e-01
-1.44465280e+00 -8.11431944e-01 -3.70998561e-01 -2.12434649e-01
-1.07907248e+00 -6.61308289e-01 3.65829885e-01 -2.84294784e-01
8.51357520e-01 4.21679169e-01 3.50720674e-01 1.92757332e+00
2.49033734e-01 6.69549108e-01 1.35621440e+00 -6.60408437e-01
3.78510281e-02 5.80567479e-01 -2.71641314e-01 3.25477660e-01
7.26631761e-01 3.12367588e-01 -4.16398764e-01 -5.10380089e-01
3.30207944e-01 -4.16962773e-01 -2.53208995e-01 -8.52981135e-02
-1.38455820e+00 9.53446209e-01 -1.63130686e-01 4.01167870e-01
-1.65429041e-01 -1.42786443e-01 6.18652105e-01 5.11265039e-01
6.49349928e-01 1.13568652e+00 -6.31292522e-01 -1.82796746e-01
-7.95114279e-01 2.29789376e-01 7.55650401e-01 1.01735759e+00
5.64253569e-01 -2.89594475e-02 -2.99980454e-02 8.21746171e-01
-1.55089840e-01 6.34903908e-01 7.90890753e-01 -5.89754283e-01
1.07900500e+00 1.76728815e-01 2.60374665e-01 -5.78512311e-01
1.02095559e-01 -2.23200366e-01 -2.54178166e-01 -1.96088448e-01
6.49243176e-01 -6.98826909e-01 -6.72721803e-01 1.85948861e+00
7.32971206e-02 -2.59148479e-01 3.80419999e-01 5.83906770e-01
3.71020973e-01 6.70885444e-01 1.20106399e-01 -9.93831754e-02
1.40831888e+00 -7.90408373e-01 -4.98898745e-01 -3.36860776e-01
1.18811905e+00 -1.01828158e+00 1.54617500e+00 2.04306585e-03
-9.45731103e-01 -1.51676506e-01 -1.02385414e+00 1.94264241e-02
-4.65596288e-01 9.68387723e-02 4.94300902e-01 1.03541982e+00
-6.77579224e-01 6.91913247e-01 -7.67010629e-01 -5.75335324e-01
3.91576827e-01 2.45472103e-01 -4.24805284e-01 -1.74825713e-01
-1.30252779e+00 1.18448341e+00 2.69300133e-01 -2.04859331e-01
-6.84380829e-01 -5.29793441e-01 -6.28814518e-01 -9.99218747e-02
2.65570372e-01 -6.63815022e-01 1.19619906e+00 -1.11423385e+00
-1.25101423e+00 7.44046807e-01 9.48348343e-02 -4.08674419e-01
7.16308355e-01 -2.03109950e-01 -2.46954724e-01 -3.70377123e-01
2.78456509e-01 5.90636790e-01 5.74871778e-01 -1.14232683e+00
-2.55389392e-01 -3.19834113e-01 -2.32785195e-01 1.19947553e-01
-5.19818366e-01 3.92502457e-01 9.25994739e-02 -9.51451361e-01
-6.02077603e-01 -1.17154777e+00 -3.03650886e-01 -4.54784304e-01
-5.43072462e-01 1.52747288e-01 4.55080718e-01 -8.96144390e-01
7.81786740e-01 -1.84164417e+00 -1.45351082e-01 -8.57755244e-02
-2.74672985e-01 7.09507167e-01 -6.08535290e-01 7.54124105e-01
1.03641674e-01 7.76166558e-01 -2.03761265e-01 -2.05327883e-01
-2.09956303e-01 -2.02886071e-02 -5.82350433e-01 1.23514704e-01
4.86845464e-01 9.59485114e-01 -8.54238808e-01 -3.02630812e-01
-3.15310329e-01 2.83165038e-01 -4.44724113e-01 2.30731174e-01
-3.66356671e-01 4.85938936e-01 -4.61350203e-01 4.24958229e-01
4.22008812e-01 1.10923303e-02 2.10797280e-01 3.26539159e-01
-4.51490171e-02 8.50980282e-01 -3.84462178e-01 1.37860560e+00
-4.51386034e-01 7.51664162e-01 -2.06066862e-01 -6.84667766e-01
6.92765057e-01 4.87018079e-01 2.07939725e-02 -7.53684878e-01
3.35308313e-01 3.99821103e-01 2.31833532e-01 -5.28878093e-01
6.35202825e-01 -3.63051236e-01 -1.53964143e-02 9.59487796e-01
-4.92918827e-02 -7.11115971e-02 2.93744296e-01 6.68633729e-02
1.00622880e+00 7.58651346e-02 -6.08193539e-02 -1.40959635e-01
6.89989999e-02 3.64876658e-01 4.49742049e-01 5.72469056e-01
1.77203678e-02 3.87893051e-01 4.03005779e-01 -3.54808718e-01
-1.46620393e+00 -6.48936510e-01 3.64848077e-02 8.07141423e-01
-2.67743021e-01 -5.00923753e-01 -1.12161005e+00 -8.94259334e-01
-1.88034996e-01 1.14852798e+00 -3.65437120e-01 -3.44572783e-01
-7.23198116e-01 -1.12281728e+00 1.07654154e+00 2.07711756e-01
3.38494152e-01 -9.32140708e-01 -3.63953203e-01 1.38609573e-01
-5.95206499e-01 -1.14579034e+00 -5.53333104e-01 1.00180469e-01
-1.01960766e+00 -6.52811944e-01 -6.59569263e-01 -5.67912877e-01
7.95947134e-01 3.16724747e-01 9.59452629e-01 1.26420885e-01
1.60128735e-02 -1.62948415e-01 -3.89304399e-01 -6.64106965e-01
-1.09081662e+00 4.13088650e-01 1.89089075e-01 -3.08179170e-01
5.34600854e-01 -5.34703612e-01 -2.23542839e-01 4.79310423e-01
-9.79224145e-01 1.02043301e-01 8.40008914e-01 9.34774399e-01
6.34694621e-02 -3.72324556e-01 6.07186317e-01 -1.25482512e+00
1.09507418e+00 -4.23234880e-01 -5.67706287e-01 4.67307776e-01
-7.56294191e-01 2.54421294e-01 8.60240281e-01 -8.36608648e-01
-1.00718880e+00 -2.94112027e-01 -1.67412236e-01 2.64819730e-02
-4.04528938e-02 4.08676356e-01 -1.45547688e-01 2.14614384e-02
1.17731607e+00 3.43710870e-01 6.23891987e-02 -5.93486249e-01
5.22771299e-01 7.96469092e-01 -2.93480176e-02 -8.69245112e-01
1.01750207e+00 1.31374434e-01 -5.13307035e-01 -7.29931653e-01
-4.63412464e-01 4.34965566e-02 -4.43432063e-01 4.18975711e-01
5.76694548e-01 -9.18940604e-01 -7.92219639e-02 2.26804867e-01
-1.18207216e+00 -4.86793637e-01 -2.18564361e-01 4.73606795e-01
-3.63177836e-01 4.63728517e-01 -8.87093186e-01 -4.76916194e-01
-5.21543801e-01 -1.28207028e+00 7.60147214e-01 -1.27989367e-01
-6.04489028e-01 -7.44252741e-01 3.52872819e-01 8.65740240e-01
3.69767398e-01 -4.06051651e-02 1.12349677e+00 -9.69459653e-01
-6.14568353e-01 -2.77801991e-01 3.34188342e-02 4.58903641e-01
2.29841322e-01 1.74563341e-02 -7.98596203e-01 -4.62762266e-01
1.46169811e-01 -6.73966229e-01 3.48944545e-01 -2.66727477e-01
6.04177296e-01 -8.58331442e-01 -2.91959405e-01 3.08048278e-01
1.08062899e+00 4.01574522e-02 5.84984958e-01 3.83514464e-01
5.15146673e-01 7.82419682e-01 4.61435437e-01 -4.27744538e-02
9.87177938e-02 5.12066603e-01 -7.49801695e-02 -5.14231920e-02
2.63708737e-02 -6.77778006e-01 6.29489779e-01 8.56790602e-01
1.90814272e-01 -4.64629799e-01 -7.94920683e-01 6.49383008e-01
-1.38192618e+00 -9.44618404e-01 6.26384318e-02 2.37471604e+00
1.19405806e+00 1.33299723e-01 2.13264823e-01 -2.11131319e-01
7.08148837e-01 -1.28079832e-01 -2.85295099e-01 -6.38974011e-01
-1.67286023e-01 2.58608758e-01 7.87276745e-01 2.59051651e-01
-7.39803135e-01 1.16631794e+00 6.88299847e+00 7.72616982e-01
-1.39247644e+00 2.05946848e-01 7.38361657e-01 -1.51510432e-01
-6.24227703e-01 1.59744531e-01 -6.92056954e-01 6.28906727e-01
1.45290244e+00 -3.38566989e-01 5.75994730e-01 7.35133052e-01
2.14118853e-01 2.54018247e-01 -1.15121806e+00 3.55226725e-01
-1.62023457e-03 -1.38710463e+00 2.34663218e-01 4.92163658e-01
7.08168864e-01 3.00241828e-01 1.27183959e-01 3.72401834e-01
6.48979008e-01 -1.00371563e+00 7.09660947e-01 -3.06694955e-01
8.16747904e-01 -5.38210988e-01 6.56065941e-01 8.21059406e-01
-3.56393576e-01 1.48336157e-01 -5.28377593e-01 -1.64300576e-01
3.75358909e-02 3.99044335e-01 -1.35654926e+00 4.34961677e-01
1.73360482e-01 1.68355897e-01 -5.85807800e-01 6.00065887e-01
-2.69109398e-01 9.00564015e-01 -3.25113118e-01 -3.17817479e-01
8.63803253e-02 -2.27229506e-01 3.59186053e-01 1.26079476e+00
4.72684324e-01 -1.13162205e-01 -3.48385483e-01 9.91173387e-01
-4.11246389e-01 3.92303765e-01 -1.11723757e+00 -7.47452497e-01
5.63078523e-01 9.81141746e-01 -5.40488720e-01 -2.41353780e-01
-5.95303953e-01 9.39553082e-01 4.54872727e-01 3.94196332e-01
-6.51212037e-01 -1.90544531e-01 6.88752770e-01 2.33252555e-01
-9.38110147e-03 -3.68401945e-01 -6.79578841e-01 -1.46873498e+00
9.52722877e-02 -1.54569900e+00 1.70191780e-01 -4.69015509e-01
-1.17330706e+00 7.20972657e-01 -1.07496113e-01 -1.07291651e+00
-2.83678740e-01 -3.94207954e-01 -5.49766898e-01 1.01703072e+00
-1.14033151e+00 -1.13184607e+00 4.66972977e-01 1.19512193e-01
6.08229399e-01 -2.51107752e-01 9.74508226e-01 2.29749128e-01
-6.75969541e-01 1.05007184e+00 3.15100819e-01 3.09428275e-01
9.54650104e-01 -6.54466391e-01 1.11115801e+00 9.69524622e-01
4.97986317e-01 1.06564832e+00 7.26085365e-01 -7.29917347e-01
-1.28783810e+00 -1.09999466e+00 1.19953251e+00 -9.88740444e-01
6.00818515e-01 -6.77031100e-01 -8.21442962e-01 7.77750850e-01
3.70029688e-01 -7.04518199e-01 9.58871603e-01 1.49928316e-01
-5.28357923e-01 2.84728795e-01 -1.26857376e+00 1.02400446e+00
8.65425587e-01 -5.65700352e-01 -6.14009142e-01 5.57175994e-01
9.66801405e-01 -9.47652534e-02 -6.78547978e-01 2.08651006e-01
3.96979362e-01 -4.67528522e-01 7.82722414e-01 -9.72583234e-01
8.09235752e-01 1.35109901e-01 -2.76882090e-02 -1.57899678e+00
6.76221848e-02 -7.97610939e-01 5.08458018e-01 1.29328597e+00
1.07257473e+00 -7.75666893e-01 8.22273135e-01 8.28874528e-01
2.04946637e-01 -5.50353348e-01 -6.91953599e-01 -1.12251794e+00
6.83123767e-01 -1.73665613e-01 6.93444610e-01 1.20218170e+00
1.72805816e-01 6.84345424e-01 -6.01138771e-01 -9.46616828e-02
4.19582605e-01 -2.22726211e-01 1.06226516e+00 -6.43186808e-01
-2.83802092e-01 -1.15605928e-01 9.68236029e-02 -8.24050248e-01
2.14502648e-01 -9.69679415e-01 -3.88412587e-02 -9.75870550e-01
2.69746542e-01 -5.06206393e-01 3.08583140e-01 4.36030537e-01
-3.46345514e-01 1.97340995e-01 2.38390788e-01 4.70924765e-01
1.18954383e-01 5.10411680e-01 1.27229989e+00 -4.24155332e-02
3.03879268e-02 -5.43278158e-02 -1.03088403e+00 3.32315177e-01
9.75363851e-01 -9.93576467e-01 -5.11572063e-01 -7.79655099e-01
8.47829655e-02 -1.04581982e-01 -1.56534072e-02 -4.70954776e-01
-1.77851334e-01 -4.82484549e-01 2.59026606e-02 -9.24622566e-02
1.73208624e-01 -6.13375843e-01 1.44516021e-01 4.90628123e-01
-5.53696692e-01 3.33545864e-01 3.35125357e-01 3.07923555e-01
1.73698261e-01 -4.01659459e-01 6.15874588e-01 -3.82014066e-01
7.11463168e-02 -8.62227306e-02 -5.28167248e-01 2.45756671e-01
7.78656483e-01 -6.36401474e-02 -6.43624544e-01 -3.33176374e-01
-1.29609123e-01 -1.45398267e-02 9.20184016e-01 5.66328526e-01
8.90528038e-02 -1.11855423e+00 -8.95139754e-01 2.59355158e-01
1.65472388e-01 -4.20417756e-01 -3.25065374e-01 4.10906702e-01
-5.07078707e-01 5.36547959e-01 -3.35698456e-01 -2.14917883e-01
-1.21459222e+00 6.50078416e-01 -1.37493142e-03 -3.07617724e-01
-4.02245224e-01 6.12203896e-01 1.65871069e-01 -6.69515967e-01
-2.03967504e-02 -2.39171814e-02 3.85490090e-01 -3.42080265e-01
3.53760153e-01 1.33813843e-01 2.46892229e-01 -5.50283968e-01
-8.29242542e-02 -2.45974232e-02 -4.27690744e-01 -6.37571514e-01
1.15481830e+00 9.13556889e-02 -1.35176741e-02 7.95838535e-02
1.17342854e+00 3.14845800e-01 -7.99689591e-01 -2.55937040e-01
1.39771536e-01 -6.39595091e-01 -5.63589334e-01 -9.64666486e-01
-6.82728171e-01 9.32242334e-01 1.29205063e-01 1.19910030e-04
6.43281937e-01 -2.87929893e-01 1.12958515e+00 5.65162778e-01
6.56268418e-01 -9.59162414e-01 2.16850638e-02 3.14662814e-01
7.09570169e-01 -1.13886392e+00 -8.82169306e-02 -2.96742111e-01
-7.69704282e-01 7.06820607e-01 5.15301883e-01 1.89286336e-01
5.52691258e-02 1.92957267e-01 4.09100473e-01 1.44882739e-01
-9.00918782e-01 1.46774322e-01 -1.46396935e-01 5.87277770e-01
5.73673308e-01 1.01976290e-01 -7.92264044e-01 4.31973010e-01
-5.36828995e-01 -1.36868715e-01 7.03346848e-01 9.43249106e-01
1.41416611e-02 -1.85537744e+00 -4.65821564e-01 3.00297827e-01
-9.11790490e-01 -4.53356057e-01 -1.00798523e+00 7.36075044e-01
-1.15839586e-01 1.04260933e+00 -3.52296978e-01 -6.23787940e-01
1.42655775e-01 4.07027006e-01 3.15182835e-01 -8.10661852e-01
-8.20230186e-01 -9.34650153e-02 7.21390009e-01 -8.07726979e-02
2.41115540e-01 -7.22941637e-01 -5.99673450e-01 -6.03907466e-01
-5.56414723e-01 3.50786179e-01 8.23680937e-01 8.14630270e-01
6.46644592e-01 -1.11486219e-01 5.27158380e-01 -4.84028399e-01
-9.77243602e-01 -1.11056805e+00 1.23858906e-01 5.12198269e-01
-4.94741537e-02 -1.41553298e-01 -2.74517924e-01 1.00641251e-01]
|
[11.571885108947754, 10.20050048828125]
|
2816c351-be36-47e2-a095-db2d1129b9a9
|
kg-cruse-recurrent-walks-over-knowledge-graph
| null | null |
https://aclanthology.org/2022.nlp4convai-1.9
|
https://aclanthology.org/2022.nlp4convai-1.9.pdf
|
KG-CRuSE: Recurrent Walks over Knowledge Graph for Explainable Conversation Reasoning using Semantic Embeddings
|
Knowledge-grounded dialogue systems utilise external knowledge such as knowledge graphs to generate informative and appropriate responses. A crucial challenge of such systems is to select facts from a knowledge graph pertinent to the dialogue context for response generation. This fact selection can be formulated as path traversal over a knowledge graph conditioned on the dialogue context. Such paths can originate from facts mentioned in the dialogue history and terminate at the facts to be mentioned in the response. These walks, in turn, provide an explanation of the flow of the conversation. This work proposes KG-CRuSE, a simple, yet effective LSTM based decoder that utilises the semantic information in the dialogue history and the knowledge graph elements to generate such paths for effective conversation explanation. Extensive evaluations showed that our model outperforms the state-of-the-art models on the OpenDialKG dataset on multiple metrics.
|
['John McCrae', 'Mihael Arcan', 'Rajdeep Sarkar']
| null | null | null | null |
nlp4convai-acl-2022-5
|
['fact-selection']
|
['natural-language-processing']
|
[ 3.14851999e-01 1.07915437e+00 -9.61387604e-02 -5.18213451e-01
-7.30426610e-01 -5.48848331e-01 8.39460850e-01 2.51649171e-01
-2.10415736e-01 1.13871253e+00 9.20155942e-01 -2.30407923e-01
-4.72013094e-02 -1.03243387e+00 -4.55329359e-01 -9.67178345e-02
1.55957162e-01 1.06466067e+00 2.75608987e-01 -8.64819705e-01
4.05290395e-01 -4.32298146e-02 -8.80443335e-01 9.33487236e-01
6.24124765e-01 5.17644048e-01 1.80051446e-01 1.00367057e+00
-8.70803297e-01 1.18468237e+00 -7.99692690e-01 -7.45933652e-01
-4.75478858e-01 -1.04508948e+00 -1.82521331e+00 5.93390651e-02
-1.50707781e-01 -3.37967604e-01 -4.03680742e-01 6.36903524e-01
4.71728325e-01 5.93106866e-01 4.66261834e-01 -8.58658195e-01
-3.29734474e-01 1.40537882e+00 3.66723269e-01 2.35998288e-01
1.03336430e+00 9.79727358e-02 1.14895904e+00 -5.53166807e-01
9.28986132e-01 1.55195928e+00 3.80286813e-01 8.60114157e-01
-8.41130614e-01 1.08659104e-01 1.41783372e-01 5.74883163e-01
-6.71057284e-01 -6.73910022e-01 6.99537516e-01 -1.00595117e-01
1.38649690e+00 2.67240107e-01 7.26558030e-01 1.16006005e+00
1.00882195e-01 7.70577490e-01 6.25806749e-01 -6.21502936e-01
1.92494690e-01 1.02599479e-01 2.89223552e-01 8.59410465e-01
-5.32693565e-01 -3.43808353e-01 -1.13524926e+00 -2.98319012e-01
6.45677567e-01 -9.67830896e-01 -4.77712393e-01 2.11983889e-01
-1.04209006e+00 1.13141012e+00 3.51275623e-01 7.22032860e-02
-6.33073270e-01 -6.90514222e-02 5.67283869e-01 4.02091205e-01
2.34543294e-01 8.43776941e-01 -4.78330463e-01 -4.53665257e-01
-3.04995537e-01 3.94222409e-01 1.68749189e+00 8.39275837e-01
7.71011651e-01 -1.85243618e-02 -6.14928007e-01 6.70907080e-01
3.73219520e-01 6.33960962e-02 1.89966783e-01 -1.06084788e+00
7.54257560e-01 7.98288167e-01 2.16505364e-01 -1.02885306e+00
-4.02948707e-01 -8.10792949e-03 -2.10036114e-01 -4.79838192e-01
6.10130429e-01 -5.16530931e-01 -6.14551067e-01 1.60971808e+00
5.83293498e-01 1.05327338e-01 6.29965603e-01 8.82806778e-01
1.32758772e+00 8.64934027e-01 1.86641261e-01 -1.99076131e-01
1.34821880e+00 -8.83418918e-01 -8.25715363e-01 -4.16962385e-01
6.66056216e-01 -5.30365765e-01 7.38072455e-01 -4.82967272e-02
-1.06043839e+00 -1.95258975e-01 -6.92219257e-01 -9.21393409e-02
3.87618691e-02 -3.56975704e-01 5.95312774e-01 1.41734421e-01
-1.09432459e+00 4.70123559e-01 -2.91691333e-01 -5.55393934e-01
-1.61560506e-01 1.42712459e-01 -2.10999325e-03 1.82409342e-02
-1.90212262e+00 1.19502091e+00 7.99260855e-01 3.52885336e-01
-8.74259472e-01 -2.40630567e-01 -1.00857234e+00 1.40572533e-01
7.75263608e-01 -9.96280193e-01 1.71616387e+00 -7.42723107e-01
-2.09416723e+00 8.09298873e-01 -2.85341859e-01 -7.54832387e-01
3.05364966e-01 -1.29163712e-01 -1.45251483e-01 3.76520306e-01
-9.95599404e-02 5.18296540e-01 5.18424571e-01 -1.12315512e+00
-6.84793353e-01 -6.67142570e-02 6.18881881e-01 8.61170888e-01
3.81258875e-01 -1.18690291e-02 -6.71699286e-01 1.83196187e-01
-2.17662185e-01 -8.68875146e-01 -2.86034375e-01 -9.83948767e-01
-1.00053310e+00 -4.62390393e-01 3.48626941e-01 -9.79936481e-01
1.16416371e+00 -1.39963222e+00 3.02155912e-01 1.54205024e-01
6.48530871e-02 2.66007781e-01 7.70107703e-03 1.09548748e+00
4.85807538e-01 -6.57708123e-02 8.30782503e-02 2.89258622e-02
1.01674967e-01 2.11492881e-01 -5.17132163e-01 -1.27005935e-01
1.61432736e-02 1.06942809e+00 -1.17184663e+00 -5.63403606e-01
8.69908929e-02 2.06794739e-01 -3.06094140e-01 6.36976421e-01
-1.00908899e+00 4.74296838e-01 -8.26879680e-01 -2.52730370e-01
-2.32521355e-01 -3.56272578e-01 4.91344094e-01 -2.85886936e-02
3.15420389e-01 1.13264775e+00 -7.83716440e-01 1.68312347e+00
-4.81015325e-01 5.52002966e-01 -1.40189290e-01 -6.03334606e-01
8.97316873e-01 6.68878853e-01 -2.02272058e-01 -4.52652514e-01
1.39385149e-01 1.34428022e-02 1.03070430e-01 -6.34500444e-01
8.58273685e-01 -1.19671285e-01 -2.68756092e-01 7.66107082e-01
2.38751292e-01 -1.22289598e-01 1.60508275e-01 7.61236727e-01
8.90389740e-01 1.87921673e-01 3.98700535e-01 2.08673507e-01
7.06007957e-01 4.13416415e-01 1.64854124e-01 9.14127231e-01
2.21975625e-01 2.31181849e-02 7.06067860e-01 -3.28701913e-01
-4.92034018e-01 -5.71396947e-01 6.39030337e-01 1.14155328e+00
-5.84408864e-02 -5.30729651e-01 -1.01227617e+00 -8.91712308e-01
-4.78515178e-01 1.25293219e+00 -3.71437430e-01 -2.19739825e-01
-7.72276402e-01 -1.81776658e-01 6.05288625e-01 1.46865577e-01
5.83831310e-01 -1.64022648e+00 -4.79431421e-01 6.89378381e-01
-9.26213145e-01 -1.35158491e+00 -2.59047508e-01 -1.40359297e-01
-5.25980890e-01 -1.28905082e+00 -2.58478783e-02 -6.29315555e-01
3.30547929e-01 -1.73762187e-01 1.38149631e+00 1.86430454e-01
2.61588007e-01 5.50996363e-01 -5.44044673e-01 -1.51929691e-01
-1.16011178e+00 2.51893580e-01 -5.35123646e-01 1.37072191e-01
2.70180523e-01 -1.32813156e-01 -3.07268769e-01 1.57934561e-01
-4.12291259e-01 5.46346784e-01 3.14544217e-04 8.08799624e-01
2.01466188e-01 -6.03344850e-02 1.01075149e+00 -1.17895818e+00
1.18403494e+00 -6.49964392e-01 -8.57096836e-02 4.19882298e-01
-1.31359011e-01 5.03116727e-01 7.60158956e-01 1.05879009e-01
-1.61552429e+00 -2.61515707e-01 -3.39960337e-01 3.40145260e-01
-1.03700988e-01 1.03115273e+00 -1.49476171e-01 4.43649560e-01
7.91353464e-01 2.47918695e-01 -2.94427276e-01 -1.73859209e-01
8.09499741e-01 5.39287686e-01 5.95905304e-01 -7.62656868e-01
4.51640375e-02 3.74404751e-02 -1.50222644e-01 -8.27283502e-01
-1.08425272e+00 -3.66533250e-01 -4.97733355e-01 -5.40870070e-01
8.28085840e-01 -4.61969525e-01 -8.82815719e-01 3.23760927e-01
-1.55773795e+00 -5.81629038e-01 -2.11438373e-01 1.98355749e-01
-6.55650556e-01 1.97738603e-01 -6.95392489e-01 -9.88055706e-01
-5.72622120e-01 -7.36134946e-01 5.87004781e-01 4.48646218e-01
-7.02180386e-01 -1.47176993e+00 -1.13132142e-01 6.15967751e-01
2.30657145e-01 1.63451552e-01 1.12697804e+00 -1.25494075e+00
-6.84045017e-01 -3.66575941e-02 1.50784597e-01 -2.61341095e-01
5.85604012e-02 -1.51624560e-01 -8.39606047e-01 2.64412880e-01
-1.47101507e-01 -5.76273561e-01 5.39890707e-01 1.13186635e-01
2.35548452e-01 -9.40999925e-01 -2.89844543e-01 -5.06399088e-02
9.00548220e-01 2.42535606e-01 5.67828119e-01 1.97141990e-01
5.95825493e-01 1.06765926e+00 4.90423352e-01 3.01149935e-01
9.56877828e-01 8.63685310e-01 1.73017293e-01 2.33747840e-01
-1.59510955e-01 -7.14669704e-01 3.24335963e-01 7.76548862e-01
2.36488611e-01 -5.88272393e-01 -8.34324837e-01 8.14131677e-01
-2.03464437e+00 -1.17794657e+00 -4.81765836e-01 1.91496170e+00
1.35630548e+00 1.71308935e-01 3.12537514e-02 -2.21097276e-01
8.47819090e-01 2.88979799e-01 -3.50495666e-01 -9.90731478e-01
3.99855785e-02 8.28192197e-03 -3.13000120e-02 1.22309172e+00
-4.69175935e-01 1.39275515e+00 5.51225376e+00 4.15776879e-01
-7.42298961e-01 -3.94283757e-02 3.98186356e-01 2.01027095e-01
-4.66362715e-01 3.05857807e-01 -1.06501698e+00 4.89689447e-02
1.32473195e+00 -5.54078877e-01 4.15311754e-01 3.79134327e-01
3.38016689e-01 -2.21361682e-01 -1.18755329e+00 2.84077227e-01
7.24187344e-02 -1.57606375e+00 3.13943297e-01 -3.41308147e-01
4.15942103e-01 -2.14323312e-01 -6.21973157e-01 3.79754424e-01
9.14944530e-01 -8.96861196e-01 3.09761494e-01 6.19207084e-01
3.68657500e-01 -7.65533805e-01 6.24882221e-01 5.08079827e-01
-8.11868310e-01 2.38603339e-01 -1.23827584e-01 3.80212069e-02
6.99564159e-01 2.45953813e-01 -1.94863677e+00 7.69384682e-01
4.47238013e-02 8.30679312e-02 7.61252129e-03 6.86459184e-01
-1.07483804e+00 7.73650229e-01 -6.53291717e-02 -2.78608203e-01
4.83046353e-01 1.03256583e-01 8.10868680e-01 1.27214980e+00
-1.14421263e-01 4.47019547e-01 3.33958000e-01 7.84173071e-01
-2.26420105e-01 1.88417554e-01 -4.87058967e-01 -2.00462833e-01
5.17512739e-01 1.02089953e+00 -4.00026709e-01 -4.58666533e-01
-8.61365348e-02 1.09240174e+00 4.34269190e-01 4.17292565e-01
-3.84211987e-01 -3.17995012e-01 2.57765234e-01 -1.88539624e-01
1.96049139e-02 9.34478641e-02 6.44386262e-02 -8.09224904e-01
-2.99768388e-01 -9.83717620e-01 7.10483074e-01 -9.06434178e-01
-8.62267733e-01 8.50687087e-01 -5.19043878e-02 -3.91367108e-01
-1.10623109e+00 -9.39398706e-02 -8.22718203e-01 1.23313236e+00
-1.43045139e+00 -1.02548611e+00 -9.31988284e-02 6.20137930e-01
7.94381917e-01 -1.96261182e-02 1.16699898e+00 -4.49196845e-01
-2.37050980e-01 8.82693902e-02 -6.62578285e-01 2.74302274e-01
3.86436999e-01 -1.31713939e+00 8.54894102e-01 7.02087283e-01
2.24308491e-01 6.15589917e-01 8.43647242e-01 -1.03082526e+00
-1.28759646e+00 -7.66400039e-01 1.50615489e+00 -3.94017160e-01
5.15653193e-01 -2.89392266e-02 -1.15201116e+00 7.08838046e-01
6.19115353e-01 -9.74648237e-01 7.32822835e-01 2.84606218e-01
1.25808129e-02 4.71795827e-01 -8.21759939e-01 7.56732643e-01
8.24986875e-01 -5.66802144e-01 -1.16216862e+00 3.34536016e-01
8.54229927e-01 -9.07302141e-01 -6.33131266e-01 -1.96466804e-01
2.32948616e-01 -6.91505313e-01 6.40852511e-01 -1.11176884e+00
6.74966991e-01 7.93993324e-02 2.57038474e-01 -1.74068999e+00
-3.73925455e-03 -1.31457460e+00 -3.92913371e-01 1.16649544e+00
9.04616654e-01 -3.14676613e-01 6.02656364e-01 9.11109626e-01
-3.51964563e-01 -6.05284631e-01 -7.30171025e-01 -7.47223869e-02
-4.26386923e-01 -2.14687377e-01 5.61211705e-01 8.95456374e-01
6.91301286e-01 1.17523897e+00 -4.56979901e-01 3.57389078e-02
3.59752923e-01 2.72333056e-01 8.60705554e-01 -9.83856201e-01
-3.93612236e-01 -2.32260719e-01 1.70462176e-01 -1.18647349e+00
4.41528499e-01 -8.70781243e-01 3.15259755e-01 -2.46181273e+00
-2.14697897e-01 -1.75800487e-01 4.34780121e-01 5.31032145e-01
-5.51077187e-01 -6.39814198e-01 1.48500130e-01 -2.95370049e-03
-6.73398614e-01 5.26161194e-01 1.33128393e+00 3.54656428e-02
-6.33370817e-01 7.48010501e-02 -7.76792645e-01 5.99819481e-01
9.10540938e-01 -2.41870299e-01 -7.55173445e-01 -3.45922440e-01
3.68682414e-01 1.03440177e+00 1.10961892e-01 -4.35826391e-01
6.80178821e-01 -3.87826711e-01 -9.08129886e-02 -2.80578583e-01
6.12323999e-01 -1.03266932e-01 -5.72269224e-02 1.52480319e-01
-9.53065753e-01 -2.28980303e-01 2.09963024e-01 5.59610546e-01
-1.65618286e-01 -3.16019595e-01 3.02190781e-01 -5.17993629e-01
-8.89869690e-01 -1.33585125e-01 -6.39608145e-01 5.14854968e-01
4.87424612e-01 -2.73829520e-01 -6.08360291e-01 -1.12946343e+00
-8.68467629e-01 5.28017759e-01 -8.58261734e-02 4.37846184e-01
8.44642758e-01 -9.66454744e-01 -1.00061858e+00 -3.41496050e-01
-8.15741941e-02 1.58897489e-02 2.40385801e-01 4.78978604e-01
-2.63679653e-01 7.90031314e-01 -8.88130143e-02 -6.25965148e-02
-1.17121434e+00 3.94770550e-03 5.87192297e-01 -5.90196431e-01
-7.73252606e-01 1.00992501e+00 -1.10780904e-02 -4.10258919e-01
5.14510311e-02 9.10204835e-03 -8.27938497e-01 1.59864113e-01
6.24589622e-01 1.77765220e-01 1.01630233e-01 -6.61764860e-01
-1.15833707e-01 -1.91011086e-01 1.19465236e-04 -7.01417506e-01
1.01276708e+00 -4.00180489e-01 -9.39476937e-02 4.94625330e-01
6.95440233e-01 1.97371002e-02 -9.53161776e-01 -5.95090687e-01
2.35367224e-01 -2.51127571e-01 -2.84380049e-01 -1.31805015e+00
-5.65395653e-01 5.86200058e-01 -6.25418007e-01 6.34545445e-01
5.80027401e-01 6.73740059e-02 1.05378628e+00 7.02030063e-01
4.66427982e-01 -1.08583498e+00 1.12870194e-01 1.07600057e+00
1.04828370e+00 -8.65537941e-01 -2.90152103e-01 -4.79812354e-01
-1.32315981e+00 1.24612880e+00 7.18306482e-01 4.31138605e-01
-2.04833448e-01 -2.47339725e-01 2.65414417e-01 -6.55633390e-01
-1.19323635e+00 -2.39441842e-01 2.71167815e-01 5.42299628e-01
4.44041520e-01 -7.63236880e-02 -1.80572778e-01 5.39264917e-01
-5.06752014e-01 -9.65750292e-02 8.65267158e-01 6.06768191e-01
-5.73724210e-01 -1.03484905e+00 -1.11945689e-01 3.47559482e-01
-3.83554488e-01 -2.64270902e-01 -1.20497561e+00 3.49753648e-01
-6.82793856e-01 1.55685639e+00 -3.90232772e-01 -2.56953061e-01
4.46603537e-01 7.08462775e-01 3.74348432e-01 -1.11057365e+00
-1.07565212e+00 -3.20689619e-01 1.46585178e+00 -5.11349678e-01
-2.89447308e-01 -2.36139268e-01 -1.81022406e+00 -3.00254107e-01
-3.27857435e-01 7.52901793e-01 4.56809014e-01 1.35060823e+00
3.56079906e-01 5.74249208e-01 3.98819685e-01 -4.05361444e-01
-4.04048234e-01 -8.70582402e-01 -1.36924675e-02 2.68772513e-01
4.22105156e-02 -2.69763499e-01 -8.03116933e-02 8.75770375e-02]
|
[12.460657119750977, 8.137899398803711]
|
7971a063-78c5-4faa-93f2-88da483a5817
|
learning-from-task-descriptions
|
2011.08115
| null |
https://arxiv.org/abs/2011.08115v1
|
https://arxiv.org/pdf/2011.08115v1.pdf
|
Learning from Task Descriptions
|
Typically, machine learning systems solve new tasks by training on thousands of examples. In contrast, humans can solve new tasks by reading some instructions, with perhaps an example or two. To take a step toward closing this gap, we introduce a framework for developing NLP systems that solve new tasks after reading their descriptions, synthesizing prior work in this area. We instantiate this framework with a new English language dataset, ZEST, structured for task-oriented evaluation on unseen tasks. Formulating task descriptions as questions, we ensure each is general enough to apply to many possible inputs, thus comprehensively evaluating a model's ability to solve each task. Moreover, the dataset's structure tests specific types of systematic generalization. We find that the state-of-the-art T5 model achieves a score of 12% on ZEST, leaving a significant challenge for NLP researchers.
|
['Matthew E. Peters', 'Matt Gardner', 'Nicholas Lourie', 'Orion Weller']
|
2020-11-16
| null |
https://aclanthology.org/2020.emnlp-main.105
|
https://aclanthology.org/2020.emnlp-main.105.pdf
|
emnlp-2020-11
|
['systematic-generalization']
|
['reasoning']
|
[ 3.64643455e-01 5.36972821e-01 -6.92601427e-02 -6.70177341e-01
-8.87567937e-01 -1.17028701e+00 5.98689735e-01 -6.65507913e-02
-4.85853910e-01 8.74612629e-01 1.19840220e-01 -5.36839426e-01
-1.35572493e-01 -5.21160245e-01 -6.61240876e-01 -9.87555459e-02
3.75861436e-01 9.58915234e-01 3.65894079e-01 -4.10993725e-01
2.30826348e-01 1.30249515e-01 -1.46000695e+00 7.72642255e-01
9.53528047e-01 7.18829453e-01 4.74960417e-01 5.32084465e-01
-2.96166569e-01 5.86240590e-01 -7.85358131e-01 -7.13995278e-01
2.93188334e-01 -1.53753191e-01 -1.63623905e+00 -3.11261296e-01
7.99721122e-01 -2.58998841e-01 -1.39312938e-01 5.29714525e-01
1.55664682e-01 4.58158433e-01 7.82967746e-01 -1.19362009e+00
-1.19523370e+00 7.82579243e-01 3.17516834e-01 3.34385931e-01
5.91117561e-01 5.33896685e-01 1.36479723e+00 -9.47035134e-01
8.59189272e-01 1.05797482e+00 4.46009725e-01 1.20527005e+00
-1.27347088e+00 -5.55609107e-01 2.15661138e-01 2.27453604e-01
-8.50661278e-01 -3.81282866e-01 2.81803310e-01 -3.30420077e-01
1.61336923e+00 3.95139128e-01 2.30903536e-01 1.51449227e+00
4.46993187e-02 1.02687252e+00 1.01540864e+00 -3.13969880e-01
5.53817004e-02 2.53410012e-01 4.71743107e-01 2.35875860e-01
2.59175956e-01 4.29981016e-02 -3.95948410e-01 2.16224387e-01
3.23477715e-01 -4.59942222e-01 -3.13403904e-01 -1.22340359e-01
-1.31289673e+00 6.34140551e-01 3.40667218e-01 4.35302943e-01
9.26810503e-02 5.93709946e-03 2.82757580e-01 6.80693448e-01
2.80239075e-01 1.55974150e+00 -1.31378043e+00 -1.77670449e-01
-6.28935218e-01 5.90674222e-01 1.11024809e+00 1.14438117e+00
6.96238160e-01 -1.78035900e-01 -3.55047464e-01 6.72092974e-01
-1.34126231e-01 3.09615403e-01 7.55540192e-01 -1.15655208e+00
8.50829422e-01 3.91200393e-01 2.15777218e-01 -4.26757842e-01
-5.67508698e-01 -5.86606681e-01 -3.46594572e-01 -3.96313459e-01
7.40523517e-01 -3.73468161e-01 -7.56902575e-01 1.88157105e+00
-1.91118252e-02 -3.21649462e-02 3.94827276e-01 4.66374010e-01
9.41942930e-01 9.05669987e-01 1.53771251e-01 -1.11508138e-01
1.31721187e+00 -1.19328833e+00 -5.38946569e-01 -1.01320076e+00
9.10371065e-01 -5.31868100e-01 1.73598456e+00 6.27410710e-01
-1.02068651e+00 -7.43130803e-01 -6.57228112e-01 -6.03710413e-01
-6.74756587e-01 -6.18823878e-02 6.52055621e-01 5.19032717e-01
-1.24631405e+00 3.22078854e-01 -3.36692721e-01 -4.78858918e-01
3.41855623e-02 2.11998239e-01 -2.45721608e-01 -3.86917442e-01
-1.36506391e+00 1.40314388e+00 8.78736854e-01 -3.09030503e-01
-1.01792157e+00 -8.72678101e-01 -9.37887788e-01 3.06621343e-01
6.64330661e-01 -9.52560842e-01 1.87001657e+00 -4.51989233e-01
-1.20323157e+00 8.21901977e-01 -4.68810260e-01 -3.82274300e-01
1.91010714e-01 -1.62983969e-01 -3.57617289e-01 -1.02848381e-01
3.18530381e-01 1.02415562e+00 5.85553885e-01 -1.17978430e+00
-6.66061163e-01 -1.27380759e-01 4.16424811e-01 1.79874465e-01
-4.16300774e-01 -8.85825008e-02 -5.91110997e-02 -3.86901945e-01
-7.68400133e-02 -8.05426121e-01 -6.82589188e-02 -5.80395520e-01
-1.96352571e-01 -7.79154360e-01 4.05092418e-01 -4.58829224e-01
1.06885350e+00 -1.87208700e+00 2.43866220e-01 -4.21473503e-01
4.08064008e-01 1.74073458e-01 -7.30121434e-01 3.62698376e-01
-7.04125762e-02 6.41101360e-01 -1.67375035e-03 -3.46546918e-01
3.82185191e-01 2.90332019e-01 -6.64642870e-01 -4.98382151e-01
5.12131155e-01 1.39597178e+00 -1.23897076e+00 -5.33704907e-02
-1.02168016e-01 -9.74960402e-02 -6.47245884e-01 1.74678937e-01
-8.29713166e-01 3.30763042e-01 -4.77185547e-01 3.99237871e-01
3.00868422e-01 -5.81677377e-01 5.32072261e-02 2.28713825e-01
9.03468207e-02 9.35822129e-01 -7.06344306e-01 1.68037593e+00
-6.95016384e-01 7.60462523e-01 -3.95396262e-01 -8.89828503e-01
8.50390017e-01 3.39586496e-01 -2.80633736e-02 -4.52193111e-01
-1.26153290e-01 2.41691574e-01 1.24368317e-01 -8.21132123e-01
4.83523488e-01 -1.68681264e-01 -2.46029139e-01 7.27945507e-01
4.90599394e-01 -6.55604541e-01 4.53593373e-01 1.45283818e-01
1.43093586e+00 -1.53483391e-01 2.89997935e-01 -2.55361587e-01
2.51421511e-01 3.49239498e-01 1.92180485e-01 1.22526860e+00
-3.25574875e-01 3.06429654e-01 2.97017306e-01 -8.07102025e-01
-9.30250943e-01 -1.12107503e+00 -4.65836329e-03 1.50799239e+00
-3.88543487e-01 -4.43033278e-01 -6.79823637e-01 -1.11976731e+00
-6.09387234e-02 1.28363156e+00 -6.19338572e-01 -4.08901237e-02
-5.93349457e-01 -4.83678460e-01 4.95574951e-01 5.48620939e-01
6.38832271e-01 -1.51115203e+00 -5.42671561e-01 2.68430173e-01
-6.19287491e-01 -1.22383606e+00 -2.28918061e-01 4.32246715e-01
-1.00487959e+00 -1.05282354e+00 -5.13757110e-01 -1.10553586e+00
4.11800027e-01 2.92442173e-01 1.66417098e+00 -2.54471824e-02
4.40034643e-02 5.84046543e-01 -4.07229960e-01 -4.91547406e-01
-3.19060564e-01 6.82250559e-01 4.38983701e-02 -7.76195049e-01
9.20621336e-01 -3.87306213e-01 -1.68125644e-01 3.13851207e-01
-9.40565050e-01 -5.63401282e-02 7.16263235e-01 7.58589447e-01
1.70301259e-01 -3.98301221e-02 9.07799423e-01 -9.89350319e-01
1.06639671e+00 -6.45821691e-01 -5.05239666e-01 6.21660888e-01
-6.11676335e-01 2.96276540e-01 8.88493061e-01 -4.93038535e-01
-1.11465693e+00 5.31620793e-02 -1.59606710e-02 1.51880339e-01
-5.07759750e-01 5.70966005e-01 1.38285784e-02 1.14024736e-01
1.11045074e+00 4.37445283e-01 -4.85661507e-01 -4.77807790e-01
6.92104399e-01 2.96377718e-01 5.95323622e-01 -1.05113935e+00
9.90187645e-01 -1.90582067e-01 -5.28567612e-01 -5.62325358e-01
-1.51870024e+00 -3.83298904e-01 -6.19253099e-01 2.58996040e-01
9.03028607e-01 -5.34561098e-01 -7.58996308e-01 1.65724177e-02
-1.51939380e+00 -7.59096682e-01 -3.11633855e-01 2.21637219e-01
-5.65101504e-01 5.80754206e-02 -5.97625196e-01 -3.02897751e-01
-9.40636471e-02 -1.11583030e+00 9.83406186e-01 -1.61544904e-02
-7.12634385e-01 -1.21260679e+00 2.77540445e-01 8.48555505e-01
6.21074140e-01 -2.27087051e-01 1.27541506e+00 -1.23845458e+00
-6.08018696e-01 2.22185493e-01 -1.21958494e-01 5.26644826e-01
-4.43078158e-03 -5.55029392e-01 -1.11045218e+00 -1.14652924e-01
4.11943734e-01 -8.50804567e-01 8.27776015e-01 -2.98430771e-02
1.30474651e+00 -5.74781179e-01 -2.33905792e-01 3.79998416e-01
1.11459458e+00 5.65653220e-02 5.83905458e-01 5.01709282e-01
3.15400362e-01 7.39062607e-01 4.91828501e-01 2.80635990e-03
6.34430289e-01 3.51133019e-01 5.34182079e-02 4.63240415e-01
-4.87765260e-02 -2.80895054e-01 3.66718829e-01 6.10821068e-01
2.74798870e-01 -4.42588598e-01 -1.37646532e+00 6.72582805e-01
-1.42040837e+00 -7.13347077e-01 5.77401966e-02 1.72167897e+00
1.18300676e+00 2.40671083e-01 -4.05057639e-01 -1.99044019e-01
2.62489557e-01 -1.64448675e-02 -5.76789975e-01 -5.29407382e-01
-4.99850847e-02 6.53666794e-01 -1.02423064e-01 6.22866452e-01
-9.49758172e-01 1.41837263e+00 7.72119570e+00 5.97131133e-01
-6.91300988e-01 -1.67643562e-01 3.81664425e-01 1.58391073e-01
-4.29612786e-01 -4.82273512e-02 -1.18706870e+00 5.28714769e-02
1.00187957e+00 -4.25778776e-01 7.43410051e-01 7.95780003e-01
-1.09737836e-01 1.90999851e-01 -1.76302373e+00 7.31810987e-01
2.09773928e-01 -1.04262352e+00 4.31237757e-01 -3.80534470e-01
8.03510785e-01 1.37681440e-01 3.62342626e-01 9.57538605e-01
7.11909115e-01 -1.27556336e+00 2.67623246e-01 -2.25830618e-02
5.26456892e-01 -6.76545352e-02 5.55656552e-01 7.69158959e-01
-5.73587000e-01 -3.29804212e-01 -5.40156245e-01 -6.20494902e-01
-8.32216442e-02 1.61636889e-01 -1.36832500e+00 2.14598387e-01
7.31022477e-01 4.71511245e-01 -1.07960343e+00 7.54453719e-01
-8.16344023e-01 7.16660142e-01 -2.20726863e-01 -2.69925416e-01
4.59665477e-01 3.55915397e-01 3.09728444e-01 1.05556977e+00
3.19262534e-01 3.67471457e-01 6.03542849e-02 1.18675339e+00
-3.03774059e-01 -1.06040388e-02 -8.09124649e-01 -7.69072771e-02
6.32591724e-01 1.13751805e+00 -2.53098398e-01 -4.74154085e-01
-5.15750825e-01 8.43890250e-01 6.27106488e-01 6.98158741e-01
-3.69463146e-01 -3.34745377e-01 3.98364782e-01 -4.95568030e-02
-3.73444892e-02 -2.77585655e-01 -4.63907719e-01 -1.47780275e+00
1.45265564e-01 -1.28880692e+00 4.24132526e-01 -1.14517307e+00
-1.74570560e+00 5.09868324e-01 2.39770725e-01 -6.97467625e-01
-5.77185035e-01 -1.11027586e+00 -4.08169836e-01 7.17511296e-01
-1.55009437e+00 -7.87238121e-01 -3.16275656e-01 4.03453380e-01
9.05043781e-01 -2.44640633e-01 1.00020695e+00 -1.31479383e-01
-2.89512753e-01 5.22106886e-01 -2.64738411e-01 5.28641082e-02
1.09938669e+00 -1.45305574e+00 1.03042734e+00 5.24591625e-01
3.16217154e-01 8.33679378e-01 7.60205388e-01 -4.89999086e-01
-1.08687198e+00 -8.63268614e-01 1.58337212e+00 -1.43532372e+00
7.55395651e-01 -5.69832087e-01 -1.03209388e+00 1.40737009e+00
3.30449998e-01 -3.07650298e-01 6.22131407e-01 4.97043133e-01
-5.96920192e-01 2.17266828e-01 -1.15480173e+00 7.11018682e-01
1.25684834e+00 -3.81369650e-01 -1.59964657e+00 6.96299613e-01
1.18358409e+00 -4.61397618e-01 -8.01411450e-01 4.32074577e-01
2.68797010e-01 -6.22455001e-01 9.13002074e-01 -1.28223240e+00
6.81310713e-01 7.70572275e-02 1.23718393e-03 -1.72558999e+00
-4.64175940e-01 -7.19265997e-01 8.92672315e-02 8.26702714e-01
1.01471615e+00 -7.08868980e-01 6.55649304e-01 1.12191534e+00
-4.76502389e-01 -5.36918104e-01 -7.24266350e-01 -1.02576804e+00
6.78093851e-01 -5.20858109e-01 5.81123829e-01 1.20889378e+00
2.86956787e-01 9.69812155e-01 1.03269160e-01 -1.15092024e-01
1.61002070e-01 -3.96264791e-01 9.76335764e-01 -1.34582031e+00
-4.88595754e-01 -3.06364179e-01 3.14867407e-01 -1.53733242e+00
4.76841629e-01 -1.12572372e+00 1.06806559e-02 -1.68149948e+00
2.40404099e-01 -2.57890046e-01 -1.17057733e-01 8.92017722e-01
-2.36213297e-01 -1.30066589e-01 1.64694741e-01 2.73130406e-02
-7.89789796e-01 1.25210807e-01 1.43258035e+00 -2.63570786e-01
-1.82272226e-01 -1.26029029e-01 -1.04494762e+00 6.47311687e-01
9.78769839e-01 -3.42030823e-01 -7.73236334e-01 -1.31616855e+00
7.24296033e-01 -3.09321642e-01 1.58178031e-01 -1.00897515e+00
2.58183867e-01 -4.03766811e-01 3.81965429e-01 -3.05558324e-01
1.93249434e-01 -7.28451669e-01 -3.76552016e-01 1.77194744e-01
-9.21318889e-01 1.95419297e-01 5.53957701e-01 9.71472114e-02
-1.53483987e-01 -5.05993783e-01 3.40951085e-01 -5.08324802e-01
-9.68697071e-01 4.33383249e-02 -2.91816920e-01 6.84496939e-01
8.06539357e-01 1.58125386e-01 -9.79142427e-01 -4.50135738e-01
-8.48828137e-01 7.04782248e-01 1.37769371e-01 7.39529133e-01
6.78464949e-01 -1.00607288e+00 -9.33794498e-01 -7.58312792e-02
4.47418720e-01 7.35512227e-02 1.32276341e-01 1.82911664e-01
-2.80545712e-01 9.34349597e-01 -3.68613563e-02 -2.13992819e-01
-6.98912799e-01 8.07728171e-01 3.14759463e-01 -5.92385173e-01
-2.96057492e-01 9.74935710e-01 5.36278725e-01 -9.32356596e-01
4.93975170e-02 -7.79654026e-01 -3.14664692e-01 -1.90470755e-01
6.29812062e-01 6.75261915e-02 4.52608392e-02 1.32119253e-01
-8.81405100e-02 3.63940567e-01 -3.79339933e-01 -2.46456340e-01
1.14725363e+00 -4.78948131e-02 4.77045216e-02 4.62059945e-01
1.05099165e+00 -3.31109583e-01 -8.80497456e-01 -4.90219414e-01
4.30212408e-01 -1.50256753e-01 -4.55511242e-01 -1.54373324e+00
-3.98004383e-01 1.06015134e+00 -1.02877259e-01 2.03080118e-01
8.78910840e-01 2.74855703e-01 1.00057769e+00 1.45823145e+00
3.53606433e-01 -9.80047286e-01 6.11569524e-01 1.21747172e+00
9.95987177e-01 -1.36533654e+00 -4.09520149e-01 -4.23296124e-01
-5.31953990e-01 1.03570235e+00 1.01339436e+00 1.11958593e-01
1.42910317e-01 7.76414052e-02 2.29666643e-02 -1.97231233e-01
-1.31124759e+00 6.03954755e-02 5.09068131e-01 8.59675586e-01
4.88567770e-01 -1.50316894e-01 -3.80546488e-02 8.85610759e-01
-7.27121711e-01 6.32113218e-02 4.75468308e-01 6.56018496e-01
-7.22381413e-01 -1.11946654e+00 8.74683354e-03 4.55360323e-01
-2.75630325e-01 -5.64054430e-01 -7.89624751e-01 1.02677774e+00
-1.18605606e-01 1.01952970e+00 -1.29818514e-01 -1.94880858e-01
4.29397315e-01 6.99903190e-01 6.34552836e-01 -1.34139681e+00
-6.31434739e-01 -6.52675629e-01 1.95874602e-01 -3.86501461e-01
2.26418689e-01 -4.12509084e-01 -8.80231440e-01 -2.19821613e-02
1.29874632e-01 1.58987999e-01 3.48321617e-01 1.21338511e+00
2.09006503e-01 2.48355806e-01 2.31033117e-01 -3.37061465e-01
-1.08971703e+00 -1.17618799e+00 -7.11685568e-02 4.14951950e-01
2.45020568e-01 -3.61838132e-01 -5.99021375e-01 1.35092974e-01]
|
[11.039196968078613, 8.16624641418457]
|
36cf18b1-6b02-49f6-a984-092dd22098b2
|
english-to-chinese-transliteration-with-1
|
2112.10321
| null |
https://arxiv.org/abs/2112.10321v2
|
https://arxiv.org/pdf/2112.10321v2.pdf
|
English-to-Chinese Transliteration with Phonetic Back-transliteration
|
Transliteration is a task of translating named entities from a language to another, based on phonetic similarity. The task has embraced deep learning approaches in recent years, yet, most ignore the phonetic features of the involved languages. In this work, we incorporate phonetic information into neural networks in two ways: we synthesize extra data using forward and back-translation but in a phonetic manner; and we pre-train models on a phonetic task before learning transliteration. Our experiments include three language pairs and six directions, namely English to and from Chinese, Hebrew and Thai. Results indicate that our proposed approach brings benefits to the model and achieves better or similar performance when compared to state of the art.
|
['Songpeng Yan', 'Zhuofei Ding', 'Shi Cheng']
|
2021-12-20
| null | null | null | null |
['transliteration']
|
['natural-language-processing']
|
[ 4.08770293e-02 -6.06082603e-02 -3.38864595e-01 -6.51177526e-01
-8.81532490e-01 -6.36400223e-01 7.02738464e-01 -5.72831094e-01
-5.35243392e-01 9.60058928e-01 6.79180980e-01 -7.95001268e-01
5.00036359e-01 -7.98874021e-01 -8.71234298e-01 -2.96564817e-01
5.25453925e-01 6.56001985e-01 -4.83835936e-02 -2.25480065e-01
9.45010632e-02 4.96492028e-01 -8.28850746e-01 4.64971721e-01
1.21367633e+00 5.62310040e-01 7.88712278e-02 2.49028817e-01
-5.64756811e-01 6.02818072e-01 -4.51213568e-01 -7.29136944e-01
3.39841813e-01 -6.98195577e-01 -1.13144541e+00 -2.17704847e-01
1.63440838e-01 -5.83516583e-02 -1.62530288e-01 9.95674908e-01
5.89667976e-01 8.60457048e-02 6.45618677e-01 -6.75341189e-01
-1.47606540e+00 1.10849547e+00 -1.31097332e-01 5.27439639e-02
2.68934965e-01 -1.31319463e-02 8.81106436e-01 -1.10844898e+00
5.96956909e-01 1.33422160e+00 9.93418634e-01 5.48760116e-01
-8.94414902e-01 -7.64395833e-01 -5.20555042e-02 2.42528990e-01
-1.30648422e+00 -5.45648277e-01 5.80278158e-01 -3.26853752e-01
1.27121508e+00 -6.49463460e-02 3.00352424e-01 1.22566485e+00
2.41271198e-01 6.01741493e-01 1.30891097e+00 -4.50001150e-01
-2.23558564e-02 3.05450886e-01 -2.98757911e-01 3.77107322e-01
-2.39004478e-01 1.93169221e-01 -3.34268868e-01 2.59660572e-01
5.21801889e-01 -3.37400198e-01 4.14210334e-02 2.69937724e-01
-1.47746193e+00 7.70922959e-01 3.81358534e-01 6.31033421e-01
-4.69067097e-01 -1.26184404e-01 4.78988707e-01 3.58062297e-01
4.14571047e-01 5.43810070e-01 -7.38902211e-01 -2.13026121e-01
-8.59854817e-01 -1.29494518e-01 9.24389899e-01 1.09816134e+00
6.98429465e-01 4.39734250e-01 -7.87276551e-02 9.93916452e-01
2.44226351e-01 7.62556255e-01 1.00260603e+00 -6.37159884e-01
7.41120756e-01 1.22094937e-01 5.35933394e-03 -5.88338435e-01
-7.76279047e-02 -2.16756001e-01 -6.66081548e-01 -3.40799242e-01
1.88554570e-01 -4.43521798e-01 -1.08058178e+00 1.57651556e+00
1.08840488e-01 -1.63386002e-01 4.42253262e-01 6.59043670e-01
9.71609533e-01 1.00460219e+00 7.60378987e-02 2.17804741e-02
1.11507642e+00 -1.42090428e+00 -9.73217070e-01 -2.70889163e-01
5.97110450e-01 -1.16953826e+00 1.38221264e+00 1.36946261e-01
-1.25600302e+00 -1.03190517e+00 -7.17238545e-01 -2.94998705e-01
-7.58814514e-01 3.83163989e-01 2.98889756e-01 7.96994090e-01
-1.20199502e+00 6.05693460e-01 -8.10455680e-01 -5.60719252e-01
9.47593823e-02 2.97029257e-01 -3.09093058e-01 2.38039717e-01
-1.60908079e+00 1.34950495e+00 7.85980403e-01 1.73911780e-01
-5.90809464e-01 -4.31182444e-01 -8.94245505e-01 1.65729616e-02
-5.78433834e-02 -6.73098683e-01 1.49663842e+00 -1.16727161e+00
-2.27456856e+00 7.21711934e-01 -4.68969852e-01 -3.26160967e-01
5.10972261e-01 -2.61944294e-01 -7.11888909e-01 -3.16147953e-01
1.47997975e-01 7.03507841e-01 2.74297744e-01 -9.76447105e-01
-7.13707387e-01 -7.43724629e-02 -2.43405789e-01 3.75905037e-01
-1.39059737e-01 5.21545589e-01 -3.21533859e-01 -6.67733014e-01
-1.05645852e-02 -1.01794600e+00 -1.43106863e-01 -4.78400975e-01
-4.96029049e-01 -2.59710073e-01 6.10591531e-01 -1.19962037e+00
9.48271215e-01 -1.78318107e+00 4.57578227e-02 -9.94061455e-02
-5.26269078e-01 5.12158453e-01 -1.12156004e-01 6.10884309e-01
-1.75786167e-02 3.21431130e-01 -2.32170701e-01 -4.70409870e-01
1.44109055e-01 4.94754136e-01 -3.98989499e-01 1.93750739e-01
3.54375690e-01 1.17216730e+00 -7.13507056e-01 -4.01178867e-01
7.19100386e-02 5.62043846e-01 -3.73962790e-01 2.44190454e-01
4.79129218e-02 8.17393243e-01 -1.51711389e-01 4.46213901e-01
5.97212911e-01 2.58365482e-01 1.85299173e-01 -1.06287003e-01
-4.39380616e-01 1.20867872e+00 -7.47867525e-01 1.80474615e+00
-9.87815738e-01 4.13891703e-01 -4.79592294e-01 -1.11062026e+00
1.19691575e+00 6.29194915e-01 1.83764882e-02 -8.05936992e-01
1.91272870e-01 7.55311191e-01 1.68518379e-01 -5.45450330e-01
6.16849244e-01 -3.61272544e-01 -7.28129745e-02 4.45487171e-01
1.25493914e-01 -2.86047578e-01 3.92148718e-02 -6.19047225e-01
3.93789619e-01 5.68070829e-01 3.94728571e-01 -2.53313035e-01
5.26833355e-01 8.03638697e-02 7.62348115e-01 4.96437281e-01
-1.03786312e-01 4.83323365e-01 -1.20171316e-01 -3.76734465e-01
-1.17630517e+00 -1.18067658e+00 3.22914869e-02 1.06620288e+00
-2.08370984e-01 9.64293405e-02 -8.37511659e-01 -7.23778188e-01
-3.61982912e-01 9.11089182e-01 -2.83595085e-01 -1.25705481e-01
-1.22102690e+00 -3.57814223e-01 9.86794353e-01 7.28766441e-01
7.91509271e-01 -1.46041572e+00 3.56798768e-02 3.60007823e-01
-3.84851098e-01 -1.35799134e+00 -8.29256773e-01 1.27190962e-01
-8.62937748e-01 -1.88379258e-01 -9.40396249e-01 -1.41257250e+00
3.44009995e-01 -1.85783386e-01 1.06526566e+00 -5.80687165e-01
4.67789978e-01 -4.03037637e-01 -3.53648573e-01 -3.49009871e-01
-8.37879419e-01 5.85278034e-01 1.39260978e-01 -5.41221499e-02
5.32039583e-01 -5.69565177e-01 -1.62874982e-01 3.22992682e-01
-5.06267726e-01 1.44011639e-02 8.53119969e-01 6.43434405e-01
5.90229809e-01 -1.40316218e-01 6.63489997e-01 -1.06344819e+00
5.82080841e-01 -3.64443243e-01 -3.13688040e-01 2.85547435e-01
-4.98333186e-01 1.67828977e-01 1.14095128e+00 -4.33422089e-01
-1.38514102e+00 8.33050907e-02 -7.60434747e-01 -1.11705326e-01
-4.79735672e-01 5.82606971e-01 -5.67065775e-01 1.16616033e-01
4.81775254e-01 4.05211419e-01 -5.11762083e-01 -6.67280197e-01
6.43824100e-01 1.10364068e+00 7.32480049e-01 -5.85508823e-01
7.75437415e-01 1.64059848e-01 -5.59544981e-01 -5.31338692e-01
-5.59649289e-01 -1.24619901e-01 -1.13489354e+00 2.31289938e-01
1.03259456e+00 -8.03619385e-01 -3.28420490e-01 4.74353731e-01
-1.51863706e+00 -2.32442394e-01 -2.08644405e-01 8.25101197e-01
-5.16422868e-01 1.89941868e-01 -7.82647848e-01 -2.99663901e-01
-5.20796657e-01 -1.28620124e+00 8.86348307e-01 2.31309995e-01
-2.57337183e-01 -1.24210429e+00 2.84826100e-01 3.29420388e-01
7.55094826e-01 -9.44417194e-02 8.89639735e-01 -9.66530800e-01
-2.58940846e-01 1.95850030e-01 -9.59925279e-02 6.07276678e-01
6.01221621e-01 -1.16973389e-02 -9.52832282e-01 1.59485135e-02
-3.56826037e-02 -2.39675567e-02 4.62428302e-01 1.48400441e-01
7.57002413e-01 -5.31033814e-01 -1.35505617e-01 7.71980464e-01
1.11197305e+00 7.62258172e-01 6.25549138e-01 3.02590370e-01
8.79302204e-01 5.95147014e-01 2.71264017e-01 -2.25646734e-01
7.93580949e-01 5.62709808e-01 -1.47721037e-01 -4.36635435e-01
-3.57533902e-01 -4.85663831e-01 6.01435781e-01 1.66142213e+00
-3.30257602e-02 -1.41645893e-01 -1.23251188e+00 7.91476607e-01
-1.52063346e+00 -7.33626246e-01 1.38364173e-02 2.00644684e+00
1.23802960e+00 2.89609898e-02 -8.81984904e-02 -1.50184169e-01
8.10077369e-01 -4.72404808e-02 -3.90252024e-01 -1.07347572e+00
-2.84647465e-01 4.86633629e-01 3.66549522e-01 6.08730495e-01
-9.28007126e-01 1.54836869e+00 6.69095421e+00 5.11836648e-01
-1.66281557e+00 4.00763005e-01 4.72893029e-01 6.24262333e-01
-3.92213970e-01 7.79896453e-02 -7.77866185e-01 4.69727427e-01
1.32732511e+00 -1.58652663e-01 4.09964770e-01 6.96274340e-01
3.56929362e-01 4.64891613e-01 -1.31370497e+00 6.13957286e-01
2.13475436e-01 -1.08856273e+00 2.85320848e-01 -1.87644631e-01
1.08033323e+00 3.21230531e-01 8.70150253e-02 5.43584049e-01
6.30219162e-01 -1.07450116e+00 8.05430114e-01 4.73942399e-01
6.22763753e-01 -9.17088807e-01 8.68389845e-01 2.63429642e-01
-1.15104699e+00 3.75847965e-01 -3.51756394e-01 -5.31949811e-02
4.11751360e-01 2.22058222e-01 -1.13385344e+00 7.72157788e-01
5.59500217e-01 6.26354396e-01 -1.79302171e-01 9.07107055e-01
-4.74293083e-01 8.95726919e-01 -2.73082927e-02 -3.36637758e-02
3.95468116e-01 -3.88045609e-01 3.67499739e-02 1.41259849e+00
5.99777460e-01 -3.70952517e-01 2.52670825e-01 9.14935052e-01
-2.32055470e-01 4.22084033e-01 -6.04276955e-01 -1.25257164e-01
5.63036203e-01 7.31957555e-01 -4.20342833e-01 -4.78840262e-01
-5.70981562e-01 1.33566558e+00 2.16154575e-01 4.44316179e-01
-9.66525853e-01 -7.31079102e-01 4.38164145e-01 -2.25622252e-01
4.30723250e-01 -2.68144697e-01 -4.54288751e-01 -1.14198065e+00
1.02035023e-01 -1.01996231e+00 -2.06635788e-01 -7.27215528e-01
-1.34160483e+00 1.05725658e+00 -3.33992958e-01 -1.08570409e+00
-4.99487489e-01 -5.75412691e-01 -6.06791019e-01 1.35572660e+00
-1.62532878e+00 -1.47681761e+00 1.57304808e-01 3.70824128e-01
7.54452586e-01 -2.08678469e-01 8.77756536e-01 7.91412592e-01
-3.14545363e-01 7.81409562e-01 3.23971480e-01 5.33447981e-01
9.06763315e-01 -1.05949557e+00 1.19239962e+00 9.42052722e-01
3.78227085e-01 8.06368411e-01 3.94870430e-01 -5.05694866e-01
-9.53437805e-01 -1.26770079e+00 1.88090467e+00 -3.06673080e-01
6.13527715e-01 -2.86739826e-01 -9.27798092e-01 9.74412858e-01
7.14130342e-01 -2.83949614e-01 6.31947935e-01 -1.38377286e-02
-2.37916216e-01 4.96308208e-02 -9.74373102e-01 7.13542640e-01
1.03185499e+00 -9.78966713e-01 -1.02414978e+00 2.45281547e-01
9.74797845e-01 -5.23819208e-01 -8.82632256e-01 3.25966448e-01
5.97382009e-01 -5.98102629e-01 6.59112692e-01 -7.97284722e-01
4.29077357e-01 -3.54711622e-01 -3.04994762e-01 -1.72863042e+00
-3.07665050e-01 -6.33864045e-01 4.56611365e-01 1.56149769e+00
7.94285178e-01 -8.31039846e-01 4.55885559e-01 2.97137443e-02
-6.02456033e-01 -5.00959277e-01 -9.21725750e-01 -9.46896195e-01
7.42876530e-01 -2.24856660e-01 1.00493479e+00 1.33787799e+00
-2.64291912e-01 5.21231771e-01 -4.53087181e-01 1.74012274e-01
-1.56095102e-01 3.62023525e-02 5.01378477e-01 -8.59599590e-01
-2.64741212e-01 -5.01104653e-01 1.66757628e-02 -1.13887382e+00
5.02118766e-01 -1.19836962e+00 2.64162511e-01 -1.74048817e+00
-1.84131145e-01 -4.01039660e-01 -1.97239488e-01 7.29113996e-01
-2.76326150e-01 2.94199109e-01 7.40254596e-02 1.90076455e-01
4.21675704e-02 6.18670583e-01 1.21408010e+00 -1.19728893e-01
-2.61824429e-01 1.29807755e-01 -5.22689342e-01 6.28016770e-01
9.95677650e-01 -4.82468665e-01 -7.52914697e-02 -1.03986323e+00
-1.08529709e-01 -1.59525886e-01 -3.82317394e-01 -9.83210802e-01
3.91819589e-02 -8.21226165e-02 1.86681867e-01 -5.04531324e-01
1.74742013e-01 -6.94868922e-01 1.87004834e-01 4.03062612e-01
-4.52898920e-01 6.00734413e-01 1.46775812e-01 -1.13222040e-01
-5.12190700e-01 -1.07653968e-01 8.09582651e-01 -2.05931872e-01
-6.15739882e-01 3.64742503e-02 -5.50922751e-01 2.11084518e-03
5.27311504e-01 -1.01873361e-01 -3.20607089e-02 -1.54012755e-01
-6.50328338e-01 -1.93145305e-01 2.34056875e-01 7.79060781e-01
2.28667140e-01 -1.65738773e+00 -8.93291414e-01 2.80702621e-01
-1.18579566e-01 -3.47720802e-01 -2.18724161e-01 7.48496592e-01
-6.61433637e-01 8.07763994e-01 -4.76323903e-01 -4.08293337e-01
-8.47206712e-01 4.07620609e-01 5.06725907e-01 -2.12644815e-01
-2.68478334e-01 6.74825430e-01 8.13325942e-02 -1.38988054e+00
1.23934194e-01 -6.13660276e-01 -1.71360850e-01 -1.99666366e-01
4.15031612e-02 1.77284077e-01 2.28914738e-01 -1.11300874e+00
-2.33876020e-01 7.35009730e-01 1.26751177e-02 -3.11949939e-01
9.81775522e-01 -1.82735175e-01 -8.30546767e-02 6.40961111e-01
1.35916519e+00 2.54379451e-01 -6.38727009e-01 -4.92616683e-01
1.79028392e-01 -1.54212087e-01 -3.95525485e-01 -9.20712948e-01
-9.55781937e-01 1.15881228e+00 4.41154301e-01 -2.51894683e-01
8.95426869e-01 -1.82610109e-01 1.21382868e+00 6.85526013e-01
2.16776535e-01 -1.24771166e+00 -3.69267046e-01 1.15491021e+00
5.26526928e-01 -1.20429027e+00 -6.34056628e-01 -1.54804736e-01
-7.84807146e-01 9.61951673e-01 4.70144242e-01 -9.66182128e-02
4.53440487e-01 1.02821797e-01 7.07109213e-01 4.74455714e-01
-3.15771401e-01 -5.08484878e-02 3.44395757e-01 6.41207635e-01
8.54306281e-01 2.71563735e-02 -5.15794575e-01 4.94950891e-01
-7.58239508e-01 1.02098830e-01 3.14871609e-01 7.38866687e-01
-2.53096431e-01 -1.47317731e+00 -2.96672404e-01 -2.28936803e-02
-6.36583567e-01 -5.27836919e-01 -4.72806364e-01 7.95251667e-01
4.59738612e-01 8.43427241e-01 -7.88163617e-02 -3.17466319e-01
4.34292257e-01 3.53559732e-01 7.24237040e-02 -6.54631615e-01
-9.53152537e-01 -1.10890558e-02 3.03738058e-01 -2.59447932e-01
-3.59193146e-01 -5.06044626e-01 -1.32449830e+00 -2.01554626e-01
-2.31680647e-01 4.64900047e-01 8.89489591e-01 1.17897999e+00
3.54181737e-01 5.62216401e-01 7.38478899e-01 -6.91393435e-01
-5.55369437e-01 -1.15669167e+00 -1.78246096e-01 1.59824401e-01
6.02033101e-02 -1.84442401e-01 -3.16685475e-02 2.98869878e-01]
|
[11.510931968688965, 10.30536937713623]
|
c606e087-ea79-4a4c-9c6d-1e12e9776c4b
|
mvss-net-multi-view-multi-scale-supervised
|
2112.08935
| null |
https://arxiv.org/abs/2112.08935v3
|
https://arxiv.org/pdf/2112.08935v3.pdf
|
MVSS-Net: Multi-View Multi-Scale Supervised Networks for Image Manipulation Detection
|
As manipulating images by copy-move, splicing and/or inpainting may lead to misinterpretation of the visual content, detecting these sorts of manipulations is crucial for media forensics. Given the variety of possible attacks on the content, devising a generic method is nontrivial. Current deep learning based methods are promising when training and test data are well aligned, but perform poorly on independent tests. Moreover, due to the absence of authentic test images, their image-level detection specificity is in doubt. The key question is how to design and train a deep neural network capable of learning generalizable features sensitive to manipulations in novel data, whilst specific to prevent false alarms on the authentic. We propose multi-view feature learning to jointly exploit tampering boundary artifacts and the noise view of the input image. As both clues are meant to be semantic-agnostic, the learned features are thus generalizable. For effectively learning from authentic images, we train with multi-scale (pixel / edge / image) supervision. We term the new network MVSS-Net and its enhanced version MVSS-Net++. Experiments are conducted in both within-dataset and cross-dataset scenarios, showing that MVSS-Net++ performs the best, and exhibits better robustness against JPEG compression, Gaussian blur and screenshot based image re-capturing.
|
['Xirong Li', 'Juan Cao', 'Ruohan Hu', 'Xinru Chen', 'Chengbo Dong']
|
2021-12-16
| null | null | null | null |
['image-manipulation-detection']
|
['computer-vision']
|
[ 6.49316609e-01 -4.73868966e-01 6.59227669e-02 -6.71333149e-02
-9.89812136e-01 -8.55923891e-01 5.39898396e-01 -5.19362204e-02
-2.72721440e-01 4.88977820e-01 -1.28642190e-02 -2.20132068e-01
-1.95133965e-02 -5.42043805e-01 -1.18101025e+00 -7.03979075e-01
-1.27980337e-01 -5.12857251e-02 2.65239686e-01 -1.48134986e-02
7.02165067e-01 6.90653861e-01 -1.49421370e+00 7.75091410e-01
4.03976321e-01 9.30137753e-01 1.15949549e-01 9.00979996e-01
2.30925888e-01 9.16346014e-01 -9.34824407e-01 -8.10657799e-01
5.71784437e-01 -2.72365212e-01 -7.75144756e-01 5.35293281e-01
7.08356977e-01 -6.22568369e-01 -5.41468143e-01 1.23314750e+00
5.85189223e-01 -1.26415610e-01 3.16143304e-01 -1.31884229e+00
-9.24868524e-01 2.67822325e-01 -8.04865062e-01 6.40143037e-01
3.98564965e-01 6.23157263e-01 5.76475441e-01 -9.00214791e-01
7.47144759e-01 1.18702435e+00 8.13529432e-01 4.14137334e-01
-1.33656442e+00 -5.94150722e-01 -3.16556394e-01 5.91059804e-01
-1.09323335e+00 -6.54938698e-01 9.08693433e-01 -2.96814412e-01
5.76118350e-01 2.51530260e-01 1.94669798e-01 1.65010047e+00
2.15379089e-01 1.00386012e+00 1.51757693e+00 -3.63620669e-01
2.64668670e-02 2.89998800e-01 -3.04930687e-01 5.11206090e-01
3.92292112e-01 2.61717021e-01 -7.52014577e-01 -1.06201157e-01
6.10481501e-01 4.48386278e-03 -6.30258501e-01 -4.60479259e-01
-9.36741650e-01 7.14826345e-01 2.21369058e-01 2.11553350e-01
-1.66365743e-01 1.19984835e-01 7.06468523e-01 6.53083324e-01
1.90008923e-01 7.24450111e-01 -5.13482511e-01 -5.43924198e-02
-1.13732100e+00 1.27426684e-01 4.70325232e-01 7.80622184e-01
6.68176711e-01 2.23534197e-01 5.46620451e-02 5.58746099e-01
-2.94151843e-01 3.55587959e-01 3.89512628e-01 -8.49348128e-01
5.26278555e-01 1.97098717e-01 -8.32039788e-02 -1.34793401e+00
-5.37885763e-02 -3.56452703e-01 -6.50429726e-01 4.99156356e-01
4.58997846e-01 1.74753413e-01 -9.10032511e-01 1.41941547e+00
1.70816183e-01 3.68036032e-01 1.38043026e-02 6.69948697e-01
4.26942199e-01 3.26779217e-01 -1.92319259e-01 -3.31783406e-02
1.15087402e+00 -6.31209254e-01 -5.24625838e-01 -4.47587430e-01
6.33584499e-01 -8.75326574e-01 1.17947030e+00 7.97354400e-01
-1.14762557e+00 -7.19902873e-01 -1.17081392e+00 -1.08653203e-01
-6.15394652e-01 -1.76267892e-01 1.57842353e-01 8.81064355e-01
-8.27241719e-01 8.83936644e-01 -4.37482655e-01 -5.60833476e-02
8.30458641e-01 2.54140794e-01 -7.34445333e-01 -3.96429420e-01
-1.10456204e+00 8.80528927e-01 3.67767066e-01 -2.55987607e-02
-9.75079954e-01 -6.25786483e-01 -5.90837002e-01 -8.20248760e-03
4.48696136e-01 -1.98712260e-01 7.61584759e-01 -1.17702830e+00
-9.75772679e-01 1.09037256e+00 2.23806798e-01 -5.77402234e-01
9.00554240e-01 -3.08179319e-01 -7.65208125e-01 5.04475296e-01
1.64815605e-01 3.86963725e-01 1.63299692e+00 -1.61500716e+00
-4.46914583e-01 -3.74830633e-01 -7.87428916e-02 -2.84476727e-01
-5.11432290e-01 1.65680408e-01 -3.64391059e-01 -8.43040824e-01
-2.42515534e-01 -6.95617497e-01 1.97536692e-01 2.63235837e-01
-5.14091194e-01 4.79426950e-01 1.49633932e+00 -1.01765203e+00
9.19244826e-01 -2.48099041e+00 -3.02155614e-01 7.59507865e-02
1.63248941e-01 7.28057623e-01 -3.37064534e-01 5.16360939e-01
-3.90359879e-01 1.88776925e-01 -3.03793907e-01 -1.73653066e-01
-3.47374827e-01 8.23019892e-02 -5.59759855e-01 1.03912079e+00
5.65447628e-01 8.57918918e-01 -8.62561762e-01 -3.41168970e-01
3.76368403e-01 3.91593248e-01 -4.46277231e-01 2.19354793e-01
4.16189665e-03 4.68075931e-01 -7.89328218e-02 8.42674196e-01
9.37638938e-01 -1.45174906e-01 -2.11818844e-01 -4.18622315e-01
3.78975570e-01 -1.06653251e-01 -1.25183940e+00 1.44580626e+00
-5.16999781e-01 1.11555362e+00 1.32523283e-01 -1.15271783e+00
7.29515195e-01 6.31724894e-02 1.52172476e-01 -8.88559341e-01
6.89574108e-02 1.87845141e-01 -1.07730724e-01 -1.05005431e+00
4.84993190e-01 5.87612428e-02 1.64150953e-01 4.86905813e-01
2.09563717e-01 1.12308525e-01 -6.19510747e-02 2.46691510e-01
1.35657597e+00 -6.80705532e-02 1.11095428e-01 6.91685453e-02
4.38268423e-01 -4.61422890e-01 3.79680425e-01 9.33693707e-01
-2.98848391e-01 9.35074747e-01 6.21556520e-01 -5.11651576e-01
-1.21961224e+00 -9.33201909e-01 -1.85941502e-01 9.28426683e-01
2.63354570e-01 -1.72789812e-01 -6.58692420e-01 -9.73595858e-01
2.86703315e-02 6.39625609e-01 -7.39564776e-01 -2.95170337e-01
-6.20732844e-01 -4.77616251e-01 8.51378977e-01 4.89008695e-01
7.00308621e-01 -9.36739206e-01 -6.66452467e-01 6.93199113e-02
-5.44132106e-02 -1.50095844e+00 -3.46054524e-01 2.17666060e-01
-5.34410834e-01 -1.47677898e+00 -4.22436535e-01 -5.28838992e-01
4.65258330e-01 5.86360455e-01 9.18866813e-01 4.31512654e-01
-4.46803451e-01 5.77948153e-01 -4.90250200e-01 -1.47518575e-01
-7.22963929e-01 -4.73470211e-01 -1.72994137e-01 4.06107932e-01
1.57361731e-01 -6.67977810e-01 -6.50957406e-01 1.46244749e-01
-1.56512260e+00 -3.73522907e-01 7.26856828e-01 1.08391654e+00
1.13604076e-01 2.68050551e-01 2.22073346e-01 -1.04368889e+00
4.61640269e-01 -4.31563377e-01 -1.78820550e-01 2.76347607e-01
-4.33353961e-01 -1.49968565e-01 7.28516757e-01 -6.50389194e-01
-7.62700975e-01 -1.31920010e-01 -1.13676218e-02 -8.80211353e-01
-3.51977110e-01 1.14375077e-01 -3.71804953e-01 -4.02139992e-01
8.17991376e-01 3.91254038e-01 -2.96096653e-02 -3.38477403e-01
2.62323320e-01 5.64990044e-01 9.20426786e-01 -3.37712735e-01
1.10084295e+00 7.40642369e-01 -1.26549834e-02 -9.11296070e-01
-5.58109283e-01 -3.72479260e-01 -6.54767573e-01 -1.97905704e-01
6.55269563e-01 -6.98615134e-01 -3.39891553e-01 7.30269551e-01
-1.21377766e+00 -6.86867759e-02 -1.49942249e-01 -5.59519120e-02
-6.41108990e-01 9.55479741e-01 -4.64943916e-01 -5.81779480e-01
-5.97229861e-02 -1.38048792e+00 1.16979098e+00 -4.59970422e-02
-6.34424090e-02 -1.02533901e+00 -4.09427345e-01 6.15197182e-01
2.93533832e-01 4.36556727e-01 8.37035179e-01 -1.00122774e+00
-5.52991688e-01 -4.92244452e-01 -3.58929783e-01 6.96364045e-01
1.18235022e-01 -3.04118264e-02 -1.36295795e+00 -5.56026757e-01
3.90346855e-01 -6.05319202e-01 8.70990336e-01 -1.43916428e-01
1.62040460e+00 -4.34880227e-01 1.39142126e-02 8.96459281e-01
1.56256247e+00 -8.24963376e-02 1.04063261e+00 7.92009890e-01
7.34931231e-01 5.65176547e-01 2.04771340e-01 3.69506031e-01
-3.09664309e-01 5.31705916e-01 7.70537734e-01 -1.46636799e-01
-2.72535950e-01 -1.69134453e-01 6.22169137e-01 3.74247432e-02
4.19725031e-01 -4.61255252e-01 -7.36485839e-01 4.94789153e-01
-1.29835117e+00 -1.31771636e+00 -1.72262579e-01 2.13624740e+00
5.58652043e-01 2.88563609e-01 -1.98395308e-02 6.03826880e-01
9.20882583e-01 3.62141669e-01 -6.04948163e-01 -3.33330572e-01
-5.39516628e-01 3.60784024e-01 7.06417084e-01 8.98535829e-04
-1.35240018e+00 6.67966962e-01 5.77247477e+00 1.03836095e+00
-1.30093062e+00 7.55176768e-02 8.28279316e-01 5.97847141e-02
3.81502137e-02 -6.72553107e-02 -3.76548052e-01 7.39608705e-01
8.30355406e-01 4.30439204e-01 3.16106766e-01 6.65947616e-01
8.85534510e-02 -2.14763060e-02 -1.00511563e+00 1.18612766e+00
4.89675194e-01 -1.49528170e+00 1.72862332e-04 1.31735168e-02
6.19593740e-01 -2.21093431e-01 5.07745981e-01 -7.62398914e-02
-4.58935685e-02 -9.09175992e-01 8.23947608e-01 1.58126041e-01
8.75185966e-01 -6.14523411e-01 6.36760294e-01 5.62045611e-02
-6.75783396e-01 -3.20823461e-01 -1.20337136e-01 4.43515927e-01
8.81943852e-02 4.29347277e-01 -6.73955441e-01 4.32381928e-01
8.16377223e-01 7.70484984e-01 -9.53589678e-01 1.05015504e+00
-1.42924711e-01 5.89066148e-01 7.86818489e-02 6.61312640e-01
1.83910757e-01 2.08740070e-01 6.23754621e-01 1.42395377e+00
2.55872577e-01 -3.90455157e-01 -7.90451244e-02 7.37993062e-01
-2.46118963e-01 -3.00490886e-01 -7.49038339e-01 -6.81475103e-02
3.13527405e-01 1.15942276e+00 -8.93155634e-01 -1.45698041e-01
-5.29229879e-01 1.45798802e+00 9.04049128e-02 3.51371676e-01
-7.59764433e-01 -3.15581769e-01 5.39105356e-01 1.83037683e-01
7.28169084e-01 -1.65063702e-02 -2.65025496e-01 -1.19522083e+00
2.13425457e-01 -1.42040670e+00 3.32365304e-01 -8.44722688e-01
-1.43752027e+00 3.24840099e-01 -1.97718844e-01 -1.40002871e+00
1.27365440e-01 -8.40506315e-01 -7.65213549e-01 3.25255722e-01
-1.61539757e+00 -1.21622574e+00 -1.46287307e-01 7.32107401e-01
6.80075943e-01 -2.98580915e-01 4.51820284e-01 3.45396399e-01
-3.37674439e-01 8.50690842e-01 2.97077537e-01 4.74883586e-01
1.00226879e+00 -1.07074046e+00 4.64785099e-01 1.21451247e+00
2.89594173e-01 5.18198073e-01 8.40158224e-01 -5.17803788e-01
-1.67380977e+00 -1.14026189e+00 2.58338362e-01 -4.46254075e-01
8.31747055e-01 -3.58501405e-01 -1.33899927e+00 4.71858412e-01
3.05897474e-01 4.48143244e-01 5.64697504e-01 -4.45889384e-01
-1.01127040e+00 1.12090155e-01 -1.29332352e+00 3.11270595e-01
7.88621783e-01 -9.03539896e-01 -4.11975265e-01 3.90704185e-01
3.24238986e-01 -2.04579532e-01 -6.68518364e-01 2.24761397e-01
3.83226305e-01 -1.42977560e+00 1.21673942e+00 -6.25347674e-01
6.12074435e-01 -1.07727356e-01 -2.71047235e-01 -1.08311296e+00
-6.27795160e-02 -7.80068099e-01 -2.03549862e-01 1.29501510e+00
-4.71341284e-03 -3.04731578e-01 7.11371779e-01 2.75857925e-01
-6.22141315e-03 -3.79989326e-01 -8.62792730e-01 -9.79983091e-01
-5.14817163e-02 -5.19376874e-01 4.16284680e-01 1.25539649e+00
-5.71564317e-01 -9.00144503e-02 -8.09262216e-01 3.92563760e-01
6.61518335e-01 -1.26747191e-01 1.00233686e+00 -7.69370556e-01
-6.01464570e-01 -3.49373370e-01 -7.60016024e-01 -6.45978153e-01
1.16610453e-01 -5.25697172e-01 -3.23565751e-01 -7.90060818e-01
1.67177513e-01 -1.07948035e-01 -1.26451924e-01 2.48258367e-01
-3.19838017e-01 5.21811903e-01 3.14418405e-01 1.81854337e-01
-5.44075847e-01 1.85789838e-01 9.72749293e-01 -3.43277305e-01
2.75960088e-01 -8.07237178e-02 -6.15913689e-01 6.96927905e-01
6.54734910e-01 -5.75323164e-01 -2.14898750e-01 -4.98786777e-01
2.12799415e-01 -1.31868050e-01 8.02106798e-01 -1.13808799e+00
8.67483467e-02 -4.87994552e-02 7.24907994e-01 -4.20100600e-01
3.27363759e-01 -8.71855080e-01 -1.36822477e-01 3.84650141e-01
-3.99013788e-01 1.10025033e-01 2.43014708e-01 9.31120038e-01
-1.62982687e-01 -4.82557207e-01 9.38353479e-01 -2.74541527e-01
-8.99323881e-01 1.96524769e-01 -3.46780658e-01 1.48659140e-01
9.15591061e-01 -7.42727041e-01 -3.35393339e-01 -4.51352119e-01
-3.96236420e-01 -4.02037799e-01 6.63364768e-01 3.99289727e-01
8.77469182e-01 -1.00982213e+00 -5.11252463e-01 3.60385627e-01
1.87681258e-01 -4.10564601e-01 3.74359429e-01 5.65825939e-01
-5.67122817e-01 -2.06732601e-01 -3.59069526e-01 -5.20930529e-01
-1.35314739e+00 1.04525614e+00 1.80382133e-01 -1.18836299e-01
-5.78062594e-01 8.38902354e-01 7.26795793e-02 3.89123037e-02
1.09319918e-01 1.98829174e-01 2.13239029e-01 -2.63467692e-02
7.32850134e-01 4.67885435e-01 2.53778249e-01 -7.13560998e-01
1.03105279e-02 3.05830598e-01 -2.96055973e-01 2.36005425e-01
1.53370965e+00 -2.16684848e-01 7.18471706e-02 1.39479730e-02
1.73575366e+00 8.27163905e-02 -1.36942077e+00 -3.71126741e-01
1.95121422e-01 -1.05207324e+00 1.27199203e-01 -5.66037714e-01
-1.39383137e+00 1.12880707e+00 8.24977517e-01 3.32435906e-01
1.31328642e+00 -3.37349832e-01 8.68985534e-01 1.93128884e-01
9.03215781e-02 -1.06381953e+00 5.89591980e-01 1.12130567e-01
7.68494785e-01 -1.53479385e+00 2.53791034e-01 -2.43180841e-02
-6.31304443e-01 1.41452014e+00 4.88989383e-01 -2.08708987e-01
3.44157726e-01 3.90212178e-01 -3.27048190e-02 -1.51341900e-01
-4.05909657e-01 1.00697823e-01 3.90267484e-02 9.14100766e-01
-7.45628923e-02 -5.32117069e-01 4.43651766e-01 7.35958070e-02
2.89045982e-02 -3.34574431e-01 8.40535462e-01 1.11325276e+00
-2.37021506e-01 -1.06344461e+00 -8.79137695e-01 4.47220623e-01
-8.10427248e-01 3.34663354e-02 -4.62585151e-01 8.79909158e-01
2.89468765e-01 8.38645279e-01 -7.85827786e-02 -4.74560350e-01
8.31403956e-02 -8.43312964e-02 4.76874948e-01 -2.34745577e-01
-6.53076410e-01 -4.94294502e-02 -2.02369943e-01 -7.84726501e-01
-3.72119367e-01 -9.13507342e-01 -6.75132990e-01 -3.48885059e-01
-4.77072895e-01 -3.60389411e-01 6.35066271e-01 1.02673841e+00
3.58742118e-01 1.52065888e-01 8.81668687e-01 -9.62351978e-01
-7.04890072e-01 -5.80377162e-01 -5.31289399e-01 1.08909142e+00
7.60346830e-01 -6.06227934e-01 -6.06587648e-01 3.79423797e-01]
|
[12.327141761779785, 0.9770194888114929]
|
328d51a3-4607-4618-a6fd-7c78870135ca
|
improved-algorithm-on-online-clustering-of
|
1902.09162
| null |
https://arxiv.org/abs/1902.09162v2
|
https://arxiv.org/pdf/1902.09162v2.pdf
|
Improved Algorithm on Online Clustering of Bandits
|
We generalize the setting of online clustering of bandits by allowing non-uniform distribution over user frequencies. A more efficient algorithm is proposed with simple set structures to represent clusters. We prove a regret bound for the new algorithm which is free of the minimal frequency over users. The experiments on both synthetic and real datasets consistently show the advantage of the new algorithm over existing methods.
|
['Kwong-Sak Leung', 'Wei Chen', 'Shuai Li']
|
2019-02-25
| null | null | null | null |
['online-clustering']
|
['computer-vision']
|
[-2.89501995e-01 1.25414893e-01 -9.36775744e-01 -2.95648843e-01
-8.27529073e-01 -8.61827374e-01 4.29174155e-02 2.25560516e-02
-3.13774884e-01 1.03302801e+00 1.18585765e-01 -5.27587771e-01
-6.65623188e-01 -8.13403785e-01 -9.38421249e-01 -7.90823579e-01
-6.63011193e-01 8.32942069e-01 -1.02302739e-02 2.34544173e-01
6.37140647e-02 1.69392392e-01 -1.14584780e+00 2.75082827e-01
9.14981782e-01 1.01973426e+00 1.32069468e-01 4.33898926e-01
-8.73574391e-02 5.13448179e-01 -5.64967930e-01 -2.09811345e-01
8.42213452e-01 -6.36985004e-01 -9.75711465e-01 5.19258559e-01
-2.15166420e-01 -2.71372736e-01 -3.00462693e-01 1.14120376e+00
2.50023812e-01 5.11900842e-01 4.83138531e-01 -1.21194232e+00
-7.78707147e-01 1.48204064e+00 -1.07473683e+00 4.03605640e-01
3.91434878e-01 -7.17316568e-01 1.39102793e+00 1.08227618e-01
2.74216920e-01 1.28645015e+00 3.76895159e-01 3.41485888e-01
-1.51359022e+00 -8.03872883e-01 4.59149778e-01 2.65898854e-01
-1.60975933e+00 -5.02627306e-02 2.59194940e-01 -1.54819950e-01
4.56232339e-01 8.09168279e-01 7.39505708e-01 4.50125843e-01
-9.57854688e-01 1.01697469e+00 1.09768140e+00 -7.19379663e-01
6.42956853e-01 9.14811417e-02 5.37166536e-01 2.75880635e-01
7.36731648e-01 -8.29201541e-04 -1.64141461e-01 -8.55341196e-01
7.11747706e-01 1.10647075e-01 -3.82179976e-01 -5.63163161e-01
-7.43498206e-01 1.17480743e+00 2.50317782e-01 1.15264335e-03
-2.75564641e-01 3.77811074e-01 1.26863956e-01 2.74505675e-01
7.28573203e-01 5.43857329e-02 -3.58074456e-01 1.04784735e-01
-9.73290622e-01 1.03991181e-01 1.07034731e+00 1.55117393e+00
6.12664580e-01 -4.34361786e-01 -4.16382790e-01 9.22698140e-01
-1.90166514e-02 4.37936604e-01 3.47505391e-01 -1.17062557e+00
2.90150344e-01 2.10061952e-01 7.89620042e-01 -6.61020279e-01
-1.70531139e-01 -6.01284206e-01 -6.73870623e-01 -6.79179311e-01
3.50143611e-01 -2.68828899e-01 -5.54966509e-01 1.75522304e+00
5.86531043e-01 3.49109948e-01 -3.57660383e-01 9.07838345e-01
-8.32233578e-02 7.89883137e-01 -4.38807845e-01 -7.96302378e-01
9.87793386e-01 -8.99143875e-01 -8.45367014e-01 2.09322020e-01
4.98478740e-01 -4.33067113e-01 5.93149960e-01 4.23201382e-01
-1.06498969e+00 2.67365485e-01 -6.93600833e-01 4.73740190e-01
-1.19834885e-01 -2.68439293e-01 1.32500672e+00 1.38890529e+00
-9.24109280e-01 4.29151058e-01 -5.42786241e-01 -3.66802990e-01
5.64923406e-01 5.13039351e-01 3.17563117e-01 1.02456585e-01
-7.97167540e-01 1.00716375e-01 9.35472608e-01 -3.16649646e-01
-1.67843044e-01 -7.77851462e-01 -3.11501503e-01 3.38284075e-01
6.21412098e-01 -5.62331557e-01 1.48487079e+00 -1.18158901e+00
-1.42866373e+00 3.85763049e-01 -3.45764495e-02 -9.14733291e-01
5.49977243e-01 2.45835446e-02 -1.64624482e-01 1.26407132e-01
2.80708894e-02 -9.99781564e-02 2.64660954e-01 -1.28295350e+00
-1.02103436e+00 -3.31113152e-02 3.19088131e-01 7.27822036e-02
-6.17004573e-01 -5.81263751e-02 -6.05467618e-01 -5.16217351e-01
-9.27903429e-02 -1.04140890e+00 -4.16774809e-01 -6.28834903e-01
-5.77062786e-01 -4.18182641e-01 1.74785867e-01 1.10652167e-02
1.76910317e+00 -1.86169660e+00 -2.33580679e-01 9.96758044e-01
-1.56338036e-01 -3.46001655e-01 2.81254739e-01 5.42759597e-01
-3.58136110e-02 3.86772066e-01 1.05288617e-01 6.10880144e-02
4.24802721e-01 4.73530084e-01 -2.88456261e-01 8.81498218e-01
-1.08005643e+00 1.65623397e-01 -8.44940484e-01 -2.75068521e-01
-2.06275145e-03 -5.14338434e-01 -1.15849447e+00 -1.06194757e-01
-3.93161088e-01 -2.24358901e-01 -6.06365561e-01 3.36361676e-01
9.77134109e-01 -7.47244298e-01 7.11932600e-01 2.61222422e-01
8.67333338e-02 5.54207750e-02 -1.67172182e+00 1.26195490e+00
-1.95929050e-01 -1.20067568e-02 4.79573578e-01 -1.40577888e+00
4.29721884e-02 3.06856900e-01 7.06760347e-01 -1.70942694e-01
2.17877507e-01 -7.67981187e-02 -1.65005729e-01 -1.59770057e-01
1.96866602e-01 -4.82044294e-02 -1.57168254e-01 9.89890933e-01
-3.06635350e-01 7.45700300e-01 3.23853761e-01 4.82891351e-01
7.63123572e-01 -6.58240080e-01 3.45807076e-01 -7.17180371e-01
1.01123236e-01 -8.51928815e-02 5.01665056e-01 1.39503121e+00
-8.59522521e-02 -8.83610267e-03 2.86165565e-01 -3.70789379e-01
-8.72090876e-01 -8.20650578e-01 -1.85573608e-01 1.60152090e+00
3.80678087e-01 -3.91770005e-01 -8.18795860e-01 -4.88179475e-01
5.63012779e-01 5.17614603e-01 -8.71656418e-01 3.89236182e-01
5.04813455e-02 -1.10443771e+00 2.09957331e-01 1.56081721e-01
2.63858020e-01 -2.10161075e-01 -2.06212968e-01 2.66254365e-01
-3.71175617e-01 -1.14723444e+00 -9.07802761e-01 -1.86783308e-03
-7.67944813e-01 -1.06718743e+00 -3.92728984e-01 -6.64757371e-01
6.71221495e-01 7.39928365e-01 8.03965807e-01 3.76279578e-02
-4.52825613e-02 5.42778075e-01 -4.19663310e-01 -3.64758611e-01
6.16785064e-02 1.34623468e-01 1.82812408e-01 2.59963244e-01
3.21470886e-01 -4.47505802e-01 -9.53706980e-01 3.40056270e-01
-7.98566997e-01 -1.03591882e-01 8.93391892e-02 5.51622748e-01
4.90401000e-01 2.33548179e-01 6.29305840e-01 -1.40435565e+00
6.36349201e-01 -9.63226616e-01 -8.92914236e-01 3.04823756e-01
-5.25341570e-01 -6.27508312e-02 5.65320790e-01 -4.15286332e-01
-8.76796961e-01 2.18464479e-01 4.57777619e-01 -2.59924740e-01
2.67369747e-01 2.70300537e-01 2.24278286e-01 4.14052978e-02
5.11713028e-01 -7.77684376e-02 -3.46004754e-01 -5.20474076e-01
7.02362418e-01 9.23995793e-01 2.80768067e-01 -1.04606307e+00
5.27612925e-01 8.42448294e-01 -3.83003682e-01 -5.83855331e-01
-1.25655377e+00 -1.00797284e+00 4.05424647e-02 6.14790991e-02
1.49972156e-01 -8.49061131e-01 -1.33379698e+00 -4.09675866e-01
-6.66381121e-01 -4.01818156e-01 -1.22982532e-01 5.76243162e-01
-9.71168518e-01 6.64837778e-01 -5.91007292e-01 -1.44472790e+00
-2.41333842e-01 -3.90308291e-01 4.00616169e-01 4.79818098e-02
1.03514835e-01 -8.04384410e-01 7.86646642e-03 3.47255677e-01
-4.62183505e-02 1.29106864e-01 5.24209321e-01 -7.25253463e-01
-5.82892835e-01 -2.06252579e-02 -1.98324338e-01 -1.17484234e-01
3.17232013e-01 -9.72204730e-02 -5.41744053e-01 -7.27001846e-01
-4.50399876e-01 -1.31424293e-01 7.57824719e-01 7.94344306e-01
1.85891140e+00 -1.04431498e+00 -8.24511528e-01 5.67562878e-01
1.77806437e+00 2.55893201e-01 1.38309434e-01 4.08910483e-01
1.24380253e-01 1.72060087e-01 6.23451352e-01 1.11132991e+00
1.67990699e-01 4.18037564e-01 2.56002456e-01 8.63115788e-02
7.07383633e-01 -2.88500600e-02 -4.83409986e-02 3.07240218e-01
-9.23089758e-02 -3.80845636e-01 -5.94165266e-01 1.03254426e+00
-2.50505114e+00 -1.30428183e+00 4.35673296e-02 2.43985105e+00
1.02378297e+00 -2.75294688e-02 9.24398065e-01 1.83005050e-01
1.30037069e+00 -3.39068681e-01 -4.68764037e-01 -4.54252362e-01
-6.93814233e-02 3.48331444e-02 1.33543515e+00 5.58076441e-01
-1.14393091e+00 7.12485373e-01 8.19517708e+00 1.22685504e+00
-2.39905953e-01 2.46365502e-01 6.40080392e-01 -7.05409110e-01
-1.46714911e-01 -9.32138115e-02 -5.36732674e-01 7.28413582e-01
1.09505498e+00 -8.59550595e-01 1.06880641e+00 1.00717521e+00
4.08556879e-01 -6.85797632e-02 -1.01253200e+00 1.16810787e+00
-4.47469801e-01 -1.64166451e+00 -1.45460263e-01 2.74943650e-01
1.23586774e+00 5.03886305e-02 -1.63938820e-01 -4.33880500e-02
1.25747454e+00 -5.93654692e-01 6.71724141e-01 3.18750888e-02
6.57362282e-01 -1.42914975e+00 2.99389482e-01 5.88705659e-01
-9.33570564e-01 -5.19225240e-01 -6.45986915e-01 -2.33860880e-01
-1.10209428e-01 6.37056351e-01 -7.27396548e-01 5.52806199e-01
9.60767031e-01 1.83663666e-01 2.51061052e-01 1.47570431e+00
5.26671231e-01 9.52545464e-01 -8.93864751e-01 -2.37491816e-01
3.85182679e-01 -5.01637101e-01 1.55994192e-01 1.58215845e+00
2.37281337e-01 6.29651308e-01 6.32106841e-01 5.11437893e-01
-3.02990586e-01 3.76878768e-01 -3.42103988e-01 8.37700292e-02
1.13644397e+00 1.01676309e+00 -8.97920251e-01 -5.63545763e-01
-3.10310632e-01 7.46933460e-01 4.27864164e-01 3.75538796e-01
-9.20767188e-01 -1.72720745e-01 6.80715621e-01 2.05505695e-02
7.36184895e-01 8.72078165e-02 -1.86882079e-01 -7.14890957e-01
-3.15746278e-01 -4.05234545e-01 1.20860350e+00 -1.58101782e-01
-1.55566168e+00 -2.48509631e-01 1.83646172e-01 -9.07184541e-01
9.15505588e-02 -1.42850518e-01 -4.01706249e-01 3.27144682e-01
-1.20164227e+00 -3.63579661e-01 3.60382855e-01 1.03619111e+00
1.64500773e-01 9.17618945e-02 5.45597434e-01 4.56469923e-01
-5.62604964e-01 8.73476386e-01 1.27450836e+00 -9.42148864e-02
4.85423177e-01 -1.56629360e+00 -4.00925905e-01 5.85837543e-01
-1.77598819e-01 5.50948262e-01 8.68918240e-01 -2.51352757e-01
-1.20948124e+00 -1.15835798e+00 1.35828227e-01 1.82853088e-01
7.51466095e-01 -4.47295219e-01 -2.07886398e-01 9.22761858e-01
3.32683653e-01 -1.19824484e-01 1.33219039e+00 5.50972998e-01
-1.52636349e-01 -2.61330366e-01 -1.33169770e+00 3.44642997e-01
1.21860194e+00 1.19589753e-01 -1.36850491e-01 9.93352890e-01
8.09215248e-01 -2.03318045e-01 -8.01057518e-01 -1.21252820e-01
5.49558580e-01 -7.49310017e-01 7.80855536e-01 -7.18077242e-01
-3.23250502e-01 -4.39939946e-02 -2.07951531e-01 -1.21956789e+00
-8.89723122e-01 -1.32170951e+00 -4.25882608e-01 9.89275336e-01
4.35209304e-01 -6.98731959e-01 1.04742479e+00 4.24575537e-01
3.46332759e-01 -3.83673787e-01 -1.00653660e+00 -1.15089214e+00
-9.15681496e-02 -1.93912327e-01 7.97530890e-01 1.24616325e+00
6.06445968e-01 -1.53839141e-02 -6.08042121e-01 3.65162820e-01
9.99049067e-01 7.88278401e-01 5.13631821e-01 -1.22186279e+00
-6.75839841e-01 -5.38548887e-01 1.60647273e-01 -1.24097610e+00
-1.53886238e-02 -9.09580052e-01 -2.78255671e-01 -1.48391998e+00
6.51436746e-01 -6.10189438e-01 -4.65127826e-01 2.74476856e-01
1.69039860e-01 2.74464637e-02 -9.69878659e-02 1.12998493e-01
-1.35376930e+00 2.35148117e-01 8.25589359e-01 1.33842170e-01
-3.02532703e-01 3.84083003e-01 -1.05518723e+00 4.85849768e-01
1.00536931e+00 -5.53052187e-01 -4.40522969e-01 -9.10000280e-02
3.42388362e-01 1.87835827e-01 -3.30648959e-01 -4.65750009e-01
3.09834689e-01 -6.00885212e-01 1.89373881e-01 -7.99444735e-01
-1.95253462e-01 -1.10468006e+00 2.91835159e-01 4.39872295e-01
-3.84153873e-01 -3.66720587e-01 6.88944533e-02 1.38557398e+00
3.63523573e-01 -2.16913279e-02 7.82117307e-01 -1.24426119e-01
-2.19609984e-03 3.57870549e-01 -3.90095681e-01 5.52139170e-02
1.32083869e+00 -2.49134064e-01 -1.36812404e-01 -7.66470432e-01
-8.81873786e-01 7.80190408e-01 2.14064032e-01 -1.77393463e-02
-2.26218566e-01 -1.51285982e+00 -5.07576942e-01 -3.45531851e-01
-1.09279469e-01 -3.67450058e-01 4.80701439e-02 4.88577783e-01
-4.49623406e-01 7.17814326e-01 2.22157344e-01 -3.73225600e-01
-1.21584368e+00 1.05933249e+00 2.35218659e-01 -4.60772850e-02
-2.27441207e-01 7.02807188e-01 -1.33909667e-02 -1.51410729e-01
5.42843223e-01 -7.20287040e-02 3.12122315e-01 1.57648306e-02
4.00768906e-01 7.12804675e-01 -1.47729054e-01 -6.10879436e-02
-1.83611602e-01 -3.40351127e-02 -2.21523568e-01 -1.27856433e-01
1.45183694e+00 -5.68343759e-01 -2.12507024e-01 3.33899319e-01
9.20647204e-01 1.44626394e-01 -8.03133130e-01 -6.09296083e-01
7.58085921e-02 -7.79796481e-01 -4.53572609e-02 -4.02527869e-01
-1.08094513e+00 -2.02967152e-01 4.42170501e-01 1.13343573e+00
1.01101017e+00 1.99550658e-01 5.02988279e-01 5.60305774e-01
7.81300902e-01 -1.56321609e+00 -6.08533144e-01 1.99136194e-02
3.00880998e-01 -9.28821385e-01 2.52230078e-01 -5.69947004e-01
-2.32006475e-01 6.96664929e-01 1.50717437e-01 -3.62333894e-01
9.64012742e-01 4.52609211e-02 -5.65256715e-01 6.91834390e-02
-8.01375031e-01 -4.70775247e-01 -1.92351818e-01 2.86358833e-01
3.25580686e-01 6.28689706e-01 -9.93005395e-01 1.00431454e+00
-3.36308748e-01 -5.35233952e-02 7.01763391e-01 9.10096049e-01
-7.12181509e-01 -9.56536472e-01 -6.93908453e-01 6.98221922e-01
-1.01119745e+00 1.81973770e-01 -3.89435321e-01 6.01041436e-01
-3.17647755e-01 1.32575309e+00 2.98414707e-01 1.37913674e-02
-1.20645650e-01 -2.06651226e-01 5.28161943e-01 -5.97682416e-01
-3.62610281e-01 5.52859783e-01 1.15172975e-01 -4.13292199e-01
-7.59480000e-01 -3.65158528e-01 -9.98656631e-01 -7.93746650e-01
-8.28289151e-01 8.47524643e-01 1.68672711e-01 5.19552112e-01
3.71717960e-01 1.65885478e-01 1.32396376e+00 -3.86958659e-01
-6.94811881e-01 -7.21069634e-01 -1.17530417e+00 4.74090844e-01
3.24358761e-01 -6.07860088e-01 -4.61509645e-01 -1.60461038e-01]
|
[4.565731048583984, 3.3419275283813477]
|
acabeb00-6e88-43db-aff5-3c8baf7fcfde
|
ernie-sat-speech-and-text-joint-pretraining-1
|
2211.03545
| null |
https://arxiv.org/abs/2211.03545v2
|
https://arxiv.org/pdf/2211.03545v2.pdf
|
ERNIE-SAT: Speech and Text Joint Pretraining for Cross-Lingual Multi-Speaker Text-to-Speech
|
Speech representation learning has improved both speech understanding and speech synthesis tasks for single language. However, its ability in cross-lingual scenarios has not been explored. In this paper, we extend the pretraining method for cross-lingual multi-speaker speech synthesis tasks, including cross-lingual multi-speaker voice cloning and cross-lingual multi-speaker speech editing. We propose a speech-text joint pretraining framework, where we randomly mask the spectrogram and the phonemes given a speech example and its transcription. By learning to reconstruct the masked parts of the input in different languages, our model shows great improvements over speaker-embedding-based multi-speaker TTS methods. Moreover, our framework is end-to-end for both the training and the inference without any finetuning effort. In cross-lingual multi-speaker voice cloning and cross-lingual multi-speaker speech editing tasks, our experiments show that our model outperforms speaker-embedding-based multi-speaker TTS methods.
|
['Hua Wu', 'Yu Sun', 'Liang Huang', 'Zeyu Chen', 'Junkun Chen', 'Shuohuan Wang', 'Pengfei Zhu', 'Renjie Zheng', 'He Bai', 'Tian Yuan', 'Chao Pang', 'Xiaoran Fan']
|
2022-11-07
|
ernie-sat-speech-and-text-joint-pretraining
|
https://arxiv.org/abs/2211.03545
|
https://arxiv.org/pdf/2211.03545.pdf
| null |
['text-to-speech-synthesis', 'voice-cloning']
|
['speech', 'speech']
|
[ 3.02243203e-01 2.50674844e-01 -2.23600909e-01 -6.40853643e-01
-1.50735867e+00 -5.36226869e-01 4.88353133e-01 -4.92428422e-01
-1.71512023e-01 4.40216959e-01 5.38456500e-01 -7.61763394e-01
5.31461000e-01 -1.41446814e-01 -9.81721997e-01 -4.24248606e-01
4.20398802e-01 5.57650566e-01 -2.53848374e-01 -4.76796064e-04
-5.11860192e-01 2.55708188e-01 -1.34364212e+00 5.63387513e-01
8.18685472e-01 3.75005484e-01 6.46105289e-01 1.06398594e+00
-2.97676563e-01 4.20202851e-01 -7.22196221e-01 -4.61329520e-01
-1.15003653e-01 -6.23352826e-01 -5.90489864e-01 1.97871640e-01
5.94130039e-01 -2.03820214e-01 -2.11133286e-01 8.20513785e-01
8.76016021e-01 1.57793790e-01 4.37811762e-01 -9.92559433e-01
-7.36449301e-01 1.22351229e+00 -2.75255173e-01 2.54224595e-02
1.24006264e-01 -1.21141292e-01 1.02931118e+00 -1.27915001e+00
1.51896745e-01 2.00051403e+00 4.14635390e-01 8.03149879e-01
-1.32945549e+00 -8.90812159e-01 2.76767403e-01 -7.10758790e-02
-1.39867532e+00 -1.36506987e+00 7.50177622e-01 -1.56070501e-01
1.25514233e+00 3.25918496e-01 -6.99813366e-02 1.43603122e+00
-1.25161454e-01 1.04714465e+00 9.11108136e-01 -7.28917599e-01
-1.42340615e-01 4.39666510e-01 -5.65767944e-01 6.08065903e-01
-4.83800888e-01 2.33835593e-01 -7.01756239e-01 9.25783738e-02
5.37332535e-01 -5.18843114e-01 -1.67604610e-01 2.72218943e-01
-1.36335552e+00 8.23226392e-01 -3.51820290e-01 4.01428133e-01
-1.16727293e-01 1.60483211e-01 6.93255484e-01 4.98060554e-01
6.57147050e-01 -6.82399422e-02 -6.83809936e-01 -1.04816467e-01
-1.25481105e+00 -1.69620275e-01 5.72644591e-01 1.12265468e+00
5.04062533e-01 9.12961304e-01 -1.64492860e-01 1.50512910e+00
3.09633762e-01 8.56978416e-01 8.90315711e-01 -6.20052397e-01
7.16127574e-01 -3.36050779e-01 -4.52765852e-01 -1.53885052e-01
6.00567162e-02 -3.72486055e-01 -8.78382266e-01 -1.51866779e-01
-8.50379243e-02 -3.19119185e-01 -9.35741246e-01 1.87632418e+00
3.01531285e-01 4.65057522e-01 5.31528115e-01 3.87046099e-01
7.15996444e-01 1.17787075e+00 -1.26045689e-01 -3.88143301e-01
1.53285110e+00 -1.67865896e+00 -1.10795462e+00 -5.02297103e-01
5.05465090e-01 -1.22991979e+00 1.42426884e+00 1.56438157e-01
-1.27502441e+00 -1.09244740e+00 -8.66612196e-01 -2.83374161e-01
-3.49415004e-01 6.30307198e-01 -3.06310336e-04 8.68569791e-01
-9.43245828e-01 5.26581183e-02 -7.18047559e-01 1.71502288e-02
-4.43375595e-02 7.15496615e-02 -2.39646122e-01 -6.66288361e-02
-1.24187446e+00 9.41586971e-01 3.71501833e-01 -1.93055555e-01
-1.17142344e+00 -8.42609406e-01 -1.12877715e+00 1.55451044e-01
2.22184867e-01 -4.44030076e-01 1.55350482e+00 -6.62781060e-01
-2.17159462e+00 7.36652970e-01 -9.42064643e-01 -5.16175210e-01
2.81762570e-01 -1.34026214e-01 -9.89692748e-01 -3.32645357e-01
1.16100676e-01 7.18645215e-01 1.29064488e+00 -1.19508755e+00
-8.30743551e-01 -8.22061524e-02 -5.29315591e-01 2.78944790e-01
-2.17572898e-02 2.96780825e-01 -4.48258877e-01 -1.01224899e+00
-1.05582476e-01 -9.52255070e-01 1.15737021e-01 -5.76924264e-01
-5.85187435e-01 -2.51602054e-01 9.36866820e-01 -9.32974517e-01
1.04840481e+00 -2.18340850e+00 4.07684177e-01 -3.53991687e-01
-5.81977785e-01 3.13931912e-01 -3.85148942e-01 5.19531310e-01
-1.84351206e-01 1.00861877e-01 -2.97570109e-01 -1.34926319e+00
3.00259143e-01 5.17336965e-01 -8.09926808e-01 1.31223202e-01
3.18316370e-01 8.26151907e-01 -4.43807364e-01 -3.62487167e-01
4.31307405e-01 8.03100348e-01 -4.63449895e-01 4.47094202e-01
-1.20909281e-01 6.34849072e-01 2.29830921e-01 5.27953148e-01
4.04275596e-01 5.09240508e-01 1.70175105e-01 -1.94892108e-01
-1.51785985e-01 1.02088964e+00 -1.01857996e+00 2.07221222e+00
-1.37535870e+00 5.68182230e-01 3.62710446e-01 -7.53250957e-01
7.48867214e-01 1.00893033e+00 -4.97358292e-02 -4.61248398e-01
-1.16269201e-01 3.70269179e-01 8.24700370e-02 -2.50943929e-01
5.09966731e-01 -5.78241706e-01 -1.06472261e-01 6.15481675e-01
4.03021663e-01 -5.36034346e-01 -8.49137902e-02 -1.64771661e-01
3.06057006e-01 3.74085382e-02 2.27190629e-01 6.32481053e-02
6.05064034e-01 -6.60168529e-01 2.90044487e-01 4.83030826e-01
1.24817006e-01 6.09791338e-01 7.77972396e-03 2.41594180e-01
-9.86573160e-01 -1.16469800e+00 -1.53567046e-01 1.82867539e+00
-5.15621901e-01 -4.27217454e-01 -9.01924312e-01 -4.52845216e-01
2.35086828e-02 1.32242167e+00 -1.40394121e-01 -9.50378105e-02
-7.60156512e-01 -2.60338515e-01 1.13256907e+00 3.13584447e-01
-1.53686434e-01 -9.73340690e-01 5.20078540e-01 6.07376039e-01
-4.18998241e-01 -1.54892755e+00 -1.19734097e+00 2.56464005e-01
-6.18147433e-01 -1.71343401e-01 -8.65700245e-01 -1.27181029e+00
2.09098771e-01 1.06373802e-01 7.38016069e-01 -5.06278634e-01
1.75339371e-01 5.87541834e-02 -4.79543433e-02 -3.33605796e-01
-1.28011680e+00 3.60205591e-01 5.15096664e-01 2.60369062e-01
-1.42057583e-01 -5.78566670e-01 2.45639414e-01 4.41273659e-01
-7.92091489e-01 8.77130851e-02 5.66520751e-01 8.32345605e-01
6.41888559e-01 -8.58814344e-02 1.07358372e+00 -6.86177909e-01
7.34352052e-01 -1.95790797e-01 -3.98223191e-01 4.57481861e-01
-2.24626496e-01 3.08724374e-01 8.95505548e-01 -6.69012785e-01
-1.24640954e+00 -2.14654561e-02 -6.38273239e-01 -5.46124637e-01
-1.94578275e-01 2.90606111e-01 -5.58859766e-01 4.68741804e-01
2.42395207e-01 6.51344955e-01 -1.19878344e-01 -9.17366087e-01
8.64456117e-01 1.22051775e+00 6.75355494e-01 -6.48509502e-01
6.90508962e-01 -8.87389034e-02 -7.97081172e-01 -1.27387011e+00
-6.25441790e-01 -3.76762778e-01 -6.46992028e-01 2.88459808e-01
8.90414596e-01 -1.38651097e+00 -4.79764551e-01 4.38174993e-01
-1.66405916e+00 -2.34780967e-01 -1.86792389e-01 7.94904232e-01
-6.37232840e-01 3.11461061e-01 -4.72745538e-01 -8.57318580e-01
-2.84221828e-01 -1.68062162e+00 1.42022359e+00 -4.16201591e-01
-2.03828707e-01 -1.02910817e+00 1.38450056e-01 5.55013061e-01
5.24189472e-01 -7.53439784e-01 1.10759366e+00 -6.31272018e-01
-3.69503468e-01 1.32566765e-01 2.07484141e-01 8.62203956e-01
6.42174184e-01 -1.36596441e-01 -1.48375893e+00 -5.52399039e-01
-1.69294104e-01 -3.03536743e-01 6.30818248e-01 4.10703838e-01
1.02820981e+00 -5.05193830e-01 -1.09024554e-01 6.83689833e-01
7.49932885e-01 2.07111910e-01 2.79410869e-01 -4.32731330e-01
9.52628613e-01 6.67900622e-01 3.67413163e-02 1.08861625e-01
5.47260284e-01 8.06188941e-01 -1.09015927e-01 -2.37319157e-01
-6.43419683e-01 -4.02800471e-01 1.00192845e+00 1.76673496e+00
6.42357647e-01 -5.58216929e-01 -4.21339840e-01 7.53107965e-01
-1.32014954e+00 -7.40934014e-01 3.17314744e-01 2.09358144e+00
1.15682483e+00 -6.30862340e-02 6.89254403e-02 -5.46187386e-02
1.09004903e+00 5.23555815e-01 -5.84676325e-01 -8.53059232e-01
-2.00460687e-01 4.15094465e-01 2.67122716e-01 1.16792083e+00
-9.21921730e-01 1.56597888e+00 6.04180861e+00 1.20638525e+00
-1.21666181e+00 5.17935216e-01 2.35926583e-01 -1.33416802e-01
-6.02292657e-01 -2.24758387e-01 -1.12084365e+00 2.97943890e-01
1.59802806e+00 -1.66453242e-01 8.44101727e-01 6.31475568e-01
2.76297808e-01 5.49755991e-01 -1.42187405e+00 1.04407728e+00
2.83678532e-01 -1.10368633e+00 3.55011672e-01 -1.76054776e-01
6.34961188e-01 2.87161291e-01 8.98584649e-02 7.03212559e-01
2.99627960e-01 -1.11986721e+00 1.03726435e+00 -1.72111243e-01
1.15493667e+00 -8.46601725e-01 9.75176990e-02 4.56175268e-01
-1.43497050e+00 3.19850445e-01 -1.16743542e-01 6.00699723e-01
7.18451262e-01 2.83187389e-01 -1.34409070e+00 5.31363964e-01
2.15429872e-01 3.97764921e-01 -1.65397435e-01 2.53044605e-01
-2.15936780e-01 1.04595006e+00 -2.43216306e-01 3.32184732e-01
1.90107718e-01 2.02800222e-02 6.55251503e-01 1.51785028e+00
5.08201540e-01 -6.01746380e-01 1.52740046e-01 7.75498629e-01
-2.96040565e-01 1.96659088e-01 -4.78489131e-01 -4.56438631e-01
7.27726817e-01 7.77206659e-01 -1.16253629e-01 -6.08866632e-01
-4.28857923e-01 1.41226304e+00 3.06314975e-01 4.61490512e-01
-8.34918380e-01 -3.56996745e-01 1.17448866e+00 -1.63863987e-01
6.10510290e-01 -3.26403081e-01 -1.43165499e-01 -1.15354073e+00
-2.83163618e-02 -1.15159273e+00 -1.59020528e-01 -6.42265558e-01
-1.10715544e+00 1.01108146e+00 -2.25184828e-01 -8.67691994e-01
-8.20648789e-01 -4.25479293e-01 -5.55369616e-01 1.28667223e+00
-1.77030694e+00 -1.50159943e+00 7.29775786e-01 4.33463186e-01
1.35777557e+00 -6.27226412e-01 1.14293396e+00 5.69179535e-01
-6.71263576e-01 9.08616662e-01 4.12996143e-01 6.84233233e-02
9.53276336e-01 -1.04767084e+00 1.14927971e+00 9.06681478e-01
4.83531117e-01 5.36896646e-01 4.68086809e-01 -3.24043840e-01
-1.23203850e+00 -1.31908453e+00 1.29489732e+00 -1.73335165e-01
5.85754037e-01 -8.51032794e-01 -1.14958882e+00 1.09946465e+00
6.27650201e-01 -1.78104967e-01 8.75935793e-01 1.92971691e-01
-3.39905351e-01 -1.09870657e-01 -7.19934642e-01 7.21169174e-01
7.40957320e-01 -1.23391914e+00 -6.71198606e-01 3.46168667e-01
1.44779348e+00 -4.44617361e-01 -6.83344424e-01 3.01619042e-02
4.66616392e-01 -3.85417879e-01 1.08869433e+00 -7.08452523e-01
-1.59055348e-02 -2.14842618e-01 -6.22518539e-01 -1.85005784e+00
-7.40379691e-02 -8.07638109e-01 2.26716474e-01 1.65287650e+00
7.94236958e-01 -7.21000612e-01 1.06392600e-01 -1.86629161e-01
-8.94072115e-01 -2.26574689e-01 -1.34578776e+00 -9.97940004e-01
3.52448106e-01 -6.78790629e-01 8.76727343e-01 8.31852078e-01
-3.92482966e-01 6.23594940e-01 -7.99547613e-01 6.50867641e-01
5.15303493e-01 1.15384562e-02 8.23401868e-01 -6.03282154e-01
-6.95094764e-01 -3.60417604e-01 3.04464608e-01 -1.33917141e+00
8.08685303e-01 -1.23647630e+00 4.04930443e-01 -1.26667023e+00
-5.56130946e-01 -1.16570026e-01 -2.10142434e-02 4.35429662e-01
-1.02126367e-01 -3.17596436e-01 1.43526644e-01 -2.50120014e-01
-2.53354404e-02 1.00366902e+00 1.11876297e+00 -2.61096209e-01
-2.17806846e-01 1.58505186e-01 -3.37125778e-01 4.41758633e-01
6.69655561e-01 -8.58229220e-01 -5.22178292e-01 -7.09804893e-01
-7.29020357e-01 3.45169902e-01 -9.44436863e-02 -7.85696387e-01
2.38932222e-02 -8.51975661e-03 -1.34224251e-01 -7.38773525e-01
8.15901160e-01 -6.73669398e-01 -3.68228741e-02 1.91999182e-01
-6.54365361e-01 -3.71845327e-02 5.12454629e-01 3.57906401e-01
-4.31248635e-01 -1.71320200e-01 9.39586461e-01 -2.98012071e-03
-2.05137029e-01 1.26129895e-01 -8.06508183e-01 2.61185855e-01
5.65531015e-01 1.13235675e-02 3.60087901e-02 -3.24458271e-01
-8.93550694e-01 -4.28167060e-02 -1.56631857e-01 9.33062315e-01
4.31973308e-01 -1.42220962e+00 -1.06636286e+00 7.63748705e-01
8.18599239e-02 -9.26808715e-02 3.27493668e-01 5.17076910e-01
2.35255197e-01 6.53132677e-01 3.32432896e-01 -5.86281359e-01
-1.41973817e+00 4.00889874e-01 4.48873371e-01 9.88405570e-02
-1.19472221e-01 1.07530379e+00 5.23764133e-01 -1.21362352e+00
5.30050278e-01 -7.14093804e-01 5.19603193e-01 -8.16432536e-02
4.35193360e-01 1.91146582e-01 2.51550049e-01 -1.08263516e+00
-3.61766428e-01 3.96642119e-01 -2.91577071e-01 -6.28777742e-01
1.01612091e+00 -5.19084752e-01 2.12334529e-01 9.35266376e-01
1.53962326e+00 5.48577249e-01 -1.02523851e+00 -5.54726720e-01
-3.71354550e-01 -6.34551570e-02 2.41298288e-01 -7.65707374e-01
-7.93922007e-01 1.26432776e+00 2.36192361e-01 -9.01922137e-02
5.67324162e-01 5.55824228e-02 1.24963474e+00 4.03032660e-01
2.89009772e-02 -1.14989090e+00 -1.48429796e-01 7.01454222e-01
1.18705785e+00 -1.31049585e+00 -5.94305754e-01 -2.83687234e-01
-1.03985715e+00 8.99708509e-01 3.48364472e-01 4.06394541e-01
7.08171368e-01 5.07417023e-01 3.66085976e-01 3.76306325e-01
-1.00278163e+00 -1.34293959e-01 4.39038843e-01 4.73034412e-01
6.35246396e-01 4.35272902e-01 5.79411864e-01 5.41445255e-01
-7.76848376e-01 -5.29758930e-01 2.09576994e-01 4.04519737e-01
-4.04237151e-01 -1.40653777e+00 -5.42810023e-01 -2.49598905e-01
-3.91026914e-01 -6.27330661e-01 -2.42903605e-01 4.22713816e-01
-3.18031609e-02 1.11215079e+00 1.37366846e-01 -2.31399611e-01
4.34719592e-01 7.57163584e-01 3.87172520e-01 -9.18594301e-01
-4.97159094e-01 5.14347494e-01 2.66810685e-01 -1.39153987e-01
-1.31606981e-01 -5.83636224e-01 -1.34243846e+00 -1.87663004e-01
-4.98218983e-01 1.96369663e-01 1.17836320e+00 9.97241437e-01
4.22783911e-01 9.33961928e-01 9.23069537e-01 -9.68202710e-01
-6.84226930e-01 -1.26582468e+00 -5.90904295e-01 -1.57327116e-01
8.29540014e-01 -2.63044775e-01 -5.16019642e-01 4.01032090e-01]
|
[14.722244262695312, 6.840377330780029]
|
cc2ca970-12db-445d-9f31-51fc10894290
|
on-granularity-of-prosodic-representations-in
|
2301.11446
| null |
https://arxiv.org/abs/2301.11446v1
|
https://arxiv.org/pdf/2301.11446v1.pdf
|
On granularity of prosodic representations in expressive text-to-speech
|
In expressive speech synthesis it is widely adopted to use latent prosody representations to deal with variability of the data during training. Same text may correspond to various acoustic realizations, which is known as a one-to-many mapping problem in text-to-speech. Utterance, word, or phoneme-level representations are extracted from target signal in an auto-encoding setup, to complement phonetic input and simplify that mapping. This paper compares prosodic embeddings at different levels of granularity and examines their prediction from text. We show that utterance-level embeddings have insufficient capacity and phoneme-level tend to introduce instabilities when predicted from text. Word-level representations impose balance between capacity and predictability. As a result, we close the gap in naturalness by 90% between synthetic speech and recordings on LibriTTS dataset, without sacrificing intelligibility.
|
['Viacheslav Klimkov', 'Daniel Korzekwa', 'Rafal Sienkiewicz', 'Raahil Shah', 'Kamil Pokora', 'Mikolaj Babianski']
|
2023-01-26
| null | null | null | null |
['expressive-speech-synthesis']
|
['speech']
|
[ 1.77925587e-01 5.34414411e-01 -1.45722955e-01 -4.97600615e-01
-7.52568245e-01 -4.61277068e-01 4.74397212e-01 -1.97950244e-01
-1.42678618e-02 6.95782959e-01 9.09370601e-01 -1.06979506e-02
1.97179750e-01 -6.06911004e-01 -4.76795256e-01 -5.53104043e-01
2.46225655e-01 2.58418024e-01 -3.27683636e-03 -2.56879002e-01
-3.01382840e-01 1.96392149e-01 -1.59760797e+00 4.71331835e-01
4.86402333e-01 7.64976323e-01 4.22420144e-01 7.40787446e-01
-4.49352682e-01 4.94883955e-01 -8.49815369e-01 -2.13451177e-01
1.15393385e-01 -5.20919919e-01 -5.28438747e-01 1.39657676e-01
-3.40034440e-02 -8.63099098e-02 -1.26733139e-01 1.06673181e+00
5.22544563e-01 1.36674151e-01 8.01376879e-01 -8.25445890e-01
-6.75773501e-01 1.19948125e+00 -7.32822064e-03 6.35626784e-04
3.26356769e-01 -9.67148393e-02 1.15412974e+00 -1.00733209e+00
4.43628937e-01 1.54436386e+00 5.12489021e-01 7.01876283e-01
-1.59226704e+00 -3.91468197e-01 -6.54052868e-02 -2.09505260e-01
-1.14125001e+00 -9.00364280e-01 9.60712254e-01 -5.00757873e-01
1.19976437e+00 4.14900929e-01 2.94475347e-01 1.66754127e+00
8.83070379e-02 5.02082586e-01 8.93271506e-01 -6.85992718e-01
7.95593634e-02 4.09797907e-01 1.55646160e-01 2.39025354e-02
-3.87264341e-01 3.04343224e-01 -6.55091405e-01 2.53412835e-02
6.97305799e-01 -6.04028642e-01 -4.74370450e-01 3.10392492e-02
-8.67580593e-01 8.22738767e-01 -9.76384357e-02 4.17089015e-01
-4.08001661e-01 4.78860438e-02 6.73391938e-01 4.40220416e-01
1.88140869e-01 4.75074053e-01 -5.47452569e-01 -4.91636097e-01
-8.88037026e-01 9.41503718e-02 7.82726586e-01 9.51603413e-01
3.82630944e-01 9.13189948e-01 -8.34087580e-02 1.46638894e+00
1.62412718e-01 2.73702204e-01 1.15577292e+00 -7.44600177e-01
3.61521184e-01 -3.86534743e-02 -6.36062175e-02 -6.92256570e-01
-1.31448165e-01 -2.69687682e-01 -7.81444371e-01 4.81472202e-02
9.22924802e-02 -2.21402213e-01 -8.49175572e-01 1.95074105e+00
-1.59225345e-01 -2.66315881e-02 4.84425008e-01 7.04521477e-01
6.12191260e-01 1.36400747e+00 -6.76362365e-02 -6.41752481e-01
1.28360915e+00 -1.11469758e+00 -1.27551103e+00 -2.34518737e-01
1.89227536e-01 -9.07065332e-01 1.63342416e+00 3.62379849e-01
-1.33597803e+00 -1.04045236e+00 -1.15116632e+00 -5.40167838e-02
-4.97818857e-01 7.65950531e-02 -1.62557483e-01 8.98676753e-01
-8.40978265e-01 5.70188165e-01 -4.76131886e-01 9.48306397e-02
-4.81580287e-01 2.16773123e-01 -3.29831928e-01 7.36660659e-01
-1.48594642e+00 8.56052399e-01 6.08613431e-01 -1.99145675e-01
-4.29523528e-01 -8.85412574e-01 -1.00787783e+00 3.55674982e-01
1.12201483e-03 -1.66655943e-01 1.34731245e+00 -8.54808271e-01
-2.23151231e+00 5.74775755e-01 -1.62110254e-01 -5.95102310e-01
1.53015286e-01 -2.39709258e-01 -8.07626903e-01 -2.51500368e-01
-2.17663720e-01 7.68925071e-01 1.00573349e+00 -1.20843494e+00
-2.54414856e-01 2.60744065e-01 -5.93872368e-01 2.74146825e-01
-2.03474075e-01 1.08170211e-01 -1.33494198e-01 -9.19597268e-01
2.66086832e-02 -7.40712821e-01 7.10831955e-02 -5.65165222e-01
-3.78493696e-01 -2.11267993e-01 8.63875329e-01 -6.96115255e-01
1.31086528e+00 -2.38798022e+00 2.96620876e-01 -2.40551382e-01
-3.24120075e-01 1.94060460e-01 -3.68835665e-02 5.50744057e-01
-2.16563091e-01 3.18876982e-01 -1.06437705e-01 -6.42121017e-01
2.86370993e-01 5.91601789e-01 -7.92663753e-01 1.22752562e-02
4.42401171e-01 7.37978756e-01 -4.85792249e-01 -3.30166310e-01
4.02755141e-01 5.97733378e-01 -5.25899768e-01 3.47383648e-01
-2.24330261e-01 4.52887028e-01 1.69981346e-01 1.74521208e-01
2.46929765e-01 5.13098240e-01 1.79504901e-01 -9.62521136e-02
-1.91846341e-01 1.04569423e+00 -1.02979684e+00 1.48821628e+00
-8.12834084e-01 6.24237955e-01 1.50183335e-01 -8.98138404e-01
1.34693635e+00 8.75074685e-01 7.88900182e-02 -5.13381600e-01
7.18643069e-02 2.18089968e-01 2.42169499e-01 -1.88432187e-01
7.50311136e-01 -5.99361539e-01 -4.18175429e-01 -5.26922494e-02
3.77476454e-01 -5.57064295e-01 -2.21688971e-01 -5.77890217e-01
5.42512536e-01 3.59894931e-02 5.52108884e-01 -4.81980175e-01
3.52933526e-01 -5.32602012e-01 6.42824054e-01 3.40687752e-01
3.50259314e-03 6.90587163e-01 6.34696662e-01 -6.80070072e-02
-1.21500373e+00 -1.32979262e+00 -4.71371889e-01 1.16781211e+00
-4.24861580e-01 -4.91035014e-01 -7.55057991e-01 -4.47302014e-02
-2.67607361e-01 1.14726281e+00 -4.37212169e-01 -2.01808587e-01
-8.35636139e-01 -2.02577159e-01 6.95649564e-01 4.57553416e-01
-1.49280712e-01 -1.33792996e+00 -3.41535121e-01 6.28502071e-01
-4.83579524e-02 -1.24269474e+00 -6.24044299e-01 6.84692383e-01
-4.98499870e-01 -1.36179909e-01 -7.03751326e-01 -8.25570405e-01
2.45694607e-03 -4.91122752e-01 1.08595264e+00 -7.23459363e-01
1.30455285e-01 -1.50198981e-01 -5.82936883e-01 -3.07694197e-01
-1.14842618e+00 -5.04968166e-02 5.83976448e-01 -8.76225233e-02
8.76175985e-02 -8.52123141e-01 2.07042899e-02 2.99110591e-01
-7.64139175e-01 -7.34817609e-02 2.11973712e-01 1.00333548e+00
6.61193967e-01 1.11451009e-02 9.62159455e-01 -5.66293240e-01
1.02099335e+00 -2.81866312e-01 -4.38216835e-01 -9.64904502e-02
-2.24989027e-01 1.92189857e-01 1.12906706e+00 -6.75910771e-01
-1.06161022e+00 -4.44676690e-02 -4.90055680e-01 -6.94329381e-01
-2.50687957e-01 4.24349457e-01 -5.20965219e-01 8.01118553e-01
6.92998409e-01 2.43417293e-01 -2.32712571e-02 -5.78034163e-01
7.15113938e-01 1.05668354e+00 6.76626563e-01 -6.34004414e-01
4.73358929e-01 -4.33664739e-01 -6.11369133e-01 -1.31264293e+00
-5.49031794e-01 -9.82250497e-02 -6.78403556e-01 8.04045647e-02
8.68024349e-01 -6.77114785e-01 -1.99921161e-01 6.15968928e-02
-1.29873025e+00 -3.35272521e-01 -8.36785376e-01 5.33296883e-01
-8.56717229e-01 7.63935074e-02 -8.06750178e-01 -8.56974483e-01
-1.12466507e-01 -1.44328189e+00 9.31792021e-01 4.94560599e-03
-6.28434479e-01 -8.98689806e-01 1.18235022e-01 -8.87916163e-02
4.22500789e-01 4.55612093e-02 1.05724096e+00 -6.33871436e-01
-1.82647761e-02 7.67940357e-02 2.96535820e-01 9.06595230e-01
5.98522723e-01 1.21959120e-01 -1.53815222e+00 5.30644432e-02
3.46224189e-01 -2.58784235e-01 4.77978379e-01 4.75627810e-01
9.97807384e-01 -6.37020767e-01 2.04747647e-01 4.66532767e-01
9.02079105e-01 4.77730036e-01 5.77310205e-01 -2.96629816e-01
3.61615717e-01 9.01916146e-01 2.72023141e-01 2.28481114e-01
-2.36916497e-01 9.03989911e-01 -1.72304899e-01 2.29393050e-01
-4.54055190e-01 -4.81040180e-01 7.12247431e-01 1.81342590e+00
4.76585150e-01 -3.49875450e-01 -8.18601787e-01 5.02689362e-01
-1.44187868e+00 -7.77339280e-01 3.29329759e-01 1.96584773e+00
1.23276043e+00 5.39231598e-01 5.35057262e-02 3.71446848e-01
6.30046546e-01 3.72379631e-01 -1.07978515e-01 -1.18874335e+00
-1.84789106e-01 1.57332808e-01 1.06283650e-01 8.71206820e-01
-7.14847386e-01 1.20863056e+00 6.56499529e+00 1.12733400e+00
-1.31697822e+00 1.57171980e-01 4.22531545e-01 8.35751668e-02
-4.14314628e-01 -3.11896324e-01 -8.49925637e-01 6.02976859e-01
1.54584908e+00 -2.75009722e-01 3.48845571e-01 8.76164496e-01
3.28721881e-01 4.74498123e-01 -1.37070489e+00 9.41063941e-01
-1.46381810e-01 -1.26662648e+00 1.03384860e-01 -7.92955160e-02
5.31028032e-01 -2.92178154e-01 2.33272389e-01 6.62524223e-01
-1.62078999e-02 -1.34649587e+00 9.43069637e-01 3.06389540e-01
9.87674713e-01 -6.80692255e-01 4.66810822e-01 4.09547985e-01
-1.27988911e+00 1.49208084e-01 -4.58534598e-01 -5.40235825e-02
6.10268295e-01 1.56769574e-01 -1.18071365e+00 3.45803946e-01
2.60596037e-01 1.80181757e-01 1.10441029e-01 3.09800655e-01
-1.32835852e-02 8.68573129e-01 -2.55133808e-01 -3.92936915e-02
2.34453633e-01 -2.21855819e-01 7.16639638e-01 1.46633756e+00
3.17240328e-01 -1.92258194e-01 1.44486859e-01 1.04053712e+00
-4.00238633e-02 1.53606191e-01 -7.46616542e-01 -2.83279181e-01
7.76382327e-01 5.76139629e-01 -2.84270644e-01 -2.45797560e-01
-2.42937863e-01 9.84571159e-01 8.45736638e-02 2.36719176e-01
-7.46495545e-01 -2.20027238e-01 9.20158565e-01 -1.97327249e-02
1.64297938e-01 -3.15148294e-01 -3.76947135e-01 -6.58973277e-01
-1.20722204e-01 -7.82654881e-01 -3.25289875e-01 -8.22798789e-01
-1.31743276e+00 1.13479447e+00 1.52254403e-01 -1.12007594e+00
-8.42466235e-01 -5.95908344e-01 -5.28441787e-01 1.17541718e+00
-1.12580824e+00 -8.08533072e-01 3.38742405e-01 7.65423924e-02
1.26647198e+00 -3.10701609e-01 1.39091766e+00 7.54457563e-02
-3.08052182e-01 7.16245294e-01 6.41756728e-02 1.30379468e-01
4.42736804e-01 -1.35793746e+00 6.20886922e-01 4.41697657e-01
4.52640355e-01 4.87743556e-01 9.59406078e-01 -2.53027648e-01
-7.38494396e-01 -9.43996131e-01 8.24271262e-01 -3.33935916e-01
8.29905033e-01 -6.90455198e-01 -1.20818353e+00 5.49018860e-01
5.68465173e-01 -1.96522117e-01 8.78414512e-01 1.26858667e-01
-1.71513379e-01 -4.21617106e-02 -6.95731461e-01 8.97241890e-01
6.29532516e-01 -9.57336545e-01 -1.21400869e+00 -8.56372565e-02
1.52886891e+00 -2.76823819e-01 -9.62849379e-01 3.04865479e-01
3.81451964e-01 -9.47531939e-01 7.72595167e-01 -4.38786954e-01
3.33078533e-01 4.73705083e-02 -6.44882917e-01 -1.63121355e+00
-1.74894661e-01 -9.24857914e-01 7.63369650e-02 1.55939972e+00
7.51978815e-01 -4.40414608e-01 4.66653109e-01 1.61272347e-01
-5.75060248e-01 -6.26883745e-01 -1.19620275e+00 -1.04766440e+00
4.73860115e-01 -5.24557233e-01 7.29731500e-01 8.68104815e-01
2.87619889e-01 6.61310077e-01 -5.11592388e-01 1.95376948e-01
4.13092040e-02 -2.61875331e-01 5.04113376e-01 -9.10568953e-01
-5.86398065e-01 -6.37127995e-01 -2.28606299e-01 -1.20747483e+00
4.26862895e-01 -6.28269494e-01 3.22056651e-01 -9.05050397e-01
-7.28239894e-01 -3.24391663e-01 -2.19943389e-01 2.42070019e-01
1.86685681e-01 -1.51542619e-01 3.36974561e-01 -3.50223407e-02
5.01261771e-01 9.93201017e-01 8.90834272e-01 3.56803574e-02
-6.40339017e-01 -5.04678302e-02 -2.27668062e-02 8.31177294e-01
1.20755732e+00 -3.74794871e-01 -6.52904928e-01 -1.74495280e-01
-3.08521807e-01 7.18396068e-01 -1.99719429e-01 -1.00090313e+00
-1.50915653e-01 -1.43873960e-01 -8.27177912e-02 -5.56614995e-01
1.16760516e+00 -7.53008664e-01 1.99554354e-01 1.05796538e-01
-6.49605274e-01 -9.51116160e-02 3.56639326e-01 3.22206646e-01
-7.26959109e-01 -5.30292153e-01 9.56933141e-01 2.19188537e-02
-3.48472238e-01 -2.18836784e-01 -5.87170720e-01 -7.14539438e-02
6.64936006e-01 -4.62879092e-01 7.06369523e-03 -3.23967248e-01
-9.89903688e-01 -2.73479223e-01 6.57567531e-02 7.82289147e-01
3.88065487e-01 -1.44074500e+00 -6.09799683e-01 6.73526287e-01
-6.07047267e-02 -2.47845963e-01 1.09672703e-01 3.22946638e-01
-1.35910302e-01 5.48249900e-01 -2.38858357e-01 -5.19062161e-01
-1.10765886e+00 3.50384653e-01 4.00357068e-01 3.87649648e-02
-7.10966110e-01 1.00006163e+00 3.45935911e-01 -6.31502986e-01
4.30203199e-01 -7.39421189e-01 -1.21818341e-01 2.95549542e-01
2.30756760e-01 -4.64052223e-02 -1.25099391e-01 -7.85915375e-01
-1.27355486e-01 3.54158282e-01 1.21811800e-01 -5.83224237e-01
8.54879260e-01 -1.46162778e-01 4.24068958e-01 1.22433412e+00
1.23408759e+00 5.58310986e-01 -1.26617825e+00 4.52230796e-02
1.35911644e-01 -5.68588227e-02 -3.94050777e-02 -5.52942753e-01
-4.13857788e-01 1.19643676e+00 3.00006509e-01 7.36106038e-01
7.96396077e-01 -1.13259837e-01 8.03391218e-01 6.47363886e-02
9.11164004e-03 -1.25298584e+00 2.76766531e-02 7.36377597e-01
1.33182657e+00 -9.14070129e-01 -6.51514947e-01 -3.69713545e-01
-1.03164780e+00 1.17026484e+00 3.81322622e-01 -1.63232118e-01
7.75816977e-01 7.17744112e-01 1.59453750e-01 4.47812170e-01
-9.46199715e-01 1.36370808e-01 2.28094742e-01 6.83936536e-01
5.73819637e-01 3.49390298e-01 -1.51466317e-02 1.05534613e+00
-1.05130565e+00 -6.59619808e-01 6.56456947e-01 3.10064852e-01
-7.03613460e-01 -1.39112115e+00 -3.83749187e-01 6.09000921e-02
-4.39769745e-01 -2.74699628e-01 -2.42903203e-01 6.57583714e-01
-6.96625337e-02 9.61218536e-01 2.93561012e-01 -4.36140686e-01
4.04750764e-01 6.55459821e-01 3.23887765e-02 -8.66376579e-01
-4.54343081e-01 6.04275227e-01 4.59274381e-01 -2.44064972e-01
-3.65302861e-02 -3.70262206e-01 -1.34737575e+00 1.34999305e-01
-3.09353650e-01 3.73431623e-01 6.26709044e-01 6.10717297e-01
7.75561258e-02 1.06795084e+00 7.01508224e-01 -8.98456454e-01
-8.62374425e-01 -1.30969644e+00 -7.61901319e-01 1.98719911e-02
4.89461988e-01 -5.02064586e-01 -4.58110809e-01 3.03800642e-01]
|
[14.916479110717773, 6.626264572143555]
|
5fb66e03-5866-4a4a-a413-95c6074a128c
|
deep-sparse-coding-for-non-intrusive-load
|
1912.12128
| null |
https://arxiv.org/abs/1912.12128v1
|
https://arxiv.org/pdf/1912.12128v1.pdf
|
Deep Sparse Coding for Non-Intrusive Load Monitoring
|
Energy disaggregation is the task of segregating the aggregate energy of the entire building (as logged by the smartmeter) into the energy consumed by individual appliances. This is a single channel (the only channel being the smart-meter) blind source (different electrical appliances) separation problem. The traditional way to address this is via stochastic finite state machines (e.g. Factorial Hidden Markov Model). In recent times dictionary learning based approaches have shown promise in addressing the disaggregation problem. The usual technique is to learn a dictionary for every device and use the learnt dictionaries as basis for blind source separation during disaggregation. Prior studies in this area are shallow learning techniques, i.e. they learn a single layer of dictionary for every device. In this work, we propose a deep learning approach, instead of learning one level of dictionary, we learn multiple layers of dictionaries for each device. These multi-level dictionaries are used as a basis for source separation during disaggregation. Results on two benchmark datasets and one actual implementation show that our method outperforms state-of-the-art techniques.
|
['Angshul Majumdar', 'Shikha Singh']
|
2019-12-11
| null | null | null | null |
['non-intrusive-load-monitoring', 'non-intrusive-load-monitoring', 'non-intrusive-load-monitoring']
|
['knowledge-base', 'miscellaneous', 'time-series']
|
[ 1.40714899e-01 -1.21754780e-01 -3.33036363e-01 -8.51591676e-02
-1.19953346e+00 -8.75489116e-01 6.06703699e-01 2.99765915e-01
-8.69824216e-02 4.20084059e-01 5.35386443e-01 -2.65581965e-01
2.82696158e-01 -8.20872843e-01 -6.55365109e-01 -1.26563680e+00
1.38453931e-01 4.42245543e-01 -3.04664284e-01 8.97007510e-02
-1.99487045e-01 3.34533006e-01 -1.36382508e+00 1.01085082e-01
5.46321690e-01 1.19639456e+00 8.57673734e-02 8.54675829e-01
-8.77342001e-02 1.14252687e+00 -8.21890771e-01 3.98834348e-02
3.70235741e-01 -5.86754501e-01 -6.73424065e-01 -5.93506731e-02
1.64858416e-01 -6.00174904e-01 -4.55011576e-01 1.33134627e+00
7.93135285e-01 -1.68603972e-01 5.13442218e-01 -1.22707462e+00
-5.83774269e-01 1.16502440e+00 -2.59401053e-01 2.98188269e-01
3.42454523e-01 -1.23326577e-01 1.16384423e+00 -3.53569269e-01
-3.41150552e-01 7.93773532e-01 5.87732792e-01 2.28180751e-01
-1.51532066e+00 -6.96840167e-01 -8.08680721e-04 3.61662686e-01
-1.46902347e+00 -6.76602483e-01 1.06685913e+00 -4.82160479e-01
1.37796533e+00 2.19985604e-01 4.84982282e-01 1.07844567e+00
1.10522963e-01 1.12320888e+00 1.25586379e+00 -4.11327064e-01
7.18932271e-01 1.74757555e-01 2.92454779e-01 4.72817630e-01
3.85751069e-01 -9.02188495e-02 -3.41363966e-01 -4.42984313e-01
3.21532458e-01 1.80004120e-01 -8.58651623e-02 -3.70956868e-01
-1.14847267e+00 8.34268034e-01 1.70138866e-01 5.17891884e-01
-5.45933843e-01 2.76799291e-01 3.36267948e-01 7.95729160e-02
2.38686666e-01 -2.11917534e-01 -5.74689925e-01 -1.02156997e-01
-1.20425332e+00 -1.99995920e-01 1.36500633e+00 8.01168919e-01
7.91599810e-01 4.82592076e-01 -1.57111973e-01 4.53690827e-01
4.32240814e-01 8.65058839e-01 7.40144789e-01 -6.23742640e-01
4.53913003e-01 4.14064348e-01 1.00649238e-01 -5.39502561e-01
-4.98608321e-01 -3.61246884e-01 -1.29429221e+00 1.11790664e-01
3.74499857e-01 -4.70426649e-01 -9.82601166e-01 1.49809945e+00
1.67680979e-01 4.36481267e-01 1.88824654e-01 5.36634803e-01
6.95464790e-01 7.90987134e-01 3.70208099e-02 -2.04966158e-01
1.24914193e+00 -8.24277043e-01 -9.15243924e-01 -2.00374007e-01
3.60137135e-01 -4.36489642e-01 4.23969179e-01 6.65425599e-01
-8.06597888e-01 -3.94360602e-01 -1.28455079e+00 -9.36449319e-03
-8.74868870e-01 2.07298175e-01 4.26879585e-01 1.06373131e+00
-8.07789862e-01 4.09524769e-01 -1.09332013e+00 -1.77077398e-01
3.03805977e-01 5.68628252e-01 1.65664315e-01 3.79270017e-01
-1.11490643e+00 7.97515810e-01 4.13639516e-01 -8.71853456e-02
-1.34292006e+00 -3.18211764e-01 -9.46635127e-01 2.53847182e-01
-2.20168266e-03 -5.07763743e-01 1.24035895e+00 -8.85234952e-01
-1.62098527e+00 5.23519516e-01 -3.37186307e-02 -7.46526718e-01
5.61009645e-02 -5.36697507e-02 -7.68472075e-01 -2.76978105e-01
1.28453583e-01 -4.17472124e-02 1.29122543e+00 -1.34073889e+00
-9.51593399e-01 -4.48757976e-01 -2.87660751e-02 -1.38160303e-01
-4.34310228e-01 -2.41879150e-01 -1.19043663e-02 -7.99099565e-01
-2.62282905e-03 -9.23030555e-01 -1.20516591e-01 -8.90952587e-01
-7.86482275e-01 -1.66721612e-01 6.98360503e-01 -1.25492477e+00
1.42916024e+00 -2.01529551e+00 1.41661927e-01 2.11564407e-01
1.40184298e-01 1.06802948e-01 2.77379543e-01 3.58977795e-01
-3.58404249e-01 -4.10938084e-01 -3.71627361e-01 -9.27316785e-01
6.19953513e-01 3.67231339e-01 -4.02594358e-01 8.22587073e-01
-4.96420473e-01 1.01174915e+00 -1.02537072e+00 -1.17936805e-01
5.56959629e-01 6.59258842e-01 6.68357685e-02 2.07026809e-01
-4.63249302e-03 3.95790607e-01 -1.04548015e-01 7.25653946e-01
6.35343254e-01 -2.85959274e-01 2.10968569e-01 -4.90439415e-01
1.20728649e-01 6.26973212e-01 -1.72144318e+00 1.88951361e+00
-7.22813666e-01 4.97135729e-01 2.11739823e-01 -1.15104187e+00
3.43317360e-01 7.08035767e-01 8.31372917e-01 -6.95579231e-01
2.86171257e-01 1.48353443e-01 -4.48963225e-01 -4.83798757e-02
2.12878957e-01 -1.58279493e-01 -5.12318552e-01 7.03119397e-01
3.31763417e-01 1.41022116e-01 -6.40241653e-02 -1.16654634e-02
1.14663827e+00 -4.10946548e-01 4.79372799e-01 -2.35766560e-01
4.27219093e-01 -2.52510369e-01 4.69217569e-01 7.08589673e-01
2.76834499e-02 4.19281214e-01 -5.08546829e-02 -2.75888205e-01
-6.90092385e-01 -9.81710434e-01 7.93364421e-02 8.17805111e-01
-1.76179990e-01 -4.05642271e-01 -8.48425329e-01 -7.79414594e-01
1.55029193e-01 9.51553702e-01 -3.86770338e-01 -1.27079442e-01
-3.97702217e-01 -1.15127504e+00 5.12531102e-01 6.45559847e-01
5.40653646e-01 -6.62938297e-01 -6.43832684e-01 4.23904419e-01
-3.33091259e-01 -1.11624134e+00 -3.85730535e-01 1.02464902e+00
-5.12634456e-01 -8.61972690e-01 -5.46607316e-01 -6.91940963e-01
3.13232362e-01 8.80739912e-02 1.32049918e+00 -5.24251342e-01
4.78285551e-03 5.07400692e-01 -1.36928007e-01 -4.77003515e-01
-5.74926198e-01 1.43000722e-01 2.71626025e-01 5.37749171e-01
6.66564882e-01 -9.35548663e-01 -4.43499833e-01 -1.70788601e-01
-9.22082126e-01 -3.64032775e-01 5.02925634e-01 3.32439959e-01
7.50094593e-01 9.33440864e-01 2.78034627e-01 -5.53430140e-01
4.62615907e-01 -6.94894373e-01 -8.16331029e-01 2.61431728e-02
-7.49864995e-01 1.05392404e-01 1.02440941e+00 -2.62653738e-01
-5.44947803e-01 5.08062065e-01 -8.64407569e-02 -3.57072622e-01
-4.25624222e-01 -6.13085330e-02 -5.99676669e-01 8.68416205e-02
7.36875162e-02 6.28668666e-01 -7.30372846e-01 -9.37065661e-01
4.64118153e-01 9.62503493e-01 5.15954494e-01 -5.27594611e-02
9.79385197e-01 6.27554059e-01 -1.60005853e-01 -6.67992949e-01
-7.47759759e-01 -5.73650777e-01 -6.63730502e-01 2.47786313e-01
1.22220361e+00 -1.38500226e+00 -6.53107226e-01 7.52070308e-01
-9.60517466e-01 -4.21900600e-01 -6.85624301e-01 3.77540648e-01
-2.61190802e-01 1.67125806e-01 -3.62392396e-01 -7.91985929e-01
-3.52145344e-01 -1.07507908e+00 1.57069075e+00 1.94451079e-01
-1.68796524e-01 -1.27654517e+00 3.23416024e-01 2.99718320e-01
2.32453674e-01 3.63784850e-01 6.82055771e-01 -7.88868845e-01
-4.57393736e-01 -3.82632256e-01 2.44904250e-01 8.32350731e-01
7.52441525e-01 -8.43619645e-01 -1.37105012e+00 -5.87191701e-01
6.54323161e-01 1.11938737e-01 8.76935065e-01 4.07268852e-01
1.00035405e+00 -6.85092926e-01 -2.74275005e-01 6.16519094e-01
1.82861006e+00 2.21623421e-01 2.94234753e-01 2.07360717e-03
1.21318555e+00 -2.39232898e-01 -3.86648118e-01 5.58416128e-01
9.36156690e-01 5.62891066e-01 3.47451627e-01 -2.18013808e-01
-3.65151167e-01 -2.97978401e-01 7.07344949e-01 9.99014437e-01
4.71519142e-01 -2.87317932e-01 -8.21060777e-01 9.21966672e-01
-1.73976493e+00 -9.09561515e-01 3.95031795e-02 2.07558203e+00
8.07446182e-01 4.52465797e-03 2.84152925e-01 7.49691427e-01
2.86058366e-01 4.09374923e-01 -7.49247551e-01 -1.16396137e-01
-2.28798300e-01 9.83588323e-02 1.00715029e+00 6.73662841e-01
-1.34922731e+00 3.58222634e-01 6.39369726e+00 8.22761536e-01
-9.68950331e-01 5.49411237e-01 3.21546406e-01 -3.04038357e-02
-1.38830528e-01 -1.47141725e-01 -9.12594914e-01 9.72479999e-01
1.26768363e+00 2.33806416e-01 9.24225926e-01 6.58090174e-01
6.37621284e-02 -3.36911112e-01 -1.50274110e+00 1.48939431e+00
1.01392515e-01 -9.87178802e-01 -2.37821966e-01 2.26676986e-01
9.74003792e-01 3.89185190e-01 -1.31172851e-01 1.03859186e-01
9.52561677e-01 -8.77408922e-01 8.56423378e-01 4.11939830e-01
3.74884129e-01 -5.90075314e-01 7.79137075e-01 4.44839895e-01
-1.49883902e+00 -3.98655266e-01 2.25018531e-01 3.42991017e-02
2.26026133e-01 1.03807902e+00 -3.11166465e-01 6.35448873e-01
1.01848185e+00 7.06179202e-01 -5.40213704e-01 4.53401983e-01
-2.56513774e-01 9.42519724e-01 -6.74554884e-01 3.97946864e-01
-6.94988072e-02 -1.19615331e-01 4.53094810e-01 1.41816318e+00
-3.85988913e-02 -1.74832359e-01 3.25135499e-01 6.70201004e-01
-1.61921289e-02 -5.24950802e-01 -4.56728071e-01 2.78228223e-02
2.37718657e-01 1.17140794e+00 -7.14313567e-01 -6.08791888e-01
-7.08399236e-01 1.31705964e+00 -2.61076152e-01 4.56311464e-01
-8.94967258e-01 -3.25641841e-01 7.43364811e-01 -2.80986875e-01
8.57757151e-01 -2.22880363e-01 -5.74907005e-01 -1.34241867e+00
-1.99557945e-01 -1.02587831e+00 3.65452141e-01 -4.74707186e-01
-1.34672463e+00 1.42872915e-01 -7.83718079e-02 -1.02119434e+00
-3.26728672e-01 -1.41329035e-01 -3.90156984e-01 9.34800684e-01
-1.64045143e+00 -9.93082047e-01 -2.17182636e-01 1.16330445e+00
5.25129914e-01 -1.74084753e-01 1.09199595e+00 6.29524231e-01
-7.07529068e-01 3.61255586e-01 6.95409000e-01 5.51864684e-01
1.16734028e-01 -1.83526909e+00 5.09007454e-01 1.11891913e+00
7.65945017e-01 6.73828796e-02 7.59626806e-01 -4.43380684e-01
-1.80712128e+00 -1.14832473e+00 8.42357934e-01 -8.55461121e-01
8.24310601e-01 -7.11124539e-01 -3.85155350e-01 8.18551481e-01
5.14680028e-01 -8.51625353e-02 1.00441015e+00 -4.33841228e-01
-2.04552174e-01 -3.95613909e-01 -1.18023050e+00 -4.58126999e-02
4.30753917e-01 -1.13183284e+00 -5.56829929e-01 3.44677180e-01
2.97138482e-01 -1.08229674e-01 -9.04075921e-01 -7.58197755e-02
2.99332529e-01 -7.76108623e-01 9.49315131e-01 5.34444000e-04
-4.26148683e-01 -6.59636915e-01 -6.79726064e-01 -1.48921204e+00
-3.93784225e-01 -7.79281318e-01 -1.21679676e+00 1.54928672e+00
4.87612411e-02 -6.12751961e-01 6.23990059e-01 4.82973099e-01
2.32510120e-01 -1.89721480e-01 -9.85623300e-01 -7.98441589e-01
1.73870474e-03 -4.01461303e-01 1.03697860e+00 9.25199389e-01
-2.16113493e-01 5.76649547e-01 -2.82711118e-01 7.12541580e-01
1.14546454e+00 4.20692861e-01 4.85199869e-01 -1.02677798e+00
-5.93992233e-01 -2.61693776e-01 -1.31244838e-01 -1.26061738e+00
1.20749526e-01 -9.23856199e-01 8.84641036e-02 -1.89774644e+00
5.61962985e-02 -1.03181347e-01 -6.54885113e-01 5.80243528e-01
1.48368493e-01 1.21699281e-01 1.44461349e-01 3.74403968e-02
-5.71272731e-01 3.48257571e-01 1.74545556e-01 -8.14420342e-01
-2.10496232e-01 2.12118313e-01 -8.32606316e-01 6.81936800e-01
1.03471446e+00 -7.67043531e-01 -3.01066369e-01 -7.19146609e-01
1.36932237e-02 -2.62270123e-01 4.73403871e-01 -1.24850762e+00
5.28875053e-01 3.40174019e-01 3.94854933e-01 -6.35538101e-01
2.14901686e-01 -1.43065870e+00 4.85768467e-01 3.51266861e-01
1.51742488e-01 -2.35099077e-01 3.12997289e-02 6.77107513e-01
-4.83623557e-02 1.66746721e-01 4.55590636e-01 -5.30927703e-02
-5.97228944e-01 6.82592764e-02 -6.00696683e-01 -2.17607543e-01
6.97164893e-01 2.13519903e-03 2.61916637e-01 -3.95276099e-01
-1.01319349e+00 2.89145373e-02 2.09307641e-01 4.88128930e-01
2.20439560e-03 -1.54250991e+00 -4.77279723e-01 3.61777335e-01
-2.17976436e-01 -4.50902544e-02 -1.58417091e-01 6.36422217e-01
2.23437950e-01 6.21577442e-01 3.25410128e-01 -4.74816322e-01
-9.69852030e-01 7.99404681e-01 6.33370817e-01 -4.26770061e-01
-4.49261338e-01 5.13598502e-01 -5.20256162e-02 -6.36815056e-02
6.70713782e-01 -8.61240745e-01 -7.72276595e-02 4.36942369e-01
6.17781520e-01 7.09174514e-01 5.66510618e-01 -9.80338633e-01
-4.94430810e-01 6.29331291e-01 4.88453329e-01 9.41565037e-02
1.39561927e+00 -4.34670925e-01 -1.61937073e-01 7.95973241e-01
1.49748480e+00 4.60086279e-02 -1.07340729e+00 -4.13796842e-01
-2.97131352e-02 1.22394748e-01 6.01794422e-01 -8.67432594e-01
-1.42297065e+00 7.01567352e-01 1.16267884e+00 1.00376499e+00
1.46437109e+00 7.72280544e-02 1.11012053e+00 1.67816043e-01
5.54434955e-01 -1.27586532e+00 -6.71114862e-01 1.39958501e-01
1.54485822e-01 -1.31364083e+00 -1.79581508e-01 2.61015981e-01
-8.13417360e-02 6.36928082e-01 -2.51416415e-01 -1.15692563e-01
1.26048470e+00 6.50848925e-01 -1.27950236e-01 -1.63472980e-01
-1.59872279e-01 -5.39793193e-01 1.03088312e-01 6.02316976e-01
-1.71638086e-01 3.99056941e-01 5.09650350e-01 8.61792088e-01
-1.56932384e-01 -1.09346390e-01 2.60909885e-01 1.06493914e+00
-9.55163911e-02 -1.07212663e+00 -5.84723413e-01 3.65191460e-01
-5.95710874e-01 -1.80299118e-01 -2.74109215e-01 -1.47109717e-01
4.77011025e-01 1.48365903e+00 -1.21526428e-01 -3.49464655e-01
4.45139825e-01 1.54307276e-01 3.53073925e-01 -5.66651642e-01
-7.20495701e-01 3.63378711e-02 -2.63172001e-01 -6.61112726e-01
-5.86736500e-01 -8.56531501e-01 -1.03769767e+00 -2.95718700e-01
-2.29324818e-01 1.57627329e-01 8.22736263e-01 9.69639242e-01
1.43599793e-01 7.92061150e-01 8.64336908e-01 -1.11309874e+00
-5.86902738e-01 -5.02097547e-01 -1.01436591e+00 2.13977903e-01
1.20491052e+00 -2.08423883e-01 -5.74981153e-01 4.06240851e-01]
|
[16.09954071044922, 7.610836029052734]
|
3e16c48d-b988-45c5-8616-81213acca2de
|
ice-core-dating-using-probabilistic
|
2210.16568
| null |
https://arxiv.org/abs/2210.16568v1
|
https://arxiv.org/pdf/2210.16568v1.pdf
|
Ice Core Dating using Probabilistic Programming
|
Ice cores record crucial information about past climate. However, before ice core data can have scientific value, the chronology must be inferred by estimating the age as a function of depth. Under certain conditions, chemicals locked in the ice display quasi-periodic cycles that delineate annual layers. Manually counting these noisy seasonal patterns to infer the chronology can be an imperfect and time-consuming process, and does not capture uncertainty in a principled fashion. In addition, several ice cores may be collected from a region, introducing an aspect of spatial correlation between them. We present an exploration of the use of probabilistic models for automatic dating of ice cores, using probabilistic programming to showcase its use for prototyping, automatic inference and maintainability, and demonstrate common failure modes of these tools.
