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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
baf41ad0-c510-4f9d-9828-07cbdb38c593
|
animatable-neural-radiance-fields-from
|
2106.13629
| null |
https://arxiv.org/abs/2106.13629v2
|
https://arxiv.org/pdf/2106.13629v2.pdf
|
Animatable Neural Radiance Fields from Monocular RGB Videos
|
We present animatable neural radiance fields (animatable NeRF) for detailed human avatar creation from monocular videos. Our approach extends neural radiance fields (NeRF) to the dynamic scenes with human movements via introducing explicit pose-guided deformation while learning the scene representation network. In particular, we estimate the human pose for each frame and learn a constant canonical space for the detailed human template, which enables natural shape deformation from the observation space to the canonical space under the explicit control of the pose parameters. To compensate for inaccurate pose estimation, we introduce the pose refinement strategy that updates the initial pose during the learning process, which not only helps to learn more accurate human reconstruction but also accelerates the convergence. In experiments we show that the proposed approach achieves 1) implicit human geometry and appearance reconstruction with high-quality details, 2) photo-realistic rendering of the human from novel views, and 3) animation of the human with novel poses.
|
['Huchuan Lu', 'Xu Jia', 'Linchao Bao', 'Xuefei Zhe', 'Di Kang', 'Ying Zhang', 'Jianchuan Chen']
|
2021-06-25
| null | null | null | null |
['3d-human-reconstruction']
|
['computer-vision']
|
[ 8.11079666e-02 1.77091941e-01 4.02204335e-01 -2.13622853e-01
-2.54820466e-01 -4.08344686e-01 5.47958434e-01 -5.28563917e-01
-3.56100947e-01 7.63816535e-01 1.35490939e-01 4.37512219e-01
1.69573873e-01 -7.65317798e-01 -9.76812899e-01 -7.04632044e-01
1.38816327e-01 5.43246865e-01 6.33018985e-02 -2.14098871e-01
-2.19277501e-01 8.12519491e-01 -1.34998167e+00 -3.90694261e-01
6.01693690e-01 8.22914779e-01 8.51473957e-02 7.97832727e-01
5.20781577e-01 7.68138468e-01 -4.72685516e-01 -3.99138361e-01
6.44071102e-01 -4.98644680e-01 -5.54209948e-01 5.11628389e-01
7.77178049e-01 -5.71638763e-01 -5.04807234e-01 8.16536844e-01
4.03182179e-01 6.17990732e-01 3.45239222e-01 -8.81763577e-01
-3.38382214e-01 -8.58693421e-02 -7.97670126e-01 -4.56289560e-01
6.59800410e-01 1.88185126e-01 5.94049513e-01 -7.84411132e-01
1.16108644e+00 1.37122703e+00 7.68446684e-01 8.81262362e-01
-1.06491387e+00 -4.98705804e-01 1.61072999e-01 -3.68794543e-03
-1.51843870e+00 -1.81387380e-01 1.05199146e+00 -3.82569164e-01
3.79407674e-01 3.84697646e-01 1.38199317e+00 1.01109171e+00
2.79659331e-01 4.90557820e-01 6.58195853e-01 -3.89222711e-01
1.16586894e-01 -2.26821616e-01 -2.74374634e-01 1.06619763e+00
7.64895156e-02 2.33838841e-01 -4.14604545e-01 -1.74537659e-01
1.75268209e+00 4.75257188e-02 -6.37110174e-01 -1.00413084e+00
-1.31359565e+00 5.88429391e-01 5.09043694e-01 -2.01476127e-01
-5.40020347e-01 5.52293479e-01 2.61222031e-02 -2.19222248e-01
2.45310277e-01 4.50391531e-01 -2.87721187e-01 1.05137482e-01
-4.13705468e-01 6.84423625e-01 3.81702453e-01 9.44098353e-01
7.47187436e-01 5.34970641e-01 5.95497228e-02 4.04973269e-01
1.99091151e-01 7.22510040e-01 5.02776094e-02 -1.53990948e+00
5.10177501e-02 4.66251403e-01 4.50235367e-01 -1.06326890e+00
-5.08546531e-01 -5.10732174e-01 -1.03389108e+00 2.79457957e-01
3.68541211e-01 -3.33675832e-01 -9.70154524e-01 1.96065521e+00
9.00963664e-01 2.17799187e-01 -2.00233936e-01 1.30050886e+00
7.02351689e-01 6.05447471e-01 -1.47889676e-02 -1.13142125e-01
1.32311368e+00 -7.92837739e-01 -7.40992069e-01 -7.94519782e-02
1.63282499e-01 -4.53559607e-01 9.67541933e-01 3.58840972e-01
-1.26484859e+00 -6.77135468e-01 -8.42133582e-01 -2.12443203e-01
3.65010530e-01 1.68142125e-01 5.57255685e-01 4.00359154e-01
-7.54641712e-01 4.87077415e-01 -8.79393637e-01 -1.99921653e-01
-4.51047793e-02 2.89148390e-01 -6.82847738e-01 2.10576922e-01
-1.12967300e+00 6.59774601e-01 -2.26015616e-02 2.23323658e-01
-8.87320578e-01 -8.48162889e-01 -1.16401470e+00 -2.65950441e-01
3.41628700e-01 -1.35557711e+00 9.43391085e-01 -1.32759178e+00
-1.92938280e+00 7.65486658e-01 1.27904445e-01 -1.36263639e-01
9.61040378e-01 -5.91280460e-01 2.94302344e-01 2.57746369e-01
-3.49575728e-01 8.62483799e-01 9.96212542e-01 -1.56942511e+00
-9.00574997e-02 -4.92175788e-01 2.55337656e-01 7.36606121e-01
1.49136066e-01 -4.02171701e-01 -5.08246124e-01 -8.54820192e-01
3.34692985e-01 -1.07048190e+00 -5.14576793e-01 6.07063651e-01
-2.27697909e-01 2.53290027e-01 7.83940196e-01 -9.63141143e-01
6.67223930e-01 -1.84173930e+00 7.65102208e-01 2.44641051e-01
3.26242507e-01 -2.18634367e-01 2.90594791e-04 -1.46798149e-01
2.06461512e-02 -5.06509244e-01 -3.47737633e-02 -4.10895169e-01
-3.26449126e-01 2.30587989e-01 -1.52764749e-02 8.16374004e-01
-2.49268696e-01 1.04436469e+00 -8.01275909e-01 -3.51563513e-01
2.20023453e-01 1.03271210e+00 -8.47823560e-01 5.47402918e-01
-1.67366341e-01 1.22112978e+00 -4.79155749e-01 2.75456131e-01
7.21608043e-01 -1.57588392e-01 1.76944718e-01 -4.44904119e-01
-1.58329103e-02 -3.08791667e-01 -1.47148192e+00 1.86989319e+00
-3.24634075e-01 1.47276983e-01 4.43185568e-01 -4.15197045e-01
8.25869739e-01 2.87971020e-01 7.94369996e-01 -4.44172323e-01
3.47680807e-01 -3.27870160e-01 -4.48080480e-01 -2.78103948e-01
4.71008509e-01 -2.79585183e-01 3.43138613e-02 1.71956807e-01
-1.29994214e-01 -1.98066205e-01 -4.99335736e-01 -6.08913302e-02
4.63034511e-01 8.45655859e-01 2.55171388e-01 -1.42688647e-01
6.57896340e-01 -2.43145764e-01 7.44617224e-01 1.27286628e-01
2.24312425e-01 8.09856236e-01 -6.00460470e-02 -9.72363949e-01
-1.35487473e+00 -1.30943966e+00 1.23515196e-01 5.89540958e-01
3.70839477e-01 -1.37086958e-01 -9.83437359e-01 -3.08275700e-01
-1.27015144e-01 2.23901302e-01 -7.36936748e-01 -2.21578367e-02
-1.22494137e+00 -2.96390712e-01 1.20838113e-01 5.39745212e-01
6.57215774e-01 -9.28534865e-01 -1.18264866e+00 -1.28983676e-01
-4.76028681e-01 -1.16718054e+00 -6.14912331e-01 -3.56066227e-01
-8.31150651e-01 -1.04133582e+00 -1.12468839e+00 -6.75848663e-01
9.33029890e-01 -1.23943478e-01 7.83270717e-01 -1.75975114e-02
-3.76776397e-01 5.48170745e-01 1.58308193e-01 1.94392204e-01
-2.37980470e-01 -3.43769878e-01 1.88891113e-01 1.37939513e-01
-4.64538723e-01 -5.38815081e-01 -7.85272896e-01 3.29836726e-01
-5.46259820e-01 6.76080942e-01 -1.96100734e-02 7.70025551e-01
6.83187008e-01 -4.13918823e-01 5.60053177e-02 -6.52701199e-01
-5.61432634e-03 1.72504142e-01 -7.52289832e-01 1.21452600e-01
1.17889643e-01 -9.59119275e-02 6.56019151e-01 -6.66642010e-01
-1.27004945e+00 6.11497462e-01 -1.68560922e-01 -8.11628342e-01
-6.16444759e-02 -2.30388626e-01 -2.76518732e-01 -3.80343169e-01
8.34658861e-01 1.66966036e-01 -6.08496144e-02 -4.39175904e-01
5.32194555e-01 -8.27339962e-02 8.63937378e-01 -8.53114247e-01
1.20791614e+00 6.14823937e-01 3.05925876e-01 -8.21541369e-01
-5.19758642e-01 1.12984873e-01 -1.13866758e+00 -6.25341952e-01
1.16324389e+00 -1.16524088e+00 -1.04615307e+00 6.49386346e-01
-1.30462587e+00 -4.52040493e-01 -3.89698595e-01 5.49044311e-01
-9.73274171e-01 5.50796807e-01 -7.11560130e-01 -8.12925816e-01
-4.53804404e-01 -9.39812779e-01 1.25558829e+00 1.10296451e-01
-2.91121155e-01 -8.92800927e-01 1.90403447e-01 2.45395362e-01
-5.81580363e-02 1.04516232e+00 7.35109329e-01 5.61166525e-01
-8.72721672e-01 9.45010129e-03 3.06451678e-01 2.64088251e-02
-8.22010338e-02 -2.18644470e-01 -7.87325084e-01 -4.76219922e-01
7.13955029e-04 -6.39475808e-02 1.36971578e-01 4.58877236e-01
8.50660980e-01 -3.95798355e-01 -1.59815833e-01 1.09466398e+00
1.26808155e+00 9.98143330e-02 6.94063544e-01 1.40852138e-01
1.26934588e+00 7.09128916e-01 4.78988439e-01 7.03391671e-01
3.73091787e-01 1.11073124e+00 3.49966437e-01 -1.48022711e-01
-2.48502448e-01 -4.86880869e-01 2.37804100e-01 6.43713176e-01
-5.80236793e-01 2.16296583e-01 -5.69313169e-01 1.63607553e-01
-1.79672146e+00 -5.89915335e-01 1.08110599e-01 2.34119678e+00
6.70671880e-01 -3.61385643e-01 2.55627811e-01 -1.39177054e-01
5.77402413e-01 -1.24190990e-02 -6.19098544e-01 -2.29476579e-02
5.91059029e-02 9.51610580e-02 2.21994773e-01 7.31686890e-01
-7.02382922e-01 1.02874553e+00 6.18533707e+00 1.06446594e-01
-1.14329684e+00 -5.14635108e-02 3.55284780e-01 1.11039262e-02
-4.00993109e-01 -9.16413292e-02 -4.31598008e-01 -1.91912323e-01
3.12509388e-01 -4.80320416e-02 4.49914426e-01 8.78582239e-01
1.50842100e-01 1.28371090e-01 -8.73796880e-01 1.14026868e+00
1.25782266e-01 -1.18973708e+00 3.08976322e-01 1.08496472e-02
9.08157408e-01 -7.37559557e-01 -1.00158475e-01 -2.38050506e-01
9.16334838e-02 -7.72395611e-01 1.07608199e+00 8.64668608e-01
9.11378145e-01 -9.86412346e-01 1.37284040e-01 4.35260594e-01
-1.25605476e+00 1.08107261e-01 -3.46553952e-01 -5.28629348e-02
4.23370510e-01 6.87219724e-02 -3.44029546e-01 4.92945641e-01
5.62463284e-01 6.67932630e-01 -3.09585273e-01 7.37302601e-01
-1.85854122e-01 -1.62895828e-01 -2.18078449e-01 3.37155312e-01
-2.26312324e-01 -4.84028399e-01 7.38180757e-01 5.62399685e-01
2.22152486e-01 4.57129270e-01 3.01903278e-01 7.91931868e-01
1.07004819e-02 1.75179362e-01 -4.09904271e-01 5.45901656e-01
-8.54861885e-02 1.20514214e+00 -3.59436750e-01 -1.07691921e-01
1.12056024e-01 1.22798693e+00 3.53889734e-01 5.58535397e-01
-9.46320713e-01 2.06289385e-02 6.77451432e-01 5.16826034e-01
8.19810927e-02 -4.66093183e-01 -8.77808779e-03 -1.34219515e+00
3.01454253e-02 -7.16933668e-01 1.56298354e-02 -1.01584864e+00
-5.76787412e-01 7.97366023e-01 1.16127573e-01 -1.10589826e+00
-4.54207808e-01 -3.01774323e-01 -1.81958467e-01 7.55551696e-01
-9.55157042e-01 -1.38320684e+00 -7.30805695e-01 8.40427399e-01
3.66321623e-01 2.02144265e-01 7.54823029e-01 8.56586844e-02
-1.93553820e-01 5.58968186e-01 -5.53218842e-01 9.71355960e-02
5.49870729e-01 -1.00494409e+00 5.83469987e-01 6.24089837e-01
-9.76713598e-02 4.86785263e-01 7.66732275e-01 -7.06132412e-01
-1.45731688e+00 -1.08819866e+00 3.45025986e-01 -5.54545999e-01
-4.51251790e-02 -4.80096936e-01 -7.58659780e-01 9.52846885e-01
-8.76132473e-02 5.27798235e-02 1.52596280e-01 -3.83391708e-01
-2.27893189e-01 -1.01933479e-02 -1.32985246e+00 8.92243207e-01
1.18812346e+00 -2.51583964e-01 -2.00310707e-01 1.83778495e-01
6.15224242e-01 -1.01301003e+00 -9.43784714e-01 3.85392308e-01
8.23938191e-01 -6.99155986e-01 1.17896736e+00 -5.70570767e-01
1.28845990e-01 -4.96306360e-01 -2.79419590e-02 -1.09907961e+00
-5.60850620e-01 -9.67103004e-01 -2.36295611e-01 5.27243733e-01
-3.37094367e-01 -1.65671006e-01 1.01011109e+00 6.40542150e-01
2.27763906e-01 -8.00672710e-01 -8.48157883e-01 -3.82577956e-01
-1.58164985e-02 5.37901074e-02 4.88656640e-01 7.90576696e-01
-4.78660226e-01 1.48433179e-01 -9.40190792e-01 3.50892425e-01
1.03201389e+00 2.52693892e-02 1.20014584e+00 -1.26195598e+00
-4.94699329e-01 2.81290740e-01 -5.68006396e-01 -1.40472221e+00
3.10620844e-01 -2.70371646e-01 4.38654795e-02 -1.25680423e+00
1.71086952e-01 -1.78586096e-01 5.37308395e-01 2.25190207e-01
-7.79721290e-02 4.83223081e-01 2.43729919e-01 6.21386766e-02
-3.02172631e-01 8.27592969e-01 1.89928532e+00 3.84455502e-01
-2.15571344e-01 -2.90159397e-02 -1.74172550e-01 1.03009331e+00
2.97088027e-01 -2.29482818e-02 -4.76436883e-01 -5.07038891e-01
2.12823153e-01 3.69862169e-01 7.34777093e-01 -1.15839374e+00
-6.33695349e-02 -3.14459950e-01 8.34476173e-01 -3.39896679e-01
8.05217326e-01 -9.02527034e-01 8.70715320e-01 6.70023322e-01
-2.52146930e-01 3.66371036e-01 1.17683187e-02 6.12147331e-01
3.82658243e-01 2.52212703e-01 1.23250484e+00 -3.73778909e-01
-5.31835377e-01 5.16653836e-01 8.26343298e-02 1.14542842e-02
9.11026418e-01 -2.90695757e-01 1.44712582e-01 -6.89769030e-01
-1.08068252e+00 -9.73720476e-02 9.73939419e-01 3.81431550e-01
7.96507120e-01 -1.55201888e+00 -5.79707980e-01 4.64820355e-01
-1.06723815e-01 4.18101281e-01 4.53725219e-01 4.29881960e-01
-1.11350417e+00 1.23952795e-02 -4.27167594e-01 -7.05107510e-01
-1.37045479e+00 3.51295382e-01 8.42927039e-01 6.63702264e-02
-1.20600152e+00 5.88967979e-01 7.45930433e-01 -6.09690309e-01
3.13149750e-01 6.70666853e-03 -1.07219741e-02 -5.62751472e-01
4.69740987e-01 4.87030357e-01 -3.56358111e-01 -1.20065367e+00
-1.68016538e-01 1.17099202e+00 2.73302555e-01 -2.15012670e-01
1.33372653e+00 -3.47949922e-01 1.15681715e-01 1.88428506e-01
1.00271797e+00 1.24807142e-01 -1.92467237e+00 -1.97308123e-01
-7.71400273e-01 -7.08541155e-01 -2.21381336e-01 -5.70770442e-01
-1.31898797e+00 6.19285762e-01 5.42174578e-01 -8.56549025e-01
8.93711448e-01 -2.93740720e-01 9.06411052e-01 3.21041912e-01
6.42271459e-01 -7.91091263e-01 3.50190073e-01 3.78875583e-01
1.16763723e+00 -5.53021312e-01 1.39116690e-01 -6.60520792e-01
-6.03522360e-01 1.15505576e+00 9.02887166e-01 -4.41931307e-01
5.71403682e-01 7.68006667e-02 2.32887506e-01 -1.36163056e-01
-3.84516865e-01 3.68308991e-01 5.23511529e-01 7.48585820e-01
1.49023443e-01 -6.44372404e-02 1.27801541e-02 2.58854419e-01
-3.41097116e-01 -1.85163245e-01 4.84103858e-01 6.23258412e-01
-2.54735112e-01 -8.09816122e-01 -5.58162034e-01 -4.65500832e-01
-8.66785571e-02 4.59727019e-01 -2.64237255e-01 8.32085431e-01
1.85603052e-01 1.93996847e-01 8.37552994e-02 -8.77823383e-02
6.90113425e-01 -1.41668066e-01 1.04898453e+00 -3.98600191e-01
-6.05850518e-01 3.37963283e-01 -1.90800548e-01 -8.10148537e-01
-2.62914479e-01 -5.14352262e-01 -1.45480919e+00 -3.65952104e-01
-4.45518009e-02 -3.07883281e-04 3.55545074e-01 5.58349192e-01
1.79997206e-01 4.28600580e-01 5.27169585e-01 -1.27822375e+00
-3.04635584e-01 -5.09348214e-01 -5.47949731e-01 7.64967203e-01
3.82724822e-01 -8.21548402e-01 -4.38267067e-02 2.43854627e-01]
|
[7.21701192855835, -1.2745369672775269]
|
7345cacd-0723-4451-bb61-c32131281431
|
compartmental-models-for-covid-19-and-control
|
2203.02860
| null |
https://arxiv.org/abs/2203.02860v1
|
https://arxiv.org/pdf/2203.02860v1.pdf
|
Compartmental Models for COVID-19 and Control via Policy Interventions
|
We demonstrate an approach to replicate and forecast the spread of the SARS-CoV-2 (COVID-19) pandemic using the toolkit of probabilistic programming languages (PPLs). Our goal is to study the impact of various modeling assumptions and motivate policy interventions enacted to limit the spread of infectious diseases. Using existing compartmental models we show how to use inference in PPLs to obtain posterior estimates for disease parameters. We improve popular existing models to reflect practical considerations such as the under-reporting of the true number of COVID-19 cases and motivate the need to model policy interventions for real-world data. We design an SEI3RD model as a reusable template and demonstrate its flexibility in comparison to other models. We also provide a greedy algorithm that selects the optimal series of policy interventions that are likely to control the infected population subject to provided constraints. We work within a simple, modular, and reproducible framework to enable immediate cross-domain access to the state-of-the-art in probabilistic inference with emphasis on policy interventions. We are not epidemiologists; the sole aim of this study is to serve as an exposition of methods, not to directly infer the real-world impact of policy-making for COVID-19.