|
['Markus Kaiser', 'Neil D. Lawrence', 'J. Scott Hosking', 'Richard E. Turner', 'Will Tebbutt', 'Ieva Kazlauskaite', 'Tom R. Andersson', 'Aditya Ravuri']
|
2022-10-29
| null | null | null | null |
['probabilistic-programming']
|
['methodology']
|
[-4.06985320e-02 -8.04198757e-02 2.36836940e-01 -3.45601022e-01
-5.41509926e-01 -9.59812999e-01 5.75368762e-01 2.85571009e-01
-3.56072485e-01 1.03847456e+00 6.50412291e-02 -5.91573775e-01
1.50590986e-01 -1.01639116e+00 -5.28087854e-01 -6.07670724e-01
-5.55496812e-01 9.48894739e-01 4.10864353e-01 1.31060556e-01
3.25880110e-01 5.23810029e-01 -1.45399606e+00 -2.10221276e-01
5.66573381e-01 3.41573507e-01 3.19525480e-01 7.40771532e-01
-2.27438673e-01 2.25206673e-01 -6.46345317e-01 -5.69035783e-02
-6.03818037e-02 -8.63486454e-02 -5.36667347e-01 -3.63791019e-01
-8.20826218e-02 -4.59135979e-01 1.83328792e-01 7.85518706e-01
-9.81249958e-02 -3.24414104e-01 7.65874565e-01 -8.61933112e-01
1.55928507e-01 7.96193480e-01 -5.30672550e-01 -2.60905474e-02
3.19875449e-01 2.23136753e-01 8.52645397e-01 -5.81428111e-01
5.07901549e-01 1.09885275e+00 9.36408222e-01 -3.28973494e-02
-1.47787321e+00 -7.00939298e-01 -1.09943949e-01 -4.71730322e-01
-1.38288426e+00 -5.44145107e-01 3.34744871e-01 -5.95535278e-01
8.02139401e-01 4.03365433e-01 9.94997323e-01 5.94731271e-01
4.80748773e-01 2.31158763e-01 1.46496117e+00 -2.47771427e-01
4.89268303e-01 -1.27871409e-01 -1.49230689e-01 2.54791707e-01
7.83542931e-01 2.67066449e-01 -5.57124376e-01 -6.21300757e-01
5.29515445e-01 -1.74582943e-01 -4.28579189e-02 9.04822424e-02
-7.19721615e-01 4.50926185e-01 -8.79049450e-02 1.67316794e-02
-7.53875971e-02 5.39406002e-01 3.43775958e-01 -7.29044387e-03
4.84789550e-01 4.62760985e-01 -5.99363089e-01 -5.67449212e-01
-1.42386973e+00 6.69269264e-01 1.17268085e+00 7.88846552e-01
9.06211555e-01 -2.13850424e-01 7.11869478e-01 2.89842725e-01
8.52793515e-01 9.07004893e-01 -7.25032296e-03 -9.27033842e-01
-1.12709440e-01 2.17962593e-01 5.56300700e-01 -7.73337126e-01
-4.50090200e-01 1.78651512e-01 -3.47640753e-01 4.67758924e-01
7.03351438e-01 -7.28014112e-02 -9.21957016e-01 1.19233012e+00
2.21508473e-01 -1.60614103e-01 -2.06540763e-01 6.18765414e-01
2.07115158e-01 5.49852252e-01 3.56009394e-01 -2.69292057e-01
1.36389446e+00 4.30551052e-01 -5.07770836e-01 -6.45889103e-01
2.46212497e-01 -8.22094381e-01 6.12004757e-01 4.17104065e-01
-8.55545223e-01 2.06424370e-01 -1.22782552e+00 3.12118918e-01
-2.88326800e-01 -3.78866315e-01 8.55309546e-01 9.18625176e-01
-7.72989571e-01 1.01082015e+00 -1.43996942e+00 -2.28397474e-01
-5.22531271e-02 1.04080513e-01 -6.49414212e-02 2.42580071e-01
-1.09574199e+00 8.83666396e-01 3.33492696e-01 4.82354343e-01
-7.33811617e-01 -7.41150498e-01 -7.96967626e-01 -2.49244124e-01
-2.06307188e-01 -2.15577558e-01 1.23745763e+00 -2.18902588e-01
-1.04422832e+00 9.81687069e-01 -2.74728358e-01 -1.75535694e-01
6.10722184e-01 -2.25635231e-01 -2.17755720e-01 -9.22262147e-02
2.51880437e-01 2.67232955e-01 5.24561286e-01 -1.25152469e+00
-4.19823527e-01 -4.30460453e-01 -4.00075793e-01 -2.31532261e-01
3.50335866e-01 1.59354627e-01 -1.05160363e-01 -2.66625911e-01
6.18574679e-01 -8.82631004e-01 -4.31005836e-01 -1.67885851e-02
-4.21177074e-02 1.95617795e-01 6.37340307e-01 -8.96520257e-01
1.18349922e+00 -1.95429957e+00 -4.45554942e-01 4.81495440e-01
-5.04075401e-02 -4.79773819e-01 7.20035732e-01 6.11697018e-01
3.25424522e-01 5.58643401e-01 -9.37113404e-01 -8.93921033e-02
3.49422917e-02 4.45429027e-01 -3.63870978e-01 7.37272263e-01
1.37462616e-01 2.50354946e-01 -1.18833959e+00 -6.74078703e-01
1.41395524e-01 1.69801593e-01 -2.38340981e-02 2.61354796e-03
-4.40704435e-01 6.69364035e-02 -2.09252298e-01 9.51810360e-01
8.60492170e-01 2.16348767e-01 7.70537913e-01 3.19292337e-01
-5.42701423e-01 6.47919178e-01 -1.21843517e+00 1.41966403e+00
-3.55363071e-01 7.26709902e-01 2.32176095e-01 -1.61119714e-01
1.19468236e+00 1.84757605e-01 2.56851763e-01 -2.16350853e-01
-2.60199755e-01 4.45414245e-01 -1.79559603e-01 -4.47213560e-01
1.07775497e+00 -7.35818923e-01 -2.81878471e-01 8.85138273e-01
-5.25352716e-01 -6.69080436e-01 1.57188311e-01 -4.55595329e-02
1.05307782e+00 5.88490605e-01 1.04360953e-01 -7.02788591e-01
-3.14991593e-01 3.18518788e-01 8.99572015e-01 6.10978246e-01
-2.35114843e-02 7.73502231e-01 5.74251831e-01 -5.74582636e-01
-1.51135945e+00 -1.25625229e+00 -6.34145796e-01 3.11928004e-01
-1.60915047e-01 -5.71520627e-01 -2.42710829e-01 -4.55272645e-02
2.66912580e-01 6.38158143e-01 -5.30076683e-01 2.22253233e-01
-2.26378798e-01 -1.28014350e+00 6.80915296e-01 6.36024594e-01
-6.85109943e-02 -6.49402976e-01 -1.16173089e+00 4.82170492e-01
-1.08935453e-01 -6.96075916e-01 8.18478540e-02 4.60855812e-01
-1.29236114e+00 -1.04663050e+00 -3.13815773e-01 -1.74248800e-01
6.26726747e-01 -1.69070750e-01 1.46375287e+00 1.29322544e-01
-2.62627602e-01 -3.68500762e-02 -1.63410380e-01 -6.37791812e-01
-5.66276312e-01 -1.29393563e-01 7.22795799e-02 -8.30817878e-01
6.23005986e-01 -1.00193882e+00 -5.16446412e-01 2.84133822e-01
-7.54279912e-01 -2.04215527e-01 2.46115625e-01 5.23728430e-01
2.96080440e-01 8.47442150e-02 2.03890786e-01 -9.33572173e-01
2.18519881e-01 -6.15411043e-01 -9.35484469e-01 1.37021601e-01
-5.70979059e-01 3.32205564e-01 3.29205357e-02 1.88404366e-01
-9.18763399e-01 3.06524456e-01 1.47947550e-01 2.00317070e-01
-4.59036268e-02 8.46206546e-01 6.07981607e-02 4.23053086e-01
5.96920609e-01 4.86685103e-03 1.66577891e-01 -4.85507280e-01
-1.16595194e-01 8.07977378e-01 5.62388361e-01 -1.07434845e+00
7.69771814e-01 7.95261264e-01 -1.15211673e-01 -8.75984609e-01
-8.45982656e-02 -2.56291091e-01 -6.22872591e-01 -3.93229336e-01
3.21798980e-01 -1.09143972e+00 -7.72813678e-01 3.83763552e-01
-1.01765871e+00 -4.63821799e-01 6.99477121e-02 3.14288259e-01
-2.09565148e-01 2.91529357e-01 -3.47615272e-01 -1.33810401e+00
-8.01446289e-02 -8.47012520e-01 9.28396463e-01 1.39232516e-01
-8.09777796e-01 -8.00832629e-01 4.23727006e-01 -2.40053147e-01
2.37679571e-01 5.59980989e-01 6.01959705e-01 -1.03544839e-01
-6.20948374e-01 -3.69048774e-01 2.47978102e-02 -2.91051239e-01
1.77357540e-01 8.33243430e-01 -1.19254053e+00 -1.49857551e-01
-2.59096891e-01 9.48001072e-03 6.24030173e-01 3.60302895e-01
4.52514857e-01 -1.32418517e-02 -4.77846622e-01 3.75148386e-01
1.68633342e+00 1.66106701e-01 8.62960219e-01 6.38728142e-01
1.40966102e-01 7.47687876e-01 4.06727046e-01 7.75805116e-01
2.90117621e-01 5.39954975e-02 2.64582932e-01 4.12045807e-01
5.14774442e-01 -1.44092932e-01 2.42201865e-01 6.08758032e-01
-2.22449630e-01 5.09516656e-01 -1.36073506e+00 9.21420634e-01
-1.43992436e+00 -9.80719805e-01 -3.06194454e-01 2.54724216e+00
1.06755090e+00 5.58419526e-01 2.58115306e-02 7.49245286e-02
6.35070682e-01 1.69934526e-01 -2.37384796e-01 -4.90549952e-01
9.57896709e-02 1.15635440e-01 1.12535453e+00 6.06540263e-01
-7.80850708e-01 4.71624494e-01 8.31756401e+00 2.20306758e-02
-7.32695043e-01 -2.73607820e-01 4.92693126e-01 3.96490172e-02
-8.84627104e-01 7.59217143e-01 -7.56729841e-01 6.44898057e-01
1.12693775e+00 -1.50941685e-01 1.05875328e-01 6.71544254e-01
4.88749653e-01 -9.15398657e-01 -8.84655654e-01 2.78028578e-01
-5.17693758e-01 -1.19480550e+00 -6.42342806e-01 6.40115142e-02
3.56713593e-01 9.85702947e-02 -6.33277655e-01 -2.76367545e-01
9.94207621e-01 -9.72617805e-01 1.00351346e+00 6.82785928e-01
9.35849786e-01 -7.19468713e-01 5.48749626e-01 -3.20975110e-02
-1.22374749e+00 3.25598538e-01 -3.76705527e-01 -6.38832033e-01
2.93900013e-01 1.15816641e+00 -9.72905457e-01 3.23096216e-01
1.09482253e+00 3.87452126e-01 -2.81151116e-01 1.23532677e+00
-4.34681028e-01 8.36777210e-01 -1.21940696e+00 -1.96724623e-01
-8.12027156e-02 -3.96353751e-01 4.67264593e-01 1.12640464e+00
3.30902904e-01 -2.41031405e-02 -2.43443325e-01 1.03385937e+00
5.03212333e-01 -5.68612516e-01 -8.01350474e-01 5.83159849e-02
9.72646832e-01 9.97014105e-01 -1.22209013e+00 1.55226011e-02
-1.23898566e-01 2.00288951e-01 -1.60984159e-01 2.99954345e-03
-3.38102192e-01 -2.64341205e-01 7.85827219e-01 3.84397328e-01
-3.05908807e-02 -7.86867797e-01 -8.93386126e-01 -6.78074002e-01
2.87472367e-01 -6.27338707e-01 1.97121978e-01 -7.76963890e-01
-8.61622393e-01 5.89048453e-02 1.70994133e-01 -1.11594379e+00
-3.30924928e-01 -4.02643859e-01 -6.58441067e-01 1.00809872e+00
-9.07702684e-01 -7.48265624e-01 2.35283840e-02 -5.05636811e-01
-4.20719571e-02 4.89231199e-01 9.75149274e-01 -1.30472004e-01
-1.22698225e-01 -4.58140820e-02 5.63221693e-01 -1.49232969e-01
5.93077600e-01 -1.58225739e+00 7.29003489e-01 8.62322032e-01
-2.69482195e-01 7.26672888e-01 1.50782382e+00 -1.17487717e+00
-1.43013358e+00 -6.27992392e-01 9.47637200e-01 -6.53078437e-01
9.41874921e-01 -4.75997567e-01 -9.85176980e-01 6.50582373e-01
-2.51925290e-01 -4.89520878e-01 5.97138703e-01 7.18490124e-01
-3.02829683e-01 -1.25753239e-01 -1.00807512e+00 4.44784731e-01
4.10767853e-01 -7.32828438e-01 -6.59208715e-01 1.58438832e-02
1.44946367e-01 -3.31229061e-01 -1.14056540e+00 1.33379772e-01
1.26624227e+00 -8.13932121e-01 5.37400484e-01 9.22499746e-02
5.23486614e-01 -7.28569806e-01 -6.31570742e-02 -9.30150509e-01
1.23604182e-02 -6.83114409e-01 5.17504156e-01 1.43341875e+00
5.35809338e-01 -2.94735342e-01 1.04182053e+00 1.32948434e+00
2.69038021e-03 -7.52898976e-02 -9.56483305e-01 -7.79265761e-01
2.39572719e-01 -7.62796044e-01 6.87749863e-01 7.28284955e-01
3.33549440e-01 -3.98849726e-01 -1.07703865e-01 5.53026617e-01
8.02718043e-01 7.61901811e-02 5.62770307e-01 -1.52156997e+00
-2.13979006e-01 -1.77536935e-01 -4.09789532e-01 -3.23925942e-01
-3.46448541e-01 -2.83420354e-01 6.85508609e-01 -1.25174129e+00
4.61676493e-02 -8.94006312e-01 3.47993225e-01 3.07022631e-01
-1.50723100e-01 3.67884077e-02 -1.42152205e-01 4.38649803e-01
2.60882974e-01 -1.84883662e-02 4.46085840e-01 -1.58372242e-02
-1.41439617e-01 -1.92524210e-01 -2.89583653e-01 7.90755153e-01
6.59082592e-01 -9.73700702e-01 7.17761815e-02 -3.27714771e-01
9.44244206e-01 1.16784617e-01 2.84091264e-01 -9.26709831e-01
2.57465303e-01 -2.54279256e-01 4.67370868e-01 -8.81396592e-01
1.53961539e-01 -7.68079400e-01 8.48329842e-01 5.76825261e-01
2.50085354e-01 1.77929267e-01 4.42051291e-01 6.09938264e-01
8.45660642e-03 -6.08796179e-01 3.67324442e-01 -5.68119109e-01
-4.93913025e-01 -2.24586233e-01 -8.59243989e-01 -3.28763336e-01
5.66787839e-01 -2.64649272e-01 -1.17045239e-01 -3.49774323e-02
-5.53901672e-01 5.20115256e-01 1.06484842e+00 -1.31343663e-01
3.15401435e-01 -6.75655425e-01 -7.77447462e-01 8.63455608e-02
2.35317573e-02 2.77090847e-01 7.56923109e-02 3.17551017e-01
-1.20249200e+00 -1.95997477e-01 -1.13951333e-01 -7.53702998e-01
-8.45590889e-01 4.54808436e-02 2.94202089e-01 -6.18573166e-02
-7.15978265e-01 6.07758760e-01 -4.01719540e-01 -3.75637144e-01
-2.50681579e-01 -1.80009469e-01 3.00508380e-01 2.75455117e-01
5.89751124e-01 3.94021496e-02 2.39255968e-02 1.09434001e-01
-6.61622703e-01 3.64589155e-01 -6.64122254e-02 -7.12577999e-01
1.45790601e+00 -2.08208576e-01 -5.02543867e-01 1.14331770e+00
4.16045994e-01 -5.55834845e-02 -1.49453402e+00 2.56652921e-01
3.89968872e-01 -4.86708194e-01 -1.64929032e-01 -6.23228192e-01
-6.02675498e-01 4.78563786e-01 2.13799268e-01 2.47077689e-01
6.51354790e-01 -1.55258819e-01 2.90042490e-01 1.78520173e-01
5.10685086e-01 -1.17302382e+00 -1.17431140e+00 1.18361562e-01
6.71338737e-01 -9.74351585e-01 7.18475878e-01 -1.27190977e-01
-5.90943433e-02 1.38565087e+00 1.36219844e-01 -4.70232069e-02
9.76175904e-01 9.57955897e-01 1.42320216e-01 -2.78188229e-01
-7.72043288e-01 4.42413837e-02 -6.38076663e-01 5.86816967e-01
7.15928197e-01 4.46190983e-01 -5.63546121e-01 1.50038451e-01
-5.21016657e-01 7.41798878e-02 9.40788567e-01 1.41983438e+00
-6.05680466e-01 -1.14756393e+00 -1.03741813e+00 5.40899515e-01
-3.57656986e-01 -1.27262816e-01 -2.14619100e-01 7.74913728e-01
-3.32849890e-01 7.19861507e-01 5.05183637e-01 -2.66082823e-01
-1.64399400e-01 3.23414594e-01 2.80490488e-01 -4.13830370e-01
-2.68302798e-01 1.84388123e-02 5.56029141e-01 -7.94187784e-02
-3.66752982e-01 -1.27871084e+00 -1.16092074e+00 -7.05696940e-01
-1.87968343e-01 4.33347434e-01 1.04486001e+00 9.33875322e-01
-1.53912365e-01 6.33882806e-02 4.39141810e-01 -8.43958259e-01
-3.23417813e-01 -9.73301470e-01 -1.02453208e+00 -1.22464128e-01
6.44040406e-02 -4.47713286e-01 -5.98705888e-01 3.00739676e-01]
|
[6.62251615524292, 3.6701743602752686]
|
fc213ae7-f6d4-4d5f-a8ee-0f5f14e77605
|
why-only-micro-f1-class-weighting-of-measures-1
|
2205.09460
| null |
https://arxiv.org/abs/2205.09460v1
|
https://arxiv.org/pdf/2205.09460v1.pdf
|
Why only Micro-F1? Class Weighting of Measures for Relation Classification
|
Relation classification models are conventionally evaluated using only a single measure, e.g., micro-F1, macro-F1 or AUC. In this work, we analyze weighting schemes, such as micro and macro, for imbalanced datasets. We introduce a framework for weighting schemes, where existing schemes are extremes, and two new intermediate schemes. We show that reporting results of different weighting schemes better highlights strengths and weaknesses of a model.
|
['Christoph Alt', 'Leonhard Hennig', 'Yuxuan Chen', 'David Harbecke']
|
2022-05-19
|
why-only-micro-f1-class-weighting-of-measures
|
https://aclanthology.org/2022.nlppower-1.4
|
https://aclanthology.org/2022.nlppower-1.4.pdf
|
nlppower-acl-2022-5
|
['relation-classification']
|
['natural-language-processing']
|
[ 1.56225994e-01 4.11105812e-01 -8.23436975e-01 -7.45594680e-01
-5.34091055e-01 -4.64227438e-01 3.99997503e-01 1.01544201e+00
-3.28094661e-01 1.06389797e+00 2.23899856e-01 -5.04838228e-01
-7.07948387e-01 -8.39921236e-01 -5.01314960e-02 -3.22237343e-01
3.78472358e-03 4.25273597e-01 2.67936051e-01 -3.10493201e-01
2.84663439e-01 4.01988328e-01 -1.53906417e+00 3.77405196e-01
8.52346599e-01 9.79362071e-01 -6.92120850e-01 4.79289055e-01
-1.57242492e-01 7.81373203e-01 -9.30711925e-01 -9.21005607e-01
-7.08585456e-02 -3.07498693e-01 -1.05225265e+00 -7.75659800e-01
-1.22105859e-01 -1.39808925e-02 -1.25970706e-01 7.07653522e-01
3.91569018e-01 -1.32125840e-01 9.85077441e-01 -1.73123646e+00
-3.31446558e-01 7.02444077e-01 -5.06488919e-01 2.22575203e-01
8.92626584e-01 -3.28227401e-01 1.21917260e+00 -4.38254803e-01
6.66506767e-01 1.02203035e+00 9.71328020e-01 1.81658462e-01
-1.40626848e+00 -3.29501837e-01 7.17782974e-02 2.53370017e-01
-1.33934796e+00 -2.05345199e-01 3.95976335e-01 -4.96609300e-01
1.17718935e+00 6.46656573e-01 2.39228085e-01 4.30400848e-01
4.47438568e-01 5.06721556e-01 9.89198506e-01 -6.34233296e-01
4.33055758e-02 1.13036126e-01 8.13574553e-01 3.15275609e-01
8.99266183e-01 -2.45038848e-02 -5.55968225e-01 -6.83710337e-01
1.07398331e-01 -1.45841956e-01 4.30598296e-03 -1.85830742e-01
-1.04810798e+00 7.87649751e-01 3.88756506e-02 4.95466352e-01
-1.70947403e-01 -5.61873853e-01 4.92545128e-01 6.71453238e-01
6.72977567e-01 5.17790258e-01 -8.55958760e-01 -2.37760410e-01
-6.63295448e-01 3.95632625e-01 8.48649681e-01 7.98127234e-01
6.27685547e-01 -7.48133302e-01 -3.84615183e-01 9.99387443e-01
1.94156706e-01 -1.67314127e-01 3.14822644e-01 -6.47572815e-01
5.27808964e-01 9.77141619e-01 1.98850304e-01 -9.17050123e-01
-9.00049090e-01 -1.80448275e-02 -7.21350610e-01 -8.04919302e-02
3.96865934e-01 -3.89856920e-02 -6.22830272e-01 1.78918850e+00
3.71128291e-01 -3.81547481e-01 3.25934410e-01 3.48239928e-01
1.14809954e+00 -7.57952500e-03 3.81433308e-01 -6.51166081e-01
1.23853838e+00 -6.89947367e-01 -1.00880075e+00 2.56482065e-01
1.15544820e+00 -9.64912236e-01 7.10202456e-01 3.06230903e-01
-1.09561622e+00 -3.48021746e-01 -1.08000052e+00 -4.37387312e-03
-7.46179283e-01 -9.86821651e-02 9.10582781e-01 1.14985764e+00
-7.05875754e-01 7.65614271e-01 -5.36541939e-01 -3.70675892e-01
1.99903384e-01 3.47249717e-01 -4.67556030e-01 1.98719159e-01
-1.31481624e+00 1.19038165e+00 5.92746615e-01 -4.02931988e-01
3.98142189e-02 -7.27788150e-01 -6.80556059e-01 -2.19228834e-01
3.44325006e-02 -3.70067984e-01 9.91320670e-01 2.64348388e-02
-1.09834182e+00 1.00997818e+00 -1.87982962e-01 -3.58867884e-01
3.54029000e-01 4.21856577e-03 -8.24144244e-01 -2.14641869e-01
-1.33672044e-01 6.01325855e-02 -1.79470465e-01 -1.13128400e+00
-8.11711431e-01 -4.68061030e-01 4.81539339e-01 -9.56425071e-03
-1.40993595e-01 4.89924580e-01 2.50841558e-01 -5.37183166e-01
3.02357763e-01 -5.37820637e-01 -1.66365936e-01 -4.34335351e-01
-4.74295020e-01 -6.76186442e-01 4.10236746e-01 -3.03058594e-01
1.94949198e+00 -1.63416970e+00 -1.87786847e-01 5.45803189e-01
3.27831179e-01 3.65022391e-01 1.89373001e-01 4.34658259e-01
-4.47235316e-01 4.13454294e-01 -2.28111356e-01 5.00055067e-02
1.55880265e-02 2.26019338e-01 2.57871807e-01 2.92925745e-01
2.23615602e-01 7.70464957e-01 -9.02108371e-01 -5.36633968e-01
-3.86053789e-03 3.76276881e-01 -1.99561670e-01 1.63065776e-01
2.25662053e-01 -2.02921942e-01 -1.68627143e-01 9.21877503e-01
7.68335998e-01 -1.28742278e-01 4.29122269e-01 -3.44538450e-01
3.43818255e-02 4.92454141e-01 -1.30747604e+00 1.03733015e+00
-2.29798481e-01 2.49324858e-01 -5.74073493e-01 -1.01536083e+00
1.32554698e+00 4.74962205e-01 5.58821678e-01 -4.57169265e-01
1.19101800e-01 3.78704041e-01 3.13927203e-01 -4.03113544e-01
5.49168825e-01 -7.13228658e-02 -3.15872319e-02 6.47325039e-01
-2.47201473e-02 1.78108037e-01 5.44057667e-01 5.39324246e-02
1.36306989e+00 -2.32040882e-01 1.10753250e+00 -2.12476149e-01
4.98469055e-01 -2.35840142e-01 4.84259456e-01 5.95977068e-01
-4.49680597e-01 6.62091196e-01 1.11923730e+00 -4.35033739e-01
-8.67220163e-01 -8.78850520e-01 -7.00840771e-01 9.58016217e-01
7.80163854e-02 -1.02042985e+00 -4.47001845e-01 -1.04326272e+00
2.68042177e-01 5.44386029e-01 -7.28489220e-01 -4.78641689e-01
-3.52408946e-01 -1.37536454e+00 7.25573361e-01 5.71765780e-01
1.45175695e-01 -7.52211988e-01 -7.14623809e-01 1.77866116e-01
-2.81972319e-01 -7.89663911e-01 3.06661259e-02 4.39453125e-01
-9.37950671e-01 -1.21484387e+00 -2.45449752e-01 -3.23030591e-01
1.72231808e-01 1.19932808e-01 1.64875126e+00 4.16639507e-01
-4.98984680e-02 -2.83328593e-01 -6.93892717e-01 -4.66034442e-01
-9.55849737e-02 2.95039088e-01 1.59468893e-02 -3.13771546e-01
7.74965703e-01 -5.02211809e-01 -5.63404441e-01 5.26287735e-01
-6.61002576e-01 -5.69995642e-01 3.95344973e-01 6.99967027e-01
4.30485666e-01 -1.59318954e-01 6.65866911e-01 -1.42237771e+00
1.12536967e+00 -6.77748322e-01 1.47151276e-01 6.56417310e-01
-1.19801652e+00 -2.58314461e-01 1.24770924e-01 -3.51572931e-01
-7.74076581e-01 -4.13916945e-01 -2.37183735e-01 3.32110673e-01
-1.46732971e-01 6.18584812e-01 -2.29214609e-01 -8.79558921e-02
8.67372334e-01 -5.89582384e-01 -3.29529762e-01 -3.69558960e-01
1.21049076e-01 8.39964390e-01 1.17804721e-01 -5.77596068e-01
2.59456068e-01 5.45578226e-02 2.90472936e-02 -3.42986166e-01
-8.33719313e-01 -5.11649430e-01 -7.84834087e-01 -6.14950992e-02
3.52580756e-01 -4.08104330e-01 -7.88743794e-01 2.50143200e-01
-1.13721943e+00 6.97777495e-02 -3.49933684e-01 5.18868387e-01
-3.75533700e-01 1.57808945e-01 -4.86513108e-01 -7.90219426e-01
-2.11031795e-01 -7.78841496e-01 7.05787897e-01 3.62303823e-01
-8.48606646e-01 -1.08886206e+00 2.40862906e-01 4.05798495e-01
4.43749815e-01 7.54461884e-01 9.43540871e-01 -8.98771226e-01
4.40500200e-01 -4.36765432e-01 -4.43366647e-01 7.65879378e-02
3.84249926e-01 3.22759062e-01 -9.50977027e-01 5.83762536e-03
-3.42724413e-01 -2.91037381e-01 6.88122272e-01 2.25881636e-01
1.17580044e+00 -5.35050081e-03 -6.92408204e-01 3.18178266e-01
1.32851517e+00 5.54747224e-01 1.03081572e+00 3.92144531e-01
3.83058965e-01 7.66744375e-01 1.04131091e+00 3.54247242e-01
7.68148005e-01 6.28457844e-01 -1.75307672e-02 -8.23867694e-02
1.31277606e-01 1.86663106e-01 -2.68407553e-01 6.87492430e-01
-6.54482365e-01 -2.25789830e-01 -1.03095174e+00 4.14838821e-01
-1.90502918e+00 -7.22551823e-01 -2.87988156e-01 2.15803432e+00
9.71518636e-01 6.40720904e-01 2.84976065e-01 9.37121451e-01
6.05376065e-01 1.53406695e-01 -1.28146365e-01 -9.13075209e-01
-2.43890464e-01 5.32617867e-01 3.37113142e-01 6.55242860e-01
-1.06368423e+00 5.40122271e-01 8.11000156e+00 7.92560697e-01
-5.09576440e-01 1.51728213e-01 7.45782554e-01 1.77862182e-01
-3.27907741e-01 6.53922632e-02 -6.16733670e-01 3.78006101e-01
1.16908610e+00 -3.48298997e-01 -2.45581403e-01 4.17244911e-01
-2.62061357e-01 -3.34616542e-01 -1.20813143e+00 6.41723394e-01
2.57915322e-04 -9.61172402e-01 -2.38087308e-02 1.38447672e-01
6.24530554e-01 -5.09384573e-01 -3.42971981e-01 1.65690631e-01
4.35402334e-01 -1.09442425e+00 2.08022773e-01 5.47400653e-01
8.75572205e-01 -8.14615309e-01 1.36597466e+00 -2.60665417e-01
-1.04658401e+00 3.13218325e-01 -3.12633246e-01 -3.39303374e-01
-3.33496444e-02 1.01910210e+00 -3.80537093e-01 1.11543512e+00
5.15428960e-01 4.99071449e-01 -6.20433807e-01 9.28154945e-01
4.36541699e-02 3.93164992e-01 -2.35856920e-01 -6.36309534e-02
-2.75702387e-01 6.49937317e-02 4.34343964e-02 1.37486768e+00
1.20923869e-01 4.11988109e-01 -1.25620246e-01 4.70225960e-02
2.16511592e-01 3.96817416e-01 -6.02976084e-01 2.46467292e-01
6.45922005e-01 1.23753071e+00 -8.61119330e-01 -2.46984154e-01
-4.72876638e-01 1.11891232e-01 5.08080184e-01 -3.72680509e-03
-7.77049005e-01 -7.71251023e-01 6.45820022e-01 -1.01998970e-01
-2.93673605e-01 1.47927582e-01 -8.40281606e-01 -8.80441964e-01
1.52129292e-01 -8.84403944e-01 9.51157153e-01 -3.25285882e-01
-1.18543625e+00 6.34326458e-01 3.29026878e-01 -1.24018693e+00
1.28137887e-01 -5.45660496e-01 -3.77155781e-01 7.50356853e-01
-1.38191187e+00 -8.50601554e-01 -4.57818717e-01 2.04496056e-01
-2.41449207e-01 -2.35295162e-01 9.81763959e-01 5.29643476e-01
-5.79740107e-01 1.19349992e+00 -1.14878267e-01 -1.04971416e-01
8.27627659e-01 -1.40157485e+00 6.31459877e-02 1.42304108e-01
-8.32414553e-02 6.34447575e-01 8.34322631e-01 -5.81174433e-01
-4.77679551e-01 -5.93067527e-01 1.52091873e+00 -5.34927964e-01
4.33133632e-01 8.94341171e-02 -8.65518272e-01 5.06073594e-01
-1.94663510e-01 -1.24944240e-01 1.23155761e+00 8.34477782e-01
-5.33910871e-01 -3.14557523e-01 -1.70813465e+00 1.71233192e-01
1.17320871e+00 -1.93976909e-01 -4.98505682e-01 3.20897192e-01
5.49405038e-01 -3.25041592e-01 -1.71797442e+00 9.79514837e-01
1.10028183e+00 -1.12161267e+00 1.06258535e+00 -1.05641687e+00
3.83195162e-01 -1.53926596e-01 -3.57944995e-01 -1.29411006e+00
-3.60304236e-01 -3.40907693e-01 -4.24804032e-01 1.33599555e+00
9.83043194e-01 -9.60748374e-01 6.77697659e-01 7.43091941e-01
2.91715682e-01 -1.19268501e+00 -9.36718702e-01 -6.98498487e-01
2.18745157e-01 -2.74237394e-01 1.20701921e+00 1.23062348e+00
5.01580000e-01 3.68955374e-01 -9.09318402e-02 -2.06857190e-01
4.78036076e-01 3.42454091e-02 5.76189101e-01 -1.64823079e+00
5.81354462e-02 -5.52817881e-01 -8.63883018e-01 -1.36463404e-01
-1.83125705e-01 -5.84265709e-01 -5.40227890e-01 -1.41938162e+00
3.29933286e-01 -7.75284529e-01 -8.51252615e-01 5.38639128e-01
-5.63047230e-01 6.11520529e-01 -1.60948813e-01 2.20142111e-01
-5.46231806e-01 -5.61957732e-02 8.83289814e-01 -7.53874108e-02
-1.51402608e-01 3.69697995e-02 -8.89085948e-01 5.88084280e-01
1.24791908e+00 -6.50671124e-01 -3.92333269e-01 1.33956494e-02
4.54156071e-01 -1.45002604e-01 -3.33979547e-01 -8.45112562e-01
-1.33655325e-01 -3.94339234e-01 3.72658849e-01 -4.90704298e-01
1.05710328e-02 -3.53422731e-01 1.74095929e-01 4.45535302e-01
-5.76080978e-01 1.93505809e-01 -1.74671873e-01 1.81654513e-01
-4.27437782e-01 -4.48624611e-01 5.43014467e-01 1.66575342e-01
-1.05974458e-01 -8.79901871e-02 2.18894362e-01 5.52245937e-02
1.13392770e+00 -2.28169575e-01 -7.01330364e-01 -8.36260095e-02
-9.45043504e-01 3.22162688e-01 2.21189305e-01 4.13212091e-01
3.45993161e-01 -1.45149934e+00 -8.28041136e-01 -1.90035746e-01
6.43697560e-01 -5.54714680e-01 -4.22495604e-02 1.01402819e+00
-4.44686204e-01 4.62735474e-01 -2.31066838e-01 -3.03809166e-01
-1.72559071e+00 4.63462561e-01 1.19120672e-01 -8.36513579e-01
-9.88032017e-03 5.21731794e-01 -2.53124118e-01 -6.02057099e-01
1.66599900e-01 -4.28363413e-01 -7.72712290e-01 4.87165600e-01
6.05100632e-01 7.34272301e-01 5.90254545e-01 -6.26020014e-01
-7.71489859e-01 6.80568993e-01 -2.09803030e-01 2.41125360e-01
1.11216807e+00 1.84555296e-02 -3.81761670e-01 6.24418795e-01
9.72148418e-01 -1.80394769e-01 -2.26394266e-01 -1.87589973e-01
4.66711581e-01 -6.07536554e-01 -3.59428674e-01 -9.96112287e-01
-7.11929679e-01 3.93132418e-01 5.05691290e-01 8.94634306e-01
1.19914103e+00 8.41076300e-02 4.80302572e-01 1.84948057e-01
3.75815064e-01 -1.00098670e+00 -4.07479286e-01 4.55417335e-01
5.69869757e-01 -1.34233189e+00 3.91825378e-01 -8.91574919e-01
-5.45091927e-01 1.01131546e+00 6.13809049e-01 2.37713203e-01
9.89472389e-01 4.68438298e-01 1.74923956e-01 -1.69683605e-01
-8.70961308e-01 -3.41686070e-01 3.29106450e-01 7.61914372e-01
1.20483947e+00 5.21798611e-01 -1.39771998e+00 7.72580624e-01
-4.98564392e-01 1.03351509e-03 2.76987255e-01 1.14399922e+00
-2.32326210e-01 -1.46098757e+00 -2.79765040e-01 1.05582881e+00
-7.08751738e-01 2.92764544e-01 -5.74112177e-01 6.99396789e-01
4.41777438e-01 1.34604895e+00 -1.13676161e-01 -9.56017554e-01
7.74338126e-01 1.63598835e-01 5.51668346e-01 -3.34628850e-01
-7.44410634e-01 -5.80285251e-01 6.35392368e-01 -2.80190110e-01
-8.82755220e-01 -7.69318938e-01 -8.11873436e-01 -6.30745113e-01
-5.93343079e-01 2.88024336e-01 9.42169800e-02 8.59942317e-01
8.18322003e-02 5.16294777e-01 6.21305227e-01 -1.62027001e-01
-1.11009702e-01 -1.32011926e+00 -5.03542185e-01 4.43950057e-01
2.56403625e-01 -1.02613056e+00 -1.45769909e-01 -3.02593648e-01]
|
[9.258954048156738, 8.674219131469727]
|
4fa5bea0-8751-4596-b8a4-a72a7da98320
|
curriculum-learning-for-relative
|
2212.02733
| null |
https://arxiv.org/abs/2212.02733v2
|
https://arxiv.org/pdf/2212.02733v2.pdf
|
Curriculum Learning for Relative Overgeneralization
|
In multi-agent reinforcement learning (MARL), many popular methods, such as VDN and QMIX, are susceptible to a critical multi-agent pathology known as relative overgeneralization (RO), which arises when the optimal joint action's utility falls below that of a sub-optimal joint action in cooperative tasks. RO can cause the agents to get stuck into local optima or fail to solve cooperative tasks that require significant coordination between agents within a given timestep. Recent value-based MARL algorithms such as QPLEX and WQMIX can overcome RO to some extent. However, our experimental results show that they can still fail to solve cooperative tasks that exhibit strong RO. In this work, we propose a novel approach called curriculum learning for relative overgeneralization (CURO) to better overcome RO. To solve a target task that exhibits strong RO, in CURO, we first fine-tune the reward function of the target task to generate source tasks that are tailored to the current ability of the learning agent and train the agent on these source tasks first. Then, to effectively transfer the knowledge acquired in one task to the next, we use a transfer learning method that combines value function transfer with buffer transfer, which enables more efficient exploration in the target task. We demonstrate that, when applied to QMIX, CURO overcomes severe RO problem and significantly improves performance, yielding state-of-the-art results in a variety of cooperative multi-agent tasks, including the challenging StarCraft II micromanagement benchmarks.
|
['Bei Peng', 'Lin Shi']
|
2022-12-06
| null | null | null | null |
['starcraft-ii', 'starcraft']
|
['playing-games', 'playing-games']
|
[-5.21214753e-02 -3.45735475e-02 -1.70140743e-01 3.11800539e-01
-9.65371788e-01 -4.16954637e-01 4.57738072e-01 1.37752846e-01
-6.93757713e-01 1.30996180e+00 -1.02048337e-01 -1.68354642e-02
-4.64722663e-01 -6.81908011e-01 -8.59430730e-01 -9.68903005e-01
-2.83233136e-01 7.19674706e-01 4.60376948e-01 -6.60057127e-01
2.93669462e-01 5.94821237e-02 -1.20594084e+00 9.81473252e-02
1.44691098e+00 6.27642214e-01 5.72866619e-01 4.85442758e-01
2.41193742e-01 9.48767900e-01 -1.03811789e+00 1.99933723e-01
1.56880677e-01 -4.14215773e-01 -9.63194013e-01 1.00280978e-02
-1.41057372e-01 -1.29468367e-01 -7.14731738e-02 1.01114702e+00
5.99686444e-01 4.70938653e-01 6.29216135e-01 -1.56402922e+00
-3.42664897e-01 7.44345486e-01 -7.80712724e-01 9.02087539e-02
1.15192115e-01 4.81498718e-01 9.27231848e-01 -4.92006153e-01
5.80463827e-01 1.38289857e+00 3.70529950e-01 6.88300431e-01
-1.26222527e+00 -6.98330998e-01 3.03114593e-01 -1.52152032e-02
-1.05491257e+00 8.24285746e-02 6.45457387e-01 -2.82704353e-01
1.04642487e+00 -1.89346537e-01 5.40884137e-01 1.10222173e+00
6.41345084e-01 9.59723473e-01 1.29841852e+00 -6.52040169e-02
6.51246488e-01 -3.09862494e-01 -3.55066866e-01 6.93335533e-01
1.19059384e-01 3.16356301e-01 -6.64651215e-01 -2.56364554e-01
8.16512465e-01 -3.56585920e-01 -1.00457162e-01 -6.94940269e-01
-1.36912000e+00 1.04843104e+00 6.77752793e-01 1.40302822e-01
-7.50002861e-01 4.67978984e-01 4.56453383e-01 6.77397907e-01
1.65692598e-01 1.19595706e+00 -5.68831623e-01 -2.36604303e-01
-3.92353028e-01 6.99862778e-01 6.42607450e-01 5.16945720e-01
9.32639837e-01 1.41871259e-01 -3.14961582e-01 6.81891561e-01
2.04043780e-02 3.16293091e-01 6.67483807e-01 -1.24402070e+00
8.05442393e-01 5.70211351e-01 3.60386431e-01 -5.10538340e-01
-6.07987583e-01 -5.05265355e-01 -6.23503268e-01 7.42189586e-01
4.12591785e-01 -6.39966786e-01 -8.04433167e-01 1.86213768e+00
4.11339849e-01 1.10917196e-01 7.27553844e-01 1.14486980e+00
3.89593691e-01 6.78266704e-01 1.42941609e-01 -2.56567866e-01
9.65585053e-01 -1.26541257e+00 -4.32195514e-01 -6.10566080e-01
8.62961888e-01 -3.31710935e-01 1.05754375e+00 5.33274889e-01
-1.11369979e+00 -3.89930993e-01 -1.11728036e+00 6.32989883e-01
-2.85400581e-02 -4.22926724e-01 5.29677927e-01 1.10652730e-01
-8.74537110e-01 8.15368235e-01 -7.99529850e-01 -2.91574784e-02
4.99895215e-01 5.81013083e-01 -7.85892755e-02 -3.45955347e-03
-1.33433545e+00 9.99529421e-01 7.09430158e-01 -2.82432795e-01
-1.59314036e+00 -8.38071465e-01 -5.64032674e-01 1.86599549e-02
1.06013048e+00 -5.57512522e-01 1.37469542e+00 -1.22814977e+00
-1.84551311e+00 2.06505090e-01 5.62999606e-01 -5.33560932e-01
4.81129259e-01 -3.48327279e-01 1.31921738e-01 9.41390451e-03
1.74022794e-01 8.77622366e-01 1.01559293e+00 -1.43877065e+00
-7.99575448e-01 -6.96375966e-02 4.24651861e-01 7.58423328e-01
-3.23375642e-01 -2.90960729e-01 2.33893357e-02 -5.63096046e-01
-4.12651986e-01 -1.14918649e+00 -6.01405144e-01 -5.62784433e-01
-1.06053934e-01 -5.91836989e-01 6.01476550e-01 -4.27310392e-02
8.80095065e-01 -1.84383404e+00 1.03714907e+00 -2.44948640e-02
3.02346051e-01 3.42890292e-01 -6.18201971e-01 5.51521957e-01
2.97485054e-01 -1.70384482e-01 -2.23213688e-01 -2.20654488e-01
-4.87201177e-02 5.91650069e-01 -1.10199526e-01 2.14990363e-01
2.23812923e-01 9.44149017e-01 -1.27135348e+00 -3.16328883e-01
-1.52563721e-01 -3.90017591e-02 -7.68246293e-01 4.55516100e-01
-7.78803647e-01 6.51050448e-01 -8.32540691e-01 3.22531879e-01
4.07106817e-01 -2.38812134e-01 2.68258393e-01 3.78892750e-01
-3.50206904e-02 -3.28433327e-02 -1.13082612e+00 1.76591742e+00
-6.27500296e-01 1.81988910e-01 1.98417321e-01 -1.07763171e+00
9.22218680e-01 1.56117231e-01 5.71692407e-01 -7.78834760e-01
-3.99430506e-02 2.58255124e-01 2.56011754e-01 -2.36020416e-01
5.33722162e-01 6.52558878e-02 -2.64360934e-01 4.10107166e-01
-2.62497496e-02 -4.32945251e-01 2.92435348e-01 1.75455153e-01
1.35426855e+00 1.96861207e-01 2.57034808e-01 -2.87349492e-01
4.49303597e-01 2.61180162e-01 8.23682845e-01 9.14393663e-01
-2.04455763e-01 1.17210686e-01 8.37664604e-01 -3.82613450e-01
-8.72846067e-01 -1.01222754e+00 4.64840800e-01 1.48653424e+00
3.27834010e-01 -2.86982387e-01 -7.09195554e-01 -1.03823102e+00
2.79266059e-01 5.49808502e-01 -6.17902935e-01 -4.38218236e-01
-7.99765229e-01 -9.66669500e-01 2.77656645e-01 4.53335285e-01
6.55874312e-01 -1.61330545e+00 -1.13320434e+00 5.56977630e-01
-3.82603593e-02 -8.45173299e-01 -5.41269541e-01 4.66411263e-01
-7.74612725e-01 -9.96460140e-01 -9.18362617e-01 -6.29092872e-01
5.95548570e-01 1.38336599e-01 1.07005250e+00 2.04674229e-01
-1.52254924e-02 1.63433522e-01 -5.16794205e-01 -1.72943130e-01
-5.80272853e-01 4.40124661e-01 1.94958448e-01 -3.28572899e-01
-2.76508540e-01 -3.65397066e-01 -4.92061496e-01 6.33027792e-01
-8.21772218e-01 -8.11179578e-02 8.44498277e-01 1.25211430e+00
3.01173091e-01 1.29426524e-01 1.05594194e+00 -5.03101170e-01
1.09421372e+00 -5.45343637e-01 -8.16221297e-01 1.39256194e-01
-6.25301421e-01 3.97923380e-01 8.57267499e-01 -8.93125474e-01
-9.52630341e-01 -9.82731953e-02 3.39232743e-01 -5.02299666e-01
2.85947382e-01 7.08049953e-01 2.05380589e-01 -1.19199596e-01
8.57394874e-01 1.35576114e-01 2.70240873e-01 -8.81666020e-02
1.27557904e-01 2.46710986e-01 2.06011698e-01 -9.72245336e-01
6.79195583e-01 6.59715571e-03 -1.22073451e-02 -3.81615967e-01
-7.61044919e-01 -9.81918424e-02 -1.41825303e-01 -2.15805888e-01
7.28025794e-01 -8.73402178e-01 -9.63430107e-01 5.47780216e-01
-8.84714842e-01 -1.15331149e+00 -2.99208254e-01 3.73067528e-01
-8.91718924e-01 7.85104111e-02 -3.61493349e-01 -5.02174795e-01
-1.43626988e-01 -1.46306705e+00 7.66989529e-01 5.35026848e-01
8.83910246e-03 -1.00633371e+00 3.71149570e-01 2.56858200e-01
4.83342528e-01 2.24214017e-01 8.84170055e-01 -4.37544256e-01
-5.14770627e-01 5.43720543e-01 7.58219436e-02 1.78913906e-01
-5.61643653e-02 -3.89582187e-01 -3.04743141e-01 -9.54257131e-01
-4.35445338e-01 -9.91894126e-01 8.25953901e-01 2.95581788e-01
6.56407833e-01 -2.47455284e-01 -3.63752544e-01 3.48794103e-01
1.26989377e+00 4.27477241e-01 4.62494403e-01 8.70659411e-01
3.74367386e-01 3.90357852e-01 1.19514310e+00 7.10298419e-01
3.65229547e-01 8.54842246e-01 8.61934423e-01 4.03334387e-02
1.58998385e-01 -3.49942013e-03 7.29437292e-01 3.49351734e-01
-1.14813857e-01 -1.92798510e-01 -8.37612212e-01 5.55483937e-01
-2.43078208e+00 -6.17280066e-01 3.84939879e-01 2.05267167e+00
1.10420144e+00 2.32595056e-01 3.34346265e-01 -3.17731410e-01
3.75798821e-01 1.63424045e-01 -9.67487037e-01 -4.80749279e-01
8.53052288e-02 1.87637303e-02 4.04381752e-01 4.66644675e-01
-9.30412948e-01 1.39876509e+00 5.77343321e+00 9.07549977e-01
-8.83499265e-01 2.45317996e-01 3.26199025e-01 -1.57040596e-01
-3.87546308e-02 -1.62950635e-01 -6.69868350e-01 3.46777052e-01
6.14473522e-01 -2.43341431e-01 8.68312716e-01 8.52636099e-01
8.94366130e-02 -3.66112381e-01 -9.80249643e-01 5.38149297e-01
-3.12272608e-01 -1.20157588e+00 -2.97112197e-01 8.55649039e-02
1.16111171e+00 1.43580332e-01 2.20764428e-01 8.60621512e-01
9.32064056e-01 -1.07486391e+00 4.69481975e-01 1.25783727e-01
2.88197637e-01 -1.07812357e+00 6.48393869e-01 5.78484833e-01
-9.75576282e-01 -4.77364928e-01 -5.10575831e-01 -8.15598294e-02
-7.30433315e-02 3.29171829e-02 -8.45007539e-01 4.90958035e-01
5.66962123e-01 3.35756332e-01 -3.22308481e-01 8.41517091e-01
-3.45478326e-01 1.86349854e-01 4.06055450e-02 -3.18395764e-01
7.97874570e-01 -9.99278799e-02 7.82575428e-01 5.60655892e-01
1.85636029e-01 -4.03544344e-02 6.41884506e-01 8.44121218e-01
1.31965633e-02 -2.45188922e-02 -4.23406869e-01 5.25500700e-02
4.45886970e-01 1.19648111e+00 -4.28938895e-01 -1.56046107e-01
-5.78779392e-02 9.68922973e-01 8.28130603e-01 4.02393371e-01
-9.71991479e-01 -2.51649678e-01 8.26642632e-01 -3.66434991e-01
3.46661955e-01 -2.44992360e-01 1.39715940e-01 -8.30003321e-01
-1.60959750e-01 -1.21818292e+00 4.10193145e-01 -5.40831149e-01
-1.01475084e+00 6.12615526e-01 -4.21655178e-02 -1.06901073e+00
-5.24128854e-01 -2.92479694e-01 -6.66728318e-01 5.10855019e-01
-1.61290741e+00 -8.07026803e-01 -2.44106099e-01 6.54044747e-01
6.85657799e-01 -7.05961823e-01 5.87632298e-01 -1.18029609e-01
-5.88142157e-01 4.11410838e-01 2.70521998e-01 -3.43881011e-01
7.78166473e-01 -1.50475478e+00 -2.34181527e-02 2.43838206e-01
-3.31057727e-01 4.57217917e-02 7.43202329e-01 -7.24198997e-01
-1.56872809e+00 -1.08153570e+00 5.00127487e-02 -2.72872206e-03
7.91961789e-01 -7.20699877e-02 -9.14719701e-01 4.83372658e-01
2.71908879e-01 -6.39088303e-02 -1.95713881e-02 1.24801785e-01
9.61356610e-02 -1.22036375e-01 -9.22747195e-01 6.82509720e-01
7.85347700e-01 4.04652767e-02 -6.62923872e-01 5.29564023e-01
8.08709621e-01 -5.68592250e-01 -9.13416147e-01 2.15916440e-01
6.57898784e-02 -8.33638132e-01 9.06185210e-01 -7.57604659e-01
5.47964633e-01 -1.86914980e-01 9.03508738e-02 -2.27076364e+00
-4.24031109e-01 -9.80334461e-01 -1.31514668e-01 8.37999523e-01
3.87345999e-01 -7.46205926e-01 8.59511316e-01 4.91559543e-02
-1.97881147e-01 -8.88826013e-01 -1.11605024e+00 -1.10604548e+00
5.71242094e-01 3.03434372e-01 4.61104840e-01 8.55172753e-01
1.79385006e-01 3.06850910e-01 -4.60071772e-01 9.21480954e-02
7.72540808e-01 6.46257252e-02 9.09359038e-01 -1.00809348e+00
-5.22217691e-01 -4.08130974e-01 2.13780507e-01 -8.62668753e-01
4.64967012e-01 -6.69524610e-01 2.88489014e-01 -1.32407999e+00
1.36845604e-01 -7.43495166e-01 -3.87802184e-01 6.47468388e-01
-4.10115361e-01 -1.23743884e-01 5.14428198e-01 3.08753461e-01
-1.01339865e+00 8.24918509e-01 1.72805965e+00 -1.78657830e-01
-7.18668759e-01 -8.08349624e-02 -6.32114947e-01 5.42743146e-01
9.10512626e-01 -6.43201947e-01 -3.92266601e-01 -3.21625680e-01
3.34022641e-01 6.29801869e-01 1.95475474e-01 -1.05748701e+00
1.72622174e-01 -5.89640141e-01 4.40893322e-02 -7.37984180e-02
3.69557977e-01 -5.01062393e-01 -2.06330851e-01 9.93191838e-01
-3.58993918e-01 2.18145728e-01 1.99909776e-01 6.01620376e-01
-1.31798357e-01 -3.04367572e-01 8.34608495e-01 -1.06375150e-01
-7.66699135e-01 1.10463887e-01 -6.73538923e-01 3.94866168e-01
1.50395060e+00 2.80341178e-01 -6.04175031e-01 -3.46310824e-01
-6.19999886e-01 1.11322272e+00 3.33426386e-01 4.52042013e-01
6.37749493e-01 -1.27380240e+00 -8.25520813e-01 -1.60578370e-01
-6.64112493e-02 3.64991017e-02 1.79089144e-01 8.10491562e-01
-2.01534227e-01 1.93314105e-01 -6.11338258e-01 -5.14190435e-01
-1.08039856e+00 5.98393083e-01 6.07308686e-01 -8.89327228e-01
-5.38759887e-01 6.16517246e-01 3.70180488e-01 -4.80126351e-01
1.80637166e-01 -2.28822574e-01 -2.52667814e-01 3.35385799e-02
2.28532553e-01 4.58107293e-01 -2.94974595e-01 -1.08058803e-01
-3.83621514e-01 3.93830210e-01 -3.46074909e-01 -1.84363961e-01
1.32094860e+00 1.96744502e-01 2.38508508e-01 -8.24631527e-02
7.75506914e-01 -5.06346583e-01 -1.97730911e+00 -3.25186342e-01
-2.11281218e-02 -2.16385558e-01 8.59109536e-02 -9.59230900e-01
-1.07401049e+00 4.47502881e-01 2.67883748e-01 1.49348825e-01
9.75898206e-01 -2.00579651e-02 6.80636764e-01 7.02404916e-01
5.01273990e-01 -1.49790370e+00 1.00045753e+00 7.95118749e-01
1.12265110e+00 -1.29172969e+00 -3.03160846e-02 1.71980634e-02
-1.25960517e+00 9.65348959e-01 1.13754380e+00 -3.73315811e-01
3.03795226e-02 2.03488424e-01 -5.43471724e-02 -1.79936215e-01
-1.18069136e+00 -1.95050582e-01 -9.79494974e-02 6.64059043e-01
-2.01621443e-01 4.92997952e-02 -1.33431599e-01 3.42791855e-01
1.48656562e-01 -8.66498202e-02 6.10472322e-01 1.05788147e+00
-7.00644195e-01 -1.12705278e+00 -4.98661369e-01 1.78939357e-01
-1.67954937e-01 2.74272949e-01 -1.73438609e-01 8.94021690e-01
-1.38249263e-01 7.79377282e-01 7.52590448e-02 -2.22812533e-01
3.35180163e-01 -3.53822172e-01 6.12345219e-01 -4.77037579e-01
-9.87044215e-01 7.55905733e-02 4.19381224e-02 -6.99208558e-01
-2.56535888e-01 -5.77716291e-01 -1.66531277e+00 -1.61666602e-01
-4.47456315e-02 3.00324261e-01 2.37958372e-01 9.73427653e-01
2.72228867e-01 9.71830130e-01 7.56868720e-01 -1.00078928e+00
-1.11268640e+00 -8.23098540e-01 -4.87340808e-01 2.80151963e-01
4.07423139e-01 -1.10629034e+00 -1.53027251e-01 -6.12046957e-01]
|
[3.7998526096343994, 1.9158456325531006]
|
93a96fe9-f580-4e60-8812-67bc94348c58
|
pai-at-semeval-2023-task-2-a-universal-system
|
2305.06099
| null |
https://arxiv.org/abs/2305.06099v1
|
https://arxiv.org/pdf/2305.06099v1.pdf
|
PAI at SemEval-2023 Task 2: A Universal System for Named Entity Recognition with External Entity Information
|
The MultiCoNER II task aims to detect complex, ambiguous, and fine-grained named entities in low-context situations and noisy scenarios like the presence of spelling mistakes and typos for multiple languages. The task poses significant challenges due to the scarcity of contextual information, the high granularity of the entities(up to 33 classes), and the interference of noisy data. To address these issues, our team {\bf PAI} proposes a universal Named Entity Recognition (NER) system that integrates external entity information to improve performance. Specifically, our system retrieves entities with properties from the knowledge base (i.e. Wikipedia) for a given text, then concatenates entity information with the input sentence and feeds it into Transformer-based models. Finally, our system wins 2 first places, 4 second places, and 1 third place out of 13 tracks. The code is publicly available at \url{https://github.com/diqiuzhuanzhuan/semeval-2023}.
|
['Xuan Li', 'Weiguo Gao', 'Hailong Huang', 'Tianbo Che', 'Kai Lu', 'Long Ma']
|
2023-05-10
| null | null | null | null |
['named-entity-recognition-ner']
|
['natural-language-processing']
|
[-3.33115280e-01 -6.38777688e-02 3.37436087e-02 -3.26572388e-01
-1.13132989e+00 -9.54958081e-01 4.53080684e-01 3.05815816e-01
-7.36316741e-01 1.00134850e+00 4.90674555e-01 -3.13703120e-01
6.27475381e-02 -8.81350815e-01 -6.41825378e-01 -1.39177993e-01
2.99356639e-01 5.15915453e-01 3.50841790e-01 -3.87964427e-01
9.36060473e-02 3.46319169e-01 -1.31657851e+00 4.47029471e-01
1.17131734e+00 7.42115378e-01 3.25150073e-01 3.96652669e-01
-4.45805013e-01 7.55730271e-01 -6.44419551e-01 -8.57498586e-01
1.31383911e-02 5.63130295e-03 -9.78542507e-01 -6.53023899e-01
2.69346148e-01 5.56984432e-02 -3.24087441e-01 1.14623916e+00
6.41808927e-01 2.26110652e-01 4.16091204e-01 -1.05101025e+00
-6.13129973e-01 9.24004495e-01 -1.11794725e-01 3.50019246e-01
2.60567218e-01 -1.24851502e-01 1.24466443e+00 -1.30030453e+00
8.67443264e-01 9.66526866e-01 7.64986038e-01 4.33885992e-01
-7.55530417e-01 -8.24637115e-01 1.29071549e-01 1.40301228e-01
-1.70598066e+00 -6.10545814e-01 6.42309012e-03 -2.84923047e-01
1.16192162e+00 3.34224820e-01 1.66522786e-01 1.19145620e+00
-4.01319295e-01 8.59564245e-01 9.18698609e-01 -3.90045822e-01
1.36841297e-01 1.56619787e-01 3.92908484e-01 4.27039176e-01
6.65321946e-01 -1.87397882e-01 -5.69077671e-01 -1.95775539e-01
3.27319235e-01 -1.26114234e-01 -2.04917207e-01 2.71059275e-01
-1.41304684e+00 2.83931136e-01 2.71977097e-01 6.63180351e-01
-4.76615429e-01 -3.08975190e-01 3.04791778e-01 1.06704533e-01
2.71027744e-01 8.00090313e-01 -9.84915257e-01 -4.19505030e-01
-9.72280681e-01 1.88054115e-01 1.01914704e+00 1.32240725e+00
7.40734041e-01 -2.75149465e-01 -1.82546884e-01 1.00406897e+00
1.62834167e-01 7.50299633e-01 3.19186807e-01 -4.36368853e-01
9.50750172e-01 7.40467429e-01 3.96193504e-01 -7.82598734e-01
-5.80239713e-01 -4.06226516e-01 -5.18302798e-01 -4.95508105e-01
5.32106400e-01 -4.56179738e-01 -7.86114633e-01 1.62559617e+00
4.41372305e-01 -1.60577476e-01 3.74800600e-02 7.41741717e-01
1.20081413e+00 4.83520567e-01 4.08537060e-01 1.72544375e-01
1.61045039e+00 -7.80068040e-01 -6.63596749e-01 -3.50898057e-01
7.51699924e-01 -8.90555739e-01 7.64519095e-01 -3.74189727e-02
-7.56572068e-01 -1.84220791e-01 -6.37936175e-01 -2.85501301e-01
-7.55607665e-01 5.26683271e-01 3.12979877e-01 3.55034500e-01
-7.12704182e-01 3.55372161e-01 -6.86616600e-01 -4.62381184e-01
1.84369702e-02 6.74941167e-02 -5.42975962e-01 -9.65130478e-02
-1.64313614e+00 9.74682331e-01 6.41077936e-01 2.23779812e-01
-3.29242319e-01 -7.62044251e-01 -9.50219691e-01 6.58492744e-02
5.98621607e-01 -4.45615381e-01 1.13750613e+00 -4.82353926e-01
-8.90964508e-01 7.74532974e-01 -2.50989079e-01 -6.02595620e-02
4.29799616e-01 -4.53712791e-01 -8.70826125e-01 -2.46198311e-01
5.88536799e-01 9.87905785e-02 -5.76501936e-02 -7.74844587e-01
-9.16867733e-01 -4.02626008e-01 1.02894433e-01 1.86549351e-01
-2.30140939e-01 3.98037910e-01 -5.32725990e-01 -7.81491816e-01
-3.80333103e-02 -8.06244314e-01 -2.61490673e-01 -8.73014927e-01
-7.55087018e-01 -3.20605725e-01 4.51266356e-02 -9.37438428e-01
1.62017930e+00 -2.19113517e+00 -2.36967459e-01 1.70807466e-01
9.43232104e-02 3.07648361e-01 -1.67182162e-02 7.98369408e-01
-1.26499310e-03 4.67146367e-01 8.45055096e-03 -6.01999573e-02
2.12903515e-01 -1.78099960e-01 -2.67040461e-01 1.42840771e-02
2.47116923e-01 9.59800363e-01 -9.98945892e-01 -4.65013891e-01
-2.29404539e-01 3.14914078e-01 -1.18235007e-01 7.95290843e-02
-6.82071075e-02 2.79428959e-01 -6.23392761e-01 8.46835256e-01
5.33659995e-01 -2.42745072e-01 2.77934730e-01 -2.38039061e-01
-5.65643489e-01 8.56864810e-01 -1.73906898e+00 1.40614200e+00
-4.42353159e-01 2.43105277e-01 -1.19669758e-01 -3.57303351e-01
7.08187342e-01 4.08895165e-01 9.32494476e-02 -6.79001808e-01
4.82291989e-02 7.17474401e-01 -3.76191854e-01 -4.79853362e-01
8.75856996e-01 1.28867492e-01 -4.98558491e-01 1.24106660e-01
2.92333663e-01 3.16423774e-01 5.39179623e-01 2.74790615e-01
1.34179211e+00 -1.13800980e-01 5.37683606e-01 -1.41793117e-01
3.65511179e-01 1.08814023e-01 1.03951418e+00 8.41638446e-01
-1.81550756e-01 4.51149106e-01 2.40256101e-01 -2.58124590e-01
-1.10605145e+00 -8.52460384e-01 -9.63512734e-02 1.30280435e+00
9.19756815e-02 -7.15151727e-01 -3.85375619e-01 -7.49777019e-01
2.51633804e-02 9.02926922e-01 -3.86140972e-01 2.32991412e-01
-5.80136120e-01 -4.28936183e-01 1.03502548e+00 5.89840531e-01
3.56380552e-01 -1.14716983e+00 -3.17718871e-02 3.08390409e-01
-7.45755732e-01 -1.26543081e+00 -4.84855264e-01 1.30794302e-01
-2.61089712e-01 -1.01742244e+00 -5.27851582e-01 -7.53750324e-01
3.94399613e-01 -9.93067175e-02 1.31824338e+00 -2.97238608e-03
-9.27318707e-02 2.92333215e-02 -4.85995054e-01 -3.88293386e-01
-2.10850880e-01 5.06047070e-01 3.44221145e-02 -2.23659381e-01
7.05762088e-01 -3.10030967e-01 -4.05422807e-01 3.36422414e-01
-7.11425006e-01 -1.93786040e-01 5.61421514e-01 7.72922933e-01
5.43654680e-01 -6.83924109e-02 6.22996867e-01 -1.07301450e+00
3.53512108e-01 -7.52781630e-01 -5.48132122e-01 5.44192970e-01
-4.16283041e-01 -2.96716369e-03 5.91157973e-01 -9.47432071e-02
-1.31027114e+00 -5.37108555e-02 -4.40737337e-01 1.43359095e-01
-4.97469008e-01 5.77969134e-01 -4.70442563e-01 4.30537969e-01
7.68538058e-01 7.84242004e-02 -1.00963402e+00 -7.76010990e-01
3.96903962e-01 9.79724050e-01 5.00086010e-01 -6.88570440e-01
7.32562900e-01 8.28358382e-02 -6.98127866e-01 -7.12982476e-01
-1.17821312e+00 -7.50182688e-01 -6.82462990e-01 5.06038219e-02
6.63253725e-01 -1.27080715e+00 -3.61600220e-01 4.02750641e-01
-1.20415115e+00 6.48766607e-02 -1.28813609e-01 7.01021969e-01
6.99839219e-02 2.68144369e-01 -9.22267437e-01 -7.32323289e-01
-5.02612352e-01 -6.87569499e-01 8.55185211e-01 4.72024292e-01
-1.87563628e-01 -5.23735583e-01 3.34904231e-02 5.09554505e-01
3.16299796e-01 1.67787392e-02 6.27769589e-01 -1.31882894e+00
-4.19751853e-01 -2.40879118e-01 -1.46002606e-01 6.86476678e-02
-1.09174727e-02 -5.67210326e-03 -7.57269382e-01 1.06595689e-02
-6.06508613e-01 -1.58686593e-01 5.81276059e-01 -1.62839994e-01
6.09273911e-01 -2.87175924e-01 -4.32827026e-01 2.02818543e-01
1.30369163e+00 7.93002024e-02 4.35710549e-01 6.66604638e-01
6.78391039e-01 4.18043375e-01 6.66702509e-01 4.97812003e-01
7.81083405e-01 4.35648203e-01 -2.33554617e-02 6.84611052e-02
-6.28400547e-03 -4.37925607e-01 1.08353443e-01 9.48625624e-01
-3.05874925e-02 -3.01213413e-01 -1.25271451e+00 9.66277540e-01
-1.60406935e+00 -1.05082333e+00 -3.04305822e-01 2.18999100e+00
1.07701957e+00 5.71697466e-02 -1.07557125e-01 -3.63460153e-01
1.09886932e+00 -9.18792561e-02 -3.60546499e-01 -9.72859096e-03
-2.85880029e-01 1.20963521e-01 6.88846886e-01 1.29550949e-01
-1.20800388e+00 1.18419981e+00 5.04754305e+00 8.84044588e-01
-8.18356037e-01 9.16704759e-02 1.96837902e-01 4.57701609e-02
-3.60708833e-01 5.69612049e-02 -1.40199828e+00 5.89391589e-01
1.09086943e+00 -3.36195886e-01 2.53078341e-01 6.63513064e-01
-6.89174831e-02 1.84000999e-01 -6.99396610e-01 6.53145552e-01
-1.66753456e-01 -1.13076210e+00 -2.42006764e-01 -2.09444463e-01
6.23474419e-01 6.68780923e-01 -3.80893677e-01 6.54176354e-01
5.76271534e-01 -7.46993780e-01 9.95060325e-01 5.69952071e-01
8.12221229e-01 -7.07332790e-01 9.01944697e-01 4.89195377e-01
-1.38034511e+00 6.72159940e-02 -1.70342326e-01 2.37706602e-01
2.34129548e-01 8.52278411e-01 -6.74198985e-01 8.11750114e-01
1.00731492e+00 3.34714979e-01 -5.31548679e-01 1.24562097e+00
-6.19275570e-01 7.02807128e-01 -5.41250706e-01 -2.05478325e-01
6.19490957e-03 1.42867252e-01 6.94005191e-01 1.67632616e+00
2.92433351e-01 3.18719566e-01 1.13053411e-01 5.56191087e-01
-5.25707901e-01 5.33436894e-01 -4.86288190e-01 -2.81377763e-01
1.09814882e+00 1.47315300e+00 -5.93124866e-01 -4.27553087e-01
-3.82962048e-01 8.50666046e-01 7.44830847e-01 2.62702972e-01
-6.70919955e-01 -9.76259649e-01 5.74869335e-01 -1.36574939e-01
4.82093811e-01 -2.15540379e-01 -1.76300049e-01 -1.64561093e+00
4.11592633e-01 -9.00968909e-01 7.50508726e-01 -5.72416782e-01
-1.48863685e+00 8.15063357e-01 -3.57588202e-01 -1.02117002e+00
-3.69512523e-03 -4.26888496e-01 -2.73326546e-01 8.58259261e-01
-1.42836106e+00 -1.06076419e+00 -1.64206967e-01 3.69461060e-01
2.93528169e-01 4.77506816e-02 8.66396308e-01 9.02677536e-01
-5.79791546e-01 8.11487377e-01 2.74862200e-01 7.72621095e-01
1.05644119e+00 -1.29282129e+00 7.10996628e-01 1.03874946e+00
1.84565574e-01 9.35950398e-01 5.11892557e-01 -8.84686351e-01
-1.00775647e+00 -1.25785577e+00 1.83666348e+00 -7.98585892e-01
9.01550055e-01 -4.98247027e-01 -9.54154253e-01 7.69838929e-01
-5.02599701e-02 2.79321223e-02 1.02403378e+00 3.87121826e-01
-6.95063770e-01 2.21037760e-01 -9.43142414e-01 5.34305930e-01
1.17774093e+00 -6.14677906e-01 -8.01680148e-01 1.10454597e-01
8.55528355e-01 -6.79170609e-01 -1.21165812e+00 2.81626314e-01
4.24725682e-01 -3.92372072e-01 5.32892466e-01 -8.34599555e-01
1.05272621e-01 -5.34870982e-01 -2.66891837e-01 -1.34660673e+00
-3.34098607e-01 -2.18156710e-01 2.07738027e-01 1.76587176e+00
1.20063055e+00 -5.69915652e-01 1.90016747e-01 8.37142885e-01
-3.85678671e-02 -3.38258654e-01 -8.82328868e-01 -6.97724640e-01
-6.09571412e-02 -3.23770106e-01 8.15172493e-01 1.22210944e+00
1.46628201e-01 4.09362078e-01 -1.81613505e-01 6.13684416e-01
2.37207636e-01 2.71132350e-01 4.03796762e-01 -1.05405319e+00
2.95089688e-02 -3.34113985e-02 -2.47657701e-01 -7.96278536e-01
-1.07815219e-02 -1.01516736e+00 1.85343176e-01 -1.69247901e+00
8.67092758e-02 -7.64906764e-01 -4.45457280e-01 8.78671288e-01
-4.47157532e-01 5.18771037e-02 3.64056587e-01 3.15891355e-01
-1.00773239e+00 2.83116162e-01 6.13005877e-01 1.05155565e-01
-1.12709336e-01 2.14155968e-02 -9.05862570e-01 5.08608460e-01
6.79302871e-01 -6.86559021e-01 4.39624488e-01 -6.65432572e-01
6.02902055e-01 -1.61404923e-01 -2.68986579e-02 -8.37188601e-01
4.57574725e-01 -9.77222919e-02 4.01435524e-01 -5.67887902e-01
-8.95031616e-02 -5.83188593e-01 2.20138401e-01 -3.60408123e-03
-3.02887738e-01 2.88436502e-01 8.53863433e-02 3.21064562e-01
-3.13404441e-01 -2.95755208e-01 3.74066651e-01 -3.29614460e-01
-8.22470486e-01 1.71856999e-01 -3.29297066e-01 6.47450864e-01
6.42275572e-01 3.19310725e-01 -6.26294494e-01 1.94989637e-01
-7.26296008e-01 4.65586841e-01 3.43696415e-01 7.64658511e-01
4.53566834e-02 -1.28168333e+00 -8.87180030e-01 -1.72329679e-01
4.54846114e-01 -1.99593045e-02 2.97240794e-01 7.05660582e-01
-2.66279668e-01 3.88512105e-01 1.80909559e-02 -3.66913602e-02
-1.03501630e+00 2.44011760e-01 2.87813783e-01 -4.05531853e-01
-3.93291861e-01 7.73167491e-01 -1.60585687e-01 -9.78409827e-01
-6.05809502e-02 -2.31151395e-02 -3.78707856e-01 3.00789148e-01
6.34879053e-01 4.63126600e-01 4.55738336e-01 -7.87752450e-01
-7.01633573e-01 5.45860231e-02 -2.63772488e-01 -6.77991137e-02
1.35207093e+00 -1.69189885e-01 -1.86706871e-01 3.79892677e-01
9.34927344e-01 5.22519350e-01 -4.61375952e-01 -6.21793151e-01
6.41879380e-01 -1.71423838e-01 -3.73175502e-01 -1.11956441e+00
-6.93032742e-01 3.63818228e-01 8.21969956e-02 8.32100883e-02
7.66406357e-01 1.76409408e-01 8.70446622e-01 6.12209439e-01
4.92370963e-01 -1.32278883e+00 -6.63695693e-01 1.10369813e+00
4.80802387e-01 -1.12181020e+00 -4.00207818e-01 -3.31930220e-01
-7.35829115e-01 7.60675550e-01 6.79910481e-01 3.99091840e-01
5.61683774e-01 2.92192698e-01 1.45989612e-01 -7.64954612e-02
-7.01024115e-01 -5.66115558e-01 2.45179921e-01 2.34097004e-01
5.09968758e-01 3.33514571e-01 -5.92546344e-01 1.15363991e+00
-3.62676710e-01 -1.00925922e-01 3.36368859e-01 9.66563106e-01
-3.44468564e-01 -1.03068841e+00 -1.73288852e-01 5.13617039e-01
-8.64976048e-01 -5.97171307e-01 -2.50983447e-01 5.64613283e-01
2.83158511e-01 9.88772154e-01 -3.10357660e-01 -2.51884967e-01
9.27998066e-01 1.39613271e-01 -1.95198163e-01 -7.03456700e-01
-8.43449056e-01 -3.02491307e-01 5.79902887e-01 -4.80484188e-01
-1.17467768e-01 -7.21268058e-01 -1.36796081e+00 -8.61119479e-02
-4.65201348e-01 5.74973941e-01 6.65932119e-01 8.67175519e-01
8.63120854e-01 1.51724830e-01 4.04150963e-01 -9.61146504e-02
-3.79779279e-01 -9.99918520e-01 -5.09961009e-01 4.60277677e-01
-4.79196534e-02 -5.34166694e-01 -3.00343633e-01 -1.50560856e-01]
|
[9.629878044128418, 9.476466178894043]
|
5bb2f581-ce6e-4c2a-9971-05be4e90efb9
|
intelligent-frame-selection-as-a-privacy
|
2101.07529
| null |
https://arxiv.org/abs/2101.07529v2
|
https://arxiv.org/pdf/2101.07529v2.pdf
|
Intelligent Frame Selection as a Privacy-Friendlier Alternative to Face Recognition
|
The widespread deployment of surveillance cameras for facial recognition gives rise to many privacy concerns. This study proposes a privacy-friendly alternative to large scale facial recognition. While there are multiple techniques to preserve privacy, our work is based on the minimization principle which implies minimizing the amount of collected personal data. Instead of running facial recognition software on all video data, we propose to automatically extract a high quality snapshot of each detected person without revealing his or her identity. This snapshot is then encrypted and access is only granted after legal authorization. We introduce a novel unsupervised face image quality assessment method which is used to select the high quality snapshots. For this, we train a variational autoencoder on high quality face images from a publicly available dataset and use the reconstruction probability as a metric to estimate the quality of each face crop. We experimentally confirm that the reconstruction probability can be used as biometric quality predictor. Unlike most previous studies, we do not rely on a manually defined face quality metric as everything is learned from data. Our face quality assessment method outperforms supervised, unsupervised and general image quality assessment methods on the task of improving face verification performance by rejecting low quality images. The effectiveness of the whole system is validated qualitatively on still images and videos.
|
['Pieter Simoens', 'Sam Leroux', 'Mattijs Baert']
|
2021-01-19
| null | null | null | null |
['face-image-quality', 'face-image-quality-assessment']
|
['computer-vision', 'computer-vision']
|
[ 2.72535205e-01 1.02363363e-01 -6.69824630e-02 -8.21424663e-01
-7.58289695e-01 -4.31752384e-01 3.17843080e-01 -3.37544888e-01
-6.53617918e-01 5.07596612e-01 5.35633527e-02 8.43563825e-02
-2.56888062e-01 -7.76687920e-01 -5.30747294e-01 -9.20934260e-01
2.39686400e-01 7.70656113e-03 -4.43340480e-01 3.19992304e-01
1.08957946e-01 7.09093392e-01 -1.78035271e+00 1.90209597e-01
5.90675473e-01 1.31024718e+00 -5.11120021e-01 4.61742193e-01
3.67762268e-01 3.44835818e-01 -6.77182913e-01 -1.04857051e+00
7.20519662e-01 -4.06822652e-01 -5.86272597e-01 5.12210608e-01
9.10162389e-01 -9.26719248e-01 -2.30409756e-01 1.33181739e+00
5.76588631e-01 -9.02110189e-02 4.95402902e-01 -1.62216163e+00
-4.35864180e-01 2.16929503e-02 -2.89348632e-01 -1.59637943e-01
5.01203358e-01 1.97800234e-01 9.45590258e-01 -8.02074730e-01
6.65152550e-01 9.82986152e-01 5.28900146e-01 7.23589778e-01
-1.22829294e+00 -7.17788398e-01 -2.81280428e-01 3.25037211e-01
-1.66048169e+00 -1.03888774e+00 7.87593067e-01 -3.70056063e-01
3.75998437e-01 3.13634694e-01 6.48063421e-01 9.92247343e-01
-1.54203877e-01 3.69087219e-01 1.18090022e+00 -3.64071906e-01
2.54537821e-01 4.70495850e-01 7.91734159e-02 8.90858531e-01
3.41542542e-01 4.34129685e-01 -6.19529426e-01 -3.58845890e-01
5.36360145e-01 5.05989194e-02 -3.46664548e-01 -3.87327135e-01
-6.83218360e-01 6.84864759e-01 2.18744087e-03 1.12457305e-01
-5.00559211e-01 -1.56078801e-01 3.20503041e-02 6.08082294e-01
3.86836886e-01 -9.73335281e-02 -1.54771417e-01 2.26501346e-01
-1.27962375e+00 -4.59070876e-02 8.61166179e-01 5.20922422e-01
9.42389667e-01 -1.14374675e-01 -2.18102410e-01 6.50986671e-01
5.53041577e-01 7.62282908e-01 1.58551767e-01 -1.29104567e+00
-7.13547245e-02 6.04433239e-01 1.37842953e-01 -1.19741666e+00
1.51036501e-01 1.81173325e-01 -7.20861733e-01 6.26642346e-01
4.64450449e-01 -1.85311928e-01 -6.77116156e-01 1.69920278e+00
4.02963430e-01 1.29571157e-02 -1.79675762e-02 8.17880213e-01
5.98882437e-01 3.22050005e-01 -1.37640685e-01 -6.09184384e-01
1.27698016e+00 -3.52448076e-01 -8.29009235e-01 4.55563158e-01
-1.43689245e-01 -5.46386778e-01 6.61027014e-01 7.46327221e-01
-8.25035334e-01 -3.12281340e-01 -9.36149001e-01 1.03743061e-01
-1.78237915e-01 3.31075937e-01 4.68076974e-01 1.56433427e+00
-1.43789315e+00 5.33350587e-01 -5.20030797e-01 -3.63159239e-01
8.61362875e-01 7.29267657e-01 -1.00419891e+00 6.07845448e-02
-8.42516899e-01 6.17363691e-01 -1.03780396e-01 1.55682519e-01
-1.07202041e+00 -3.93796951e-01 -7.80027628e-01 5.79720996e-02
3.30148160e-01 -4.50006425e-01 7.59312749e-01 -1.46700752e+00
-1.82869530e+00 1.24148083e+00 -2.67399311e-01 -3.95618200e-01
5.66771984e-01 -6.04529865e-02 -4.77932453e-01 5.04512370e-01
-1.72808439e-01 5.19923747e-01 1.48319352e+00 -1.08590603e+00
-4.16400075e-01 -6.95162177e-01 -4.22554240e-02 -1.62132546e-01
-6.48332357e-01 2.89291501e-01 -4.23793316e-01 -2.13838711e-01
-1.83938071e-01 -7.18834400e-01 2.32690960e-01 5.20163238e-01
-1.75345391e-01 4.70596515e-02 8.30609202e-01 -8.41080725e-01
9.61303830e-01 -2.17801833e+00 2.88254563e-02 4.37251896e-01
2.00992718e-01 3.14973652e-01 -1.42784566e-01 -8.01861212e-02
1.61293611e-01 1.75773829e-01 -3.21194440e-01 -5.67956865e-01
4.98884823e-03 -3.97650450e-02 6.51643947e-02 8.52324367e-01
3.42381775e-01 6.16863310e-01 -4.44267839e-01 -5.49923122e-01
2.67111808e-01 8.10717940e-01 -6.85236454e-01 4.08917964e-01
2.16113389e-01 2.39789769e-01 -8.41968730e-02 9.56412733e-01
1.05735278e+00 1.20671749e-01 2.05824077e-01 -2.87936091e-01
2.81504661e-01 -2.78315812e-01 -1.24246454e+00 1.33723402e+00
2.91007268e-03 4.66641277e-01 4.49064165e-01 -8.60668421e-01
7.66083479e-01 5.66570342e-01 7.53884733e-01 -4.40993339e-01
4.18566912e-01 -1.52705178e-01 -3.20862651e-01 -5.71690440e-01
5.33372648e-02 -6.66605085e-02 2.91373044e-01 4.74349052e-01
4.64583486e-01 3.90568525e-01 -1.54789820e-01 -5.66902868e-02
9.02737975e-01 -5.03299199e-02 2.03777254e-01 -1.96519569e-01
6.90705180e-01 -6.53051436e-01 7.05638170e-01 4.64537978e-01
-7.15222716e-01 5.09162784e-01 5.78941524e-01 -3.77283096e-01
-8.35472167e-01 -1.04742754e+00 -2.67051995e-01 7.87697315e-01
-9.89300609e-02 -4.28188324e-01 -1.21561158e+00 -1.01077497e+00
7.71294385e-02 6.60165995e-02 -8.01035464e-01 -5.90795800e-02
-1.12748332e-01 -6.16028309e-01 6.82100713e-01 -1.06494233e-01
5.60399592e-01 -9.15565670e-01 -5.89792013e-01 -3.18197638e-01
-6.71416000e-02 -8.90671372e-01 -4.54529434e-01 -4.54215795e-01
-6.00014985e-01 -1.40634227e+00 -4.76307601e-01 -3.07670683e-01
8.12259078e-01 -3.75947319e-02 8.68274748e-01 3.68362039e-01
-3.54793757e-01 8.46099377e-01 -8.11274126e-02 -3.11007768e-01
-2.30493918e-01 -4.02852118e-01 3.43739033e-01 9.39951718e-01
7.06442058e-01 -3.67521197e-01 -5.78058362e-01 2.51981795e-01
-8.79770815e-01 -6.75101459e-01 4.46634918e-01 8.33209276e-01
5.06374896e-01 1.83563977e-01 3.12519759e-01 -8.37431312e-01
5.45441210e-01 -1.00472540e-01 -7.78356373e-01 4.45997149e-01
-8.75470221e-01 3.26361693e-02 2.64480084e-01 -3.45823169e-01
-1.03173161e+00 3.35898697e-01 -2.15024725e-01 -5.93870759e-01
-4.21147257e-01 -3.56691927e-02 -6.98135376e-01 -4.37421143e-01
2.92977601e-01 2.42567763e-01 3.60945046e-01 -2.33415335e-01
1.72907442e-01 7.15485215e-01 2.55428553e-01 -2.34926462e-01
8.15728009e-01 6.64666891e-01 3.91010195e-03 -1.11188567e+00
-3.41831148e-01 -2.75292873e-01 -7.08226919e-01 -5.65941751e-01
7.75228918e-01 -6.94621444e-01 -1.26827860e+00 6.55338764e-01
-9.16625619e-01 2.58131325e-01 -7.65966773e-02 3.65805179e-01
-4.20461774e-01 5.65637052e-01 -3.38201046e-01 -1.29955125e+00
-4.33541358e-01 -1.21385956e+00 1.21429420e+00 1.05316870e-01
1.46890372e-01 -5.95655262e-01 -7.37878829e-02 4.48611379e-01
3.05352628e-01 2.08186075e-01 2.37031087e-01 -2.47235760e-01
-6.81332588e-01 -5.58909297e-01 -6.99511170e-02 7.87554324e-01
3.09230447e-01 2.23413244e-01 -1.32752216e+00 -4.28272784e-01
4.32578921e-01 -3.70313913e-01 7.01187670e-01 3.92540067e-01
1.12042832e+00 -6.73572421e-01 1.20738722e-01 8.72372687e-01
1.46849930e+00 6.42347485e-02 9.52374935e-01 -1.56614527e-01
4.44865972e-01 9.43137109e-01 2.57535726e-01 5.42379439e-01
7.10137710e-02 5.69493532e-01 3.95292342e-01 2.10165367e-01
3.56950074e-01 -3.51185240e-02 7.32619822e-01 2.13697672e-01
-2.74020642e-01 -7.25990087e-02 -4.45591986e-01 1.78132743e-01
-1.54632521e+00 -1.31017017e+00 5.62414944e-01 2.61136818e+00
6.99827254e-01 -4.13138658e-01 3.27921301e-01 1.17377721e-01
6.61413610e-01 -3.41043109e-04 -2.75462359e-01 -9.46895778e-02
-8.46126303e-02 2.97969639e-01 3.90064687e-01 6.39365911e-01
-1.09381080e+00 7.19401777e-01 6.60919046e+00 4.24087226e-01
-1.15056813e+00 1.19508415e-01 9.27853823e-01 -2.99537957e-01
-3.26248147e-02 -1.37545034e-01 -7.03967094e-01 4.09415126e-01
9.78568435e-01 1.29328063e-02 5.66328704e-01 8.84252667e-01
1.15785316e-01 -6.46123588e-02 -1.17161286e+00 1.31991947e+00
4.24207211e-01 -9.02671754e-01 6.71502873e-02 5.05987942e-01
2.85446793e-01 -5.77880979e-01 2.71468371e-01 -9.23723057e-02
-2.74421331e-02 -1.20810187e+00 4.93910372e-01 9.70648527e-01
8.76035988e-01 -8.52931559e-01 5.80308676e-01 7.04376176e-02
-7.28634417e-01 -5.28844781e-02 -5.06907105e-01 2.66918033e-01
-1.40207142e-01 6.18256390e-01 -3.86363059e-01 3.77547085e-01
8.36852133e-01 4.94858533e-01 -5.98211348e-01 7.36802280e-01
-7.26905763e-02 4.75408018e-01 -4.41510111e-01 2.29012191e-01
-3.69801193e-01 -4.66852874e-01 4.99408334e-01 8.31740737e-01
3.59341979e-01 3.05824250e-01 -1.83330551e-01 8.27190101e-01
-3.64236683e-01 1.92894071e-01 -8.20461571e-01 -2.82351784e-02
2.44827837e-01 1.30816817e+00 -2.88329154e-01 -9.65925902e-02
-4.44022477e-01 1.36113906e+00 7.99400508e-02 1.78401589e-01
-5.12091160e-01 -7.06857145e-02 8.21046054e-01 2.21114904e-02
2.14328393e-01 1.91851169e-01 9.19404402e-02 -1.43298626e+00
2.31533676e-01 -1.17013550e+00 4.10959542e-01 -3.67302239e-01
-1.15807855e+00 7.45058596e-01 -2.71947980e-01 -1.11607909e+00
-1.85633734e-01 -6.08164310e-01 -3.13959777e-01 7.07385242e-01
-1.31225634e+00 -1.14988887e+00 -2.77034521e-01 9.83309925e-01
-5.05740056e-03 -5.06846368e-01 9.72881675e-01 4.33142543e-01
-6.63412809e-01 9.68906105e-01 -1.70704171e-01 4.17620867e-01
7.83286631e-01 -9.57102835e-01 -3.58972579e-01 1.01516449e+00
3.40517849e-01 7.67694175e-01 3.74449313e-01 -5.05511343e-01
-1.55444860e+00 -9.03404534e-01 8.07175219e-01 -6.83557451e-01
4.01172563e-02 -3.77083749e-01 -7.16753244e-01 5.03355265e-01
3.63590747e-01 2.29078859e-01 1.01257527e+00 -1.04298845e-01
-5.17295301e-01 -6.11838698e-01 -1.85921311e+00 1.26311928e-01
8.25757325e-01 -7.00590312e-01 -2.89711386e-01 2.96241138e-03
1.35972455e-01 3.59880060e-01 -1.05176353e+00 3.31246793e-01
8.43039870e-01 -1.35020697e+00 8.03569317e-01 -3.91667902e-01
8.91794264e-02 -2.23731652e-01 -2.81006634e-01 -6.95195675e-01
2.25231685e-02 -5.80193937e-01 -1.80104271e-01 1.46912301e+00
1.44780040e-01 -5.24602771e-01 1.07614946e+00 9.72099066e-01
8.03076684e-01 -2.90217608e-01 -1.12007833e+00 -5.64949095e-01
-3.89431268e-01 -2.52860844e-01 7.18910635e-01 8.11259806e-01
-3.53676319e-01 -1.76087588e-01 -7.38147795e-01 4.53989923e-01
1.26351774e+00 -1.45397842e-01 8.00450921e-01 -1.31914556e+00
-1.73011854e-01 -2.81693071e-01 -7.37258792e-01 -3.23871493e-01
3.47836047e-01 -5.21318197e-01 -1.88275695e-01 -7.68904030e-01
2.68604666e-01 7.48811141e-02 -4.92103636e-01 4.03379530e-01
1.26702860e-01 4.97391641e-01 3.41690108e-02 2.81016249e-02
-4.38791424e-01 5.74473739e-01 6.95055783e-01 -1.98177606e-01
7.78328031e-02 1.49748936e-01 -6.53835595e-01 4.48632717e-01
5.75999677e-01 -4.03641939e-01 -3.71842682e-01 -1.91126466e-01
-1.02752380e-01 1.62351981e-01 5.92483580e-01 -1.07180214e+00
1.65476054e-01 -2.39039287e-01 6.06422663e-01 -2.32982263e-01
4.20711130e-01 -1.36573768e+00 2.50987202e-01 2.23689362e-01
-2.53481120e-01 -2.97272831e-01 -1.09193303e-01 5.36216259e-01
-2.87047267e-01 -2.08881482e-01 1.07559669e+00 -7.42880628e-02
-3.90557498e-01 6.88061774e-01 -1.93469241e-01 -5.89151621e-01
1.08052015e+00 -3.77625853e-01 1.00916110e-01 -5.18912673e-01
-7.69676208e-01 -1.34157404e-01 7.13086963e-01 2.10802481e-01
8.71765494e-01 -1.35259640e+00 -7.40238249e-01 7.28846967e-01
9.58839655e-02 -6.69026911e-01 1.40371710e-01 5.02236664e-01
-3.37979496e-01 1.98044926e-01 -3.90836984e-01 -4.66650099e-01
-1.75420034e+00 6.74600422e-01 4.98076499e-01 2.60770500e-01
-2.10529879e-01 7.28252947e-01 -1.92096427e-01 -2.00425342e-01
3.76579493e-01 1.26531541e-01 -1.75642073e-01 2.04466417e-01
9.00431275e-01 2.59197712e-01 1.57151788e-01 -1.06499922e+00
-4.79899168e-01 4.93605375e-01 1.34042591e-01 -3.68082911e-01
1.37568915e+00 -2.84174621e-01 -4.09350336e-01 -8.43798295e-02
1.43429744e+00 1.49096757e-01 -1.31663811e+00 -1.33061232e-02
-9.23688710e-03 -1.00777268e+00 1.66028127e-01 -5.35655558e-01
-1.57605159e+00 6.51161194e-01 1.06589866e+00 5.01889586e-02
1.56245971e+00 -3.17856967e-01 3.16123098e-01 3.39039654e-01
4.96327549e-01 -1.05638325e+00 -9.54333618e-02 -1.14143968e-01
6.94511890e-01 -1.68173742e+00 -5.81460539e-03 -2.44716927e-01
-5.20681858e-01 9.53873158e-01 3.88478667e-01 -3.67982127e-02
8.73518765e-01 1.31683126e-01 8.80979151e-02 -1.35504991e-01
-5.59696496e-01 2.80877226e-03 4.45393920e-01 8.29472363e-01
1.22792475e-01 -6.19149096e-02 -1.82564303e-01 5.17329395e-01
-1.32670611e-01 2.72521526e-01 3.09011549e-01 6.95559621e-01
-1.97687179e-01 -1.40165615e+00 -4.52300280e-01 5.55661678e-01
-7.70445287e-01 1.41453356e-01 -6.34221137e-01 2.84435749e-01
2.56728202e-01 1.05016160e+00 -1.90602183e-01 -4.13039535e-01
1.46342263e-01 3.21165621e-01 6.12788081e-01 -3.00911307e-01
-6.06376469e-01 -1.08085915e-01 -2.60638833e-01 -8.83712769e-01
-8.12761068e-01 -8.48058581e-01 -3.51710737e-01 -5.61579525e-01
-3.28194261e-01 -2.79606064e-03 6.94793582e-01 6.42678678e-01
2.91093111e-01 -2.67810404e-01 9.39880967e-01 -5.67370534e-01
-4.00536150e-01 -4.70742345e-01 -7.35819519e-01 6.90369606e-01
6.79472148e-01 -4.40863848e-01 -3.40187252e-01 3.89485598e-01]
|
[13.10615348815918, 0.7857871055603027]
|
0ff3b347-15fc-4f36-b606-f5544ff7e767
|
ensembling-instance-and-semantic-segmentation
|
2304.10326
| null |
https://arxiv.org/abs/2304.10326v1
|
https://arxiv.org/pdf/2304.10326v1.pdf
|
Ensembling Instance and Semantic Segmentation for Panoptic Segmentation
|
We demonstrate our solution for the 2019 COCO panoptic segmentation task. Our method first performs instance segmentation and semantic segmentation separately, then combines the two to generate panoptic segmentation results. To enhance the performance, we add several expert models of Mask R-CNN in instance segmentation to tackle the data imbalance problem in the training data; also HTC model is adopted yielding our best instance segmentation results. In semantic segmentation, we trained several models with various backbones and use an ensemble strategy which further boosts the segmentation results. In the end, we analyze various combinations of instance and semantic segmentation, and report on their performance for the final panoptic segmentation results. Our best model achieves $PQ$ 47.1 on 2019 COCO panoptic test-dev data.
|
['Yogesh Langhe', 'Mehmet Yildirim']
|
2023-04-20
| null | null | null | null |
['panoptic-segmentation']
|
['computer-vision']
|
[-4.72328905e-03 4.42598127e-02 -3.93566191e-01 -5.66280425e-01
-8.75669241e-01 -6.91869199e-01 3.40186536e-01 -5.08921184e-02
-2.34708995e-01 4.75966662e-01 -2.34205917e-01 -4.26301390e-01
3.10363099e-02 -9.52721655e-01 -7.08172321e-01 -7.87070096e-01
5.69604188e-02 7.84663856e-01 2.75068015e-01 1.86014399e-01
4.62754041e-01 3.33496988e-01 -1.15444767e+00 2.64860570e-01
1.29506421e+00 1.35440552e+00 -1.24037869e-01 7.06764877e-01
-2.49699205e-01 3.57784003e-01 -9.06543851e-01 -3.18831712e-01
7.63521373e-01 -3.96389574e-01 -7.86710203e-01 2.74279684e-01
4.67474848e-01 7.52692670e-02 2.70692736e-01 1.06732881e+00
3.89567554e-01 9.06465724e-02 6.57082617e-01 -1.05313933e+00
-1.09367751e-01 9.31778729e-01 -1.10584724e+00 4.94522184e-01
-5.53248465e-01 3.27758014e-01 1.17686617e+00 -5.48138201e-01
3.92062098e-01 1.04396439e+00 6.74165785e-01 2.39569440e-01
-1.02200186e+00 -9.12358820e-01 1.77205607e-01 -3.04468423e-01
-1.17106009e+00 -2.42823124e-01 5.63717604e-01 -2.46381924e-01
1.06319737e+00 3.82335603e-01 1.00790691e+00 2.82535970e-01
1.36156052e-01 1.09102714e+00 1.28095448e+00 2.59763817e-03
1.46201968e-01 -1.17500722e-01 3.02869588e-01 3.48421216e-01
7.17142448e-02 -1.81846663e-01 5.89422472e-02 2.58894801e-01
6.25576138e-01 -3.31789047e-01 5.42906746e-02 4.00012195e-01
-7.55245090e-01 8.64197254e-01 6.71094954e-01 2.17338905e-01
-3.45026165e-01 1.73524305e-01 2.78127819e-01 1.83554515e-01
1.07235634e+00 8.13433766e-01 -8.75104666e-01 2.31854647e-01
-1.72070551e+00 5.15032172e-01 6.37164772e-01 7.65100479e-01
7.18036950e-01 1.01326287e-01 -2.24270016e-01 1.17119873e+00
3.83595765e-01 5.54094613e-01 3.42291057e-01 -1.09060216e+00
8.46175253e-01 7.68347561e-01 -1.37773871e-01 -6.29809976e-01
-6.11819863e-01 -8.81358564e-01 -7.30327070e-01 -1.43904269e-01
2.59586155e-01 -5.08377135e-01 -1.86095166e+00 1.29432642e+00
3.47939879e-01 1.90068483e-01 -9.83799696e-02 7.17547178e-01
7.72732437e-01 1.19394171e+00 1.06654815e-01 -2.43845731e-01
1.22585487e+00 -1.35599267e+00 -4.50525492e-01 -1.42133176e-01
4.07475710e-01 -7.43003488e-01 6.14909589e-01 4.47958708e-01
-1.13334692e+00 -6.03657663e-01 -1.11474848e+00 2.78456122e-01
-5.07626355e-01 9.38734859e-02 2.69546658e-01 8.04239511e-01
-1.02146029e+00 5.86706638e-01 -7.72377431e-01 -1.66494213e-02
7.43239462e-01 6.23947144e-01 4.07046348e-01 2.34799594e-01
-1.09093630e+00 4.86951023e-01 7.46337056e-01 2.22677618e-01
-7.46401787e-01 -9.15311635e-01 -6.38308108e-01 1.25723854e-01
2.96310812e-01 -4.01019335e-01 1.05682552e+00 -8.54823172e-01
-1.34356260e+00 9.21309412e-01 3.98639798e-01 -8.22805941e-01
4.48138565e-01 -3.88896614e-02 -1.22022077e-01 1.49832249e-01
7.68921003e-02 1.52699518e+00 6.45396471e-01 -1.24952388e+00
-1.11266494e+00 -1.89882025e-01 -1.31068721e-01 3.61727566e-01
3.40630203e-01 4.77406792e-02 -9.41960871e-01 -8.59842420e-01
3.41133535e-01 -9.62641776e-01 -6.49243534e-01 -8.33637238e-01
-6.64137065e-01 -2.32970923e-01 7.29430079e-01 -9.34117377e-01
1.37255168e+00 -2.06170201e+00 -5.88799939e-02 5.70276260e-01
6.91983849e-02 2.81001091e-01 -6.52641878e-02 -2.89327711e-01
-3.50281335e-02 6.69705987e-01 -9.88438189e-01 -5.75184166e-01
-1.12134032e-01 2.48794779e-01 -1.93245113e-01 1.90054253e-01
2.04669207e-01 8.94470036e-01 -2.40455732e-01 -7.31485963e-01
1.53694913e-01 1.13107793e-01 -7.72298574e-01 2.68932223e-01
-8.46916556e-01 4.65966254e-01 -3.67225796e-01 9.35785532e-01
1.17387128e+00 -5.66533245e-02 -1.47637129e-01 2.58180380e-01
-2.27101728e-01 1.19515091e-01 -1.12117100e+00 1.63203251e+00
-1.90102115e-01 3.87551397e-01 1.93599798e-02 -1.10619843e+00
9.47576880e-01 1.88763410e-01 6.54253542e-01 -1.00087285e+00
1.91222027e-01 3.91913623e-01 1.05768979e-01 -2.92595178e-01
7.05170095e-01 -8.04152489e-02 -1.90146804e-01 3.50403577e-01
7.36370236e-02 -7.73886025e-01 4.59458530e-01 -8.94649774e-02
2.95871228e-01 2.35904798e-01 -1.86381653e-01 -5.50483167e-01
2.60942042e-01 3.76700759e-01 8.38335812e-01 6.87614918e-01
-2.48405799e-01 1.15859234e+00 9.40029979e-01 -3.82821590e-01
-1.08079219e+00 -7.90933073e-01 -4.43179220e-01 8.91641140e-01
6.20031124e-03 -4.12528925e-02 -1.17456210e+00 -9.20143366e-01
-2.91266948e-01 6.78718686e-01 -7.98814058e-01 6.05146885e-01
-7.97083497e-01 -2.01969266e+00 5.82462192e-01 3.27047437e-01
8.59667778e-01 -1.27401757e+00 -4.14644241e-01 2.37372831e-01
-1.63561255e-01 -1.01242542e+00 -2.79761016e-01 3.72552603e-01
-1.06399000e+00 -9.24772024e-01 -9.75593805e-01 -6.51882529e-01
4.46157098e-01 -2.67104298e-01 1.20817053e+00 5.79536371e-02
-1.71968058e-01 -3.17046821e-01 -2.73609132e-01 -4.63963658e-01
-2.49798998e-01 6.34353757e-01 -6.98779225e-01 -1.02217160e-01
-1.52617514e-01 -3.06476295e-01 -8.65778744e-01 1.73395455e-01
-8.93025875e-01 2.01691970e-01 2.90327102e-01 3.92459810e-01
9.35037255e-01 2.44473904e-01 5.34830093e-01 -1.23285949e+00
2.27421001e-01 -7.23306179e-01 -1.09070277e+00 2.45508358e-01
-8.02629590e-01 -2.79046178e-01 4.22659785e-01 1.21768557e-01
-1.06233442e+00 -1.30467474e-01 -5.13869822e-01 -1.68910593e-01
-9.42032933e-02 5.21439195e-01 5.18460125e-02 3.04004103e-01
3.35212559e-01 -1.56868413e-01 -5.88630795e-01 -7.01722383e-01
3.34904790e-01 5.44367075e-01 2.43321985e-01 -4.91442859e-01
4.13022727e-01 1.11878522e-01 -3.39202344e-01 -3.89173359e-01
-1.07268846e+00 -2.76728094e-01 -5.37676096e-01 -4.58382890e-02
1.26958239e+00 -1.01282835e+00 1.66727602e-02 8.06051195e-01
-8.57083440e-01 -8.13680470e-01 -3.24704677e-01 1.53845429e-01
-2.61677265e-01 -2.14855164e-01 -7.40392387e-01 -5.53748667e-01
-7.36994267e-01 -1.50872827e+00 1.02170503e+00 6.40464425e-01
2.47858405e-01 -8.05852115e-01 3.25241864e-01 6.66021645e-01
2.85049021e-01 3.54897439e-01 8.23043466e-01 -8.80997598e-01
-5.96019566e-01 -7.78434053e-02 -5.88896990e-01 5.37540853e-01
-2.62723118e-01 3.65050822e-01 -1.00697136e+00 -5.30398600e-02
-6.43314943e-02 -7.45236427e-02 1.21282554e+00 1.00652468e+00
1.50732875e+00 -6.72893450e-02 -1.60923466e-01 7.93597221e-01
1.56383920e+00 5.71519792e-01 7.37689912e-01 2.12227881e-01
7.73340464e-01 6.85606420e-01 7.31724441e-01 1.47899851e-01
4.15256321e-01 2.57632494e-01 3.83855551e-01 -5.10920048e-01
-1.12233736e-01 1.60608307e-01 -6.30302876e-02 9.86950874e-01
-5.54931462e-02 -6.97991967e-01 -1.16724169e+00 8.15281808e-01
-1.57508826e+00 -3.24871242e-01 -3.35709989e-01 1.64517975e+00
7.91071653e-01 3.07177037e-01 3.40519428e-01 5.57550229e-02
7.09852576e-01 5.04525483e-01 -4.15437549e-01 -6.01138294e-01
-1.94688857e-01 6.98023081e-01 7.67012060e-01 4.63374138e-01
-1.35953450e+00 1.35709798e+00 6.97348022e+00 1.16938078e+00
-1.14609921e+00 2.54494101e-01 1.68388510e+00 -1.13995150e-01
-2.44983643e-01 -1.05327606e-01 -8.84509981e-01 5.94684839e-01
9.50151801e-01 4.63856369e-01 2.03696221e-01 4.25557494e-01
7.92244375e-02 -4.11751330e-01 -4.52156603e-01 3.64825189e-01
-3.11109960e-01 -1.37589371e+00 4.31329124e-02 1.92199331e-02
1.57151651e+00 5.54604828e-01 1.41629577e-01 1.94150612e-01
4.72736299e-01 -1.28974402e+00 5.95146418e-01 4.91034351e-02
5.28633714e-01 -1.05387557e+00 9.78307426e-01 5.59346229e-02
-1.11732256e+00 -1.70107819e-02 -2.20777988e-01 2.66697317e-01
4.74421717e-02 7.87626922e-01 -6.73594713e-01 8.09851289e-01
7.58790195e-01 5.33857763e-01 -6.24758124e-01 1.40699899e+00
-5.40255941e-02 1.03806961e+00 -7.60753214e-01 5.13989329e-01
7.66961813e-01 -4.12049502e-01 2.13967323e-01 1.40392756e+00
2.66004086e-01 -9.46796387e-02 2.20344245e-01 9.38325465e-01
-3.88551593e-01 3.19240808e-01 1.19927809e-01 1.27417430e-01
1.37204230e-01 1.20774341e+00 -1.47756743e+00 -6.57287180e-01
-1.29449949e-01 3.49220634e-01 -1.36436209e-01 3.74174207e-01
-1.18640447e+00 -1.10921733e-01 1.91742301e-01 -2.71732420e-01
3.56753975e-01 2.48226166e-01 -9.47912276e-01 -9.41388011e-01
-3.84543359e-01 -8.79153728e-01 7.54263878e-01 -5.66930234e-01
-7.17010260e-01 5.29298663e-01 2.06388906e-01 -7.97597110e-01
1.46545261e-01 -3.05113673e-01 -8.23878050e-01 6.40966475e-01
-1.49599922e+00 -1.08058703e+00 6.35560304e-02 -5.28855361e-02
9.91693974e-01 8.38814005e-02 1.78928822e-01 5.32112241e-01
-1.03624284e+00 3.36624682e-01 -5.31509444e-02 8.92084986e-02
1.33879170e-01 -1.43844604e+00 7.60826230e-01 9.92846191e-01
1.59892157e-01 4.03323434e-02 4.83109444e-01 -7.46685565e-01
-4.11915302e-01 -1.27655733e+00 6.06213272e-01 -1.15892000e-01
3.59926075e-01 -6.83659390e-02 -7.43267059e-01 6.21212304e-01
4.23860520e-01 -1.53965503e-01 4.48233724e-01 -1.95532084e-01
8.02487656e-02 -2.44916514e-01 -1.26409936e+00 2.84800470e-01
4.73696023e-01 1.04208328e-01 -4.00948167e-01 2.29667619e-01
9.77111638e-01 -6.82644248e-01 -9.39419091e-01 8.43040407e-01
3.35651457e-01 -8.17701101e-01 8.30576062e-01 -1.68350831e-01
6.42483592e-01 -3.99620891e-01 -3.53786424e-02 -1.29506767e+00
1.87189639e-01 -3.85320395e-01 3.80719751e-01 1.37181211e+00
1.12334204e+00 -3.62313241e-01 8.74261558e-01 2.65760690e-01
-3.17531765e-01 -1.03265202e+00 -9.31617022e-01 -2.36406580e-01
5.95609784e-01 -6.03132606e-01 9.34169769e-01 9.65915024e-01
-7.22818017e-01 -2.35299356e-02 -2.31805518e-01 2.03821971e-03
4.28659946e-01 5.41725874e-01 4.67701465e-01 -9.06500041e-01
-1.33415028e-01 -7.96941817e-01 3.26056391e-01 -1.09099340e+00
-1.34534627e-01 -9.08134997e-01 2.80379981e-01 -1.49296558e+00
1.86364993e-01 -8.85779381e-01 -5.03005564e-01 3.66501987e-01
-2.68461108e-01 8.74225438e-01 3.75109315e-01 4.81350943e-02
-5.52578807e-01 2.23737746e-01 1.52263737e+00 -2.04706222e-01
-5.33360124e-01 3.55731636e-01 -6.46627486e-01 5.20983696e-01
1.14180946e+00 -7.51523197e-01 -2.02398062e-01 -5.35663307e-01
1.30616933e-01 7.50306025e-02 -6.20565601e-02 -1.00194716e+00
-1.34131774e-01 -2.07425416e-01 4.00978535e-01 -1.08601475e+00
-1.16133146e-01 -5.68947434e-01 5.65726236e-02 4.54678357e-01
-1.13053575e-01 5.97123199e-06 3.51865798e-01 2.96992771e-02
-2.95402437e-01 -2.49216855e-01 1.18959463e+00 -3.59580368e-01
-3.64788175e-01 6.19675577e-01 -1.94443420e-01 2.96959579e-01
7.57472277e-01 -2.80539114e-02 -2.97582567e-01 2.30789825e-01
-9.69228804e-01 8.48642588e-01 2.02947930e-01 1.27968341e-01
4.62691002e-02 -8.18808138e-01 -7.84704328e-01 1.65903196e-01
-4.10618752e-01 5.99977732e-01 2.06041232e-01 1.03087211e+00
-1.12336028e+00 2.80975461e-01 -1.04927935e-01 -7.51086712e-01
-7.25728571e-01 1.00795299e-01 6.87278032e-01 -7.93699682e-01
-3.91451091e-01 9.59686279e-01 3.72625679e-01 -6.79798543e-01
1.21013023e-01 -6.62412107e-01 -3.96910548e-01 6.26387775e-01
-4.43923473e-02 3.86634320e-01 3.86705389e-03 -3.58283937e-01
-2.14455530e-01 7.51707077e-01 1.34578824e-01 -3.23100686e-01
1.41152596e+00 -5.53565919e-02 -2.72530794e-01 1.98352292e-01
1.03772163e+00 -3.11978668e-01 -1.47690809e+00 1.94280654e-01
-1.99767631e-02 -4.44878936e-02 4.70943674e-02 -1.17903936e+00
-1.88074994e+00 9.54183102e-01 5.57823837e-01 4.13301378e-01
1.43618929e+00 -1.67212009e-01 9.78505969e-01 -1.72894970e-01
-3.20772618e-01 -1.45377815e+00 -4.12625462e-01 5.15763998e-01
4.92415190e-01 -1.14997554e+00 1.82797432e-01 -4.23136503e-01
-5.46456695e-01 8.27017903e-01 7.22250342e-01 -4.31089252e-01
7.42869616e-01 3.67988825e-01 1.80723786e-01 -3.10900211e-01
-5.30035377e-01 -2.10155487e-01 4.26406175e-01 -2.66298484e-02
1.82024032e-01 2.84566313e-01 -4.07558382e-01 5.43460369e-01
-3.73580188e-01 -3.67163539e-01 1.36689678e-01 3.89645576e-01
-5.20462453e-01 -8.20240915e-01 -3.29339951e-01 6.26178861e-01
-1.10668433e+00 -3.61755222e-01 -2.43269771e-01 8.11133385e-01
6.72579288e-01 7.44852185e-01 6.12332761e-01 -1.09639741e-01
6.45572841e-02 6.24732338e-02 1.26029670e-01 -3.67413074e-01
-1.28064620e+00 8.11546803e-01 2.03669891e-01 -2.74565518e-01
-5.49018562e-01 -5.57463527e-01 -1.30309486e+00 -2.59777427e-01
-3.82675469e-01 3.40429872e-01 8.11114132e-01 1.05795252e+00
-1.35248333e-01 6.89305544e-01 8.36124361e-01 -6.58403099e-01
2.06242472e-01 -1.20475125e+00 -3.50780427e-01 4.42494266e-02
1.14840426e-01 -5.29870428e-02 -3.21764112e-01 7.29151666e-02]
|
[9.515697479248047, 0.25471174716949463]
|
4db95ac9-b8db-4045-9ac7-39497a8b49e6
|
non-autoregressive-sign-language-production
|
2208.06183
| null |
https://arxiv.org/abs/2208.06183v1
|
https://arxiv.org/pdf/2208.06183v1.pdf
|
Non-Autoregressive Sign Language Production via Knowledge Distillation
|
Sign Language Production (SLP) aims to translate expressions in spoken language into corresponding ones in sign language, such as skeleton-based sign poses or videos. Existing SLP models are either AutoRegressive (AR) or Non-Autoregressive (NAR). However, AR-SLP models suffer from regression to the mean and error propagation during decoding. NSLP-G, a NAR-based model, resolves these issues to some extent but engenders other problems. For example, it does not consider target sign lengths and suffers from false decoding initiation. We propose a novel NAR-SLP model via Knowledge Distillation (KD) to address these problems. First, we devise a length regulator to predict the end of the generated sign pose sequence. We then adopt KD, which distills spatial-linguistic features from a pre-trained pose encoder to alleviate false decoding initiation. Extensive experiments show that the proposed approach significantly outperforms existing SLP models in both Frechet Gesture Distance and Back-Translation evaluation.
|
['Jong C. Park', 'Suk min Cho', 'Jung Ho Kim', 'Eui Jun Hwang']
|
2022-08-12
| null | null | null | null |
['sign-language-production']
|
['natural-language-processing']
|
[ 2.78850734e-01 2.23612972e-02 -2.68330663e-01 -5.68048239e-01
-1.06248927e+00 -4.28533643e-01 6.20558321e-01 -8.78270924e-01
-2.92174518e-01 5.92225373e-01 7.74667919e-01 -9.30181146e-02
5.51987626e-02 -3.29834968e-01 -7.01320589e-01 -7.06547022e-01
1.80728227e-01 4.81037915e-01 3.05849671e-01 -1.23247884e-01
3.50043654e-01 6.07034028e-01 -1.25537908e+00 9.81420055e-02
8.95210922e-01 6.57178402e-01 1.60601050e-01 8.74002635e-01
-1.81274265e-01 1.07972121e+00 -6.52060866e-01 -1.54418245e-01
1.80455208e-01 -7.82602549e-01 -4.89176631e-01 -8.23976696e-02
2.57500172e-01 -7.03137398e-01 -3.51126909e-01 8.03992927e-01
7.53124774e-01 4.17466350e-02 8.83713961e-01 -1.15098464e+00
-7.08205342e-01 4.74059612e-01 -5.96955061e-01 -3.43751162e-01
5.22516370e-01 2.96312451e-01 7.89996743e-01 -9.73087847e-01
6.81144714e-01 1.47280586e+00 6.52951360e-01 8.69997442e-01
-5.39699316e-01 -7.60150075e-01 1.50808722e-01 4.43233289e-02
-1.23421073e+00 -5.25464952e-01 7.21674442e-01 -3.53952974e-01
8.22922349e-01 1.45548761e-01 5.57626843e-01 1.27161467e+00
-6.89471364e-02 1.10232615e+00 1.11682999e+00 -5.97517550e-01
5.49936667e-02 -6.20747864e-01 -2.29509920e-01 7.17334986e-01
-2.55335987e-01 1.59799844e-01 -8.62223923e-01 1.46538973e-01
1.06179619e+00 -3.42040926e-01 -3.14276874e-01 -1.79371536e-01
-1.20167422e+00 3.49618942e-01 1.39273450e-01 2.73142934e-01
-5.26289105e-01 2.00409383e-01 2.78401911e-01 2.57716149e-01
-1.35914385e-01 7.68787637e-02 -4.44932461e-01 -6.52395487e-01
-6.97483599e-01 2.74323136e-01 6.35371327e-01 1.07795727e+00
-1.82824805e-02 3.38428646e-01 -3.76739353e-01 9.65317070e-01
8.07300210e-01 6.95216775e-01 6.38423264e-01 -7.92326510e-01
6.88195169e-01 3.50303411e-01 3.20890695e-02 -5.84345341e-01
-3.46933365e-01 -6.49200231e-02 -5.32433331e-01 3.81131411e-01
5.86822033e-01 -2.16538295e-01 -1.43955064e+00 1.74999404e+00
6.89835027e-02 3.29164684e-01 1.88272327e-01 1.47338927e+00
7.40679741e-01 6.36684716e-01 3.06615710e-01 -1.48596302e-01
1.00537455e+00 -1.05880404e+00 -8.48642886e-01 -4.26842794e-02
6.12579107e-01 -8.06324244e-01 1.20323575e+00 4.45936084e-01
-9.56099808e-01 -2.29718342e-01 -7.26828754e-01 -2.51026273e-01
1.42414331e-01 4.54069316e-01 2.89186478e-01 5.37399530e-01
-8.22558820e-01 2.64692307e-01 -1.07233298e+00 -3.38702708e-01
1.95316691e-02 1.88529745e-01 -2.02432588e-01 8.97564068e-02
-9.53100860e-01 9.65167701e-01 1.26754925e-01 5.72420239e-01
-5.53728521e-01 -3.27726990e-01 -9.12033975e-01 -3.65424573e-01
2.50827551e-01 -4.37090665e-01 1.41275787e+00 -9.90105748e-01
-2.41547823e+00 5.61530888e-01 -3.53899419e-01 -1.22742586e-01
8.79606187e-01 -6.08840168e-01 -3.85125697e-01 -7.64709041e-02
-2.44209141e-01 6.94046080e-01 9.69130635e-01 -1.15663564e+00
-5.08056879e-01 -2.45074734e-01 -1.05441034e-01 5.27563572e-01
2.32053384e-01 3.53529692e-01 -6.84916019e-01 -8.46960902e-01
4.99208987e-01 -1.02027547e+00 -8.29887316e-02 2.13697478e-01
-3.88255507e-01 -1.62762657e-01 7.60377824e-01 -1.08154893e+00
1.14477062e+00 -1.79978573e+00 4.53907996e-01 2.88239181e-01
-2.97514468e-01 5.69186091e-01 -3.75792176e-01 3.55467319e-01
2.40662932e-01 -2.99870104e-01 -2.88140923e-01 -2.33666778e-01
1.83987431e-02 6.78329170e-01 -3.77380997e-01 3.67514253e-01
2.40684345e-01 1.10865116e+00 -9.37656879e-01 -6.52006865e-01
2.88444757e-01 6.71633720e-01 -5.95592022e-01 2.34816596e-01
-3.45238209e-01 1.05173981e+00 -3.27817738e-01 7.69700885e-01
4.06092554e-01 2.22915411e-01 1.67758480e-01 -3.34150232e-02
-2.64310300e-01 2.90544420e-01 -9.87109005e-01 1.81088185e+00
-4.62547541e-01 5.23461819e-01 -1.94527462e-01 -7.04440236e-01
9.11995888e-01 3.38946849e-01 3.40436608e-01 -4.31690067e-01
2.93951809e-01 5.96840680e-01 1.35113433e-01 -8.22972476e-01
3.11142385e-01 -2.65155196e-01 6.68056980e-02 2.11166114e-01
-1.49803355e-01 -6.48166090e-02 -3.53606418e-02 -4.05761153e-01
1.00255454e+00 7.35184073e-01 1.68119967e-01 4.83948231e-01
5.84763169e-01 -2.10593164e-01 6.33203924e-01 5.16328752e-01
-2.19798177e-01 8.66206586e-01 4.57083732e-01 -8.54982883e-02
-7.35459864e-01 -1.12696910e+00 3.46919924e-01 1.15398884e+00
6.08636290e-02 -8.74348357e-02 -6.11892998e-01 -5.87183356e-01
-2.11528689e-01 8.26740742e-01 -2.17215329e-01 8.63961056e-02
-1.10429156e+00 -2.99574822e-01 9.47669387e-01 8.59707117e-01
4.27826911e-01 -1.42528903e+00 -5.32611191e-01 1.10472307e-01
-2.16404021e-01 -1.17018163e+00 -6.47165120e-01 -4.19363946e-01
-6.63926601e-01 -9.20987725e-01 -1.30409706e+00 -1.05501068e+00
7.91936815e-01 -2.90039599e-01 4.00610119e-01 -1.85981676e-01
9.01346952e-02 7.20944703e-02 -8.13051999e-01 -3.50205451e-01
-5.87437332e-01 -2.07596466e-01 1.41550019e-01 8.56492436e-04
5.08898854e-01 -4.98807490e-01 -3.39041501e-01 4.56864923e-01
-6.40048146e-01 1.53108731e-01 1.03780174e+00 9.60654020e-01
4.88462538e-01 -7.84422994e-01 4.29305941e-01 -2.04041854e-01
7.16219664e-01 8.82419124e-02 -4.72708315e-01 3.25628847e-01
-2.72833347e-01 3.63675445e-01 4.84120458e-01 -6.19644701e-01
-1.06538033e+00 2.65155256e-01 -5.78811109e-01 -5.90937495e-01
-1.07864738e-01 4.77264524e-01 -1.66156828e-01 -3.19087990e-02
3.73210132e-01 6.33342505e-01 1.75946593e-01 -6.89409614e-01
4.32865053e-01 8.64648402e-01 7.55684078e-01 -4.87401485e-01
7.17810810e-01 1.64432466e-01 -3.31451371e-02 -1.00420845e+00
-5.08544385e-01 -2.29772836e-01 -8.20911705e-01 -3.30347687e-01
6.87884629e-01 -6.50707006e-01 -7.82944918e-01 8.07687581e-01
-1.33109438e+00 -2.93976396e-01 -2.35785633e-01 8.44708741e-01
-9.91783440e-01 5.48772812e-01 -5.49605846e-01 -1.04103899e+00
-2.36851782e-01 -1.02684510e+00 1.40470672e+00 5.32495640e-02
-3.59093189e-01 -2.46857330e-01 1.11482143e-01 3.23697776e-01
1.37380987e-01 2.53409415e-01 5.84448695e-01 -2.81194270e-01
-5.32818317e-01 -3.13876778e-01 -2.65730947e-01 4.17906284e-01
3.92792188e-02 -7.01828822e-02 -5.38361251e-01 1.64303258e-02
-5.05054891e-01 -3.59566778e-01 5.82634628e-01 4.25373465e-01
7.41953731e-01 -4.26428407e-01 1.17336646e-01 6.02249920e-01
9.05135214e-01 2.28690028e-01 8.06291163e-01 1.97746381e-02
7.14267671e-01 5.53393602e-01 8.13910007e-01 3.50883037e-01
4.34022605e-01 9.20638561e-01 -1.24104472e-03 2.53267407e-01
-4.62799072e-01 -7.85084426e-01 8.50564122e-01 1.28222871e+00
-6.79351151e-01 -1.31342292e-01 -8.95749629e-01 3.97589743e-01
-2.00233889e+00 -7.14104950e-01 -1.40145257e-01 1.89220583e+00
8.69769394e-01 -9.40257832e-02 7.97943994e-02 5.73127754e-02
3.48506749e-01 7.99120516e-02 -5.90155900e-01 -4.27258283e-01
-2.77259111e-01 1.91023618e-01 6.32567585e-01 6.48607731e-01
-7.82593608e-01 1.34137857e+00 6.11731195e+00 5.06097019e-01
-1.27251542e+00 7.43085966e-02 -2.54157186e-01 -6.96152896e-02
-1.02106361e-02 -1.49640843e-01 -7.40882337e-01 4.25694942e-01
5.87612808e-01 1.76987186e-01 2.08990678e-01 7.39344120e-01
5.22045493e-01 2.00049907e-01 -9.38203156e-01 1.14367187e+00
2.58987278e-01 -7.71417975e-01 2.25019902e-01 -8.82274583e-02
5.12723267e-01 1.42550794e-02 -1.79533452e-01 5.20200014e-01
3.03621113e-01 -9.34074998e-01 9.04082179e-01 8.26192796e-01
9.92526174e-01 -3.73803169e-01 5.85974991e-01 5.40852249e-01
-1.16947889e+00 -4.30070758e-02 1.07093230e-01 -1.04094170e-01
8.29466403e-01 -2.74304718e-01 -8.07794094e-01 2.60941178e-01
9.30159688e-02 5.44740975e-01 -3.14617828e-02 9.58499134e-01
-8.27885628e-01 8.68438065e-01 -4.90000635e-01 -2.94557393e-01
3.01846713e-01 -8.15466344e-02 9.02705848e-01 1.07866359e+00
5.67191243e-01 1.54440969e-01 1.04364626e-01 5.23166597e-01
2.18067169e-01 3.13124657e-01 -4.43877488e-01 -1.31425574e-01
1.97062626e-01 3.55044335e-01 -2.80231893e-01 -2.46173799e-01
-3.19886386e-01 1.43032849e+00 -6.34628907e-02 5.59427440e-01
-8.48285377e-01 -3.05854023e-01 6.53842390e-01 -8.25334981e-04
1.75504804e-01 -5.74504912e-01 -9.16743726e-02 -1.16736674e+00
3.30903113e-01 -7.66741812e-01 -7.83844963e-02 -8.63940120e-01
-8.70357275e-01 2.87124008e-01 -7.04387873e-02 -1.62643790e+00
-9.04652774e-01 -6.88350081e-01 -2.61346906e-01 7.15546429e-01
-1.66509914e+00 -1.55495203e+00 -2.34104916e-01 3.30729008e-01
7.94825912e-01 -3.17183286e-02 7.40005970e-01 4.04165387e-01
-2.46467590e-01 8.05202544e-01 -8.75591561e-02 4.44734871e-01
6.11126840e-01 -9.80036616e-01 3.93294722e-01 7.56262064e-01
1.94466829e-01 4.99931514e-01 7.57882357e-01 -6.42265677e-01
-1.24028933e+00 -8.73598278e-01 1.11192429e+00 -2.76131332e-01
5.06698847e-01 1.35480344e-01 -6.73491240e-01 6.96134090e-01
-4.62181240e-01 -1.94688275e-01 2.20488191e-01 -2.59520769e-01
-3.27671915e-01 2.85200804e-01 -7.87323177e-01 8.79041910e-01
1.49310935e+00 -3.60596895e-01 -9.02362764e-01 2.18559265e-01
4.76818770e-01 -7.94741333e-01 -5.84373534e-01 5.67003191e-01
1.17901981e+00 -7.03425109e-01 6.98334575e-01 -7.30681002e-01
4.46718812e-01 -4.16360229e-01 -4.70454842e-02 -1.17693090e+00
8.18709284e-02 -8.17199230e-01 -4.08727825e-01 9.87994492e-01
2.74984181e-01 -4.80118096e-01 8.81262243e-01 4.23186421e-01
-2.63121247e-01 -6.31254077e-01 -1.09151554e+00 -1.12650979e+00
-1.50016900e-02 -4.76005971e-01 3.97496432e-01 4.69682574e-01
-9.34115872e-02 1.90654263e-01 -8.98968041e-01 1.91806868e-01
4.52342510e-01 -7.84993470e-02 1.10981941e+00 -8.56164813e-01
-3.28039795e-01 -4.11764324e-01 -6.25239193e-01 -1.80972564e+00
1.62721857e-01 -5.78712523e-01 6.46038234e-01 -1.64794803e+00
-3.84269565e-01 -1.10074848e-01 -3.06621622e-02 6.62519038e-01
1.17354095e-01 5.60341962e-02 2.73301780e-01 4.02454525e-01
-2.70331919e-01 6.71826065e-01 1.56820846e+00 2.35735610e-01
-5.69480658e-01 6.97595105e-02 -7.07391575e-02 9.63728368e-01
6.73446894e-01 -2.57369727e-01 -1.90672964e-01 -5.51561534e-01
-1.30462483e-01 2.22570121e-01 3.04827660e-01 -8.56305003e-01
2.33514905e-01 -1.20844267e-01 -3.23554804e-03 -7.14662433e-01
5.18044353e-01 -5.97516656e-01 -1.37472376e-01 4.70112085e-01
-3.91939789e-01 -1.34778559e-01 -2.68487155e-01 4.32872444e-01
-3.01571012e-01 -9.64614376e-02 6.55728757e-01 6.89018816e-02
-9.17939723e-01 2.69333762e-03 -3.50127727e-01 -1.06240667e-01
8.49389493e-01 -4.59306180e-01 -2.93428712e-02 -6.32064044e-01
-6.28877819e-01 2.12054417e-01 1.17935941e-01 6.97567523e-01
8.24059367e-01 -1.33290172e+00 -7.75321007e-01 3.39979202e-01
2.29429826e-01 7.80758411e-02 5.29847899e-03 1.23453546e+00
-8.30549955e-01 5.53424597e-01 -1.03412077e-01 -4.80773151e-01
-1.44435871e+00 -1.29818663e-01 7.34876022e-02 -1.16047502e-01
-9.07965422e-01 9.28861439e-01 -2.63674349e-01 -6.30997837e-01
4.31990087e-01 -4.86182094e-01 -1.13004267e-01 -3.26976240e-01
3.83540869e-01 1.99791074e-01 -4.20324892e-01 -1.18723845e+00
-2.65611976e-01 1.03116882e+00 9.82282832e-02 -4.59626108e-01
1.12644660e+00 1.40160499e-02 2.12556884e-01 4.03261900e-01
9.30833638e-01 1.10938072e-01 -1.45076823e+00 -1.84722811e-01
1.17623441e-01 -4.02236074e-01 -3.02415401e-01 -9.70002115e-01
-7.26432204e-01 8.57405543e-01 3.97365868e-01 -7.48809278e-01
1.12840784e+00 -1.09602049e-01 1.18624699e+00 4.50873017e-01
4.46176350e-01 -1.23506355e+00 -1.56650260e-01 8.21239948e-01
1.34592903e+00 -9.72455919e-01 -2.95295388e-01 -4.09576714e-01
-8.63128960e-01 1.06807613e+00 4.85236406e-01 -1.15681097e-01
5.23618639e-01 2.59303689e-01 5.49721003e-01 1.08740829e-01
-3.29904139e-01 -2.75388479e-01 4.17163104e-01 6.57720089e-01
4.16230708e-01 9.66219977e-02 -8.43208253e-01 4.23199683e-01
-4.52285767e-01 5.14007330e-01 8.34553540e-02 1.18998647e+00
-3.61412346e-01 -1.17410970e+00 -4.29232836e-01 1.60801306e-01
-4.33682688e-02 1.30219292e-02 -4.78398204e-01 6.01101518e-01
-2.49198842e-04 5.45906901e-01 -1.46416619e-01 -4.65234280e-01
5.97349226e-01 3.02502900e-01 7.36928821e-01 -2.78956622e-01
-1.50950775e-01 2.80866444e-01 6.57228082e-02 -5.02614558e-01
-5.53087592e-01 -6.83335125e-01 -1.55359185e+00 1.78957522e-01
-3.15639257e-01 -3.88272643e-01 7.23124266e-01 1.07640398e+00
1.24689735e-01 4.35557574e-01 1.83898658e-01 -9.59523082e-01
-8.56763959e-01 -1.26449943e+00 -4.26551819e-01 3.97237957e-01
3.19680780e-01 -7.10383713e-01 -1.93730563e-01 1.53043360e-01]
|
[9.194982528686523, -6.4980573654174805]
|
e2f3ed5c-0924-4ec8-84cb-f52faf76f268
|
improved-vocal-effort-transfer-vector
|
2305.02147
| null |
https://arxiv.org/abs/2305.02147v3
|
https://arxiv.org/pdf/2305.02147v3.pdf
|
Improved Vocal Effort Transfer Vector Estimation for Vocal Effort-Robust Speaker Verification
|
Despite the maturity of modern speaker verification technology, its performance still significantly degrades when facing non-neutrally-phonated (e.g., shouted and whispered) speech. To address this issue, in this paper, we propose a new speaker embedding compensation method based on a minimum mean square error (MMSE) estimator. This method models the joint distribution of the vocal effort transfer vector and non-neutrally-phonated embedding spaces and operates in a principal component analysis domain to cope with non-neutrally-phonated speech data scarcity. Experiments are carried out using a cutting-edge speaker verification system integrating a powerful self-supervised pre-trained model for speech representation. In comparison with a state-of-the-art embedding compensation method, the proposed MMSE estimator yields superior and competitive equal error rate results when tackling shouted and whispered speech, respectively.
|
['Eduardo Lleida', 'Alfonso Ortega', 'Santi Prieto', 'Iván López-Espejo']
|
2023-05-03
| null | null | null | null |
['speaker-verification']
|
['speech']
|
[ 1.64597839e-01 4.22335044e-02 5.66442870e-03 -3.24885577e-01
-9.37471867e-01 -3.36207986e-01 2.93837428e-01 -4.06353921e-01
-4.43834960e-01 3.98728579e-01 3.89939040e-01 -3.40924323e-01
1.51149612e-02 2.01464798e-02 -1.39605343e-01 -6.79485798e-01
2.03330696e-01 -1.31112054e-01 -2.03721046e-01 -9.34334546e-02
6.67146817e-02 2.96651334e-01 -1.86239576e+00 -1.49954349e-01
8.21888685e-01 9.59064662e-01 3.12954724e-01 7.78240442e-01
1.06510714e-01 2.62330562e-01 -8.04551601e-01 -7.34167099e-01
9.18824002e-02 -4.01978105e-01 -1.64951265e-01 9.05474946e-02
5.52996516e-01 -1.01210624e-01 -4.22425717e-01 1.35036981e+00
1.07435918e+00 1.43159077e-01 6.73883796e-01 -1.26457143e+00
-8.07355464e-01 5.31557798e-01 -2.14182287e-01 2.29552522e-01
2.58874178e-01 -9.27025974e-02 9.62713301e-01 -1.23658848e+00
1.34534180e-01 1.24094939e+00 7.83881068e-01 8.96819532e-01
-1.19871545e+00 -8.60289335e-01 -5.67455180e-02 3.74582201e-01
-1.41254389e+00 -1.03754640e+00 1.26071727e+00 -2.73175299e-01
7.64049828e-01 5.57913601e-01 3.42493534e-01 1.26852417e+00
1.12772979e-01 9.05787885e-01 1.04691494e+00 -4.38943535e-01
2.03546882e-01 6.73850656e-01 4.92017111e-03 4.01283860e-01
-1.46470413e-01 3.41771424e-01 -7.72389770e-01 -3.97071272e-01
1.02026328e-01 -6.09437466e-01 -5.56366563e-01 -1.91175282e-01
-8.89752686e-01 6.67909801e-01 -6.00477457e-02 4.81531829e-01
-4.63160425e-01 -2.48603582e-01 4.78132486e-01 4.41053241e-01
7.00504005e-01 6.04436547e-02 -3.60390306e-01 -2.80386001e-01
-1.47219419e+00 6.72907308e-02 8.28981698e-01 6.85171664e-01
3.09112817e-01 8.69035959e-01 -2.47533470e-02 1.29805410e+00
9.49091733e-01 7.62338161e-01 8.68341863e-01 -5.92511117e-01
3.67269248e-01 -4.17483225e-02 -5.53288050e-02 -8.14661980e-01
-2.08981950e-02 -7.16220617e-01 -8.23426664e-01 2.40233779e-01
1.12157516e-01 -1.42528594e-01 -5.24877131e-01 1.87307346e+00
3.35314482e-01 1.41828254e-01 5.44997692e-01 9.08346236e-01
6.97660387e-01 7.17834532e-01 -1.63237855e-01 -5.34774005e-01
1.56597412e+00 -9.63685453e-01 -1.38000607e+00 -3.81144941e-01
-1.42296568e-01 -1.04462218e+00 1.28485274e+00 4.30595785e-01
-1.00419152e+00 -7.70348549e-01 -1.26705289e+00 1.15443125e-01
-2.27937624e-01 2.04689473e-01 -1.76838443e-01 1.31474876e+00
-1.13350165e+00 3.33293766e-01 -4.43337053e-01 1.41977534e-01
-8.44501182e-02 1.85653001e-01 -4.36341435e-01 2.72534698e-01
-1.39214373e+00 1.06202149e+00 5.32007925e-02 2.28274852e-01
-5.90038300e-01 -7.36261368e-01 -1.06418312e+00 1.78183600e-01
-2.53136158e-01 -3.08099210e-01 1.31887186e+00 -8.19623232e-01
-2.14932108e+00 5.31221032e-01 -3.69706661e-01 -3.52818877e-01
4.87002194e-01 -2.41428390e-01 -9.11351979e-01 -8.91510174e-02
-2.47749522e-01 3.19637924e-01 1.62395811e+00 -1.10140002e+00
-1.91883743e-01 -5.34344554e-01 -8.96481454e-01 1.55465290e-01
-7.55605459e-01 3.52765739e-01 1.67895913e-01 -7.84563482e-01
3.47853124e-01 -6.92482829e-01 4.01802838e-01 -6.18881620e-02
-3.87988627e-01 -5.91820776e-02 1.30329669e+00 -1.27126145e+00
1.25433004e+00 -2.53320122e+00 1.27512857e-01 -1.03084661e-01
-2.99362957e-01 5.09193659e-01 -4.37733755e-02 3.55871111e-01
-2.10356772e-01 -1.85949862e-01 -3.55300307e-01 -1.00468552e+00
5.24720550e-01 3.98302488e-02 -4.23398286e-01 6.84254408e-01
1.71174243e-01 2.79122680e-01 -6.72349632e-01 -4.65130776e-01
2.92373449e-01 1.17839050e+00 -3.75392467e-01 3.25955302e-01
6.20998859e-01 7.23626092e-02 3.30476642e-01 6.81270599e-01
9.21069026e-01 7.01302230e-01 -1.23371474e-01 -2.21677318e-01
-1.60580650e-01 3.32053512e-01 -1.43932354e+00 1.38310528e+00
-5.78308821e-01 9.52478528e-01 7.82840669e-01 -5.96470654e-01
1.23817372e+00 9.41159785e-01 4.60163094e-02 -1.86665669e-01
1.44924715e-01 5.04851878e-01 7.85696059e-02 -3.26054394e-01
5.66671550e-01 -6.48341477e-01 2.28782788e-01 6.88032955e-02
3.33999544e-01 -3.99173915e-01 -3.83964539e-01 -3.94755095e-01
6.24026477e-01 -3.70764881e-01 3.02806497e-01 -3.73846591e-01
1.04653776e+00 -1.03603077e+00 4.79073375e-01 2.22218782e-01
-8.54207814e-01 5.10464907e-01 9.52938348e-02 4.38897789e-01
-9.86865938e-01 -1.28044295e+00 -5.43247342e-01 9.09393132e-01
-1.92825481e-01 -2.04208031e-01 -9.29604471e-01 -3.51855397e-01
9.40402597e-02 9.12565410e-01 -1.41157150e-01 -3.11996579e-01
-1.31520286e-01 -4.00565803e-01 8.72622371e-01 3.03517252e-01
4.81537342e-01 -8.81964326e-01 -8.51259604e-02 2.73132414e-01
-2.00381190e-01 -9.73779082e-01 -9.36572015e-01 1.18178770e-01
-5.18870592e-01 -2.76887208e-01 -1.10275936e+00 -9.15360749e-01
2.08901644e-01 -1.15204521e-03 4.86357331e-01 -6.12709403e-01
7.78689384e-02 3.80091965e-01 3.13056484e-02 -6.11112356e-01
-7.68171072e-01 -1.44586504e-01 6.22942865e-01 6.00735962e-01
3.41230839e-01 -5.02289653e-01 -3.50329250e-01 5.62576890e-01
-5.41027188e-01 -5.38703918e-01 4.18164223e-01 1.11678171e+00
1.79704159e-01 2.40019113e-01 1.06362092e+00 6.66497722e-02
1.13150454e+00 -1.48406938e-01 -5.21041095e-01 -6.77661672e-02
-8.60255599e-01 6.48373878e-03 4.76743311e-01 -7.04651475e-01
-1.17666543e+00 -3.02004702e-02 -6.68452501e-01 -4.82254416e-01
-1.19544432e-01 2.41887584e-01 -4.14559126e-01 -8.90404135e-02
4.45818335e-01 6.70414448e-01 3.85185808e-01 -6.20730758e-01
2.13204622e-01 1.62854242e+00 7.00265527e-01 -5.73714543e-03
9.13873613e-01 -6.78787455e-02 -5.34965634e-01 -1.25130749e+00
-1.61174953e-01 -7.45181143e-01 -3.48665982e-01 -3.53919059e-01
5.28935134e-01 -8.07847023e-01 -7.54242957e-01 7.91921616e-01
-1.12618792e+00 3.84711176e-02 -2.66922563e-01 8.73893678e-01
-5.40318489e-01 7.92924762e-01 -6.29133701e-01 -1.63026834e+00
-7.49108851e-01 -1.17640090e+00 1.13454020e+00 6.40262961e-02
-2.26601496e-01 -7.48398364e-01 2.63002694e-01 5.76983988e-01
8.73504281e-01 -4.46184069e-01 3.40261310e-01 -6.62344337e-01
2.05094457e-01 -6.13896191e-01 1.55341953e-01 1.19494677e+00
2.16816381e-01 -4.19752039e-02 -1.51990151e+00 -3.86131883e-01
5.95799208e-01 1.89815499e-02 5.55564821e-01 1.38104826e-01
6.73119307e-01 -5.43285191e-01 3.19227904e-01 3.68472815e-01
8.21431160e-01 4.16073576e-02 5.13118386e-01 -6.05612919e-02
5.43093719e-02 6.96521759e-01 3.21756423e-01 3.77546757e-01
1.37170628e-01 7.18825698e-01 1.98608160e-01 2.39807591e-01
-3.06936353e-01 -2.96808660e-01 9.12707269e-01 1.45931935e+00
4.06705976e-01 -1.85228184e-01 -5.77746868e-01 5.17889678e-01
-1.23404002e+00 -1.25818241e+00 1.99614674e-01 2.09300113e+00
9.53705668e-01 8.19153786e-02 2.38166656e-02 1.02015340e+00
7.88050890e-01 3.90374094e-01 -5.40870905e-01 -6.80015802e-01
-3.49561423e-01 2.38904823e-02 1.21427231e-01 7.42750525e-01
-1.11171246e+00 5.22089839e-01 6.25067043e+00 9.49934661e-01
-1.45177853e+00 4.41844583e-01 -1.09630659e-01 3.34491968e-01
-1.16329938e-01 -6.53130114e-01 -6.60022795e-01 6.44772112e-01
1.36530304e+00 -2.57128656e-01 4.54457253e-01 1.15909529e+00
1.29490420e-01 4.78778005e-01 -7.83528090e-01 1.31663597e+00
5.11211216e-01 -5.50123394e-01 -5.20368695e-01 6.84053749e-02
2.17977151e-01 1.16485721e-02 5.54187477e-01 5.12565255e-01
-5.75170159e-01 -6.08089685e-01 9.58374560e-01 1.86299205e-01
7.38778710e-01 -5.44429004e-01 8.17694902e-01 5.30476093e-01
-1.13083780e+00 -2.75559068e-01 -2.43953392e-01 2.15735540e-01
1.88562617e-01 5.05309463e-01 -9.66982543e-01 3.19662124e-01
3.62610579e-01 2.42653236e-01 -8.94781947e-03 8.80299687e-01
-2.18541026e-01 7.09737778e-01 -1.84676439e-01 6.98045129e-03
-2.32682616e-01 1.00006253e-01 9.35743272e-01 1.43203855e+00
5.39923906e-01 -3.75383228e-01 -4.56443071e-01 8.04869950e-01
7.99381286e-02 2.68890262e-01 -4.57462251e-01 -1.19755775e-01
4.44119930e-01 1.19089830e+00 4.33996096e-02 -2.01419041e-01
-2.25982577e-01 1.11753500e+00 -2.06216142e-01 3.93832713e-01
-7.36266434e-01 -6.45018160e-01 9.71051157e-01 -1.16884778e-03
2.83094466e-01 -1.74078196e-01 -1.15645066e-01 -1.28457880e+00
1.01603501e-01 -9.73129690e-01 -1.00636937e-01 -7.38787770e-01
-1.46343279e+00 8.72724056e-01 -2.24156663e-01 -1.13732088e+00
-5.18661141e-01 -6.35519683e-01 -5.71030617e-01 1.23329580e+00
-1.85032511e+00 -7.99050629e-01 5.67878522e-02 5.53416908e-01
6.48717463e-01 -5.42814612e-01 1.09268641e+00 6.55764163e-01
-5.58052421e-01 1.09161901e+00 3.23686242e-01 -7.19938725e-02
6.85146332e-01 -1.23858535e+00 3.74209195e-01 7.62436509e-01
4.67459597e-02 4.96961057e-01 9.70197260e-01 -1.34832308e-01
-1.43257940e+00 -6.79203987e-01 1.19418085e+00 2.43257415e-02
4.62200612e-01 -7.05640674e-01 -1.16376138e+00 1.22370563e-01
2.92035878e-01 1.83043368e-02 9.48155940e-01 -1.21493272e-01
-6.60982788e-01 -3.33585858e-01 -1.52339423e+00 3.76183778e-01
5.06763935e-01 -1.17042232e+00 -1.01916587e+00 -4.19045575e-02
4.45791245e-01 -9.18538198e-02 -8.04809988e-01 3.60625535e-01
6.59757912e-01 -7.45134413e-01 1.08754814e+00 -2.44221225e-01
-3.57425511e-01 -2.53600359e-01 -5.67314804e-01 -1.54296875e+00
-9.80286226e-02 -9.36189294e-01 -3.69486839e-01 1.38694477e+00
3.97973269e-01 -7.85389662e-01 6.89587057e-01 3.58439833e-01
-3.99666309e-01 -3.48988295e-01 -1.57880521e+00 -1.09270871e+00
-1.29735813e-01 -5.38137317e-01 3.27316880e-01 7.53254294e-01
4.51016396e-01 4.12318438e-01 -6.14802122e-01 3.56346548e-01
8.69447827e-01 -4.56433862e-01 4.65069562e-01 -1.13060820e+00
-1.93550617e-01 -5.16586959e-01 -7.36740649e-01 -9.04709816e-01
6.21028423e-01 -8.13488185e-01 3.89725804e-01 -9.87808883e-01
-3.73758316e-01 1.39951035e-01 -4.22736913e-01 8.69610235e-02
-2.54647166e-01 1.11871339e-01 1.30917281e-01 -2.17287689e-01
2.37609282e-01 9.85911191e-01 6.61067069e-01 -2.82983482e-01
-1.04694709e-01 2.68049836e-01 -3.57046217e-01 5.72919548e-01
8.31723034e-01 -3.70310396e-01 -4.64188635e-01 1.06274128e-01
-5.31795561e-01 1.50007755e-01 1.55357137e-01 -1.18048680e+00
2.45699018e-01 5.01976073e-01 -1.03519492e-01 -5.58272183e-01
9.89824712e-01 -8.47252548e-01 -1.80599242e-02 4.99995232e-01
-3.91331077e-01 -1.26483083e-01 2.06565917e-01 5.15002370e-01
-5.75422525e-01 -4.14151311e-01 1.01774299e+00 3.92733008e-01
-9.92807150e-02 -1.50400817e-01 -6.19844139e-01 -2.56323963e-01
6.61041617e-01 -2.89010286e-01 -2.20143616e-01 -6.08287096e-01
-6.34179294e-01 -2.98975557e-01 3.73577140e-02 4.89433587e-01
9.55861807e-01 -1.37391233e+00 -9.59727883e-01 8.67394030e-01
-3.15872878e-02 -8.50608110e-01 5.71319044e-01 7.22068071e-01
1.07728586e-01 4.56465811e-01 1.53825907e-02 -4.96495932e-01
-1.61018550e+00 5.39235234e-01 4.20267493e-01 2.41582096e-01
-1.48366734e-01 8.52220654e-01 -2.49385893e-01 -7.14737177e-01
6.39688313e-01 -1.77963093e-01 2.29653120e-02 1.68130144e-01
7.28311062e-01 4.72974479e-01 3.46781433e-01 -1.10453069e+00
-4.53178316e-01 4.07705724e-01 2.88237780e-01 -5.85129619e-01
1.00291693e+00 -2.74283767e-01 2.10941866e-01 6.78524256e-01
1.45820558e+00 3.89435679e-01 -9.64779913e-01 -2.55171269e-01
-1.18329756e-01 -4.61413860e-01 5.49160898e-01 -5.80116868e-01
-7.82204747e-01 1.18978405e+00 9.68080997e-01 3.85203362e-01
9.41446304e-01 -3.98355246e-01 7.87325978e-01 1.45567253e-01
-1.86603013e-02 -1.18401194e+00 -2.95301042e-02 2.85937995e-01
1.14255309e+00 -1.24640346e+00 -4.33332562e-01 -2.63343871e-01
-6.48702860e-01 1.01862180e+00 2.03403369e-01 3.25407416e-01
1.02725613e+00 3.63850415e-01 4.15030032e-01 3.89484584e-01
-6.63507462e-01 6.54307976e-02 6.12275839e-01 7.55400956e-01
4.85480696e-01 4.49507125e-02 -7.65634775e-02 6.46719635e-01
-5.42231202e-01 -3.90161783e-01 2.37863675e-01 5.49436510e-01
-4.54244614e-01 -9.50207055e-01 -7.99548447e-01 -3.53204831e-02
-4.95539188e-01 -1.81075066e-01 -2.47614637e-01 3.19582671e-01
-1.76106989e-01 1.31711650e+00 -9.71949026e-02 -5.85760415e-01
5.97763658e-01 7.23638117e-01 6.83036298e-02 -2.34108612e-01
-6.34675920e-01 3.35750312e-01 -3.49551216e-02 -2.72039995e-02
-1.87307492e-01 -8.62610877e-01 -8.04121137e-01 -1.34713545e-01
-8.16544771e-01 2.65225023e-01 1.47601867e+00 4.04473752e-01
2.52278298e-01 4.00920242e-01 8.45591605e-01 -8.74496281e-01
-1.36641097e+00 -1.34258127e+00 -8.29490781e-01 3.03038776e-01
7.79161751e-01 -5.44566333e-01 -9.76060987e-01 -1.50919482e-01]
|
[14.488577842712402, 6.090006351470947]
|
e38cc29d-d356-4839-a2f6-12845fbbea26
|
event-transformer
|
2204.05172
| null |
https://arxiv.org/abs/2204.05172v1
|
https://arxiv.org/pdf/2204.05172v1.pdf
|
Event Transformer
|
The event camera is a bio-vision inspired camera with high dynamic range, high response speed, and low power consumption, recently attracting extensive attention for its use in vast vision tasks. Unlike the conventional cameras that output intensity frame at a fixed time interval, event camera records the pixel brightness change (a.k.a., event) asynchronously (in time) and sparsely (in space). Existing methods often aggregate events occurred in a predefined temporal duration for downstream tasks, which apparently overlook varying behaviors of fine-grained temporal events. This work proposes the Event Transformer to directly process the event sequence in its native vectorized tensor format. It cascades a Local Transformer (LXformer) for exploiting the local temporal correlation, a Sparse Conformer (SCformer) for embedding the local spatial similarity, and a Global Transformer (GXformer) for further aggregating the global information in a serial means to effectively characterize the time and space correlations from input raw events for the generation of effective spatiotemporal features used for tasks. %In both LXformer and SCformer, Experimental studies have been extensively conducted in comparison to another fourteen existing algorithms upon five different datasets widely used for classification. Quantitative results report the state-of-the-arts classification accuracy and the least computational resource requirements, of the Event Transformer, making it practically attractive for event-based vision tasks.
|
['Zhan Ma', 'M. Salman Asif', 'Zhihao LI']
|
2022-04-11
| null | null | null | null |
['event-based-vision']
|
['computer-vision']
|
[ 4.53431487e-01 -8.56036365e-01 1.17703013e-01 -2.77396381e-01
-3.17794740e-01 -3.40660304e-01 9.47956860e-01 1.31525472e-01
-4.97858196e-01 4.91950423e-01 2.84242809e-01 2.59759724e-01
-3.53005797e-01 -5.59304893e-01 -3.44250172e-01 -1.18170404e+00
-2.60615975e-01 -2.73742467e-01 3.58706862e-01 3.18385035e-01
3.59721333e-01 4.39433604e-01 -1.75429869e+00 4.40958679e-01
2.84573197e-01 1.40453744e+00 3.34676862e-01 6.80280447e-01
9.08872038e-02 1.24831450e+00 -3.32073063e-01 1.18591660e-03
2.16048166e-01 -2.45550558e-01 -2.91037798e-01 1.51487455e-01
4.13822532e-01 -2.82054007e-01 -8.26848149e-01 8.45465779e-01
2.66251236e-01 3.38331968e-01 3.60415936e-01 -1.24160647e+00
-4.62716997e-01 2.21013203e-01 -5.59627891e-01 9.90028858e-01
4.09288913e-01 4.99423504e-01 7.28174269e-01 -1.10499322e+00
6.62709951e-01 8.09465408e-01 4.66927141e-01 1.59592312e-02
-9.97767627e-01 -4.89399105e-01 6.26808926e-02 7.36999273e-01
-1.16502082e+00 -1.68397978e-01 9.27365422e-01 -6.02495551e-01
1.10017729e+00 2.91578114e-01 9.33793664e-01 1.28306484e+00
5.66573739e-01 5.48906267e-01 1.27411580e+00 5.81494197e-02
4.89204526e-01 -5.46205103e-01 2.89598167e-01 4.43103373e-01
-1.75366059e-01 3.21679384e-01 -1.25866640e+00 -3.56755368e-02
8.38047564e-01 6.68075144e-01 -4.58982289e-01 9.83395427e-02
-1.79802942e+00 5.11464477e-01 4.65799123e-01 1.78426698e-01
-8.60588253e-01 3.18531752e-01 5.24890423e-01 2.08609164e-01
2.50685930e-01 -7.73374783e-03 -2.66139597e-01 -4.93836910e-01
-8.72874975e-01 -4.01530741e-03 5.01801848e-01 6.96537137e-01
6.85841382e-01 2.16294959e-01 -3.89788330e-01 3.58526468e-01
1.67596259e-03 5.42341709e-01 8.58096719e-01 -8.79576623e-01
1.77083567e-01 6.08754158e-01 -8.47465768e-02 -1.11547756e+00
-2.36580595e-01 -3.34102511e-01 -1.18671417e+00 -4.18883488e-02
3.20493937e-01 1.69858828e-01 -6.86811507e-01 1.49508893e+00
4.34226096e-01 9.60515261e-01 -2.04173759e-01 1.06753004e+00
7.23008871e-01 9.98379886e-01 8.85010883e-02 -5.75744152e-01
1.72643793e+00 -5.19503117e-01 -5.79537094e-01 4.30026576e-02
-1.60595447e-01 -7.27238119e-01 8.87205422e-01 5.74253261e-01
-7.81518281e-01 -6.68479502e-01 -8.19072902e-01 5.41039417e-03
-3.71042907e-01 1.66571662e-01 9.49275315e-01 1.26873985e-01
-7.98327982e-01 4.43531901e-01 -1.04534316e+00 -3.81403774e-01
3.67498368e-01 7.34594911e-02 -5.36950111e-01 1.64367836e-02
-8.73943985e-01 4.33008850e-01 2.64880657e-01 2.63159484e-01
-1.00108421e+00 -8.84488285e-01 -6.09024584e-01 -3.89019214e-02
7.11942604e-03 -5.27000606e-01 9.14722085e-01 -6.91043794e-01
-1.47114778e+00 5.08568227e-01 -2.72204876e-01 -6.42824471e-01
1.85664833e-01 4.02187221e-02 -3.14429194e-01 5.85608959e-01
3.90956476e-02 2.92656690e-01 1.21058309e+00 -3.77924055e-01
-7.78908432e-01 -4.42896217e-01 -1.56907752e-01 1.92290887e-01
-5.56301951e-01 6.50124773e-02 -4.76578981e-01 -9.43709075e-01
2.46540263e-01 -6.54836774e-01 -5.83036877e-02 6.95581436e-02
1.15999818e-01 -2.64309227e-01 1.01671457e+00 -2.34384850e-01
1.20573521e+00 -2.37094903e+00 2.26570144e-01 -1.65846467e-01
2.94017911e-01 2.29809713e-02 1.44991770e-01 5.05800486e-01
-1.67229667e-01 -6.23706222e-01 -2.58732177e-02 -1.51295096e-01
-5.31983137e-01 1.46165505e-01 -7.53110766e-01 7.91428387e-01
3.10228676e-01 7.71908045e-01 -1.18575323e+00 -4.35242295e-01
8.03089261e-01 5.72916389e-01 -1.64229810e-01 1.28720492e-01
1.25424176e-01 4.78333473e-01 -6.21775270e-01 5.94587684e-01
2.00834796e-01 -4.48707342e-01 -4.14790690e-01 -8.40058088e-01
-4.43387121e-01 -1.09538473e-01 -1.09564674e+00 1.66157901e+00
-2.85167456e-01 9.14982498e-01 -4.76941586e-01 -9.11661506e-01
8.02626193e-01 3.29417974e-01 9.95446980e-01 -7.57580400e-01
2.38115266e-01 -9.94452015e-02 -3.54570270e-01 -5.11971772e-01
3.80654573e-01 1.43108204e-01 3.69721651e-02 1.75383151e-01
3.14197809e-01 2.76846796e-01 2.39431843e-01 2.55061835e-01
1.47492456e+00 3.48682068e-02 4.82956260e-01 -3.52213942e-02
4.23904836e-01 -2.89541818e-02 3.41427475e-01 5.19629478e-01
-4.25185174e-01 6.44730985e-01 -7.42515624e-02 -7.82591879e-01
-7.51160026e-01 -1.29884815e+00 -1.69544309e-01 8.71913493e-01
3.65587533e-01 -4.49856281e-01 -2.43125275e-01 -7.52262473e-02
-2.43870586e-01 2.18646109e-01 -6.64037466e-01 -2.32108533e-01
-5.49881876e-01 -8.45647335e-01 3.94374877e-01 6.76412404e-01
7.44627476e-01 -1.01020443e+00 -1.29941666e+00 4.32419628e-01
-1.95813164e-01 -1.38035452e+00 -5.66985190e-01 2.08174467e-01
-8.18005621e-01 -8.56564105e-01 -3.62497270e-01 -4.32285964e-01
3.90752286e-01 5.34085333e-01 6.49336934e-01 -5.85679889e-01
-7.44447112e-01 6.28960133e-01 -4.03831452e-01 -2.74015486e-01
6.12972438e-01 -7.39687026e-01 -5.54354861e-02 7.00364947e-01
4.23397809e-01 -8.29135180e-01 -1.02993894e+00 1.61181495e-01
-1.08450747e+00 1.46020502e-01 5.43603718e-01 9.51141477e-01
7.89924324e-01 2.40620002e-02 2.04474226e-01 -2.91437447e-01
1.90547824e-01 -5.04634023e-01 -4.88943338e-01 -1.19880438e-01
-1.94640234e-01 -3.45418453e-01 8.00876379e-01 -9.46877897e-01
-1.14145029e+00 2.41370991e-01 4.95573491e-01 -8.42144966e-01
6.32040668e-04 5.50365329e-01 3.18504900e-01 -3.24276648e-02
5.36814809e-01 7.20974088e-01 -2.79463828e-01 -1.13309875e-01
2.88251996e-01 3.17135036e-01 9.39523518e-01 -4.22678590e-01
5.44212282e-01 1.06749070e+00 1.93684161e-01 -1.05179131e+00
-4.74963456e-01 -9.20112848e-01 -3.22445005e-01 -6.03197873e-01
9.58874345e-01 -1.16276729e+00 -7.52474368e-01 8.03946853e-01
-9.19397175e-01 9.20514110e-04 -6.01305246e-01 7.58508265e-01
-4.00485009e-01 2.25077704e-01 -6.81154788e-01 -5.87447524e-01
-2.20358014e-01 -7.89889574e-01 1.25128615e+00 5.03026366e-01
-1.09189183e-01 -7.91718364e-01 3.11316606e-02 4.04827595e-02
2.60155141e-01 4.91540402e-01 2.63568968e-01 -1.70870960e-01
-9.19527531e-01 -2.07120270e-01 -3.02577913e-01 2.55555864e-02
1.41875744e-01 3.59680429e-02 -8.97907615e-01 -1.09208450e-01
5.04568696e-01 5.15074357e-02 7.84255564e-01 6.11547709e-01
1.19392121e+00 -1.01943508e-01 -1.84536248e-01 7.69669950e-01
1.55204713e+00 3.08328122e-01 6.15234613e-01 1.46166742e-01
6.62009358e-01 1.82581142e-01 5.36600292e-01 9.41352367e-01
3.44112605e-01 5.99604666e-01 2.64496684e-01 1.74711376e-01
-1.81469262e-01 5.22352010e-02 6.36329830e-01 8.23302984e-01
-2.80689836e-01 -1.59413442e-01 -6.71932697e-01 7.66480386e-01
-1.96574640e+00 -1.36339569e+00 -2.82013476e-01 2.09120369e+00
7.37105727e-01 -1.04006588e-01 -2.25245938e-01 3.00785661e-01
6.00087583e-01 5.02831221e-01 -6.30401015e-01 4.48387861e-02
-2.93428510e-01 3.97410184e-01 4.74791646e-01 -1.19538873e-01
-1.05373895e+00 5.35512745e-01 5.15031624e+00 7.02427745e-01
-1.48325205e+00 2.33343884e-01 3.65967304e-01 -4.28743720e-01
2.56418377e-01 7.06289560e-02 -5.67740023e-01 8.15904260e-01
9.25411224e-01 -1.88883170e-01 4.68759090e-01 4.92327601e-01
6.03147030e-01 -2.09616750e-01 -1.06766880e+00 1.50784099e+00
-7.99693819e-03 -1.43878448e+00 1.86926231e-01 -1.89046428e-01
5.38923860e-01 1.64983273e-01 1.69877969e-02 -2.10465148e-01
-5.13274260e-02 -6.18119836e-01 8.77802610e-01 8.60809565e-01
6.23666108e-01 -1.64796114e-01 3.07953715e-01 -1.90866948e-03
-1.65316033e+00 -1.42334878e-01 -7.96446428e-02 -3.74154747e-01
2.39331931e-01 9.00868118e-01 -3.73060971e-01 5.32492876e-01
1.07285583e+00 1.28802407e+00 -3.95139366e-01 9.18539107e-01
2.30279118e-01 7.44688511e-01 -4.98669595e-01 -3.10634132e-02
3.77118051e-01 -3.30890030e-01 7.36775577e-01 1.15273201e+00
2.54932314e-01 4.73066449e-01 2.29567274e-01 4.91720825e-01
3.95716757e-01 -4.00719464e-01 -3.50296885e-01 -2.03441568e-02
5.43363571e-01 1.44516230e+00 -8.12170327e-01 -4.24134910e-01
-5.53738475e-01 1.19278550e+00 -1.05947137e-01 4.38482732e-01
-9.69929039e-01 -2.91214377e-01 5.74302018e-01 -2.62965977e-01
5.02274454e-01 -6.13641977e-01 -1.82183236e-01 -1.40287173e+00
3.53401124e-01 -4.07879949e-01 5.27160347e-01 -1.05000579e+00
-1.44330263e+00 4.74123865e-01 3.11958212e-02 -1.62946188e+00
-1.06090851e-01 -5.50672650e-01 -6.57625079e-01 6.37599409e-01
-1.38791454e+00 -1.08702505e+00 -8.26677501e-01 1.31354570e+00
8.35536957e-01 -1.17037380e-02 3.30653131e-01 3.38747889e-01
-6.04358733e-01 -5.47531582e-02 1.53663959e-02 4.24151681e-02
5.50456047e-01 -1.07016075e+00 -7.68967271e-02 1.13846624e+00
3.62258136e-01 5.82406342e-01 4.98172730e-01 -5.23200691e-01
-2.16781592e+00 -1.29038393e+00 6.28065228e-01 -2.72780061e-01
9.88473058e-01 -3.45471948e-02 -5.78458905e-01 5.00735521e-01
-2.60824524e-03 5.54509044e-01 3.34452510e-01 -5.99500060e-01
-4.64012325e-01 -6.39981925e-01 -8.17120612e-01 4.86305684e-01
9.42112029e-01 -8.76779795e-01 -6.55186892e-01 2.79406011e-01
4.36274439e-01 -2.43313476e-01 -9.35163617e-01 1.04307130e-01
6.19939566e-01 -1.07563210e+00 1.04401994e+00 -1.11966610e-01
5.07973790e-01 -7.92811453e-01 -2.91490823e-01 -8.24051559e-01
-5.17359614e-01 -6.73802257e-01 -3.18548054e-01 1.07155740e+00
-3.61148655e-01 -6.59762084e-01 4.23207581e-01 2.20019177e-01
-9.92730912e-03 -6.26991093e-01 -1.17397904e+00 -6.53309464e-01
-8.70645583e-01 -5.76805174e-01 1.93583250e-01 7.64969409e-01
-1.89888105e-01 2.18150362e-01 -2.57204711e-01 3.20878088e-01
8.08995962e-01 3.37692529e-01 2.71228731e-01 -1.01303315e+00
-1.70279115e-01 -2.64893711e-01 -9.32031810e-01 -8.49467635e-01
-4.44703013e-01 -5.26969314e-01 -2.62502432e-02 -9.27752733e-01
3.05824131e-01 1.01452738e-01 -5.39572656e-01 2.24126667e-01
-2.30066195e-01 4.41554219e-01 1.54369742e-01 5.59396386e-01
-5.14792144e-01 5.00236630e-01 8.01497400e-01 -4.31534648e-02
-1.44105516e-02 -4.46017653e-01 -5.20826206e-02 6.60938144e-01
2.66295582e-01 -1.80785149e-01 -6.56661570e-01 -3.52086663e-01
-9.88860726e-02 2.19211936e-01 9.05114710e-01 -1.25236261e+00
7.20013499e-01 -2.77833670e-01 5.00172079e-01 -5.28766811e-01
5.38977444e-01 -8.72435689e-01 4.89778399e-01 2.11948618e-01
-7.27918595e-02 2.58084387e-01 -1.01642676e-01 1.03143692e+00
-5.53240716e-01 3.03455174e-01 5.57039142e-01 -7.38133937e-02
-1.22120535e+00 6.81767702e-01 -4.21266079e-01 -1.90127358e-01
1.20895767e+00 -5.83875835e-01 -2.79462487e-01 -3.45957465e-02
-4.51828837e-01 -2.73491114e-01 2.04752624e-01 3.71472001e-01
9.24542248e-01 -1.19175506e+00 -4.57093090e-01 3.47597361e-01
2.37325624e-01 -3.14097047e-01 4.41581935e-01 9.97458816e-01
-3.20925146e-01 1.14843853e-01 -4.21492904e-01 -1.10503721e+00
-1.03017020e+00 4.24034953e-01 3.45242508e-02 -2.77052564e-03
-1.01731145e+00 8.00310850e-01 4.12402213e-01 6.21614635e-01
3.89481895e-02 -6.07120693e-01 -1.71524957e-01 2.76380450e-01
8.33994687e-01 4.80656266e-01 8.26534778e-02 -5.60966074e-01
-4.24914032e-01 7.08781779e-01 2.41833985e-01 -2.02884004e-01
1.37458754e+00 -2.77828902e-01 -1.88266650e-01 6.55013621e-01
1.15027237e+00 -3.81644368e-01 -1.54960036e+00 -4.09791052e-01
-1.20629303e-01 -5.09489536e-01 5.10771096e-01 -4.25580889e-01
-1.11297607e+00 7.53327429e-01 7.49652028e-01 2.03892171e-01
1.73650527e+00 -3.01072627e-01 8.34346771e-01 1.05887994e-01
5.05465984e-01 -6.17544830e-01 2.88680285e-01 3.99578601e-01
6.97661698e-01 -9.49844897e-01 7.53159374e-02 -2.59903669e-01
-6.03079319e-01 1.28095078e+00 1.78085431e-01 -3.96114439e-01
7.72853613e-01 3.93290043e-01 -2.68957019e-01 -3.17781657e-01
-1.01501334e+00 -6.78448379e-02 1.87451914e-01 4.47910517e-01
-4.36995402e-02 -7.52144381e-02 -1.28728881e-01 4.08644348e-01
-1.29530411e-02 2.18409792e-01 3.91642451e-01 9.46468294e-01
-9.64054018e-02 -5.22108972e-01 -3.21324378e-01 5.83520949e-01
-2.97775298e-01 -1.36086151e-01 -9.45876073e-03 3.79106939e-01
2.91105002e-01 9.38414097e-01 4.00706679e-01 -3.96800995e-01
1.77416652e-01 -2.46734723e-01 4.23303217e-01 -2.07931042e-01
-6.95459604e-01 4.59893905e-02 -3.62642109e-01 -1.01494908e+00
-8.37162077e-01 -9.52825844e-01 -1.02385533e+00 -2.66627949e-02
3.15653980e-02 -3.74760509e-01 5.57597816e-01 7.66599357e-01
4.75820899e-01 5.08310437e-01 7.70250618e-01 -9.99053299e-01
-2.58590370e-01 -6.38011932e-01 -5.16796827e-01 6.94586515e-01
5.01191080e-01 -6.77410960e-01 -4.74935532e-01 7.17465818e-01]
|
[8.638672828674316, -1.258387565612793]
|
33ebef80-364c-4b6c-bb3c-4031b2da8c94
|
hit-qmul-at-semeval-2022-task-9-label
| null | null |
https://aclanthology.org/2022.semeval-1.177
|
https://aclanthology.org/2022.semeval-1.177.pdf
|
HIT&QMUL at SemEval-2022 Task 9: Label-Enclosed Generative Question Answering (LEG-QA)
|
This paper presents the second place system for the R2VQ: competence-based multimodal question answering shared task. The purpose of this task is to involve semantic&cooking roles and text-images objects when querying how well a system understands the procedure of a recipe. This task is approached with text-to-text generative model based on transformer architecture. As a result, the model can well generalise to soft constrained and other competence-based question answering problem. We propose label enclosed input method which help the model achieve significant improvement from 65.34 (baseline) to 91.3. In addition to describing the submitted system, the impact of model architecture and label selection are investigated along with remarks regarding error analysis. Finally, future works are presented.
|
['Bingquan Liu', 'Arkaitz Zubiaga', 'Mingqiang Feng', 'Weihe Zhai']
| null | null | null | null |
semeval-naacl-2022-7
|
['generative-question-answering']
|
['natural-language-processing']
|
[ 5.00466764e-01 5.61002731e-01 6.22654736e-01 -9.60383177e-01
-1.32825887e+00 -9.15571034e-01 8.85738969e-01 1.25175178e-01
-4.93363827e-01 3.84581417e-01 4.44131553e-01 -1.79907084e-01
3.28937247e-02 -5.45440435e-01 -6.20963693e-01 -5.80225646e-01
5.29373288e-01 9.34896648e-01 -8.50712694e-03 -9.08936322e-01
1.74709350e-01 -3.43890667e-01 -1.27654469e+00 9.18301821e-01
6.50259137e-01 8.57813239e-01 3.65008295e-01 1.25912571e+00
-1.80352375e-01 1.49606967e+00 -6.53751433e-01 -6.81758046e-01
-3.36913466e-01 -4.82919335e-01 -1.50282896e+00 -4.99181636e-03
7.25807607e-01 -2.00814337e-01 8.35666135e-02 8.49219620e-01
6.40399992e-01 6.41324937e-01 9.02686000e-01 -1.25577831e+00
-8.93330872e-01 6.72819555e-01 3.10980052e-01 -6.90365061e-02
1.03761518e+00 4.75341380e-02 1.09912634e+00 -8.95909548e-01
5.86255670e-01 1.48117423e+00 1.27808124e-01 8.94858539e-01
-8.92743886e-01 -1.08811222e-01 1.10196494e-01 2.80722529e-01
-1.18472683e+00 -4.46414381e-01 6.48217201e-01 -3.04357588e-01
1.25051308e+00 6.99438751e-01 -4.90551069e-02 1.17557645e+00
-3.98564547e-01 9.94105101e-01 1.23345971e+00 -5.53220093e-01
-1.03232171e-02 5.52286565e-01 5.67150056e-01 8.45351458e-01
-6.68664098e-01 -4.19601738e-01 -5.49976110e-01 -3.16366851e-02
2.45053157e-01 -5.74510396e-01 -1.45196050e-01 -1.45988643e-01
-1.16484535e+00 1.02756548e+00 2.12975547e-01 2.62782902e-01
-1.39251560e-01 2.39441723e-01 5.16771019e-01 3.78595889e-01
1.09113179e-01 5.89070737e-01 -5.46848893e-01 -1.04873233e-01
-7.30620444e-01 4.54365253e-01 9.93443787e-01 1.29721904e+00
4.66366529e-01 -5.92972003e-02 -6.86907172e-01 1.04138970e+00
8.34604681e-01 6.02025747e-01 -1.06084328e-02 -1.18758821e+00
5.72651684e-01 7.55014539e-01 4.48671542e-02 -7.70270467e-01
-3.01581979e-01 2.31983960e-02 -5.44529617e-01 -2.91887283e-01
3.58165354e-01 3.61969285e-02 -8.70003104e-01 1.57980621e+00
2.31566265e-01 -5.87139428e-01 5.84904730e-01 1.00816655e+00
1.78229117e+00 1.00668955e+00 7.54827738e-01 2.39303142e-01
2.01307559e+00 -1.26198840e+00 -9.73736048e-01 -6.87671974e-02
4.95243311e-01 -8.21546853e-01 1.31053913e+00 3.94127637e-01
-1.17018569e+00 -8.10608029e-01 -4.68208402e-01 -6.55672193e-01
-8.02667618e-01 3.49380255e-01 4.45401162e-01 6.62260175e-01
-1.31216526e+00 -1.68171763e-01 -2.37024665e-01 -7.95115829e-01
-2.18996838e-01 1.92874566e-01 -2.55566597e-01 -2.92698532e-01
-1.40680301e+00 1.13444531e+00 7.03368723e-01 3.90465632e-02
-1.23869097e+00 -3.25623006e-01 -1.15335953e+00 -5.21044880e-02
5.29204845e-01 -9.33769047e-01 1.68003058e+00 -7.52096534e-01
-1.57233226e+00 1.25680494e+00 -3.25062931e-01 -1.97885066e-01
1.53249860e-01 -2.56965101e-01 -4.15546060e-01 5.20909965e-01
-7.47590065e-02 1.08786845e+00 5.32994568e-01 -1.31373703e+00
-3.46043676e-01 -2.38111153e-01 5.81887066e-01 6.30089581e-01
2.89834470e-01 3.51328671e-01 -5.62424421e-01 -2.84602851e-01
8.54819939e-02 -8.49437237e-01 -2.32193675e-02 -6.37771130e-01
-2.54473805e-01 -8.88176203e-01 4.83523309e-01 -1.11535728e+00
1.08099270e+00 -1.64688170e+00 2.89017260e-01 -8.73559564e-02
-1.64339900e-01 -7.18217269e-02 -1.91956863e-01 8.69722664e-01
1.36400172e-02 1.14760302e-01 -2.61477053e-01 -3.55282247e-01
5.41321695e-01 3.20548594e-01 -3.87618482e-01 -7.24460855e-02
4.03591782e-01 1.38237095e+00 -7.23543227e-01 -7.95763671e-01
2.83386588e-01 4.22917277e-01 -3.16493303e-01 7.85413742e-01
-6.84213877e-01 3.78324658e-01 -3.92796695e-01 7.77336061e-01
3.44703019e-01 -2.37726822e-01 1.17274513e-02 -5.25845408e-01
3.66974711e-01 1.87937081e-01 -1.00941253e+00 2.27417111e+00
-3.30203027e-01 3.44039172e-01 2.43643165e-01 -9.39001322e-01
9.47608709e-01 7.47784257e-01 -1.60250962e-01 -6.85217857e-01
2.16128528e-01 -1.89086691e-01 -4.17424291e-01 -1.04672325e+00
9.76201057e-01 -3.11625510e-01 -5.45736134e-01 2.17963278e-01
5.05973995e-01 -4.39579487e-01 2.72281826e-01 8.02077591e-01
5.97590744e-01 6.60207450e-01 6.20200709e-02 -2.97423929e-01
8.92749846e-01 2.35994533e-01 -3.04021686e-01 5.85512519e-01
-2.65257895e-01 5.30581236e-01 2.95309365e-01 -6.39575347e-02
-7.43187189e-01 -6.95759416e-01 2.41568223e-01 1.84240186e+00
-2.47868281e-02 -3.84888381e-01 -9.60281193e-01 -9.29204404e-01
-5.28389513e-01 1.10873997e+00 -6.82044566e-01 1.33725600e-02
-4.26012516e-01 -4.72277582e-01 8.23033154e-01 4.01823193e-01
4.82473224e-01 -1.15303433e+00 -5.52175879e-01 4.41269279e-02
-8.89250040e-01 -1.25389946e+00 -2.18801305e-01 1.24739356e-01
-6.06999338e-01 -1.09485114e+00 -7.38885999e-01 -1.07740510e+00
4.25953299e-01 -1.11928575e-01 1.74131584e+00 6.60281479e-02
1.34060875e-01 1.27527869e+00 -9.12537038e-01 -1.76819816e-01
-7.06260443e-01 -3.14961672e-02 -8.23403358e-01 -1.82478651e-01
5.54746091e-01 2.17565879e-01 -4.14983064e-01 3.19406807e-01
-1.07228184e+00 2.77030617e-01 2.04084262e-01 5.52985668e-01
3.81161094e-01 -4.33806986e-01 5.05934417e-01 -7.14290321e-01
8.85514677e-01 -5.30276656e-01 -1.35713100e-01 9.47573900e-01
-3.01949888e-01 1.78214356e-01 1.16445504e-01 -1.77716643e-01
-1.35976720e+00 9.32329372e-02 -4.05136645e-01 1.40512422e-01
-5.90108812e-01 2.18052313e-01 -4.64824408e-01 3.31941932e-01
5.85093558e-01 2.30919376e-01 -3.34183246e-01 -3.93739492e-01
8.43444228e-01 7.06530929e-01 5.68118632e-01 -8.36038470e-01
3.37468207e-01 -5.93214892e-02 -3.48990053e-01 -6.52106166e-01
-8.36202443e-01 -9.47339952e-01 -2.94033408e-01 -5.12580812e-01
1.54870927e+00 -1.04461741e+00 -1.10476840e+00 6.10301364e-03
-1.27511442e+00 -3.95624548e-01 2.54653599e-02 -2.61078170e-03
-8.17752540e-01 3.64008933e-01 -7.39969671e-01 -1.22477496e+00
-5.85265040e-01 -1.12626135e+00 1.40512645e+00 1.93271443e-01
-3.16600174e-01 -1.00774181e+00 8.04562494e-02 1.52624691e+00
3.55353773e-01 2.60831416e-01 9.52724874e-01 -1.06120789e+00
-4.09556568e-01 1.14816703e-01 -1.16975345e-01 3.18736792e-01
-5.44291556e-01 -3.57589185e-01 -1.10006094e+00 -1.99368879e-01
-2.02888288e-02 -1.06132638e+00 6.76122546e-01 -2.50499025e-02
6.39650404e-01 -2.02688158e-01 1.34597525e-01 -8.26467872e-02
1.37277746e+00 8.87341946e-02 6.37011349e-01 -1.18420888e-02
5.65677345e-01 1.18494725e+00 8.04799318e-01 -1.65709525e-01
1.04705691e+00 6.64263785e-01 5.63912630e-01 4.50279266e-02
-2.33002365e-01 -3.58192533e-01 3.34070206e-01 9.01735187e-01
3.20986658e-02 -5.90852439e-01 -9.41611409e-01 4.67956752e-01
-1.79910982e+00 -9.32112932e-01 -6.28031611e-01 1.60478902e+00
8.49111319e-01 -5.74749410e-01 7.04298615e-02 -1.97796330e-01
4.19526488e-01 1.05009153e-02 -8.99071917e-02 -6.90236211e-01
-1.95036173e-01 4.96924967e-02 -1.31465241e-01 8.58871937e-01
-1.17521572e+00 1.21859813e+00 6.35353136e+00 6.15233243e-01
-1.45414546e-01 5.04295170e-01 4.96832848e-01 4.39977497e-01
-4.70805854e-01 -8.32364112e-02 -7.95359731e-01 -9.14403349e-02
1.16521335e+00 5.27804017e-01 4.94582087e-01 5.98375201e-01
-1.84552729e-01 -4.37964499e-01 -1.27121556e+00 9.96084392e-01
6.83008134e-01 -6.70061052e-01 2.07988083e-01 -5.40423512e-01
3.83207530e-01 -3.20138127e-01 -1.60057992e-01 1.00231230e+00
4.35942411e-01 -1.34071958e+00 7.65995204e-01 7.53543854e-01
5.81721723e-01 -6.67433500e-01 8.94243240e-01 4.35308456e-01
-1.09495997e+00 2.52074242e-01 -1.27397105e-01 1.87956586e-01
3.01510394e-01 -1.80193409e-01 -1.08442795e+00 7.77798593e-01
8.52687657e-01 8.60123634e-02 -8.58468652e-01 2.21889719e-01
-4.20438677e-01 6.28992260e-01 -1.00949928e-02 -3.94698948e-01
2.79142886e-01 -9.34592634e-02 2.54399806e-01 1.70578682e+00
-7.24350065e-02 4.53637809e-01 3.37899178e-02 1.05937743e+00
1.69131979e-01 4.33693588e-01 -4.11488712e-01 -1.09799318e-01
-5.55531308e-02 1.44885480e+00 -4.28447902e-01 -5.18117845e-01
-2.03161210e-01 1.41861486e+00 2.28518486e-01 5.93924344e-01
-8.07287455e-01 -1.70460954e-01 -1.10597365e-01 -2.76949227e-01
6.82835951e-02 -1.62277836e-02 3.28162104e-01 -1.16748047e+00
-2.26702824e-01 -1.29108465e+00 6.79124713e-01 -1.47955370e+00
-1.06562102e+00 6.60378158e-01 2.35977322e-01 -4.88197833e-01
-4.14008737e-01 -6.65415227e-01 6.03306899e-03 8.62397611e-01
-1.38244689e+00 -1.55698073e+00 -3.86618257e-01 7.64700413e-01
9.41579580e-01 -1.08587705e-01 1.28661215e+00 2.15158224e-01
-6.19776100e-02 4.26568538e-01 -5.85472167e-01 2.52523690e-01
9.07626390e-01 -1.68310046e+00 -8.68853554e-02 5.39292455e-01
4.32011575e-01 6.57664895e-01 8.49046171e-01 -4.15952563e-01
-1.49740064e+00 -7.74294376e-01 1.19266582e+00 -1.13089514e+00
3.07513922e-01 -4.58067864e-01 -8.59960794e-01 5.93417346e-01
1.06147218e+00 -5.65645695e-01 1.01011872e+00 8.65384284e-03
-4.99025285e-01 1.79918125e-01 -1.27668631e+00 2.56496966e-01
4.37517613e-01 -8.12648475e-01 -1.08161104e+00 6.00992441e-01
8.72333825e-01 -3.84087384e-01 -1.10401273e+00 2.82696009e-01
4.85081552e-03 -6.60943449e-01 9.35043752e-01 -8.47088397e-01
4.07183945e-01 -3.97375852e-01 -8.38380277e-01 -8.68352473e-01
-3.96628715e-02 -4.73991036e-01 -8.72319192e-02 1.29839742e+00
5.45422316e-01 3.14500965e-02 4.71436441e-01 8.58969390e-01
-1.43806905e-01 -2.87435651e-01 -7.72094309e-01 -1.45068318e-01
8.26451555e-02 -5.15120447e-01 2.96533883e-01 9.98044908e-01
2.50426438e-02 1.01660120e+00 -3.52525502e-01 2.71696776e-01
3.25593740e-01 -3.09024543e-01 6.38946235e-01 -8.19208384e-01
-3.01987678e-01 -2.10809588e-01 8.24496672e-02 -1.21649706e+00
2.82505881e-02 -1.09313488e+00 2.74484724e-01 -2.01203680e+00
4.15495187e-01 1.20059744e-01 -1.20566525e-01 4.57314193e-01
-3.86830479e-01 1.58043623e-01 4.50928926e-01 -2.13963628e-01
-1.26123726e+00 4.34316993e-01 1.19492590e+00 -2.39988893e-01
2.44736418e-01 -2.35757917e-01 -6.39015973e-01 3.68023336e-01
8.67460787e-01 -1.53933510e-01 -6.44477904e-01 -5.59315979e-01
7.04076409e-01 3.82984281e-01 7.39925981e-01 -4.74911362e-01
2.75693506e-01 1.46769017e-01 -8.44761878e-02 -8.09492767e-01
5.99020064e-01 -1.01053894e+00 -7.18308836e-02 2.23465618e-02
-9.13598537e-01 3.15836549e-01 1.10793628e-01 3.77204537e-01
-3.47942173e-01 -5.73952496e-01 3.09579551e-01 -4.61551279e-01
-9.21912670e-01 -3.14262509e-01 -4.52841461e-01 1.70217708e-01
8.91138196e-01 4.80825752e-02 -2.86797106e-01 -8.04724932e-01
-1.02994490e+00 5.83088219e-01 -4.52400520e-02 6.98250353e-01
6.24717236e-01 -1.21222878e+00 -1.04686081e+00 -4.13903594e-01
5.09278119e-01 -1.41713530e-01 4.83659536e-01 5.03717542e-01
-4.50866610e-01 9.47128177e-01 2.60627717e-01 -5.73251486e-01
-1.63617873e+00 4.29974914e-01 4.73288804e-01 -3.14142615e-01
2.19999105e-02 1.13005662e+00 -3.00828777e-02 -1.04884005e+00
5.92718303e-01 -1.53722405e-01 -8.41373920e-01 1.05184756e-01
5.60730398e-01 3.00551474e-01 7.19464347e-02 -9.47888315e-01
-3.23433518e-01 5.31274080e-01 1.75091088e-01 -4.81535375e-01
7.97688484e-01 -3.44891697e-01 -2.58547992e-01 5.71651518e-01
1.17130041e+00 -5.37146389e-01 -6.76128685e-01 -5.30624129e-02
-8.63772258e-03 3.52225423e-01 -1.36765063e-01 -1.79028356e+00
-5.83223283e-01 1.10121584e+00 8.65300000e-01 2.84922928e-01
9.66306627e-01 5.55135667e-01 4.37233120e-01 6.89063072e-01
9.92592573e-02 -1.21482038e+00 2.52616256e-01 7.27295518e-01
1.15219021e+00 -1.72521818e+00 -3.01156640e-01 -1.78459004e-01
-1.22193444e+00 8.96338940e-01 6.80626750e-01 3.13635647e-01
-2.80490387e-02 -4.51606005e-01 3.64049733e-01 -6.92354679e-01
-8.28865707e-01 -2.86549181e-01 7.03284323e-01 6.56693876e-01
7.47678161e-01 2.75550395e-01 -1.01520881e-01 7.79057324e-01
-2.33942568e-01 -3.72038037e-01 1.46248981e-01 8.21581602e-01
-4.06887412e-01 -8.51081908e-01 -4.19428557e-01 -2.16115862e-01
-4.37643021e-01 -3.04692537e-01 -6.89474344e-01 6.96011782e-01
3.29600163e-02 1.81825650e+00 -3.02122772e-01 -1.78924337e-01
7.19817758e-01 7.49837458e-01 4.78564799e-01 -7.73368895e-01
-1.27384174e+00 5.43120578e-02 5.80058277e-01 -5.38994968e-01
-9.40398097e-01 -5.08754730e-01 -1.17902327e+00 3.21176380e-01
-2.35062897e-01 4.45520937e-01 8.54687929e-01 8.45097542e-01
1.74587872e-02 5.61916292e-01 -7.55204707e-02 -7.73405060e-02
-4.89496976e-01 -1.45832646e+00 -4.21716534e-02 6.70152545e-01
-1.63873583e-02 -1.30960301e-01 -1.75890565e-01 5.40058315e-01]
|
[11.28055191040039, 7.898998260498047]
|
333c6c85-8079-4909-9d7a-a54d4aae6043
|
video-denoising-and-enhancement-via-dynamic
|
1710.02213
| null |
http://arxiv.org/abs/1710.02213v1
|
http://arxiv.org/pdf/1710.02213v1.pdf
|
Video Denoising and Enhancement via Dynamic Video Layering
|
Video denoising refers to the problem of removing "noise" from a video
sequence. Here the term "noise" is used in a broad sense to refer to any
corruption or outlier or interference that is not the quantity of interest. In
this work, we develop a novel approach to video denoising that is based on the
idea that many noisy or corrupted videos can be split into three parts - the
"low-rank layer", the "sparse layer", and a small residual (which is small and
bounded). We show, using extensive experiments, that our denoising approach
outperforms the state-of-the-art denoising algorithms.
|
['Namrata Vaswani', 'Han Guo']
|
2017-10-05
| null | null | null | null |
['video-denoising']
|
['computer-vision']
|
[ 3.34370047e-01 -3.96050394e-01 4.63313907e-01 -6.28974065e-02
-8.26565504e-01 -2.12983564e-01 2.92736501e-01 -1.25676781e-01
-2.00351357e-01 5.95887542e-01 5.51603258e-01 1.57360017e-01
1.02650054e-01 -4.08802271e-01 -9.75680530e-01 -1.12021863e+00
-2.83646822e-01 -3.58192921e-01 1.16327778e-01 -3.04953873e-01
1.34565607e-01 1.52926043e-01 -1.38252807e+00 3.55747312e-01
5.66497684e-01 9.06494141e-01 5.76879084e-02 5.85107386e-01
2.18367606e-01 1.03974783e+00 -7.77394533e-01 -1.44115612e-01
4.53532249e-01 -6.42001152e-01 -3.56498599e-01 3.42431217e-01
4.51974273e-01 -4.28517580e-01 -8.62804115e-01 1.51378798e+00
4.97777432e-01 2.61193693e-01 1.57149225e-01 -7.55981863e-01
-2.75765538e-01 3.00772250e-01 -7.40321755e-01 4.22751009e-01
5.93409836e-01 -1.49030507e-01 3.93981636e-01 -1.10762393e+00
8.39989781e-01 1.30340219e+00 8.96931946e-01 3.78506988e-01
-1.39865017e+00 -4.24508899e-01 2.97087491e-01 2.61839926e-01
-1.41651261e+00 -6.93555176e-01 8.39435875e-01 -5.26482403e-01
5.47386885e-01 2.85943449e-01 2.74026692e-01 1.28617251e+00
5.35760939e-01 6.50311410e-01 9.85798061e-01 -3.10155302e-01
2.52073407e-01 -5.31127930e-01 1.31668150e-01 5.28152883e-01
4.06293362e-01 9.10877734e-02 -7.44290411e-01 -2.46733904e-01
6.03834152e-01 3.02728098e-02 -9.14597631e-01 -1.69897467e-01
-1.01751065e+00 6.53629184e-01 3.22880894e-02 2.68098682e-01
-7.17046440e-01 3.22873443e-01 6.84131980e-01 5.83765090e-01
8.38072062e-01 -8.05846602e-02 -1.31574839e-01 -1.72678784e-01
-1.17439401e+00 1.29564852e-01 9.61417913e-01 6.90859079e-01
5.96126139e-01 4.16245103e-01 8.19680393e-02 9.21769381e-01
1.53749287e-01 4.68909860e-01 3.68453264e-01 -1.27399290e+00
5.25036752e-01 -1.64075270e-01 1.93921700e-01 -1.18551803e+00
-3.13136466e-02 -3.58560115e-01 -1.37751174e+00 2.64923453e-01
1.70314103e-01 -2.46707320e-01 -9.20407712e-01 1.75087416e+00
-2.41010319e-02 8.30955684e-01 -1.77879199e-01 1.05059934e+00
8.56094837e-01 8.65712941e-01 -3.07099998e-01 -6.74912751e-01
7.79893458e-01 -8.03968906e-01 -1.07159269e+00 -1.05538301e-01
-4.37456630e-02 -1.06202114e+00 4.28872913e-01 9.56873119e-01
-1.38936460e+00 -6.06346250e-01 -1.06185508e+00 5.89054264e-02
-1.14292257e-01 -4.15592104e-01 2.20511302e-01 2.73757011e-01
-1.27573454e+00 8.84522140e-01 -6.02544367e-01 -8.50820243e-02
1.34643510e-01 7.65049383e-02 -6.07680798e-01 -5.59420764e-01
-9.17749941e-01 5.26477695e-01 -1.32787764e-01 4.92083639e-01
-1.28055930e+00 -3.98231536e-01 -8.70668054e-01 -3.99594940e-02
6.45281553e-01 -7.82108605e-01 9.52007592e-01 -1.13475561e+00
-1.10245812e+00 6.49729073e-01 -7.31048882e-01 -3.41563374e-01
4.94802594e-01 -5.18179536e-01 -5.11450887e-01 4.44725633e-01
9.27951559e-03 -2.25102559e-01 1.57847583e+00 -1.66768873e+00
-4.31370348e-01 -3.34102064e-01 -2.23809540e-01 -7.74601623e-02
2.17260525e-01 1.87188029e-01 -7.64094472e-01 -1.38804781e+00
6.78826034e-01 -8.50558043e-01 -4.28513885e-01 -3.29922587e-01
-2.37873867e-01 2.92822987e-01 9.21610892e-01 -1.12340450e+00
1.34168375e+00 -2.49369836e+00 5.22416532e-01 5.41854024e-01
3.56759310e-01 1.17659606e-01 -2.57934272e-01 3.63207668e-01
-4.35275018e-01 2.45523006e-01 -1.83808178e-01 -4.56083417e-01
-4.09014016e-01 4.53236163e-01 -3.28308254e-01 9.10568357e-01
-2.06215024e-01 2.63669044e-01 -1.12642109e+00 -1.17021531e-01
1.90432295e-01 6.94246888e-01 -3.23344022e-01 1.89077720e-01
3.46406817e-01 6.45128012e-01 -2.21026585e-01 8.25254083e-01
8.75537932e-01 2.36368880e-01 7.97938257e-02 -5.46783030e-01
-1.04079917e-01 -7.01807514e-02 -1.53777373e+00 1.65906692e+00
1.03297845e-01 7.23367929e-01 9.19929564e-01 -9.55357790e-01
4.49499249e-01 5.72856784e-01 6.27204120e-01 -3.87907326e-01
1.81898579e-01 1.74718723e-01 -2.83968717e-01 -6.56120300e-01
4.52955306e-01 -1.30440816e-01 2.88772434e-01 -7.41764158e-02
5.60488887e-02 1.39135107e-01 5.56218565e-01 3.41259331e-01
1.58674085e+00 -1.05682136e-02 1.61008865e-01 -3.19484890e-01
6.98954165e-01 -3.84687722e-01 9.25994396e-01 1.06542039e+00
-2.94292659e-01 9.29829180e-01 4.87567067e-01 -5.34263849e-02
-9.33916390e-01 -1.08977580e+00 2.16058001e-01 8.56883287e-01
1.02343537e-01 -5.59522986e-01 -9.42352235e-01 -1.91656664e-01
-5.42828292e-02 -6.57171085e-02 -7.06281304e-01 3.39016616e-02
-7.32371449e-01 -4.92593974e-01 2.13310003e-01 2.43237093e-01
3.00426424e-01 -7.50076771e-01 1.77407727e-01 3.81291002e-01
-4.71277118e-01 -1.03005195e+00 -7.22216010e-01 2.29647905e-01
-1.02993524e+00 -1.04633641e+00 -7.73811042e-01 -6.94933057e-01
7.63532758e-01 8.33712995e-01 1.31631088e+00 4.19565022e-01
4.97344546e-02 5.09113252e-01 -7.15844691e-01 1.67472344e-02
-2.62956202e-01 -6.52125835e-01 3.42771769e-01 2.43379340e-01
2.93536987e-02 -7.04270184e-01 -4.65772837e-01 5.07753864e-02
-1.24042392e+00 -5.02357006e-01 3.84161949e-01 8.91556859e-01
8.95738304e-01 5.41666031e-01 1.17243439e-01 -1.00418782e+00
6.43775165e-01 -6.50041938e-01 -2.73544699e-01 -3.96609642e-02
-9.54773352e-02 -2.38815665e-01 6.76177621e-01 -3.36327493e-01
-8.21370125e-01 -4.14694361e-02 -2.39876211e-01 -7.64389277e-01
-6.01145513e-02 6.59306705e-01 -3.68560046e-01 -4.36011106e-01
3.16871464e-01 2.70599574e-01 -5.40166833e-02 -8.80933762e-01
2.99205691e-01 4.05214250e-01 9.80768144e-01 -4.28653449e-01
1.00192463e+00 7.35623896e-01 9.55696627e-02 -1.19277310e+00
-8.27665627e-01 -9.44988072e-01 -6.42345309e-01 -3.37943316e-01
4.82582927e-01 -1.21334469e+00 -2.01449811e-01 6.11721337e-01
-1.07856631e+00 -1.25546679e-02 -3.23879957e-01 5.03627300e-01
-4.17420208e-01 9.04711366e-01 -1.01066267e+00 -6.07489049e-01
3.89070669e-03 -1.03885210e+00 7.92819142e-01 -2.18370538e-02
-2.43811155e-04 -8.24521661e-01 2.42407583e-02 2.64552623e-01
2.63365269e-01 4.42013234e-01 3.11604410e-01 -1.99454799e-01
-4.16566998e-01 -2.91472018e-01 1.24032311e-02 1.09814239e+00
-5.70299104e-04 -8.44519138e-02 -7.78267860e-01 -5.82444191e-01
7.15839922e-01 9.58008394e-02 1.29486012e+00 5.60125411e-01
1.01977324e+00 -2.94390947e-01 5.17857773e-03 8.05279851e-01
1.75751770e+00 9.47207063e-02 1.08946264e+00 2.66288817e-01
7.11932838e-01 2.06023723e-01 4.84292597e-01 5.78728378e-01
-1.74242511e-01 3.29229921e-01 3.49517852e-01 -1.38891980e-01
-1.08850181e-01 8.83919299e-02 6.66771233e-01 1.25941992e+00
-3.74584407e-01 -3.50616395e-01 -4.44947422e-01 6.01022422e-01
-2.01483369e+00 -1.32983339e+00 -4.14167821e-01 2.13069892e+00
6.42383993e-01 5.90691417e-02 -1.64126694e-01 3.67314905e-01
8.06634665e-01 5.32267928e-01 -7.63498917e-02 -3.17965746e-01
-4.25565541e-01 3.03408623e-01 6.81643784e-01 6.14692330e-01
-1.27564728e+00 5.55018008e-01 7.37088203e+00 7.70070136e-01
-8.28953922e-01 1.62150532e-01 2.92405069e-01 9.07341242e-02
5.83532602e-02 -1.86560869e-01 -1.72219589e-01 6.33587301e-01
7.61917710e-01 3.61268893e-02 6.39451504e-01 5.24684250e-01
8.35835516e-01 -2.63732672e-01 -9.01355445e-01 1.20636129e+00
3.27569306e-01 -1.05818343e+00 5.46656698e-02 -1.66929364e-01
1.00068998e+00 5.21905534e-02 -1.06688567e-01 -5.47083952e-02
2.83726249e-02 -9.10614491e-01 8.59398305e-01 9.67751086e-01
4.27005202e-01 -6.12251818e-01 1.04914260e+00 1.22898974e-01
-1.25471485e+00 9.71005298e-03 -3.56540084e-01 -1.53819667e-02
4.58225161e-01 9.53943908e-01 3.02066714e-01 7.01757550e-01
1.17286527e+00 1.01434064e+00 -1.42463014e-01 1.29284465e+00
-2.09848151e-01 8.60957503e-01 -1.41440898e-01 7.78464496e-01
2.88289636e-01 -6.48579597e-01 9.07023847e-01 1.54759324e+00
4.68499154e-01 5.36701083e-01 3.54139000e-01 1.32348523e-01
-2.23481819e-01 -1.68152407e-01 -6.07736051e-01 4.00462985e-01
7.85020366e-02 9.41640913e-01 -4.18780118e-01 -6.05968952e-01
-7.56581366e-01 1.23633349e+00 -4.87495601e-01 7.46187329e-01
-5.80856562e-01 -2.82574743e-01 9.10867989e-01 1.13672152e-01
5.73146641e-01 -3.34895492e-01 -1.64203778e-01 -1.45449853e+00
1.97981566e-01 -1.22327256e+00 1.39778480e-01 -8.22320282e-01
-1.47957122e+00 4.42435026e-01 -3.75385612e-01 -1.43836296e+00
-2.95756962e-02 -4.86938953e-01 -4.68169510e-01 8.28895986e-01
-1.34984970e+00 -4.91638809e-01 -5.91410816e-01 8.22417259e-01
5.49607098e-01 -3.79354693e-02 3.88046652e-01 5.30004442e-01
-5.83089173e-01 -2.02588849e-02 6.90869868e-01 1.34494096e-01
6.72290564e-01 -9.70938504e-01 -6.90552592e-02 1.44932854e+00
-1.95889741e-01 5.81214368e-01 1.28309238e+00 -8.04378688e-01
-1.48433506e+00 -1.00957668e+00 7.48049855e-01 2.25817990e-02
7.52194345e-01 -1.06945187e-01 -1.28448546e+00 7.03302085e-01
1.88413545e-01 1.82169631e-01 4.31482524e-01 -2.38605171e-01
-2.99379617e-01 -1.61369011e-01 -1.13494372e+00 5.24503469e-01
1.04051900e+00 -5.26211023e-01 -6.74125552e-01 2.75011390e-01
4.89342988e-01 -3.86926353e-01 -9.15719211e-01 2.66124338e-01
2.86336720e-01 -1.15285385e+00 1.17928386e+00 -4.00250196e-01
2.93011069e-01 -5.59896290e-01 -4.80498195e-01 -1.34464681e+00
-5.27952909e-01 -1.01014197e+00 -5.31511784e-01 1.00943828e+00
-2.39079177e-01 -1.06452905e-01 7.05667555e-01 5.15805222e-02
-2.48905808e-01 -2.01448828e-01 -1.09733140e+00 -9.76553619e-01
-4.17494059e-01 -5.13470054e-01 4.85220253e-02 8.29747975e-01
-2.81026185e-01 9.16524753e-02 -8.40917051e-01 2.06160873e-01
9.35270429e-01 -5.16213119e-01 4.93424177e-01 -1.10966051e+00
4.53005508e-02 -1.15907773e-01 -5.69400549e-01 -1.20923865e+00
-3.12725292e-03 -2.98409283e-01 5.24777889e-01 -1.35785091e+00
1.96593925e-01 2.25583300e-01 -5.50033808e-01 -6.24068035e-03
-1.25909284e-01 5.23627102e-01 7.85678476e-02 2.98961192e-01
-7.06520140e-01 3.94904673e-01 8.53660941e-01 -1.64599657e-01
9.82952863e-02 -1.88875198e-01 -6.04796529e-01 1.09773207e+00
5.07472694e-01 -6.11674845e-01 -4.25537154e-02 -4.90473062e-01
-2.37804540e-02 1.23287275e-01 3.42169940e-01 -9.98658359e-01
2.69357979e-01 1.45370245e-01 2.52875715e-01 -4.43623692e-01
3.90150219e-01 -1.00813735e+00 2.50849605e-01 3.82356137e-01
1.16452970e-01 9.60200876e-02 -9.67183411e-02 8.04357827e-01
-7.15659976e-01 -3.35815936e-01 8.29643250e-01 -3.75750333e-01
-7.72480667e-01 1.44264743e-01 -7.34500408e-01 -9.29050818e-02
6.22779906e-01 -2.09254205e-01 -1.45362332e-01 -8.49199891e-01
-1.00619709e+00 -1.32024035e-01 6.01593137e-01 1.50448769e-01
8.98314714e-01 -1.32394671e+00 -9.95933950e-01 9.16619152e-02
-3.00568849e-01 -2.29261011e-01 2.72117794e-01 9.00165796e-01
-6.61478162e-01 2.71143857e-02 2.42238060e-01 -4.03986245e-01
-1.49462068e+00 6.23535514e-01 8.47314075e-02 -1.67326465e-01
-8.22206378e-01 8.45898807e-01 5.23016118e-02 3.16924930e-01
5.89635432e-01 -1.38016611e-01 9.98422224e-03 5.87725360e-03
8.83883238e-01 6.81629062e-01 7.80112168e-04 -1.09781849e+00
-2.54113048e-01 6.55453384e-01 3.05481017e-01 4.83005270e-02
1.47018051e+00 -4.83217180e-01 -7.22498000e-01 4.61548090e-01
1.28252697e+00 2.64257103e-01 -1.21368074e+00 -2.45672449e-01
8.12193006e-03 -8.11727226e-01 2.25034669e-01 -3.61779690e-01
-1.32929885e+00 3.19523543e-01 6.70756698e-01 4.63306367e-01
1.57022238e+00 -2.86628902e-01 8.61788452e-01 3.72287631e-01
4.12046522e-01 -1.34882390e+00 -7.09899664e-02 7.56217480e-01
1.10880721e+00 -1.09078503e+00 2.46309057e-01 -5.80019891e-01
-1.85182974e-01 9.45572019e-01 -2.90612057e-02 -4.96137977e-01
1.00983703e+00 4.87181485e-01 1.93264812e-01 -1.37948859e-02
-5.46660900e-01 -3.20931882e-01 2.38732249e-01 6.18513644e-01
4.53965187e-01 -3.58184576e-01 -3.30140501e-01 5.15729427e-01
2.23201692e-01 2.02063918e-01 6.92658901e-01 1.11362410e+00
-6.44230545e-01 -9.72276986e-01 -1.01636899e+00 4.79782164e-01
-8.75213861e-01 -2.06079394e-01 -9.70034748e-02 5.80079556e-01
4.44031179e-01 1.32295358e+00 -2.94039011e-01 -5.09447098e-01
5.03094316e-01 -3.46921414e-01 3.52716833e-01 -4.86800790e-01
-4.85291541e-01 6.28693044e-01 2.33926736e-02 -1.08921230e+00
-8.20739567e-01 -6.19203091e-01 -7.47316837e-01 -7.19407141e-01
-1.75674886e-01 8.95488188e-02 4.11698997e-01 8.39018941e-01
-1.78726450e-01 6.11991346e-01 6.51838720e-01 -1.20278227e+00
-2.38509014e-01 -8.42668653e-01 -1.05490422e+00 7.38313973e-01
7.39239454e-01 -2.61954129e-01 -8.04363012e-01 5.03858924e-01]
|
[11.404804229736328, -2.210583448410034]
|
aad3544c-43df-4889-8149-790ed8d8aad6
|
in-silico-identification-of-tipifarnib-like
|
2305.16156
| null |
https://arxiv.org/abs/2305.16156v1
|
https://arxiv.org/pdf/2305.16156v1.pdf
|
In silico Identification of tipifarnib-like compounds by structure-based pharmacophore, virtual screening and molecular docking against K-Ras post-translation in colorectal cancer
|
Colorectal cancer is a public health problem.Approximately 30 to 50 \% of colorectal tumors are caused by mutations in the KRAS gene.These mutations induce uncontrolled proliferation.To date,There is no approved effective treatment for the mutated KRAS oncogene.Farnesyltransferase (FTI) inhibitors are considered a therapeutic target against the mutated KRAS oncogene.Tipifarnib is a farnesyltransferase inhibitor that was analyzed in a Phase II trial.In the present study, the three-dimensional structure of farnesyltransferase complexed with tipifarnib [1SA4] was used as a basis to exploit the characteristics of tipifarnib.A pharmacophore model was generated based on the structure using the Asinex (Gold and Platinum Collections) database.A total of 299 molecules were obtained after screening.The 299 molecules were anchored to the tipifarnib binding site in the farnesyltransferase crystal structure for docking analysis.During the molecular docking process, the pharmacophore that was modeled, and was used as a constraint to eliminate the molecules that do not satisfy the pharmacophore.Finally, four Hits identified as farnesyltransferase inhibitors for biological tests. Keywords: colorectal cancer, structure-based pharmacophore, molecular docking, KRAS, farnesyltransferase inhibitors, Virtual Screening.
|
['Houda Filali', 'Imane Rahmoune', 'Youness Kadil1', 'Mohammed Mouhcine']
|
2023-05-07
| null | null | null | null |
['molecular-docking']
|
['medical']
|
[ 3.67027998e-01 -2.09830031e-01 -7.78778017e-01 2.50675917e-01
-3.98542494e-01 -7.43001223e-01 4.95390072e-02 5.60292244e-01
-5.21324337e-01 1.29392624e+00 2.80455977e-01 -7.35546768e-01
2.10945278e-01 -6.39211237e-01 -5.48590183e-01 -1.07714665e+00
1.64428234e-01 4.01420385e-01 1.94148690e-01 -2.10241646e-01
4.19615895e-01 7.32390285e-01 -6.97940409e-01 1.42821506e-01
1.46989000e+00 -7.32553303e-02 3.21213543e-01 5.51154315e-01
1.73654258e-02 3.17017622e-02 -3.99791956e-01 7.76677057e-02
-4.03545946e-02 -6.01218581e-01 -4.03406769e-01 -4.54544544e-01
-3.39660794e-02 9.33415070e-02 2.02171188e-02 8.46339166e-01
4.02880788e-01 -8.07395205e-02 7.46548951e-01 -4.55304086e-01
-1.56157196e-01 1.13532223e-01 -3.18036675e-01 1.66612118e-01
6.56667113e-01 3.38525251e-02 9.09055293e-01 -1.38472664e+00
5.64556658e-01 6.34936512e-01 4.05752331e-01 5.15634596e-01
-9.58317220e-01 -7.44708717e-01 -2.59731531e-01 7.03055039e-02
-1.54799247e+00 -1.91315950e-03 1.77548066e-01 -4.56278861e-01
1.39812541e+00 5.90107143e-01 1.30356658e+00 5.29998302e-01
7.70918846e-01 3.14235836e-01 8.10166597e-01 -5.73046625e-01
3.09975982e-01 -1.18768103e-01 1.29140615e-01 8.56048584e-01
9.20376897e-01 3.02849621e-01 -7.06722319e-01 -7.58907914e-01
3.91628683e-01 7.60546178e-02 -4.10462081e-01 -2.65618652e-01
-9.15904582e-01 9.86710787e-01 4.89244580e-01 5.19230604e-01
-2.43914619e-01 -1.29711092e-01 2.61235088e-01 -4.15654629e-01
-2.99664527e-01 4.92966533e-01 -3.48717779e-01 2.58247733e-01
-3.70243549e-01 7.78100491e-02 7.42455781e-01 2.02519432e-01
4.81965661e-01 -2.91741759e-01 2.57242084e-01 1.51959404e-01
8.25622022e-01 3.74841243e-01 6.20538890e-01 3.17559987e-01
-4.46571112e-02 1.06662297e+00 2.80551374e-01 -5.27163982e-01
-6.44801676e-01 -4.57530003e-03 -3.60371977e-01 1.86291650e-01
7.27448702e-01 -1.83609620e-01 -9.80217576e-01 1.19196463e+00
4.83841032e-01 4.03411299e-01 3.22626084e-01 9.94595706e-01
7.49438584e-01 6.01313472e-01 4.66767937e-01 -5.21161795e-01
1.53835511e+00 -7.02397525e-01 -4.80541825e-01 4.26604301e-01
8.06159675e-01 -1.04101896e+00 5.33545434e-01 4.66225207e-01
-7.57585824e-01 4.48419839e-01 -1.32559323e+00 2.97898859e-01
-4.15477425e-01 6.57623168e-03 8.57785106e-01 1.20965350e+00
-5.74946165e-01 3.37844223e-01 -1.27789009e+00 -4.57561433e-01
-6.48002978e-03 8.41365516e-01 -4.29445624e-01 4.69701318e-03
-1.21818578e+00 9.17250574e-01 9.17466760e-01 -1.58777937e-01
-7.28952885e-01 -7.28154302e-01 -5.20430863e-01 -9.30291787e-02
9.08106565e-02 -1.03081477e+00 7.56292343e-01 -6.64020479e-01
-1.68717396e+00 6.16874218e-01 -2.99338013e-01 -2.76278794e-01
-8.67450535e-02 2.70185202e-01 -2.67175734e-01 2.89437408e-03
-9.39455181e-02 -2.05694690e-01 -1.88446175e-02 -5.11052907e-01
-7.09561765e-01 -5.79245925e-01 -4.09443557e-01 6.90884233e-01
1.98030442e-01 1.28702983e-01 -1.10668458e-01 -5.22880912e-01
7.14432150e-02 -1.11360335e+00 -5.59911072e-01 -4.14604753e-01
-5.43715715e-01 -2.82581866e-01 3.13808560e-01 -4.79396909e-01
1.16216385e+00 -1.84047246e+00 2.37140685e-01 7.14240909e-01
6.40959898e-03 7.32472062e-01 2.99603045e-01 9.22226727e-01
-3.48956406e-01 3.95429343e-01 8.83915499e-02 1.09214389e+00
-4.66027200e-01 -2.30282038e-01 3.27626079e-01 1.00648916e+00
-2.91372001e-01 6.78295672e-01 -1.13473964e+00 -2.72216201e-02
2.39471510e-01 5.44918776e-01 -6.73853695e-01 -3.33259702e-01
-4.90595132e-01 6.55091166e-01 -1.18085325e+00 9.39997792e-01
6.08048797e-01 -1.26114398e-01 6.56830788e-01 -8.41989815e-02
-4.54524636e-01 1.01123825e-01 -7.19187617e-01 1.66637850e+00
3.24001014e-01 -7.68135190e-02 -5.70347726e-01 -3.03272069e-01
6.97110653e-01 7.25749135e-01 3.60462785e-01 -5.49688280e-01
1.41520798e-01 8.37534666e-01 3.49038988e-01 -1.29742816e-01
-2.16398150e-01 -4.37848806e-01 3.15459102e-01 -3.42483103e-01
-5.62668681e-01 4.39982206e-01 5.89544117e-01 -8.10784008e-03
9.90269482e-01 -5.80287538e-02 1.04699123e+00 -5.76406717e-01
1.04094052e+00 5.61125994e-01 6.07561648e-01 2.05509588e-01
2.49584198e-01 3.33085796e-03 2.94909060e-01 -7.15089560e-01
-8.05724442e-01 -7.90960312e-01 -4.18047756e-01 3.68108332e-01
1.97620034e-01 -5.18169820e-01 -5.68613231e-01 -2.16647059e-01
-1.60276800e-01 5.11881173e-01 -5.10473251e-01 9.89159662e-03
-3.93158734e-01 -1.11369610e+00 6.76421881e-01 -2.44225398e-01
3.93667191e-01 -3.83769929e-01 -2.89361000e-01 6.45083547e-01
2.57293761e-01 -2.63180044e-02 -4.10379589e-01 1.77569777e-01
-6.31590009e-01 -1.76886368e+00 -8.49781930e-01 -7.06441283e-01
6.65627897e-01 9.17226821e-02 3.08201313e-01 4.08489704e-01
-3.99282366e-01 -4.32082087e-01 -1.42552599e-01 -7.19560623e-01
-4.63962048e-01 -2.40381941e-01 3.82247157e-02 -5.48191249e-01
6.48672104e-01 -4.37111944e-01 -8.53485584e-01 2.74001569e-01
-4.37110782e-01 3.36234905e-02 7.31952608e-01 9.66655016e-01
8.38290036e-01 -9.92295071e-02 3.29522014e-01 -9.89822507e-01
5.05214751e-01 -5.40420890e-01 -9.35173333e-01 4.30335663e-02
-4.20413345e-01 -1.51396543e-01 6.83055997e-01 -2.15487644e-01
-8.88198197e-01 3.46356034e-01 -3.12627524e-01 3.32069397e-01
1.87193677e-01 1.11949599e+00 -4.64064002e-01 -3.76249045e-01
8.61036122e-01 1.17854178e-01 1.01689912e-01 -3.04980010e-01
-3.65956992e-01 2.55533248e-01 4.41359077e-03 -1.59120262e-01
5.15735567e-01 2.93467939e-01 2.70769924e-01 -9.68643963e-01
-4.51629698e-01 -7.38693595e-01 -9.49948374e-03 3.45871955e-01
1.03012383e+00 -9.14402902e-01 -1.20534599e+00 3.06215018e-01
-8.49993050e-01 1.40542919e-02 4.67019290e-01 1.11576331e+00
-1.75655186e-01 5.09644032e-01 -3.97453308e-01 -2.50961721e-01
-5.90121686e-01 -1.20822942e+00 1.94138736e-01 6.41667068e-01
-1.61330685e-01 -1.01797700e+00 7.86959112e-01 2.40027905e-01
-3.08902040e-02 7.89548576e-01 1.27860987e+00 -9.50375676e-01
-7.05542326e-01 -3.31495434e-01 1.82246059e-01 -4.52854365e-01
2.64121175e-01 2.41797552e-01 -3.74241829e-01 -3.90120566e-01
-3.74196082e-01 1.03632301e-01 5.02994716e-01 5.14941037e-01
4.08700258e-01 -2.11724237e-01 -8.79994512e-01 7.86415160e-01
1.71627009e+00 9.26284671e-01 7.18416750e-01 5.41866004e-01
5.37356436e-01 -3.66305619e-01 8.00000370e-01 4.71659839e-01
-4.30760890e-01 6.11712694e-01 4.52511877e-01 -3.80924225e-01
2.06012830e-01 -3.22171748e-01 9.46863368e-02 -4.03544866e-02
-5.76852083e-01 -4.70488220e-01 -1.11740148e+00 7.14993626e-02
-1.43327260e+00 -9.45674002e-01 -7.24145055e-01 2.22639298e+00
1.09548485e+00 -1.10244930e-01 3.20574902e-02 -6.32367060e-02
4.54798251e-01 -6.81276321e-01 -7.64895499e-01 -2.94673443e-01
-3.24784801e-03 5.10013163e-01 9.51857924e-01 7.89807022e-01
-4.86541212e-01 7.91296840e-01 6.20464277e+00 7.55441546e-01
-1.14071250e+00 -5.12853920e-01 1.02072254e-01 3.50048512e-01
-2.39502698e-01 6.31030142e-01 -8.68059158e-01 2.50567615e-01
8.04290771e-01 -7.92423308e-01 1.42266182e-02 4.35407996e-01
6.44119740e-01 -4.05759186e-01 -9.36159730e-01 6.49857759e-01
-4.28225428e-01 -1.86675131e+00 1.72950961e-02 6.15181923e-01
6.06785119e-01 -3.17956358e-02 -2.51110315e-01 -3.01884562e-01
2.64582276e-01 -1.27059615e+00 7.29737878e-02 3.65196139e-01
7.48125374e-01 -1.00345242e+00 7.30690360e-01 3.39168668e-01
-9.50390160e-01 1.67988300e-01 -3.40631515e-01 2.13592887e-01
-2.83975124e-01 4.85157967e-01 -1.42567098e+00 5.16382873e-01
6.87854039e-03 3.73272002e-01 -2.54247814e-01 1.35573208e+00
-1.96044341e-01 5.15431345e-01 -3.59072477e-01 -6.02094471e-01
2.99478769e-01 -4.24336255e-01 3.83538276e-01 7.17220962e-01
2.01881379e-01 4.98245001e-01 3.56603116e-01 4.06695455e-01
1.72334045e-01 7.62317717e-01 -3.06053787e-01 -1.98953450e-01
4.62772250e-01 8.27756047e-01 -6.63866341e-01 -1.37307227e-01
-5.93062580e-01 7.19981730e-01 -1.76126152e-01 3.72384161e-01
-6.18313491e-01 -2.34421268e-01 6.31851733e-01 1.62485108e-01
-8.96721184e-02 4.42189902e-01 1.01003021e-01 -7.68128514e-01
-6.61145270e-01 -1.13590896e+00 7.62939572e-01 -1.79380968e-01
-6.40081942e-01 3.45224887e-02 -1.15589693e-01 -8.48875999e-01
2.88567126e-01 -6.41293466e-01 -7.57477343e-01 1.38034010e+00
-1.37922502e+00 -1.02856290e+00 -1.03935294e-01 5.90451241e-01
1.71018422e-01 -5.63095696e-02 1.42210031e+00 -1.98606960e-02
-6.96499288e-01 1.37965590e-01 3.96161169e-01 -4.78010207e-01
7.36057997e-01 -1.13227856e+00 -3.89570326e-01 4.35260296e-01
-7.86540627e-01 1.04892552e+00 9.51702178e-01 -1.04935193e+00
-1.63341320e+00 -8.30885172e-01 7.74586797e-01 -6.68129101e-02
5.63925266e-01 2.63108999e-01 -5.55990815e-01 6.15160525e-01
1.26323983e-01 -3.64035845e-01 1.71433556e+00 -5.75379074e-01
1.54318839e-01 7.35991120e-01 -1.28659880e+00 8.97973657e-01
1.99659824e-01 1.17993385e-01 -8.39376226e-02 8.86924744e-01
1.93120494e-01 -7.60606587e-01 -1.14318538e+00 -1.99483503e-02
6.60501242e-01 -6.83949172e-01 9.87251878e-01 -4.20782894e-01
-2.43017465e-01 -8.85204434e-01 2.61684299e-01 -1.08879781e+00
-4.81661797e-01 -9.35478747e-01 3.66943061e-01 2.28066877e-01
7.45526493e-01 -6.33169174e-01 1.18722904e+00 4.56432998e-01
-9.34893116e-02 -4.93978530e-01 -1.14185345e+00 -4.78630364e-01
1.27717748e-01 6.84199393e-01 4.85801429e-01 6.96947634e-01
7.98243701e-01 5.88652015e-01 1.00877941e-01 9.71987769e-02
3.12005967e-01 -3.47903788e-01 5.41577399e-01 -1.13856721e+00
-3.58486921e-02 -1.85168102e-01 -3.80061030e-01 -2.45833799e-01
-2.46493742e-01 -1.11789763e+00 -7.03386843e-01 -1.62356985e+00
3.21298391e-01 -1.50365263e-01 1.47693306e-02 4.23706353e-01
-4.90271188e-02 -1.37382686e-01 -3.26124340e-01 1.56575128e-01
3.48764807e-01 8.72809254e-03 1.08189940e+00 -1.16228528e-01
-6.54414356e-01 2.09511921e-01 -7.70220757e-01 7.61602998e-01
9.56869185e-01 -4.60758984e-01 -3.46042156e-01 7.40376174e-01
4.80565995e-01 2.64114141e-01 -2.79801965e-01 -3.92558873e-01
1.29385278e-01 -8.96403968e-01 5.19482791e-01 -8.07223022e-01
5.03180437e-02 -9.62039292e-01 9.77220774e-01 1.47912920e+00
4.29237157e-01 -3.57707769e-01 2.50697523e-01 5.45837402e-01
1.27070531e-01 -5.84040463e-01 8.93362045e-01 -2.10405231e-01
-5.18305242e-01 1.59281164e-01 -1.11920404e+00 -5.18548191e-01
1.39901924e+00 -8.69555175e-01 -1.43135041e-01 5.40894628e-01
-8.96205425e-01 -2.74725817e-02 8.13909948e-01 -3.54228437e-01
6.49638236e-01 -1.07907188e+00 -6.58129215e-01 9.52385738e-02
4.60799873e-01 -1.22332223e-01 1.20301619e-01 9.54003274e-01
-1.44023037e+00 8.63050699e-01 -1.46472633e-01 -4.12165135e-01
-1.64684916e+00 8.41191232e-01 7.09550858e-01 -4.48811352e-01
-4.41583544e-01 7.43287206e-01 3.02027434e-01 -1.38790496e-02
-1.34484526e-02 -1.80912852e-01 -3.94500613e-01 -3.25246185e-01
7.39253163e-01 1.21260524e-01 1.92236185e-01 -4.34265196e-01
-7.27861583e-01 3.76186192e-01 -2.31300905e-01 4.91114885e-01
1.16578138e+00 3.02588880e-01 -1.21879578e-01 -3.62555683e-01
9.53236997e-01 3.73398185e-01 -4.56032217e-01 2.08182827e-01
1.10334203e-01 -5.42140424e-01 -1.15458928e-01 -1.08433652e+00
-3.43599468e-01 -1.26118869e-01 6.42304182e-01 -6.87324047e-01
7.96074152e-01 -3.96926939e-01 3.25166494e-01 3.61887306e-01
1.82010591e-01 -6.78334057e-01 -1.33909628e-01 1.17872156e-01
7.17052400e-01 -7.63252974e-01 2.44694799e-02 -8.07061911e-01
-1.82480782e-01 1.28080392e+00 3.32876325e-01 -1.20467000e-01
3.63509178e-01 1.09191261e-01 -7.59289265e-02 -2.74227649e-01
-5.75333536e-01 3.91893774e-01 2.47244630e-02 4.79599446e-01
9.64972854e-01 6.45172819e-02 -1.45947182e+00 3.98059011e-01
2.12140828e-01 2.70189017e-01 7.20530689e-01 1.16788518e+00
-6.71298444e-01 -1.63333130e+00 -7.01071620e-01 3.92334685e-02
-8.32456768e-01 -2.62802005e-01 -7.45244801e-01 9.81519043e-01
-1.64410904e-01 6.97245121e-01 -6.38566256e-01 4.33080822e-01
4.38036650e-01 -1.02858342e-01 4.91106570e-01 -5.65823495e-01
-8.29703927e-01 7.47329593e-01 3.09694916e-01 -3.55927870e-02
-3.46209258e-01 -4.10806865e-01 -1.96760881e+00 -3.60035002e-01
-9.52199876e-01 7.79691815e-01 7.62090206e-01 6.26614153e-01
-1.10487007e-01 3.12778115e-01 4.99493629e-01 -3.32341999e-01
-1.28615528e-01 -5.17190099e-01 -8.41246784e-01 -2.85348982e-01
5.67920413e-03 -4.23238486e-01 -2.18821868e-01 -1.97727289e-02]
|
[4.663742542266846, 5.130850791931152]
|
90a34453-1452-41a7-8531-fbf6f345782d
|
automated-control-and-optimisation-of-laser
|
2303.00823
| null |
https://arxiv.org/abs/2303.00823v1
|
https://arxiv.org/pdf/2303.00823v1.pdf
|
Automated control and optimisation of laser driven ion acceleration
|
The interaction of relativistically intense lasers with opaque targets represents a highly non-linear, multi-dimensional parameter space. This limits the utility of sequential 1D scanning of experimental parameters for the optimisation of secondary radiation, although to-date this has been the accepted methodology due to low data acquisition rates. High repetition-rate (HRR) lasers augmented by machine learning present a valuable opportunity for efficient source optimisation. Here, an automated, HRR-compatible system produced high fidelity parameter scans, revealing the influence of laser intensity on target pre-heating and proton generation. A closed-loop Bayesian optimisation of maximum proton energy, through control of the laser wavefront and target position, produced proton beams with equivalent maximum energy to manually-optimized laser pulses but using only 60% of the laser energy. This demonstration of automated optimisation of laser-driven proton beams is a crucial step towards deeper physical insight and the construction of future radiation sources.
|
['C. A. J. Palmer', 'N. Xu', 'F. Treffert', 'A. G. R. Thomas', 'D. R. Symes', 'C. Spindloe', 'P. Parsons', 'C. Parisuaña', 'Z. Najmudin', 'P. McKenna', 'O. McCusker', 'D. Margarone', 'M. King', 'V. Istokskaia', 'C. Hyland', 'G. S. Hicks', 'R. J. Gray', 'J. S. Green', 'S. H. Glenzer', 'G. D. Glenn', 'L. Giuffrida', 'M. Gauthier', 'O. C. Ettlinger', 'T. Dzelzanis', 'N. P. Dover', 'S. DiIorio', 'S. J. D. Dann', 'C. B. Curry', 'N. Bourgeois', 'M. Borghesi', 'M. Balcazar', 'S. Astbury', 'H. Ahmed', 'M. J. V. Streeter', 'B. Loughran']
|
2023-03-01
| null | null | null | null |
['bayesian-optimisation']
|
['methodology']
|
[ 4.64070708e-01 2.03097854e-02 -8.07332546e-02 -2.37323225e-01
-9.09918845e-01 -5.26807643e-02 7.95720100e-01 6.72335699e-02
-8.52425754e-01 6.17297053e-01 2.19134018e-01 -5.33017039e-01
-5.47143579e-01 -5.54567695e-01 -2.28392839e-01 -1.27357841e+00
7.43094534e-02 1.28582096e+00 2.96223909e-01 9.52972472e-02
7.81981170e-01 1.23013568e+00 -1.44661021e+00 1.22864932e-01
8.67088512e-02 4.89781857e-01 5.99180460e-01 8.73088956e-01
2.82262892e-01 1.07862711e-01 1.15166586e-02 2.84932256e-01
-1.06158711e-01 -5.50391555e-01 -6.97928667e-01 -3.13120902e-01
-5.87220788e-01 2.01461017e-01 -2.21032336e-01 6.14386797e-01
1.00052047e+00 2.67567307e-01 1.07646310e+00 -5.44918597e-01
1.24550872e-01 5.44176757e-01 -4.67721790e-01 3.91827643e-01
1.73691213e-01 4.51890469e-01 5.29102266e-01 -5.33195972e-01
4.70053107e-01 7.91320622e-01 1.25070438e-01 4.23084378e-01
-1.16826391e+00 -3.93259406e-01 -1.14131773e+00 3.97797316e-01
-1.09624493e+00 -5.72189450e-01 3.21208090e-01 -6.15367234e-01
1.36561251e+00 4.97358948e-01 5.71888506e-01 6.43443048e-01
5.29784143e-01 -4.55969870e-01 1.13234603e+00 -9.75734234e-01
3.42445672e-01 2.95463383e-01 -4.79857385e-01 3.53613853e-01
1.92670077e-02 1.04663777e+00 -4.54962015e-01 -7.37494975e-02
7.61450469e-01 -6.86123967e-01 -6.90048873e-01 -4.27180707e-01
-1.18860686e+00 8.61100614e-01 2.82852918e-01 5.89290142e-01
-4.34866369e-01 -1.86563775e-01 2.27048174e-01 -7.02020675e-02
1.46218225e-01 1.12786829e+00 -5.76288700e-01 -4.72966880e-01
-5.84100068e-01 2.74804175e-01 4.61695522e-01 4.04963344e-01
5.44187367e-01 -1.99290574e-01 -1.86629578e-01 4.26887423e-01
6.02950931e-01 5.55792391e-01 3.14565688e-01 -4.84099895e-01
1.29119650e-01 1.36730775e-01 2.22262982e-02 -1.73435584e-01
-8.92957270e-01 -2.35042036e-01 -5.92388749e-01 6.60958171e-01
4.18806106e-01 -2.48569444e-01 -9.94921267e-01 9.88931298e-01
7.26078629e-01 -2.15703905e-01 2.27246493e-01 1.08107829e+00
4.06736404e-01 6.32415295e-01 1.68837532e-01 -8.82141054e-01
1.21552122e+00 -2.53864288e-01 -4.71866190e-01 -1.80837095e-01
8.33436191e-01 -1.06922305e+00 3.86936069e-01 1.75563857e-01
-1.31331730e+00 -1.27657026e-01 -6.71596289e-01 3.19943875e-01
1.33763611e-01 -3.22048903e-01 5.51349998e-01 7.82462478e-01
-6.37821257e-01 8.55221331e-01 -6.93614662e-01 -1.65076956e-01
3.77693504e-01 5.34666657e-01 -2.20104828e-01 3.29795927e-01
-9.22389388e-01 1.35351861e+00 6.53438747e-01 7.30899861e-03
-3.23662966e-01 -1.10434687e+00 -4.79839832e-01 -2.38612145e-01
2.51497984e-01 -6.92393899e-01 1.61736536e+00 3.54117036e-01
-2.03421569e+00 6.08190298e-01 -1.58447802e-01 -4.16668892e-01
2.63499379e-01 1.41902581e-01 -2.04871684e-01 2.45498389e-01
-2.78537989e-01 -5.21786287e-02 7.83025265e-01 -1.00782430e+00
-4.73923266e-01 -1.84391841e-01 -8.53780270e-01 3.63210976e-01
6.91229939e-01 7.69345522e-01 1.78873405e-01 3.91628563e-01
3.98978084e-01 -7.91863620e-01 -5.60971022e-01 -9.29178238e-01
-3.18314791e-01 -1.03942581e-01 6.25417650e-01 -1.61305815e-01
5.29510856e-01 -1.55917430e+00 4.75094408e-01 4.28091556e-01
-2.00612873e-01 -8.65432806e-03 5.10041416e-01 6.18628204e-01
-6.31595612e-01 -9.35070291e-02 -3.52772295e-01 6.80400804e-02
-3.70424092e-01 -4.46055740e-01 -4.77404818e-02 7.84206271e-01
-2.48865232e-01 8.66902113e-01 -7.57736921e-01 -3.02914053e-01
5.19002497e-01 3.95776778e-01 -2.51260340e-01 2.76498795e-01
-3.12060893e-01 9.16383147e-01 -3.76840830e-01 2.36234978e-01
5.54735661e-01 4.46384192e-01 -2.77099371e-01 -5.07360995e-01
-8.85834217e-01 -8.21164902e-03 -6.35759592e-01 1.29819357e+00
-3.99799138e-01 3.08599651e-01 2.66714208e-02 -4.76282179e-01
8.57289076e-01 5.01982331e-01 6.96465909e-01 -1.07015216e+00
4.89286572e-01 2.78423727e-01 2.96018392e-01 -7.56285310e-01
4.97976154e-01 -1.16857302e+00 1.95873469e-01 5.65450311e-01
-2.67404586e-01 -1.25106466e+00 -2.98069984e-01 -7.40319788e-02
9.31270123e-01 9.27284583e-02 1.95126772e-01 -1.34183004e-01
3.16790998e-01 4.56941910e-02 -1.87848121e-01 7.85563886e-01
3.28215450e-01 8.07326257e-01 -2.58731358e-02 1.48768183e-02
-1.66297662e+00 -5.81606686e-01 -8.14671755e-01 4.22847718e-01
-8.97549242e-02 -1.55104890e-01 -3.09573382e-01 6.60863519e-02
-2.91540712e-01 1.23044586e+00 1.13822734e-02 -3.67280036e-01
-6.16929650e-01 -1.49458706e+00 -1.65359914e-01 7.20936432e-02
-3.23619634e-01 -1.55390418e+00 -9.34668779e-01 1.70717195e-01
2.64834136e-01 -8.43866467e-01 3.97164047e-01 4.59156334e-01
-7.01159418e-01 -1.09277463e+00 -2.67092973e-01 1.23022139e-01
3.27035367e-01 -2.43310332e-01 9.43534255e-01 -9.65205058e-02
-7.99689770e-01 3.82384151e-01 -3.31177533e-01 -8.16742599e-01
-1.21935916e+00 -1.86372697e-01 2.74649620e-01 -6.89446867e-01
1.49635583e-01 -3.78895640e-01 -3.75224978e-01 3.12276125e-01
-5.19638658e-01 5.16244397e-03 1.01171780e+00 5.04100084e-01
6.62188649e-01 6.82082295e-01 4.63149995e-01 -5.88493645e-01
2.92106807e-01 -1.73969090e-01 -8.99411619e-01 -2.61104167e-01
-5.60102999e-01 3.08012456e-01 1.10417672e-01 4.83889841e-02
-1.55710876e+00 -9.29262117e-02 -5.50613105e-01 2.43174955e-01
-5.78819871e-01 3.02550822e-01 -4.83518839e-02 -4.70665872e-01
1.01809621e+00 -7.37819076e-02 1.81360513e-01 -3.59733671e-01
2.90027261e-01 6.75179124e-01 5.72952330e-01 -3.13713700e-01
1.02379596e+00 1.66880816e-01 5.61512470e-01 -1.17232263e+00
-6.64410412e-01 -8.58547270e-01 -8.31758559e-01 -2.40987465e-01
1.02777338e+00 -4.30300266e-01 -9.35099423e-01 3.78604323e-01
-7.82075107e-01 -3.64409387e-01 -4.95325476e-01 1.08364272e+00
-8.22416246e-01 5.01074493e-01 9.90960523e-02 -1.19668174e+00
-3.50830823e-01 -1.00450838e+00 9.19477403e-01 2.81578302e-01
-1.55081794e-01 -5.45462549e-01 2.77523726e-01 5.00882626e-01
5.09126127e-01 1.84609786e-01 1.16016352e+00 -9.41933021e-02
-6.30647779e-01 -3.45697999e-01 1.48746178e-01 -2.73191780e-01
-1.04235813e-01 -2.20124722e-02 -7.59478033e-01 -3.76795530e-01
6.94594026e-01 -3.18175048e-01 6.29808843e-01 6.48205400e-01
1.00520980e+00 4.11164433e-01 -8.13758433e-01 4.63175982e-01
1.57940900e+00 3.19737107e-01 6.88014627e-01 6.62307024e-01
5.02453029e-01 3.61813277e-01 6.12799585e-01 5.09267449e-01
-3.84930640e-01 9.82801974e-01 4.08917964e-01 7.00794160e-02
2.30922610e-01 6.47819579e-01 -1.00231148e-01 6.98162019e-02
-6.33916795e-01 -7.48012261e-03 -9.32157457e-01 1.04074433e-01
-1.20319104e+00 -1.42148232e+00 -8.99283946e-01 2.67840934e+00
1.01461005e+00 3.72252762e-01 -2.95464247e-02 2.71553874e-01
5.12349367e-01 -3.68167877e-01 -1.20725341e-01 -5.03541946e-01
1.59773290e-01 7.49889553e-01 8.58171582e-01 7.49845684e-01
-4.28797573e-01 2.91803062e-01 6.72228098e+00 5.52717865e-01
-9.74951565e-01 -5.95924109e-02 -2.52497852e-01 -4.82038587e-01
-2.61545688e-01 1.62174061e-01 -9.95622516e-01 5.19625008e-01
1.42743576e+00 -3.55618626e-01 3.71354550e-01 1.81175679e-01
6.66309953e-01 -9.03981745e-01 -7.13223338e-01 6.04454517e-01
-3.66985947e-01 -1.37060928e+00 -5.65257430e-01 2.44957507e-01
3.27553004e-01 2.60575742e-01 -3.72473687e-01 1.16944648e-01
-3.15590411e-01 -9.51888859e-01 5.31360447e-01 6.54833376e-01
7.87652314e-01 -1.06291950e+00 6.94894254e-01 6.08357728e-01
-2.95818985e-01 -4.35145386e-02 -5.50914705e-01 1.86820731e-01
9.69588518e-01 9.15278852e-01 -1.54072714e+00 8.77127349e-01
6.38025820e-01 -1.72415167e-01 -3.40782523e-01 1.10196018e+00
9.27698705e-03 5.09051323e-01 -7.74141192e-01 2.98699364e-02
-1.91117644e-01 -4.69167084e-01 7.98275292e-01 9.93953884e-01
3.58194530e-01 5.13647974e-01 -5.01226306e-01 5.25898755e-01
6.29183114e-01 -1.46805108e-01 -4.91867304e-01 1.27981380e-02
1.40775308e-01 1.36713004e+00 -6.37233496e-01 4.81304526e-01
6.65267035e-02 1.51593402e-01 1.48454998e-02 -5.25328517e-02
-6.70573652e-01 6.22431226e-02 2.08545476e-01 3.24712992e-01
2.03852504e-01 -3.60021621e-01 -2.60794044e-01 -1.26653224e-01
-3.94768745e-01 -9.43862349e-02 2.16772839e-01 -8.06829631e-01
-7.57828414e-01 2.22934142e-01 6.47150159e-01 -3.51658523e-01
-5.28862834e-01 -5.30640125e-01 -9.12875593e-01 1.35342813e+00
-1.48690093e+00 -6.69991374e-01 -7.17070699e-02 -4.08100151e-02
5.25924526e-02 -1.91562399e-01 1.03718972e+00 -3.04406077e-01
-2.88570374e-01 -2.16255933e-01 3.85305911e-01 -9.72832620e-01
6.06197000e-01 -1.33408594e+00 3.18025202e-02 2.90556252e-01
-1.50896370e-01 -2.75339484e-01 1.58106112e+00 -7.66000569e-01
-1.69358420e+00 -6.81891680e-01 4.69817787e-01 -5.85227072e-01
5.68584502e-01 -1.30160287e-01 -8.63502502e-01 2.34255642e-01
3.28475028e-01 -3.94989669e-01 6.90913320e-01 1.19995423e-01
6.91545367e-01 4.20690149e-01 -1.21312809e+00 2.50057966e-01
6.79833055e-01 -3.25936824e-01 -5.72967529e-01 8.52280378e-01
3.38470072e-01 -8.19984376e-01 -8.75021994e-01 9.58275259e-01
6.20176904e-02 -1.09543312e+00 1.04770350e+00 -4.48590517e-01
1.32322788e-01 -2.41333693e-02 5.20427346e-01 -1.35644877e+00
-6.62797272e-01 -6.58171237e-01 1.90343320e-01 8.11655045e-01
5.67636847e-01 -3.27432811e-01 9.60807562e-01 6.01307690e-01
-5.14929831e-01 -7.91651726e-01 -1.20123065e+00 -2.27494746e-01
-4.10525799e-02 -4.90196884e-01 6.05680831e-02 2.20763952e-01
1.43813193e-01 2.61884421e-01 8.72765332e-02 6.23888254e-01
7.12009490e-01 1.25423104e-01 3.92468929e-01 -1.14113104e+00
-4.27150756e-01 -2.91222721e-01 -2.92497903e-01 9.09455586e-03
-3.55393700e-02 -9.70838070e-01 2.99483925e-01 -1.36914623e+00
1.24651670e-01 -7.54860342e-01 1.84616417e-01 -6.74558580e-02
1.58647880e-01 7.68775046e-02 -3.95069331e-01 1.89891428e-01
3.30140769e-01 6.29558921e-01 1.28064311e+00 3.43225926e-01
-5.27393222e-01 4.44397867e-01 -2.44293809e-01 4.65829611e-01
9.34678257e-01 -6.22138262e-01 -1.11327671e-01 3.55164856e-01
3.51898164e-01 1.11297533e-01 3.23578238e-01 -9.51701641e-01
2.31332898e-01 -4.67681140e-01 4.46489692e-01 -7.26582527e-01
3.54381263e-01 -7.09948301e-01 6.64922953e-01 1.86543614e-01
-3.44934128e-03 -5.51142871e-01 2.13771001e-01 3.72876763e-01
2.24846989e-01 -1.10575795e+00 1.16295123e+00 -4.03395385e-01
-2.01838046e-01 5.75616397e-02 -3.79332095e-01 -4.73427564e-01
1.25076151e+00 1.45010874e-02 -7.88633153e-02 2.33143345e-01
-6.82243824e-01 -2.51492620e-01 4.83083457e-01 8.40640739e-02
4.06569451e-01 -8.53143275e-01 -7.33655810e-01 1.73071846e-01
-7.09692240e-02 4.67950135e-01 5.86542547e-01 9.52724099e-01
-5.66772103e-01 6.65252805e-01 1.11793526e-01 -7.79174328e-01
-1.42129683e+00 4.74268019e-01 6.69398487e-01 -5.62520586e-02
-8.46868098e-01 9.32906210e-01 -2.57160425e-01 -1.97533518e-02
-4.40538347e-01 3.22647065e-01 -3.80099922e-01 -1.62568405e-01
4.06644911e-01 1.06217965e-01 6.24947667e-01 -6.85616791e-01
-1.40450979e-02 5.00490427e-01 -6.76992610e-02 -5.90453967e-02
1.48155999e+00 6.53824257e-03 -5.32928482e-02 1.98437348e-01
7.26072431e-01 -7.05994964e-02 -9.73679900e-01 1.51826218e-01
5.26911467e-02 -5.94708622e-01 6.48309529e-01 -8.07108581e-01
-2.11921543e-01 3.20763677e-01 5.09241760e-01 9.35620517e-02
8.27437103e-01 5.52693427e-01 -1.00248000e-02 3.38084787e-01
3.97833943e-01 -9.87207532e-01 -2.61244953e-01 3.59351277e-01
9.83191550e-01 -1.01686108e+00 4.94827390e-01 -4.93394583e-01
-3.99742335e-01 1.23855853e+00 4.09943849e-01 6.45935714e-01
4.89257395e-01 4.19281423e-01 -3.17297399e-01 -5.44405580e-01
-6.63847625e-01 -2.80953087e-02 8.11132789e-02 6.64298534e-01
1.38760820e-01 -1.72535926e-01 -3.50245208e-01 -3.88634145e-01
-4.84274983e-01 -1.72940940e-01 8.63296032e-01 1.04748654e+00
-8.29712987e-01 -1.27946508e+00 -7.19499469e-01 5.02742589e-01
-2.41531983e-01 5.82612418e-02 2.31455460e-01 4.70306158e-01
-2.18206063e-01 5.85618317e-01 -1.07808337e-01 4.02724475e-01
5.85535169e-01 5.01830041e-01 9.80921984e-01 -8.17110479e-01
-3.57347906e-01 -9.85199288e-02 1.43438637e-01 -2.06260026e-01
-6.09742403e-01 -9.00018573e-01 -1.42109275e+00 1.14807487e-01
-8.92716587e-01 4.84032631e-01 1.52101362e+00 1.32403004e+00
-1.08496226e-01 3.38486522e-01 8.50977063e-01 -1.44674325e+00
-7.44925857e-01 -9.93967533e-01 -6.33850455e-01 -7.11797476e-02
1.25481850e-02 -7.51040399e-01 -9.41442847e-01 -3.60187322e-01]
|
[6.289713382720947, 3.683126449584961]
|
ea3ebcbf-6d36-4027-a7a0-91ab79cad62a
|
action-agnostic-human-pose-forecasting
|
1810.09676
| null |
http://arxiv.org/abs/1810.09676v1
|
http://arxiv.org/pdf/1810.09676v1.pdf
|
Action-Agnostic Human Pose Forecasting
|
Predicting and forecasting human dynamics is a very interesting but
challenging task with several prospective applications in robotics,
health-care, etc. Recently, several methods have been developed for human pose
forecasting; however, they often introduce a number of limitations in their
settings. For instance, previous work either focused only on short-term or
long-term predictions, while sacrificing one or the other. Furthermore, they
included the activity labels as part of the training process, and require them
at testing time. These limitations confine the usage of pose forecasting models
for real-world applications, as often there are no activity-related annotations
for testing scenarios. In this paper, we propose a new action-agnostic method
for short- and long-term human pose forecasting. To this end, we propose a new
recurrent neural network for modeling the hierarchical and multi-scale
characteristics of the human dynamics, denoted by triangular-prism RNN
(TP-RNN). Our model captures the latent hierarchical structure embedded in
temporal human pose sequences by encoding the temporal dependencies with
different time-scales. For evaluation, we run an extensive set of experiments
on Human 3.6M and Penn Action datasets and show that our method outperforms
baseline and state-of-the-art methods quantitatively and qualitatively. Codes
are available at https://github.com/eddyhkchiu/pose_forecast_wacv/
|
['De-An Huang', 'Borui Wang', 'Hsu-kuang Chiu', 'Juan Carlos Niebles', 'Ehsan Adeli']
|
2018-10-23
| null | null | null | null |
['human-pose-forecasting', 'human-dynamics']
|
['computer-vision', 'computer-vision']
|
[ 6.21562488e-02 -1.01608373e-01 -1.91812932e-01 -4.46963012e-01
-3.91528100e-01 -1.46136060e-01 3.41548622e-01 -3.79025877e-01
-3.71052116e-01 6.55196786e-01 6.47111893e-01 2.23363653e-01
-6.95304526e-03 -4.61120814e-01 -7.84959376e-01 -5.45914888e-01
-3.88146073e-01 5.20343065e-01 4.25979644e-01 -3.47106785e-01
-1.21811837e-01 1.77003592e-01 -1.68047547e+00 3.48957479e-01
5.17232239e-01 9.59313869e-01 1.54778019e-01 5.40282905e-01
4.61559147e-01 1.00390172e+00 -5.22021174e-01 -1.53965667e-01
1.04456022e-01 -4.53098804e-01 -7.43444026e-01 1.42386988e-01
9.12640691e-02 -3.47229809e-01 -6.94022238e-01 5.50593853e-01
7.67260730e-01 3.90863746e-01 5.08213043e-01 -1.39831781e+00
-2.88316876e-01 3.62835199e-01 -3.57799530e-01 4.49527539e-02
6.26742840e-01 2.56989121e-01 6.60340607e-01 -6.73798382e-01
5.51827908e-01 1.19373667e+00 8.47888529e-01 7.21753120e-01
-7.19421685e-01 -5.75784326e-01 3.79797608e-01 4.60477561e-01
-1.25991249e+00 -3.36728245e-01 8.97965193e-01 -6.22216940e-01
9.55406308e-01 5.60874157e-02 9.94126499e-01 1.76530492e+00
5.09925187e-01 1.14378273e+00 8.40988457e-01 -5.83377369e-02
-3.58186923e-02 -4.55621481e-01 -1.16427034e-01 7.42200136e-01
-1.46138981e-01 2.09636733e-01 -6.15322232e-01 7.88809955e-02
8.68260086e-01 2.64271200e-01 -2.57868141e-01 -4.30436969e-01
-1.43157375e+00 6.05834305e-01 5.19682169e-01 2.27107361e-01
-4.88696754e-01 2.71534801e-01 6.09220743e-01 -1.03201777e-01
4.83209729e-01 4.15005684e-02 -5.82737803e-01 -3.91710967e-01
-5.52601397e-01 3.80647600e-01 5.53400695e-01 9.32554185e-01
1.07981227e-01 -1.85977682e-01 -2.58235514e-01 7.86121011e-01
1.88115597e-01 3.20257217e-01 7.04158008e-01 -9.49412525e-01
5.01378179e-01 4.61698383e-01 2.22910658e-01 -1.20761859e+00
-9.96465802e-01 -2.70046949e-01 -1.03622484e+00 -2.02568755e-01
4.32590574e-01 -2.16704026e-01 -9.65295315e-01 1.88136590e+00
4.31953281e-01 2.29944959e-01 -2.56552607e-01 1.18273556e+00
6.53377771e-01 5.82918882e-01 1.63800761e-01 -1.22206993e-01
1.30498040e+00 -1.32928133e+00 -8.70732725e-01 -3.34309310e-01
5.71665943e-01 -4.24900055e-01 9.45170581e-01 4.72975105e-01
-8.80420864e-01 -8.43529582e-01 -7.93962896e-01 -7.63719752e-02
-1.54462576e-01 3.32217008e-01 6.82726324e-01 1.03292353e-01
-6.53109848e-01 8.09743881e-01 -1.33621812e+00 -4.93211418e-01
3.52365784e-02 2.10034370e-01 -3.08201581e-01 3.11220884e-01
-1.39905870e+00 9.74774003e-01 1.18541077e-01 6.44986093e-01
-7.76158750e-01 -2.03864519e-02 -9.84822452e-01 -3.95558327e-01
5.56110561e-01 -6.48590803e-01 1.42204618e+00 -5.40721118e-01
-1.66456699e+00 4.47265774e-01 -1.36322916e-01 -4.95229602e-01
8.55828702e-01 -8.12988639e-01 -4.00151372e-01 -9.79431197e-02
1.28493845e-01 7.24333823e-01 5.63919961e-01 -8.47153842e-01
-5.55872023e-01 -4.12423879e-01 5.47705777e-02 3.26281905e-01
-2.62159407e-01 -1.60712972e-01 -6.75501406e-01 -9.39735293e-01
1.55798346e-01 -1.37554610e+00 -4.05294150e-01 -1.34474263e-01
-3.99314970e-01 -2.53281772e-01 5.23526430e-01 -8.55416596e-01
1.39264572e+00 -1.87597895e+00 5.98447382e-01 -1.98838934e-01
-2.79661249e-02 8.33697170e-02 1.68753285e-02 4.44044292e-01
1.05390608e-01 -2.98191607e-01 -2.04378888e-01 -4.97491091e-01
-1.46470899e-02 3.28259617e-01 -1.94587678e-01 4.94737625e-01
-2.23131418e-01 1.05019462e+00 -7.62908161e-01 -5.35056531e-01
4.38144773e-01 7.06357896e-01 -4.61122572e-01 2.36624673e-01
-1.89737260e-01 1.00020003e+00 -3.32849383e-01 5.68603516e-01
2.64267661e-02 -3.03930998e-01 1.79694280e-01 -1.98722675e-01
-3.07770651e-02 2.21775204e-01 -1.14326692e+00 1.90548718e+00
-2.95806795e-01 3.48080337e-01 -4.70711857e-01 -8.76953483e-01
5.19081175e-01 4.99953419e-01 9.29702699e-01 -5.96034825e-01
3.32431495e-01 1.03373222e-01 7.70851132e-03 -7.36531973e-01
4.65549022e-01 1.22559845e-01 -2.74626315e-01 1.40576228e-01
-3.53416204e-02 4.78302598e-01 1.19034328e-01 -2.59953231e-01
1.08009791e+00 6.06590867e-01 2.34869257e-01 1.10941835e-01
3.78557742e-01 -2.37626880e-01 8.92590344e-01 3.55891705e-01
-6.08360529e-01 8.07962954e-01 2.34944999e-01 -7.04568088e-01
-7.68106222e-01 -8.11756849e-01 1.89990163e-01 1.29836965e+00
2.50076532e-01 -5.51442444e-01 -6.95345521e-01 -6.69294417e-01
-1.51663810e-01 1.68674216e-01 -7.89482594e-01 -2.07691699e-01
-1.02079713e+00 -5.34060001e-01 6.01546407e-01 1.00831974e+00
4.84648257e-01 -1.58710098e+00 -9.82011378e-01 3.52557689e-01
-7.88896859e-01 -1.28829443e+00 -6.47937596e-01 -5.48991896e-02
-8.31613064e-01 -1.02850235e+00 -9.07670617e-01 -7.93208659e-01
4.87491369e-01 -6.50765374e-02 8.80167723e-01 -2.49521956e-01
-2.49034032e-01 3.35390031e-01 -5.13960421e-01 -2.91570276e-01
1.37232035e-01 1.21627696e-01 5.05773246e-01 -2.74953842e-02
1.62475660e-01 -6.02232099e-01 -7.84043968e-01 6.71801150e-01
-4.08350885e-01 1.64927989e-01 6.01896226e-01 8.29808056e-01
7.36661553e-01 -1.50103226e-01 3.68273139e-01 -6.43700004e-01
4.34240967e-01 -1.79888010e-01 -2.28639603e-01 1.30349919e-01
-2.04772249e-01 -6.04735278e-02 4.99904364e-01 -7.84652650e-01
-9.17917490e-01 4.69493717e-01 -3.20343733e-01 -5.66515028e-01
-1.42157078e-01 5.29092968e-01 -1.58450469e-01 2.57343084e-01
5.77970028e-01 1.55359402e-01 -8.71028677e-02 -4.45869267e-01
1.03921749e-01 3.38550717e-01 6.51535153e-01 -5.01187146e-01
5.01932919e-01 3.88144940e-01 -8.60877782e-02 -6.02603555e-01
-1.02761185e+00 -5.00482619e-01 -9.51288283e-01 -5.30560017e-01
9.62414265e-01 -1.00161195e+00 -9.11240637e-01 7.76757836e-01
-1.18775499e+00 -6.55692518e-01 -6.23432584e-02 5.81718445e-01
-9.33827877e-01 2.24038094e-01 -9.16971743e-01 -8.73028338e-01
-1.95934579e-01 -1.10176206e+00 1.12542331e+00 -8.06607679e-02
-5.31464756e-01 -6.34567738e-01 1.68846145e-01 4.70754832e-01
1.38274074e-01 6.77035868e-01 2.37913430e-01 -2.36387476e-01
-2.16750786e-01 -3.11647832e-01 3.43862623e-01 1.20050341e-01
8.91335234e-02 -3.97883117e-01 -6.58384264e-01 -2.52350330e-01
-1.04569174e-01 -5.01586318e-01 6.89609528e-01 6.73727334e-01
1.40129459e+00 -2.95739591e-01 -4.50551003e-01 4.34813172e-01
6.55840456e-01 4.92594764e-02 6.96616530e-01 2.08602667e-01
1.00244319e+00 9.74278331e-01 8.93792987e-01 6.10446215e-01
6.79907382e-01 1.00959778e+00 1.72815293e-01 1.25980973e-01
1.14331782e-01 -5.41835785e-01 5.36443114e-01 9.91463900e-01
-6.06405139e-01 -1.70211941e-01 -9.57630634e-01 3.54276866e-01
-2.39819646e+00 -1.04316294e+00 8.21068604e-03 2.05301881e+00
6.56125784e-01 2.48551115e-01 5.84889770e-01 1.85557932e-01
5.91338575e-01 2.83788979e-01 -6.85927868e-01 1.58138588e-01
2.29235128e-01 -2.59088814e-01 2.83966780e-01 3.48927289e-01
-1.50458133e+00 9.00236309e-01 5.58989763e+00 4.38997000e-01
-1.05853772e+00 1.22468628e-01 4.05762941e-01 -3.67773086e-01
4.93693292e-01 -4.68785524e-01 -7.42198288e-01 5.23014367e-01
8.57540309e-01 2.40181491e-01 1.46314234e-01 8.68072510e-01
3.71259809e-01 8.40014517e-02 -1.13858831e+00 1.08215022e+00
1.08725451e-01 -7.59620488e-01 -2.02554613e-01 -3.08219390e-03
4.05049950e-01 3.01800924e-03 -1.26867160e-01 5.11374235e-01
-8.49183798e-02 -9.19296443e-01 9.46815908e-01 7.33691692e-01
5.80652416e-01 -6.97814047e-01 7.56473541e-01 6.58775628e-01
-1.56743038e+00 -2.68972695e-01 -6.56376034e-03 -2.89011389e-01
6.11373007e-01 3.37699980e-01 -2.14637890e-01 3.54173481e-01
9.34001148e-01 1.01705718e+00 -5.67239940e-01 8.29620123e-01
-4.09181744e-01 3.27834368e-01 -2.59118497e-01 -1.32679855e-02
-3.99243720e-02 3.91999148e-02 4.11949277e-01 1.07367098e+00
2.39824966e-01 2.93046087e-01 5.98799646e-01 3.63609940e-01
1.15704224e-01 -1.71396971e-01 -5.87728560e-01 7.35687166e-02
1.26055494e-01 9.86197591e-01 -7.29992211e-01 -1.06008768e-01
-1.62567586e-01 1.17477298e+00 2.64939696e-01 1.94793805e-01
-1.19836819e+00 -1.26684651e-01 4.12103236e-01 2.71128476e-01
-2.05460545e-02 -4.54125375e-01 -9.23491642e-02 -1.31526530e+00
4.82657880e-01 -8.83903861e-01 5.07027209e-01 -5.49543858e-01
-1.16580713e+00 6.27321362e-01 1.81330666e-01 -1.61108124e+00
-7.10896075e-01 -5.10061145e-01 -1.39111415e-01 2.81764686e-01
-8.74536574e-01 -1.24444604e+00 -4.57992107e-01 5.58037162e-01
7.20004380e-01 1.49868771e-01 8.08596492e-01 4.12547976e-01
-5.74225307e-01 5.11900246e-01 -4.16743398e-01 4.25034553e-01
7.04496920e-01 -9.54234898e-01 5.46865642e-01 6.55905068e-01
6.84813783e-02 5.96349120e-01 9.37811673e-01 -8.35150361e-01
-1.08984447e+00 -1.10392773e+00 9.58852530e-01 -8.14784646e-01
4.27914172e-01 -5.78742325e-01 -6.99551046e-01 9.54890430e-01
-1.40900820e-01 1.13535330e-01 4.81931478e-01 2.24453196e-01
1.65767893e-02 1.68636993e-01 -6.85331285e-01 6.97569847e-01
1.65900791e+00 -2.97217160e-01 -5.25036454e-01 4.64586854e-01
6.98760033e-01 -7.63916910e-01 -9.31940019e-01 9.30433273e-01
1.08757627e+00 -9.06966448e-01 9.32333291e-01 -4.55086052e-01
5.72152793e-01 -2.34112322e-01 -1.31935373e-01 -1.02909398e+00
-4.91395414e-01 -4.13048625e-01 -5.90533733e-01 6.89097106e-01
1.66948780e-01 -3.84235173e-01 9.85316575e-01 5.02546430e-01
-1.39438167e-01 -1.25450647e+00 -8.09658349e-01 -8.20409834e-01
-2.55630851e-01 -4.43406701e-01 2.42301866e-01 6.30074978e-01
7.23368302e-02 3.87960792e-01 -1.10447586e+00 -7.08457839e-04
3.61175954e-01 1.40604544e-02 8.76491010e-01 -1.17581713e+00
-3.93428206e-01 -2.00111225e-01 -5.33748507e-01 -1.32910383e+00
1.71952590e-01 -2.91202396e-01 4.96433347e-01 -1.59342408e+00
5.68814948e-02 -1.60368823e-03 -2.19430938e-01 6.07659400e-01
-3.65989143e-03 1.97504878e-01 2.25552768e-01 3.67142111e-01
-9.23284471e-01 9.20488656e-01 1.16691923e+00 1.08328527e-02
-3.09605241e-01 3.27796847e-01 -1.21547863e-01 1.06823123e+00
7.30492771e-01 -2.68684655e-01 -4.61320102e-01 -3.22714239e-01
1.52058275e-02 3.11312139e-01 4.93522137e-01 -1.45732129e+00
3.23206037e-01 -1.15676895e-01 5.20682693e-01 -8.28524232e-01
6.25691473e-01 -6.15715802e-01 2.24022999e-01 7.01065660e-01
-4.85008419e-01 2.62060314e-01 -7.22487867e-02 7.16478765e-01
-1.98777005e-01 3.38186294e-01 4.56916571e-01 -1.96137726e-01
-8.65850508e-01 6.76217318e-01 -2.66742885e-01 2.80354470e-02
9.64576602e-01 -2.18129382e-01 5.10119237e-02 -4.98327017e-01
-9.37537670e-01 4.34846759e-01 1.58269182e-01 8.80180657e-01
6.03068531e-01 -1.55949032e+00 -4.14509952e-01 -1.03974909e-01
3.50636870e-01 1.18266284e-01 4.82705057e-01 1.05089080e+00
-2.46338814e-01 5.07249892e-01 -3.52705121e-01 -6.39848948e-01
-1.06300032e+00 4.50348556e-01 3.30303967e-01 -3.70583594e-01
-9.14950013e-01 6.88992798e-01 1.58595949e-01 -6.99402928e-01
6.83482528e-01 -3.45104605e-01 -4.13011432e-01 -1.17277913e-02
3.51714849e-01 4.40203696e-01 -2.57728219e-01 -1.05907547e+00
-5.19114673e-01 6.58630311e-01 2.11961091e-01 -1.96565434e-01
1.31820571e+00 -1.28515378e-01 1.92835703e-01 9.29065228e-01
9.77264702e-01 -4.79539543e-01 -1.40423632e+00 -1.67508274e-01
2.75861025e-02 -1.23077311e-01 -5.47352076e-01 -7.29542136e-01
-1.05403876e+00 8.23497176e-01 6.65913343e-01 -2.49755591e-01
1.05869091e+00 -9.45230424e-02 1.17451847e+00 4.95026946e-01
8.36084545e-01 -1.41363657e+00 3.65893811e-01 7.18330979e-01
1.19940615e+00 -1.21472645e+00 -5.09872250e-02 -4.02094930e-01
-9.44812059e-01 8.37271512e-01 9.41879749e-01 -1.32993653e-01
7.00196505e-01 4.11629826e-02 1.31826460e-01 -1.08606994e-01
-7.49684155e-01 -1.82542175e-01 4.79527533e-01 3.42522413e-01
7.11420953e-01 2.56378442e-01 -4.28305507e-01 8.07080269e-01
-4.21935230e-01 3.32005948e-01 1.38830766e-01 1.06631148e+00
-2.58246541e-01 -8.69105995e-01 -1.43479392e-01 3.76968414e-01
-3.17136765e-01 4.49784577e-01 -3.04328650e-01 7.20027328e-01
2.89682746e-01 8.63305807e-01 -1.81029245e-01 -9.52756286e-01
6.90866232e-01 3.85078825e-02 5.72100103e-01 -4.03028578e-01
-4.02374715e-01 1.81706414e-01 1.36229768e-01 -9.96350706e-01
-6.05850160e-01 -8.38335991e-01 -1.41217864e+00 -1.51287317e-01
-9.01247486e-02 -1.12873569e-01 2.78118163e-01 9.95236516e-01
3.95329893e-01 8.04884195e-01 2.43974403e-01 -1.28919375e+00
-4.30072635e-01 -1.26736856e+00 -3.57869476e-01 6.21906221e-01
2.14099601e-01 -1.16970277e+00 -1.70587972e-02 8.57076943e-02]
|
[7.2829718589782715, -0.2952585816383362]
|
ecb53d1d-0e00-47ac-ba0b-abfbf9987756
|
improving-code-switching-and-named-entity
|
2306.08588
| null |
https://arxiv.org/abs/2306.08588v1
|
https://arxiv.org/pdf/2306.08588v1.pdf
|
Improving Code-Switching and Named Entity Recognition in ASR with Speech Editing based Data Augmentation
|
Recently, end-to-end (E2E) automatic speech recognition (ASR) models have made great strides and exhibit excellent performance in general speech recognition. However, there remain several challenging scenarios that E2E models are not competent in, such as code-switching and named entity recognition (NER). Data augmentation is a common and effective practice for these two scenarios. However, the current data augmentation methods mainly rely on audio splicing and text-to-speech (TTS) models, which might result in discontinuous, unrealistic, and less diversified speech. To mitigate these potential issues, we propose a novel data augmentation method by applying the text-based speech editing model. The augmented speech from speech editing systems is more coherent and diversified, also more akin to real speech. The experimental results on code-switching and NER tasks show that our proposed method can significantly outperform the audio splicing and neural TTS based data augmentation systems.
|
['Xie Chen', 'Kai Yu', 'Chenpeng Du', 'Ziyang Ma', 'Zheshu Song', 'Zheng Liang']
|
2023-06-14
| null | null | null | null |
['named-entity-recognition-ner', 'automatic-speech-recognition']
|
['natural-language-processing', 'speech']
|
[ 3.59299839e-01 1.27186313e-01 1.76597178e-01 -5.11152148e-01
-8.11437309e-01 -1.87404767e-01 5.66015422e-01 -2.82880843e-01
-3.97385240e-01 4.77467895e-01 5.44487059e-01 -5.74178994e-01
3.90108883e-01 -1.95136502e-01 -4.28528696e-01 -3.74407470e-01
3.00422519e-01 1.65424988e-01 -1.58714354e-02 -4.38764155e-01
-6.18138686e-02 2.28384972e-01 -1.50430059e+00 2.72478074e-01
1.25767446e+00 6.43566847e-01 3.51394087e-01 6.14134252e-01
-7.79612422e-01 5.79719841e-01 -9.33908343e-01 -4.50201392e-01
1.07028179e-01 -4.05741453e-01 -4.83184308e-01 2.98005074e-01
7.58600906e-02 -1.31014302e-01 -6.60079420e-01 1.13378096e+00
7.57170618e-01 4.01108921e-01 8.51984546e-02 -1.11554813e+00
-8.47987711e-01 8.78580153e-01 -4.67073560e-01 1.10831358e-01
3.15117896e-01 -8.99234563e-02 5.22898674e-01 -1.32789207e+00
2.67582804e-01 1.25869513e+00 5.87834358e-01 8.78812969e-01
-8.53066146e-01 -9.06373918e-01 3.95346642e-01 -3.50327194e-02
-1.38911045e+00 -1.13459158e+00 7.59953916e-01 -1.43016502e-01
9.25865471e-01 3.56499434e-01 3.80110502e-01 1.34699416e+00
-5.64982653e-01 1.13537872e+00 8.86847079e-01 -4.80768740e-01
8.98318887e-02 1.86865389e-01 -1.02105163e-01 1.29821599e-01
-3.71898115e-01 1.89943090e-01 -6.75534308e-01 7.66411126e-02
6.31263196e-01 -5.96161373e-02 -4.49688017e-01 4.25019264e-01
-1.26583135e+00 3.54143918e-01 -1.25028893e-01 4.95992541e-01
-3.72603565e-01 -3.32688957e-01 6.85904920e-01 3.98278892e-01
4.74084705e-01 1.81624770e-01 -5.89833260e-01 -6.09281719e-01
-1.11343491e+00 -2.83342272e-01 4.81871277e-01 1.33986950e+00
2.92012066e-01 8.81627262e-01 -2.16244265e-01 1.53483558e+00
3.44522059e-01 5.24376690e-01 8.74674201e-01 -3.33140969e-01
8.01196754e-01 2.62700617e-01 -1.54723242e-01 -4.03439432e-01
1.74574722e-02 -6.58701301e-01 -1.19676447e+00 -1.51184827e-01
-1.53269812e-01 -2.46984556e-01 -1.35337698e+00 1.72852695e+00
1.52358621e-01 4.39137161e-01 2.82762110e-01 6.10994101e-01
9.37201202e-01 9.47615325e-01 3.44141603e-01 -3.81378800e-01
9.92069900e-01 -1.09727097e+00 -1.43845391e+00 -3.91440332e-01
8.10597062e-01 -9.99320924e-01 1.28897572e+00 2.49014765e-01
-9.57142711e-01 -8.01072836e-01 -8.41598213e-01 7.15938210e-02
-2.65325487e-01 4.69348341e-01 2.86267132e-01 9.38170671e-01
-8.59478712e-01 9.03978795e-02 -6.86626315e-01 -2.12771744e-01
1.43669173e-01 1.66371554e-01 -4.33893889e-01 1.44475773e-01
-1.44589281e+00 6.30398750e-01 4.57830310e-01 4.11891371e-01
-6.99365079e-01 -5.33097386e-01 -9.11494374e-01 -5.97550860e-03
4.77288127e-01 -1.58898845e-01 1.47045934e+00 -1.02499354e+00
-1.80153847e+00 4.90515500e-01 -3.37197304e-01 -3.17729592e-01
2.18805701e-01 -1.98153108e-01 -1.12924492e+00 -4.73389536e-01
-3.28170419e-01 3.87171090e-01 7.79501379e-01 -1.15419316e+00
-5.72158754e-01 -3.70949060e-01 -4.54288512e-01 3.45492333e-01
-5.55781841e-01 4.97295111e-01 -4.30397242e-01 -1.24851346e+00
2.08539471e-01 -7.55998850e-01 -9.83350873e-02 -4.52595294e-01
-5.71907222e-01 -1.05221674e-01 9.68955100e-01 -1.11606812e+00
1.66122675e+00 -2.51394200e+00 -1.71065405e-01 -3.09941154e-02
-2.23458022e-01 9.35323656e-01 -3.31519455e-01 4.27889049e-01
-3.45568359e-01 3.45380604e-01 -3.93480390e-01 -6.77524269e-01
-1.80518731e-01 2.29317054e-01 -5.13319433e-01 -2.17022881e-01
1.68812141e-01 4.82939601e-01 -6.22589827e-01 -2.27562770e-01
2.28874683e-01 4.87012774e-01 -2.75840044e-01 3.83617938e-01
2.38027927e-02 7.11294770e-01 -1.73141047e-01 5.50205350e-01
7.09747314e-01 4.43990171e-01 -1.05944477e-01 1.08552083e-01
-2.94785738e-01 6.01007283e-01 -1.23384631e+00 1.75663650e+00
-7.59337962e-01 5.51047623e-01 3.32576126e-01 -8.36799800e-01
1.23861217e+00 7.26625919e-01 -2.75625731e-03 -6.07971191e-01
6.50695562e-02 3.13837230e-01 9.94852483e-02 -5.43915093e-01
5.94646692e-01 -2.24547148e-01 2.61844724e-01 -1.07696643e-02
3.81091461e-02 1.75983012e-02 -1.45255178e-01 1.10358015e-01
9.81872261e-01 -1.70654044e-01 8.26510638e-02 2.40711436e-01
6.04250789e-01 -4.72611487e-01 8.92668009e-01 5.01825929e-01
-2.93525100e-01 7.30188251e-01 -7.43029937e-02 1.19733982e-01
-1.13207901e+00 -8.83030891e-01 1.22682564e-01 1.06243789e+00
-3.00836504e-01 -4.40720350e-01 -7.96055198e-01 -6.09312057e-01
-4.96824503e-01 9.91430163e-01 -5.54064773e-02 -3.15888822e-01
-5.26756108e-01 -3.47706497e-01 1.03750134e+00 6.10222936e-01
7.04044104e-01 -1.08737600e+00 6.49284840e-01 5.63694119e-01
-5.74640572e-01 -1.31139708e+00 -9.38183784e-01 4.01225872e-02
-8.45212102e-01 -3.35243821e-01 -9.08173501e-01 -9.31687415e-01
5.02069712e-01 4.48039442e-01 5.68355501e-01 2.00326107e-02
1.90248817e-01 -5.36522944e-04 -8.39642107e-01 -4.99488741e-01
-9.91102755e-01 1.10667355e-01 3.57081860e-01 3.60884666e-01
2.92272836e-01 -5.59000373e-01 -1.83916822e-01 4.81498688e-01
-1.07251251e+00 3.74658369e-02 8.80856633e-01 8.66123676e-01
5.40650368e-01 -2.60812207e-03 1.12525249e+00 -7.49071360e-01
7.72235751e-01 -2.96737164e-01 -2.96783537e-01 4.46081877e-01
-4.93747413e-01 -8.09779763e-02 9.83093262e-01 -6.87322557e-01
-1.63407540e+00 1.36797771e-01 -8.43285978e-01 -5.63871920e-01
-5.08302927e-01 6.80432200e-01 -6.86952412e-01 9.52515230e-02
4.78048116e-01 7.15933263e-01 -1.20158747e-01 -8.66585910e-01
2.51074880e-01 1.59759104e+00 7.60978103e-01 -3.30549270e-01
8.30815434e-01 -1.35897428e-01 -8.03873599e-01 -1.35862410e+00
-4.02129918e-01 -6.34844840e-01 -4.43630785e-01 -5.73468022e-02
5.41882992e-01 -1.02969515e+00 -8.16813186e-02 8.08901072e-01
-1.44141281e+00 -8.01079124e-02 -2.43114874e-01 7.82415688e-01
-1.59969166e-01 6.66378736e-01 -4.52056497e-01 -1.24197352e+00
-3.42041343e-01 -1.10979414e+00 9.40504611e-01 1.87591374e-01
7.59058446e-02 -5.78458250e-01 -1.51118889e-01 5.68274617e-01
6.51906788e-01 -4.57429945e-01 6.95747554e-01 -1.11272860e+00
-2.44800746e-01 -9.32881236e-02 1.21092774e-01 7.97809720e-01
3.00759077e-01 1.84252709e-01 -1.13870978e+00 -1.12921141e-01
7.36049339e-02 -4.74854670e-02 4.43101108e-01 -8.93219262e-02
1.23283899e+00 -3.64361733e-01 -2.34245081e-02 4.66367155e-01
6.65215433e-01 8.18766475e-01 8.98047388e-01 -2.74614971e-02
5.77451229e-01 6.52877629e-01 6.76157534e-01 3.75478327e-01
1.06827863e-01 6.94683790e-01 2.52725333e-02 -2.63958991e-01
-4.09153283e-01 -4.79166865e-01 6.42645419e-01 1.79144979e+00
3.26411754e-01 -3.15692484e-01 -1.03217256e+00 7.42004871e-01
-1.49357581e+00 -9.17617500e-01 -2.17984304e-01 2.33666253e+00
1.03267086e+00 8.75824764e-02 -7.43788853e-02 3.75337332e-01
1.28220534e+00 -5.07411268e-03 -4.55815345e-01 -3.60853016e-01
-3.26639384e-01 1.74368814e-01 1.07560195e-01 1.82983443e-01
-9.05218482e-01 1.02940798e+00 6.13259459e+00 1.19006491e+00
-9.56099093e-01 2.77821332e-01 4.02950019e-01 2.23090842e-01
-3.67850900e-01 -3.65376957e-02 -8.33700299e-01 6.46153867e-01
1.14569759e+00 -3.50564681e-02 5.06939292e-01 7.79005349e-01
3.90062809e-01 5.36719918e-01 -8.38788033e-01 1.21171319e+00
-1.53958932e-01 -9.80517626e-01 3.28527302e-01 -1.75332829e-01
5.68346858e-01 1.10306427e-01 4.09128293e-02 7.30460286e-01
1.23027600e-01 -7.48701155e-01 6.53729081e-01 1.67009220e-01
1.09174252e+00 -6.33649111e-01 7.69962072e-01 5.18732488e-01
-1.32830071e+00 -2.10764762e-02 8.68830271e-03 4.43420887e-01
3.96337807e-01 5.45734167e-01 -8.33397985e-01 5.81839383e-01
3.87411058e-01 2.53333122e-01 -2.79895931e-01 1.18047154e+00
-1.11523315e-01 9.52968895e-01 -3.34081918e-01 1.39627919e-01
5.69969453e-02 -1.18772119e-01 7.95227706e-01 1.35955822e+00
4.95224863e-01 2.01415971e-01 -6.17532469e-02 5.28768003e-01
-3.13848585e-01 3.05492073e-01 -6.21946216e-01 -5.95912158e-01
9.93048608e-01 8.81642044e-01 -2.79736549e-01 -2.28498235e-01
-7.10930288e-01 1.02440572e+00 -2.22225469e-02 5.20430744e-01
-6.89723969e-01 -8.49875867e-01 6.52409077e-01 -1.80249996e-02
-9.19814184e-02 -4.17656660e-01 -8.52051899e-02 -1.36791110e+00
3.11385989e-01 -1.23566592e+00 -4.42878809e-03 -8.16819608e-01
-1.22158313e+00 8.73839676e-01 -4.37821716e-01 -1.36953592e+00
-3.37285153e-03 -2.10862681e-01 -6.78772211e-01 7.38110185e-01
-1.38447893e+00 -9.79288399e-01 -1.52834773e-01 5.60795784e-01
1.17036033e+00 -5.65702856e-01 6.42305553e-01 8.50227416e-01
-7.65905082e-01 1.06474793e+00 3.54181379e-01 3.52002293e-01
6.25158906e-01 -9.19074118e-01 8.29273343e-01 1.18777573e+00
2.52070397e-01 7.87469327e-01 4.75503504e-01 -7.65442312e-01
-1.12980676e+00 -1.34114945e+00 7.48611391e-01 1.82968415e-02
5.13127983e-01 -5.29483497e-01 -1.29516649e+00 7.00507998e-01
9.83076468e-02 -3.10737848e-01 8.23364079e-01 1.00671155e-02
-1.98042229e-01 -1.03902750e-01 -8.51905406e-01 7.30992913e-01
1.26405239e+00 -8.53198946e-01 -7.99854934e-01 1.05920225e-01
1.12637055e+00 -3.06123048e-01 -6.77701771e-01 4.59334821e-01
1.43066466e-01 -4.92864162e-01 6.07561052e-01 -5.99493682e-01
-8.26795772e-02 -3.69936168e-01 -1.65093705e-01 -1.55745554e+00
1.70752719e-01 -1.07447886e+00 1.29650176e-01 1.93500090e+00
5.12285173e-01 -5.67418814e-01 5.30539215e-01 4.80685204e-01
-8.39060545e-01 -3.17324907e-01 -1.15248656e+00 -1.14585519e+00
-3.06811392e-01 -6.34974360e-01 7.49468684e-01 1.19496524e+00
-5.78590222e-02 3.08532864e-01 -5.25133729e-01 2.62613297e-01
1.97306246e-01 -6.23269677e-01 6.54540837e-01 -1.00859416e+00
-2.44282186e-02 -3.16562504e-01 -2.20242232e-01 -1.45785022e+00
2.09315076e-01 -9.23755884e-01 3.58211458e-01 -1.22095931e+00
-2.70272911e-01 -5.50805151e-01 -1.20576911e-01 5.19883633e-01
-1.55012786e-01 -4.65912998e-01 1.01134159e-01 2.00784907e-01
-2.53836870e-01 1.28934169e+00 9.49531019e-01 -1.33045232e-02
-3.94814461e-01 2.23973200e-01 -3.93990487e-01 5.29397845e-01
7.82957733e-01 -4.03000206e-01 -3.90068889e-01 -5.57831824e-01
-1.62467644e-01 3.79822254e-01 -2.63059169e-01 -8.80697370e-01
4.83571589e-01 4.40359302e-02 -2.43626982e-01 -5.97076058e-01
3.13438207e-01 -8.67046118e-01 1.30822629e-01 2.53159795e-02
-3.61764133e-01 -1.73962578e-01 3.70863080e-01 6.40739083e-01
-7.88375616e-01 -1.89049050e-01 7.57707059e-01 1.70436963e-01
-6.31216109e-01 4.18116748e-01 -6.71388865e-01 8.58891681e-02
6.02789342e-01 -4.28258926e-01 -1.12334594e-01 -6.50600016e-01
-9.03163493e-01 1.09304175e-01 -4.14758101e-02 8.53075027e-01
8.21960151e-01 -1.43887901e+00 -8.24818254e-01 5.56738496e-01
2.01034680e-01 -1.26774758e-01 4.58165109e-01 6.17520809e-01
1.31301554e-02 3.78052682e-01 1.90613478e-01 -1.95333019e-01
-1.28920567e+00 2.75108516e-01 2.52531439e-01 1.45138845e-01
-3.12113792e-01 7.06077874e-01 2.04924345e-01 -9.33164597e-01
5.44440508e-01 -1.17987126e-01 -2.04161406e-01 -2.85590917e-01
5.03152966e-01 4.37576383e-01 3.94496351e-01 -7.49930680e-01
-1.73956603e-01 1.08197019e-01 -3.56782675e-01 -2.73631424e-01
1.07940876e+00 -4.36010748e-01 2.41049364e-01 3.24035496e-01
8.40861797e-01 1.81062073e-01 -8.24186206e-01 -4.54713434e-01
1.10555835e-01 -3.63442719e-01 1.03603445e-01 -8.96763265e-01
-1.00878334e+00 1.13621020e+00 4.67629194e-01 3.25686514e-01
1.11714721e+00 -4.34947759e-01 1.26846874e+00 4.74199951e-01
2.67945647e-01 -1.38254690e+00 -2.07707569e-01 8.07449579e-01
1.02879703e+00 -1.12374699e+00 -8.23941052e-01 -4.85364735e-01
-8.41354311e-01 1.01208436e+00 8.89678478e-01 7.12264478e-01
5.86560071e-01 3.83272976e-01 2.33695582e-01 3.76744509e-01
-6.61041439e-01 -1.62561014e-01 3.05479616e-02 7.26113677e-01
5.64106107e-01 -1.69923156e-01 -2.15238422e-01 7.88092971e-01
-2.25837514e-01 -1.81453958e-01 6.40742123e-01 7.15064883e-01
-3.35860997e-01 -1.24732208e+00 -4.42987800e-01 3.12778771e-01
-5.65189719e-01 -4.74371463e-01 -4.47757274e-01 5.47321796e-01
-2.16321498e-01 1.37314093e+00 -1.12523466e-01 -6.63038731e-01
5.99369168e-01 4.78014201e-01 -1.81772083e-01 -8.37351441e-01
-5.88384330e-01 1.79367244e-01 1.59288853e-01 -2.03321233e-01
-1.79559320e-01 -3.90961826e-01 -1.41132140e+00 -4.35211360e-02
-8.82588983e-01 3.70872945e-01 1.08837056e+00 8.17734659e-01
6.23999536e-01 6.34114444e-01 8.17080855e-01 -3.54640424e-01
-6.23890400e-01 -1.26204193e+00 -6.83738649e-01 3.46553147e-01
3.45816195e-01 -3.61047149e-01 -4.78104115e-01 9.95225236e-02]
|
[14.606691360473633, 6.682539463043213]
|
42c29c4e-6695-42ab-ae9d-5497a680f822
|
a-ccnn-adaptive-ccnn-for-density-estimation
|
1804.06958
| null |
http://arxiv.org/abs/1804.06958v2
|
http://arxiv.org/pdf/1804.06958v2.pdf
|
A-CCNN: adaptive ccnn for density estimation and crowd counting
|
Crowd counting, for estimating the number of people in a crowd using
vision-based computer techniques, has attracted much interest in the research
community. Although many attempts have been reported, real-world problems, such
as huge variation in subjects' sizes in images and serious occlusion among
people, make it still a challenging problem. In this paper, we propose an
Adaptive Counting Convolutional Neural Network (A-CCNN) and consider the scale
variation of objects in a frame adaptively so as to improve the accuracy of
counting. Our method takes advantages of contextual information to provide more
accurate and adaptive density maps and crowd counting in a scene. Extensively
experimental evaluation is conducted using different benchmark datasets for
object-counting and shows that the proposed approach is effective and
outperforms state-of-the-art approaches.
|
['Michelle Zeibots', 'Saeed Amirgholipour Kasmani', 'Xiangjian He', 'Wenjing Jia', 'Dadong Wang']
|
2018-04-19
| null | null | null | null |
['object-counting']
|
['computer-vision']
|
[-3.27848554e-01 -8.50588441e-01 3.07113200e-01 -3.65481019e-01
3.29436399e-02 -1.52328178e-01 5.13887525e-01 2.70019710e-01
-1.05700350e+00 8.74233007e-01 8.67378712e-02 2.34334879e-02
3.70826513e-01 -8.51718664e-01 -3.42107177e-01 -4.88173187e-01
1.09433001e-02 6.62049770e-01 7.76038766e-01 8.65144879e-02
4.65381950e-01 4.07315493e-01 -1.50615561e+00 -2.26699591e-01
8.49112749e-01 7.88207173e-01 2.06227243e-01 7.92063594e-01
-4.63064581e-01 1.10856020e+00 -9.93850052e-01 -5.16837239e-01
1.83667898e-01 -1.36129901e-01 -3.65595967e-01 2.68536776e-01
6.40246034e-01 -6.75932646e-01 -4.29234356e-01 1.16243064e+00
6.41540647e-01 2.23856196e-01 6.24848127e-01 -9.78238463e-01
-7.80480087e-01 7.58121386e-02 -1.17706418e+00 1.09011710e+00
1.72689781e-01 2.74199843e-01 1.74624562e-01 -8.48199248e-01
4.18870077e-02 1.44483221e+00 7.31258810e-01 4.45644200e-01
-5.31530857e-01 -7.74721146e-01 1.71617180e-01 3.73440295e-01
-1.57401800e+00 -2.28814855e-01 5.36026776e-01 -5.78560531e-01
5.85634112e-01 -1.48195833e-01 9.38843966e-01 4.16091383e-01
-9.24933981e-03 6.40658855e-01 8.74840796e-01 -5.72468221e-01
4.59499180e-01 -1.65690109e-02 7.69756064e-02 7.40255177e-01
8.77589703e-01 -5.58135271e-01 -2.05204129e-01 -1.04060493e-01
1.04133832e+00 2.81799018e-01 -2.83998232e-02 -3.52032036e-02
-1.11629021e+00 8.69198084e-01 5.66919684e-01 4.98581022e-01
-3.86213303e-01 3.07836115e-01 5.93269527e-01 -5.69907367e-01
7.30776012e-01 -1.19852930e-01 4.91062552e-02 -1.78302035e-01
-9.81411695e-01 5.87203860e-01 5.70287406e-01 8.09474528e-01
3.44676644e-01 1.20125987e-01 -5.29131651e-01 7.05658197e-01
-5.83136752e-02 7.52327859e-01 3.64111722e-01 -9.03986454e-01
6.52297854e-01 7.78949142e-01 4.87459958e-01 -1.52307153e+00
-4.72007215e-01 -8.56749862e-02 -1.09461999e+00 4.75999899e-02
7.58673728e-01 -3.84740978e-01 -6.46639228e-01 1.31380522e+00
5.51548719e-01 7.23376453e-01 -4.67161149e-01 9.67254639e-01
8.18571448e-01 5.25101602e-01 3.20853978e-01 -2.82871962e-01
1.48522210e+00 -7.52438784e-01 -6.62096858e-01 -3.77449185e-01
9.94897112e-02 -4.95255113e-01 6.93843007e-01 3.07656638e-02
-1.02460182e+00 -6.76356673e-01 -8.00618649e-01 2.79403850e-02
-3.59885901e-01 2.31321111e-01 8.53965104e-01 7.59802341e-01
-7.92892277e-01 7.09127039e-02 -7.29352236e-01 -4.60790157e-01
9.30500865e-01 4.28879917e-01 9.17694066e-03 -1.19714871e-01
-7.84537435e-01 6.03099048e-01 4.02730793e-01 1.74122199e-01
-4.47145492e-01 -3.06577384e-01 -6.92797303e-01 2.54691001e-02
2.57288814e-01 -7.33252168e-01 1.08321536e+00 -5.95175505e-01
-1.01296282e+00 8.22549820e-01 -3.54433924e-01 -4.65851516e-01
8.60894263e-01 -9.20175314e-02 -1.52801871e-01 1.38194919e-01
2.77275801e-01 5.23586154e-01 5.21285057e-01 -9.94786441e-01
-9.91915643e-01 -5.94655097e-01 8.59661028e-02 2.01658681e-01
-4.64914501e-01 3.00375640e-01 -6.16487861e-01 -3.75609368e-01
-4.10549194e-01 -6.10504448e-01 -2.47787580e-01 5.52897491e-02
-9.06624049e-02 -5.83337605e-01 8.04375708e-01 -5.57144761e-01
1.21236956e+00 -1.57182097e+00 -3.96813482e-01 -3.88118178e-01
6.23781919e-01 8.05608809e-01 2.81654149e-01 -2.46010363e-01
6.88188791e-01 -2.41798297e-01 -9.93538946e-02 -6.09286249e-01
-3.88042539e-01 9.62489694e-02 2.75426984e-01 8.11532795e-01
1.23804614e-01 8.26367736e-01 -1.01900673e+00 -1.12874699e+00
7.28061438e-01 7.47598886e-01 -3.76466095e-01 2.55627781e-01
-1.61312521e-02 7.76008725e-01 -4.89682049e-01 7.46492445e-01
1.10909021e+00 -3.21426451e-01 -2.51145631e-01 1.89544186e-01
-2.29303986e-01 -6.56821966e-01 -1.24253988e+00 1.09079099e+00
-3.93173456e-01 7.84296751e-01 -2.86473542e-01 -8.44468594e-01
7.00770617e-01 -4.00589071e-02 2.28769973e-01 -4.99398679e-01
5.59001744e-01 1.49235338e-01 -4.21899520e-02 -6.06087148e-01
8.11202109e-01 1.70709357e-01 1.00874171e-01 4.44452241e-02
-1.90725163e-01 1.88048869e-01 7.44153082e-01 -1.03760958e-02
8.43185544e-01 -3.45452726e-01 7.14819431e-01 -1.67595446e-01
8.19409430e-01 -1.86017081e-01 6.61073625e-01 8.39745760e-01
-9.16511118e-01 4.83202517e-01 2.83850916e-02 -1.06719744e+00
-1.15010083e+00 -6.33617640e-01 -2.96189822e-02 9.57655191e-01
2.75341094e-01 2.47384354e-01 -1.03526962e+00 -4.11397845e-01
-1.85587630e-02 1.65052116e-01 -7.91181624e-01 5.61626375e-01
-1.15463150e+00 -8.63955915e-01 3.94339353e-01 8.62843096e-01
1.35134041e+00 -1.03543854e+00 -9.83018816e-01 2.06036955e-01
-4.30475801e-01 -1.56153738e+00 -6.94564402e-01 -6.98031306e-01
-6.42624557e-01 -1.12202263e+00 -1.30340850e+00 -7.33543515e-01
5.84506333e-01 8.02530169e-01 1.30767941e+00 5.21845937e-01
-5.04362702e-01 1.93401873e-01 -7.09528252e-02 -9.29587781e-01
8.66101980e-02 -9.54206567e-03 4.81114313e-02 -1.44698760e-02
1.09471846e+00 -3.99806410e-01 -9.83172774e-01 1.96749464e-01
-7.06424892e-01 -4.39424068e-01 1.87138557e-01 4.85386074e-01
3.05792540e-01 8.58776346e-02 5.88668644e-01 -6.24726534e-01
6.43113256e-01 -4.29189593e-01 -1.01927876e+00 2.68484861e-01
1.34149686e-01 -4.63184237e-01 5.15031874e-01 -5.72977066e-01
-9.68720138e-01 -8.99705440e-02 2.19625026e-01 -4.17890698e-01
-2.38541335e-01 -2.87654698e-01 7.38969147e-02 -2.19694614e-01
5.41208327e-01 2.38448128e-01 -3.08652282e-01 3.32393833e-02
-3.95948887e-02 7.00345933e-01 6.05802476e-01 -3.24011296e-01
6.00711882e-01 8.29657078e-01 1.12695247e-01 -9.75990713e-01
-9.78804946e-01 -9.23506796e-01 -8.64327610e-01 -4.67090279e-01
1.15562332e+00 -1.14876831e+00 -1.19904327e+00 8.09920371e-01
-1.50102019e+00 1.55959070e-01 9.74197537e-02 4.49846804e-01
-1.28621832e-01 4.42145169e-01 -4.04574335e-01 -1.42648780e+00
-4.42553759e-01 -1.05831563e+00 1.10603678e+00 9.86851990e-01
1.86827436e-01 -1.20970690e+00 7.30700195e-02 3.47348899e-01
6.68193996e-01 2.96939671e-01 1.77997634e-01 -5.11155367e-01
-6.83942437e-01 -3.99537265e-01 -6.13417625e-01 4.32245880e-02
3.72126251e-02 -1.46221444e-02 -8.91050816e-01 -7.80941397e-02
-2.39653379e-01 -1.50767356e-01 8.86952758e-01 8.50807071e-01
1.40783846e+00 -8.92816037e-02 -4.61882949e-01 3.95777106e-01
1.38608587e+00 1.66723263e-02 6.29783392e-01 3.04078013e-01
8.16500545e-01 3.10165793e-01 2.98125029e-01 9.50768948e-01
7.29948103e-01 5.05163908e-01 3.15908492e-01 4.00977321e-02
-1.69885177e-02 1.71032082e-02 -4.29810256e-01 7.24143088e-01
-7.81771541e-01 -2.75817096e-01 -8.21133971e-01 6.64490759e-01
-1.84931254e+00 -1.21996725e+00 -1.47608861e-01 2.11791205e+00
3.99383843e-01 -6.73198104e-02 6.08735323e-01 6.08084127e-02
1.26870489e+00 2.17115194e-01 -4.24235851e-01 9.88569409e-02
-1.91572905e-02 -1.08420886e-01 5.77171385e-01 2.23829567e-01
-1.29696560e+00 8.32874477e-01 6.01014423e+00 7.23680854e-01
-8.04678440e-01 2.30948448e-01 1.05431843e+00 3.55599448e-02
5.52193880e-01 -6.30473554e-01 -1.13420081e+00 8.37848842e-01
3.29977542e-01 -4.33645993e-02 2.33066067e-01 8.26605439e-01
1.37847468e-01 -5.77213109e-01 -5.31955600e-01 1.28704333e+00
3.55239630e-01 -1.30915284e+00 -4.07307968e-02 -8.06632265e-02
8.44539583e-01 -1.79592088e-01 -9.95112285e-02 3.76939029e-01
3.13452691e-01 -8.71969044e-01 6.02451563e-01 5.77637017e-01
6.07038736e-01 -9.08254743e-01 1.34557652e+00 4.85133082e-01
-1.57170820e+00 -1.42053053e-01 -8.42287421e-01 -4.28689450e-01
3.08267415e-01 7.36937344e-01 -6.05145574e-01 -1.06740646e-01
8.37129831e-01 2.51199126e-01 -8.11330795e-01 1.51237118e+00
2.63062537e-01 3.35007966e-01 -3.73736680e-01 -6.29481196e-01
8.84950608e-02 -1.01314463e-01 2.20594347e-01 1.41129160e+00
1.30143836e-01 3.60280871e-01 4.23775554e-01 6.81791246e-01
-2.22326383e-01 8.74056220e-02 -5.16034365e-01 3.46236706e-01
6.33555233e-01 1.22952461e+00 -1.18499732e+00 -5.57298660e-01
-3.78205210e-01 7.53088474e-01 6.16815925e-01 1.50893018e-01
-1.01647115e+00 -8.33899900e-02 2.81092107e-01 2.71904588e-01
3.70466620e-01 -4.54650193e-01 -4.67399955e-02 -1.19316173e+00
1.20917775e-01 -2.42624640e-01 2.80413598e-01 -3.36261511e-01
-1.22032034e+00 4.12427634e-01 1.59126207e-01 -1.02271676e+00
4.91466261e-02 -5.10655999e-01 -9.35026884e-01 6.50394678e-01
-1.43605959e+00 -1.03942800e+00 -9.87003505e-01 5.65807343e-01
9.07896996e-01 -4.07102257e-01 3.47536922e-01 4.76613432e-01
-5.25449038e-01 4.44465876e-01 -9.64639485e-02 6.74990714e-01
1.75757736e-01 -1.10257041e+00 5.10047555e-01 8.67410839e-01
-1.34282857e-01 4.18618113e-01 5.71926713e-01 -7.18321383e-01
-8.33510101e-01 -1.13645613e+00 7.69698083e-01 -5.32141924e-01
3.23885351e-01 -1.80503070e-01 -7.22431242e-01 3.68161201e-01
1.34015843e-01 7.15286076e-01 3.61563265e-01 -2.89110333e-01
-2.93471366e-02 1.38916709e-02 -1.36490321e+00 2.71153331e-01
8.48401666e-01 2.54218038e-02 -1.61998883e-01 3.83912504e-01
4.40501720e-01 -4.87577736e-01 -2.15388060e-01 2.87951641e-02
4.36176747e-01 -1.27482748e+00 1.08983481e+00 -7.65247121e-02
2.54816562e-01 -3.67011011e-01 7.97149166e-03 -9.91703391e-01
-3.41818213e-01 -2.54427711e-03 -2.04535097e-01 1.18221891e+00
-4.64850694e-01 -4.73136991e-01 8.60477328e-01 4.95699018e-01
4.51539069e-01 -4.76542830e-01 -1.14617527e+00 -5.56638479e-01
1.10456392e-01 5.63201830e-02 6.98438406e-01 6.58785999e-01
-4.66190904e-01 1.14207588e-01 -4.37481642e-01 1.08466893e-01
9.07106042e-01 -4.22156066e-01 1.07701528e+00 -1.35053277e+00
1.30435258e-01 -4.00412440e-01 -8.91232610e-01 -9.34291005e-01
1.39150480e-02 2.37426143e-02 -2.65929531e-02 -1.67732525e+00
6.33897543e-01 -3.48806262e-01 1.98379129e-01 -1.66454822e-01
-7.69932747e-01 4.79645371e-01 3.23307127e-01 2.33770773e-01
-1.21505558e+00 3.78750712e-01 1.18844795e+00 -3.03040862e-01
-1.70579813e-02 1.51770204e-01 -2.66574860e-01 1.09200680e+00
8.11906993e-01 -2.94550329e-01 -4.18381765e-02 -7.10074067e-01
-1.08016104e-01 -8.57305676e-02 4.38921601e-01 -1.65098810e+00
4.79852945e-01 -1.17004104e-01 9.83022869e-01 -7.15157032e-01
3.42885137e-01 -5.70150673e-01 -4.90265518e-01 5.88214934e-01
2.19043627e-01 3.26732755e-01 9.41731110e-02 8.25557947e-01
-8.36031660e-02 -2.82470018e-01 9.58218515e-01 -5.66236973e-01
-7.06854761e-01 5.94165921e-01 -2.96182111e-02 3.40464830e-01
1.12592638e+00 -3.31903398e-01 -4.56692785e-01 -2.80731708e-01
-3.45244035e-02 2.17905521e-01 2.40000993e-01 1.21148676e-01
4.98894900e-01 -1.22421014e+00 -1.01846075e+00 -2.18755692e-01
1.54261261e-01 3.03305596e-01 3.82066429e-01 6.18966877e-01
-1.01172340e+00 3.71089607e-01 -1.29592896e-01 -8.10032964e-01
-1.23985732e+00 5.23022771e-01 4.00846034e-01 -3.12296718e-01
-3.76266420e-01 9.60148931e-01 1.28666490e-01 -1.60278827e-01
3.30915719e-01 -2.93557614e-01 -6.43022239e-01 -1.94259211e-01
1.24719942e+00 8.22310746e-01 -3.56566131e-01 -9.95367527e-01
-4.02070343e-01 7.78845906e-01 -2.46901140e-02 3.26893121e-01
1.05558658e+00 -3.31405848e-01 1.33031324e-01 4.69152838e-01
7.24131525e-01 -5.10173738e-02 -1.32573020e+00 -4.06223625e-01
-2.28253871e-01 -1.08071780e+00 -1.66135430e-01 -1.34345591e-01
-1.20984244e+00 1.00346255e+00 8.69438052e-01 2.84179837e-01
7.38417387e-01 -1.40699491e-01 8.31499815e-01 3.45483720e-01
7.34730065e-01 -1.10708892e+00 2.99335748e-01 4.95834947e-01
4.10320014e-01 -1.89156234e+00 2.97090352e-01 -2.86287904e-01
-5.39436996e-01 1.03582704e+00 9.27376509e-01 -3.54541719e-01
4.83631700e-01 2.49554724e-01 -4.03571278e-02 -8.10474232e-02
-4.41245437e-02 -3.16998303e-01 -4.29895259e-02 9.48820174e-01
3.62921149e-01 2.28621170e-01 -2.29110822e-01 9.45207179e-02
6.67188242e-02 4.56570119e-01 5.03150284e-01 7.65833080e-01
-8.49114954e-01 -3.48088384e-01 -8.77846301e-01 5.58587849e-01
-6.48518026e-01 1.13182507e-01 6.94984794e-02 8.34727168e-01
5.65122843e-01 1.05690205e+00 5.49482226e-01 1.92130476e-01
1.63558975e-01 -4.94994730e-01 6.17608547e-01 -3.27560574e-01
-3.99024904e-01 -3.56389880e-01 -3.55716497e-01 -5.16559510e-03
-9.59602714e-01 -5.90682566e-01 -1.08564210e+00 -8.58701289e-01
-5.25608778e-01 -1.74678773e-01 4.10093844e-01 1.03435075e+00
-1.02248006e-01 4.30860281e-01 3.08016002e-01 -1.25128818e+00
-2.03393802e-01 -1.25666893e+00 -5.49389005e-01 4.48431760e-01
1.45986617e-01 -1.02888227e+00 -2.17502400e-01 3.03349970e-03]
|
[8.419402122497559, -0.3017589747905731]
|
a19f5585-3943-4c77-a4b4-d680e05559b1
|
learning-to-segment-moving-objects
|
1712.01127
| null |
http://arxiv.org/abs/1712.01127v1
|
http://arxiv.org/pdf/1712.01127v1.pdf
|
Learning to Segment Moving Objects
|
We study the problem of segmenting moving objects in unconstrained videos.
Given a video, the task is to segment all the objects that exhibit independent
motion in at least one frame. We formulate this as a learning problem and
design our framework with three cues: (i) independent object motion between a
pair of frames, which complements object recognition, (ii) object appearance,
which helps to correct errors in motion estimation, and (iii) temporal
consistency, which imposes additional constraints on the segmentation. The
framework is a two-stream neural network with an explicit memory module. The
two streams encode appearance and motion cues in a video sequence respectively,
while the memory module captures the evolution of objects over time, exploiting
the temporal consistency. The motion stream is a convolutional neural network
trained on synthetic videos to segment independently moving objects in the
optical flow field. The module to build a 'visual memory' in video, i.e., a
joint representation of all the video frames, is realized with a convolutional
recurrent unit learned from a small number of training video sequences.