|
['Noah Kasmanoff', 'Swapneel Mehta']
|
2022-03-06
| null | null | null | null |
['probabilistic-programming']
|
['methodology']
|
[ 8.07958543e-02 -6.58731982e-02 -1.48390740e-01 -2.94102747e-02
-5.52726388e-01 -5.41265130e-01 7.31756687e-01 3.27360064e-01
-5.21107376e-01 9.08995450e-01 3.03481996e-01 -1.02623248e+00
-3.69307667e-01 -7.49130726e-01 -5.79917371e-01 -5.87326050e-01
-3.83031547e-01 1.17475927e+00 1.25244126e-01 3.51875299e-03
-1.82363912e-01 6.12722635e-01 -1.11001492e+00 2.95994002e-02
7.33041406e-01 -4.59806882e-02 1.32542789e-01 8.70410621e-01
1.80645734e-01 5.19269049e-01 -7.13509083e-01 -7.58923218e-02
-2.73758769e-02 -1.57153562e-01 -4.22371060e-01 -5.92383623e-01
-6.37161314e-01 -6.21840239e-01 -1.51671181e-02 3.83024156e-01
5.84851921e-01 -3.00808847e-01 1.13165629e+00 -1.46129930e+00
-1.03647143e-01 4.91650671e-01 -4.24365669e-01 4.89695013e-01
3.68441790e-01 3.27755213e-01 2.60870636e-01 -2.52266139e-01
8.49477053e-01 1.42717922e+00 1.01475465e+00 3.88148934e-01
-1.54043818e+00 -4.32486057e-01 9.21636075e-02 -3.24405253e-01
-1.55061102e+00 -3.62891316e-01 -7.14126276e-03 -8.70065212e-01
1.19173503e+00 2.79265434e-01 6.01526737e-01 1.20291984e+00
6.33900404e-01 3.21611077e-01 1.13255644e+00 -2.77916193e-01
3.61892819e-01 1.36162236e-01 4.60286029e-02 2.68699467e-01
6.62643671e-01 3.13745230e-01 4.50490117e-02 -1.16940331e+00
7.04747736e-01 3.71171355e-01 -1.06405891e-01 2.99510383e-03
-9.85227287e-01 9.56218600e-01 -2.77823925e-01 -1.46229309e-03
-7.77403951e-01 2.55418986e-01 1.48607552e-01 -4.06388104e-01
6.82518363e-01 -4.70003411e-02 -7.27818012e-01 -7.40133971e-02
-8.99925172e-01 7.38427281e-01 8.84712279e-01 5.47411919e-01
1.17655993e-01 -1.43859610e-01 -4.10188168e-01 3.38323385e-01
7.42971301e-01 1.22127438e+00 -4.64297533e-01 -8.57351303e-01
1.78070888e-01 -7.94510320e-02 7.04040766e-01 -7.47123480e-01
-4.98542815e-01 -5.85710108e-02 -5.25340438e-01 -1.86558560e-01
3.25555354e-01 -7.42152691e-01 -8.37198734e-01 1.87669992e+00
4.26990628e-01 2.65112668e-01 -1.50283426e-01 2.54268438e-01
1.39842838e-01 1.04802787e+00 6.57186210e-01 -5.50726295e-01
1.40430903e+00 1.10377170e-01 -6.68038905e-01 7.52996951e-02
6.51472330e-01 -5.65804124e-01 3.03408891e-01 -5.73776998e-02
-9.93725121e-01 2.30958134e-01 -2.77371883e-01 7.74415135e-01
-5.20438135e-01 -3.24950010e-01 5.01109660e-01 7.45928586e-01
-1.14451468e+00 8.53523836e-02 -1.28381002e+00 -7.55043685e-01
1.71255767e-01 1.63821161e-01 1.44356474e-01 1.39161140e-01
-1.23124874e+00 1.26593816e+00 3.30365151e-01 1.55315548e-02
-9.56369281e-01 -1.13189375e+00 -5.65550148e-01 -1.13980070e-01
1.74051329e-01 -9.98684466e-01 1.04489553e+00 -5.28384596e-02
-8.74819338e-01 6.64653957e-01 -4.40636903e-01 -5.36082387e-01
5.43037117e-01 2.93017149e-01 -3.98406625e-01 1.92392226e-02
1.74437955e-01 4.26626772e-01 1.44857466e-01 -1.27119946e+00
-6.25093639e-01 -2.09375799e-01 -2.38390476e-01 3.73344682e-02
3.24152321e-01 8.30734849e-01 -2.23012328e-01 -6.06454313e-01
-7.16657400e-01 -9.06201065e-01 -6.03137672e-01 -4.47328657e-01
-4.06244129e-01 -1.44785702e-01 4.82894599e-01 -7.81875610e-01
1.35579443e+00 -1.66416764e+00 -1.61188841e-01 1.98476076e-01
-1.50554655e-02 2.23646849e-01 1.52147457e-01 1.04350317e+00
3.35547239e-01 3.43374014e-01 -6.41918361e-01 -2.27766931e-01
5.71656115e-02 3.61513048e-01 -5.47535121e-01 7.21164167e-01
2.03440726e-01 5.21619916e-01 -8.69184256e-01 -6.10378504e-01
2.55601853e-01 7.44652748e-01 -7.34535396e-01 3.46949250e-01
-4.75551873e-01 4.74347174e-01 -5.70465744e-01 3.08574438e-01
8.82319331e-01 -2.87139893e-01 5.66126108e-01 3.89824152e-01
-3.32113534e-01 2.11625129e-01 -8.54950666e-01 6.36899412e-01
-1.41927242e-01 -6.65737391e-02 3.17131847e-01 -6.40527129e-01
2.81343818e-01 4.98422891e-01 7.44008064e-01 2.06512183e-01
-7.46666119e-02 -1.00324422e-01 -6.86285943e-02 -4.50218052e-01
2.18990352e-02 -1.49622843e-01 -9.58291367e-02 9.97632444e-01
-4.58284944e-01 -9.29239467e-02 1.48245886e-01 1.61516026e-01
8.81829381e-01 9.56056863e-02 4.23702687e-01 -6.40051663e-01
-2.33907704e-05 3.88182223e-01 5.97591519e-01 1.07531333e+00
-2.33898491e-01 3.29670906e-02 6.12008095e-01 -3.55543762e-01
-9.39723611e-01 -1.43971145e+00 -5.38052082e-01 1.00472283e+00
-5.90388417e-01 -1.10735655e-01 -7.07287610e-01 -3.20774287e-01
7.56708235e-02 9.74298239e-01 -7.52374053e-01 4.40125108e-01
-4.74435598e-01 -1.63581669e+00 6.83199584e-01 1.35016367e-01
-1.80269867e-01 -9.15601850e-01 -9.64623153e-01 4.35147732e-01
-1.01762019e-01 -8.23062360e-01 -3.31281692e-01 -6.52798787e-02
-4.95157093e-01 -1.26385891e+00 -8.02186668e-01 -2.79822141e-01
7.29001582e-01 -1.73039123e-01 9.17808950e-01 -1.30102351e-01
-2.02245116e-01 7.16059864e-01 1.63225427e-01 -7.89499819e-01
-7.67345548e-01 -3.73626053e-01 4.20474529e-01 -6.11742556e-01
5.73483527e-01 -1.11398116e-01 -7.52759039e-01 1.11749254e-01
-1.00405407e+00 -1.70953199e-02 -1.96345560e-02 4.40407842e-01
4.90399003e-01 -5.05604520e-02 5.74038446e-01 -1.00425017e+00
9.22946870e-01 -1.05116558e+00 -8.75795603e-01 5.42182267e-01
-5.27481079e-01 -1.83685154e-01 2.83314735e-01 -2.83806771e-01
-1.20954788e+00 -3.13103288e-01 -1.01405978e-01 1.02603778e-01
-2.03696936e-01 8.14561069e-01 3.61052960e-01 7.03065455e-01
3.46660465e-01 1.62103921e-01 6.07951283e-02 -3.97006452e-01
2.13766932e-01 7.70555377e-01 2.05388576e-01 -6.60495996e-01
2.87837416e-01 6.72819555e-01 -1.27707675e-01 -7.52296031e-01
-1.95335597e-01 -3.15025032e-01 -5.77392802e-02 -3.30917090e-02
8.54050398e-01 -9.22132730e-01 -9.60499406e-01 4.55851763e-01
-1.31302059e+00 -7.13690400e-01 1.87886745e-01 6.06445909e-01
-5.83954096e-01 7.91741908e-02 -6.53153181e-01 -1.26446235e+00
-3.02289784e-01 -1.29673815e+00 9.12709594e-01 -2.02586055e-01
-3.27689201e-01 -1.51028228e+00 8.44498158e-01 3.94485779e-02
7.38369286e-01 3.74388695e-01 1.01477504e+00 -6.87320292e-01
-1.96908504e-01 1.93178281e-01 -1.75688639e-01 -3.35678101e-01
1.93604648e-01 4.72511888e-01 -6.25834405e-01 -3.75632316e-01
-1.21180810e-01 3.16508472e-01 4.43907470e-01 9.90109265e-01
7.24815071e-01 -7.34674990e-01 -9.23235297e-01 3.10771227e-01
1.37895942e+00 3.94662052e-01 2.95594245e-01 8.33905339e-02
1.77636921e-01 8.30751717e-01 2.56897926e-01 7.48670995e-01
7.93688238e-01 7.59011447e-01 -1.26973495e-01 -8.23260993e-02
4.17953193e-01 -3.31021428e-01 2.50735492e-01 3.25195134e-01
-9.13335755e-02 -3.25489372e-01 -1.53342676e+00 6.77356422e-01
-1.71299803e+00 -1.05581319e+00 -1.28899664e-01 2.31723285e+00
9.81638908e-01 -7.59508088e-02 5.74502647e-01 -6.69399083e-01
6.38739705e-01 -1.36934131e-01 -1.73426330e-01 -5.58257997e-01
2.18870044e-01 -1.53403739e-02 7.99942732e-01 7.74254382e-01
-9.44417417e-01 6.55418694e-01 8.47925568e+00 5.66765845e-01
-7.57527351e-01 2.38761112e-01 7.41352022e-01 8.23929086e-02
-5.46945691e-01 -3.22681665e-02 -8.75228643e-01 6.84542060e-01
1.48904479e+00 -2.77502358e-01 4.24146920e-01 1.70105889e-01
1.05767906e+00 -5.97232059e-02 -7.36397326e-01 1.79439694e-01
-3.05864871e-01 -1.42778480e+00 -2.07923964e-01 2.24389404e-01
7.01463759e-01 3.89560521e-01 -4.84099761e-02 5.87379336e-02
1.10889399e+00 -8.93397808e-01 3.74704063e-01 7.85106957e-01
7.01882303e-01 -7.90662885e-01 8.26998174e-01 3.66098821e-01
-7.18332469e-01 3.26385707e-01 -4.76202788e-03 1.28472105e-01
7.34228492e-01 5.71479619e-01 -1.28088033e+00 1.32227689e-01
6.22008264e-01 1.93586022e-01 1.33990765e-01 1.00397050e+00
1.44847352e-02 1.00358760e+00 -6.79365516e-01 8.10410529e-02
1.36273980e-01 4.13351506e-02 6.22052312e-01 1.65399635e+00
2.38988891e-01 2.76072592e-01 1.94794804e-01 8.61733258e-01
5.34266233e-01 -1.74295917e-01 -8.70696425e-01 -2.29534522e-01
6.57932758e-01 4.45457995e-01 -6.45702600e-01 -3.31445307e-01
-1.92738380e-02 2.17236623e-01 -1.48963749e-01 7.25951493e-01
-7.95186877e-01 4.45272736e-02 7.37551987e-01 2.65749633e-01
3.22234690e-01 -1.83776066e-01 -5.37175685e-02 -7.77401507e-01
-6.02326989e-01 -8.81718874e-01 5.67295194e-01 -5.45617104e-01
-1.23302364e+00 2.71929801e-01 8.93626869e-01 -4.15978700e-01
-8.04526806e-01 -3.91762197e-01 -6.72826052e-01 1.26564491e+00
-1.21352708e+00 -1.09103727e+00 6.65400684e-01 2.74174750e-01
-3.36491987e-02 3.30707133e-01 9.85399306e-01 -1.28923357e-01
-6.01284921e-01 -4.63171396e-03 5.39885759e-01 -3.63741577e-01
2.93915093e-01 -9.50263500e-01 3.35760504e-01 7.53531754e-01
-7.87994504e-01 1.08157694e+00 1.24905837e+00 -1.27913761e+00
-1.15606225e+00 -9.82626200e-01 1.04929030e+00 -7.87996769e-01
8.80892277e-01 -4.03372049e-01 -4.74212736e-01 9.82563615e-01
1.26452848e-01 -7.21024334e-01 7.28533983e-01 4.47639264e-03
-1.64779842e-01 2.33360648e-01 -1.55914676e+00 7.68723309e-01
5.03521979e-01 -2.97912121e-01 -6.39743149e-01 6.46878183e-01
7.43627429e-01 1.06025189e-01 -8.30991805e-01 3.06977391e-01
4.98560190e-01 -5.83006620e-01 1.12797070e+00 -8.54173958e-01
-1.16467156e-01 -4.09593612e-01 -8.32665563e-02 -1.13749802e+00
-1.43082663e-01 -7.45371699e-01 1.57623440e-01 8.71053696e-01
4.45649803e-01 -1.21855211e+00 2.83017904e-01 6.39332116e-01
4.36858594e-01 -5.38756609e-01 -9.50953007e-01 -5.08989394e-01
3.07882696e-01 -5.06956041e-01 6.96936548e-01 8.14189613e-01
-1.09507054e-01 -4.31857347e-01 -4.81501490e-01 6.04390621e-01
8.73596311e-01 -1.66771427e-01 5.83686411e-01 -9.48778510e-01
-3.54627579e-01 -1.88180029e-01 2.75543369e-02 -3.77326041e-01
-1.49503484e-01 -2.78037339e-01 6.94503486e-02 -1.65428519e+00
7.21203268e-01 -7.24439919e-01 -2.32877344e-01 5.72498500e-01
-1.15193591e-01 -2.02525020e-01 -3.77125363e-03 2.51570404e-01
-2.40909144e-01 1.27073631e-01 5.85369408e-01 6.96418211e-02
-3.70886028e-01 2.24639148e-01 -6.26206458e-01 5.94778419e-01
8.67981672e-01 -9.12806094e-01 -5.45722663e-01 -2.67122328e-01
1.81480259e-01 3.20098758e-01 5.53059876e-01 -2.78720826e-01
1.67320028e-01 -8.68864119e-01 2.34883111e-02 -1.07604039e+00
2.95945965e-02 -6.95934832e-01 8.84283125e-01 1.03873122e+00
-6.07347712e-02 3.28194350e-01 6.71940267e-01 5.82029521e-01
6.07108533e-01 3.03167850e-02 3.25348258e-01 5.94872385e-02
2.56105095e-01 2.15070128e-01 -1.15322721e+00 1.80071592e-01
1.12421167e+00 3.82334650e-01 -6.82252944e-01 -3.32150221e-01
-4.95395511e-01 3.60335618e-01 6.10250831e-01 5.34659345e-03
1.63129225e-01 -5.00683010e-01 -1.19051969e+00 -1.95368275e-01
-1.96735039e-01 -3.98795605e-01 3.87653232e-01 7.77628064e-01
-9.07566428e-01 1.00959134e+00 3.94790582e-02 -5.86752951e-01
-9.25185502e-01 8.39227438e-01 4.57052499e-01 -5.86443186e-01
-3.42928827e-01 2.15708330e-01 3.01058471e-01 -6.19368017e-01
3.23940180e-02 -5.18538579e-02 7.97061026e-02 -2.53675997e-01
7.21363664e-01 4.57064152e-01 -4.59571391e-01 -4.96415079e-01
-8.09703887e-01 4.69339229e-02 -3.72648016e-02 -4.02776062e-01
1.54563153e+00 -1.42012179e-01 -3.29433084e-01 2.63629258e-01
6.30345345e-01 2.44536940e-02 -1.22234893e+00 2.87239581e-01
-2.33187467e-01 4.65271622e-02 -1.25019878e-01 -1.09994316e+00
-3.09252918e-01 3.55411977e-01 5.87747455e-01 2.81671792e-01
8.34216177e-01 3.01522128e-02 1.92786962e-01 -2.15654522e-02
4.35502857e-01 -5.75279474e-01 -1.09619701e+00 7.27398545e-02
6.57512009e-01 -7.34263718e-01 2.25908443e-01 -1.99121803e-01
-3.67265075e-01 3.19450080e-01 -2.16544494e-01 2.96938092e-01
1.06340945e+00 5.34372449e-01 7.95276463e-03 -3.05628896e-01
-9.81260777e-01 2.63253953e-02 -5.34702577e-02 1.09892237e+00
2.20013171e-01 6.41919792e-01 -6.24723852e-01 2.13210598e-01
3.04718524e-01 3.60365093e-01 3.95434529e-01 9.63593841e-01
-1.37476474e-01 -1.11308300e+00 -7.78084219e-01 6.08288705e-01
-7.76049912e-01 -3.86572093e-01 -4.54760753e-02 8.56974959e-01
1.15264192e-01 1.14540637e+00 1.90059930e-01 3.67160529e-01
-8.73745233e-02 -8.03689957e-02 9.50893387e-02 -3.89001966e-01
-4.72308069e-01 1.03499465e-01 2.14285702e-01 -4.68195640e-02
-4.42176133e-01 -9.36359048e-01 -9.17414308e-01 -7.28805661e-01
2.17067599e-01 1.98671356e-01 6.49813950e-01 1.00219953e+00
6.02801263e-01 2.20161498e-01 3.19872737e-01 -5.76736033e-01
-4.98123854e-01 -6.44607425e-01 -4.20978338e-01 -1.18283525e-01
2.38540292e-01 -6.16924703e-01 -3.96037340e-01 -2.81516835e-02]
|
[6.034331321716309, 4.379302501678467]
|
f6ff5763-0b99-4bd8-ae78-903703a818ae
|
two-stage-convolutional-neural-network-for
|
1803.04054
| null |
http://arxiv.org/abs/1803.04054v2
|
http://arxiv.org/pdf/1803.04054v2.pdf
|
Two-Stage Convolutional Neural Network for Breast Cancer Histology Image Classification
|
This paper explores the problem of breast tissue classification of microscopy
images. Based on the predominant cancer type the goal is to classify images
into four categories of normal, benign, in situ carcinoma, and invasive
carcinoma. Given a suitable training dataset, we utilize deep learning
techniques to address the classification problem. Due to the large size of each
image in the training dataset, we propose a patch-based technique which
consists of two consecutive convolutional neural networks. The first
"patch-wise" network acts as an auto-encoder that extracts the most salient
features of image patches while the second "image-wise" network performs
classification of the whole image. The first network is pre-trained and aimed
at extracting local information while the second network obtains global
information of an input image. We trained the networks using the ICIAR 2018
grand challenge on BreAst Cancer Histology (BACH) dataset. The proposed method
yields 95 % accuracy on the validation set compared to previously reported 77 %
accuracy rates in the literature. Our code is publicly available at
https://github.com/ImagingLab/ICIAR2018
|
['Mehran Ebrahimi', 'Kamyar Nazeri', 'Azad Aminpour']
|
2018-03-11
| null | null | null | null |
['breast-cancer-histology-image-classification']
|
['medical']
|
[ 4.66629356e-01 2.50875175e-01 -6.60335049e-02 -2.94020861e-01
-1.01505446e+00 -1.54372200e-01 4.63001609e-01 4.33263749e-01
-4.89184946e-01 5.21625996e-01 -1.80431649e-01 -4.57492054e-01
1.26686826e-01 -7.52424598e-01 -8.29559207e-01 -1.29642487e+00
1.92110557e-02 2.16485217e-01 1.01228349e-01 2.13795722e-01
1.25106201e-01 6.71427250e-01 -1.09774184e+00 6.79843664e-01
4.57347929e-01 1.45631814e+00 3.66481841e-01 8.94482732e-01
8.86447653e-02 9.88580287e-01 -3.22065532e-01 -5.31194583e-02
1.64258733e-01 -3.13156724e-01 -1.00128686e+00 1.78133566e-02
3.77971768e-01 -2.29209840e-01 -3.99004161e-01 1.06288862e+00
4.79423165e-01 -4.45553601e-01 7.33260214e-01 -7.37444997e-01
-2.89322138e-01 4.15133268e-01 -5.67829549e-01 4.62357432e-01
-2.81048745e-01 -1.19942963e-01 7.07718730e-01 -7.39301622e-01
8.61497104e-01 5.31599700e-01 6.80658460e-01 5.88733613e-01
-1.24777782e+00 -5.15219092e-01 -4.24621552e-01 3.32047582e-01
-1.40832734e+00 -5.62522769e-01 7.24834561e-01 -6.38924301e-01
5.45103908e-01 2.82397866e-01 6.26569748e-01 9.68833148e-01
6.56212330e-01 7.56104052e-01 1.16255462e+00 -4.42981422e-01
3.11424285e-01 3.25199902e-01 1.28316388e-01 7.12603450e-01
1.48001999e-01 -6.63295388e-02 -1.23339735e-01 -1.54536247e-01
6.46977246e-01 1.62156656e-01 -3.51413339e-01 -1.92317009e-01
-1.06662512e+00 6.30407751e-01 8.41799378e-01 5.84383607e-01
-5.22087932e-01 -2.54682340e-02 4.65504825e-01 1.40107274e-01
4.64162499e-01 1.85509071e-01 -2.61161596e-01 4.91358966e-01
-1.04619777e+00 -4.64750491e-02 6.74865425e-01 4.87022161e-01
7.25268662e-01 -4.96142447e-01 -8.49532225e-05 8.71526837e-01
2.24494517e-01 1.99434087e-01 5.90583324e-01 -5.43556511e-01
-7.34182447e-02 7.12004483e-01 -1.75287992e-01 -9.10409987e-01
-4.90824878e-01 -6.84103727e-01 -1.25944245e+00 1.27343893e-01
4.86469716e-01 -1.22802965e-02 -1.03375733e+00 1.29193389e+00
3.62457275e-01 -1.28154337e-01 1.24050520e-01 6.58180952e-01
1.09782815e+00 5.20180285e-01 7.73582086e-02 -3.52352411e-02
1.46136248e+00 -8.42683852e-01 -4.65310365e-01 2.72099376e-02
5.83391130e-01 -6.16468430e-01 3.07339311e-01 2.57698238e-01
-8.05717826e-01 -3.20500821e-01 -1.10856152e+00 -1.00650005e-01
-6.58249736e-01 6.20865941e-01 2.62178898e-01 1.09416530e-01
-1.25362599e+00 3.14645916e-01 -9.25346196e-01 -5.39397061e-01
8.64426196e-01 4.14154321e-01 -7.13331819e-01 -1.19271360e-01
-6.64364278e-01 6.25415444e-01 2.47391596e-01 1.23132981e-01
-1.10595191e+00 -8.43128860e-01 -6.56926095e-01 4.41121273e-02
-2.74682701e-01 -4.39656287e-01 1.09431100e+00 -1.33843160e+00
-1.06879663e+00 1.33681655e+00 -3.07297409e-01 -5.62722623e-01
3.55254859e-01 4.68989760e-01 1.77623592e-02 5.62968850e-01
1.31454483e-01 8.44466031e-01 5.68138897e-01 -1.23563361e+00
-7.34386683e-01 -4.81177449e-01 -1.90592274e-01 -1.90358475e-01
-3.08935583e-01 -2.83183455e-01 -4.43546861e-01 -4.34021264e-01
3.53760482e-03 -8.48629534e-01 -1.94364712e-01 2.05401108e-01
-5.80350339e-01 3.46245468e-02 8.73075724e-01 -8.11343133e-01
7.30986476e-01 -2.51672244e+00 1.01939440e-01 2.04796106e-01
3.94702941e-01 -6.60528487e-04 -1.15894675e-01 3.59718591e-01
-3.65411282e-01 -2.38445438e-02 5.25654620e-03 -2.15724409e-01
-5.31727910e-01 -1.23480543e-01 1.75706059e-01 8.15584242e-01
3.02557409e-01 8.44713926e-01 -6.16773129e-01 -6.84481680e-01
7.11906478e-02 6.87997758e-01 -2.70973980e-01 3.10579866e-01
7.13363439e-02 6.34865105e-01 -2.09343880e-01 8.40873659e-01
7.93563604e-01 -6.20520115e-01 3.26064199e-01 -6.32802606e-01
-4.40372154e-02 4.42218632e-02 -4.16595519e-01 1.50532210e+00
-2.83106863e-01 8.84493172e-01 3.39481294e-01 -1.40013814e+00
7.23598421e-01 4.69599515e-01 7.68063068e-01 -6.12240314e-01
3.88938099e-01 3.58172476e-01 2.53140211e-01 -5.24233103e-01
-3.69153880e-02 -5.57878576e-02 3.15914094e-01 9.75328460e-02
3.37372720e-01 2.84150481e-01 2.68282771e-01 5.18294126e-02
1.48914385e+00 -1.08387984e-01 5.34617484e-01 -6.13093138e-01
6.72023118e-01 2.16222689e-01 2.99705833e-01 4.16464746e-01
-3.43294382e-01 5.92347085e-01 6.68074906e-01 -7.76726425e-01
-9.15970802e-01 -7.17702210e-01 -5.09832978e-01 7.07141459e-01
6.78849220e-02 -3.36291380e-02 -8.36370885e-01 -7.99392164e-01
-8.09898302e-02 1.50804054e-02 -1.23421037e+00 -2.50881235e-03
-3.26585054e-01 -8.36097479e-01 3.83350462e-01 3.78963023e-01
5.93561232e-01 -9.68837380e-01 -6.46124423e-01 -6.31040260e-02
-1.91928744e-01 -8.73772621e-01 -1.08801305e-01 5.98735452e-01
-6.08560562e-01 -1.28756499e+00 -7.58860528e-01 -1.05770695e+00
1.26832366e+00 1.76797137e-01 8.84220362e-01 2.58730888e-01
-8.87324810e-01 -4.71531413e-02 -3.91267359e-01 -4.78307843e-01
-3.94105732e-01 1.83909968e-01 -6.34751976e-01 3.01033467e-01
4.57924008e-01 -2.22639591e-01 -1.02509308e+00 -9.24553573e-02
-8.75193894e-01 2.56836955e-02 1.12614059e+00 1.12977171e+00
9.58087802e-01 1.21221177e-01 3.10804665e-01 -1.11639166e+00
2.75255013e-02 -7.66179740e-01 -3.00709665e-01 6.77853525e-02
-9.60394964e-02 -3.19760054e-01 6.62457824e-01 -2.03429371e-01
-6.34509027e-01 3.99363905e-01 -2.88001239e-01 -1.86028123e-01
-4.09307986e-01 6.37515128e-01 1.11444429e-01 -3.95285040e-01
5.61746895e-01 4.78671432e-01 1.10560901e-01 -2.30637953e-01
-5.07065296e-01 8.06312859e-01 5.04427254e-01 -6.69146478e-02
2.68345654e-01 7.81712174e-01 9.22674015e-02 -8.19401801e-01
-7.49458909e-01 -7.10261106e-01 -6.79144084e-01 -2.92479575e-01
1.06347549e+00 -7.97201514e-01 -4.50326294e-01 5.66534460e-01
-9.10768211e-01 -4.76584315e-01 -1.35505497e-01 1.23049192e-01
-4.01340991e-01 8.40505213e-03 -7.21055210e-01 -3.85995954e-01
-6.39283121e-01 -1.23888433e+00 1.31285310e+00 3.06018710e-01
-4.92352620e-02 -1.08655930e+00 7.93059096e-02 3.29932094e-01
4.20386255e-01 4.68574882e-01 1.00846004e+00 -7.74047792e-01
-4.31296438e-01 -5.65088391e-01 -4.19791818e-01 3.11186165e-01
3.39111328e-01 2.53758766e-02 -1.08428097e+00 -4.91170466e-01
-4.24047187e-02 -3.69943172e-01 1.04789627e+00 6.31962836e-01
1.51410854e+00 -2.81045854e-01 -7.12806404e-01 7.27538645e-01
1.72791696e+00 2.72336960e-01 7.42206573e-01 3.98943484e-01
2.72375882e-01 4.69022512e-01 2.83432156e-01 1.17962301e-01
2.92304128e-01 3.52388054e-01 5.45328796e-01 -5.52114487e-01
-2.63243824e-01 1.45696819e-01 -1.17611334e-01 3.92238319e-01
1.66364282e-01 -1.71121225e-01 -1.08873582e+00 7.19645202e-01
-1.48336661e+00 -7.65283167e-01 2.71117598e-01 1.93785357e+00
7.17689455e-01 -2.07644761e-01 -1.27310887e-01 1.13664493e-01
4.80513483e-01 -9.29525420e-02 -5.36589384e-01 -7.13633299e-02
8.10676441e-02 2.61505514e-01 5.52356780e-01 2.51657993e-01
-1.48880911e+00 3.69203538e-01 5.54384184e+00 7.77576923e-01
-1.73728323e+00 6.23770170e-02 1.31854200e+00 1.99447185e-01
2.04575703e-01 -3.58457744e-01 -5.55108011e-01 4.95965213e-01
9.36945677e-01 1.99079245e-01 -7.99755976e-02 6.15359068e-01
7.13521987e-02 -2.72468835e-01 -1.13656998e+00 7.40129173e-01
-2.89344192e-02 -1.51347005e+00 1.97731610e-02 2.77018547e-01
5.90605199e-01 2.02610716e-01 -9.00619081e-04 -4.55655567e-02
-2.21686050e-01 -1.21166015e+00 3.42859477e-01 7.20890701e-01
1.01385176e+00 -5.40737092e-01 1.15486133e+00 3.89243960e-01
-9.62543905e-01 -2.59628296e-02 -3.77002954e-01 3.98932993e-01
-6.29615843e-01 6.56681001e-01 -9.96378601e-01 5.29076576e-01
8.71885955e-01 8.63111675e-01 -6.45973086e-01 9.90226150e-01
3.79817605e-01 5.84054351e-01 -6.62006438e-02 1.61009103e-01
1.18587710e-01 4.21946980e-02 2.03629404e-01 1.39365220e+00
2.61973023e-01 -1.01345420e-01 -4.67791408e-03 6.20007217e-01
-2.43564546e-01 1.51744157e-01 -5.04753351e-01 -1.81244597e-01
2.43623331e-01 1.79201448e+00 -8.61110032e-01 -3.76268983e-01
-3.60774457e-01 8.93650830e-01 4.88675565e-01 2.28400946e-01
-4.10494804e-01 -4.40171808e-01 1.33226991e-01 2.51243591e-01
4.19977129e-01 3.19033474e-01 -1.80145741e-01 -8.64482939e-01
-2.41895080e-01 -8.19582582e-01 3.41758132e-01 -4.88580942e-01
-1.11433458e+00 8.06339741e-01 -4.10686404e-01 -1.11728239e+00
2.12579928e-02 -9.54355299e-01 -6.49716258e-01 6.99959278e-01
-1.52625525e+00 -1.29998374e+00 -6.76424503e-01 2.56970227e-01
4.19142872e-01 -1.25224695e-01 1.06103146e+00 2.95855165e-01
-6.30564094e-01 4.80687588e-01 3.09739172e-01 4.17306304e-01
5.48794568e-01 -1.34945750e+00 -3.60011607e-01 4.50391531e-01
-3.52199316e-01 5.16509593e-01 3.35349292e-01 -2.79235780e-01
-1.27840865e+00 -1.31613958e+00 8.70594859e-01 3.56155187e-02
4.39828962e-01 -2.72680879e-01 -7.19483793e-01 5.41284263e-01
4.17003900e-01 5.35299182e-01 1.01306057e+00 -4.66839850e-01
-2.24052027e-01 -4.28287745e-01 -1.41694570e+00 1.55478537e-01
2.02434823e-01 -3.35580707e-01 1.44690990e-01 4.75480586e-01
4.47187126e-02 -4.29834396e-01 -1.18244648e+00 5.10683537e-01
7.27227688e-01 -8.75144780e-01 6.94381356e-01 -2.05741018e-01
9.22229946e-01 -1.54546306e-01 -1.90905079e-01 -1.15479648e+00
-4.57203269e-01 5.85317202e-02 2.72091120e-01 7.44918585e-01
4.48995024e-01 -5.23817182e-01 9.49129403e-01 -8.56480971e-02
4.52961354e-03 -1.35084653e+00 -9.17389750e-01 -2.19866470e-01
2.70313978e-01 3.75311196e-01 1.59544364e-01 7.40804613e-01
5.56210056e-03 -8.34497681e-04 1.21632338e-01 1.14348240e-01
6.38463974e-01 1.85639992e-01 5.29049873e-01 -8.60390067e-01
-2.29050204e-01 -4.19505268e-01 -7.02869594e-01 -6.13273382e-01
-2.21036617e-02 -1.07134175e+00 1.13955483e-01 -1.39052796e+00
7.43446648e-01 -3.80028218e-01 -5.60262620e-01 6.18739665e-01
-1.98516119e-02 6.51095986e-01 -3.14133495e-01 3.47256601e-01
-4.59576607e-01 -1.22597121e-01 1.10687256e+00 -6.06113255e-01
3.49379241e-01 -1.27621487e-01 -7.16238618e-01 3.01800936e-01
9.53752398e-01 -2.06782654e-01 -1.98811479e-02 -3.38900805e-01
-3.01471740e-01 -2.40521654e-02 5.02807975e-01 -1.11992443e+00
3.57410401e-01 1.25694960e-01 8.56528342e-01 -4.82754111e-01
2.81139493e-01 -9.17145848e-01 2.48363048e-01 9.29494083e-01
-5.16576886e-01 -4.78501469e-01 2.18723059e-01 3.43708366e-01
-4.27542388e-01 -1.71010077e-01 1.15401638e+00 -1.95878342e-01
-2.45301306e-01 3.89245927e-01 -4.54744518e-01 -5.81354558e-01
1.11460102e+00 4.25036699e-02 -4.38373834e-01 -9.41622108e-02
-6.77782238e-01 -1.41855538e-01 4.24375087e-01 -1.93914369e-01
5.34419358e-01 -1.17098927e+00 -7.68194795e-01 2.68593222e-01
2.77344197e-01 3.87029871e-02 3.78457487e-01 1.25199580e+00
-7.60930240e-01 4.84419346e-01 -3.08404475e-01 -9.09601033e-01
-1.37118208e+00 3.59545708e-01 7.97645450e-01 -4.66373593e-01
-4.24078822e-01 9.13331270e-01 4.12737697e-01 -1.74556121e-01
1.41729429e-01 -6.28171787e-02 -3.99414808e-01 -1.79771841e-01
6.10688031e-01 -1.34786159e-01 3.15340966e-01 -6.74087465e-01
-4.08940524e-01 3.36912066e-01 -5.58686614e-01 4.01613444e-01
1.40504348e+00 2.52457947e-01 -5.14089525e-01 3.11560929e-01
1.70620155e+00 -3.34819466e-01 -8.32151592e-01 -1.61520675e-01
-1.43763036e-01 -1.78599581e-01 3.17128360e-01 -8.45992386e-01
-1.26743209e+00 8.81461501e-01 8.35094988e-01 2.58880794e-01
1.42837131e+00 3.27770561e-02 5.08854747e-01 2.34899402e-01
2.92128511e-02 -6.45360291e-01 -1.71912745e-01 1.45438179e-01
8.40352535e-01 -1.34700823e+00 -2.76247151e-02 -3.46150577e-01
-1.82798833e-01 1.27385998e+00 4.53999907e-01 -3.57595295e-01
8.98064375e-01 5.27381241e-01 1.96293354e-01 -3.75690222e-01
-9.79540408e-01 7.71703869e-02 2.69356579e-01 3.97716701e-01
7.54530728e-01 3.34599949e-02 -2.74801016e-01 4.95255411e-01
2.16345750e-02 1.96270600e-01 2.61032552e-01 9.65582848e-01
-2.86644727e-01 -7.80157626e-01 -4.92596924e-02 7.67590404e-01
-7.85118282e-01 1.02353849e-01 -5.43137968e-01 7.45541930e-01
2.23568752e-01 5.51640928e-01 3.00203025e-01 -2.93844849e-01
-1.16094269e-01 -1.77483320e-01 5.11040509e-01 -4.45988268e-01
-5.19941568e-01 1.23401493e-01 -2.17728630e-01 -3.60786825e-01
-5.12145698e-01 -6.04130864e-01 -1.05116987e+00 -7.42155612e-02
-1.12603202e-01 1.73191443e-01 8.80062282e-01 7.61528313e-01
3.56652617e-01 7.02651501e-01 7.45559931e-01 -9.02313650e-01
-2.32031122e-01 -1.16210008e+00 -4.45149392e-01 3.45146656e-01
6.48980141e-01 -3.06482404e-01 -2.77447045e-01 4.31370854e-01]
|
[15.093503952026367, -2.9432520866394043]
|
0a0bb4aa-fe18-46d4-a008-863606607f5d
|
liver-segmentation-from-multimodal-images
|
1910.10504
| null |
http://arxiv.org/abs/1910.10504v1
|
http://arxiv.org/pdf/1910.10504v1.pdf
|
Liver Segmentation from Multimodal Images using HED-Mask R-CNN
|
Precise segmentation of the liver is critical for computer-aided diagnosis
such as pre-evaluation of the liver for living donor-based transplantation
surgery. This task is challenging due to the weak boundaries of organs,
countless anatomical variations, and the complexity of the background. Computed
tomography (CT) scanning and magnetic resonance imaging (MRI) images have
different parameters and settings. Thus, images acquired from different
modalities differ from one another making liver segmentation challenging task.