For every pixel in a frame of a test video, our approach assigns an object or
background label based on the learned spatio-temporal features as well as the
'visual memory' specific to the video. We evaluate our method extensively on
three benchmarks, DAVIS, Freiburg-Berkeley motion segmentation dataset and
SegTrack. In addition, we provide an extensive ablation study to investigate
both the choice of the training data and the influence of each component in the
proposed framework.
|
['Karteek Alahari', 'Pavel Tokmakov', 'Cordelia Schmid']
|
2017-12-01
| null | null | null | null |
['unsupervised-video-object-segmentation']
|
['computer-vision']
|
[ 2.88224816e-01 -2.56092042e-01 -3.24484617e-01 -2.29772836e-01
-3.05878490e-01 -4.73901629e-01 3.80075246e-01 -3.09817970e-01
-5.48122942e-01 4.03401345e-01 -1.26289099e-01 5.44582270e-02
2.83439010e-01 -5.47604322e-01 -1.13373661e+00 -8.87813628e-01
-1.85942709e-01 6.74695075e-02 7.07775116e-01 3.82021874e-01
2.14683384e-01 3.42104733e-01 -1.51654994e+00 3.24100316e-01
4.45080131e-01 1.28057837e+00 5.75150847e-01 9.52119052e-01
3.46445814e-02 1.41926610e+00 -3.03842068e-01 1.12551533e-01
2.83525437e-01 -5.48502803e-01 -1.13374376e+00 6.36970341e-01
4.59032625e-01 -4.74893808e-01 -5.99184155e-01 9.47061598e-01
-5.67385033e-02 5.39551795e-01 4.85361427e-01 -1.20655489e+00
-3.43529522e-01 3.46599549e-01 -5.45715094e-01 5.82508266e-01
8.25416297e-02 4.62423384e-01 7.34846234e-01 -5.50348043e-01
8.94934297e-01 9.68704283e-01 2.17848822e-01 5.66368580e-01
-9.78540778e-01 -1.74303457e-01 4.76299614e-01 4.96529549e-01
-1.09875393e+00 -4.80568200e-01 7.64170349e-01 -8.90421867e-01
6.00529909e-01 5.43062314e-02 7.99696803e-01 8.64271700e-01
1.54241458e-01 1.03250277e+00 3.34034115e-01 -3.10437959e-02
3.82791430e-01 -7.04307407e-02 3.58766526e-01 8.10784400e-01
-1.01725452e-01 4.59323823e-02 -3.76596779e-01 2.98325807e-01
1.06406224e+00 1.89814508e-01 -5.24249315e-01 -4.22029555e-01
-1.11782742e+00 4.69077826e-01 4.26716208e-01 2.63592780e-01
-4.10494000e-01 4.39448893e-01 2.60496289e-01 8.30257684e-02
1.72317326e-01 -1.87942043e-01 -4.00654465e-01 1.12740785e-01
-1.04913342e+00 -1.56387631e-02 6.50373459e-01 7.55185723e-01
9.78066802e-01 1.60579160e-01 -2.46726915e-01 4.79158193e-01
4.47495818e-01 2.92394161e-01 5.76317787e-01 -1.34408677e+00
4.36257750e-01 3.94035071e-01 2.53794491e-01 -9.58040059e-01
-1.34591609e-01 -1.52669415e-01 -7.60515451e-01 1.53316662e-01
4.87484783e-01 -1.48218632e-01 -1.09412289e+00 1.94316602e+00
3.23068649e-01 7.78639853e-01 4.06897515e-02 1.20767522e+00
7.17133224e-01 9.28742290e-01 6.64747879e-02 -4.06196862e-01
1.04012299e+00 -1.35447550e+00 -4.55089509e-01 -4.03744310e-01
4.70610112e-01 -6.29246473e-01 4.49442536e-01 4.78121340e-02
-1.34047341e+00 -9.20662880e-01 -8.71928513e-01 5.17122969e-02
2.39117793e-03 1.54116914e-01 1.94280133e-01 1.35847777e-01
-8.56894135e-01 7.15336978e-01 -1.08044994e+00 -9.37181562e-02
3.21474642e-01 2.89890766e-01 -3.51493269e-01 -6.17156774e-02
-9.46149051e-01 4.60048020e-01 4.74291801e-01 3.71232539e-01
-1.35631382e+00 -4.31241989e-01 -1.00204420e+00 7.03854207e-03
2.31749132e-01 -8.59328389e-01 1.00251436e+00 -1.83666825e+00
-1.26230252e+00 8.05186331e-01 -4.35769677e-01 -5.38706779e-01
6.33019090e-01 -1.95121691e-01 -1.06936134e-01 5.18069625e-01
1.80412650e-01 9.07044053e-01 1.18513751e+00 -1.19414604e+00
-9.15891945e-01 -1.48692071e-01 3.12302820e-02 9.85050276e-02
7.29351118e-02 -9.34693143e-02 -1.07121933e+00 -6.54385924e-01
-7.88387358e-02 -1.00606787e+00 -2.51662970e-01 -4.15714234e-02
-3.64901990e-01 1.85138866e-01 1.02732718e+00 -9.56075847e-01
1.10752356e+00 -2.32736731e+00 6.19607151e-01 -5.13906367e-02
-1.05668448e-01 2.55455136e-01 -2.75297850e-01 -2.96029717e-01
-1.08157746e-01 -7.54083470e-02 -4.17545587e-01 -4.60285097e-01
-5.65071881e-01 3.24656636e-01 -1.61701873e-01 5.82284629e-01
4.51561242e-01 8.66546333e-01 -8.97287905e-01 -5.33340931e-01
3.56582195e-01 3.94692242e-01 -5.28621197e-01 6.25620663e-01
-3.93619835e-01 7.91396260e-01 -3.84713024e-01 4.11092818e-01
4.35332358e-01 -4.41168994e-01 4.57350425e-02 -2.12967575e-01
-1.50391892e-01 -3.10563091e-02 -1.43794489e+00 1.74848759e+00
-8.20029080e-02 7.02792823e-01 1.24163497e-02 -1.14130461e+00
4.90092099e-01 2.95147181e-01 6.64669931e-01 -3.61631840e-01
1.62476093e-01 -1.83597673e-02 -6.18798733e-02 -9.56858635e-01
2.55171180e-01 2.76701868e-01 3.36991698e-01 3.54734659e-01
2.50504911e-01 4.18285519e-01 5.23639441e-01 4.09381613e-02
1.01653671e+00 2.90672332e-01 -2.13475287e-01 1.31817150e-03
7.27909982e-01 -2.23033667e-01 8.28985691e-01 6.16584122e-01
-3.36491466e-01 8.35982382e-01 3.53915989e-01 -6.66475892e-01
-8.50893140e-01 -8.48490179e-01 1.78893983e-01 8.79125834e-01
5.85592866e-01 6.78859875e-02 -6.83868825e-01 -6.37932301e-01
-3.04908037e-01 2.00100958e-01 -8.16923797e-01 -1.55206084e-01
-9.28186417e-01 -5.72915077e-01 -1.66809056e-02 6.89941823e-01
6.86253786e-01 -1.24085498e+00 -1.04563081e+00 1.42744005e-01
-4.52346057e-01 -1.28963351e+00 -8.35534394e-01 4.80093062e-02
-8.56440663e-01 -1.22738111e+00 -7.53675580e-01 -1.07364428e+00
6.52875245e-01 4.57089216e-01 9.23440754e-01 9.72476751e-02
-2.05654711e-01 3.81432623e-01 -1.29831478e-01 4.22084242e-01
-2.86895394e-01 -9.37030539e-02 -2.63345540e-01 6.72285736e-01
-2.09690183e-01 -3.13045949e-01 -8.93081129e-01 4.59368020e-01
-1.13989723e+00 1.94472089e-01 4.66272384e-01 5.90046585e-01
6.58003509e-01 -1.90339148e-01 7.54416361e-02 -5.76249003e-01
-3.33032578e-01 -6.44847691e-01 -4.96953070e-01 2.72047937e-01
2.50305265e-01 1.11449640e-02 3.56972873e-01 -4.98804927e-01
-8.61442566e-01 5.96623778e-01 1.59124136e-01 -9.15582359e-01
-1.50735155e-01 2.91778713e-01 -2.70139545e-01 7.07483739e-02
1.59075290e-01 3.66559058e-01 -1.07555062e-01 -2.00260103e-01
3.04309189e-01 2.55487949e-01 9.43270922e-01 -4.08799052e-01
4.21335846e-01 6.33007586e-01 -1.12206057e-01 -9.08071101e-01
-6.75377250e-01 -7.23129928e-01 -8.91353369e-01 -5.15363336e-01
1.27912462e+00 -9.53346908e-01 -4.35388893e-01 6.30110681e-01
-1.33114326e+00 -6.65120125e-01 -2.37429887e-01 6.04793310e-01
-7.23131359e-01 3.68271410e-01 -7.55142689e-01 -5.77794135e-01
-6.09512627e-03 -1.38808525e+00 9.49574471e-01 5.28435528e-01
-8.43113437e-02 -1.08536887e+00 8.97834599e-02 3.34054172e-01
-8.70369524e-02 3.82859826e-01 6.42918527e-01 -3.32830787e-01
-1.20853639e+00 3.66752921e-03 -1.78799465e-01 6.43212557e-01
1.33679286e-01 3.65060270e-01 -8.87638092e-01 -2.83095896e-01
1.93396613e-01 -1.26839012e-01 1.21636236e+00 8.08904767e-01
1.03020573e+00 -2.42591336e-01 -3.89404446e-01 7.91979611e-01
1.36195505e+00 5.28621852e-01 6.48107469e-01 1.78532809e-01
9.30597365e-01 5.19652665e-01 4.75591362e-01 1.69738322e-01
1.34494022e-01 6.39917254e-01 4.74777788e-01 -6.60192147e-02
-1.83936462e-01 8.22843015e-02 5.76526582e-01 7.30520070e-01
-2.52389759e-01 -2.68415302e-01 -7.19667971e-01 5.43251812e-01
-2.15798473e+00 -1.13837028e+00 3.07497587e-02 2.30166125e+00
4.35851157e-01 1.15981832e-01 1.62127435e-01 -8.40385333e-02
9.64830577e-01 3.32773089e-01 -6.54983997e-01 9.14017856e-02
-1.77349582e-01 -1.59350097e-01 2.07528844e-01 6.55879378e-01
-1.40708160e+00 8.32126319e-01 5.22181129e+00 2.53745586e-01
-1.48584831e+00 -2.42521800e-02 9.06441033e-01 -9.89561826e-02
9.60004926e-02 -8.44119303e-03 -6.66491091e-01 7.61187494e-01
6.41455293e-01 2.12517485e-01 2.63945788e-01 6.47928774e-01
2.98436314e-01 -1.73480615e-01 -1.28597844e+00 7.31506288e-01
1.50788233e-01 -1.51130223e+00 1.43933028e-01 -3.04479986e-01
9.24412489e-01 1.47894248e-01 7.23899007e-02 -5.02527133e-02
-1.90233365e-01 -8.21473897e-01 1.12039030e+00 7.95461297e-01
3.01607907e-01 -4.75264341e-01 5.09083867e-01 2.90542364e-01
-1.31307971e+00 -1.79346696e-01 -1.39358878e-01 1.32962972e-01
2.98611164e-01 2.60125101e-01 -1.35735184e-01 3.80493581e-01
6.29977822e-01 1.18937898e+00 -4.48849916e-01 1.23118782e+00
-1.45961717e-01 5.31949461e-01 -1.33887991e-01 5.11363566e-01
4.91966277e-01 -3.16278368e-01 5.90415418e-01 1.22842824e+00
1.09032519e-01 1.54751036e-02 3.51504058e-01 8.33137333e-01
1.19614054e-03 -2.18469799e-01 -2.68638343e-01 1.85729206e-01
1.57266125e-01 1.14867818e+00 -1.01744854e+00 -5.20270705e-01
-5.06421208e-01 1.14052320e+00 2.80687630e-01 8.00213635e-01
-8.96081328e-01 6.89968690e-02 6.70687616e-01 -1.40831142e-03
7.23336875e-01 -2.23410651e-01 1.21487521e-01 -1.30126870e+00
1.36590719e-01 -4.42082912e-01 3.52351099e-01 -6.93914056e-01
-7.79144347e-01 6.01684630e-01 -3.18686903e-01 -1.20241952e+00
-3.82647663e-01 -4.49539632e-01 -8.84170353e-01 6.62559986e-01
-1.43239808e+00 -8.15156162e-01 -4.09215182e-01 6.22276366e-01
8.16115379e-01 7.20148236e-02 1.93198979e-01 4.47960794e-01
-1.14182258e+00 -1.07957423e-03 -2.19289400e-02 5.56313813e-01
3.89503360e-01 -8.65053892e-01 2.67622828e-01 1.13121331e+00
2.46156201e-01 2.46191904e-01 4.12828416e-01 -6.77411795e-01
-1.19192755e+00 -1.48466182e+00 4.59320933e-01 -3.01009178e-01
4.10408944e-01 -4.79456782e-02 -1.15009832e+00 9.40972447e-01
2.63462346e-02 5.03594637e-01 2.02406526e-01 -7.37008989e-01
2.63772383e-02 1.23068102e-01 -5.71593106e-01 2.68567443e-01
8.74806225e-01 -3.16189468e-01 -4.40979213e-01 9.56375599e-02
6.40347719e-01 -4.55311090e-01 -5.50555348e-01 4.79062170e-01
5.16101301e-01 -9.74070311e-01 9.96434152e-01 -8.23624790e-01
7.41773963e-01 -6.58771753e-01 -1.26943782e-01 -8.86066318e-01
-2.94017434e-01 -4.62997854e-01 -3.11181366e-01 1.11266625e+00
9.64483246e-02 3.05571351e-02 8.97478044e-01 7.61149704e-01
-3.83497439e-02 -7.01706767e-01 -8.16938400e-01 -5.00105202e-01
-2.57942587e-01 -3.65972489e-01 5.20952903e-02 7.64554918e-01
-6.17712379e-01 3.17861378e-01 -4.38497424e-01 2.91874349e-01
4.06861216e-01 2.90432036e-01 6.63250923e-01 -6.56387508e-01
-5.16588151e-01 -3.79703999e-01 -6.09944522e-01 -1.43277121e+00
4.71990615e-01 -7.35427141e-01 2.54250079e-01 -1.27740586e+00
3.21725309e-01 -1.66294739e-01 -2.91820377e-01 1.73136979e-01
-3.72579873e-01 1.45680174e-01 3.59293878e-01 4.44818288e-01
-8.42974722e-01 3.64116699e-01 1.16696298e+00 -4.27065998e-01
-3.77564073e-01 2.07441881e-01 7.44409934e-02 9.01011646e-01
4.77965415e-01 -2.82106847e-01 -4.33688432e-01 -6.72280312e-01
-3.08344126e-01 4.75594640e-01 6.34835184e-01 -9.66066182e-01
3.62205505e-01 -1.78855285e-01 6.25733256e-01 -3.41701478e-01
2.27114335e-01 -7.08765566e-01 3.02185416e-01 7.09079206e-01
-3.65378052e-01 -2.44827978e-02 4.92773466e-02 7.16069818e-01
-4.24658239e-01 -3.55077088e-01 1.01672232e+00 -2.22336218e-01
-1.15392256e+00 5.09015203e-01 -4.60039109e-01 1.39203385e-01
1.07339966e+00 -1.61894500e-01 -5.15531860e-02 -1.58999652e-01
-9.82169867e-01 3.54698420e-01 4.79972094e-01 4.72635508e-01
5.82584739e-01 -1.14958727e+00 -4.52401757e-01 3.67044419e-01
-2.62662143e-01 2.23697200e-01 4.38744068e-01 8.24370623e-01
-6.33284271e-01 1.40077084e-01 -3.94256741e-01 -9.44041431e-01
-1.00633442e+00 7.45236814e-01 5.85740030e-01 6.37513250e-02
-4.95623887e-01 9.06737089e-01 5.80393612e-01 3.74643296e-01
4.32821989e-01 -5.15298367e-01 -2.67952830e-01 6.66658282e-02
5.87353587e-01 3.90637189e-01 -2.93222874e-01 -1.03439820e+00
-1.66006669e-01 7.25548387e-01 8.80478173e-02 -5.73965907e-02
8.88165474e-01 -2.92876691e-01 -9.63601992e-02 7.28593469e-01
1.50491643e+00 -5.46317220e-01 -1.78748786e+00 -1.40349969e-01
-9.54386313e-03 -3.96084487e-01 -1.45428151e-01 -2.72820503e-01
-1.56470311e+00 8.09341788e-01 5.28477192e-01 9.71035473e-03
1.10051906e+00 -1.56896144e-01 8.54075015e-01 3.07932179e-02
2.72832625e-02 -8.86002660e-01 2.99781978e-01 4.40383852e-01
5.10275841e-01 -1.24135149e+00 -4.47312653e-01 -2.62932867e-01
-6.20221436e-01 1.17078674e+00 6.05609715e-01 -2.01349512e-01
6.37641430e-01 -8.82145576e-03 3.31545956e-02 1.73646703e-01
-8.31151962e-01 -2.49623820e-01 4.29770738e-01 2.31859699e-01
1.92077667e-01 -2.89706528e-01 1.73089609e-01 3.74788791e-01
4.17227417e-01 2.24602118e-01 3.78536761e-01 8.18632245e-01
-4.43947256e-01 -6.69801891e-01 -1.20709322e-01 1.06499627e-01
-3.34004939e-01 3.77660930e-01 -1.67868078e-01 5.84865749e-01
3.63850504e-01 6.91675901e-01 4.26562667e-01 -2.54336506e-01
4.93916944e-02 -2.72922460e-02 3.72375429e-01 -4.40964609e-01
-3.11860055e-01 1.95836753e-01 -2.94353902e-01 -7.65452385e-01
-8.28526199e-01 -8.02060664e-01 -1.33099186e+00 7.99033046e-02
4.25575254e-03 9.26736742e-02 2.68444568e-01 1.04254901e+00
1.23369440e-01 6.35872364e-01 7.56589055e-01 -1.16520262e+00
5.73500879e-02 -5.74069202e-01 -5.19316316e-01 7.00226426e-01
6.67593777e-01 -5.15295982e-01 -3.44484180e-01 7.84515858e-01]
|
[9.068084716796875, -0.1794736385345459]
|
f495199e-d9b3-43ed-82a9-a90bc9e62359
|
estimating-the-value-of-evidence-based
|
2306.13681
| null |
https://arxiv.org/abs/2306.13681v1
|
https://arxiv.org/pdf/2306.13681v1.pdf
|
Estimating the Value of Evidence-Based Decision Making
|
Business/policy decisions are often based on evidence from randomized experiments and observational studies. In this article we propose an empirical framework to estimate the value of evidence-based decision making (EBDM) and the return on the investment in statistical precision.
|
['Serguei Stepaniants', 'James McQueen', 'Siwei Jia', 'Guido Imbens', 'Anish Agarwal', 'Alberto Abadie']
|
2023-06-21
| null | null | null | null |
['decision-making']
|
['reasoning']
|
[-2.96260267e-01 2.94350475e-01 -1.21273005e+00 -3.33364367e-01
-4.16312665e-01 -1.39979139e-01 8.13549519e-01 4.77374643e-01
-7.91747212e-01 9.80649412e-01 3.12788248e-01 -1.40954196e+00
-4.87907708e-01 -6.40858293e-01 -7.16048896e-01 -1.77869052e-01
1.60641819e-01 1.39671028e-01 -1.11407273e-01 4.36675727e-01
9.00327444e-01 4.20301825e-01 -8.31296861e-01 -1.15522325e-01
6.88665152e-01 1.13239443e+00 -1.37912154e-01 3.13571632e-01
-3.57175767e-02 8.54615092e-01 -2.46585667e-01 -6.86575770e-01
4.01474714e-01 -2.31446356e-01 -1.61599591e-01 -3.76119226e-01
-3.23520273e-01 -6.23138487e-01 -5.94057739e-02 7.50465453e-01
4.70890701e-01 2.71543283e-02 7.19190955e-01 -1.24942744e+00
-6.36199117e-01 8.76292050e-01 -4.70963359e-01 6.60634220e-01
-1.14656381e-01 5.64210355e-01 7.82913685e-01 -6.09155416e-01
7.51214206e-01 1.29951084e+00 2.86120206e-01 -1.48663279e-02
-9.80374455e-01 -7.45182872e-01 2.24928468e-01 1.32889077e-01
-8.58499527e-01 -6.37514055e-01 4.56517220e-01 -6.59421384e-01
7.27656603e-01 -1.57485723e-01 7.12066412e-01 8.96917343e-01
7.71889806e-01 -7.20025226e-02 1.64926422e+00 -9.04050589e-01
7.02827394e-01 3.32817763e-01 1.46265209e-01 3.25742871e-01
1.63530612e+00 1.05541658e+00 -4.47390646e-01 -1.80753857e-01
8.96612763e-01 1.17978364e-01 2.17458501e-01 5.58615588e-02
-9.07274425e-01 1.00731051e+00 1.13805100e-01 9.54624712e-02
-9.41635370e-01 4.49516803e-01 2.33761713e-01 3.19604039e-01
2.75367528e-01 4.83942628e-01 -5.35399377e-01 -6.63146302e-02
-6.48176491e-01 3.12466323e-01 5.86098492e-01 2.91751564e-01
1.78205565e-01 -1.73671305e-01 -2.80384213e-01 1.51881784e-01
6.89745724e-01 1.04450977e+00 2.12421507e-01 -1.22529721e+00
3.47950965e-01 2.33287930e-01 8.05889606e-01 -7.51197517e-01
-2.31821582e-01 9.09292996e-02 -2.67853886e-01 3.00759614e-01
1.99553758e-01 -5.52914858e-01 -6.83709919e-01 1.05410528e+00
-2.18630526e-02 -1.02079980e-01 -1.30844563e-01 6.27861857e-01
4.71750945e-01 4.57449794e-01 4.12762493e-01 -3.92231435e-01
9.06745911e-01 -3.80932003e-01 -1.03073275e+00 -3.23565267e-02
5.15447795e-01 -2.27736205e-01 6.91214323e-01 4.08535242e-01
-1.17352426e+00 -7.38366544e-02 -8.35708976e-01 4.10311520e-01
-8.95975381e-02 -1.74264550e-01 7.59303451e-01 5.34841895e-01
-5.67103922e-01 4.21423674e-01 -7.46344090e-01 5.92465177e-02
4.54295248e-01 -1.31544635e-01 2.54473090e-01 -3.74405161e-02
-7.26667941e-01 1.40191889e+00 3.00568044e-01 2.45529786e-01
-7.39341795e-01 -5.68629265e-01 -5.70133984e-01 -6.87536448e-02
6.33848548e-01 -7.96716750e-01 1.31312621e+00 -4.69002992e-01
-1.29642749e+00 5.84910929e-01 2.35935569e-01 -9.40554976e-01
4.16972727e-01 1.43449664e-01 -3.05794597e-01 6.99594170e-02
2.03851879e-01 8.92233104e-02 2.81064123e-01 -9.54466999e-01
-9.14308250e-01 -6.59455121e-01 1.80086996e-02 -4.50482458e-01
2.97108322e-01 7.79023886e-01 6.42714858e-01 -6.52427733e-01
-1.72821656e-01 -8.12118828e-01 -7.79772103e-01 -6.67703271e-01
-9.37866941e-02 -2.31469542e-01 -1.35737956e-01 -6.65580928e-01
1.17792058e+00 -1.73793364e+00 -6.28077328e-01 2.49657854e-01
1.48027211e-01 -7.38972947e-02 3.77280653e-01 3.81649703e-01
3.27508509e-01 7.19654799e-01 3.08963150e-01 5.06698787e-01
1.76265910e-01 2.43395150e-01 -5.02400041e-01 3.35270524e-01
1.48831323e-01 1.08276975e+00 -5.21837711e-01 -3.69328618e-01
4.05287623e-01 -2.96622843e-01 -4.41658109e-01 5.22398464e-02
6.77096844e-02 2.14124873e-01 -7.65003026e-01 7.75276423e-01
4.80864644e-01 -2.04121813e-01 6.67753816e-01 2.87793517e-01
-6.04167402e-01 7.91833937e-01 -5.18132865e-01 7.25067854e-01
-3.73724073e-01 5.67403853e-01 1.85136572e-02 -1.28275120e+00
5.89765310e-01 2.60318369e-01 -9.86825526e-02 -1.13156772e+00
3.47579807e-01 1.81977674e-01 5.13051391e-01 -8.02740276e-01
-1.10335939e-01 -6.22491598e-01 -4.97779027e-02 5.43995798e-01
-3.56029421e-01 -2.29051381e-01 -1.33188851e-02 -2.79790819e-01
7.46578753e-01 -2.49731794e-01 6.69295192e-01 -6.27309024e-01
-1.83641523e-01 9.85013205e-04 6.84216142e-01 9.86376166e-01
-1.85754389e-01 -6.00968480e-01 6.90239847e-01 -5.16263366e-01
-1.04381824e+00 -5.69309413e-01 -3.76085877e-01 6.48599982e-01
-3.83947909e-01 1.37412846e-01 -3.31528157e-01 -4.54595476e-01
6.95080817e-01 1.03500783e+00 -7.21849620e-01 -1.12454660e-01
-1.68536618e-01 -8.05010617e-01 4.03668396e-02 7.46347189e-01
3.42056185e-01 -9.63400662e-01 -1.25026202e+00 7.00406432e-02
5.93517065e-01 -1.03360653e+00 3.61706197e-01 -6.47983700e-02
-1.14940310e+00 -1.20432329e+00 -4.15492088e-01 3.89780462e-01
4.07710344e-01 1.07263975e-01 8.29821944e-01 -1.46001652e-02
3.34820002e-02 1.34757549e-01 -3.29052210e-01 -1.32216120e+00
-3.80927473e-01 -5.54251790e-01 2.22101018e-01 -4.12590832e-01
5.36699653e-01 -8.07748213e-02 -7.11884677e-01 3.13455075e-01
-4.62732613e-01 -5.10188878e-01 6.81169987e-01 6.80150509e-01
3.34009081e-01 4.96855862e-02 8.72971356e-01 -7.25021839e-01
9.83996451e-01 -5.14018178e-01 -1.16890144e+00 3.05497229e-01
-1.38856626e+00 5.10391220e-02 -2.27747217e-01 -2.47113928e-01
-1.33723843e+00 -1.23485613e+00 4.38334107e-01 6.92143571e-03
-1.43209221e-02 1.20375788e+00 3.24111223e-01 2.31915098e-02
6.35944128e-01 -7.63450265e-01 4.97649796e-02 -4.85834718e-01
1.66664511e-01 6.53533340e-01 -5.51711880e-02 -7.85808504e-01
8.43286216e-02 3.44819099e-01 1.59259424e-01 -2.83866495e-01
-8.15450728e-01 1.36398122e-01 1.38603687e-01 -6.62490055e-02
4.90758717e-01 -6.64219260e-01 -7.86816657e-01 -4.43896651e-01
-6.59958661e-01 -6.13875329e-01 -3.35808486e-01 1.45343304e+00
-3.93960565e-01 -2.52823800e-01 -1.45508349e-01 -1.28337896e+00
9.65151414e-02 -1.05052805e+00 2.78308511e-01 2.30287671e-01
-3.80279958e-01 -7.40090013e-01 9.40924063e-02 3.10615480e-01
4.34462249e-01 4.41080093e-01 6.78579330e-01 -5.96458435e-01
-5.83473444e-01 -3.44587445e-01 -3.00787151e-01 1.36461228e-01
-1.37021497e-01 6.27777874e-01 -2.95529008e-01 1.56153247e-01
3.75685036e-01 -1.29647940e-01 6.94873273e-01 1.60598171e+00
1.01989734e+00 -7.27596879e-01 -2.82384425e-01 1.13403864e-01
1.39433634e+00 7.09860742e-01 3.54583383e-01 6.92105711e-01
-8.22196603e-02 1.18409610e+00 1.05743587e+00 7.87104487e-01
1.83724523e-01 3.26342314e-01 -7.86472335e-02 3.22420359e-01
4.43216294e-01 -3.96078348e-01 -9.63115096e-02 1.62480146e-01
-6.31826103e-01 5.89074148e-03 -1.24225891e+00 3.19824934e-01
-1.89406371e+00 -8.59216630e-01 -6.62264088e-03 2.31851435e+00
5.72247565e-01 6.77204311e-01 5.29769540e-01 -1.85719416e-01
6.67333841e-01 -3.26134026e-01 -2.58808255e-01 -1.07861233e+00
-4.94456757e-03 -3.88990268e-02 1.03480434e+00 1.47431388e-01
-2.34929025e-01 3.66803139e-01 9.05554104e+00 4.46219087e-01
-9.20767367e-01 6.53183684e-02 1.20049477e+00 -1.64388120e-01
-4.54040289e-01 1.93121836e-01 -7.40192592e-01 5.50674617e-01
1.78032315e+00 -7.52185643e-01 -1.07276104e-01 6.90070331e-01
8.61358583e-01 -8.02070677e-01 -7.34923840e-01 2.92364240e-01
-6.79528475e-01 -1.57364666e+00 -1.52100727e-01 5.77906489e-01
7.77839959e-01 -2.28778914e-01 2.18052849e-01 7.34320655e-03
1.01291668e+00 -9.96033072e-01 9.74789202e-01 5.84488153e-01
7.56643534e-01 -5.46036959e-01 1.23068929e+00 3.07218850e-01
9.03412178e-02 -5.69187880e-01 -4.91536260e-01 -7.28397429e-01
-9.10527632e-02 9.71827745e-01 -9.07872915e-01 6.32028505e-02
8.04690182e-01 1.17137618e-02 -3.58972847e-02 7.68720865e-01
-5.79721391e-01 1.20385897e+00 2.53119301e-02 -3.08876872e-01
4.48762119e-01 -4.56641048e-01 -9.10806954e-02 1.09367299e+00
-1.18714282e-02 6.34651124e-01 -3.46559525e-01 8.00497830e-01
1.27462894e-01 -1.97102614e-02 -8.95056844e-01 -7.32840657e-01
6.90712988e-01 4.68562692e-01 -8.84336233e-01 -3.97901386e-01
-4.97770607e-01 -3.70529532e-01 -1.88887656e-01 2.58490145e-01
-4.49464083e-01 1.50098249e-01 1.11417115e-01 1.53083652e-01
3.65467459e-01 1.13112524e-01 -8.24952841e-01 -8.17418694e-01
-2.24073958e-02 -8.95034611e-01 4.48881567e-01 -4.80916440e-01
-9.67352986e-01 -3.98547858e-01 4.68349308e-01 -3.28978837e-01
-1.18763961e-01 -6.16512239e-01 -4.47280049e-01 8.51041555e-01
-1.33162582e+00 -3.56725007e-01 4.30551291e-01 -4.48908120e-01
-5.37353940e-02 1.39226422e-01 2.97586322e-01 -2.64039427e-01
-7.61495769e-01 1.79665416e-01 3.66598099e-01 -1.65730178e-01
4.20298129e-01 -9.13447857e-01 8.60534683e-02 6.43093109e-01
-6.39221907e-01 6.47865474e-01 8.31251442e-01 -8.61612320e-01
-1.09279454e+00 -2.03838497e-01 6.91759944e-01 -3.88949126e-01
8.97487402e-01 1.50312185e-01 -1.85849205e-01 8.13257396e-01
6.43041208e-02 -3.42884302e-01 6.72635734e-01 4.88226980e-01
-1.93398669e-01 -2.86599904e-01 -1.35552943e+00 4.45045084e-01
6.41327202e-01 -2.77976751e-01 -9.42940533e-01 -1.59614667e-01
6.78443432e-01 4.89678502e-01 -9.13640916e-01 6.78072155e-01
8.72495830e-01 -8.15426886e-01 7.76686490e-01 -9.98198271e-01
7.62836218e-01 3.49157482e-01 -2.77629018e-01 -1.00812292e+00
-3.03407431e-01 -4.83593941e-01 2.03540191e-01 6.81504905e-01
5.60474753e-01 -9.24646139e-01 3.33919108e-01 1.05756497e+00
4.54103887e-01 -5.91253638e-01 -8.67414832e-01 -9.01108027e-01
1.94799170e-01 -2.70185560e-01 7.78478861e-01 1.07938623e+00
1.49158970e-01 -3.82832587e-02 3.04745678e-02 -3.18010181e-01
6.19570553e-01 2.61465371e-01 6.82034373e-01 -1.12273073e+00
-3.11466437e-02 -7.02832580e-01 -2.71071613e-01 1.71978295e-01
-2.06948489e-01 -1.25977293e-01 -2.71833718e-01 -1.39068079e+00
3.29084486e-01 -3.46130550e-01 -5.99275410e-01 -3.51469517e-02
-1.54744610e-01 -6.91093445e-01 1.49196461e-01 -7.05264658e-02
-3.02815974e-01 2.62988746e-01 7.71817386e-01 2.41773412e-01
-4.13349532e-02 -8.18616450e-02 -1.01107633e+00 8.91492248e-01
9.76143420e-01 -5.87457418e-01 5.74428588e-02 1.30252436e-01
6.15492880e-01 7.10935593e-01 5.67415774e-01 -1.62604555e-01
-2.08839014e-01 -1.10367060e+00 -4.92555797e-02 -4.96625096e-01
-5.80163240e-01 -7.36813426e-01 3.64014328e-01 1.00138426e+00
-7.01684594e-01 2.11943746e-01 1.59476668e-01 6.42922044e-01
8.77540112e-02 -4.60682005e-01 2.91664302e-01 -1.30398154e-01
-1.43019214e-01 -2.11475432e-01 -3.62025827e-01 -6.11271188e-02
8.72672260e-01 8.90486836e-02 -4.73557591e-01 -1.47665188e-01
-3.80113065e-01 4.37948592e-02 4.75506961e-01 -1.84884354e-01
2.47764200e-01 -9.79264736e-01 -9.90178466e-01 -5.20381570e-01
3.59239825e-03 -6.75094008e-01 -1.18660592e-01 9.47661161e-01
-6.91710651e-01 1.11127162e+00 -2.93297797e-01 2.39975885e-01
-5.40673614e-01 8.46725941e-01 4.52765018e-01 -2.52573431e-01
-3.91766965e-01 5.46214461e-01 1.23171709e-01 3.04638475e-01
1.20859057e-01 -6.21692240e-01 1.30118858e-02 -3.10284328e-02
6.20835423e-01 1.02414620e+00 -2.34299123e-01 2.30401441e-01
-3.71336937e-01 -1.06979609e-02 1.48091137e-01 -8.80537391e-01
1.31194949e+00 -1.18404113e-01 -3.23016286e-01 7.66356528e-01
5.63782513e-01 8.84278268e-02 -9.55181479e-01 -2.90689524e-02
4.18454140e-01 -8.78659308e-01 6.11258984e-01 -1.10086787e+00
-5.90227604e-01 5.51509023e-01 2.57490188e-01 3.71476889e-01
6.63630307e-01 -2.84328222e-01 -1.28240570e-01 4.68860865e-01
3.08077812e-01 -1.65564632e+00 -4.35452342e-01 -7.73644686e-01
1.00690675e+00 -1.23095381e+00 4.61092442e-01 -1.40315309e-01
-5.50300539e-01 5.53514302e-01 -3.43919061e-02 -2.70314336e-01
1.26718771e+00 -1.32002458e-02 -2.75177956e-01 -3.32690507e-01
-9.73293006e-01 8.21152180e-02 1.04476206e-01 3.54497313e-01
1.88284069e-01 7.93237448e-01 -1.30124176e+00 1.02398741e+00
9.51027274e-02 7.93903649e-01 3.77196610e-01 9.64414418e-01
-5.87076962e-01 -6.61233664e-01 -6.02819204e-01 8.99812937e-01
-1.20279348e+00 -1.14037894e-01 -4.51508164e-01 8.58916223e-01
4.22443449e-02 1.35505986e+00 3.08778644e-01 3.54079306e-02
2.66345978e-01 -2.19566271e-01 2.98556805e-01 -2.30271220e-01
-1.82858616e-01 1.93397939e-01 5.64971030e-01 -3.83595765e-01
-8.15272927e-01 -9.24011827e-01 -7.95414746e-01 -6.66293502e-01
-4.66546357e-01 1.51968062e-01 1.10234761e+00 9.63114619e-01
4.64851648e-01 2.04050198e-01 6.67449474e-01 -1.62947532e-02
-9.02922511e-01 -9.00766134e-01 -8.29410553e-01 -1.43424571e-01
-1.00896927e-02 -8.43136728e-01 -8.11440825e-01 -6.10280752e-01]
|
[7.976164817810059, 5.229262351989746]
|
cd5fa524-d137-4010-bb22-e039b6108d18
|
layout-design-for-intelligent-warehouse-by
|
1811.05685
| null |
http://arxiv.org/abs/1811.05685v1
|
http://arxiv.org/pdf/1811.05685v1.pdf
|
Layout Design for Intelligent Warehouse by Evolution with Fitness Approximation
|
With the rapid growth of the express industry, intelligent warehouses that
employ autonomous robots for carrying parcels have been widely used to handle
the vast express volume. For such warehouses, the warehouse layout design plays
a key role in improving the transportation efficiency. However, this work is
still done by human experts, which is expensive and leads to suboptimal
results. In this paper, we aim to automate the warehouse layout designing
process. We propose a two-layer evolutionary algorithm to efficiently explore
the warehouse layout space, where an auxiliary objective fitness approximation
model is introduced to predict the outcome of the designed warehouse layout and
a two-layer population structure is proposed to incorporate the approximation
model into the ordinary evolution framework. Empirical experiments show that
our method can efficiently design effective warehouse layouts that outperform
both heuristic-designed and vanilla evolution-designed warehouse layouts.
|
['Wei-Nan Zhang', 'Zilong Guo', 'Yong Yu', 'Jun Wang', 'Wenxin Li', 'Han Cai', 'Haifeng Zhang', 'Chris Wang']
|
2018-11-14
| null | null | null | null |
['layout-design']
|
['computer-vision']
|
[-2.68963188e-01 -1.53788209e-01 1.53925091e-01 -2.95264900e-01
4.68332181e-03 -3.72035950e-01 3.81793967e-03 2.42654026e-01
-4.19346988e-01 6.89441502e-01 -1.22196428e-01 -3.29567820e-01
-3.90734702e-01 -1.26537931e+00 -4.06187981e-01 -8.42109740e-01
1.86077014e-01 5.00303328e-01 1.26774982e-01 -6.36456788e-01
4.26799029e-01 3.77910137e-01 -1.47291267e+00 -3.26223493e-01
1.52544582e+00 9.72489297e-01 5.97049773e-01 2.65469670e-01
-5.61358154e-01 1.43042728e-02 -7.28796661e-01 -4.65042561e-01
2.81297207e-01 -5.11444032e-01 -1.80925399e-01 2.35570714e-01
-9.75162148e-01 -9.06668380e-02 3.85261737e-02 1.02268004e+00
5.46111286e-01 6.36064187e-02 4.78575557e-01 -1.36454618e+00
-6.99537277e-01 7.85009742e-01 -7.75797188e-01 -3.33283484e-01
-6.72711655e-02 1.01008587e-01 6.95810914e-01 -3.94403607e-01
3.69612157e-01 1.07117283e+00 2.96494186e-01 6.57066852e-02
-9.95142639e-01 -4.13358629e-01 2.42808431e-01 1.17421411e-01
-1.51861465e+00 9.63460878e-02 8.10748458e-01 1.08889572e-01
5.62313676e-01 2.50660658e-01 1.25786138e+00 3.14824522e-01
6.44237936e-01 9.19090509e-01 5.21326184e-01 -3.48985136e-01
6.72563434e-01 1.16217449e-01 -2.70045459e-01 7.35561371e-01
8.89572918e-01 -1.76721707e-01 9.76315960e-02 1.93367958e-01
4.97016877e-01 3.35786268e-02 -2.11438894e-01 -7.03740537e-01
-8.33116293e-01 9.91426349e-01 4.61454868e-01 2.74137557e-01
-7.75619447e-01 -1.42612472e-01 2.02038229e-01 3.96177880e-02
5.55445254e-02 9.57967997e-01 -4.84326184e-01 -4.03449088e-01
-4.72799748e-01 2.58294255e-01 9.56893682e-01 1.13007247e+00
3.17253768e-01 -4.48128320e-02 1.07712105e-01 6.81994855e-01
3.49750817e-01 4.56086993e-01 2.60165334e-01 -7.82180011e-01
3.71376425e-01 1.06726873e+00 3.42394918e-01 -8.81041169e-01
-6.23134077e-01 -3.59685183e-01 -8.30400705e-01 -1.68719918e-01
-5.47825769e-02 -4.26512241e-01 -6.32752061e-01 1.13170302e+00
5.14063597e-01 -6.15342677e-01 1.12002030e-01 9.59638178e-01
1.57381862e-01 6.81158304e-01 -1.51624218e-01 -3.62034798e-01
1.26991248e+00 -1.15230668e+00 -1.08941531e+00 1.97733745e-01
7.59001076e-01 -5.99340498e-01 8.61086845e-01 4.18928713e-01
-1.10906076e+00 -2.66103566e-01 -1.20390856e+00 4.48402017e-01
-5.01282811e-01 3.07507217e-01 9.73804474e-01 8.59213650e-01
-6.18831456e-01 1.98780745e-01 -6.83856070e-01 -4.47643608e-01
1.83096185e-01 3.09312522e-01 5.88126034e-02 -2.30612367e-01
-1.02308381e+00 9.67528760e-01 6.24940515e-01 6.50988460e-01
-1.56017216e-02 -4.02009070e-01 -6.56317174e-01 3.67606044e-01
6.59468472e-01 -4.92281407e-01 9.44502831e-01 -1.73791811e-01
-1.68670940e+00 2.88499929e-02 7.85288773e-03 -7.27126049e-03
5.32964170e-01 1.93722099e-01 -3.89255464e-01 -2.70524144e-01
-2.54259765e-01 3.86345714e-01 2.60880411e-01 -1.46971285e+00
-1.06715167e+00 -2.54910469e-01 6.94813281e-02 2.94787496e-01
-3.82556260e-01 -5.10138154e-01 -5.67517877e-01 -3.80413175e-01
3.46574664e-01 -8.27547133e-01 -7.85349727e-01 -2.83218205e-01
-3.04658890e-01 -6.03025779e-02 6.18729115e-01 -4.99744296e-01
1.28453004e+00 -1.93880284e+00 3.30546111e-01 4.08873081e-01
-1.88930362e-01 1.60813287e-01 -7.16764554e-02 5.79494834e-01
6.49852574e-01 -6.66096136e-02 -3.31870586e-01 6.94392547e-02
4.18808073e-01 5.21019220e-01 1.80195093e-01 -1.40085835e-02
1.47367388e-01 1.15116370e+00 -7.81802893e-01 -4.27198529e-01
1.61266118e-01 6.84480742e-02 -6.72726750e-01 3.60610962e-01
-2.20415935e-01 6.01538047e-02 -7.57538021e-01 9.57830310e-01
1.02907956e+00 1.60487086e-01 3.69218379e-01 -1.64668217e-01
-5.21063924e-01 -5.91379821e-01 -1.12883055e+00 1.49283767e+00
-4.90836352e-01 -1.37124389e-01 2.63537198e-01 -9.26681340e-01
1.46493864e+00 -3.06069762e-01 4.47611302e-01 -8.29936087e-01
4.66510177e-01 3.50614756e-01 3.46022636e-01 -7.15884387e-01
8.68290842e-01 2.21411854e-01 -2.68151760e-01 4.02405143e-01
-5.72220027e-01 -2.95403510e-01 4.84004945e-01 -3.60685259e-01
9.15611923e-01 4.17043626e-01 1.97007477e-01 -2.16477394e-01
6.25832677e-01 3.32496911e-01 8.44244599e-01 3.33986700e-01
-1.23469748e-01 1.47833481e-01 5.24483979e-01 -4.70129937e-01
-9.72395718e-01 -7.69107401e-01 1.88453525e-01 6.10719621e-01
7.50552356e-01 -1.06992319e-01 -7.81490505e-01 -5.03526986e-01
1.49033844e-01 1.16086721e+00 -3.60606432e-01 -6.44924998e-01
-4.49504703e-01 -9.37034547e-01 1.14540122e-01 4.41866696e-01
8.02954614e-01 -8.56809020e-01 -1.20111239e+00 5.36816239e-01
-1.75007343e-01 -6.73148870e-01 -3.85528773e-01 4.54665393e-01
-7.83352852e-01 -7.51421094e-01 -8.26612711e-01 -7.14702785e-01
1.04851747e+00 3.65322948e-01 5.33529878e-01 3.13281626e-01
-4.31865454e-01 -1.73185140e-01 -5.90246916e-01 -5.45254350e-01
-2.31160834e-01 5.51334202e-01 -1.67861402e-01 -1.83684483e-01
4.61513579e-01 -2.63416946e-01 -5.25170684e-01 5.16341925e-01
-8.42412591e-01 1.72772184e-01 1.09033644e+00 9.18951035e-01
3.54557365e-01 9.77888525e-01 6.59694135e-01 -3.31745327e-01
1.08388281e+00 -4.00223076e-01 -9.64747369e-01 6.97811842e-01
-7.81830370e-01 4.00510877e-01 8.25009525e-01 -2.58464843e-01
-1.11247587e+00 -9.80009232e-03 -1.18512630e-01 3.39674354e-02
7.04390183e-02 7.29418576e-01 -6.58302367e-01 -8.36725757e-02
-4.41280067e-01 2.24024475e-01 1.76544204e-01 -3.91757458e-01
1.08313479e-01 5.69742143e-01 1.88537106e-01 -5.93029499e-01
8.58109117e-01 -8.93309042e-02 2.52002001e-01 -5.35051346e-01
-1.54843420e-01 9.26662423e-03 -6.35890305e-01 -2.12704211e-01
7.31825113e-01 -1.60695426e-02 -1.16391551e+00 1.83913797e-01
-1.00544405e+00 5.04420809e-02 -3.18424881e-01 4.24390823e-01
-3.56678873e-01 2.17314763e-03 -7.09096566e-02 -1.23753965e+00
-1.30971432e-01 -1.47994769e+00 9.12526369e-01 5.41733861e-01
8.56707469e-02 -5.17553627e-01 -2.60997832e-01 4.15164195e-02
6.27570093e-01 2.12671876e-01 1.33635426e+00 -2.80724704e-01
-7.51479506e-01 -2.62664407e-01 -1.19186737e-01 -2.18291610e-01
2.75769651e-01 1.10251240e-01 -2.68526655e-03 -4.00039107e-02
-2.00308323e-01 9.26686749e-02 2.97935188e-01 3.68911952e-01
8.30066621e-01 -2.74748961e-03 -4.59615707e-01 5.64239323e-01
1.32124090e+00 1.09426081e+00 7.04560757e-01 7.44755328e-01
1.91978589e-01 9.10479605e-01 1.34717238e+00 7.78894603e-01
6.19917870e-01 4.34034735e-01 4.42344010e-01 5.37400842e-02
8.05195689e-01 -3.55274498e-01 -2.14787256e-02 1.05574691e+00
-1.43103257e-01 -5.88727176e-01 -6.37673616e-01 3.18459600e-01
-2.21576333e+00 -5.80478191e-01 1.12047054e-01 1.94486701e+00
3.72983575e-01 1.29267469e-01 -7.27309808e-02 1.79530561e-01
7.59240746e-01 -3.13815147e-01 -6.27074778e-01 -6.55858576e-01
-1.35783032e-01 -2.58771449e-01 6.93811893e-01 8.36144909e-02
-5.35653889e-01 8.12640369e-01 5.80007792e+00 5.29768050e-01
-6.85605884e-01 -4.38118190e-01 3.20391387e-01 5.09397425e-02
-5.13224661e-01 2.86079571e-02 -5.11468232e-01 7.04032481e-01
4.73313749e-01 -3.88655990e-01 6.56502187e-01 6.35649025e-01
4.21079814e-01 -3.73817176e-01 -4.85065222e-01 7.07031608e-01
-2.94471264e-01 -1.15539289e+00 -1.31629005e-01 4.38086480e-01
6.22878313e-01 -1.02558661e+00 -1.84025586e-01 4.00878698e-01
2.59857714e-01 -6.17102563e-01 6.98289037e-01 6.67867720e-01
1.54352546e-01 -1.36171544e+00 1.17443728e+00 3.91841650e-01
-1.40478361e+00 -5.11718154e-01 -6.75073147e-01 4.35691029e-01
8.08094501e-01 2.85125524e-01 -5.79013467e-01 1.03944135e+00
6.71342254e-01 -6.86784983e-02 -3.03515524e-01 1.44003749e+00
-1.51501238e-01 -1.14279337e-01 -3.25912476e-01 -7.85749137e-01
5.15134573e-01 -9.34862018e-01 2.67080367e-01 4.55540001e-01
7.63585150e-01 2.30146974e-01 2.40529552e-01 8.94602835e-01
2.74122059e-01 1.55655041e-01 -3.81156951e-01 -3.18696678e-01
5.80475390e-01 1.31687438e+00 -1.11664987e+00 3.19869429e-01
8.30719247e-02 9.40430045e-01 -6.85290098e-02 2.37836063e-01
-1.15777731e+00 -1.00991964e+00 4.50577050e-01 -2.50196755e-01
5.70992470e-01 -4.61519659e-01 -3.39911550e-01 -5.81182897e-01
-1.58839956e-01 -5.54882944e-01 -2.91221738e-02 -7.35803246e-01
-8.23927760e-01 3.75245392e-01 -1.90618634e-01 -7.98541486e-01
-1.36632800e-01 -5.08415699e-01 -6.54554665e-01 6.53521419e-01
-1.51394320e+00 -8.93173277e-01 -4.72251892e-01 -9.62130576e-02
5.52739084e-01 -3.47049326e-01 5.52336633e-01 1.87190950e-01
-9.44650054e-01 5.05729973e-01 4.25760716e-01 -4.39734519e-01
1.58273682e-01 -1.02143335e+00 1.61996111e-01 5.54726422e-01
-6.93364561e-01 7.00741410e-01 8.70575726e-01 -7.73133814e-01
-2.25596833e+00 -7.36567676e-01 6.08919263e-01 5.02665937e-01
4.41513687e-01 -3.38928342e-01 -6.86766624e-01 1.12924255e-01
2.94766039e-01 -6.76494837e-01 4.99968559e-01 -9.84031856e-02
6.63147151e-01 -2.59008139e-01 -1.34206939e+00 9.60222721e-01
1.13668382e+00 5.90431511e-01 -3.40419143e-01 -1.42780125e-01
8.54572296e-01 -1.78889811e-01 -9.01462495e-01 3.60048652e-01
7.64656723e-01 -5.75229645e-01 3.99931073e-01 -3.91294658e-01
5.79602160e-02 -3.62686038e-01 -6.91637769e-02 -1.65519595e+00
-4.40992743e-01 -6.92886770e-01 -1.15756534e-01 1.36132550e+00
3.09585124e-01 -1.05241215e+00 7.75037110e-01 8.37224126e-01
-1.50393516e-01 -1.15206516e+00 -5.29136419e-01 -7.55676806e-01
-4.79197353e-01 2.91358799e-01 1.42203319e+00 3.63367677e-01
-1.74609885e-01 9.01007205e-02 5.97645007e-02 1.73247293e-01
4.80235994e-01 3.39497536e-01 7.61455238e-01 -1.24836087e+00
6.39095828e-02 -6.62027121e-01 -2.49465123e-01 -8.58621538e-01
-4.17458154e-02 -4.28497136e-01 3.62057120e-01 -1.92739105e+00
-2.53137469e-01 -9.19556975e-01 -1.30445838e-01 5.08689284e-02
-7.49934837e-02 -4.82103318e-01 1.77691698e-01 -1.86944976e-01
-3.64878595e-01 1.09481192e+00 1.43767798e+00 -1.02019161e-01
-4.97616291e-01 -3.62484977e-02 -9.84802604e-01 3.75392050e-01
8.83133888e-01 -2.22913444e-01 -6.00446224e-01 -4.76006120e-01
2.77439982e-01 1.53140813e-01 -4.46091175e-01 -6.85182691e-01
4.05502766e-01 -4.13541108e-01 3.25977504e-01 -1.02225924e+00
1.69663325e-01 -1.24251926e+00 3.23989511e-01 9.70134199e-01
-7.73636028e-02 7.35885620e-01 9.49113891e-02 5.14132440e-01
-4.05142345e-02 -5.53752005e-01 2.44579032e-01 8.01056921e-02
-5.52424610e-01 1.93588167e-01 -4.87598389e-01 -4.84460473e-01
1.37916613e+00 -4.52343225e-01 -2.18926921e-01 -1.84978485e-01
-5.83736673e-02 9.72964644e-01 4.97483730e-01 6.11045837e-01
6.16619587e-01 -1.25271034e+00 -4.32795197e-01 2.76532263e-01
-1.02280162e-01 1.56003078e-02 1.78279012e-01 7.54749238e-01
-8.82886052e-01 4.64424521e-01 -3.51359755e-01 -3.10478568e-01
-7.31197000e-01 6.45122349e-01 -7.39543885e-02 -3.84350389e-01
-3.66924703e-01 5.46255708e-01 -2.17133552e-01 -5.14036119e-01
1.34476379e-01 -1.54982030e-01 -2.21161783e-01 9.09764767e-02
2.05475181e-01 5.76049566e-01 -1.72566727e-01 -2.35834986e-01
-1.89624339e-01 5.72684288e-01 8.18843320e-02 -4.81959768e-02
1.52961850e+00 -4.29599702e-01 -1.53706357e-01 2.42510319e-01
7.40726829e-01 -1.96573406e-01 -1.02546775e+00 4.93318439e-01
2.80366600e-01 -6.65360093e-01 4.38149050e-02 -7.62301445e-01
-1.21043110e+00 4.68578577e-01 2.91972816e-01 3.01496416e-01
1.42338860e+00 -6.23772800e-01 1.08836734e+00 5.96391439e-01
9.26426947e-01 -1.29655886e+00 -2.13414639e-01 3.63595337e-01
5.82352817e-01 -9.17331874e-01 -1.10551991e-01 -4.46844012e-01
-7.57698119e-01 1.12334204e+00 8.22732687e-01 1.88720211e-01
2.07224101e-01 5.43347299e-01 -1.79004207e-01 -1.10111557e-01
-5.37530720e-01 -3.06322277e-01 -3.24630976e-01 5.17077982e-01
1.00195609e-01 -1.00663761e-02 -9.05534625e-01 8.93308163e-01
-1.49874955e-01 -7.29328319e-02 1.94199681e-01 1.45691383e+00
-5.76733530e-01 -1.29449248e+00 -4.18866336e-01 3.33522677e-01
2.41245180e-01 3.57211560e-01 -7.34654814e-02 9.53494966e-01
2.87550747e-01 9.30296957e-01 1.02169953e-01 -3.28713030e-01
8.38222384e-01 -1.37121186e-01 4.11779940e-01 5.64589091e-02
-5.97647369e-01 5.70313074e-02 4.75246869e-02 -2.23882481e-01
1.01953171e-01 -5.33777714e-01 -1.47751033e+00 -4.44276154e-01
-6.28087282e-01 7.35174537e-01 9.73931134e-01 5.45271695e-01
4.71159458e-01 8.93355012e-01 9.74239230e-01 -6.48074210e-01
-5.56583881e-01 -4.56904888e-01 -8.96088719e-01 -2.55752891e-01
-4.40609306e-01 -9.98788953e-01 2.00809106e-01 -5.69889009e-01]
|
[5.702325820922852, 3.4919850826263428]
|
1ff008a3-9762-4d7b-b88e-daebd671ff08
|
mixskd-self-knowledge-distillation-from-mixup
|
2208.05768
| null |
https://arxiv.org/abs/2208.05768v1
|
https://arxiv.org/pdf/2208.05768v1.pdf
|
MixSKD: Self-Knowledge Distillation from Mixup for Image Recognition
|
Unlike the conventional Knowledge Distillation (KD), Self-KD allows a network to learn knowledge from itself without any guidance from extra networks. This paper proposes to perform Self-KD from image Mixture (MixSKD), which integrates these two techniques into a unified framework. MixSKD mutually distills feature maps and probability distributions between the random pair of original images and their mixup images in a meaningful way. Therefore, it guides the network to learn cross-image knowledge by modelling supervisory signals from mixup images. Moreover, we construct a self-teacher network by aggregating multi-stage feature maps for providing soft labels to supervise the backbone classifier, further improving the efficacy of self-boosting. Experiments on image classification and transfer learning to object detection and semantic segmentation demonstrate that MixSKD outperforms other state-of-the-art Self-KD and data augmentation methods. The code is available at https://github.com/winycg/Self-KD-Lib.
|
['Qian Zhang', 'Yongjun Xu', 'Jiwen Wu', 'Xiang Zhi', 'Linhang Cai', 'Helong Zhou', 'Zhulin An', 'Chuanguang Yang']
|
2022-08-11
| null | null | null | null |
['self-knowledge-distillation']
|
['computer-vision']
|
[ 6.33320063e-02 4.48715657e-01 -3.70834559e-01 -4.61582094e-01
-6.57205641e-01 -3.97396356e-01 7.34141886e-01 -1.82241082e-01
-3.73632848e-01 6.20413244e-01 -1.01828285e-01 -1.51243106e-01
8.47679526e-02 -7.33450234e-01 -9.07929361e-01 -9.47260559e-01
4.05660719e-01 3.78071487e-01 4.02301401e-01 1.05762206e-01
-4.14713055e-01 1.42385274e-01 -1.43021035e+00 2.70580709e-01
1.06529641e+00 8.97854090e-01 4.20245558e-01 4.61078554e-01
-1.70711666e-01 8.84746969e-01 -4.08823907e-01 -4.19566989e-01
1.51758626e-01 -4.41247791e-01 -7.92827904e-01 5.21652326e-02
6.27276480e-01 -1.87960550e-01 -3.98661166e-01 1.14542174e+00
4.45770562e-01 3.10997404e-02 7.29633808e-01 -1.51022279e+00
-9.94640529e-01 9.31830883e-01 -5.66666424e-01 1.90311342e-01
-1.92336306e-01 1.73838541e-01 6.24190271e-01 -8.13426077e-01
2.71018296e-01 1.32020128e+00 6.95051968e-01 5.97798288e-01
-1.54301298e+00 -1.08818161e+00 2.23737508e-01 2.63419300e-01
-1.29477358e+00 -3.29925686e-01 1.00669837e+00 -4.35525358e-01
2.40744278e-01 -1.83447991e-02 6.60673141e-01 1.21065605e+00
-4.71964687e-01 1.38124526e+00 1.41959798e+00 -4.98809934e-01
1.66574240e-01 5.63290298e-01 3.24187487e-01 7.74151623e-01
-1.71616394e-02 8.11664835e-02 -5.05508065e-01 3.33799869e-02
9.05857205e-01 -7.87489936e-02 -1.63221240e-01 -6.99155629e-01
-1.05360281e+00 6.48635924e-01 6.30939364e-01 3.13381732e-01
-2.46782854e-01 1.68506771e-01 6.24703839e-02 1.67318180e-01
5.65194309e-01 6.23999052e-02 -5.30577600e-01 3.67509037e-01
-7.73007929e-01 -2.64693927e-02 5.55807412e-01 8.04960668e-01
1.13930511e+00 1.92835435e-01 -1.48288667e-01 9.23493028e-01
3.60493422e-01 5.69755435e-01 6.90880775e-01 -1.05320954e+00
2.92931538e-04 6.68640435e-01 -2.88121134e-01 -6.32485271e-01
-2.32477516e-01 -7.42223740e-01 -9.67175663e-01 2.34038681e-01
2.46062472e-01 -2.29774237e-01 -1.19835651e+00 1.89611220e+00
6.23918474e-01 8.24768186e-01 3.15811336e-01 6.94624126e-01
1.14468575e+00 4.55729872e-01 1.77057326e-01 2.22976543e-02
1.13479996e+00 -1.28183031e+00 -5.92274487e-01 -4.18180078e-01
4.48403120e-01 -4.23926532e-01 9.77960348e-01 3.48357260e-01
-8.96571159e-01 -1.07723975e+00 -1.03197455e+00 3.54395270e-01
-4.08009142e-01 2.13156685e-01 6.83412194e-01 4.91880953e-01
-1.21012616e+00 4.32946861e-01 -8.53877127e-01 -6.58114953e-03
9.47343051e-01 3.52895558e-01 -4.17111158e-01 -5.57240695e-02
-1.15184999e+00 8.44089985e-01 7.85210729e-01 -1.69778034e-01
-1.12237036e+00 -9.13558304e-01 -9.14228320e-01 -3.70417297e-01
4.08275783e-01 -7.58534372e-01 1.39970672e+00 -1.31193972e+00
-1.59842706e+00 9.98720467e-01 1.49701595e-01 -4.82294887e-01
4.53271508e-01 -2.83542573e-01 -1.91617474e-01 1.74276486e-01
2.54062772e-01 1.29261279e+00 1.24571133e+00 -1.70326996e+00
-7.10543156e-01 -3.25391740e-01 -1.07515909e-01 3.54745477e-01
-2.63492137e-01 -4.60122079e-01 -6.51801884e-01 -6.44501626e-01
2.44156439e-02 -7.75421858e-01 -1.53049275e-01 -2.78990000e-01
-5.99614859e-01 -3.76887709e-01 9.71463859e-01 -2.98483163e-01
7.07038999e-01 -2.27012372e+00 3.72671448e-02 7.68981650e-02
4.13670152e-01 5.55976927e-01 -3.69582653e-01 -2.09604919e-01
-3.37643892e-01 -2.32303411e-01 -3.14843446e-01 -4.70405906e-01
-8.50975960e-02 5.22942483e-01 -2.02984810e-01 3.28622520e-01
4.06692207e-01 1.11653697e+00 -1.07184184e+00 -5.81155896e-01
4.35213476e-01 5.22463083e-01 -2.65200168e-01 4.12017018e-01
-1.86582297e-01 7.60096729e-01 -2.42916018e-01 3.93312186e-01
8.72695923e-01 -4.13470745e-01 1.17174029e-01 -3.29479039e-01
2.88620919e-01 -2.98263151e-02 -1.16165090e+00 1.70215464e+00
-2.81532228e-01 3.50987315e-01 1.41743710e-02 -1.32300544e+00
9.36486423e-01 1.85517162e-01 3.77150446e-01 -5.83762288e-01
3.09164941e-01 -4.23631705e-02 -1.16689391e-01 -2.00661585e-01
-2.80204862e-02 -1.42415494e-01 2.01996326e-01 4.44538474e-01
8.55596185e-01 -1.30217224e-01 1.24603761e-02 2.79910088e-01
7.65609384e-01 4.05784339e-01 1.23539276e-01 -1.68173149e-01
4.58631724e-01 -1.04234859e-01 7.26780891e-01 8.59424293e-01
-2.17467889e-01 4.53927547e-01 1.52495369e-01 -7.21546486e-02
-6.25260353e-01 -1.47369397e+00 -2.57823765e-01 1.18798316e+00
2.64695376e-01 2.10092925e-02 -9.85656857e-01 -1.10710824e+00
2.02892736e-01 7.58861244e-01 -8.55176806e-01 -3.53970855e-01
-8.80052224e-02 -8.92141938e-01 6.04134142e-01 5.72598219e-01
9.48615015e-01 -9.51862931e-01 2.92483307e-02 7.84621909e-02
-1.28814161e-01 -1.17269337e+00 -1.38098463e-01 6.33833528e-01
-6.90303624e-01 -1.12320566e+00 -7.39883661e-01 -8.89902532e-01
8.47963214e-01 4.45093066e-01 1.10033870e+00 -3.23016733e-01
-2.32667893e-01 6.60732090e-01 -2.92946905e-01 -4.84050095e-01
-6.12109363e-01 5.74757233e-02 8.96803960e-02 1.80303022e-01
5.65425098e-01 -7.96190977e-01 -4.48477089e-01 4.19038206e-01
-8.98980141e-01 4.34850365e-01 8.68305922e-01 8.22184205e-01
6.57972813e-01 2.08729550e-01 8.19720984e-01 -1.00083661e+00
2.54205048e-01 -5.68403900e-01 -4.19367284e-01 2.18630686e-01
-7.21307814e-01 -2.78245229e-02 2.99905568e-01 -6.98533118e-01
-1.26727164e+00 2.33824268e-01 -5.64435720e-02 -7.57605731e-01
-5.40472984e-01 -1.73633192e-02 -3.14163417e-01 -8.99279118e-02
7.34681308e-01 4.42990869e-01 1.68427944e-01 -4.64824885e-01
1.07158637e+00 6.50519073e-01 1.07023156e+00 -5.73878527e-01
1.00441146e+00 5.42246401e-01 -5.78113377e-01 -4.99035865e-01
-1.32352400e+00 -4.46348339e-01 -9.56964493e-01 -2.00530112e-01
7.95686424e-01 -1.25280118e+00 -3.57839733e-01 8.59569371e-01
-7.74124384e-01 -7.69822717e-01 -5.99403322e-01 4.12963420e-01
-6.12693012e-01 8.33530352e-02 -4.73524213e-01 -5.85600197e-01
-2.80113518e-02 -9.47902918e-01 6.93198144e-01 6.19827032e-01
1.03200980e-01 -1.14825881e+00 1.12076789e-01 6.55380905e-01
1.33578479e-01 -5.35912290e-02 4.35201794e-01 -8.49310875e-01
-3.23032886e-01 5.67467064e-02 -3.91120195e-01 1.04175961e+00
3.85694355e-01 -1.66421130e-01 -1.39773369e+00 -3.52396332e-02
-1.04126729e-01 -7.47873902e-01 1.14841759e+00 5.90414226e-01
9.88811195e-01 -1.47411361e-01 -3.76395702e-01 6.32520497e-01
1.20937610e+00 -1.48612456e-02 3.93547118e-01 2.96445906e-01
9.41624641e-01 6.29865706e-01 4.68187362e-01 1.03075420e-02
6.94927454e-01 3.15213084e-01 4.32687402e-01 -6.08830571e-01
-5.71543694e-01 -3.07671547e-01 3.35722208e-01 7.92961657e-01
3.83961588e-01 1.57789841e-01 -8.26925814e-01 6.69474959e-01
-2.00383949e+00 -5.79568326e-01 6.78518414e-02 1.73067343e+00
1.26423132e+00 3.10762599e-02 1.01631761e-01 -3.40799015e-04
9.03919637e-01 -4.78193816e-03 -8.54055583e-01 2.18903288e-01
-3.54340374e-01 4.18167301e-02 4.30990428e-01 4.79579657e-01
-1.49640155e+00 1.18490243e+00 5.84174824e+00 1.24587357e+00
-9.08286035e-01 4.97227520e-01 8.38837385e-01 2.67645776e-01
-1.14133649e-01 -9.87181962e-02 -9.50694323e-01 3.69111508e-01
6.40170038e-01 1.97693035e-01 1.64693058e-01 1.06241715e+00
-2.62456179e-01 -2.79227257e-01 -8.36710572e-01 8.47537160e-01
4.22170423e-02 -1.34474885e+00 -2.01152451e-02 -2.02905491e-01
1.08560193e+00 3.09397787e-01 1.68744490e-01 5.12458742e-01
1.11580622e+00 -7.19256163e-01 5.78319550e-01 4.34532046e-01
4.67749447e-01 -6.15684032e-01 6.43303156e-01 4.57853615e-01
-8.47986579e-01 3.55921723e-02 -2.80343235e-01 3.26929152e-01
-1.98290423e-01 8.77368629e-01 -8.82859528e-01 6.34170592e-01
7.35000193e-01 8.48058164e-01 -7.01245368e-01 6.67368829e-01
-6.60604775e-01 8.65379810e-01 -1.59925327e-01 5.73858440e-01
1.55857146e-01 -4.76439185e-02 3.29218507e-01 1.31985843e+00
-2.33277276e-01 -1.04164138e-01 3.77874941e-01 9.83851910e-01
-1.63906932e-01 -2.15789020e-01 -4.59394217e-01 2.04904944e-01
5.29392660e-01 1.48573196e+00 -7.61830688e-01 -6.80370927e-01
-3.31774741e-01 1.00053811e+00 4.67691779e-01 3.91882479e-01
-7.66810179e-01 -6.07886538e-02 5.17545044e-01 -2.64552981e-01
4.21348542e-01 1.68343619e-01 -1.54162407e-01 -1.05706298e+00
-2.61637628e-01 -6.58385217e-01 2.51851022e-01 -9.33797538e-01
-1.48835349e+00 5.36978364e-01 1.56207532e-01 -1.04655600e+00
-4.89323922e-02 -4.41084266e-01 -5.91337323e-01 6.65808797e-01
-1.79634857e+00 -1.37448370e+00 -5.40258646e-01 7.64232635e-01
2.42239058e-01 -2.76029050e-01 7.23578930e-01 2.38070339e-01
-7.24605858e-01 4.53148514e-01 1.67544484e-01 3.91008168e-01
8.40090990e-01 -1.43039441e+00 1.85746506e-01 6.57792866e-01
3.01463366e-01 2.37230062e-01 3.62756193e-01 -5.69927394e-01
-7.61968017e-01 -1.33211696e+00 -2.72394083e-02 -5.57483852e-01
6.07684493e-01 -2.04659998e-01 -1.08693254e+00 6.58580124e-01
2.73814082e-01 3.17233950e-01 8.60992193e-01 6.47668839e-02
-5.93059123e-01 -3.78031731e-01 -9.76330400e-01 4.73041505e-01
7.99359858e-01 -4.45200443e-01 -6.83330774e-01 1.78300440e-01
8.18114996e-01 -2.53017545e-01 -8.71657372e-01 4.72589791e-01
2.00474292e-01 -9.95377719e-01 1.07684362e+00 -2.99013436e-01
2.42845297e-01 -4.73929912e-01 -8.92169029e-03 -1.54511976e+00
-3.41835648e-01 -3.63289386e-01 -3.71940136e-01 1.61407673e+00
2.14283720e-01 -6.29789650e-01 8.22010338e-01 1.88951656e-01
-1.12165816e-01 -6.14540517e-01 -7.21373916e-01 -9.64684486e-01
1.39524177e-01 -5.86961031e-01 4.69543785e-01 1.37249219e+00
-2.76739120e-01 4.41717535e-01 -1.67965963e-01 2.52492696e-01
1.01194572e+00 -1.13031283e-01 9.19265211e-01 -1.13580549e+00
-3.42586160e-01 -4.47887897e-01 -4.01545912e-01 -1.02569366e+00
3.81357342e-01 -1.21320081e+00 1.03660822e-01 -1.42664957e+00
4.17371213e-01 -7.22354114e-01 -5.78060985e-01 1.05901134e+00
-4.63560671e-01 6.27657712e-01 1.03032045e-01 1.37278408e-01
-8.20038438e-01 7.43528426e-01 1.40561581e+00 -3.38672698e-01
-2.82047451e-01 -7.05918074e-02 -1.10060251e+00 8.94457579e-01
1.02591920e+00 -6.57202840e-01 -7.07471550e-01 -1.87409192e-01
-3.30424905e-01 -4.36317265e-01 5.76764166e-01 -1.10097563e+00
1.73253074e-01 6.94907159e-02 5.84075391e-01 -4.91260558e-01
2.18793169e-01 -5.88588715e-01 -5.19613326e-02 2.95860112e-01
-2.77870297e-01 -6.96463108e-01 3.23724359e-01 6.20950818e-01
-2.85996914e-01 -2.25308791e-01 9.92618322e-01 -1.21084139e-01
-7.77813673e-01 1.85446724e-01 -1.52863592e-01 1.35371402e-01
1.08108199e+00 -2.64713377e-01 -4.36122239e-01 -2.28300214e-01
-9.84188199e-01 4.51530188e-01 1.14787392e-01 4.33580339e-01
5.24611473e-01 -1.42127597e+00 -6.71228468e-01 3.58461618e-01
4.63445298e-02 6.16050303e-01 4.14978832e-01 8.70425344e-01
6.86710253e-02 9.91506651e-02 -2.50352919e-01 -9.34288383e-01
-1.13383412e+00 3.97161633e-01 3.32536191e-01 -1.33715779e-01
-6.02817476e-01 1.19484103e+00 6.96642995e-01 -8.66425097e-01
3.08247894e-01 -1.61765218e-01 -1.71695948e-01 1.30269080e-01
5.55872083e-01 1.27371937e-01 -2.11567283e-01 -5.02341926e-01
-7.27295503e-02 3.54919702e-01 -3.60053748e-01 -2.78521925e-02
1.22141290e+00 -1.88038260e-01 -3.33300866e-02 4.74854410e-01
1.08407187e+00 -4.27654058e-01 -1.82361341e+00 -8.08956325e-01
-1.85130373e-01 -8.92206728e-02 2.87815809e-01 -9.46269870e-01
-1.39335763e+00 7.48846233e-01 6.89366996e-01 -7.81191885e-02
1.25032091e+00 4.31826711e-01 4.29600477e-01 3.06902677e-01
-2.09266134e-02 -9.01618481e-01 4.32625681e-01 2.44913191e-01
5.78785062e-01 -1.36857343e+00 -1.90115064e-01 -3.78931403e-01
-8.29400718e-01 7.04834878e-01 1.00440538e+00 -2.66446382e-01
8.90928864e-01 2.55352795e-01 5.54112077e-01 2.14525573e-02
-5.39893508e-01 -5.41568756e-01 3.52116793e-01 1.04440320e+00
-5.23940660e-02 4.17024381e-02 4.23913538e-01 7.92209804e-01
-8.44402760e-02 5.29061332e-02 1.92190066e-01 6.95675075e-01
-3.80963355e-01 -1.18180764e+00 -4.48699296e-01 3.32313031e-01
-1.17219895e-01 -9.55435932e-02 -2.37452775e-01 6.44338787e-01
5.67027450e-01 7.13029265e-01 1.33181408e-01 -5.98442733e-01
-2.14904137e-02 1.55941024e-01 5.36102235e-01 -6.09491408e-01
-4.27658588e-01 1.94242358e-01 -2.64412701e-01 -3.26827854e-01
-7.73731828e-01 -4.96967226e-01 -1.20769560e+00 1.11295350e-01
-4.72814977e-01 1.72077835e-01 6.69238329e-01 9.25584733e-01
2.93991596e-01 8.40858519e-01 5.78416944e-01 -9.14278507e-01
-3.07985961e-01 -1.02501070e+00 -6.59522951e-01 1.96608022e-01
3.03315163e-01 -8.43433678e-01 -2.38278940e-01 3.38000715e-01]
|
[9.415118217468262, 1.656624674797058]
|
066c6410-f9d4-44bc-bb57-43e9c259a1f8
|
machine-learning-strategies-to-improve
|
2212.08744
| null |
https://arxiv.org/abs/2212.08744v1
|
https://arxiv.org/pdf/2212.08744v1.pdf
|
Machine Learning Strategies to Improve Generalization in EEG-based Emotion Assessment: \\a Systematic Review
|
A systematic review on machine-learning strategies for improving generalizability (cross-subjects and cross-sessions) electroencephalography (EEG) based in emotion classification was realized. In this context, the non-stationarity of EEG signals is a critical issue and can lead to the Dataset Shift problem. Several architectures and methods have been proposed to address this issue, mainly based on transfer learning methods. 418 papers were retrieved from the Scopus, IEEE Xplore and PubMed databases through a search query focusing on modern machine learning techniques for generalization in EEG-based emotion assessment. Among these papers, 75 were found eligible based on their relevance to the problem. Studies lacking a specific cross-subject and cross-session validation strategy and making use of other biosignals as support were excluded. On the basis of the selected papers' analysis, a taxonomy of the studies employing Machine Learning (ML) methods was proposed, together with a brief discussion on the different ML approaches involved. The studies with the best results in terms of average classification accuracy were identified, supporting that transfer learning methods seem to perform better than other approaches. A discussion is proposed on the impact of (i) the emotion theoretical models and (ii) psychological screening of the experimental sample on the classifier performances.
|
['Roberto Prevete', 'Nicola Moccaldi', 'Giovanna Mastrati', 'Davide Marocco', "Giovanni D'Errico", 'Pasquale Arpaia', 'Andrea Apicella']
|
2022-12-16
| null | null | null | null |
['emotion-classification', 'emotion-classification']
|
['computer-vision', 'natural-language-processing']
|
[ 5.46534993e-02 -1.98402286e-01 -5.69189608e-01 -3.99102926e-01
-5.68648636e-01 -3.05485606e-01 7.19575882e-02 4.38733637e-01
-7.75735140e-01 1.06141019e+00 -2.04332530e-01 -2.35064298e-01
-7.61398792e-01 -2.50746518e-01 -5.93659043e-01 -6.91443086e-01
-4.13513184e-01 -5.42651340e-02 -2.06649229e-01 2.04932913e-01
7.71223009e-01 5.50637960e-01 -1.76305532e+00 6.55651271e-01
1.08276260e+00 1.04958498e+00 2.59343594e-01 -1.62731800e-02
1.62695661e-01 2.75012702e-01 -8.47988129e-01 -3.29523444e-01
-3.47993046e-01 -6.71298504e-01 -5.98362148e-01 -4.48969781e-01
-8.10100138e-02 2.37862304e-01 1.78791046e-01 8.49621356e-01
1.02352035e+00 -2.08452359e-01 8.52915227e-01 -1.46548581e+00
-6.07274711e-01 3.91685754e-01 -1.38509870e-01 5.32732725e-01
5.64340472e-01 -2.52804816e-01 3.42255920e-01 -1.11642480e+00
3.15724611e-01 5.87917387e-01 7.88069248e-01 4.27984685e-01
-1.08584249e+00 -1.26583385e+00 5.48570529e-02 8.64765227e-01
-1.25221515e+00 -1.98441446e-01 8.30125213e-01 -6.13488674e-01
1.29909527e+00 3.03354055e-01 1.12076545e+00 1.32609582e+00
9.19996738e-01 2.52310280e-02 1.74336874e+00 -6.42534673e-01
6.16351902e-01 8.38036358e-01 2.27073058e-01 -9.53488126e-02
4.43539441e-01 6.28661662e-02 -7.40165710e-01 -3.23572993e-01
2.17367128e-01 -5.44209301e-01 -3.52059096e-01 -2.17783600e-01
-7.58441031e-01 7.55689323e-01 1.92641214e-01 8.08986962e-01
-6.36315227e-01 -5.84409654e-01 1.04404724e+00 5.88241220e-01
6.77497983e-01 7.14613557e-01 -6.40582383e-01 -1.89782798e-01
-9.68000710e-01 -1.83898464e-01 7.47592986e-01 2.57217228e-01
2.48671636e-01 2.04803050e-01 3.46792489e-02 8.57964158e-01
1.10899834e-02 2.05602050e-01 9.14689243e-01 -3.27449560e-01
2.06589043e-01 4.59050208e-01 -2.24092290e-01 -1.07770753e+00
-8.96087527e-01 -4.10386950e-01 -7.50547171e-01 2.44936496e-01
-2.71227896e-01 -4.02880162e-01 -3.50196600e-01 1.51590180e+00
-2.29666159e-01 -1.29948765e-01 -3.44785526e-02 7.83296943e-01
8.96698773e-01 2.38841906e-01 5.18371105e-01 -7.86898196e-01
1.36379826e+00 -4.36822832e-01 -1.02385747e+00 -7.04736337e-02
7.68326342e-01 -3.14561874e-01 8.14830601e-01 9.39429998e-01
-1.03888226e+00 -5.16360939e-01 -1.23592734e+00 4.35450971e-01
-7.62832701e-01 2.91636586e-01 5.98932624e-01 1.11625624e+00
-8.77377152e-01 6.41636789e-01 -6.17054164e-01 -5.71909845e-01
2.41214484e-01 6.30030930e-01 -4.67810124e-01 4.36363906e-01
-1.45208704e+00 1.59648967e+00 4.80324805e-01 1.35050207e-01
-9.09468755e-02 -7.48366416e-01 -3.79103452e-01 -7.38882348e-02
-1.94985732e-01 -4.92360145e-01 4.38120335e-01 -1.44541883e+00
-1.36654878e+00 9.96547520e-01 -4.68735918e-02 -2.95732051e-01
1.02110229e-01 2.20186822e-02 -6.67138994e-01 4.64847922e-01
-3.34787033e-02 2.35771805e-01 5.54748297e-01 -8.48845661e-01
-4.88949716e-01 -7.57190943e-01 -5.26549816e-01 1.72512829e-01
-3.94551486e-01 4.10979778e-01 3.90991092e-01 -6.63459182e-01
-3.15800011e-02 -6.90871119e-01 2.63615191e-01 -6.10346973e-01
2.65631348e-01 -5.80605388e-01 3.56916070e-01 -7.24950016e-01
1.43259287e+00 -2.07145429e+00 1.25485003e-01 2.73337275e-01
-8.83468017e-02 1.80861756e-01 2.89614320e-01 6.31873608e-01
-7.30746746e-01 7.47278333e-02 4.15712856e-02 1.50578052e-01
-8.14621523e-02 -2.61498153e-01 -3.60063612e-02 7.50532746e-01
-1.08287614e-02 7.34740376e-01 -5.13008296e-01 -2.44138747e-01
4.63541359e-01 4.93771940e-01 -1.60518348e-01 9.09509361e-02
7.85525620e-01 3.93447608e-01 -1.23856752e-03 2.75838882e-01
5.42048514e-01 2.69166559e-01 1.22917302e-01 -3.19337279e-01
-2.28975028e-01 4.82961029e-01 -9.01522338e-01 1.38392067e+00
-4.20324534e-01 7.58397281e-01 -1.92685246e-01 -1.27228105e+00
1.11088490e+00 8.06212664e-01 6.16646171e-01 -7.51657009e-01
3.34494829e-01 6.62280738e-01 5.01868844e-01 -8.10197234e-01
-2.60219932e-01 -3.64516228e-01 2.37227112e-01 1.57759652e-01
5.26083708e-01 -7.61693045e-02 -2.01411262e-01 -3.89266670e-01
5.27261436e-01 5.48701957e-02 5.67237139e-01 -7.54192114e-01
4.67581779e-01 -3.24871033e-01 5.18893898e-01 6.22929633e-01
-1.32558674e-01 8.53781104e-02 5.84831953e-01 -2.35980317e-01
-4.19445544e-01 -4.69244987e-01 -7.90077031e-01 7.00279951e-01
-3.29340994e-01 -2.00568795e-01 -9.41670001e-01 -1.55274525e-01
-2.77354777e-01 1.05512559e+00 -9.09336150e-01 -7.97704637e-01
1.35237113e-01 -1.30996048e+00 4.09578711e-01 4.44748759e-01
1.79157197e-01 -1.42765141e+00 -1.26367927e+00 1.09313846e-01
4.25714906e-03 -8.43851388e-01 6.29609704e-01 6.73630953e-01
-1.17582190e+00 -9.45004761e-01 -6.64913237e-01 -8.18158567e-01
3.10661048e-01 -2.72954315e-01 6.06187940e-01 -2.07438096e-01
-1.94983527e-01 7.23709464e-01 -4.69273239e-01 -8.67273688e-01
-3.88718098e-02 8.72864202e-02 3.54948163e-01 -2.22395822e-01
1.01519859e+00 -7.51039565e-01 -4.49474812e-01 3.59677500e-03
-6.10107422e-01 -5.22650182e-01 7.70988822e-01 8.65890384e-01
2.97495797e-02 -5.37572503e-02 1.33342993e+00 -4.03303444e-01
1.05126631e+00 -7.54822135e-01 -1.44570917e-01 1.76609412e-01
-1.25289142e+00 -7.16976821e-01 4.13102247e-02 -6.83144689e-01
-7.94890344e-01 -5.46470761e-01 8.25968310e-02 -1.26658902e-01
-4.39725906e-01 7.23646820e-01 -1.25091463e-01 -4.24952567e-01
7.08761752e-01 1.44478753e-01 2.96273883e-02 -1.45053014e-01
-5.69983125e-01 8.53162706e-01 -1.21763922e-01 -4.93214071e-01
-1.52262181e-01 -1.81110740e-01 -1.79429606e-01 -8.15814197e-01
-1.00823946e-01 -2.16969118e-01 -7.24133432e-01 -4.16768253e-01
8.95115435e-01 -8.29370916e-01 -6.94163501e-01 3.86937588e-01
-9.02924538e-01 -2.21342698e-01 1.42162129e-01 1.40258718e+00
-6.35199904e-01 -6.48521855e-02 -3.55924278e-01 -1.00299489e+00
-7.55449593e-01 -1.24275553e+00 3.79782230e-01 1.54509649e-01
-6.63248241e-01 -8.67690802e-01 2.24072278e-01 -2.84637362e-01
2.81341106e-01 1.35868430e-01 1.18031776e+00 -1.08843982e+00
7.25505650e-01 -3.94524902e-01 1.25158638e-01 2.24354565e-01
2.81137735e-01 -2.37013802e-01 -1.10070741e+00 -1.45507604e-01
5.58328152e-01 -1.59425110e-01 2.22096428e-01 6.98239565e-01
1.15505040e+00 5.75914383e-02 -3.48336041e-01 1.34187162e-01
1.28046691e+00 9.58831549e-01 8.62330973e-01 6.01692319e-01
-4.11963724e-02 1.02752340e+00 3.32535326e-01 4.07813221e-01
4.16893810e-02 6.05108857e-01 2.27056164e-02 1.00363400e-02
2.74208397e-01 3.64254892e-01 3.79365742e-01 7.20365644e-01
-2.60442704e-01 1.51358515e-01 -7.89784610e-01 3.64532769e-01
-1.33213723e+00 -7.90226996e-01 -3.42287630e-01 2.43896627e+00
3.68779093e-01 2.03831628e-01 3.43892276e-01 5.08472383e-01
5.78872085e-01 -4.77061063e-01 -3.78346086e-01 -8.81940186e-01
-1.75658137e-01 5.14721930e-01 1.28212899e-01 -9.12314802e-02
-6.94523931e-01 2.23893672e-01 6.55816078e+00 5.42464674e-01
-1.47394633e+00 1.98076770e-01 4.21262056e-01 -2.52492458e-01
3.78430188e-01 -3.78749192e-01 -2.42832705e-01 7.86720991e-01
1.48455977e+00 -3.02457482e-01 1.67029008e-01 5.93224347e-01
4.58460540e-01 -6.39380872e-01 -9.11185265e-01 1.29384363e+00
3.67956787e-01 -7.86801159e-01 -2.75868922e-01 -7.81037882e-02
2.67148584e-01 -1.29028767e-01 7.94560388e-02 4.42679375e-01
-1.16904664e+00 -7.94387400e-01 5.34623921e-01 5.40954709e-01
8.34562421e-01 -6.45958960e-01 1.09199417e+00 1.09648570e-01
-4.43038225e-01 -1.63785264e-01 -3.05747002e-01 -3.25919688e-01
-2.03345627e-01 5.38917124e-01 -3.43344003e-01 8.17146719e-01
1.24954593e+00 6.43798113e-01 -5.40656805e-01 1.22164428e+00
-1.25965804e-01 8.92446101e-01 1.46916648e-03 -4.74321604e-01
-7.35912174e-02 -2.75812984e-01 3.62183660e-01 1.39386320e+00
4.20545608e-01 3.03207040e-01 -7.19632447e-01 8.22272122e-01
4.88761455e-01 5.17119646e-01 -4.81376588e-01 1.57276839e-01
1.84528306e-01 1.19063020e+00 -9.32887375e-01 -1.74686611e-01
-7.30467319e-01 9.13522482e-01 -8.60452950e-02 3.30788523e-01
-5.66724002e-01 -7.09388077e-01 5.25505580e-02 2.10838541e-02
-2.92360276e-01 3.90062928e-01 -4.90841687e-01 -8.67366135e-01
2.07937002e-01 -1.01547778e+00 5.70439875e-01 -9.56434548e-01
-1.26822317e+00 7.53468990e-01 6.07794881e-01 -9.58920896e-01
3.06734219e-02 -7.63989568e-01 -3.29053789e-01 1.07063949e+00
-8.43632102e-01 -6.10976756e-01 7.48766661e-02 5.31601608e-01
2.82816082e-01 -1.28829777e-01 1.20991111e+00 3.13430309e-01
-5.21815002e-01 3.04617971e-01 1.87813025e-02 -3.81897092e-01
9.67002392e-01 -8.88444066e-01 -7.37112224e-01 1.39844507e-01
-4.72614527e-01 5.63506424e-01 5.35021365e-01 -6.26324832e-01
-8.48442674e-01 -1.82252645e-01 9.36959267e-01 -1.48350433e-01
4.33346629e-01 -3.95252615e-01 -1.09462905e+00 4.99343097e-01
5.72705984e-01 -6.79418027e-01 1.24521041e+00 3.05000484e-01
1.10282846e-01 -1.09681025e-01 -1.14570606e+00 3.32620651e-01
4.80869949e-01 -3.75127584e-01 -1.01638508e+00 7.08261132e-02
-2.58890063e-01 -6.69915751e-02 -1.19314671e+00 7.21387982e-01
8.79842103e-01 -9.92171288e-01 5.26669323e-01 -7.13805795e-01
1.54945582e-01 5.24951935e-01 3.40196103e-01 -1.50289285e+00
-4.37963814e-01 -2.00431734e-01 4.22598630e-01 9.81432021e-01
3.07732850e-01 -1.03730559e+00 2.67344207e-01 8.63095462e-01
-1.72102988e-01 -9.35548186e-01 -1.21162319e+00 -6.00359797e-01
3.99543315e-01 -4.06444728e-01 1.55925050e-01 9.83848572e-01
9.51213002e-01 2.77278543e-01 -4.15868200e-02 -2.58772105e-01
2.41883774e-03 -2.79734194e-01 -1.43006817e-02 -1.38591981e+00
3.16153079e-01 -8.06704819e-01 -6.97968721e-01 2.14783743e-01
4.06183809e-01 -8.95607352e-01 -3.60465944e-01 -1.55588126e+00
1.80364430e-01 -3.62748560e-03 -6.10331118e-01 2.43937299e-01
-4.00533788e-02 -1.41847119e-01 -1.86770752e-01 -3.07633393e-02
-5.83553538e-02 4.79406685e-01 4.93519694e-01 3.08421314e-01
-4.55053508e-01 1.96924031e-01 -6.17118359e-01 5.87713957e-01
1.24020123e+00 -7.68830180e-01 -5.44962943e-01 1.46184370e-01
3.71178329e-01 -5.78647554e-02 2.63065040e-01 -1.00442612e+00
4.32820283e-02 2.31169418e-01 5.91139913e-01 -3.55432659e-01
3.46826464e-01 -8.81821811e-01 3.18575561e-01 6.69027805e-01
-3.53946865e-01 3.74386430e-01 8.80949795e-01 1.76470820e-02
-8.47896710e-02 -7.72475481e-01 5.10715246e-01 1.52231365e-01
-4.06730592e-01 -3.35842699e-01 -1.00213742e+00 -2.60457844e-01
1.15507603e+00 -4.58788365e-01 1.79768994e-01 -3.52345586e-01
-9.62906718e-01 -3.11355054e-01 -4.52951975e-02 4.74444300e-01
4.64940369e-01 -1.16410196e+00 -4.88799691e-01 -5.08090742e-02
3.01452696e-01 -1.14904642e+00 6.81544542e-01 1.73273778e+00
6.69689775e-02 7.86084771e-01 -8.98268700e-01 -2.50771701e-01
-1.08311021e+00 8.18511009e-01 5.75726926e-01 2.80896962e-01
-3.64792019e-01 4.77660447e-01 1.79727767e-02 -1.79394618e-01
3.88128579e-01 -1.60630152e-01 -4.91857409e-01 6.36414170e-01
5.32186270e-01 6.50277376e-01 7.71345973e-01 -3.58416349e-01
-7.63743639e-01 4.97926474e-01 3.70544255e-01 -2.77738810e-01
1.37613606e+00 5.19527541e-03 -4.82802749e-01 1.11877394e+00
1.12366438e+00 -3.47440571e-01 -2.71105498e-01 4.01992559e-01
1.75241921e-02 1.06864594e-01 -1.62666366e-02 -1.14756072e+00
-6.04337513e-01 1.05513775e+00 1.16155696e+00 -6.80138096e-02
1.52772534e+00 -4.00542676e-01 -3.98632258e-01 1.13068715e-01
6.09950721e-01 -1.43102634e+00 -2.00262755e-01 -1.57947212e-01
1.20375812e+00 -1.03646541e+00 6.16437569e-02 -1.46282881e-01
-7.88152754e-01 1.47126222e+00 3.57245207e-01 -1.59104452e-01
1.08095741e+00 2.05095802e-02 -1.50304228e-01 -3.00575078e-01
-5.24749160e-01 3.78818214e-01 4.67756689e-01 8.51870716e-01
5.67924201e-01 -2.86417957e-02 -1.45124161e+00 1.32148623e+00
-6.54819161e-02 3.92018527e-01 2.41560489e-01 6.49050832e-01
8.16641301e-02 -7.99844503e-01 -5.74514449e-01 7.73406267e-01
-6.61176741e-01 8.02701991e-03 -5.22275150e-01 1.10987115e+00
2.61664778e-01 1.13550663e+00 -8.24209601e-02 -3.65476727e-01
4.20803905e-01 4.46173489e-01 7.00334191e-01 -3.82201940e-01
-1.07677054e+00 -2.64860261e-02 -1.97751187e-02 -3.64749938e-01
-8.06592643e-01 -8.70042682e-01 -8.18035781e-01 3.47787350e-01
-4.75219846e-01 5.83531678e-01 1.08781612e+00 1.02756810e+00
4.79789853e-01 5.97191513e-01 2.21805722e-01 -6.75056934e-01
-9.67527479e-02 -1.37715828e+00 -7.40021706e-01 -6.96881786e-02
-2.06986323e-01 -1.00269878e+00 -6.93639636e-01 -2.76976228e-01]
|
[13.253623962402344, 3.361152172088623]
|
545c5383-2802-44a9-98ae-7eb4dbff74dc
|
dense-and-aligned-captions-dac-promote
|
2305.19595
| null |
https://arxiv.org/abs/2305.19595v2
|
https://arxiv.org/pdf/2305.19595v2.pdf
|
Dense and Aligned Captions (DAC) Promote Compositional Reasoning in VL Models
|
Vision and Language (VL) models offer an effective method for aligning representation spaces of images and text, leading to numerous applications such as cross-modal retrieval, visual question answering, captioning, and more. However, the aligned image-text spaces learned by all the popular VL models are still suffering from the so-called `object bias' - their representations behave as `bags of nouns', mostly ignoring or downsizing the attributes, relations, and states of objects described/appearing in texts/images. Although some great attempts at fixing these `compositional reasoning' issues were proposed in the recent literature, the problem is still far from being solved. In this paper, we uncover two factors limiting the VL models' compositional reasoning performance. These two factors are properties of the paired VL dataset used for finetuning and pre-training the VL model: (i) the caption quality, or in other words `image-alignment', of the texts; and (ii) the `density' of the captions in the sense of mentioning all the details appearing on the image. We propose a fine-tuning approach for automatically treating these factors leveraging a standard VL dataset (CC3M). Applied to CLIP, we demonstrate its significant compositional reasoning performance increase of up to $\sim27\%$ over the base model, up to $\sim20\%$ over the strongest baseline, and by $6.7\%$ on average.
|
['Paola Cascante-Bonilla', 'Donghyun Kim', 'Leonid Karlinsky', 'Shimon Ullman', 'Rameswar Panda', 'Rogerio Feris', 'Raja Giryes', 'Roei Herzig', 'Amit Alfassy', 'Sivan Harary', 'Assaf Arbelle', 'Sivan Doveh']
|
2023-05-31
| null | null | null | null |
['cross-modal-retrieval']
|
['miscellaneous']
|
[ 2.95347929e-01 2.67505944e-02 -2.15755880e-01 -3.31820816e-01
-8.03153515e-01 -4.81141567e-01 1.01582849e+00 3.18527073e-01
-4.79015470e-01 3.33726555e-01 4.78658289e-01 -1.78732052e-01
1.14948131e-01 -4.53030556e-01 -8.94967735e-01 -6.81062818e-01
3.56937855e-01 4.77322757e-01 2.99681902e-01 -1.18720382e-01
1.88554093e-01 1.12220921e-01 -1.81863558e+00 5.95301092e-01
5.96598446e-01 1.01800311e+00 2.49212250e-01 3.33531708e-01
-5.49161017e-01 9.90876257e-01 -3.38317871e-01 -7.42214322e-01
5.92001379e-02 -4.74772751e-01 -6.50550008e-01 3.55946451e-01
9.42600012e-01 2.65849140e-02 -3.53327036e-01 1.23061097e+00
1.57441646e-01 3.32252793e-02 7.19397306e-01 -1.24829483e+00
-1.09984124e+00 5.33145666e-01 -7.31177747e-01 1.85202479e-01
1.87892258e-01 2.20137835e-01 1.26269555e+00 -9.41322505e-01
7.44094312e-01 1.45477188e+00 3.19279790e-01 6.44284844e-01
-1.43368709e+00 -4.31548804e-01 3.29051584e-01 3.66908848e-01
-1.39729428e+00 -5.03245771e-01 5.85097730e-01 -6.65083647e-01
7.14307427e-01 2.99246162e-01 2.76602596e-01 1.30965853e+00
-8.79315734e-02 1.03671849e+00 1.16107523e+00 -5.41709840e-01
1.42659664e-01 6.34939015e-01 2.07041264e-01 5.46147525e-01
1.55697852e-01 -3.36145461e-01 -5.12437761e-01 1.94843374e-02
6.23743594e-01 -2.82612354e-01 -3.49833369e-01 -6.05786562e-01
-1.36127877e+00 7.58363128e-01 3.76141936e-01 3.77523392e-01
-3.02570194e-01 2.70053893e-01 4.44709957e-01 -3.16521786e-02
2.66373217e-01 4.32039201e-01 -2.34996393e-01 2.43748114e-01
-7.70183444e-01 2.53550738e-01 3.66443127e-01 1.08415639e+00
7.01945543e-01 -1.87948391e-01 -5.18793523e-01 8.37659419e-01
3.40514809e-01 5.79182208e-01 4.11297679e-01 -7.18154073e-01
7.16797888e-01 6.84122145e-01 -4.07058895e-02 -1.08741796e+00
-7.23303994e-03 -2.92074203e-01 -8.90272558e-01 -1.71953321e-01
5.09762645e-01 3.74040693e-01 -8.18412960e-01 2.10152698e+00
3.63584496e-02 -1.17978387e-01 9.36670378e-02 8.51898015e-01
8.14099789e-01 6.70453012e-01 3.95476520e-01 -2.92550951e-01
1.71332037e+00 -1.01510286e+00 -8.00691724e-01 -5.87667882e-01
2.83575594e-01 -8.91788602e-01 1.37191188e+00 1.16037630e-01
-1.16215074e+00 -6.01046979e-01 -7.40205228e-01 -2.65601873e-01
-2.93412834e-01 4.13082466e-02 4.28246766e-01 2.50953764e-01
-1.03294611e+00 1.83657125e-01 -3.11622351e-01 -3.66439879e-01
5.98828018e-01 1.90332159e-02 -3.79850388e-01 -2.86710888e-01
-9.44931805e-01 9.39395607e-01 1.32634282e-01 -1.27493083e-01
-7.64883220e-01 -5.83172500e-01 -8.10759962e-01 5.82606122e-02
4.90089178e-01 -7.40909278e-01 9.80186105e-01 -1.09380555e+00
-8.70893598e-01 1.26330829e+00 -3.99044991e-01 -4.79330659e-01
5.94707966e-01 -2.93487608e-01 -2.76719451e-01 4.20611799e-02
2.40898982e-01 8.34238052e-01 9.89225149e-01 -1.56331062e+00
-5.71035504e-01 -4.34883207e-01 1.82335258e-01 3.61835659e-01
-1.85233921e-01 5.67670576e-02 -8.80886912e-01 -7.22268999e-01
-1.30514726e-02 -9.07763124e-01 -4.28159498e-02 2.45498404e-01
-2.47228086e-01 -3.12402934e-01 4.94961828e-01 -5.14143586e-01
1.05455434e+00 -2.38338423e+00 2.86098897e-01 -1.38340607e-01
1.20091960e-01 3.77516359e-01 -3.67837936e-01 3.69397193e-01
-6.02308065e-02 7.56799728e-02 -1.24235407e-01 -5.20086169e-01
2.17655793e-01 2.82257646e-01 -6.62531197e-01 4.65322793e-01
1.41377479e-01 1.07232499e+00 -7.84624398e-01 -7.51020193e-01
3.10859412e-01 4.89830852e-01 -3.15286607e-01 2.15927020e-01
-4.73307908e-01 2.96278924e-01 -1.87952876e-01 2.19840467e-01
4.62377012e-01 -3.97105873e-01 8.80149826e-02 -6.03835344e-01
9.14390832e-02 1.70000449e-01 -1.00697589e+00 1.64696634e+00
-3.19276899e-01 7.92587519e-01 -3.15852426e-02 -8.82709861e-01
7.14848578e-01 2.55229861e-01 2.98393965e-01 -9.46918130e-01
1.39953077e-01 -5.44942468e-02 -1.22346908e-01 -6.96964383e-01
4.06284064e-01 -1.59757704e-01 5.50469896e-03 3.72201890e-01
-1.78468525e-02 -9.44368765e-02 2.62256444e-01 3.34995598e-01
6.96093976e-01 8.18627328e-02 1.02951296e-01 -2.92498052e-01
7.26129115e-01 2.94706551e-03 2.48025760e-01 7.64944553e-01
-2.55080462e-01 6.02129400e-01 5.16158879e-01 -2.94279784e-01
-1.13711214e+00 -1.03035998e+00 -1.36666372e-01 1.20235085e+00
3.84061575e-01 -3.71309012e-01 -7.38887668e-01 -4.23465461e-01
-1.17109641e-01 1.03234267e+00 -6.46338284e-01 -4.23385426e-02
-4.20992970e-01 -5.83886266e-01 4.74077314e-01 5.70053399e-01
5.05908608e-01 -1.06999862e+00 -4.73088533e-01 -1.67823926e-01
-4.79558557e-01 -1.56498551e+00 -5.22338390e-01 -1.18928596e-01
-6.18900418e-01 -7.95857549e-01 -4.84284163e-01 -7.65035331e-01
7.16499388e-01 4.52081233e-01 1.26922119e+00 8.94209668e-02
-2.24493295e-01 4.10625905e-01 -3.42941046e-01 -2.66514152e-01
-3.85017663e-01 -3.84816498e-01 -4.68233787e-02 2.57066399e-01
2.41445914e-01 -2.38971770e-01 -6.99874580e-01 9.55856293e-02
-1.21153307e+00 3.09711605e-01 7.40973353e-01 6.86713696e-01
5.78691483e-01 -1.03692427e-01 1.07379653e-01 -8.70980501e-01
3.24673623e-01 -1.45952091e-01 -4.20389563e-01 4.98576045e-01
-4.93211538e-01 3.37151438e-01 4.48247224e-01 -6.17666006e-01
-7.83930898e-01 -8.80921781e-02 1.15881709e-03 -5.53386986e-01
-2.02546984e-01 1.73523799e-01 -3.12884629e-01 3.09792191e-01
4.25501913e-01 3.71116459e-01 -2.24660616e-02 -2.49898583e-01
7.15808213e-01 4.65566814e-01 7.16692865e-01 -6.17380798e-01
7.96953738e-01 6.10872447e-01 -1.28843933e-01 -7.77501225e-01
-9.41825867e-01 -4.75992680e-01 -4.11496103e-01 2.95258090e-02
1.05943406e+00 -9.48472977e-01 -5.03642559e-01 1.14539355e-01
-1.23209536e+00 -1.03128873e-01 -2.85350263e-01 2.65120447e-01
-5.38016140e-01 5.25296092e-01 -3.50546837e-01 -7.27565289e-01
-1.90963641e-01 -1.35127997e+00 1.07358241e+00 5.42143732e-02
-1.76896065e-01 -8.35512400e-01 -2.42990673e-01 6.89980447e-01
4.14879501e-01 1.23794889e-02 1.38397479e+00 -4.56718773e-01
-5.55642962e-01 -5.04976287e-02 -7.82679737e-01 4.60140795e-01
-8.73711184e-02 -1.78804860e-01 -1.14525652e+00 -1.48131177e-01
4.17666659e-02 -1.95984006e-01 7.91260242e-01 2.62702465e-01
1.08545041e+00 -4.33427900e-01 -2.49373689e-01 2.22522408e-01
1.52486324e+00 -2.98480932e-02 7.69075572e-01 2.10397810e-01
8.06691885e-01 9.04100835e-01 4.19664174e-01 2.15759827e-03
4.24426794e-01 9.13106084e-01 6.24417007e-01 -8.56004208e-02
-3.70314091e-01 -3.34322691e-01 3.33655924e-01 6.42800450e-01
9.76344720e-02 -3.16122204e-01 -9.24047828e-01 7.42211640e-01
-1.91858101e+00 -9.63955164e-01 -1.84880689e-01 2.22985029e+00
7.85978317e-01 2.97555756e-02 -1.59602955e-01 -1.56655699e-01
7.63681710e-01 3.19679558e-01 -3.57413918e-01 -2.48248711e-01
-2.62929410e-01 1.04038964e-03 2.90821046e-01 4.09262389e-01
-1.09320033e+00 1.01112378e+00 5.40715075e+00 8.24534178e-01
-1.07804632e+00 1.34617686e-01 5.28633475e-01 -4.09245156e-02
-4.53498662e-01 9.05920491e-02 -7.88058698e-01 3.69050890e-01
6.20171726e-01 1.62876844e-02 4.05569404e-01 6.14200354e-01
9.56706610e-03 -1.51121438e-01 -1.17119455e+00 1.35714662e+00
5.17823815e-01 -1.29791212e+00 5.36997139e-01 7.04733655e-02
5.67851782e-01 -1.76501453e-01 2.02581689e-01 2.75245756e-01
3.99405025e-02 -9.45962906e-01 1.13698280e+00 3.34674984e-01
7.07005084e-01 -4.34825391e-01 6.25504851e-01 3.47015381e-01
-9.37286079e-01 2.62703397e-03 -2.98706681e-01 2.72283137e-01
7.74351135e-02 4.43513751e-01 -5.10537207e-01 2.79912114e-01
7.23551571e-01 3.84153932e-01 -8.99770200e-01 6.00383818e-01
-2.05404639e-01 2.99384415e-01 1.44667804e-01 -2.00000871e-02
3.14893365e-01 -7.39225671e-02 4.61442381e-01 1.20007634e+00
-1.70211978e-02 -9.64469910e-02 -4.83617671e-02 1.02510190e+00
-1.02484658e-01 1.97538674e-01 -4.65238363e-01 -2.11048037e-01
2.08316848e-01 1.06844378e+00 -5.90916872e-01 -3.19285631e-01
-5.74768186e-01 9.65697825e-01 2.71881729e-01 3.97408783e-01
-1.01673830e+00 1.18908510e-01 6.17357135e-01 2.31950670e-01
3.62770975e-01 -1.17543496e-01 -1.67972177e-01 -1.25340652e+00
1.88016772e-01 -9.65641320e-01 2.11686328e-01 -9.75340545e-01
-1.42404842e+00 6.89100862e-01 8.22392330e-02 -1.01481915e+00
-2.16196291e-02 -8.08909714e-01 -7.54415244e-02 6.46660209e-01
-1.40619504e+00 -1.31498682e+00 -3.03080648e-01 5.27181089e-01
6.98015094e-01 -9.61777847e-03 6.54754519e-01 3.13331485e-01
-4.91375953e-01 4.52239066e-01 -1.30379219e-02 1.15405731e-01
8.14860284e-01 -1.13395190e+00 2.52604514e-01 7.97190130e-01
5.89356780e-01 6.67769074e-01 1.04505253e+00 -2.15032995e-01
-1.48136747e+00 -1.03084099e+00 1.01737404e+00 -6.80754662e-01
7.96096385e-01 -4.94990796e-01 -1.09727418e+00 6.04855418e-01
5.07462502e-01 -2.48469636e-02 4.56993639e-01 -6.77505061e-02
-8.05157363e-01 -1.95660532e-01 -9.22163010e-01 9.66911554e-01
9.34124351e-01 -7.54047155e-01 -7.66665757e-01 3.11628699e-01
7.22179294e-01 -1.88146412e-01 -5.71079135e-01 3.64676774e-01
4.53202695e-01 -1.00831902e+00 1.16173339e+00 -4.98531908e-01
5.88190556e-01 -2.70555794e-01 -5.39547145e-01 -7.78363764e-01
-3.71755123e-01 -1.57259390e-01 7.51131251e-02 1.58571637e+00
2.27525145e-01 -3.13708901e-01 3.49429816e-01 6.86116755e-01
7.61081055e-02 -6.20293379e-01 -8.76344204e-01 -4.55413252e-01
1.01671271e-01 -5.93637049e-01 4.08735961e-01 8.51014197e-01
-3.25375080e-01 7.23066568e-01 -1.27683178e-01 1.74134076e-01
5.89543819e-01 1.08629569e-01 6.65033519e-01 -8.91477942e-01
-2.73135036e-01 -7.71537602e-01 -3.58884811e-01 -1.00467324e+00
1.86815172e-01 -8.09367359e-01 2.22300008e-01 -1.51140714e+00
5.50974965e-01 -3.12888473e-01 -1.41508624e-01 3.84751022e-01
-1.98433518e-01 2.57145166e-01 3.48528892e-01 4.26908851e-01
-7.92811275e-01 2.99234331e-01 1.12585056e+00 -3.17511678e-01
2.02895850e-01 -4.28427964e-01 -7.73508430e-01 7.23756135e-01
4.34177905e-01 -1.56022891e-01 -4.75608170e-01 -7.84944117e-01
3.84294808e-01 -1.40360996e-01 8.04060876e-01 -7.82106161e-01
7.05834553e-02 -2.28434533e-01 7.31406063e-02 -5.53598106e-01
4.07525599e-01 -8.68750870e-01 1.17246673e-01 1.31661028e-01
-6.18301213e-01 6.50857314e-02 3.25611264e-01 6.59458697e-01
-2.96519905e-01 -2.42356628e-01 9.07095671e-01 -2.00272992e-01
-8.62367153e-01 4.38103452e-02 -1.29437476e-01 2.01591045e-01
9.61696863e-01 -1.76849179e-02 -5.87319016e-01 -3.63076150e-01
-4.44094777e-01 1.16912894e-01 6.34496570e-01 6.02589726e-01
4.80824083e-01 -1.21354103e+00 -6.05874598e-01 5.14961146e-02
4.85251784e-01 -1.48786325e-02 3.49918157e-01 9.85511422e-01
-3.52751911e-01 5.78891218e-01 1.64514147e-02 -7.39126742e-01
-1.20463681e+00 8.20828319e-01 1.12163648e-01 -2.20479608e-01
-6.51016712e-01 8.41494501e-01 6.98846042e-01 5.04698381e-02
4.16770339e-01 -2.45502487e-01 -1.57690600e-01 1.16958775e-01
4.98857290e-01 2.45083179e-02 -1.32377028e-01 -1.02716935e+00
-3.82212281e-01 6.60151958e-01 -1.69737428e-01 -1.48245260e-01
1.10261369e+00 -4.06194538e-01 -3.05630922e-01 6.02401555e-01
1.09380627e+00 -1.22813947e-01 -1.03355634e+00 -5.22825837e-01
3.07005960e-02 -3.29917312e-01 -6.05653189e-02 -5.90380967e-01
-8.70057583e-01 1.09900558e+00 5.27805209e-01 1.87008098e-01
9.32499051e-01 4.38457459e-01 5.44898093e-01 7.60474429e-02
3.08535576e-01 -7.85473406e-01 2.37292826e-01 2.83316821e-01
1.00749409e+00 -1.24704158e+00 -3.75951640e-02 -3.68288904e-01
-9.72514272e-01 7.33067393e-01 4.85288471e-01 -3.09731364e-02
1.67489558e-01 -1.56090796e-01 1.47552282e-01 -1.73799902e-01
-8.33879352e-01 -4.28247213e-01 5.80460429e-01 4.77879167e-01
3.63507003e-01 -1.15826257e-01 -2.21312627e-01 2.95860767e-01
5.07487841e-02 -3.17180634e-01 1.73440561e-01 4.53968078e-01
-2.32355416e-01 -9.08124804e-01 -3.96348625e-01 2.08957404e-01
-3.22665066e-01 -2.44808152e-01 -3.77746701e-01 6.98091269e-01
4.10250396e-01 8.21676552e-01 1.65539041e-01 -3.76077145e-02
3.48960310e-01 1.83533221e-01 6.02728844e-01 -5.24866521e-01
-3.42576623e-01 1.17754914e-01 -8.05147588e-02 -5.14349043e-01
-4.83049899e-01 -7.08496392e-01 -1.01028931e+00 -1.89690292e-02
-1.41406253e-01 -1.47637613e-02 5.68629026e-01 1.13387275e+00
2.35409588e-01 3.19501877e-01 2.26982549e-01 -6.37958705e-01
-5.36919534e-01 -7.90853083e-01 -4.07263905e-01 1.00549304e+00
2.32162520e-01 -6.90493345e-01 -4.36039209e-01 3.24390411e-01]
|
[10.775211334228516, 1.651843786239624]
|
72b836fe-b63e-4951-bbf4-5d83b4f5cc96
|
convolutional-neural-network-architecture-for
|
1703.05593
| null |
http://arxiv.org/abs/1703.05593v2
|
http://arxiv.org/pdf/1703.05593v2.pdf
|
Convolutional neural network architecture for geometric matching
|
We address the problem of determining correspondences between two images in
agreement with a geometric model such as an affine or thin-plate spline
transformation, and estimating its parameters. The contributions of this work
are three-fold. First, we propose a convolutional neural network architecture
for geometric matching. The architecture is based on three main components that
mimic the standard steps of feature extraction, matching and simultaneous
inlier detection and model parameter estimation, while being trainable
end-to-end. Second, we demonstrate that the network parameters can be trained
from synthetically generated imagery without the need for manual annotation and
that our matching layer significantly increases generalization capabilities to
never seen before images. Finally, we show that the same model can perform both
instance-level and category-level matching giving state-of-the-art results on
the challenging Proposal Flow dataset.
|
['Relja Arandjelović', 'Ignacio Rocco', 'Josef Sivic']
|
2017-03-16
|
convolutional-neural-network-architecture-for-1
|
http://openaccess.thecvf.com/content_cvpr_2017/html/Rocco_Convolutional_Neural_Network_CVPR_2017_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2017/papers/Rocco_Convolutional_Neural_Network_CVPR_2017_paper.pdf
|
cvpr-2017-7
|
['geometric-matching']
|
['computer-vision']
|
[ 2.09408119e-01 -1.83065999e-02 4.95155230e-02 -4.98344362e-01
-8.44499409e-01 -6.36606932e-01 5.70235491e-01 -1.70145005e-01
-4.02251959e-01 3.18033010e-01 -9.50062945e-02 -1.08225979e-01
1.62087251e-02 -5.32849908e-01 -8.88626397e-01 -5.66722266e-02
-2.34014496e-01 6.27784669e-01 3.13743353e-01 -1.04019143e-01
4.46448147e-01 1.02480781e+00 -1.54438627e+00 4.90414687e-02
7.19029307e-01 1.14271522e+00 -2.33970001e-01 9.27154005e-01
3.62891793e-01 5.05891979e-01 -2.38520324e-01 -4.76673335e-01
8.80392492e-01 -5.98635375e-02 -9.55589771e-01 3.05127889e-01
1.48751867e+00 -8.00400674e-01 -5.63638270e-01 8.69414985e-01
3.51849675e-01 2.37705052e-01 6.98238730e-01 -1.30243909e+00
-2.22489268e-01 -4.27705795e-02 -4.50070858e-01 -1.75360441e-01
2.42224306e-01 3.99378300e-01 9.17734861e-01 -1.15519834e+00
7.11573720e-01 1.27521002e+00 1.23483884e+00 2.87618339e-01
-1.38752401e+00 -4.20387030e-01 -2.53447950e-01 5.93309700e-02
-1.28861439e+00 -7.25405157e-01 7.10899055e-01 -8.48254383e-01
8.56982768e-01 -6.68449104e-02 6.95886970e-01 4.51702267e-01
-6.17397949e-02 6.52700305e-01 5.19952178e-01 -4.76373464e-01
-1.49005130e-01 -2.45654643e-01 -1.86895400e-01 9.70192432e-01
3.61616910e-02 4.05855834e-01 -1.31158546e-01 -5.50049655e-02
1.27463222e+00 -2.46164441e-01 -3.51265371e-01 -8.37815821e-01
-1.42828953e+00 6.72627091e-01 7.76944399e-01 -4.95905913e-02
-1.87001660e-01 6.39380157e-01 4.32442039e-01 8.02351758e-02
2.73249269e-01 5.85869908e-01 -3.54508430e-01 8.86921119e-03
-1.25361347e+00 3.58698457e-01 6.06745720e-01 1.22410631e+00
9.52953696e-01 4.86144945e-02 2.19574481e-01 6.30166650e-01
3.17392260e-01 3.67978632e-01 1.04356289e-01 -1.48580730e+00
4.42294002e-01 3.03294063e-01 1.13032907e-01 -1.10660708e+00
-4.84233052e-01 -3.20291221e-01 -7.20894456e-01 6.70519352e-01
8.22589636e-01 -4.45602983e-02 -7.98202872e-01 1.52548563e+00
1.92547798e-01 5.46256781e-01 -1.41757593e-01 8.33876371e-01
7.44946241e-01 1.21885061e-01 -1.94503263e-01 5.36421120e-01
1.18805778e+00 -1.20955718e+00 -1.37512729e-01 -2.99774915e-01
6.11651778e-01 -9.91541862e-01 7.64918506e-01 2.76776049e-02
-1.38501549e+00 -8.86312723e-01 -1.22151661e+00 -3.31701547e-01
-1.38685599e-01 4.05687243e-01 6.41505480e-01 2.85734445e-01
-1.26773405e+00 9.23002958e-01 -7.67699897e-01 -3.09689492e-01
6.90527022e-01 4.32998568e-01 -6.49514735e-01 8.16171989e-02
-7.35521257e-01 8.75737727e-01 1.82864562e-01 1.92109570e-01
-6.61778688e-01 -1.17285848e+00 -1.17040527e+00 1.50193140e-01
-3.10930014e-02 -1.07874489e+00 1.39669859e+00 -9.89067018e-01
-1.33628392e+00 1.19634306e+00 -2.04872310e-01 -4.48759377e-01
1.10993469e+00 -1.65744200e-01 1.20915599e-01 3.23697209e-01
6.65838197e-02 1.26207185e+00 8.35719109e-01 -1.12648189e+00
-6.37827158e-01 -8.25185403e-02 -7.71939605e-02 -9.00590792e-02
1.24461353e-01 -1.59834489e-01 -4.22773987e-01 -6.55329287e-01
2.42745310e-01 -1.01147103e+00 -2.90524691e-01 7.86694765e-01
-3.45725626e-01 1.25045732e-01 8.62353742e-01 -6.42049849e-01
3.77210706e-01 -2.06438899e+00 -2.07858384e-01 2.52228439e-01
1.34628132e-01 4.27117616e-01 -2.49676302e-01 1.98640838e-01
-2.70318210e-01 -2.99301781e-02 -3.35565090e-01 -4.84478444e-01
7.87082836e-02 -2.78562814e-01 -5.05606353e-01 6.99669302e-01
4.87817228e-01 1.14155257e+00 -6.91185653e-01 -3.49617034e-01
6.09052598e-01 3.64078373e-01 -5.14454663e-01 3.53960931e-01
1.79662253e-03 4.86958027e-01 1.09409571e-01 4.17353362e-01
8.77968490e-01 -1.79319501e-01 -1.02456130e-01 -6.67387247e-01
9.61904321e-03 2.89916277e-01 -1.44817436e+00 1.92109072e+00
-4.53844279e-01 1.07520211e+00 -1.82754975e-02 -8.29702795e-01
9.33911145e-01 1.45786315e-01 5.54768980e-01 -3.63070607e-01
1.38322845e-01 5.17782450e-01 -4.33537327e-02 -3.28857362e-01
3.87478948e-01 1.89594254e-01 2.27565885e-01 3.09372813e-01
2.48933077e-01 -5.69345832e-01 1.90652132e-01 -5.46765476e-02
7.18284726e-01 4.48334366e-01 2.56909341e-01 -2.71302104e-01
6.94473386e-01 1.90590501e-01 4.22849238e-01 5.81282496e-01
-3.05010766e-01 1.04809594e+00 3.64975125e-01 -9.06318247e-01
-1.45232630e+00 -9.72584963e-01 -2.54145384e-01 3.57324392e-01
2.57124633e-01 -5.26449941e-02 -6.55839145e-01 -6.26444221e-01
2.47264892e-01 2.46688232e-01 -5.66169262e-01 1.65442675e-01
-1.00090611e+00 3.47161964e-02 7.21682727e-01 9.07104969e-01
8.67514253e-01 -8.43245804e-01 -7.64357209e-01 6.15022331e-02
-1.76334288e-02 -1.43915784e+00 -7.38091469e-01 -2.98514187e-01
-9.45439696e-01 -1.45746112e+00 -5.19812644e-01 -9.92622256e-01
9.02067602e-01 1.48107603e-01 1.47831118e+00 4.05104131e-01
-5.76616466e-01 3.25105727e-01 2.87686408e-01 -7.84712583e-02
-2.77772695e-01 1.20667301e-01 -2.45948628e-01 -4.33658138e-02
-8.06978568e-02 -7.04568803e-01 -8.67968917e-01 5.28749108e-01
-5.27383864e-01 1.96954757e-01 4.08281416e-01 7.68878162e-01
4.68257695e-01 -5.78777671e-01 3.01789373e-01 -4.68431026e-01
1.63777456e-01 3.46275419e-01 -9.52295661e-01 1.31302491e-01
-2.51871705e-01 -3.01586445e-02 5.88838398e-01 -2.03611434e-01
-7.42606819e-01 7.00891733e-01 -2.18101338e-01 -7.28859603e-01
-3.95151466e-01 -1.18145995e-01 1.11754261e-01 -7.42169976e-01
6.92649782e-01 -1.36636466e-01 1.58325508e-01 -1.84695169e-01
5.46667576e-01 2.78142720e-01 1.06913662e+00 -5.81294417e-01
1.14464164e+00 7.97345996e-01 4.00066704e-01 -4.86265361e-01
-6.34425044e-01 -5.18995821e-01 -1.31198668e+00 -1.80263996e-01
7.53034234e-01 -8.76097560e-01 -9.55960572e-01 6.26073182e-01
-1.43001080e+00 -4.77376014e-01 -4.47369784e-01 6.52690411e-01
-1.07839561e+00 3.97748649e-01 -5.80474257e-01 -3.45249414e-01
-3.32158625e-01 -1.27533245e+00 1.15403330e+00 1.28224701e-01
-1.91148952e-01 -1.23311186e+00 -3.47288959e-02 1.21496029e-01
4.43307430e-01 5.39110899e-01 5.69015324e-01 -4.58274156e-01
-9.99993861e-01 -3.67414176e-01 -5.70750237e-01 2.94556886e-01
-6.60115927e-02 2.70751774e-01 -1.07916951e+00 -3.58357221e-01
-4.36883658e-01 -5.03759980e-01 7.79811084e-01 4.14464802e-01
9.71621871e-01 -3.62099744e-02 -2.64489532e-01 1.26516581e+00
1.43319845e+00 -2.01080367e-01 7.09698260e-01 4.77111042e-01
8.45917046e-01 5.89220166e-01 4.23890620e-01 -1.01314522e-01
2.87683308e-01 8.61539900e-01 6.04156196e-01 -3.71102661e-01
-4.66246516e-01 -2.30284184e-01 -1.88245609e-01 8.20080489e-02
-1.77255867e-03 3.14172745e-01 -9.70134795e-01 7.56915987e-01
-1.88208115e+00 -9.90216732e-01 -3.58662635e-01 2.25469565e+00
3.78780633e-01 -5.27236387e-02 1.44873857e-01 -1.14885978e-01
6.38420999e-01 -7.09161684e-02 -4.72154707e-01 -2.00306877e-01
9.42844432e-03 9.58136842e-02 7.04841971e-01 8.13238084e-01
-1.36144161e+00 1.00727844e+00 6.97914124e+00 4.60131973e-01
-1.07057416e+00 -3.09356362e-01 6.51371539e-01 2.76098520e-01
9.47370902e-02 1.62699968e-01 -5.57127595e-01 -5.03258221e-02
2.38162383e-01 2.18814258e-02 3.35206509e-01 8.09609115e-01
1.54777244e-01 1.14140972e-01 -1.49151671e+00 1.08173573e+00
1.68136612e-01 -1.75254595e+00 -6.03928380e-02 -1.47598684e-01
7.00805366e-01 1.22406520e-01 -1.06070675e-01 -1.24325447e-01
2.83297688e-01 -1.06727028e+00 8.24566662e-01 6.23795986e-01
9.52584624e-01 -5.75889468e-01 6.16227686e-01 9.71911326e-02
-1.23083043e+00 2.16964453e-01 -3.11488450e-01 1.36344880e-01
3.28423470e-01 1.75787941e-01 -9.61628795e-01 5.34809291e-01
4.31272984e-01 9.14399087e-01 -6.79154515e-01 1.61757386e+00
-1.66044682e-01 1.58872917e-01 -2.45782778e-01 6.71319246e-01
2.06499249e-01 -2.29409248e-01 4.54293072e-01 1.35014522e+00
2.14331940e-01 -3.51938546e-01 2.64024347e-01 1.11600780e+00
-1.62572101e-01 -1.69928998e-01 -6.24592662e-01 5.95484436e-01
3.52665097e-01 1.39864016e+00 -5.83953202e-01 -3.83712947e-01
-1.53331712e-01 8.63426089e-01 4.17985469e-01 2.00057134e-01
-6.76376939e-01 -6.26290143e-01 7.47947752e-01 1.96645454e-01
4.10002708e-01 -3.53102088e-01 -5.15983462e-01 -1.14109421e+00
3.05520803e-01 -5.25366485e-01 9.70320776e-02 -8.14035952e-01
-1.24512064e+00 5.76692522e-01 -8.88518691e-02 -1.55036020e+00
-5.37790000e-01 -7.04164088e-01 -8.42776000e-01 9.13972795e-01
-1.69938755e+00 -1.39182377e+00 -7.00451016e-01 4.54460919e-01
3.90553534e-01 -4.97756014e-03 7.11126685e-01 3.94795299e-01
-9.90775824e-02 6.84117675e-01 -2.47979879e-01 7.42697954e-01
7.55339384e-01 -1.10823870e+00 9.12480712e-01 1.00841999e+00
4.91465479e-02 3.83452088e-01 3.75074983e-01 -2.67480224e-01
-9.84451830e-01 -1.21563816e+00 8.73311281e-01 -3.75279874e-01
5.03044248e-01 -3.06959808e-01 -8.78174663e-01 8.64125073e-01
1.78704038e-02 6.70130491e-01 -3.47225666e-02 -3.50394785e-01
-6.10287428e-01 -5.32426387e-02 -1.37402976e+00 4.94496822e-01
1.17660081e+00 -4.50323701e-01 -4.56886888e-01 4.38520610e-01
3.89051288e-01 -9.24920022e-01 -8.68661642e-01 4.70088482e-01
6.42146349e-01 -1.14661956e+00 1.31324363e+00 -6.55090213e-01
3.94176722e-01 -4.38347965e-01 -3.96809028e-03 -1.21367276e+00
-2.26239547e-01 -8.07139516e-01 1.66402921e-01 9.24327552e-01
3.46760869e-01 -4.75898981e-01 8.91734064e-01 7.57391632e-01
-3.10230702e-01 -5.99675953e-01 -9.50692713e-01 -7.77108014e-01
3.44683863e-02 -3.64634395e-01 5.69675565e-01 9.48924184e-01
-3.89325440e-01 5.99495992e-02 -1.76931098e-01 1.60646781e-01
7.86569357e-01 1.75384358e-01 1.30202353e+00 -1.34492564e+00
-1.41837552e-01 -6.05807364e-01 -8.54562819e-01 -1.40950203e+00
3.66939545e-01 -8.02944005e-01 8.53088424e-02 -1.44070685e+00
-2.47471839e-01 -7.04471529e-01 4.41560149e-01 3.95441204e-01
-4.21606898e-02 4.99911815e-01 3.36976081e-01 3.37926537e-01
-3.14155489e-01 3.19478780e-01 1.27922761e+00 -1.56603232e-01
-9.66278911e-02 1.16111031e-02 -1.45175025e-01 8.58787596e-01
6.06588304e-01 -2.38048524e-01 -1.93100527e-01 -5.81265688e-01
-1.53587356e-01 3.70269194e-02 8.58604312e-01 -1.19941962e+00
3.61209333e-01 5.90000786e-02 5.59160709e-01 -5.63826680e-01
3.31882805e-01 -9.29844081e-01 -5.17873652e-02 4.69840258e-01
-4.38177735e-01 3.16760272e-01 4.57883269e-01 2.09741354e-01
-1.33309752e-01 -3.00569326e-01 1.15364671e+00 -3.31767509e-03
-7.39281058e-01 5.51887989e-01 1.64288044e-01 2.17986107e-01
9.80703473e-01 -4.84217286e-01 -3.32064748e-01 -3.95327717e-01
-4.45167780e-01 1.44729659e-01 6.02398038e-01 3.76019090e-01
6.30364776e-01 -1.50567102e+00 -9.19886410e-01 3.79567891e-01
1.53646648e-01 2.95007557e-01 2.93367524e-02 7.77701378e-01
-1.08945048e+00 3.36340129e-01 -3.77739817e-01 -9.91341412e-01
-1.07131815e+00 3.14102113e-01 9.92578268e-01 -7.84685984e-02
-7.63354719e-01 6.35313630e-01 5.10705262e-02 -7.32225955e-01
2.15477705e-01 -2.44614154e-01 2.18142807e-01 -3.49225581e-01
4.25644338e-01 3.26417297e-01 2.95739114e-01 -8.26277494e-01
-3.40875179e-01 1.03844798e+00 1.70046389e-01 -1.60364941e-01
1.19971955e+00 1.32406637e-01 6.39645159e-02 -7.66132176e-02
1.49847543e+00 -1.82258382e-01 -1.70223355e+00 -1.31253734e-01
-2.08034679e-01 -9.36436832e-01 -1.26188844e-01 -4.68790650e-01
-1.29366517e+00 1.07097173e+00 5.66773891e-01 -1.62576243e-01
6.96310878e-01 -3.86770010e-01 7.57482409e-01 4.72657561e-01
2.34694742e-02 -8.38907182e-01 -6.99938536e-02 5.11767447e-01
9.13453043e-01 -1.34688342e+00 -5.77298813e-02 -6.16251111e-01
-1.84418529e-01 1.61073387e+00 6.98721588e-01 -6.63855910e-01
5.74920714e-01 2.01310307e-01 2.90594488e-01 -1.33541688e-01
-3.76275927e-01 -1.81085363e-01 6.38134778e-01 8.30077946e-01
4.09952164e-01 -3.48555475e-01 2.82664627e-01 -4.98409927e-01
-3.08081150e-01 8.55657384e-02 5.42491496e-01 6.32821381e-01
-1.44769996e-01 -9.23597872e-01 -2.96707839e-01 1.32513240e-01
-1.56390101e-01 -4.96738181e-02 -1.83771819e-01 1.04092860e+00
-1.18500590e-01 6.88445210e-01 5.14987051e-01 -1.94762632e-01
7.06836879e-01 -1.36278629e-01 6.36943400e-01 -2.06125885e-01
-7.48645961e-01 -2.47692421e-01 6.44892827e-03 -7.83352494e-01
-3.41738790e-01 -6.91221654e-01 -1.10450733e+00 -3.64761800e-01
-2.66003668e-01 -2.22296119e-01 6.18204117e-01 8.05920839e-01
5.47275603e-01 2.67194360e-01 5.95320821e-01 -1.35306776e+00
-5.73075294e-01 -7.15795398e-01 -2.69213971e-03 7.42653370e-01
6.04602873e-01 -5.27892947e-01 -2.74858743e-01 2.36079782e-01]
|
[8.533690452575684, -2.093501567840576]
|
ccecc160-66c4-463b-9aed-80d7b3fba5fd
|
indoor-scene-recognition-in-3d
|
2002.12819
| null |
https://arxiv.org/abs/2002.12819v2
|
https://arxiv.org/pdf/2002.12819v2.pdf
|
Indoor Scene Recognition in 3D
|
Recognising in what type of environment one is located is an important perception task. For instance, for a robot operating in indoors it is helpful to be aware whether it is in a kitchen, a hallway or a bedroom. Existing approaches attempt to classify the scene based on 2D images or 2.5D range images. Here, we study scene recognition from 3D point cloud (or voxel) data, and show that it greatly outperforms methods based on 2D birds-eye views. Moreover, we advocate multi-task learning as a way of improving scene recognition, building on the fact that the scene type is highly correlated with the objects in the scene, and therefore with its semantic segmentation into different object classes. In a series of ablation studies, we show that successful scene recognition is not just the recognition of individual objects unique to some scene type (such as a bathtub), but depends on several different cues, including coarse 3D geometry, colour, and the (implicit) distribution of object categories. Moreover, we demonstrate that surprisingly sparse 3D data is sufficient to classify indoor scenes with good accuracy.
|
['Mikhail Usvyatsov', 'Shengyu Huang', 'Konrad Schindler']
|
2020-02-28
| null | null | null | null |
['scene-recognition']
|
['computer-vision']
|
[ 5.03400922e-01 -4.28548068e-01 1.65576816e-01 -4.50405985e-01
-2.47062579e-01 -7.24224567e-01 5.18415034e-01 6.06896460e-01
-4.04047459e-01 2.10054427e-01 -1.37336493e-01 -2.38279641e-01
-3.42958897e-01 -7.12709665e-01 -6.13735497e-01 -8.14669847e-01
1.33782759e-01 7.57141948e-01 1.87711298e-01 -9.16100740e-02
6.02902830e-01 8.17359149e-01 -1.67731142e+00 2.90940166e-01
4.03931677e-01 1.10099435e+00 7.27960408e-01 6.04137599e-01
-2.05989420e-01 3.83222699e-01 -4.99697655e-01 2.51834989e-01
3.40627313e-01 -2.59513348e-01 -7.68485844e-01 4.39961165e-01
6.48824811e-01 3.11973877e-02 2.71222323e-01 9.23801482e-01
3.60020474e-02 2.56750077e-01 1.00286794e+00 -9.63381767e-01
-3.38358462e-01 -2.73162544e-01 -3.95027786e-01 1.14280216e-01
5.17781436e-01 -1.60365194e-01 8.96786869e-01 -7.27558553e-01
2.93218136e-01 9.40841138e-01 4.68263000e-01 5.50968498e-02
-1.32577312e+00 -1.61669418e-01 4.44520324e-01 2.99977183e-01
-1.36140990e+00 -3.92739385e-01 1.00699878e+00 -6.88798130e-01
7.24826872e-01 4.37678277e-01 5.12577772e-01 8.09487343e-01
1.80093467e-01 6.21070743e-01 1.50637019e+00 -5.73151708e-01
6.02670670e-01 2.95410007e-01 9.53904390e-02 3.90364051e-01
1.25179768e-01 -1.41742617e-01 -3.34481061e-01 1.99589774e-01
8.62696767e-01 2.86576420e-01 -2.02328071e-01 -1.01439953e+00
-1.24480820e+00 5.16245425e-01 7.07658350e-01 5.55692255e-01
-3.94951165e-01 -1.44292146e-01 8.58795270e-02 1.25022531e-01
3.39412838e-01 5.52212536e-01 -4.93568987e-01 1.01008400e-01
-6.72673285e-01 -7.20547363e-02 6.30287349e-01 7.25886285e-01
1.04178786e+00 -3.63592803e-01 4.64444935e-01 7.66484261e-01
3.37939143e-01 4.97019708e-01 3.29698503e-01 -8.70680749e-01
4.11933698e-02 7.02759981e-01 1.17271682e-02 -1.05910063e+00
-5.89581251e-01 -2.23963037e-01 -7.18536556e-01 5.06217539e-01
6.13727629e-01 4.55663592e-01 -1.08714676e+00 1.25570965e+00
3.81934553e-01 -1.41183496e-01 -7.82402158e-02 1.01276124e+00
7.48212636e-01 2.57476628e-01 -1.49084553e-01 1.38804063e-01
1.38872111e+00 -5.71640193e-01 -7.58237317e-02 -7.40947187e-01
5.24231017e-01 -6.39415979e-01 8.19212615e-01 5.06657541e-01
-3.89883935e-01 -7.91318476e-01 -7.69242406e-01 1.15890794e-01
-6.81696773e-01 7.27646947e-02 5.86662233e-01 7.37237632e-01
-7.72843301e-01 3.86143684e-01 -4.62334782e-01 -8.05069923e-01
2.74869621e-01 1.99571714e-01 -7.25413561e-01 -3.61910135e-01
-3.01849335e-01 9.80165541e-01 4.56027746e-01 -1.44650210e-02
-7.77652204e-01 -2.79192746e-01 -1.06042230e+00 -1.78223833e-01
3.75287741e-01 -5.83085656e-01 7.67480612e-01 -8.85720968e-01
-1.08501625e+00 1.22607911e+00 -3.00981343e-01 -3.12780216e-02
2.42982805e-01 -9.87825915e-02 -8.10234547e-02 1.88843176e-01
2.07412452e-01 3.27334106e-01 1.02041137e+00 -1.83821869e+00
-6.76617622e-01 -9.37894583e-01 2.61114419e-01 5.67993879e-01
2.62937874e-01 -2.41276145e-01 -3.08951996e-02 -3.76869775e-02
6.78077936e-01 -8.46085250e-01 -4.09227997e-01 -2.66617723e-02
-4.56503749e-01 -7.07292482e-02 8.29588234e-01 -3.86421114e-01
2.62404501e-01 -2.41291666e+00 1.93219051e-01 3.37918788e-01
7.59343877e-02 -1.91648394e-01 2.46283084e-01 2.69669175e-01
-6.39133826e-02 -9.42776054e-02 -3.76829445e-01 -2.31506988e-01
-7.82738924e-02 4.18566197e-01 -1.13714918e-01 7.35830307e-01
-8.28832537e-02 4.30302143e-01 -9.04382110e-01 -2.39078477e-01
7.56413341e-01 3.27672392e-01 -2.27842659e-01 -3.34557937e-03
2.29227450e-02 9.40117061e-01 -4.87779915e-01 4.59800899e-01
5.37674427e-01 -3.52763176e-01 7.51067176e-02 -1.11482359e-01
-2.53832251e-01 1.45250276e-01 -1.34080791e+00 1.78659987e+00
-6.77578449e-01 5.26778817e-01 2.04588532e-01 -1.27183521e+00
1.09000206e+00 2.46436913e-02 4.14529204e-01 -7.63043225e-01
-2.57762745e-02 2.52734512e-01 -9.58791226e-02 -4.06906188e-01
3.91062796e-01 -1.24255449e-01 -7.18459040e-02 2.87367940e-01
-7.11051300e-02 -6.91935062e-01 -6.07541427e-02 -2.66643018e-01
8.56574714e-01 1.46865189e-01 5.32369077e-01 -2.75840163e-01
2.78693318e-01 8.21531862e-02 1.88439175e-01 7.44744599e-01
-2.14990258e-01 8.24916303e-01 1.04911625e-01 -4.83487904e-01
-1.02331054e+00 -1.02804863e+00 -4.41546142e-01 9.85933959e-01
6.10456169e-01 5.19379973e-03 -2.37690836e-01 -4.32021081e-01
3.09376270e-02 7.34775364e-01 -6.39938831e-01 -5.94582548e-03
-2.73702502e-01 -4.84876454e-01 -2.27511749e-01 3.74550283e-01
4.48373467e-01 -8.62411141e-01 -1.12008142e+00 -2.16125861e-01
-4.44523990e-02 -1.11749613e+00 1.22641854e-01 7.20572770e-01
-7.70162702e-01 -1.24798918e+00 -5.14664531e-01 -7.36606956e-01
8.86245787e-01 7.78795123e-01 9.72043216e-01 -1.26897082e-01
-2.32545763e-01 8.61287296e-01 -5.55517554e-01 -2.72741228e-01
-3.27124074e-02 -2.51937568e-01 1.31040022e-01 1.98146462e-01
2.36921266e-01 -7.06077278e-01 -3.99467945e-01 4.06043231e-01
-4.81709003e-01 4.07369137e-02 6.01432920e-01 4.37230051e-01
6.79144502e-01 4.96265620e-01 -9.32668447e-02 -8.89412642e-01
-3.83784659e-02 -3.57090414e-01 -2.95445770e-01 1.79035634e-01
-1.73221335e-01 -1.99574843e-01 4.55793619e-01 5.37268445e-02
-8.80924404e-01 5.65695524e-01 3.36839864e-03 -2.22129077e-01
-1.23302686e+00 1.82115451e-01 -1.52926609e-01 -1.74893856e-01
6.38803065e-01 4.24764752e-01 -2.39404514e-01 -4.39929456e-01
1.70237750e-01 5.13781488e-01 3.39349955e-01 -4.49856132e-01
6.57102466e-01 7.23401189e-01 1.54575363e-01 -1.30896771e+00
-8.84060621e-01 -1.05051696e+00 -1.31704998e+00 -3.07668477e-01
1.00584781e+00 -8.04783881e-01 -5.82612336e-01 3.30287039e-01
-9.64694858e-01 -5.14648736e-01 -1.69544175e-01 5.55564761e-01
-6.59005165e-01 2.45746195e-01 1.23150624e-01 -1.03215837e+00
4.54089910e-01 -1.02236164e+00 1.44372690e+00 2.71685086e-02
-2.64163256e-01 -1.09466958e+00 -8.54955167e-02 6.08563006e-01
4.68602106e-02 2.89543062e-01 1.02044773e+00 -6.67589843e-01
-5.65933585e-01 -8.70101526e-02 -3.33214521e-01 1.35008812e-01
4.58806306e-01 -2.90818691e-01 -1.23430264e+00 -1.44871145e-01
3.40392530e-01 -1.85862660e-01 8.05032611e-01 4.87375677e-01
1.09617746e+00 2.22903863e-01 -3.56442183e-01 6.35570109e-01
1.58587253e+00 4.00777966e-01 3.00521106e-01 3.66001487e-01
9.37453151e-01 9.46352005e-01 4.83738899e-01 1.73602700e-01
3.89934331e-01 8.01798284e-01 7.22072005e-01 -6.29082248e-02
3.34858485e-02 5.52493036e-02 -4.98230234e-02 3.96706164e-01
-1.72809243e-01 6.87474981e-02 -1.14092827e+00 3.71324569e-01
-1.60489452e+00 -7.50748277e-01 -3.38758975e-01 2.35294652e+00
1.81395620e-01 1.50372744e-01 -2.11418197e-02 3.47343653e-01
4.48276550e-01 9.12384838e-02 -5.94357789e-01 -2.25610480e-01
-1.28410965e-01 -7.17249289e-02 5.17902434e-01 2.92861462e-01
-1.27570868e+00 5.14263272e-01 6.16145849e+00 5.27657270e-01
-1.03856552e+00 -2.36982062e-01 5.99228501e-01 3.85204107e-01
3.04445182e-03 -6.95279986e-02 -6.47198260e-01 2.56101370e-01
3.59634936e-01 2.20245004e-01 2.87030965e-01 1.06311774e+00
-8.27180296e-02 -6.65760338e-01 -1.37094760e+00 1.21038818e+00
4.40657169e-01 -8.74142885e-01 -2.06550434e-01 1.33609861e-01
3.89493376e-01 -1.42706111e-01 7.33285546e-02 -1.13469854e-01
2.23669350e-01 -1.05063331e+00 8.61367464e-01 3.13069940e-01
4.40311104e-01 -3.56605947e-01 5.82007468e-01 7.46613383e-01
-1.15570915e+00 -2.10869849e-01 -2.27952659e-01 -4.04400527e-01
-6.97262734e-02 4.78197128e-01 -1.02335906e+00 4.89622980e-01
9.57684338e-01 8.24836850e-01 -8.04383337e-01 1.26540720e+00
-2.92775542e-01 3.81951369e-02 -4.63057429e-01 -5.02729602e-02
9.99096781e-02 -3.27527255e-01 4.14939791e-01 1.00210035e+00
2.64069438e-01 2.47723848e-01 4.70248580e-01 7.60099947e-01
4.24124807e-01 -1.12392314e-01 -9.52061296e-01 4.61881727e-01
4.81671654e-02 1.23254991e+00 -1.18120766e+00 -4.37654182e-02
-2.84238160e-01 1.10492742e+00 1.11245528e-01 4.06939626e-01
-1.41576424e-01 -2.17002735e-01 4.93529588e-01 -6.59066290e-02
5.20902097e-01 -7.81318307e-01 -5.07807255e-01 -8.11006248e-01
3.04539185e-02 -3.20487738e-01 7.10660443e-02 -1.04475760e+00
-1.14184654e+00 3.42351526e-01 6.30746856e-02 -1.17425084e+00
-6.24818765e-02 -9.35373187e-01 -3.84009689e-01 7.90073156e-01
-1.42591894e+00 -1.11059141e+00 -5.06149352e-01 4.95626092e-01
5.41301668e-01 2.30465978e-01 9.74838853e-01 2.82326546e-02
-1.78620517e-01 -1.94289163e-01 2.39722446e-01 1.11381225e-01
3.39396626e-01 -1.56457913e+00 1.73648745e-02 5.45301914e-01
6.82097435e-01 3.89721751e-01 7.46887147e-01 -2.83481479e-01
-1.28491402e+00 -8.18623900e-01 6.59920871e-01 -8.27143729e-01
3.16030115e-01 -4.91569072e-01 -8.01591098e-01 4.32759970e-01
-2.78994858e-01 -7.23142251e-02 7.72356808e-01 5.04178941e-01
-3.31083715e-01 -9.39800888e-02 -1.14074457e+00 1.87470093e-01
1.00464261e+00 -7.18693495e-01 -6.89642668e-01 5.90514243e-01
2.56519467e-01 -3.80454063e-01 -5.25636911e-01 3.74698073e-01
2.29308918e-01 -1.25332916e+00 1.12869692e+00 -2.71414220e-01
2.76700824e-01 -4.43710864e-01 -5.26608884e-01 -1.50447249e+00
-5.21142364e-01 3.34817201e-01 4.48172897e-01 7.35518038e-01
2.12149128e-01 -4.33408737e-01 9.17932928e-01 4.84768778e-01
-2.69570798e-01 -2.09624931e-01 -1.04220831e+00 -7.37997353e-01
-2.74394512e-01 -5.79484701e-01 1.98224902e-01 8.33792806e-01
-4.63741809e-01 5.20163357e-01 1.17611349e-01 4.85715896e-01
6.63487315e-01 6.92201614e-01 9.32640016e-01 -1.38009524e+00
-1.86894789e-01 -4.28788513e-01 -8.12437594e-01 -1.21657515e+00
-3.05736642e-02 -7.78835654e-01 3.91189367e-01 -1.80904460e+00
9.08089206e-02 -8.69402826e-01 -2.32486010e-01 2.44679496e-01
1.25889435e-01 1.95810020e-01 1.82643890e-01 3.14281046e-01
-8.19376826e-01 3.21406931e-01 1.07964587e+00 -1.54717535e-01
-1.04989097e-01 3.12030882e-01 -3.35500866e-01 8.31430852e-01
6.18050337e-01 -5.19252755e-02 -2.04872012e-01 -5.34163058e-01
9.21877399e-02 -1.97812796e-01 7.98773766e-01 -1.30575204e+00
3.56394440e-01 -1.69768333e-01 6.17697954e-01 -6.37081444e-01
8.39214146e-01 -1.27911782e+00 2.16633156e-02 2.16818348e-01
-3.17399353e-02 -3.95374179e-01 1.51941583e-01 6.85318053e-01
-8.45107660e-02 -3.78533304e-01 6.52435958e-01 -5.97935319e-01
-1.28939700e+00 -9.96563584e-02 -3.70665133e-01 -3.91492128e-01
1.00063229e+00 -7.63109505e-01 1.14396950e-02 -1.09976090e-01
-7.72542000e-01 -2.78379545e-02 7.23804951e-01 2.49810874e-01
7.54153550e-01 -9.84364092e-01 -3.28313053e-01 4.31471378e-01
3.79789501e-01 3.28580052e-01 2.65397221e-01 6.07232988e-01
-4.11504656e-01 5.03218770e-01 -3.27997208e-01 -1.10876143e+00
-1.38170397e+00 5.35571873e-01 2.90209800e-01 3.14520359e-01
-5.18190503e-01 8.91318321e-01 6.21558249e-01 -7.33317316e-01
1.65547043e-01 -4.52899605e-01 -3.89547259e-01 6.92399591e-02
2.02701464e-01 1.20972032e-02 1.79761067e-01 -9.65311944e-01
-5.14395535e-01 1.03947699e+00 2.39643678e-01 2.47358784e-01
1.26421607e+00 -2.23727301e-01 -1.57433227e-01 1.01483428e+00
1.01378489e+00 -1.87104747e-01 -1.16985643e+00 -3.31015557e-01
-8.21353421e-02 -7.65742660e-01 -3.98767851e-02 -6.14464819e-01
-6.33236468e-01 9.47824359e-01 6.15432620e-01 5.56252003e-01
1.16160440e+00 3.12003374e-01 2.10799277e-01 5.34129560e-01
9.28526223e-01 -8.57463062e-01 1.71901792e-01 5.47955692e-01
6.69059277e-01 -1.63096476e+00 1.23983525e-01 -4.52853471e-01
-5.03639281e-01 1.13924444e+00 3.34012479e-01 -4.40689623e-02
6.66887164e-01 -2.46332273e-01 2.15965342e-02 -4.72423315e-01
-1.35972559e-01 -4.46765304e-01 3.51387322e-01 8.23517144e-01
1.38565123e-01 2.69606650e-01 7.42821693e-01 -1.10461414e-01
-3.44666928e-01 -6.21521831e-01 2.32103020e-01 9.35266018e-01
-8.01640391e-01 -7.32099831e-01 -5.30369043e-01 4.69255924e-01
9.36428830e-02 1.86203480e-01 -5.26655912e-01 5.32773852e-01
3.43463391e-01 9.74727094e-01 4.49265331e-01 -4.02396113e-01
3.51051241e-01 -3.16472575e-02 7.55048394e-01 -1.09618294e+00
-1.86054632e-01 -1.83550462e-01 -1.57396123e-01 -4.01886404e-01
-6.85350478e-01 -8.46054792e-01 -1.01293504e+00 1.03416659e-01
-7.54093826e-02 2.85117608e-02 1.01025331e+00 1.06831765e+00
8.26046802e-03 3.79183888e-01 6.83458626e-01 -1.17613399e+00
2.60167390e-01 -4.73807573e-01 -8.92990589e-01 4.35749441e-01
6.25411689e-01 -7.61956573e-01 -5.12781084e-01 2.57820070e-01]
|
[8.203076362609863, -2.3267805576324463]
|
b4b41153-4f8c-4b22-b65a-0f92f4e7332f
|
beyond-planar-symmetry-modeling-human
|
1704.03568
| null |
http://arxiv.org/abs/1704.03568v2
|
http://arxiv.org/pdf/1704.03568v2.pdf
|
Beyond Planar Symmetry: Modeling human perception of reflection and rotation symmetries in the wild
|
Humans take advantage of real world symmetries for various tasks, yet
capturing their superb symmetry perception mechanism with a computational model
remains elusive. Motivated by a new study demonstrating the extremely high
inter-person accuracy of human perceived symmetries in the wild, we have
constructed the first deep-learning neural network for reflection and rotation
symmetry detection (Sym-NET), trained on photos from MS-COCO (Microsoft-Common
Object in COntext) dataset with nearly 11K consistent symmetry-labels from more
than 400 human observers. We employ novel methods to convert discrete human
labels into symmetry heatmaps, capture symmetry densely in an image and
quantitatively evaluate Sym-NET against multiple existing computer vision
algorithms. On CVPR 2013 symmetry competition testsets and unseen MS-COCO
photos, Sym-NET significantly outperforms all other competitors. Beyond
mathematically well-defined symmetries on a plane, Sym-NET demonstrates
abilities to identify viewpoint-varied 3D symmetries, partially occluded
symmetrical objects, and symmetries at a semantic level.
|
['Yanxi Liu', 'Christopher Funk']
|
2017-04-11
|
beyond-planar-symmetry-modeling-human-1
|
http://openaccess.thecvf.com/content_iccv_2017/html/Funk_Beyond_Planar_Symmetry_ICCV_2017_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2017/papers/Funk_Beyond_Planar_Symmetry_ICCV_2017_paper.pdf
|
iccv-2017-10
|
['symmetry-detection']
|
['computer-vision']
|
[ 1.84539735e-01 6.02318831e-02 1.61286205e-01 -5.90858042e-01
-4.09789205e-01 -8.32791269e-01 9.42098796e-01 -4.08480376e-01
6.17108569e-02 -7.60763586e-02 5.75882971e-01 1.85218096e-01
-3.60410601e-01 -4.13636714e-01 -6.23338997e-01 -1.61721393e-01
-7.14952499e-03 9.05530751e-01 2.91876607e-02 -2.09549397e-01
6.12877190e-01 8.31100523e-01 -1.55081248e+00 7.50478625e-01
2.64385343e-01 9.32541072e-01 -5.28377891e-01 2.37481028e-01
6.47125721e-01 5.05698562e-01 -2.44688392e-01 -6.64890289e-01
8.13749969e-01 -3.70035380e-01 -1.01804566e+00 3.40952843e-01
1.55872226e+00 -1.04554869e-01 -4.30770248e-01 1.00573623e+00
2.52997369e-01 8.91799927e-02 8.85742486e-01 -1.64694583e+00
-1.00231361e+00 2.04305634e-01 -4.88618046e-01 4.86681145e-03
5.45951664e-01 1.69212166e-02 1.35226798e+00 -8.59797359e-01
1.22397768e+00 1.41948509e+00 1.10462415e+00 3.07175159e-01
-1.22763503e+00 -6.41877294e-01 -2.50177979e-01 4.80679393e-01
-1.47032034e+00 -3.73172462e-01 7.74712026e-01 -4.94183838e-01
1.27772665e+00 3.21867466e-01 6.22048855e-01 1.41424596e+00
7.91586414e-02 3.69941205e-01 1.30126679e+00 -3.41786183e-02
-2.27958299e-02 -5.81207931e-01 -4.07006949e-01 7.54018784e-01
5.20283356e-02 -9.99743938e-02 -7.60371983e-01 7.67710432e-02
1.13356054e+00 -2.77194381e-02 -1.91035330e-01 -1.01790249e+00
-1.58352029e+00 4.25579280e-01 1.02106309e+00 -8.29121172e-02
-1.42910466e-01 -1.47383228e-01 5.90508759e-01 3.01236749e-01
5.04724979e-02 1.16113818e+00 -3.32738370e-01 -2.29258556e-02
-6.72378480e-01 7.29407370e-01 4.22691315e-01 9.68596637e-01
3.60934168e-01 -2.44672075e-01 8.88238475e-02 7.92947292e-01
-3.46361011e-01 1.01489276e-01 2.02162161e-01 -1.44673860e+00
4.12884057e-01 7.87202716e-01 -1.34067044e-01 -1.19759083e+00
-8.24517488e-01 -4.92605776e-01 -1.17183232e+00 2.59825557e-01
5.48395097e-01 7.90153623e-01 -7.92586029e-01 1.82998395e+00
-5.62212989e-02 -4.85506743e-01 -2.75449812e-01 1.07909906e+00
6.35491431e-01 1.55546635e-01 -5.45609891e-01 7.74874866e-01
1.52348602e+00 -9.06421900e-01 1.48738459e-01 -6.46033958e-02
4.21739459e-01 -8.82011056e-01 9.22050595e-01 6.22522354e-01
-9.36082244e-01 -5.46208143e-01 -9.00791049e-01 -6.17333293e-01
-3.49037826e-01 3.00797343e-01 7.95881808e-01 3.02562624e-01
-1.22462261e+00 7.95579910e-01 -4.79548723e-01 -7.68616021e-01
7.22964644e-01 2.03350663e-01 -1.05371594e+00 -4.54649702e-02
-6.62050486e-01 8.95059109e-01 9.11256149e-02 -1.66606233e-01
-5.98177493e-01 -8.78776073e-01 -1.07082343e+00 1.87414005e-01
1.67779088e-01 -9.69396710e-01 1.22542155e+00 -9.47157502e-01
-1.04213691e+00 1.66133988e+00 -7.58028552e-02 -3.94943535e-01
9.57164228e-01 6.43438101e-02 -2.67553538e-01 2.75621474e-01
3.53970587e-01 1.32473373e+00 6.42390609e-01 -1.03271258e+00
-1.18316174e-01 -4.12848979e-01 1.11721516e-01 2.78794110e-01
2.44844601e-01 2.13286161e-01 -2.67955065e-02 -5.23102939e-01
9.13753808e-01 -1.09163618e+00 2.80709952e-01 7.22196639e-01
-7.64529526e-01 -5.28214037e-01 7.51677632e-01 -6.50105357e-01
4.70670424e-02 -1.99390137e+00 4.03072760e-02 2.51562655e-01
5.51612258e-01 -7.08332360e-02 -2.92941898e-01 5.34261286e-01
-6.59503102e-01 4.72506247e-02 7.72243515e-02 -2.96678424e-01
4.37196940e-01 -2.19053149e-01 -4.14355636e-01 8.28897297e-01
2.26489574e-01 1.01123714e+00 -5.22852838e-01 -6.68531284e-02
4.14498061e-01 4.94249046e-01 -9.72245693e-01 -1.17720455e-01
1.72375396e-01 2.48592585e-01 3.29990029e-01 8.72917116e-01
8.75363231e-01 -5.67814291e-01 -1.53939612e-02 -4.53752130e-01
-5.82333952e-02 4.09499675e-01 -9.71596122e-01 1.68159628e+00
-1.90016210e-01 9.86712754e-01 -4.23207194e-01 -9.88602579e-01
8.19937825e-01 -6.60612658e-02 4.04323637e-01 -1.01661658e+00
2.03696135e-02 5.90597056e-02 2.49373838e-01 -8.50344300e-02
4.94996428e-01 1.51591897e-01 -1.51067153e-01 4.06082511e-01
2.99221933e-01 -3.11965287e-01 7.74160624e-02 -5.31460233e-02
8.58303428e-01 1.68619439e-01 4.83148843e-01 -4.14387196e-01
3.26322407e-01 -2.04784378e-01 4.33630288e-01 7.31404543e-01
-4.17629391e-01 1.42173529e+00 7.73279071e-01 -1.23358595e+00
-1.69001842e+00 -1.62624860e+00 -2.41615906e-01 9.83897984e-01
1.08871683e-01 -6.23324811e-01 -6.12834513e-01 -2.18338415e-01
7.22554177e-02 1.69628561e-01 -8.08476448e-01 -3.51667777e-02
-5.49733818e-01 -8.06457475e-02 6.42130435e-01 5.90408504e-01
1.00969291e+00 -9.60612595e-01 -4.96481389e-01 -5.53165615e-01
-2.98458725e-01 -1.48793626e+00 -5.23650289e-01 -3.42240632e-01
-3.38713378e-01 -1.32038307e+00 -7.75768816e-01 -8.00373197e-01
6.51904166e-01 4.08083707e-01 1.22984231e+00 -3.95321786e-01
-8.48074317e-01 1.34099886e-01 1.17170237e-01 -3.58023606e-02
-2.03104645e-01 -1.79358587e-01 2.49305665e-01 -2.30858043e-01
2.69690156e-01 -9.17078078e-01 -4.98236716e-01 7.57295609e-01
-1.44667685e-01 4.06180680e-01 3.53846878e-01 6.11013889e-01
3.07205379e-01 -3.14136222e-02 -8.45246464e-02 -1.56920061e-01
1.32187173e-01 2.11258009e-01 -8.00472796e-01 2.24137336e-01
1.11012101e-01 -1.19283363e-01 4.85517144e-01 1.08935293e-02
-7.04336882e-01 1.39463201e-01 3.44530642e-01 -5.42340994e-01
-2.86851406e-01 -3.41128707e-01 1.40948659e-02 -3.23636532e-01
8.60133350e-01 2.47705385e-01 1.72918290e-03 -1.61737531e-01
2.34452173e-01 2.33934149e-01 1.10241950e+00 -5.74468315e-01
8.98945093e-01 1.03803706e+00 4.29753214e-01 -9.06167448e-01
-1.01898205e+00 -5.22305727e-01 -9.23500121e-01 1.36476591e-01
9.67321038e-01 -1.11638618e+00 -1.24373269e+00 8.33737195e-01
-1.45647728e+00 1.06051899e-01 -9.77361649e-02 3.10966760e-01
-1.07429874e+00 4.58745539e-01 -2.55374789e-01 -2.56334126e-01
-7.79710114e-02 -9.16997373e-01 1.56718898e+00 -3.77443321e-02
-9.22801435e-01 -5.13381124e-01 -1.14400350e-01 7.72508562e-01
3.17733914e-01 2.79780656e-01 9.00543690e-01 -6.78609848e-01
-1.02308607e+00 -2.29329050e-01 -9.34141517e-01 1.89279810e-01
-6.55939281e-02 -5.14157563e-02 -1.00312173e+00 -1.81565583e-01
-2.76374936e-01 -7.88044512e-01 8.80363762e-01 1.39539763e-01
1.64934540e+00 -4.78044935e-02 1.89491391e-01 1.18290460e+00
8.81815314e-01 -4.80497360e-01 5.64592421e-01 5.28247535e-01
7.06638992e-01 7.34064281e-01 -2.39470139e-01 1.72978163e-01
4.58130002e-01 8.00696731e-01 4.95944917e-01 -1.04696773e-01
-2.30346680e-01 -3.67215633e-01 -4.18128781e-02 6.60721660e-02
-5.21457911e-01 -6.02438301e-02 -1.02076912e+00 1.38005868e-01
-1.47846866e+00 -1.38971519e+00 2.86412627e-01 2.00146031e+00
1.95795432e-01 3.32642585e-01 1.65731773e-01 1.00299053e-01
8.22877884e-01 3.49839568e-01 -4.82780695e-01 -3.29272002e-01
-5.51509202e-01 1.04009137e-02 3.34614486e-01 3.81515026e-02
-1.49459124e+00 6.75935268e-01 6.74803162e+00 4.15764153e-01
-1.19543958e+00 -3.73923004e-01 3.25129598e-01 6.79176226e-02
5.06429486e-02 -6.91042468e-02 -4.28393573e-01 -1.72340974e-01
2.59269681e-02 -6.02663420e-02 5.93905509e-01 1.19552910e+00
-2.50208557e-01 4.46347952e-01 -1.60264206e+00 1.68269229e+00
5.49907804e-01 -1.79414344e+00 4.07676727e-01 2.08659261e-01
1.12052274e+00 3.03776532e-01 3.38020146e-01 -4.45450507e-02
8.33979473e-02 -1.41209650e+00 8.29777300e-01 2.56531060e-01
8.08343351e-01 -3.90669018e-01 4.18054551e-01 -9.84762143e-03
-1.08253598e+00 -1.21101514e-01 -4.17741477e-01 -3.28808337e-01
3.19723785e-02 -6.28411174e-02 -7.80216992e-01 5.07705688e-01
9.83710825e-01 1.10338223e+00 -8.54562104e-01 1.06126380e+00
-1.80145085e-01 -2.02608600e-01 -4.33738589e-01 3.79757226e-01
2.66157746e-01 -3.76477130e-02 4.40657049e-01 9.37212706e-01
2.29630858e-01 -2.68365443e-01 -1.36090279e-01 1.37602854e+00
-2.58938015e-01 -3.00835431e-01 -5.86620390e-01 -2.39601191e-02
1.73899949e-01 1.04042017e+00 -9.10014808e-01 -1.84072051e-02
-1.65724590e-01 1.20492196e+00 3.26289564e-01 2.73016512e-01
-6.66429758e-01 -1.94605187e-01 8.45065951e-01 3.29840362e-01
9.02414247e-02 -4.14708823e-01 -3.35810900e-01 -1.44107819e+00
4.25468534e-01 -9.32764113e-01 2.03912020e-01 -1.46110845e+00
-1.52442014e+00 4.82215881e-01 2.09404975e-02 -1.35818100e+00
-8.06868002e-02 -1.15281582e+00 -8.08293164e-01 6.59764349e-01
-7.70294666e-01 -1.50975978e+00 -6.53274477e-01 5.03628790e-01
4.32881564e-01 -5.56514859e-01 7.90239334e-01 -8.52522552e-02
2.30042562e-02 5.00635147e-01 -1.74277589e-01 5.68740010e-01
7.29559183e-01 -1.23532736e+00 1.23316574e+00 6.26148760e-01
6.16723299e-01 8.62744451e-01 5.45983076e-01 -1.69341505e-01
-7.70795524e-01 -6.53368175e-01 1.23920047e+00 -8.59926283e-01
4.93940264e-01 -9.31134343e-01 -5.30067503e-01 8.63366961e-01
2.29714029e-02 1.47409126e-01 3.40724766e-01 3.06783855e-01
-1.54389465e+00 1.44711435e-01 -9.73238170e-01 8.48881245e-01
1.96454167e+00 -1.14458406e+00 -1.01025093e+00 7.71351695e-01
2.71673679e-01 -6.41689301e-01 -3.57639223e-01 7.38102376e-01
7.69612670e-01 -1.69962943e+00 1.39127648e+00 -9.24382746e-01
8.99729371e-01 -7.00735748e-02 -2.64002144e-01 -9.93775427e-01
-4.54868108e-01 -7.32269883e-01 8.39185536e-01 5.30747175e-01
-1.36572674e-01 -5.68807304e-01 9.02878344e-01 2.45471016e-01
-1.68137789e-01 -4.65938002e-01 -1.22905576e+00 -1.14093721e+00
6.88995421e-03 -1.09953389e-01 4.72868264e-01 1.27798486e+00
-3.48683655e-01 -1.49555448e-02 -1.37762308e-01 7.31481314e-02
8.68461788e-01 3.97535175e-01 1.05552709e+00 -1.38396692e+00
-8.74663442e-02 -8.55257452e-01 -1.36159217e+00 -9.45659220e-01
4.45804447e-01 -1.08492231e+00 -5.77051222e-01 -1.18859041e+00
4.59223390e-01 1.90894052e-01 8.93996358e-02 4.48117882e-01
7.12834477e-01 6.92664444e-01 2.07293376e-01 1.17172286e-01
-8.05793941e-01 6.40884459e-01 1.29527140e+00 -3.58899049e-02
5.07363617e-01 -1.93650082e-01 -5.95762253e-01 1.16028929e+00
4.20806378e-01 -9.07441303e-02 -2.76484817e-01 -3.00265163e-01
5.70362449e-01 -5.57130158e-01 1.24628484e+00 -1.24584734e+00
1.65014654e-01 3.07115447e-02 9.40478921e-01 -7.50712931e-01
5.31177521e-01 -4.29976523e-01 1.36854857e-01 2.81225860e-01
-4.73316431e-01 4.25859600e-01 -8.87830183e-02 1.83977976e-01
-1.53150916e-01 4.02538449e-01 9.03385818e-01 -3.17500561e-01
-6.94971263e-01 3.22468907e-01 1.01905324e-01 5.06289601e-01
5.59321344e-01 -5.74929833e-01 -7.34827220e-01 -4.70325381e-01
-7.39568710e-01 -8.49255323e-02 9.06265259e-01 8.25996220e-01
5.03934503e-01 -1.44526160e+00 -5.46573937e-01 4.94605988e-01
5.61402440e-01 -9.65797231e-02 5.19576967e-01 4.31914359e-01
-1.05205154e+00 6.28264546e-01 -1.02174890e+00 -9.10239577e-01
-1.29583788e+00 3.69750142e-01 7.99853742e-01 3.38923365e-01
-8.06513906e-01 1.03976166e+00 5.38810432e-01 -1.05228317e+00
8.52690116e-02 -5.57870626e-01 3.47940952e-01 -3.90202463e-01
2.03300864e-01 4.19374794e-01 5.70332399e-03 -9.85473633e-01
-5.75191855e-01 1.18717718e+00 3.81586887e-02 -1.35780796e-01
1.18943977e+00 2.37860829e-01 -3.14954132e-01 -6.48297146e-02
1.50777662e+00 -7.47521296e-02 -1.24617183e+00 -2.78135985e-01
-2.54126519e-01 -5.20356119e-01 -7.88183570e-01 -6.71407282e-01
-4.68336165e-01 9.35263038e-01 1.71598122e-01 -1.12204626e-01
6.06871426e-01 1.53257892e-01 3.61124545e-01 1.04354417e+00
2.24760488e-01 -9.65862870e-01 5.01462579e-01 7.11203277e-01
1.49056900e+00 -1.50314224e+00 1.98504537e-01 -6.48283362e-01
-6.56886399e-01 1.33020937e+00 7.52121866e-01 -5.42834163e-01
4.35313284e-01 -4.33016062e-01 9.87717360e-02 -3.94842327e-01
-4.77209657e-01 1.97674006e-01 8.90642464e-01 7.76713014e-01
7.13996962e-02 -2.17627622e-02 4.48057920e-01 2.16943637e-01
-1.17188478e+00 -3.69863302e-01 4.76282477e-01 1.87470213e-01
-5.74670620e-02 -6.60598695e-01 -1.91783890e-01 -2.54545547e-02
-8.21482465e-02 1.98875412e-01 -1.01401246e+00 1.04281116e+00
2.31834099e-01 1.75546199e-01 5.24628699e-01 -1.06401131e-01
2.67048925e-01 2.27654278e-02 7.99399197e-01 -3.70563954e-01
-3.15588981e-01 -5.58086634e-01 1.13533288e-01 -1.08840346e+00
-3.13956797e-01 -9.15548980e-01 -8.37446749e-01 -2.84693778e-01
5.39504647e-01 -3.83104324e-01 5.24239838e-01 8.53171885e-01
3.23731422e-01 5.66864796e-02 3.94171447e-01 -1.14655840e+00
-8.22586775e-01 -6.18157566e-01 -5.02477527e-01 1.04545128e+00
1.98641166e-01 -8.20142210e-01 -7.24087358e-01 -2.22451925e-01]
|
[8.624198913574219, -2.051072597503662]
|
a7bafb69-e9f9-46a5-82e9-44b8d687029a
|
your-room-is-not-private-gradient-inversion
|
2306.09273
| null |
https://arxiv.org/abs/2306.09273v1
|
https://arxiv.org/pdf/2306.09273v1.pdf
|
Your Room is not Private: Gradient Inversion Attack for Deep Q-Learning
|
The prominence of embodied Artificial Intelligence (AI), which empowers robots to navigate, perceive, and engage within virtual environments, has attracted significant attention, owing to the remarkable advancements in computer vision and large language models. Privacy emerges as a pivotal concern within the realm of embodied AI, as the robot access substantial personal information. However, the issue of privacy leakage in embodied AI tasks, particularly in relation to decision-making algorithms, has not received adequate consideration in research. This paper aims to address this gap by proposing an attack on the Deep Q-Learning algorithm, utilizing gradient inversion to reconstruct states, actions, and Q-values. The choice of using gradients for the attack is motivated by the fact that commonly employed federated learning techniques solely utilize gradients computed based on private user data to optimize models, without storing or transmitting the data to public servers. Nevertheless, these gradients contain sufficient information to potentially expose private data. To validate our approach, we conduct experiments on the AI2THOR simulator and evaluate our algorithm on active perception, a prevalent task in embodied AI. The experimental results convincingly demonstrate the effectiveness of our method in successfully recovering all information from the data across all 120 room layouts.
|
['Ding Zhao', 'Wenhao Ding', 'Miao Li']
|
2023-06-15
| null | null | null | null |
['q-learning', 'navigate']
|
['methodology', 'reasoning']
|
[ 3.31606656e-01 3.30744743e-01 1.50941312e-01 -4.03350204e-01
-7.54425049e-01 -6.40108466e-01 4.39578891e-01 1.63201198e-01
-8.77681315e-01 7.95648754e-01 -1.65123846e-02 -3.47099900e-01
-8.47465545e-02 -7.20115840e-01 -8.27269733e-01 -9.64597344e-01
-2.83377916e-01 -1.22619025e-01 -4.79577422e-01 -2.02040777e-01
1.97248846e-01 3.69687170e-01 -1.59994221e+00 -1.23813644e-01
8.77475083e-01 1.25066161e+00 6.40492216e-02 4.41377133e-01
1.79579556e-01 9.54046607e-01 -5.61100304e-01 -7.94226229e-01
4.29144770e-01 -1.02604873e-01 -5.54071426e-01 -3.66475731e-01
-1.14721894e-01 -6.51436329e-01 -4.12622988e-01 1.27317333e+00
5.53679347e-01 6.88419119e-03 2.40765199e-01 -1.72549009e+00
-5.81122756e-01 5.10640919e-01 -2.01638058e-01 -3.04149568e-01
4.04374182e-01 1.98249057e-01 8.42410922e-01 -4.20077801e-01
4.56250012e-01 9.64636087e-01 1.73923150e-01 6.57049894e-01
-8.30528557e-01 -7.43584037e-01 1.64016381e-01 3.27129692e-01
-1.46087849e+00 -7.38307476e-01 8.25134158e-01 -7.56130964e-02
7.15926647e-01 4.01899368e-01 5.57283342e-01 1.55033624e+00
4.45402175e-01 9.80486274e-01 1.24103236e+00 -3.12361658e-01
8.45560670e-01 5.25104940e-01 -2.03794137e-01 9.22891974e-01
3.50688785e-01 2.90015817e-01 -7.89957404e-01 -5.87842584e-01
2.69925296e-01 -6.46919459e-02 -2.80649751e-01 -9.91575181e-01
-1.02575922e+00 7.37710834e-01 3.21877569e-01 -1.50788784e-01
-6.41763091e-01 3.88081133e-01 5.36958277e-01 2.78954983e-01
1.54711843e-01 1.17351890e-01 -2.97185153e-01 -4.36969161e-01
-6.71900958e-02 1.97206378e-01 8.87194216e-01 9.15895224e-01
6.98768198e-01 -1.58151209e-01 5.06329648e-02 2.07665041e-01
3.88622701e-01 6.44556582e-01 1.78720504e-01 -1.13829136e+00
5.13271272e-01 3.39631975e-01 4.33285654e-01 -1.27943873e+00
-1.23270534e-01 -2.39351373e-02 -8.63021374e-01 3.92177761e-01
1.11797430e-01 -4.76479083e-01 -3.57400030e-01 2.03747106e+00
4.59260046e-01 -1.78136870e-01 5.53055763e-01 1.15739918e+00
3.75754416e-01 4.36601311e-01 1.01766512e-01 -1.40459150e-01
1.15791667e+00 -6.26805663e-01 -9.05357420e-01 -1.95542052e-01
4.95203733e-01 -6.71549290e-02 7.83368826e-01 4.26167995e-01
-8.05139422e-01 -9.02824551e-02 -1.31570566e+00 2.17210059e-03
-3.91897678e-01 -1.26268223e-01 9.95244801e-01 9.68792319e-01
-9.31487024e-01 2.20419511e-01 -9.03250396e-01 -3.00214022e-01
7.57951975e-01 5.59939921e-01 -5.74645758e-01 -3.68957445e-02
-1.26729500e+00 6.32951260e-01 1.43080607e-01 3.13973844e-01
-9.93265510e-01 -1.35675669e-01 -9.28883672e-01 1.38234630e-01
6.34330988e-01 -7.29314923e-01 1.10112977e+00 -7.75098026e-01
-1.87780488e+00 7.63109267e-01 9.34928358e-02 -7.22835124e-01
7.69580066e-01 -2.45416373e-01 -1.36279240e-01 5.50508797e-02
-2.14603052e-01 4.04403239e-01 9.55377996e-01 -1.31734467e+00
-5.63093364e-01 -7.43861556e-01 3.78089786e-01 4.58157182e-01
-5.56531370e-01 -3.66237998e-01 -2.34659798e-02 2.17573233e-02
-8.63749757e-02 -1.03179753e+00 -4.80474293e-01 2.15153456e-01
-4.01847363e-01 1.14253722e-01 5.42705119e-01 -3.62104684e-01
6.95847988e-01 -2.43043160e+00 1.13490969e-01 3.07742983e-01
2.56009012e-01 -2.73778617e-01 1.26129046e-01 4.21838850e-01
6.83899045e-01 -1.01004355e-01 -3.35847974e-01 -5.72526038e-01
5.24388909e-01 2.08907351e-01 -6.70965552e-01 8.57946157e-01
-1.99352294e-01 8.45638752e-01 -8.37943733e-01 -3.22744519e-01
1.77774709e-02 3.86146903e-01 -4.36472476e-01 3.75668138e-01
-7.16105336e-03 2.55755842e-01 -8.36387038e-01 5.02053499e-01
6.08626187e-01 4.43307795e-02 3.20303053e-01 1.00079454e-01
-6.12810999e-02 -6.89386427e-02 -8.80078495e-01 2.01199055e+00
-2.91212976e-01 3.57743710e-01 5.82735121e-01 -9.49273884e-01
7.45918870e-01 3.09707046e-01 5.95778465e-01 -9.41723645e-01
3.97725523e-01 6.27722889e-02 -4.15040910e-01 -4.39274400e-01
6.06879056e-01 2.50482857e-01 -5.15760481e-01 5.27486861e-01
-5.61181486e-01 -2.20304802e-02 -5.98151743e-01 2.08685368e-01
1.30367959e+00 1.55715674e-01 2.33686537e-01 6.99743927e-02
3.40994239e-01 -9.80635732e-03 4.35604930e-01 8.88484001e-01
-6.77482367e-01 -2.61089765e-02 2.77008116e-01 -2.24616691e-01
-5.23904324e-01 -9.77532208e-01 2.62409538e-01 1.03090298e+00
6.88321888e-01 -1.07560359e-01 -9.03085113e-01 -6.90194130e-01
6.45088777e-02 7.94367015e-01 -5.60698330e-01 -5.26466787e-01
-3.99303883e-02 -4.51638758e-01 5.75849414e-01 5.11130877e-02
8.00557554e-01 -1.14751601e+00 -1.63852096e+00 3.27999219e-02
-2.27710411e-01 -9.71994340e-01 1.94516946e-02 4.19431061e-01
-4.65901136e-01 -8.20183218e-01 -3.24067086e-01 -4.03559953e-01
6.27945662e-01 3.01863164e-01 4.94018167e-01 -2.31392682e-01
-1.99093461e-01 8.92329276e-01 -2.50190645e-01 -6.46077275e-01
-2.79498786e-01 1.78504884e-02 2.45033994e-01 2.08640367e-01
3.46796721e-01 -3.52370828e-01 -5.67153037e-01 -4.87078056e-02
-8.53455603e-01 -1.56558510e-02 5.52486539e-01 7.38897860e-01
1.69126272e-01 1.63507879e-01 3.96011919e-01 -7.28454947e-01
7.46126354e-01 -5.00977993e-01 -7.39203870e-01 3.09058428e-01
-6.09367788e-01 2.04341523e-02 4.96847928e-01 -2.29156777e-01
-1.17967844e+00 1.09661810e-01 1.32563978e-01 -1.94155797e-01
-1.47160530e-01 2.97425449e-01 -5.98455727e-01 -1.94922417e-01
2.50192195e-01 1.65498659e-01 2.84632266e-01 6.25719577e-02
5.39543092e-01 9.28417921e-01 4.70205992e-01 -6.20460927e-01
5.92331231e-01 8.20845246e-01 -1.03395537e-01 -5.32036185e-01
-2.25779176e-01 -5.58236092e-02 -3.31582874e-02 -5.84997684e-02
8.04574370e-01 -9.54620361e-01 -1.45974040e+00 4.16617930e-01
-1.07937169e+00 -1.09909914e-01 -1.37806371e-01 4.49193180e-01
-9.10265148e-01 2.79754251e-01 -4.26246703e-01 -1.31866431e+00
-4.62251097e-01 -1.39189279e+00 8.04990888e-01 1.44273430e-01
-5.74299097e-02 -4.06902760e-01 -9.59021598e-02 5.85970700e-01
4.24857795e-01 3.04015189e-01 8.54634404e-01 -4.56809521e-01
-9.94284391e-01 -1.05687544e-01 7.73432851e-02 -5.91519848e-02
7.35999122e-02 -5.11457145e-01 -1.38213432e+00 -5.82568288e-01
2.20809087e-01 -7.01386094e-01 3.01744431e-01 -8.65926668e-02
1.12373245e+00 -5.38622737e-01 -2.67944962e-01 5.79538584e-01
1.24433517e+00 3.91717136e-01 4.90848660e-01 5.02117932e-01
4.13111269e-01 6.75667405e-01 7.86779225e-01 9.92352903e-01
5.25551736e-01 3.70509744e-01 1.03000319e+00 1.02502137e-01
8.11775327e-01 -2.10734770e-01 6.21859550e-01 4.71735537e-01
2.15997294e-01 -3.19358170e-01 -5.02954543e-01 2.46393561e-01
-1.91394711e+00 -7.23238349e-01 6.29163861e-01 2.19524860e+00
5.48772633e-01 -2.94841472e-02 -4.40776855e-01 -1.43528402e-01
3.36476684e-01 2.51056850e-01 -1.02064526e+00 -5.39096713e-01
2.73300661e-03 -1.88109979e-01 7.42555082e-01 6.13192730e-02
-1.00419843e+00 7.59954870e-01 5.29890013e+00 3.98565888e-01
-9.75157082e-01 1.66040361e-01 6.36880338e-01 -1.26842350e-01
-3.88942987e-01 -1.65951490e-01 -1.01562850e-01 2.76305646e-01
8.41446757e-01 -3.38416725e-01 8.32727432e-01 1.05165124e+00
-5.40408976e-02 -2.61918664e-01 -1.17499971e+00 1.21597683e+00
-7.39174336e-02 -9.40070808e-01 -4.40525413e-01 2.92587429e-01
2.74392933e-01 -7.73006305e-02 5.29794395e-01 3.08793336e-01
4.57427144e-01 -8.92868280e-01 1.14873910e+00 3.44714254e-01
5.09424090e-01 -1.00931919e+00 6.04178607e-01 5.33810973e-01
-6.99550033e-01 -3.71559232e-01 -5.08431971e-01 -3.34365889e-02
-3.99335362e-02 6.09188005e-02 -7.71671772e-01 5.54733694e-01
9.33614671e-01 1.06496759e-01 -1.32852390e-01 4.65694577e-01
-1.89077575e-02 2.63935059e-01 -3.19919944e-01 -4.41702187e-01
1.65538117e-01 -4.07121956e-01 4.13122743e-01 4.68693703e-01
2.30227798e-01 1.86223388e-01 -1.40779406e-01 9.32246745e-01
-1.89432874e-01 1.07851207e-01 -9.97317195e-01 -2.91894048e-01
4.82649058e-01 1.05115700e+00 -2.62067020e-01 2.11251348e-01
-2.32943326e-01 1.49323761e+00 4.62901771e-01 5.19977510e-01
-9.16526675e-01 -2.24793583e-01 1.00900280e+00 -5.56866407e-01
3.88731845e-02 -4.74901825e-01 -1.50624067e-01 -8.49781156e-01
1.44608527e-01 -1.02845442e+00 1.47943020e-01 -4.86412019e-01
-9.73597467e-01 4.96969193e-01 -1.62153035e-01 -8.05863917e-01
-2.35717267e-01 -4.31230128e-01 2.94646304e-02 5.41515112e-01
-1.31139123e+00 -9.10643756e-01 -2.98338890e-01 6.53462708e-01
-3.34756225e-02 -2.47485176e-01 1.11537039e+00 -3.24295275e-03
-6.00519657e-01 7.88570702e-01 2.50144541e-01 -9.54358950e-02
4.60503995e-01 -7.07571447e-01 3.21686655e-01 5.91514647e-01
1.15392968e-01 6.93267941e-01 6.21389031e-01 -3.72519821e-01
-2.34224772e+00 -5.93550920e-01 2.63495058e-01 -4.69996303e-01
4.29978728e-01 -6.62976325e-01 -4.51234728e-01 6.22787356e-01
1.35921419e-01 2.61307415e-02 7.66264975e-01 -2.63270438e-01
-9.45086703e-02 -2.59686261e-01 -1.52823806e+00 8.58730078e-01
1.10703087e+00 -8.41785371e-01 -1.68967918e-01 -1.05409667e-01
1.02094793e+00 -3.46903801e-01 -6.39225125e-01 1.49321303e-01
6.85276866e-01 -1.13252926e+00 8.11843455e-01 -4.05619204e-01
-7.05955774e-02 7.01627880e-03 -5.36306679e-01 -1.02979851e+00
1.75697371e-01 -9.38707054e-01 -1.67326078e-01 1.10806501e+00
1.28226101e-01 -1.09670210e+00 1.00423717e+00 1.31587410e+00
3.76578987e-01 -5.48210680e-01 -1.20344257e+00 -2.38154203e-01
-3.65247875e-01 -3.69271636e-01 9.69176352e-01 7.21023381e-01
1.05105378e-02 -1.64745927e-01 -3.83720845e-01 5.24689972e-01
9.35844839e-01 5.09368144e-02 9.20929611e-01 -7.30417967e-01
-6.39088377e-02 3.52275968e-02 -4.02077585e-01 -7.89241374e-01
4.20079082e-01 -5.55477798e-01 3.69143903e-01 -1.22925615e+00
2.70881131e-02 -7.62114644e-01 -5.74136555e-01 4.70744282e-01
8.90117958e-02 -4.02114503e-02 2.81694531e-01 3.14163007e-02
-6.93952799e-01 1.08847535e+00 8.80705714e-01 -3.99779588e-01
-7.40191117e-02 8.82401038e-03 -7.93868661e-01 4.60952997e-01
7.34973550e-01 -3.79649848e-01 -7.19862878e-01 -4.71829414e-01
3.19575906e-01 1.15064003e-01 4.51260060e-01 -9.35895205e-01
5.14062643e-01 -2.46773139e-01 1.18927434e-01 -2.27001294e-01
6.36118889e-01 -1.47234046e+00 1.99135050e-01 4.62181628e-01
-4.67488468e-01 4.58773151e-02 3.02122179e-02 8.93168569e-01
9.04754102e-02 8.70952010e-02 8.59149396e-02 1.59838758e-02
-9.87256706e-01 3.12296063e-01 -4.68379915e-01 -3.29731166e-01
1.25429952e+00 -1.64175615e-01 -1.59895971e-01 -5.21639943e-01
-2.11181104e-01 3.10482383e-01 7.16983676e-01 4.72835928e-01
8.10004056e-01 -9.62994814e-01 -1.80323899e-01 4.71461177e-01
2.60890633e-01 5.03409989e-02 2.37367332e-01 4.63654101e-01
-3.12563300e-01 2.91106373e-01 -2.02904582e-01 -4.10500586e-01
-9.98972893e-01 8.22610438e-01 1.06423266e-01 2.07059771e-01
-5.16960263e-01 6.74760640e-01 1.64094329e-01 -4.24351543e-01
7.90832937e-01 -1.48087874e-01 5.35098016e-02 -2.49308780e-01
2.94218123e-01 3.58249903e-01 -1.25928015e-01 -4.70163971e-01
-5.42404413e-01 -1.01431012e-01 1.39978945e-01 -3.16256344e-01
1.05891037e+00 -5.94531059e-01 -2.23220289e-02 3.29125553e-01
1.16959441e+00 1.17538929e-01 -1.43761218e+00 -5.68890162e-02
5.33333223e-04 -3.68015826e-01 1.32179752e-01 -6.20556533e-01
-8.14931512e-01 6.47412181e-01 9.08385217e-01 1.05727643e-01
9.70700800e-01 -4.54684049e-01 8.70997965e-01 7.92334795e-01
1.40814090e+00 -1.12667429e+00 -6.34992868e-02 9.73821282e-02
5.11695325e-01 -1.35974169e+00 -1.02632768e-01 -7.33171105e-02
-8.78757715e-01 5.46342731e-01 4.69631463e-01 4.41507190e-01
3.52692008e-01 2.96039820e-01 4.17241395e-01 -4.99008857e-02
-6.12138450e-01 1.84676647e-01 -3.65471095e-01 5.84836483e-01
-3.92354101e-01 2.68830985e-01 7.03481436e-02 6.70199156e-01
-1.66834310e-01 3.56851295e-02 2.86282867e-01 1.42434335e+00
-2.04131603e-01 -7.53098488e-01 -2.82102972e-01 -2.51592919e-02
-3.38561624e-01 3.16258907e-01 -3.05385888e-01 4.43136185e-01
-1.19173601e-01 1.15901053e+00 -2.82997806e-02 -4.08503354e-01
1.01559564e-01 -1.76569477e-01 3.73313874e-01 -3.59468535e-03
-5.25755048e-01 -6.20174527e-01 -8.00191686e-02 -1.01741314e+00
-3.00936699e-01 -4.40215379e-01 -1.38773608e+00 -3.33266526e-01
-7.99965486e-02 3.15816045e-01 1.17323470e+00 7.33347416e-01
5.54595172e-01 1.38276592e-01 9.20287788e-01 -6.30377948e-01
-1.14426947e+00 -3.10754687e-01 -5.73389530e-01 2.14813426e-01
5.11789799e-01 -5.02689064e-01 -4.63906139e-01 -3.94483507e-01]
|
[5.831775188446045, 6.789862632751465]
|
d7abb4c6-cec2-4eb2-871a-a5ca1adcb1b9
|
multimodal-emotion-recognition-using-transfer
|
2202.08974
| null |
https://arxiv.org/abs/2202.08974v1
|
https://arxiv.org/pdf/2202.08974v1.pdf
|
Multimodal Emotion Recognition using Transfer Learning from Speaker Recognition and BERT-based models
|
Automatic emotion recognition plays a key role in computer-human interaction as it has the potential to enrich the next-generation artificial intelligence with emotional intelligence. It finds applications in customer and/or representative behavior analysis in call centers, gaming, personal assistants, and social robots, to mention a few. Therefore, there has been an increasing demand to develop robust automatic methods to analyze and recognize the various emotions. In this paper, we propose a neural network-based emotion recognition framework that uses a late fusion of transfer-learned and fine-tuned models from speech and text modalities. More specifically, we i) adapt a residual network (ResNet) based model trained on a large-scale speaker recognition task using transfer learning along with a spectrogram augmentation approach to recognize emotions from speech, and ii) use a fine-tuned bidirectional encoder representations from transformers (BERT) based model to represent and recognize emotions from the text. The proposed system then combines the ResNet and BERT-based model scores using a late fusion strategy to further improve the emotion recognition performance. The proposed multimodal solution addresses the data scarcity limitation in emotion recognition using transfer learning, data augmentation, and fine-tuning, thereby improving the generalization performance of the emotion recognition models. We evaluate the effectiveness of our proposed multimodal approach on the interactive emotional dyadic motion capture (IEMOCAP) dataset. Experimental results indicate that both audio and text-based models improve the emotion recognition performance and that the proposed multimodal solution achieves state-of-the-art results on the IEMOCAP benchmark.
|
['Ram D. Sriram', 'Dinesh Manocha', 'Seyed Omid Sadjadi', 'Sarala Padi']
|
2022-02-16
| null | null | null | null |
['multimodal-emotion-recognition', 'emotional-intelligence', 'multimodal-emotion-recognition']
|
['computer-vision', 'natural-language-processing', 'speech']
|
[ 5.29676601e-02 -2.87927061e-01 1.04846261e-01 -5.47118366e-01
-7.28288412e-01 -1.89697102e-01 3.01338017e-01 -2.59743869e-01
-5.35019577e-01 4.75715965e-01 3.01758319e-01 1.48329765e-01
-1.85197741e-01 -3.12408209e-01 -4.09907281e-01 -7.69890308e-01
8.85975957e-02 2.52798855e-01 -3.52006704e-01 -5.26165724e-01
-1.16013521e-02 4.12123650e-01 -1.71569824e+00 5.07133126e-01
8.26699793e-01 1.72561717e+00 1.96542609e-02 7.98419952e-01
-1.32068917e-01 9.86493111e-01 -3.99390310e-01 -3.65084767e-01
-2.79552341e-01 -3.07151079e-01 -6.63718283e-01 2.85231252e-03
-3.13190252e-01 5.06086014e-02 -1.49921507e-01 7.36650527e-01
7.91641295e-01 7.73750901e-01 5.01709163e-01 -1.39872241e+00
-4.15819079e-01 2.99379826e-01 -3.78934443e-01 5.09720482e-02
2.72713631e-01 -3.18728507e-01 6.73160791e-01 -9.44175720e-01
3.28171283e-01 1.15726674e+00 6.53934360e-01 6.86069608e-01
-6.62261069e-01 -7.37156630e-01 7.33422115e-02 6.15470886e-01
-1.28054118e+00 -4.56872702e-01 1.06421888e+00 -2.68375814e-01
1.18104780e+00 1.82481095e-01 4.23707247e-01 1.38401628e+00
-5.61088212e-02 7.49789000e-01 7.51364350e-01 -4.01320100e-01
1.87193453e-01 5.78763127e-01 7.92823657e-02 6.19933486e-01
-8.64645302e-01 -1.06281295e-01 -7.61991024e-01 -1.51966125e-01
3.26859504e-01 -3.50332558e-02 -7.63568208e-02 -9.17508900e-02
-9.31743205e-01 8.67671311e-01 2.18524024e-01 5.24850965e-01
-8.44496429e-01 6.43680850e-03 9.02670979e-01 3.64560306e-01
5.63694239e-01 4.95574325e-02 -5.66582799e-01 -7.20340669e-01
-5.88795125e-01 -3.44748527e-01 7.10108578e-01 4.19299394e-01
4.93626505e-01 4.46546733e-01 1.76964805e-01 1.50916517e+00
1.93424657e-01 4.21561807e-01 1.07029545e+00 -7.59739101e-01
2.30360344e-01 5.43921232e-01 -2.22042724e-01 -1.17531109e+00
-4.89474058e-01 -2.13532537e-01 -9.45574343e-01 -1.07255364e-02
-3.42530876e-01 -3.85485709e-01 -5.83975255e-01 1.82251966e+00
2.57957548e-01 5.06354094e-01 4.92163628e-01 9.82454300e-01
9.65082824e-01 9.82420981e-01 2.95030624e-01 -1.29898027e-01
1.15448213e+00 -1.14705098e+00 -8.99085820e-01 3.89639437e-02
5.74931502e-01 -5.09528518e-01 7.49816179e-01 6.02588475e-01
-6.98247552e-01 -7.35222161e-01 -8.76462042e-01 2.59309649e-01
-5.79026222e-01 3.69032323e-01 5.51500678e-01 7.55783260e-01
-8.07389498e-01 1.75358012e-01 -7.36114383e-01 -4.81298774e-01
4.95499335e-02 6.59680486e-01 -5.23717701e-01 3.04790348e-01
-1.34614742e+00 8.43162715e-01 3.68164122e-01 3.19876343e-01
-4.90897119e-01 -1.00384817e-01 -1.01565588e+00 2.82101125e-01
9.75488424e-02 -4.09720868e-01 9.35330153e-01 -1.58738375e+00
-2.02126956e+00 3.78660411e-01 -1.66664064e-01 -2.62720764e-01
-2.95041502e-01 -1.67339340e-01 -1.05671299e+00 3.56480360e-01
-4.82921183e-01 6.99965656e-01 8.97381008e-01 -1.11295438e+00
-7.24507272e-01 -4.22604740e-01 -4.82656956e-01 4.49157476e-01
-8.96406054e-01 2.75685281e-01 -3.33265394e-01 -4.98980999e-01
-1.32560655e-01 -9.66795266e-01 8.34408626e-02 -6.24965072e-01
8.18874687e-02 -1.85530469e-01 9.79891121e-01 -6.68784320e-01
1.18445337e+00 -2.39614272e+00 4.46204871e-01 3.72895688e-01
-3.16325933e-01 3.65086019e-01 -4.33636487e-01 1.47647053e-01
-3.36336583e-01 -1.31051525e-01 2.60854214e-02 -5.08297205e-01
1.34812787e-01 2.09255725e-01 -1.99255958e-01 2.58281585e-02
2.83862919e-01 7.63490379e-01 -3.16670030e-01 -2.86907166e-01
5.09389937e-01 8.95142078e-01 -4.91569281e-01 3.03935885e-01
2.46749580e-01 4.32899058e-01 -3.15205514e-01 5.71363688e-01
2.95211107e-01 1.29561588e-01 -1.00947045e-01 -3.54406357e-01
4.85790335e-02 -1.29534677e-01 -1.14957869e+00 1.69038796e+00
-8.10738206e-01 5.52104414e-01 2.60187447e-01 -1.33869541e+00
1.23633027e+00 7.92043269e-01 8.09613526e-01 -5.97783566e-01
6.29859328e-01 9.70847309e-02 -1.92400977e-01 -9.44816351e-01
6.16015196e-01 -5.20980179e-01 -3.49035800e-01 1.12924844e-01
4.83938903e-01 1.98883116e-01 -3.18644971e-01 -2.52956688e-01
7.23281085e-01 -5.41903712e-02 4.34898548e-02 4.77763623e-01
1.01088321e+00 -2.47294128e-01 5.48798382e-01 7.63995573e-02
-3.14616203e-01 1.85320571e-01 -4.89366613e-02 -1.93999425e-01
-4.49055821e-01 -5.65847576e-01 1.67436987e-01 1.59477615e+00
1.44140841e-02 2.87815053e-02 -5.28859496e-01 -5.35925984e-01
-3.80523622e-01 6.08949840e-01 -5.21669984e-01 -5.78929842e-01
-9.81317759e-02 -6.01097286e-01 8.36550236e-01 6.50672615e-01
5.33906519e-01 -1.48656356e+00 -3.66797805e-01 3.05632830e-01
-6.83982611e-01 -1.32236063e+00 -1.68352321e-01 4.02516752e-01
-5.80035686e-01 -5.04913151e-01 -5.85811734e-01 -9.13906634e-01
7.40506425e-02 -1.62796065e-01 4.19461042e-01 -4.16289896e-01
3.33837345e-02 8.76594961e-01 -7.42523849e-01 -3.01368386e-01
-2.24373877e-01 8.36275667e-02 2.34100714e-01 7.07218766e-01
4.38842177e-01 -5.29761553e-01 -1.17313519e-01 4.11101609e-01
-8.99807513e-01 -3.55209202e-01 4.42749321e-01 1.00864971e+00
1.98636487e-01 1.93254128e-01 1.18357718e+00 -2.39107609e-01
8.88805687e-01 -6.66451991e-01 2.96672918e-02 1.40843049e-01
-2.15797022e-01 -1.41441897e-01 5.61686754e-01 -8.49099040e-01
-1.58746767e+00 1.44754440e-01 -4.56678361e-01 -7.66962528e-01
-3.58628482e-01 8.13658178e-01 -1.77978978e-01 -2.03008577e-01
2.96482950e-01 1.34841770e-01 -1.81924738e-02 -2.00123101e-01
3.29676837e-01 1.23396540e+00 7.32990384e-01 -4.67614204e-01
4.98007052e-02 2.00549796e-01 -3.87534410e-01 -8.98811162e-01
-2.12155968e-01 -7.15094984e-01 -1.88608646e-01 -5.72924316e-01
1.12852216e+00 -9.49314356e-01 -1.18485534e+00 4.56697106e-01
-9.45709944e-01 -1.16528541e-01 9.93559975e-03 9.75380540e-01
-7.34892309e-01 1.49842173e-01 -7.41072357e-01 -1.23031390e+00
-5.02092838e-01 -1.08438575e+00 1.12703097e+00 3.92761588e-01
-2.48752251e-01 -9.77076113e-01 8.59774500e-02 6.03396654e-01
4.59371299e-01 5.97412921e-02 7.17689753e-01 -9.30008590e-01
2.24864870e-01 -3.06569725e-01 5.55160455e-02 6.97406411e-01
-2.80866101e-02 -3.05308938e-01 -1.07811785e+00 1.98523358e-01
1.54145554e-01 -7.29523897e-01 6.71142995e-01 2.22484425e-01
9.70137715e-01 6.25874996e-02 4.08819132e-03 4.38211054e-01
1.04471517e+00 7.11529732e-01 6.08681321e-01 1.86612159e-01
5.75915873e-01 7.75606096e-01 6.03899300e-01 7.04646468e-01
4.28475142e-01 6.65090144e-01 3.06390971e-01 -5.17409444e-02
4.31885600e-01 2.01720119e-01 6.40491009e-01 1.28892016e+00
-9.80105475e-02 -3.04696590e-01 -7.26421058e-01 4.32610720e-01
-2.08224511e+00 -1.08005679e+00 3.08225214e-01 1.77577436e+00
2.85901159e-01 -3.35925579e-01 -2.71994565e-02 2.39843264e-01
7.62695968e-01 -1.25679240e-01 -5.10011494e-01 -1.00857806e+00
-1.05091304e-01 2.66339332e-01 -1.54582188e-01 3.24549586e-01
-1.12360835e+00 9.54770088e-01 4.79992008e+00 8.78043830e-01
-1.63583279e+00 2.85397023e-01 6.12025142e-01 -1.71089292e-01
2.76799589e-01 -7.83193588e-01 -3.73156250e-01 3.23077321e-01
1.41519153e+00 8.16699341e-02 7.00667679e-01 9.82763648e-01
3.26764792e-01 1.61804989e-01 -7.71194398e-01 1.49740374e+00
5.57085574e-01 -7.86498010e-01 -2.14617625e-01 -1.97318688e-01
4.12275821e-01 -6.71834275e-02 1.47207364e-01 8.81550431e-01
-2.80486077e-01 -8.88278425e-01 4.41288859e-01 7.43926406e-01
4.81547296e-01 -1.11025608e+00 1.06862700e+00 3.56091350e-01
-1.24382782e+00 -4.41640854e-01 1.02622375e-01 1.04988299e-01
3.34625334e-01 -6.77015856e-02 -6.03999019e-01 6.61849678e-01
8.71260941e-01 4.94179398e-01 2.86971405e-02 5.81883132e-01
4.08156157e-01 4.86133784e-01 -2.62915134e-01 -1.98479190e-01
1.97088584e-01 -1.72291741e-01 4.09294307e-01 1.25950146e+00
5.63073695e-01 3.69720638e-01 -8.07938203e-02 2.96194673e-01
-1.81741148e-01 3.71153265e-01 -4.12791908e-01 -3.58617067e-01
6.79161996e-02 1.55824125e+00 -3.13012481e-01 -4.01392609e-01
-3.18665951e-01 1.21922231e+00 1.40164465e-01 4.26043868e-01
-1.07969320e+00 -4.95754093e-01 6.76548541e-01 -7.79752374e-01
3.58679801e-01 -3.35849002e-02 2.17399653e-02 -1.09802592e+00
-1.76864848e-01 -1.02481270e+00 4.14865524e-01 -1.12556589e+00
-1.17338479e+00 1.07612717e+00 -3.21280897e-01 -1.14548159e+00
-6.05351567e-01 -6.22628987e-01 -4.25978422e-01 5.82170188e-01
-1.18425560e+00 -1.11219597e+00 -3.63891006e-01 9.69528139e-01
6.33476436e-01 -4.01468992e-01 1.01773894e+00 7.08867788e-01
-7.56351113e-01 5.10148048e-01 -2.93861236e-03 -2.70637237e-02
8.26231599e-01 -6.78139627e-01 -6.76464915e-01 3.15439194e-01
-1.12197407e-01 1.78915605e-01 2.99321711e-01 -2.25982070e-01
-1.29141867e+00 -9.87012148e-01 5.40319681e-01 6.56597838e-02
4.40706521e-01 -1.26776129e-01 -8.00144732e-01 5.39672673e-01
3.25417995e-01 -1.43381178e-01 1.01436377e+00 1.46850139e-01
-3.50378454e-02 -2.85077900e-01 -1.15519238e+00 3.05952847e-01
3.28884274e-01 -7.52873302e-01 -5.33675492e-01 -2.72945106e-01
4.16881412e-01 -4.70700227e-02 -9.76665795e-01 6.22510254e-01
8.24546635e-01 -8.25754464e-01 8.18765461e-01 -6.39619410e-01
3.14678222e-01 7.98063725e-02 -5.54267585e-01 -1.49440050e+00
-6.96040317e-02 -3.39162380e-01 3.12843025e-02 1.38913548e+00
3.18605065e-01 -5.57656288e-01 6.10138416e-01 7.67477810e-01
-2.68171728e-01 -7.44060099e-01 -1.00913763e+00 -3.49183232e-01
-4.27639455e-01 -8.01564991e-01 3.81555408e-01 1.08412755e+00
5.71997404e-01 6.56792343e-01 -5.78725994e-01 6.34386167e-02
-1.49296761e-01 -2.20012859e-01 5.29537559e-01 -1.06316435e+00
-1.42595574e-01 -4.68915015e-01 -5.80063462e-01 -7.44841158e-01
6.46734536e-01 -6.89310253e-01 2.11539000e-01 -1.06346571e+00
-5.46566956e-02 -5.80394119e-02 -7.33537614e-01 4.39380765e-01
1.53237537e-01 3.57766956e-01 2.01735169e-01 -3.34571958e-01
-7.56435990e-01 1.20796227e+00 6.69227183e-01 -4.14501950e-02
-5.01778543e-01 -1.86972529e-01 -4.97801453e-01 7.95512199e-01
7.67393231e-01 -2.03900635e-01 -2.64933616e-01 -1.33780129e-02
1.33084849e-01 4.35631663e-01 2.20576108e-01 -1.11015356e+00
3.80165696e-01 1.55714497e-01 3.63695502e-01 -3.51977289e-01
1.06760395e+00 -1.25781882e+00 2.17921231e-02 -3.66331786e-02
-3.75949681e-01 -3.19966413e-02 5.58542848e-01 5.05495191e-01
-6.79509342e-01 -1.25387078e-02 5.25884390e-01 3.16934526e-01
-9.13848579e-01 1.24699138e-02 -7.66349554e-01 -4.84771729e-01
8.80110264e-01 -1.45302415e-01 -1.82630550e-02 -8.53744030e-01
-1.18041003e+00 1.36640415e-01 -3.54398817e-01 8.45443070e-01
7.82757163e-01 -1.59731424e+00 -2.91262299e-01 1.46848604e-01
1.25206128e-01 -7.63280392e-01 6.42987967e-01 1.03961408e+00
1.05187967e-01 3.43949914e-01 -4.06918615e-01 -4.16785598e-01
-1.44008815e+00 3.12504232e-01 3.99403483e-01 -1.44999355e-01
8.26303214e-02 6.87846124e-01 5.74543737e-02 -7.88931906e-01
4.57859933e-01 -9.40261856e-02 -5.37451088e-01 1.86450765e-01
3.40510368e-01 5.80029547e-01 1.02284230e-01 -1.18277109e+00
-4.36109513e-01 5.45128167e-01 3.44671905e-01 -5.22951484e-01
1.39973497e+00 -3.97249490e-01 6.96445182e-02 7.54367411e-01
1.40002191e+00 -2.40622014e-01 -5.93813539e-01 -8.33863243e-02
-2.05829173e-01 1.23677209e-01 3.21002215e-01 -1.02051914e+00
-1.08769298e+00 1.06623352e+00 9.87241089e-01 1.70728322e-02
1.66614926e+00 -2.79982775e-01 9.79571402e-01 6.25604212e-01
1.02855917e-02 -1.43110299e+00 3.75890762e-01 6.77668154e-01
8.28313768e-01 -1.31333780e+00 -7.53035486e-01 -1.11980118e-01
-1.27291548e+00 1.06144142e+00 6.40208066e-01 1.97579026e-01
7.49670744e-01 6.02287278e-02 2.89126307e-01 -3.88452364e-03
-9.11747813e-01 -2.42961347e-01 4.54456300e-01 3.52734089e-01
3.55994046e-01 -6.01965189e-02 4.68864292e-03 1.33307219e+00
1.04735583e-01 2.10764885e-01 -3.53673920e-02 6.14921033e-01
-3.93553853e-01 -9.43429530e-01 -5.63887775e-01 1.88995734e-01
-3.89819115e-01 -1.83207858e-02 -3.54439229e-01 3.84864628e-01
1.77310646e-01 1.40621018e+00 -1.01348208e-02 -9.68371391e-01
3.76073837e-01 7.40159690e-01 1.37237400e-01 -7.05815926e-02
-8.00668597e-01 2.52057344e-01 2.81422168e-01 -5.32683611e-01
-6.81095183e-01 -2.72633821e-01 -1.47395384e+00 7.40118027e-02
-4.07334059e-01 2.82993823e-01 1.12171745e+00 1.06120849e+00
7.34088898e-01 7.94372797e-01 8.25984359e-01 -1.04625118e+00
-1.64133474e-01 -1.14757490e+00 -5.30495107e-01 4.31306213e-01
1.22628503e-01 -7.17332840e-01 -3.07820916e-01 1.98369980e-01]
|
[13.346765518188477, 5.295775890350342]
|
bd4f9c35-b372-48b4-963b-cf71eb3b122c
|
unrealcv-connecting-computer-vision-to-unreal
|
1609.01326
| null |
http://arxiv.org/abs/1609.01326v1
|
http://arxiv.org/pdf/1609.01326v1.pdf
|
UnrealCV: Connecting Computer Vision to Unreal Engine
|
Computer graphics can not only generate synthetic images and ground truth but
it also offers the possibility of constructing virtual worlds in which: (i) an
agent can perceive, navigate, and take actions guided by AI algorithms, (ii)
properties of the worlds can be modified (e.g., material and reflectance),
(iii) physical simulations can be performed, and (iv) algorithms can be learnt
and evaluated. But creating realistic virtual worlds is not easy. The game
industry, however, has spent a lot of effort creating 3D worlds, which a player
can interact with. So researchers can build on these resources to create
virtual worlds, provided we can access and modify the internal data structures
of the games. To enable this we created an open-source plugin UnrealCV
(http://unrealcv.github.io) for a popular game engine Unreal Engine 4 (UE4). We
show two applications: (i) a proof of concept image dataset, and (ii) linking
Caffe with the virtual world to test deep network algorithms.
|
['Weichao Qiu', 'Alan Yuille']
|
2016-09-05
| null | null | null | null |
['physical-simulations']
|
['miscellaneous']
|
[-2.58539975e-01 2.33870372e-01 5.68210900e-01 -8.61627609e-02
-4.51846421e-02 -7.66294122e-01 7.17572033e-01 -3.24970096e-01
-2.21698165e-01 7.44516730e-01 -1.00909606e-01 -4.41799551e-01
3.70864242e-01 -1.39184129e+00 -7.20143914e-01 -3.44814241e-01
-1.34946123e-01 4.68816549e-01 8.12409997e-01 -4.16143626e-01
-7.27458950e-03 7.01143622e-01 -2.01872420e+00 1.70932204e-01
6.24616981e-01 5.66033900e-01 5.87652087e-01 8.31206203e-01
-2.55639330e-02 8.35142434e-01 -5.02184570e-01 -3.26781422e-01
6.04878485e-01 -5.54221988e-01 -6.53234363e-01 -1.37780070e-01
-2.20712591e-02 -2.58307129e-01 -4.06801283e-01 9.94604766e-01
4.61364031e-01 3.24135661e-01 1.32901281e-01 -1.35571158e+00
-4.39463228e-01 2.33364895e-01 -2.07585245e-01 -2.61600852e-01
5.60237229e-01 8.74124169e-01 5.49429595e-01 -3.49369824e-01
1.13552380e+00 1.13802528e+00 3.07238340e-01 8.98035526e-01
-1.05224514e+00 -7.74275303e-01 -1.31574884e-01 1.17682420e-01
-1.23319304e+00 -4.08006519e-01 6.29305124e-01 -3.32610667e-01
8.74998748e-01 4.23795909e-01 1.30710876e+00 1.40236914e+00
1.40447319e-01 4.81003225e-01 9.74913120e-01 -1.53645769e-01
4.88234460e-01 2.69701570e-01 -6.46923482e-01 9.27614093e-01
-3.84977385e-02 3.87954861e-01 -4.02922750e-01 1.47043675e-01
1.22602737e+00 -5.45942605e-01 -2.05683202e-01 -6.08872592e-01
-1.45327139e+00 6.31267548e-01 5.89597404e-01 -1.60700336e-01
-5.24972558e-01 4.44215268e-01 2.94541001e-01 9.86833200e-02
1.68700784e-01 8.73109162e-01 -2.30552077e-01 -3.51618081e-01
-4.15397465e-01 5.47621548e-01 8.37706029e-01 6.92190230e-01
7.77305663e-01 3.64053190e-01 1.82920456e-01 5.17437816e-01
4.25258040e-01 4.40194696e-01 4.08247530e-01 -1.58792210e+00
3.15804631e-02 4.13374126e-01 3.53987336e-01 -9.86514330e-01
-4.79113549e-01 -7.52626508e-02 -5.42961419e-01 9.64757144e-01
5.17255425e-01 -4.53489870e-01 -9.64801013e-01 1.55172122e+00
5.30748785e-01 4.23773348e-01 2.96946149e-03 1.09129357e+00
1.23709273e+00 5.71195900e-01 -1.16855383e-01 4.78579104e-01
1.11292017e+00 -7.06444204e-01 -1.49567470e-01 -3.96458387e-01
5.07202089e-01 -5.30803025e-01 1.19175875e+00 3.82804930e-01
-1.26774037e+00 -4.21243399e-01 -1.17052162e+00 1.86277002e-01
-5.76541483e-01 -5.79468191e-01 1.23228085e+00 7.50120521e-01
-1.35040224e+00 6.82376027e-01 -1.07467890e+00 -3.43984038e-01
2.96007931e-01 2.21678540e-01 -4.48545694e-01 1.16841599e-01
-1.27893913e+00 7.52129495e-01 4.71134514e-01 -1.12414248e-01
-1.12517011e+00 -6.43365860e-01 -9.54485059e-01 -1.61492035e-01
3.22091281e-01 -1.17872798e+00 1.03328526e+00 -1.24526191e+00
-1.83633137e+00 9.47660089e-01 3.27800423e-01 -2.08454967e-01
8.36222529e-01 4.06922102e-01 -1.01193771e-01 9.49232131e-02
7.10761966e-03 9.95345294e-01 3.16121787e-01 -1.31227922e+00
-4.66238946e-01 -2.04992324e-01 7.81628489e-01 4.90926296e-01
2.59899199e-01 -2.80927075e-03 -5.83715737e-01 -2.62496740e-01
3.69310305e-02 -8.84588301e-01 -5.06843805e-01 5.21838605e-01
-4.11505252e-01 3.77632767e-01 4.29074019e-01 -4.77929235e-01
2.30598733e-01 -1.96684110e+00 -1.11502439e-01 2.26054370e-01
4.07704026e-01 3.86870056e-01 -1.59764931e-01 2.01413468e-01
-2.74276994e-02 2.88980603e-01 2.23729730e-01 -8.97302851e-03
-2.22400296e-02 6.91119209e-02 2.43523732e-01 2.25033969e-01
-2.93728560e-01 8.24543536e-01 -1.05860090e+00 -2.34789401e-01
6.79437697e-01 5.71598709e-01 -7.05508530e-01 -2.52675191e-02
-4.47744995e-01 5.88498175e-01 -4.28362966e-01 1.62105083e-01
5.78313231e-01 -3.03318858e-01 1.39658436e-01 2.33147591e-01
-2.53191054e-01 3.30972403e-01 -1.51924050e+00 1.68627000e+00
-6.10143423e-01 8.90615940e-01 5.19401319e-02 -5.73930204e-01
8.20616245e-01 2.44623736e-01 3.03008020e-01 -9.29113328e-01
1.86150998e-01 -1.69959173e-01 -9.21162516e-02 -4.39158112e-01
4.30396765e-01 5.60875796e-02 2.27123141e-01 5.77014267e-01
-2.94714928e-01 -5.51491380e-01 1.57454938e-01 2.30688855e-01
1.20387101e+00 4.56074595e-01 -7.27083087e-02 -2.15329248e-02
6.97583193e-03 2.72502840e-01 3.49643141e-01 6.61715090e-01
1.71114467e-02 6.68301463e-01 5.15628159e-01 -6.79524899e-01
-1.40220559e+00 -1.46143341e+00 2.73777336e-01 7.75275290e-01
3.97873551e-01 -1.92228809e-01 -7.98934400e-01 6.52995519e-03
-2.28760898e-01 7.99183726e-01 -6.83083832e-01 -5.51575124e-02
-2.95190871e-01 -5.36436856e-01 4.27199632e-01 3.48938167e-01
8.87460709e-01 -1.36219704e+00 -1.02363062e+00 6.56636581e-02
-2.17975378e-02 -1.09051847e+00 1.11222312e-01 -2.17826068e-01
-5.40182948e-01 -9.62855756e-01 -2.06993744e-01 -5.39653897e-01
3.23997140e-01 4.67246115e-01 1.41788793e+00 4.09076631e-01
-2.16331661e-01 3.67844313e-01 -2.38100022e-01 -4.24561232e-01
-5.09049237e-01 -1.79112032e-01 -2.34055091e-02 -4.53983903e-01
-2.24015981e-01 -8.63545239e-01 -7.14144826e-01 4.13616449e-01
-7.57095039e-01 9.64185774e-01 -3.47320456e-03 3.22004110e-01
6.05283320e-01 -8.10674205e-02 3.38956505e-01 -8.65614831e-01
4.91003394e-01 -2.57689714e-01 -9.71778989e-01 -3.29210460e-02
-1.49509370e-01 -7.19972327e-02 5.73542893e-01 -2.57083148e-01
-1.17841363e+00 1.97001070e-01 -3.56814206e-01 -1.72884747e-01
-3.67274880e-01 3.99504602e-01 -4.77894664e-01 -2.06253186e-01
9.55913424e-01 -4.46207747e-02 1.03567772e-01 4.32768371e-03
6.12901032e-01 3.08690846e-01 5.28944314e-01 -4.40328777e-01
7.35193789e-01 7.59080887e-01 -3.43390144e-02 -8.64307821e-01
-9.62237790e-02 2.51901716e-01 -3.96321416e-01 -6.41428471e-01
9.46366131e-01 -8.54593635e-01 -9.92509365e-01 5.42303920e-01
-1.00131166e+00 -9.69537437e-01 -3.04958344e-01 5.03458858e-01
-6.50588572e-01 -2.91947052e-02 -4.30898517e-01 -5.39226949e-01
8.55351016e-02 -1.16587472e+00 5.29438019e-01 6.92396879e-01
-3.08268726e-01 -1.10761702e+00 4.54335995e-02 2.18678579e-01
4.42729414e-01 7.12908685e-01 3.84310544e-01 -1.45180389e-01
-9.08539355e-01 -1.12931214e-01 -3.87625307e-01 -3.06669567e-02
-8.34348500e-02 4.26140398e-01 -1.05735421e+00 1.13772869e-01
-4.86118942e-01 -1.72279000e-01 3.08893740e-01 4.86050993e-01
1.26084960e+00 1.13844201e-01 -4.59567487e-01 1.04628432e+00
1.22629881e+00 3.93436790e-01 1.33951581e+00 5.77033758e-01
6.40739679e-01 4.84812349e-01 7.16431290e-02 3.11948836e-01
5.57986677e-01 8.65627348e-01 8.11805010e-01 -3.51588577e-01
-8.64305049e-02 -1.72450811e-01 4.62825373e-02 3.51747572e-01
-4.70675081e-01 -2.71614999e-01 -9.53372598e-01 2.16228068e-01
-1.70341468e+00 -1.18709099e+00 -3.16320151e-01 2.07838511e+00
5.77135921e-01 3.13548803e-01 4.44005951e-02 -2.77803749e-01
4.95425969e-01 2.88496930e-02 -8.06462526e-01 -5.25669694e-01
-1.19509362e-01 4.81634408e-01 3.49112123e-01 5.21107495e-01
-6.25712335e-01 1.14223754e+00 5.63159132e+00 6.37331903e-01
-1.37131131e+00 1.44752525e-02 5.14937818e-01 -2.37429336e-01
-4.63158607e-01 -3.98091562e-02 -1.08779766e-01 2.56622642e-01
8.54195416e-01 -4.52680856e-01 7.89434314e-01 8.21608782e-01
5.06614208e-01 -2.35043123e-01 -8.05029273e-01 9.00726318e-01
-3.13239306e-01 -1.73624051e+00 -1.26897559e-01 8.51610675e-02
4.37231272e-01 3.44260424e-01 -1.57186866e-01 2.13850796e-01
9.50263977e-01 -1.11407995e+00 8.50077391e-01 5.53346395e-01
7.94344485e-01 -6.71704054e-01 3.57657820e-01 4.24495757e-01
-9.53061700e-01 5.56931913e-01 -5.02150595e-01 -2.85493702e-01
2.92629469e-02 2.06972674e-01 -7.59407163e-01 3.24347049e-01
8.43369603e-01 3.82808417e-01 -4.84879762e-01 1.45561135e+00
-5.15438497e-01 3.58044446e-01 -4.21822280e-01 -3.72236192e-01
8.39597359e-02 -4.56315488e-01 4.17125434e-01 7.65892267e-01
2.11267337e-01 9.66232941e-02 -2.15111762e-01 1.31729126e+00
-9.50205326e-02 -1.56149209e-01 -7.25078583e-01 1.45197079e-01
3.72855067e-01 1.29530394e+00 -1.23372424e+00 -3.62668447e-02
-3.15105289e-01 1.02592957e+00 1.29285827e-01 4.28547114e-01
-9.59620535e-01 -4.77075428e-01 1.18716872e+00 3.09720010e-01
2.13698465e-02 -3.88873637e-01 -1.73116207e-01 -1.20228517e+00
-1.20752528e-01 -8.02547932e-01 -2.46786535e-01 -1.53447723e+00
-7.41800070e-01 6.75847530e-01 -2.24452659e-01 -9.36030626e-01
-2.94340938e-01 -6.59758389e-01 -6.75434351e-01 8.92158508e-01
-8.42809796e-01 -8.78713906e-01 -7.83448040e-01 5.70965111e-01
2.26897344e-01 -4.17391257e-03 1.02211177e+00 1.01222329e-01
-2.69418001e-01 7.14142695e-02 -6.89443499e-02 2.24818796e-01
1.07590780e-01 -1.07515466e+00 1.11768472e+00 5.61653793e-01
3.23375076e-01 1.72293380e-01 9.27700460e-01 -4.51222360e-01
-1.28775787e+00 -7.76170373e-01 -4.21509892e-02 -4.83273357e-01
4.97880787e-01 -6.88888013e-01 -7.26774573e-01 6.81474745e-01
6.50150925e-02 -6.83643147e-02 4.35459077e-01 -1.73016429e-01
-1.09647341e-01 1.87539518e-01 -1.25770175e+00 1.18270898e+00
1.28395844e+00 -2.54215747e-01 1.58157527e-01 4.44353282e-01
6.85955167e-01 -9.11515415e-01 -3.94273937e-01 -4.61644977e-02
8.68310273e-01 -1.60952675e+00 1.18002784e+00 -6.73813164e-01
3.59981477e-01 -4.40220147e-01 7.40518048e-02 -1.66707385e+00
-6.46628730e-04 -5.31032920e-01 3.15481395e-01 6.70296133e-01
6.88081980e-01 -8.24298561e-01 1.09193742e+00 9.89665866e-01
-1.16223939e-01 -6.21348619e-01 -5.99662066e-01 -6.06045544e-01
-5.65072708e-02 -8.92817736e-01 1.05021107e+00 9.07548010e-01
-1.93898410e-01 1.43026467e-02 5.49475802e-03 2.33807027e-01
2.93893486e-01 -1.22456700e-01 1.07362747e+00 -1.19257283e+00
-5.14679551e-01 -4.25217450e-01 -7.22321391e-01 -8.01501155e-01
7.86591414e-03 -8.06024611e-01 -1.89202487e-01 -2.06146359e+00
-7.85732940e-02 -6.33579552e-01 3.62667799e-01 3.82737905e-01
1.34590670e-01 3.99994582e-01 2.47019127e-01 -3.72500382e-02
-4.81674999e-01 3.85705769e-01 1.61833465e+00 1.86293617e-01
-2.44843692e-01 -1.04834892e-01 -5.77298582e-01 8.79180074e-01
1.06497109e+00 -1.82634234e-01 -5.84831059e-01 -4.71229345e-01
7.26867974e-01 1.48230612e-01 7.31061161e-01 -1.28914011e+00
-5.77302650e-02 -3.43815774e-01 5.35582125e-01 8.90569016e-02
6.28284156e-01 -3.53884876e-01 6.34847939e-01 2.53968298e-01
6.70461953e-02 -1.53989017e-01 3.94269526e-01 6.53686896e-02
3.96582782e-01 -1.45617470e-01 7.17527807e-01 -5.68738818e-01
-9.18899894e-01 2.71520823e-01 -4.58735824e-01 1.69252425e-01
1.13774192e+00 -4.95530665e-01 -5.16830802e-01 -7.47892737e-01
-7.54437208e-01 2.15945914e-01 1.01348603e+00 2.98906446e-01
6.27817571e-01 -1.12505794e+00 -5.56105137e-01 1.37084723e-01
-8.69652405e-02 1.49231553e-01 1.57214984e-01 5.49775548e-02
-1.46782219e+00 -2.66433358e-01 -6.05546236e-01 -3.53246063e-01
-9.09445584e-01 4.77754980e-01 8.71583700e-01 2.31072292e-01
-9.10851538e-01 1.00043130e+00 3.56153429e-01 -6.40390337e-01
-2.25780100e-01 -4.86408472e-02 4.15345700e-03 -4.66016710e-01
6.12770438e-01 1.63546875e-02 3.68435644e-02 -4.04542774e-01
-1.79796383e-01 9.64451432e-02 4.30512041e-01 -3.72558445e-01
1.59146547e+00 4.92352098e-02 8.62780660e-02 1.31199539e-01
5.69259584e-01 -1.32087424e-01 -1.55793381e+00 3.56346846e-01
-5.22334516e-01 -6.03759885e-01 2.06122678e-02 -7.85852194e-01
-1.45216298e+00 6.31141722e-01 5.00758111e-01 4.26453441e-01
8.58397901e-01 -4.85333428e-02 6.20293558e-01 2.67066717e-01
8.55667114e-01 -9.34713304e-01 1.41841331e-02 3.73318553e-01
7.41304874e-01 -8.35909128e-01 -2.31494337e-01 -4.14922833e-01
-7.00771391e-01 9.72066522e-01 6.50925934e-01 -3.25788081e-01
6.69534683e-01 5.64083517e-01 3.14740717e-01 -4.40274417e-01
-8.28094900e-01 -2.42681593e-01 -1.89165026e-01 1.15387726e+00
2.94441551e-01 2.20484346e-01 2.33536303e-01 2.66725928e-01
-8.37001741e-01 -2.92322040e-02 9.72244978e-01 5.52901328e-01
-2.50002295e-01 -7.84634888e-01 -2.72788525e-01 3.90346467e-01
-1.48471724e-02 7.74123222e-02 -2.98928291e-01 9.92568433e-01
1.36210978e-01 9.03308868e-01 1.25088498e-01 -4.33217496e-01
4.02193129e-01 -3.49075466e-01 6.06480658e-01 -7.81527638e-01
-3.51433188e-01 -6.05138838e-01 3.17292541e-01 -9.20131266e-01
-1.22765966e-01 -4.77506906e-01 -1.46251202e+00 -7.34728217e-01
-9.06664357e-02 1.45577807e-02 9.80918467e-01 6.16253138e-01
3.91530931e-01 7.55444109e-01 1.46008983e-01 -1.28975582e+00
4.08488125e-01 -5.84826112e-01 -7.35206664e-01 3.60525340e-01
-3.19887280e-01 -7.62615621e-01 -3.78152430e-01 2.85980776e-02]
|
[4.589593410491943, 0.5154894590377808]
|
2b20dc87-a38e-47b4-b030-917e9def0cf7
|
revisiting-dense-retrieval-with-unanswerable
|
2304.03031
| null |
https://arxiv.org/abs/2304.03031v5
|
https://arxiv.org/pdf/2304.03031v5.pdf
|
Revisiting Dense Retrieval with Unanswerable Counterfactuals
|
The retriever-reader framework is popular for open-domain question answering (ODQA), where a retriever samples for the reader a set of relevant candidate passages from a large corpus. A key assumption behind this method is that high relevance scores from the retriever likely indicate high answerability from the reader, which implies a high probability that the retrieved passages contain answers to a given question. In this work, we empirically dispel this belief and observe that recent dense retrieval models based on DPR often rank unanswerable counterfactual passages higher than their answerable original passages. To address such answer-unawareness in dense retrievers, we seek to use counterfactual samples as additional training resources to better synchronize the relevance measurement of DPR with the answerability of question-passage pairs. Specifically, we present counterfactually-Pivoting Contrastive Learning (PiCL), a novel representation learning approach for passage retrieval that leverages counterfactual samples as pivots between positive and negative samples in their learned embedding space. We incorporate PiCL into the retriever training to show the effectiveness of PiCL on ODQA benchmarks and the robustness of the learned models.
|
['Jinyeong Yeo', 'Kyungjae Lee', 'Dahyun Lee', 'Yongho Song']
|
2023-04-06
| null | null | null | null |
['passage-retrieval', 'open-domain-question-answering']
|
['natural-language-processing', 'natural-language-processing']
|
[-1.05335668e-01 1.06355637e-01 -2.69477934e-01 -5.20434231e-02
-1.78606367e+00 -8.37658644e-01 9.22934532e-01 4.88433897e-01
-4.45725828e-01 7.45251417e-01 9.77971613e-01 -3.04310411e-01
-4.53338116e-01 -1.03630483e+00 -8.63838851e-01 -2.65799791e-01
1.72558039e-01 7.26870000e-01 3.17548096e-01 -6.43737137e-01
4.50144470e-01 -2.36996990e-02 -1.43697751e+00 7.50631452e-01
8.57432008e-01 7.19761789e-01 5.94598539e-02 7.71384120e-01
-2.46793374e-01 1.20580721e+00 -8.39641273e-01 -8.43910158e-01
2.05057323e-01 -5.07946968e-01 -1.19882274e+00 -8.54122937e-01
7.58939743e-01 -3.54200035e-01 -4.74633396e-01 7.38280892e-01
5.17397642e-01 4.07607526e-01 1.01685417e+00 -8.45178783e-01
-1.26174796e+00 6.05766833e-01 -4.60499041e-02 8.42671573e-01
8.71614993e-01 2.96527520e-02 1.97560251e+00 -9.60229874e-01
6.19406581e-01 1.27928996e+00 2.57971108e-01 4.51622516e-01
-1.05267441e+00 -3.35632443e-01 -2.68627014e-02 5.54093421e-01
-1.03137660e+00 -2.48299256e-01 7.61831462e-01 -1.10267401e-01
7.81576693e-01 6.74015343e-01 4.70020592e-01 1.21865404e+00
2.92767227e-01 1.18101335e+00 7.07138002e-01 -4.47827220e-01
3.23338419e-01 2.02779621e-01 3.75364214e-01 -2.12564785e-03
3.11571628e-01 1.89772338e-01 -6.21586800e-01 -4.87646133e-01
1.39928862e-01 9.23467614e-03 -6.78991199e-01 -3.22805166e-01
-9.99956191e-01 1.21707058e+00 7.36200631e-01 1.50836691e-01
-5.39342284e-01 5.73580377e-02 1.51000112e-01 7.69955099e-01
5.37719190e-01 1.24247980e+00 -4.09573019e-01 1.57364920e-01
-7.35225081e-01 8.71798813e-01 9.84039009e-01 4.43134040e-01
6.63244128e-01 -7.64918089e-01 -9.86911356e-01 7.85757601e-01
3.39104265e-01 6.71945751e-01 7.17593133e-01 -8.77469838e-01
6.05333447e-01 5.36921203e-01 3.95712137e-01 -1.16013217e+00
2.58349925e-01 -4.21081662e-01 -7.11766630e-02 -6.69608414e-01
1.55248791e-01 2.22498029e-01 -1.72925577e-01 1.62813926e+00
2.75703698e-01 -4.60730083e-02 4.05524433e-01 1.05079353e+00
8.17262828e-01 1.00057697e+00 6.44850545e-03 -7.15917870e-02
1.26036859e+00 -9.94600356e-01 -5.75114250e-01 -2.64160812e-01
7.23371863e-01 -7.54934251e-01 1.53331542e+00 -2.38986760e-01
-8.61538649e-01 -2.27474332e-01 -8.21151435e-01 -3.48746240e-01
-1.69178814e-01 -1.83324903e-01 1.64003074e-01 1.93672314e-01
-8.61318231e-01 4.68169779e-01 2.13139206e-02 -9.66800004e-02
2.24632874e-01 -2.51719326e-01 7.26051927e-02 -4.73508805e-01
-1.78516400e+00 9.64023292e-01 2.42142789e-02 -2.45318145e-01
-9.09915328e-01 -1.07585990e+00 -5.85852146e-01 3.93519402e-01
2.43410930e-01 -9.40353274e-01 1.49005949e+00 -3.62775534e-01
-1.04071331e+00 8.51282954e-01 -1.57683015e-01 -7.32522190e-01
2.63138205e-01 -6.14464164e-01 -4.21705812e-01 5.24822116e-01
3.13726127e-01 3.61666173e-01 9.82815146e-01 -1.13841569e+00
-4.59949523e-01 -1.79344609e-01 3.97731304e-01 4.94384110e-01
-1.50042698e-01 -3.13045084e-01 4.74167150e-03 -4.96754497e-01
-1.26387045e-01 -5.13412714e-01 1.15386702e-01 -2.61640489e-01
-1.38552785e-01 -7.63525188e-01 3.33171934e-01 -5.49642205e-01
1.37162149e+00 -2.01590848e+00 -1.57662928e-01 3.75563279e-02
2.51952052e-01 2.20518038e-01 -6.88708246e-01 7.03360438e-01
8.30759704e-02 -3.51643935e-02 1.57544911e-01 2.98655421e-01
2.65208095e-01 -4.73279469e-02 -1.31325316e+00 1.94093913e-01
1.44081503e-01 1.28702331e+00 -1.29549158e+00 -3.98379773e-01
-2.42249355e-01 -8.52396935e-02 -7.56857991e-01 5.36418378e-01
-5.88810384e-01 -9.49001983e-02 -5.90565264e-01 2.29358867e-01
3.45164567e-01 -3.42554480e-01 -2.00264707e-01 -4.87142764e-02
4.52509612e-01 1.16493094e+00 -5.97694755e-01 1.22565079e+00
-5.74907959e-01 5.47954381e-01 -5.38088381e-01 -7.41991460e-01
8.04384232e-01 3.07709694e-01 -6.19179122e-02 -1.21348584e+00
-1.66015953e-01 3.80262941e-01 -1.64484754e-01 -7.39487708e-01
1.00973392e+00 -3.76940340e-01 -5.29161021e-02 9.27660942e-01
-2.42745578e-02 -4.58407402e-01 1.68625057e-01 7.36668289e-01
1.39779508e+00 -2.71958172e-01 1.76089779e-01 -2.61383146e-01
5.61096549e-01 -2.69770669e-03 1.81793608e-02 1.27402806e+00
-1.95476666e-01 5.47174215e-01 3.94697696e-01 -9.69379172e-02
-8.33709121e-01 -1.58062208e+00 5.34249321e-02 1.27392066e+00
1.87573284e-01 -2.44875073e-01 -1.91041872e-01 -8.39047015e-01
1.32190078e-01 1.13070631e+00 -6.27778888e-01 -6.13903403e-01
-6.37098014e-01 -2.83783227e-01 3.56338322e-01 1.54213428e-01
-1.91595145e-02 -1.10142124e+00 -4.63723779e-01 2.35090125e-03
-7.35541224e-01 -4.69506681e-01 -4.74527061e-01 -2.97775716e-01
-7.76564956e-01 -1.25617099e+00 -9.97322440e-01 -7.29403794e-01
2.55442023e-01 5.35272121e-01 1.79566455e+00 -8.52082297e-03
2.22146839e-01 9.21042025e-01 -6.65117562e-01 -1.78637281e-01
-5.03152609e-01 5.16299568e-02 -2.60375619e-01 -2.64108300e-01
7.33785808e-01 -2.47777119e-01 -1.07270551e+00 5.61911706e-03
-9.57550704e-01 -4.74742860e-01 5.51574051e-01 8.84083450e-01
4.45991784e-01 -6.05909824e-01 1.08397257e+00 -6.36489868e-01
1.42093158e+00 -1.03839219e+00 -2.30679974e-01 6.85497761e-01
-4.06573445e-01 3.55016112e-01 4.50605124e-01 -5.53779542e-01
-9.52014446e-01 -1.18531775e+00 -1.73427746e-01 -2.37746939e-01
4.45683390e-01 5.28989196e-01 2.46499419e-01 6.08246565e-01
1.05025506e+00 3.09457958e-01 -1.45376667e-01 -2.82166868e-01
7.15613306e-01 5.70306599e-01 3.11473787e-01 -6.72134101e-01
8.06388199e-01 4.05335933e-01 -5.28122902e-01 -4.13497746e-01
-1.49697924e+00 -9.90260303e-01 2.72090673e-01 -4.38996442e-02
3.35936785e-01 -8.25277269e-01 -4.88270164e-01 -3.48815560e-01
-1.16197217e+00 4.03120629e-02 -8.90537918e-01 4.26232159e-01
-5.08105159e-01 1.76437885e-01 -4.31663334e-01 -6.75230622e-01
-5.99339783e-01 -5.27606845e-01 1.06119466e+00 5.52986749e-02
-5.89760780e-01 -1.06539500e+00 6.75253749e-01 5.99852085e-01
5.57574630e-01 -3.04820299e-01 1.18117607e+00 -1.20825636e+00
-7.44632483e-01 -5.27600646e-01 -4.66014408e-02 3.15719694e-01
-4.15400416e-02 -5.47330737e-01 -9.10691857e-01 -3.17753255e-01
3.37822318e-01 -7.16607988e-01 1.17226005e+00 4.04603295e-02
6.27816856e-01 -8.11418056e-01 -1.13629274e-01 -3.48005831e-01
1.26864171e+00 -4.62734520e-01 6.48545444e-01 3.65481675e-01
1.27515733e-01 8.16411078e-01 7.24649787e-01 2.48318151e-01
3.71279418e-01 2.28640035e-01 1.64106831e-01 4.81599599e-01
-1.28165528e-01 -7.82346606e-01 4.11794066e-01 9.44196820e-01
7.14077950e-01 -5.06928265e-01 -5.91695249e-01 1.10259223e+00
-1.48697388e+00 -1.27672362e+00 3.85557078e-02 2.29995847e+00
1.12296724e+00 -1.86629757e-01 -1.57472134e-01 -4.21113893e-02
4.78840232e-01 5.17811000e-01 -4.09449130e-01 -1.18509054e-01
-1.23531878e-01 4.33390051e-01 -1.56628743e-01 5.96286476e-01
-5.84604621e-01 5.56081712e-01 5.69416046e+00 8.53044033e-01
-5.21912396e-01 2.01231893e-02 4.47283179e-01 -3.46749514e-01
-1.13407779e+00 1.17310286e-01 -5.78924358e-01 3.64058524e-01
1.00970256e+00 -6.56887650e-01 1.54478669e-01 7.05510676e-01
-1.53878167e-01 -1.35375008e-01 -1.46849442e+00 6.33295894e-01
2.91202068e-01 -1.59567165e+00 6.16034567e-01 -2.75806934e-01
8.36890996e-01 1.77554563e-01 3.84622395e-01 9.60222542e-01
3.47191185e-01 -7.87522256e-01 5.83360195e-01 6.52959049e-01
1.81714222e-01 -4.90666240e-01 7.92441845e-01 4.76920992e-01
-4.85649377e-01 -2.27425009e-01 -7.00157166e-01 -1.65498912e-01
-6.89595006e-03 5.91077149e-01 -9.68487740e-01 3.55359674e-01
4.90422994e-01 3.93835872e-01 -5.92967033e-01 9.49152589e-01
-4.77888793e-01 7.69310594e-01 -7.21510127e-02 -5.42143881e-01
2.17910334e-01 7.63118491e-02 7.05083430e-01 7.98326015e-01
1.45530790e-01 6.15822934e-02 -6.52008295e-01 1.06715548e+00
-5.42530596e-01 3.12144756e-01 -6.77322447e-01 -5.72423600e-02
6.99547112e-01 7.72784770e-01 1.05060913e-01 -3.79267275e-01
-3.47018838e-01 7.90026724e-01 6.31722569e-01 4.81515557e-01
-4.62305576e-01 -2.23060101e-01 4.69184697e-01 1.34635508e-01
4.42139596e-01 4.83621985e-01 2.05758527e-01 -1.38326156e+00
4.11152393e-01 -1.12750995e+00 7.47786820e-01 -7.45860100e-01
-1.85917294e+00 1.56307161e-01 -2.08031669e-01 -1.20405722e+00
-4.70489740e-01 -3.07227194e-01 -6.42950237e-01 9.56152022e-01
-2.01911640e+00 -5.36028266e-01 2.67638624e-01 3.77938002e-01
5.57337761e-01 1.02246076e-01 7.42295325e-01 -1.59240402e-02
6.48318306e-02 6.02581263e-01 3.97969872e-01 1.53428083e-02
8.80592108e-01 -1.13251770e+00 1.15070373e-01 6.56950235e-01
5.09537458e-01 1.11763906e+00 8.06723773e-01 -3.72166097e-01
-1.35238469e+00 -9.10909295e-01 1.36963212e+00 -1.15328717e+00
8.26839626e-01 1.32452026e-02 -1.20144689e+00 4.60852474e-01
3.99779797e-01 -1.87591717e-01 8.74792576e-01 3.94807339e-01
-6.57595634e-01 -1.70596629e-01 -9.21988726e-01 8.45243156e-01
3.81166130e-01 -1.00019002e+00 -1.71100187e+00 4.43021744e-01
1.02018332e+00 -6.03525601e-02 -4.88569617e-01 2.23349139e-01
3.30271155e-01 -8.80512178e-01 1.19026816e+00 -8.56721759e-01
9.00628686e-01 -5.87794520e-02 -4.30882186e-01 -1.33791697e+00
-6.77549765e-02 -2.91904330e-01 -6.47281528e-01 9.31364477e-01
5.28058827e-01 -4.83877629e-01 4.33770865e-01 5.85470736e-01
1.87282324e-01 -7.89165974e-01 -1.03051567e+00 -7.83799648e-01
6.71457469e-01 -2.05058798e-01 7.21631825e-01 6.75799131e-01
1.20827086e-01 7.59669840e-01 1.35254234e-01 6.56767488e-02
2.70092607e-01 5.28433740e-01 6.72405064e-01 -1.05430949e+00
-4.33262229e-01 -2.24204287e-01 8.18988904e-02 -1.54096615e+00
2.14132160e-01 -1.16190612e+00 1.33427829e-02 -1.71831739e+00
3.93740594e-01 -2.33553395e-01 -4.71018434e-01 -3.40846777e-01
-6.90937698e-01 -1.84790846e-02 6.33902177e-02 5.05996943e-01
-1.13846469e+00 9.17517304e-01 1.43441331e+00 -3.72076362e-01
3.13093998e-02 3.92902680e-02 -1.07803726e+00 2.21407667e-01
4.39960986e-01 -5.75474024e-01 -6.12901807e-01 -4.42042381e-01
9.91906285e-01 1.66469529e-01 6.02222025e-01 -5.33699572e-01
1.07109986e-01 1.52620777e-01 -1.75329335e-02 -5.53945720e-01
1.21000744e-01 -4.04445261e-01 -6.39817178e-01 3.26921791e-01
-1.15090811e+00 -6.04560487e-02 -1.95434481e-01 1.03961623e+00
-4.90006238e-01 -5.34932315e-01 2.22161114e-01 -1.07203938e-01
-4.42112714e-01 2.54082438e-02 -1.45804971e-01 9.79566932e-01
4.14741129e-01 2.32378453e-01 -6.94292724e-01 -6.63671911e-01
-8.57873783e-02 3.10871065e-01 9.52783599e-02 6.06551409e-01
8.75008762e-01 -1.32555890e+00 -1.01203573e+00 -2.19899088e-01
6.31854534e-01 -2.40311757e-01 3.91197503e-01 4.89358366e-01
-2.01543823e-01 7.46259093e-01 4.74154592e-01 -1.96101934e-01
-6.42515182e-01 6.19884789e-01 1.97271839e-01 -7.29144990e-01
-3.44587922e-01 9.75485384e-01 2.70479828e-01 -6.99646354e-01
-8.13465938e-02 -1.82554021e-01 -3.34731460e-01 1.54371381e-01
8.30517352e-01 4.21925485e-01 1.38636887e-01 -1.02021210e-01
-2.48008110e-02 -8.62691030e-02 -5.25890768e-01 -4.04028028e-01
1.03715444e+00 -1.95450440e-01 -1.20988972e-01 5.68033636e-01
1.59362006e+00 2.77038902e-01 -7.41774857e-01 -6.33329391e-01
1.91758826e-01 -6.71197057e-01 -1.02781229e-01 -8.55615675e-01
-2.33837768e-01 7.41551816e-01 2.42293403e-01 2.41441116e-01
7.38252640e-01 5.24776757e-01 1.05849683e+00 8.68919671e-01
1.76924765e-01 -9.42455351e-01 6.35397851e-01 6.33232176e-01
1.23245466e+00 -1.22819448e+00 -1.60112619e-01 2.64814675e-01
-4.45818186e-01 6.98454976e-01 3.90884131e-01 -2.81752259e-01
2.50007510e-01 -6.77082241e-01 1.59721345e-01 -4.63293999e-01
-1.08159816e+00 2.04113275e-02 4.82259661e-01 3.60482216e-01
4.56353188e-01 -1.29342973e-01 -4.70718950e-01 5.92949510e-01
-4.94875103e-01 -4.15818632e-01 4.74605143e-01 7.19084144e-01
-3.68137956e-01 -5.79487503e-01 -2.90774763e-01 7.61352897e-01
-4.60123628e-01 -4.75012630e-01 -3.23653430e-01 3.60766470e-01
-6.55448079e-01 1.05364239e+00 1.41919225e-01 1.43948942e-02
2.62196034e-01 1.92362249e-01 4.74242210e-01 -6.70262694e-01
-5.21288097e-01 -6.12369955e-01 -3.34726311e-02 -3.28776956e-01
-1.74897417e-01 -4.80903506e-01 -7.80984402e-01 -8.53962004e-02
-5.05704165e-01 8.26032043e-01 3.53191197e-02 1.04242349e+00
6.07789397e-01 2.55707145e-01 7.69346833e-01 1.60591125e-01
-1.28091621e+00 -1.02388406e+00 -3.74682724e-01 8.13205242e-01
7.18005002e-01 -1.94654286e-01 -7.60775328e-01 -5.45717180e-01]
|
[11.427091598510742, 7.756728649139404]
|
c2ddec18-69c8-47c2-84c6-87b0a3a423d4
|
multi-label-learning-with-missing-values
|
2008.07234
| null |
https://arxiv.org/abs/2008.07234v1
|
https://arxiv.org/pdf/2008.07234v1.pdf
|
Multi-label Learning with Missing Values using Combined Facial Action Unit Datasets
|
Facial action units allow an objective, standardized description of facial micro movements which can be used to describe emotions in human faces. Annotating data for action units is an expensive and time-consuming task, which leads to a scarce data situation. By combining multiple datasets from different studies, the amount of training data for a machine learning algorithm can be increased in order to create robust models for automated, multi-label action unit detection. However, every study annotates different action units, leading to a tremendous amount of missing labels in a combined database. In this work, we examine this challenge and present our approach to create a combined database and an algorithm capable of learning under the presence of missing labels without inferring their values. Our approach shows competitive performance compared to recent competitions in action unit detection.
|
['Jaspar Pahl', 'Dominik Seuss', 'Ines Rieger']
|
2020-08-17
| null | null | null | null |
['action-unit-detection']
|
['computer-vision']
|
[ 4.87364858e-01 1.10318139e-01 -3.85181367e-01 -5.95645785e-01
-8.77258658e-01 -6.12740636e-01 3.91829103e-01 -2.83806268e-02
-3.24792534e-01 7.55504549e-01 1.35819465e-01 4.36499208e-01
4.72829700e-01 -2.87607372e-01 -2.89831191e-01 -8.29737425e-01
1.69016853e-01 4.65849578e-01 6.94484189e-02 -3.87056284e-02
-2.31348239e-02 5.87432981e-01 -1.87842178e+00 7.01268077e-01
9.21924412e-02 1.04129612e+00 -7.55013004e-02 2.81119108e-01
-1.21803887e-01 7.80139267e-01 -6.74070776e-01 -4.12325650e-01
1.99609756e-01 -8.02326977e-01 -7.64023364e-01 1.91351429e-01
7.66123712e-01 -1.82265744e-01 4.30157244e-01 9.24866855e-01
7.61069894e-01 -9.84195620e-02 8.33739638e-01 -1.48181891e+00
9.48264748e-02 3.82276803e-01 -6.35902822e-01 -1.54143661e-01
5.99653363e-01 -1.17567569e-01 1.10829556e+00 -6.89009011e-01
1.08002865e+00 1.28858232e+00 4.26630825e-01 1.09483767e+00
-1.32518411e+00 -9.43852127e-01 -2.47318461e-03 1.99546650e-01
-1.31474161e+00 -7.91494191e-01 6.47985876e-01 -4.66461748e-01
9.16693270e-01 2.14180261e-01 6.44775927e-01 1.28378236e+00
-2.19280243e-01 6.06884062e-01 1.10010612e+00 -3.51067036e-01
1.50082439e-01 7.75981247e-02 -3.71110886e-01 6.41166806e-01
-6.01538643e-02 -1.73491821e-01 -6.61085665e-01 -6.11323640e-02
6.23169303e-01 -2.52252460e-01 9.53279734e-02 -2.42437124e-01
-7.86334872e-01 6.77260280e-01 1.68511167e-01 6.37817860e-01
-3.16371858e-01 1.70086876e-01 7.48384655e-01 3.10289979e-01
5.50468564e-01 4.09435213e-01 -4.50073808e-01 -3.69851679e-01
-9.14606035e-01 1.77495778e-02 6.82852924e-01 5.57057381e-01
7.91750908e-01 -1.54944345e-01 -4.29546423e-02 8.90448093e-01
7.29516670e-02 1.36814728e-01 6.20908439e-01 -9.77961659e-01
6.06717169e-03 8.40887904e-01 5.40483817e-02 -8.85667443e-01
-8.77785325e-01 3.11681330e-01 -5.12619138e-01 5.97547531e-01
5.50562501e-01 -1.42342165e-01 -7.61913955e-01 1.98498356e+00
5.36809206e-01 1.27652764e-01 -1.58826515e-01 8.36223483e-01
9.06908631e-01 1.06310822e-01 3.04993451e-01 -4.24615204e-01
1.52226293e+00 -6.03146493e-01 -7.96557069e-01 -6.15480766e-02
1.10183382e+00 -8.26072097e-01 7.38300860e-01 4.86748040e-01
-6.97362959e-01 -4.26727533e-01 -7.91963100e-01 9.89848562e-03
-3.62108171e-01 4.58159924e-01 7.64948368e-01 7.26219654e-01
-7.41042912e-01 4.66160506e-01 -7.18256712e-01 -4.78158981e-01
8.16344082e-01 7.05669284e-01 -9.43697512e-01 1.70464486e-01
-8.66734743e-01 1.06006539e+00 4.31073338e-01 -2.43683404e-04
-7.26997614e-01 -2.88778335e-01 -7.83107996e-01 -3.62893373e-01
6.05899990e-01 -9.26555991e-02 1.15538108e+00 -1.46439648e+00
-1.48508275e+00 1.28325915e+00 -1.41597822e-01 -8.63548443e-02
5.02941370e-01 1.67297751e-01 -3.77305388e-01 1.36449099e-01
-1.05660595e-02 1.05252445e+00 9.45028663e-01 -1.02729893e+00
-6.57167554e-01 -3.62091780e-01 -2.93029174e-02 9.05282497e-02
-2.33256713e-01 5.12213588e-01 -2.12592155e-01 -4.10204053e-01
-8.58618543e-02 -1.03293240e+00 -9.88208968e-03 1.18752336e-02
2.40282733e-02 -5.10593832e-01 7.29457557e-01 -4.69667375e-01
1.06930494e+00 -2.13591456e+00 2.54169047e-01 -1.16151243e-01
2.96390623e-01 2.25575894e-01 -1.50788784e-01 5.80676831e-02
-2.20139712e-01 1.99027345e-01 2.73688771e-02 -5.88121772e-01
-1.36521354e-01 3.58372569e-01 2.45374098e-01 5.15079260e-01
2.16943651e-01 7.83939481e-01 -7.94313669e-01 -8.97949517e-01
2.10776076e-01 4.54409540e-01 -4.28139806e-01 5.85437976e-02
-1.82493329e-01 5.64038336e-01 -2.50355870e-01 1.03509653e+00
3.86616766e-01 1.31770581e-01 3.56607378e-01 -1.65864423e-01
1.51261419e-01 -6.13899343e-02 -1.34155023e+00 1.58307528e+00
-5.55038095e-01 6.04831815e-01 1.01607166e-01 -9.06420469e-01
9.06726122e-01 5.84201515e-01 8.99506390e-01 -4.45022404e-01
5.89521945e-01 3.24683189e-01 -2.33768765e-02 -5.62984109e-01
4.85213138e-02 -6.10652089e-01 -1.75899819e-01 6.12958789e-01
4.62054372e-01 -5.28266393e-02 3.72004837e-01 -2.97435611e-01
1.00205350e+00 3.14073235e-01 4.53120738e-01 1.25664577e-01
5.16230524e-01 -3.18476707e-01 7.84121454e-01 2.18080103e-01
-4.43250537e-01 5.42347431e-01 8.61161411e-01 -6.90967321e-01
-6.70997977e-01 -3.70340914e-01 -1.40437216e-01 1.36158133e+00
-3.02460104e-01 -5.85651040e-01 -7.54431605e-01 -1.01460385e+00
-1.56364739e-01 1.06289946e-01 -1.03029323e+00 -1.35166407e-01
-3.10534567e-01 -7.57466733e-01 6.71309829e-01 4.35605973e-01
7.90732428e-02 -1.39823222e+00 -6.86921895e-01 1.53424546e-01
-3.56128514e-01 -1.26297414e+00 -1.75720409e-01 2.60837078e-01
-4.07861024e-01 -1.24657476e+00 -3.92922014e-01 -6.66077077e-01
6.58799231e-01 -2.30021387e-01 1.04504144e+00 -2.59652617e-03
-6.99674964e-01 -2.94973403e-02 -5.00283241e-01 -7.27870822e-01
-5.11349857e-01 -1.44730806e-01 2.78481454e-01 3.99312019e-01
6.99713469e-01 -2.08225638e-01 -3.04451674e-01 5.43436348e-01
-8.04714203e-01 -1.95707321e-01 5.65514386e-01 7.29713798e-01
5.31729400e-01 -2.94634134e-01 6.39403939e-01 -7.97400475e-01
3.28646392e-01 -2.08428875e-01 -4.97578502e-01 1.88822076e-01
-2.86183625e-01 -1.46443710e-01 3.32230210e-01 -6.63875639e-01
-9.02812719e-01 6.94111645e-01 -1.44868940e-01 -3.02867711e-01
-4.47722614e-01 1.76312730e-01 -3.28249931e-01 -4.02340651e-01
6.52661502e-01 -3.15379828e-01 1.96376994e-01 -4.67031151e-01
2.89103389e-01 7.26139605e-01 1.77038729e-01 -1.88389808e-01
1.10836148e-01 5.58007360e-01 2.18522996e-01 -8.11302304e-01
-8.74050975e-01 -5.83372235e-01 -1.20360494e+00 -5.86822391e-01
9.58894968e-01 -8.21782053e-01 -6.78907454e-01 3.96109819e-01
-1.01779449e+00 -1.52947590e-01 -1.45912394e-01 5.06561697e-01
-7.76833832e-01 2.38851473e-01 -3.58536869e-01 -6.74934089e-01
-8.75526592e-02 -1.28403175e+00 1.29453266e+00 -6.09255135e-02
-7.41648436e-01 -7.59730458e-01 2.03529716e-01 5.84554136e-01
4.26648036e-02 5.35945535e-01 3.01985830e-01 -6.73762262e-01
7.40981475e-02 -5.32208860e-01 -5.91981485e-02 4.57470119e-01
3.71047556e-01 1.92633510e-01 -1.11379147e+00 -3.05790454e-02
-1.43659770e-01 -9.76005256e-01 7.52374232e-01 1.94833025e-01
9.41448450e-01 -7.14519545e-02 -3.87696713e-01 1.92426413e-01
9.25375462e-01 2.79592350e-02 4.44022179e-01 -7.69126415e-03
5.27629197e-01 1.02517140e+00 8.17707300e-01 6.42932653e-01
-9.14804488e-02 1.09194040e+00 4.10686821e-01 -1.69131458e-01
-1.46511095e-02 2.58910954e-01 3.69244426e-01 1.05545983e-01
-5.46456754e-01 -4.90315929e-02 -7.40677357e-01 2.75699317e-01
-1.79697621e+00 -1.16297376e+00 -1.08791515e-02 1.92888474e+00
9.42906260e-01 -8.92586783e-02 5.46430945e-01 1.72056228e-01
6.38153732e-01 -5.26478253e-02 -3.00557137e-01 -5.12031913e-01
-4.94043082e-02 2.16492280e-01 1.81318313e-01 2.74803311e-01
-1.22849214e+00 1.15945923e+00 6.87834978e+00 7.79313207e-01
-1.37296772e+00 2.38743171e-01 3.58258516e-01 -4.06454831e-01
3.09337437e-01 -2.73280323e-01 -8.57272565e-01 4.35354024e-01
9.19587851e-01 9.56080928e-02 3.67971547e-02 7.80126333e-01
2.01822832e-01 -1.76611707e-01 -1.10363424e+00 1.34647727e+00
4.37771708e-01 -8.58157575e-01 -1.47667274e-01 1.26732484e-01
5.96551418e-01 -1.05414912e-01 -3.31238031e-01 7.70008415e-02
1.24930397e-01 -1.10187924e+00 4.03454393e-01 2.64167100e-01
9.20281351e-01 -6.98800683e-01 9.26063359e-01 5.41468635e-02
-1.05432594e+00 5.41632921e-02 -2.77135730e-01 -1.04665443e-01
9.81718525e-02 1.84580266e-01 -9.32918429e-01 1.86748043e-01
5.83600402e-01 6.68564975e-01 -5.35940945e-01 7.34714091e-01
-3.46968770e-01 2.69212008e-01 -3.88013631e-01 -1.86094955e-01
7.53737316e-02 -1.58080772e-01 8.73187706e-02 1.01272702e+00
1.20344453e-01 -2.21594721e-02 2.75949031e-01 3.59483361e-01
-2.10309520e-01 5.84955454e-01 -5.42462051e-01 -2.03737676e-01
-7.98825826e-03 1.71821809e+00 -9.47916567e-01 -2.04173759e-01
-6.22063339e-01 1.18577600e+00 4.92680043e-01 -2.29949087e-01
-7.75915265e-01 1.32147029e-01 7.53083408e-01 -6.74534664e-02
-2.17282586e-02 6.34480193e-02 8.72256532e-02 -9.31131899e-01
-8.50392133e-02 -8.79467905e-01 6.33362532e-01 -3.73316884e-01
-9.38907266e-01 5.60830414e-01 -8.67852718e-02 -1.41233039e+00
-5.14241278e-01 -7.85299480e-01 -2.22758889e-01 4.62734938e-01
-1.00810874e+00 -1.43297446e+00 -3.84158671e-01 4.81750011e-01
3.10916126e-01 -1.88103795e-01 1.12383437e+00 5.24252295e-01
-6.20419741e-01 6.58546686e-01 -3.52430522e-01 5.97345494e-02
1.25887632e+00 -1.07498336e+00 -3.18772197e-01 2.87204504e-01
5.36210895e-01 8.95045921e-02 7.14355111e-01 -4.22306359e-01
-8.38715196e-01 -7.98622310e-01 9.04272318e-01 -7.17672229e-01
6.33367598e-01 -5.92980564e-01 -7.50952423e-01 6.81523144e-01
-2.15364680e-01 2.43960127e-01 1.33135176e+00 2.30261624e-01
-4.91554171e-01 -6.04339801e-02 -1.20077407e+00 3.71102959e-01
9.64329958e-01 -5.16478479e-01 -2.67829895e-01 4.20015454e-01
-1.22957721e-01 -2.64170051e-01 -8.67865264e-01 3.30921650e-01
9.36471105e-01 -1.00383651e+00 5.24565518e-01 -6.76821470e-01
2.51094997e-01 -1.32550851e-01 8.20879042e-02 -1.18361795e+00
-1.17087858e-02 -3.70598555e-01 1.52824894e-01 1.35059655e+00
4.90494162e-01 -2.88168818e-01 8.85803223e-01 6.39381170e-01
1.98967859e-01 -5.30989707e-01 -1.22672319e+00 -6.30873859e-01
-2.62772828e-01 -4.14378583e-01 3.52421403e-01 9.92217541e-01
2.18519822e-01 4.03189987e-01 -6.14080667e-01 -4.14075971e-01
2.51365155e-01 8.09673220e-02 8.11814904e-01 -1.41949487e+00
4.37687375e-02 -6.62474811e-01 -9.95175302e-01 -1.96838617e-01
5.32962084e-01 -8.98048878e-01 -7.29081482e-02 -1.23555207e+00
2.61464864e-01 -1.60493881e-01 -2.95489192e-01 1.12105131e+00
9.69904736e-02 9.66119409e-01 5.97533584e-02 1.24275424e-01
-6.75776601e-01 2.19584793e-01 1.01567209e+00 4.20657992e-02
-1.98113352e-01 -9.23558995e-02 -4.30096775e-01 9.03304696e-01
8.48694026e-01 -5.17907619e-01 -2.18662456e-01 1.40395001e-01
2.14319661e-01 -3.11037034e-01 9.08786952e-02 -8.00797403e-01
-2.10132882e-01 -2.29811177e-01 4.26988930e-01 -2.35643044e-01
6.06047809e-01 -8.94010007e-01 6.79817647e-02 3.54376525e-01
-1.40675575e-01 -4.16121870e-01 1.73701271e-01 1.64044216e-01
-4.49187905e-01 -3.24734628e-01 1.03099561e+00 -2.70211846e-01
-8.61439645e-01 3.13530594e-01 -3.18958044e-01 -2.11430714e-01
1.42918706e+00 -1.21027090e-01 1.74515456e-01 -3.37227315e-01
-1.12571478e+00 -9.68959741e-03 4.86581922e-01 6.99914038e-01
2.09277168e-01 -1.35487390e+00 -6.58791065e-01 1.12921029e-01
4.58538204e-01 -4.38629061e-01 2.84055710e-01 8.56284499e-01
-2.58202136e-01 1.89368591e-01 -6.82961047e-01 -4.72325057e-01
-2.03314185e+00 5.03753006e-01 2.96511114e-01 -1.17000036e-01
-1.23523772e-01 8.21807861e-01 -6.96173385e-02 -2.90258110e-01
1.82103321e-01 4.21710424e-02 -6.86673105e-01 1.08019590e+00
6.64990723e-01 2.32903883e-01 1.77500710e-01 -1.15027893e+00
-5.47059178e-01 5.76900005e-01 9.85028520e-02 6.15114272e-02
1.22990465e+00 5.48918881e-02 -2.82950670e-01 5.27460635e-01
1.33596623e+00 -1.05633229e-01 -9.21019554e-01 2.16544554e-01
2.16374150e-03 -3.50834757e-01 -1.26993150e-01 -6.29426301e-01
-1.01090169e+00 8.21486473e-01 6.43734694e-01 -8.17603692e-02
1.16651320e+00 3.55692565e-01 3.20247054e-01 1.22040205e-01
5.27343094e-01 -1.44046664e+00 5.10992296e-02 4.86786030e-02
9.63337660e-01 -1.70904517e+00 7.28820041e-02 -5.25616527e-01
-7.52393365e-01 1.13291514e+00 7.56211519e-01 3.95727634e-01
4.73652899e-01 3.08216155e-01 5.04058480e-01 -2.66274631e-01
-7.11333275e-01 -5.64185917e-01 2.84216523e-01 4.78947520e-01
6.34745181e-01 1.43794581e-01 -6.67090118e-01 4.67572957e-01
-8.06216989e-03 5.51716276e-02 3.47030789e-01 9.61155593e-01
-2.43092939e-01 -1.59592521e+00 -1.48423225e-01 4.84492809e-01
-7.78651834e-01 3.82803202e-01 -9.72575486e-01 6.70470834e-01
5.18103421e-01 8.11098695e-01 -1.82229951e-01 -3.56056303e-01
2.70501435e-01 4.02831405e-01 6.84982479e-01 -7.55121469e-01
-4.20227051e-01 2.42536575e-01 3.77403706e-01 -8.52670431e-01
-9.55876052e-01 -9.73163068e-01 -1.19970703e+00 1.23510242e-01
-3.83168042e-01 -2.31815264e-01 5.76566458e-01 8.93398046e-01
1.57913372e-01 8.01269710e-02 5.48983037e-01 -1.01995778e+00
-1.05516024e-01 -1.20146620e+00 -6.90252006e-01 9.28704262e-01
-3.86557169e-02 -1.13043547e+00 -4.09881294e-01 1.90700591e-01]
|
[13.571832656860352, 1.7895984649658203]
|
e12e1459-a52c-4632-ae38-095bd265ff5b
|
the-adapter-bot-all-in-one-controllable
|
2008.12579
| null |
https://arxiv.org/abs/2008.12579v2
|
https://arxiv.org/pdf/2008.12579v2.pdf
|
The Adapter-Bot: All-In-One Controllable Conversational Model
|
Considerable progress has been made towards conversational models that generate coherent and fluent responses by training large language models on large dialogue datasets. These models have little or no control of the generated responses and miss two important features: continuous dialogue skills integration and seamlessly leveraging diverse knowledge sources. In this paper, we propose the Adapter-Bot, a dialogue model that uses a fixed backbone conversational model such as DialGPT (Zhang et al., 2019) and triggers on-demand dialogue skills (e.g., emphatic response, weather information, movie recommendation) via different adapters (Houlsby et al., 2019). Each adapter can be trained independently, thus allowing a continual integration of skills without retraining the entire model. Depending on the skills, the model is able to process multiple knowledge types, such as text, tables, and graphs, in a seamless manner. The dialogue skills can be triggered automatically via a dialogue manager, or manually, thus allowing high-level control of the generated responses. At the current stage, we have implemented 12 response styles (e.g., positive, negative etc.), 8 goal-oriented skills (e.g. weather information, movie recommendation, etc.), and personalized and emphatic responses. We evaluate our model using automatic evaluation by comparing it with existing state-of-the-art conversational models, and we have released an interactive system at adapter.bot.ust.hk.
|
['Pascale Fung', 'Yejin Bang', 'Andrea Madotto', 'Zhaojiang Lin']
|
2020-08-28
| null | null | null | null |
['movie-recommendation']
|
['miscellaneous']
|
[-1.49322018e-01 4.68544990e-01 1.54289648e-01 -3.50049198e-01
-4.37510759e-01 -1.12838817e+00 8.94201159e-01 7.07207397e-02
-3.47741097e-01 8.37688148e-01 6.46528840e-01 -3.76367986e-01
6.89817145e-02 -1.03494918e+00 -2.00178444e-01 -1.77905008e-01
2.73109972e-01 9.47277188e-01 4.94029820e-01 -8.35359931e-01
3.54754746e-01 1.05731055e-01 -1.12080312e+00 6.96777165e-01
8.45630527e-01 4.01230067e-01 3.29549074e-01 1.22601950e+00
-6.35067999e-01 1.35447705e+00 -8.06056917e-01 -5.31406164e-01
-2.20204711e-01 -5.57941973e-01 -1.39081562e+00 -1.53105885e-01
-1.60483941e-02 -4.59220290e-01 -7.28116697e-03 4.79283333e-01
6.82546198e-01 3.46938282e-01 4.89516556e-01 -1.05764639e+00
-4.34567213e-01 1.14241147e+00 2.26572007e-01 -2.24080637e-01
1.11679387e+00 5.35392761e-01 8.31039071e-01 -5.08749902e-01
6.63673460e-01 1.32932842e+00 4.40184772e-01 1.07932949e+00
-1.25570929e+00 -3.32942575e-01 1.46681592e-01 -1.75035675e-05
-5.66738725e-01 -3.54746372e-01 5.84690750e-01 -5.04353166e-01
1.15488923e+00 6.38964832e-01 5.25066674e-01 1.41289747e+00
-6.79277852e-02 7.76502550e-01 1.27070224e+00 -4.04976368e-01
6.82068765e-02 7.29728580e-01 2.18273222e-01 4.17045504e-01
-6.82911515e-01 -3.53175491e-01 -6.43483281e-01 -2.32685983e-01
6.70676529e-01 -3.98730010e-01 -3.15825343e-01 2.63192624e-01
-1.20472717e+00 7.36688077e-01 2.95989234e-02 2.77218431e-01
-3.41030031e-01 -4.26467031e-01 5.00287056e-01 9.10017669e-01
3.82786602e-01 9.88543034e-01 -6.88900828e-01 -6.17236972e-01
-2.44740486e-01 4.96705055e-01 1.83548188e+00 1.05740201e+00
6.48109138e-01 -3.37923408e-01 -6.37883067e-01 1.16257358e+00
6.35173321e-02 3.70614648e-01 4.63349909e-01 -1.05102026e+00
5.99661052e-01 7.90313661e-01 4.33258772e-01 -7.87536621e-01
-5.27479708e-01 1.86371371e-01 -6.00759208e-01 -1.97051585e-01
5.35761654e-01 -9.13126111e-01 -2.62403786e-01 1.77724028e+00
4.55569744e-01 -2.23097339e-01 4.59641159e-01 5.98646998e-01
1.47611094e+00 8.21908116e-01 1.17261231e-01 -2.53000855e-01
1.18745041e+00 -1.17944574e+00 -7.33071566e-01 -1.91851065e-01
6.24981999e-01 -8.84170055e-01 1.45069361e+00 4.58502114e-01
-1.36137915e+00 -7.08524227e-01 -2.92189002e-01 3.08517981e-02
-4.37094808e-01 -8.00985321e-02 3.81339937e-01 3.55675131e-01
-1.25917661e+00 5.20321846e-01 -2.68696785e-01 -6.29890382e-01
-7.44763672e-01 2.16901869e-01 -2.09397331e-01 2.79466838e-01
-1.68665397e+00 1.01338446e+00 3.51308025e-02 -2.21105069e-01
-5.05259991e-01 -3.74024779e-01 -7.43498564e-01 2.63282843e-02
4.68236268e-01 -7.67737985e-01 1.82645452e+00 -8.13426912e-01
-2.35949302e+00 6.88172698e-01 3.68849844e-01 -1.01832822e-01
6.49936199e-01 -3.10206532e-01 -2.89851129e-01 1.38514251e-01
-8.35916921e-02 6.32091582e-01 2.73898333e-01 -8.95125866e-01
-5.41052401e-01 8.06149244e-02 8.07674766e-01 5.38413227e-01
-4.13129091e-01 4.63340431e-01 -3.15801919e-01 -2.31155187e-01
-7.06162989e-01 -9.86676097e-01 -2.42885679e-01 -7.38464653e-01
-4.40736175e-01 -5.77510059e-01 2.90447593e-01 -6.22764289e-01
1.54298735e+00 -1.67518413e+00 4.24690038e-01 -5.23137301e-02
-3.77496369e-02 4.04726624e-01 -2.29543686e-01 1.33852375e+00
3.92463088e-01 6.02385961e-02 1.08137012e-01 -4.27490324e-01
1.68081626e-01 1.18362606e-01 -7.92350918e-02 -4.17132109e-01
2.16758683e-01 7.18059480e-01 -1.01106393e+00 -4.13548172e-01
1.09226428e-01 8.27656388e-02 -6.33561373e-01 9.98065412e-01
-6.89154863e-01 6.96975172e-01 -4.34116185e-01 -1.90238990e-02
1.93421945e-01 -2.65700400e-01 3.94052058e-01 3.03747982e-01
-3.89010161e-01 7.64523268e-01 -1.06793582e+00 1.48741615e+00
-1.13746071e+00 2.66949713e-01 3.46493304e-01 -3.79513502e-01
1.09105515e+00 6.40943468e-01 -1.22024948e-02 -3.13340515e-01
-2.54343841e-02 -1.21019989e-01 1.40396833e-01 -7.76963115e-01
6.79147959e-01 2.34309331e-01 -4.85588044e-01 9.11417305e-01
7.39794075e-02 -4.05624509e-01 4.87180173e-01 6.84228837e-01
1.04970694e+00 7.42269233e-02 2.54361391e-01 4.86256182e-02
7.70605981e-01 1.47171929e-01 1.38548240e-01 8.61105680e-01
2.34188378e-01 2.56388158e-01 7.32454717e-01 -4.61753085e-02
-6.36520743e-01 -5.25821269e-01 4.78487909e-01 1.61271119e+00
-2.94213772e-01 -6.61777437e-01 -9.39561903e-01 -7.35594153e-01
-2.37352490e-01 7.98368156e-01 -3.89333785e-01 4.15210724e-02
-6.95167720e-01 -1.70900702e-01 5.26252389e-01 3.14927429e-01
6.65341020e-01 -1.76607704e+00 -4.72610682e-01 6.21798813e-01
-6.52621984e-01 -1.00301993e+00 -5.87854862e-01 -7.03125745e-02
-3.59764844e-01 -8.34951222e-01 -3.65313590e-01 -6.65851533e-01
3.38866323e-01 4.84901518e-02 1.48399568e+00 1.79544717e-01
2.22683907e-01 7.44989634e-01 -6.69374466e-01 -1.45070180e-01
-1.03326523e+00 2.40537420e-01 -1.54428378e-01 -1.16670234e-02
4.13477719e-02 -5.13651788e-01 -5.22215366e-01 6.78361237e-01
-6.52733982e-01 4.69091535e-01 2.61510670e-01 8.13422740e-01
-1.77173957e-01 -5.82211018e-01 9.64092970e-01 -1.44906318e+00
1.55306375e+00 -6.50896013e-01 -2.12790042e-01 4.15117323e-01
-3.22093308e-01 -1.94495529e-01 1.07010889e+00 -6.54725790e-01
-1.48863971e+00 -2.64437944e-01 -5.78649819e-01 3.45469624e-01
-5.45659542e-01 6.28181338e-01 4.20541875e-02 1.39662161e-01
8.82645726e-01 3.36251557e-01 2.74840891e-02 -4.75272745e-01
7.05877542e-01 1.17028010e+00 3.03114295e-01 -9.43075061e-01
2.49269381e-01 -3.89805585e-01 -7.62691319e-01 -6.38478518e-01
-4.96548772e-01 -5.34620106e-01 -4.10689175e-01 -5.51700115e-01
6.93692744e-01 -6.14951074e-01 -1.03938973e+00 6.21784270e-01
-1.22034633e+00 -1.13056874e+00 -6.64692149e-02 2.04399675e-01
-7.06683278e-01 2.16341168e-01 -1.20716751e+00 -8.15959156e-01
-7.05467761e-01 -8.27832818e-01 4.64895159e-01 5.09655297e-01
-9.44946110e-01 -1.18664014e+00 3.01911712e-01 4.37527925e-01
8.09380114e-01 -7.23135769e-02 8.25358689e-01 -1.10830200e+00
-1.11455850e-01 -6.38597310e-02 1.29172936e-01 3.04960579e-01
1.03383705e-01 2.01558888e-01 -6.52894795e-01 2.11629719e-02
-7.24855140e-02 -1.01561332e+00 1.24041811e-01 -3.56121153e-01
6.71768665e-01 -7.53881931e-01 3.25646326e-02 -1.07408732e-01
5.53311229e-01 2.32595101e-01 3.94159257e-01 7.93632045e-02
2.96341091e-01 1.04713678e+00 7.78687418e-01 4.04730201e-01
8.81221533e-01 6.85233712e-01 -1.36373444e-02 6.59077764e-02
6.72948081e-03 -4.07070369e-01 5.04945219e-01 1.09446585e+00
-1.36741456e-02 -3.72022063e-01 -7.76251495e-01 3.10497165e-01
-2.10906029e+00 -1.02056432e+00 -2.19835192e-01 1.96251261e+00
1.46661747e+00 9.90329981e-02 3.91301721e-01 -3.20654541e-01
4.41751778e-01 1.02680758e-01 -2.95224726e-01 -1.01179135e+00
3.73059243e-01 6.31606355e-02 -3.07927400e-01 9.71897960e-01
-4.99895781e-01 1.23715854e+00 5.63511992e+00 3.79480690e-01
-1.03448761e+00 -8.33603218e-02 3.81756395e-01 3.82396542e-02
-3.70946497e-01 -1.36593670e-01 -8.04056466e-01 4.52422589e-01
1.00345695e+00 -2.65616089e-01 6.43301904e-01 6.62782907e-01
2.88274378e-01 -6.82450011e-02 -1.11798990e+00 4.01961505e-01
-1.46258339e-01 -1.17197633e+00 -1.55116826e-01 -3.70218664e-01
3.29600990e-01 -2.26234645e-01 -4.48222935e-01 9.03837383e-01
1.05853105e+00 -6.47105455e-01 3.39937925e-01 6.41341984e-01
7.00473905e-01 -4.55071270e-01 6.23312235e-01 9.52664614e-01
-7.64598548e-01 3.07913553e-02 3.53255048e-02 -3.00263464e-01
2.63321579e-01 -7.81811848e-02 -1.20569825e+00 4.52717423e-01
4.66343462e-01 3.26454073e-01 -3.46106708e-01 4.19904888e-01
-5.45490444e-01 7.00754881e-01 -2.50151515e-01 -6.65843070e-01
1.15067616e-01 -2.74073184e-01 3.84719908e-01 1.56955826e+00
-6.69147223e-02 5.93221188e-01 5.92953980e-01 6.03582621e-01
1.61432624e-01 4.92959172e-01 -5.21267891e-01 3.72239165e-02
5.85245490e-01 1.54220057e+00 -1.88773051e-01 -5.93037188e-01
-4.30997789e-01 7.90045559e-01 5.50476909e-01 3.88680428e-01
-3.83212894e-01 -3.76219541e-01 3.65482748e-01 9.11870599e-02
-2.72406757e-01 -1.00997528e-02 2.36701500e-03 -1.06593585e+00
-3.78010601e-01 -1.42513692e+00 4.12201166e-01 -9.49837327e-01
-1.32098699e+00 7.68233180e-01 7.24284127e-02 -7.87641346e-01
-8.47471297e-01 -3.17554802e-01 -1.07986832e+00 1.14092469e+00
-8.97874773e-01 -9.24440920e-01 -2.91127294e-01 8.80313933e-01
8.47089052e-01 -1.13589391e-01 1.19636333e+00 2.89508495e-02
-5.45566380e-01 4.88789707e-01 -5.38762033e-01 1.99236423e-01
1.13504946e+00 -1.65516961e+00 5.95669270e-01 2.50666201e-01
-2.80079663e-01 7.78596699e-01 8.12378883e-01 -6.17918372e-01
-1.21456015e+00 -7.00965583e-01 1.18322802e+00 -5.33515930e-01
8.41786683e-01 -4.66360360e-01 -1.00859344e+00 4.91145134e-01
7.77510166e-01 -8.93315434e-01 8.87879312e-01 2.74371982e-01
-2.22223997e-02 1.26857311e-02 -1.06254506e+00 8.47049594e-01
8.16767395e-01 -3.12930375e-01 -6.14953101e-01 6.86254144e-01
9.49938118e-01 -6.81689620e-01 -9.45217788e-01 1.32694513e-01
2.86476254e-01 -1.11338055e+00 6.20508850e-01 -9.40567553e-01
6.31299555e-01 1.02362908e-01 3.23646367e-01 -1.76566672e+00
-2.54655242e-01 -1.04888225e+00 -2.33082715e-02 1.49747980e+00
7.12502778e-01 -7.82715857e-01 2.64100432e-01 9.57977891e-01
-1.82197064e-01 -6.35523081e-01 -1.75326392e-01 -1.81373447e-01
-3.34736481e-02 -8.87981579e-02 5.30927896e-01 9.62992907e-01
9.99667645e-01 1.02913117e+00 -6.76092923e-01 -2.21130624e-01
-2.07435802e-01 3.19370657e-01 1.35708547e+00 -1.09384596e+00
-7.96373546e-01 -4.16504204e-01 6.24281764e-01 -1.33776081e+00
6.21003509e-02 -5.25330901e-01 7.99903870e-02 -1.85848117e+00
-2.03727752e-01 -4.87359643e-01 2.80238897e-01 6.03482485e-01
-3.08555782e-01 -2.69100040e-01 2.52235502e-01 2.72900965e-02
-5.66357136e-01 2.84897804e-01 1.47631347e+00 2.30253905e-01
-5.83985269e-01 5.49138486e-01 -7.57978439e-01 5.40148973e-01
1.24248695e+00 -1.57009616e-01 -6.05147541e-01 -1.48758218e-01
3.69116068e-01 7.59827197e-01 -1.65576965e-01 -5.35497546e-01
5.69585383e-01 -3.81902039e-01 -2.25647628e-01 -2.97422502e-02
2.54458755e-01 -3.28892857e-01 4.71546009e-05 2.30844036e-01
-8.86758089e-01 2.08892018e-01 6.04392849e-02 1.79692715e-01
-2.30545506e-01 -2.13220105e-01 3.88458461e-01 -6.15952909e-01
-2.92285651e-01 -6.28943220e-02 -8.15221608e-01 1.02831841e-01
8.08189213e-01 1.40248910e-01 -6.80751860e-01 -9.79100049e-01
-9.16561484e-01 8.43025029e-01 2.52238989e-01 6.93226755e-01
3.27485144e-01 -8.65039706e-01 -8.56188595e-01 -8.47014338e-02
1.01642810e-01 -3.33109275e-02 2.60149032e-01 5.64357579e-01
-3.45492840e-01 2.32406721e-01 -2.12834373e-01 -2.81424791e-01
-1.38533258e+00 3.69463921e-01 3.09185535e-01 -6.97473228e-01
-3.39987814e-01 7.03306198e-01 -1.09951109e-01 -1.08576083e+00
2.57479399e-01 -2.19603851e-02 -9.33753312e-01 2.34823644e-01
6.36370003e-01 1.99395373e-01 -7.90774226e-02 -2.16071866e-02
9.95172039e-02 2.98905343e-01 -1.44622624e-01 -3.65091562e-01
1.01196361e+00 -1.31447136e-01 -3.20334136e-01 6.01947844e-01
4.34164971e-01 2.75008112e-01 -9.89789009e-01 -4.46748853e-01
-3.07690382e-01 -1.85625583e-01 -7.61220515e-01 -1.48211181e+00
-5.78259349e-01 7.53499806e-01 -3.48604321e-01 1.00663364e+00
8.50916982e-01 -7.12343156e-02 5.27043939e-01 5.41513324e-01
5.28776467e-01 -1.20162606e+00 3.61963660e-01 8.78845632e-01
1.24018109e+00 -9.98789251e-01 -5.02267540e-01 -4.12446171e-01
-1.22571099e+00 1.22738576e+00 1.24851441e+00 2.84483045e-01
2.16487750e-01 3.12361568e-01 6.52776182e-01 -1.80024691e-02
-1.52641904e+00 -5.22664264e-02 -9.12982151e-02 4.71637219e-01
6.26574159e-01 2.35491678e-01 -6.68620646e-01 5.71394920e-01
-4.28091764e-01 -1.28490746e-01 7.76025295e-01 7.27503777e-01
-4.69467223e-01 -1.38097942e+00 -2.04060912e-01 2.03455240e-01
-2.15923399e-01 -1.22541018e-01 -1.11464024e+00 4.90269750e-01
-4.04495507e-01 1.53050780e+00 -2.77564615e-01 -5.38146555e-01
7.93716967e-01 3.29269260e-01 6.33205026e-02 -1.12190449e+00
-1.58584380e+00 -1.59934610e-01 9.94240820e-01 -3.04423690e-01
-7.04643354e-02 -4.39914703e-01 -1.02002501e+00 -4.71763939e-01
-1.38330877e-01 5.48194528e-01 2.69996047e-01 7.22972035e-01
3.45989704e-01 3.34051251e-01 8.68889391e-01 -5.59799016e-01
-8.67488921e-01 -1.57090962e+00 -1.77921355e-02 4.01881248e-01
-1.43997073e-01 -7.74266198e-02 -3.46394747e-01 -1.93716288e-01]
|
[12.795272827148438, 8.075533866882324]
|
5d41a9e0-eda1-432b-a751-3ae8203ddf09
|
incomplete-multimodal-learning-for-complex
|
2305.16222
| null |
https://arxiv.org/abs/2305.16222v1
|
https://arxiv.org/pdf/2305.16222v1.pdf
|
Incomplete Multimodal Learning for Complex Brain Disorders Prediction
|
Recent advancements in the acquisition of various brain data sources have created new opportunities for integrating multimodal brain data to assist in early detection of complex brain disorders. However, current data integration approaches typically need a complete set of biomedical data modalities, which may not always be feasible, as some modalities are only available in large-scale research cohorts and are prohibitive to collect in routine clinical practice. Especially in studies of brain diseases, research cohorts may include both neuroimaging data and genetic data, but for practical clinical diagnosis, we often need to make disease predictions only based on neuroimages. As a result, it is desired to design machine learning models which can use all available data (different data could provide complementary information) during training but conduct inference using only the most common data modality. We propose a new incomplete multimodal data integration approach that employs transformers and generative adversarial networks to effectively exploit auxiliary modalities available during training in order to improve the performance of a unimodal model at inference. We apply our new method to predict cognitive degeneration and disease outcomes using the multimodal imaging genetic data from Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort. Experimental results demonstrate that our approach outperforms the related machine learning and deep learning methods by a significant margin.
|
['Paul M. Thompson', 'Li Shen', 'Heng Huang', 'Liang Zhan', 'Reza Shirkavand']
|
2023-05-25
| null | null | null | null |
['data-integration']
|
['knowledge-base']
|
[ 3.27210486e-01 -2.40191668e-02 -1.03279296e-02 -5.93011320e-01
-8.49842250e-01 -3.80290419e-01 3.57741475e-01 -8.37063566e-02
-5.57755828e-01 9.93499637e-01 5.89974038e-02 -4.53515530e-01
-1.86427772e-01 -7.49229193e-01 -6.93640530e-01 -6.01367533e-01
-2.22665295e-02 6.52219534e-01 -1.54989481e-01 8.13076049e-02
-2.73687482e-01 3.44230264e-01 -1.31341100e+00 3.46773148e-01
1.31193519e+00 9.98980343e-01 2.90608317e-01 4.14665461e-01
-3.01369373e-02 4.42890674e-01 -3.61393243e-01 -5.62768161e-01
2.47774333e-01 -3.14957947e-01 -5.24807215e-01 -1.95706010e-01
2.66299903e-01 -6.23879850e-01 -4.57761854e-01 1.32704568e+00
1.05626822e+00 -2.86269575e-01 6.19601607e-01 -1.25440621e+00
-6.41052365e-01 5.49076259e-01 -5.80348194e-01 2.43831754e-01
1.16495319e-01 3.50015044e-01 7.98060060e-01 -6.58385992e-01
6.05192482e-01 1.04116058e+00 5.75446486e-01 8.63294244e-01
-1.34077001e+00 -6.63097680e-01 -5.76915704e-02 5.12373328e-01
-1.04056859e+00 -4.02093321e-01 8.32381904e-01 -4.86606717e-01
4.50609237e-01 9.92408544e-02 7.82261193e-01 1.50316083e+00
9.87278372e-02 6.92939162e-01 1.22313464e+00 -1.27023652e-01
1.29978880e-01 -2.89330661e-01 2.11560130e-01 8.23043287e-01
7.02907071e-02 -4.74691391e-03 -2.72984564e-01 -5.67870259e-01
6.61661625e-01 2.87094295e-01 -3.10358971e-01 5.93635738e-02
-1.45394802e+00 7.15682626e-01 4.54424441e-01 1.03890724e-01
-5.43641508e-01 -2.44526282e-01 3.70201200e-01 4.51622486e-01
3.27129126e-01 5.28687052e-02 -3.43072206e-01 2.77553767e-01
-9.39869821e-01 2.28901789e-01 4.85884011e-01 4.99279588e-01
3.81687969e-01 -2.10256323e-01 2.94772033e-02 1.15875649e+00
3.26924175e-01 6.10765398e-01 8.71716380e-01 -7.77353525e-01
5.56671619e-01 7.77439117e-01 -4.51184124e-01 -6.25691414e-01
-7.70320952e-01 -3.43984991e-01 -1.32673371e+00 9.25249681e-02
5.51761150e-01 -4.41463351e-01 -1.08317542e+00 1.86030185e+00
2.19770014e-01 7.55917802e-02 -1.59836398e-03 9.67924297e-01
8.34769964e-01 8.49221796e-02 1.71755910e-01 8.88613090e-02
1.45469260e+00 -4.09102082e-01 -4.92781401e-01 -1.76370665e-01
6.10233605e-01 -2.14076534e-01 9.70455229e-01 2.64153153e-01
-1.00310981e+00 -2.96641678e-01 -8.81679356e-01 -4.98196185e-02
-3.80115747e-01 1.83685794e-01 7.65898287e-01 6.12284899e-01
-9.80600059e-01 1.12258628e-01 -9.76614058e-01 -1.60573229e-01
9.82330680e-01 5.34845173e-01 -6.66619778e-01 -5.20516038e-01
-1.33818448e+00 8.81894112e-01 2.39779711e-01 3.80720854e-01
-1.05357552e+00 -9.34645236e-01 -6.01060569e-01 -1.31949484e-01
2.85211205e-01 -1.08066094e+00 7.19389021e-01 -7.22157240e-01
-9.94463444e-01 7.77549148e-01 3.23325954e-02 -4.39394265e-01
5.26617587e-01 -6.04229700e-03 -4.74327505e-01 4.00333405e-01
-7.91907869e-03 9.60462511e-01 5.09420335e-01 -7.92889237e-01
-3.16533566e-01 -8.99090648e-01 -9.92667601e-02 -2.08328031e-02
-4.11584795e-01 2.17427053e-02 -1.69432927e-02 -5.22956550e-01
-5.98676242e-02 -8.57845843e-01 -2.81784564e-01 2.45493561e-01
-6.97202802e-01 4.84567210e-02 7.04557300e-01 -1.15886712e+00
6.08632982e-01 -1.78701031e+00 4.08054978e-01 1.73919454e-01
6.38732791e-01 2.57603109e-01 -2.71991283e-01 -1.92563504e-01
-1.33197829e-01 1.73154876e-01 -4.33598369e-01 -2.12219015e-01
-6.94457293e-02 7.06262141e-02 -1.78447124e-02 2.90370315e-01
4.09812272e-01 1.09573686e+00 -5.37213922e-01 -4.92334992e-01
-7.30580790e-03 6.20754600e-01 -7.71756172e-01 1.87315241e-01
-1.44844338e-01 8.41232419e-01 -6.22014523e-01 1.05449963e+00
5.90089381e-01 -3.73411953e-01 1.52825877e-01 -2.70020306e-01
4.11475718e-01 -4.74563725e-02 -4.72391725e-01 1.76088607e+00
-1.09520428e-01 1.96847320e-01 -4.24155481e-02 -1.46498680e+00
5.25754690e-01 4.46875095e-01 5.79042017e-01 -7.02694118e-01
1.40099317e-01 1.53346345e-01 5.95179558e-01 -8.73200595e-01
-4.51847583e-01 -3.08018535e-01 1.53727993e-01 3.55699956e-01
9.05327052e-02 4.47222561e-01 8.52382258e-02 -7.43580163e-02
1.44815028e+00 -1.83468997e-01 -1.30385637e-01 2.40326777e-01
4.49185729e-01 -4.90718186e-02 8.58026206e-01 5.34162998e-01
-3.33076179e-01 5.85110247e-01 7.28096485e-01 -1.28479004e-01
-1.04584086e+00 -1.15357673e+00 -4.38774407e-01 7.35030413e-01
-5.29389560e-01 1.90163627e-01 -6.25600338e-01 -8.21251035e-01
-4.83422168e-02 2.33895048e-01 -6.50237203e-01 -3.58685553e-01
-2.71675497e-01 -1.47968233e+00 9.83267784e-01 5.92011094e-01
5.45921028e-01 -7.44277835e-01 -3.49209219e-01 3.12701724e-02
-2.15047464e-01 -1.04561830e+00 -1.90973282e-01 -3.42404768e-02
-1.05667984e+00 -1.26118290e+00 -1.17857301e+00 -6.24270976e-01
9.86858070e-01 -3.12063128e-01 6.88560128e-01 -5.83571717e-02
-7.22180247e-01 2.70281583e-01 -2.38353923e-01 -2.91685253e-01
-3.79487276e-01 1.32121369e-02 1.85808986e-02 1.24347627e-01
2.72618413e-01 -7.21403301e-01 -8.00846815e-01 1.47565315e-02
-1.08475780e+00 1.25537634e-01 1.01071930e+00 1.09163702e+00
4.57160085e-01 -3.95823345e-02 1.08001614e+00 -6.98511839e-01
5.95285833e-01 -8.99769723e-01 -3.94176304e-01 5.25695920e-01
-3.50650221e-01 -2.05334555e-02 5.15058279e-01 -6.31421208e-01
-1.06173980e+00 -8.27679783e-02 -2.88433641e-01 -4.79485244e-01
-4.85668063e-01 1.13347650e+00 -5.27249873e-01 -2.82253288e-02
2.79378235e-01 1.87354729e-01 3.65933478e-01 -6.44766271e-01
2.08189785e-01 5.99561036e-01 5.77192068e-01 -5.78347027e-01
3.48560035e-01 3.32784086e-01 1.97282299e-01 -5.43164551e-01
-5.11132598e-01 3.86107564e-02 -7.01809883e-01 -1.09474555e-01
8.97332549e-01 -8.74310911e-01 -5.27640522e-01 6.53712809e-01
-9.50265110e-01 -2.12080125e-02 2.51089543e-01 6.93916798e-01
-2.02914000e-01 3.32612723e-01 -4.90009934e-01 -2.39438429e-01
-4.80291337e-01 -1.44959962e+00 8.29418898e-01 8.63678828e-02
5.65338768e-02 -1.04811192e+00 6.07000999e-02 7.04500020e-01
2.85285860e-01 3.48000646e-01 1.47586894e+00 -8.13414156e-01
-4.58332211e-01 -3.61498863e-01 -3.83932203e-01 3.97274941e-01
1.10171661e-01 -4.11834985e-01 -6.68666065e-01 -7.11808577e-02
-1.21467873e-01 -6.59705341e-01 8.83143365e-01 3.34209412e-01
1.35621297e+00 8.68700370e-02 -2.60852903e-01 4.86144215e-01
1.05449986e+00 2.09923610e-01 6.23863697e-01 3.40032689e-02
8.38236928e-01 6.79285944e-01 -8.75645205e-02 2.60607511e-01
7.06882894e-01 4.74660516e-01 3.92191350e-01 -1.53660789e-01
-1.64334383e-02 1.79370254e-01 3.18884104e-01 6.25418484e-01
-1.72701851e-01 -4.27251421e-02 -1.21473193e+00 4.29120660e-01
-1.64854860e+00 -8.56895506e-01 1.22609735e-01 2.17733026e+00
1.19791198e+00 -2.19568938e-01 1.54545486e-01 -4.46891338e-02
7.52047539e-01 -4.05350029e-01 -1.00044870e+00 3.20016086e-01
-3.17494720e-01 3.39256674e-01 1.25518218e-01 -1.79751754e-01
-8.58458102e-01 3.14865738e-01 6.35133266e+00 2.92362332e-01
-1.10313821e+00 3.95326346e-01 6.97214484e-01 -4.15201783e-01
-2.70389944e-01 -4.05385435e-01 -4.37329829e-01 7.11476922e-01
1.07978511e+00 1.06721751e-01 4.07462597e-01 2.54732043e-01
9.24853832e-02 -4.93552648e-02 -1.32551503e+00 9.00701582e-01
-1.17271438e-01 -1.01697838e+00 7.02212155e-02 2.48007074e-01
3.10690194e-01 2.29095578e-01 1.89580679e-01 2.38337219e-01
9.29739922e-02 -1.13047636e+00 1.19022682e-01 1.06162524e+00
6.84597135e-01 -6.81553602e-01 7.49099493e-01 3.59710813e-01
-5.25653660e-01 -4.00811806e-02 -2.10240662e-01 3.81661147e-01
1.24708645e-01 9.42078531e-01 -7.32427537e-01 5.42009592e-01
6.29942834e-01 5.03620863e-01 -6.68724835e-01 1.26134849e+00
1.09571986e-01 5.77344000e-01 -1.64396748e-01 3.37320119e-01
-2.61186630e-01 -1.25826403e-01 4.16738927e-01 6.25177443e-01
5.61933935e-01 5.49088456e-02 1.18337512e-01 1.19937527e+00
-3.75868678e-01 -9.63349417e-02 -4.39667523e-01 -3.53620261e-01
2.49565244e-01 1.16343248e+00 -3.33817333e-01 -1.33402959e-01
-7.95615673e-01 6.38392866e-01 4.51648325e-01 2.30381936e-01
-8.32617819e-01 -1.74325973e-01 4.30357665e-01 -2.43453030e-02
-1.35170236e-01 -1.17927238e-01 -1.87388524e-01 -1.35878217e+00
7.32023492e-02 -1.01256931e+00 4.08356756e-01 -6.37850165e-01
-1.75402296e+00 4.31035310e-01 -1.03545278e-01 -9.52273786e-01
-3.00091982e-01 -6.60599709e-01 -5.29022157e-01 1.06506896e+00
-1.31426179e+00 -1.28842485e+00 -6.42878786e-02 7.89330125e-01
-1.41966164e-01 -6.50607586e-01 7.94766665e-01 7.43852198e-01
-9.53248084e-01 6.03449821e-01 5.57940975e-02 3.64293486e-01
8.68431270e-01 -1.01936769e+00 -2.11960927e-01 5.14493227e-01
-3.91205132e-01 7.51675546e-01 2.13003263e-01 -8.79833639e-01
-1.50833011e+00 -9.97828364e-01 4.10836697e-01 3.10553294e-02
7.87543774e-01 -1.60204787e-02 -1.02609801e+00 6.22018933e-01
9.47190099e-04 1.21054605e-01 1.18799031e+00 1.12250865e-01
-2.65689403e-01 -8.32931623e-02 -1.31000531e+00 5.87849557e-01
9.07918811e-01 -6.48742676e-01 -6.80572033e-01 2.01134890e-01
3.70367557e-01 -4.54530567e-02 -1.25316858e+00 6.71226859e-01
4.80777979e-01 -5.22494376e-01 1.01938486e+00 -1.01685250e+00
6.16168916e-01 -1.83928981e-01 -7.98351243e-02 -1.37523544e+00
1.69181973e-01 9.68247186e-04 -1.07005186e-01 1.21320903e+00
4.41825420e-01 -8.61799538e-01 5.88341832e-01 1.11925483e+00
2.55478360e-02 -6.81070030e-01 -1.08723474e+00 -4.16691363e-01
2.73818552e-01 -3.89101267e-01 6.35668993e-01 1.00657046e+00
-7.28643760e-02 4.92752418e-02 -2.66267240e-01 2.38274679e-01
8.50319088e-01 -6.08318299e-02 2.03678563e-01 -1.37424314e+00
-1.73246577e-01 -4.76690114e-01 -5.82391560e-01 -3.25444192e-01
4.23851460e-01 -1.36344898e+00 -4.29034889e-01 -1.47469425e+00
6.02346301e-01 -5.11291265e-01 -5.93433201e-01 7.95665979e-01
-2.50316471e-01 2.53828615e-01 -5.55405803e-02 2.64464598e-02
-1.86954454e-01 6.98554933e-01 1.34945941e+00 -4.46680754e-01
5.11880927e-02 -1.16307929e-01 -8.74453247e-01 6.11495554e-01
7.94474721e-01 -3.12171310e-01 -4.79674190e-01 -4.96787280e-01
7.95957670e-02 2.48098746e-01 1.11832952e+00 -9.06233907e-01
2.70831168e-01 2.50249326e-01 8.51078928e-01 -4.58725095e-01
2.41426423e-01 -7.19512820e-01 1.08433537e-01 2.22364813e-01
-3.37797821e-01 -1.46257833e-01 1.00894116e-01 4.76197511e-01
-1.96607649e-01 -1.20031379e-01 5.02019227e-01 -3.86756398e-02
-3.38625759e-01 5.63750625e-01 -3.00716102e-01 8.47084820e-02
7.69165754e-01 2.39321992e-01 -5.99626243e-01 -1.91730110e-03
-1.24358511e+00 3.53934199e-01 8.04561526e-02 3.05508673e-01
7.47113109e-01 -1.45967746e+00 -8.36891651e-01 2.12354124e-01
-1.24254175e-01 -1.23461068e-01 5.87900996e-01 1.33531988e+00
-8.52829814e-02 1.59656748e-01 -7.60177970e-01 -4.76972669e-01
-1.04368985e+00 2.38245979e-01 3.33641708e-01 -1.21186316e-01
-5.11704504e-01 4.82841253e-01 1.76445380e-01 -6.35141730e-01
1.70798942e-01 -6.97194710e-02 -1.97830871e-01 1.72743350e-01
5.98468721e-01 2.57631153e-01 1.69144928e-01 -3.86247039e-01
-2.58544296e-01 8.18822756e-02 -2.98352718e-01 -1.24136247e-01
1.51952982e+00 5.08601405e-02 -3.54014486e-01 2.55870581e-01
1.14765728e+00 -4.67533350e-01 -8.49025309e-01 -3.30410510e-01
-1.84516221e-01 -3.18418778e-02 3.53906602e-01 -1.13956797e+00
-1.56899667e+00 1.11850381e+00 9.51903045e-01 -7.00796628e-03
1.34186637e+00 1.03938028e-01 8.74107778e-01 4.47400928e-01
3.78757775e-01 -8.39445233e-01 -3.38589132e-01 1.57608628e-01
8.43317151e-01 -1.36182320e+00 -2.02952057e-01 -4.55660336e-02
-4.91460681e-01 1.06567442e+00 5.19237638e-01 3.12518477e-01
8.09350669e-01 2.88390815e-01 -5.59987128e-02 -1.96184829e-01
-7.19202340e-01 -1.53340667e-01 5.67557335e-01 6.50488615e-01
3.88695985e-01 1.02230096e-02 -1.95540145e-01 1.13470638e+00
1.96234211e-01 1.69103950e-01 1.39632314e-01 7.88725197e-01
-9.61602032e-02 -1.57333720e+00 -5.51611423e-01 1.10561275e+00
-7.44917631e-01 -2.06772164e-01 -3.20973366e-01 5.18753827e-01
2.70156473e-01 6.20775104e-01 -1.14424698e-01 -2.42377490e-01
1.28072932e-01 4.44970012e-01 5.99567950e-01 -2.76524782e-01
-5.26825115e-02 1.92032959e-02 -1.89313907e-02 -3.24721545e-01
-5.79203844e-01 -9.70192313e-01 -1.27084994e+00 -1.98655307e-01
6.97348360e-03 -4.92373705e-01 6.22185707e-01 1.19003820e+00
5.47951102e-01 7.84023046e-01 2.63904691e-01 -8.00412178e-01
-4.15850699e-01 -9.56049144e-01 -3.78320336e-01 1.49202749e-01
2.17797920e-01 -8.57150316e-01 1.18725888e-01 3.32758218e-01]
|
[14.231600761413574, -1.6506675481796265]
|
c751600b-5b45-4235-a8ae-8c1dfab8a83b
|
beyond-classification-financial-reasoning-in
|
2305.01505
| null |
https://arxiv.org/abs/2305.01505v2
|
https://arxiv.org/pdf/2305.01505v2.pdf
|
Beyond Classification: Financial Reasoning in State-of-the-Art Language Models
|
Large Language Models (LLMs), consisting of 100 billion or more parameters, have demonstrated remarkable ability in complex multi-step reasoning tasks. However, the application of such generic advancements has been limited to a few fields, such as clinical or legal, with the field of financial reasoning remaining largely unexplored. To the best of our knowledge, the ability of LLMs to solve financial reasoning problems has never been dealt with, and whether it can be performed at any scale remains unknown. To address this knowledge gap, this research presents a comprehensive investigation into the potential application of LLMs in the financial domain. The investigation includes a detailed exploration of a range of subjects, including task formulation, synthetic data generation, prompting methods, and evaluation capability. Furthermore, the study benchmarks various GPT variants with parameter scales ranging from 2.8B to 13B, with and without instruction tuning, on diverse dataset sizes. By analyzing the results, we reveal that the ability to generate coherent financial reasoning first emerges at 6B parameters, and continues to improve with better instruction-tuning or larger datasets. Additionally, the study provides a publicly accessible dataset named sFIOG (Synthetic-Financial Investment Opinion Generation), consisting of 11,802 synthetic investment thesis samples, to support further research in the field of financial reasoning. Overall, this research seeks to contribute to the understanding of the efficacy of language models in the field of finance, with a particular emphasis on their ability to engage in sophisticated reasoning and analysis within the context of investment decision-making.
|
['Sol Jin', 'Keonju Na', 'Moonjeong Hahm', 'Hanearl Jung', 'Guijin Son']
|
2023-04-30
| null | null | null | null |
['synthetic-data-generation', 'synthetic-data-generation']
|
['medical', 'miscellaneous']
|
[-3.56589779e-02 4.39765304e-01 -5.02041221e-01 -3.80945683e-01
-8.02264810e-01 -6.88626885e-01 4.38361347e-01 2.60250479e-01
-2.30013713e-01 7.51435995e-01 1.52149811e-01 -1.09906554e+00
-4.00080681e-01 -9.29893851e-01 -5.45428038e-01 -2.46461451e-01
2.39507839e-01 8.00826609e-01 -1.45522803e-01 -3.83628458e-01
5.34359157e-01 3.92754674e-01 -1.20854998e+00 7.26695418e-01
8.45252633e-01 8.49347115e-01 -4.00035679e-01 5.72006047e-01
-3.36170316e-01 1.23859096e+00 -9.63114083e-01 -1.01213992e+00
2.42456689e-01 6.14219904e-03 -7.86588490e-01 -2.44472191e-01
1.79390654e-01 -4.42464232e-01 -6.05210923e-02 5.14699697e-01
7.79595375e-01 -6.47791699e-02 2.82161444e-01 -1.02423251e+00
-7.01658070e-01 9.14218187e-01 -3.67743105e-01 4.06060129e-01
5.11511981e-01 4.66028780e-01 1.13977587e+00 -5.09691358e-01
4.43165988e-01 1.36261475e+00 5.13185620e-01 4.93095934e-01
-8.83049011e-01 -7.63718188e-01 7.07603469e-02 -9.88417044e-02
-9.04560328e-01 -1.84044689e-01 5.25441229e-01 -4.48982269e-01
1.48681712e+00 -2.39064563e-02 5.18392384e-01 1.11267781e+00
6.18050277e-01 7.32392311e-01 1.43275523e+00 -3.68825257e-01
2.55839795e-01 3.76273185e-01 4.03595865e-01 6.46987259e-01
7.59185433e-01 -1.94059238e-02 -8.14283133e-01 -1.51770204e-01
4.20271277e-01 -3.72891456e-01 2.87991613e-01 3.11400682e-01
-8.75119388e-01 1.18027484e+00 1.04507115e-02 3.89131367e-01
-2.75417805e-01 9.95627120e-02 3.58160794e-01 5.10407388e-01
5.53415418e-01 7.49257088e-01 -5.81196070e-01 -5.98600507e-01
-1.04066932e+00 7.06185579e-01 1.26314914e+00 7.42204428e-01
1.57257274e-01 1.17974356e-01 -4.73856241e-01 4.18819875e-01
1.80906147e-01 4.22930479e-01 4.92422998e-01 -9.81509507e-01
1.05417085e+00 7.72963405e-01 -1.17783956e-02 -9.46735322e-01
-4.57140535e-01 -4.62736934e-01 -2.83881724e-01 5.49375936e-02
5.27663529e-01 -4.49996501e-01 -3.65338534e-01 1.47248769e+00
-9.03627947e-02 -2.75403410e-01 4.11138237e-01 6.12790585e-01
9.87019002e-01 3.26452583e-01 2.83090264e-01 2.49573842e-01
1.68497908e+00 -8.55656207e-01 -3.41123879e-01 -5.85607529e-01
8.72427404e-01 -7.29435921e-01 1.20088017e+00 5.91048479e-01
-1.38172281e+00 -4.24533516e-01 -8.08100224e-01 -7.10185021e-02
-4.84020531e-01 -2.56717540e-02 1.26868236e+00 1.16161895e+00
-9.79776382e-01 3.37540120e-01 -6.85010076e-01 -1.34979278e-01
4.63364989e-01 3.19709897e-01 1.65375024e-01 -2.48765230e-01
-1.49693668e+00 9.24332201e-01 3.63009959e-01 -1.57818422e-01
-3.28016043e-01 -9.56893086e-01 -5.89799523e-01 1.65759400e-01
5.51808834e-01 -1.05622661e+00 1.30570579e+00 -4.20967519e-01
-1.46887350e+00 6.69900894e-01 2.57823579e-02 -8.61993551e-01
7.03356862e-01 -2.67112195e-01 -2.86977410e-01 3.74615528e-02
7.25931674e-02 5.15448928e-01 2.49342278e-01 -5.22329688e-01
-3.59553814e-01 -1.56390533e-01 5.16514659e-01 1.15281321e-01
-3.38953286e-01 2.43488073e-01 -2.56864309e-01 -8.20930004e-01
-3.78352523e-01 -6.96260333e-01 -1.81090742e-01 -4.89300311e-01
-1.78872049e-01 -2.72217184e-01 7.73980692e-02 -6.48014605e-01
1.42889476e+00 -1.79397178e+00 -1.79210171e-01 5.93945868e-02
5.54600246e-02 9.50892791e-02 -4.55923304e-02 4.95381892e-01
-1.81875303e-02 5.64828753e-01 2.69662868e-02 -2.63842404e-01
2.69830346e-01 1.40014723e-01 -5.87382674e-01 -2.34295756e-01
4.16473746e-01 1.36639822e+00 -6.94272041e-01 -6.24593914e-01
-7.24244341e-02 1.59673512e-01 -8.70058119e-01 -1.34014368e-01
-3.70758504e-01 1.89228170e-02 -8.60720754e-01 1.04572582e+00
3.42818230e-01 -4.74544972e-01 1.34123489e-01 1.49681494e-01
1.51777089e-01 3.83243889e-01 -7.75009036e-01 1.31403339e+00
-4.17346507e-01 3.63013357e-01 -3.81575525e-01 -8.62619102e-01
8.73222947e-01 1.54455632e-01 4.40547645e-01 -1.04784286e+00
1.06277913e-01 3.30783963e-01 2.81659365e-01 -5.29938161e-01
7.09706306e-01 -3.48455131e-01 -3.43721867e-01 7.50377238e-01
-5.30400634e-01 -3.84181261e-01 6.69895649e-01 1.17424347e-01
1.16691685e+00 -4.00111228e-02 -6.89098658e-03 1.17810152e-01
3.14180195e-01 3.63196939e-01 2.14336485e-01 8.87510002e-01
-2.30720509e-02 1.19312569e-01 8.98674071e-01 -3.65518987e-01
-6.89105809e-01 -7.77180076e-01 -2.76519537e-01 9.34677303e-01
-4.46870387e-01 -3.73543173e-01 -8.42887163e-01 -6.11431301e-01
4.66528386e-01 1.06292379e+00 -4.19156313e-01 -9.72102582e-02
-5.69676101e-01 -1.28139091e+00 8.55667353e-01 5.09392083e-01
6.85952425e-01 -1.24534965e+00 -8.71444046e-01 2.28736103e-01
-8.02698284e-02 -1.36412716e+00 1.87258169e-01 -1.13094777e-01
-1.15535462e+00 -1.19885671e+00 -4.55328315e-01 -1.35566145e-01
3.34214479e-01 -2.63356447e-01 1.48786139e+00 7.95103684e-02
-1.52461842e-01 4.67190623e-01 -1.51954800e-01 -4.81616676e-01
-6.95579350e-01 3.30139518e-01 -4.48687524e-01 -7.15532124e-01
6.41767144e-01 -8.96378886e-03 -3.57097566e-01 1.67039126e-01
-9.46073532e-01 -6.96916655e-02 9.67040241e-01 7.17837274e-01
3.79543573e-01 3.93696785e-01 6.20014966e-01 -1.28141141e+00
1.32535565e+00 -6.91745520e-01 -2.38395348e-01 4.90952045e-01
-9.20397818e-01 1.82609618e-01 5.37056506e-01 -3.12623441e-01
-1.16663694e+00 -8.70298684e-01 -2.04183217e-02 6.35544062e-02
1.03223935e-01 1.01388645e+00 2.60397792e-01 -1.15889562e-02
5.03873527e-01 -4.86656167e-02 -3.57010551e-02 -1.91101223e-01
2.56952137e-01 4.27737892e-01 1.19059309e-01 -1.13357210e+00
3.59804749e-01 1.13060616e-01 5.95320165e-02 -3.93512547e-01
-8.56467962e-01 2.30022565e-01 -6.11854643e-02 1.26059622e-01
5.40226519e-01 -8.37562799e-01 -7.23469317e-01 4.28962320e-01
-7.38212645e-01 -6.39461577e-01 -2.18368188e-01 3.79364252e-01
-4.52838600e-01 3.42311561e-01 -1.01677847e+00 -6.17215514e-01
-4.97512162e-01 -1.40845311e+00 8.77373815e-01 1.52889296e-01
-6.26165926e-01 -1.28464365e+00 -2.48518437e-01 1.35020328e+00
4.34072286e-01 2.42689729e-01 1.34504366e+00 -8.73059809e-01
-5.88192999e-01 -3.17841679e-01 -7.17739388e-02 2.33216330e-01
-1.62435174e-01 -5.37249520e-02 -5.84191263e-01 -1.17937788e-01
1.60295367e-01 -5.26977360e-01 5.35394430e-01 2.14450791e-01
1.13683391e+00 -2.30997831e-01 -3.07269357e-02 2.37469897e-01
1.06199670e+00 2.95251340e-01 5.15476882e-01 6.56757295e-01
2.98408270e-01 7.53666580e-01 9.78605092e-01 5.16366541e-01
5.91216743e-01 4.24268186e-01 3.24140228e-02 2.55451351e-01
-2.20447145e-02 -6.86929524e-02 4.09574330e-01 3.41019809e-01
-1.66115239e-01 -4.23482150e-01 -1.30874050e+00 1.78662986e-02
-1.43916297e+00 -7.87545979e-01 1.76702023e-01 1.81571734e+00
8.40565860e-01 7.83655941e-01 -1.36886820e-01 1.47499844e-01
4.29790914e-02 5.33420295e-02 -7.75460601e-01 -9.08280849e-01
-1.92976370e-01 6.24512613e-01 4.75439996e-01 2.32440665e-01
-4.81882840e-01 9.78547454e-01 6.94456244e+00 8.66305351e-01
-1.05925322e+00 -1.39631897e-01 1.23286486e+00 -4.01143312e-01
-6.40263021e-01 -2.35763695e-02 -1.04499638e+00 5.49375176e-01
1.42490447e+00 -2.45405346e-01 3.92844796e-01 7.16013968e-01
2.43577361e-01 -2.47606382e-01 -1.07558084e+00 5.66454887e-01
3.07652969e-02 -1.29674911e+00 3.15171272e-01 1.80780038e-01
6.95377231e-01 -4.39411879e-01 4.06344026e-01 8.22404742e-01
3.89013499e-01 -1.36396515e+00 8.29798818e-01 6.09050155e-01
3.70046347e-01 -8.07411611e-01 1.11308563e+00 3.11053962e-01
-7.10483491e-01 -4.34190482e-01 -8.99620876e-02 -2.82255083e-01
7.98179805e-02 6.15447104e-01 -1.03519666e+00 6.39008582e-01
5.60653508e-01 4.31976110e-01 -7.60841370e-01 3.96516412e-01
-1.50148049e-01 6.87486529e-01 -1.01430163e-01 -2.29886606e-01
2.67815262e-01 1.63432360e-02 -5.35934381e-02 1.06594288e+00
2.85487264e-01 2.50677884e-01 1.14646442e-02 1.03949487e+00
2.35375185e-02 1.08582310e-01 -1.86103523e-01 -7.73728013e-01
3.95379245e-01 9.25809264e-01 -7.37417459e-01 -4.23007965e-01
-5.59254527e-01 3.63445044e-01 2.56217390e-01 3.08503330e-01
-9.91886914e-01 2.64536645e-02 4.89129573e-01 3.48271608e-01
-1.46375149e-01 -1.26499534e-01 -8.94022107e-01 -8.76674771e-01
1.52224466e-01 -1.38393641e+00 6.89703703e-01 -8.38022649e-01
-1.16992939e+00 2.80056745e-01 2.89130092e-01 -5.51324010e-01
-5.45683444e-01 -9.47957277e-01 -4.96657282e-01 1.02482128e+00
-1.60857630e+00 -8.99099529e-01 -8.65688920e-02 2.01645598e-01
4.45798486e-01 -5.36897182e-01 5.93348086e-01 3.34627718e-01
-8.02868128e-01 8.11432540e-01 -3.67545158e-01 -3.95305194e-02
6.18728936e-01 -8.75244439e-01 5.27484238e-01 3.74703884e-01
-3.38951200e-01 9.85630274e-01 4.65924978e-01 -8.59652162e-01
-1.52493858e+00 -7.73794293e-01 9.58165169e-01 -7.11075306e-01
9.18478310e-01 2.77556572e-02 -6.90144062e-01 4.48480606e-01
-1.08154945e-01 -4.73593652e-01 8.28118384e-01 -1.36253545e-02
-4.59950976e-02 6.77148998e-02 -1.29093933e+00 6.55303180e-01
6.99156702e-01 -5.14339149e-01 -5.34982562e-01 3.57410997e-01
7.33172596e-01 -7.86984146e-01 -1.28380072e+00 2.99631268e-01
3.79314423e-01 -1.06754148e+00 1.25230420e+00 -7.71479487e-01
9.59958553e-01 4.03697759e-01 -3.72192003e-02 -6.50274992e-01
2.28143726e-02 -4.10457850e-01 -2.77472764e-01 1.12055433e+00
5.94283700e-01 -1.02377176e+00 1.08233941e+00 1.14668119e+00
-1.13928124e-01 -1.52746928e+00 -6.00501776e-01 -4.23432440e-01
5.78688323e-01 -6.50767863e-01 9.36743021e-01 7.01296747e-01
-2.11062893e-01 -8.10099766e-02 1.29564732e-01 -2.21051663e-01
1.95364252e-01 3.62790883e-01 6.85982347e-01 -9.32898521e-01
-7.55225718e-01 -7.95967281e-01 -1.01575097e-02 -6.96679831e-01
4.38071877e-01 -9.58229899e-01 -7.43132055e-01 -1.68619895e+00
1.25341350e-02 -5.16169131e-01 -6.93402886e-02 7.13080883e-01
-1.69889882e-01 -1.47703335e-01 2.08377615e-01 6.27898648e-02
-2.28868634e-01 -2.65567582e-02 1.31287587e+00 -1.37234613e-01
-3.76444012e-02 1.42891690e-01 -1.23484969e+00 5.13439834e-01
1.01477492e+00 -2.06331104e-01 -5.91901720e-01 -3.21583360e-01
6.08884633e-01 3.91157031e-01 1.70829698e-01 -7.97839403e-01
1.43505529e-01 -3.00683498e-01 1.83063716e-01 -3.82321686e-01
1.86558992e-01 -3.37434947e-01 8.60609710e-02 6.40405595e-01
-3.47379893e-01 5.07104754e-01 5.85406184e-01 8.52474794e-02
-2.76220769e-01 -4.43272650e-01 1.93944708e-01 -3.22892517e-01
-6.53762460e-01 -8.12928230e-02 -1.91294730e-01 5.40811956e-01
1.08474398e+00 -4.08306807e-01 -6.75349951e-01 -9.05227065e-02
-3.93777013e-01 4.36402768e-01 1.83000937e-01 3.75626355e-01
3.82015526e-01 -8.53586197e-01 -7.01228917e-01 7.82072842e-02
-1.83260262e-01 5.40200025e-02 3.53237182e-01 7.09690392e-01
-7.26599753e-01 1.04049408e+00 2.63642017e-02 -1.37852177e-01
-9.24767017e-01 4.36086744e-01 3.43062609e-01 -9.48258877e-01
-4.09042895e-01 6.52984798e-01 -2.01635882e-01 -2.90119171e-01
8.91338289e-02 -6.59001529e-01 -1.64857119e-01 1.93559378e-01
1.81805640e-01 4.31191325e-01 2.28044957e-01 -1.29349098e-01
-1.53159648e-01 5.05741358e-01 -1.00014314e-01 2.90998723e-03
1.21377861e+00 3.30621898e-01 -6.11212030e-02 1.51425049e-01
6.37611806e-01 2.23346725e-02 -8.78138840e-01 1.06658772e-01
2.12805390e-01 -2.37413988e-01 -1.80276722e-01 -1.14635706e+00
-1.12282002e+00 7.38994062e-01 1.67948529e-02 1.79337282e-02
9.37909424e-01 -2.99622089e-01 7.10734308e-01 4.99923468e-01
5.08832693e-01 -9.93813753e-01 3.56907427e-01 5.35209119e-01
6.53951943e-01 -1.18725586e+00 1.33532301e-01 -2.95128495e-01
-7.73285866e-01 1.16252875e+00 5.69009364e-01 3.20655257e-01
3.54189068e-01 5.70650280e-01 -1.50386393e-01 -4.08720881e-01
-1.01071167e+00 3.95157844e-01 1.48731902e-01 7.45201856e-02
8.53265107e-01 1.03674918e-01 -4.21638608e-01 1.06687677e+00
-8.50724339e-01 4.21211123e-01 5.05705893e-01 9.97266114e-01
-3.57532576e-02 -1.38367188e+00 -4.89032418e-01 7.04618394e-01
-8.93416524e-01 -4.56083506e-01 -3.55620444e-01 9.91049111e-01
-4.60223779e-02 1.03116286e+00 2.40546335e-02 -2.07952447e-02
4.69085097e-01 2.67924964e-01 2.61038363e-01 -7.17281520e-01
-1.00617862e+00 -4.48014557e-01 4.27958488e-01 -4.80059743e-01
-1.95754200e-01 -7.23149598e-01 -1.29556942e+00 -8.21887493e-01
2.61240751e-01 1.52127475e-01 3.64913702e-01 8.71715426e-01
3.21935862e-01 9.16815817e-01 -1.03413984e-01 -1.00103624e-01
-9.45572555e-01 -5.78535199e-01 -3.73238742e-01 1.44029126e-01
-3.41624171e-01 -6.06601417e-01 -1.41856104e-01 -4.48995292e-01]
|
[10.480300903320312, 7.925762176513672]
|
11310cba-59bb-4fc9-bd89-57746b214e29
|
performance-efficiency-trade-offs-in
|
2109.06870
| null |
https://arxiv.org/abs/2109.06870v1
|
https://arxiv.org/pdf/2109.06870v1.pdf
|
Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition
|
This paper is a study of performance-efficiency trade-offs in pre-trained models for automatic speech recognition (ASR). We focus on wav2vec 2.0, and formalize several architecture designs that influence both the model performance and its efficiency. Putting together all our observations, we introduce SEW (Squeezed and Efficient Wav2vec), a pre-trained model architecture with significant improvements along both performance and efficiency dimensions across a variety of training setups. For example, under the 100h-960h semi-supervised setup on LibriSpeech, SEW achieves a 1.9x inference speedup compared to wav2vec 2.0, with a 13.5% relative reduction in word error rate. With a similar inference time, SEW reduces word error rate by 25-50% across different model sizes.
|
['Yoav Artzi', 'Kilian Q. Weinberger', 'Kyu Han', 'Jing Pan', 'Kwangyoun Kim', 'Felix Wu']
|
2021-09-14
| null | null | null | null |
['unsupervised-pre-training']
|
['methodology']
|
[-1.41913608e-01 2.50767358e-02 6.41196361e-03 -4.71091926e-01
-9.61048543e-01 -4.04000193e-01 5.10209441e-01 -3.73359695e-02
-7.09331453e-01 3.34592074e-01 7.16593981e-01 -1.07263041e+00
2.86882013e-01 -3.03540856e-01 -4.35726672e-01 -2.73714602e-01
6.48091035e-03 5.62036693e-01 9.03714597e-02 -4.12255496e-01
-1.44493118e-01 3.69419873e-01 -1.12051308e+00 2.24085227e-02
1.61535516e-01 7.65323639e-01 1.61383882e-01 1.39569521e+00
-1.61422178e-01 7.45670676e-01 -7.26559460e-01 -4.22059178e-01
2.37065088e-02 2.73710378e-02 -1.09270871e+00 -2.17166822e-02
4.77502376e-01 -4.85844433e-01 -8.83485436e-01 5.22978067e-01
9.66410458e-01 3.24483871e-01 3.29543412e-01 -8.70971322e-01
-5.07721722e-01 9.79654670e-01 -1.28537357e-01 5.11587441e-01
7.78796058e-03 3.24938208e-01 1.42387807e+00 -1.03940082e+00
3.67917418e-01 1.11227882e+00 5.40026069e-01 8.44878614e-01
-1.03564942e+00 -4.95638490e-01 -1.33588314e-01 1.39831677e-01
-1.76054347e+00 -1.17274868e+00 1.24954052e-01 1.85312238e-02
1.98695934e+00 5.51355362e-01 1.68639854e-01 1.05145884e+00
-1.74508080e-01 8.98254335e-01 6.49864137e-01 -5.15629709e-01
1.34932235e-01 7.68353939e-02 7.97632933e-01 6.69749022e-01
9.49076712e-02 -1.22104928e-01 -6.47319138e-01 -1.94368660e-01
3.69711280e-01 -4.53828543e-01 -4.53135401e-01 3.75604272e-01
-8.58028471e-01 8.14659178e-01 -1.31008914e-03 2.42940187e-01
-6.44890442e-02 4.85110968e-01 6.66060865e-01 3.75953943e-01
3.22382540e-01 6.52266860e-01 -8.18333685e-01 -8.08620214e-01
-1.07067907e+00 1.53069766e-02 7.69007862e-01 1.23446989e+00
3.19785327e-01 8.20769250e-01 -4.35816944e-01 1.27409446e+00
4.14765865e-01 7.14804471e-01 7.62521565e-01 -5.21304309e-01
7.16319203e-01 -1.01605184e-01 -3.25011849e-01 -3.61967355e-01
-3.18295538e-01 -6.27360165e-01 -6.13195956e-01 -2.85798907e-01
5.09843826e-02 -4.09803480e-01 -1.25490999e+00 1.51671743e+00
-1.16859255e-02 2.85023987e-01 1.39097258e-01 7.20597148e-01
8.91600549e-01 1.12906134e+00 3.90611328e-02 -2.01591745e-01
1.48436284e+00 -1.40867853e+00 -8.45130384e-01 -5.01331091e-01
1.06097448e+00 -9.74970996e-01 1.21324241e+00 1.30187720e-01
-1.33436191e+00 -4.60973173e-01 -1.12609708e+00 -3.60496759e-01
-3.46389294e-01 7.86881987e-03 3.09995264e-01 1.03037965e+00
-1.45194447e+00 2.38948867e-01 -9.75372672e-01 -3.77761066e-01
1.30233049e-01 3.15141529e-01 -1.49341524e-01 -9.45563521e-03
-8.60879481e-01 8.94261897e-01 1.86252534e-01 -3.47821534e-01
-8.89089227e-01 -8.40423048e-01 -7.88889289e-01 5.57769775e-01
2.10405156e-01 -3.07377160e-01 1.78934169e+00 -1.47596613e-01
-1.66174400e+00 4.26022619e-01 -4.27752137e-01 -8.41358721e-01
4.57539223e-02 -2.74228841e-01 -7.92135060e-01 -4.91715878e-01
-8.30432534e-01 3.70768517e-01 3.70751113e-01 -7.28013456e-01
-4.01199341e-01 -2.89534330e-02 -3.92660588e-01 1.65841937e-01
-7.05114901e-01 3.29699129e-01 -9.67512131e-01 -5.75301409e-01
-3.26949984e-01 -7.73217618e-01 -2.80312151e-01 -6.11678541e-01
-2.59277046e-01 -2.76348948e-01 7.63226926e-01 -8.42894316e-01
1.76908958e+00 -1.99147809e+00 -5.77085130e-02 1.81143045e-01
4.21839058e-01 9.47331250e-01 -4.67795461e-01 4.58023161e-01
2.52842933e-01 3.71185809e-01 -2.94056702e-02 -7.45944440e-01
1.44462481e-01 5.23169875e-01 -3.26111108e-01 2.15190977e-01
-5.51674934e-03 1.02152944e+00 -5.60257554e-01 -9.98502299e-02
2.07295984e-01 7.76775301e-01 -6.54533029e-01 5.35618544e-01
1.21823981e-01 -5.16559064e-01 2.69736815e-02 2.42971241e-01
6.28573239e-01 -9.78635345e-03 4.19182420e-01 2.42971793e-01
-4.87226620e-02 1.06145489e+00 -8.39874685e-01 1.43177068e+00
-8.79676402e-01 1.09526300e+00 9.17959306e-03 -5.42564034e-01
8.78588140e-01 5.54385126e-01 -1.66297063e-01 -6.39741719e-01
3.18769664e-01 1.33275643e-01 9.73435119e-02 -1.18024513e-01
9.28934336e-01 3.28499168e-01 2.45228440e-01 4.46744800e-01
5.64822316e-01 -1.44227162e-01 -1.74923390e-01 2.30163440e-01
1.30698562e+00 -7.58842885e-01 2.71996528e-01 -2.83874542e-01
2.78864503e-01 -3.45314324e-01 2.57771045e-01 9.09835100e-01
-1.52874842e-01 6.39088273e-01 1.10742480e-01 -3.61410946e-01
-1.28959990e+00 -8.65831614e-01 -1.63352549e-01 1.27048600e+00
-5.31844974e-01 -8.85646105e-01 -7.56123364e-01 -5.11537731e-01
-1.87691197e-01 1.03253019e+00 -1.90027192e-01 1.42056579e-02
-7.14550316e-01 -7.22531915e-01 1.14906204e+00 8.07607889e-01
2.90170878e-01 -7.10762143e-01 -3.17784131e-01 1.89632714e-01
1.95353031e-01 -1.34104013e+00 -8.41942251e-01 2.87795067e-01
-6.24410391e-01 -2.53962487e-01 -5.22415102e-01 -7.57193744e-01
1.77736163e-01 4.60429847e-01 1.37724066e+00 2.82316804e-01
-1.75209701e-01 2.58839130e-01 -5.85648537e-01 -3.19718271e-01
-4.94252294e-01 5.16761959e-01 3.33723754e-01 -3.66873920e-01
4.24387425e-01 -1.82745889e-01 -1.69217750e-01 9.76842269e-03
-5.11629820e-01 -2.29028881e-01 2.31070861e-01 1.09011757e+00
3.61158162e-01 -3.31688762e-01 1.12981193e-01 -6.47112489e-01
6.37885213e-01 -3.19396377e-01 -5.05752623e-01 3.52054030e-01
-1.02058804e+00 2.99291611e-01 4.98880833e-01 -2.82680064e-01
-8.26917470e-01 -4.13406640e-01 -9.34744895e-01 -3.77303034e-01
4.62294696e-03 4.36957091e-01 -1.01007551e-01 1.67995542e-01
4.51235086e-01 3.81273866e-01 -1.25003252e-02 -6.22575760e-01
6.08335614e-01 1.28741384e+00 3.05671781e-01 -2.24077746e-01
4.94080454e-01 -3.17006767e-01 -6.90918505e-01 -1.39932597e+00
-3.29275101e-01 -7.20342457e-01 -8.25226232e-02 2.41882339e-01
7.61458755e-01 -9.67942655e-01 -6.19729102e-01 2.15088367e-01
-1.13184667e+00 -8.62106502e-01 -1.97587684e-01 5.80389798e-01
-4.91237044e-02 2.79132724e-01 -8.66824925e-01 -9.29691315e-01
-9.21977282e-01 -1.31124842e+00 8.93206656e-01 -2.69764930e-01
-4.31128412e-01 -1.01266682e+00 9.27207153e-03 4.02603239e-01
8.64241004e-01 -9.02887702e-01 7.61516571e-01 -9.95716155e-01
-1.65761337e-01 -1.52902827e-01 -2.54215091e-01 6.55669391e-01
-3.14057380e-01 -5.34255505e-02 -1.32640529e+00 -6.55287981e-01
-4.15291220e-01 -3.75202656e-01 7.44740248e-01 2.32469514e-01
1.26079798e+00 -4.65712190e-01 5.53189553e-02 7.61088789e-01
1.18854713e+00 2.27317855e-01 6.72581553e-01 -7.76820630e-02
6.41276300e-01 -1.43718749e-01 -8.95358846e-02 4.59363759e-01
2.05912977e-01 9.38529789e-01 -1.36762455e-01 3.81475687e-02
-6.98345244e-01 -2.66519755e-01 4.89848226e-01 1.88162911e+00
1.12148888e-01 -8.73971760e-01 -1.17329895e+00 6.16487861e-01
-1.34706795e+00 -6.71005189e-01 -1.30214021e-01 2.25677872e+00
8.63367081e-01 9.98486057e-02 -1.24570271e-02 1.32833183e-01
2.38089994e-01 5.81205010e-01 -5.87232448e-02 -1.13365030e+00
-1.11278825e-01 7.97693312e-01 8.98815513e-01 1.11691940e+00
-6.08728647e-01 1.25812316e+00 7.46582222e+00 1.16573906e+00
-1.12558246e+00 4.89246190e-01 5.39711535e-01 -2.55074501e-01
-4.44891274e-01 -4.15792465e-01 -1.13898087e+00 1.68687716e-01
1.83818924e+00 -2.33091131e-01 5.90165734e-01 9.05768335e-01
-5.72556481e-02 7.14037299e-01 -7.69256055e-01 1.14962184e+00
3.37848306e-01 -1.46144485e+00 1.15649775e-01 1.75093874e-01
4.26365376e-01 7.08694458e-01 1.24410205e-01 7.73641169e-01
6.23590052e-01 -1.18388820e+00 4.18509930e-01 -3.10797244e-01
1.11157072e+00 -8.49179029e-01 8.95609260e-01 -1.58751622e-01
-9.67618227e-01 1.26126334e-01 -6.00745201e-01 -1.56747480e-03
3.12652200e-01 2.67702311e-01 -1.24312222e+00 1.00915104e-01
5.05128324e-01 -1.70084815e-02 -4.73770827e-01 6.72415674e-01
-8.50350186e-02 1.40468228e+00 -2.80951113e-01 -2.85343140e-01
6.08633935e-01 3.81976575e-01 3.27119976e-01 1.84369659e+00
3.02095443e-01 1.68638319e-01 -2.72447288e-01 1.00868605e-01
-3.63280565e-01 2.28906289e-01 -2.79598922e-01 -1.33121356e-01
9.24695611e-01 9.16638792e-01 1.00958785e-02 -7.49494910e-01
-5.10996997e-01 1.00458968e+00 6.24936759e-01 3.96235108e-01
-9.65924859e-01 -5.66198051e-01 1.30624914e+00 -2.32413895e-02
6.01843059e-01 -7.42189407e-01 -3.82592469e-01 -1.06230438e+00
-3.19564134e-01 -9.83952940e-01 7.46139362e-02 -6.21509850e-01
-6.24373436e-01 1.01325202e+00 -3.65523934e-01 -5.57952225e-01
-2.29660094e-01 -7.43604243e-01 -5.20870984e-01 8.41275990e-01
-1.31826186e+00 -7.08673596e-01 4.55321111e-02 2.26503506e-01
1.05533683e+00 -5.61474562e-01 1.19610596e+00 4.24987823e-01
-8.96586597e-01 1.56868184e+00 4.29255664e-01 2.95568973e-01
3.45521718e-01 -1.23194468e+00 1.40659690e+00 9.48797226e-01
5.25249660e-01 7.72377253e-01 5.93951464e-01 -1.38568506e-01
-1.68029130e+00 -1.25449181e+00 1.47357440e+00 -5.52317083e-01
6.99779510e-01 -6.41144514e-01 -9.38743770e-01 8.14768016e-01
6.50570095e-01 -6.45564124e-02 7.79859781e-01 6.60368562e-01
-9.48044240e-01 -1.64715387e-02 -6.82845473e-01 9.47896600e-01
1.24322355e+00 -6.96628928e-01 -3.81997973e-01 2.08847061e-01
1.29567897e+00 -5.46455741e-01 -8.84754419e-01 1.09344356e-01
5.25288165e-01 -4.30808723e-01 9.28582132e-01 -1.02747917e+00
1.28942514e-02 2.11424410e-01 -6.26184464e-01 -1.49234080e+00
-6.24800146e-01 -8.03761125e-01 -4.32145208e-01 1.13659358e+00
8.97905886e-01 -5.83519042e-01 4.16517377e-01 5.48857927e-01
-5.36189795e-01 -8.09781134e-01 -9.60827291e-01 -1.03052497e+00
2.28635266e-01 -1.07315338e+00 7.71281421e-01 6.84904635e-01
-9.96944681e-02 6.71489954e-01 -4.86006469e-01 1.83743238e-01
5.49919568e-02 -5.54615796e-01 8.69139075e-01 -3.29424411e-01
-4.81189042e-01 -5.85311890e-01 -5.63525140e-01 -1.55293930e+00
3.79785299e-02 -8.39871645e-01 8.54686201e-02 -1.18505847e+00
-5.36314286e-02 -1.98290601e-01 -4.40536797e-01 6.01893663e-01
-2.02107519e-01 1.59345046e-01 2.49386415e-01 -1.62147895e-01
-4.17653859e-01 5.16016901e-01 4.58234042e-01 -2.76900172e-01
-1.19497344e-01 -4.37486649e-01 -5.03719568e-01 1.18356399e-01
9.66397822e-01 -9.36065614e-02 -5.61252594e-01 -1.29712188e+00
-4.52755094e-01 -1.24520347e-01 -2.54660010e-01 -1.06643116e+00
1.94127306e-01 1.39318928e-01 -2.42263809e-01 -2.12500855e-01
5.99245965e-01 -2.72640526e-01 -3.90524119e-01 2.27459863e-01
-5.85727751e-01 3.50119859e-01 4.93169874e-01 1.27254546e-01
-5.14475740e-02 -4.18890268e-02 6.90281868e-01 3.01154137e-01
-7.63444722e-01 2.72336662e-01 -8.26379061e-01 2.20676422e-01
3.46824497e-01 2.62953877e-01 -4.65018362e-01 -6.20929301e-01
-2.98271358e-01 1.09365337e-01 -1.25115976e-01 5.28943419e-01
7.51558661e-01 -1.13826680e+00 -8.92319620e-01 5.04713774e-01
2.95002669e-01 -5.29349744e-01 2.69341379e-01 4.78348464e-01
-5.17600656e-01 8.09782386e-01 4.31395173e-01 -1.87989056e-01
-1.66548610e+00 2.41952047e-01 1.40942186e-01 -1.39805853e-01
-4.60298091e-01 1.45586538e+00 -4.47242469e-01 -5.23958981e-01
5.46154439e-01 -4.58325654e-01 2.02936932e-01 -3.99800807e-01
9.12537336e-01 5.49805045e-01 5.54092646e-01 -5.67058027e-01
-4.32441980e-01 -4.68694195e-02 -4.02425081e-01 -4.16451246e-01
1.06934249e+00 3.72305745e-03 4.01060283e-01 2.22345203e-01
1.69439697e+00 7.36621693e-02 -6.23608828e-01 -4.30509299e-01
-2.43851960e-01 -2.26591051e-01 6.50130212e-01 -6.41704917e-01
-1.12506187e+00 1.15691125e+00 6.19553387e-01 -1.54543016e-02
6.57574892e-01 -8.28119740e-02 1.32375944e+00 5.99658549e-01
1.75668135e-01 -1.13307536e+00 -3.45819771e-01 9.99819338e-01
6.65546060e-01 -1.00840080e+00 -1.69718266e-01 2.57665459e-02
-8.61992836e-01 8.53433788e-01 5.03761232e-01 3.47195685e-01
7.05656767e-01 5.91598570e-01 3.13730747e-01 -2.07538277e-01
-1.16892350e+00 -3.67606938e-01 1.78323045e-01 3.69654626e-01
8.51508379e-01 5.62307298e-01 -1.97449937e-01 4.41273510e-01
-5.37524164e-01 -5.56433082e-01 4.19295192e-01 5.98067343e-01
-5.38262606e-01 -1.05062151e+00 -4.42458056e-02 4.25493985e-01
-2.67888427e-01 -5.77006757e-01 -1.47590578e-01 5.62482715e-01
-5.22961378e-01 1.31964469e+00 4.60413456e-01 -9.71360147e-01
4.56510723e-01 3.93045604e-01 -9.36040655e-03 -6.18054926e-01
-7.34313786e-01 1.51917279e-01 7.19045043e-01 -5.49671650e-01
4.40634459e-01 -2.92687207e-01 -1.02040410e+00 -7.79877901e-01
-5.20901501e-01 2.30498135e-01 8.34811211e-01 6.35660112e-01
7.32623816e-01 7.03162611e-01 5.55509388e-01 -1.20845191e-01
-8.09836149e-01 -1.27469337e+00 -4.80903655e-01 6.56880364e-02
2.38430068e-01 -1.66218415e-01 -3.24703038e-01 -2.27692276e-01]
|
[14.35218334197998, 6.678980827331543]
|
a81f68f4-1123-42ab-95fd-a12f2425b1ab
|
parallel-scale-wise-attention-network-for
|
2104.12076
| null |
https://arxiv.org/abs/2104.12076v1
|
https://arxiv.org/pdf/2104.12076v1.pdf
|
Parallel Scale-wise Attention Network for Effective Scene Text Recognition
|
The paper proposes a new text recognition network for scene-text images. Many state-of-the-art methods employ the attention mechanism either in the text encoder or decoder for the text alignment. Although the encoder-based attention yields promising results, these schemes inherit noticeable limitations. They perform the feature extraction (FE) and visual attention (VA) sequentially, which bounds the attention mechanism to rely only on the FE final single-scale output. Moreover, the utilization of the attention process is limited by only applying it directly to the single scale feature-maps. To address these issues, we propose a new multi-scale and encoder-based attention network for text recognition that performs the multi-scale FE and VA in parallel. The multi-scale channels also undergo regular fusion with each other to develop the coordinated knowledge together. Quantitative evaluation and robustness analysis on the standard benchmarks demonstrate that the proposed network outperforms the state-of-the-art in most cases.