We propose an efficient liver segmentation with the combination of
holistically-nested edge detection (HED) and the Mask-region-convolutional
neural network (R-CNN) to address these challenges. The proposed HED-Mask R-CNN
approach is based on effective identification of edge maps from multimodal
images. The proposed system firstly applies a preprocessing step of image
enhancement to get the 'primal sketches' of the abdomen. Then the HED network
is applied to enhanced CT and MRI modality images to get a better edge map.
Finally, the Mask R-CNN is used to segment the liver from edge map images. We
used a dataset of 20 CT patients and 9 MR patients from the CHAOS challenge.
The system is trained on CT and MRI images separately and then converted to 2D
slices. We significantly improved the segmentation accuracy of CT and MRI
images on a database with a Dice value of 0.94 for CT, 0.89 for T2-weighted MRI
and 0.91 for T1-weighted MRI.
|
[]
|
2019-10-23
| null | null | null | null |
['liver-segmentation']
|
['medical']
|
[-4.37919721e-02 -1.66686177e-01 3.70693915e-02 -1.94977000e-01
-3.27023804e-01 -4.89906520e-01 1.06120773e-01 1.51724964e-01
-5.76217949e-01 3.12309295e-01 1.39893517e-01 -4.26801473e-01
-3.92512083e-02 -6.06898189e-01 -1.13743402e-01 -7.91221857e-01
-7.02115834e-01 5.11041224e-01 1.03469014e-01 1.11803293e-01
-4.57825027e-02 8.08490038e-01 -5.57572365e-01 9.85109583e-02
7.93209016e-01 1.10764802e+00 2.92819351e-01 8.23897362e-01
-1.48567751e-01 3.65382880e-01 -1.42923191e-01 -7.36002624e-02
7.00970411e-01 -6.79778814e-01 -6.59040153e-01 2.88360924e-01
-1.43502563e-01 -3.43063474e-01 -3.05084229e-01 1.21323609e+00
7.81000733e-01 -7.10292980e-02 6.46110535e-01 -8.72924149e-01
-5.26267052e-01 5.78061342e-01 -6.50265992e-01 4.83059824e-01
-2.07700059e-01 1.17267713e-01 1.15693286e-01 -1.02691782e+00
6.10787094e-01 4.69911337e-01 8.32807243e-01 4.39066440e-01
-8.57013524e-01 -4.61063951e-01 -6.09754801e-01 2.02987507e-01
-1.48380899e+00 4.27669361e-02 7.17859745e-01 -4.86467659e-01
5.17004013e-01 2.13559896e-01 1.19000626e+00 1.50177311e-02
3.95534396e-01 4.29400265e-01 1.18939841e+00 -3.10719371e-01
-2.16653142e-02 -1.38331488e-01 -2.38549799e-01 1.01467168e+00
2.76968956e-01 1.63099989e-01 2.36680284e-01 1.62354991e-01
1.22876477e+00 2.90578038e-01 -6.85155272e-01 -3.60960186e-01
-1.66688752e+00 5.75120986e-01 9.32587326e-01 7.41119921e-01
-7.82688856e-01 -1.29049838e-01 3.69624794e-01 1.71973780e-02
1.27247900e-01 2.30363503e-01 -3.47532809e-01 3.37895483e-01
-1.20532131e+00 -6.29335463e-01 9.89814878e-01 6.56419516e-01
3.42380583e-01 5.08741327e-02 -9.36578363e-02 5.09795845e-01
2.15024844e-01 6.08432591e-02 7.87817538e-01 -5.32070160e-01
-1.42380610e-01 5.58611989e-01 -3.10174644e-01 -9.00659740e-01
-7.23640203e-01 -6.38511002e-01 -1.62386584e+00 2.40001887e-01
5.92108369e-01 -2.73924947e-01 -1.31245208e+00 1.12525487e+00
5.01165867e-01 2.58052766e-01 -1.08296752e-01 1.48244333e+00
1.24189627e+00 3.44347179e-01 1.11901373e-01 -1.53666556e-01
1.68906665e+00 -1.11243951e+00 -6.17435038e-01 1.01626091e-01
5.38107157e-01 -8.40774298e-01 4.77077752e-01 7.70669952e-02
-1.02684617e+00 -9.56641808e-02 -1.00610566e+00 2.91559577e-01
-2.39582002e-01 2.84104168e-01 4.68155652e-01 5.86533844e-01
-1.17580616e+00 4.34949994e-01 -9.40334558e-01 -1.99669570e-01
5.01073122e-01 6.89288199e-01 -6.39383793e-01 -1.13094591e-01
-7.85172641e-01 1.03181016e+00 6.64571822e-01 3.65302682e-01
-7.12069750e-01 -8.80495548e-01 -9.48805630e-01 9.80271772e-02
1.59875393e-01 -7.59279847e-01 7.84900367e-01 -9.58228528e-01
-1.35940456e+00 1.03534365e+00 3.40650052e-01 -3.52719724e-01
7.47454345e-01 5.33266127e-01 -1.05170839e-01 5.90460896e-01
-3.44974577e-01 8.22116256e-01 6.03212416e-01 -1.16175318e+00
-2.75685281e-01 -3.26855958e-01 -5.14045715e-01 3.08438659e-01
3.65303278e-01 2.46389568e-01 -4.81204212e-01 -7.16936171e-01
5.01953840e-01 -8.46763372e-01 -4.72380072e-01 1.22859143e-01
-4.00311828e-01 4.28501159e-01 7.06720710e-01 -1.33617210e+00
9.26857710e-01 -1.84289968e+00 6.86349021e-03 2.95934498e-01
4.73796219e-01 2.64544159e-01 -2.73368582e-02 -4.48070586e-01
-4.15675044e-01 1.55595750e-01 -5.36131561e-01 -2.30237264e-02
-4.84088629e-01 4.90997210e-02 6.04653478e-01 8.11217248e-01
-6.37518317e-02 1.30184996e+00 -8.37110102e-01 -9.96789515e-01
4.49691653e-01 6.93079174e-01 -2.66615778e-01 3.63492608e-01
4.50426221e-01 1.05560493e+00 4.80785519e-02 8.89140904e-01
7.72412419e-01 -2.42391661e-01 2.02942058e-01 -4.94545192e-01
-9.31620076e-02 -3.54199469e-01 -1.02351940e+00 1.70412457e+00
-3.27234149e-01 5.10058582e-01 5.24283946e-01 -8.55714560e-01
6.85167968e-01 7.20460474e-01 1.01555347e+00 -6.16545081e-01
4.35680985e-01 3.50548416e-01 4.83773649e-01 -7.04933047e-01
-1.36656463e-01 -4.21234548e-01 3.34003478e-01 5.56107700e-01
-2.18236819e-02 -3.72440845e-01 1.26580223e-01 -1.05963126e-01
5.36952019e-01 -3.37019175e-01 3.80864292e-01 -4.88233268e-01
6.45130932e-01 1.34407640e-01 5.79250932e-01 2.63174146e-01
-6.37717128e-01 1.01508033e+00 7.72982463e-02 -8.82005572e-01
-1.15257132e+00 -7.28886664e-01 -2.96160579e-01 3.38984042e-01
3.38792980e-01 2.55088627e-01 -8.39078009e-01 -8.24295342e-01
-1.68838069e-01 9.31419507e-02 -5.94938934e-01 2.82480329e-01
-8.47129166e-01 -1.05892110e+00 3.25172573e-01 3.54460537e-01
7.80748904e-01 -9.29858565e-01 -9.84107018e-01 2.46138722e-01
-2.95836866e-01 -9.83647823e-01 -9.39617932e-01 1.41607866e-01
-1.05736184e+00 -1.22207665e+00 -1.34221470e+00 -1.26931965e+00
1.21440077e+00 1.69397458e-01 1.20474720e+00 5.64811170e-01
-6.91964626e-01 1.20864086e-01 2.89556812e-02 8.16061074e-05
-3.44018281e-01 -6.13608435e-02 -1.22993641e-01 -1.99423090e-01
-1.87978268e-01 -4.25287664e-01 -1.18312156e+00 3.89366031e-01
-1.01968074e+00 3.17497402e-01 9.26047742e-01 1.11409795e+00
6.00212038e-01 9.54384729e-02 1.39096558e-01 -4.29735065e-01
5.19243538e-01 -3.81498009e-01 -5.65300524e-01 4.15828466e-01
-5.88492692e-01 -2.35861003e-01 3.98120463e-01 -5.47994852e-01
-6.74410284e-01 3.62935096e-01 2.16824487e-02 -4.90193099e-01
-1.70515552e-01 5.88386655e-01 3.01467478e-01 -6.75761104e-01
3.52946967e-01 3.11894894e-01 2.71763533e-01 -1.40470073e-01
4.07987423e-02 4.18125093e-01 7.81579137e-01 -1.25671357e-01
4.26768631e-01 3.65882933e-01 2.06040919e-01 -4.77939188e-01
2.47455612e-02 -4.37288254e-01 -8.51352930e-01 -3.94579321e-01
1.07434380e+00 -5.98316967e-01 -6.30121708e-01 4.63784724e-01
-1.01613152e+00 -4.60720062e-01 -1.70475528e-01 9.88147378e-01
-7.92551041e-02 6.07337236e-01 -1.14072549e+00 -2.45048791e-01
-9.83370245e-01 -1.67876506e+00 4.05614555e-01 6.58962309e-01
2.23860443e-01 -1.35799444e+00 -1.75633147e-01 6.06732033e-02
9.48011100e-01 5.12346089e-01 8.35104942e-01 -7.77765334e-01
-5.93850017e-01 -2.64163584e-01 -5.28827012e-01 3.43276113e-01
2.07890600e-01 -3.53784740e-01 -3.18760335e-01 -3.71850520e-01
1.87912196e-01 2.92848885e-01 5.46209574e-01 9.16326463e-01
9.33781862e-01 -2.40522519e-01 2.14263592e-02 1.09319782e+00
1.64520729e+00 4.49023664e-01 4.73102897e-01 1.18005544e-01
5.91180265e-01 4.40950155e-01 -1.29213068e-03 2.79623359e-01
2.07587510e-01 2.33139038e-01 5.83068311e-01 -7.50890672e-01
-3.92150313e-01 3.47188622e-01 -2.28672728e-01 1.10754609e+00
-1.30719945e-01 3.68113279e-01 -1.29900455e+00 8.26504707e-01
-1.22054911e+00 -5.23567021e-01 -2.01965839e-01 2.05449224e+00
6.74719691e-01 -5.05660295e-01 -2.93528810e-02 -2.17233062e-01
1.06301749e+00 -4.65239644e-01 -3.51552218e-01 -2.79185213e-02
1.46590546e-01 2.99922466e-01 6.87645614e-01 5.81403553e-01
-1.20312977e+00 3.63877743e-01 5.77961111e+00 3.88403237e-01
-1.62353826e+00 1.94306314e-01 1.05673468e+00 4.44301188e-01
6.95901960e-02 -1.47497490e-01 3.81943793e-03 4.78831679e-01
2.54607499e-01 -2.42252037e-01 6.61878943e-01 4.83420789e-01
1.38962530e-02 -4.30494815e-01 -8.30170512e-01 1.18670177e+00
8.06684867e-02 -1.23161125e+00 -3.30470800e-01 -5.73543720e-02
7.98281789e-01 2.44596139e-01 -1.28377825e-01 -1.18139945e-01
-7.12866485e-02 -1.36654305e+00 2.58794695e-01 3.59471947e-01
9.57607448e-01 -5.50883591e-01 1.18775845e+00 3.94229025e-01
-1.20855796e+00 2.26670325e-01 -4.58914079e-02 4.21400815e-01
3.80436480e-02 6.70281887e-01 -1.36412275e+00 6.05739474e-01
5.57982266e-01 2.31138051e-01 -3.03050697e-01 1.61708212e+00
-4.30774540e-02 1.90390319e-01 -5.62261581e-01 3.02703083e-01
1.68617830e-01 -6.68407619e-01 4.57167417e-01 1.51221502e+00
5.63800633e-01 4.51666176e-01 2.22050443e-01 8.74888480e-01
-2.44474337e-01 2.67372549e-01 -3.89020741e-01 3.27271819e-01
8.24896693e-02 1.91106963e+00 -1.46642125e+00 -5.75404644e-01
-1.96837440e-01 8.23882401e-01 -3.87930065e-01 2.53889114e-01
-6.69201672e-01 -2.51479357e-01 -1.54446870e-01 -1.20674588e-01
8.49782750e-02 -2.41591260e-01 -5.23603797e-01 -1.31924570e+00
-4.01309311e-01 -5.85478663e-01 5.38275003e-01 -5.74192226e-01
-1.11030066e+00 9.50101972e-01 -1.66963339e-01 -1.22539020e+00
-3.27817388e-02 -4.03723449e-01 -8.97900224e-01 1.11418855e+00
-1.73616827e+00 -1.03347731e+00 -7.71216512e-01 6.13535345e-01
3.14934611e-01 1.59823686e-01 6.26449049e-01 5.80786169e-01
-3.84660453e-01 1.67772219e-01 -2.41525784e-01 7.99293756e-01
3.96095276e-01 -1.39304233e+00 -7.93937519e-02 8.26623142e-01
-3.37584615e-01 3.22829694e-01 1.76271141e-01 -7.31714010e-01
-1.30945969e+00 -9.71924245e-01 6.35873318e-01 1.60115257e-01
1.85114041e-01 3.08252424e-01 -7.86619544e-01 5.97415507e-01
5.46004176e-01 7.65296221e-01 7.14500725e-01 -9.77552533e-01
3.70489061e-01 2.67872512e-01 -1.72195458e+00 3.29426974e-01
3.14661413e-01 -1.01167411e-01 -4.92485195e-01 2.14217350e-01
1.80969059e-01 -9.30974782e-01 -1.38798797e+00 6.09231830e-01
5.73163688e-01 -7.97730863e-01 1.18165445e+00 -2.24618524e-01
1.70561746e-01 -3.41832370e-01 2.33408213e-01 -1.40732336e+00
-2.26232633e-01 -4.15699571e-01 2.38136232e-01 4.42563087e-01
1.62677705e-01 -6.31979942e-01 7.08877683e-01 7.75269389e-01
-2.50876129e-01 -9.54476893e-01 -9.72388268e-01 -2.97863424e-01
1.51887000e-01 -3.04186381e-02 4.48654741e-01 1.25430655e+00
-8.47479925e-02 -1.85669228e-01 1.20971940e-01 1.50717393e-01
8.73859227e-01 1.87765062e-01 2.25447848e-01 -9.53750849e-01
-2.27757590e-03 -7.88405716e-01 -3.59889805e-01 -6.25187159e-01
-4.76452917e-01 -1.32013524e+00 6.27445057e-02 -1.69114113e+00
4.28344220e-01 -5.77989340e-01 -3.62753510e-01 4.00035858e-01
-1.63277641e-01 5.38966775e-01 4.30917829e-01 3.82682681e-01
-9.82635617e-02 -2.06661224e-02 1.81584001e+00 -2.34303832e-01
-2.38481700e-01 -7.56196156e-02 -1.58315659e-01 6.52655900e-01
9.32137072e-01 -4.07283783e-01 1.20350145e-01 -3.09583753e-01
-3.14705819e-01 6.74516022e-01 3.43853682e-01 -8.74489903e-01
6.78253412e-01 1.96526274e-01 9.21542883e-01 -5.66214025e-01
-1.77720621e-01 -1.19417143e+00 5.02418637e-01 1.17008579e+00
-5.19271754e-03 3.76818389e-01 1.08478114e-01 -1.44379362e-01
-2.54379153e-01 -2.02921078e-01 1.12480640e+00 -7.75970638e-01
-3.39241564e-01 6.76651061e-01 -1.57910347e-01 -1.01209134e-01
1.16193736e+00 -1.74682453e-01 1.79715976e-01 -3.34076464e-01
-1.03983831e+00 1.87428623e-01 4.01301563e-01 -3.17514658e-01
9.78409469e-01 -1.31841815e+00 -8.76546800e-01 5.18117487e-01
-5.02952039e-01 3.49315166e-01 1.78773195e-01 1.78639948e+00
-1.35602212e+00 2.12971240e-01 -4.93603647e-01 -8.39798510e-01
-1.19434702e+00 5.16059697e-01 9.33016419e-01 -5.39243519e-01
-7.99411833e-01 5.75979650e-01 8.55827779e-02 -4.91539955e-01
1.75234854e-01 -6.41388416e-01 -6.07629493e-02 -2.09422886e-01
3.03567588e-01 1.41437545e-01 1.91217318e-01 -8.02129149e-01
-3.39834124e-01 7.00026929e-01 2.94183731e-01 4.28962376e-04
1.24319959e+00 -3.38225216e-02 -5.64086080e-01 -4.89674717e-01
1.26103163e+00 -2.62675822e-01 -9.62647140e-01 -2.16863707e-01
-6.01438172e-02 -4.64436322e-01 4.36815172e-01 -1.00532639e+00
-1.76498199e+00 1.04695332e+00 9.05065238e-01 1.47305116e-01
1.34493208e+00 -3.78473699e-01 1.03227186e+00 -3.07446450e-01
1.61604226e-01 -5.56573272e-01 -1.20216683e-01 2.66397983e-01
7.03066468e-01 -1.40903342e+00 4.51700762e-02 -3.39477599e-01
-5.87937593e-01 1.48395109e+00 2.59775460e-01 8.71414691e-02
7.38903701e-01 4.60244536e-01 3.58702838e-01 -1.19773068e-01
-6.77061304e-02 -1.25368968e-01 5.44252276e-01 3.54052126e-01
3.31417412e-01 2.25212514e-01 -2.79333234e-01 4.11653847e-01
2.92744547e-01 2.57981151e-01 3.01099658e-01 8.95417213e-01
-2.78466433e-01 -6.31171763e-01 -6.71277225e-01 3.16525906e-01
-6.79554224e-01 -2.61344492e-01 1.82288722e-03 8.40316713e-01
1.79386124e-01 4.02681470e-01 1.12780273e-01 -4.39911522e-02
-1.09171383e-01 6.66082576e-02 5.51045060e-01 -1.41628608e-01
-1.16010237e+00 2.23706290e-01 -4.24466372e-01 -2.21117988e-01
-2.64762253e-01 -2.59594232e-01 -1.49923706e+00 -2.87291288e-01
-2.87047178e-01 2.72678703e-01 1.22925472e+00 5.97062588e-01
1.77833289e-01 5.03234863e-01 7.27685153e-01 -1.14313912e+00
-2.84069210e-01 -7.96598852e-01 -4.61727083e-01 5.20763636e-01
3.88244897e-01 -1.26459286e-01 -2.86918849e-01 2.28955001e-01]
|
[14.477005958557129, -2.6889665126800537]
|
e2403e54-b533-4c6d-897b-f744b84e91c6
|
a-state-transition-model-for-mobile
|
2207.03099
| null |
https://arxiv.org/abs/2207.03099v1
|
https://arxiv.org/pdf/2207.03099v1.pdf
|
A State Transition Model for Mobile Notifications via Survival Analysis
|
Mobile notifications have become a major communication channel for social networking services to keep users informed and engaged. As more mobile applications push notifications to users, they constantly face decisions on what to send, when and how. A lack of research and methodology commonly leads to heuristic decision making. Many notifications arrive at an inappropriate moment or introduce too many interruptions, failing to provide value to users and spurring users' complaints. In this paper we explore unique features of interactions between mobile notifications and user engagement. We propose a state transition framework to quantitatively evaluate the effectiveness of notifications. Within this framework, we develop a survival model for badging notifications assuming a log-linear structure and a Weibull distribution. Our results show that this model achieves more flexibility for applications and superior prediction accuracy than a logistic regression model. In particular, we provide an online use case on notification delivery time optimization to show how we make better decisions, drive more user engagement, and provide more value to users.
|
['Romer Rosales', 'Shipeng Yu', 'Shaunak Chatterjee', 'Jing Zhang', 'Yiping Yuan']
|
2022-07-07
| null | null | null | null |
['survival-analysis']
|
['miscellaneous']
|
[ 3.00548255e-01 -1.73028689e-02 -7.02462614e-01 -4.90284443e-01
-5.82244277e-01 -5.68910599e-01 3.77559692e-01 2.15739533e-01
-3.85447353e-01 7.85328150e-01 2.45153040e-01 -9.70059812e-01
-3.11139941e-01 -6.61796212e-01 -1.80994734e-01 -1.72349960e-01
-2.03840863e-02 3.12007219e-01 2.23218232e-01 -9.80135873e-02
1.88347593e-01 4.02923971e-01 -1.30575407e+00 -2.85864938e-02
7.76760697e-01 9.20373440e-01 2.44768068e-01 7.84875393e-01
-1.88289464e-01 5.18160701e-01 -7.19718039e-01 -2.69936293e-01
-2.09306508e-01 -6.24168962e-02 -4.43761468e-01 -5.88065200e-02
-3.54884684e-01 -7.05582380e-01 -5.94049739e-03 3.16250384e-01
5.15017211e-01 1.72504842e-01 5.10231256e-01 -1.73287642e+00
-3.39147449e-01 3.41789782e-01 -2.33176634e-01 3.98372203e-01
7.35009909e-01 -5.70145287e-02 6.33187890e-01 -1.09217517e-01
2.01982215e-01 9.82458532e-01 7.60805964e-01 4.42382395e-01
-1.25631189e+00 -5.93219697e-01 4.03274238e-01 -1.18464157e-01
-1.08227074e+00 -7.24531531e-01 3.18014443e-01 -4.49521244e-01
7.53773808e-01 1.00246274e+00 5.70155621e-01 1.23023367e+00
4.03322101e-01 4.87083882e-01 5.75704694e-01 -2.29276076e-01
2.65020430e-01 4.82433915e-01 3.98751795e-01 -3.75309326e-02
2.81690657e-01 -2.67451346e-01 -3.22591424e-01 -6.86500609e-01
4.64655340e-01 5.68328381e-01 -6.76290542e-02 2.33868584e-01
-5.26712477e-01 5.23557127e-01 -2.10373238e-01 4.04276937e-01
-5.90132058e-01 1.04269840e-01 -1.17467217e-01 3.09229106e-01
5.82839668e-01 5.76120950e-02 -4.81596023e-01 -9.97696161e-01
-5.78172803e-01 1.90499306e-01 8.81560981e-01 8.67640674e-01
3.48952085e-01 -3.87312889e-01 -4.76139307e-01 8.43516886e-01
3.45815808e-01 4.96029347e-01 2.95928586e-02 -9.05292690e-01
2.22202554e-01 4.20913696e-01 5.49891233e-01 -1.02752161e+00
-6.63066089e-01 -4.24867183e-01 -4.57902700e-01 -3.38078707e-01
4.25025523e-01 -3.31732184e-01 -1.22114249e-01 1.70244777e+00
-8.38753581e-02 1.24487415e-01 -3.74080330e-01 3.03680241e-01
1.60054013e-01 5.14345706e-01 1.09131426e-01 -7.29203641e-01
1.23407662e+00 -1.93248674e-01 -1.01231849e+00 -1.51932895e-01
7.37051547e-01 -9.16942954e-01 1.31940949e+00 3.78083974e-01
-1.12373018e+00 -5.30666858e-02 -3.03319782e-01 4.39259291e-01
-5.97933754e-02 -2.58579016e-01 7.91855872e-01 1.15959144e+00
-9.77843165e-01 4.34016347e-01 -9.76670682e-01 -5.06879747e-01
3.81410390e-01 5.04685640e-01 4.11311865e-01 3.32536176e-02
-8.36247146e-01 6.07880950e-01 -4.98757750e-01 -2.19483569e-01
-1.19715907e-01 -7.56671488e-01 -2.89649129e-01 2.35700995e-01
3.58866215e-01 -6.15661383e-01 1.80846417e+00 -5.99132836e-01
-1.32995939e+00 1.77389786e-01 -4.11925822e-01 -1.65619776e-01
4.55005050e-01 -7.49348179e-02 -6.65230751e-01 -3.85949701e-01
1.68569088e-01 3.24248187e-02 5.05253077e-01 -9.82506692e-01
-8.68230402e-01 -1.21756002e-01 1.84236541e-01 2.17420496e-02
-6.88263416e-01 4.07028049e-01 -4.41412628e-01 -3.11199784e-01
-1.82037398e-01 -8.81124973e-01 -2.89053172e-01 -3.88236642e-01
-2.97371060e-01 -2.27197215e-01 7.44705141e-01 -4.31650788e-01
1.93698919e+00 -2.00346303e+00 -5.37289143e-01 2.54910797e-01
2.29695991e-01 -8.66929069e-02 1.77419022e-01 6.44991875e-01
4.15146917e-01 5.60994208e-01 2.37350345e-01 -4.74606872e-01
5.20261824e-02 1.95521176e-01 -1.97695628e-01 6.17660955e-02
-1.99116305e-01 5.96356511e-01 -9.64790046e-01 -1.00173049e-01
7.89827257e-02 6.37763500e-01 -6.76333964e-01 2.69314408e-01
1.67940110e-01 4.00031477e-01 -6.02055848e-01 7.26041555e-01
5.11005700e-01 -5.07157385e-01 2.08944201e-01 5.28429091e-01
-2.27658331e-01 6.60817981e-01 -7.27391541e-01 7.31866717e-01
-8.28211486e-01 5.59125304e-01 9.81218070e-02 -4.37598556e-01
5.97064197e-01 2.21485853e-01 6.56524062e-01 -7.54190028e-01
1.56709105e-01 -8.55720192e-02 -2.18944058e-01 -4.77261305e-01
4.13396746e-01 1.70099154e-01 -8.88682902e-02 6.75323844e-01
-7.63231695e-01 3.37333113e-01 -3.20940465e-02 2.97657788e-01
1.47491241e+00 -5.22845805e-01 1.71336770e-01 1.13973930e-01
-1.19586408e-01 -5.92476726e-01 3.89070600e-01 1.08401060e+00
-2.20645770e-01 1.67931095e-01 8.41767490e-01 9.33364406e-03
-3.21651101e-01 -9.35616672e-01 -9.82893929e-02 1.30062270e+00
-3.44062522e-02 -4.91655946e-01 -6.58572078e-01 -4.10654128e-01
4.24116887e-02 1.02851450e+00 -2.21006870e-01 -2.81654388e-01
-2.33885758e-02 -7.61538267e-01 1.86544776e-01 2.01442391e-01
-5.51939495e-02 -8.44300687e-01 -5.23417592e-01 3.60104114e-01
-4.37256008e-01 -1.04835439e+00 -6.77107513e-01 -2.79348165e-01
-9.00574028e-01 -8.11202824e-01 -4.40696210e-01 -1.25473440e-01
5.45411348e-01 6.74446762e-01 8.67927253e-01 2.30681911e-01
2.35103056e-01 7.64481604e-01 -2.03459367e-01 -4.11620378e-01
-3.71764064e-01 2.30326355e-01 2.15542093e-01 1.80685803e-01
2.15143383e-01 -7.39892542e-01 -6.70187771e-01 7.58065403e-01
-7.92115927e-01 -3.38028848e-01 2.82579660e-01 2.06205279e-01
1.02954909e-01 1.24939412e-01 1.00388098e+00 -9.49443161e-01
1.12568247e+00 -8.79848242e-01 -4.85929549e-01 -1.46798706e-02
-1.11942887e+00 -4.94628996e-01 1.59618720e-01 -5.97989440e-01
-9.66231942e-01 -3.18651706e-01 -2.93586910e-01 2.89261013e-01
-7.92958364e-02 3.77511412e-01 -1.82018876e-01 1.41234294e-01
4.06279862e-01 -2.44701773e-01 8.47640913e-03 -6.39210939e-01
5.53556904e-02 1.31876147e+00 -8.09643976e-03 -2.07643285e-01
2.04342917e-01 3.84459466e-01 -4.88180369e-01 -1.08640814e+00
-4.40324634e-01 -7.30640948e-01 9.31216329e-02 -4.99244720e-01
3.03891808e-01 -4.79427159e-01 -1.34084547e+00 2.34288469e-01
-1.03937030e+00 -3.93939108e-01 1.57693386e-01 2.60671198e-01
-4.42515999e-01 1.34132475e-01 -4.05358940e-01 -1.52647519e+00
2.15613171e-01 -9.04075921e-01 8.08596671e-01 3.20588320e-01
-6.07260287e-01 -1.12437892e+00 -1.39800176e-01 4.52977329e-01
8.76997888e-01 -5.54767027e-02 4.81163144e-01 -5.23775518e-01
-6.00114048e-01 -5.63484669e-01 -1.51277527e-01 -1.60459593e-01
6.68734908e-01 5.57637997e-02 -8.69347811e-01 -2.09608167e-01
-3.92131582e-02 5.98636270e-01 1.20773874e-01 7.72745848e-01
1.30999339e+00 -8.51375341e-01 -7.97166228e-01 3.96849692e-01
8.57387543e-01 6.53869331e-01 7.64858544e-01 3.20760339e-01
2.52762794e-01 4.99012411e-01 5.40730834e-01 6.78124070e-01
6.85436845e-01 9.77061212e-01 3.88815105e-01 1.16675355e-01
4.76035982e-01 -2.32809022e-01 4.73512381e-01 5.15517890e-01
1.22671516e-03 -6.36775434e-01 -8.49552393e-01 3.29707652e-01
-1.95416045e+00 -8.47783864e-01 -2.62059420e-01 2.59472156e+00
6.59033000e-01 6.74994230e-01 5.58917761e-01 1.44149348e-01
5.87174535e-01 -3.64106327e-01 -2.25599736e-01 -4.11257148e-01
3.17471534e-01 -2.81792313e-01 7.03833282e-01 8.50136757e-01
-4.81305301e-01 5.18339694e-01 6.93115950e+00 6.11726344e-01
-9.71024394e-01 3.68809074e-01 7.90676236e-01 -2.22857296e-01
-5.85830510e-01 9.59883332e-02 -8.64389718e-01 8.88848245e-01
1.35372818e+00 -8.53435695e-02 5.89338243e-01 6.23263001e-01
1.07937300e+00 -2.57013649e-01 -9.18753147e-01 8.86849999e-01
-4.27411616e-01 -1.13769889e+00 -3.93081039e-01 5.03804982e-01
2.96851724e-01 -2.69648463e-01 4.62818034e-02 2.15388998e-01
8.16675499e-02 -7.96304464e-01 3.30877483e-01 1.05142224e+00
5.97432971e-01 -6.57359540e-01 4.66702640e-01 4.92832661e-01
-7.91406631e-01 -2.95229584e-01 2.44607672e-01 -7.02527583e-01
6.60151482e-01 8.55886817e-01 -1.00974333e+00 -4.28719334e-02
5.64780474e-01 2.92908013e-01 -4.23643649e-01 1.10716593e+00
3.41251075e-01 8.50280523e-01 -4.75465745e-01 -4.78812516e-01
-2.21661165e-01 -1.49506973e-02 4.86167878e-01 1.10104477e+00
4.06925887e-01 1.61849428e-02 -1.14380894e-03 4.36818331e-01
2.19000489e-01 -4.39470373e-02 -7.02382326e-01 -2.58680433e-01
7.74677515e-01 1.17225432e+00 -9.46447194e-01 -6.90383986e-02
-4.76708144e-01 6.22490108e-01 -2.55748749e-01 6.33496761e-01
-7.60091066e-01 -3.53385657e-01 9.54529583e-01 8.04663420e-01
-2.11316109e-01 -3.29021364e-01 -4.52341586e-01 -6.91353321e-01
9.10195708e-02 -4.66467768e-01 2.99629048e-02 -5.98859906e-01
-1.01217699e+00 2.26522699e-01 -3.13092917e-02 -9.13180888e-01
-3.75348181e-01 -2.60687292e-01 -7.76621401e-01 5.96401274e-01
-1.21119595e+00 -4.84131098e-01 -1.24391362e-01 1.61179945e-01
1.81388974e-01 2.57407427e-01 6.03919804e-01 7.34707296e-01
-4.77456629e-01 6.90747976e-01 8.80242512e-02 -5.56253552e-01
5.46845734e-01 -8.77524853e-01 3.57014745e-01 3.82867247e-01
-3.10141444e-01 9.08877909e-01 7.45294750e-01 -7.38122940e-01
-1.15127993e+00 -7.32489705e-01 1.23961914e+00 -1.03751957e+00
4.07841295e-01 -4.15063292e-01 -7.54592538e-01 6.14066184e-01
-2.10746780e-01 -5.13610840e-01 8.66902113e-01 6.89831376e-01
3.81497145e-01 -1.73815250e-01 -1.03084624e+00 8.20847631e-01
1.01816392e+00 -4.84208912e-01 1.39215216e-01 3.22119713e-01
8.05256903e-01 -6.98848348e-03 -6.68549597e-01 -7.70870894e-02
7.07451880e-01 -1.09364486e+00 6.58760548e-01 -5.20583034e-01
-3.40086162e-01 1.17004760e-01 1.28265365e-03 -8.82423818e-01
-2.10372940e-01 -1.42475617e+00 -2.76335597e-01 1.49848771e+00
6.23393595e-01 -8.27642977e-01 5.98020792e-01 1.07876265e+00
1.48956969e-01 -8.88205528e-01 -8.28231812e-01 -7.69705594e-01
-5.88625669e-01 -1.07755911e+00 7.94379890e-01 7.61379361e-01
1.31533891e-01 3.87199163e-01 -4.17957217e-01 5.23873828e-02
2.83510268e-01 -4.78054672e-01 6.74698591e-01 -1.32701361e+00
-1.15056485e-01 -7.49192297e-01 -6.46135509e-02 -1.31220627e+00
-4.18789208e-01 -4.15387183e-01 -2.16847152e-01 -1.50620198e+00
9.48827863e-02 -8.96498501e-01 -2.60686815e-01 5.19520700e-01
1.36815459e-01 -9.33707505e-02 -1.22276314e-01 1.04003482e-01
-7.26017773e-01 2.07128897e-01 5.87947011e-01 3.01247984e-01
-8.07941318e-01 1.20192587e+00 -1.25642455e+00 6.19937003e-01
9.45338905e-01 -4.53172386e-01 -4.37192649e-01 2.79137976e-02
8.01433265e-01 3.38008970e-01 2.33602479e-01 -4.59481716e-01
1.55879736e-01 -4.62550163e-01 -1.22401662e-01 -3.29877526e-01
2.10198313e-01 -9.22631800e-01 1.75537795e-01 2.85328895e-01
-1.36218309e-01 8.11923891e-02 1.99754149e-01 7.90641129e-01
4.55233783e-01 1.04416594e-01 1.85120746e-01 6.47772014e-01
2.65799612e-01 1.99969262e-01 -9.90896404e-01 -1.66784272e-01
7.19041824e-01 -2.35997230e-01 -7.41932169e-02 -1.14989316e+00
-7.86325455e-01 4.85398859e-01 4.03582841e-01 7.25650728e-01
3.31198007e-01 -1.27827418e+00 -6.39836639e-02 6.26196042e-02
-1.09894909e-01 -5.43350160e-01 9.95074511e-02 9.92320776e-01
1.39080927e-01 4.75473672e-01 2.39121601e-01 -6.05236292e-01
-1.31297112e+00 8.12572688e-02 1.05341360e-01 8.10364187e-02
-1.35144427e-01 3.97652090e-01 -2.93504238e-01 -1.60407767e-01
6.10702991e-01 -5.91722131e-01 -7.43262991e-02 8.75016153e-02
7.52900541e-01 8.31038594e-01 1.39456987e-01 -1.58598050e-01
-5.75270474e-01 -6.80909753e-02 9.88143235e-02 -3.25714141e-01
1.03367519e+00 -6.69641614e-01 1.13509096e-01 5.81369102e-01
8.94390643e-01 3.85946333e-01 -1.33619487e+00 1.69762477e-01
1.94820687e-01 -7.15401053e-01 6.79039955e-02 -7.38339067e-01
-6.36406302e-01 3.77253264e-01 6.78044558e-01 1.10503840e+00
1.02399337e+00 3.24569680e-02 8.44305217e-01 2.12919340e-01
1.26179561e-01 -9.42890882e-01 -1.52633846e-01 2.85657287e-01
4.42514300e-01 -1.17403996e+00 -3.84154648e-01 -3.55250299e-01
-4.31355119e-01 5.65208375e-01 2.64659792e-01 7.19732761e-01
9.70392108e-01 3.16530913e-01 1.19903695e-03 1.89929336e-01
-8.35194170e-01 -6.23118766e-02 -4.63796519e-02 8.00088227e-01
6.07005060e-01 2.82382369e-01 -4.60739136e-01 1.11276472e+00
-1.98662821e-02 2.00909555e-01 6.93849504e-01 8.73685122e-01
-5.22955060e-01 -1.45937943e+00 -2.33231068e-01 8.96217525e-01
-6.34629965e-01 4.46781563e-03 -2.78943628e-01 4.27811623e-01
-2.12659150e-01 1.61112642e+00 6.60683960e-02 -3.57387990e-01
4.99449611e-01 -2.15081349e-02 -2.49887090e-02 -6.38939381e-01
-2.56846368e-01 4.33670878e-02 3.01114708e-01 -4.28566784e-01
6.23465702e-02 -9.15356338e-01 -1.10155857e+00 -7.90844023e-01
-5.35447598e-01 3.18787515e-01 9.18978274e-01 9.45487499e-01
7.21478403e-01 6.97182834e-01 8.63448858e-01 -5.98213971e-01
-3.06692570e-01 -7.12156296e-01 -3.47310603e-01 -2.76335090e-01
4.50286716e-01 -5.95756233e-01 -4.23846096e-01 -2.67544657e-01]
|
[10.222745895385742, 5.852547645568848]
|
463fb70a-8e46-4805-b493-25b946184054
|
repainting-and-imitating-learning-for-lane
|
2210.05097
| null |
https://arxiv.org/abs/2210.05097v1
|
https://arxiv.org/pdf/2210.05097v1.pdf
|
Repainting and Imitating Learning for Lane Detection
|
Current lane detection methods are struggling with the invisibility lane issue caused by heavy shadows, severe road mark degradation, and serious vehicle occlusion. As a result, discriminative lane features can be barely learned by the network despite elaborate designs due to the inherent invisibility of lanes in the wild. In this paper, we target at finding an enhanced feature space where the lane features are distinctive while maintaining a similar distribution of lanes in the wild. To achieve this, we propose a novel Repainting and Imitating Learning (RIL) framework containing a pair of teacher and student without any extra data or extra laborious labeling. Specifically, in the repainting step, an enhanced ideal virtual lane dataset is built in which only the lane regions are repainted while non-lane regions are kept unchanged, maintaining the similar distribution of lanes in the wild. The teacher model learns enhanced discriminative representation based on the virtual data and serves as the guidance for a student model to imitate. In the imitating learning step, through the scale-fusing distillation module, the student network is encouraged to generate features that mimic the teacher model both on the same scale and cross scales. Furthermore, the coupled adversarial module builds the bridge to connect not only teacher and student models but also virtual and real data, adjusting the imitating learning process dynamically. Note that our method introduces no extra time cost during inference and can be plug-and-play in various cutting-edge lane detection networks. Experimental results prove the effectiveness of the RIL framework both on CULane and TuSimple for four modern lane detection methods. The code and model will be available soon.