|
['Guanghui Wang', 'Taejoon Kim', 'Jin Zhang', 'Michael Chow', 'Usman Sajid']
|
2021-04-25
| null | null | null | null |
['scene-text-recognition']
|
['computer-vision']
|
[ 3.26968879e-01 -4.74797726e-01 8.49813148e-02 -3.74458879e-01
-8.38787317e-01 -9.34266225e-02 7.97541320e-01 -1.13565013e-01
-5.26886165e-01 2.60229349e-01 2.90082425e-01 -2.42263842e-02
2.44568422e-01 -4.24246430e-01 -6.75504565e-01 -7.37680078e-01
7.97836244e-01 2.11640075e-01 2.46676117e-01 -7.18211243e-03
5.00510156e-01 4.43224698e-01 -1.37701190e+00 4.05584157e-01
8.84623468e-01 9.87562120e-01 3.18194926e-01 8.94219220e-01
-3.56171489e-01 7.55392969e-01 -4.93554771e-01 -5.83803654e-01
7.17806667e-02 -4.07356769e-01 -4.42081183e-01 2.83830911e-01
5.85805297e-01 -5.23569167e-01 -6.24148190e-01 1.18116677e+00
8.15571070e-01 -1.94264054e-02 6.19457960e-01 -8.92944038e-01
-1.02339506e+00 2.98106402e-01 -8.81194293e-01 1.61358356e-01
3.42195153e-01 -1.49445226e-02 1.05407321e+00 -1.15313113e+00
2.64610827e-01 9.92589235e-01 4.84410733e-01 1.79662675e-01
-7.68505514e-01 -4.68623847e-01 2.67815322e-01 4.23690587e-01
-1.54793060e+00 -7.02709436e-01 8.67968142e-01 -3.87295216e-01
1.25336170e+00 1.84179276e-01 2.68805802e-01 1.02772319e+00
3.40197861e-01 1.12596893e+00 7.00544357e-01 -5.72355390e-01
-2.50276864e-01 -1.77350670e-01 2.06592754e-01 8.30408454e-01
1.26945019e-01 -2.89347589e-01 -6.70068681e-01 4.14551497e-01
7.08769023e-01 3.30639929e-02 -3.77264619e-01 -5.95191792e-02
-1.35139525e+00 6.66977227e-01 3.42627496e-01 4.51460928e-01
-3.23554665e-01 1.60010934e-01 5.37748098e-01 1.89575478e-01
2.86398590e-01 1.24875791e-01 -1.96361005e-01 -1.22155264e-01
-1.25140131e+00 -3.81789684e-01 4.53567415e-01 9.75441098e-01
5.61627388e-01 2.98199564e-01 -5.06134033e-01 8.30523551e-01
3.90485764e-01 5.37850440e-01 7.93883801e-01 5.51560745e-02
9.64493692e-01 6.88511491e-01 -1.70120791e-01 -1.11336493e+00
-2.67469317e-01 -5.09565592e-01 -1.06539273e+00 -7.09524751e-02
1.51676282e-01 -7.43372515e-02 -1.16554773e+00 1.24469590e+00
8.25802833e-02 1.93436161e-01 -1.55492639e-03 1.01824248e+00
8.66651475e-01 7.70365000e-01 -3.17789674e-01 -1.47728458e-01
1.18431318e+00 -1.43128586e+00 -1.09946239e+00 -1.51403114e-01
4.01765525e-01 -1.11777937e+00 1.05502498e+00 3.12866926e-01
-1.02234387e+00 -8.80372226e-01 -1.48099208e+00 -3.78853977e-01
-4.31388736e-01 8.16200078e-01 1.34783477e-01 5.49745202e-01
-9.11906362e-01 2.85024643e-01 -8.14555109e-01 -5.72515547e-01
2.13720962e-01 5.33451796e-01 -4.69914973e-01 1.66184783e-01
-9.28091109e-01 8.83136153e-01 8.59982986e-03 4.16024923e-01
-6.26687407e-01 2.50306636e-01 -7.43670225e-01 3.76708657e-01
2.34536558e-01 -4.98957425e-01 9.68046367e-01 -1.23919904e+00
-1.88159513e+00 5.91513216e-01 -2.09575161e-01 -1.69789508e-01
6.78738117e-01 -4.84278083e-01 -3.88035834e-01 1.48892745e-01
-2.89878342e-02 4.11068231e-01 1.22046959e+00 -8.93897593e-01
-5.95604360e-01 -2.16514707e-01 -2.49243602e-01 5.22090852e-01
-6.20034099e-01 2.47682586e-01 -1.02106154e+00 -9.48905528e-01
2.46302515e-01 -4.66255784e-01 2.57346600e-01 -5.66028943e-03
-5.70493579e-01 7.41880462e-02 1.14804566e+00 -8.84813309e-01
1.40697455e+00 -2.10786009e+00 3.86323005e-01 -2.27895305e-01
-3.75139974e-02 3.06576550e-01 -2.39624888e-01 4.67240870e-01
-8.80538896e-02 -8.43329951e-02 7.17400983e-02 -7.00281262e-01
3.10220364e-02 -1.86330050e-01 -3.46635997e-01 7.04893172e-01
2.82480389e-01 1.07194066e+00 -3.16726476e-01 -6.95094466e-01
4.32164788e-01 4.22817409e-01 -3.94401520e-01 3.84475380e-01
5.05646504e-02 2.65885144e-01 -5.87116838e-01 5.84591627e-01
6.57129467e-01 -2.80380934e-01 5.50188532e-04 -3.97405922e-01
-1.86622337e-01 1.01952866e-01 -1.10694671e+00 1.66885459e+00
-1.88409716e-01 9.23310459e-01 -7.79198250e-03 -9.87969041e-01
9.93440926e-01 3.09374124e-01 3.16609591e-01 -5.75503647e-01
4.26731855e-01 -6.74100071e-02 -4.08008918e-02 -7.00394213e-01
7.42447853e-01 2.13805482e-01 3.78584675e-02 4.82308179e-01
1.24328800e-01 1.23768121e-01 -7.95864612e-02 -9.46989432e-02
8.63161325e-01 2.85989732e-01 3.30567747e-01 6.57413434e-03
1.08717644e+00 -4.63934779e-01 3.81147504e-01 6.26614273e-01
-1.95106104e-01 7.97464967e-01 3.99491847e-01 -2.75982469e-01
-1.32951057e+00 -5.74470997e-01 -2.36857347e-02 1.13920701e+00
2.40037531e-01 -4.33321387e-01 -8.70005369e-01 -8.75337720e-01
-2.35142708e-01 2.46057406e-01 -7.02714384e-01 1.50509188e-02
-4.55603480e-01 -5.65402985e-01 7.02254593e-01 7.77724206e-01
1.04093957e+00 -8.93109798e-01 -3.67269635e-01 1.29145220e-01
-3.17643076e-01 -1.19246495e+00 -7.66885400e-01 1.56239450e-01
-4.56650645e-01 -7.52520144e-01 -9.15928841e-01 -9.16120291e-01
7.42339134e-01 1.26723990e-01 3.42272818e-01 1.49652272e-01
-9.32187289e-02 1.36545852e-01 -5.83335936e-01 -7.40653202e-02
-6.83272630e-02 3.43869925e-01 -1.36895329e-01 5.40104747e-01
2.37045392e-01 -1.55833244e-01 -2.10076764e-01 4.41205680e-01
-7.55108774e-01 3.97737861e-01 8.42610359e-01 1.21511054e+00
4.57637876e-01 1.01917237e-02 5.07391393e-01 -3.27714860e-01
5.67445278e-01 8.38107094e-02 -6.68735325e-01 5.23133337e-01
-2.53004849e-01 8.44032988e-02 8.10291350e-01 -3.39709610e-01
-1.05931056e+00 3.07053655e-01 -1.44197777e-01 -4.49324727e-01
-8.94764587e-02 5.23711979e-01 -4.16433454e-01 -1.91816807e-01
1.59902588e-01 7.32644677e-01 -3.94949138e-01 -4.80347395e-01
3.26862978e-03 1.20501554e+00 4.32104647e-01 -1.94422364e-01
7.20379055e-01 2.74426460e-01 -1.65383786e-01 -1.00199127e+00
-7.41673589e-01 -3.69886249e-01 -9.62732136e-01 -8.53087679e-02
1.16184402e+00 -8.66165578e-01 -6.84639037e-01 9.12083745e-01
-1.36630678e+00 -5.03285378e-02 2.33401448e-01 5.53995490e-01
-4.84014988e-01 7.44693100e-01 -5.79853296e-01 -6.93371415e-01
-5.71168363e-01 -1.45963395e+00 1.32803774e+00 1.12958938e-01
1.80808514e-01 -6.23395681e-01 -1.82630911e-01 1.85381889e-01
2.98544586e-01 -3.15306544e-01 6.47006929e-01 -7.75996566e-01
-6.81469858e-01 -4.16323811e-01 -5.95467627e-01 2.48923764e-01
3.74091268e-02 2.82192022e-01 -1.07393134e+00 -4.66473818e-01
-5.75515181e-02 -2.23498046e-01 9.77209866e-01 2.55338252e-01
1.19517159e+00 -4.25245352e-02 -6.72115833e-02 8.68432224e-01
1.29099333e+00 1.82245687e-01 7.46739030e-01 3.43225539e-01
9.38085198e-01 7.31275305e-02 4.02276576e-01 4.32326168e-01
1.69201523e-01 8.73464406e-01 2.75420010e-01 -1.83089241e-01
-5.35809323e-02 -6.38643801e-02 5.40956438e-01 1.20349109e+00
8.13355595e-02 -7.14217424e-01 -7.99133778e-01 3.39955300e-01
-2.12694573e+00 -8.93444538e-01 1.36536574e-02 2.15739655e+00
4.56875414e-01 1.26357958e-01 -3.96170944e-01 2.21202388e-01
1.04673839e+00 4.14836109e-01 -3.73445511e-01 -4.11200106e-01
-3.19442809e-01 -1.90616548e-02 4.51636195e-01 4.63016242e-01
-1.29646730e+00 1.03035522e+00 6.22727966e+00 8.77321720e-01
-1.40209615e+00 -8.90491530e-02 3.63261431e-01 8.84677246e-02
1.79818466e-01 -3.07725906e-01 -9.16914880e-01 3.97060335e-01
5.55295229e-01 8.42249095e-02 4.07741010e-01 5.75437844e-01
-2.29216367e-02 4.97481488e-02 -1.05774426e+00 1.19045293e+00
4.39577371e-01 -1.15237629e+00 3.40098470e-01 -2.35015363e-01
7.52107322e-01 8.01673830e-02 6.54637367e-02 6.86763823e-02
-1.86813593e-01 -1.03558433e+00 7.50208080e-01 6.63035750e-01
1.10204530e+00 -7.06189573e-01 1.01241684e+00 2.31466085e-01
-1.57854700e+00 -5.90652488e-02 -4.75160539e-01 1.99570894e-01
-1.50071746e-02 2.49556959e-01 -5.51266670e-01 9.02688682e-01
3.73529255e-01 9.83124971e-01 -7.55816102e-01 8.92207503e-01
-3.07502896e-01 3.28789443e-01 -1.72239840e-01 -2.50503540e-01
4.03808385e-01 -2.15180330e-02 4.39735889e-01 1.50846493e+00
3.67686182e-01 -3.16349000e-01 1.59476146e-01 5.93302906e-01
-2.59006709e-01 3.95930707e-01 -3.22124749e-01 -1.90645963e-01
1.24137580e-01 1.19871175e+00 -5.88371396e-01 -6.09444678e-01
-7.74279237e-01 1.51097488e+00 3.92717510e-01 3.36709321e-01
-1.00895083e+00 -9.25059438e-01 2.88934588e-01 -2.98997074e-01
8.22094083e-01 -2.40668416e-01 -5.55131435e-01 -1.50773406e+00
2.50906438e-01 -9.15994346e-01 1.29108086e-01 -1.16268420e+00
-1.11906397e+00 6.62842035e-01 -5.65100729e-01 -1.28729630e+00
1.19944356e-01 -6.09632730e-01 -6.83367968e-01 1.08631384e+00
-1.50957227e+00 -1.39506352e+00 -4.56310362e-01 8.18678379e-01
9.60658908e-01 -4.52489555e-01 6.60570085e-01 3.21308851e-01
-9.73386943e-01 1.08955276e+00 3.81150097e-01 4.78996009e-01
9.34146643e-01 -8.64403844e-01 5.76588273e-01 1.17723393e+00
4.77180816e-02 3.56684476e-01 2.47055769e-01 -7.94901848e-01
-1.48767090e+00 -9.47885156e-01 9.18635547e-01 -1.31584510e-01
6.39007390e-01 -3.87593746e-01 -9.13345397e-01 7.14300513e-01
4.59283769e-01 -7.22709820e-02 1.58862203e-01 -2.79261619e-01
-2.48067111e-01 -8.89144465e-02 -8.17774415e-01 6.23006165e-01
6.65876865e-01 -6.63715005e-01 -6.51749313e-01 -4.32494730e-02
5.56757152e-01 -4.86454934e-01 -5.96834779e-01 2.30839983e-01
6.51841223e-01 -8.84221911e-01 5.49337924e-01 -3.38269085e-01
6.72318876e-01 -5.17905533e-01 -3.00543636e-01 -8.57420743e-01
-4.54418987e-01 -6.44735694e-01 9.39458162e-02 1.31941772e+00
2.53042638e-01 -6.00708067e-01 5.39082587e-01 2.74285953e-02
-3.24499071e-01 -5.69657445e-01 -1.10028899e+00 -4.20986950e-01
-1.94951907e-01 -2.94136703e-01 5.56324720e-01 9.42921638e-01
1.06263258e-01 6.10827684e-01 -7.08033025e-01 2.68126696e-01
2.86923409e-01 1.04036793e-01 8.44989777e-01 -9.19839323e-01
-2.71495640e-01 -4.66336429e-01 -6.71406806e-01 -1.40951908e+00
1.13516241e-01 -6.67822003e-01 1.63514167e-01 -1.45508134e+00
3.99919808e-01 2.19960600e-01 -2.78334975e-01 4.33561802e-01
-3.25013012e-01 1.13752298e-02 2.53805459e-01 1.99205667e-01
-7.12501168e-01 8.67525578e-01 1.31898618e+00 -2.76410073e-01
2.90243588e-02 -4.55114722e-01 -3.07257742e-01 6.68544948e-01
7.07290947e-01 -1.53147161e-01 -1.01258725e-01 -7.92644262e-01
-7.15205222e-02 -4.09046514e-03 1.55664101e-01 -1.21548533e+00
7.00132787e-01 2.18537431e-02 8.36831450e-01 -9.87778425e-01
3.38943392e-01 -9.37522292e-01 -2.75719702e-01 2.35966623e-01
-4.08267349e-01 2.43661359e-01 2.15904117e-01 5.28288066e-01
-3.61311495e-01 -1.25903293e-01 8.32840383e-01 2.96578765e-01
-3.98787141e-01 2.84015626e-01 -4.10270452e-01 -2.24444449e-01
7.48400033e-01 -3.51034850e-01 -3.70162338e-01 -4.89122570e-01
-5.36553741e-01 2.16965983e-03 4.76884633e-01 3.64972323e-01
6.75050855e-01 -1.26527393e+00 -7.54780829e-01 6.42559528e-01
1.03801012e-01 -2.16904163e-01 2.42320523e-01 1.02563345e+00
-4.13859814e-01 6.93177700e-01 -2.95615226e-01 -6.01343989e-01
-1.30999160e+00 6.22848988e-01 5.11984289e-01 -5.42688847e-01
-4.11805123e-01 7.01494515e-01 3.34739894e-01 -2.07634736e-02
3.99495482e-01 -1.08137183e-01 -1.87456712e-01 -1.11971416e-01
3.96989584e-01 2.18818679e-01 1.70318484e-01 -9.37928855e-01
-3.06182712e-01 1.17586613e+00 -3.76914650e-01 -1.40603229e-01
1.04560935e+00 -3.22237879e-01 5.70989773e-02 2.84819841e-01
1.08001304e+00 1.10587683e-02 -1.27052248e+00 -3.01874280e-01
-3.30052793e-01 -4.44166154e-01 2.17653736e-01 -7.24727988e-01
-1.00038993e+00 1.25433993e+00 3.22489083e-01 -1.55282140e-01
1.27688920e+00 -6.87062562e-01 6.10331714e-01 6.63778484e-01
-9.83273610e-02 -1.43591273e+00 3.15741301e-01 8.41948807e-01
9.78858888e-01 -1.20536816e+00 2.75450766e-01 -2.62582749e-01
-6.37646556e-01 1.33215904e+00 7.89169729e-01 -6.18397072e-02
4.04102147e-01 4.16819841e-01 2.53932010e-02 1.68337733e-01
-7.66933858e-01 -2.63381720e-01 5.33577025e-01 2.50914395e-01
7.10539818e-01 -4.32459742e-01 -1.78958341e-01 6.19316161e-01
1.76685154e-01 -3.56110670e-02 3.39915603e-01 8.14522564e-01
-5.23531735e-01 -1.01475096e+00 -5.17611682e-01 4.03348297e-01
-5.51149726e-01 -3.71412575e-01 -6.03899777e-01 5.86371243e-01
-1.83695793e-01 8.12448204e-01 5.36331907e-02 -6.45483673e-01
2.76314974e-01 1.70170754e-01 3.49395871e-01 -3.46919239e-01
-8.08935165e-01 5.01379073e-01 -2.18414739e-02 -3.11753124e-01
-1.46557689e-01 -4.89116281e-01 -1.07682884e+00 -2.74549812e-01
-7.45057344e-01 -2.02344418e-01 6.57752812e-01 8.45961988e-01
4.38406289e-01 7.79911041e-01 7.34165192e-01 -9.32396293e-01
-4.99121875e-01 -1.27423561e+00 -3.46141905e-01 1.84525400e-01
4.76398528e-01 -4.39219654e-01 -1.75823778e-01 2.35067397e-01]
|
[11.866340637207031, 2.144838809967041]
|
f04d0d26-69b6-456f-8d07-e803789d634c
|
learnable-multi-level-frequency-decomposition
|
2109.07950
| null |
https://arxiv.org/abs/2109.07950v3
|
https://arxiv.org/pdf/2109.07950v3.pdf
|
Learnable Multi-level Frequency Decomposition and Hierarchical Attention Mechanism for Generalized Face Presentation Attack Detection
|
With the increased deployment of face recognition systems in our daily lives, face presentation attack detection (PAD) is attracting much attention and playing a key role in securing face recognition systems. Despite the great performance achieved by the hand-crafted and deep-learning-based methods in intra-dataset evaluations, the performance drops when dealing with unseen scenarios. In this work, we propose a dual-stream convolution neural networks (CNNs) framework. One stream adapts four learnable frequency filters to learn features in the frequency domain, which are less influenced by variations in sensors/illuminations. The other stream leverages the RGB images to complement the features of the frequency domain. Moreover, we propose a hierarchical attention module integration to join the information from the two streams at different stages by considering the nature of deep features in different layers of the CNN. The proposed method is evaluated in the intra-dataset and cross-dataset setups, and the results demonstrate that our proposed approach enhances the generalizability in most experimental setups in comparison to state-of-the-art, including the methods designed explicitly for domain adaption/shift problems. We successfully prove the design of our proposed PAD solution in a step-wise ablation study that involves our proposed learnable frequency decomposition, our hierarchical attention module design, and the used loss function. Training codes and pre-trained models are publicly released
|
['Arjan Kuijper', 'Florian Kirchbuchner', 'Naser Damer', 'Meiling Fang']
|
2021-09-16
| null | null | null | null |
['face-presentation-attack-detection']
|
['computer-vision']
|
[ 2.79151589e-01 -2.36890450e-01 -9.59587283e-03 -4.08951163e-01
-3.99143010e-01 -4.87925351e-01 4.98472244e-01 -1.67258278e-01
-4.69393075e-01 2.51417190e-01 -1.15058914e-01 -9.47450846e-02
-2.49529094e-01 -6.67984545e-01 -7.37655103e-01 -7.01179147e-01
-2.58826673e-01 -2.17085943e-01 9.55646783e-02 -2.17758834e-01
6.56886101e-02 8.80281687e-01 -1.64121389e+00 3.37575197e-01
5.40660143e-01 1.64789081e+00 -3.57405037e-01 4.19789076e-01
2.03988373e-01 3.46118480e-01 -6.79537117e-01 -7.24352777e-01
4.75559711e-01 -1.15433097e-01 -3.48697811e-01 -4.49451059e-02
6.80078745e-01 -4.95179206e-01 -6.08663142e-01 9.36399281e-01
8.89090955e-01 -2.12575138e-01 2.76831537e-01 -1.25211465e+00
-5.79011321e-01 1.34994552e-01 -4.82250780e-01 1.76580310e-01
2.35192806e-01 3.10613602e-01 4.69001710e-01 -9.06440318e-01
1.19010732e-01 1.24025202e+00 8.95696938e-01 7.38821208e-01
-1.00995088e+00 -9.67698574e-01 1.17769517e-01 5.62366486e-01
-1.41838634e+00 -6.43739223e-01 1.00150573e+00 -1.70328856e-01
8.84465933e-01 9.68178809e-02 3.05481553e-01 1.52815413e+00
1.49076274e-02 4.09510702e-01 9.03504312e-01 -3.56180578e-01
8.79299492e-02 3.23325574e-01 -6.52117804e-02 5.36761165e-01
3.32901329e-01 2.04063714e-01 -6.77226126e-01 -1.50798202e-01
6.51810586e-01 2.05948040e-01 -4.50767100e-01 -1.50728434e-01
-6.48613513e-01 7.29978740e-01 4.71742302e-01 2.95288205e-01
-3.08626354e-01 1.28967330e-01 4.98903632e-01 2.87018657e-01
3.96996528e-01 1.17927857e-01 -3.86148691e-01 -8.25689454e-03
-9.92052615e-01 -8.19225162e-02 6.26267791e-01 5.97379744e-01
4.92890656e-01 1.55608505e-01 -3.09893340e-01 6.75063968e-01
1.42182902e-01 4.70398813e-01 4.48288292e-01 -3.80048662e-01
4.00581449e-01 4.70304430e-01 -2.49678101e-02 -1.10746181e+00
-5.52263200e-01 -6.34552777e-01 -8.77333403e-01 8.24837610e-02
2.62789845e-01 -2.32380345e-01 -8.76944721e-01 1.92457128e+00
4.13141578e-01 5.18306613e-01 -9.27369520e-02 7.47661531e-01
6.41474247e-01 1.98846340e-01 8.70104283e-02 -1.67234465e-01
1.48492038e+00 -6.11344695e-01 -7.38437474e-01 1.61482599e-02
1.70828119e-01 -6.55856967e-01 8.92940700e-01 5.24436593e-01
-7.41618276e-01 -8.99203658e-01 -1.26503050e+00 8.12017098e-02
-3.72309893e-01 2.30840281e-01 4.30650175e-01 1.34427798e+00
-1.10476243e+00 6.54364586e-01 -6.71691537e-01 -3.43571186e-01
7.93194890e-01 6.90986335e-01 -3.81920844e-01 1.27679920e-02
-1.34834349e+00 5.62014163e-01 8.51799920e-02 3.26593786e-01
-9.69873011e-01 -7.94799387e-01 -6.78146720e-01 4.47262943e-01
2.36766189e-01 -3.82815868e-01 7.99731195e-01 -9.72982049e-01
-1.76690924e+00 5.05803227e-01 3.09362002e-02 -5.27389884e-01
3.57261181e-01 -4.40216452e-01 -6.69602692e-01 5.64215481e-01
-4.99741256e-01 4.11536247e-01 1.53687119e+00 -9.21301961e-01
-3.97486567e-01 -3.54963183e-01 5.17407022e-02 -4.04610753e-01
-1.15571940e+00 9.60620940e-02 -4.18549955e-01 -8.41398478e-01
-3.19439799e-01 -6.96992576e-01 3.03395540e-01 2.07221821e-01
-1.49813563e-01 -6.17341660e-02 1.12979543e+00 -6.50862873e-01
1.23845577e+00 -2.46656919e+00 -9.49557871e-02 2.18066782e-01
2.31914371e-02 7.24389195e-01 -1.87496275e-01 3.04160684e-01
-4.61225539e-01 -1.29157782e-01 -1.96529910e-01 -7.21958637e-01
9.63327438e-02 -1.58301488e-01 -4.96126562e-01 6.56499922e-01
7.81448722e-01 3.72422814e-01 -4.55468774e-01 6.98682386e-03
2.05255628e-01 1.03934920e+00 -7.43452847e-01 3.99092257e-01
8.31247866e-02 5.07866502e-01 -2.26930007e-01 5.94381094e-01
1.19128489e+00 -1.58824380e-02 3.20910178e-02 -5.47876358e-01
2.49224633e-01 1.30456105e-01 -1.19021845e+00 1.57332134e+00
-4.99524772e-01 5.31613111e-01 1.78820908e-01 -1.10768163e+00
8.07567596e-01 4.99016583e-01 3.85406941e-01 -7.67550647e-01
1.67199016e-01 7.78169706e-02 3.47890891e-02 -5.44361234e-01
3.46751697e-02 3.76629643e-02 1.59479946e-01 3.00510764e-01
4.17992353e-01 3.61716121e-01 -2.07675174e-01 -1.04808658e-01
1.12584841e+00 -1.32478606e-02 -2.12726705e-02 -1.08655520e-01
8.41108203e-01 -7.74858773e-01 4.55489546e-01 4.76321876e-01
-3.55475992e-01 5.94276726e-01 5.70270956e-01 -5.34513950e-01
-6.23085201e-01 -7.30746508e-01 -4.75690663e-01 1.05243874e+00
4.77239415e-02 -3.90457332e-01 -1.01541698e+00 -8.71803701e-01
1.44095853e-01 3.44571412e-01 -7.86385834e-01 -3.97828132e-01
-5.65813899e-01 -8.15632343e-01 9.23330843e-01 4.78267521e-01
8.26980352e-01 -9.19432878e-01 -8.05381238e-01 6.44167373e-03
3.16321999e-01 -1.42814124e+00 -3.61803591e-01 9.98198763e-02
-3.51776928e-01 -1.16976297e+00 -5.95736921e-01 -4.53571826e-01
4.82278943e-01 6.64199591e-02 5.50413966e-01 3.01199090e-02
-4.84884351e-01 6.66889191e-01 -3.69336665e-01 -3.19767684e-01
9.85763222e-02 9.55370739e-02 2.05392107e-01 8.44454110e-01
4.47821707e-01 -7.75055945e-01 -9.76971209e-01 1.54624969e-01
-1.05993438e+00 -5.87689698e-01 5.44792473e-01 8.61068726e-01
-1.37104075e-02 1.31998420e-01 5.53147376e-01 -7.12662280e-01
4.48718071e-01 -4.20277566e-01 -4.94090468e-01 1.06238686e-01
-3.88066202e-01 -1.93464151e-03 7.64005363e-01 -6.72357261e-01
-8.89998674e-01 -8.10779706e-02 -3.10370922e-01 -6.23158038e-01
-2.15763733e-01 1.79377776e-02 -6.03059590e-01 -4.84668046e-01
6.07376397e-01 9.58415940e-02 -1.97531953e-02 -4.36102450e-01
8.09352100e-02 6.29774630e-01 4.68520999e-01 -4.56250310e-01
1.05830312e+00 5.77368677e-01 1.53144868e-02 -7.55174458e-01
-3.36608946e-01 -8.09522718e-02 -3.62345040e-01 -1.34933025e-01
5.24901569e-01 -8.24575543e-01 -1.11969185e+00 8.00901830e-01
-1.26534116e+00 9.43016447e-03 -1.11618586e-01 3.20267022e-01
-9.85701457e-02 5.02133787e-01 -5.29397845e-01 -8.88062060e-01
-4.69644547e-01 -1.31690359e+00 1.22726369e+00 3.46403658e-01
1.31436244e-01 -8.28613818e-01 -2.67865062e-01 -2.10384093e-02
7.94155538e-01 2.61088133e-01 8.02632153e-01 -7.63440490e-01
-3.81575048e-01 -2.97789991e-01 -2.49777153e-01 5.94481230e-01
1.51556298e-01 -1.70965150e-01 -1.74318230e+00 -7.66740322e-01
3.61203790e-01 -2.88673908e-01 1.06872845e+00 1.88877117e-02
1.50621724e+00 -4.68146354e-01 -6.57661855e-02 1.00561523e+00
1.32281291e+00 7.37664253e-02 7.46608436e-01 1.78799570e-01
5.28940260e-01 6.21978402e-01 -8.41261540e-03 6.76479518e-01
-7.36906305e-02 8.90800834e-01 6.60133421e-01 -2.55917907e-01
-1.78810164e-01 4.49800976e-02 5.59846640e-01 1.21317461e-01
3.10300082e-01 -1.85129136e-01 -6.19258761e-01 2.88705379e-01
-1.49885809e+00 -9.56233859e-01 5.33691764e-01 2.21888113e+00
6.24465048e-01 1.71042457e-01 1.09192640e-01 4.95062470e-01
5.91728806e-01 9.02933180e-02 -5.82543612e-01 -4.03309286e-01
-8.38476196e-02 9.73076224e-01 2.64316201e-01 3.18465084e-01
-1.30912423e+00 5.93403578e-01 5.32840300e+00 1.01981759e+00
-1.70279646e+00 2.44034603e-01 5.82837105e-01 -1.21563002e-01
4.01445851e-02 -4.41983879e-01 -8.25826466e-01 5.27973592e-01
1.00906754e+00 3.95159513e-01 3.74348640e-01 7.60751605e-01
-7.36838579e-02 3.47251713e-01 -1.26476789e+00 1.15042591e+00
2.47804493e-01 -9.21635151e-01 2.52299737e-02 -2.88967546e-02
2.67222762e-01 -4.97290641e-01 7.35176265e-01 1.79883629e-01
-5.57309806e-01 -1.16371226e+00 6.34741247e-01 3.05939585e-01
1.05070484e+00 -8.23184669e-01 7.45132685e-01 -1.60441205e-01
-1.12853324e+00 -4.44801539e-01 -2.14702711e-01 8.69578645e-02
-4.05142546e-01 7.46759772e-01 -6.90518141e-01 5.36233008e-01
7.92652905e-01 4.94053572e-01 -6.10979140e-01 7.78789282e-01
-1.29607007e-01 7.33062863e-01 -3.81071568e-01 3.39947045e-01
-5.08550927e-02 3.69171649e-01 4.61025059e-01 1.32495368e+00
3.11342597e-01 -8.77023116e-02 -3.34181994e-01 7.00713336e-01
-2.25562274e-01 -2.38506153e-01 -2.94757813e-01 1.18294790e-01
3.89721870e-01 1.48713231e+00 -3.47364247e-01 6.59076795e-02
-4.34843987e-01 1.01887095e+00 1.59533679e-01 4.92924958e-01
-1.07929742e+00 -7.19853401e-01 1.00959897e+00 2.10836962e-01
7.48006999e-01 -6.75845146e-02 -2.16116369e-01 -1.02605283e+00
3.23912263e-01 -9.24149752e-01 3.67870688e-01 -1.34570301e-01
-1.33473086e+00 9.55368161e-01 -1.27543911e-01 -1.14738977e+00
-1.87406391e-01 -9.89217758e-01 -6.37446344e-01 9.39723194e-01
-1.82636285e+00 -1.16284549e+00 -4.45769101e-01 8.56965601e-01
2.38518700e-01 -3.01373094e-01 9.25890505e-01 7.30802834e-01
-8.43176484e-01 1.42202425e+00 -1.11778401e-01 2.59664029e-01
9.41984653e-01 -7.21334100e-01 3.25941652e-01 1.01953721e+00
-1.43094882e-01 7.90300310e-01 3.54068011e-01 -2.46989384e-01
-1.55586743e+00 -1.12319398e+00 3.53668272e-01 2.26048823e-03
3.13302189e-01 -7.78836608e-01 -1.01524448e+00 1.95350811e-01
2.23303288e-01 4.01091605e-01 8.05950999e-01 -1.35762170e-01
-8.98740590e-01 -6.15336537e-01 -1.47841775e+00 1.52999640e-01
9.72820699e-01 -7.18426347e-01 -7.26668686e-02 1.47150964e-01
6.09904110e-01 -3.16187084e-01 -6.91495359e-01 6.80411458e-01
6.27712905e-01 -1.11239123e+00 1.00939035e+00 -3.92954707e-01
1.42333671e-01 -1.50332585e-01 -1.29981458e-01 -1.00509286e+00
-2.22911388e-01 -9.15938377e-01 -3.48790437e-01 1.28975499e+00
4.25427072e-02 -7.88935483e-01 7.23073959e-01 4.12055373e-01
6.42685443e-02 -7.81411052e-01 -1.21535623e+00 -5.26797950e-01
-1.62359521e-01 -4.01627004e-01 7.47569978e-01 7.13069141e-01
-1.54914960e-01 2.55589690e-02 -4.10022259e-01 5.92868268e-01
5.86817682e-01 -2.09100217e-01 6.49559379e-01 -9.78200138e-01
-5.50201595e-01 -2.81256080e-01 -7.30107188e-01 -7.56063640e-01
1.96608566e-02 -4.89655137e-01 -3.80284041e-01 -6.38621807e-01
-8.72405693e-02 -2.43664369e-01 -3.99150819e-01 6.53712988e-01
-1.94173068e-01 4.38230991e-01 2.96995044e-01 -3.11386329e-03
-2.83036709e-01 7.18306363e-01 6.66091204e-01 -2.29867071e-01
-2.10374258e-02 -1.90097883e-01 -7.22968102e-01 4.42484111e-01
5.48314989e-01 -2.81550407e-01 -3.37719321e-01 -4.91622537e-01
-1.73499435e-01 -3.39784086e-01 5.66489518e-01 -1.44499969e+00
3.09584707e-01 2.43244946e-01 8.28924954e-01 -2.52355449e-02
4.41740721e-01 -1.12848413e+00 -9.84629467e-02 5.76233268e-01
-1.01613902e-01 -2.15101600e-01 8.28616858e-01 3.80289435e-01
-1.13907985e-01 1.60407454e-01 9.71580863e-01 5.51718771e-01
-3.07744861e-01 4.84305680e-01 5.96941039e-02 -2.92465091e-01
9.63007450e-01 -2.39635274e-01 -3.12957466e-01 -2.87256896e-01
-5.33568561e-01 -2.33532116e-01 6.05915263e-02 4.41145152e-01
6.17224514e-01 -1.28984785e+00 -6.54836416e-01 8.83292496e-01
-6.83766678e-02 -2.52357244e-01 4.18564051e-01 6.91068232e-01
-1.87776089e-01 3.76161635e-01 -3.65359366e-01 -5.94345748e-01
-9.70617294e-01 6.76524580e-01 4.63799566e-01 -1.96784019e-01
-3.83657157e-01 8.92104685e-01 2.87079692e-01 2.27734055e-02
6.20060861e-01 -1.24311157e-01 -1.76002458e-01 4.94665205e-02
8.42547238e-01 1.68534279e-01 5.90693176e-01 -4.11915451e-01
-5.37978292e-01 6.08359158e-01 -2.01745898e-01 1.15148708e-01
1.39284825e+00 2.16724649e-01 1.23866379e-01 -2.30122417e-01
1.45643759e+00 1.33646205e-02 -1.40219331e+00 -1.24286607e-01
-2.58921266e-01 -4.90420997e-01 -3.58739458e-02 -7.20768213e-01
-1.33261490e+00 1.25776827e+00 1.20417190e+00 2.96611097e-02
1.72901893e+00 -5.34805000e-01 8.27260077e-01 1.04464732e-01
2.28796616e-01 -7.26434827e-01 3.82741272e-01 2.38398850e-01
7.94914544e-01 -9.90836620e-01 -2.31160119e-01 -4.47075546e-01
-8.83978009e-02 1.31567776e+00 5.92795730e-01 -8.03498328e-02
8.46512079e-01 3.53613883e-01 -2.38943413e-01 5.46186864e-02
-5.23563385e-01 9.10919253e-03 3.09918314e-01 7.50751853e-01
2.92800397e-01 -4.96827722e-01 4.02585231e-02 9.12887275e-01
1.09354749e-01 2.40171375e-03 2.95949876e-02 8.06112647e-01
-2.32797815e-03 -1.19586086e+00 -5.74979484e-01 1.12383135e-01
-5.91041267e-01 -3.51020843e-02 -2.37096220e-01 6.59659386e-01
5.19849598e-01 9.49171126e-01 6.78666309e-02 -5.41120768e-01
3.91735554e-01 1.04609951e-01 5.87020993e-01 -3.12809438e-01
-8.56491864e-01 -1.04951024e-01 -3.90950859e-01 -7.68236697e-01
-2.69868433e-01 -3.96641582e-01 -8.26990008e-01 -3.21876347e-01
-1.74219817e-01 -1.26100585e-01 5.68137705e-01 7.49353349e-01
7.64108539e-01 6.20374858e-01 1.04926836e+00 -9.69227314e-01
-7.20326543e-01 -9.29800332e-01 -3.02106112e-01 5.33234298e-01
6.58751309e-01 -8.06075931e-01 -5.18279254e-01 -3.57106365e-02]
|
[13.097590446472168, 0.8885894417762756]
|
1f560371-e91a-4053-acbd-3e3abbb76eaf
|
variance-dependent-regret-bounds-for-linear
|
2302.10371
| null |
https://arxiv.org/abs/2302.10371v1
|
https://arxiv.org/pdf/2302.10371v1.pdf
|
Variance-Dependent Regret Bounds for Linear Bandits and Reinforcement Learning: Adaptivity and Computational Efficiency
|
Recently, several studies (Zhou et al., 2021a; Zhang et al., 2021b; Kim et al., 2021; Zhou and Gu, 2022) have provided variance-dependent regret bounds for linear contextual bandits, which interpolates the regret for the worst-case regime and the deterministic reward regime. However, these algorithms are either computationally intractable or unable to handle unknown variance of the noise. In this paper, we present a novel solution to this open problem by proposing the first computationally efficient algorithm for linear bandits with heteroscedastic noise. Our algorithm is adaptive to the unknown variance of noise and achieves an $\tilde{O}(d \sqrt{\sum_{k = 1}^K \sigma_k^2} + d)$ regret, where $\sigma_k^2$ is the variance of the noise at the round $k$, $d$ is the dimension of the contexts and $K$ is the total number of rounds. Our results are based on an adaptive variance-aware confidence set enabled by a new Freedman-type concentration inequality for self-normalized martingales and a multi-layer structure to stratify the context vectors into different layers with different uniform upper bounds on the uncertainty. Furthermore, our approach can be extended to linear mixture Markov decision processes (MDPs) in reinforcement learning. We propose a variance-adaptive algorithm for linear mixture MDPs, which achieves a problem-dependent horizon-free regret bound that can gracefully reduce to a nearly constant regret for deterministic MDPs. Unlike existing nearly minimax optimal algorithms for linear mixture MDPs, our algorithm does not require explicit variance estimation of the transitional probabilities or the use of high-order moment estimators to attain horizon-free regret. We believe the techniques developed in this paper can have independent value for general online decision making problems.
|
['Quanquan Gu', 'Tong Zhang', 'Dongruo Zhou', 'Jiafan He', 'Heyang Zhao']
|
2023-02-21
| null | null | null | null |
['multi-armed-bandits']
|
['miscellaneous']
|
[ 4.21365872e-02 7.28436708e-02 -5.34080803e-01 -2.88940161e-01
-1.30604053e+00 -5.87632656e-01 1.47844478e-01 1.93061437e-02
-6.78941667e-01 1.17489398e+00 -2.20670074e-01 -8.25709224e-01
-8.28527331e-01 -7.94523239e-01 -8.05409849e-01 -9.92721617e-01
-2.47588217e-01 4.88337606e-01 -1.34967407e-02 1.26929432e-01
1.20476745e-01 3.13739568e-01 -1.16104794e+00 -2.03900069e-01
8.34908843e-01 1.63827848e+00 1.22903533e-01 7.19741404e-01
-8.50742832e-02 7.56591141e-01 -3.74421299e-01 -6.23051703e-01
5.90235353e-01 -5.50123513e-01 -4.25280303e-01 1.40058808e-03
-1.89524442e-01 -4.82015014e-01 -1.16668999e-01 1.11257851e+00
5.27608395e-01 1.98074400e-01 7.12883353e-01 -9.86271501e-01
-1.92406341e-01 8.98526669e-01 -9.30174828e-01 2.71821797e-01
-1.94418699e-01 -1.79160669e-01 9.58944738e-01 -2.91659515e-02
7.32665658e-02 1.22228146e+00 3.72873515e-01 6.87275529e-01
-1.23539770e+00 -9.46456134e-01 5.58221400e-01 6.37814105e-02
-1.12478459e+00 -2.68300921e-01 2.97987610e-01 -3.34756017e-01
5.29844820e-01 3.42351973e-01 4.64724153e-01 8.02408278e-01
8.67590085e-02 1.03228033e+00 1.45219064e+00 -4.12210107e-01
8.46388817e-01 1.93237048e-02 -7.36842453e-02 3.47731024e-01
4.13297445e-01 4.70208287e-01 -1.49165869e-01 -1.71728298e-01
9.24427211e-01 1.43540382e-01 2.58844607e-02 -3.34634602e-01
-4.58706081e-01 1.05254889e+00 -3.06393914e-02 -1.73520029e-01
-4.82178688e-01 5.51218629e-01 1.11979432e-01 4.48342383e-01
5.69137275e-01 -3.39049958e-02 -5.36003351e-01 -4.86858100e-01
-1.02634919e+00 4.10602778e-01 8.74905884e-01 1.04120874e+00
4.24017072e-01 5.02256341e-02 -5.04456341e-01 5.92683315e-01
2.90762246e-01 7.60224640e-01 1.18825287e-02 -1.29214716e+00
7.98046827e-01 -1.77196145e-01 7.74968028e-01 -3.30642313e-01
-1.92980945e-01 -6.69340909e-01 -6.76683784e-01 1.90295070e-01
7.08012581e-01 -6.94641590e-01 -6.80999696e-01 1.98879099e+00
2.57884592e-01 -2.19858542e-01 9.95395929e-02 7.35551655e-01
-3.67945492e-01 4.77280110e-01 -2.47530431e-01 -7.75861144e-01
9.89327312e-01 -5.99946082e-01 -7.30815947e-01 -2.52597511e-01
2.67839968e-01 -5.39535582e-01 6.69045389e-01 5.65897286e-01
-1.33866477e+00 1.55751735e-01 -7.97496140e-01 6.66175544e-01
2.54338890e-01 -2.47610644e-01 7.59352028e-01 1.16983342e+00
-6.99380100e-01 6.07074678e-01 -8.70723307e-01 2.83885509e-01
4.79220331e-01 3.77964437e-01 3.36114287e-01 -6.13079183e-02
-9.20773804e-01 5.94725430e-01 2.72637665e-01 1.52283430e-01
-1.00570202e+00 -3.88897508e-01 -4.63815778e-01 1.45188779e-01
1.07004833e+00 -5.18535435e-01 1.56958628e+00 -9.85225022e-01
-1.82335150e+00 1.91489309e-01 -3.06122098e-02 -9.39024270e-01
8.81559670e-01 -1.55364022e-01 3.11266221e-02 -3.53053436e-02
-1.40524488e-02 -1.38195634e-01 8.12641144e-01 -8.65983486e-01
-9.79550481e-01 -4.26486194e-01 2.33153462e-01 3.12927067e-01
4.65766601e-02 -2.25869581e-01 -1.55986562e-01 -6.43232286e-01
-1.17149856e-02 -9.57283318e-01 -6.30231261e-01 -3.74871343e-01
-1.85662568e-01 1.96161434e-01 -7.86272660e-02 -3.71191770e-01
1.38721204e+00 -2.03776431e+00 -1.03927456e-01 3.07643592e-01
-3.67596924e-01 -1.32590383e-01 1.47304997e-01 4.28537309e-01
3.40142041e-01 1.60557017e-01 -3.46218318e-01 -3.34235758e-01
3.56038928e-01 3.46231550e-01 -5.39345443e-01 4.88226563e-01
-3.49259943e-01 4.33207095e-01 -8.09321880e-01 -1.35869816e-01
-4.63214479e-02 -1.43632665e-01 -5.33184767e-01 -1.33868113e-01
-2.90666670e-01 6.38506711e-02 -5.92486501e-01 4.33410257e-01
6.83804333e-01 -2.06574425e-01 3.70398670e-01 5.48215032e-01
-9.54061747e-02 -2.19224114e-02 -1.64658749e+00 1.05152309e+00
-6.14378691e-01 2.58188713e-02 4.80093271e-01 -1.24691808e+00
4.10736352e-01 1.54664382e-01 4.84866709e-01 -4.98674959e-01
2.84903467e-01 2.73254484e-01 -1.93189994e-01 -7.58546172e-03
2.59869218e-01 -7.06846893e-01 -3.28399301e-01 4.42146331e-01
-1.28741771e-01 -6.10466003e-02 2.99434990e-01 -1.69118151e-01
1.04694164e+00 -3.47187072e-02 2.22241372e-01 -2.83844471e-01
8.87735784e-02 -4.60524380e-01 8.11339438e-01 1.38867426e+00
-3.00716013e-01 1.02925882e-01 8.55840087e-01 -2.97402591e-02
-8.63346279e-01 -1.17727244e+00 -2.50864685e-01 1.08386528e+00
4.35681865e-02 2.03443065e-01 -5.48990965e-01 -5.70695341e-01
3.54057580e-01 9.97975588e-01 -8.02826226e-01 1.25315607e-01
1.56565398e-01 -1.00060785e+00 3.17080587e-01 3.61611009e-01
5.62571943e-01 -3.58628750e-01 -4.92690355e-01 4.97656345e-01
7.84226581e-02 -8.15994680e-01 -3.34416240e-01 5.52714467e-01
-9.08603549e-01 -7.74121821e-01 -7.74358928e-01 3.94499004e-02
4.57727730e-01 -6.43848106e-02 4.81387526e-01 -9.48931456e-01
1.98692039e-01 6.52341068e-01 -1.14910744e-01 -8.01239669e-01
-8.26149993e-03 -3.11211199e-01 5.97464591e-02 9.24286544e-02
1.67443082e-02 -3.47429931e-01 -8.25964630e-01 2.38617823e-01
-9.54569817e-01 -1.96929216e-01 6.27625048e-01 8.06308925e-01
7.43346214e-01 1.35339379e-01 6.24699593e-01 -6.72818661e-01
6.72993422e-01 -4.88028735e-01 -1.17584050e+00 2.78528184e-01
-7.70532310e-01 4.26038414e-01 3.87080014e-01 -6.15110219e-01
-1.11503947e+00 -6.43250793e-02 2.16288418e-01 -2.50569403e-01
3.16750795e-01 4.73914534e-01 8.97384062e-02 2.45827556e-01
2.59137660e-01 3.35527182e-01 -9.36230347e-02 -4.07606989e-01
3.72650534e-01 5.39839566e-01 7.88604468e-02 -9.25225914e-01
4.17542756e-01 4.91460592e-01 2.91551203e-01 -1.99008167e-01
-1.13113642e+00 -1.29473507e-01 2.75723130e-01 -2.51033790e-02
4.32394743e-01 -9.12393808e-01 -1.14345133e+00 2.41390109e-01
-4.78582501e-01 -6.62927687e-01 -6.02719843e-01 7.69517362e-01
-1.26343429e+00 2.74239123e-01 -6.80529714e-01 -1.83389926e+00
-8.24689344e-02 -8.54561508e-01 3.76487702e-01 2.45751023e-01
3.82305682e-01 -7.55874395e-01 -5.42611349e-03 3.89711708e-01
3.61760974e-01 1.00629099e-01 6.19893968e-01 -5.12760222e-01
-6.26258433e-01 -6.26652911e-02 -7.80042261e-02 6.09073818e-01
2.48181112e-02 -3.92950326e-01 -3.83209020e-01 -3.61180335e-01
2.04904571e-01 -2.52856672e-01 8.99676442e-01 8.25525105e-01
1.14994812e+00 -9.30121601e-01 -6.52116910e-02 2.88399816e-01
1.39033616e+00 6.67255104e-01 2.66856015e-01 5.87737858e-01
-1.91062450e-01 2.41691574e-01 8.18318129e-01 1.11156583e+00
2.60485470e-01 2.61222601e-01 5.65417469e-01 4.29191858e-01
7.44420052e-01 -1.18178993e-01 5.06916165e-01 1.04043782e-01
-1.35352284e-01 -3.10356319e-01 -5.25718987e-01 6.59686983e-01
-2.05324697e+00 -1.20953214e+00 2.83161879e-01 2.87385345e+00
1.07081103e+00 2.96700597e-01 2.89441884e-01 -3.28510925e-02
7.49911785e-01 -1.35834038e-01 -8.38798940e-01 -7.41485834e-01
-2.31955759e-02 2.78159887e-01 1.24108779e+00 5.94753385e-01
-9.60069597e-01 5.30278385e-01 5.83266020e+00 1.36451793e+00
-6.19537413e-01 2.12012127e-01 8.51603508e-01 -7.51948476e-01
-1.40343100e-01 1.30883232e-01 -9.01897311e-01 8.90097916e-01
1.08674514e+00 -3.44534993e-01 6.91276968e-01 9.90195096e-01
1.08530082e-01 -5.88071942e-01 -9.39272106e-01 8.89861345e-01
-5.14157712e-01 -9.91119683e-01 -5.68474472e-01 4.23824698e-01
9.91925776e-01 -1.58835381e-01 4.49485391e-01 4.36518431e-01
9.88492548e-01 -8.01532984e-01 7.81349361e-01 5.75100243e-01
6.19520843e-01 -1.26078558e+00 6.81016922e-01 6.78788900e-01
-7.65435398e-01 -6.80520236e-01 -2.85269499e-01 -3.12685370e-01
1.82396010e-01 7.78019905e-01 -4.27033871e-01 5.66873252e-01
6.34629071e-01 -1.12996876e-01 3.55951130e-01 1.02992320e+00
-1.23257115e-01 6.44983232e-01 -7.01315701e-01 -3.31831753e-01
5.52761436e-01 -3.70094806e-01 3.77706409e-01 9.49110270e-01
5.28950572e-01 2.67169863e-01 9.23191682e-02 4.24154192e-01
1.26403272e-01 9.51705128e-02 -5.67526594e-02 -6.06941991e-02
4.21293885e-01 7.67136574e-01 -7.27153063e-01 -2.79564053e-01
-2.47297153e-01 7.16976702e-01 1.16610602e-01 5.08733511e-01
-8.98447573e-01 -2.47442320e-01 7.50031471e-01 -1.18500412e-01
7.98814714e-01 -1.58040717e-01 -2.84043550e-01 -9.15511549e-01
9.72454064e-03 -5.21178842e-01 6.42557085e-01 -2.99135577e-02
-1.40218592e+00 -7.06792697e-02 1.85991019e-01 -8.84998560e-01
-4.09747988e-01 -3.84694755e-01 -7.01939240e-02 8.31790626e-01
-1.47779477e+00 -4.50395912e-01 7.57391036e-01 6.02848649e-01
2.00573951e-01 -9.30027105e-03 4.54253107e-01 6.26042858e-02
-5.38082302e-01 6.26617014e-01 1.16365743e+00 -1.36707395e-01
3.71407390e-01 -1.29025424e+00 -3.77380967e-01 5.86391747e-01
-3.06957811e-01 3.22228670e-01 8.37433040e-01 -5.01339912e-01
-1.11582041e+00 -8.52989972e-01 2.22155988e-01 -4.36548032e-02
9.34094787e-01 -1.38183028e-01 -2.48496726e-01 6.49002731e-01
-2.71015137e-01 -1.03177316e-01 8.78001571e-01 1.38466895e-01
-9.28176194e-02 -4.59774703e-01 -1.38258052e+00 4.77800548e-01
8.24944317e-01 -1.83244035e-01 -1.26778111e-01 2.17164531e-01
4.84811842e-01 -4.67812210e-01 -9.16686356e-01 3.94217193e-01
6.10716164e-01 -9.78789985e-01 5.88697731e-01 -5.83751798e-01
4.10479829e-02 1.37871519e-01 -4.54282522e-01 -9.72701490e-01
-4.02634591e-02 -9.13745761e-01 -3.70745510e-01 1.01893246e+00
5.09910822e-01 -7.61449218e-01 6.59764946e-01 8.86816561e-01
2.26736695e-01 -8.69941294e-01 -1.52259958e+00 -1.15676701e+00
4.08282936e-01 -7.96718836e-01 4.09103274e-01 3.69379580e-01
-1.07662357e-01 -2.89743066e-01 -5.82743824e-01 1.08832791e-01
9.41661119e-01 2.95258105e-01 3.38098019e-01 -7.65782714e-01
-8.45017493e-01 -6.02810264e-01 1.15302160e-01 -1.29732239e+00
-1.08371854e-01 -3.09528142e-01 4.05062810e-02 -1.23161769e+00
3.30253661e-01 -6.66473806e-01 -6.53431952e-01 3.04004818e-01
8.54520425e-02 -4.38001394e-01 3.60851437e-01 -1.71702236e-01
-8.35115433e-01 6.72405005e-01 1.06410193e+00 1.55802825e-02
-2.75198638e-01 6.69682920e-01 -7.50402570e-01 6.58867896e-01
8.07378650e-01 -5.85738003e-01 -4.68707174e-01 -5.64805418e-02
4.74559814e-01 6.46734178e-01 8.86963159e-02 -6.72099352e-01
1.48153445e-02 -8.04301322e-01 7.13888630e-02 -5.57112992e-01
2.75481135e-01 -8.15122485e-01 1.30878866e-01 6.21708035e-01
-5.17517507e-01 -4.00762498e-01 1.16998374e-01 1.14541590e+00
2.23543897e-01 -5.89732766e-01 7.75943398e-01 -2.39639357e-01
-3.79857384e-02 2.81104594e-01 -6.40677273e-01 1.27405614e-01
1.22523439e+00 5.17045371e-02 -2.78710928e-02 -8.18656027e-01
-8.17078888e-01 5.33866525e-01 1.05345562e-01 -1.07574202e-01
2.04506725e-01 -1.07700026e+00 -3.73677045e-01 -3.96346420e-01
-4.98827219e-01 8.42035096e-03 5.87418854e-01 8.49611878e-01
2.22390778e-02 4.31452721e-01 2.17493340e-01 -1.98675111e-01
-7.86431849e-01 7.33243227e-01 4.18946594e-01 -6.81547940e-01
1.58166528e-01 8.69046807e-01 6.04546117e-03 1.80555716e-01
4.04132545e-01 -3.74183685e-01 3.90708894e-01 1.25505030e-01
5.37047386e-01 6.53219402e-01 -2.48479620e-01 1.26467824e-01
-1.23059019e-01 1.51121944e-01 -8.86438191e-02 -7.33398855e-01
1.17750096e+00 -5.14176846e-01 2.89217919e-01 6.78588629e-01
6.21485829e-01 -1.06288262e-01 -1.80286241e+00 -4.33377326e-01
6.83718622e-02 -3.45893234e-01 1.47855386e-01 -9.75809693e-01
-9.14782941e-01 7.54189670e-01 6.80216551e-01 3.30302209e-01
1.08216631e+00 -1.45419836e-01 3.89216810e-01 2.95460999e-01
6.75996006e-01 -1.52351868e+00 -2.98925400e-01 5.51090121e-01
4.95297164e-01 -1.03270411e+00 -1.30377486e-02 2.30190292e-01
-6.75695181e-01 8.99846733e-01 1.36775732e-01 3.08966283e-02
5.27891576e-01 3.38724941e-01 -4.32634830e-01 5.22411346e-01
-7.87527621e-01 -5.43609381e-01 -8.15699995e-02 1.86929867e-01
-2.72780210e-02 4.30557251e-01 -8.03035378e-01 1.03746438e+00
-3.45146507e-02 1.08497560e-01 4.29039806e-01 1.14422929e+00
-6.34779751e-01 -1.07490504e+00 -5.02249658e-01 5.71160674e-01
-1.03781843e+00 -3.73124890e-02 2.07365319e-01 6.05869114e-01
-2.29671776e-01 1.08003831e+00 6.87478334e-02 1.80558190e-01
1.96321040e-01 7.60760158e-02 7.44481206e-01 -2.20728606e-01
-1.33180723e-01 4.69090015e-01 9.42429826e-02 -3.68528426e-01
-3.86697769e-01 -7.77821720e-01 -9.61878598e-01 -4.96331662e-01
-3.51320028e-01 4.15650934e-01 7.28124499e-01 1.02010572e+00
1.73296317e-01 2.17833117e-01 8.36428642e-01 -4.25123692e-01
-1.37821531e+00 -8.40359986e-01 -9.98622477e-01 -1.64554000e-01
3.52203488e-01 -6.62802398e-01 -5.33502221e-01 -4.99294430e-01]
|
[4.470541477203369, 3.1813783645629883]
|
7e653ce8-fd90-4f1e-a772-5a43500df10d
|
memonet-memorizing-representations-of-all
|
2211.01334
| null |
https://arxiv.org/abs/2211.01334v2
|
https://arxiv.org/pdf/2211.01334v2.pdf
|
MemoNet:Memorizing Representations of All Cross Features Efficiently via Multi-Hash Codebook Network for CTR Prediction
|
New findings in natural language processing(NLP) demonstrate that the strong memorization capability contributes a lot to the success of large language models.This inspires us to explicitly bring an independent memory mechanism into CTR ranking model to learn and memorize all cross features'representations. In this paper,we propose multi-Hash Codebook NETwork(HCNet) as the memory mechanism for efficiently learning and memorizing representations of all cross features in CTR tasks.HCNet uses multi-hash codebook as the main memory place and the whole memory procedure consists of three phases: multi-hash addressing,memory restoring and feature shrinking.HCNet can be regarded as a general module and can be incorporated into any current deep CTR model.We also propose a new CTR model named MemoNet which combines HCNet with a DNN backbone.Extensive experimental results on three public datasets show that MemoNet reaches superior performance over state-of-the-art approaches and validate the effectiveness of HCNet as a strong memory module.Besides, MemoNet shows the prominent feature of big models in NLP,which means we can enlarge the size of codebook in HCNet to sustainably obtain performance gains.Our work demonstrates the importance and feasibility of learning and memorizing representations of all cross features ,which sheds light on a new promising research direction.
|
['Junlin Zhang', 'PengTao Zhang']
|
2022-10-25
| null | null | null | null |
['click-through-rate-prediction']
|
['miscellaneous']
|
[-4.45616633e-01 -2.25905269e-01 -1.91234201e-01 -1.80562809e-01
-3.65816444e-01 -3.69558781e-01 7.26141989e-01 2.27812544e-01
-7.08713710e-01 4.81481135e-01 4.08493906e-01 -2.07613975e-01
-9.13360640e-02 -1.14955056e+00 -5.40149927e-01 -6.11765563e-01
1.83929980e-01 3.36105347e-01 6.45287931e-01 -5.58894336e-01
3.73018622e-01 3.52963179e-01 -1.50198948e+00 4.78686005e-01
7.47138679e-01 8.04155946e-01 7.44881451e-01 7.48473927e-02
-5.63602507e-01 8.29346418e-01 -4.21287864e-01 -2.64313340e-01
-1.03107095e-02 1.43424511e-01 -9.26523268e-01 -5.62911272e-01
2.24639028e-01 -2.37726301e-01 -7.28198946e-01 6.26917720e-01
7.23145247e-01 3.16789269e-01 2.96187878e-01 -1.02611041e+00
-1.05421317e+00 1.00508189e+00 -3.94841045e-01 5.89410067e-01
1.22590199e-01 1.47798285e-01 1.07365215e+00 -1.23177862e+00
5.21940410e-01 1.31403792e+00 4.01773632e-01 4.28707123e-01
-6.31051481e-01 -1.04900897e+00 2.69811422e-01 4.36430097e-01
-1.56548834e+00 -2.99578875e-01 3.01213056e-01 -7.16478899e-02
1.48768353e+00 2.24514157e-02 6.20541692e-01 8.13317716e-01
4.57075149e-01 9.80197549e-01 7.77545512e-01 -3.17851096e-01
-9.64687169e-02 -5.83302230e-02 5.08124292e-01 6.65041327e-01
2.16907546e-01 -1.08758137e-01 -6.54888511e-01 -3.71115096e-02
8.24918747e-01 2.98160613e-01 -8.57623965e-02 1.68131664e-01
-1.28095317e+00 8.27839375e-01 7.38698959e-01 8.65942657e-01
-2.05209434e-01 2.41724595e-01 4.19097781e-01 4.71482307e-01
2.19128087e-01 2.49622792e-01 -4.33904767e-01 3.02250683e-02
-6.82766914e-01 -7.16044828e-02 4.67599243e-01 1.02826917e+00
7.94992149e-01 3.31976414e-02 -2.25602746e-01 1.03691649e+00
2.38908261e-01 6.85482979e-01 9.67769802e-01 -4.57942158e-01
6.35015666e-01 9.93480206e-01 -5.81336200e-01 -1.14312482e+00
-5.50019979e-01 -6.76633120e-01 -1.30210829e+00 -7.70277500e-01
-4.17061210e-01 2.80072629e-01 -8.19023669e-01 1.87486744e+00
-6.45991340e-02 2.02250585e-01 4.76372913e-02 4.63899821e-01
1.27109182e+00 1.09501922e+00 3.84377897e-01 -1.23773254e-01
1.21543765e+00 -1.00276744e+00 -4.98688191e-01 -1.79140121e-01
9.07123029e-01 -7.53526986e-01 1.11317337e+00 1.35916322e-01
-1.04066288e+00 -1.02832079e+00 -9.32013452e-01 -3.03271860e-01
-6.71344101e-01 2.03787424e-02 7.99919903e-01 1.66830137e-01
-1.25710344e+00 3.27788949e-01 -5.05585611e-01 -3.77200961e-01
2.19743922e-01 6.46059513e-01 -5.16037762e-01 -3.57104540e-01
-1.59120560e+00 9.04274106e-01 8.54346037e-01 2.70489007e-01
-7.04360664e-01 -5.58297813e-01 -8.21819067e-01 2.12757811e-01
-4.21536490e-02 -7.23311067e-01 9.10936654e-01 -4.25021052e-01
-9.27565575e-01 5.72850585e-01 -1.60274655e-01 -3.41372073e-01
-2.05868438e-01 -2.62887090e-01 -8.06288600e-01 8.94710347e-02
-1.27854258e-01 1.00844312e+00 6.85017169e-01 -8.63905787e-01
-5.77439964e-01 -2.21698303e-02 3.60186584e-02 1.23360738e-01
-9.49570894e-01 -1.52879171e-02 -4.25829500e-01 -6.01453424e-01
2.28050023e-01 -6.48419201e-01 -5.24537750e-02 -5.43274343e-01
-1.92324609e-01 -6.92950130e-01 6.57501578e-01 -2.08779439e-01
1.93115389e+00 -2.29106712e+00 -5.59752658e-02 2.77290076e-01
6.00197911e-01 5.55595398e-01 -5.98561466e-01 7.45891392e-01
1.69524863e-01 2.71966308e-01 1.32330790e-01 -1.37708768e-01
2.49713659e-02 3.66750896e-01 -6.39290214e-01 1.70040742e-01
9.45225582e-02 1.24866152e+00 -7.71387815e-01 -6.68404698e-01
2.39654537e-02 5.54392576e-01 -5.30261934e-01 1.16610743e-01
1.11285541e-02 -1.09894555e-02 -5.15060604e-01 3.73716027e-01
1.00894260e+00 -5.52242935e-01 1.22986056e-01 -8.50381479e-02
-3.29767466e-01 4.28861231e-01 -1.01571417e+00 1.64858902e+00
-3.26089472e-01 2.24926353e-01 -4.41691965e-01 -7.49101520e-01
1.19730580e+00 1.15943298e-01 9.56321284e-02 -1.25597596e+00
9.32855234e-02 1.92659244e-01 -1.21991318e-02 -2.15886861e-01
8.71869862e-01 -6.95668161e-02 -1.11906327e-01 5.56812525e-01
2.84579039e-01 4.35802281e-01 1.91797867e-01 5.81111491e-01
1.06788683e+00 -5.69446802e-01 1.68134719e-01 -3.25876057e-01
9.40502822e-01 -3.31509799e-01 5.27024627e-01 8.51507962e-01
6.59610331e-02 2.84141034e-01 3.83729637e-02 -5.73146224e-01
-7.57374108e-01 -1.04224372e+00 -7.01987445e-02 1.69873416e+00
8.55588689e-02 -7.52119899e-01 -2.68580526e-01 -3.97933275e-01
2.75535770e-02 2.62647688e-01 -5.45765817e-01 -5.37003636e-01
-1.02303016e+00 -7.35907733e-01 6.47473991e-01 8.76567602e-01
9.34564054e-01 -1.51806366e+00 -2.64076144e-01 2.58854032e-01
-4.09045905e-01 -8.87607396e-01 -4.18598771e-01 2.07292095e-01
-9.26245868e-01 -6.99957371e-01 -4.48068529e-01 -1.28185058e+00
5.23942173e-01 6.51492715e-01 1.16238034e+00 6.77047551e-01
-3.01081408e-02 1.70762420e-01 -5.65201521e-01 -9.66913328e-02
-1.54592276e-01 7.37924874e-01 2.39476487e-01 -3.68911684e-01
4.60097730e-01 -7.19525754e-01 -6.49494350e-01 3.27847213e-01
-1.08200347e+00 6.82893991e-02 9.76839125e-01 6.64350986e-01
4.06046152e-01 1.35957256e-01 7.57943034e-01 -6.98172808e-01
7.99738288e-01 -6.32714987e-01 -2.63472676e-01 3.91872466e-01
-4.58939463e-01 1.45426631e-01 7.83653975e-01 -5.13704240e-01
-9.24042463e-01 -3.65463048e-01 -2.58583009e-01 -6.65798262e-02
2.39503488e-01 7.99826860e-01 -1.56372100e-01 -8.99830535e-02
2.14587599e-01 8.17851305e-01 -3.23316246e-01 -6.61957145e-01
4.45996225e-01 5.66858411e-01 3.24460208e-01 -7.27305353e-01
8.35377455e-01 1.53585285e-01 -2.43131831e-01 -6.64799154e-01
-7.40523517e-01 -5.52706599e-01 -8.30489278e-01 1.17999442e-01
4.13405955e-01 -1.18621254e+00 -9.05213296e-01 3.80312145e-01
-1.23332310e+00 -3.39805298e-02 -5.33692390e-02 2.78982192e-01
-7.01694340e-02 2.82184958e-01 -9.88610208e-01 -2.56507039e-01
-7.10151017e-01 -8.09188008e-01 7.82926500e-01 2.08121687e-01
1.91175848e-01 -9.74512577e-01 1.92320898e-01 1.28273696e-01
7.44962692e-01 -6.11722469e-01 1.23882639e+00 -6.68957174e-01
-8.72435451e-01 2.24998906e-01 -4.87436175e-01 3.11652601e-01
-4.73049939e-01 -4.06252444e-01 -9.22896624e-01 -5.72697997e-01
-2.19109103e-01 -5.88987470e-01 1.35295963e+00 -6.03665039e-02
1.19103825e+00 -2.46394128e-01 -3.68936837e-01 4.45844173e-01
1.59742117e+00 -7.67896045e-03 9.27173257e-01 5.98523915e-01
5.92628539e-01 1.90713719e-01 5.53441584e-01 4.53049332e-01
6.65302813e-01 3.00056040e-01 2.26935461e-01 -8.78247768e-02
-1.09367676e-01 -5.13527095e-01 5.14227450e-01 2.03569174e+00
1.13868974e-01 -3.83240193e-01 -9.06720996e-01 4.73238260e-01
-1.76541173e+00 -1.05239856e+00 2.84617525e-02 1.82878530e+00
8.65534186e-01 1.66327834e-01 -2.99511284e-01 1.06726035e-01
6.62021518e-01 3.97251636e-01 -1.67598560e-01 -4.19290274e-01
-5.51471293e-01 3.78425330e-01 9.89422277e-02 1.63582623e-01
-9.37563241e-01 1.30388927e+00 6.25372553e+00 1.22759914e+00
-1.22409737e+00 3.45967442e-01 3.06129217e-01 2.55994171e-01
-3.24967355e-01 -9.98949185e-02 -1.40205657e+00 3.42873335e-01
1.07344329e+00 -1.99827537e-01 1.86936140e-01 5.79896331e-01
-1.95980608e-01 2.01111600e-01 -8.54630888e-01 1.02299833e+00
2.92511638e-02 -1.44678438e+00 9.17455971e-01 8.77053067e-02
5.47269762e-01 3.97892177e-01 2.17565522e-01 8.79299700e-01
6.47189096e-02 -1.05386221e+00 3.99029195e-01 5.30985773e-01
7.40794659e-01 -9.12985444e-01 8.88738036e-01 5.01514077e-01
-1.75119054e+00 -4.02606875e-01 -1.16066384e+00 -7.19921514e-02
2.62442119e-02 4.81645733e-01 -4.62081224e-01 4.29006577e-01
4.69175965e-01 9.12642717e-01 -9.30345535e-01 9.71711695e-01
-7.33230636e-02 4.73386735e-01 -2.53357917e-01 -5.89928441e-02
4.32375491e-01 3.56059909e-01 -4.68408652e-02 1.57822442e+00
2.59935588e-01 3.22412044e-01 3.64082873e-01 6.14355266e-01
-3.75998139e-01 4.61067259e-01 -4.85654444e-01 -2.92881839e-02
9.24972951e-01 1.35321987e+00 -5.84798515e-01 -4.03073549e-01
-4.73474115e-01 5.97715437e-01 7.64339387e-01 3.07385158e-02
-6.81990743e-01 -4.72661197e-01 3.15131605e-01 1.38073936e-01
4.67548430e-01 -4.28948194e-01 1.51918873e-01 -1.19998276e+00
-6.22439235e-02 -6.87097847e-01 5.18392205e-01 -6.58982933e-01
-1.15063632e+00 9.35411990e-01 -2.19620392e-01 -1.19251740e+00
2.08871678e-01 -5.41071177e-01 -6.27725184e-01 6.11615658e-01
-1.84935749e+00 -1.29193091e+00 -2.38184407e-01 1.13742995e+00
4.37689036e-01 -3.52757782e-01 8.66085947e-01 4.78641182e-01
-5.89936495e-01 7.90364802e-01 1.44203365e-01 1.66961521e-01
5.93290925e-01 -7.41506934e-01 3.44298720e-01 6.49178088e-01
1.54782951e-01 1.38924038e+00 -4.66533229e-02 -6.03254497e-01
-1.57043600e+00 -1.19856954e+00 1.43734789e+00 -2.58877307e-01
6.91300869e-01 -5.10251522e-01 -1.08576918e+00 6.33161783e-01
3.34817022e-01 4.48014103e-02 7.16286182e-01 1.51894644e-01
-6.04843855e-01 -3.59618157e-01 -6.47876263e-01 2.48477042e-01
1.19345152e+00 -6.34211838e-01 -9.67272818e-01 2.12756798e-01
1.15661967e+00 1.23313516e-02 -8.79337728e-01 3.54076654e-01
3.44403416e-01 -6.46697342e-01 1.05885577e+00 -4.64780480e-01
3.35926503e-01 -1.50605217e-01 -2.44523436e-01 -9.72691894e-01
-7.58099616e-01 -3.02777588e-01 -1.58260047e-01 1.41620564e+00
2.50190973e-01 -7.87007213e-01 3.61867726e-01 1.28463477e-01
-1.33646682e-01 -8.26765418e-01 -9.07718599e-01 -8.62564981e-01
4.05185193e-01 -3.10265064e-01 7.39282072e-01 9.65982139e-01
7.41595998e-02 6.50364578e-01 -2.35954195e-01 -9.83850956e-02
2.18782574e-01 1.63519815e-01 4.92238581e-01 -1.47354972e+00
1.95114203e-02 -3.37088376e-01 -1.95649818e-01 -1.42654037e+00
3.46473604e-01 -1.32209647e+00 -3.53633374e-01 -1.41272843e+00
7.71409750e-01 -8.24073613e-01 -9.19511616e-01 7.86633015e-01
2.74599884e-02 8.82844403e-02 4.09637868e-01 5.03192246e-01
-1.16241515e+00 8.16512644e-01 1.36259770e+00 -1.59782901e-01
-3.01899314e-02 -4.97523040e-01 -9.75860000e-01 4.39636320e-01
8.45058203e-01 -4.69677299e-01 -6.12144232e-01 -8.76832306e-01
4.91520792e-01 -3.62519324e-01 1.83833405e-01 -1.07120705e+00
7.77939677e-01 2.22934061e-03 5.01911581e-01 -8.29432011e-01
2.78116822e-01 -7.35260904e-01 -1.67903692e-01 5.57465971e-01
-3.56652170e-01 6.85513675e-01 3.49155605e-01 2.95210421e-01
-4.26838905e-01 -2.28196643e-02 4.75868911e-01 -3.66936177e-01
-1.07490456e+00 5.15154064e-01 -2.89264649e-01 2.57396013e-01
4.86172795e-01 -4.48588617e-02 -7.78712451e-01 7.32537881e-02
-4.15851921e-01 3.32821459e-01 -1.02453224e-01 7.60167837e-01
1.13440824e+00 -1.45981824e+00 -6.17925167e-01 4.82033938e-01
1.66518465e-01 -2.81491727e-01 4.12137836e-01 6.09661460e-01
-2.40937248e-01 8.99722993e-01 -1.94155142e-01 -3.16621393e-01
-1.05212486e+00 6.33275390e-01 -1.09471038e-01 -5.55068552e-01
-7.02915490e-01 7.76718080e-01 3.99510920e-01 -5.19962132e-01
2.17425793e-01 -3.04647893e-01 -5.04575968e-01 8.57806057e-02
9.19549763e-01 2.47934669e-01 6.40939772e-02 -7.31761217e-01
-4.29628074e-01 7.03577518e-01 -6.92789853e-01 3.24712187e-01
1.52620578e+00 -2.66917139e-01 -6.65730357e-01 2.40219891e-01
1.35462224e+00 -2.09347412e-01 -3.23772699e-01 -6.00498736e-01
-1.02535598e-01 -1.57199576e-01 -5.48851267e-02 -7.07751036e-01
-1.28604770e+00 1.08780682e+00 4.01249349e-01 -2.88601637e-01
1.11624610e+00 1.66965965e-02 1.32054365e+00 8.37177455e-01
5.85528433e-01 -1.10431683e+00 3.92212272e-01 1.21981573e+00
9.24248457e-01 -8.46918881e-01 -2.28363127e-01 -1.74436629e-01
-3.25343430e-01 1.12352192e+00 1.02700150e+00 -5.56666069e-02
9.47072029e-01 1.40759617e-01 -1.11195900e-01 -2.43650347e-01
-1.17005360e+00 -8.69716555e-02 2.54367262e-01 2.61862934e-01
5.32941639e-01 -1.21571459e-01 -5.96949339e-01 8.00582051e-01
-2.42723331e-01 -6.21834956e-02 8.04754570e-02 1.08853531e+00
-1.07395291e+00 -1.27886248e+00 3.32708769e-02 3.38225812e-01
-3.70203592e-02 -7.06641018e-01 -1.78519897e-02 8.12860668e-01
4.23016250e-01 7.47197032e-01 -2.21191184e-03 -7.03624606e-01
1.61295488e-01 2.32282002e-02 1.74472302e-01 -5.76270163e-01
-9.49781358e-01 -2.15222865e-01 -4.25600111e-01 -6.45434439e-01
-2.17433140e-01 -6.75236136e-02 -1.51672518e+00 -8.22241902e-01
-3.70683759e-01 3.43914300e-01 1.88088223e-01 8.62728953e-01
5.07775486e-01 3.46241683e-01 4.75694954e-01 -3.01484764e-01
-5.10309994e-01 -1.00982559e+00 -5.51545203e-01 1.73786297e-01
5.11669703e-02 -6.23503447e-01 -1.03217401e-01 -4.21439290e-01]
|
[11.02379035949707, 8.364947319030762]
|
0fe42637-038c-45b1-a0b0-1b79a2f2844d
|
glass-global-to-local-attention-for-scene
|
2208.03364
| null |
https://arxiv.org/abs/2208.03364v1
|
https://arxiv.org/pdf/2208.03364v1.pdf
|
GLASS: Global to Local Attention for Scene-Text Spotting
|
In recent years, the dominant paradigm for text spotting is to combine the tasks of text detection and recognition into a single end-to-end framework. Under this paradigm, both tasks are accomplished by operating over a shared global feature map extracted from the input image. Among the main challenges that end-to-end approaches face is the performance degradation when recognizing text across scale variations (smaller or larger text), and arbitrary word rotation angles. In this work, we address these challenges by proposing a novel global-to-local attention mechanism for text spotting, termed GLASS, that fuses together global and local features. The global features are extracted from the shared backbone, preserving contextual information from the entire image, while the local features are computed individually on resized, high-resolution rotated word crops. The information extracted from the local crops alleviates much of the inherent difficulties with scale and word rotation. We show a performance analysis across scales and angles, highlighting improvement over scale and angle extremities. In addition, we introduce an orientation-aware loss term supervising the detection task, and show its contribution to both detection and recognition performance across all angles. Finally, we show that GLASS is general by incorporating it into other leading text spotting architectures, improving their text spotting performance. Our method achieves state-of-the-art results on multiple benchmarks, including the newly released TextOCR.
|
['R. Manmatha', 'Amir Markovitz', 'Inbal Lavi', 'Oron Anschel', 'Shahar Tsiper', 'Roi Ronen']
|
2022-08-05
| null | null | null | null |
['text-spotting']
|
['computer-vision']
|
[ 7.06125259e-01 -6.78946793e-01 -1.28839344e-01 -2.00827360e-01
-8.97681653e-01 -5.64987540e-01 7.85872340e-01 7.35350624e-02
-5.53423822e-01 7.71561041e-02 3.55135739e-01 9.49738026e-02
2.10637361e-01 -4.49276149e-01 -5.03187656e-01 -6.76787674e-01
5.84096074e-01 6.98995069e-02 2.51410484e-01 -1.38671637e-01
5.47371149e-01 4.52941775e-01 -1.51374841e+00 5.33830166e-01
7.20024228e-01 9.66619909e-01 3.84751737e-01 9.89456236e-01
-1.61925510e-01 3.81888717e-01 -5.90231717e-01 -4.06917393e-01
3.86936843e-01 -1.21500351e-01 -5.61809003e-01 2.58886844e-01
1.14530361e+00 -5.38902283e-01 -4.34371263e-01 8.55931044e-01
6.63109541e-01 1.64951719e-02 4.27420735e-01 -8.07013810e-01
-7.90274978e-01 3.38689417e-01 -8.48339677e-01 2.92817666e-03
4.35243547e-01 1.52133703e-01 1.46728325e+00 -1.50434291e+00
5.90958774e-01 1.22494733e+00 6.85903907e-01 1.16918929e-01
-1.25713766e+00 -3.54633331e-01 4.85251486e-01 5.58293425e-02
-1.54720199e+00 -3.78318816e-01 5.06220639e-01 -3.18061322e-01
1.30574107e+00 4.86818701e-01 2.85633415e-01 1.04240716e+00
2.95551628e-01 1.25042236e+00 7.81594396e-01 -7.38603473e-01
-2.41303280e-01 -9.56331044e-02 1.85068492e-02 6.14545703e-01
2.59430915e-01 -3.60507041e-01 -9.86727178e-01 1.86148390e-01
5.25974095e-01 1.63940430e-01 -3.51436943e-01 -4.00022507e-01
-1.45783138e+00 7.69873440e-01 3.18286031e-01 3.03361148e-01
3.56657431e-02 1.25047371e-01 4.13575739e-01 2.37428293e-01
6.86620176e-01 1.92446247e-01 -3.82298738e-01 -1.21198155e-01
-1.26408350e+00 1.37802348e-01 5.08067131e-01 7.39443898e-01
6.78250432e-01 -1.98290229e-01 -5.77821851e-01 1.06063986e+00
2.06332952e-01 9.55073655e-01 6.00868821e-01 2.24501695e-02
8.44714522e-01 6.92450523e-01 -1.51289582e-01 -1.21575296e+00
-4.97274846e-01 -3.48250717e-01 -5.91066897e-01 5.93661563e-03
2.87097633e-01 5.31574525e-02 -1.27063203e+00 1.36984062e+00
7.37101510e-02 -1.92276537e-01 -3.60027254e-01 8.52186203e-01
5.27981877e-01 5.82444727e-01 -1.65308416e-01 2.47270644e-01
1.51601911e+00 -1.24398732e+00 -7.52084732e-01 -7.10701883e-01
7.02419817e-01 -1.21537924e+00 1.23745799e+00 3.80532533e-01
-7.77059674e-01 -4.27276403e-01 -1.19900203e+00 -5.85678220e-01
-7.52517879e-01 6.24246478e-01 1.24227758e-02 7.40605474e-01
-1.12326407e+00 3.29223692e-01 -8.20835471e-01 -7.96673000e-01
9.75863859e-02 3.14549267e-01 -1.94334850e-01 -1.62207246e-01
-8.18302989e-01 7.42466867e-01 7.24030435e-02 4.48540822e-02
-2.92981207e-01 -2.62444049e-01 -6.84134245e-01 1.98492721e-01
4.35499787e-01 -4.26259637e-01 9.69872713e-01 -9.04105306e-01
-1.37842524e+00 8.22336197e-01 -4.23485875e-01 -2.33866289e-01
5.89849114e-01 -6.91764176e-01 -3.36931288e-01 1.70203149e-01
2.57186234e-01 6.00071192e-01 1.47828102e+00 -7.95593023e-01
-6.24106765e-01 -4.47680324e-01 -5.32923043e-01 4.86417830e-01
-7.99373865e-01 2.30266944e-01 -9.21931863e-01 -1.17792737e+00
2.52031125e-02 -7.70164430e-01 2.92960256e-01 2.53248870e-01
-5.16679525e-01 -2.38444693e-02 1.32129097e+00 -8.54538083e-01
1.34971213e+00 -2.16055036e+00 2.75935858e-01 2.37824675e-02
1.49160728e-01 2.04580128e-01 -4.00693893e-01 5.51194608e-01
-5.53356893e-02 9.14661139e-02 -1.31370932e-01 -7.42022574e-01
1.04100749e-01 -2.62224525e-01 -5.94257534e-01 5.04166961e-01
4.80880469e-01 1.08784616e+00 -4.32033747e-01 -4.60177362e-01
4.75816190e-01 5.36984563e-01 -2.75528342e-01 -1.77332014e-01
-1.72220364e-01 -1.64764538e-01 -3.42390418e-01 7.95450985e-01
6.86998129e-01 -3.84946615e-01 7.60649815e-02 -1.90823734e-01
-7.30167925e-02 2.21747443e-01 -1.19991958e+00 1.76170361e+00
-3.80401790e-01 1.09422278e+00 1.98662758e-01 -7.10485876e-01
8.67494524e-01 9.13786329e-03 2.87455946e-01 -8.66631269e-01
7.52457082e-02 1.44866422e-01 -5.49639046e-01 -2.74620622e-01
9.83398020e-01 3.36314499e-01 -1.10605314e-01 6.49897277e-01
-1.44812195e-02 -1.40149906e-01 2.05363914e-01 3.29286426e-01
1.05664504e+00 1.13904424e-01 1.45778015e-01 -6.45009354e-02
3.88470262e-01 -2.00966105e-01 -9.64746401e-02 8.88090074e-01
2.48253383e-02 1.03874505e+00 3.56146812e-01 -4.59507495e-01
-1.08060014e+00 -7.43010461e-01 -1.81269303e-01 1.69965899e+00
7.86300153e-02 -7.66874909e-01 -5.34394860e-01 -8.72800767e-01
1.81671768e-01 3.25118929e-01 -8.10909152e-01 5.28363138e-02
-7.14063525e-01 -9.53544796e-01 5.82181633e-01 5.98824859e-01
5.22075593e-01 -6.73152924e-01 -6.57677174e-01 9.06424597e-02
-2.01209828e-01 -1.33694983e+00 -1.06412721e+00 2.96272844e-01
-5.89991391e-01 -5.81520021e-01 -1.04275417e+00 -6.67883813e-01
4.36807960e-01 7.55548060e-01 8.71318400e-01 6.69319369e-03
-6.60815537e-01 4.59974289e-01 -6.74012244e-01 -1.30561948e-01
9.22665223e-02 2.94641048e-01 -3.16981435e-01 4.02648151e-01
1.40430793e-01 5.77669311e-03 -5.70182145e-01 4.67060924e-01
-1.06991458e+00 2.15673044e-01 7.42971420e-01 9.88247097e-01
3.21086854e-01 -2.84758031e-01 1.30266324e-01 -4.68746930e-01
6.13471806e-01 1.90617993e-01 -4.01626199e-01 3.84243011e-01
-4.64528680e-01 6.49828613e-02 5.12126446e-01 -3.50157559e-01
-7.43395925e-01 2.66729921e-01 2.19109938e-01 -1.94394007e-01
1.22440875e-01 4.16493386e-01 -9.52375308e-02 -2.43902996e-01
6.38817549e-01 5.06789207e-01 -1.78045735e-01 -5.77561021e-01
5.08582413e-01 8.31984580e-01 2.46668056e-01 -3.11257601e-01
8.76032829e-01 6.58963561e-01 -1.73474953e-01 -1.16555727e+00
-8.17710221e-01 -7.42796004e-01 -7.25788295e-01 1.43185496e-01
7.21292257e-01 -9.69682693e-01 -2.05091119e-01 8.53165150e-01
-1.03130865e+00 -3.17027211e-01 -1.08902864e-01 4.19216827e-02
-3.37394238e-01 7.44682074e-01 -4.20148939e-01 -6.10850811e-01
-5.87687969e-01 -1.09569287e+00 1.98327947e+00 -7.80516351e-03
-6.46652281e-02 -9.06618953e-01 -2.17480343e-02 1.52855933e-01
6.04606807e-01 -1.72124460e-01 6.70782685e-01 -6.73777401e-01
-5.78313172e-01 -4.72798020e-01 -7.08256781e-01 1.96110353e-01
1.38823867e-01 1.35886716e-02 -9.47336972e-01 -5.92974782e-01
-3.70206267e-01 -2.93770164e-01 1.40175664e+00 7.98877999e-02
8.59947503e-01 -6.13681301e-02 -3.82831454e-01 6.09821618e-01
1.37448847e+00 -3.69294852e-01 3.66899222e-01 3.60544741e-01
1.06902552e+00 2.94034898e-01 4.68580574e-01 4.11108762e-01
7.03108013e-02 9.87101018e-01 1.72255322e-01 -3.32704812e-01
-4.80702281e-01 -8.15094635e-02 5.55162132e-01 5.27432680e-01
4.49283779e-01 -5.44723988e-01 -9.53580260e-01 4.72927094e-01
-2.03260899e+00 -6.83159292e-01 6.98684901e-02 2.13663387e+00
6.08647168e-01 3.65619585e-02 9.51453522e-02 1.46645889e-01
8.32981765e-01 6.19652450e-01 -6.54885113e-01 -2.47496054e-01
-5.99020660e-01 3.91297698e-01 6.63114846e-01 5.11122644e-01
-1.43820786e+00 1.26414967e+00 6.62011194e+00 1.05619109e+00
-1.53914285e+00 -1.83981076e-01 4.14163619e-01 -3.08805406e-01
1.65237546e-01 -2.39754632e-01 -9.95283008e-01 1.25361517e-01
5.23223758e-01 2.95754056e-02 4.42927122e-01 5.56541741e-01
5.56007251e-02 -8.61977637e-02 -9.79646444e-01 9.88228083e-01
5.85462630e-01 -1.01592159e+00 2.35323012e-01 -7.66866282e-02
6.99903727e-01 2.57929921e-01 3.16429973e-01 -5.76342130e-03
-3.10025760e-03 -1.13476408e+00 8.15218925e-01 2.25130081e-01
1.17452490e+00 -4.11169142e-01 3.76151413e-01 -4.43219244e-02
-1.51672232e+00 -1.22878753e-01 -1.19646341e-01 1.74488351e-01
-1.89365581e-01 6.95280313e-01 -9.06351388e-01 5.59158802e-01
5.38114548e-01 1.01380599e+00 -1.06770861e+00 8.10088456e-01
-9.11362618e-02 3.06962550e-01 -6.15406454e-01 -1.87518701e-01
2.62418389e-01 2.07514063e-01 5.52031696e-01 1.70831800e+00
3.18556577e-01 -6.30068898e-01 2.37756133e-01 7.46200085e-01
-2.15684876e-01 3.86780590e-01 -3.61993253e-01 -1.79486066e-01
1.34349242e-01 1.34684920e+00 -9.92791533e-01 -2.64452964e-01
-5.30899465e-01 1.42570329e+00 2.00635225e-01 3.97384226e-01
-5.73629379e-01 -8.97828102e-01 3.90327156e-01 -1.47320688e-01
7.86399484e-01 -2.78490573e-01 -6.40396833e-01 -1.45086336e+00
5.18981397e-01 -9.89887655e-01 2.80615956e-01 -7.54614174e-01
-1.01920533e+00 3.03699046e-01 -3.95466715e-01 -9.75635290e-01
8.06139261e-02 -9.68988001e-01 -5.05440772e-01 9.91791546e-01
-1.49246442e+00 -1.50720036e+00 -4.60028142e-01 5.65225124e-01
9.53152657e-01 6.72052577e-02 6.06260657e-01 1.28591359e-01
-8.20646942e-01 9.92427289e-01 4.05198544e-01 3.03548515e-01
1.25889432e+00 -1.19643509e+00 1.03834391e+00 1.08290136e+00
4.42358166e-01 3.96107048e-01 4.53083515e-01 -7.13061929e-01
-1.58867121e+00 -1.05855858e+00 8.64918828e-01 -6.20479107e-01
7.18086600e-01 -8.38461578e-01 -8.94084811e-01 6.10892534e-01
1.40126437e-01 -2.54203775e-03 1.40198469e-01 1.42905101e-01
-7.93515742e-01 8.18891600e-02 -7.71204352e-01 7.34444559e-01
8.24130595e-01 -7.92680800e-01 -2.32402161e-01 4.47732091e-01
6.86194360e-01 -4.09835666e-01 -3.69544685e-01 5.49406856e-02
7.91157603e-01 -8.35096478e-01 9.00138855e-01 -1.61065564e-01
3.01855534e-01 -2.02815831e-01 -2.70654410e-01 -9.57787752e-01
-2.31188208e-01 -7.72348523e-01 9.47200507e-02 9.55610573e-01
4.68173832e-01 -6.08255446e-01 8.26684654e-01 1.01104714e-01
5.15775867e-02 -5.34820616e-01 -1.02946448e+00 -5.52466512e-01
2.01381773e-01 -3.18249732e-01 2.55389392e-01 7.86904573e-01
1.18372120e-01 5.22890627e-01 -5.59222519e-01 -9.34813693e-02
2.48539343e-01 2.34995365e-01 7.27125943e-01 -9.74033833e-01
-2.02699095e-01 -6.91123962e-01 -4.06178027e-01 -1.49272597e+00
-2.15759441e-01 -8.47148836e-01 2.19178647e-01 -1.36167109e+00
2.83854097e-01 1.02848917e-01 -1.36959255e-02 6.30772769e-01
-5.57637453e-01 4.30056930e-01 5.39008439e-01 3.00057620e-01
-8.10068190e-01 5.35853446e-01 1.14214373e+00 -2.58517295e-01
-9.06461626e-02 -4.53216791e-01 -5.35942078e-01 4.91390556e-01
6.65965497e-01 -3.48730713e-01 1.39657572e-01 -8.87239873e-01
3.14190000e-01 -4.26007718e-01 4.08045471e-01 -9.67503071e-01
4.45977420e-01 1.21863224e-01 5.86503088e-01 -7.45571494e-01
4.16796237e-01 -6.62670195e-01 -7.00162590e-01 1.32379904e-01
-3.74913990e-01 6.62096664e-02 3.88907164e-01 6.44468665e-01
3.69416848e-02 4.78487611e-02 8.31274211e-01 4.95537609e-01
-4.08502102e-01 -3.80819812e-02 -1.77395150e-01 -5.14279231e-02
6.56790376e-01 -2.85972804e-01 -5.31661808e-01 -2.54274547e-01
-4.14396048e-01 8.20636377e-02 5.93903124e-01 6.86941087e-01
5.67996025e-01 -1.00956941e+00 -8.89573514e-01 4.59543526e-01
3.13098997e-01 -1.77594751e-01 7.01255677e-03 9.46289480e-01
-4.52080965e-01 7.63300598e-01 8.78607258e-02 -8.40961337e-01
-1.42912924e+00 4.16188270e-01 2.21753806e-01 -5.96629083e-01
-7.51730323e-01 7.70996153e-01 3.10067326e-01 -2.25631714e-01
3.77632171e-01 -4.17184711e-01 2.46528044e-01 1.53136253e-01
7.06512094e-01 2.79260010e-01 5.44460475e-01 -6.29740417e-01
-3.08236122e-01 1.22271121e+00 -5.58724761e-01 -2.33745128e-01
9.92841363e-01 -3.69763166e-01 1.55183688e-01 2.75072813e-01
1.20047724e+00 3.36864531e-01 -1.13160598e+00 -5.92615247e-01
-3.27658653e-02 -5.67708790e-01 2.40500063e-01 -1.10037029e+00
-9.26717103e-01 1.05949545e+00 6.85261369e-01 2.51835078e-01
1.18130422e+00 -3.81120026e-01 8.31770897e-01 5.45547009e-01
-1.17824651e-01 -1.32162285e+00 3.34207267e-01 7.62811959e-01
9.43181753e-01 -1.22038138e+00 3.17286342e-01 -3.31381530e-01
-4.30967718e-01 1.29162073e+00 1.47659734e-01 -9.82069150e-02
1.52616963e-01 4.29923594e-01 -3.03134471e-02 5.10754576e-03
-6.95825994e-01 -3.00432205e-01 4.68079269e-01 3.12774539e-01
4.26347017e-01 -9.47013944e-02 2.12629586e-02 1.77710444e-01
1.68744802e-01 -3.62100571e-01 2.48044580e-01 1.05137527e+00
-6.66170955e-01 -1.07389784e+00 -7.05371499e-01 5.53464234e-01
-5.08229673e-01 -4.59386617e-01 -9.47300851e-01 7.01034486e-01
-2.77630836e-01 8.46764922e-01 7.28835911e-02 -3.68696660e-01
2.68692553e-01 2.56443948e-01 2.77159810e-01 -4.97928023e-01
-6.20004535e-01 4.65128869e-01 -1.16257086e-01 -5.33825934e-01
4.35069017e-02 -6.94714427e-01 -8.63223076e-01 -1.94669247e-01
-6.38511002e-01 -4.91352201e-01 9.08442557e-01 6.96835399e-01
6.14514887e-01 4.53896314e-01 6.12962902e-01 -1.14649832e+00
-5.65500319e-01 -9.90436494e-01 -4.38108981e-01 3.22800189e-01
6.17432475e-01 -3.54342490e-01 -5.16037583e-01 1.72119796e-01]
|
[11.918320655822754, 2.2436444759368896]
|
704d1284-a71d-475e-bada-6dcd15b477c2
|
mi-segnet-mutual-information-based-us
|
2303.12649
| null |
https://arxiv.org/abs/2303.12649v1
|
https://arxiv.org/pdf/2303.12649v1.pdf
|
MI-SegNet: Mutual Information-Based US Segmentation for Unseen Domain Generalization
|
Generalization capabilities of learning-based medical image segmentation across domains are currently limited by the performance degradation caused by the domain shift, particularly for ultrasound (US) imaging. The quality of US images heavily relies on carefully tuned acoustic parameters, which vary across sonographers, machines, and settings. To improve the generalizability on US images across domains, we propose MI-SegNet, a novel mutual information (MI) based framework to explicitly disentangle the anatomical and domain feature representations; therefore, robust domain-independent segmentation can be expected. Two encoders are employed to extract the relevant features for the disentanglement. The segmentation only uses the anatomical feature map for its prediction. In order to force the encoders to learn meaningful feature representations a cross-reconstruction method is used during training. Transformations, specific to either domain or anatomy are applied to guide the encoders in their respective feature extraction task. Additionally, any MI present in both feature maps is punished to further promote separate feature spaces. We validate the generalizability of the proposed domain-independent segmentation approach on several datasets with varying parameters and machines. Furthermore, we demonstrate the effectiveness of the proposed MI-SegNet serving as a pre-trained model by comparing it with state-of-the-art networks.
|
['Nassir Navab', 'Angelos Karlas', 'Reza Ghotbi', 'Ricarda Clarenbach', 'Zhongliang Jiang', 'Yuan Bi']
|
2023-03-22
| null | null | null | null |
['anatomy']
|
['miscellaneous']
|
[ 6.16588950e-01 3.56046021e-01 -1.64000809e-01 -5.15118539e-01
-1.16413689e+00 -5.52441120e-01 2.14350000e-01 4.57049683e-02
-4.96069998e-01 3.73569995e-01 2.04510957e-01 -2.53648728e-01
-2.86491871e-01 -5.17062545e-01 -7.84787774e-01 -9.05157506e-01
-2.71442056e-01 1.56660497e-01 2.54132569e-01 9.19837505e-02
-1.26245722e-01 1.89348236e-01 -1.08209443e+00 1.09182328e-01
1.17703927e+00 1.12291193e+00 4.61218059e-01 5.35600543e-01
1.35150507e-01 3.36926222e-01 -4.58055615e-01 -2.92829663e-01
3.96265715e-01 -4.75804865e-01 -7.81604886e-01 -8.20938647e-02
2.85022020e-01 -3.56037229e-01 -4.25535202e-01 1.13448262e+00
6.27488494e-01 1.75842404e-01 9.19918776e-01 -7.50313222e-01
-4.17004645e-01 7.13870406e-01 -3.97547036e-01 2.04895586e-01
1.73705965e-01 8.98222178e-02 1.03993130e+00 -4.33927596e-01
6.05473936e-01 8.29943180e-01 3.82534683e-01 6.66998625e-01
-1.24636626e+00 -7.57272720e-01 3.33230346e-02 3.67677808e-02
-1.19942188e+00 -2.90490627e-01 9.07076955e-01 -4.70104754e-01
4.50309306e-01 2.58587569e-01 3.99905413e-01 1.07964516e+00
3.66429508e-01 7.84657001e-01 8.64918709e-01 -2.55334616e-01
2.25122154e-01 3.28805268e-01 -1.02644593e-01 8.45642686e-01
1.08790502e-01 1.17480919e-01 -3.89536679e-01 1.82845630e-04
1.01730847e+00 -3.10151637e-01 -5.27096868e-01 -6.75729990e-01
-1.24373937e+00 6.42901063e-01 5.36613584e-01 4.74878192e-01
-3.56325656e-01 -1.08584851e-01 4.87643182e-01 1.90312609e-01
1.21379077e-01 7.41884470e-01 -3.38377088e-01 1.55281872e-02
-8.33749533e-01 -1.74466729e-01 6.79253519e-01 8.84074986e-01
5.24836719e-01 -5.67908958e-02 -2.09379226e-01 8.49231780e-01
2.25990921e-01 3.32309574e-01 7.39501476e-01 -8.09212208e-01
4.01683509e-01 3.26216221e-01 -2.78985828e-01 -1.15231371e+00
-5.15970886e-01 -7.74124265e-01 -9.02884364e-01 -3.11751068e-01
3.03555250e-01 -2.16190919e-01 -1.21017504e+00 1.90705228e+00
2.15936542e-01 5.80853939e-01 2.91129291e-01 1.14414001e+00
1.15741277e+00 1.37791917e-01 2.57304400e-01 1.16075054e-01
1.30949998e+00 -6.38998508e-01 -4.91851568e-01 -3.15567195e-01
7.21715927e-01 -6.11597419e-01 9.07144666e-01 3.60874921e-01
-9.66017246e-01 -6.70390844e-01 -1.28188705e+00 1.55696645e-01
7.46903569e-02 1.10337898e-01 3.73158813e-01 6.08271122e-01
-7.07176447e-01 7.48628139e-01 -1.28727841e+00 -2.62830593e-02
3.91144097e-01 6.08295560e-01 -5.76591372e-01 1.38311312e-01
-1.38869810e+00 8.84213865e-01 3.38044316e-01 3.87481600e-02
-8.99257720e-01 -7.35078335e-01 -1.10909188e+00 1.26680136e-02
1.00569285e-01 -6.94307446e-01 8.66932631e-01 -7.86064029e-01
-1.55394423e+00 9.06640053e-01 9.46004838e-02 -4.06809121e-01
3.66276473e-01 -2.68174652e-02 -4.15594727e-01 5.72417498e-01
1.49643451e-01 6.04460120e-01 9.75984812e-01 -1.28186774e+00
-3.76541495e-01 -1.58043981e-01 -2.26570033e-02 3.17449450e-01
-4.54302371e-01 -4.05216515e-01 -4.20363545e-01 -7.37046599e-01
5.84599674e-01 -9.48856831e-01 -3.33610743e-01 -2.13753164e-01
-5.84000647e-01 2.59183347e-01 2.50953943e-01 -8.00589621e-01
1.08016562e+00 -2.43259335e+00 3.31982076e-01 3.94995391e-01
2.75786787e-01 -3.44095416e-02 -3.78763258e-01 -2.61406302e-01
-1.19722456e-01 -2.15290859e-03 -5.72416663e-01 -2.52152026e-01
-2.63445258e-01 3.84941667e-01 2.07000256e-01 6.10238910e-01
3.37660521e-01 5.15858293e-01 -9.38646317e-01 -7.44862378e-01
2.81323195e-01 2.95765340e-01 -7.30320454e-01 4.32455659e-01
2.67767221e-01 1.06124091e+00 -6.68586075e-01 2.45855898e-01
7.85209000e-01 -2.44803771e-01 3.28314245e-01 -6.92785800e-01
3.56166720e-01 4.74394232e-01 -1.02625954e+00 2.19287586e+00
-7.67705917e-01 2.98876762e-01 7.53055960e-02 -1.49369216e+00
8.38066936e-01 5.11056781e-01 7.40365863e-01 -6.34141207e-01
3.00168157e-01 3.72865885e-01 5.10909021e-01 -7.58167207e-01
-4.35720161e-02 -2.95168847e-01 -1.52443975e-01 2.97129124e-01
5.83244681e-01 -2.05447361e-01 -2.93371212e-02 3.75248641e-02
9.17539239e-01 2.38382770e-03 1.24745861e-01 -4.39217299e-01
6.24893725e-01 -3.47776771e-01 7.24661291e-01 7.27242529e-01
-3.92670006e-01 9.07076955e-01 3.27259570e-01 -7.07456619e-02
-5.66896319e-01 -1.25548840e+00 -5.89322448e-01 8.47923517e-01
5.48232794e-01 1.62284493e-01 -6.61200523e-01 -1.09868550e+00
-1.89171806e-01 7.00921059e-01 -7.14771867e-01 -4.57458794e-01
-6.55426979e-01 -7.17295766e-01 5.39936066e-01 5.08569777e-01
4.26565051e-01 -6.45837307e-01 -7.29431450e-01 2.40129620e-01
-3.09861839e-01 -1.37750578e+00 -3.72549832e-01 3.88592184e-01
-9.03864264e-01 -1.02212322e+00 -7.65918255e-01 -8.83983910e-01
6.33729458e-01 -1.83938831e-01 8.61882031e-01 -3.41219962e-01
-2.28824556e-01 3.21736276e-01 -3.22735161e-01 2.88527645e-02
-7.22008109e-01 4.19805527e-01 -1.71236128e-01 7.74517469e-03
7.44688064e-02 -7.50359535e-01 -9.67701018e-01 2.92480230e-01
-9.66612577e-01 -1.02079548e-02 7.56288767e-01 9.97701347e-01
4.05480027e-01 -2.37043768e-01 7.04213679e-01 -9.44842458e-01
3.62012327e-01 -6.10836744e-01 -6.95333853e-02 1.39960617e-01
-2.64462054e-01 2.92746007e-01 5.07384002e-01 -4.01105165e-01
-1.10842264e+00 3.46215144e-02 -2.96760857e-01 -2.37011060e-01
-3.61065298e-01 6.57216549e-01 -3.68880518e-02 -1.46034971e-01
7.41188705e-01 4.46035504e-01 6.84809685e-02 -3.30300689e-01
2.20813081e-01 7.86275804e-01 6.46232963e-01 -5.58907866e-01
5.18065095e-01 1.91550344e-01 -1.16489507e-01 -6.99262857e-01
-5.92520535e-01 -4.04641241e-01 -7.85829782e-01 2.11217776e-02
9.64185536e-01 -8.60306084e-01 -1.37411714e-01 1.22727007e-01
-1.00581014e+00 -7.54607394e-02 1.46077927e-02 9.49952543e-01
-5.74272990e-01 4.97518927e-01 -6.17343545e-01 -3.64856809e-01
-2.64035136e-01 -1.69668174e+00 9.77043986e-01 2.49521464e-01
-2.88119018e-01 -1.17233253e+00 -1.23929329e-01 2.41938651e-01
2.36028492e-01 2.63952732e-01 1.08350623e+00 -1.07306743e+00
-2.52600789e-01 -7.89207071e-02 -1.52146339e-01 5.90142727e-01
5.47024310e-01 -3.73034239e-01 -1.00320530e+00 -2.77677625e-01
2.41477266e-01 -1.25164673e-01 7.95554161e-01 5.83127797e-01
1.28183794e+00 -5.82252443e-02 -4.25260246e-01 8.32949817e-01
1.05526412e+00 3.45205903e-01 3.36488724e-01 1.42103672e-01
5.59166908e-01 4.85407203e-01 4.24296975e-01 2.63122439e-01
3.66035432e-01 3.43181401e-01 2.72169054e-01 -3.62614810e-01
-2.26937398e-01 7.15779141e-03 -7.26483092e-02 1.09202409e+00
9.60431397e-02 3.03248502e-02 -7.68184423e-01 6.34023428e-01
-1.37048030e+00 -2.05152601e-01 4.28431571e-01 2.24505162e+00
1.19223881e+00 1.48128092e-01 -2.51773775e-01 -1.18059725e-01
4.53952044e-01 1.17402270e-01 -5.96422493e-01 -1.83562890e-01
2.82881379e-01 4.34087247e-01 5.19972384e-01 3.44413668e-01
-1.30650663e+00 6.49548173e-01 5.98513126e+00 8.53222728e-01
-1.49962616e+00 7.24671632e-02 6.24412894e-01 2.08491325e-01
-3.23211014e-01 -3.57653439e-01 -1.87857360e-01 3.96265030e-01
6.71110034e-01 1.88489810e-01 4.24996093e-02 6.98803186e-01
1.53663475e-02 -1.53010506e-02 -1.32471347e+00 7.71672010e-01
-6.65640235e-02 -9.59847510e-01 -1.40127614e-01 -2.72844017e-01
6.34082317e-01 -1.34669870e-01 3.25753033e-01 1.43634707e-01
-1.54721318e-02 -9.90117311e-01 3.12499195e-01 3.12194437e-01
9.50321078e-01 -4.96460617e-01 7.90167034e-01 1.61834911e-01
-8.50749969e-01 1.44470096e-01 -8.65092427e-02 5.03199577e-01
2.51828898e-02 5.16911149e-01 -1.16092443e+00 7.91584253e-01
4.50984955e-01 5.87621868e-01 -2.37731531e-01 8.76690507e-01
-1.37277067e-01 6.58761859e-01 -3.31248522e-01 3.71332318e-01
1.91600055e-01 -1.41026676e-02 6.64421320e-01 1.40852880e+00
1.94872037e-01 -1.78939924e-02 1.66147858e-01 9.00636375e-01
4.87852432e-02 5.65780476e-02 -3.44919592e-01 1.60457850e-01
4.73873854e-01 1.16567636e+00 -5.76600790e-01 -2.41170228e-01
-4.33435142e-01 1.04167509e+00 -6.44852966e-02 4.52156693e-01
-9.43059623e-01 -5.08449078e-01 5.54013133e-01 -9.51392949e-02
2.15093866e-01 -1.10954098e-01 -4.94818807e-01 -1.14373994e+00
-2.21160889e-01 -8.11075389e-01 3.00038338e-01 -1.40920982e-01
-1.16169286e+00 8.88182819e-01 8.95269215e-02 -1.46842778e+00
-5.46038389e-01 -4.52171683e-01 -3.06943089e-01 9.79131699e-01
-1.68113792e+00 -1.06053782e+00 -2.02316299e-01 4.33773667e-01
3.22144598e-01 -4.63335253e-02 8.68560076e-01 4.73196983e-01
-5.43951750e-01 9.92508173e-01 -5.47684059e-02 4.28806543e-01
8.41656148e-01 -1.25393951e+00 9.56512894e-03 7.31816411e-01
-7.91975707e-02 7.53233194e-01 7.00876117e-01 -4.91785944e-01
-1.20812297e+00 -8.59780788e-01 1.76713660e-01 -2.76470259e-02
4.72463101e-01 -3.07062775e-01 -9.49498892e-01 5.76957166e-01
-3.26407515e-02 1.76539436e-01 9.27619219e-01 -5.34152240e-02
-1.74725920e-01 -5.57442531e-02 -1.27555466e+00 4.64781761e-01
8.31999183e-01 -4.55006301e-01 -5.88175178e-01 -5.27828187e-02
7.32486367e-01 -7.25393414e-01 -1.25738823e+00 6.72659218e-01
7.52074838e-01 -8.21423054e-01 9.83573616e-01 -3.73189688e-01
6.46943808e-01 9.03427303e-02 -1.08356237e-01 -1.43011665e+00
-1.71247169e-01 -3.04125488e-01 2.01890036e-01 8.30263376e-01
6.07196093e-01 -7.74685502e-01 7.75116324e-01 6.07306719e-01
-3.33341658e-01 -7.31279373e-01 -1.13656926e+00 -5.39520264e-01
2.97058940e-01 -4.76153940e-01 4.75725770e-01 1.00176132e+00
1.86987177e-01 1.34528175e-01 -6.34055883e-02 4.43784416e-01
5.54400623e-01 1.60588592e-01 3.52756858e-01 -9.18293357e-01
-6.28055930e-01 -3.91787946e-01 -4.73118216e-01 -1.22786868e+00
3.45537513e-02 -1.03361654e+00 3.41074944e-01 -1.17089438e+00
2.96507422e-02 -8.40035319e-01 -6.18476033e-01 2.43831396e-01
-2.44842410e-01 1.16590314e-01 8.66996199e-02 7.05899745e-02
-2.91692615e-01 4.22667533e-01 1.79000592e+00 -8.81092548e-02
-2.91792452e-01 7.14127719e-02 -6.54861927e-01 7.13151217e-01
5.54276168e-01 -4.54846472e-01 -4.74843860e-01 -4.60630924e-01
-3.02266657e-01 4.47472334e-01 1.58299953e-01 -1.03158844e+00
1.95828751e-01 1.34517223e-01 2.66234875e-01 -4.54273038e-02
1.97212994e-01 -8.10462236e-01 -3.25310796e-01 5.42857766e-01
-5.90915143e-01 -6.33388400e-01 2.27617174e-01 3.43487710e-01
-4.47501361e-01 -3.34750295e-01 9.15973783e-01 -1.57301724e-01
-4.72813904e-01 3.08120728e-01 -1.79864928e-01 4.19659019e-02
6.90684974e-01 -2.72353768e-01 3.29478562e-01 -2.54975587e-01
-1.11797845e+00 7.62572289e-02 -1.14699610e-01 2.40950018e-01
6.37690246e-01 -9.91789401e-01 -6.68738842e-01 5.89773059e-01
7.98298884e-03 2.37973660e-01 5.73635399e-01 1.14020920e+00
-3.54095846e-01 3.14739615e-01 -1.70683607e-01 -8.93510640e-01
-8.57507050e-01 1.15766332e-01 5.15434146e-01 -3.59008193e-01
-3.80589813e-01 9.83526826e-01 7.02453256e-01 -5.40922105e-01
1.78871661e-01 -6.48960233e-01 -2.94385254e-01 -1.94665149e-01
1.33551478e-01 8.14860016e-02 2.43039146e-01 -4.39006686e-01
-4.76723224e-01 5.53753555e-01 -2.38117099e-01 -9.72558558e-02
1.14293623e+00 -1.92880169e-01 2.90085852e-01 7.22464770e-02
1.53869009e+00 -2.67951757e-01 -1.23391628e+00 -3.27112705e-01
-2.42410064e-01 -2.40257204e-01 3.11897397e-01 -8.85882020e-01
-1.24844778e+00 1.03875041e+00 8.83086562e-01 -1.06133902e-02
1.40773261e+00 9.54672229e-03 9.40808892e-01 -1.95712045e-01
1.19374856e-01 -7.95287788e-01 -6.51350915e-02 1.02244228e-01
7.10469604e-01 -1.34251332e+00 -3.86706293e-01 -6.42774045e-01
-7.71749437e-01 1.06415868e+00 5.34870267e-01 -2.54420161e-01
7.80075669e-01 3.82072568e-01 1.44021958e-01 -8.68499279e-04
-6.11894131e-02 -7.25749880e-02 7.41016805e-01 6.89554214e-01
5.72928727e-01 1.10164598e-01 -2.87810653e-01 1.04488778e+00
-2.22134203e-01 -6.88782483e-02 1.90395817e-01 6.44912243e-01
-1.93270698e-01 -9.75338519e-01 -1.04797795e-01 4.99768734e-01
-5.98129153e-01 -5.11036664e-02 3.52610469e-01 7.30492532e-01
2.23145857e-01 6.22608542e-01 5.70419729e-02 -5.00444293e-01
2.43578404e-01 -2.81074017e-01 6.61413312e-01 -6.59942746e-01
-4.48948056e-01 1.82434708e-01 -9.00260434e-02 -5.45117974e-01
-3.22131991e-01 -5.77035367e-01 -1.52054071e+00 4.35056567e-01
-3.93055677e-01 4.55627292e-02 6.33782625e-01 1.07410383e+00
3.68551075e-01 8.41115654e-01 7.53196716e-01 -6.32248998e-01
-7.39831507e-01 -1.01538348e+00 -5.23544371e-01 5.35703838e-01
5.16371369e-01 -8.39890063e-01 -2.55263686e-01 5.05957790e-02]
|
[14.621288299560547, -2.066981792449951]
|
a2e7fb91-a7ec-4242-990e-d450710885c4
|
not-enough-data-deep-learning-to-the-rescue
|
1911.03118
| null |
https://arxiv.org/abs/1911.03118v2
|
https://arxiv.org/pdf/1911.03118v2.pdf
|
Not Enough Data? Deep Learning to the Rescue!
|
Based on recent advances in natural language modeling and those in text generation capabilities, we propose a novel data augmentation method for text classification tasks. We use a powerful pre-trained neural network model to artificially synthesize new labeled data for supervised learning. We mainly focus on cases with scarce labeled data. Our method, referred to as language-model-based data augmentation (LAMBADA), involves fine-tuning a state-of-the-art language generator to a specific task through an initial training phase on the existing (usually small) labeled data. Using the fine-tuned model and given a class label, new sentences for the class are generated. Our process then filters these new sentences by using a classifier trained on the original data. In a series of experiments, we show that LAMBADA improves classifiers' performance on a variety of datasets. Moreover, LAMBADA significantly improves upon the state-of-the-art techniques for data augmentation, specifically those applicable to text classification tasks with little data.
|
['Naama Tepper', 'Esther Goldbraich', 'Ateret Anaby-Tavor', 'George Kour', 'Segev Shlomov', 'Naama Zwerdling', 'Boaz Carmeli', 'Amir Kantor']
|
2019-11-08
| null | null | null | null |
['lambada']
|
['natural-language-processing']
|
[ 7.30604768e-01 5.69721699e-01 -3.45633626e-01 -4.11226422e-01
-6.77460313e-01 -4.18979943e-01 9.71821547e-01 5.22306442e-01
-5.82834303e-01 9.15367544e-01 3.22184145e-01 -4.40791488e-01
6.07152224e-01 -9.35885191e-01 -5.63101053e-01 -3.80484313e-01
3.63625288e-01 7.53781557e-01 -2.09018111e-01 -4.70666498e-01
1.07935116e-01 2.72538453e-01 -1.56224883e+00 4.76971269e-01
8.66307020e-01 5.01027465e-01 3.22259367e-02 7.66344488e-01
-7.27449179e-01 9.31394935e-01 -8.35331082e-01 -2.67881006e-01
2.16109063e-02 -7.33318210e-01 -1.13046110e+00 3.05932581e-01
9.26376507e-02 -3.87129784e-02 -1.01173725e-02 6.49591446e-01
2.68446743e-01 2.38773555e-01 6.91730917e-01 -1.10786855e+00
-5.61600566e-01 9.36585724e-01 -2.45312735e-01 1.02020614e-02
2.60741949e-01 -2.44208753e-01 7.27721691e-01 -1.16327119e+00
5.64839721e-01 1.17105138e+00 6.33354485e-01 1.14394379e+00
-1.46842706e+00 -5.15144765e-01 3.64451349e-01 -5.64337134e-01
-9.33746397e-01 -4.82110292e-01 9.22227740e-01 -4.36786503e-01
7.89531529e-01 9.92132202e-02 2.83419371e-01 1.15829325e+00
-2.38126516e-01 8.11388493e-01 1.07666671e+00 -1.25181460e+00
3.44940960e-01 4.52616006e-01 3.98089260e-01 5.98297179e-01
1.20239429e-01 -1.64213672e-01 -3.76462013e-01 -1.96224153e-01
4.44745004e-01 -2.13447586e-01 -6.92044059e-03 -1.44683346e-01
-1.19948649e+00 1.10073781e+00 1.05400600e-01 5.86172819e-01
-3.48202646e-01 -1.52908951e-01 5.29682100e-01 3.59641522e-01
1.03512025e+00 8.52077603e-01 -7.46037304e-01 6.17106222e-02
-8.68664920e-01 2.93639064e-01 9.70969856e-01 8.00456762e-01
7.99097896e-01 1.03739195e-01 -4.01819974e-01 9.45373237e-01
1.44247249e-01 2.57155001e-01 8.84720147e-01 -2.32082039e-01
7.66480386e-01 9.30932522e-01 7.29725650e-03 -3.06748420e-01
-3.96810949e-01 -3.00107986e-01 -9.80327308e-01 1.32589653e-01
3.38643700e-01 -3.58678997e-01 -1.28379357e+00 1.84863675e+00
1.78647831e-01 -2.36116707e-01 4.75908667e-01 1.25631556e-01
9.09820557e-01 8.20741415e-01 3.33687633e-01 -3.55990231e-01
1.12735474e+00 -1.13602543e+00 -7.18027413e-01 -5.38478017e-01
1.14528680e+00 -4.84657556e-01 1.51942945e+00 2.30534866e-01
-8.12989712e-01 -7.55812287e-01 -9.17634726e-01 5.64176477e-02
-7.07517624e-01 2.98939407e-01 5.60273230e-01 8.23311448e-01
-9.83518362e-01 1.68027192e-01 -5.63503504e-01 -4.31934386e-01
4.27689135e-01 3.39172661e-01 -2.98890740e-01 4.24005650e-02
-1.25891149e+00 8.07026923e-01 5.78160107e-01 -4.10903305e-01
-7.68985391e-01 -5.76167822e-01 -1.06875110e+00 -6.12840801e-02
4.26284343e-01 -6.99511945e-01 1.62823975e+00 -1.02645528e+00
-1.60672712e+00 1.02855337e+00 -3.22497725e-01 -6.00381613e-01
3.75254035e-01 -1.43190563e-01 -3.77334297e-01 -2.08500430e-01
1.45211905e-01 8.89280975e-01 8.69067371e-01 -1.57654977e+00
-4.63801891e-01 -8.13579485e-02 -8.58288035e-02 2.01393869e-02
-7.62228191e-01 6.51681647e-02 6.33003116e-02 -1.08058631e+00
-2.66037285e-01 -7.13216066e-01 -5.87483764e-01 -3.28523993e-01
-5.37821889e-01 -4.37363356e-01 9.34471309e-01 -3.59435260e-01
1.30252039e+00 -1.79307497e+00 -1.55736402e-01 1.30969048e-01
1.71149760e-01 5.98080873e-01 -4.53021705e-01 5.29207408e-01
-3.62639993e-01 4.61852342e-01 -4.23687965e-01 -8.22710991e-01
-1.93550095e-01 2.42647156e-01 -5.79182208e-01 -3.31812143e-01
4.76326108e-01 9.14328694e-01 -9.38130140e-01 -2.48853564e-01
3.27433161e-02 1.16076894e-01 -5.25822222e-01 4.82883126e-01
-7.26397038e-01 5.80021918e-01 -4.58508968e-01 8.71181116e-02
2.97924608e-01 -2.54514426e-01 -1.82080403e-01 2.96150833e-01
3.68809402e-02 5.55653632e-01 -8.65094423e-01 1.72111905e+00
-7.48949766e-01 5.80245435e-01 -3.09320301e-01 -1.15353942e+00
1.21070158e+00 3.89166862e-01 1.31470919e-01 -1.18554756e-01
2.93773055e-01 6.45145550e-02 -9.13686901e-02 -5.07521152e-01
6.02884471e-01 -2.75586188e-01 -1.65917635e-01 9.44253981e-01
3.01970273e-01 -2.29396805e-01 6.37387931e-01 3.29779357e-01
9.97096479e-01 -2.43658237e-02 4.48666841e-01 -1.12674750e-01
7.82188535e-01 1.69766322e-01 -3.08599952e-03 9.95528936e-01
2.29632244e-01 4.01977450e-01 2.40236998e-01 -4.69696343e-01
-1.30420315e+00 -3.64391744e-01 7.67279565e-02 1.48944187e+00
-6.11944854e-01 -5.13367474e-01 -9.56232727e-01 -1.12809896e+00
-9.70445350e-02 1.06984890e+00 -9.38097298e-01 -2.34717667e-01
-4.85737830e-01 -8.78003478e-01 5.54005861e-01 5.80753505e-01
4.00279641e-01 -1.47481859e+00 -1.22460909e-01 2.46201381e-01
-7.37459809e-02 -1.11433423e+00 -2.21438766e-01 4.35269177e-01
-8.43374610e-01 -6.74578130e-01 -5.85338056e-01 -1.16866350e+00
1.17680717e+00 -3.22221667e-02 1.18314123e+00 1.44419745e-01
1.16403073e-01 2.17898592e-01 -7.24700928e-01 -6.91714048e-01
-1.42047942e+00 5.71144819e-01 -1.81888938e-02 1.09270737e-01
3.63830954e-01 -2.04144880e-01 1.55987412e-01 -4.91533913e-02
-1.12464523e+00 4.82670933e-01 5.20338058e-01 1.11565948e+00
2.87928969e-01 8.83998945e-02 1.07275295e+00 -1.40465713e+00
1.14903462e+00 -3.05869013e-01 -3.71996045e-01 2.19313979e-01
-6.84385955e-01 5.15967429e-01 9.63851333e-01 -7.71811068e-01
-1.10655916e+00 1.56385615e-01 -2.51289397e-01 3.50282714e-02
-4.80285794e-01 8.08375239e-01 -1.66212991e-01 1.92018211e-01
1.10478389e+00 2.35568553e-01 1.23211101e-01 -5.26514411e-01
5.84879398e-01 9.37951028e-01 4.99926418e-01 -4.96606439e-01
8.76757205e-01 2.32031032e-01 -1.78031117e-01 -5.79259872e-01
-1.19323862e+00 -2.18666807e-01 -9.82803643e-01 8.69612172e-02
6.63157701e-01 -6.74401760e-01 -1.56463161e-01 5.01689017e-01
-1.19847226e+00 -6.89677179e-01 -6.95687175e-01 2.14010000e-01
-2.71058649e-01 -1.36171198e-02 -3.83047581e-01 -9.12517846e-01
-6.35100126e-01 -7.14926302e-01 9.63554084e-01 6.04321696e-02
-3.61314565e-01 -1.20053506e+00 2.30502278e-01 2.88417727e-01
2.62720704e-01 -6.26447350e-02 1.16491008e+00 -1.39485395e+00
1.00150496e-01 -6.30155146e-01 1.65872321e-01 6.17242038e-01
3.64140123e-01 -2.02819973e-01 -1.14465868e+00 -1.85417652e-01
-2.52625309e-02 -6.86816096e-01 8.00993800e-01 -3.80698442e-02
1.41426253e+00 -4.27195996e-01 -3.53506118e-01 1.02385757e-02
7.93592393e-01 1.94109738e-01 2.52960086e-01 3.39244783e-01
6.51707351e-01 5.74064612e-01 4.69518006e-01 2.93579519e-01
2.33179137e-01 3.78155440e-01 1.62214637e-02 -4.97668087e-01
-9.95693505e-02 -5.22572517e-01 6.24832660e-02 7.88517296e-01
3.54779154e-01 -5.74725211e-01 -1.02018356e+00 4.46404606e-01
-1.65601325e+00 -6.54913187e-01 1.42374970e-02 2.07105422e+00
1.34294856e+00 4.40903902e-01 1.49724707e-01 6.06305063e-01
7.73180306e-01 -1.10677026e-01 -3.38001877e-01 -4.31001782e-01
-5.50457463e-02 5.03085554e-01 1.82683572e-01 5.06874740e-01
-1.26752675e+00 1.10882092e+00 6.65915537e+00 6.32690609e-01
-1.01888621e+00 2.51901709e-02 8.69875491e-01 3.21429193e-01
-3.60468417e-01 -1.41264766e-01 -9.57798302e-01 1.58363208e-01
1.03976476e+00 -3.23811352e-01 3.02127242e-01 8.10674608e-01
3.23311418e-01 9.30015743e-02 -1.31337535e+00 6.54459298e-01
4.63799059e-01 -1.46256256e+00 5.45075476e-01 -1.57661483e-01
8.96912754e-01 -3.12784582e-01 -2.50427932e-01 6.34266734e-01
5.39818823e-01 -9.97273803e-01 3.07663918e-01 1.88056499e-01
8.67722690e-01 -7.19037533e-01 8.18522036e-01 6.82169080e-01
-6.98830307e-01 -9.12440345e-02 3.15974876e-02 -2.07233682e-01
-2.45993156e-02 6.05358601e-01 -1.30352688e+00 2.21523479e-01
8.51816311e-02 4.99031872e-01 -8.36195469e-01 4.88005638e-01
-4.40690994e-01 8.81759286e-01 -8.94415230e-02 -1.95581526e-01
1.36319518e-01 2.27074206e-01 2.43542179e-01 1.24641728e+00
-8.74848291e-02 7.32703209e-02 4.73070115e-01 7.98790991e-01
-5.28594494e-01 3.77221763e-01 -6.33778393e-01 -3.25248688e-01
3.98993403e-01 1.27919233e+00 -6.01192892e-01 -7.94009089e-01
-4.31174964e-01 7.80347764e-01 2.95925707e-01 2.36597270e-01
-1.39741078e-01 -4.32228982e-01 -5.49595617e-02 4.27173451e-02
-6.04074821e-02 3.54555398e-02 -6.19478643e-01 -1.24514246e+00
-5.81269823e-02 -1.07120550e+00 3.09908301e-01 -7.49860346e-01
-1.27554786e+00 9.38067734e-01 1.52278244e-02 -1.12701118e+00
-7.73442864e-01 -4.98113573e-01 -6.28117979e-01 9.87631619e-01
-1.16707480e+00 -1.30143499e+00 -2.70867616e-01 3.94538462e-01
8.69568169e-01 -6.44150853e-01 1.32895708e+00 -9.67489090e-03
-4.44523484e-01 7.07756519e-01 -1.06300689e-01 4.79180485e-01
6.32824361e-01 -1.26255357e+00 9.77654517e-01 8.22645485e-01
1.78563595e-01 3.66665900e-01 5.91687202e-01 -7.89780021e-01
-7.59953558e-01 -1.29478049e+00 1.30276811e+00 -4.37445700e-01
6.96626604e-01 -9.49484944e-01 -1.11129463e+00 6.72096193e-01
1.83969423e-01 -2.18062680e-02 8.67233038e-01 1.05442889e-01
-2.38420874e-01 2.30426908e-01 -1.15762973e+00 6.26774609e-01
7.42301047e-01 -3.99022639e-01 -7.95498550e-01 5.10936499e-01
8.40960681e-01 -2.65745759e-01 -4.24552947e-01 4.07095820e-01
1.07409090e-01 9.54719074e-03 4.75789368e-01 -1.26189613e+00
5.83591163e-01 -4.00281027e-02 1.99454919e-01 -1.59375370e+00
-1.25744399e-02 -7.53743947e-01 2.26598270e-02 1.52254808e+00
7.91128814e-01 -6.12327874e-01 8.81854355e-01 5.55562973e-01
-1.37839854e-01 -4.99068677e-01 -4.83967543e-01 -6.37472928e-01
3.39227885e-01 -3.54242831e-01 5.41109264e-01 1.16786933e+00
2.36782029e-01 8.89352798e-01 -2.08068430e-01 -3.65916163e-01
2.34897628e-01 -1.43215820e-01 1.13614643e+00 -1.47390544e+00
-1.11578675e-02 -2.24930093e-01 -6.74675852e-02 -8.88879538e-01
5.16020954e-01 -1.06930530e+00 2.28168964e-01 -1.43611979e+00
2.43212983e-01 -5.21047711e-01 2.80517600e-02 9.79445100e-01
-5.21121919e-01 1.64899439e-01 1.12052478e-01 1.40656576e-01
-2.05834016e-01 6.05137050e-01 1.20055664e+00 -1.94078550e-01
-6.39491558e-01 3.43369633e-01 -9.22516048e-01 6.42093539e-01
9.86287415e-01 -6.15969539e-01 -5.03382325e-01 -1.76294953e-01
5.72890900e-02 -2.22509220e-01 -2.88079903e-02 -8.84530365e-01
-6.22097664e-02 1.39684202e-02 2.67562062e-01 -4.82624680e-01
7.95236155e-02 -4.61221099e-01 -4.99287963e-01 5.04559577e-01
-1.16005945e+00 5.09865582e-03 3.81872505e-01 2.69338131e-01
-1.77924886e-01 -5.27889371e-01 8.26579034e-01 -6.96839988e-02
-3.03687721e-01 -1.99359190e-02 -8.28916848e-01 1.75308913e-01
7.53420472e-01 1.07413687e-01 -3.81739825e-01 -4.98461306e-01
-9.66856897e-01 1.05350725e-01 1.79916069e-01 5.77244282e-01
3.04894030e-01 -1.35051370e+00 -9.96937633e-01 6.08575463e-01
3.21026057e-01 1.49254307e-01 -4.73403841e-01 3.01936865e-02
-2.55810898e-02 5.37289977e-01 1.60862863e-01 -3.20617527e-01
-1.15017736e+00 7.33398259e-01 1.41731203e-01 -7.62939453e-01
-3.02044123e-01 6.77935600e-01 1.36308089e-01 -8.07835877e-01
1.48507565e-01 -5.14267325e-01 -5.00096679e-01 -4.26946767e-02
5.62972605e-01 -3.48224826e-02 3.20134848e-01 -4.81966823e-01
3.80028561e-02 -6.64707869e-02 -4.09759939e-01 -4.30374950e-01
1.14783096e+00 1.48265645e-01 3.73864248e-02 6.75470233e-01
8.76214564e-01 -2.97194626e-03 -8.00256193e-01 -6.08286500e-01
-1.43350549e-02 -2.69230306e-02 -3.38871889e-02 -1.11267340e+00
-6.68075800e-01 7.07924902e-01 2.63709337e-01 5.05537510e-01
1.09564602e+00 -1.04754921e-02 4.25261289e-01 6.93328142e-01
1.23364199e-02 -1.01276207e+00 3.42222482e-01 8.13674688e-01
9.04641032e-01 -1.27253497e+00 -1.49874076e-01 -5.98963261e-01
-6.14632308e-01 1.08031642e+00 7.14018703e-01 1.66434020e-01
6.61372662e-01 3.23389590e-01 3.03289086e-01 1.57036588e-01
-8.76513004e-01 -1.83727443e-01 2.45588377e-01 5.89564621e-01
6.69331431e-01 -2.21488371e-01 -2.96280682e-01 5.06941497e-01
-5.00056684e-01 2.55637854e-01 7.00357258e-01 1.11063313e+00
-4.45932895e-01 -1.59533370e+00 -3.96631390e-01 7.80755877e-01
-3.59266758e-01 -3.88237804e-01 -6.48049057e-01 8.14208508e-01
2.89978068e-02 1.15320468e+00 2.10070580e-01 -2.75101125e-01
2.50541180e-01 5.69656432e-01 1.73374802e-01 -1.30361724e+00
-6.74743652e-01 -1.46419838e-01 3.08331996e-01 3.36611480e-01
-5.07241845e-01 -5.98464608e-01 -1.32289958e+00 1.89471737e-01
-4.93849874e-01 4.07642066e-01 7.51596570e-01 1.24931085e+00
1.75344080e-01 5.05536556e-01 8.25299084e-01 -8.18268597e-01
-5.53656578e-01 -1.44527721e+00 -1.70961276e-01 5.94322860e-01
3.61635447e-01 -3.94908041e-01 -2.54328281e-01 5.26151419e-01]
|
[10.872079849243164, 8.32581615447998]
|
db906030-4956-4d21-bbd1-57a04204d241
|
nilc-at-cwi-2018-exploring-feature
| null | null |
https://aclanthology.org/W18-0540
|
https://aclanthology.org/W18-0540.pdf
|
NILC at CWI 2018: Exploring Feature Engineering and Feature Learning
|
This paper describes the results of NILC team at CWI 2018. We developed solutions following three approaches: (i) a feature engineering method using lexical, n-gram and psycholinguistic features, (ii) a shallow neural network method using only word embeddings, and (iii) a Long Short-Term Memory (LSTM) language model, which is pre-trained on a large text corpus to produce a contextualized word vector. The feature engineering method obtained our best results for the classification task and the LSTM model achieved the best results for the probabilistic classification task. Our results show that deep neural networks are able to perform as well as traditional machine learning methods using manually engineered features for the task of complex word identification in English.
|
['ro Borges', 'Nathan Hartmann', 'Le dos Santos']
|
2018-06-01
| null | null | null |
ws-2018-6
|
['complex-word-identification']
|
['natural-language-processing']
|
[-3.16267721e-02 1.56421121e-02 1.29367903e-01 -3.88733625e-01
-6.94338799e-01 -2.64922321e-01 6.86306059e-01 3.86982530e-01
-1.17142010e+00 5.28015733e-01 5.00542521e-01 -7.35271692e-01
-4.61052656e-02 -8.33932638e-01 -3.31873626e-01 -1.52229533e-01
8.97388626e-03 5.76801479e-01 -1.16254009e-01 -3.61374170e-01
5.59977293e-01 4.82898176e-01 -1.44462466e+00 2.74586916e-01
7.29575634e-01 1.03341782e+00 4.13059920e-01 7.61973262e-01
-6.43848896e-01 7.39542544e-01 -5.15749753e-01 -2.47699380e-01
-1.54340968e-01 9.25391540e-02 -1.19336283e+00 -5.76452672e-01
1.70985952e-01 -9.66846123e-02 -5.05275838e-02 8.71919155e-01
4.99552876e-01 3.94713998e-01 5.88812590e-01 -4.68816817e-01
-9.90350306e-01 1.04746020e+00 1.38352796e-01 1.90006897e-01
5.20088136e-01 3.90352942e-02 1.27164435e+00 -1.37463379e+00
5.77611268e-01 1.48884737e+00 7.65674055e-01 4.46111888e-01
-9.60289538e-01 -4.48744297e-01 -8.62479657e-02 4.82743859e-01
-1.23587060e+00 -1.76601887e-01 2.80669451e-01 -4.56363559e-01
2.01211929e+00 -2.49384344e-01 6.32443190e-01 1.36020780e+00
2.67274320e-01 6.52959645e-01 1.18484342e+00 -1.13792586e+00
1.72462642e-01 3.43692571e-01 9.22326922e-01 6.13581300e-01
5.86941279e-02 2.42908448e-01 -6.09975398e-01 -2.69410193e-01
1.07686251e-01 -3.70059848e-01 1.03809498e-01 4.40816522e-01
-1.09968603e+00 1.22627294e+00 -9.31882933e-02 9.01363492e-01
-4.65033293e-01 1.12040758e-01 7.05777347e-01 3.42211485e-01
6.92980766e-01 9.68168914e-01 -1.05244553e+00 -4.42840666e-01
-7.78612196e-01 1.04252629e-01 1.04863989e+00 5.83166659e-01
7.52398372e-01 3.57780665e-01 -2.78939635e-01 1.16226733e+00
3.97082239e-01 4.51165318e-01 1.23226297e+00 -3.85968059e-01
2.61260539e-01 4.00830984e-01 -1.33805275e-01 -9.29509282e-01
-6.56444013e-01 -8.67588371e-02 -1.29943311e-01 1.98012702e-02
3.02017868e-01 -4.45432365e-01 -1.06866586e+00 1.66641569e+00
-2.69375861e-01 -4.39761996e-01 1.87665775e-01 3.21020842e-01
1.05278337e+00 8.73103499e-01 4.26012933e-01 2.24696979e-01
1.55488288e+00 -7.24815309e-01 -9.81030166e-01 -3.21214378e-01
1.02396464e+00 -7.74559081e-01 1.31685853e+00 3.94724846e-01
-9.24816251e-01 -6.33111715e-01 -1.14484060e+00 -9.30578187e-02
-1.37971151e+00 1.51744515e-01 4.73234504e-01 1.03648269e+00
-1.27190328e+00 6.10312998e-01 -4.19713318e-01 -6.13307595e-01
3.04252356e-02 3.54795426e-01 -5.15355349e-01 1.44272700e-01
-1.62365878e+00 1.54825568e+00 8.81352544e-01 -4.42411229e-02
-5.48446536e-01 -5.03887475e-01 -1.34171700e+00 1.50074333e-01
1.15954049e-01 -1.09653629e-01 9.64837730e-01 -8.07010531e-01
-1.77198040e+00 9.46513772e-01 -2.60516524e-01 -5.06239116e-01
-2.36893371e-01 -4.70275074e-01 -4.24709350e-01 -2.75637507e-01
-5.89938834e-02 5.26808083e-01 4.86886084e-01 -7.61930406e-01
-6.76874340e-01 -3.38499069e-01 -2.63758868e-01 -1.16471812e-01
-9.43101108e-01 5.17291784e-01 1.87342554e-01 -7.29292095e-01
-3.56901854e-01 -7.65514374e-01 -3.23329866e-02 -6.78312480e-01
-5.16225338e-01 -9.34743285e-01 6.05315804e-01 -1.13235176e+00
1.25591159e+00 -1.87952888e+00 -1.85831450e-02 2.50133067e-01
7.73374038e-03 6.97491944e-01 -4.62156564e-01 5.97123444e-01
-2.23188877e-01 5.44650376e-01 1.31980047e-01 -4.04735178e-01
2.87366331e-01 2.04092190e-01 -9.96750444e-02 1.15332432e-01
3.03251266e-01 9.95731652e-01 -8.58606637e-01 -4.17360723e-01
2.57245213e-01 3.89520109e-01 -1.69649631e-01 2.64120221e-01
-9.85512435e-02 -3.73056799e-01 1.31132141e-01 2.45824829e-01
3.35711986e-01 2.87433743e-01 4.63600345e-02 -9.83130857e-02
-2.59755999e-01 5.63373327e-01 -9.38857436e-01 1.60277545e+00
-8.38885307e-01 8.90025198e-01 -4.07075882e-01 -8.86732161e-01
1.05639660e+00 4.91850942e-01 -8.56965855e-02 -5.54958582e-01
3.46130341e-01 3.82341117e-01 1.31576106e-01 -7.03997493e-01
7.32095182e-01 -1.87367976e-01 -2.23379150e-01 5.18794298e-01
7.04149604e-01 -7.94234313e-03 1.84941381e-01 2.78529618e-03
1.05636919e+00 1.20686144e-01 3.44900608e-01 -6.94817364e-01
4.60536271e-01 -2.84125283e-02 2.82522082e-01 7.52545238e-01
-1.97827861e-01 4.02063549e-01 2.90267318e-01 -7.53907502e-01
-8.99635494e-01 -6.57348454e-01 -1.10469192e-01 1.61298585e+00
-6.69253945e-01 -6.14548147e-01 -8.70231867e-01 -5.17496109e-01
-1.86334610e-01 1.28297973e+00 -8.50895166e-01 -4.13468570e-01
-4.66712296e-01 -5.09552717e-01 7.21218228e-01 5.68028808e-01
6.54145842e-03 -1.73154449e+00 -4.72483993e-01 3.52919191e-01
-1.04059637e-01 -1.20097244e+00 -2.55801558e-01 7.20363915e-01
-3.18001390e-01 -6.44283652e-01 -3.77657175e-01 -1.05778360e+00
6.09070435e-02 -4.89304513e-01 1.03641069e+00 -7.44538680e-02
-4.44081873e-01 1.53555959e-01 -7.74522126e-01 -7.60900080e-01
-5.94702542e-01 3.24120998e-01 1.73921153e-01 -3.01323384e-01
1.05568063e+00 -1.95711359e-01 8.29203147e-03 -3.37266296e-01
-7.77417600e-01 -3.11434448e-01 4.23132479e-01 1.04595196e+00
-1.03605732e-01 -3.26303363e-01 5.91442466e-01 -7.94473529e-01
1.04911256e+00 -3.40206236e-01 -1.18293762e-01 3.36634040e-01
-7.90538430e-01 2.95649081e-01 5.51717758e-01 -5.35730958e-01
-8.64937484e-01 -1.34271413e-01 -7.74753451e-01 2.02610940e-01
-2.76620686e-01 8.50454271e-01 1.65054277e-02 1.17208160e-01
7.53249586e-01 2.33248204e-01 -2.68617067e-02 -6.36525452e-01
4.53340352e-01 9.54930246e-01 -3.38255963e-03 -6.25422060e-01
3.56219292e-01 -2.24022418e-01 -5.50406277e-01 -1.18088055e+00
-9.50814903e-01 -1.89045757e-01 -1.04922545e+00 -7.08859190e-02
1.27846313e+00 -5.27460396e-01 -3.07738066e-01 6.29048824e-01
-1.51660597e+00 -3.85104150e-01 -3.33584845e-01 6.42513454e-01
-1.31919652e-01 2.05276892e-01 -7.73106694e-01 -9.02063489e-01
-6.67208314e-01 -9.42072928e-01 8.01518321e-01 1.37378901e-01
-7.79670954e-01 -1.30195105e+00 2.67856628e-01 1.70887820e-02
8.25463235e-01 7.64808133e-02 1.36043799e+00 -1.33366942e+00
5.37761331e-01 -3.95748913e-01 -2.32597351e-01 8.05491090e-01
-9.08502191e-02 2.35379592e-01 -1.15998638e+00 -3.78339738e-02
-2.64522970e-01 -6.95918500e-01 8.41030657e-01 3.28788280e-01
9.68467534e-01 -4.87674922e-02 -6.15652390e-02 2.81796694e-01
1.34429550e+00 1.08380251e-01 5.49071670e-01 4.44572598e-01
8.15581024e-01 7.12794483e-01 2.78316587e-01 3.24220568e-01
4.05561060e-01 3.74717295e-01 5.68479896e-02 1.60408020e-01
-6.79856613e-02 -1.92153677e-01 3.79416525e-01 1.09729826e+00
1.70256451e-01 -1.59155980e-01 -1.38559580e+00 7.74430215e-01
-1.64451349e+00 -5.07849634e-01 -1.38468087e-01 1.89882457e+00
8.58143508e-01 2.41440460e-01 -1.51315838e-01 1.87381059e-01
4.91294235e-01 6.57668635e-02 2.08909176e-02 -1.32580042e+00
-1.55744895e-01 9.22147214e-01 2.78189033e-01 5.30704677e-01
-9.92573380e-01 1.52891064e+00 7.11179829e+00 1.12863076e+00
-9.18774307e-01 5.16558230e-01 3.57071608e-01 3.12050819e-01
-3.41986418e-01 -1.61636844e-01 -9.27341342e-01 3.59659731e-01
1.61845088e+00 -9.28670075e-03 2.66838521e-01 7.77894139e-01
-9.23838466e-02 -1.69562042e-01 -7.83415198e-01 6.76822007e-01
3.21715832e-01 -1.15753257e+00 2.53839076e-01 -5.29203676e-02
1.30593807e-01 4.07682687e-01 -2.39379272e-01 8.07054043e-01
4.75774050e-01 -1.38713145e+00 7.18619347e-01 4.99763638e-01
6.66047096e-01 -8.01319242e-01 9.47016478e-01 3.86613011e-01
-9.35706437e-01 2.68650223e-02 -5.84263146e-01 -2.54646927e-01
-7.51560405e-02 5.82695842e-01 -6.19323671e-01 1.40299171e-01
7.79380620e-01 5.13780713e-01 -7.18811810e-01 4.80929315e-01
-1.96158051e-01 8.07377160e-01 -3.05216700e-01 -7.64808834e-01
5.17186105e-01 4.17117449e-03 4.68778253e-01 1.79341483e+00
1.09556220e-01 -4.17650521e-01 -5.28876558e-02 7.47935772e-01
1.67048737e-01 5.73741555e-01 -7.68124759e-01 -2.70831943e-01
3.38041872e-01 1.41329265e+00 -5.08934915e-01 -3.46094728e-01
-2.54124522e-01 8.74004185e-01 7.21291125e-01 2.56906658e-01
-2.99426317e-01 -8.76438200e-01 4.88985240e-01 -3.40553015e-01
2.74548382e-01 -4.17161137e-01 -4.04758006e-01 -8.23418558e-01
-2.46889561e-01 -6.13937199e-01 2.64766365e-01 -6.08583510e-01
-1.37653935e+00 1.11540508e+00 -5.09726480e-02 -3.50522399e-01
-5.58022857e-01 -1.37433946e+00 -6.38819456e-01 1.25965583e+00
-1.25502253e+00 -1.13698137e+00 1.21210113e-01 2.75338411e-01
4.87360448e-01 -6.57449245e-01 1.52401555e+00 9.05818343e-02
-4.70614016e-01 5.36681712e-01 1.58542991e-01 3.49810719e-01
5.36953509e-01 -1.64091873e+00 4.88976240e-01 5.12516797e-01
2.34381661e-01 7.54126012e-01 4.89464670e-01 -5.24883270e-01
-1.30141437e+00 -7.24874675e-01 1.59147847e+00 -4.97992545e-01
9.17061031e-01 -5.48961163e-01 -7.88714409e-01 5.73345363e-01
6.16007686e-01 -1.56003088e-01 1.08814359e+00 4.72460806e-01
-3.62860143e-01 2.51013368e-01 -1.07658851e+00 3.26554358e-01
5.17836630e-01 -8.62158954e-01 -1.09298933e+00 3.73071223e-01
9.67510760e-01 4.37993445e-02 -9.39580619e-01 1.73050418e-01
7.87716150e-01 -4.18816686e-01 7.30511308e-01 -8.90711963e-01
5.06125391e-01 4.23359245e-01 -2.72837132e-01 -1.65555251e+00
-2.61899054e-01 -2.69800842e-01 1.06122352e-01 1.29894829e+00
9.09194827e-01 -8.27766061e-01 1.83231890e-01 6.62691414e-01
5.22572473e-02 -8.05998087e-01 -8.91764581e-01 -6.90744460e-01
5.45230806e-01 -8.30704629e-01 2.38598019e-01 9.13297594e-01
1.14252977e-01 5.64516544e-01 -1.76221713e-01 -5.00219941e-01
1.26320422e-01 -4.95526820e-01 1.44312128e-01 -1.19608092e+00
5.25888465e-02 -5.93029380e-01 -4.47565794e-01 -3.60521942e-01
7.70051837e-01 -9.27302599e-01 2.26682574e-01 -1.55434930e+00
3.61789688e-02 -9.56848189e-02 -4.03680086e-01 7.00623870e-01
-2.35922411e-01 -7.56074581e-03 -7.35529000e-03 -4.73358452e-01
-1.66181147e-01 3.93950850e-01 5.00549495e-01 1.06387645e-01
3.52889253e-03 -3.90638351e-01 -6.75959110e-01 7.38235474e-01
1.00433612e+00 -5.28427482e-01 1.25541195e-01 -4.90367413e-01
3.71995121e-01 -6.01672411e-01 -7.71860108e-02 -8.31877768e-01
1.53160943e-02 1.37826890e-01 4.78507638e-01 -4.57590848e-01
3.11858565e-01 -4.60159093e-01 -7.37618625e-01 6.14809811e-01
-3.71955872e-01 2.67053783e-01 4.72670645e-01 7.13198707e-02
-2.03982562e-01 -8.06003034e-01 6.97696209e-01 -1.93536565e-01
-8.54610622e-01 -5.52613959e-02 -1.12912893e+00 4.58934084e-02
5.40700078e-01 -1.01341300e-01 -1.15704618e-01 -1.70218855e-01
-7.65026331e-01 -1.08335186e-02 -1.41030803e-01 7.56379783e-01
5.88629425e-01 -1.24394119e+00 -8.67804229e-01 2.99331248e-01
1.65661380e-01 -7.29758322e-01 -3.09738964e-01 3.16097856e-01
-5.96204638e-01 5.37986219e-01 -1.11871846e-01 -2.20345676e-01
-1.05366468e+00 3.62024009e-01 2.99052984e-01 -4.97350901e-01
-4.20728505e-01 9.85150039e-01 -5.41524768e-01 -9.07551169e-01
3.16883832e-01 -3.24880958e-01 -7.74059892e-01 3.91206890e-01
6.34702682e-01 2.14010745e-01 2.69416362e-01 -8.16531897e-01
-4.39576745e-01 4.92604494e-01 -1.93216741e-01 -5.03798306e-01
1.50199234e+00 8.44548121e-02 -3.67935240e-01 8.27346921e-01
1.57113254e+00 -3.25547904e-01 -8.71540159e-02 -4.54041243e-01
5.82881212e-01 5.64513505e-02 4.67797548e-01 -8.75252664e-01
-4.83649135e-01 1.02850509e+00 7.19065845e-01 5.10171473e-01
5.53056836e-01 -3.11060101e-01 1.01972437e+00 7.96864748e-01
1.58103436e-01 -1.70839798e+00 -1.35400683e-01 1.33263564e+00
7.68009067e-01 -1.19714808e+00 -3.65400791e-01 2.36169294e-01
-4.83343959e-01 1.64085364e+00 4.75988954e-01 -3.00971895e-01
1.25494719e+00 4.23515230e-01 3.13110091e-02 -3.12191367e-01
-6.82220101e-01 -4.60014701e-01 4.80614156e-01 5.10814905e-01
6.95898354e-01 1.33701280e-01 -5.44921935e-01 9.31818426e-01
-4.24955636e-01 -2.48141468e-01 2.37757951e-01 7.45256901e-01
-4.79676425e-01 -1.15292335e+00 -1.54601574e-01 7.63236940e-01
-5.35752952e-01 -5.70335865e-01 -5.06649017e-01 7.13622332e-01
4.90863547e-02 1.04853570e+00 -3.65656279e-02 -6.10927224e-01
2.82163382e-01 7.05806732e-01 1.61394626e-01 -1.08336043e+00
-9.84322965e-01 -4.17161494e-01 4.14743960e-01 -5.75525463e-01
-6.30385503e-02 -5.02010286e-01 -1.10728598e+00 -1.31162331e-01
-4.75145578e-01 2.10994333e-01 1.16970003e+00 1.48127639e+00
2.64987699e-03 4.28895086e-01 3.23967397e-01 -1.03722024e+00
-5.79034567e-01 -1.69535863e+00 -5.36158442e-01 2.50745445e-01
-3.59545089e-02 -3.47111911e-01 -4.29521322e-01 -8.57679769e-02]
|
[10.507673263549805, 10.323017120361328]
|
cce88547-9c3a-4707-b699-b97c9b7263cd
|
abstractive-timeline-summarization
| null | null |
https://aclanthology.org/D19-5403
|
https://aclanthology.org/D19-5403.pdf
|
Abstractive Timeline Summarization
|
Timeline summarization (TLS) automatically identifies key dates of major events and provides short descriptions of what happened on these dates. Previous approaches to TLS have focused on extractive methods. In contrast, we suggest an abstractive timeline summarization system. Our system is entirely unsupervised, which makes it especially suited to TLS where there are very few gold summaries available for training of supervised systems. In addition, we present the first abstractive oracle experiments for TLS. Our system outperforms extractive competitors in terms of ROUGE when the number of input documents is high and the output requires strong compression. In these cases, our oracle experiments confirm that our approach also has a higher upper bound for ROUGE scores than extractive methods. A study with human judges shows that our abstractive system also produces output that is easy to read and understand.
|
['Katja Markert', 'Julius Steen']
|
2019-11-01
| null | null | null |
ws-2019-11
|
['timeline-summarization']
|
['natural-language-processing']
|
[ 2.11458534e-01 2.85911947e-01 -4.49739426e-01 -1.07881568e-01
-1.37658095e+00 -9.36376810e-01 8.90235066e-01 8.59057307e-01
-4.81990933e-01 9.53410923e-01 8.23802948e-01 -2.73389757e-01
-3.35049070e-02 -6.11120462e-01 -5.15331864e-01 -2.02556625e-01
-3.86122940e-03 7.98820436e-01 3.74497086e-01 -1.70494005e-01
8.09445739e-01 3.29376757e-01 -1.16805482e+00 3.29324991e-01
9.99422848e-01 3.77019554e-01 9.77104232e-02 1.14868021e+00
-1.97868258e-01 7.93857515e-01 -1.28940928e+00 -3.74506146e-01
4.05212492e-02 -7.33853996e-01 -1.05620337e+00 -1.05972558e-01
4.64997262e-01 -4.70483303e-01 -4.72853452e-01 7.29614496e-01
4.35139894e-01 2.10400268e-01 8.31612289e-01 -1.06151330e+00
-2.72750378e-01 1.00441217e+00 -5.21632969e-01 6.17028296e-01
7.53924251e-01 -1.50903404e-01 1.33048356e+00 -6.74632430e-01
7.72646725e-01 8.93268406e-01 6.26229167e-01 8.55276063e-02
-1.06787181e+00 -1.06024109e-01 -1.78004783e-02 -8.53984058e-02
-1.02522862e+00 -5.69988966e-01 2.86408722e-01 -9.73924324e-02
1.37598050e+00 7.67353415e-01 6.26938283e-01 8.03202331e-01
2.14493513e-01 1.13380992e+00 5.05875945e-01 -6.50913239e-01
3.14403474e-01 -6.31074533e-02 4.60920006e-01 2.85965621e-01
5.57238460e-01 -5.73938906e-01 -7.04414845e-01 -4.99923170e-01
1.80472210e-01 -8.09371248e-02 -1.78767651e-01 3.25710416e-01
-1.30791867e+00 8.38353932e-01 -2.47798637e-01 3.40748310e-01
-5.52356362e-01 5.20164967e-02 8.10266316e-01 3.51983279e-01
7.08564222e-01 9.94543433e-01 -3.69108498e-01 -4.89604026e-01
-1.60138416e+00 4.86054480e-01 1.33574510e+00 1.13272858e+00
2.30853081e-01 -7.96524957e-02 -3.87309611e-01 6.38656497e-01
-3.22300643e-01 5.75821638e-01 6.09462023e-01 -1.11804533e+00
7.89465666e-01 5.10570884e-01 3.73032570e-01 -8.14650357e-01
-2.77623981e-01 -3.17032516e-01 -4.27023828e-01 -3.22265804e-01
2.16519803e-01 -3.58848795e-02 -7.38503098e-01 1.20288467e+00
-2.02636003e-01 -2.81512707e-01 3.79470348e-01 2.36781031e-01
8.92406106e-01 1.08833706e+00 -2.63444632e-01 -8.86689126e-01
1.26891339e+00 -9.38476503e-01 -1.00630462e+00 -3.06517720e-01
7.83556223e-01 -1.00927329e+00 1.09450352e+00 4.09972429e-01
-1.31687582e+00 -6.87353918e-03 -1.25956309e+00 -2.75910143e-02
-2.67723382e-01 2.65069693e-01 6.04589343e-01 4.29628283e-01
-9.32852030e-01 8.16807270e-01 -9.69078660e-01 -7.81901419e-01
9.93417874e-02 9.14114341e-02 -1.25406593e-01 1.36457607e-01
-1.00348234e+00 8.94014239e-01 7.14291215e-01 -4.30699795e-01
-4.20931011e-01 -4.15749967e-01 -6.75160468e-01 1.69044420e-01
6.00150466e-01 -5.58567524e-01 1.94149792e+00 -2.91287720e-01
-1.05815291e+00 5.36857009e-01 -3.84513557e-01 -8.21062207e-01
3.90943557e-01 -4.26933497e-01 -3.04948002e-01 6.21024072e-01
4.81931895e-01 2.58546352e-01 4.08438802e-01 -6.66840315e-01
-9.03674364e-01 2.46591121e-02 1.15701146e-02 4.56131279e-01
-5.27758837e-01 2.62388080e-01 -4.80491340e-01 -6.71353996e-01
-1.64760742e-02 -5.60495377e-01 -1.69084936e-01 -8.28796744e-01
-7.19933987e-01 -4.86359537e-01 5.91899157e-01 -6.28518045e-01
2.05380702e+00 -1.72236979e+00 -1.99932292e-01 -2.95401528e-03
2.04863027e-01 1.09857440e-01 1.59063533e-01 1.26644194e+00
2.92812228e-01 4.59125727e-01 -1.47576019e-01 -1.67693764e-01
5.70079610e-02 7.23839998e-02 -8.51089299e-01 8.02714005e-02
-1.85398087e-01 6.80711687e-01 -1.10883796e+00 -7.53298223e-01
-2.48801365e-01 -3.91641051e-01 -3.54403764e-01 2.01458514e-01
-2.61810571e-01 -3.09571207e-01 -4.92865860e-01 4.82644737e-01
4.07924540e-02 -2.99984574e-01 -1.08737983e-02 1.07103184e-01
-3.26829523e-01 7.07175374e-01 -8.47741187e-01 1.41556656e+00
-1.54935583e-01 9.21644270e-01 -6.98352993e-01 -5.67180634e-01
5.66167951e-01 4.25437272e-01 4.70804930e-01 -2.69588083e-01
-2.46156320e-01 4.67987090e-01 -5.48130333e-01 -4.39621508e-01
1.31513095e+00 2.52747416e-01 -5.66864848e-01 9.48467672e-01
2.92290058e-02 -4.64462489e-01 1.00754619e+00 1.13877821e+00
1.35246110e+00 -2.60094047e-01 7.89891481e-01 -2.47827712e-02
3.87936682e-02 5.54551423e-01 4.09106255e-01 1.26231122e+00
2.68796682e-01 8.10748875e-01 8.64804447e-01 -2.68452317e-01
-1.31906652e+00 -8.41344535e-01 1.76384330e-01 9.36361730e-01
2.42952220e-02 -1.33427155e+00 -8.99570227e-01 -7.10117519e-01
-2.73984492e-01 1.26531005e+00 -3.05367231e-01 5.41209355e-02
-6.39437377e-01 -6.66905403e-01 7.65004456e-01 5.81612825e-01
9.40977484e-02 -1.09548521e+00 -7.90476859e-01 4.13079023e-01
-6.49350464e-01 -1.02609181e+00 -8.17274690e-01 1.24821268e-01
-1.17146897e+00 -1.02142298e+00 -6.07035398e-01 -5.04344285e-01
5.43428183e-01 3.54498804e-01 1.08886707e+00 -2.35193953e-01
1.06150441e-01 3.91705155e-01 -5.47013283e-01 -8.40904355e-01
-6.31322443e-01 5.87759972e-01 2.21835580e-02 -7.58322001e-01
4.81122911e-01 -4.28190410e-01 -4.45524067e-01 -1.17737897e-01
-1.10234272e+00 -5.39352559e-02 5.56132376e-01 6.35890305e-01
4.01357204e-01 1.49824440e-01 7.85499632e-01 -1.23900533e+00
1.30341017e+00 -3.59897405e-01 -2.31207773e-01 4.27386552e-01
-9.47549403e-01 9.81924683e-02 6.37145221e-01 -1.98115930e-01
-8.90256643e-01 -2.44963259e-01 5.30718081e-02 2.11859122e-01
2.47178916e-02 8.26913238e-01 4.76374120e-01 8.81953895e-01
9.95553017e-01 2.99650550e-01 -3.23055446e-01 -5.41116059e-01
2.18893304e-01 9.36576366e-01 7.91319191e-01 -2.31323600e-01
4.95255172e-01 2.41404310e-01 -6.17463946e-01 -9.97568607e-01
-9.81470108e-01 -8.39034498e-01 -4.38708603e-01 -2.52834558e-01
3.28277558e-01 -7.15916216e-01 -2.99304008e-01 1.25900328e-01
-1.16497397e+00 -5.32527044e-02 -6.64566875e-01 5.43375134e-01
-7.67894387e-01 5.24431348e-01 -7.24416137e-01 -7.49600589e-01
-7.73440897e-01 -5.69621027e-01 1.01477098e+00 3.91106635e-01
-9.33744371e-01 -8.22597742e-01 1.21290684e-01 9.71903652e-02
1.15408711e-01 3.78355682e-01 6.03275120e-01 -1.21626401e+00
-1.58202171e-01 -6.85752630e-01 2.00580701e-01 -2.94846594e-02
7.39539713e-02 2.07086712e-01 -5.82195342e-01 -1.82825193e-01
-1.21809080e-01 -3.09470743e-01 1.03381884e+00 5.69972456e-01
7.33324707e-01 -8.75184536e-01 -3.60541254e-01 -3.92264053e-02
1.12707925e+00 2.75888503e-01 4.87418771e-01 5.06361306e-01
2.11898252e-01 3.96055609e-01 9.24569726e-01 6.06987774e-01
2.13869289e-01 3.32430065e-01 -2.42817745e-01 1.10624172e-01
1.05892152e-01 -5.13350129e-01 5.50336421e-01 1.18081510e+00
1.21748215e-02 -7.40148544e-01 -8.28264713e-01 8.60802829e-01
-2.03490019e+00 -1.28694797e+00 -1.90803707e-01 2.09830451e+00
9.33048368e-01 6.48710012e-01 2.39673793e-01 2.05858827e-01
6.44390881e-01 4.23905045e-01 -3.04934621e-01 -7.35034943e-01
-7.39133209e-02 7.37963691e-02 5.88799655e-01 3.24035615e-01
-9.87938881e-01 8.98040771e-01 7.41109705e+00 8.20656776e-01
-6.11896038e-01 -1.90411463e-01 2.45808691e-01 -4.54548836e-01
-3.88231337e-01 1.92591444e-01 -8.18385601e-01 3.81917596e-01
1.31991851e+00 -9.99327540e-01 -1.10746361e-01 7.21484005e-01
3.55515420e-01 -5.32619894e-01 -1.21440339e+00 6.87989593e-01
1.80425584e-01 -1.67630768e+00 2.10079625e-02 -7.68891573e-02
7.68348932e-01 -4.11989912e-02 -3.65689367e-01 2.36504808e-01
4.43517178e-01 -5.80992043e-01 7.95725167e-01 4.43092048e-01
8.10575962e-01 -8.10620844e-01 9.06819761e-01 6.23969734e-01
-8.23240340e-01 1.85568348e-01 -3.67246091e-01 1.00158401e-01
4.43696946e-01 6.06342316e-01 -1.28103197e+00 7.58205473e-01
3.38564336e-01 5.37600040e-01 -6.09312892e-01 1.16977119e+00
-2.44499117e-01 9.69530940e-01 -5.57845533e-01 -3.50056440e-01
4.32479143e-01 1.02019854e-01 9.33040142e-01 1.79773426e+00
4.02863801e-01 2.59254456e-01 3.30297917e-01 1.19168252e-01
-3.13778341e-01 3.74757200e-01 -8.37863684e-01 -4.30936992e-01
7.57872045e-01 9.56863999e-01 -1.20212686e+00 -8.73238087e-01
2.42179874e-02 8.65419984e-01 -7.74455443e-02 2.47016326e-01
-4.86050338e-01 -9.71421599e-01 -1.59494340e-01 -2.86520571e-02
2.89085776e-01 -1.89752191e-01 -2.45722517e-01 -1.28579700e+00
8.16494152e-02 -8.91086757e-01 7.36525059e-01 -7.61364162e-01
-7.79794812e-01 6.82238460e-01 4.42342222e-01 -1.27606428e+00
-8.57721150e-01 1.80260256e-01 -8.09723794e-01 2.85883129e-01
-8.69649172e-01 -5.69154322e-01 1.34829342e-01 1.18135512e-02
1.03692782e+00 -2.00335100e-01 8.08779597e-01 -6.81693405e-02
-4.23124701e-01 4.00905937e-01 4.08789307e-01 -5.19713238e-02
8.87148380e-01 -1.55166960e+00 6.90790892e-01 1.19905722e+00
4.03430521e-01 7.97508538e-01 1.33750939e+00 -8.77394497e-01
-1.08191097e+00 -8.38527143e-01 1.41857791e+00 -4.87519205e-01
6.09791040e-01 8.62313155e-03 -7.47786939e-01 7.16771007e-01
6.40329123e-01 -8.65022182e-01 6.91363156e-01 3.30017567e-01
-1.33426890e-01 -5.64317703e-02 -7.60154843e-01 6.14188313e-01
7.73245215e-01 -4.56629306e-01 -1.13180029e+00 8.69964957e-01
6.92385554e-01 -4.66165632e-01 -7.39244878e-01 4.92580049e-02
4.01316106e-01 -6.38364315e-01 4.58825976e-01 -5.34563899e-01
6.04200065e-01 -1.22294195e-01 1.48234770e-01 -1.28806376e+00
7.90664554e-02 -9.09356534e-01 -4.31231856e-01 1.36233926e+00
5.72809398e-01 -4.77202564e-01 4.92341369e-01 4.20634210e-01
-3.17285150e-01 -5.42066157e-01 -4.39756334e-01 -9.51717257e-01
-3.32497507e-01 -2.56839067e-01 3.87499839e-01 6.69094563e-01
5.51189959e-01 6.93659365e-01 -3.34690481e-01 -2.22563654e-01
4.34961945e-01 4.37245995e-01 7.61562049e-01 -1.15937519e+00
-2.26493515e-02 -5.03042400e-01 -8.39109998e-03 -1.18391502e+00
-1.97999969e-01 -7.84286439e-01 1.14542790e-01 -1.97297895e+00
4.70865607e-01 2.44267248e-02 1.12692870e-01 4.25067484e-01
-1.44463211e-01 8.24538991e-02 2.11972669e-02 5.54552078e-01
-1.05699134e+00 2.12106287e-01 7.07392573e-01 -9.56906304e-02
-5.88186979e-01 1.83124214e-01 -1.05305982e+00 7.96867847e-01
8.83142292e-01 -8.39710057e-01 -4.83891010e-01 -2.10327953e-01
2.15791449e-01 2.41490811e-01 -3.41173857e-01 -8.87903035e-01
7.58477449e-01 -2.14675635e-01 2.72133816e-02 -1.19501770e+00
-2.05005229e-01 -2.56508049e-02 -1.71672564e-03 3.43755275e-01
-6.81306422e-01 3.92320961e-01 7.36870915e-02 6.05292857e-01
-4.76363152e-01 -6.27972543e-01 2.80982316e-01 -2.14333251e-01
-4.76866245e-01 3.58490869e-02 -8.51865709e-01 3.75151783e-01
6.27587497e-01 -3.54515880e-01 -3.98153514e-01 -7.21408367e-01
-3.31884384e-01 3.37640345e-01 4.29684728e-01 8.25235471e-02
5.55953026e-01 -9.70243335e-01 -9.98552382e-01 -4.47446227e-01
2.28197709e-01 1.36777398e-03 -2.00894624e-01 6.46425724e-01
-9.07153368e-01 7.02315867e-01 1.00345448e-01 -3.27678293e-01
-1.39337313e+00 3.48656952e-01 -3.95550907e-01 -5.79970717e-01
-1.02553701e+00 3.91224653e-01 -1.35460570e-01 1.33749425e-01
3.93593550e-01 -2.82662719e-01 -3.11379492e-01 5.71281493e-01
9.09856856e-01 5.36960185e-01 3.07026029e-01 -1.68310151e-01
-1.16430961e-01 -5.63171618e-02 -6.44323945e-01 -5.10204315e-01
1.61805677e+00 -1.11466669e-01 1.16167739e-02 6.99384093e-01
9.82134879e-01 3.61436695e-01 -7.18419254e-01 -2.12109163e-01
5.12746751e-01 -2.53052771e-01 -8.41727927e-02 -5.84732950e-01
-2.99751520e-01 3.08952808e-01 -3.61453265e-01 7.36071944e-01
1.19730198e+00 1.25816181e-01 1.04066861e+00 9.55000162e-01
9.26419944e-02 -1.49768639e+00 1.12577803e-01 6.24150932e-01
1.01520097e+00 -8.51448357e-01 5.89000285e-01 -3.92445065e-02
-6.90938413e-01 1.33360386e+00 1.26996145e-01 2.43940298e-02
-3.22008617e-02 1.84102938e-01 -1.74427733e-01 -2.82220066e-01
-1.10312760e+00 4.14509289e-02 1.05279475e-01 3.65146273e-03
2.44824946e-01 -8.04663524e-02 -8.76709163e-01 4.18658972e-01
-6.53832018e-01 -2.03086317e-01 1.21563554e+00 1.18432546e+00
-8.04282725e-01 -8.10884714e-01 -2.90937215e-01 8.31858754e-01
-8.55348647e-01 -1.25947714e-01 -7.00906277e-01 8.90079618e-01
-7.29792237e-01 1.06161380e+00 3.61747034e-02 -1.04807816e-01
5.22905767e-01 2.39559025e-01 3.25613320e-01 -8.89649510e-01
-6.88562393e-01 3.09325695e-01 5.51550508e-01 -3.10056537e-01
-2.32390150e-01 -9.42067027e-01 -1.26490831e+00 -5.57761192e-01
-4.77906108e-01 8.35264802e-01 4.31298226e-01 8.06284070e-01
7.97987804e-02 3.55395198e-01 6.44203007e-01 -4.70004350e-01
-6.27133012e-01 -1.08767116e+00 -5.04059434e-01 2.45582595e-01
3.03347081e-01 4.13615908e-03 -3.07278752e-01 3.62820178e-01]
|
[12.565291404724121, 9.568561553955078]
|
448c490c-0fd7-4974-9d9e-422877dad6ec
|
how-to-turn-your-knowledge-graph-embeddings
|
2305.15944
| null |
https://arxiv.org/abs/2305.15944v1
|
https://arxiv.org/pdf/2305.15944v1.pdf
|
How to Turn Your Knowledge Graph Embeddings into Generative Models via Probabilistic Circuits
|
Some of the most successful knowledge graph embedding (KGE) models for link prediction -- CP, RESCAL, TuckER, ComplEx -- can be interpreted as energy-based models. Under this perspective they are not amenable for exact maximum-likelihood estimation (MLE), sampling and struggle to integrate logical constraints. This work re-interprets the score functions of these KGEs as circuits -- constrained computational graphs allowing efficient marginalisation. Then, we design two recipes to obtain efficient generative circuit models by either restricting their activations to be non-negative or squaring their outputs. Our interpretation comes with little or no loss of performance for link prediction, while the circuits framework unlocks exact learning by MLE, efficient sampling of new triples, and guarantee that logical constraints are satisfied by design. Furthermore, our models scale more gracefully than the original KGEs on graphs with millions of entities.
|
['Antonio Vergari', 'Robert Peharz', 'Nicola Di Mauro', 'Lorenzo Loconte']
|
2023-05-25
| null | null | null | null |
['graph-embedding', 'link-prediction', 'knowledge-graph-embedding', 'knowledge-graph-embeddings', 'knowledge-graph-embeddings']
|
['graphs', 'graphs', 'graphs', 'graphs', 'methodology']
|
[ 1.76350009e-02 9.61994469e-01 -6.12991095e-01 -7.06790686e-02
-4.41597998e-01 -6.94806695e-01 6.74482405e-01 1.83513135e-01
1.28477842e-01 9.36624348e-01 2.59958476e-01 -6.18004441e-01
-3.63904864e-01 -1.23739290e+00 -1.22009075e+00 -4.40682620e-01
-4.33967382e-01 8.47194791e-01 2.78907806e-01 2.85072532e-02
-1.71204478e-01 2.80625165e-01 -1.08754551e+00 -2.88920384e-02
7.30470061e-01 5.32812595e-01 -3.28409970e-01 6.38515353e-01
6.09641410e-02 9.30309355e-01 4.12342995e-02 -9.17335391e-01
-3.96465287e-02 -4.56621796e-01 -7.70108104e-01 -6.73806429e-01
3.81467223e-01 -1.09203279e-01 -8.10501873e-01 8.41599107e-01
2.13749871e-01 -1.51224777e-01 9.04418766e-01 -1.64979231e+00
-9.63989019e-01 1.02102697e+00 -1.88820109e-01 -2.76667684e-01
3.33154172e-01 -1.44627914e-02 1.78383815e+00 -7.33637214e-01
1.16961789e+00 1.19300938e+00 1.14041328e+00 5.07433534e-01
-1.96546030e+00 -1.22836232e-01 4.95725661e-04 4.09021854e-01
-1.49690890e+00 -4.10236329e-01 7.11521924e-01 -2.84557223e-01
1.43935657e+00 4.02406901e-01 1.00031841e+00 1.24130929e+00
2.79329032e-01 7.96101868e-01 5.33092499e-01 -4.71670419e-01
3.84319156e-01 3.15452158e-01 -3.64057980e-02 1.08601487e+00
7.60794878e-01 7.06172064e-02 -7.13470399e-01 -3.68798524e-01
5.96829057e-01 -2.38783494e-01 -3.71061683e-01 -9.83710706e-01
-1.02378201e+00 8.07524741e-01 5.37742257e-01 6.79493695e-02
1.11518681e-01 8.66883993e-01 3.02351743e-01 1.39297843e-01
2.17073351e-01 3.71210992e-01 -5.53348124e-01 1.93134218e-01
-8.25238407e-01 2.91572124e-01 1.12175679e+00 1.17537701e+00
7.14104235e-01 -1.45239949e-01 -9.21453685e-02 2.55610228e-01
5.57376146e-01 2.60754347e-01 -3.56926262e-01 -6.68555915e-01
2.89599031e-01 4.91405636e-01 -5.45239784e-02 -1.08426142e+00
-2.92211682e-01 -4.72678512e-01 -7.23750353e-01 2.11488083e-02
3.11858654e-01 4.54361700e-02 -7.82705247e-01 1.92934752e+00
3.98701206e-02 1.52989939e-01 -1.68195814e-01 3.40335369e-01
2.92490005e-01 7.22351849e-01 4.07764375e-01 -1.06931061e-01
1.04437459e+00 -8.31618249e-01 -5.32101572e-01 -2.90854990e-01
6.23801351e-01 -5.00744730e-02 9.44796860e-01 2.09796637e-01
-1.16911328e+00 -8.22542012e-02 -1.21141756e+00 -2.99502164e-01
-4.50090408e-01 -2.17219770e-01 1.26504934e+00 7.67639339e-01
-1.34783840e+00 9.22263622e-01 -1.07517731e+00 -2.52262652e-01
6.63282573e-01 5.35251796e-01 -4.38058227e-01 -4.44935597e-02
-1.12704301e+00 1.09543073e+00 5.24309695e-01 -1.65931076e-01
-7.75859058e-01 -8.94430101e-01 -8.85770380e-01 5.32908857e-01
3.09796512e-01 -1.10098302e+00 5.20618558e-01 -6.46418273e-01
-1.13916445e+00 7.37389743e-01 -4.27979119e-02 -4.78962839e-01
2.78193533e-01 2.32477739e-01 -2.98674166e-01 -8.13535452e-02
-3.79163444e-01 8.20331693e-01 5.19845665e-01 -1.20437157e+00
1.60434872e-01 -1.60698652e-01 8.11563954e-02 -7.38681182e-02
-5.30010521e-01 -5.70287943e-01 -4.54554230e-01 -3.05264264e-01
8.51967484e-02 -8.32834959e-01 -1.24293551e-01 1.29728958e-01
-6.99571311e-01 -1.40160203e-01 3.37536693e-01 -6.14728928e-01
1.47284806e+00 -1.71706486e+00 5.34439087e-01 6.09942317e-01
5.09281993e-01 -8.87039155e-02 -3.40330303e-02 5.77184796e-01
-1.76688150e-01 4.35437053e-01 -3.92854176e-02 -1.01243921e-01
5.80673397e-01 4.31750804e-01 -1.86033323e-01 3.45499605e-01
4.30414766e-01 1.67209780e+00 -8.88568461e-01 -6.34812832e-01
5.04212966e-03 7.42971122e-01 -1.05125368e+00 -2.91933447e-01
-5.05968332e-01 -2.22288117e-01 -2.62450248e-01 5.67194700e-01
4.20904428e-01 -4.73158717e-01 9.67357039e-01 -3.69792491e-01
4.20821488e-01 5.01591325e-01 -1.15930974e+00 1.55182290e+00
-1.00131109e-01 5.88046908e-01 -1.12962931e-01 -8.73715341e-01
4.77641582e-01 5.21899574e-02 2.31014565e-01 -3.90658468e-01
-1.88320428e-01 6.21616282e-02 -2.06810534e-01 -7.41227865e-02
2.28003293e-01 2.17425730e-02 1.79681908e-02 1.94772393e-01
4.91577357e-01 6.59205243e-02 8.60339180e-02 6.90093338e-01
1.29633713e+00 5.74758649e-01 6.42405525e-02 -3.72404844e-01
-7.72180408e-02 1.12863993e-02 5.29621363e-01 5.75947225e-01
3.14821303e-01 2.08640054e-01 1.09864354e+00 -2.04401240e-01
-1.39258397e+00 -1.27947819e+00 5.24742603e-02 8.03660095e-01
-9.16877314e-02 -8.71185958e-01 -7.19692349e-01 -6.12863004e-01
2.12789938e-01 8.58910382e-01 -7.57011533e-01 -5.32756805e-01
-2.36095622e-01 -9.90363300e-01 5.73300242e-01 6.17023110e-01
-2.51496255e-01 -5.36566675e-01 -1.39592722e-01 2.09580824e-01
1.32902160e-01 -5.79401374e-01 -3.17984968e-01 6.01221025e-01
-8.86982918e-01 -1.03870332e+00 -2.01485887e-01 -7.12766707e-01
8.17109644e-01 -4.49778408e-01 1.46358871e+00 1.16054140e-01
-4.71857756e-01 3.62719685e-01 8.24127421e-02 1.46961063e-01
-2.97272205e-01 2.05210194e-01 -1.06666386e-01 -4.47278142e-01
2.15879589e-01 -1.10249949e+00 -4.98104900e-01 -8.02979618e-02
-5.23419976e-01 8.96967798e-02 6.38108850e-01 8.89373124e-01
6.83732688e-01 6.18755408e-02 3.43734801e-01 -1.06256104e+00
3.72226745e-01 -5.34027994e-01 -5.29386699e-01 7.52845585e-01
-1.25282693e+00 5.65183938e-01 4.51435953e-01 -3.62765282e-01
-5.53862572e-01 6.71339501e-03 3.72399800e-02 -2.07810074e-01
4.45487589e-01 4.57896739e-01 -3.49236399e-01 -2.24256083e-01
6.58380747e-01 -2.61602141e-02 -3.93148452e-01 -2.44794503e-01
7.85763443e-01 -1.15786731e-01 4.25015360e-01 -9.25426543e-01
1.03355515e+00 1.19171478e-01 3.94293308e-01 -2.27702558e-01
-3.49419981e-01 1.92433029e-01 -4.99045610e-01 1.61061287e-02
7.81141102e-01 -9.24052656e-01 -1.07674098e+00 -2.33988062e-01
-8.54860067e-01 -3.36571604e-01 -3.01464796e-01 2.43970409e-01
-5.12425601e-01 3.97452384e-01 -7.35480905e-01 -9.86497879e-01
-2.62960583e-01 -5.98577619e-01 6.98116541e-01 8.24124739e-02
-3.40100676e-01 -1.20678806e+00 7.19376057e-02 1.38717620e-02
4.92445737e-01 3.29237819e-01 1.71230364e+00 -3.80922675e-01
-9.58562553e-01 -2.20634624e-01 -1.48900449e-01 -6.47510514e-02
-5.14859617e-01 2.83971310e-01 -9.15991127e-01 -1.00870937e-01
-9.63856876e-01 -3.10187370e-01 9.51416373e-01 1.39180154e-01
9.26832676e-01 -5.71265042e-01 -8.84800076e-01 6.22785211e-01
1.78533483e+00 -3.94862235e-01 7.52939045e-01 -6.47590756e-02
8.35933745e-01 2.20146090e-01 -1.77199289e-01 2.40131337e-02
6.30757213e-01 5.68340421e-01 5.59857607e-01 1.39548957e-01
-1.91956207e-01 -1.13064504e+00 4.56814080e-01 7.82491028e-01
-2.39940867e-01 -5.10575116e-01 -8.09029341e-01 6.54533505e-01
-2.05976343e+00 -1.06646943e+00 -3.54233027e-01 2.20920825e+00
1.07616961e+00 4.68911380e-01 -3.98619622e-02 -3.30962092e-02
3.04487735e-01 -1.56218195e-02 -5.41722715e-01 -1.01397730e-01
-2.04414696e-01 5.78075051e-01 7.23102152e-01 7.69776225e-01
-6.38252437e-01 8.79406214e-01 7.30841637e+00 9.01684344e-01
-3.68496925e-01 3.64314228e-01 6.09113455e-01 -7.32359216e-02
-1.21362209e+00 5.93596697e-01 -6.41209960e-01 3.14447880e-01
1.19106865e+00 2.14604698e-02 8.10724974e-01 9.25503016e-01
-4.74149555e-01 1.21855311e-01 -1.46923840e+00 5.98372340e-01
-3.55755657e-01 -1.66936636e+00 -4.01415378e-02 1.25151739e-01
6.46100938e-01 -7.19368272e-03 -1.91175595e-01 5.27960122e-01
6.58475339e-01 -1.28076887e+00 7.71427989e-01 7.02451766e-01
8.40123057e-01 -6.47925019e-01 3.63140494e-01 4.37215157e-02
-1.02890837e+00 5.79174347e-02 -5.34409821e-01 1.56051442e-01
3.14776376e-02 8.44363034e-01 -6.67527974e-01 5.23639798e-01
1.80350587e-01 3.18320602e-01 -5.59569836e-01 6.14010096e-01
-5.53021550e-01 7.90472984e-01 -5.23978293e-01 -2.04886436e-01
-1.87360898e-01 -2.23955795e-01 3.08022976e-01 1.20621872e+00
1.78965285e-01 -2.19429553e-01 -4.60801180e-03 1.31773162e+00
-3.10625285e-01 -8.89070407e-02 -5.27507305e-01 -3.59735966e-01
6.39805138e-01 1.12887669e+00 -5.67393005e-01 -9.69161168e-02
-2.83478439e-01 8.97585332e-01 7.20608234e-01 4.71576095e-01
-1.03207958e+00 -3.52490366e-01 4.11472768e-01 2.64251649e-01
3.63685161e-01 -2.66029179e-01 -2.73387998e-01 -1.25018466e+00
-4.57406752e-02 -3.96662802e-01 1.70357555e-01 -8.51281643e-01
-1.21292984e+00 3.04998439e-02 -8.15311596e-02 -3.48167181e-01
-1.06209554e-01 -8.41756463e-01 -6.12198353e-01 6.35543227e-01
-1.41356683e+00 -1.39912021e+00 2.84318000e-01 3.31389964e-01
-3.53522748e-01 2.53799528e-01 1.07697690e+00 4.23111409e-01
-5.54169893e-01 8.44891667e-01 4.62212376e-02 -1.55947460e-02
2.21556321e-01 -1.62765408e+00 4.22885329e-01 5.79781473e-01
4.70988125e-01 7.80931890e-01 7.77184308e-01 -7.18721092e-01
-1.97462404e+00 -7.49251723e-01 1.13815796e+00 -7.59853661e-01
8.04515004e-01 -8.40819478e-01 -7.09586799e-01 1.14425421e+00
1.25154212e-01 1.28930688e-01 6.29611850e-01 6.72417581e-01
-6.91575587e-01 5.03366701e-02 -9.87755537e-01 5.52169621e-01
1.46035087e+00 -7.27228045e-01 -2.30460748e-01 5.77051818e-01
7.09707439e-01 -5.36251329e-02 -1.32164860e+00 2.18085736e-01
6.93964899e-01 -7.32477844e-01 1.23370123e+00 -8.41546416e-01
3.30697864e-01 -2.95583665e-01 -2.11460516e-01 -1.06487811e+00
-6.05301738e-01 -7.30828285e-01 -8.06212366e-01 1.24592793e+00
8.85915041e-01 -5.06682038e-01 8.77692401e-01 6.92510962e-01
1.07126594e-01 -9.56828952e-01 -7.53038228e-01 -6.49827480e-01
-5.96340261e-02 -3.09033662e-01 5.52071214e-01 1.14996243e+00
4.76343304e-01 5.40319562e-01 -2.49507383e-01 1.43937752e-01
8.20501089e-01 -1.02370486e-01 4.08521593e-01 -1.31622255e+00
-8.77922833e-01 -6.33164525e-01 -5.97484171e-01 -6.83250964e-01
1.85900927e-01 -1.41902614e+00 -4.46894884e-01 -1.80396271e+00
5.26118815e-01 -5.77681601e-01 -2.04177395e-01 9.17515099e-01
3.37523632e-02 2.09993750e-01 -1.95377663e-01 -1.11878045e-01
-6.13420069e-01 5.41483700e-01 6.47015333e-01 -3.01585227e-01
1.75116032e-01 -3.79048318e-01 -7.90893078e-01 3.44724894e-01
4.99322772e-01 -4.85997945e-01 -6.62749171e-01 -1.99029729e-01
1.22912049e+00 -2.82728449e-02 5.75304806e-01 -6.51685476e-01
2.27076113e-01 1.58976912e-02 5.62841475e-01 -2.27746740e-01
1.87926292e-01 -6.60628259e-01 8.41136813e-01 4.33698654e-01
-3.20356220e-01 -3.84480208e-01 9.98619571e-02 9.06375408e-01
2.57717729e-01 -2.42874131e-01 3.23211432e-01 -9.10269618e-02
-3.00389558e-01 2.92851418e-01 6.89147040e-02 -2.86091473e-02
8.36430013e-01 -1.40055194e-01 -4.65777636e-01 -2.78662950e-01
-1.07538736e+00 2.57734179e-01 6.48169398e-01 -6.01682439e-02
2.50941277e-01 -1.45073617e+00 -5.37675917e-01 -6.74180984e-02
-2.57845665e-03 -1.76519632e-01 -1.07253445e-02 8.09234619e-01
-4.66826230e-01 4.09189701e-01 3.35291587e-02 -1.67157531e-01
-7.22271681e-01 6.28773212e-01 3.10367614e-01 -6.42893553e-01
-4.49943453e-01 1.01849115e+00 -3.23832661e-01 -4.91451085e-01
1.46192953e-01 -1.35344684e-01 4.65468973e-01 -6.76801279e-02
-1.22540668e-01 2.76997477e-01 -1.44422259e-02 7.43943900e-02
-4.61201489e-01 3.12078178e-01 1.10743642e-01 8.33587646e-02
1.62852633e+00 1.80078760e-01 -3.73092741e-01 1.35216758e-01
1.04592824e+00 2.20407248e-01 -1.16986573e+00 6.93073496e-02
3.06535792e-02 9.24519375e-02 7.87257180e-02 -1.05556417e+00
-7.30430245e-01 8.10073256e-01 1.79426387e-01 1.84783146e-01
6.10268831e-01 1.73721164e-01 4.71945167e-01 5.71040392e-01
5.16282558e-01 -1.01189315e+00 -3.83790553e-01 1.25361830e-01
3.23746175e-01 -7.75851071e-01 2.94429600e-01 -4.61542606e-01
-1.29961863e-01 9.67606366e-01 3.47366899e-01 -3.06077115e-02
5.47882378e-01 2.94366360e-01 -1.03197837e+00 -2.88778305e-01
-1.09089172e+00 3.51983197e-02 3.17923278e-01 7.62916923e-01
3.76144230e-01 3.30129415e-01 -1.15871333e-01 6.55601621e-01
-6.59731999e-02 -1.32175133e-01 2.76038617e-01 6.15532458e-01
-3.61638635e-01 -1.30234587e+00 2.21595556e-01 4.73067284e-01
-1.48686022e-01 -4.08095181e-01 -4.24912244e-01 9.94634688e-01
5.72898649e-02 4.32699412e-01 -1.96533114e-01 -4.93076056e-01
8.81477296e-02 6.11419857e-01 1.01256454e+00 -3.40788871e-01
-1.19098641e-01 -2.25703984e-01 3.67651194e-01 -6.00886166e-01
7.27883074e-03 -4.14067864e-01 -1.24267840e+00 -7.38559902e-01
-6.25398159e-01 1.59536630e-01 5.04630625e-01 5.49236298e-01
6.38798177e-01 5.23738921e-01 1.39107600e-01 -5.58835447e-01
-5.43640375e-01 -5.40709019e-01 -5.87976277e-01 6.98147416e-02
-1.89885274e-01 -4.62334216e-01 -2.25614637e-01 -7.29514658e-02]
|
[8.647440910339355, 7.475470542907715]
|
09039f11-6d03-4ba6-aa61-281955c0c278
|
transferring-rich-deep-features-for-facial
|
1803.07253
| null |
http://arxiv.org/abs/1803.07253v1
|
http://arxiv.org/pdf/1803.07253v1.pdf
|
Transferring Rich Deep Features for Facial Beauty Prediction
|
Feature extraction plays a significant part in computer vision tasks. In this
paper, we propose a method which transfers rich deep features from a pretrained
model on face verification task and feeds the features into Bayesian ridge
regression algorithm for facial beauty prediction. We leverage the deep neural
networks that extracts more abstract features from stacked layers. Through
simple but effective feature fusion strategy, our method achieves improved or
comparable performance on SCUT-FBP dataset and ECCV HotOrNot dataset. Our
experiments demonstrate the effectiveness of the proposed method and clarify
the inner interpretability of facial beauty perception.
|
['Lu Xu', 'Jinhai Xiang', 'Xiaohui Yuan']
|
2018-03-20
| null | null | null | null |
['facial-beauty-prediction']
|
['computer-vision']
|
[-9.71135795e-02 -2.04952713e-02 -2.22607553e-01 -1.07793117e+00
-3.23060155e-01 5.78219332e-02 6.63127542e-01 -6.19867265e-01
1.29066664e-03 4.74809378e-01 1.31594449e-01 6.56519830e-02
-3.17846358e-01 -5.65381885e-01 -5.56040287e-01 -6.25720978e-01
6.74127694e-03 -1.83623478e-01 -4.90111709e-01 -2.31914908e-01
1.32085308e-01 7.27531970e-01 -1.58477664e+00 3.08406264e-01
4.40003335e-01 1.88002932e+00 -5.01432300e-01 2.44112968e-01
1.18098602e-01 6.76129162e-01 -1.66521370e-01 -9.60094035e-01
4.51923311e-01 1.20158093e-02 -4.99123633e-01 1.11970887e-01
6.93778574e-01 -6.14638388e-01 -4.99798566e-01 9.52351272e-01
3.24199468e-01 -2.15986684e-01 8.18796217e-01 -1.58201075e+00
-9.60753858e-01 3.55837554e-01 -9.87497389e-01 -9.77684185e-02
1.36771262e-01 7.39106461e-02 9.97574627e-01 -1.09270632e+00
2.95700759e-01 1.58283627e+00 8.22694004e-01 6.41713858e-01
-8.91658664e-01 -1.32979918e+00 -4.77860235e-02 2.83253580e-01
-1.37969732e+00 -9.15793836e-01 9.21385407e-01 -3.78542066e-01
3.45840752e-01 1.95393369e-01 5.90243459e-01 1.14866388e+00
2.08198875e-01 8.94212365e-01 1.34646404e+00 -3.77067663e-02
-2.86283135e-01 2.88598891e-02 1.74000472e-01 1.35376227e+00
1.89893290e-01 4.70113695e-01 -7.71089673e-01 -1.61582023e-01
5.75552106e-01 3.63677591e-01 -4.61301729e-02 7.48849800e-03
-4.74400699e-01 9.74564254e-01 7.20928729e-01 -6.78973868e-02
-2.67518461e-01 3.12619627e-01 2.24759698e-01 1.63302556e-01
3.19552600e-01 -3.18558514e-02 -4.55317855e-01 3.51193666e-01
-9.09436822e-01 2.07923483e-02 2.45319933e-01 8.07712793e-01
6.22845769e-01 2.75454354e-02 -3.52535188e-01 9.34418917e-01
8.57411087e-01 6.12055540e-01 1.59620911e-01 -7.90523350e-01
-1.54471964e-01 6.63498998e-01 -3.50127161e-01 -1.32094121e+00
-3.38273317e-01 -2.26895809e-01 -9.07319009e-01 4.80324715e-01
-2.36079581e-02 1.02840915e-01 -1.09102499e+00 1.46568632e+00
3.37778628e-01 1.51886299e-01 -1.71604946e-01 8.68507385e-01
1.36825514e+00 3.42583150e-01 3.27923656e-01 1.55199960e-01
1.55560565e+00 -7.80466557e-01 -6.60081208e-01 2.78938949e-01
-5.21608107e-02 -7.44944632e-01 5.33000410e-01 6.05621099e-01
-7.29654133e-01 -6.08454466e-01 -1.13051891e+00 -3.48676473e-01
-2.21592575e-01 6.45041883e-01 1.13116300e+00 9.26120341e-01
-8.37679267e-01 7.43447661e-01 -4.75677609e-01 1.97565630e-02
1.41346240e+00 6.40534759e-01 -7.55303979e-01 -9.02582332e-02
-8.19177032e-01 7.43934155e-01 1.32610664e-01 7.07822561e-01
-9.30425823e-01 -6.10197723e-01 -1.18695879e+00 -5.39970249e-02
3.60858664e-02 -4.65477645e-01 8.94910991e-01 -1.02909768e+00
-1.74024117e+00 7.18096137e-01 -2.25744367e-01 -4.49539460e-02
2.70222574e-01 -2.29527205e-01 -4.24328536e-01 -9.62963887e-03
-3.91720533e-01 9.06410992e-01 1.33242249e+00 -1.11825776e+00
-3.65724206e-01 -8.56771410e-01 -3.60269010e-01 -4.21198100e-01
-4.08772886e-01 3.14699173e-01 -2.30587661e-01 -4.17587221e-01
1.88787177e-01 -8.35865617e-01 1.09362200e-01 6.11168027e-01
-5.09425223e-01 -5.63742518e-01 1.17360485e+00 -7.89723873e-01
7.19812572e-01 -2.18859196e+00 -9.08467174e-02 4.33884293e-01
4.15332258e-01 1.52364895e-01 -3.26343268e-01 -2.16091588e-01
-5.57138547e-02 1.36997253e-01 -1.09818932e-02 -5.72365224e-01
2.32494682e-01 -1.54053941e-02 -3.35460067e-01 5.66568315e-01
7.54094541e-01 1.24989903e+00 -3.18273753e-01 -6.67958975e-01
2.27573328e-02 1.00506294e+00 -5.44561267e-01 1.26885250e-01
4.14833799e-02 2.96460658e-01 -5.27393699e-01 1.15752363e+00
1.17696738e+00 1.04550198e-01 -1.24005392e-01 -6.48615241e-01
3.65927994e-01 -1.57532424e-01 -6.32356048e-01 1.61544800e+00
-3.51122230e-01 7.17779517e-01 -4.42620404e-02 -5.15159667e-01
1.31059086e+00 -9.01999325e-02 1.49005696e-01 -6.96880341e-01
7.17019677e-01 -1.90261900e-01 -9.25148353e-02 -5.70058584e-01
3.59181970e-01 -3.72902811e-01 1.16651729e-01 1.00728005e-01
5.76657712e-01 6.90293163e-02 -4.28146154e-01 -2.08083421e-01
6.38012826e-01 4.74040389e-01 3.75370532e-01 -8.38302076e-02
5.64875066e-01 -7.21126139e-01 7.71523297e-01 8.97012278e-02
-5.51779211e-01 5.02594829e-01 5.59487820e-01 -6.53486848e-01
-6.09608889e-01 -8.75353634e-01 -4.40041810e-01 1.25186706e+00
-2.20596775e-01 -5.28373837e-01 -3.02397668e-01 -1.13618159e+00
3.55672956e-01 1.24891408e-01 -1.27046359e+00 -1.96882144e-01
-1.40424103e-01 -7.28635609e-01 7.61583686e-01 5.96325576e-01
8.27801168e-01 -8.76506567e-01 -2.02023014e-01 -5.71459830e-01
3.61938119e-01 -9.88438845e-01 -1.34271756e-01 -4.51701909e-01
-5.33008873e-01 -1.11129844e+00 -4.76772487e-01 -4.87346977e-01
6.60210848e-01 -1.40166050e-02 6.22574925e-01 4.17488277e-01
-6.81707978e-01 -1.64036125e-01 -1.06234692e-01 -6.20639682e-01
4.00102348e-04 -4.61520016e-01 -2.81048957e-02 5.21176577e-01
5.48453331e-01 -2.62387693e-01 -6.77541137e-01 2.05623791e-01
-5.19106030e-01 -1.26063183e-01 7.26779699e-01 8.65466416e-01
1.18408054e-01 -1.04491308e-01 3.76005381e-01 -6.58897936e-01
3.39875311e-01 -3.24064791e-01 -7.11394727e-01 2.73478329e-01
-4.53169495e-01 1.88560709e-01 1.36718750e-01 2.96511431e-03
-1.23778510e+00 3.37928236e-01 -2.81851798e-01 -6.43152237e-01
-1.04202814e-01 1.74685970e-01 -4.19779122e-01 -4.83085424e-01
3.28473926e-01 4.10753116e-02 2.65939832e-01 -4.89838362e-01
3.99879456e-01 6.68749750e-01 3.79373997e-01 -6.57707632e-01
9.28818345e-01 5.40775239e-01 3.63779485e-01 -4.65909392e-01
-7.84105361e-01 8.15762654e-02 -7.20900297e-01 -2.49746159e-01
6.59271896e-01 -8.82620037e-01 -1.43753314e+00 5.32469153e-01
-1.18310928e+00 3.81837070e-01 3.12258899e-01 2.23761037e-01
-1.73018754e-01 1.81270748e-01 -5.45428872e-01 -9.86986995e-01
-5.74195206e-01 -1.10363603e+00 1.37140632e+00 4.94709671e-01
1.42803103e-01 -4.38956678e-01 -7.09869266e-02 5.41169286e-01
3.46010596e-01 3.82547677e-01 6.18073821e-01 -3.89747828e-01
-4.53520596e-01 -1.55300558e-01 -8.95204782e-01 4.19767290e-01
2.20040515e-01 5.81120193e-01 -1.56629384e+00 2.19247267e-02
-3.37459177e-01 -6.94829464e-01 1.36358798e+00 1.77366912e-01
1.66553974e+00 -3.47161859e-01 -1.53770000e-01 9.84055459e-01
1.29951608e+00 -1.48604780e-01 6.46107793e-01 -2.25965694e-01
6.63346410e-01 9.23636615e-01 3.44837129e-01 5.54755390e-01
3.00757319e-01 3.57808143e-01 4.51535940e-01 -1.43686593e-01
-2.07466632e-01 -2.06108868e-01 2.59862065e-01 3.16847295e-01
-2.95459300e-01 5.79384506e-01 -6.52707696e-01 9.43881124e-02
-1.63586116e+00 -9.16727662e-01 2.35839993e-01 1.46867001e+00
7.32110620e-01 -2.73245305e-01 -5.51191941e-02 -2.66072173e-02
4.89478767e-01 1.50330409e-01 -4.22557712e-01 -4.50110227e-01
-7.49021173e-02 4.53774571e-01 6.12921603e-02 1.42173171e-01
-1.29530489e+00 9.50598955e-01 6.57475424e+00 7.21132994e-01
-1.25876641e+00 1.32011324e-01 1.09024131e+00 -1.18466839e-01
-8.30511525e-02 -4.14343208e-01 -8.10552001e-01 3.02348912e-01
6.49236143e-01 3.38319570e-01 1.52836025e-01 9.75229084e-01
-1.89614832e-01 1.65549293e-01 -1.13058197e+00 1.30870688e+00
4.36723053e-01 -1.25651705e+00 2.27439284e-01 1.37285858e-01
4.86451566e-01 -3.07788104e-01 6.54520392e-01 3.34683925e-01
2.98182517e-01 -1.75561392e+00 4.36323166e-01 6.78758025e-01
8.71878564e-01 -8.31097662e-01 6.16141200e-01 -3.47619832e-01
-9.15397704e-01 -2.87634194e-01 -5.81665039e-01 -2.31111031e-02
-3.21398079e-01 4.58512425e-01 -8.51261556e-01 4.96019572e-01
8.76975596e-01 8.84614408e-01 -8.36192548e-01 8.32324088e-01
-2.92760193e-01 2.58546561e-01 -1.71221375e-01 1.03699520e-01
-8.17212760e-02 2.13922914e-02 -9.80717987e-02 1.07678211e+00
2.50652671e-01 1.52395055e-01 -3.66109520e-01 1.08141255e+00
-4.37181145e-01 6.83395639e-02 -7.53835559e-01 -8.19609761e-02
5.71004823e-02 1.86211658e+00 -2.70015746e-01 3.27900089e-02
-3.72844636e-01 6.48261487e-01 3.49641085e-01 6.24207407e-02
-9.28550184e-01 -1.92160040e-01 8.46214890e-01 -3.73347044e-01
3.70439529e-01 1.12869412e-01 -5.37052691e-01 -1.06493640e+00
-1.52848110e-01 -6.81976557e-01 2.54085869e-01 -7.41358876e-01
-1.51782155e+00 7.69101024e-01 -3.51416081e-01 -9.01974440e-01
2.45807171e-01 -1.09248233e+00 -6.87823713e-01 5.73114932e-01
-1.83370519e+00 -1.86972332e+00 -4.85041648e-01 8.74449670e-01
1.79197848e-01 -5.70468485e-01 7.12777734e-01 9.72383097e-02
-8.95034373e-01 9.37964559e-01 -5.37110865e-01 6.75852060e-01
5.67914784e-01 -7.09236681e-01 1.09988406e-01 6.00075424e-01
9.81087536e-02 7.75505602e-01 2.47586146e-01 -4.76021022e-01
-1.57292247e+00 -1.15148318e+00 4.71027106e-01 -4.55203742e-01
4.46656585e-01 -4.24073309e-01 -5.51752448e-01 5.55457294e-01
2.92049170e-01 3.69494677e-01 1.02261853e+00 4.33009118e-01
-1.05990052e+00 -5.04063547e-01 -1.46545041e+00 2.69469976e-01
9.42777574e-01 -4.45930034e-01 -6.00720108e-01 -5.02875149e-02
4.44565862e-01 4.80452329e-02 -8.58716190e-01 8.77114475e-01
1.26190054e+00 -9.32156920e-01 8.32987964e-01 -1.03297687e+00
8.50201726e-01 -1.86909568e-02 -3.54099721e-01 -8.76754522e-01
-2.82329440e-01 -5.39600074e-01 -5.66881290e-03 1.30372381e+00
4.24318045e-01 -3.26302230e-01 8.82543743e-01 6.48749232e-01
3.31563652e-01 -1.01088738e+00 -9.91438806e-01 -2.76461780e-01
3.07988096e-03 -3.93885106e-01 9.21487391e-01 8.79053712e-01
-2.72253215e-01 9.25737768e-02 -6.20925009e-01 9.04895961e-02
8.78660381e-01 1.81902945e-01 7.25979507e-01 -1.49891448e+00
-2.48419605e-02 -4.89863813e-01 -7.26388752e-01 -3.66726547e-01
6.37478292e-01 -9.55662251e-01 -6.85363114e-02 -8.20665181e-01
7.15444624e-01 -2.96552628e-01 -6.64226532e-01 8.91000152e-01
7.49305403e-03 8.42766523e-01 3.44817072e-01 -4.02738869e-01
-4.90189224e-01 1.00737309e+00 1.30354977e+00 -4.02164578e-01
2.88094401e-01 -2.68388718e-01 -9.20235753e-01 7.91332126e-01
6.80911839e-01 -1.66859373e-01 -9.09437090e-02 -4.02641565e-01
4.57103252e-02 -3.29816163e-01 6.84520304e-01 -6.41084790e-01
1.28798271e-02 -1.33020595e-01 1.30216217e+00 -3.21163118e-01
7.27536380e-01 -8.79659534e-01 -2.07123071e-01 3.35900605e-01
-2.11484522e-01 -3.02683949e-01 4.04644608e-01 3.60166341e-01
-7.54934996e-02 3.34656626e-01 8.59204173e-01 3.85713130e-01
-6.54967070e-01 7.21527278e-01 3.71514142e-01 -6.93272054e-01
8.98429751e-01 3.90090495e-02 -4.08928752e-01 -1.25378415e-01
-6.99669957e-01 1.20262511e-01 -6.87999977e-03 7.70094514e-01
1.18031394e+00 -1.78641486e+00 -9.51194227e-01 8.03020060e-01
2.36569554e-01 -5.56744933e-01 1.73162192e-01 7.54501343e-01
-1.89817384e-01 2.54088938e-01 -7.44720280e-01 -6.92064345e-01
-1.57339442e+00 3.06026429e-01 3.28934371e-01 5.85997641e-01
-3.09810042e-01 1.26648629e+00 1.28348336e-01 -3.23313624e-01
1.06027052e-01 -1.15528375e-01 -4.02028114e-01 3.73073779e-02
7.58963227e-01 4.37324867e-02 -9.65145752e-02 -9.70745623e-01
-4.97984201e-01 7.82420337e-01 -3.50536138e-01 1.68210402e-01
1.51223147e+00 1.96206033e-01 -3.79824668e-01 -1.24385245e-01
1.44817436e+00 -1.69728845e-01 -1.33908165e+00 -1.40892968e-01
-1.29474059e-01 -9.24039364e-01 2.81321824e-01 -8.82310748e-01
-1.67292619e+00 9.70486462e-01 7.78715193e-01 -5.10787785e-01
1.35518301e+00 1.74135178e-01 7.28766322e-01 3.70862693e-01
2.35074610e-01 -6.45567954e-01 -6.16585985e-02 2.57812813e-02
1.20863378e+00 -1.67298496e+00 2.39452541e-01 -4.23060328e-01
-5.71490824e-01 1.34701002e+00 6.92927241e-01 -2.23162770e-01
1.17733109e+00 2.35829189e-01 1.15394332e-01 -3.34095925e-01
-7.07281828e-01 -6.12151474e-02 7.04782069e-01 4.34756726e-01
3.44875574e-01 9.28557366e-02 -5.86991869e-02 1.03007817e+00
-3.88205230e-01 2.84569502e-01 -5.86472750e-02 5.54826796e-01
-3.24702173e-01 -9.34616089e-01 -2.80857265e-01 4.48646843e-01
-4.82163727e-01 -5.65158986e-02 -7.08454847e-01 7.38706589e-01
3.57558280e-01 8.19108605e-01 -8.16761330e-02 -7.83292711e-01
-1.56382367e-01 1.52604625e-01 8.69652390e-01 -1.03323378e-01
-4.34680760e-01 -4.61304188e-02 -8.79368186e-02 -9.88462269e-01
-3.37368399e-01 -4.96005356e-01 -9.32644129e-01 -4.79748636e-01
-1.96489498e-01 -2.81621277e-01 1.00727677e+00 1.03602397e+00
3.66143197e-01 2.22770974e-01 8.95101070e-01 -7.27595150e-01
-3.95061046e-01 -1.02638018e+00 -5.64367652e-01 3.28232527e-01
5.98317504e-01 -1.04758596e+00 1.76780799e-03 -1.16684780e-01]
|
[13.369916915893555, 0.8413964509963989]
|
36b206b0-f1b6-4093-9365-9e840aa0e421
|
fast-global-registration
| null | null |
http://vladlen.info/papers/fast-global-registration.pdf
|
http://vladlen.info/papers/fast-global-registration.pdf
|
Fast Global Registration
|
We present an algorithm for fast global registration of partially overlapping 3D surfaces. The algorithm operates on candidate matches that cover the
surfaces. A single objective is optimized to align the surfaces and disable false
matches. The objective is defined densely over the surfaces and the optimization achieves tight alignment with no initialization. No correspondence updates
or closest-point queries are performed in the inner loop. An extension of the algorithm can perform joint global registration of many partially overlapping surfaces. Extensive experiments demonstrate that the presented approach matches
or exceeds the accuracy of state-of-the-art global registration pipelines, while being at least an order of magnitude faster. Remarkably, the presented approach is
also faster than local refinement algorithms such as ICP. It provides the accuracy achieved by well-initialized local refinement algorithms, without requiring
an initialization and at lower computational cost.
|
['Vladlen Koltun', 'Jaesik Park', 'Qian-Yi Zhou']
|
2016-10-08
| null | null | null |
eccv-2016-10
|
['point-cloud-registration']
|
['computer-vision']
|
[ 4.04216439e-01 3.86538535e-01 4.73014742e-01 -9.66773629e-02
-1.27312148e+00 -2.83444017e-01 6.23016298e-01 5.00967622e-01
-3.30843627e-01 3.65211546e-01 -3.40178460e-01 2.08099231e-01
2.36673607e-03 -8.27781677e-01 -8.39798450e-01 -4.21693951e-01
-1.11669958e-01 1.26769412e+00 9.02402043e-01 -2.79534012e-01
4.67646509e-01 8.32616329e-01 -1.70062256e+00 -1.80222079e-01
7.03323245e-01 9.63859081e-01 1.30581468e-01 5.12460709e-01
-3.99663821e-02 -4.41823572e-01 -3.49705487e-01 -1.88007519e-01
5.35200298e-01 5.13327494e-02 -8.97646070e-01 -8.52887426e-03
1.05102336e+00 -9.64095369e-02 3.12579662e-01 1.02457976e+00
3.58139098e-01 1.05315149e-02 4.51007336e-01 -7.83413231e-01
1.80652097e-01 -2.35032409e-01 -5.99563360e-01 -3.96053761e-01
8.63953352e-01 -2.62057006e-01 8.03012550e-01 -1.41656089e+00
8.25410962e-01 1.10983324e+00 1.18718839e+00 -4.67118947e-03
-1.40562177e+00 -4.76305246e-01 -1.11185707e-01 -5.37829459e-01
-1.81480718e+00 -5.74861705e-01 4.76103127e-01 -3.59365284e-01
1.08346188e+00 5.21734595e-01 5.71397185e-01 1.70538351e-01
2.81904161e-01 2.71359459e-02 6.95499122e-01 -4.91414905e-01
1.95978492e-01 -3.74892682e-01 3.83364875e-03 6.28642023e-01
4.53308046e-01 3.77604887e-02 -6.72156334e-01 -7.76713014e-01
1.24112833e+00 -1.43581361e-01 -5.13403676e-03 -7.46149063e-01
-1.37936878e+00 3.83329064e-01 1.83504269e-01 3.43068182e-01
-4.46304381e-01 1.60542801e-01 4.24147360e-02 2.06836283e-01
7.31546581e-01 3.49126458e-01 -4.59350407e-01 6.36748821e-02
-1.14848077e+00 4.55065757e-01 7.28210628e-01 1.14427888e+00
1.53118896e+00 -6.79575264e-01 4.61348116e-01 5.12783647e-01
2.94397771e-01 6.16466463e-01 -2.64755338e-01 -1.18858051e+00
2.34758928e-01 7.87435234e-01 3.01680505e-01 -1.36626220e+00
-5.66672683e-01 8.00975971e-03 -5.40692687e-01 6.31231844e-01
1.87357575e-01 4.22614455e-01 -8.99864376e-01 1.05456948e+00
9.53933001e-01 4.58105892e-01 -4.17611003e-01 5.43564022e-01
5.54883242e-01 1.49346590e-01 -5.16443670e-01 -1.74466819e-01
1.17908061e+00 -5.31039953e-01 -5.59885740e-01 -1.50521800e-01
3.99927884e-01 -1.44263947e+00 5.67174137e-01 1.75918207e-01
-1.55232811e+00 -4.67535883e-01 -1.10953534e+00 -2.69513428e-01
-3.34000625e-02 -3.47453326e-01 3.33397359e-01 2.52903402e-01
-1.45084572e+00 9.27982926e-01 -1.19277644e+00 -5.72778702e-01
1.97720855e-01 8.00893188e-01 -6.30347490e-01 1.83317527e-01
-4.79928732e-01 8.74596834e-01 -1.52163938e-01 -5.32092005e-02
-3.74040157e-01 -9.21994865e-01 -9.33525562e-01 -4.09711272e-01
3.33274975e-02 -7.15549409e-01 1.12111950e+00 -5.42622745e-01
-1.54665363e+00 1.44337034e+00 -6.83801413e-01 1.24485888e-01
6.91932201e-01 -4.88398463e-01 1.54456079e-01 5.08370586e-02
3.42014372e-01 3.35038424e-01 4.36717033e-01 -1.51900446e+00
-4.06783372e-01 -5.28478682e-01 -5.33913255e-01 2.51751661e-01
2.84501791e-01 2.04297498e-01 -8.77407968e-01 -4.17525202e-01
9.12128508e-01 -8.84611547e-01 -5.93747079e-01 4.48162764e-01
-4.06522393e-01 4.27407026e-03 8.27396452e-01 -5.24162591e-01
8.06087255e-01 -2.05528784e+00 -5.66323400e-02 9.04117107e-01
2.77894676e-01 -3.26298177e-01 -8.73201862e-02 3.73632133e-01
3.05154353e-01 -8.73805508e-02 -7.35346735e-01 -9.79601741e-01
-2.04033285e-01 2.77708527e-02 1.06566846e-01 1.03219295e+00
-1.63343266e-01 5.92301846e-01 -6.98774338e-01 -4.12421077e-01
3.34909528e-01 5.87724626e-01 -4.44042623e-01 3.30392629e-01
-7.04391450e-02 6.22712910e-01 -4.29379672e-01 7.27330267e-01
1.12421632e+00 -3.14973116e-01 1.66386902e-01 -2.00823203e-01
-4.30553287e-01 4.97209758e-01 -1.75678265e+00 1.99774230e+00
-1.10901065e-01 1.15483478e-01 5.72314143e-01 -6.08767688e-01
1.33749974e+00 4.24694926e-01 9.57663476e-01 -3.43975991e-01
3.77745703e-02 7.11771429e-01 -6.35667205e-01 2.98601657e-01
4.51833963e-01 2.69446783e-02 1.86342821e-01 5.97503126e-01
-3.09208006e-01 -7.21593559e-01 -3.99719179e-01 -1.95950538e-01
1.15197456e+00 4.59980190e-01 7.06824601e-01 -6.60820305e-01
6.10078096e-01 3.05804819e-01 4.40448254e-01 4.73750502e-01
3.90111804e-01 9.71345127e-01 -2.80094206e-01 -5.89295745e-01
-1.03437901e+00 -1.12370431e+00 -5.17629206e-01 5.46189189e-01
8.45561862e-01 -5.55340290e-01 -7.60369718e-01 -5.42590171e-02
2.16528267e-01 -1.37818828e-01 -5.69497049e-01 6.06314480e-01
-9.64796960e-01 -2.45617241e-01 6.65195361e-02 2.84219146e-01
1.87789440e-01 -6.19409621e-01 -5.60702562e-01 2.20783323e-01
3.18047911e-01 -1.07227027e+00 -3.76030535e-01 -2.11265743e-01
-1.33661377e+00 -1.23392665e+00 -5.51512539e-01 -8.69437039e-01
1.18213463e+00 3.03117305e-01 1.42408240e+00 5.89795113e-01
-2.21769303e-01 3.52401137e-01 5.78515455e-02 -9.74967927e-02
-3.04508746e-01 9.26453844e-02 1.03538536e-01 -2.12741390e-01
2.36244842e-01 -9.23416793e-01 -5.12034237e-01 8.29736710e-01
-3.67068678e-01 -7.81176984e-02 3.05203408e-01 3.73862088e-01
1.35449696e+00 -2.68271297e-01 -2.17837363e-01 -8.28323007e-01
5.00377156e-02 -1.02927759e-01 -9.05537784e-01 1.81883499e-02
-3.96013647e-01 1.24542685e-02 -9.17936713e-02 -9.72268656e-02
-8.59572232e-01 5.57588696e-01 -1.54954106e-01 -2.55099207e-01
-2.37038463e-01 3.00708711e-02 -1.19361579e-01 -7.70950615e-01
6.21679127e-01 -1.96586430e-01 2.73442149e-01 -8.79410326e-01
1.28767863e-01 3.62170517e-01 7.01322079e-01 -7.13860750e-01
1.20485294e+00 1.12930202e+00 2.35589013e-01 -7.88738728e-01
-3.97373796e-01 -7.47689247e-01 -1.39215803e+00 -1.99725740e-02
7.07800567e-01 -6.59889579e-01 -5.35569429e-01 4.82403040e-01
-1.54425251e+00 -2.64884382e-01 -2.29132205e-01 1.01887576e-01
-7.48093367e-01 3.38122368e-01 -2.63385624e-01 -5.25776446e-01
-2.90159881e-01 -1.15462148e+00 1.69638443e+00 -9.26634222e-02
-6.03321552e-01 -9.41909254e-01 5.01997769e-01 8.92250892e-03
2.93718487e-01 6.41389966e-01 2.08328471e-01 -3.78321141e-01
-9.86732364e-01 -3.39212507e-01 1.00349426e-01 -3.58124793e-01
4.77640808e-01 1.23324104e-01 -9.37329233e-01 -3.79499644e-01
-6.83977082e-02 2.01974615e-01 2.09698483e-01 2.98086047e-01
6.01921737e-01 -6.29202882e-03 -8.29167902e-01 7.24124551e-01
1.41043031e+00 -2.96809882e-01 7.52995253e-01 4.64146018e-01
6.11742318e-01 6.74951196e-01 8.08395863e-01 2.12052956e-01
3.58736932e-01 1.04547250e+00 5.23192644e-01 -4.74982768e-01
2.47988496e-02 1.62576422e-01 1.80605799e-02 7.21406519e-01
-5.31497478e-01 5.32177091e-01 -1.25467789e+00 4.39509749e-01
-1.91438198e+00 -4.88446325e-01 -6.31891549e-01 2.58654857e+00
8.59075606e-01 -3.70923765e-02 -1.12608440e-01 -1.55900523e-01
6.86158061e-01 -8.80489647e-02 -1.96100011e-01 -1.35378703e-01
-1.60872340e-02 7.83050537e-01 6.23089254e-01 1.25103986e+00
-1.23760414e+00 1.09680557e+00 7.60402822e+00 3.19181591e-01
-7.33431518e-01 1.51549056e-01 1.19067654e-01 2.35332564e-01
-3.42738777e-01 1.77865341e-01 -8.18559766e-01 -2.77724534e-01
4.27111894e-01 9.14154723e-02 1.67825967e-01 7.17217982e-01
-5.24219249e-05 -2.33965591e-01 -1.05926406e+00 8.31272364e-01
6.53674901e-02 -1.51038110e+00 -1.78888485e-01 2.99582988e-01
1.03626275e+00 3.69873583e-01 -4.60760415e-01 -6.69110656e-01
3.98992598e-01 -9.48847413e-01 6.77155793e-01 6.62775099e-01
9.97120023e-01 -5.71096241e-01 7.09557712e-01 2.51683950e-01
-1.66725469e+00 8.41983855e-01 -4.71097082e-01 -5.88601828e-02
4.89799857e-01 7.65564442e-01 -5.39848268e-01 6.34152353e-01
7.79799938e-01 5.04060924e-01 -3.63646269e-01 1.04587209e+00
6.43452480e-02 -1.14534609e-01 -1.00043917e+00 5.09702861e-01
-2.30467424e-01 -5.20565093e-01 8.70756686e-01 9.91709888e-01
4.62614357e-01 2.55998075e-01 5.13125598e-01 6.78263903e-01
1.37142777e-01 2.96839893e-01 -5.96861959e-01 8.08831990e-01
8.39248359e-01 1.10666335e+00 -7.48365402e-01 -6.66851029e-02
-4.05934244e-01 1.15163982e+00 3.05343360e-01 2.73451488e-02
-3.58709425e-01 -6.13053665e-02 9.35150623e-01 5.89730680e-01
6.25815913e-02 -6.44988298e-01 -7.90696084e-01 -6.76685095e-01
1.84562981e-01 -5.83012879e-01 1.79841623e-01 -3.50601941e-01
-1.17408836e+00 5.38286209e-01 -1.15296580e-01 -1.14227235e+00
9.45039168e-02 -3.74743491e-01 -6.47282779e-01 9.75337267e-01
-1.27675462e+00 -1.16476822e+00 -7.26841033e-01 5.43826759e-01
5.18981740e-02 3.47874522e-01 1.01325607e+00 1.48604929e-01
1.15484469e-01 3.63432795e-01 -1.21297471e-01 -2.66145259e-01
7.33267367e-01 -9.87982810e-01 8.60713720e-01 7.33272076e-01
-5.11053056e-02 9.72228527e-01 6.85764432e-01 -1.17361462e+00
-1.24769974e+00 -7.03223705e-01 9.52643216e-01 -6.57021046e-01
3.78788948e-01 -4.39257711e-01 -1.16858160e+00 7.37388015e-01
-3.62134799e-02 1.39285401e-01 3.09884101e-01 2.41277695e-01
4.55818698e-02 8.95520747e-02 -1.28944218e+00 2.08711594e-01
1.21971536e+00 -3.83130312e-01 -6.28951669e-01 4.42861617e-01
4.46078748e-01 -1.09805441e+00 -1.12046349e+00 6.10350668e-01
4.41577137e-01 -9.02114570e-01 1.32908165e+00 1.45976037e-01
-2.49689445e-01 -6.29370749e-01 -2.58224547e-01 -6.05527461e-01
-3.50388199e-01 -1.15119433e+00 1.04074337e-01 9.86443043e-01
3.55800778e-01 -8.04591179e-01 9.45134699e-01 7.03303158e-01
-3.52546751e-01 -5.95130622e-01 -1.50129330e+00 -7.02531040e-01
-2.63958484e-01 -2.89292842e-01 7.64744043e-01 9.72759962e-01
-2.12073758e-01 -2.55010962e-01 2.02045694e-01 8.16147566e-01
1.09924817e+00 1.98794886e-01 1.35064256e+00 -1.76995385e+00
2.03987420e-01 -3.37692916e-01 -5.74163854e-01 -1.14329958e+00
-8.68560374e-02 -6.81110978e-01 3.07930589e-01 -1.63143992e+00
-1.16511911e-01 -9.54892278e-01 2.29613930e-01 4.83626574e-01
-1.37506306e-01 6.10838354e-01 -5.23522556e-01 6.42843783e-01
-4.11775798e-01 3.83390963e-01 1.05156982e+00 3.63188326e-01
-3.90395522e-01 -1.42852604e-01 -1.12625659e-01 9.42133904e-01
4.85416830e-01 -4.94519114e-01 3.35348248e-01 -3.98227364e-01
1.95074975e-01 -4.24152434e-01 2.54261613e-01 -1.01486218e+00
5.30283272e-01 -1.94852233e-01 -1.77277196e-02 -7.41500914e-01
4.72084790e-01 -9.45203304e-01 8.02734554e-01 2.49173641e-01
3.95320654e-01 2.55979300e-01 1.93084896e-01 3.56284559e-01
-1.24678761e-01 5.35112247e-03 1.04968572e+00 -3.58896814e-02
-2.30908364e-01 4.58955288e-01 2.57218152e-01 -3.85461926e-01
1.03501797e+00 -5.81928492e-01 2.10568070e-01 3.39448126e-03
-7.17545867e-01 2.94260327e-02 1.37096465e+00 1.84411779e-02
4.81035262e-01 -1.30707812e+00 -8.37825835e-01 4.05538112e-01
-3.17539833e-02 7.46455729e-01 -2.18854010e-01 8.87487292e-01
-1.01570749e+00 6.29642680e-02 1.89286113e-01 -1.17023563e+00
-1.52657807e+00 -7.06943870e-02 6.13920748e-01 -8.34766030e-02
-1.01055956e+00 8.64839017e-01 1.67002618e-01 -8.26789558e-01
-1.63614556e-01 -2.63806522e-01 3.39444637e-01 -3.91914278e-01
3.72320056e-01 6.14339590e-01 5.20871818e-01 -8.65551710e-01
-7.91701257e-01 1.43581200e+00 2.27438107e-01 -1.58592850e-01
1.47646248e+00 -1.49430856e-01 -6.24232948e-01 1.58934817e-01
9.55181062e-01 5.79060972e-01 -1.08183706e+00 -1.63671792e-01
8.83020535e-02 -6.43954635e-01 3.11710755e-03 -1.32440984e-01
-6.70210660e-01 4.36706126e-01 3.68670344e-01 -6.14736155e-02
5.83541691e-01 4.23044592e-01 8.40929449e-01 1.91289067e-01
1.01969087e+00 -8.66959155e-01 -2.49709621e-01 6.02163851e-01
1.12062240e+00 -9.53549027e-01 6.20789587e-01 -1.09768105e+00
1.97975114e-01 1.06313550e+00 3.31605732e-01 -5.95679879e-01
7.80785978e-01 7.70033598e-01 1.33889228e-01 -7.02906430e-01
-1.40083944e-02 9.92229655e-02 5.04492581e-01 5.75021923e-01
3.29801440e-01 -2.82343209e-01 -2.57541418e-01 -7.63196796e-02
-3.29408497e-01 -4.28529680e-01 -3.43787335e-02 1.00637007e+00
-5.89205444e-01 -1.37832844e+00 -8.54187489e-01 1.09594181e-01
-9.27725732e-02 1.22383691e-01 -3.61069292e-01 8.45503628e-01
-1.92440867e-01 5.85828960e-01 5.03606796e-01 -9.06223878e-02
6.06005013e-01 -2.14042798e-01 4.73115027e-01 -7.77640641e-01
-7.05302417e-01 3.90476406e-01 9.34138671e-02 -1.24788845e+00
-5.59342623e-01 -1.06990409e+00 -1.66453969e+00 -2.15319648e-01
-5.54960966e-01 1.61301252e-02 5.73033929e-01 8.85276675e-01
7.08061397e-01 -2.39229038e-01 4.89978373e-01 -1.57165670e+00
-5.27874492e-02 -6.11367762e-01 -3.35827440e-01 3.45841318e-01
2.15397477e-01 -9.04683650e-01 -4.13628280e-01 8.60022157e-02]
|
[7.735663414001465, -2.9240684509277344]
|
fd46d1cf-75d7-40ea-9ed2-6ca004a75063
|
integral-probability-metrics-pac-bayes-bounds
|
2207.00614
| null |
https://arxiv.org/abs/2207.00614v8
|
https://arxiv.org/pdf/2207.00614v8.pdf
|
Integral Probability Metrics PAC-Bayes Bounds
|
We present a PAC-Bayes-style generalization bound which enables the replacement of the KL-divergence with a variety of Integral Probability Metrics (IPM). We provide instances of this bound with the IPM being the total variation metric and the Wasserstein distance. A notable feature of the obtained bounds is that they naturally interpolate between classical uniform convergence bounds in the worst case (when the prior and posterior are far away from each other), and improved bounds in favorable cases (when the posterior and prior are close). This illustrates the possibility of reinforcing classical generalization bounds with algorithm- and data-dependent components, thus making them more suitable to analyze algorithms that use a large hypothesis space.
|
['Ron Meir', 'Shay Moran', 'Baruch Epstein', 'Ron Amit']
|
2022-07-01
| null | null | null | null |
['style-generalization']
|
['computer-vision']
|
[ 1.56921238e-01 -5.19637503e-02 -3.10151517e-01 -1.84108868e-01
-7.33654261e-01 -8.17651868e-01 6.29724085e-01 4.07674521e-01
-6.36412382e-01 9.31383312e-01 -2.85661697e-01 -2.72312850e-01
-4.93579000e-01 -7.58571148e-01 -5.11836886e-01 -1.00004494e+00
-4.43852127e-01 6.20911062e-01 6.78941369e-01 -1.08480595e-01
3.49621326e-01 7.14857817e-01 -1.69577920e+00 -5.11121452e-01
7.03966320e-01 1.04071331e+00 -2.02409908e-01 8.89577508e-01
8.53276253e-02 -2.18620241e-01 -4.40875739e-01 -6.39229953e-01
4.96489704e-01 -3.77098352e-01 -6.70635343e-01 -1.99371070e-01
2.52803653e-01 4.95097749e-02 6.55487180e-02 1.34627593e+00
2.44432300e-01 3.25796813e-01 1.03896332e+00 -1.44286692e+00
-4.76953506e-01 2.65105695e-01 -6.72580302e-01 4.08077717e-01
4.46105093e-01 -4.46767390e-01 1.13812304e+00 -3.21457118e-01
4.48982984e-01 1.17732775e+00 9.20268476e-01 3.59527647e-01
-1.53644097e+00 -1.59767449e-01 -1.23618007e-01 1.85294643e-01
-1.47441137e+00 8.97033811e-02 1.46467194e-01 -6.41902328e-01
4.22822058e-01 4.62623864e-01 3.54000032e-01 7.56097794e-01
3.94011676e-01 6.83250427e-01 1.11406636e+00 -6.56068623e-01
5.13182878e-01 2.94031590e-01 3.04042071e-01 5.70629001e-01
8.03192139e-01 2.27372169e-01 1.70222316e-02 -5.01952708e-01
5.99642754e-01 -4.71135303e-02 -5.95871925e-01 -1.00808966e+00
-8.99250627e-01 9.63194788e-01 -8.63453671e-02 3.72902185e-01
-4.23698090e-02 4.13478725e-03 2.03043520e-01 2.39276305e-01
4.12246585e-01 3.41361225e-01 -3.62007976e-01 -3.44918489e-01
-8.27000797e-01 4.53615040e-01 1.14981675e+00 1.11019337e+00
6.44383073e-01 -4.50684190e-01 -1.39039710e-01 5.92657745e-01
2.79349506e-01 4.34566289e-01 3.02978635e-01 -7.44745374e-01
8.04979727e-02 3.76382493e-03 5.28690040e-01 -7.37926066e-01
-2.45446905e-01 -4.78050113e-01 -3.73646021e-01 1.99587986e-01
9.14754093e-01 1.04844652e-01 -4.63430494e-01 1.86921132e+00
3.77408445e-01 9.38712209e-02 -6.84619546e-02 6.55162334e-01
-3.08631390e-01 3.68999928e-01 -2.21198186e-01 -5.55739999e-01
9.28458154e-01 -5.67414880e-01 -7.44349360e-01 3.20954561e-01
6.04350209e-01 -6.33424520e-01 1.13007796e+00 6.29784465e-01
-9.71698105e-01 -1.71150073e-01 -1.32115066e+00 2.49068856e-01
-4.76538837e-01 -5.73756933e-01 3.98884028e-01 1.03187311e+00
-8.78579021e-01 1.04762018e+00 -7.77357161e-01 -3.58644724e-01
-1.09955527e-01 4.02935930e-02 -6.57743663e-02 -3.11207809e-02
-8.70354712e-01 8.92221153e-01 5.72175086e-01 -1.31561965e-01
-7.14390457e-01 -5.84950387e-01 -4.44516599e-01 5.02659827e-02
1.51160896e-01 -2.01254800e-01 1.16515207e+00 -4.32714790e-01
-1.36451995e+00 6.70970023e-01 2.98269272e-01 -5.45451760e-01
9.40462530e-01 -1.50886208e-01 -3.40175390e-01 7.21594766e-02
-1.66206911e-01 -6.83667185e-03 6.51959300e-01 -1.02222824e+00
-6.11184359e-01 -4.85758185e-01 6.99944496e-02 1.60399899e-02
-1.60026655e-01 -2.82652587e-01 -2.29827911e-01 -3.60828370e-01
1.17411494e-01 -8.31258774e-01 -1.58720776e-01 1.13541260e-01
-2.37928405e-01 -4.62593913e-01 5.36602557e-01 -2.51269788e-01
1.22395921e+00 -2.18532085e+00 2.01613560e-01 4.16520029e-01
-3.65871303e-02 3.34455036e-02 2.02318221e-01 5.53737700e-01
-1.41968042e-01 2.14850619e-01 -3.27151030e-01 -1.22418463e-01
3.80959690e-01 4.77514803e-01 -1.71817809e-01 9.89537716e-01
-2.08670944e-01 8.71226862e-02 -1.07491267e+00 -4.09555197e-01
2.00637355e-02 1.51522607e-01 -3.62891823e-01 -5.97329461e-04
-1.40965030e-01 7.48624429e-02 -2.12834805e-01 -7.41187632e-02
8.68155420e-01 7.00411722e-02 -8.06534588e-02 3.01315278e-01
-1.50610670e-01 1.49581656e-02 -1.67132223e+00 1.40052092e+00
-2.01149464e-01 5.55863082e-01 1.38318375e-01 -1.06931615e+00
8.42187345e-01 1.77227020e-01 4.72869873e-01 4.00883183e-02
1.96199745e-01 2.36705154e-01 -2.01526836e-01 -2.62094170e-01
2.02110022e-01 -7.06482291e-01 1.38259634e-01 2.67112523e-01
8.58440250e-02 -2.70341694e-01 4.41219151e-01 -1.04694858e-01
8.54414999e-01 4.15561460e-02 4.46065217e-01 -7.83529401e-01
5.71026921e-01 -4.62038875e-01 6.23763025e-01 7.51944005e-01
-4.45068508e-01 5.66377461e-01 8.66429746e-01 8.88545364e-02
-1.01763844e+00 -1.55669045e+00 -9.35860157e-01 8.93053412e-01
4.06930238e-01 -3.53897691e-01 -7.24968255e-01 -6.97907269e-01
2.43979096e-01 7.32861638e-01 -8.23546827e-01 -2.00663373e-01
1.33920655e-01 -7.43409634e-01 4.29933399e-01 5.90756297e-01
7.00979978e-02 -4.44811247e-02 -4.78760600e-01 -4.61966209e-02
2.28279114e-01 -8.40635836e-01 -3.90820384e-01 3.88038576e-01
-8.29569519e-01 -1.28610373e+00 -7.84419656e-01 -3.05446684e-01
2.36896768e-01 -4.52634692e-02 8.69788766e-01 -2.38491908e-01
-1.92932025e-01 6.63208187e-01 -2.71472424e-01 -3.90609682e-01
-3.11371356e-01 -4.25232202e-01 3.64870965e-01 -1.23692602e-01
4.31083053e-01 -5.08098602e-01 -1.97389096e-01 5.14218450e-01
-9.45536017e-01 -6.54992878e-01 1.35130867e-01 8.48664045e-01
6.50396228e-01 2.58438259e-01 2.83973515e-01 -6.16361678e-01
4.46385562e-01 -5.10833502e-01 -9.83736455e-01 3.64232302e-01
-8.00119519e-01 2.99787492e-01 4.06774610e-01 -4.84711617e-01
-6.75990343e-01 -3.60780299e-01 1.09970920e-01 -3.23494941e-01
-1.23094149e-01 -5.70850782e-02 -9.49950665e-02 -2.35951636e-02
7.47458398e-01 -1.17462173e-01 -1.71785533e-01 -5.76605976e-01
2.85619259e-01 6.78546011e-01 4.74309236e-01 -8.83823037e-01
6.97427869e-01 3.60944957e-01 3.02236646e-01 -8.01987886e-01
-9.65718329e-01 -6.98615551e-01 -6.79263234e-01 7.27160573e-02
9.12298083e-01 -1.35608405e-01 -5.79761744e-01 -1.83074246e-03
-8.31043959e-01 1.80305522e-02 -5.94419301e-01 8.95609796e-01
-9.75126207e-01 6.33486688e-01 -4.64714587e-01 -1.20842659e+00
3.08996618e-01 -1.07791114e+00 8.58437836e-01 2.24523887e-01
8.28753255e-05 -1.40660501e+00 4.01562810e-01 -1.15923837e-01
6.51826560e-02 4.98473912e-01 8.33493650e-01 -1.15292180e+00
3.89763825e-02 -6.98330641e-01 -2.57203430e-01 7.25408912e-01
-1.06736839e-01 1.98981851e-01 -6.65735304e-01 -5.14825404e-01
1.80572331e-01 2.34168544e-01 6.33213580e-01 3.80324364e-01
1.07934773e+00 -1.42401457e-01 -4.24372286e-01 4.45200562e-01
1.61050057e+00 1.20394699e-01 2.51724511e-01 1.74792409e-01
1.18338108e-01 3.70030820e-01 5.19981980e-01 4.72120821e-01
-1.82813555e-01 6.34903312e-01 3.15641731e-01 4.72547233e-01
2.94121414e-01 9.52391699e-03 2.01702476e-01 5.23733497e-01
-1.56449284e-02 -1.78010866e-01 -7.71669924e-01 5.16000450e-01
-1.87405550e+00 -7.64911473e-01 -4.29560900e-01 2.87055969e+00
9.49286819e-01 3.43461841e-01 3.73493046e-01 2.99004972e-01
1.02144921e+00 -2.48226792e-01 -3.54951859e-01 -7.07388282e-01
-1.88913965e-03 2.16944829e-01 8.00034583e-01 8.21148157e-01
-1.01183403e+00 1.62234053e-01 7.90944338e+00 1.01201630e+00
-3.12085599e-01 2.57403523e-01 3.92219350e-02 -1.41645921e-02
-2.78079748e-01 8.66983756e-02 -8.39316308e-01 6.68550909e-01
1.09872675e+00 -3.37862313e-01 2.64661700e-01 1.02559030e+00
-3.69343102e-01 -3.81233603e-01 -1.47309661e+00 8.08278501e-01
-6.42615631e-02 -6.34990215e-01 -3.21225256e-01 4.17684495e-01
6.95221484e-01 -2.76091695e-01 -1.15903497e-01 1.84470080e-02
4.44044262e-01 -8.33012521e-01 4.55928922e-01 4.88040417e-01
5.71445704e-01 -8.05306077e-01 9.84623671e-01 3.61887038e-01
-8.94666791e-01 6.97474182e-03 -4.96625930e-01 -6.95575178e-02
-6.23873025e-02 9.25765753e-01 -4.38636035e-01 5.30308962e-01
5.96960843e-01 3.26473057e-01 -1.20289482e-01 1.48597717e+00
-5.89397773e-02 3.52095634e-01 -8.55858028e-01 -2.76665062e-01
1.72307715e-01 -6.27664447e-01 9.27155614e-01 1.44638705e+00
2.80802459e-01 -3.29693854e-01 6.91317469e-02 6.62566364e-01
3.04192036e-01 1.02616981e-01 -5.46692431e-01 9.49605480e-02
5.70051134e-01 9.94013369e-01 -8.62353623e-01 -2.47837588e-01
-2.89253414e-01 6.90024495e-01 1.17463626e-01 2.60712504e-01
-9.36392128e-01 -7.74996996e-01 8.41338217e-01 -1.04537137e-01
4.63214904e-01 -3.92035663e-01 -1.86651021e-01 -1.00992632e+00
3.20490748e-01 -1.94403723e-01 6.23828113e-01 -2.27365240e-01
-1.34950984e+00 1.03968583e-01 5.36347866e-01 -1.04754424e+00
1.91427898e-02 -1.12729478e+00 -3.68514299e-01 7.44508564e-01
-1.03517759e+00 -3.62238199e-01 1.82769239e-01 2.56868362e-01
-1.10313967e-01 1.49492711e-01 7.77515233e-01 8.10229480e-02
-4.57437783e-01 6.15272462e-01 7.08613336e-01 -2.42819831e-01
5.33258379e-01 -1.72629702e+00 -5.23250759e-01 8.85194600e-01
5.61630093e-02 4.41076845e-01 1.33635437e+00 -2.33751804e-01
-1.02896535e+00 -4.71307546e-01 4.16125029e-01 -5.88377297e-01
9.76416290e-01 -2.61039466e-01 -8.56730044e-01 5.56627691e-01
-1.67372346e-01 -1.11813165e-01 8.67661297e-01 4.33090389e-01
-5.49045801e-01 -1.23040318e-01 -1.46093047e+00 4.49065775e-01
8.92186522e-01 -3.17251265e-01 -6.66687131e-01 6.09182119e-01
3.08637381e-01 -1.04409240e-01 -1.26265812e+00 3.15967202e-01
5.26497185e-01 -1.22672832e+00 6.03938043e-01 -6.86839402e-01
-2.68650740e-01 -1.94445625e-01 -5.79522133e-01 -1.17173803e+00
-6.10566251e-02 -7.91199148e-01 -1.43690094e-01 1.09572268e+00
3.25865507e-01 -8.59748483e-01 3.01893204e-01 5.79681933e-01
6.25484297e-03 -9.90107775e-01 -1.29491091e+00 -1.43883514e+00
4.39663768e-01 -5.60592294e-01 4.06098485e-01 7.72379041e-01
3.33067268e-01 -1.84449956e-01 1.64559186e-01 1.08341880e-01
7.09792197e-01 -1.41350716e-01 2.79479146e-01 -1.58798826e+00
-5.81573665e-01 -9.44892704e-01 -9.33237553e-01 -1.15983927e+00
-7.41448775e-02 -7.27136493e-01 2.40078926e-01 -1.03803813e+00
1.44648612e-01 -3.80186468e-01 -3.55103433e-01 1.35465525e-02
7.37683102e-02 -2.58174930e-02 -1.90819278e-01 1.71262138e-02
-4.09645915e-01 5.26017249e-01 1.01206064e+00 4.00019050e-01
-7.79299811e-02 2.49080464e-01 -4.34625566e-01 9.31325555e-01
5.75998306e-01 -4.55640435e-01 -3.02180856e-01 9.42600220e-02
2.57551461e-01 -2.20368564e-01 1.36215180e-01 -1.07409298e+00
-7.68144205e-02 -2.59243160e-01 1.24294490e-01 -3.88753653e-01
1.20724685e-01 -8.24701190e-01 -1.74057290e-01 3.72122496e-01
-4.58335698e-01 -3.42745781e-02 2.86937747e-02 1.04470408e+00
6.63736984e-02 -8.00373554e-01 1.17749453e+00 2.92475581e-01
-1.85540199e-01 2.76713461e-01 -2.39012763e-01 1.93011373e-01
1.31294012e+00 -3.58886063e-01 -6.26382604e-02 -2.82675385e-01
-7.55774200e-01 6.48559630e-02 5.85766792e-01 6.29374161e-02
1.49105430e-01 -1.40571296e+00 -4.81019497e-01 3.09369876e-03
1.24220431e-01 -1.20002858e-01 -1.56030253e-01 1.21418929e+00
-5.34003913e-01 2.69076884e-01 8.00100341e-02 -6.37465417e-01
-8.62734079e-01 8.98501039e-01 4.72366542e-01 -3.16538438e-02
-5.66279292e-01 9.13156986e-01 -2.13403497e-02 -7.27704316e-02
5.55689871e-01 -3.89163405e-01 3.80611598e-01 -5.71702346e-02
7.32560217e-01 8.53360593e-01 -1.20772701e-02 -2.37686649e-01
-3.86962324e-01 4.48628098e-01 2.13032573e-01 -2.85867870e-01
1.08227122e+00 -1.49628699e-01 -7.27300644e-02 9.38030124e-01
1.39381599e+00 8.50994885e-02 -1.20180047e+00 -2.24355638e-01
2.96374321e-01 -5.61230779e-01 -2.73003001e-02 -4.30386245e-01
-4.94054615e-01 9.12304938e-01 6.53956652e-01 9.61112797e-01
9.33848977e-01 6.38842881e-02 3.55025977e-01 2.96571910e-01
3.25334966e-01 -1.26194906e+00 -2.82051384e-01 4.01151270e-01
8.01316381e-01 -6.87483191e-01 1.67879507e-01 -1.77634865e-01
-2.70930499e-01 1.33918548e+00 1.98230475e-01 -4.16508645e-01
9.65176344e-01 3.21936727e-01 -3.48885566e-01 3.23294133e-01
-4.17956829e-01 -2.94306993e-01 4.98875082e-01 7.35321939e-01
4.09609944e-01 1.71611845e-01 -7.48666346e-01 4.13196683e-01
-3.03438127e-01 -4.72134590e-01 4.39795345e-01 7.86780417e-01
-6.15825355e-01 -1.01294851e+00 -2.69459277e-01 2.48933807e-01
-4.37472403e-01 3.56382787e-01 -1.04853965e-01 1.01062143e+00
-7.62354434e-02 8.63302886e-01 7.30527863e-02 -4.51874770e-02
1.72673225e-01 3.97993833e-01 7.70278573e-01 -4.69514489e-01
3.05481464e-01 2.49757227e-02 -6.13777945e-03 -5.89365542e-01
-3.66770893e-01 -8.01761210e-01 -9.93974626e-01 -4.47496653e-01
-7.93723047e-01 5.10079861e-01 7.18248904e-01 1.08065224e+00
-7.55806565e-02 -1.20206900e-01 5.58471143e-01 -6.65055156e-01
-1.14678967e+00 -9.87479866e-01 -1.20719516e+00 3.14973831e-01
3.91568422e-01 -9.51662123e-01 -8.31882834e-01 -1.93097323e-01]
|
[7.352199554443359, 4.131488800048828]
|
734371ad-7713-44b0-9911-cad86a5d831e
|
extract-and-attend-improving-entity
|
2306.02242
| null |
https://arxiv.org/abs/2306.02242v1
|
https://arxiv.org/pdf/2306.02242v1.pdf
|
Extract and Attend: Improving Entity Translation in Neural Machine Translation
|
While Neural Machine Translation(NMT) has achieved great progress in recent years, it still suffers from inaccurate translation of entities (e.g., person/organization name, location), due to the lack of entity training instances. When we humans encounter an unknown entity during translation, we usually first look up in a dictionary and then organize the entity translation together with the translations of other parts to form a smooth target sentence. Inspired by this translation process, we propose an Extract-and-Attend approach to enhance entity translation in NMT, where the translation candidates of source entities are first extracted from a dictionary and then attended to by the NMT model to generate the target sentence. Specifically, the translation candidates are extracted by first detecting the entities in a source sentence and then translating the entities through looking up in a dictionary. Then, the extracted candidates are added as a prefix of the decoder input to be attended to by the decoder when generating the target sentence through self-attention. Experiments conducted on En-Zh and En-Ru demonstrate that the proposed method is effective on improving both the translation accuracy of entities and the overall translation quality, with up to 35% reduction on entity error rate and 0.85 gain on BLEU and 13.8 gain on COMET.
|
['Tie-Yan Liu', 'Tao Qin', 'Xu Tan', 'Junliang Guo', 'Yichong Leng', 'Rui Wang', 'Zixin Zeng']
|
2023-06-04
| null | null | null | null |
['nmt']
|
['computer-code']
|
[ 3.93116891e-01 1.07564680e-01 -1.96318537e-01 -3.22283566e-01
-1.11259544e+00 -5.40341198e-01 4.51630145e-01 2.35083282e-01
-5.30326247e-01 9.96535480e-01 3.31625283e-01 -2.69207776e-01
4.42212820e-01 -8.70967865e-01 -9.06450510e-01 -3.19654107e-01
5.75909913e-01 6.14851058e-01 -1.64295226e-01 -2.87178993e-01
-2.58720554e-02 -1.65735692e-01 -7.61097610e-01 1.83056727e-01
1.45588017e+00 6.77224100e-01 3.98453772e-01 -1.81577858e-02
-4.65326458e-01 3.86169404e-01 -6.42790616e-01 -8.84968579e-01
1.84428558e-01 -8.25143039e-01 -6.90598786e-01 -3.45874503e-02
4.99529317e-02 -3.05516928e-01 -2.32278615e-01 1.25505924e+00
6.98534310e-01 -6.69725379e-03 4.79988575e-01 -7.98483968e-01
-1.24210227e+00 1.02749491e+00 -6.16311729e-01 1.53225496e-01
3.90241504e-01 -1.57445699e-01 9.97361183e-01 -1.75173151e+00
8.85523975e-01 1.03624392e+00 3.03423762e-01 6.33675396e-01
-9.27382708e-01 -7.06671715e-01 -6.16501383e-02 1.02556817e-01
-1.41452682e+00 -6.59095824e-01 3.77179414e-01 4.05625962e-02
1.11202800e+00 2.58813232e-01 3.14728200e-01 9.80497658e-01
9.40360129e-02 9.07273591e-01 5.76358795e-01 -5.54432333e-01
-6.85804412e-02 2.35163048e-01 -7.15775713e-02 5.15641809e-01
2.86732942e-01 -2.20619857e-01 -4.79086250e-01 1.76405802e-01
6.88363850e-01 -1.16099589e-01 -1.24880478e-01 3.42752606e-01
-1.77075434e+00 5.21646976e-01 5.52867115e-01 3.16911459e-01
-9.53936517e-01 -3.13716054e-01 2.62346685e-01 4.11460936e-01
5.20592213e-01 4.82080996e-01 -3.84801418e-01 1.21466862e-02
-1.00638652e+00 9.96417850e-02 5.12168467e-01 1.49081051e+00
8.22670043e-01 -2.33045354e-01 -4.95990038e-01 8.47752273e-01
1.29153982e-01 9.34434354e-01 5.07167161e-01 -2.06076652e-01
1.33776224e+00 9.26728487e-01 3.16052675e-01 -9.73010361e-01
4.85998439e-03 -7.58018076e-01 -9.16636407e-01 -8.01794171e-01
6.50881417e-03 -6.58056200e-01 -8.41480076e-01 1.76400173e+00
3.00365925e-01 -2.08148852e-01 3.80106926e-01 9.89365518e-01
8.34006488e-01 1.04185474e+00 6.02023713e-02 -4.14975137e-01
1.45378053e+00 -1.27747130e+00 -9.19224858e-01 -6.60286844e-01
6.87521875e-01 -1.11127138e+00 7.17394054e-01 -3.74267280e-01
-1.23643053e+00 -6.54864311e-01 -7.95224547e-01 -1.99209183e-01
-1.97644293e-01 7.30244458e-01 4.34480458e-02 -1.82332955e-02
-5.69384575e-01 3.77258062e-01 -7.52484262e-01 -4.37437832e-01
1.62488222e-01 3.63917679e-01 -3.14360887e-01 -1.60185575e-01
-1.51588082e+00 1.06701088e+00 4.95931983e-01 3.57254833e-01
-4.23920095e-01 -3.75262082e-01 -8.35635781e-01 1.48464486e-01
2.24795654e-01 -1.04745924e+00 1.15581465e+00 -1.20293987e+00
-1.24950695e+00 4.95836735e-01 -6.66170835e-01 -3.73322248e-01
4.36864525e-01 -3.02076936e-01 -7.17047572e-01 -1.04843467e-01
5.03563702e-01 7.29602754e-01 4.63606298e-01 -6.55237556e-01
-9.31856334e-01 -1.97329193e-01 -5.93477003e-02 7.33585298e-01
-4.03508395e-01 3.23171020e-01 -7.64209569e-01 -5.79394579e-01
3.54653746e-01 -9.97792423e-01 -8.73611718e-02 -5.82283437e-01
-5.97374201e-01 -1.46818489e-01 4.45649326e-01 -1.16831410e+00
1.54562581e+00 -1.92249393e+00 4.38686877e-01 -8.37610960e-02
-8.74719694e-02 2.03490376e-01 -2.78002441e-01 6.53799236e-01
8.23138654e-02 1.40981689e-01 -4.20431107e-01 -2.61277914e-01
-1.29188925e-01 -1.75602525e-01 -3.70081931e-01 1.20676113e-02
5.04352927e-01 1.25669849e+00 -1.02858388e+00 -5.39058208e-01
-2.52623260e-01 3.73208612e-01 -2.52078116e-01 1.92020893e-01
-4.31087129e-02 4.83535260e-01 -7.13740885e-01 6.35867715e-01
4.83063728e-01 -3.33107501e-01 6.00048490e-02 -6.77622184e-02
-5.19525222e-02 8.39972854e-01 -9.30194318e-01 1.54291987e+00
-5.49643159e-01 5.68086743e-01 -3.60524356e-01 -4.80465531e-01
1.00311720e+00 6.18435562e-01 6.89845979e-02 -8.38986278e-01
1.15724087e-01 6.50270224e-01 2.02784464e-01 -4.68060255e-01
7.29374766e-01 -8.60596448e-02 -1.81773677e-01 4.54472870e-01
5.76518923e-02 4.07723159e-01 3.78204823e-01 7.33923614e-02
7.87886322e-01 1.39711559e-01 2.38294393e-01 1.04872666e-01
6.06873870e-01 3.81252021e-01 7.25227177e-01 1.83976233e-01
2.93144763e-01 4.27669764e-01 -5.96429333e-02 -3.10565233e-01
-1.29267251e+00 -7.52507806e-01 2.92437613e-01 8.13333988e-01
3.27877998e-01 -3.80721152e-01 -9.69097912e-01 -7.38358796e-01
-3.98608685e-01 9.17135060e-01 -3.54090065e-01 -3.49608541e-01
-9.81988966e-01 -6.80741608e-01 3.91355097e-01 5.34719169e-01
7.17007279e-01 -1.17313135e+00 -1.75397903e-01 6.35468662e-01
-9.78934586e-01 -1.11375391e+00 -8.97837698e-01 -1.27337158e-01
-9.55370128e-01 -3.12784076e-01 -9.29410577e-01 -1.12604403e+00
1.13013196e+00 3.00412267e-01 1.05573082e+00 1.84766501e-02
4.66973990e-01 -5.60756505e-01 -3.68602812e-01 -2.83611119e-01
-5.69995105e-01 5.10171592e-01 2.04147130e-01 2.47283891e-01
6.69565797e-01 -3.38889480e-01 -5.21732688e-01 3.51385236e-01
-5.58257341e-01 4.17161286e-01 1.06032181e+00 8.38822544e-01
8.05478573e-01 -3.00235510e-01 7.89434791e-01 -7.30278194e-01
7.58857906e-01 -6.12995267e-01 -3.72704595e-01 4.25612777e-01
-5.26337028e-01 -2.12040991e-02 7.98035562e-01 -5.74815273e-01
-9.77136433e-01 1.81211419e-02 -5.15171736e-02 -1.81691214e-01
2.94189692e-01 8.98937404e-01 -3.13425988e-01 4.49448943e-01
6.84919715e-01 6.85344398e-01 -5.17792165e-01 -3.44150096e-01
1.48940250e-01 9.88878906e-01 4.45440114e-01 -3.64643663e-01
1.07838798e+00 -2.73617685e-01 -5.70316136e-01 -3.04247111e-01
-6.99985743e-01 -1.38960004e-01 -6.80627525e-01 8.27994421e-02
7.63633192e-01 -1.16046238e+00 1.99844390e-02 9.25987884e-02
-1.46410441e+00 3.12791348e-01 -3.51590328e-02 6.97430551e-01
-1.79087892e-01 3.33674997e-02 -6.82860494e-01 -4.51917648e-01
-9.56700981e-01 -1.17885029e+00 1.06315756e+00 5.37973225e-01
-2.87153721e-01 -6.91507697e-01 -2.58310050e-01 3.43956143e-01
4.96553957e-01 -2.11967275e-01 9.71805096e-01 -8.70292783e-01
-7.40604401e-01 -4.21848387e-01 -2.71584362e-01 5.40038832e-02
3.24657798e-01 -3.49887162e-01 -3.45135868e-01 -1.91644207e-01
-1.15189955e-01 1.31564945e-01 3.71212840e-01 -9.89796519e-02
2.11278155e-01 -5.64433277e-01 -4.90941852e-01 3.54120880e-01
1.30850136e+00 2.88842350e-01 4.12235469e-01 2.51079232e-01
6.85405314e-01 4.08410966e-01 8.74241948e-01 -2.34210473e-02
5.34720838e-01 6.56664908e-01 -5.55556407e-03 -1.31493136e-01
-2.38863692e-01 -7.18999743e-01 5.69391668e-01 1.47363424e+00
-1.22127473e-01 -3.67133081e-01 -7.61464059e-01 8.02831948e-01
-1.81420732e+00 -7.06960082e-01 -4.39814031e-02 2.15028477e+00
1.08926404e+00 6.31246641e-02 -2.52559930e-01 -4.48226035e-01
1.13664317e+00 -2.98626065e-01 -6.51206613e-01 -4.48649339e-02
-2.34119162e-01 9.66545939e-03 4.45033759e-01 4.03186142e-01
-6.84267163e-01 1.33752346e+00 5.05846405e+00 7.27326810e-01
-1.20986736e+00 2.14900583e-01 4.57818598e-01 4.44059595e-02
-2.77292699e-01 1.86121643e-01 -1.17424691e+00 5.69922030e-01
1.05937481e+00 -6.72217369e-01 5.16590416e-01 5.40770292e-01
2.29443222e-01 3.20966005e-01 -1.12317300e+00 8.46999645e-01
1.22964747e-01 -1.16883445e+00 3.90617311e-01 1.65510876e-03
8.13975692e-01 -2.06414144e-02 -1.36904046e-01 5.15776694e-01
-7.55819902e-02 -7.09047556e-01 8.65594566e-01 3.37489814e-01
8.06434989e-01 -8.28120589e-01 8.91056418e-01 7.16464818e-01
-1.11671090e+00 2.00006843e-01 -7.08848059e-01 5.55017069e-02
2.68597841e-01 4.68036294e-01 -1.05769217e+00 8.88970017e-01
1.80177450e-01 6.39094710e-01 -2.19141871e-01 8.66335690e-01
-6.55341268e-01 6.35689080e-01 -2.81117886e-01 -2.65170217e-01
3.62310201e-01 -3.84803861e-01 6.35924816e-01 1.14935303e+00
8.97822320e-01 -1.03386780e-02 5.82566783e-02 1.01864147e+00
-7.39803910e-01 6.80639148e-01 -3.73695225e-01 -2.32881710e-01
7.62089252e-01 1.25822556e+00 -4.97075379e-01 -6.75873876e-01
-4.72262204e-01 1.49102533e+00 4.00049746e-01 5.06394684e-01
-9.76797879e-01 -6.81879878e-01 4.03205395e-01 -7.94403702e-02
3.76959532e-01 1.80911236e-02 -2.28472605e-01 -1.40702426e+00
5.48499465e-01 -1.03067601e+00 -5.76408319e-02 -7.62240410e-01
-9.57967401e-01 1.07723129e+00 -5.02793133e-01 -1.63923466e+00
-2.95680255e-01 1.16306864e-01 -3.69763583e-01 1.44153333e+00
-1.27024376e+00 -1.14337540e+00 1.72102898e-01 2.46678188e-01
8.56720567e-01 -9.71420556e-02 8.23003352e-01 6.47788525e-01
-8.03733468e-01 8.10408056e-01 1.16592593e-01 5.83603799e-01
8.52266431e-01 -8.69608462e-01 1.05982494e+00 1.22558403e+00
3.33988458e-01 1.11283064e+00 4.99871045e-01 -1.02741003e+00
-1.35426617e+00 -1.43931174e+00 1.97034049e+00 -4.74599063e-01
3.03129554e-01 -3.41563731e-01 -8.30094516e-01 7.97478676e-01
5.27202129e-01 -3.18983793e-01 4.05050069e-01 -2.80138373e-01
-1.37780011e-01 1.10111728e-01 -8.86412323e-01 8.93633068e-01
9.65732098e-01 -5.74316561e-01 -9.39073443e-01 4.07355160e-01
1.00581527e+00 -8.27901363e-01 -7.48512030e-01 1.95665374e-01
2.71758735e-01 -2.45288457e-03 6.33850932e-01 -6.55154467e-01
8.13524067e-01 -5.40120661e-01 -4.87890430e-02 -1.54101145e+00
-3.65987360e-01 -6.26824975e-01 -4.12827693e-02 1.45826995e+00
1.29342723e+00 -3.34711134e-01 3.34122002e-01 6.05901837e-01
-2.20422015e-01 -8.19132268e-01 -7.68413782e-01 -5.66016555e-01
-9.78088677e-02 1.80566922e-01 9.07319427e-01 9.83469903e-01
6.98255599e-02 1.14087570e+00 -4.76388931e-01 3.17711085e-01
2.75137693e-01 4.51059669e-01 5.37785649e-01 -6.66917086e-01
-4.24183197e-02 -1.42344460e-01 2.23171860e-01 -1.51300859e+00
-1.32837698e-01 -1.15104377e+00 2.20160201e-01 -1.93746185e+00
3.30194235e-01 -1.70434862e-01 -1.67966187e-01 5.56504428e-01
-6.92344010e-01 1.60706997e-01 9.15230364e-02 5.20531654e-01
-4.49757963e-01 5.39574385e-01 1.37973726e+00 -2.19765052e-01
-3.02005291e-01 2.24151034e-02 -8.62383008e-01 3.47240150e-01
5.83938837e-01 -7.98436403e-01 -4.50835600e-02 -1.08450663e+00
3.58779907e-01 3.29398245e-01 -2.72991300e-01 -8.12125266e-01
4.02860105e-01 -4.95387055e-02 5.25875688e-01 -6.79282486e-01
7.08265901e-02 -8.37423086e-01 1.51917160e-01 3.02163273e-01
-4.63900149e-01 5.07062435e-01 5.01800030e-02 2.63458818e-01
-4.09965068e-01 -1.70343280e-01 3.44824106e-01 -1.20748930e-01
-3.17485243e-01 3.22429299e-01 -6.76048547e-02 -8.18179697e-02
6.78845227e-01 -6.11022934e-02 -2.39580467e-01 -2.44989723e-01
-5.20176053e-01 3.45947146e-01 3.42711151e-01 6.02707088e-01
4.70314771e-01 -1.69056880e+00 -1.12474453e+00 1.95679814e-01
1.56899408e-01 -5.65170646e-02 4.39487100e-02 1.01925051e+00
-1.96656555e-01 5.04572272e-01 -1.84182093e-01 -2.68350750e-01
-1.07445717e+00 5.14165163e-01 1.48033112e-01 -3.16763908e-01
-4.69534636e-01 7.53980935e-01 -1.89180709e-02 -4.66724873e-01
-1.20094426e-01 -3.19787294e-01 -7.88307339e-02 3.23266201e-02
5.83977640e-01 -3.42626944e-02 1.84828579e-01 -9.40923929e-01
-3.78076077e-01 4.73513424e-01 -2.63402373e-01 -1.99836046e-01
1.23568201e+00 -4.65006620e-01 -2.90373504e-01 1.05381519e-01
9.94036078e-01 2.99335301e-01 -6.11718655e-01 -7.09477305e-01
1.02888934e-01 -1.60269052e-01 -3.45325112e-01 -1.10416400e+00
-9.06894863e-01 8.50253820e-01 1.91076025e-01 -1.78550839e-01
1.13363421e+00 -1.34389564e-01 1.50720918e+00 5.45338929e-01
4.33262378e-01 -8.86866570e-01 -4.95692015e-01 6.94076180e-01
7.58056343e-01 -1.18816376e+00 -4.18694764e-01 -3.99778247e-01
-6.70743942e-01 9.94276226e-01 5.65800369e-01 1.93646133e-01
-2.63546892e-02 -7.65152127e-02 1.54371008e-01 2.41836905e-01
-6.68649793e-01 -6.78103492e-02 3.78831804e-01 1.12003088e-01
6.29782736e-01 1.10655285e-01 -5.77215016e-01 8.04860413e-01
-3.18228334e-01 2.03853771e-02 2.45627850e-01 6.44778728e-01
-4.31993186e-01 -1.14068413e+00 -4.39733192e-02 2.43813276e-01
-6.37298524e-01 -6.26838624e-01 -6.27555609e-01 3.34431201e-01
1.20284960e-01 9.28028703e-01 -1.55183464e-01 -5.47954559e-01
6.37568355e-01 3.82348269e-01 1.81209251e-01 -8.50761235e-01
-8.35772693e-01 1.89814523e-01 3.02542418e-01 -1.04509324e-01
-1.64426744e-01 -6.17298543e-01 -1.42803657e+00 -4.56114225e-02
-5.52842796e-01 4.89065945e-01 6.45299613e-01 1.01009572e+00
7.47689843e-01 4.39225733e-01 6.86658621e-01 -2.43378952e-01
-4.31515038e-01 -1.33551359e+00 2.00603500e-01 3.74172628e-01
1.39664873e-01 -1.22149482e-01 -8.14776495e-03 1.82947189e-01]
|
[11.661026000976562, 10.266135215759277]
|
9938e3f4-9429-41a7-a607-0eac71b6af62
|
a-hierarchical-subspace-model-for-language
|
2011.03115
| null |
https://arxiv.org/abs/2011.03115v2
|
https://arxiv.org/pdf/2011.03115v2.pdf
|
A Hierarchical Subspace Model for Language-Attuned Acoustic Unit Discovery
|
In this work, we propose a hierarchical subspace model for acoustic unit discovery. In this approach, we frame the task as one of learning embeddings on a low-dimensional phonetic subspace, and simultaneously specify the subspace itself as an embedding on a hyper-subspace. We train the hyper-subspace on a set of transcribed languages and transfer it to the target language. In the target language, we infer both the language and unit embeddings in an unsupervised manner, and in so doing, we simultaneously learn a subspace of units specific to that language and the units that dwell on it. We conduct our experiments on TIMIT and two low-resource languages: Mboshi and Yoruba. Results show that our model outperforms major acoustic unit discovery techniques, both in terms of clustering quality and segmentation accuracy.
|
['Murat Saraclar', 'Jan Cernocky', 'Lukas Burget', 'Lucas Ondel', 'Bolaji Yusuf']
|
2020-11-04
| null | null | null | null |
['acoustic-unit-discovery']
|
['speech']
|
[ 4.16837931e-02 1.42206997e-02 -3.24006349e-01 -3.12685519e-01
-9.95651245e-01 -9.18488145e-01 3.70783687e-01 -3.46669257e-01
-3.96706998e-01 2.40847185e-01 6.25246823e-01 -3.48277211e-01
3.79201323e-01 -3.81288439e-01 -7.73846984e-01 -7.56657302e-01
1.63491279e-01 9.26437020e-01 -7.05980733e-02 4.40633953e-01
-1.01697840e-01 3.05080116e-01 -1.06059110e+00 8.29891488e-02
7.13794470e-01 1.74613610e-01 2.22493082e-01 5.89111984e-01
-9.23814531e-03 2.43380025e-01 -2.16494530e-01 1.10434137e-01
1.31390174e-03 -4.64349091e-01 -8.60782087e-01 3.77906770e-01
3.00391823e-01 -6.74176663e-02 -5.76457798e-01 8.27980697e-01
2.60474682e-01 2.33766198e-01 1.10292578e+00 -6.31579459e-01
-5.78035831e-01 1.10902345e+00 -2.58728713e-01 8.67889747e-02
-2.90909540e-02 -8.70452598e-02 1.41978145e+00 -1.22274542e+00
5.94918132e-01 1.24082208e+00 3.99003595e-01 5.65033853e-01
-1.69393802e+00 -5.69477081e-01 1.81300014e-01 -1.81565434e-02
-1.42288327e+00 -8.85926068e-01 8.45210791e-01 -8.50186110e-01
9.49692905e-01 9.32257343e-03 3.82417023e-01 1.14366078e+00
-3.61464769e-01 1.08088684e+00 7.49770880e-01 -5.64450979e-01
6.00201786e-01 6.94436282e-02 4.84373540e-01 3.78115356e-01
-1.20785095e-01 -2.86858678e-01 -5.15768111e-01 -2.43811980e-01
7.13344514e-01 -3.62502754e-01 -1.56270117e-01 -6.23648286e-01
-1.24460328e+00 8.88001859e-01 -8.99395496e-02 5.06018579e-01
-2.80846179e-01 5.56450970e-02 3.11740816e-01 -1.59550637e-01
6.55508101e-01 3.20277870e-01 -3.79243851e-01 -2.70825326e-01
-9.11625922e-01 -2.49363139e-01 9.40085292e-01 8.70241106e-01
9.70122337e-01 1.97703376e-01 1.53276905e-01 1.22197998e+00
5.10217130e-01 4.94278640e-01 5.71795583e-01 -9.23227727e-01
3.43660533e-01 3.17200065e-01 -2.71197826e-01 -2.79428810e-01
-3.92777473e-02 -2.03571081e-01 -5.45502186e-01 -4.72547889e-01
9.41215008e-02 -1.77085131e-01 -1.11751652e+00 1.90230429e+00
2.40361854e-01 7.86963582e-01 1.28031746e-01 6.13349020e-01
4.61571902e-01 1.04628408e+00 -2.49501094e-01 -1.49493247e-01
1.24795055e+00 -1.05155528e+00 -6.28727198e-01 -3.93285960e-01
5.64755499e-01 -5.19536674e-01 1.29479146e+00 2.89474666e-01
-8.02985191e-01 -6.36238635e-01 -7.93631077e-01 -1.27788290e-01
-1.53624624e-01 4.43826616e-01 1.41285032e-01 3.78366232e-01
-9.94762778e-01 1.32810235e-01 -1.16474092e+00 -5.08671820e-01
-1.35221810e-03 3.45473528e-01 -3.25645655e-01 -4.23461385e-02
-9.49123323e-01 3.88028294e-01 5.16232848e-01 -7.00502247e-02
-1.21922481e+00 -4.79896158e-01 -1.04347467e+00 1.47573054e-01
1.90153252e-02 -3.82232279e-01 1.28060460e+00 -5.13021231e-01
-1.60382497e+00 8.19458961e-01 -6.18906021e-01 -3.15627664e-01
-1.46937937e-01 -3.43859285e-01 -2.86680490e-01 -7.63846412e-02
9.23254490e-02 4.60997581e-01 8.24758768e-01 -1.55407751e+00
-3.61880571e-01 -4.75123107e-01 -3.86930406e-01 3.75858366e-01
-6.37228251e-01 3.33948731e-02 -1.04830801e+00 -5.15639782e-01
4.68204737e-01 -1.25112581e+00 -1.57399759e-01 -8.28871548e-01
-6.86204970e-01 -3.46204668e-01 7.17588902e-01 -7.31882393e-01
1.47292149e+00 -2.42138076e+00 7.72733927e-01 3.33734959e-01
-3.98839265e-03 3.05615012e-02 -1.34211322e-02 2.82934695e-01
1.47337526e-01 5.01309410e-02 -7.03866243e-01 -8.09423923e-01
9.58233476e-02 8.44264328e-01 -6.19659305e-01 6.32586062e-01
-2.19685473e-02 7.39311874e-01 -8.59678447e-01 -4.06146467e-01
1.98331252e-01 5.42533517e-01 -8.10349405e-01 3.51876795e-01
-8.51834342e-02 5.78365386e-01 -2.66929924e-01 4.19984788e-01
3.16100270e-01 2.98350126e-01 6.01031363e-01 7.99943879e-02
-1.45240366e-01 6.91381514e-01 -1.25409281e+00 1.95435226e+00
-4.90515828e-01 6.28588438e-01 1.58329248e-01 -1.00022924e+00
8.07603955e-01 5.29419780e-01 4.23961788e-01 1.55043766e-01
-1.92417398e-01 2.87344575e-01 1.04712754e-01 -3.34873378e-01
3.35155606e-01 -3.22199911e-02 -3.22276741e-01 6.76140130e-01
5.09790361e-01 4.60631633e-03 -9.89938378e-02 9.57887024e-02
8.40154111e-01 1.67606756e-01 7.56764188e-02 -3.46874923e-01
2.57664055e-01 -2.83763081e-01 7.25453556e-01 5.01462638e-01
-3.01111918e-02 5.99182248e-01 2.93193996e-01 2.90019810e-02
-1.14783597e+00 -1.40944541e+00 -3.95139307e-01 1.45331812e+00
-2.88904309e-01 -6.38960123e-01 -8.10505748e-01 -5.24853170e-01
4.27176477e-03 8.13283503e-01 -6.05095029e-01 8.24618489e-02
-9.15730715e-01 -3.97771627e-01 6.69417083e-01 5.62477767e-01
-2.35876232e-01 -1.12794363e+00 1.06174136e-02 2.42691666e-01
-1.87329069e-01 -1.29591990e+00 -8.54341269e-01 4.36368883e-01
-8.80827427e-01 -5.37524819e-01 -5.32227635e-01 -1.26974142e+00
6.96284115e-01 1.97219625e-02 9.58525062e-01 -6.30979002e-01
9.29392055e-02 4.16371107e-01 -1.52168706e-01 -6.94782138e-02
-4.52295274e-01 4.79033172e-01 5.78137100e-01 3.20981622e-01
7.07019448e-01 -6.03159547e-01 -1.40049076e-02 1.29038036e-01
-7.17126906e-01 -3.50263268e-01 2.81688750e-01 6.86245084e-01
8.95936787e-01 -2.34033078e-01 2.17035130e-01 -1.04242563e+00
3.41767311e-01 -5.88362515e-01 -5.54864526e-01 8.60859826e-02
-2.75967330e-01 1.01411648e-01 4.46499974e-01 -4.52400506e-01
-7.63142526e-01 5.73749483e-01 -1.12275463e-02 -7.94921815e-01
-4.44754601e-01 4.40823704e-01 -5.50452292e-01 6.61181509e-01
3.82392406e-01 5.26452422e-01 -4.43992049e-01 -1.06315005e+00
8.31887960e-01 1.14241564e+00 7.76709676e-01 -8.23417544e-01
9.36207831e-01 2.77293384e-01 -7.86651015e-01 -1.35227573e+00
-4.56068456e-01 -9.45899308e-01 -1.43451142e+00 8.79943594e-02
1.10500765e+00 -1.02974439e+00 -1.07592210e-01 1.80911031e-02
-1.20124030e+00 -2.97557443e-01 -3.83714736e-01 7.17970550e-01
-5.32965720e-01 3.77143174e-01 -7.47247458e-01 -8.79039884e-01
-8.73083901e-03 -1.32089055e+00 1.23040283e+00 -4.53650951e-02
-3.82944286e-01 -1.12303555e+00 7.83414841e-01 2.33199954e-01
-1.16145268e-01 -3.26021343e-01 1.03799832e+00 -9.29534554e-01
-3.16866010e-01 1.28488034e-01 2.92961210e-01 5.96003473e-01
2.78729230e-01 7.82027915e-02 -1.48160672e+00 -3.24462146e-01
-1.42207876e-01 -9.34221745e-02 9.48082626e-01 5.41179299e-01
9.54140186e-01 -1.42192990e-01 -4.70243812e-01 7.75330126e-01
1.01979327e+00 3.02355886e-01 1.06164932e-01 -8.54083672e-02
1.05831635e+00 6.98103786e-01 2.31641736e-02 -3.03681921e-02
3.43123198e-01 6.29867375e-01 -1.77217454e-01 -1.50715141e-02
7.36676827e-02 -5.18248677e-01 7.84460843e-01 1.88787091e+00
3.09647530e-01 9.20377672e-02 -1.12950563e+00 1.12668526e+00
-1.87303758e+00 -5.20949841e-01 2.96987593e-01 2.25795984e+00
9.17102158e-01 -2.57099606e-02 2.93416500e-01 -1.79508105e-01
6.97988033e-01 1.28533334e-01 -5.97950578e-01 -5.24594486e-01
-2.00958863e-01 9.89288762e-02 1.24852099e-01 8.68943393e-01
-1.19712245e+00 1.48233461e+00 6.88127518e+00 3.47887516e-01
-9.93310094e-01 2.40482211e-01 4.17459995e-01 -6.04453310e-02
-5.41737616e-01 1.89639434e-01 -8.01580846e-01 2.38811150e-01
1.28148258e+00 -6.22963496e-02 6.42607749e-01 8.23868871e-01
1.20183691e-01 5.53296030e-01 -1.43382728e+00 7.70040989e-01
1.75317526e-01 -1.09836400e+00 2.11759843e-02 4.65711415e-01
7.67847478e-01 4.85965014e-01 5.52570485e-02 4.66083467e-01
3.49099785e-01 -1.09720266e+00 4.14770961e-01 7.35425800e-02
8.31611454e-01 -5.86594403e-01 2.99178004e-01 4.41243708e-01
-1.15405464e+00 3.07454795e-01 -6.10186577e-01 2.69397914e-01
1.77968234e-01 2.60969609e-01 -1.17362475e+00 1.68303713e-01
4.65178758e-01 7.39320576e-01 -1.36509806e-01 7.37098575e-01
-3.74100238e-01 1.48040700e+00 -3.23763102e-01 4.38779235e-01
3.34027439e-01 -7.26000547e-01 5.95496595e-01 1.65172303e+00
9.49449986e-02 1.46733345e-02 4.41293687e-01 9.42555010e-01
-3.23411196e-01 3.38256389e-01 -9.44850385e-01 -2.56597579e-01
8.25455487e-01 1.16288948e+00 -4.41456765e-01 -5.51905930e-01
-3.48792225e-01 1.14131677e+00 5.15812635e-01 7.05232739e-01
-4.98316079e-01 -1.52159795e-01 1.08142316e+00 -3.10823858e-01
5.71300864e-01 -5.59610903e-01 -2.93637097e-01 -1.41418517e+00
-1.60276577e-01 -6.86637521e-01 2.10726768e-01 -2.78804332e-01
-1.03637195e+00 6.50839269e-01 -2.28869304e-01 -9.77235615e-01
-5.61196387e-01 -6.29876435e-01 -6.49921656e-01 9.67681408e-01
-9.87471938e-01 -1.19228113e+00 4.70866919e-01 4.92102176e-01
8.36234391e-01 -3.22117895e-01 1.12838626e+00 1.96610257e-01
-9.38967943e-01 5.25299013e-01 6.45985842e-01 5.02375603e-01
5.90376377e-01 -1.59914219e+00 6.21416330e-01 9.21833873e-01
1.14468479e+00 8.62140596e-01 4.49416637e-01 -3.94217730e-01
-1.37554264e+00 -1.08574438e+00 1.00904191e+00 -7.63067722e-01
8.67215872e-01 -1.14240801e+00 -1.18730569e+00 1.25187027e+00
3.50911021e-01 -1.32467434e-01 1.14061320e+00 6.13803625e-01
-5.40492654e-01 2.12916508e-01 -4.05400902e-01 5.05881369e-01
8.95693243e-01 -1.32456815e+00 -9.72824097e-01 3.77956837e-01
9.60823715e-01 -1.84372604e-01 -9.74290192e-01 -1.08955599e-01
4.75456059e-01 -3.82857859e-01 7.65786111e-01 -9.51158345e-01
-1.39485374e-02 -3.80097628e-01 -6.20832443e-01 -1.58285737e+00
-5.45371532e-01 -6.05891526e-01 -2.19071016e-01 1.47799313e+00
7.29623795e-01 -3.06979835e-01 7.50811398e-01 6.88086972e-02
-4.71338242e-01 -5.20318210e-01 -1.19587922e+00 -7.42251217e-01
4.68494266e-01 -7.68303037e-01 4.11148161e-01 1.03599656e+00
1.83684915e-01 9.13437665e-01 -3.23396057e-01 4.94334340e-01
5.58135986e-01 1.87896222e-01 7.78251052e-01 -1.11551273e+00
-6.19306087e-01 -2.55856633e-01 -2.06682324e-01 -1.30126953e+00
8.26060414e-01 -1.11264157e+00 4.96554404e-01 -1.31471431e+00
2.37176389e-01 -2.94639200e-01 -5.60087502e-01 4.26615834e-01
-1.11112960e-01 6.46198988e-02 -8.14225972e-02 6.79750383e-01
-3.60320747e-01 5.23519754e-01 4.29182649e-01 -1.93007767e-01
-5.12810349e-01 -1.17944874e-01 -4.30535704e-01 6.88415527e-01
6.10109866e-01 -3.72994691e-01 -3.45911443e-01 -8.98723483e-01
-4.22796667e-01 -3.36069375e-01 -3.20986718e-01 -6.38820112e-01
1.93447754e-01 8.20663348e-02 7.35725835e-02 -5.11343718e-01
4.52744782e-01 -4.60025400e-01 -2.68897831e-01 1.43438429e-01
-3.47870797e-01 -2.60146648e-01 2.15797395e-01 5.26410282e-01
-3.52341652e-01 -9.80111808e-02 5.37273586e-01 2.69259244e-01
-5.91853976e-01 2.61981398e-01 -5.10248601e-01 1.44784227e-01
7.40454733e-01 -3.15504335e-02 2.78220445e-01 -9.99725163e-02
-9.32572544e-01 5.49363382e-02 6.26827359e-01 6.16999924e-01
4.37883735e-01 -1.43873060e+00 -7.86032856e-01 5.73718309e-01
3.89555186e-01 6.90385699e-02 -2.23554894e-01 5.07532239e-01
-4.49066572e-02 5.50931811e-01 3.17151159e-01 -9.56923544e-01
-9.88618016e-01 3.89087647e-01 2.67038792e-01 1.43493610e-02
-5.97746074e-01 8.56898248e-01 7.27211356e-01 -1.10328186e+00
3.48355651e-01 -4.95317817e-01 -2.50652254e-01 1.05041817e-01
1.97875738e-01 7.20753744e-02 -2.92991549e-01 -1.18452525e+00
-5.04784584e-01 6.55744791e-01 2.27795411e-02 -6.63423300e-01
1.26947153e+00 -1.42114788e-01 -2.67169237e-01 1.31873965e+00
1.45705462e+00 3.76418799e-01 -1.09679985e+00 -5.69818914e-01
4.25521940e-01 1.58934649e-02 7.72215575e-02 -1.68197826e-01
-5.97305715e-01 1.01842475e+00 3.76458853e-01 -1.12047270e-01
6.39162362e-01 5.40862262e-01 6.28287375e-01 2.18838096e-01
3.61684081e-03 -1.23281705e+00 -2.74935812e-01 8.61632407e-01
4.68230247e-01 -9.00397897e-01 -5.61642826e-01 -1.04495980e-01
-6.56339169e-01 9.64019001e-01 2.99070418e-01 -1.90580711e-01
7.90223777e-01 1.89194843e-01 2.66019493e-01 -8.82906914e-02
-7.57991076e-01 -1.83576554e-01 6.94109499e-01 5.25834680e-01
7.68934250e-01 5.74074388e-01 2.72900939e-01 6.37820542e-01
-2.78803796e-01 -4.65429246e-01 2.64411837e-01 5.06043613e-01
-6.18475795e-01 -1.10028148e+00 -4.94149983e-01 2.25921020e-01
-1.16727144e-01 -2.10729510e-01 -5.97804666e-01 4.32071120e-01
1.67524796e-02 7.93976724e-01 4.51971501e-01 -5.14087498e-01
3.23230416e-01 6.13790691e-01 2.83982661e-02 -1.23041713e+00
-1.51999220e-01 7.60257483e-01 5.62902763e-02 -3.19746524e-01
1.80438347e-02 -9.72326338e-01 -1.35876346e+00 3.70335191e-01
-4.07997109e-02 5.27304649e-01 6.81110442e-01 9.28894997e-01
1.60385981e-01 4.45204139e-01 6.86584949e-01 -7.88326085e-01
-6.12201810e-01 -1.22505260e+00 -8.23565722e-01 3.57980460e-01
4.00868922e-01 -2.91543454e-01 -3.72676611e-01 1.93844453e-01]
|
[14.478325843811035, 6.642317771911621]
|
02b228f4-3fbe-4e07-9d15-9ed6f4825d9f
|
cyclical-self-supervision-for-semi-supervised
|
2210.11291
| null |
https://arxiv.org/abs/2210.11291v2
|
https://arxiv.org/pdf/2210.11291v2.pdf
|
Cyclical Self-Supervision for Semi-Supervised Ejection Fraction Prediction from Echocardiogram Videos
|
Left-ventricular ejection fraction (LVEF) is an important indicator of heart failure. Existing methods for LVEF estimation from video require large amounts of annotated data to achieve high performance, e.g. using 10,030 labeled echocardiogram videos to achieve mean absolute error (MAE) of 4.10. Labeling these videos is time-consuming however and limits potential downstream applications to other heart diseases. This paper presents the first semi-supervised approach for LVEF prediction. Unlike general video prediction tasks, LVEF prediction is specifically related to changes in the left ventricle (LV) in echocardiogram videos. By incorporating knowledge learned from predicting LV segmentations into LVEF regression, we can provide additional context to the model for better predictions. To this end, we propose a novel Cyclical Self-Supervision (CSS) method for learning video-based LV segmentation, which is motivated by the observation that the heartbeat is a cyclical process with temporal repetition. Prediction masks from our segmentation model can then be used as additional input for LVEF regression to provide spatial context for the LV region. We also introduce teacher-student distillation to distill the information from LV segmentation masks into an end-to-end LVEF regression model that only requires video inputs. Results show our method outperforms alternative semi-supervised methods and can achieve MAE of 4.17, which is competitive with state-of-the-art supervised performance, using half the number of labels. Validation on an external dataset also shows improved generalization ability from using our method. Our code is available at https://github.com/xmed-lab/CSS-SemiVideo.
|
['Kwang-Ting Cheng', 'Xinpeng Ding', 'Xiaomeng Li', 'Weihang Dai']
|
2022-10-20
| null | null | null | null |
['video-prediction']
|
['computer-vision']
|
[ 1.03522837e-01 1.57458082e-01 -4.58699375e-01 -6.00241184e-01
-7.54445195e-01 -5.47620177e-01 -1.76953040e-02 -1.64940134e-01
-1.04594551e-01 8.00920606e-01 1.11027114e-01 -2.69317031e-01
3.54172975e-01 -4.03814048e-01 -6.63875401e-01 -5.16634643e-01
-1.56444609e-01 5.07947803e-01 2.16358528e-01 3.71689558e-01
-1.01167336e-01 3.19294095e-01 -1.08521068e+00 3.11968178e-01
7.79258072e-01 9.10810351e-01 3.28195632e-01 1.24165058e+00
2.27125525e-01 1.12508714e+00 -5.78779519e-01 2.89638966e-01
6.06631674e-02 -6.96507514e-01 -1.01892889e+00 3.78852814e-01
6.63272977e-01 -6.04195833e-01 -3.03547263e-01 4.33732361e-01
6.60322905e-01 -2.37524077e-01 5.01508594e-01 -1.00879693e+00
-1.90692857e-01 5.20536900e-01 -1.67129114e-01 6.49945021e-01
5.39721884e-02 1.03236154e-01 9.16351676e-01 -8.83794248e-01
9.40792620e-01 6.82372153e-01 7.35679507e-01 5.66509902e-01
-1.32416141e+00 -5.04715621e-01 -9.36404392e-02 3.13513093e-02
-1.12873304e+00 -2.73221046e-01 7.95516968e-01 -7.38443315e-01
5.89618981e-01 2.22110972e-01 9.76439595e-01 6.02005661e-01
6.44786507e-02 1.00242233e+00 9.34215903e-01 -2.87929058e-01
-5.76703325e-02 1.24787152e-01 -6.29828274e-02 9.24761653e-01
-2.93957800e-01 6.25272002e-03 -2.43375987e-01 9.19295922e-02
1.06712425e+00 1.49197310e-01 -4.47044641e-01 -4.30509329e-01
-1.52388978e+00 5.68014324e-01 2.84081012e-01 1.85695335e-01
-5.14925905e-02 2.37067178e-01 5.65746844e-01 3.30152750e-01
6.08895004e-01 2.03658059e-01 -8.30501437e-01 -3.19967538e-01
-1.43219960e+00 4.23157290e-02 5.98433912e-01 7.84237266e-01
5.44067502e-01 2.01820329e-01 -1.13455124e-01 7.89755225e-01
2.15482756e-01 4.93963420e-01 3.31388235e-01 -1.41461825e+00
3.47998112e-01 3.98664653e-01 -1.44242510e-01 -5.93218744e-01
-5.42152286e-01 -6.42387509e-01 -8.60955000e-01 5.79096377e-02
5.67372143e-01 -3.56289625e-01 -8.32955658e-01 1.37271249e+00
3.31487775e-01 6.81165576e-01 -6.82603046e-02 9.96139884e-01
1.12138021e+00 6.16104424e-01 3.59825976e-02 -4.87100661e-01
1.09695601e+00 -1.19794750e+00 -4.64549571e-01 -1.41766174e-02
1.27525878e+00 -4.99748051e-01 9.46250260e-01 2.21479505e-01
-8.36357951e-01 -7.27618277e-01 -7.53799796e-01 1.49243176e-01
3.34100276e-01 4.28229481e-01 3.06325197e-01 4.37639147e-01
-9.89200115e-01 9.10775661e-01 -1.10953939e+00 -1.26552954e-01
7.07663953e-01 3.30751181e-01 -2.97625273e-01 9.83670205e-02
-9.07536089e-01 5.23235977e-01 2.90818572e-01 -1.79378949e-02
-7.84803689e-01 -1.09525919e+00 -8.49346876e-01 -1.32598042e-01
3.66697103e-01 -7.12341487e-01 1.11086071e+00 -1.17028248e+00
-1.32030249e+00 1.15935314e+00 -1.19557105e-01 -5.90681255e-01
6.56441867e-01 -2.54292101e-01 -1.17814742e-01 7.41501868e-01
1.75465513e-02 1.06595814e+00 8.27595830e-01 -9.79235888e-01
-4.92546350e-01 -7.47484416e-02 -3.56684238e-01 1.39896065e-01
-9.68510285e-02 -6.81344746e-03 -4.72952694e-01 -9.61718082e-01
9.80527028e-02 -1.28366923e+00 -1.79205760e-01 6.60020709e-02
-2.36296922e-01 -2.71194037e-02 1.04617441e+00 -1.04944336e+00
1.38021958e+00 -1.98792446e+00 2.42047515e-02 -5.49665578e-02
6.31848395e-01 3.82685065e-01 2.18896091e-01 -1.33733181e-02
-2.31959254e-01 2.53250569e-01 -2.70330846e-01 -3.16675842e-01
-6.66646600e-01 3.08794647e-01 -6.61688671e-02 4.03128684e-01
5.15445694e-02 1.05231500e+00 -9.39975381e-01 -1.03222466e+00
4.12593395e-01 4.83360618e-01 -6.48713350e-01 3.03749561e-01
-6.53935745e-02 1.20656991e+00 -2.66181797e-01 6.44994736e-01
3.89689803e-01 -7.21186519e-01 4.38282251e-01 -1.40717596e-01
2.95076281e-01 -1.17469870e-01 -8.05707455e-01 1.62153876e+00
-2.29003549e-01 8.19379926e-01 -2.51162112e-01 -1.27090859e+00
8.40575099e-01 6.10143781e-01 1.02381456e+00 -5.47344908e-02
-6.95440546e-03 2.30365410e-01 -9.77827013e-02 -5.39681733e-01
-1.46062583e-01 -1.22054875e-01 2.14439422e-01 2.72797436e-01
2.25457966e-01 -1.02576368e-01 2.69133180e-01 1.85008198e-01
9.67648983e-01 5.66028893e-01 2.43419558e-01 -1.64339587e-01
5.56406379e-01 9.40673891e-03 1.01694083e+00 6.00733280e-01
-5.79308808e-01 9.94837999e-01 5.70044279e-01 -7.58330941e-01
-1.19789946e+00 -8.43048990e-01 -3.94767225e-01 7.44162977e-01
-7.72294775e-02 -5.35363913e-01 -6.36365533e-01 -1.08795273e+00
-7.83118531e-02 1.77639186e-01 -3.68624151e-01 3.32815975e-01
-1.05763757e+00 -4.57271218e-01 4.07936513e-01 1.08766699e+00
4.23531681e-01 -1.03227210e+00 -8.59360993e-01 2.09705442e-01
-4.30251807e-01 -1.42400265e+00 -5.43184578e-01 -8.84310082e-02
-1.52262890e+00 -1.04707205e+00 -1.06842518e+00 -9.62423623e-01
7.29514420e-01 -1.83645129e-01 1.35825658e+00 4.08809245e-01
-4.55230087e-01 4.70675886e-01 -2.57947415e-01 -2.50414666e-02
-6.24982417e-01 -9.89691541e-03 -1.65374160e-01 -2.61736840e-01
-2.36346513e-01 -5.11119187e-01 -1.02508378e+00 4.61070120e-01
-4.03987885e-01 3.96982998e-01 2.36173674e-01 1.11427665e+00
9.26827133e-01 -3.86646599e-01 7.25631535e-01 -1.22678018e+00
-2.20203519e-01 -2.97989041e-01 -5.17859757e-01 -3.02347969e-02
-7.31484115e-01 -3.95132601e-01 8.12781215e-01 -4.13959175e-01
-7.83378482e-01 4.47639287e-01 3.13766371e-03 -9.04191792e-01
-2.14642838e-01 4.41664875e-01 4.30341989e-01 1.61969870e-01
4.41895038e-01 1.64638653e-01 3.29106867e-01 -2.94158220e-01
1.89549342e-01 5.99833012e-01 5.23245037e-01 -2.11495847e-01
4.85510468e-01 3.35159510e-01 2.17223212e-01 -7.69523978e-01
-9.21141505e-01 -5.54769993e-01 -9.00992632e-01 -6.52652919e-01
8.26103866e-01 -1.05081725e+00 -5.21609426e-01 2.36875609e-01
-7.22447276e-01 -6.85693622e-01 -3.75628978e-01 7.18093395e-01
-9.44836915e-01 5.84055483e-01 -9.24742997e-01 -4.35906470e-01
-3.90543491e-01 -8.88903856e-01 1.01088047e+00 -1.56706590e-02
-3.16250861e-01 -1.26548100e+00 -2.49016425e-03 9.11398411e-01
8.20078254e-02 4.15585577e-01 5.83444059e-01 -6.75626397e-01
-5.70692182e-01 -8.99568200e-02 -1.14605211e-01 7.33567953e-01
9.27486792e-02 -1.96089186e-02 -7.16955841e-01 -2.76121825e-01
-2.07272276e-01 -4.28994179e-01 8.96840394e-01 8.55126619e-01
1.41021287e+00 -1.24220043e-01 -3.78765613e-01 6.96953714e-01
1.16324925e+00 -1.75781641e-02 4.57386374e-01 1.72971278e-01
9.94031906e-01 3.55133146e-01 9.56912816e-01 5.77217460e-01
4.08955723e-01 5.24688601e-01 1.86459005e-01 -6.38640761e-01
-3.57616931e-01 -1.53976539e-03 1.77808076e-01 9.54589844e-01
-3.24072897e-01 2.86219064e-02 -1.14948058e+00 4.09970373e-01
-1.86875284e+00 -8.83463919e-01 -2.43394345e-01 2.07798529e+00
9.54475760e-01 1.04813166e-01 3.25003475e-01 1.38734490e-01
5.00904083e-01 1.91699207e-01 -4.79917943e-01 3.06223542e-03
1.73261613e-01 -5.38667254e-02 3.56340259e-01 3.79499674e-01
-1.54611230e+00 8.11057866e-01 6.19695044e+00 3.49706978e-01
-1.24276233e+00 4.42926679e-03 1.19368064e+00 -5.32200113e-02
1.73932001e-01 5.77093177e-02 -6.43813610e-01 4.83104259e-01
9.68159974e-01 2.61760741e-01 7.34641328e-02 7.48976052e-01
5.09078801e-01 -1.55608146e-03 -1.14368296e+00 9.62553144e-01
8.55811760e-02 -1.54018438e+00 -4.76428747e-01 -1.93537846e-01
7.77956665e-01 1.63824726e-02 -3.31996858e-01 2.58843154e-01
-3.18334013e-01 -8.72265041e-01 2.83392012e-01 5.38858891e-01
1.17903948e+00 -2.86006421e-01 8.78086090e-01 3.94824475e-01
-1.21197391e+00 -3.45099270e-02 1.36456192e-02 1.95019990e-01
3.51201445e-01 6.43894434e-01 -1.20097756e+00 2.25560069e-01
6.71048403e-01 1.26930070e+00 -4.06041831e-01 7.77860165e-01
-1.28683392e-02 1.17697692e+00 -1.18042789e-01 5.29066443e-01
-1.47908479e-01 -7.26358742e-02 5.95200300e-01 1.23168445e+00
1.84211522e-01 1.16362505e-01 6.24000847e-01 6.33013129e-01
-1.03833586e-01 3.37924182e-01 -5.25748730e-01 -5.35573959e-02
1.31873399e-01 1.19273949e+00 -9.80374396e-01 -7.24326074e-01
-5.89164436e-01 8.82455528e-01 -1.87357347e-02 3.35904360e-01
-9.96887028e-01 -1.75739415e-02 -6.48817718e-02 5.72110057e-01
5.08559346e-01 2.19176691e-02 -2.79437870e-01 -1.35713291e+00
-6.62534013e-02 -7.48485804e-01 5.37526667e-01 -6.96188509e-01
-6.95198715e-01 4.28308755e-01 -1.09135650e-01 -1.59898412e+00
-3.43308449e-01 -3.31103027e-01 -3.46633375e-01 5.23447871e-01
-1.44993913e+00 -1.17423844e+00 -4.44047570e-01 1.03774525e-01
8.84736359e-01 -1.11465469e-01 8.38261008e-01 3.62654805e-01
-5.99304438e-01 2.81357318e-01 -1.70101106e-01 5.55997133e-01
9.14948761e-01 -1.51516366e+00 -3.23445648e-02 6.48766220e-01
7.85703883e-02 7.28370696e-02 4.97333616e-01 -8.39349568e-01
-9.16326523e-01 -1.29285300e+00 9.28107202e-01 -5.40613472e-01
3.23463172e-01 8.01605955e-02 -7.59712040e-01 9.60676372e-01
-8.12417865e-02 9.18006182e-01 7.30100513e-01 -2.41349995e-01
2.02968940e-01 -1.96975917e-01 -8.69467258e-01 3.31479669e-01
9.27916944e-01 -2.77214825e-01 -2.95996070e-01 4.82761741e-01
4.61007446e-01 -8.24108243e-01 -1.45144999e+00 8.19559455e-01
5.28222084e-01 -8.60455811e-01 1.03602660e+00 -5.22300243e-01
7.66979098e-01 -3.32176924e-01 2.05402419e-01 -8.84886444e-01
-7.27151111e-02 -5.77817380e-01 -5.69203496e-01 1.02299023e+00
3.31434429e-01 -3.57917786e-01 1.21015620e+00 3.79808366e-01
-9.76054296e-02 -1.28236425e+00 -5.94154835e-01 -6.74122691e-01
-9.47709233e-02 -2.17005044e-01 -1.38877913e-01 1.04431605e+00
-4.97084409e-02 2.28428394e-01 -5.92612207e-01 2.06680316e-03
5.25663018e-01 3.51791769e-01 7.02521682e-01 -1.19785702e+00
-4.04100955e-01 -9.44242701e-02 -5.54311752e-01 -1.26943624e+00
1.49695665e-01 -1.13470531e+00 -1.17179736e-01 -1.41599607e+00
3.20911407e-01 -5.61781704e-01 -3.28784853e-01 4.88080263e-01
-2.43826777e-01 5.54916203e-01 3.21526974e-01 4.44856465e-01
-6.45184338e-01 2.57907838e-01 1.52932262e+00 2.51001455e-02
-5.45441747e-01 2.97410607e-01 -3.02563962e-02 9.05337632e-01
1.02385688e+00 -4.83516693e-01 -4.22058016e-01 9.83445868e-02
-1.82740808e-01 8.00824642e-01 4.48007315e-01 -1.00630701e+00
-1.53989047e-01 1.58692181e-01 6.15797341e-01 -7.05639184e-01
1.00465253e-01 -6.11119509e-01 -3.98263633e-02 6.95448577e-01
-4.05685782e-01 -1.01974852e-01 -1.28902748e-01 4.69535410e-01
-3.89925092e-01 -2.49910623e-01 7.24263489e-01 -3.63300562e-01
-5.87849915e-01 4.03616041e-01 -4.80821520e-01 5.69229007e-01
1.06036413e+00 -3.11975449e-01 7.38656940e-03 -4.89736199e-01
-1.39169002e+00 4.04133976e-01 1.45408422e-01 1.25025958e-01
6.46048248e-01 -1.05833912e+00 -8.53753507e-01 1.15408055e-01
-1.00552492e-01 9.50033441e-02 3.23430300e-01 1.35240698e+00
-9.58564878e-01 3.18151593e-01 -1.19343856e-02 -1.32249928e+00
-1.79310572e+00 4.09486622e-01 5.03738046e-01 -4.22774464e-01
-1.21611762e+00 5.44074237e-01 1.43631488e-01 -3.84047091e-01
2.68677890e-01 -4.98976529e-01 -3.47211689e-01 -1.44568235e-01
2.09581375e-01 4.28152621e-01 -3.25520664e-01 -5.59504747e-01
-3.22231084e-01 6.89165831e-01 1.65747121e-01 2.84098297e-01
1.26501298e+00 -2.55893350e-01 1.92751288e-01 5.50773025e-01
1.20181775e+00 -9.74655151e-02 -1.59926987e+00 -2.40116760e-01
-4.67356145e-02 -4.25689608e-01 1.45262122e-01 -7.39811420e-01
-1.38309598e+00 8.97596896e-01 7.99359441e-01 -1.45083770e-01
1.20081198e+00 1.04591981e-01 8.43604326e-01 3.73279154e-02
-9.89580825e-02 -1.03829670e+00 2.56707430e-01 1.75241500e-01
5.90539873e-01 -1.50650072e+00 1.39401957e-01 -7.59453654e-01
-1.05469191e+00 1.29890740e+00 5.93420565e-01 8.44505243e-03
8.48394036e-01 2.78953880e-01 4.70110744e-01 5.00738546e-02
-7.23114550e-01 1.94250017e-01 4.14053380e-01 3.48493814e-01
7.18286633e-01 5.29851243e-02 -2.06744596e-01 3.16771567e-01
2.11329386e-01 4.91104037e-01 4.28457618e-01 1.00821137e+00
-3.73457283e-01 -1.01916409e+00 -4.00332734e-03 6.57764435e-01
-7.02381253e-01 -6.12505227e-02 2.71575779e-01 5.17974257e-01
8.70232284e-02 4.59490001e-01 -4.34051566e-02 -8.73877332e-02
-2.42226068e-02 2.84365684e-01 4.10554975e-01 -7.05015302e-01
-4.00141031e-01 4.22116876e-01 1.69872835e-01 -5.64049304e-01
-7.17144310e-01 -8.05048466e-01 -1.59871888e+00 1.06803551e-01
-1.60194203e-01 -2.22200349e-01 6.52892590e-02 8.61658990e-01
4.13494438e-01 6.67516589e-01 7.25723505e-01 -8.01941812e-01
-2.58897930e-01 -7.90876091e-01 -4.25705969e-01 4.92760211e-01
4.63300645e-01 -3.77162397e-01 -3.29716623e-01 8.25065732e-01]
|
[14.340490341186523, -2.2567079067230225]
|
95d736ee-c7f1-4343-8403-9bfdc2f230f3
|
graph-neural-networks-for-image
|
2203.03457
| null |
https://arxiv.org/abs/2203.03457v2
|
https://arxiv.org/pdf/2203.03457v2.pdf
|
Graph Neural Networks for Image Classification and Reinforcement Learning using Graph representations
|
In this paper, we will evaluate the performance of graph neural networks in two distinct domains: computer vision and reinforcement learning. In the computer vision section, we seek to learn whether a novel non-redundant representation for images as graphs can improve performance over trivial pixel to node mapping on a graph-level prediction graph, specifically image classification. For the reinforcement learning section, we seek to learn if explicitly modeling solving a Rubik's cube as a graph problem can improve performance over a standard model-free technique with no inductive bias.
|
['David Steiner', 'Naman Goyal']
|
2022-03-07
| null | null | null | null |
['rubik-s-cube']
|
['graphs']
|
[ 5.94683886e-01 8.33211124e-01 -2.10338384e-01 -3.50286961e-01
-3.62452060e-01 -2.65473634e-01 3.42492908e-01 2.71058261e-01
-1.85772315e-01 5.53062201e-01 -3.92714776e-02 -5.94602048e-01
-1.09505035e-01 -8.49541426e-01 -1.11571670e+00 -5.15612721e-01
-3.81232500e-01 3.67932618e-01 -1.57547474e-01 -4.67978865e-02
2.57038444e-01 5.72983682e-01 -1.09323263e+00 1.31742150e-01
5.30341387e-01 1.03148484e+00 6.37891367e-02 8.26774538e-01
1.81054801e-01 1.30275011e+00 -3.84409994e-01 -1.60591155e-01
2.38451377e-01 -2.93325454e-01 -8.79857838e-01 1.84638053e-01
1.04898739e+00 -1.39495581e-01 -7.71421075e-01 1.30731857e+00
7.13165402e-02 2.97114234e-02 7.72849083e-01 -1.40113783e+00
-1.03712749e+00 5.67399859e-01 -6.33869708e-01 3.90920550e-01
1.93110138e-01 2.07248673e-01 1.13185108e+00 -5.91371000e-01
8.32371414e-01 1.40462315e+00 8.02555263e-01 4.67605025e-01
-1.52457666e+00 -5.61324418e-01 4.03326303e-01 4.77009863e-01
-1.25938284e+00 -2.98855245e-01 1.02654493e+00 -4.59620118e-01
1.37472916e+00 2.51535214e-02 7.58610964e-01 7.23814785e-01
5.82375944e-01 6.96879327e-01 1.13067126e+00 -4.83062983e-01
7.47649223e-02 -2.96397090e-01 1.28737345e-01 1.22584832e+00
2.37149626e-01 5.20666778e-01 -2.93401659e-01 2.70400882e-01
8.71782541e-01 -3.48873734e-01 -1.34824857e-01 -7.23990798e-01
-7.85838783e-01 1.23475206e+00 1.27978718e+00 -1.00981534e-01
-2.75877148e-01 9.64733183e-01 2.90688872e-02 4.45564449e-01
4.43926722e-01 9.23916399e-01 -8.13600048e-02 6.32258773e-01
-5.59841871e-01 -2.49519031e-02 8.67585540e-01 9.15907502e-01
1.07503414e+00 5.30321956e-01 -3.53602879e-02 3.73091429e-01
2.56674767e-01 1.71775326e-01 1.19342677e-01 -1.09722185e+00
1.62950709e-01 5.68145871e-01 -5.36818683e-01 -1.16143763e+00
-7.96134472e-01 -3.57362658e-01 -8.87078047e-01 4.42505002e-01
2.26745352e-01 -3.13987017e-01 -1.29473388e+00 1.75566971e+00
-7.05807582e-02 2.40577996e-01 9.68395099e-02 5.92317641e-01
1.10756671e+00 4.22161281e-01 2.31758147e-01 -2.38366351e-01
8.85043681e-01 -1.14352572e+00 -3.39881122e-01 -7.38937140e-01
7.72249758e-01 -1.96357906e-01 8.07212710e-01 2.16072083e-01
-9.28087473e-01 -5.26165962e-01 -1.12226152e+00 -1.08956337e-01
-3.91779035e-01 -2.14682519e-01 9.35552359e-01 2.62859523e-01
-1.59086680e+00 8.09505999e-01 -5.80708086e-01 -5.06333530e-01
7.01464295e-01 4.39237297e-01 -5.68453729e-01 -2.73664087e-01
-1.04345691e+00 1.10359108e+00 4.72979575e-01 -2.19374612e-01
-1.06030226e+00 -4.51758921e-01 -1.25397694e+00 1.50094256e-01
5.48341036e-01 -7.78805256e-01 1.16147017e+00 -1.29602098e+00
-1.01626790e+00 9.42338884e-01 1.06554311e-02 -9.55963433e-01
-1.78352697e-04 2.40761071e-01 -9.73325297e-02 1.99371725e-01
-1.16035067e-01 1.21803725e+00 1.07055831e+00 -1.28620267e+00
-2.87949085e-01 -5.51472187e-01 2.09890604e-01 3.80210102e-01
-5.34994267e-02 -5.00695169e-01 -2.92488426e-01 -6.55255377e-01
6.52414560e-02 -1.16749263e+00 -5.68096578e-01 1.65608544e-02
-3.16226453e-01 -2.58445203e-01 5.97940862e-01 -5.74638963e-01
6.89975798e-01 -1.89959252e+00 2.55614579e-01 3.12444806e-01
6.47891700e-01 -7.63457566e-02 -5.26578844e-01 1.91114694e-01
-4.77586627e-01 2.05592260e-01 -1.49917230e-01 2.08465040e-01
-4.30349439e-01 4.06593144e-01 1.62684955e-02 4.51665998e-01
5.72666943e-01 1.41613996e+00 -9.04071331e-01 -4.85000104e-01
8.10631216e-02 2.40966469e-01 -5.24988353e-01 -9.13150012e-02
-3.44231933e-01 1.23361893e-01 -1.87277108e-01 4.53286618e-01
3.14568907e-01 -7.72758543e-01 3.75923097e-01 -1.66286051e-01
6.14210606e-01 -6.28846213e-02 -6.77932620e-01 1.64486039e+00
-2.11464033e-01 9.27645802e-01 4.79870774e-02 -1.48617816e+00
6.69078946e-01 -2.40315050e-01 2.90703714e-01 -1.16091311e+00
-4.35577184e-02 -4.01052326e-01 3.94381016e-01 -1.16937555e-01
2.47059539e-01 -1.45161211e-01 2.82362610e-01 2.29635432e-01
1.85261160e-01 -2.41696373e-01 3.12129874e-02 2.58655876e-01
1.62423909e+00 -1.01407878e-01 2.41906598e-01 -4.06303227e-01
8.61819759e-02 2.82655418e-01 2.06396118e-01 8.37319374e-01
-2.19138265e-01 2.18360901e-01 5.28437972e-01 -6.11586094e-01
-7.25598574e-01 -1.18027258e+00 1.44447744e-01 1.03106594e+00
1.60059884e-01 -2.26964280e-01 -5.36304951e-01 -7.37018049e-01
3.40422601e-01 6.62170827e-01 -8.72846305e-01 -6.43356979e-01
-4.04314071e-01 -4.54619706e-01 1.66694671e-01 7.77221441e-01
2.64315784e-01 -1.11262214e+00 -2.52512395e-01 7.98003450e-02
3.50693882e-01 -1.11959302e+00 -4.36890960e-01 7.47957826e-01
-1.11486316e+00 -1.28650045e+00 -3.53722483e-01 -1.22204101e+00
9.67684865e-01 3.06519628e-01 1.62262952e+00 4.15048927e-01
-4.72858399e-01 7.78917253e-01 -7.91091919e-02 -5.64808488e-01
-3.76214832e-01 5.17586432e-02 -2.69544065e-01 -5.04467487e-01
1.75659999e-01 -3.86838168e-01 -4.70019311e-01 4.64398935e-02
-5.43295503e-01 1.85078770e-01 6.38616920e-01 8.62172902e-01
6.59231186e-01 3.50875854e-02 4.62759256e-01 -1.30697024e+00
8.50040317e-01 -2.77038366e-01 -7.56261826e-01 3.38596225e-01
-1.04638255e+00 4.95548069e-01 5.06990016e-01 -2.08367810e-01
-4.05375600e-01 5.87924361e-01 3.58771116e-01 -8.10748637e-01
-7.71084102e-03 8.41651261e-01 2.72275269e-01 -5.92688262e-01
9.87906456e-01 1.42692894e-01 2.01913178e-01 1.26250893e-01
5.97431958e-01 -1.52123839e-01 5.56064069e-01 -3.45832467e-01
8.75854969e-01 8.92076418e-02 5.14984131e-01 -7.52910554e-01
-6.81536496e-01 -1.78407490e-01 -5.48699319e-01 -2.60324776e-01
1.09302747e+00 -8.01043212e-01 -7.08566248e-01 1.02804422e-01
-1.03919816e+00 -8.55469882e-01 -3.42131257e-01 2.78146803e-01
-9.05828178e-01 1.93461701e-01 -5.31688392e-01 -4.46477056e-01
-7.46623948e-02 -9.62139666e-01 8.27402294e-01 2.37762600e-01
7.73283318e-02 -1.30179095e+00 -6.87457025e-02 1.28508240e-01
1.49713278e-01 3.74398530e-01 1.21240807e+00 -4.32951957e-01
-7.06978917e-01 -6.07066303e-02 -5.64339876e-01 2.74892539e-01
3.72815435e-03 -2.77168572e-01 -9.74909365e-01 -4.86453176e-01
-3.80777627e-01 -7.63797462e-01 1.20188856e+00 6.30490899e-01
1.58026063e+00 -3.03555638e-01 -3.29900831e-01 8.48310113e-01
1.51085246e+00 1.84660584e-01 6.12478256e-01 2.41742004e-02
9.67795014e-01 5.98600745e-01 1.45100668e-01 -2.46714741e-01
3.65886211e-01 2.34812692e-01 9.11439359e-01 -4.28868443e-01
-4.60203260e-01 -4.25851554e-01 8.14512223e-02 3.88119400e-01
2.18582422e-01 -3.12217414e-01 -1.13984942e+00 3.23327601e-01
-1.74449384e+00 -8.52569997e-01 1.06154256e-01 1.72989035e+00
3.45932990e-01 2.50038326e-01 -4.95808870e-02 -2.94321209e-01
7.02480376e-01 3.79500240e-01 -8.67700398e-01 -5.27818620e-01
1.08501753e-02 2.53748357e-01 9.21692252e-01 6.74035430e-01
-1.07594645e+00 1.26544404e+00 8.08908653e+00 4.18609738e-01
-1.05659366e+00 -4.64632183e-01 9.73061800e-01 2.97064960e-01
-1.54587969e-01 -7.51470495e-03 -3.25128287e-01 -1.94648325e-01
1.07590830e+00 -1.13622442e-01 1.01239657e+00 1.02404499e+00
-2.79464126e-01 -3.02192699e-02 -1.69527745e+00 1.01691532e+00
4.14982677e-01 -1.63320553e+00 1.25726402e-01 2.78364867e-01
6.31494462e-01 2.36262739e-01 1.90283179e-01 4.63847280e-01
8.84356499e-01 -1.78817677e+00 1.69039786e-01 4.84268308e-01
5.93511403e-01 -4.56531554e-01 3.09050381e-01 2.04677835e-01
-1.03163052e+00 5.40515222e-02 -5.41585565e-01 -1.82746351e-01
-4.69084352e-01 1.00980543e-01 -1.24033916e+00 2.30565131e-01
4.83831257e-01 9.32207286e-01 -9.71791923e-01 9.04057026e-01
-2.43136436e-01 6.20627582e-01 -1.09117180e-02 -4.27307263e-02
3.34391326e-01 -1.69259906e-01 1.99692026e-01 9.25735354e-01
-1.38866305e-01 1.12483785e-01 4.38714474e-01 8.70824933e-01
-6.20157242e-01 -1.50679424e-01 -1.27954173e+00 -3.20246041e-01
1.25426715e-02 1.12546587e+00 -8.24151099e-01 -1.97053611e-01
-4.70471829e-01 7.77793646e-01 8.86261225e-01 5.78271031e-01
-6.18673146e-01 -1.42853722e-01 1.87208742e-01 1.66558653e-01
4.07688171e-01 -1.90236807e-01 -3.28760356e-01 -8.80410254e-01
-4.46373194e-01 -7.01828003e-01 5.48335910e-01 -1.24214697e+00
-1.27891552e+00 3.60978603e-01 -8.65474865e-02 -6.99844241e-01
-4.92903739e-01 -1.04886317e+00 -5.90068698e-01 7.62515724e-01
-1.60422063e+00 -1.18867671e+00 -4.40009683e-01 8.25095117e-01
1.86371788e-01 -3.93729091e-01 7.89405584e-01 -3.42915118e-01
-3.10989946e-01 6.01390243e-01 -1.57010615e-01 4.43504184e-01
3.29077244e-01 -1.56876040e+00 6.81413710e-01 7.65847623e-01
8.78160655e-01 1.27605855e-01 5.29117465e-01 -6.35306180e-01
-1.74524319e+00 -1.21570158e+00 1.07151434e-01 -3.51548165e-01
7.28536785e-01 -1.59344792e-01 -9.06093001e-01 1.10928798e+00
2.75488168e-01 5.21914780e-01 3.06518406e-01 3.27275097e-01
-5.85250258e-01 -9.40917879e-02 -9.78189349e-01 4.90712643e-01
1.08951521e+00 -6.83353305e-01 -3.51209402e-01 6.24992073e-01
6.10349059e-01 -3.76678288e-01 -8.84049296e-01 3.96367252e-01
2.08047986e-01 -3.47159624e-01 9.43520367e-01 -1.22028303e+00
6.31583095e-01 6.03341609e-02 -2.49581113e-01 -1.68276262e+00
-7.56322801e-01 -2.27629602e-01 -1.07784480e-01 3.29195708e-01
5.44856846e-01 -4.86935884e-01 1.34800076e+00 3.42892438e-01
-8.98883343e-02 -7.13271737e-01 -8.13881874e-01 -7.58035660e-01
3.42995554e-01 -2.77111620e-01 -6.11097403e-02 9.40919697e-01
-3.07369053e-01 9.12414610e-01 -1.71513900e-01 2.39115998e-01
7.30348587e-01 4.83007692e-02 7.93183923e-01 -1.32769835e+00
-4.14439380e-01 -6.66456282e-01 -9.60967183e-01 -7.79414535e-01
6.33448780e-01 -1.41555357e+00 9.44679752e-02 -2.01145959e+00
1.16687328e-01 2.01487280e-02 -4.17126805e-01 7.48896539e-01
-8.84525180e-02 2.39113137e-01 3.37747812e-01 -1.52482569e-01
-7.47922957e-01 4.17004585e-01 1.52789915e+00 -7.30186760e-01
2.40339264e-01 -3.02070141e-01 -9.75954592e-01 6.81933701e-01
6.84569538e-01 -4.82342094e-01 -8.23158622e-01 -5.75169265e-01
1.40096292e-01 3.58890027e-01 5.18289566e-01 -9.56659377e-01
2.53324479e-01 -2.70656735e-01 6.63515449e-01 -9.59572122e-02
1.79324269e-01 -7.47121453e-01 -1.59206107e-01 8.25646877e-01
-4.62302566e-01 4.24443245e-01 5.30940950e-01 9.85059679e-01
-7.90640339e-02 -1.16709538e-01 8.58421028e-01 -3.75480443e-01
-1.06954324e+00 3.74180138e-01 -9.17702243e-02 2.49880895e-01
1.15918612e+00 -4.11870152e-01 -5.91020644e-01 -7.46215940e-01
-1.03017807e+00 4.21802133e-01 4.05383945e-01 2.97710896e-01
8.58260989e-01 -1.31537724e+00 -5.78002214e-01 5.82420006e-02
2.74468720e-01 -2.45272100e-01 -1.08512275e-01 4.79456455e-01
-5.95887363e-01 4.22330171e-01 -5.28635979e-01 -6.27115548e-01
-1.28274453e+00 9.52034116e-01 6.11964881e-01 -3.37096781e-01
-5.70853412e-01 9.76881206e-01 4.82029915e-01 -3.11572462e-01
4.05426413e-01 -3.25037777e-01 -3.05766817e-02 -3.73998702e-01
-1.15790498e-02 -3.57360877e-02 3.87088060e-02 -4.69956666e-01
-2.28849053e-01 3.57614428e-01 -1.42746836e-01 3.06467891e-01
1.22947192e+00 2.99184322e-01 9.75043550e-02 4.55591649e-01
1.24357104e+00 -7.24020123e-01 -1.26790273e+00 -2.14317814e-01
1.01436479e-02 -9.66857672e-02 4.48746502e-01 -8.40671599e-01
-1.36600876e+00 9.95055318e-01 7.89620221e-01 2.92966187e-01
1.13688946e+00 3.75438817e-02 1.01265289e-01 8.84868443e-01
2.17376992e-01 -1.00471282e+00 3.64716798e-01 5.22158384e-01
1.14462578e+00 -1.68786621e+00 3.80692780e-01 -4.16289359e-01
-6.36300266e-01 1.14906192e+00 9.69772637e-01 -6.50803089e-01
9.92331326e-01 5.99716865e-02 -2.68487841e-01 -5.68369448e-01
-9.23358977e-01 -3.35490495e-01 6.80987775e-01 1.04369342e+00
3.17858368e-01 5.18093780e-02 2.48391092e-01 -1.61491245e-01
-1.94697976e-01 -1.95807412e-01 6.23581886e-01 6.32273316e-01
-5.97580314e-01 -5.97638786e-01 -5.51507138e-02 1.00014389e+00
1.22799240e-01 -1.55599326e-01 -8.29545498e-01 1.09958661e+00
-3.83840114e-01 7.26923406e-01 4.44794059e-01 -5.57608545e-01
1.32129371e-01 -1.37819257e-02 9.04935241e-01 -8.58871400e-01
-2.20974356e-01 -2.86957830e-01 1.75899476e-01 -6.97108448e-01
-4.02624816e-01 -2.85125017e-01 -1.23237252e+00 -2.26763874e-01
-2.36639187e-01 -1.73164517e-01 5.27289629e-01 7.38807440e-01
1.98363587e-01 7.94421673e-01 4.16219920e-01 -5.64146817e-01
-6.07021272e-01 -7.32937098e-01 -6.39778495e-01 3.48446667e-01
2.20792294e-01 -5.82436740e-01 -1.13697216e-01 -4.97375727e-02]
|
[6.951810836791992, 6.215753078460693]
|
7165f2c2-ed2a-41d1-bc2a-6bc49a6a10ad
|
a-text-editing-approach-to-joint-japanese
| null | null |
https://aclanthology.org/2021.wnut-1.9
|
https://aclanthology.org/2021.wnut-1.9.pdf
|
A Text Editing Approach to Joint Japanese Word Segmentation, POS Tagging, and Lexical Normalization
|
Lexical normalization, in addition to word segmentation and part-of-speech tagging, is a fundamental task for Japanese user-generated text processing. In this paper, we propose a text editing model to solve the three task jointly and methods of pseudo-labeled data generation to overcome the problem of data deficiency. Our experiments showed that the proposed model achieved better normalization performance when trained on more diverse pseudo-labeled data.