|
['Errui Ding', 'Xiao Tan', 'Wei zhang', 'Zhikang Zou', 'Liang Du', 'Xiaoqing Ye', 'Minyue Jiang', 'Yue He']
|
2022-10-11
| null | null | null | null |
['lane-detection']
|
['computer-vision']
|
[-1.50379553e-01 2.78137714e-01 -1.46447822e-01 -4.19817120e-01
-6.29577339e-01 -7.81490266e-01 4.10336196e-01 -5.71973801e-01
-2.11559221e-01 6.18307471e-01 -2.79797614e-01 -4.40987676e-01
1.87673524e-01 -7.46362984e-01 -9.54711556e-01 -8.31923544e-01
4.10208479e-02 1.18909925e-01 5.21091521e-01 -3.23557407e-01
-4.15463783e-02 8.02610755e-01 -1.43763447e+00 -2.54028767e-01
1.30609000e+00 6.71566844e-01 2.50845164e-01 5.74567020e-01
-6.56948835e-02 6.44747674e-01 -3.68324906e-01 -2.35229194e-01
6.02927744e-01 -1.23103961e-01 9.09987018e-02 1.58834830e-01
9.70045090e-01 -4.64279175e-01 -8.59901607e-01 1.14691472e+00
5.12630224e-01 2.85774380e-01 4.15925801e-01 -1.80660355e+00
-2.51765758e-01 1.97766811e-01 -7.74835050e-01 -2.40134057e-02
6.63263956e-03 5.95953703e-01 6.30267799e-01 -8.75526249e-01
4.75879759e-01 1.38004923e+00 8.05644989e-01 4.80910063e-01
-1.15342879e+00 -9.78597581e-01 4.22865897e-01 2.77297497e-01
-1.36887121e+00 -2.82477170e-01 1.23922527e+00 -2.07793266e-01
1.14157207e-01 1.15094870e-01 5.63683033e-01 1.05691028e+00
1.93975866e-01 9.82899070e-01 8.44491780e-01 -9.27746948e-03
-6.52646720e-02 5.02507150e-01 -2.49886159e-02 1.07672799e+00
6.99581429e-02 3.46463382e-01 1.49895409e-02 1.14091739e-01
7.74790525e-01 1.66289464e-01 -2.49303713e-01 -7.76107073e-01
-8.44879806e-01 7.89182365e-01 7.89498150e-01 -2.53341734e-01
7.33576017e-03 4.73396555e-02 2.48364240e-01 1.36702180e-01
1.03558518e-01 2.91981131e-01 -1.69278160e-01 2.28690431e-01
-7.72127807e-01 3.15475702e-01 5.80745816e-01 1.00594401e+00
1.23237312e+00 3.08122277e-01 5.02767600e-02 6.06056035e-01
1.71269253e-01 8.20710063e-01 1.60895482e-01 -1.01804829e+00
5.92920005e-01 5.17022789e-01 -1.06446140e-01 -1.26696396e+00
-3.59380513e-01 -6.41234696e-01 -7.67070830e-01 5.83441675e-01
4.92217958e-01 -5.19112229e-01 -1.04757833e+00 1.96400499e+00
5.29402256e-01 8.41046751e-01 1.65737085e-02 8.96331787e-01
4.48911846e-01 7.43371725e-01 -1.82790413e-01 3.47818285e-01
1.14281547e+00 -1.35340345e+00 -4.91942167e-01 -5.65787196e-01
6.28871143e-01 -6.26772225e-01 1.06107593e+00 -6.99757040e-02
-6.58310413e-01 -7.68137455e-01 -1.36336994e+00 -3.61786666e-03
-6.53663337e-01 1.51244566e-01 4.75021213e-01 5.30199945e-01
-7.65373647e-01 3.47757101e-01 -3.54501635e-01 -2.32558623e-02
3.80073249e-01 1.59081429e-01 -4.84520257e-01 -2.13887095e-01
-1.27139711e+00 7.85407364e-01 1.50007397e-01 3.67992908e-01
-9.38665152e-01 -9.65530336e-01 -1.19578314e+00 -1.56061262e-01
6.10658467e-01 -3.70057225e-01 8.20480227e-01 -9.39499378e-01
-1.39837945e+00 4.47013378e-01 1.26439318e-01 -1.76830471e-01
8.84668469e-01 -1.14434041e-01 -6.46010280e-01 -7.17645884e-02
1.97626933e-01 8.09400797e-01 1.05595124e+00 -1.63925838e+00
-8.19151342e-01 -1.49450943e-01 7.06055537e-02 5.51438704e-02
1.78362250e-01 -7.79077470e-01 -5.06201208e-01 -5.88922799e-01
-2.14368179e-02 -1.08409560e+00 -2.28146434e-01 2.60816723e-01
-6.55935287e-01 6.04066812e-02 1.56238651e+00 -4.04050708e-01
9.21460152e-01 -2.53730726e+00 -3.67422074e-01 5.16474962e-01
3.27142060e-01 3.54892492e-01 -6.44079268e-01 8.59725326e-02
-6.02634996e-02 -8.84910300e-02 -7.21135959e-02 -2.14093968e-01
1.12260506e-01 4.34065640e-01 -6.13259554e-01 7.04308391e-01
4.76532996e-01 9.68178988e-01 -1.00940800e+00 -4.03893918e-01
3.87215734e-01 3.41220319e-01 -4.60803896e-01 5.21625519e-01
1.47620976e-01 3.58078152e-01 -5.81375420e-01 4.78618711e-01
1.13564408e+00 3.15759867e-01 -1.78002775e-01 -3.33746523e-01
-2.43644282e-01 -4.73063625e-02 -1.30211687e+00 1.27305794e+00
-5.48560560e-01 8.70240748e-01 1.92356095e-01 -7.90630639e-01
1.13962758e+00 -2.83998132e-01 -5.70305437e-02 -7.99606383e-01
1.62979364e-02 9.28885862e-02 -2.81950925e-02 -4.53438193e-01
4.17515576e-01 1.67852715e-01 -3.15622091e-01 1.52633205e-01
-3.26300442e-01 -1.67010382e-01 -2.25638911e-01 1.63165107e-01
8.74612153e-01 9.94239077e-02 -4.24443275e-01 -2.41082031e-02
5.36075592e-01 -1.80426285e-01 8.70179951e-01 6.67340696e-01
-5.06830156e-01 4.93758410e-01 5.25325716e-01 -4.21342641e-01
-1.19624054e+00 -1.41917503e+00 7.58530498e-02 9.32843745e-01
6.00841284e-01 2.00189993e-01 -7.49914706e-01 -1.02910089e+00
3.95906836e-01 6.51073933e-01 -7.53318667e-01 -4.28773254e-01
-7.18653619e-01 -2.37500702e-04 8.19377840e-01 5.69520712e-01
7.18879759e-01 -5.90952575e-01 -3.61405134e-01 1.11843660e-01
2.90077865e-01 -1.15394843e+00 -9.22809243e-01 7.52394944e-02
-3.56409341e-01 -1.10758841e+00 -3.71169686e-01 -8.61336529e-01
8.62464249e-01 6.05768144e-01 6.10375166e-01 4.13646549e-02
-1.94755465e-01 -1.92501340e-02 6.86218664e-02 -3.35909188e-01
-3.79509956e-01 3.39693129e-02 -6.20025322e-02 2.12229729e-01
-6.62519410e-03 -6.73872292e-01 -6.13819599e-01 5.26624084e-01
-7.09869742e-01 2.00994775e-01 6.76711082e-01 1.01327431e+00
3.86164516e-01 -1.28532732e-02 7.00799823e-01 -8.79157364e-01
1.89488754e-01 -5.33757269e-01 -7.96580374e-01 -1.01464897e-01
-5.01988113e-01 2.61896215e-02 1.09112144e+00 -6.91641569e-01
-1.04939580e+00 1.75182864e-01 -3.83380763e-02 -8.16035986e-01
-2.02073604e-01 -2.90776640e-02 -6.27555430e-01 -2.00062022e-01
3.33102733e-01 5.80375046e-02 4.14426744e-01 -2.41590187e-01
5.74412584e-01 3.93272102e-01 9.27139044e-01 -2.41943106e-01
1.60042906e+00 5.60344815e-01 3.14498097e-02 -5.51049173e-01
-7.81507730e-01 -1.94710299e-01 -5.40900111e-01 -4.39243168e-01
5.65573633e-01 -8.82616043e-01 -5.15518308e-01 5.41773140e-01
-8.59044552e-01 -4.56298411e-01 -3.49250555e-01 2.06348136e-01
-4.13932830e-01 3.23394179e-01 -2.17215180e-01 -5.79154789e-01
9.95641053e-02 -1.25142920e+00 9.24470305e-01 7.00344443e-01
2.90957212e-01 -1.07642937e+00 -5.21695986e-02 2.24215344e-01
3.62636715e-01 5.12233675e-01 9.03037846e-01 -4.91956145e-01
-6.97956979e-01 -5.70928514e-01 -4.68361109e-01 3.29981625e-01
1.10664800e-01 1.31626902e-02 -1.12383497e+00 -3.71278197e-01
-6.24749064e-01 -1.68510959e-01 7.98697352e-01 -3.11567578e-02
1.09682846e+00 -2.28839725e-01 -3.78538877e-01 8.34044874e-01
1.16745734e+00 1.04148179e-01 6.97310805e-01 2.95073092e-01
1.10765517e+00 7.67638624e-01 6.46850288e-01 -2.40839738e-02
4.61490482e-01 5.85486293e-01 5.88708460e-01 -7.20555604e-01
-2.59529471e-01 -8.37060690e-01 4.31517780e-01 2.49490738e-01
6.55977845e-01 -6.72783852e-02 -6.07996941e-01 5.55855334e-01
-1.75611067e+00 -1.08185077e+00 3.29250470e-02 2.05544519e+00
4.87992287e-01 3.59539956e-01 1.35962054e-01 -1.23317227e-01
6.97816849e-01 4.61978614e-01 -7.75210917e-01 -3.97688687e-01
-3.98237966e-02 -4.09231693e-01 7.04643071e-01 6.42807603e-01
-1.12802696e+00 1.10833681e+00 4.86699152e+00 1.03220260e+00
-1.45134997e+00 -1.94491461e-01 5.35675049e-01 2.60160565e-01
-4.26609159e-01 9.44482982e-02 -8.36839676e-01 5.59458256e-01
6.44132495e-01 -1.53021086e-02 4.12378341e-01 1.02135968e+00
5.66508174e-01 8.03053565e-03 -8.31325769e-01 6.43809736e-01
-1.20730931e-02 -1.19052601e+00 -1.78967342e-01 -2.18950547e-02
6.04549289e-01 -1.22070769e-02 2.57340163e-01 6.59878731e-01
1.49502039e-01 -9.21053112e-01 7.14962721e-01 4.86795247e-01
7.97370970e-01 -9.77011919e-01 6.88451231e-01 5.78773737e-01
-1.34198189e+00 -1.69461802e-01 -2.14821577e-01 3.20370346e-01
1.16020039e-01 3.07148069e-01 -9.11121786e-01 3.84129375e-01
2.26435438e-01 6.65859878e-01 -5.70741057e-01 1.00993967e+00
-2.30781898e-01 4.64370131e-01 -2.60321230e-01 2.34854802e-01
6.29051208e-01 -3.67172003e-01 7.32200205e-01 1.29530215e+00
5.83729036e-02 -3.31221998e-01 6.34039044e-01 9.19258893e-01
-3.27695347e-02 -2.56239504e-01 -1.01788926e+00 5.47903299e-01
6.23559833e-01 1.36814129e+00 -2.79477596e-01 -2.03192890e-01
-4.11546588e-01 8.70563686e-01 4.15599465e-01 5.42839229e-01
-1.18241560e+00 -6.68985367e-01 9.12451744e-01 2.11565718e-01
3.16436261e-01 -8.87391195e-02 2.89237499e-02 -8.79278302e-01
-1.33798748e-01 -6.55477226e-01 -8.26056227e-02 -6.47289634e-01
-1.33781135e+00 4.09237742e-01 -8.35847631e-02 -1.48893178e+00
-1.72360301e-01 -4.24918145e-01 -1.05503297e+00 9.06920314e-01
-1.87659752e+00 -1.29713356e+00 -5.78945875e-01 5.90971589e-01
5.60760140e-01 -2.07234740e-01 1.33989930e-01 3.64209026e-01
-9.50413764e-01 1.03071117e+00 1.42552391e-01 3.48678678e-01
7.94303000e-01 -1.23753512e+00 2.88405865e-01 8.81993592e-01
-2.07742140e-01 2.90609539e-01 7.94559121e-01 -4.27150518e-01
-1.42108953e+00 -1.61590958e+00 2.83295721e-01 -2.10552633e-01
8.45262408e-01 -5.59911966e-01 -1.05588603e+00 5.58386445e-01
-1.07374609e-01 3.79773527e-01 1.67193949e-01 -5.13576150e-01
-4.81787592e-01 -4.56937671e-01 -1.17569351e+00 8.67164612e-01
8.64438832e-01 -6.02301240e-01 -2.79614329e-01 -9.22429934e-03
6.14973843e-01 -5.17389715e-01 -4.10099626e-01 3.98214340e-01
6.32840216e-01 -7.48507559e-01 8.65244985e-01 -4.90380377e-01
3.78457196e-02 -7.32699633e-01 1.42056316e-01 -1.45377386e+00
-1.51047394e-01 -6.00282729e-01 1.45559162e-01 1.32436526e+00
4.35794234e-01 -8.67637813e-01 9.69940841e-01 3.88423413e-01
-4.65632200e-01 -8.89123321e-01 -9.56420541e-01 -8.20082843e-01
1.93985980e-02 -3.69572222e-01 6.60662472e-01 8.49283695e-01
-5.36018550e-01 2.83052921e-01 -3.70288521e-01 6.82254016e-01
7.63927579e-01 1.48815304e-01 1.29449821e+00 -9.44887102e-01
-1.36206523e-01 -2.30624110e-01 -4.09385532e-01 -1.32285953e+00
5.85694075e-01 -8.07701111e-01 2.58607894e-01 -8.84277642e-01
-5.41298091e-02 -7.41999924e-01 -1.53860450e-01 4.54668462e-01
-3.75727475e-01 1.14616968e-01 3.60085964e-01 -1.08286060e-01
-4.06243861e-01 6.04603350e-01 1.47618496e+00 -3.74651968e-01
-2.28394017e-01 9.08754915e-02 -5.13328135e-01 7.50410855e-01
7.38616943e-01 -1.88081786e-01 -7.09368169e-01 -2.13747874e-01
-3.80661458e-01 -1.48842528e-01 6.87130570e-01 -9.64917779e-01
2.63301045e-01 -9.79952663e-02 3.96555752e-01 -6.81789339e-01
2.24749044e-01 -1.07415438e+00 -4.89783175e-02 2.40574926e-01
-9.59183797e-02 1.40388012e-01 4.44834560e-01 7.75434077e-01
-4.48956639e-02 -1.97781008e-02 9.80614364e-01 4.96520907e-01
-1.07850420e+00 3.75734895e-01 -1.88489184e-01 -9.11827758e-03
1.20397222e+00 -5.54657817e-01 -6.03768170e-01 -5.24604619e-01
-2.24319413e-01 7.54961431e-01 6.74803793e-01 5.93558729e-01
6.23305798e-01 -1.28645396e+00 -4.07078177e-01 6.27932191e-01
1.42149493e-01 1.65350884e-01 4.83690739e-01 6.27492249e-01
-3.54108065e-01 -6.81789443e-02 -1.97986230e-01 -4.59549010e-01
-8.38064909e-01 7.01045573e-01 6.49646938e-01 -2.09178939e-01
-6.90214753e-01 3.26524615e-01 5.12751341e-01 -6.40459239e-01
2.69396424e-01 -2.69437861e-02 -6.44661710e-02 -5.86119257e-02
3.88292372e-01 3.62283885e-01 -3.23890418e-01 -6.61788940e-01
-9.40334499e-02 6.32142365e-01 -1.47255942e-01 3.06357354e-01
9.84807372e-01 -2.37356007e-01 4.10992593e-01 3.02651286e-01
1.40614116e+00 4.33615744e-01 -2.01019812e+00 1.60842445e-02
-3.12756687e-01 -4.90107358e-01 1.08630873e-01 -5.10314047e-01
-1.28178453e+00 7.34571397e-01 9.00779366e-01 5.58955537e-04
8.56006503e-01 -2.96113014e-01 1.03320241e+00 1.86115965e-01
1.61705360e-01 -1.02421522e+00 -1.50969177e-02 4.11726534e-01
6.31024718e-01 -1.26660156e+00 -4.26760346e-01 -5.62092781e-01
-5.04776239e-01 9.03614759e-01 9.50053275e-01 -5.02050936e-01
5.02645075e-01 5.49743891e-01 4.08028811e-01 3.65770347e-02
-5.04247606e-01 -3.29850987e-02 1.87238470e-01 7.56896973e-01
-2.64230162e-01 1.18588492e-01 2.33932719e-01 2.32596189e-01
-1.22450136e-01 -5.27932286e-01 5.13376474e-01 6.97183788e-01
-5.38937449e-01 -8.39601159e-01 -4.29220468e-01 3.07497531e-01
3.06803435e-01 2.52524823e-01 6.64529344e-03 1.12114561e+00
3.85479629e-01 6.38307333e-01 2.42280975e-01 -6.15049899e-01
6.87395990e-01 -1.84927151e-01 -1.78852305e-01 -2.04365686e-01
-1.95583373e-01 -1.45727992e-01 -6.28296211e-02 -7.51360655e-01
3.75700623e-01 -3.64716262e-01 -1.41297674e+00 -4.40264642e-01
-2.25718364e-01 1.59620479e-01 4.73532677e-01 7.81288683e-01
4.99882013e-01 4.91978317e-01 1.32280600e+00 -1.11625051e+00
-6.03937268e-01 -5.88495195e-01 -6.59137011e-01 3.74573261e-01
7.43232548e-01 -8.48030150e-01 -5.69172442e-01 -3.18244040e-01]
|
[8.022945404052734, -1.4933087825775146]
|
d01f2963-47e9-44e6-af33-0cc69791f530
|
plot2api-recommending-graphic-api-from-plot
|
2104.01032
| null |
https://arxiv.org/abs/2104.01032v1
|
https://arxiv.org/pdf/2104.01032v1.pdf
|
Plot2API: Recommending Graphic API from Plot via Semantic Parsing Guided Neural Network
|
Plot-based Graphic API recommendation (Plot2API) is an unstudied but meaningful issue, which has several important applications in the context of software engineering and data visualization, such as the plotting guidance of the beginner, graphic API correlation analysis, and code conversion for plotting. Plot2API is a very challenging task, since each plot is often associated with multiple APIs and the appearances of the graphics drawn by the same API can be extremely varied due to the different settings of the parameters. Additionally, the samples of different APIs also suffer from extremely imbalanced. Considering the lack of technologies in Plot2API, we present a novel deep multi-task learning approach named Semantic Parsing Guided Neural Network (SPGNN) which translates the Plot2API issue as a multi-label image classification and an image semantic parsing tasks for the solution. In SPGNN, the recently advanced Convolutional Neural Network (CNN) named EfficientNet is employed as the backbone network for API recommendation. Meanwhile, a semantic parsing module is complemented to exploit the semantic relevant visual information in feature learning and eliminate the appearance-relevant visual information which may confuse the visual-information-based API recommendation. Moreover, the recent data augmentation technique named random erasing is also applied for alleviating the imbalance of API categories. We collect plots with the graphic APIs used to drawn them from Stack Overflow, and release three new Plot2API datasets corresponding to the graphic APIs of R and Python programming languages for evaluating the effectiveness of Plot2API techniques. Extensive experimental results not only demonstrate the superiority of our method over the recent deep learning baselines but also show the practicability of our method in the recommendation of graphic APIs.
|
['Dan Yang', 'Bei Wang', 'Xin Xia', 'Meng Yan', 'Zhongxin Liu', 'Sheng Huang', 'Zeyu Wang']
|
2021-04-02
| null | null | null | null |
['multi-label-image-classification']
|
['computer-vision']
|
[-5.89345843e-02 -3.19998860e-01 -1.65501863e-01 -4.72294539e-01
-2.40199491e-01 -5.29698193e-01 2.54304856e-01 -1.14823140e-01
1.60858154e-01 -3.67371738e-02 -8.40913802e-02 -7.77809322e-01
-1.75401866e-01 -6.63162947e-01 -7.65002251e-01 -5.17477632e-01
2.51723796e-01 5.85857481e-02 3.30111645e-02 7.05128461e-02
2.93597519e-01 5.85013211e-01 -1.67649066e+00 6.31705642e-01
9.27604735e-01 1.24891400e+00 8.61472711e-02 3.53780329e-01
-7.61945605e-01 4.37918931e-01 -9.05551672e-01 -4.82927084e-01
1.29180640e-01 -1.06086722e-02 -1.37713760e-01 -1.65537536e-01
6.77001715e-01 -1.03315450e-01 6.56927973e-02 1.18740416e+00
3.90176505e-01 -2.00517699e-01 4.05370533e-01 -1.88112652e+00
-7.24479973e-01 5.66173255e-01 -1.07236719e+00 -7.14633688e-02
-9.40283835e-02 2.01707661e-01 8.26995492e-01 -7.90759146e-01
3.81121159e-01 1.24684799e+00 6.14185631e-01 1.00844175e-01
-8.92013133e-01 -1.00525284e+00 4.42923844e-01 2.50864476e-01
-1.06928444e+00 8.13914314e-02 1.00881135e+00 -7.28335142e-01
6.20456755e-01 6.21392906e-01 6.02895498e-01 1.03818166e+00
8.65147337e-02 6.09392941e-01 7.53885925e-01 -1.05247624e-01
-2.20537577e-02 1.03968211e-01 5.40572107e-01 8.68715346e-01
1.61508814e-01 -2.37550810e-01 -3.03787559e-01 -1.00075714e-01
7.11026192e-01 6.03419185e-01 -1.74299657e-01 -4.33568716e-01
-1.01419389e+00 4.73141104e-01 8.46878290e-01 2.16128398e-02
1.48164690e-01 1.86357304e-01 5.73375762e-01 2.36946225e-01
4.02585149e-01 3.57303411e-01 -4.53746200e-01 -1.66653097e-02
-6.54687226e-01 -3.02116312e-02 5.04740059e-01 1.24670875e+00
8.93289685e-01 6.47999048e-02 -2.74762660e-01 9.25418437e-01
3.45505923e-01 5.90569556e-01 4.19530720e-01 -4.54419434e-01
8.75955522e-01 1.33199072e+00 -2.47224540e-01 -1.44043720e+00
-6.10985339e-01 -3.38593870e-01 -7.83137143e-01 6.61077976e-01
3.82443815e-01 5.36580645e-02 -8.67540359e-01 1.35413849e+00
1.69854850e-01 -2.92650849e-01 -3.50963622e-01 7.22968876e-01
1.29380560e+00 5.21157801e-01 1.62227869e-01 3.22511733e-01
1.58301210e+00 -1.53212571e+00 -5.50305486e-01 -2.51850128e-01
5.55660307e-01 -8.97157669e-01 1.84315014e+00 3.66590887e-01
-4.30347085e-01 -7.23930061e-01 -1.41104102e+00 -2.73617625e-01
-8.71709347e-01 5.67150295e-01 9.02817786e-01 5.68488896e-01
-6.12045348e-01 5.63230634e-01 -7.44152248e-01 2.63629425e-02
7.71362960e-01 1.81050375e-01 -2.52752721e-01 4.15166095e-02
-4.65490282e-01 2.18093559e-01 5.31379245e-02 2.90367007e-01
-2.28798613e-01 -8.34240854e-01 -6.52447999e-01 3.52733970e-01
3.47262055e-01 -3.88024718e-01 7.95889914e-01 -1.00093043e+00
-1.19840705e+00 6.13606215e-01 2.40333065e-01 3.65746111e-01
5.93505383e-01 -4.25355524e-01 -5.17248809e-01 -4.01162744e-01
1.65862478e-02 1.02172151e-01 8.96473825e-01 -9.10108447e-01
-4.86686230e-01 -3.76203239e-01 -2.66112760e-03 -2.97001526e-02
-4.62054849e-01 5.13763353e-02 -9.91128147e-01 -7.09610820e-01
-1.78208947e-01 -6.40467823e-01 1.46166161e-01 1.40966520e-01
-8.10270309e-01 -1.86317965e-01 1.05604720e+00 -5.42103708e-01
1.48677719e+00 -2.74880171e+00 -6.55297786e-02 3.56990516e-01
5.86190403e-01 1.30945191e-01 -2.54983038e-01 3.79674472e-02
-4.14649218e-01 3.99453402e-01 1.73002899e-01 -2.24465773e-01
9.32262391e-02 -1.68358207e-01 -3.71096790e-01 2.43677706e-01
1.68725327e-02 8.01338851e-01 -6.14683151e-01 -4.66120631e-01
2.99826264e-01 3.48031282e-01 -3.23875010e-01 3.84735614e-01
-3.70460361e-01 1.56622380e-01 -6.44084156e-01 7.50571132e-01
9.79035616e-01 -7.83516765e-01 -1.87199432e-02 -6.82554066e-01
-3.14136177e-01 1.51422806e-03 -1.01964176e+00 1.60969877e+00
-4.91930753e-01 5.18984199e-01 -4.02856171e-01 -5.72783172e-01
1.01650620e+00 -2.58561313e-01 1.82491198e-01 -9.90907371e-01
-2.52775680e-02 6.01516888e-02 -2.10324451e-01 -6.57134533e-01
3.24166626e-01 7.86162853e-01 -1.62207425e-01 4.90673333e-01
-3.41335475e-01 3.21521521e-01 4.81900424e-02 7.82028884e-02
1.02944219e+00 3.75894248e-01 -1.07683718e-01 -2.78538987e-02
2.01113045e-01 3.08472360e-03 4.24405813e-01 5.47515512e-01
2.67271310e-01 5.54268777e-01 9.35423911e-01 -8.09091568e-01
-9.69837129e-01 -8.54344487e-01 -6.48367777e-02 1.33821070e+00
1.84590071e-01 -7.20545352e-01 -5.95680237e-01 -9.57826495e-01
-3.70554030e-02 5.52139401e-01 -7.20167100e-01 -5.72617799e-02
-5.39194047e-01 -6.75692916e-01 2.57003516e-01 5.59207678e-01
4.20020372e-01 -1.07850242e+00 -6.22263074e-01 -2.13350192e-01
2.53254771e-01 -7.78209567e-01 -6.20915055e-01 1.85383528e-01
-5.36400020e-01 -1.56508434e+00 -5.24392247e-01 -5.73892176e-01
8.89702439e-01 4.57406342e-01 1.16765440e+00 4.77269679e-01
-2.26280123e-01 -8.22266042e-02 -2.53520638e-01 -3.69395643e-01
-1.13825202e-01 1.47188395e-01 -7.26425111e-01 -1.78275034e-01
1.82959855e-01 -4.97130573e-01 -7.22590983e-01 5.06754041e-01
-8.60982955e-01 7.90912747e-01 4.95481879e-01 6.41409695e-01
5.84655166e-01 -2.14782804e-01 4.05558869e-02 -1.41972458e+00
5.20050585e-01 -4.33424383e-01 -9.67432261e-01 5.61463892e-01
-6.56393766e-01 3.29078436e-02 1.01909804e+00 -3.60412657e-01
-8.31422687e-01 -1.49628580e-01 7.12630004e-02 -5.67201495e-01
-6.15219511e-02 5.28978348e-01 -5.61693788e-01 4.22072560e-02
4.11013097e-01 -2.57297933e-01 -7.02933595e-02 -8.12230825e-01
5.69356263e-01 5.74020624e-01 3.24306667e-01 -4.00196344e-01
6.97303951e-01 1.76213145e-01 -4.17169705e-02 -2.19671130e-01
-6.20014906e-01 -4.60730568e-02 -2.04127267e-01 -1.53520703e-01
8.01932991e-01 -5.98588586e-01 -9.70676839e-01 5.32953858e-01
-1.23693120e+00 -4.07815307e-01 2.86614567e-01 -1.56992510e-01
-1.44574568e-01 2.28465170e-01 -3.33560497e-01 -2.76312768e-01
-5.08675516e-01 -1.81780577e+00 1.05557382e+00 5.69444180e-01
7.76258931e-02 -6.08084440e-01 -3.89381170e-01 1.99534476e-01
3.86082530e-01 4.62106436e-01 1.58968747e+00 -7.97891736e-01
-6.43376887e-01 -1.90498561e-01 -9.91491973e-01 2.83721015e-02
1.93184689e-01 5.75999737e-01 -1.02511764e+00 5.84682859e-02
-4.67300177e-01 2.40993220e-02 4.80658889e-01 3.21599133e-02
1.83196008e+00 -1.84269339e-01 -4.92762506e-01 1.24051917e+00
1.45028305e+00 4.45786059e-01 3.53814006e-01 7.08471596e-01
1.45140219e+00 3.98816854e-01 4.65567172e-01 3.09448093e-01
4.13557470e-01 7.42580295e-01 6.06772959e-01 -5.47083199e-01
-1.83330059e-01 -2.20269352e-01 -2.72317789e-02 6.73405766e-01
1.65147021e-01 -1.83616146e-01 -8.77719760e-01 -8.69696140e-02
-2.06406283e+00 -2.93993145e-01 -4.70683604e-01 2.10493588e+00
4.29233164e-01 -2.83090882e-02 -4.67856266e-02 -1.73616350e-01
7.47280180e-01 1.55558914e-01 -7.46570766e-01 -3.79519224e-01
1.97356772e-02 1.53337792e-01 3.28771144e-01 -2.23838896e-01
-1.11502492e+00 5.76605439e-01 4.64443350e+00 9.43262517e-01
-1.47611654e+00 -1.11141056e-01 7.96223402e-01 1.46636248e-01
-2.33821318e-01 -2.83955336e-01 -6.58498466e-01 8.99402976e-01
5.71293712e-01 1.79061927e-02 4.84397858e-01 1.32006669e+00
1.10783599e-01 1.05728805e-01 -1.22525263e+00 1.37929273e+00
-1.78710803e-01 -1.37913823e+00 1.73731402e-01 -7.32592493e-02
2.71478981e-01 -5.82663827e-02 4.34887469e-01 2.46430755e-01
2.06729211e-02 -9.52614844e-01 7.88708866e-01 4.26576555e-01
9.92989600e-01 -7.93999135e-01 5.35372734e-01 -3.48083675e-01
-1.24112165e+00 3.49178873e-02 -3.46371770e-01 4.08038288e-01
-2.56945610e-01 5.18631101e-01 -4.81617153e-01 6.03732586e-01
1.03765059e+00 8.66183221e-01 -1.16805816e+00 1.15516651e+00
-2.09938183e-01 2.75883198e-01 -1.21894725e-01 -1.17792763e-01
-1.19825630e-02 -2.39198759e-01 1.50244936e-01 1.11880183e+00
3.97656471e-01 -6.90398335e-01 -9.88516025e-03 1.10265064e+00
-1.60545796e-01 4.06254232e-01 -4.89287794e-01 -1.09821692e-01
3.63676190e-01 1.51244402e+00 -8.57230127e-01 -2.25191548e-01
-8.02214384e-01 8.22107375e-01 4.80518490e-01 4.70227718e-01
-1.06955123e+00 -6.89448118e-01 5.40587842e-01 -1.44869819e-01
3.33024472e-01 4.29413952e-02 -7.06150591e-01 -9.75485861e-01
2.42583886e-01 -9.63340580e-01 4.02141005e-01 -9.11635518e-01
-1.21497297e+00 9.16621864e-01 -1.43787667e-01 -1.47156250e+00
2.87922084e-01 -8.19519401e-01 -8.12076330e-01 9.53495502e-01
-1.25911689e+00 -1.29928637e+00 -8.74287844e-01 4.35078442e-01
3.55106533e-01 -3.85485858e-01 6.32708907e-01 6.21417761e-01
-1.07801342e+00 8.81218195e-01 1.77886248e-01 1.86971799e-01
8.21404457e-01 -1.41744399e+00 6.01337492e-01 6.87214375e-01
1.17026217e-01 7.46062040e-01 2.84059584e-01 -5.17044306e-01
-1.59416449e+00 -1.19049144e+00 7.84965083e-02 -3.03898126e-01
8.25120747e-01 -6.06900513e-01 -9.93933558e-01 5.49154460e-01
8.24734569e-03 1.39945492e-01 9.20116484e-01 3.63290578e-01
-7.52571762e-01 -2.10605547e-01 -6.01515114e-01 8.92960250e-01
9.12626088e-01 -2.29538321e-01 2.77806800e-02 4.36055422e-01
8.45463574e-01 -6.15496159e-01 -6.25680387e-01 1.00507438e-01
6.48410916e-01 -1.12226307e+00 9.96742964e-01 -5.02648175e-01
6.97675288e-01 -4.82432485e-01 8.92349929e-02 -1.09041214e+00
-1.72581941e-01 -3.82891268e-01 1.49910226e-01 1.42223763e+00
4.57253456e-01 -4.26092684e-01 6.36806965e-01 6.48060381e-01
-4.58172411e-01 -7.58323669e-01 -5.12196302e-01 -3.94217938e-01
-4.26213890e-01 -5.08117378e-01 1.03422904e+00 9.04701054e-01
-2.34906361e-01 3.31538826e-01 -2.26665080e-01 5.45098335e-02
3.12435746e-01 6.37056708e-01 1.15968847e+00 -1.21770012e+00
-3.74025971e-01 -8.62383902e-01 -3.19662094e-01 -7.96537936e-01
-1.22119926e-01 -9.19565737e-01 -2.68499851e-01 -1.33526993e+00
1.90161526e-01 -7.71974623e-01 -3.90751302e-01 7.56137192e-01
-4.11733359e-01 3.43292207e-02 4.47988093e-01 2.77337730e-01
-6.42052054e-01 1.60160482e-01 1.22479749e+00 -3.52091819e-01
-3.01418036e-01 1.36364087e-01 -7.38351822e-01 7.01570630e-01
3.54965061e-01 -3.58120680e-01 -5.09699762e-01 -7.41533697e-01
6.37843430e-01 -3.35984409e-01 4.13347721e-01 -8.53233874e-01
1.47925988e-01 2.38393955e-02 4.93510097e-01 -8.04202616e-01
-3.00874680e-01 -1.14238524e+00 3.50020915e-01 -3.38544808e-02
1.55088678e-03 5.20728528e-01 5.68521142e-01 4.29112166e-01
1.30997702e-01 -7.73812160e-02 4.50967163e-01 1.19282633e-01
-6.41170204e-01 3.29043955e-01 2.62499154e-01 -1.60012186e-01
7.38693714e-01 -1.71668336e-01 -9.97930586e-01 2.29644641e-01
-4.12011027e-01 1.15700305e-01 5.74770331e-01 6.38557196e-01
4.60247844e-01 -1.30083120e+00 2.96581574e-02 3.65861118e-01
5.57587206e-01 2.67410219e-01 3.56936842e-01 6.82900310e-01
-7.56711543e-01 -4.89310250e-02 -4.38869506e-01 -5.47930479e-01
-1.31899786e+00 7.78604269e-01 6.93337545e-02 -2.14015052e-01
-1.00213003e+00 5.40959179e-01 6.79864645e-01 -1.51219904e-01
4.58813250e-01 -5.57313144e-01 -3.50447863e-01 3.96902449e-02
7.93042958e-01 3.00141543e-01 2.87387282e-01 -1.33970112e-01
-2.71766335e-01 5.54882824e-01 -2.88113207e-01 6.52532458e-01
1.49786663e+00 1.81435049e-01 -1.91942021e-01 5.28116584e-01
1.22472978e+00 1.83909893e-01 -1.26078176e+00 1.51824623e-01
-4.40086536e-02 -4.34208512e-01 -2.02703968e-01 -1.00380206e+00
-1.52339280e+00 1.13816881e+00 7.66587496e-01 4.09152061e-01
1.03811073e+00 -2.03225985e-01 3.66381645e-01 8.45317766e-02
7.97410831e-02 -6.86523139e-01 2.40601543e-02 4.85059805e-02
1.04166222e+00 -1.24059200e+00 6.29800856e-02 -5.19850254e-01
-3.73514086e-01 1.64462078e+00 9.81198490e-01 1.30571455e-01
5.11755228e-01 3.86246681e-01 2.94171542e-01 -5.84929526e-01
-2.36024439e-01 3.64041388e-01 4.74600285e-01 4.93325830e-01
5.67306399e-01 -8.96624662e-03 2.19649430e-02 1.02182245e+00
-1.56918779e-01 -2.76327521e-01 1.71015009e-01 6.47832930e-01
1.91266105e-01 -9.40773129e-01 -1.92945942e-01 7.31386065e-01
-1.97302595e-01 -2.95448542e-01 -1.86177805e-01 9.49601173e-01
1.89947665e-01 4.58028078e-01 1.98659956e-01 -5.90558648e-01
5.08342743e-01 -1.19417004e-01 -5.94716519e-02 -4.41360891e-01
-7.60545135e-01 1.48286358e-01 -2.34833077e-01 -7.61232972e-01
-2.74890736e-02 -1.49382249e-01 -1.18623316e+00 -2.51643181e-01
-3.70103270e-02 -2.05614269e-01 8.55888069e-01 7.17299759e-01
6.81630015e-01 1.14437735e+00 5.00522017e-01 -8.03407192e-01
5.72514627e-03 -7.85966456e-01 -3.09841275e-01 5.18200636e-01
1.49616271e-01 -8.93621325e-01 -2.92264789e-01 -1.05611503e-01]
|
[11.321520805358887, 2.192783832550049]
|
7ec11269-83ac-4f38-9601-ad89370b05b7
|
predicting-clinical-outcome-of-stroke
|
1907.10419
| null |
https://arxiv.org/abs/1907.10419v3
|
https://arxiv.org/pdf/1907.10419v3.pdf
|
Predicting Clinical Outcome of Stroke Patients with Tractographic Feature
|
The volume of stroke lesion is the gold standard for predicting the clinical outcome of stroke patients. However, the presence of stroke lesion may cause neural disruptions to other brain regions, and these potentially damaged regions may affect the clinical outcome of stroke patients. In this paper, we introduce the tractographic feature to capture these potentially damaged regions and predict the modified Rankin Scale (mRS), which is a widely used outcome measure in stroke clinical trials. The tractographic feature is built from the stroke lesion and average connectome information from a group of normal subjects. The tractographic feature takes into account different functional regions that may be affected by the stroke, thus complementing the commonly used stroke volume features. The proposed tractographic feature is tested on a public stroke benchmark Ischemic Stroke Lesion Segmentation 2017 and achieves higher accuracy than the stroke volume and the state-of-the-art feature on predicting the mRS grades of stroke patients. In addition, the tractographic feature also yields a lower average absolute error than the commonly used stroke volume feature.