|
['Eiichiro Sumita', 'Taro Watanabe', 'Masao Utiyama', 'Shohei Higashiyama']
| null | null | null | null |
wnut-acl-2021-11
|
['lexical-normalization']
|
['natural-language-processing']
|
[ 3.13601255e-01 -6.22954732e-03 -2.81130284e-01 -6.81295574e-01
-8.77933979e-01 -4.03097510e-01 1.71441957e-01 5.30081950e-02
-1.04655087e+00 1.08971906e+00 5.62400579e-01 -3.63304257e-01
5.60264349e-01 -5.25800586e-01 -5.55377938e-02 -3.07769746e-01
9.10590231e-01 4.86740023e-01 2.57849187e-01 -4.55816239e-01
4.50542867e-01 1.34913802e-01 -1.01843536e+00 2.45160684e-01
1.36108041e+00 5.98782338e-02 4.50160474e-01 5.93590498e-01
-9.08652782e-01 4.76465166e-01 -1.06098711e+00 -4.78985935e-01
3.06117740e-02 -6.11571908e-01 -8.65852237e-01 3.91493291e-01
-2.34374031e-02 -2.41516018e-03 1.14981443e-01 1.33717394e+00
6.08129919e-01 5.29513121e-01 5.37244976e-01 -8.04087818e-01
-8.53770614e-01 1.20606554e+00 -3.40793818e-01 3.76782507e-01
1.21428065e-01 -3.67356062e-01 8.93001258e-01 -8.59654903e-01
6.17922246e-01 9.39052463e-01 3.25731039e-01 1.21889186e+00
-8.75082731e-01 -5.44769406e-01 2.71048099e-01 -2.88099557e-01
-1.25392473e+00 -3.32108527e-01 5.49136817e-01 -1.36777192e-01
1.04528964e+00 1.90442011e-01 2.59402215e-01 8.49687040e-01
-9.87578109e-02 6.86908603e-01 5.92591345e-01 -9.85903800e-01
-2.40695626e-02 2.52539098e-01 5.83725929e-01 3.33506405e-01
5.45676947e-01 -7.50255466e-01 -1.22053213e-01 3.44611555e-02
8.25811446e-01 -3.49348009e-01 -1.32956266e-01 4.15726542e-01
-1.17893565e+00 8.56934130e-01 -5.37241936e-01 6.65496349e-01
-1.83912799e-01 -3.50729197e-01 5.39917350e-01 -1.17837951e-01
8.04653645e-01 6.59734547e-01 -1.05467606e+00 -3.96367639e-01
-7.33253360e-01 -1.10375144e-01 7.91235328e-01 1.40710473e+00
5.96608043e-01 3.78832757e-01 -3.69698048e-01 1.25671351e+00
3.16565156e-01 6.60936534e-01 1.15887249e+00 -5.01224220e-01
8.82598996e-01 7.27490723e-01 1.28550798e-01 -3.91945243e-01
-3.66863847e-01 -1.37079909e-01 -6.68678403e-01 -6.28660381e-01
4.20130819e-01 -6.95336938e-01 -1.31767654e+00 1.70202303e+00
2.87067503e-01 -4.29399699e-01 3.76418769e-01 5.10503590e-01
9.06631470e-01 9.43995178e-01 5.31031787e-01 -5.92078984e-01
1.39728773e+00 -1.11961126e+00 -1.50915432e+00 -4.56653148e-01
1.04792559e+00 -1.24863970e+00 1.34736860e+00 9.36715007e-02
-8.93558264e-01 -6.53722167e-01 -7.94862151e-01 -2.85423100e-01
-4.92553979e-01 3.76490980e-01 3.50189954e-01 9.62417603e-01
-6.33869648e-01 -3.26582044e-03 -6.79269671e-01 -6.06852412e-01
-8.47664550e-02 1.34873763e-01 -2.30858982e-01 1.98453739e-01
-1.52256906e+00 7.78319120e-01 7.79728174e-01 -1.79800447e-02
-7.93060660e-02 -1.81116700e-01 -9.67242897e-01 -1.21218488e-01
2.65856534e-01 -3.12600523e-01 1.69625688e+00 -7.56063461e-01
-1.57560587e+00 8.04828942e-01 -5.07295370e-01 1.14835978e-01
7.86409006e-02 -3.97523910e-01 -8.14915299e-01 -5.02060652e-01
2.70098686e-01 4.52754378e-01 5.43249130e-01 -9.22033668e-01
-9.18076873e-01 -3.14258933e-01 -6.35845244e-01 4.64571953e-01
-5.58296859e-01 4.32538420e-01 -6.66851401e-01 -1.14890540e+00
2.10593298e-01 -6.85412645e-01 -6.37603104e-01 -1.03683364e+00
-4.13807601e-01 -5.48946977e-01 5.13184249e-01 -9.42090869e-01
1.78966224e+00 -1.92367899e+00 -1.20121017e-01 1.74428038e-02
-2.57566214e-01 7.14559972e-01 -2.14142174e-01 3.47486645e-01
-1.12577714e-01 6.81465209e-01 -3.41462225e-01 -4.46168274e-01
-8.49428847e-02 5.13181865e-01 -7.85930455e-02 -2.53472060e-01
8.99631977e-02 9.18512464e-01 -9.70644116e-01 -7.30538130e-01
-9.56368595e-02 2.68985685e-02 -1.20562278e-01 3.66714209e-01
-2.41139337e-01 4.45464820e-01 -5.31333447e-01 4.49138701e-01
6.02313876e-01 3.89015138e-01 1.78122237e-01 7.79300630e-02
-8.13304782e-02 4.24988031e-01 -1.14783061e+00 1.78533196e+00
-2.71861702e-01 3.11739534e-01 -2.16688067e-02 -5.34783304e-01
9.15644526e-01 6.09177351e-01 1.03920065e-01 -5.84400773e-01
6.05960369e-01 2.35939398e-01 1.05549926e-02 -7.65042067e-01
1.25992906e+00 -1.80607647e-01 -4.76430595e-01 5.83997488e-01
2.60428220e-01 -2.84735888e-01 6.95194304e-01 9.45126042e-02
5.12547791e-01 1.13668740e-01 6.91948414e-01 -8.83704871e-02
5.71434855e-01 2.95179397e-01 1.00546622e+00 5.77988386e-01
-2.05922619e-01 6.62940860e-01 2.11609110e-01 -4.55609011e-03
-1.04219198e+00 -4.63505298e-01 1.44608319e-01 1.42336583e+00
-1.90573841e-01 -4.17739689e-01 -1.34112442e+00 -1.05572975e+00
-5.69665849e-01 1.24780107e+00 -3.33958298e-01 9.87841710e-02
-1.01208079e+00 -1.13847136e+00 7.59045064e-01 5.01026332e-01
8.27129185e-02 -1.29721308e+00 1.66492805e-01 5.82540512e-01
-6.35900497e-01 -1.18972552e+00 -9.74324763e-01 3.09863895e-01
-8.31177592e-01 -8.62215519e-01 -7.21447051e-01 -1.35171282e+00
1.13746142e+00 3.69964331e-01 7.51732588e-01 4.59263362e-02
2.12175578e-01 -1.99214160e-01 -8.48109782e-01 -6.14589930e-01
-8.77588093e-01 5.07258058e-01 4.51731794e-02 -1.90930352e-01
6.01021290e-01 -1.33474767e-01 1.33941650e-01 4.18166518e-01
-1.34592760e+00 8.43700543e-02 5.37384510e-01 7.46774614e-01
6.84284627e-01 6.13611005e-03 5.26505232e-01 -1.63175726e+00
1.16819859e+00 -5.51366359e-02 -3.92205954e-01 6.03547633e-01
-5.86011529e-01 1.53884754e-01 6.86768770e-01 -5.44559777e-01
-1.74801898e+00 2.16754273e-01 -6.51854753e-01 4.78436977e-01
-4.36219633e-01 5.92730582e-01 -7.76454151e-01 1.02010265e-01
5.52441895e-01 2.12320536e-01 -3.13143045e-01 -8.84143889e-01
6.61260188e-01 1.06538653e+00 5.08391619e-01 -5.40597737e-01
6.10242784e-01 -2.61898696e-01 -6.76449716e-01 -9.23045993e-01
-1.07575607e+00 -5.73528886e-01 -1.13837457e+00 2.06467137e-01
1.08169305e+00 -6.87216043e-01 2.00202763e-01 7.57022023e-01
-1.51784289e+00 -1.81621775e-01 -3.48223150e-01 6.61538005e-01
1.55169144e-01 5.57931542e-01 -6.25710845e-01 -5.84076464e-01
-4.16547388e-01 -7.74302125e-01 5.30803084e-01 6.55604601e-01
-2.61676192e-01 -1.04983032e+00 2.81736732e-01 4.47070241e-01
2.04928279e-01 -3.48161608e-01 9.45185661e-01 -1.30885601e+00
2.44771615e-01 -3.87049437e-01 -6.84020743e-02 6.28114879e-01
5.90990305e-01 2.31067941e-01 -7.10033715e-01 2.38494277e-01
1.28986999e-01 -4.11914336e-03 4.61790442e-01 2.07948565e-01
8.49609256e-01 -2.53764957e-01 6.16481602e-02 2.97344983e-01
9.46269572e-01 5.87835729e-01 4.86632109e-01 1.47171646e-01
1.15122509e+00 6.86008573e-01 8.88647437e-01 3.14278394e-01
4.45457637e-01 2.58807838e-01 -1.95257783e-01 -2.57129759e-01
-1.48179203e-01 -2.30244011e-01 7.02118576e-02 1.68912852e+00
2.46081427e-01 -7.34194040e-01 -8.86241853e-01 4.66500878e-01
-1.91480160e+00 -5.43797612e-01 -7.11111844e-01 1.79764128e+00
1.31865990e+00 2.78925747e-01 -1.55771226e-01 -9.11404341e-02
1.18024731e+00 -4.38475274e-02 -1.02950178e-01 -5.57572067e-01
-4.29134309e-01 2.15777546e-01 6.78289652e-01 6.31175220e-01
-1.05123162e+00 1.59245872e+00 7.05097485e+00 9.32190776e-01
-8.17020476e-01 3.73349845e-01 4.48512703e-01 3.92700553e-01
-4.38794702e-01 -1.59641445e-01 -1.25231373e+00 3.38092327e-01
8.48166287e-01 -3.11051995e-01 6.28687218e-02 7.92729855e-01
3.12512189e-01 -8.49951506e-02 -6.01145148e-01 6.88010514e-01
3.86464417e-01 -6.54901445e-01 4.19019252e-01 -2.57190943e-01
8.38059366e-01 -2.46935815e-01 -6.43955350e-01 3.94205034e-01
4.18540299e-01 -4.70034122e-01 5.02284288e-01 5.13289906e-02
4.92420882e-01 -6.60212815e-01 1.06947029e+00 4.71906275e-01
-9.91545796e-01 5.45612037e-01 -3.78020883e-01 -9.12317336e-02
6.80767298e-01 5.04570663e-01 -7.98786044e-01 4.45340306e-01
2.63668578e-02 3.27603847e-01 -7.54060447e-01 8.36019099e-01
-8.11548650e-01 9.48175430e-01 -1.15054045e-02 -3.07999015e-01
-6.98640123e-02 -2.05946997e-01 3.20900679e-01 1.58923304e+00
2.73055434e-01 3.62561017e-01 1.50777996e-01 2.11356267e-01
-2.13248178e-01 7.79756486e-01 -4.78845052e-02 -4.20283824e-01
5.85058093e-01 1.32323551e+00 -1.12709582e+00 -5.86248875e-01
-4.51766551e-01 9.85141873e-01 3.27778429e-01 4.05668288e-01
-7.15485036e-01 -8.97876322e-01 4.72294450e-01 -2.79486388e-01
1.03323720e-01 -4.04222608e-01 -6.98553085e-01 -1.42728567e+00
-7.73473531e-02 -7.07488358e-01 4.08066154e-01 -6.03590786e-01
-1.30030239e+00 7.98162818e-01 -1.24831893e-01 -9.33959603e-01
-3.55992131e-02 -5.94533384e-01 -5.18123627e-01 1.03068578e+00
-1.29329705e+00 -8.06595862e-01 -1.99253671e-02 2.73681998e-01
8.56556177e-01 -1.28112197e-01 9.53168273e-01 4.74486440e-01
-1.16764677e+00 7.50427127e-01 3.66518945e-02 5.68876803e-01
9.89219785e-01 -1.18081045e+00 8.72498930e-01 1.31250632e+00
2.07010567e-01 7.30847001e-01 5.08024812e-01 -1.24717677e+00
-5.20748019e-01 -1.11729324e+00 1.58702016e+00 -2.43595541e-01
4.63410169e-01 -4.46049422e-01 -1.16845667e+00 5.90257049e-01
3.72612357e-01 -4.67079818e-01 1.15674531e+00 -2.12739632e-02
2.36262992e-01 1.56453446e-01 -1.00122213e+00 5.80149710e-01
8.96141350e-01 -1.83087379e-01 -1.17005944e+00 2.23927140e-01
1.22352898e+00 -5.39676368e-01 -5.32455623e-01 1.30505279e-01
-2.69351471e-02 -9.10056457e-02 2.67894804e-01 -7.47201383e-01
-1.61029667e-01 -2.88165867e-01 -2.56350078e-02 -1.46107948e+00
-3.47249031e-01 -8.26760054e-01 5.48511624e-01 1.82604206e+00
9.18013632e-01 -3.69372070e-01 7.07711577e-01 8.29688907e-01
-4.59283620e-01 -2.17211731e-02 -6.35579467e-01 -5.08316755e-01
-1.77136108e-01 -4.89276290e-01 6.10946119e-01 1.07770300e+00
1.00031562e-01 8.36835563e-01 -3.93974960e-01 -1.39603630e-01
1.85549885e-01 -5.27354121e-01 3.89195055e-01 -1.09415650e+00
2.56225377e-01 -3.33754331e-01 2.21022919e-01 -1.09244037e+00
2.58216530e-01 -6.25415325e-01 5.50810158e-01 -1.73355067e+00
-8.65235999e-02 -2.47961923e-01 -1.74010172e-01 6.16108537e-01
-8.33844483e-01 1.61982149e-01 -7.37393424e-02 9.12585184e-02
-5.35056531e-01 6.52979970e-01 1.38774312e+00 2.22493649e-01
-6.51111007e-01 7.13960379e-02 -8.15256119e-01 6.51599586e-01
1.05540979e+00 -9.26885605e-01 -2.35498816e-01 -9.22789216e-01
1.12364814e-01 -1.76355198e-01 -8.12255859e-01 -5.87888718e-01
1.45031616e-01 -5.95986128e-01 1.99372485e-01 -6.20432675e-01
-1.39565527e-01 -5.02215862e-01 -3.04497212e-01 1.36798233e-01
-3.61672223e-01 2.81159371e-01 2.17957571e-01 1.40691161e-01
-1.64970368e-01 -7.36564338e-01 5.66388369e-01 -2.72433639e-01
-8.14173341e-01 -1.29992347e-02 -8.70339453e-01 5.40188134e-01
8.39807093e-01 -3.03444326e-01 -2.35039681e-01 -1.08009249e-01
-5.98539591e-01 2.91461855e-01 2.60890752e-01 5.45393229e-01
3.34921867e-01 -1.15235937e+00 -7.58741081e-01 3.21284980e-01
1.64265350e-01 9.37996209e-02 -4.79720309e-02 1.76085994e-01
-6.36676133e-01 5.19381762e-01 -2.10728288e-01 2.25089446e-01
-1.24917972e+00 1.47467718e-01 -5.14672697e-02 -3.52814138e-01
1.26533344e-01 9.63452995e-01 -1.92072242e-01 -8.74017179e-01
4.07422364e-01 -4.53698158e-01 -7.19892323e-01 2.44323358e-01
4.75852311e-01 3.69977087e-01 3.29916775e-01 -9.06129062e-01
-8.04805234e-02 3.15576315e-01 -2.83901989e-01 -3.56094807e-01
9.28298056e-01 -5.16228199e-01 -4.07911509e-01 3.93745333e-01
6.88746631e-01 3.85694206e-01 -6.47696078e-01 -3.99872005e-01
4.54049408e-01 -1.55513287e-01 -1.41011849e-01 -8.94163191e-01
-7.84869373e-01 6.62380338e-01 2.05278341e-02 1.45551801e-01
8.74726057e-01 -3.16352576e-01 1.19779742e+00 5.74849844e-01
5.70621677e-02 -1.78954339e+00 -2.10190699e-01 1.08599854e+00
3.22832167e-01 -1.37033784e+00 -2.54196525e-01 -7.08077788e-01
-8.90518069e-01 1.14382493e+00 1.23462057e+00 5.57713449e-01
5.46273589e-01 5.13334349e-02 7.49521673e-01 2.45581806e-01
-3.05509508e-01 -3.53289396e-01 2.72341818e-01 6.70868814e-01
8.11621964e-01 3.87039222e-02 -1.11974478e+00 1.07822371e+00
-1.75267458e-01 -2.37463593e-01 6.51151240e-01 1.23984718e+00
-6.59675896e-01 -1.84007502e+00 -4.05431896e-01 3.39035481e-01
-7.20761597e-01 -4.49674487e-01 -5.34038484e-01 5.93485534e-01
1.64029703e-01 1.13985384e+00 3.40450108e-02 -2.57001728e-01
6.27984583e-01 3.22082072e-01 9.76066366e-02 -1.38234389e+00
-6.69166267e-01 5.25945008e-01 2.66834795e-01 1.83518864e-02
-4.35824990e-01 -5.51001072e-01 -1.65736926e+00 6.02085665e-02
-5.76029181e-01 7.60249853e-01 8.22896540e-01 9.78418112e-01
5.00157289e-03 5.53294063e-01 4.98952121e-01 -4.48636383e-01
-3.24897021e-01 -1.54681981e+00 -7.08589494e-01 4.58288729e-01
-2.27711990e-01 5.93455732e-02 -1.78010762e-01 4.35223550e-01]
|
[10.219980239868164, 10.090128898620605]
|
95d4ee66-d66c-4c73-8931-6d9624424d92
|
meta-learning-pathologies-from-radiology
|
2210.13979
| null |
https://arxiv.org/abs/2210.13979v2
|
https://arxiv.org/pdf/2210.13979v2.pdf
|
Meta-learning Pathologies from Radiology Reports using Variance Aware Prototypical Networks
|
Large pretrained Transformer-based language models like BERT and GPT have changed the landscape of Natural Language Processing (NLP). However, fine tuning such models still requires a large number of training examples for each target task, thus annotating multiple datasets and training these models on various downstream tasks becomes time consuming and expensive. In this work, we propose a simple extension of the Prototypical Networks for few-shot text classification. Our main idea is to replace the class prototypes by Gaussians and introduce a regularization term that encourages the examples to be clustered near the appropriate class centroids. Experimental results show that our method outperforms various strong baselines on 13 public and 4 internal datasets. Furthermore, we use the class distributions as a tool for detecting potential out-of-distribution (OOD) data points during deployment.
|
['Benjamin Odry', 'Anasuya Das', 'Nabila Abraham', 'Kawshik Kannan', 'Arijit Sehanobish']
|
2022-10-22
| null | null | null | null |
['few-shot-text-classification']
|
['natural-language-processing']
|
[-2.21742198e-01 1.58373490e-02 -3.19021761e-01 -7.18502045e-01
-8.09476912e-01 -6.03563368e-01 6.23424947e-01 5.57874262e-01
-5.62286556e-01 4.19803679e-01 1.22981519e-01 -3.44290942e-01
-6.06056675e-02 -7.32677042e-01 -4.44726944e-01 -5.16489685e-01
-3.94633785e-02 8.31894755e-01 3.65720898e-01 -3.09215933e-01
1.13118097e-01 4.67885852e-01 -1.31042993e+00 4.84557182e-01
4.71316308e-01 1.01812804e+00 5.34886569e-02 4.43361819e-01
-6.93391263e-01 8.26100826e-01 -1.05669570e+00 -5.09151876e-01
2.16332540e-01 6.15411513e-02 -5.82922459e-01 -3.71179521e-01
8.43190029e-02 -1.72958836e-01 -1.48752272e-01 9.28777695e-01
8.00225198e-01 5.53159475e-01 7.64240146e-01 -1.49808919e+00
-6.04498208e-01 8.71635854e-01 -6.82098866e-01 3.74873310e-01
-2.06020266e-01 -2.24959105e-02 9.72153366e-01 -1.06181860e+00
3.44717592e-01 1.34196305e+00 9.29322422e-01 5.06926358e-01
-1.17125106e+00 -8.16048622e-01 2.45377377e-01 7.11705089e-02
-1.58139706e+00 -4.95999455e-01 5.99682510e-01 -3.59874070e-01
1.22560477e+00 -1.48459584e-01 3.83721218e-02 1.36111879e+00
2.54969820e-02 9.82892215e-01 3.25252682e-01 -4.75173444e-01
5.37104309e-01 4.18673098e-01 4.79164392e-01 5.11828184e-01
1.90262478e-02 -3.37844908e-01 -4.20592785e-01 -6.04610205e-01
1.40502185e-01 2.03098759e-01 1.62324980e-01 -4.36628163e-01
-6.51331842e-01 1.08582675e+00 2.50376999e-01 5.84715486e-01
-2.82117844e-01 1.24658771e-01 5.30061483e-01 -2.59228237e-02
8.91989410e-01 4.48384404e-01 -5.63407242e-01 -3.78971964e-01
-1.11481249e+00 2.15859748e-02 9.80604112e-01 1.11812806e+00
7.88969636e-01 -1.90403730e-01 -7.17520416e-01 1.18715250e+00
7.49014169e-02 1.31977290e-01 4.78778988e-01 -4.29773182e-01
6.62186801e-01 5.98180056e-01 7.89559782e-02 -1.09720278e+00
-3.97061944e-01 -4.75462377e-01 -6.70867145e-01 -2.49719009e-01
2.75901318e-01 -3.67812246e-01 -1.00183845e+00 1.39324808e+00
1.30379662e-01 6.74555451e-02 -8.51056054e-02 5.55800200e-01
5.28327286e-01 9.84042704e-01 4.60908562e-01 1.01051979e-01
1.14475989e+00 -9.29154754e-01 -6.87414169e-01 -5.88906050e-01
8.99794817e-01 -5.29666662e-01 1.52414858e+00 1.73039809e-01
-6.85142279e-01 -3.55183303e-01 -7.20378697e-01 -5.87627701e-02
-7.76674211e-01 3.84174794e-01 4.99375403e-01 4.74700570e-01
-8.07970464e-01 4.67697620e-01 -6.09705806e-01 -5.77054739e-01
7.39417315e-01 1.84402645e-01 5.57027757e-02 -8.16832297e-03
-1.23270285e+00 6.25184059e-01 3.34985107e-01 -4.15306678e-03
-9.44029927e-01 -9.35756862e-01 -6.91508234e-01 4.04599369e-01
3.39270204e-01 -1.70992434e-01 1.38972723e+00 -4.67552513e-01
-1.14752305e+00 9.11258578e-01 -1.34997323e-01 -5.13381243e-01
2.74637908e-01 -3.13456088e-01 -3.75003487e-01 -9.08621326e-02
3.08412820e-01 6.46115065e-01 8.78458679e-01 -1.05885160e+00
-7.52717614e-01 -1.38952732e-01 4.83162776e-02 -3.48241441e-02
-7.93994725e-01 4.33466583e-01 -5.01587033e-01 -5.28549552e-01
-4.79669571e-01 -4.19380367e-01 -2.71966130e-01 -9.02704448e-02
-5.22253156e-01 -6.45391226e-01 8.19008827e-01 -4.66993958e-01
1.51549053e+00 -2.31784344e+00 -4.42614466e-01 1.98327407e-01
3.43423456e-01 3.31122518e-01 -1.88322946e-01 4.13962662e-01
5.35169989e-02 4.18065906e-01 4.55059446e-02 -5.22062719e-01
3.03218067e-01 2.44490832e-01 -5.71026027e-01 2.45780408e-01
2.03694806e-01 9.34445024e-01 -1.11434901e+00 -4.31776285e-01
6.56920001e-02 3.83519441e-01 -3.43755394e-01 2.25551680e-01
-3.94210488e-01 -1.48957878e-01 -4.32180136e-01 4.50357348e-01
5.19735694e-01 -2.88607001e-01 -1.71287775e-01 4.00190726e-02
6.01542331e-02 4.93760496e-01 -7.79731333e-01 1.61282694e+00
-4.48469639e-01 7.23794281e-01 -1.08397603e-01 -1.14266813e+00
1.06993568e+00 9.22685191e-02 2.38509506e-01 -3.34174514e-01
3.80632311e-01 -2.89055645e-01 -3.50303739e-01 -5.48585057e-01
5.02536774e-01 -2.03205273e-01 -4.51115876e-01 5.56668282e-01
2.08003566e-01 8.82029086e-02 3.61893803e-01 2.98748910e-01
1.27361834e+00 -1.79745346e-01 1.55145586e-01 -1.34430677e-01
1.36239901e-01 1.69011831e-01 6.24178052e-01 1.04316115e+00
-2.83385664e-01 5.60007155e-01 7.18589246e-01 -5.69068789e-01
-8.95261109e-01 -8.29578638e-01 -3.44504379e-02 1.67139554e+00
-1.83527440e-01 -6.01513505e-01 -5.36120057e-01 -1.10747588e+00
-1.38446271e-01 1.28088641e+00 -5.72279692e-01 -3.70612770e-01
-2.69359320e-01 -8.88508201e-01 1.00188410e+00 6.89175904e-01
1.92573398e-01 -9.48393106e-01 -4.13982034e-01 3.18339974e-01
-4.08531249e-01 -1.02603877e+00 -4.35772985e-01 7.32185602e-01
-3.53969038e-01 -7.84795582e-01 -4.53453124e-01 -8.59709501e-01
5.66372454e-01 2.65920013e-01 1.26924276e+00 -2.25347236e-01
-5.29201865e-01 1.25199556e-01 -4.94228005e-01 -6.96531653e-01
-4.25486207e-01 3.28682959e-01 3.07716839e-02 -1.02196284e-01
1.02119470e+00 -4.19658780e-01 -1.27455413e-01 2.14877829e-01
-7.30967760e-01 -2.90297627e-01 2.73646742e-01 7.46503174e-01
3.76901150e-01 2.77417302e-01 4.91006374e-01 -8.80236804e-01
1.11047125e+00 -7.12460697e-01 -4.96636122e-01 4.96896684e-01
-3.71105373e-01 -1.89244002e-02 7.80549765e-01 -6.81340873e-01
-1.12672412e+00 -2.10918054e-01 -3.52474861e-02 -7.16782033e-01
-4.17142570e-01 5.58422804e-01 8.52421597e-02 3.84614676e-01
1.14803481e+00 1.97755024e-02 -4.21539217e-01 -6.00227833e-01
3.97226334e-01 9.40374553e-01 1.80690527e-01 -5.38126230e-01
1.00255167e+00 3.80730152e-01 -5.16475856e-01 -9.19167817e-01
-1.32805216e+00 -8.17717373e-01 -7.63108671e-01 6.92932084e-02
7.37032175e-01 -8.33371639e-01 -3.03105712e-01 1.66856736e-01
-1.32846916e+00 -5.08621931e-01 -6.05363965e-01 4.21698764e-02
-1.61039725e-01 2.48532981e-01 -5.30091882e-01 -9.91642177e-01
-5.33823907e-01 -5.91184795e-01 1.31781745e+00 1.06427856e-01
-3.35503131e-01 -1.00869989e+00 -3.83344130e-03 -5.92206568e-02
5.58551788e-01 -2.38857567e-01 1.09340918e+00 -1.20642805e+00
9.72259268e-02 -6.72742248e-01 -2.37277687e-01 3.09442133e-01
-2.29135174e-02 -8.30404833e-02 -1.07248557e+00 -2.26472780e-01
-1.07800486e-02 -4.60217476e-01 5.21658897e-01 3.37919474e-01
1.47753417e+00 -8.97953585e-02 -7.28076100e-01 4.15665418e-01
9.14606273e-01 -6.96161855e-03 3.49242181e-01 2.31570140e-01
5.51278591e-01 7.33076632e-01 8.04303050e-01 6.36824250e-01
1.69807732e-01 4.28614944e-01 1.36390343e-01 -1.15052257e-02
2.57024825e-01 -4.96067673e-01 4.00412440e-01 3.10770452e-01
5.32525361e-01 -7.84496307e-01 -1.32551813e+00 7.37974346e-01
-1.93936968e+00 -8.56158078e-01 7.24158287e-02 1.96526229e+00
6.80741966e-01 2.14876160e-01 -4.43193968e-03 -2.14743361e-01
9.72184598e-01 9.60388407e-02 -4.41740453e-01 -2.02036887e-01
2.76984647e-02 2.85657048e-01 3.75815570e-01 7.78108314e-02
-1.32922792e+00 1.15058815e+00 6.39172935e+00 1.15581930e+00
-9.38561440e-01 3.05163950e-01 8.22867811e-01 -4.68792826e-01
1.78971723e-01 -5.78647964e-02 -1.25902212e+00 4.27028120e-01
1.18214977e+00 -4.11528796e-01 2.98390090e-01 1.18099248e+00
2.88584352e-01 1.44010544e-01 -1.03953075e+00 9.50295091e-01
1.62911683e-01 -1.16591346e+00 -1.25390679e-01 -2.24563226e-01
3.74557376e-01 4.77146506e-01 -1.70544580e-01 8.96123707e-01
6.76737845e-01 -8.06728840e-01 6.00814939e-01 1.36030480e-01
5.19587874e-01 -6.78403854e-01 6.83153510e-01 8.04259777e-01
-9.32548761e-01 -2.75756657e-01 -8.29979718e-01 1.60848439e-01
1.75020009e-01 6.49581730e-01 -1.07670856e+00 7.19639659e-02
1.03440750e+00 5.74978948e-01 -7.00699985e-01 1.09100807e+00
-3.46573859e-01 7.36748278e-01 -4.94411826e-01 -2.77276367e-01
5.20183146e-01 1.63113952e-01 4.61219013e-01 1.42878711e+00
2.21344918e-01 -1.78382784e-01 3.01530123e-01 1.14205933e+00
-2.46578187e-01 1.68124154e-01 -6.47681653e-01 -3.00375849e-01
5.99359989e-01 1.53393984e+00 -7.40007460e-01 -5.14132440e-01
-2.55082309e-01 7.89844692e-01 4.81783509e-01 4.37301546e-01
-1.04450607e+00 -7.78075993e-01 5.69121540e-01 9.15889964e-02
2.44417727e-01 5.58799617e-02 1.97229479e-02 -1.15935397e+00
1.19085833e-02 -4.64441508e-01 6.27383709e-01 -7.07763374e-01
-1.78178632e+00 5.63360393e-01 5.37216477e-02 -1.10724235e+00
-3.01818371e-01 -5.33656359e-01 -8.86323154e-01 7.61822343e-01
-1.33984721e+00 -9.93941545e-01 -2.51205742e-01 3.12746465e-01
6.68475926e-01 -2.18973130e-01 8.37589324e-01 4.43296134e-01
-5.54897904e-01 6.47631168e-01 1.86415270e-01 4.90631431e-01
8.19920123e-01 -1.15638793e+00 6.57130182e-01 7.80265749e-01
1.20152816e-01 5.64548671e-01 6.29783511e-01 -4.17664677e-01
-7.83393085e-01 -1.32129419e+00 1.02797210e+00 -6.48845255e-01
1.04777658e+00 -9.77419674e-01 -8.99730504e-01 7.48062670e-01
1.95128378e-02 1.20401554e-01 8.73194098e-01 4.06512409e-01
-3.76354784e-01 -7.97515810e-02 -1.04223597e+00 3.89570862e-01
9.11272645e-01 -5.18185258e-01 -6.50178313e-01 7.55959153e-01
8.28997433e-01 1.74665581e-02 -5.18390000e-01 4.25393023e-02
8.93588662e-02 -5.92688382e-01 8.40236604e-01 -8.87534976e-01
1.44122690e-01 -1.53599456e-01 -9.05230176e-03 -1.29865968e+00
-3.21403325e-01 -5.44476807e-01 -1.24991819e-01 1.69382274e+00
5.43609500e-01 -3.82239759e-01 7.64141500e-01 8.69072199e-01
-1.71296671e-01 -5.09739816e-01 -7.39811182e-01 -9.82039869e-01
1.15045533e-01 -6.76105082e-01 4.80472028e-01 1.04132724e+00
2.23046318e-01 7.85205543e-01 -3.02063406e-01 1.70281097e-01
5.26619256e-01 -1.67105928e-01 7.83272684e-01 -1.33403754e+00
-1.63563415e-01 -3.50212336e-01 -2.11743996e-01 -1.10336304e+00
4.01423931e-01 -9.07981575e-01 4.49262589e-01 -1.41518831e+00
1.54076755e-01 -8.52544904e-01 -3.19413811e-01 9.17839289e-01
8.40745345e-02 -1.60561457e-01 -1.37209132e-01 1.34121016e-01
-9.48819816e-01 5.70887983e-01 3.33554298e-01 -3.95563483e-01
-3.39624673e-01 2.33394191e-01 -6.81076705e-01 8.26157331e-01
9.31120336e-01 -8.88352096e-01 -3.50026488e-01 -6.20547831e-01
3.29021424e-01 -4.25410450e-01 1.65686801e-01 -9.63087499e-01
4.51546460e-01 -1.96949113e-02 2.05681518e-01 -8.26220870e-01
2.54153937e-01 -8.28974485e-01 -4.82803702e-01 -6.72886148e-02
-6.11391187e-01 -1.36023685e-01 3.74914378e-01 6.74021780e-01
-7.33343810e-02 -6.13659441e-01 6.64693594e-01 2.74215695e-02
-6.62158132e-01 3.74678701e-01 -6.16429925e-01 3.11895669e-01
1.12054133e+00 1.13458671e-01 -3.72602403e-01 -2.94470072e-01
-6.82677329e-01 4.72639710e-01 8.65626559e-02 8.19124341e-01
3.25637758e-01 -9.28113163e-01 -3.08389008e-01 1.10169709e-01
4.90973175e-01 2.63457187e-02 2.08948180e-02 5.83492935e-01
-1.54010087e-01 2.68343180e-01 3.34858298e-01 -6.22082233e-01
-1.05924702e+00 6.80282116e-01 4.19756383e-01 -4.49623436e-01
-6.15905821e-01 9.71765697e-01 2.27417156e-01 -7.81498015e-01
6.10438704e-01 -1.99509963e-01 -2.33961627e-01 1.60461023e-01
6.61555767e-01 1.42603949e-01 5.91187030e-02 -2.84987539e-01
-3.42053026e-01 -1.32133305e-01 -2.76135087e-01 2.07308173e-01
1.59807575e+00 -1.04435906e-01 -2.20241725e-01 7.21779168e-01
1.14583218e+00 -3.41928393e-01 -1.00064051e+00 -3.97998244e-01
3.52660567e-01 -2.52373338e-01 1.19780950e-01 -7.96166003e-01
-6.45647228e-01 1.32249928e+00 4.98245478e-01 3.61650527e-01
7.95522988e-01 2.42015347e-01 6.19083643e-01 7.65198946e-01
3.28880042e-01 -1.36192584e+00 4.66091335e-02 7.90010273e-01
4.02872264e-01 -1.07404602e+00 -4.54860806e-01 -2.69743711e-01
-6.65458262e-01 1.12005627e+00 7.57219613e-01 1.54041052e-01
8.27044308e-01 4.62329924e-01 1.08343966e-01 -4.96786445e-01
-9.71735716e-01 -1.33271977e-01 -5.40339313e-02 6.00171149e-01
4.21597421e-01 -2.02959761e-01 1.41213432e-01 9.35776412e-01
-4.22964944e-03 2.03849524e-02 1.90728977e-01 1.07049429e+00
-5.78713238e-01 -1.04951084e+00 -1.06184699e-01 7.30850339e-01
-5.75723350e-01 -5.00433624e-01 -3.09193611e-01 5.81805766e-01
2.48675197e-02 1.02129900e+00 2.56021202e-01 -3.06706190e-01
4.15141493e-01 5.08583069e-01 -1.61948830e-01 -1.12344098e+00
-6.88502848e-01 -1.38060316e-01 1.12644970e-01 -3.00434560e-01
9.08367932e-02 -4.21896905e-01 -1.30843139e+00 -1.06674381e-01
-4.62088883e-01 2.73650855e-01 5.97909749e-01 8.54699492e-01
6.81663632e-01 5.01653552e-01 3.79949659e-01 -7.36623704e-01
-8.25196803e-01 -1.12588096e+00 -5.37105143e-01 3.49811107e-01
-1.21174254e-01 -6.77776277e-01 -5.30903578e-01 2.92810351e-02]
|
[10.4635648727417, 7.350512981414795]
|
a9ae8ae6-cfac-40a6-8a9b-08ab35dc89d0
|
yake-keyword-extraction-from-single-documents
| null | null |
https://repositorio.inesctec.pt/server/api/core/bitstreams/ef121a01-a0a6-4be8-945d-3324a58fc944/content
|
https://repositorio.inesctec.pt/server/api/core/bitstreams/ef121a01-a0a6-4be8-945d-3324a58fc944/content
|
YAKE! Keyword extraction from single documents using multiple local features
|
In this paper, we present YAKE!, a novel feature-based system for
multi-lingual keyword extraction from single documents, which supports texts
of different sizes, domains or languages. Unlike most systems, YAKE! does not
rely on dictionaries or thesauri, neither it is trained against any corpora. Instead,
we follow an unsupervised approach which builds upon features extracted from
the text, making it thus applicable to documents written in many different languages without the need for external knowledge. This can be beneficial for a large number of tasks and a plethora of situations where the access to training
corpora is either limited or restricted. In this demo, we offer an easy to use,
interactive session, where users from both academia and industry can try our
system, either by using a sample document or by introducing their own text. As
an add-on, we compare our extracted keywords against the output produced by
the IBM Natural Language Understanding (IBM NLU) and Rake system.
YAKE! demo is available at http://bit.ly/YakeDemoECIR2018. A python
implementation of YAKE! is also available at PyPi repository (https://pypi.
python.org/pypi/yake/).
|
['A. Jatowt', 'C. Nunes', 'A. Jorge', 'Arian Pasquali', 'Vítor Mangaravite', 'Ricardo Campos']
|
2018-03-01
| null | null | null |
ecir-2018-2018-3
|
['keyword-extraction']
|
['natural-language-processing']
|
[-3.57338399e-01 -1.55971006e-01 -2.53776401e-01 -1.85879514e-01
-9.60480213e-01 -9.34443116e-01 8.12479079e-01 2.96049058e-01
-6.70875072e-01 7.35597670e-01 1.52750790e-01 -6.19990468e-01
9.21302065e-02 -6.07757270e-01 -4.34749216e-01 -4.00909901e-01
3.29126626e-01 5.89458168e-01 2.65813291e-01 -4.58314598e-01
3.43785316e-01 2.47396797e-01 -1.42919528e+00 2.98831224e-01
7.42764652e-01 5.55931032e-01 7.04389989e-01 7.13625491e-01
-5.65364659e-01 4.94909942e-01 -3.54753971e-01 -3.95596981e-01
1.86038718e-01 -1.13262527e-01 -8.95362973e-01 -2.97498643e-01
1.76503919e-02 8.79878923e-03 5.74851930e-02 9.05842185e-01
4.02512312e-01 -1.61083728e-01 5.30525744e-01 -9.31037664e-01
-5.84998667e-01 9.69560087e-01 -1.03515700e-01 1.15121767e-01
6.65713191e-01 5.38458824e-02 1.04596615e+00 -1.22264814e+00
6.79005682e-01 1.01191640e+00 3.99367452e-01 4.37316626e-01
-9.67957735e-01 -8.00823271e-01 -2.29644179e-01 1.93566144e-01
-1.50183952e+00 -6.56620383e-01 6.09545648e-01 -4.14584726e-01
1.13743579e+00 2.52931029e-01 1.68667361e-01 1.04132318e+00
-2.30672672e-01 9.29448724e-01 1.13499081e+00 -1.08563685e+00
-4.73975390e-02 8.45244467e-01 2.31231540e-01 4.81505036e-01
-1.15257189e-01 -3.91187400e-01 -4.41738844e-01 -3.35492551e-01
3.80190611e-01 -3.37436348e-01 -5.68944156e-01 -1.10847391e-01
-1.30234373e+00 7.83378363e-01 -3.41220558e-01 7.57295489e-01
-2.40477771e-01 -2.46296123e-01 6.49278939e-01 5.49855769e-01
2.88510114e-01 6.53590381e-01 -8.52232099e-01 -3.54781151e-01
-9.30238545e-01 2.16841310e-01 1.20723593e+00 1.03378999e+00
7.35917091e-01 -4.38576788e-01 5.03154576e-01 1.05122960e+00
3.40032279e-01 5.94866514e-01 9.58126426e-01 -5.49400687e-01
4.08746511e-01 5.36704123e-01 2.16268227e-01 -7.94725776e-01
-1.32701978e-01 6.13559708e-02 -3.72893095e-01 -1.80719987e-01
4.57612485e-01 -2.00812295e-01 -6.71051860e-01 1.22987592e+00
3.39588374e-01 -5.88881552e-01 2.66312033e-01 4.46339518e-01
1.05942976e+00 9.00231659e-01 -1.34354979e-01 -3.03073019e-01
1.49923861e+00 -9.25810575e-01 -8.57755780e-01 -1.88445270e-01
9.96644974e-01 -1.28164446e+00 1.22432768e+00 6.57867968e-01
-1.00968397e+00 -2.35614017e-01 -8.32393527e-01 -1.82846278e-01
-1.02188957e+00 3.36589783e-01 3.49924952e-01 5.67326546e-01
-1.05387318e+00 4.76740748e-01 -6.10975564e-01 -6.75963223e-01
-1.52863890e-01 1.45981267e-01 -4.07215744e-01 -6.79252371e-02
-1.47910929e+00 1.01620591e+00 5.84618807e-01 -3.08697641e-01
-2.84366310e-01 -3.21329504e-01 -8.66203010e-01 -2.18217224e-01
8.31203163e-01 -2.09216908e-01 1.53691244e+00 -8.93007994e-01
-1.44548059e+00 8.43211412e-01 -3.67488861e-01 -3.09105545e-01
4.09896821e-01 -5.02783895e-01 -6.14195645e-01 -8.01736563e-02
8.42920095e-02 1.33506984e-01 5.62158287e-01 -9.29082453e-01
-4.78163183e-01 1.08884368e-03 -7.94156790e-02 1.29250884e-01
-3.17577243e-01 7.22817302e-01 -8.27261925e-01 -5.31635821e-01
-5.80100954e-01 -7.88526356e-01 -7.46413618e-02 -4.42635149e-01
-4.87536341e-01 -5.51994026e-01 7.11751521e-01 -8.64881694e-01
1.55406547e+00 -2.05721712e+00 -6.54589161e-02 3.53679895e-01
-1.76918417e-01 5.85488796e-01 1.33125931e-01 1.16749477e+00
1.27074510e-01 4.55914319e-01 -1.51875183e-01 -1.25099033e-01
9.77229849e-02 6.51488453e-02 -2.72276461e-01 5.60642034e-02
-1.08089387e-01 7.51125932e-01 -9.32679594e-01 -4.94881630e-01
2.80484498e-01 2.60192454e-01 -1.92787290e-01 8.49894956e-02
-2.41735235e-01 1.79972067e-01 -5.07661939e-01 5.54240048e-01
2.56242335e-01 -2.99743354e-01 2.98874348e-01 1.76536888e-01
-5.24306238e-01 7.23431408e-01 -1.26958048e+00 1.66995656e+00
-9.28507149e-01 5.80035269e-01 -5.49400225e-02 -8.25164020e-01
8.66294444e-01 6.86410844e-01 1.56198248e-01 -3.68903220e-01
1.97737619e-01 7.28598177e-01 -3.02550077e-01 -3.29152912e-01
6.90181732e-01 2.71640778e-01 -2.51534283e-01 7.93698192e-01
1.97986946e-01 -2.77126044e-01 6.75288141e-01 4.70382750e-01
8.87407064e-01 7.48945698e-02 8.77810180e-01 -4.32099462e-01
6.97579741e-01 2.74714202e-01 1.32352829e-01 5.84449410e-01
2.40983605e-01 1.61509380e-01 1.57198682e-01 -1.83130279e-01
-1.18104899e+00 -5.80671370e-01 -3.98618758e-01 1.09081292e+00
-2.15412751e-01 -1.01522768e+00 -6.48060322e-01 -6.04113519e-01
-7.53508210e-02 8.66321087e-01 -1.62687674e-01 3.58111948e-01
-5.01625240e-01 -2.82495469e-01 3.99801940e-01 2.97135673e-02
1.53803840e-01 -1.36875451e+00 -3.23211670e-01 2.74050057e-01
-2.89711982e-01 -1.08219111e+00 -5.09685934e-01 2.76539981e-01
-3.13541502e-01 -8.54919195e-01 -6.50893271e-01 -8.04693580e-01
2.74292558e-01 -2.57495251e-02 1.15823495e+00 2.30615456e-02
-1.06200092e-01 5.51541686e-01 -7.84944177e-01 -6.03194416e-01
-7.47303605e-01 3.43678266e-01 1.54037222e-01 -4.70273882e-01
9.63912427e-01 -4.29801553e-01 -2.51773775e-01 3.65551651e-01
-9.65454221e-01 4.46817800e-02 4.39817190e-01 6.10796034e-01
3.20105344e-01 -7.89312199e-02 7.01431990e-01 -1.00554848e+00
7.74151385e-01 -5.49397886e-01 -6.34690464e-01 3.89307410e-01
-8.08286786e-01 1.09730847e-01 8.48068655e-01 -3.93397957e-01
-6.77011967e-01 -5.32564446e-02 -3.90087247e-01 -3.62600684e-02
-3.32105279e-01 9.32362139e-01 -3.51890117e-01 3.04300964e-01
6.28421485e-01 4.32948261e-01 -1.05495401e-01 -8.50598872e-01
6.27766907e-01 1.32961380e+00 3.37189138e-01 -3.75157595e-01
7.94279099e-01 6.86986884e-03 -9.11962628e-01 -9.86149669e-01
-4.77455288e-01 -9.17224824e-01 -7.22452581e-01 1.23522738e-02
4.66748953e-01 -7.96082497e-01 -2.49937534e-01 4.45096195e-01
-9.94295239e-01 -3.73040974e-01 -9.20301378e-02 4.00556535e-01
-2.17151865e-01 4.96004254e-01 -2.79777825e-01 -4.92096484e-01
-5.58758020e-01 -8.51280749e-01 7.55484879e-01 2.31229380e-01
-4.88497049e-01 -1.16874707e+00 2.71672904e-02 2.15468422e-01
3.83767784e-01 -2.59652942e-01 6.47152781e-01 -1.21238446e+00
-1.46578699e-01 -4.54061836e-01 9.32306498e-02 5.35766363e-01
4.29554850e-01 1.69374332e-01 -8.37100625e-01 -1.86596408e-01
-1.83154330e-01 -6.15161002e-01 4.05883849e-01 -2.61441842e-02
8.58532012e-01 -6.00013018e-01 -3.14098001e-01 1.36744842e-01
1.38129091e+00 2.34939605e-01 3.55036795e-01 7.23775208e-01
4.98678237e-01 6.80413485e-01 5.90446949e-01 4.37804997e-01
5.11901736e-01 7.32111454e-01 -1.92665979e-01 1.02891468e-01
1.87340721e-01 -1.44684315e-01 5.01407862e-01 1.23478734e+00
2.30720878e-01 -3.13229918e-01 -1.23172438e+00 8.71162474e-01
-1.75440884e+00 -6.73903525e-01 -5.89347221e-02 2.14189649e+00
1.31581807e+00 7.24421069e-02 1.06532902e-01 -6.11697249e-02
4.21330363e-01 -8.48052576e-02 -8.12588111e-02 -7.28860199e-01
-3.09710335e-02 3.75351012e-01 4.90888804e-01 5.25741398e-01
-9.33509529e-01 1.15484786e+00 5.29057693e+00 1.17848754e+00
-1.30575919e+00 6.45735189e-02 3.55389714e-03 6.21176213e-02
-3.45305264e-01 1.98278397e-01 -1.10362697e+00 5.03010213e-01
1.33867931e+00 -8.76589060e-01 4.41739887e-01 1.00261688e+00
2.34863013e-01 -2.13729694e-01 -9.74060893e-01 8.05904984e-01
1.35049531e-02 -1.23971033e+00 -2.95051277e-01 5.05820699e-02
9.44021344e-02 4.51853961e-01 -3.23423237e-01 3.54105800e-01
5.86455107e-01 -9.30730462e-01 5.78693390e-01 1.93747312e-01
8.96972775e-01 -6.97192490e-01 7.05655456e-01 7.97046185e-01
-1.08897293e+00 3.34659904e-01 -1.22092679e-01 1.12570882e-01
2.76391864e-01 5.80850244e-01 -1.06711853e+00 7.48758674e-01
8.02937031e-01 5.00539541e-01 -5.45481026e-01 7.32887983e-01
-3.29838187e-01 6.38142943e-01 -5.20471096e-01 -3.22684288e-01
1.54361099e-01 -2.10518967e-02 4.85807270e-01 1.74622548e+00
2.76066035e-01 -4.79986742e-02 3.73199821e-01 2.87764847e-01
-3.39645371e-02 8.99674952e-01 -7.83999681e-01 -3.59103441e-01
5.65359890e-01 1.43612146e+00 -4.81110930e-01 -5.59008718e-01
-5.97306430e-01 8.08081388e-01 1.39690503e-01 2.43054762e-01
-3.24842930e-01 -8.17077041e-01 2.84278065e-01 1.35872990e-01
4.96496290e-01 -4.37204331e-01 1.93400592e-01 -1.41880715e+00
1.36837319e-01 -1.30533838e+00 3.06941777e-01 -6.70606077e-01
-1.17274272e+00 7.83660710e-01 2.09499851e-01 -1.12154269e+00
-8.17786217e-01 -6.91859901e-01 -2.17556521e-01 1.15664661e+00
-1.49950624e+00 -8.53981555e-01 2.52737492e-01 5.64008772e-01
6.85585320e-01 -2.29682833e-01 1.08190548e+00 4.20306355e-01
-1.84002206e-01 4.69527364e-01 4.70716059e-01 3.37256044e-01
1.00529444e+00 -1.28899789e+00 4.56597060e-01 6.46230400e-01
4.48983163e-01 9.75914419e-01 8.08429182e-01 -5.91594279e-01
-1.22475278e+00 -6.99952483e-01 1.60619020e+00 -6.14356577e-01
1.27130294e+00 -5.28402150e-01 -1.09196782e+00 6.25796080e-01
5.88470817e-01 -2.76942492e-01 9.98229742e-01 2.82108575e-01
-2.66160548e-01 1.42068863e-01 -8.31517816e-01 5.13794303e-01
4.94308978e-01 -6.14453018e-01 -7.75116920e-01 4.25088435e-01
5.28578699e-01 -2.36530781e-01 -9.40628648e-01 5.99847846e-02
5.31340539e-01 -5.24581850e-01 5.34398913e-01 -2.51147062e-01
3.25625390e-01 -3.09603155e-01 -1.72088236e-01 -1.31996548e+00
8.36294964e-02 -6.74190342e-01 5.27693778e-02 1.50982118e+00
8.19629431e-01 -7.53162920e-01 1.35599226e-01 4.23571318e-01
1.07316308e-01 -6.09186053e-01 -7.30714083e-01 -9.22006905e-01
2.04514414e-01 -8.11996639e-01 4.77441818e-01 1.06210768e+00
4.47348803e-01 3.96209598e-01 -3.33134919e-01 1.03020086e-03
1.71814173e-01 8.17885324e-02 1.00214970e+00 -8.66203010e-01
-4.55803663e-01 -2.68615961e-01 1.82154179e-02 -1.01449358e+00
2.61061750e-02 -9.83307719e-01 4.16925289e-02 -1.39947462e+00
7.40323588e-02 -5.05813539e-01 -4.50781286e-02 8.68931472e-01
1.68243468e-01 8.29287618e-02 1.05123013e-01 5.30610263e-01
-3.22680771e-01 2.35947162e-01 8.13359141e-01 1.01619169e-01
-3.27719659e-01 -7.07175285e-02 -7.77036309e-01 8.54770184e-01
9.57652807e-01 -5.36578119e-01 -1.67105779e-01 -2.77956665e-01
2.50790060e-01 -3.16396654e-01 -4.50775772e-02 -5.60857773e-01
2.64254123e-01 -1.02631055e-01 -1.36756510e-01 -3.87276918e-01
-3.54879759e-02 -6.11998975e-01 -2.06361134e-02 1.36634437e-02
-6.95727244e-02 1.96732998e-01 3.17915171e-01 -2.36817803e-02
-3.42040122e-01 -7.31719851e-01 3.64918739e-01 -5.35014510e-01
-8.19836199e-01 -9.75728482e-02 -5.20625353e-01 -1.67708006e-02
7.43074179e-01 3.60189425e-03 -2.24571824e-01 -3.36403638e-01
-5.96243441e-01 3.69245768e-01 7.57075787e-01 5.11034906e-01
3.03920060e-01 -9.91838932e-01 -6.08010411e-01 -1.73653401e-02
4.62007046e-01 -2.51634210e-01 -2.43087620e-01 6.35143578e-01
-4.61962342e-01 6.99905634e-01 2.21385539e-01 -2.26451740e-01
-1.34218311e+00 4.81800824e-01 3.25163803e-03 -3.94336969e-01
-6.22299552e-01 6.54119074e-01 -2.42871176e-02 -6.90925896e-01
4.68484536e-02 -1.37917966e-01 -5.32443583e-01 1.12846076e-01
8.24552655e-01 -2.33263671e-01 3.20628673e-01 -6.59371912e-01
-5.06052911e-01 2.32250914e-01 -4.49915618e-01 -4.30504173e-01
1.37248456e+00 -3.55965346e-01 -5.02153151e-02 7.65523970e-01
9.75526571e-01 5.09234428e-01 -4.14822221e-01 -6.08392477e-01
4.97585595e-01 -3.93912464e-01 -8.40853713e-03 -9.49243784e-01
-3.80201161e-01 5.00060439e-01 1.45766079e-01 4.02105004e-01
1.10039496e+00 3.87686461e-01 7.75221527e-01 5.91864049e-01
4.14311528e-01 -1.14039290e+00 -4.29182529e-01 7.14049578e-01
9.19481874e-01 -1.40636611e+00 1.58050470e-02 -1.13546617e-01
-8.43644917e-01 1.14737213e+00 1.05253525e-01 2.12759465e-01
8.72819722e-01 3.18732530e-01 5.60645938e-01 -2.18327060e-01
-8.48872006e-01 -1.95987061e-01 3.77875865e-01 2.79947847e-01
7.73132503e-01 -6.03448264e-02 -8.61469209e-01 5.87657273e-01
-5.76810837e-01 1.05775364e-01 7.31201172e-01 1.23561239e+00
-3.54257733e-01 -1.82319224e+00 -3.15446973e-01 3.78548712e-01
-8.69582355e-01 -5.79842210e-01 -5.29293716e-01 8.79895151e-01
-9.68583673e-02 8.64168942e-01 -4.19921815e-01 -9.21157077e-02
1.43400311e-01 4.32380527e-01 1.31532297e-01 -1.05928648e+00
-5.75523376e-01 2.86517888e-01 4.61168110e-01 -2.46710509e-01
-4.02407467e-01 -9.00424063e-01 -9.97281790e-01 -1.48610443e-01
-4.55224484e-01 7.04526007e-01 9.91577804e-01 9.93167818e-01
1.66682452e-01 -1.47635669e-01 5.47445834e-01 -5.79521000e-01
-2.43857205e-01 -1.15046215e+00 -5.27514875e-01 3.15188840e-02
1.46864936e-01 -3.28627855e-01 -4.36208546e-01 2.66072869e-01]
|
[10.203770637512207, 9.687365531921387]
|
c7d0b6bb-8bf9-484a-94ba-cb180073628e
|
navigation-as-the-attacker-wishes-towards
|
2211.14769
| null |
https://arxiv.org/abs/2211.14769v3
|
https://arxiv.org/pdf/2211.14769v3.pdf
|
Navigation as Attackers Wish? Towards Building Byzantine-Robust Embodied Agents under Federated Learning
|
Federated embodied agent learning protects the data privacy of individual visual environments by keeping data locally at each client (the individual environment) during training. However, since the local data is inaccessible to the server under federated learning, attackers may easily poison the training data of the local client to build a backdoor in the agent without notice. Deploying such an agent raises the risk of potential harm to humans, as the attackers may easily navigate and control the agent as they wish via the backdoor. Towards Byzantine-robust federated embodied agent learning, in this paper, we study the attack and defense for the task of vision-and-language navigation (VLN), where the agent is required to follow natural language instructions to navigate indoor environments. First, we introduce a simple but effective attack strategy, Navigation as Wish (NAW), in which the malicious client manipulates local trajectory data to implant a backdoor into the global model. Results on two VLN datasets (R2R and RxR) show that NAW can easily navigate the deployed VLN agent regardless of the language instruction, without affecting its performance on normal test sets. Then, we propose a new Prompt-Based Aggregation (PBA) to defend against the NAW attack in federated VLN, which provides the server with a ''prompt'' of the vision-and-language alignment variance between the benign and malicious clients so that they can be distinguished during training. We validate the effectiveness of the PBA method on protecting the global model from the NAW attack, which outperforms other state-of-the-art defense methods by a large margin in the defense metrics on R2R and RxR.
|
['Xin Eric Wang', 'Cihang Xie', 'Kaiwen Zhou', 'Zonglin Di', 'Yunchao Zhang']
|
2022-11-27
| null | null | null | null |
['vision-and-language-navigation']
|
['robots']
|
[-3.42562586e-01 4.25563082e-02 2.41383567e-01 -1.71355128e-01
-5.92888415e-01 -1.32804060e+00 8.01934719e-01 -4.40205634e-01
-7.86874354e-01 2.41306692e-01 -2.28587419e-01 -6.96686029e-01
-8.26672316e-02 -7.49735177e-01 -9.63003039e-01 -1.20039809e+00
-4.59979564e-01 9.15872827e-02 1.38912931e-01 -1.81407735e-01
-9.07587633e-02 8.45479488e-01 -1.35903049e+00 9.79555771e-03
4.58647907e-01 9.04220939e-01 -1.44792289e-01 8.01558197e-01
3.23046863e-01 1.22482526e+00 -8.46575320e-01 -4.42994297e-01
7.77786016e-01 -1.14384249e-01 -4.41107869e-01 -3.56549710e-01
3.33775759e-01 -6.90288484e-01 -6.21789694e-01 1.28908110e+00
5.07368326e-01 1.53950587e-01 2.15073466e-01 -2.08585882e+00
-5.39160132e-01 3.34784806e-01 -3.21315765e-01 -3.01915616e-01
1.70507878e-01 7.48072922e-01 4.89605576e-01 -2.74148732e-01
5.52470148e-01 1.16140115e+00 3.97201538e-01 1.02290499e+00
-8.75193775e-01 -1.02369440e+00 4.98903692e-01 -1.72284350e-03
-1.22268224e+00 -4.82167482e-01 4.63324964e-01 -2.46109158e-01
5.52891195e-01 6.10224605e-01 1.33022532e-01 1.65611053e+00
3.14493150e-01 5.23083985e-01 1.15546227e+00 1.11507349e-01
6.79470479e-01 1.42717704e-01 1.14070363e-01 1.04164422e+00
3.82022738e-01 7.51863837e-01 -4.97622252e-01 -8.04328561e-01
3.87941033e-01 -8.62236321e-02 -5.36719143e-01 -7.79720783e-01
-8.61840069e-01 8.65284622e-01 4.59266365e-01 -4.09713477e-01
-3.19458544e-01 1.97866485e-01 4.86508936e-01 3.31624925e-01
-1.70036167e-01 2.48176664e-01 -4.47108507e-01 6.94235340e-02
4.97637615e-02 2.04234853e-01 9.70753491e-01 7.75799394e-01
5.20057440e-01 1.08651377e-01 1.44195750e-01 -9.17455032e-02
4.42460030e-01 8.35773408e-01 3.35436404e-01 -1.20164716e+00
3.28368753e-01 3.81837547e-01 1.38020352e-01 -1.02142453e+00
-1.24516271e-01 -5.74692041e-02 -5.01451313e-01 1.11810315e+00
3.44312698e-01 -6.08665764e-01 -5.57586432e-01 2.26586103e+00
8.34252357e-01 5.10647781e-02 7.49154627e-01 1.07854474e+00
4.31856036e-01 5.13467073e-01 2.17998773e-02 9.88762602e-02
1.33878505e+00 -9.68022645e-01 -4.69303519e-01 -1.51319563e-01
7.15179205e-01 -1.30227268e-01 9.71279442e-01 2.39766881e-01
-4.25827742e-01 6.87394664e-02 -1.08821857e+00 3.29375505e-01
-5.22549748e-01 -4.99428660e-01 6.07197702e-01 9.74069178e-01
-9.60007608e-01 -8.39647725e-02 -9.89170015e-01 -2.65710086e-01
5.61794102e-01 5.77954948e-01 -7.34231889e-01 -1.91787910e-02
-1.00213110e+00 6.53984308e-01 -1.22201644e-01 -1.73299968e-01
-1.89188504e+00 -5.48793852e-01 -9.18060362e-01 -1.74069896e-01
6.96741104e-01 -6.03528321e-01 1.10875809e+00 -6.56281769e-01
-1.43697059e+00 5.76890469e-01 5.24366379e-01 -6.66860044e-01
9.01584864e-01 4.24633175e-02 -2.33854681e-01 1.82001948e-01
-1.01752214e-01 2.94586539e-01 1.05362248e+00 -1.73571360e+00
-6.64877951e-01 -7.42180645e-01 6.04243577e-01 2.30198845e-01
-3.20414960e-01 -6.20691255e-02 -1.97453443e-02 -3.09792578e-01
-4.15533602e-01 -1.14413297e+00 -1.79010168e-01 3.25775862e-01
-4.29276049e-01 4.91991863e-02 1.55376124e+00 -3.08908850e-01
4.15708989e-01 -2.54661632e+00 -2.46551648e-01 2.66019672e-01
4.79458779e-01 2.73798198e-01 -4.73034292e-01 2.90596038e-01
2.89365888e-01 1.24231339e-01 7.74331391e-02 -4.01137859e-01
3.27211082e-01 4.70535547e-01 -8.32122266e-01 1.07948399e+00
-7.53022373e-01 6.20137513e-01 -8.17838907e-01 -5.48640937e-02
-7.36852661e-02 4.76589411e-01 -4.30341274e-01 6.72271490e-01
-2.28777111e-01 3.25711131e-01 -6.48309171e-01 5.90971410e-01
6.78420603e-01 2.54007399e-01 1.15898535e-01 6.21197075e-02
-1.40818283e-01 -1.43881276e-01 -7.66238689e-01 1.25236261e+00
-3.78252327e-01 4.41906631e-01 7.56350994e-01 -2.62562573e-01
7.23545790e-01 2.45043725e-01 2.96481997e-01 -5.91953337e-01
3.42408776e-01 -7.09642246e-02 -1.85515195e-01 -5.05865932e-01
1.57984328e-02 3.91376972e-01 -2.59861171e-01 8.33803117e-01
-4.37963605e-01 4.55900639e-01 -7.44370222e-01 3.00607383e-01
1.63004518e+00 -1.65152341e-01 -1.51333585e-01 1.99180353e-03
5.73660910e-01 -1.57111540e-01 4.02146876e-01 1.20470417e+00
-7.27116466e-01 -1.36440426e-01 3.95491928e-01 -4.18341964e-01
-5.37991643e-01 -1.30369258e+00 4.94215608e-01 1.24428058e+00
5.37221372e-01 -3.02008301e-01 -1.15209150e+00 -1.32641292e+00
1.22227035e-01 9.37171221e-01 -6.28384352e-01 -5.76454580e-01
-4.11561251e-01 -2.04875678e-01 1.28138828e+00 1.28264233e-01
8.89596701e-01 -9.02068377e-01 -1.18581271e+00 -4.83560264e-01
3.07175494e-03 -1.08723831e+00 -5.95204413e-01 2.12846935e-01
-6.14070110e-02 -1.38310122e+00 1.05856843e-01 -4.90087748e-01
7.58471549e-01 6.05458319e-01 2.68094301e-01 2.28885636e-01
-3.34718414e-02 1.03040600e+00 -2.91058779e-01 -1.95193350e-01
-6.44999564e-01 -3.74371797e-01 4.63481784e-01 2.40297318e-01
2.11050496e-01 -2.52645552e-01 -4.57739651e-01 5.31027794e-01
-8.95470977e-01 -5.35468698e-01 9.26200375e-02 6.37978494e-01
9.64320377e-02 2.42082492e-01 9.49548557e-02 -5.39114952e-01
6.50852203e-01 -4.72102344e-01 -1.11120152e+00 3.17049414e-01
-3.93846214e-01 -9.15296823e-02 1.02824962e+00 -7.86618173e-01
-7.10794091e-01 2.87485868e-02 2.00222015e-01 -7.44848430e-01
-2.92190582e-01 -1.48081869e-01 -7.88649201e-01 -8.08671176e-01
6.67493880e-01 3.34249467e-01 3.82653207e-01 -2.76444674e-01
4.93191242e-01 6.58059478e-01 7.08743930e-01 -7.62513459e-01
1.18379748e+00 8.39174092e-01 1.77717339e-02 -5.88461697e-01
-1.77316070e-01 1.95518762e-01 2.41481960e-01 -2.74817467e-01
8.94645095e-01 -7.18851626e-01 -1.57147741e+00 7.53422379e-01
-1.21900046e+00 -4.61886227e-01 -9.24387425e-02 2.20973879e-01
-6.04013443e-01 3.36243600e-01 -3.28289002e-01 -9.44415808e-01
-2.90427327e-01 -1.46031547e+00 6.31962478e-01 1.00337081e-01
1.66974291e-01 -6.92154527e-01 8.86841342e-02 4.66432244e-01
3.72734100e-01 4.50188994e-01 9.96886432e-01 -1.24903941e+00
-8.61560702e-01 -2.57634252e-01 2.83636808e-01 1.87861845e-01
1.31843790e-01 -1.89816579e-01 -1.05778205e+00 -8.15431952e-01
7.22531199e-01 -4.67895418e-01 1.55257002e-01 -3.01001936e-01
9.18660581e-01 -1.07413125e+00 -1.59335285e-01 1.02975512e+00
1.26782215e+00 5.26531458e-01 4.23025519e-01 8.14526081e-01
7.14836061e-01 6.95912302e-01 4.28075254e-01 3.81649315e-01
3.77813756e-01 3.05800349e-01 1.25882638e+00 2.36999467e-01
3.59017730e-01 -3.27417642e-01 9.41566467e-01 -1.89749092e-01
3.87446731e-01 -3.43197316e-01 -5.86251855e-01 3.26909535e-02
-1.79831314e+00 -9.15186107e-01 4.24917519e-01 2.26691341e+00
3.64780545e-01 -2.52931267e-01 -2.33937427e-02 -5.21925151e-01
4.88678992e-01 4.95261520e-01 -9.45735812e-01 -3.85610551e-01
-3.62793729e-02 -3.99589062e-01 9.28315639e-01 5.59187651e-01
-1.00353825e+00 9.56443250e-01 5.13016033e+00 6.23293757e-01
-1.15859997e+00 4.23850387e-01 2.98428297e-01 -3.37042034e-01
-1.35055110e-01 3.76743451e-02 -6.37485445e-01 4.13332880e-01
8.73747647e-01 -7.23335519e-02 1.10495996e+00 1.28194547e+00
2.51688482e-03 2.11700678e-01 -1.17437494e+00 7.95673251e-01
9.75294635e-02 -1.04488504e+00 -1.13730669e-01 4.24119622e-01
1.84231967e-01 2.61642456e-01 4.76919234e-01 3.50187242e-01
9.34793293e-01 -8.84651721e-01 8.71062338e-01 5.55546954e-02
5.39993882e-01 -9.99248087e-01 2.13808432e-01 6.91305637e-01
-8.31880212e-01 -5.17611682e-01 -1.63563371e-01 3.51372808e-01
-3.14025491e-01 -4.91509408e-01 -4.57328349e-01 3.40833455e-01
8.21233988e-01 -3.60216439e-01 -4.56681490e-01 3.21539938e-01
-1.59763843e-01 2.21912116e-01 -4.05703396e-01 5.84950596e-02
4.69086617e-01 -1.97173923e-01 1.08236802e+00 4.62903380e-01
-1.73753630e-02 2.22079441e-01 3.07579756e-01 7.15123057e-01
-1.17939413e-01 -2.55429447e-01 -1.21592176e+00 1.38081282e-01
6.23724997e-01 1.20074451e+00 -2.08461493e-01 2.21290678e-01
-2.41021216e-01 1.08667707e+00 4.72174495e-01 6.21913433e-01
-1.19634891e+00 -2.57027507e-01 1.49751437e+00 -3.80062401e-01
1.85957506e-01 -3.87702763e-01 2.39783540e-01 -8.28234434e-01
6.20981352e-03 -1.58733726e+00 4.90679473e-01 -4.76379037e-01
-1.26598895e+00 1.05696857e+00 -3.49220097e-01 -1.01307487e+00
-1.37716606e-01 -5.43048143e-01 -5.05161345e-01 4.43130761e-01
-1.07036221e+00 -1.28567851e+00 -2.83249527e-01 1.20976412e+00
1.06588453e-02 -7.28553355e-01 1.03717220e+00 -1.50331378e-01
-7.04498887e-01 9.65999007e-01 -3.90369482e-02 4.28019822e-01
5.23096800e-01 -6.97995424e-01 3.08702677e-01 1.22107279e+00
1.08244203e-01 9.34241474e-01 6.30521655e-01 -6.82846785e-01
-2.07983732e+00 -1.24931085e+00 -3.21833603e-02 -5.44853985e-01
7.58441806e-01 -6.95801795e-01 -6.19155645e-01 9.10013735e-01
1.36825979e-01 2.94512153e-01 5.92831731e-01 -5.43092906e-01
-1.01602161e+00 -1.16183646e-01 -1.79029584e+00 1.07857692e+00
7.80637264e-01 -8.49103928e-01 -1.47132233e-01 3.56783539e-01
1.31626797e+00 -2.75902778e-01 -2.45013833e-01 1.09929644e-01
4.50131029e-01 -1.09074748e+00 8.84750128e-01 -9.41983044e-01
-4.20780838e-01 -5.97333550e-01 -7.56377101e-01 -1.03939450e+00
1.03447452e-01 -1.11417508e+00 -3.24142575e-01 1.13106751e+00
-1.43186048e-01 -1.27237546e+00 8.34880054e-01 9.35291946e-01
2.61114568e-01 -3.97047281e-01 -1.39151227e+00 -1.09699464e+00
-1.06954817e-02 -4.07289624e-01 9.22056854e-01 8.71513307e-01
-4.22283828e-01 -2.69020140e-01 -3.40646625e-01 1.13876677e+00
1.04208875e+00 -3.61468047e-01 1.12812328e+00 -5.39172113e-01
-1.32701024e-01 -9.69965011e-02 -3.43700111e-01 -6.93076372e-01
5.98803282e-01 -7.79620111e-01 9.44963470e-02 -7.64482439e-01
-1.82774141e-01 -4.08918798e-01 -9.69159603e-02 8.65429044e-01
3.10693443e-01 -2.30485916e-01 5.06152809e-01 2.13984832e-01
-5.77071428e-01 7.06861854e-01 6.77910447e-01 -3.03629458e-01
1.94804871e-03 3.02271508e-02 -6.97644651e-01 6.77464485e-01
6.61749363e-01 -6.84045017e-01 -7.44516134e-01 -4.51574624e-01
-2.74292588e-01 -6.35298043e-02 8.80276322e-01 -8.77776384e-01
6.57332301e-01 -2.57304192e-01 -1.22727670e-01 -2.07238123e-01
2.64336675e-01 -1.59290600e+00 3.14615935e-01 6.96748853e-01
-2.75325686e-01 2.55645663e-01 4.59847040e-03 6.73689246e-01
4.30058628e-01 -5.13859019e-02 7.76670933e-01 4.05269116e-02
-4.63909894e-01 4.90210652e-01 -5.38597584e-01 -1.71700761e-01
1.57036650e+00 1.25259295e-01 -1.19958448e+00 -4.61784661e-01
-1.31801561e-01 5.43296218e-01 1.06803882e+00 4.03911769e-01
7.59050846e-01 -1.03429115e+00 -2.48798892e-01 5.77405572e-01
7.49322549e-02 -1.28451943e-01 1.43655509e-01 3.45216542e-01
-4.60284829e-01 -4.72048074e-02 -6.98592216e-02 -2.81538785e-01
-1.45349884e+00 1.30288339e+00 7.97617078e-01 2.25553185e-01
-6.78920507e-01 8.23786318e-01 5.21823406e-01 -7.81290174e-01
8.62146676e-01 2.44411469e-01 2.41408914e-01 -3.79724205e-01
7.44200945e-01 3.88574451e-01 -2.89604098e-01 -7.48928189e-01
-6.51883066e-01 1.16835669e-01 -1.56898022e-01 -1.69393599e-01
9.74771798e-01 -1.80321187e-01 -3.14992934e-01 -2.34280899e-01
1.37958908e+00 2.76041508e-01 -1.63281536e+00 1.11004971e-01
-4.11351383e-01 -6.42321646e-01 -1.17462948e-01 -7.42111146e-01
-1.21320462e+00 2.55362600e-01 8.94276142e-01 3.11466753e-01
8.70851457e-01 -2.16743782e-01 7.87441194e-01 6.70010149e-01
9.18240607e-01 -5.34663796e-01 1.48164690e-01 3.82379293e-01
6.87702775e-01 -9.79642332e-01 -4.18235004e-01 -3.82888690e-02
-8.11827719e-01 7.05604196e-01 9.87361133e-01 -9.38699173e-04
4.56155986e-01 6.13165975e-01 6.58077180e-01 -3.06069434e-01
-7.85958827e-01 4.21501875e-01 -4.15476620e-01 1.15667939e+00
-1.03839958e+00 6.19432004e-03 6.27298295e-01 7.34846354e-01
-2.22471505e-01 -6.96833193e-01 5.76292276e-01 1.14441168e+00
-3.08970392e-01 -8.23012590e-01 -7.09233463e-01 -3.57904762e-01
-1.85669497e-01 3.17712724e-01 -5.89653969e-01 8.21936786e-01
5.43919504e-02 1.23006821e+00 -2.02229023e-01 -7.72129476e-01
1.96678296e-01 -2.02004537e-01 -1.03251524e-02 -5.13357185e-02
-7.68487334e-01 -2.68632650e-01 -1.65455565e-01 -1.13472772e+00
1.04526080e-01 -3.26846480e-01 -1.47113132e+00 -6.35692179e-01
5.57893440e-02 1.48737013e-01 7.99772739e-01 6.24061048e-01
2.96613246e-01 1.17680170e-01 1.27125335e+00 -5.43402910e-01
-1.21351171e+00 -6.91209314e-03 -4.86503869e-01 1.67165190e-01
8.77663255e-01 -4.36622173e-01 -9.91936803e-01 -4.81582999e-01]
|
[5.698072910308838, 7.090514183044434]
|
6c913da0-1c81-4f53-ad58-2cb7ff9e4d73
|
adding-3d-geometry-control-to-diffusion
|
2306.08103
| null |
https://arxiv.org/abs/2306.08103v1
|
https://arxiv.org/pdf/2306.08103v1.pdf
|
Adding 3D Geometry Control to Diffusion Models
|
Diffusion models have emerged as a powerful method of generative modeling across a range of fields, capable of producing stunning photo-realistic images from natural language descriptions. However, these models lack explicit control over the 3D structure of the objects in the generated images. In this paper, we propose a novel method that incorporates 3D geometry control into diffusion models, making them generate even more realistic and diverse images. To achieve this, our method exploits ControlNet, which extends diffusion models by using visual prompts in addition to text prompts. We generate images of 3D objects taken from a 3D shape repository (e.g., ShapeNet and Objaverse), render them from a variety of poses and viewing directions, compute the edge maps of the rendered images, and use these edge maps as visual prompts to generate realistic images. With explicit 3D geometry control, we can easily change the 3D structures of the objects in the generated images and obtain ground-truth 3D annotations automatically. This allows us to use the generated images to improve a lot of vision tasks, e.g., classification and 3D pose estimation, in both in-distribution (ID) and out-of-distribution (OOD) settings. We demonstrate the effectiveness of our method through extensive experiments on ImageNet-50, ImageNet-R, PASCAL3D+, ObjectNet3D, and OOD-CV datasets. The results show that our method significantly outperforms existing methods across multiple benchmarks (e.g., 4.6 percentage points on ImageNet-50 using ViT and 3.5 percentage points on PASCAL3D+ and ObjectNet3D using NeMo).
|
['Alan Yuille', 'Adam Kortylewski', 'Yaoyao Liu', 'Angtian Wang', 'Jiahao Wang', 'Qihao Liu', 'Wufei Ma']
|
2023-06-13
| null | null | null | null |
['pose-estimation', '3d-pose-estimation']
|
['computer-vision', 'computer-vision']
|
[-2.17905432e-01 3.10888216e-02 2.31841043e-01 -1.88911334e-01
-3.71487468e-01 -9.58231390e-01 8.47696126e-01 -3.84189159e-01
-1.31370947e-01 2.34649479e-01 1.87559444e-02 -1.65090159e-01
3.52527142e-01 -8.95267785e-01 -9.13765609e-01 -5.72943449e-01
2.86596328e-01 8.33087027e-01 5.35429120e-01 -3.16134572e-01
1.87965706e-01 8.12006533e-01 -1.52920628e+00 9.95274559e-02
7.88056195e-01 7.66721666e-01 4.40846533e-01 7.88595915e-01
-2.33071178e-01 4.20322865e-01 -8.87010038e-01 -2.98692256e-01
4.44179296e-01 -9.78408232e-02 -4.12255734e-01 6.70925736e-01
7.48136044e-01 -5.21596253e-01 -2.20496938e-01 1.03221405e+00
7.37403691e-01 5.70447892e-02 9.14249003e-01 -1.39944589e+00
-8.00283194e-01 -5.98718747e-02 -8.68750751e-01 -2.18246639e-01
4.75891560e-01 6.19010091e-01 3.53028148e-01 -9.95416284e-01
1.05974209e+00 1.70849681e+00 3.19781929e-01 7.20548332e-01
-1.47926855e+00 -6.56385481e-01 1.19044334e-01 -1.32288516e-01
-1.36809921e+00 -2.39422232e-01 8.87120187e-01 -6.37943208e-01
9.52502549e-01 1.93131700e-01 7.27197886e-01 1.35460150e+00
2.16887757e-01 7.45845675e-01 1.11920857e+00 -2.47661620e-01
2.99591452e-01 6.30939752e-02 -5.16191244e-01 6.71079397e-01
1.48401819e-02 3.69459689e-01 -3.38448375e-01 -1.23931460e-01
1.19925988e+00 -1.90942615e-01 -1.45283356e-01 -7.35423386e-01
-1.15405929e+00 6.40215158e-01 6.08842254e-01 -3.01176339e-01
-2.98936129e-01 2.54155993e-01 1.79957487e-02 -1.96311802e-01
6.33622944e-01 1.50505722e-01 -1.74300969e-01 1.13879994e-01
-4.23329175e-01 6.93010986e-01 6.87278390e-01 1.37401164e+00
7.28263676e-01 3.61435145e-01 -4.22955036e-01 8.28517258e-01
3.40994596e-01 1.01031482e+00 1.37621194e-01 -1.16065729e+00
3.31939816e-01 5.65398574e-01 2.44643137e-01 -1.10570586e+00
-1.50277615e-01 -2.49013633e-01 -9.59500432e-01 6.53866231e-01
2.48677358e-01 -2.15380453e-02 -1.49172950e+00 1.46524024e+00
6.99342966e-01 -3.49909924e-02 -7.10011572e-02 1.02588534e+00
9.72900033e-01 8.55200231e-01 1.12743869e-01 4.11097407e-01
1.12114584e+00 -9.28677738e-01 -1.63477734e-01 -3.15435082e-01
3.37991804e-01 -9.28338528e-01 1.17002308e+00 2.94909984e-01
-1.00762296e+00 -7.73620784e-01 -7.52241313e-01 -8.90870988e-02
-1.38475642e-01 2.28819274e-03 4.37066793e-01 4.68999892e-01
-1.12228739e+00 1.83008358e-01 -6.43492222e-01 -2.73960568e-02
4.57735747e-01 -3.89405899e-02 -2.57200330e-01 -3.34743351e-01
-6.97759986e-01 7.09479868e-01 2.96493739e-01 -2.77764857e-01
-1.31264818e+00 -8.21958065e-01 -1.04350686e+00 -1.60035521e-01
3.83231908e-01 -8.97690117e-01 1.25990820e+00 -7.02812135e-01
-1.24573720e+00 9.67785358e-01 2.08504498e-01 -1.07659608e-01
9.23515320e-01 5.51118553e-02 5.01297377e-02 -2.81508025e-02
2.05495626e-01 1.42127132e+00 8.44241560e-01 -1.80253720e+00
-3.29151869e-01 -3.52460146e-01 2.49996215e-01 4.19020146e-01
2.58254051e-01 -2.26767063e-01 -9.13671136e-01 -7.23909318e-01
-3.85031626e-02 -1.13935566e+00 -4.66815025e-01 4.31418002e-01
-7.07077444e-01 -2.27120697e-01 1.09777510e+00 -4.08538550e-01
3.43253255e-01 -2.17577314e+00 1.44287691e-01 1.21812649e-01
3.25844109e-01 3.18776935e-01 -2.99753666e-01 1.30770311e-01
1.90360382e-01 3.16475391e-01 -3.35849524e-02 -6.18766546e-01
6.63712919e-02 3.32841039e-01 -1.23301283e-01 1.63716465e-01
2.31291056e-01 1.06834948e+00 -8.41227114e-01 -4.59032774e-01
6.05936170e-01 7.75984764e-01 -6.90767229e-01 2.61368096e-01
-7.72588193e-01 6.16717458e-01 -3.32462579e-01 5.58325946e-01
9.05175388e-01 -2.83244014e-01 -2.11541265e-01 -2.29691535e-01
-2.22403579e-03 -1.99511871e-01 -1.27442181e+00 1.67691755e+00
-4.89406615e-01 4.58454221e-01 -2.69897163e-01 -2.09562913e-01
1.07624972e+00 4.54392061e-02 1.81165919e-01 -7.49592841e-01
1.74087286e-02 -7.48504475e-02 -1.04354747e-01 -3.50736946e-01
5.89063168e-01 2.44302481e-01 -1.55233154e-02 4.31853175e-01
-1.37346610e-01 -8.90890837e-01 2.67583877e-01 3.57482523e-01
6.45304978e-01 2.99881488e-01 -2.10493505e-02 -5.08216545e-02
8.03838149e-02 1.05412155e-01 2.81009853e-01 6.32379293e-01
1.81615517e-01 1.10899210e+00 3.31300616e-01 -4.46156979e-01
-1.37940383e+00 -1.30781484e+00 1.23050719e-01 4.67757821e-01
4.41968918e-01 -3.57615113e-01 -7.41864681e-01 -6.83785498e-01
3.98816839e-02 9.34332490e-01 -5.55872619e-01 -1.15456298e-01
-2.79695213e-01 -4.57245857e-01 3.82410526e-01 6.07907414e-01
7.50153840e-01 -1.05366814e+00 -5.31566143e-01 -2.86105797e-02
-1.64273992e-01 -1.23547077e+00 -7.67646968e-01 -2.91430771e-01
-6.23256564e-01 -8.46600056e-01 -9.55403030e-01 -6.13451660e-01
9.28220868e-01 3.87661844e-01 1.26866364e+00 -1.29652023e-01
-4.60164189e-01 4.33347613e-01 -2.41808638e-01 -5.01700461e-01
-8.08079302e-01 -1.69644102e-01 -2.07518399e-01 -1.33866847e-01
-1.29253447e-01 -5.00686646e-01 -7.31761456e-01 7.23632216e-01
-1.10544753e+00 6.07918143e-01 5.05934000e-01 4.63884264e-01
9.54264402e-01 -8.08880329e-02 9.11324695e-02 -8.23649883e-01
4.34117258e-01 -2.30346575e-01 -8.26656878e-01 1.87883750e-02
-3.99877220e-01 6.84722066e-02 3.60662788e-01 -7.47848749e-01
-1.11815238e+00 1.86186612e-01 -1.28498733e-01 -9.10862505e-01
-3.20316434e-01 1.04456870e-02 -3.84594530e-01 1.37350122e-02
1.01220047e+00 3.89103353e-01 -1.60306275e-01 -4.26356465e-01
4.95095402e-01 4.52616394e-01 4.87008929e-01 -7.04523563e-01
1.06586409e+00 5.45063317e-01 -1.57158136e-01 -8.25042486e-01
-6.00817621e-01 -7.54480809e-02 -4.18965548e-01 -3.75030071e-01
9.49127555e-01 -9.09702957e-01 -3.85594010e-01 7.79666662e-01
-1.40678334e+00 -6.16206825e-01 -1.93564504e-01 1.82613030e-01
-7.03109145e-01 1.61605716e-01 -2.71764934e-01 -4.90822494e-01
-2.20431492e-01 -1.39661002e+00 1.51016152e+00 3.03131282e-01
-1.82248369e-01 -7.95362711e-01 -2.05288440e-01 2.15696380e-01
3.29009354e-01 4.84026104e-01 9.77636337e-01 3.11601050e-02
-1.00391030e+00 -3.57369855e-02 -3.91085953e-01 3.92559171e-01
5.13077006e-02 8.67383182e-02 -8.15943599e-01 -9.45590511e-02
-4.66100216e-01 -2.47661039e-01 3.91911715e-01 3.36168975e-01
1.34539461e+00 -1.47095934e-01 -3.61468226e-01 7.85632968e-01
1.22046590e+00 1.71652272e-01 9.32550251e-01 -2.19685733e-02
9.68109369e-01 3.65290314e-01 4.73780781e-01 3.88715744e-01
4.89439487e-01 9.27796543e-01 6.56212211e-01 -2.96499729e-01
-7.28612304e-01 -5.61774969e-01 1.16951376e-01 2.77704775e-01
-1.78248063e-01 -5.26757538e-01 -8.85350823e-01 4.76025999e-01
-1.46050549e+00 -7.46141553e-01 -1.95815891e-01 2.09125328e+00
7.47112751e-01 1.41519308e-01 4.93668541e-02 -2.97271609e-01
6.93132102e-01 1.04994662e-01 -8.22190285e-01 -1.29694417e-01
-8.98268167e-03 -7.40864500e-02 4.48109806e-01 3.36062312e-01
-7.17420459e-01 1.08709788e+00 5.75388765e+00 8.59259307e-01
-1.20652676e+00 -1.22231640e-01 8.96224022e-01 4.38481756e-02
-4.88120466e-01 -1.15251616e-01 -8.63998294e-01 3.86248082e-01
1.97343394e-01 -2.18455419e-01 3.57707888e-01 1.03112304e+00
3.04893672e-01 -1.40756294e-01 -1.06313670e+00 1.32817686e+00
1.16986088e-01 -1.47622740e+00 3.88511986e-01 2.95651436e-01
1.10796905e+00 4.20742631e-02 1.61219135e-01 9.47490409e-02
6.52460754e-01 -9.93286669e-01 1.08661103e+00 3.48827899e-01
1.12403309e+00 -6.82787120e-01 2.60558426e-01 6.51383340e-01
-8.94144595e-01 5.70969582e-01 -3.69913220e-01 3.29127133e-01
3.13066006e-01 5.28306842e-01 -1.03238618e+00 3.76088709e-01
8.51760745e-01 4.27961618e-01 -6.36873364e-01 1.00589669e+00
-4.30057496e-01 2.17590868e-01 -3.74054313e-01 -4.53113467e-02
1.36472657e-01 -1.45797003e-02 6.25278056e-01 9.84796047e-01
4.28674310e-01 -7.15791211e-02 3.19598436e-01 1.37757432e+00
-1.65072635e-01 -2.34320924e-01 -6.91967309e-01 1.98890373e-01
3.62680584e-01 1.13152492e+00 -7.88421035e-01 -2.71907479e-01
6.70067370e-02 1.09110618e+00 3.79005959e-03 4.72485721e-01
-1.02657890e+00 -1.19684517e-01 6.23026311e-01 3.95069987e-01
3.56957793e-01 -5.16078949e-01 2.75667645e-02 -9.79137480e-01
2.09994428e-02 -9.44265366e-01 -1.96112677e-01 -1.33777249e+00
-1.25706100e+00 7.81087697e-01 5.34716964e-01 -1.14095640e+00
-5.11642039e-01 -5.27339280e-01 -3.97218674e-01 9.69059646e-01
-1.10853052e+00 -1.18025553e+00 -7.38293052e-01 4.98829484e-01
7.15018153e-01 2.12864131e-01 4.91513342e-01 2.62436032e-01
-2.52016746e-02 2.29952037e-01 -3.12057406e-01 1.12752467e-01
5.05880892e-01 -1.29946923e+00 1.09081447e+00 4.27200884e-01
3.03464502e-01 1.23384550e-01 6.93710208e-01 -6.74699008e-01
-1.18683362e+00 -1.42241681e+00 3.66642356e-01 -6.64907455e-01
-1.01907635e-02 -6.01546168e-01 -6.65682852e-01 4.82355803e-01
9.49120671e-02 1.05795071e-01 -5.07318415e-02 -5.47900498e-01
-3.99568975e-01 1.39479890e-01 -1.16217721e+00 9.90396857e-01
1.49392962e+00 -1.45073712e-01 -3.44483927e-02 4.48075056e-01
8.02851796e-01 -1.08561170e+00 -5.10162592e-01 2.79235095e-01
3.36520672e-01 -1.07069039e+00 1.21969163e+00 -2.62899160e-01
4.66071218e-01 -5.90339065e-01 -2.10404336e-01 -1.76534808e+00
-4.08080928e-02 -4.47868139e-01 8.08251351e-02 1.11520398e+00
2.23217711e-01 -4.64338034e-01 7.48522878e-01 5.45593739e-01
-1.22875802e-01 -5.70470273e-01 -5.76462030e-01 -8.82723093e-01
-4.85861152e-02 -5.50794125e-01 7.45413840e-01 5.89675426e-01
-1.00566208e+00 2.60307789e-01 -2.12929398e-01 5.58183007e-02
7.32872427e-01 4.99901511e-02 1.39908898e+00 -1.00391340e+00
-2.90415317e-01 -3.63118947e-01 -5.60310721e-01 -1.49292386e+00
1.46858105e-02 -7.90482938e-01 -9.24799591e-02 -1.71745563e+00
3.94715369e-02 -7.29908824e-01 6.17105782e-01 3.87015104e-01
5.18319272e-02 4.55693454e-01 4.09302205e-01 1.54450491e-01
-2.04633996e-01 6.97586179e-01 1.89562595e+00 -3.85770082e-01
-2.73538888e-01 -5.23951240e-02 -5.37334979e-01 8.19517970e-01
6.33558691e-01 -4.61624026e-01 -7.73585796e-01 -7.25961208e-01
1.38689829e-02 -1.17053464e-01 8.47229779e-01 -9.28865314e-01
-1.01846814e-01 -3.08663487e-01 7.44777322e-01 -8.55333865e-01
5.69024324e-01 -6.58138633e-01 4.34798896e-01 2.32239544e-01
-4.01187688e-02 1.36076227e-01 3.51926655e-01 5.47534883e-01
7.85616189e-02 5.87953664e-02 7.59113014e-01 -3.08692902e-01
-8.84607673e-01 6.29204690e-01 4.09820601e-02 2.62335449e-01
1.15284729e+00 -3.60785425e-01 -4.69727188e-01 -6.32472694e-01
-6.14817560e-01 1.76469952e-01 8.18146527e-01 8.08822036e-01
8.66252065e-01 -1.47416615e+00 -7.67463982e-01 3.67092997e-01
3.71494561e-01 7.36406803e-01 4.56550539e-01 3.43783021e-01
-8.20865810e-01 -6.07189424e-02 -1.48474261e-01 -1.17307973e+00
-1.20576930e+00 4.06654507e-01 4.72115576e-01 7.17848241e-02
-7.18402028e-01 5.82017541e-01 8.59275281e-01 -6.48415685e-01
1.42608777e-01 -4.52651769e-01 2.90975988e-01 -4.87201929e-01
3.96930754e-01 3.22459713e-02 -4.45838086e-02 -6.58357859e-01
-1.24524429e-01 6.19253635e-01 -2.41650864e-02 -2.24295527e-01
1.19292521e+00 6.83389008e-02 3.62075448e-01 2.90202908e-02
1.00920844e+00 1.23586424e-01 -1.83404732e+00 1.47444801e-02
-5.39177477e-01 -7.72670746e-01 -8.87557939e-02 -9.38080251e-01
-1.24046969e+00 8.18072200e-01 5.81798494e-01 -1.23694718e-01
7.42914081e-01 3.34366888e-01 5.50182998e-01 6.65697977e-02
5.43106019e-01 -7.36599982e-01 4.62038130e-01 3.45546901e-01
1.30553055e+00 -9.33277071e-01 -3.01291943e-01 -6.47303104e-01
-7.74580777e-01 6.75758302e-01 9.73190546e-01 -1.16763823e-01
3.66000831e-01 2.56583065e-01 2.33632997e-01 -2.50504702e-01
-6.12691820e-01 -5.55508099e-02 3.98824632e-01 1.12487257e+00
6.30518049e-02 4.33228724e-02 3.26423138e-01 -4.18041758e-02
-2.43911207e-01 -2.31726468e-01 6.26071036e-01 6.85061991e-01
-1.54298916e-01 -9.54531193e-01 -5.81713855e-01 5.77464215e-02
1.54672042e-01 6.82478175e-02 -5.52405238e-01 9.80391085e-01
1.58511937e-01 6.92235768e-01 8.53397027e-02 -3.25727403e-01
5.56857705e-01 -3.40623856e-01 7.59765267e-01 -7.61520684e-01
2.44136108e-03 1.16217256e-01 -6.37573674e-02 -4.01426524e-01
-2.01421440e-01 -3.31582397e-01 -1.15883291e+00 -3.86474162e-01
-1.57859266e-01 -3.35393578e-01 9.81079042e-01 5.47688186e-01
6.85092688e-01 5.54963112e-01 3.92596841e-01 -1.32931602e+00
-5.73165834e-01 -8.58421206e-01 -4.31060314e-01 5.94914675e-01
-9.57465321e-02 -8.77039850e-01 -2.28676870e-01 2.05170661e-01]
|
[9.288121223449707, -3.042825222015381]
|
19c4f579-6a07-4491-88ac-d91e679c0c31
|
a-hybrid-approach-for-learning-program
|
1907.02136
| null |
https://arxiv.org/abs/1907.02136v2
|
https://arxiv.org/pdf/1907.02136v2.pdf
|
Learning Blended, Precise Semantic Program Embeddings
|
Learning neural program embeddings is key to utilizing deep neural networks in program languages research --- precise and efficient program representations enable the application of deep models to a wide range of program analysis tasks. Existing approaches predominately learn to embed programs from their source code, and, as a result, they do not capture deep, precise program semantics. On the other hand, models learned from runtime information critically depend on the quality of program executions, thus leading to trained models with highly variant quality. This paper tackles these inherent weaknesses of prior approaches by introducing a new deep neural network, \liger, which learns program representations from a mixture of symbolic and concrete execution traces. We have evaluated \liger on \coset, a recently proposed benchmark suite for evaluating neural program embeddings. Results show \liger (1) is significantly more accurate than the state-of-the-art syntax-based models Gated Graph Neural Network and code2vec in classifying program semantics, and (2) requires on average 10x fewer executions covering 74\% fewer paths than the state-of-the-art dynamic model \dypro. Furthermore, we extend \liger to predict the name for a method from its body's vector representation. Learning on the same set of functions (more than 170K in total), \liger significantly outperforms code2seq, the previous state-of-the-art for method name prediction.
|
['Zhendong Su', 'Ke Wang']
|
2019-07-03
| null | null | null | null |
['method-name-prediction']
|
['natural-language-processing']
|
[-2.10063964e-01 -1.54893667e-01 -9.12811935e-01 -4.07221675e-01
-4.59840417e-01 -5.05748510e-01 2.30511829e-01 4.96821612e-01
-3.43638271e-01 5.74491099e-02 3.01414430e-01 -9.14978862e-01
4.37051773e-01 -1.08158898e+00 -1.08477354e+00 -9.00634155e-02
-2.48451605e-01 8.46660510e-02 2.29123216e-02 -2.64020652e-01
3.23212206e-01 1.43777847e-01 -1.58333755e+00 4.96142507e-01
5.92872500e-01 6.42073870e-01 -1.98112354e-02 9.69290733e-01
-6.63868725e-01 1.23227465e+00 -5.01617551e-01 -4.26124275e-01
-1.58864200e-01 -2.09986530e-02 -8.29967737e-01 -8.45685244e-01
4.62123752e-01 -5.39538026e-01 -7.75831521e-01 1.16372764e+00
9.88341123e-02 5.15443087e-03 3.82597297e-01 -1.09386420e+00
-1.05695772e+00 1.16525531e+00 -5.78565359e-01 4.24948722e-01
2.11663097e-01 2.95391440e-01 1.35755336e+00 -5.65961540e-01
5.61437011e-01 1.09384406e+00 1.13544631e+00 6.53437197e-01
-1.60501969e+00 -4.76529866e-01 -1.24353752e-01 -1.27811730e-01
-1.05354226e+00 -2.56899953e-01 6.21818244e-01 -7.08497345e-01
1.92971098e+00 2.22639367e-02 3.18520546e-01 1.08774841e+00
4.62978721e-01 7.66043484e-01 4.58008081e-01 -2.75797218e-01
4.82765347e-04 -6.36375472e-02 9.57949340e-01 1.08211136e+00
4.49011594e-01 4.07938749e-01 -6.62907064e-02 -7.05407441e-01
2.57889211e-01 3.09819490e-01 -3.26934129e-01 -3.95106912e-01
-1.19744408e+00 1.05188727e+00 6.05761647e-01 6.55027866e-01
-4.05009426e-02 1.15749407e+00 1.29532087e+00 4.03462470e-01
1.56136915e-01 6.32529914e-01 -6.94761276e-01 -7.27768064e-01
-9.56599653e-01 4.64501590e-01 1.09554422e+00 9.04784501e-01
9.35880125e-01 6.71220899e-01 -1.63193747e-01 6.46651268e-01
2.49237910e-01 2.66351193e-01 7.08935976e-01 -5.11243463e-01
6.73784971e-01 9.84624863e-01 -5.50087750e-01 -1.02934885e+00
-2.73988903e-01 -3.44837099e-01 -5.00519454e-01 3.89052123e-01
1.54025868e-01 1.04504190e-01 -6.54891551e-01 1.63656938e+00
-3.41506153e-01 -1.23667106e-01 -8.14425647e-02 3.56651187e-01
1.03215456e+00 6.62439108e-01 9.18198526e-02 4.72505033e-01
9.78641510e-01 -1.08098674e+00 -2.68172055e-01 -3.60405326e-01
1.33793843e+00 -3.60984653e-01 1.27276361e+00 1.82095319e-01
-6.74962461e-01 -7.25220382e-01 -1.17499876e+00 -4.95233722e-02
-5.38712382e-01 7.54559934e-02 1.17985368e+00 8.13158572e-01
-1.31627893e+00 7.76366949e-01 -1.16356766e+00 -1.08365022e-01
4.29754794e-01 4.38920110e-01 -4.26000774e-01 -3.28770652e-02
-5.95469773e-01 4.80669230e-01 6.83320284e-01 -4.11414862e-01
-1.07075226e+00 -1.09885693e+00 -1.37995827e+00 4.78985667e-01
1.15546554e-01 -3.24626237e-01 1.32910395e+00 -1.09576499e+00
-1.09656274e+00 8.74175847e-01 -1.59116194e-01 -5.85339248e-01
-2.29284495e-01 -2.60581344e-01 -4.45884019e-01 -4.87466753e-01
3.44280563e-02 2.26538591e-02 3.48887682e-01 -9.63794589e-01
-5.26656583e-02 -2.14379713e-01 2.83382952e-01 -6.72040582e-01
-6.67320788e-01 1.42020300e-01 -4.41604495e-01 -6.85369015e-01
-7.20319450e-01 -7.22692013e-01 -1.88614368e-01 -2.35026017e-01
-3.00756395e-01 -3.60643417e-01 7.51813114e-01 -6.17447615e-01
1.73352599e+00 -2.38614988e+00 2.17920557e-01 -1.26145706e-01
7.36086965e-01 4.52070862e-01 -2.96255738e-01 4.88261908e-01
-2.93991148e-01 3.92813534e-01 -2.60851175e-01 -2.18181029e-01
4.35136914e-01 2.39238575e-01 -5.30251265e-01 5.58664739e-01
6.82960451e-02 1.22779286e+00 -9.41305161e-01 -1.70844316e-01
-4.07867786e-03 3.92646641e-01 -1.02321339e+00 3.32343161e-01
-6.86422169e-01 -4.46043909e-01 -4.79558617e-01 6.30052805e-01
3.23631614e-01 -4.34850097e-01 1.73388347e-01 9.43689346e-02
9.22187045e-03 2.84781843e-01 -5.70645630e-01 1.91071486e+00
-8.62562418e-01 9.98398483e-01 -1.66593894e-01 -1.10997522e+00
8.35560083e-01 1.41228363e-01 1.32051364e-01 -6.29890025e-01
8.76194760e-02 2.71425992e-01 -8.24163482e-02 -6.86457634e-01
6.89888895e-01 5.31829238e-01 -5.30071020e-01 6.83674574e-01
3.47652018e-01 1.84355065e-01 1.23736463e-01 1.82012185e-01
1.65515471e+00 3.28067303e-01 2.82019258e-01 -3.18993151e-01
4.92317408e-01 1.35335147e-01 5.14612317e-01 8.67990911e-01
-1.43593475e-01 1.32496044e-01 1.12896252e+00 -8.53509009e-01
-9.55603480e-01 -7.33183801e-01 2.11098135e-01 1.64858389e+00
-4.68374312e-01 -8.81815970e-01 -7.69983888e-01 -9.14111376e-01
3.74079913e-01 9.56201136e-01 -9.50989187e-01 -2.29514867e-01
-1.03951669e+00 -5.20026445e-01 1.11496127e+00 1.05911851e+00
6.95759207e-02 -8.63472283e-01 -5.47844112e-01 2.87190497e-01
4.94488597e-01 -5.71915865e-01 -6.90958321e-01 4.42454278e-01
-8.12170267e-01 -1.31075108e+00 -3.63087237e-01 -7.18297541e-01
3.14082682e-01 -1.58513486e-01 1.93720412e+00 5.39094687e-01
-2.48479679e-01 3.37402314e-01 -1.60737440e-01 4.64732721e-02
-1.06453657e+00 2.84298301e-01 -3.35690767e-01 -6.77240133e-01
8.24987710e-01 -6.27307653e-01 -1.59185171e-01 -3.55792910e-01
-9.35814440e-01 -3.88480127e-01 4.74976420e-01 1.15392888e+00
3.18914115e-01 -2.66525522e-02 2.64171481e-01 -1.30539894e+00
6.22647047e-01 -7.16742277e-01 -9.13098872e-01 5.20287305e-02
-9.36088145e-01 5.39129496e-01 1.08056200e+00 -3.74286979e-01
-7.77134657e-01 -2.78728634e-01 -3.79510760e-01 -6.58183932e-01
-8.41217339e-02 8.61683011e-01 1.25502467e-01 -2.23158970e-01
1.19247198e+00 2.41477251e-01 -6.41406253e-02 -4.04161304e-01
6.34047806e-01 4.16595340e-01 5.92328489e-01 -8.69904518e-01
5.44142306e-01 -4.76218648e-02 -1.91162184e-01 -5.34815133e-01
-1.39821433e-02 -3.27409893e-01 -3.80555153e-01 3.79920632e-01
7.83972979e-01 -6.13937020e-01 -7.53843427e-01 3.99812371e-01
-1.34658837e+00 -6.24749601e-01 -5.21662980e-02 2.04523206e-01
-3.57039601e-01 4.36303467e-01 -8.78014505e-01 -2.64629215e-01
-4.86081541e-01 -1.63257980e+00 9.43264842e-01 -1.29352257e-01
-4.24519539e-01 -1.29010975e+00 4.65377808e-01 -1.83523327e-01
7.18279004e-01 5.38592637e-01 1.62077904e+00 -8.58201683e-01
-6.02776110e-01 -3.41858029e-01 -3.28538537e-01 5.08323908e-01
-2.13625774e-01 2.00494602e-01 -9.94756281e-01 -2.39560917e-01
-4.18165416e-01 -3.31382483e-01 1.03556120e+00 2.03419760e-01
1.42571652e+00 -2.77768135e-01 -4.91986394e-01 1.02967882e+00
1.92292595e+00 -4.44131084e-02 3.97457957e-01 1.93105295e-01
1.09175479e+00 -7.47386366e-02 -1.10163599e-01 3.15495133e-01
2.65951067e-01 4.71433431e-01 6.86163187e-01 2.42473155e-01
-1.66983679e-01 -2.37017497e-01 6.14119828e-01 8.72222245e-01
2.36895189e-01 -5.33803254e-02 -1.38498986e+00 8.26836586e-01
-1.57603216e+00 -7.50882030e-01 -2.35374853e-01 1.97376966e+00
9.58878398e-01 1.80274561e-01 5.94894551e-02 -9.63769406e-02
1.78090304e-01 5.49484551e-01 -3.94253373e-01 -1.00756931e+00
4.25565392e-01 7.55021095e-01 8.38420868e-01 1.99515983e-01
-9.20226812e-01 9.18723404e-01 6.02107430e+00 7.51135170e-01
-1.38598537e+00 2.76521236e-01 1.59735531e-01 2.18113288e-01
-7.26429999e-01 1.77614950e-02 -7.37922370e-01 3.54172766e-01
1.49552214e+00 -4.33662683e-01 6.44527555e-01 1.53971481e+00
-4.84934598e-01 2.20741466e-01 -1.74102950e+00 7.99060822e-01
1.73181847e-01 -1.64876485e+00 -2.47800693e-01 3.10104154e-02
6.20576262e-01 5.81509173e-01 1.73705816e-02 1.12542582e+00
5.90272605e-01 -1.36736071e+00 6.07131422e-01 2.91204423e-01
9.46213603e-01 -7.47187376e-01 7.24491358e-01 8.59918371e-02
-1.36103868e+00 -2.89257407e-01 -1.83707535e-01 1.03722252e-02
-4.14872169e-01 4.89968896e-01 -6.32306933e-01 3.40029448e-01
6.92963064e-01 9.33646321e-01 -8.94435644e-01 5.04885733e-01
-5.64006828e-02 8.84077728e-01 2.45613158e-01 -3.41715217e-01
2.91219115e-01 4.61880356e-01 2.51964271e-01 1.61221969e+00
9.30831432e-02 -4.53241348e-01 1.26848206e-01 1.57406557e+00
-4.67236370e-01 -7.13366345e-02 -1.01724899e+00 -8.11046183e-01
2.20648944e-01 7.77008712e-01 -2.88513660e-01 -4.54030871e-01
-9.92510915e-01 5.75713396e-01 6.18044317e-01 3.52460444e-01
-9.56798553e-01 -8.03611755e-01 8.91485512e-01 -1.56201184e-01
3.90795439e-01 -3.67721587e-01 -3.70697558e-01 -1.25635445e+00
7.01419711e-02 -1.17930460e+00 2.47683495e-01 -1.61556005e-01
-8.76408339e-01 8.00334692e-01 -1.62605748e-01 -5.84899247e-01
-5.03794134e-01 -8.88574839e-01 -8.10298383e-01 9.94897425e-01
-1.40152001e+00 -9.25196350e-01 -1.10829957e-01 1.34355828e-01
3.87156278e-01 -4.42891717e-01 1.18594432e+00 4.49471742e-01
-5.31451464e-01 1.07538366e+00 2.82223165e-01 5.45106411e-01
3.12255293e-01 -1.43175256e+00 1.02604163e+00 6.94575310e-01
8.27125367e-03 1.03322279e+00 6.04202449e-01 -5.06143272e-01
-2.23940539e+00 -1.36112595e+00 6.25598729e-01 -6.45610571e-01
8.94287348e-01 -2.86694527e-01 -1.28825271e+00 9.95019078e-01
1.83669105e-01 4.07725185e-01 7.08421111e-01 4.30666417e-01
-1.10072517e+00 3.94999236e-02 -7.22831070e-01 3.38768274e-01
7.55603135e-01 -1.00005412e+00 -6.28338933e-01 6.31895959e-02
1.13445425e+00 -4.95326877e-01 -1.15189636e+00 1.80766881e-01
5.17651021e-01 -1.08889914e+00 8.57039452e-01 -7.31324434e-01
8.90171587e-01 7.81566128e-02 -3.96921515e-01 -9.73773718e-01
-1.37963519e-01 -1.97014362e-01 -6.37504160e-01 1.09721434e+00
2.77573466e-01 -5.82954645e-01 8.45821977e-01 3.08444858e-01
-4.14544731e-01 -9.52409506e-01 -3.66539627e-01 -7.51255572e-01
5.30562818e-01 -7.22272635e-01 9.36154664e-01 9.03174698e-01
-1.06647201e-02 -1.91272832e-02 1.07226454e-01 -2.73900554e-02
3.49131733e-01 2.76436746e-01 9.24177587e-01 -8.63320351e-01
-9.13862348e-01 -1.00772905e+00 -6.88410342e-01 -8.78245771e-01
9.34940875e-01 -1.37912118e+00 -1.87475756e-01 -1.20773888e+00
2.15531126e-01 -2.37101346e-01 -2.81612068e-01 7.21985281e-01
-8.33331496e-02 -3.42053205e-01 -2.00362384e-01 -1.51272804e-01
-2.45187864e-01 3.90021026e-01 3.42880905e-01 -5.96506000e-01
-1.24931842e-01 -2.58588105e-01 -6.15863264e-01 4.58748758e-01
3.83767962e-01 -5.25460780e-01 -2.20700875e-01 -7.88153768e-01
5.09378076e-01 1.95071682e-01 4.49390054e-01 -9.70778048e-01
-7.45687410e-02 4.97175120e-02 -9.80849192e-02 -3.65186840e-01
-1.98748991e-01 -3.86572599e-01 -5.78029230e-02 5.55844963e-01
-4.87842143e-01 4.37249243e-01 7.01661229e-01 7.35062778e-01
-2.12353408e-01 -3.94359916e-01 4.91420001e-01 -2.76496500e-01
-1.16727555e+00 3.20958465e-01 -2.91236252e-01 2.79128432e-01
6.15840077e-01 5.34005016e-02 -4.58201617e-01 2.91895926e-01
-1.71629056e-01 -1.19055204e-01 8.15665245e-01 6.49952471e-01
4.88405108e-01 -1.19936991e+00 -3.25257510e-01 3.77499223e-01
5.30269325e-01 -3.36682916e-01 9.72314328e-02 4.70349431e-01
-9.02608454e-01 5.48157215e-01 4.66729142e-02 -5.45824707e-01
-1.16164541e+00 1.03825235e+00 4.12489235e-01 -4.17902589e-01
-6.20849133e-01 8.60520601e-01 2.08170488e-01 -8.62624466e-01
1.70296848e-01 -8.13220322e-01 1.13733582e-01 -5.51761746e-01
5.36447287e-01 2.54339576e-01 1.67653337e-01 -4.17262614e-01
-3.20307225e-01 2.73485541e-01 -1.48648873e-01 5.43828011e-01
1.41677070e+00 8.98317993e-01 -6.65717781e-01 4.88978922e-01
1.77682149e+00 7.72547722e-03 -7.78025925e-01 -3.42497140e-01
1.55721277e-01 -4.25794333e-01 2.38331586e-01 -4.30500150e-01
-1.20566726e+00 1.14680970e+00 4.46845740e-01 5.39989173e-02
7.17474222e-01 -1.44267425e-01 9.00725484e-01 5.14171720e-01
4.50385213e-01 -5.64055204e-01 4.19861935e-02 7.04316556e-01
4.78292882e-01 -1.18146253e+00 -2.27380544e-01 2.71318972e-01
-1.66983111e-03 1.48168433e+00 7.52979577e-01 -3.28013778e-01
6.58409119e-01 6.84305847e-01 -3.77358615e-01 -3.93544614e-01
-8.24457169e-01 3.27994496e-01 1.54775932e-01 5.04364252e-01
9.63265121e-01 1.76462963e-01 7.22439885e-02 8.08810771e-01
8.48993361e-02 1.97358027e-01 5.09913683e-01 1.12946260e+00
-1.25141382e-01 -1.20925963e+00 -1.05892904e-01 8.25602293e-01
-6.68490291e-01 -3.69453073e-01 1.30598307e-01 9.21698868e-01
-1.31999910e-01 4.33412284e-01 -2.27159947e-01 -7.07117915e-01
1.99834138e-01 -1.53776305e-02 1.98711529e-01 -9.93405461e-01
-1.01849318e+00 -6.56516373e-01 -8.89172256e-02 -1.03953600e+00
2.30596170e-01 -2.92107165e-01 -1.24874783e+00 -7.11853981e-01
-1.79173186e-01 -4.25881408e-02 5.14553607e-01 3.75034273e-01
5.14255702e-01 1.05576086e+00 2.04372644e-01 -7.49670267e-01
-9.52298880e-01 -6.18534148e-01 -3.01233381e-01 2.94012845e-01
6.00624084e-01 -3.59907180e-01 -3.32092375e-01 -4.89646755e-03]
|
[7.469923496246338, 7.803621292114258]
|
6eacd036-1b55-4854-86fb-3f2f0f08aade
|
cross-paradigm-pretraining-of-convolutional
|
1806.09532
| null |
http://arxiv.org/abs/1806.09532v2
|
http://arxiv.org/pdf/1806.09532v2.pdf
|
Cross-paradigm pretraining of convolutional networks improves intracranial EEG decoding
|
When it comes to the classification of brain signals in real-life
applications, the training and the prediction data are often described by
different distributions. Furthermore, diverse data sets, e.g., recorded from
various subjects or tasks, can even exhibit distinct feature spaces. The fact
that data that have to be classified are often only available in small amounts
reinforces the need for techniques to generalize learned information, as
performances of brain-computer interfaces (BCIs) are enhanced by increasing
quantity of available data. In this paper, we apply transfer learning to a
framework based on deep convolutional neural networks (deep ConvNets) to prove
the transferability of learned patterns in error-related brain signals across
different tasks. The experiments described in this paper demonstrate the
usefulness of transfer learning, especially improving performances when only
little data can be used to distinguish between erroneous and correct
realization of a task. This effect could be delimited from a transfer of merely
general brain signal characteristics, underlining the transfer of
error-specific information. Furthermore, we could extract similar patterns in
time-frequency analyses in identical channels, leading to selective high signal
correlations between the two different paradigms. Classification on the
intracranial data yields in median accuracies up to $(81.50 \pm 9.49)\,\%$.
Decoding on only $10\%$ of the data without pre-training reaches performances
of $(54.76 \pm 3.56)\,\%$, compared to $(64.95 \pm 0.79)\,\%$ with
pre-training.
|
[]
|
2018-07-20
| null | null | null | null |
['eeg-decoding', 'eeg-decoding']
|
['medical', 'time-series']
|
[ 3.53490412e-01 -3.04919444e-02 3.60000283e-01 -4.68072146e-01
-5.75367808e-01 -2.22558200e-01 4.34251904e-01 2.45598584e-01
-7.95494914e-01 1.17660677e+00 -3.62099379e-01 -1.31477267e-01
-5.14378607e-01 -7.02842891e-01 -7.83894360e-01 -7.78777719e-01
-4.82127517e-01 7.68263713e-02 1.19861346e-02 -2.13687003e-01
1.62746206e-01 4.53739345e-01 -1.67546523e+00 5.36880851e-01
8.04373324e-01 1.36364877e+00 4.33193028e-01 3.08072925e-01
9.44851637e-02 1.95237011e-01 -9.28565979e-01 -2.57232368e-01
3.00436858e-02 -5.21337390e-01 -4.07023072e-01 -2.98932672e-01
2.21470505e-01 6.01649955e-02 -2.03618348e-01 1.17365277e+00
6.51723325e-01 -7.97684025e-03 8.81744564e-01 -1.10067630e+00
-3.32666546e-01 5.00585318e-01 -1.90819070e-01 5.27595401e-01
2.56586462e-01 1.98073223e-01 6.54317439e-01 -7.76537240e-01
4.00345832e-01 3.90498877e-01 5.90820014e-01 5.24871469e-01
-1.50403631e+00 -1.00157392e+00 -1.35832697e-01 4.66811061e-01
-1.41832352e+00 -3.68362278e-01 7.97949672e-01 -7.02554226e-01
9.20912325e-01 1.23723239e-01 7.64759183e-01 1.36030519e+00
3.16067040e-01 3.36308777e-01 1.38905180e+00 -2.57487953e-01
2.62100548e-01 5.11142015e-01 3.91137302e-02 1.08070038e-01
4.31053072e-01 2.66431153e-01 -5.49691617e-01 1.90684825e-01
5.15700519e-01 -2.09364012e-01 -5.61274588e-01 2.65215114e-02
-1.13600183e+00 4.83299315e-01 5.02435327e-01 1.00222063e+00
-4.67624515e-01 -1.97762966e-01 5.21454453e-01 5.94835818e-01
2.96410382e-01 6.37832999e-01 -6.37925923e-01 -1.87615767e-01
-8.02811265e-01 1.06062345e-01 5.18673360e-01 7.01615632e-01
7.67275095e-01 2.54714131e-01 -6.58655446e-03 8.79432082e-01
-1.95653960e-01 4.41875905e-01 9.04493809e-01 -5.93726218e-01
2.81240046e-01 3.56451720e-01 -9.54011977e-02 -1.09122324e+00
-6.71772182e-01 -7.86438107e-01 -1.05126667e+00 3.12417805e-01
6.98914468e-01 -2.61426389e-01 -4.80652034e-01 2.03981805e+00
-3.24375153e-01 -1.95731983e-01 -4.37426157e-02 7.18906522e-01
4.57421184e-01 2.61713147e-01 1.05561592e-01 -3.30936283e-01
1.05347848e+00 7.63443159e-03 -5.70618510e-01 -3.53887498e-01
6.77432895e-01 -3.73641819e-01 1.02110314e+00 7.09344864e-01
-9.61288452e-01 -7.48544812e-01 -1.03940666e+00 5.31906426e-01
-4.89075184e-01 1.19482778e-01 4.00047630e-01 5.27543902e-01
-9.61497068e-01 9.43333864e-01 -4.22369689e-01 -1.36345372e-01
5.93564212e-01 6.35087252e-01 -5.30471146e-01 2.74216712e-01
-1.28031337e+00 1.01515615e+00 6.43310487e-01 7.23659694e-02
-6.57904088e-01 -7.53023922e-01 -4.24233496e-01 2.47393116e-01
-6.10985197e-02 -1.16648182e-01 7.60788143e-01 -1.30904162e+00
-1.18453157e+00 8.40225875e-01 1.85997456e-01 -5.08866787e-01
4.80224907e-01 1.07881688e-01 -7.86215365e-01 4.79167625e-02
-3.43242064e-02 5.30516982e-01 8.31281066e-01 -1.02764750e+00
-6.19064152e-01 -5.25306582e-01 -4.00474787e-01 -3.30834061e-01
-3.37320745e-01 -1.56463176e-01 5.27510226e-01 -6.52336836e-01
-4.54351194e-02 -7.04460084e-01 2.61964470e-01 -3.16258341e-01
-3.74557450e-02 -8.60197321e-02 4.03005600e-01 -6.15476310e-01
1.00643981e+00 -2.28808522e+00 9.13036689e-02 4.17462230e-01
2.06076801e-01 3.07166189e-01 -1.35231584e-01 2.60489166e-01
-5.85309267e-01 -4.74672467e-02 -3.84241670e-01 1.90663084e-01
-7.06931949e-02 -9.45080519e-02 -2.21082922e-02 4.97063756e-01
2.79253423e-01 6.07117593e-01 -6.62363350e-01 -9.41371603e-04
4.14838046e-02 3.03024411e-01 -2.94367909e-01 -1.64036844e-02
2.03876838e-01 8.36840272e-01 -1.31545156e-01 1.45360917e-01
4.52968895e-01 5.92353148e-03 1.14490777e-01 -2.63916194e-01
-2.90548112e-02 1.81354776e-01 -8.52688491e-01 1.68747532e+00
-5.05093873e-01 1.03500092e+00 -1.63025841e-01 -1.74565911e+00
1.00017715e+00 5.14602244e-01 5.68922102e-01 -1.17739511e+00
5.21915317e-01 6.43487632e-01 9.31238472e-01 -4.60393578e-01
2.03500427e-02 -3.18757206e-01 9.43102986e-02 3.95492941e-01
5.29676974e-01 -1.12966366e-01 1.45258188e-01 -4.00688827e-01
9.91214395e-01 -2.94256389e-01 9.31756720e-02 -5.94721615e-01
5.45405746e-01 -3.35061848e-01 2.50138134e-01 5.50702512e-01
-2.09952697e-01 2.69125491e-01 5.91910839e-01 -2.12313399e-01
-8.15285802e-01 -9.05179262e-01 -7.62556851e-01 7.64323950e-01
-1.96878672e-01 6.39640167e-02 -8.07758331e-01 -3.30644667e-01
-1.00470968e-01 6.06978476e-01 -6.46328509e-01 -5.48753500e-01
-3.84687185e-01 -7.67301381e-01 5.08991897e-01 4.65716004e-01
3.95196259e-01 -1.34403968e+00 -9.22511041e-01 2.29438812e-01
7.45163858e-02 -1.12090981e+00 3.21969658e-01 5.71348310e-01
-1.00443208e+00 -1.01219773e+00 -7.32169807e-01 -6.13930523e-01
4.99220252e-01 -1.64022356e-01 8.82966816e-01 -9.03963968e-02
-4.71380293e-01 1.58782020e-01 -2.92224765e-01 -5.56952477e-01
-2.61065215e-01 1.19654931e-01 2.62511015e-01 9.60003212e-02
5.64075828e-01 -9.62677658e-01 -6.55486047e-01 2.90104270e-01
-7.45511770e-01 -2.10064992e-01 7.16880441e-01 1.12611628e+00
2.81996846e-01 7.61002302e-02 1.14485860e+00 -4.17526811e-01
7.21961021e-01 -4.53897983e-01 -2.86514968e-01 -7.96591565e-02
-5.52268028e-01 8.88367817e-02 7.44280159e-01 -7.27564514e-01
-6.60162151e-01 -4.02549207e-01 -1.31063491e-01 -1.40016347e-01
-6.22744083e-01 5.47980428e-01 -1.09674044e-01 -2.01352760e-02
1.04282200e+00 4.01285142e-01 -3.86312045e-02 -2.69288361e-01
-1.71667472e-01 7.48605847e-01 3.50756288e-01 -5.01956880e-01
3.78251463e-01 9.74435955e-02 -1.66498318e-01 -9.20915425e-01
-2.40540847e-01 4.44024019e-02 -5.13129234e-01 -1.59118980e-01
6.07272208e-01 -5.10513663e-01 -7.30320513e-01 5.34941018e-01
-9.29724991e-01 -4.24735010e-01 -5.50947368e-01 9.66952384e-01
-6.68035567e-01 1.43772515e-03 -3.25151384e-01 -7.16203630e-01
-1.30232200e-01 -1.17367244e+00 4.79653358e-01 1.09412323e-03
-4.30363238e-01 -6.93244040e-01 -2.12103352e-01 -1.71277612e-01
5.75808108e-01 1.93952098e-01 1.07808101e+00 -1.00777543e+00
5.64350281e-03 -3.42642218e-01 -2.17220739e-01 6.07794344e-01
2.86722362e-01 -4.73720938e-01 -1.22715545e+00 -3.81176919e-01
1.93431973e-01 -1.95260018e-01 5.52608728e-01 3.33256453e-01
1.44712746e+00 7.51079619e-02 -1.49019212e-01 1.97181076e-01
1.19125092e+00 6.76961124e-01 7.67282844e-01 1.47035822e-01
1.34832342e-03 8.94144356e-01 3.61127369e-02 4.08497393e-01
-4.04967040e-01 6.85310483e-01 1.69166833e-01 2.24859983e-01
2.19382569e-02 1.18771568e-01 1.46392971e-01 6.81368470e-01
-3.25807989e-01 1.49235368e-01 -8.14130008e-01 5.18925846e-01
-1.31847155e+00 -8.85721147e-01 -1.38860941e-01 2.56033158e+00
8.77516747e-01 2.94575840e-01 -5.29964603e-02 6.51776969e-01
7.29615569e-01 -2.55131930e-01 -5.23465335e-01 -3.21687967e-01
-2.57570632e-02 8.42087567e-01 2.21151322e-01 -6.73110336e-02
-7.14451313e-01 3.89360368e-01 5.63996315e+00 9.08070445e-01
-1.63339758e+00 1.30955920e-01 6.02623880e-01 -2.16601267e-01
4.38117795e-02 -4.97070253e-01 -3.20670158e-01 8.12671125e-01
1.35573900e+00 -3.21762174e-01 4.12608862e-01 4.89542514e-01
4.19520289e-02 -3.94752830e-01 -1.34608591e+00 1.41643071e+00
-8.46478622e-03 -8.90453875e-01 -4.03990865e-01 6.83271885e-02
4.29114431e-01 -1.45359524e-02 2.25785360e-01 3.33323359e-01
-4.74462688e-01 -1.30170679e+00 7.42424190e-01 4.93305266e-01
9.75593090e-01 -6.20275199e-01 8.01669419e-01 4.87122804e-01
-6.59871697e-01 -1.39404759e-01 -3.17688823e-01 -2.15929195e-01
-1.68409601e-01 6.47189140e-01 -5.35333455e-01 4.58771586e-01
9.70925033e-01 5.18505216e-01 -3.33736420e-01 1.00548780e+00
-1.59795880e-01 4.49395567e-01 -1.93104535e-01 -2.42023081e-01
-2.56147146e-01 -1.16202243e-01 3.26403320e-01 1.11729467e+00
6.92346394e-01 -5.99731132e-02 -5.34694016e-01 9.82495070e-01
-1.31622481e-03 1.96587473e-01 -6.69218957e-01 -2.92231515e-02
1.22738935e-01 1.01341379e+00 -5.60398996e-01 -1.75927997e-01
-3.72278720e-01 7.84774482e-01 2.89726585e-01 4.32013601e-01
-7.84248233e-01 -6.90616667e-01 4.08040404e-01 1.62632421e-01
1.58797935e-01 -2.28191137e-01 -3.66410553e-01 -1.02431262e+00
1.92792848e-01 -7.64223337e-01 -6.43832311e-02 -7.10343361e-01
-1.34334207e+00 9.30996656e-01 1.28652960e-01 -1.26575792e+00
-3.28770310e-01 -1.01314771e+00 -2.79273123e-01 1.13261485e+00
-1.21085203e+00 -3.54664177e-01 -2.60910481e-01 7.49970853e-01
2.16154456e-01 -2.17541218e-01 9.10525739e-01 5.54306984e-01
-1.84257433e-01 6.21492982e-01 9.96096656e-02 4.97748852e-02
7.07290411e-01 -9.33321774e-01 -2.97511250e-01 4.56097096e-01
1.59048274e-01 4.90040541e-01 5.79339504e-01 -3.24137881e-02
-7.91911423e-01 -6.39868975e-01 6.66674376e-01 1.68828536e-02
5.94244719e-01 -4.40599561e-01 -1.19368351e+00 3.41640770e-01
1.49453253e-01 4.63379323e-02 8.86857629e-01 1.39806449e-01
-2.25364774e-01 -4.39056158e-01 -1.23051989e+00 4.64400917e-01
9.94446933e-01 -6.77904785e-01 -6.45453572e-01 -4.82016839e-02
1.07761398e-02 -9.81800184e-02 -1.07589161e+00 4.12157953e-01
6.72829151e-01 -1.13798559e+00 6.41944110e-01 -5.91294527e-01
2.92105794e-01 2.42828518e-01 -2.07662895e-01 -1.66204870e+00
-1.54579774e-01 -2.22062636e-02 2.24715799e-01 8.71454775e-01
5.88869512e-01 -1.04739106e+00 3.99237961e-01 6.73399806e-01
-1.81456819e-01 -6.52528286e-01 -1.20141220e+00 -7.47463584e-01
3.73498321e-01 -7.53988445e-01 5.50101757e-01 8.42437148e-01
3.74978364e-01 -5.83819253e-03 1.42250638e-02 -2.25310430e-01
1.02760844e-01 -2.41553321e-01 2.68831104e-01 -1.55296636e+00
-3.63405824e-01 -7.47415185e-01 -7.29228437e-01 -4.07286137e-01
3.35059226e-01 -9.50393796e-01 -8.20265263e-02 -9.17662203e-01
-6.44563138e-02 -5.43148875e-01 -7.73018360e-01 3.11938405e-01
1.99312747e-01 2.76796252e-01 5.80398478e-02 1.33085370e-01
9.15724039e-02 4.09784049e-01 9.70208645e-01 -2.01944187e-01
-1.38468787e-01 -5.19250333e-03 -4.87368524e-01 6.01836503e-01
1.15651667e+00 -4.28405941e-01 -3.86589617e-01 -2.42409542e-01
1.32583380e-01 2.45092362e-02 4.88401294e-01 -1.51281238e+00
-9.52432752e-02 3.45324367e-01 7.38843322e-01 5.75717352e-02
2.83088773e-01 -1.00168991e+00 2.58034199e-01 6.23643994e-01
-3.72421950e-01 -1.82301491e-01 5.49887657e-01 3.36565197e-01
-4.15939808e-01 -3.40468317e-01 7.59405553e-01 -9.43746790e-02
-5.91690421e-01 7.99915344e-02 -4.91761893e-01 -4.41629179e-02
1.04229105e+00 -4.82703388e-01 -1.10139579e-01 -1.76441059e-01
-1.09135616e+00 -4.16233897e-01 -6.30539469e-03 2.97047943e-01
4.17922527e-01 -1.15983212e+00 -5.73227227e-01 5.23170352e-01
1.74849987e-01 -4.15798724e-01 5.24670541e-01 1.21379840e+00
-4.57699317e-03 4.52603161e-01 -7.50727892e-01 -6.57913148e-01
-8.70089591e-01 4.67261404e-01 5.23579657e-01 8.38043615e-02
-3.73380095e-01 5.74282110e-01 1.76672965e-01 2.99838535e-03
-4.91525531e-02 -4.95573014e-01 -2.54696369e-01 3.14557642e-01
4.37176973e-01 2.26313025e-01 6.42217815e-01 -4.70935583e-01
-2.92179286e-01 4.02696937e-01 6.02809824e-02 -2.18002558e-01
1.32388270e+00 3.06184113e-01 -1.38322292e-02 6.89742267e-01
1.32909656e+00 -3.38058352e-01 -1.00268912e+00 3.09750848e-02
-4.15873230e-02 -3.30921203e-01 -1.48059651e-01 -9.56387281e-01
-1.18127298e+00 1.28472185e+00 1.03617418e+00 3.86548251e-01
1.28453696e+00 -6.52244166e-02 3.27331692e-01 1.87960297e-01
8.38223338e-01 -1.09115660e+00 -1.46022007e-01 2.32925639e-01
1.03176343e+00 -1.10031414e+00 -3.79558742e-01 8.75763744e-02
-3.73852372e-01 1.23310673e+00 3.39821249e-01 -3.32124680e-01
7.40872145e-01 2.11212970e-02 -2.86414206e-01 -1.50781095e-01
-3.59614164e-01 -1.07501306e-01 2.57837504e-01 8.34874213e-01
5.50903976e-01 1.70770586e-01 -7.77209580e-01 1.01502466e+00
-3.45057130e-01 1.94729149e-01 3.35464925e-01 7.21755266e-01
-2.67335206e-01 -1.04524028e+00 -2.73944438e-01 8.58662665e-01
-5.57335496e-01 -4.86691818e-02 2.01394316e-02 9.72688675e-01
3.71298403e-01 9.00306344e-01 2.20510870e-01 -5.74101150e-01
3.49247754e-01 4.34928596e-01 6.36644006e-01 -4.14314359e-01
-5.85730791e-01 -2.26826385e-01 -1.62902951e-01 -4.30399537e-01
-4.09197509e-01 -5.64992309e-01 -1.17184949e+00 -4.36953418e-02
-1.62793502e-01 2.58724958e-01 7.26538777e-01 9.63437736e-01
4.44267213e-01 8.20830226e-01 3.06672364e-01 -8.83258700e-01
-4.59961116e-01 -1.26796055e+00 -9.99997675e-01 5.25048435e-01
1.69517353e-01 -8.63251030e-01 -4.66306806e-01 3.96875925e-02]
|
[13.068241119384766, 3.432602643966675]
|
b12f85d8-30b2-424b-873e-414df891533f
|
molecular-dynamics-simulations-reveal-the
|
1808.08375
| null |
http://arxiv.org/abs/1808.08375v1
|
http://arxiv.org/pdf/1808.08375v1.pdf
|
Molecular dynamics simulations reveal the role of ceramicine B as novel PPAR{\gamma} partial agonist against type 2 diabetes
|
Peroxisome proliferator-activated receptors gamma (PPAR{\gamma}) are
ligand-activated controllers of various metabolic actions and insulin
sensitivity. PPAR{\gamma} is thus considered as an important target to treat
type 2 diabetes. Available PPAR{\gamma} drugs (full agonists) have robust
insulin-sensitizing properties but are accompanied by severe side effects
leading to complicated health problems. Here, we have used molecular docking
and a molecular dynamics simulation study to find a novel PPAR{\gamma} ligand
from a natural product. Our study suggests that the inhibition of ceramicine B
in the PPAR{\gamma} ligand-binding domain (LBD) could act as a partial agonist
and block cdk5-mediated phosphorylation. This result may provide an opportunity
for the development of new anti-diabetic drugs by targeting PPAR{\gamma} while
avoiding the side effects associated with full agonists.
|
[]
|
2018-08-25
| null | null | null | null |
['molecular-docking']
|
['medical']
|
[ 4.27158743e-01 -4.04195517e-01 -7.10927367e-01 -8.80741999e-02
-5.38894892e-01 -6.44781351e-01 5.92255220e-02 4.92685884e-01
-3.39606255e-01 1.19533300e+00 2.46008441e-01 -5.62589943e-01
8.91143084e-02 -5.85286021e-01 -5.25806069e-01 -1.14631355e+00
-1.23738974e-01 3.19063991e-01 -3.59754893e-03 -8.09501484e-02
2.35523403e-01 6.42138422e-01 -1.02155280e+00 1.14629358e-01
1.30566049e+00 5.78358233e-01 2.05002606e-01 8.66915360e-02
-2.28473470e-01 -3.05806875e-01 -2.76299119e-01 1.26638398e-01
4.71266620e-02 -7.49935687e-01 -1.35987639e-01 -3.22398484e-01
1.25839636e-01 -1.13037517e-02 1.42624870e-01 1.02101099e+00
8.20277154e-01 4.78553399e-02 4.65571880e-01 -7.22000122e-01
-4.11398649e-01 1.30001932e-01 -6.80669606e-01 1.72578797e-01
4.71982718e-01 3.74248624e-01 3.83705854e-01 -9.11010742e-01
2.48621657e-01 1.29501176e+00 3.09468418e-01 6.21003091e-01
-1.66164577e+00 -9.02140439e-01 1.89864468e-02 1.22928299e-01
-1.65016365e+00 -4.80883747e-01 3.42951655e-01 -4.66742873e-01
7.37113357e-01 3.71074468e-01 7.03712821e-01 7.12187529e-01
4.50696141e-01 2.25984856e-01 1.32866859e+00 -1.87327459e-01
3.17710161e-01 -5.52003503e-01 -1.54146537e-01 4.93202895e-01
4.99762625e-01 5.59764393e-02 -3.30601394e-01 -5.40589213e-01
8.71026516e-01 -4.31797728e-02 -5.37498116e-01 -5.26808500e-02
-8.16569149e-01 8.32761288e-01 1.18722819e-01 2.45970398e-01
-4.50139374e-01 1.91180781e-01 4.51782912e-01 -5.36086380e-01
-3.81162092e-02 -3.00332811e-03 -6.12825274e-01 1.83077186e-01
8.86797234e-02 2.62264758e-01 4.55101281e-01 3.46303046e-01
5.22713125e-01 2.66583264e-01 -1.18508423e-02 6.89119458e-01
7.33314753e-01 6.02843702e-01 1.18229792e-01 -7.79044390e-01
-1.97425947e-01 7.48592734e-01 1.24403067e-01 -5.59525132e-01
-5.92901349e-01 -1.52506426e-01 -7.36764610e-01 3.68325382e-01
1.14571452e+00 -2.27816775e-02 -4.94282424e-01 1.94815576e+00
5.50392568e-01 1.13703951e-01 -8.99306312e-02 1.16835010e+00
8.12003493e-01 9.06386793e-01 1.20076954e+00 -7.85612702e-01
2.05002475e+00 -1.69134781e-01 -5.91931760e-01 3.37293297e-01
3.86858106e-01 -8.03630710e-01 7.00580060e-01 4.84520257e-01
-7.79672325e-01 1.89652946e-02 -7.76795268e-01 2.81795084e-01
-1.85544908e-01 1.00912094e-01 6.10135436e-01 8.38690579e-01
-4.18999285e-01 4.02401447e-01 -8.45389187e-01 -4.36113209e-01
2.50137061e-01 7.08573341e-01 -2.82878280e-01 -2.04263672e-01
-9.53280389e-01 7.64414847e-01 4.76201057e-01 -4.17670533e-02
-4.30938154e-01 -7.16057062e-01 -5.79513907e-01 -1.05964195e-03
3.14131141e-01 -1.05760002e+00 5.76082885e-01 -7.39009976e-01
-1.42317128e+00 4.39296335e-01 -1.45747513e-01 -9.31551233e-02
-4.05026436e-01 7.58381933e-02 -4.47349459e-01 1.20047942e-01
1.48226261e-01 7.53533170e-02 -2.69923527e-02 -8.86298418e-01
-3.00454080e-01 -8.82413447e-01 -4.73125845e-01 1.07743956e-01
7.26365447e-01 4.08706576e-01 1.62659737e-03 -8.46296668e-01
1.58712044e-01 -9.83728886e-01 -5.08531809e-01 6.79753870e-02
-1.93499818e-01 -4.89290267e-01 3.13721299e-01 -4.29176241e-01
9.30476010e-01 -1.75543356e+00 -1.76814482e-01 1.55982733e-01
-2.35053033e-01 9.16703880e-01 9.30544958e-02 7.14066982e-01
-3.30032915e-01 5.04232943e-01 2.39547685e-01 1.25641680e+00
-3.15818667e-01 -5.15978001e-02 1.31919414e-01 1.04741073e+00
-2.86180764e-01 5.37326455e-01 -1.04944551e+00 1.04165517e-01
2.38511071e-01 7.33733773e-01 -4.00406748e-01 -6.58684015e-01
-8.21033835e-01 1.14445210e+00 -9.20550644e-01 9.53187943e-01
9.13797259e-01 5.31828701e-02 1.07801378e+00 -3.51983339e-01
-4.61171269e-01 1.82441115e-01 -9.07913208e-01 1.46232760e+00
3.15933704e-01 -1.62010193e-01 4.17155698e-02 -6.98192060e-01
9.39243078e-01 7.17778921e-01 5.20502865e-01 -1.08536446e+00
2.93424577e-01 3.52415502e-01 7.79483244e-02 -4.49062049e-01
-5.44794977e-01 -7.70515382e-01 1.90045729e-01 1.56729091e-02
-4.47405666e-01 9.76052523e-01 2.89784551e-01 -3.46496761e-01
7.53774762e-01 3.34398299e-01 6.72341824e-01 -7.84664989e-01
8.15493464e-01 2.54690237e-02 1.06651378e+00 1.51887238e-01
1.05830029e-01 -1.08850993e-01 3.37218821e-01 -6.00377619e-01
-7.53893197e-01 -9.16866064e-01 -2.68884808e-01 7.44445741e-01
2.08620176e-01 -2.12388292e-01 -1.12087078e-01 8.85049929e-04
5.88973202e-02 7.15402842e-01 -2.23247986e-02 2.61675324e-02
-7.91349232e-01 -1.06785083e+00 8.03762734e-01 1.12472676e-01
4.60609853e-01 -2.84340739e-01 -1.20741770e-01 6.15018547e-01
-3.19259614e-03 -3.89418453e-01 -3.71860892e-01 2.63872985e-02
-8.22749197e-01 -1.32209587e+00 -7.33024836e-01 -6.27565622e-01
3.82771194e-01 1.63188070e-01 3.05943340e-01 -3.74889337e-02
-1.93214998e-01 -2.10488543e-01 -4.42164987e-02 -5.20681322e-01
-2.53932953e-01 -5.48719227e-01 3.68914843e-01 -4.05127138e-01
3.39751393e-01 -1.08980703e+00 -1.32260752e+00 5.64190149e-01
-7.46262610e-01 -2.77160645e-01 6.26021802e-01 2.80800372e-01
8.73996496e-01 -3.27961445e-01 1.02071035e+00 -7.06539512e-01
4.50061619e-01 -5.22519290e-01 -1.07834375e+00 -3.05122197e-01
-3.47614318e-01 -3.35538201e-02 7.70775974e-01 -3.27581078e-01
-1.21915281e+00 2.77275592e-01 -2.75690436e-01 1.81137368e-01
-2.03089848e-01 5.41060388e-01 -9.37125921e-01 1.42048197e-02
6.07673407e-01 1.49330184e-01 2.46616960e-01 -8.73409331e-01
-8.67276266e-03 -1.17659979e-01 3.52520019e-01 -7.35687435e-01
2.04646185e-01 3.24180394e-01 5.73785663e-01 -9.82682467e-01
-4.15816486e-01 -4.93517756e-01 3.70631188e-01 1.40210211e-01
1.03563368e+00 -1.13998950e+00 -1.63611126e+00 -1.28290087e-01
-9.95015442e-01 -7.19453171e-02 5.49972117e-01 1.12099862e+00
-1.48093745e-01 5.10946631e-01 -3.94189984e-01 -3.17673594e-01
-3.13732624e-01 -1.06262064e+00 2.28007630e-01 4.67406064e-01
-1.77708194e-01 -6.75374210e-01 4.97693211e-01 4.56497371e-01
-2.22233217e-02 6.42682731e-01 1.53748858e+00 -6.91219568e-01
-6.44979298e-01 1.70902938e-01 -3.80403459e-01 -2.03923479e-01
1.58240035e-01 -6.42590150e-02 -2.94141710e-01 8.06286559e-02
-5.01882017e-01 -3.62757742e-02 5.95069349e-01 6.06350303e-01
7.75519907e-01 -7.85222411e-01 -5.91089547e-01 4.61921990e-01
1.62975633e+00 8.55472863e-01 9.76357937e-01 1.45171180e-01
2.73842961e-01 -1.76863655e-01 7.25176513e-01 5.25466979e-01
-2.03111112e-01 8.54074538e-01 4.13151503e-01 -1.77761510e-01
7.80064687e-02 -1.18829452e-01 1.24028236e-01 -2.56517202e-01
-5.64411700e-01 -2.98433900e-01 -7.11070001e-01 -4.66866698e-03
-1.54211378e+00 -1.12567413e+00 -1.02058733e+00 2.27904558e+00
1.16939652e+00 -6.05918646e-01 4.02663141e-01 -4.30008799e-01
7.37170279e-01 -4.25605744e-01 -4.46656227e-01 -4.82760310e-01
-1.70269310e-01 7.11235285e-01 7.47491717e-01 3.32670331e-01
-5.09734571e-01 4.32410985e-01 5.98263121e+00 6.49484038e-01
-1.11784041e+00 -3.31216156e-01 2.32717231e-01 2.37421483e-01
-2.71977872e-01 6.85028791e-01 -7.36437678e-01 6.51088476e-01
5.98241985e-01 -3.51947606e-01 -6.82987943e-02 5.11039734e-01
9.68479216e-01 -6.80895090e-01 -5.89394748e-01 7.53841937e-01
-5.46127319e-01 -1.70302713e+00 -2.24393792e-02 6.20445609e-01
3.84191692e-01 -4.91259456e-01 -2.83254385e-01 -4.73799333e-02
2.31965348e-01 -9.54300106e-01 -5.74572757e-02 7.31684625e-01
9.67358887e-01 -9.13354695e-01 3.01752537e-01 2.29901522e-01
-8.98570001e-01 2.59424061e-01 -2.58272886e-01 -7.14431629e-02
3.77188742e-01 6.51944399e-01 -8.43594491e-01 1.85313106e-01
7.84125999e-02 -3.72601449e-01 9.31479875e-03 1.25141311e+00
-1.55099496e-01 4.10356581e-01 -1.33577615e-01 9.86386091e-02
-5.02529368e-02 -5.99501848e-01 6.10228181e-01 7.52299190e-01
8.92067701e-02 1.07001531e+00 7.09270835e-01 6.30732954e-01
-1.54264882e-01 4.66432899e-01 -2.93716788e-01 1.55843347e-02
2.19920218e-01 8.92863750e-01 -7.93675125e-01 -1.31092399e-01
-5.12473822e-01 4.21198249e-01 -3.17497641e-01 4.35436279e-01
-7.12839544e-01 1.19186886e-01 9.18192923e-01 5.74748635e-01
1.38081864e-01 -1.96547538e-01 6.77924007e-02 -4.52583432e-01
-5.41738629e-01 -1.05323994e+00 6.30475938e-01 -2.20707282e-01
-7.57797182e-01 -5.02573013e-01 7.66089484e-02 -7.99584150e-01
4.60007131e-01 -1.89183488e-01 -4.47559386e-01 1.00517166e+00
-1.12074697e+00 -1.08168519e+00 2.24439085e-01 3.54991466e-01
2.87161440e-01 3.93496990e-01 1.15586591e+00 3.03746104e-01
-7.46346295e-01 4.55298573e-02 2.64807016e-01 -2.69261241e-01
8.62983942e-01 -7.51944959e-01 -8.49626839e-01 1.10922545e-01
-7.44762123e-01 1.05071914e+00 1.04202592e+00 -8.82161617e-01
-1.69604599e+00 -1.07279718e+00 8.24558198e-01 3.94167364e-01
2.23280966e-01 2.09146246e-01 -6.07946515e-01 5.49263179e-01
1.06314652e-01 -3.54533583e-01 1.43916094e+00 -2.55683154e-01
-7.84365088e-02 3.24356779e-02 -1.29049826e+00 7.80416250e-01
3.87970328e-01 3.05554539e-01 2.09183730e-02 4.13225502e-01
1.05226338e-01 -3.73628289e-02 -1.00395155e+00 4.24478590e-01
6.99788988e-01 -7.22881019e-01 1.04916036e+00 -4.31805015e-01
-1.60490483e-01 -1.00123179e+00 5.15557304e-02 -1.00479662e+00
-4.83124554e-01 -8.43886971e-01 3.11790973e-01 9.38739479e-01
1.48625553e-01 -4.74063367e-01 4.61506605e-01 6.44746721e-01
-4.42047305e-02 -4.40584153e-01 -1.35697794e+00 -7.43631065e-01
-1.80097818e-01 2.74472356e-01 1.33408114e-01 8.67417395e-01
5.19710541e-01 5.51900864e-01 -1.72682360e-01 -1.86943747e-02
5.58555424e-01 -1.91040635e-01 6.96544349e-01 -1.11751854e+00
1.13749646e-01 -3.18430662e-01 -2.13668644e-01 -5.41345000e-01
-1.15615159e-01 -7.44335532e-01 -5.20261884e-01 -1.50671816e+00
2.07794324e-01 -5.30738473e-01 7.43909031e-02 5.04988551e-01
2.43118525e-01 2.92026550e-01 -1.88232139e-01 -2.62506455e-01
-2.08399117e-01 2.88966179e-01 1.10327053e+00 2.48918757e-02
-5.74034572e-01 8.11097547e-02 -1.09330404e+00 8.23958576e-01
1.02551687e+00 -5.90664208e-01 -1.06952891e-01 3.84262085e-01
4.50327665e-01 3.58350456e-01 1.43594876e-01 -1.42563179e-01
-1.44648194e-01 -9.59701359e-01 6.28424883e-01 -3.66127431e-01
2.19443500e-01 -6.20609641e-01 9.01151478e-01 7.87760198e-01
2.72800446e-01 -3.15808177e-01 2.77336895e-01 5.62849045e-01
5.52239597e-01 3.59374098e-02 1.12678409e+00 -4.34947014e-01
-4.04652029e-01 -3.59924175e-02 -1.48080873e+00 -3.61135334e-01
1.22499979e+00 -3.78849000e-01 -2.26056919e-01 3.60883586e-02
-1.16857004e+00 2.41969619e-02 6.76234841e-01 -1.37662351e-01
2.95975149e-01 -1.26157832e+00 -5.51189125e-01 -2.64080822e-01
2.84945428e-01 -3.19478095e-01 3.59350204e-01 1.11254227e+00
-8.41499507e-01 7.61133850e-01 -3.26894641e-01 -5.44142962e-01
-1.53352559e+00 8.94006014e-01 2.57039517e-01 1.60745494e-02
-3.30708742e-01 2.74660379e-01 3.10223281e-01 2.92291403e-01
1.12296820e-01 1.43037125e-01 -1.07040107e-01 -2.49987096e-01
6.89020157e-01 7.58511066e-01 -2.62055665e-01 -7.38904119e-01
-5.73646426e-01 1.41055688e-01 2.21380219e-01 5.87785780e-01
1.20642042e+00 -8.12980533e-02 -2.90906638e-01 -2.02609152e-01
7.25188375e-01 2.77046233e-01 -6.83885574e-01 -6.38564229e-02
-1.28442585e-01 -6.33481681e-01 -1.78038418e-01 -9.94721174e-01
-6.47558749e-01 4.00804132e-02 8.35757732e-01 -7.69309759e-01
8.00041199e-01 1.57398328e-01 7.35923946e-01 -4.04434614e-02
2.03586251e-01 -9.07231212e-01 -4.57008220e-02 -2.14677546e-02
7.90109634e-01 -5.52706003e-01 3.27192634e-01 -7.70846009e-01
-2.22305253e-01 1.06741893e+00 6.62166536e-01 -5.09081334e-02
2.86454886e-01 -1.81047797e-01 -2.59298980e-02 -2.29591623e-01
-6.61364615e-01 -1.53736353e-01 -1.05070457e-01 7.27597713e-01
8.90777767e-01 1.37179205e-02 -1.75074697e+00 6.45928204e-01
7.90668964e-01 3.67238939e-01 2.81171083e-01 7.53550768e-01
-6.37700260e-01 -1.78705931e+00 -7.11800516e-01 1.05145350e-02
-7.65507042e-01 8.32565278e-02 -2.58962661e-01 8.29326153e-01
4.65871990e-01 7.65838981e-01 -4.61466342e-01 6.08209491e-01
2.65941679e-01 3.97281826e-01 7.22626567e-01 -3.76807958e-01
-4.02826130e-01 1.18561232e+00 3.61675560e-01 -3.86100262e-01
-4.17837173e-01 -3.26502204e-01 -1.86040640e+00 -4.75659013e-01
-3.11364561e-01 2.27412879e-02 6.89350009e-01 8.77572417e-01
4.44335252e-01 3.68693657e-02 3.92459571e-01 -3.92116576e-01
-2.91397244e-01 -5.25157928e-01 -9.04825628e-01 -3.53872813e-02
-2.08091214e-01 -6.39243841e-01 2.71091759e-01 1.83853149e-01]
|
[4.702845573425293, 5.105762958526611]
|
89c4e303-d7f4-4181-b737-22a2bdb482a1
|
knowledge-based-review-generation-by
|
2105.03815
| null |
https://arxiv.org/abs/2105.03815v1
|
https://arxiv.org/pdf/2105.03815v1.pdf
|
Knowledge-based Review Generation by Coherence Enhanced Text Planning
|
As a natural language generation task, it is challenging to generate informative and coherent review text. In order to enhance the informativeness of the generated text, existing solutions typically learn to copy entities or triples from knowledge graphs (KGs). However, they lack overall consideration to select and arrange the incorporated knowledge, which tends to cause text incoherence. To address the above issue, we focus on improving entity-centric coherence of the generated reviews by leveraging the semantic structure of KGs. In this paper, we propose a novel Coherence Enhanced Text Planning model (CETP) based on knowledge graphs (KGs) to improve both global and local coherence for review generation. The proposed model learns a two-level text plan for generating a document: (1) the document plan is modeled as a sequence of sentence plans in order, and (2) the sentence plan is modeled as an entity-based subgraph from KG. Local coherence can be naturally enforced by KG subgraphs through intra-sentence correlations between entities. For global coherence, we design a hierarchical self-attentive architecture with both subgraph- and node-level attention to enhance the correlations between subgraphs. To our knowledge, we are the first to utilize a KG-based text planning model to enhance text coherence for review generation. Extensive experiments on three datasets confirm the effectiveness of our model on improving the content coherence of generated texts.
|
['Ji-Rong Wen', 'Nicholas Jing Yuan', 'Zhicheng Wei', 'Wayne Xin Zhao', 'Junyi Li']
|
2021-05-09
| null | null | null | null |
['review-generation']
|
['natural-language-processing']
|
[ 2.58812726e-01 1.04189730e+00 -3.65609020e-01 -4.55446333e-01
-5.80556571e-01 -1.55026495e-01 8.15595567e-01 2.52718747e-01
1.50580645e-01 9.97002542e-01 9.24497128e-01 -7.12672621e-02
-6.42739758e-02 -1.12909877e+00 -8.57681334e-01 -2.81432569e-01
3.14376950e-01 3.55485797e-01 1.15473501e-01 -3.80586445e-01
4.01949733e-01 -2.63175428e-01 -1.13691521e+00 4.04700309e-01
1.58389282e+00 4.61162776e-01 5.06810844e-01 3.94082963e-01
-4.67954487e-01 1.20014691e+00 -5.09043992e-01 -5.16249835e-01
-8.94758627e-02 -8.56828988e-01 -1.07864857e+00 3.97094190e-01
1.27825677e-01 -3.32392097e-01 -3.01233768e-01 8.72503996e-01
2.42248759e-01 1.89855605e-01 7.46749997e-01 -1.12883103e+00
-1.03095376e+00 1.66051638e+00 -5.36292374e-01 -7.79331028e-02
4.82675493e-01 1.75618306e-02 1.51710033e+00 -7.30116248e-01
8.06970775e-01 1.31299436e+00 1.40430316e-01 3.25884134e-01
-7.97784984e-01 -3.59898388e-01 8.17076504e-01 1.40842363e-01
-1.24985337e+00 -1.48267329e-01 9.82976317e-01 -1.11646054e-03
1.16781390e+00 1.04364686e-01 9.11298633e-01 8.22268367e-01
3.74392658e-01 1.06781220e+00 6.22168720e-01 -4.42552745e-01
1.59740552e-01 1.00110784e-01 2.91813046e-01 7.26565361e-01
5.94613075e-01 -4.79458719e-01 -8.88123453e-01 1.79869160e-01
4.09516394e-01 -2.28741705e-01 -3.46374124e-01 -1.31862164e-01
-1.14139593e+00 8.49902213e-01 7.12676048e-01 2.61468023e-01
-6.60303593e-01 2.58903831e-01 1.41960695e-01 -2.72833910e-02
5.01381874e-01 7.09994376e-01 -3.41775507e-01 2.26428196e-01
-7.52817631e-01 3.16106081e-01 9.05407250e-01 1.42326057e+00
9.48042214e-01 -6.00796714e-02 -6.39239788e-01 4.74943310e-01
6.29290342e-01 4.03335094e-01 5.28849244e-01 -4.33267593e-01
8.73356402e-01 9.45686042e-01 -4.26282883e-02 -1.44928563e+00
-3.06464016e-01 -4.53858465e-01 -9.01475728e-01 -7.78706133e-01
-3.83219093e-01 -4.12622333e-01 -8.31746459e-01 1.79519320e+00
3.15220267e-01 -6.92650080e-02 1.63938016e-01 6.07807994e-01
1.19401467e+00 7.99382031e-01 1.14289775e-01 -3.85505140e-01
1.19394851e+00 -1.38571560e+00 -9.58559752e-01 -4.08656687e-01
8.18382978e-01 -4.39376622e-01 9.33749855e-01 -1.27634972e-01
-1.04017484e+00 -3.91559660e-01 -1.02086985e+00 -7.82300383e-02
-2.23934412e-01 1.38912261e-01 5.33325791e-01 1.08220443e-01
-1.07981420e+00 3.72652978e-01 -4.57104325e-01 -4.31221753e-01
2.53444880e-01 8.65852386e-02 -8.32786188e-02 -1.74826071e-01
-1.56349623e+00 6.60694003e-01 7.33000100e-01 1.13419831e-01
-5.04968464e-01 -6.15790367e-01 -1.04280901e+00 3.25851113e-01
7.43249416e-01 -1.29537022e+00 1.10435247e+00 -7.45542765e-01
-1.50353444e+00 2.19412774e-01 -2.38322452e-01 -2.24883467e-01
-2.76994426e-02 -2.20500708e-01 -1.33282200e-01 2.16808990e-01
3.54828477e-01 8.92097533e-01 6.17466509e-01 -1.40133214e+00
-4.93858993e-01 -1.01639368e-01 2.73643047e-01 9.13119555e-01
-4.49811310e-01 -4.37331766e-01 -5.60441613e-01 -7.29400933e-01
-6.40799329e-02 -7.52617776e-01 -3.86512458e-01 -7.96412528e-01
-1.10913265e+00 -4.51641440e-01 3.69806707e-01 -6.75648510e-01
1.51256168e+00 -1.52208364e+00 2.52624899e-01 1.36669874e-01
2.84929574e-01 -2.57080961e-02 -3.18735152e-01 8.93825114e-01
2.42110252e-01 5.30274332e-01 -2.05520108e-01 -3.93748790e-01
2.11708948e-01 1.07891001e-01 -6.24055266e-01 -2.16627359e-01
3.43478858e-01 1.33612359e+00 -1.26228595e+00 -6.65499449e-01
-2.60045201e-01 8.69855881e-02 -5.78307331e-01 1.02138638e-01
-7.02736497e-01 1.87518910e-01 -1.01669753e+00 2.20726252e-01
4.78864938e-01 -5.92114210e-01 2.98298627e-01 -3.16634655e-01
2.15511382e-01 7.31289923e-01 -9.30055916e-01 1.65394747e+00
-3.99108618e-01 1.94357976e-01 -4.49596941e-01 -6.88701570e-01
8.86422634e-01 1.88168392e-01 2.14336142e-01 -6.50304198e-01
-1.41938433e-01 -1.09559692e-01 -2.29512274e-01 -2.55532533e-01
1.23127985e+00 -9.08205882e-02 -2.17857301e-01 9.03989971e-01
-6.17910875e-03 -5.20711303e-01 6.01480961e-01 1.12706828e+00
1.14036679e+00 -2.99148932e-02 4.35295403e-01 -2.07917437e-01
4.12539542e-01 8.45896006e-02 3.82846683e-01 8.28189015e-01
3.52922022e-01 4.60690409e-01 8.61456513e-01 1.85403954e-02
-7.26704299e-01 -7.37366617e-01 4.83639151e-01 6.12670243e-01
3.33423078e-01 -1.02638555e+00 -6.73390806e-01 -9.43052590e-01
-2.36055300e-01 1.04798222e+00 -5.02826929e-01 -4.42185104e-01
-4.47218716e-01 -5.88333368e-01 1.75568104e-01 5.56073844e-01
5.76612890e-01 -1.25891078e+00 7.24715889e-02 2.87219673e-01
-6.44715190e-01 -1.21664178e+00 -1.01874864e+00 -2.99944937e-01
-6.50797069e-01 -9.77412641e-01 -2.08229467e-01 -6.37226403e-01
1.15363419e+00 4.86973345e-01 1.41023910e+00 2.95986593e-01
2.91393429e-01 6.02507889e-01 -8.17389131e-01 -3.73469770e-01
-3.42647672e-01 5.55066466e-01 -1.55304208e-01 -1.01134926e-01
4.74932529e-02 -4.40401852e-01 -5.46218872e-01 -2.33900268e-02
-1.12826443e+00 6.39896572e-01 7.61424899e-01 6.51522160e-01
5.11155605e-01 4.41331416e-01 1.00718248e+00 -1.10492134e+00
1.21445167e+00 -5.88765681e-01 -1.57324389e-01 5.56572914e-01
-8.26561332e-01 3.38897467e-01 7.67246962e-01 9.32977572e-02
-1.44781053e+00 -2.58503199e-01 2.25942537e-01 1.58396378e-01
3.65288883e-01 1.17476749e+00 -1.84078112e-01 5.19953609e-01
3.84714693e-01 4.31503266e-01 -4.45597231e-01 1.48469925e-01
8.28309298e-01 3.99534434e-01 4.89593409e-02 -5.55596054e-01
7.55533159e-01 1.07550189e-01 -1.12256281e-01 -5.07530212e-01
-1.19961655e+00 -2.41923198e-01 -4.80610996e-01 -1.57804206e-01
8.61709297e-01 -1.00856102e+00 -2.02179328e-01 2.33280510e-01
-1.22243237e+00 -2.15576902e-01 -4.44128692e-01 2.38327950e-01
-3.48341584e-01 5.46664417e-01 -4.72571462e-01 -5.22223949e-01
-7.90931046e-01 -7.77756035e-01 1.23505056e+00 4.62834835e-01
-1.89618841e-01 -1.10133374e+00 1.10671893e-01 3.86197954e-01
2.48773515e-01 9.55952425e-03 9.46537554e-01 -5.60989022e-01
-9.72068787e-01 -5.08874189e-03 -2.53966630e-01 -5.05570583e-02
3.87521297e-01 7.25911036e-02 -4.33687419e-01 5.06898202e-02
-3.22760403e-01 -4.79123622e-01 9.50565457e-01 2.31564134e-01
7.12134242e-01 -8.41594160e-01 -4.57263619e-01 -6.08067513e-02
1.07661259e+00 -1.39535233e-01 6.29712760e-01 1.28659159e-02
1.00001764e+00 7.38204896e-01 7.01200247e-01 5.53484440e-01
1.18116391e+00 4.05375004e-01 3.03917348e-01 1.14455685e-01
-2.27368534e-01 -9.08131421e-01 3.97459626e-01 1.34121513e+00
2.58219868e-01 -6.35447979e-01 -6.44782722e-01 6.10646963e-01
-2.06980515e+00 -9.91776824e-01 -1.55754119e-01 1.63853431e+00
1.16152501e+00 -2.49543432e-02 -3.36946964e-01 -3.30797404e-01
7.18910933e-01 4.39936608e-01 -3.39933604e-01 -2.15314791e-01
-4.25994247e-02 -2.86023229e-01 -4.15901355e-02 6.29062831e-01
-7.08947659e-01 1.35607302e+00 5.21311474e+00 8.23059380e-01
-6.41796052e-01 -3.77272993e-01 5.55049121e-01 -3.34017016e-02
-1.10991406e+00 2.29107827e-01 -1.06037331e+00 4.06996578e-01
4.88004416e-01 -7.30065942e-01 1.98460653e-01 6.93847001e-01
2.12563053e-01 -1.70231327e-01 -8.28449249e-01 3.68207067e-01
2.98034191e-01 -1.46333683e+00 7.20857918e-01 -1.43752888e-01
1.29066622e+00 -4.10509765e-01 -3.02101791e-01 5.88589787e-01
6.95164263e-01 -8.47707629e-01 5.43224454e-01 6.29394650e-01
3.04941118e-01 -8.73075068e-01 7.16351628e-01 4.07017529e-01
-1.47814882e+00 2.53125161e-01 -4.26479071e-01 1.91879764e-01
5.10582805e-01 9.47600126e-01 -1.09701622e+00 1.19978130e+00
1.64679781e-01 1.03375816e+00 -6.60456002e-01 4.49884892e-01
-8.80202353e-01 3.74264210e-01 -2.89724562e-02 -3.18379879e-01
3.58991116e-01 -2.40717858e-01 5.18021822e-01 1.05316031e+00
3.07592243e-01 2.95594573e-01 4.24992859e-01 1.03486526e+00
-4.98203158e-01 3.12640160e-01 -6.13829076e-01 -4.43186343e-01
4.40341383e-01 1.51599908e+00 -6.91542685e-01 -4.64917988e-01
-1.67999327e-01 8.52938175e-01 5.52902341e-01 4.20443386e-01
-6.23113930e-01 -4.90496933e-01 1.01409711e-01 -7.20307678e-02
4.70342487e-01 7.62001285e-03 -4.14042324e-01 -1.30921483e+00
1.80389643e-01 -6.68576896e-01 2.72931546e-01 -1.02315998e+00
-1.13760436e+00 6.00259721e-01 3.05500161e-02 -7.95358062e-01
-3.63345951e-01 1.76671252e-01 -9.18223560e-01 7.85072505e-01
-1.66226423e+00 -1.30994606e+00 -3.98626596e-01 3.61182630e-01
5.08560598e-01 2.18072236e-02 3.97008866e-01 -3.15014988e-01
-4.93720889e-01 3.78323048e-01 -4.82106119e-01 -2.39958733e-01
4.98715043e-01 -1.39162159e+00 4.84295279e-01 1.08851349e+00
1.27025634e-01 8.33466768e-01 5.12452960e-01 -1.28468585e+00
-1.28081512e+00 -1.39950144e+00 1.27348435e+00 -1.40352607e-01
6.18108571e-01 -3.34267356e-02 -7.82718360e-01 7.94921637e-01
6.72281504e-01 -6.24268234e-01 6.69585526e-01 2.10999206e-01
-8.43841061e-02 -4.57180850e-02 -5.92551410e-01 1.07777035e+00
1.09865236e+00 -3.18767935e-01 -6.56902552e-01 5.81418276e-01
1.36495817e+00 -3.08683932e-01 -7.31192172e-01 4.11800265e-01
-6.59884289e-02 -6.31079197e-01 4.69321817e-01 -4.08185810e-01
1.00272918e+00 -4.75144863e-01 3.09187084e-01 -1.76163363e+00
-3.83261293e-01 -5.89663148e-01 -3.34391683e-01 1.47846067e+00
6.81050658e-01 -3.55350345e-01 6.04668319e-01 6.29547656e-01
-5.88588893e-01 -1.00021005e+00 -3.89970541e-01 -3.31042647e-01
-2.15550616e-01 -1.14636205e-01 8.06230783e-01 8.17891836e-01
5.41549861e-01 9.92704332e-01 -4.17992234e-01 1.40983507e-01
2.10895464e-01 3.76458138e-01 8.61025631e-01 -6.72925234e-01
-2.80727088e-01 -4.21540231e-01 2.89413929e-01 -1.29544628e+00
2.91400194e-01 -1.09140432e+00 3.35739225e-01 -2.42182755e+00
5.36106348e-01 -2.44740620e-01 1.88463092e-01 5.09218156e-01
-8.23747754e-01 -4.63391721e-01 3.21578920e-01 2.29862034e-02
-1.09179616e+00 1.18359458e+00 1.72696483e+00 -1.34321645e-01
-3.85834455e-01 -2.68885672e-01 -1.47397041e+00 3.92902434e-01
8.02323163e-01 -3.27563554e-01 -1.08625579e+00 -3.68477672e-01
8.07339251e-01 7.93079808e-02 -1.06089629e-01 -4.59929556e-01
6.32642031e-01 -3.14869404e-01 3.06264497e-03 -8.68315220e-01
-4.78615351e-02 -3.45998347e-01 2.97620296e-02 1.37427032e-01
-6.01154029e-01 -1.11768723e-01 -4.09992635e-02 8.77142966e-01
-2.76540101e-01 -2.57292122e-01 1.13921098e-01 -3.33841056e-01
-4.64525372e-01 3.52589637e-01 -3.99873883e-01 1.61061525e-01
7.73057878e-01 1.97645426e-01 -8.28563213e-01 -6.36826336e-01
-2.49906376e-01 7.81699181e-01 2.41707727e-01 4.76918846e-01
7.71230817e-01 -1.37994969e+00 -7.44632185e-01 -2.43326783e-01
2.83834934e-01 4.75988001e-01 4.59335446e-01 7.26478040e-01
-1.12586014e-01 8.23273301e-01 2.40123153e-01 -1.34920344e-01
-8.58270109e-01 4.44337994e-01 -1.38522321e-02 -1.18844092e+00
-4.81045306e-01 7.36417711e-01 4.31041747e-01 -4.84513313e-01
-1.83613807e-01 -3.57359529e-01 -5.01154959e-01 7.16361031e-02
5.32655478e-01 1.59679819e-02 -1.20836683e-01 -2.22672194e-01
-5.45458123e-02 2.84397781e-01 -5.68190634e-01 -6.36783838e-02
1.14373922e+00 -4.64470685e-01 -4.48886275e-01 8.74210969e-02
6.65300071e-01 1.52640477e-01 -1.00690019e+00 -2.93181777e-01
-2.96906736e-02 -1.73798800e-01 -2.51869410e-02 -7.96813846e-01
-1.00452447e+00 2.86458015e-01 -7.87532151e-01 3.21173936e-01
9.95681226e-01 9.19359326e-02 9.38010633e-01 5.74468017e-01
2.84465879e-01 -1.13729978e+00 6.14301205e-01 7.43674159e-01
1.21195936e+00 -9.80780482e-01 2.30559111e-01 -8.06354940e-01
-1.20230877e+00 8.88987303e-01 9.85843778e-01 2.40565225e-01
4.18183535e-01 -1.93875700e-01 -4.13225532e-01 -4.13804770e-01
-1.19684911e+00 -3.01152855e-01 4.66316998e-01 5.29582620e-01
4.29551929e-01 1.96666978e-02 -5.35607934e-01 8.99778128e-01
-3.36703151e-01 -4.46329378e-02 8.14318299e-01 8.64273787e-01
-6.35380208e-01 -1.09643042e+00 1.12973422e-01 6.69858992e-01
4.77468967e-02 -5.55414855e-01 -7.41829991e-01 4.30676043e-01
-2.83279806e-01 1.10469139e+00 -2.27170974e-01 -3.98775220e-01
1.79233626e-01 -8.89705792e-02 2.38520354e-01 -1.12805724e+00
-5.86571693e-01 -2.29174808e-01 4.54271317e-01 -2.79695362e-01
-4.02954906e-01 -3.89061302e-01 -1.60069144e+00 -3.14881623e-01
-4.25851136e-01 3.66131365e-01 3.00289243e-01 1.15685844e+00
7.73774505e-01 9.15888309e-01 7.48608410e-01 -3.97117734e-01
-3.40801686e-01 -1.01281607e+00 -7.39865541e-01 2.33728513e-01
-2.13522583e-01 -2.73797095e-01 -1.28082335e-01 6.45682439e-02]
|
[11.910406112670898, 8.908956527709961]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.