|
['Po-Yu Kao', 'Jefferson W. Chen', 'B. S. Manjunath']
|
2019-07-22
| null | null | null | null |
['ischemic-stroke-lesion-segmentation']
|
['medical']
|
[-2.54709363e-01 -4.92669344e-01 -4.64473099e-01 -2.02406064e-01
-5.32515526e-01 -6.19656086e-01 4.56886798e-01 3.07344168e-01
-7.18626618e-01 7.90810466e-01 8.96927476e-01 -3.53355855e-01
-3.42267036e-01 -1.00174391e+00 -2.48670563e-01 -6.33577526e-01
-3.20714235e-01 5.44168949e-01 4.19017851e-01 8.87848362e-02
4.14978057e-01 1.05723894e+00 -8.23884189e-01 3.14809680e-01
1.10594130e+00 8.68642151e-01 4.48129743e-01 1.27909780e-01
-1.82081938e-01 2.80172318e-01 -4.12341118e-01 2.13467911e-01
1.16771139e-01 -6.51472151e-01 -8.13339174e-01 -5.38103700e-01
-1.24264784e-01 -5.15694320e-01 -6.51596129e-01 8.64526987e-01
8.29717815e-01 5.35603911e-02 1.13356316e+00 -8.29785109e-01
-2.92774867e-02 4.49906886e-01 -3.68802339e-01 1.01772094e+00
1.60888303e-02 1.97596833e-01 6.04651749e-01 -8.27099562e-01
8.05567205e-01 9.18843091e-01 3.29432070e-01 -3.37523445e-02
-7.99661338e-01 -6.45966470e-01 -1.27954975e-01 8.81161451e-01
-1.10075676e+00 3.17510888e-02 4.74749804e-01 -8.30576777e-01
7.31212497e-01 1.18426755e-01 1.08541417e+00 8.36206019e-01
7.39449441e-01 5.10219574e-01 1.19972289e+00 2.45690703e-01
1.18986890e-01 -6.24675333e-01 4.50947106e-01 4.61651772e-01
2.52349734e-01 1.17523231e-01 -9.31769088e-02 -3.35278273e-01
6.65595651e-01 4.82023329e-01 -7.24759758e-01 -3.48162025e-01
-1.43706834e+00 7.85956800e-01 1.07972479e+00 5.49508989e-01
-6.79708004e-01 -2.26185560e-01 7.52585471e-01 1.51794329e-01
1.56158417e-01 1.15653418e-01 -3.18856359e-01 -2.11132661e-01
-1.06310737e+00 -3.82801183e-02 2.96495795e-01 1.41331116e-02
2.88446009e-01 -4.45750654e-02 -6.72349274e-01 8.19519639e-01
1.57361671e-01 3.40052545e-01 1.10554564e+00 -4.84582186e-01
5.64081609e-01 1.03527141e+00 -2.40574583e-01 -5.70926905e-01
-9.13122058e-01 -7.06186116e-01 -1.00253487e+00 4.78393108e-01
6.72824025e-01 -2.28818282e-01 -7.67528653e-01 1.39663935e+00
-9.91960755e-04 5.40691577e-02 -2.34488249e-01 1.26303458e+00
8.63922358e-01 1.57182857e-01 1.28954440e-01 -1.98999465e-01
1.04272139e+00 -6.93397403e-01 -2.93197602e-01 -1.80795074e-01
1.03601873e+00 -3.54451030e-01 9.54005241e-01 -7.18086958e-02
-8.54628801e-01 1.11912027e-01 -7.90943861e-01 1.57242760e-01
-3.11143368e-01 2.19576016e-01 3.44951928e-01 3.91081661e-01
-7.93247581e-01 9.04157758e-01 -8.68909240e-01 -6.79949582e-01
8.67991686e-01 1.22825444e-01 -4.77248400e-01 -8.33251253e-02
-1.07875359e+00 1.65452945e+00 2.57770211e-01 -1.89323753e-01
-6.59872830e-01 -9.14931059e-01 -3.48130196e-01 2.39329696e-01
-5.23415357e-02 -6.03714883e-01 3.94891798e-01 -2.90327609e-01
-1.27174187e+00 6.36334002e-01 -3.89411867e-01 -3.36945653e-01
9.57415104e-01 -1.64077133e-01 -1.56967595e-01 3.67502540e-01
8.85002762e-02 1.00462645e-01 1.62335336e-01 -5.46336353e-01
-1.17919229e-01 -7.97398031e-01 -3.53873104e-01 5.72762251e-01
-1.45088062e-01 1.39729664e-01 3.72091413e-01 -7.15241015e-01
1.75675288e-01 -6.39176309e-01 -3.01255167e-01 2.08705395e-01
-2.12467164e-01 -1.22278951e-01 6.35544419e-01 -8.79872143e-01
8.41100395e-01 -2.02569461e+00 2.50651509e-01 4.23620641e-01
5.11240125e-01 2.43603230e-01 -9.66344476e-02 2.72892952e-01
-1.02532156e-01 2.90301859e-01 -3.00239861e-01 6.89756691e-01
-5.59661269e-01 -1.55379474e-01 -8.26225728e-02 7.51604736e-01
-1.97291777e-01 1.10496175e+00 -7.77069211e-01 -3.06306571e-01
2.42024750e-01 3.25703740e-01 -3.24706733e-01 5.25649525e-02
6.06768489e-01 7.36754775e-01 -7.21265197e-01 -4.60846089e-02
3.34064543e-01 1.56568125e-01 -1.61062226e-01 -5.05112074e-02
-1.97113249e-02 2.37921074e-01 -5.28253734e-01 1.56017172e+00
1.56665280e-01 6.08826339e-01 -3.19487602e-01 -1.06927836e+00
9.43487048e-01 2.45151699e-01 8.48949313e-01 -5.28256238e-01
5.75980961e-01 2.88161010e-01 7.40120649e-01 -4.88977522e-01
-6.30022228e-01 5.08170668e-03 3.01900446e-01 6.55151784e-01
-3.72190505e-01 1.05456874e-01 2.27022111e-01 2.55204141e-01
1.31500471e+00 -3.05752695e-01 5.72459102e-01 -5.30388772e-01
5.59269488e-01 3.78985405e-02 5.91018260e-01 6.71322346e-01
-7.09534585e-01 5.20669639e-01 9.00194466e-01 -1.21306367e-01
-7.35078394e-01 -1.42746592e+00 -3.85106087e-01 5.32629311e-01
2.04368874e-01 2.24988293e-02 -7.00425804e-01 -5.49061894e-01
5.08757196e-02 2.83971995e-01 -5.55396914e-01 -5.16549766e-01
-6.67241931e-01 -9.55482423e-01 5.70130289e-01 8.31341445e-01
6.82630002e-01 -9.28705156e-01 -6.80481255e-01 1.70789003e-01
-3.96677136e-01 -5.67729712e-01 -6.60068095e-01 -2.50983953e-01
-1.23000145e+00 -1.56913614e+00 -1.34461868e+00 -8.17209482e-01
4.61637020e-01 2.97152966e-01 3.67617846e-01 3.11685055e-01
-4.53892171e-01 -1.74235091e-01 -4.73891795e-01 -1.52382120e-01
-1.66793689e-01 1.29565388e-01 -1.48435533e-01 2.97685601e-02
3.54795605e-02 -8.25895488e-01 -1.33122659e+00 2.21414626e-01
-6.01992905e-01 1.74815692e-02 8.45840335e-01 5.68427742e-01
5.25769949e-01 -4.95765030e-01 1.12312937e+00 -5.11875570e-01
6.05325818e-01 -7.77654707e-01 7.83745125e-02 1.38353780e-01
-8.05473268e-01 -1.59136742e-01 4.78131920e-01 -4.20196503e-01
-6.55420721e-01 -1.01777196e-01 1.67602167e-01 1.47158459e-01
-1.18147597e-01 8.29794109e-01 -3.71539295e-01 -5.27813146e-03
7.37801015e-01 1.82215869e-01 2.51836836e-01 -5.75693786e-01
2.45133892e-01 5.07659256e-01 4.64717239e-01 -4.58801500e-02
3.16592991e-01 3.59570205e-01 3.46137822e-01 -6.14422798e-01
-3.06616992e-01 -7.80403793e-01 -1.05286312e+00 -2.58807778e-01
7.75532246e-01 -5.02751648e-01 -1.98753372e-01 4.04561371e-01
-1.06844425e+00 -4.95336652e-01 -7.32170567e-02 9.76440132e-01
-4.73906875e-01 4.95050460e-01 -5.08223295e-01 1.10050358e-01
-5.60189426e-01 -1.37997997e+00 4.19290066e-01 9.60460156e-02
-1.10959508e-01 -6.65399730e-01 1.23055547e-01 4.01535025e-03
5.96716821e-01 6.11359954e-01 1.55954874e+00 -6.33434832e-01
-6.94549531e-02 -3.09785903e-01 -6.99751496e-01 -1.37916967e-01
1.10799089e-01 -4.33014959e-01 -1.47554293e-01 -2.57636845e-01
-2.15262353e-01 2.73507118e-01 1.07212520e+00 7.08217502e-01
8.57849717e-01 2.18083039e-01 -5.11998296e-01 4.26773816e-01
1.25168884e+00 1.46771997e-01 7.24880934e-01 2.83782452e-01
5.60850322e-01 3.36346000e-01 7.06964433e-02 4.39325534e-02
1.85396299e-01 6.27552569e-01 2.50130594e-01 -8.34933817e-02
-5.81799507e-01 3.36510301e-01 2.01137692e-01 4.52735603e-01
-3.57270092e-01 1.21996649e-01 -1.14665031e+00 4.78539020e-01
-1.82801497e+00 -1.11931396e+00 -7.08522916e-01 2.18631268e+00
6.13651097e-01 -2.12501451e-01 4.11462814e-01 9.58279520e-02
8.61698329e-01 -4.56716754e-02 -7.94480622e-01 -1.95546359e-01
-2.01934338e-01 1.79934189e-01 5.91292620e-01 2.80612022e-01
-6.60017729e-01 6.96821213e-01 6.41234636e+00 3.19616884e-01
-1.29256082e+00 6.49489462e-01 3.12366962e-01 -4.02094364e-01
2.24633545e-01 -4.20728810e-02 -1.82190016e-02 3.97529244e-01
7.42705107e-01 -8.15987468e-01 4.17313337e-01 4.89226073e-01
8.05647075e-01 -1.76730141e-01 -7.13527501e-01 5.63194692e-01
-1.82415485e-01 -1.13686192e+00 1.04345717e-02 8.82012025e-02
4.86060739e-01 6.57487154e-01 -3.94518793e-01 -9.62232519e-03
-1.58822104e-01 -8.55099797e-01 5.52187383e-01 8.71902227e-01
8.51317823e-01 -6.11921966e-01 9.57593143e-01 3.60687196e-01
-1.04901540e+00 -1.01826087e-01 -1.54715374e-01 -3.05629931e-02
4.67485309e-01 7.39643276e-01 -4.54990953e-01 1.86532453e-01
7.22170174e-01 8.90251696e-01 -6.60640359e-01 2.02440071e+00
-2.09257066e-01 6.36545062e-01 -2.18120590e-02 5.18127792e-02
8.51700976e-02 -2.85201788e-01 9.10139859e-01 9.44546521e-01
3.90596926e-01 2.75855601e-01 2.49159321e-01 6.81121111e-01
3.37536670e-02 6.65083051e-01 -5.81710339e-01 3.45508009e-01
2.53854722e-01 1.02951026e+00 -1.12864888e+00 -3.83079350e-01
-2.31036842e-01 6.97889805e-01 3.11723381e-01 4.99798983e-01
-3.57993335e-01 -3.26441795e-01 3.97853881e-01 2.94538975e-01
-1.90647304e-01 -1.18041083e-01 -7.76804805e-01 -1.22810733e+00
6.07071370e-02 -3.86416167e-01 2.90804088e-01 -6.04915679e-01
-1.12486374e+00 5.42720079e-01 -4.74762954e-02 -1.12789798e+00
5.20695709e-02 -4.79565561e-01 -1.33494163e+00 1.39783955e+00
-1.33768976e+00 -5.22990227e-01 -6.00015581e-01 5.98555028e-01
1.88439742e-01 -3.41029018e-01 7.85718739e-01 1.95720062e-01
-6.88054621e-01 9.65002105e-02 2.48990104e-01 3.47113699e-01
5.75094402e-01 -9.83204663e-01 -9.20564383e-02 7.65886009e-01
-6.45608246e-01 4.29350913e-01 2.16373950e-01 -1.00570881e+00
-6.45974517e-01 -1.15269482e+00 5.24351120e-01 -1.98287264e-01
8.12082767e-01 2.32133105e-01 -9.33936119e-01 3.08751225e-01
-2.50447124e-01 -5.32815941e-02 5.89763701e-01 -3.44248682e-01
-1.96445912e-01 1.51115581e-01 -1.24639487e+00 6.78367972e-01
1.15882385e+00 -3.83310705e-01 -9.42335904e-01 3.90814155e-01
-3.36872377e-02 8.69316459e-02 -1.05889773e+00 6.34317815e-01
8.57068241e-01 -5.87574422e-01 1.04755461e+00 -9.33719695e-01
2.07005873e-01 2.48388685e-02 3.62954706e-01 -1.61812270e+00
-6.40325904e-01 5.06690629e-02 1.29254237e-01 7.77800500e-01
2.58334607e-01 -9.82163191e-01 6.79561436e-01 6.29638255e-01
-2.68049657e-01 -9.61036623e-01 -1.10658073e+00 -9.10269260e-01
9.41370845e-01 1.59103602e-01 4.91962463e-01 5.85749090e-01
3.51431578e-01 6.44731671e-02 3.96603703e-01 -3.28304499e-01
2.44341508e-01 2.26293057e-02 1.20976701e-01 -1.47703326e+00
3.69613677e-01 -1.32375240e+00 -7.66762972e-01 -2.29980439e-01
3.61884922e-01 -1.84384692e+00 -2.84978122e-01 -2.31055403e+00
6.16080523e-01 -2.32944638e-01 -6.20573103e-01 4.83184189e-01
-2.52513945e-01 1.75797835e-01 1.68510139e-01 4.69623983e-01
3.14002037e-01 5.97865820e-01 1.40871072e+00 -1.72107145e-01
-5.61597407e-01 -2.24716452e-04 -5.02934039e-01 7.91099310e-01
1.32054007e+00 -7.01606095e-01 -2.70125151e-01 -1.38722286e-01
-4.28855747e-01 9.90939662e-02 6.30372703e-01 -9.45314288e-01
-6.54086098e-03 -2.41742238e-01 6.98579192e-01 -4.55870330e-01
-7.13308081e-02 -5.25433302e-01 -3.27427506e-01 9.03997183e-01
-3.05742443e-01 -5.37404791e-03 -1.29477665e-01 3.32449913e-01
8.18415508e-02 -2.08708063e-01 9.09132898e-01 1.77676141e-01
-3.26640815e-01 5.75253487e-01 -1.06151831e+00 1.14283986e-01
1.02041280e+00 -1.96690246e-01 -4.96129423e-01 -7.20584169e-02
-8.03055763e-01 -4.77970913e-02 1.43825248e-01 6.08453035e-01
7.07929015e-01 -1.41080773e+00 -1.00753558e+00 -2.06865981e-01
-1.62171781e-01 -7.33431339e-01 2.16403663e-01 1.58582866e+00
-7.55638778e-01 4.70726460e-01 -7.85373330e-01 -3.03604543e-01
-1.10239697e+00 1.59596547e-01 4.60728645e-01 -2.93396801e-01
-1.30806100e+00 8.66189674e-02 1.62381828e-01 2.00813681e-01
-1.94979440e-02 -3.14800858e-01 -7.56039679e-01 2.57359505e-01
5.95339596e-01 8.22921097e-01 -3.49731445e-02 -9.13952053e-01
-4.97286648e-01 5.16898155e-01 4.13753688e-02 -3.04647684e-01
1.23486114e+00 1.83499362e-02 -3.00423682e-01 1.73220396e-01
1.26470041e+00 -3.51819456e-01 -9.46789265e-01 -2.56298780e-01
-4.50990163e-02 -2.75149465e-01 5.03832042e-01 -1.14868140e+00
-1.39674032e+00 1.07228291e+00 1.14703238e+00 -2.84773380e-01
7.99871206e-01 9.82291326e-02 1.18334603e+00 1.98903352e-01
5.91732860e-01 -5.32666326e-01 -3.08515489e-01 4.00208563e-01
1.13222551e+00 -8.76261711e-01 -1.14257328e-01 -1.91428140e-01
-6.02252245e-01 1.13328767e+00 1.73827156e-01 -3.84177059e-01
1.06664658e+00 -2.76264638e-01 1.58792049e-01 -2.59975255e-01
1.71047434e-01 -2.55945235e-01 6.46550059e-01 5.53796589e-01
3.27302873e-01 4.11700010e-01 -1.24639320e+00 8.01576793e-01
-2.05847502e-01 1.12072594e-01 3.48363191e-01 5.45465887e-01
-7.18048036e-01 -9.83054519e-01 9.50818136e-03 1.41596174e+00
-4.22134340e-01 -2.05297112e-01 -6.42268956e-01 4.15191442e-01
-3.09792608e-02 8.16461742e-01 -3.06302726e-01 7.41037875e-02
4.15019810e-01 2.96742529e-01 3.90359670e-01 -6.23618841e-01
-7.22983301e-01 -2.72685438e-01 -2.63075441e-01 -6.25696301e-01
-5.08579686e-02 -1.00315404e+00 -1.85238159e+00 -3.69798578e-02
-2.36635298e-01 -8.09291378e-02 5.16557932e-01 1.35177922e+00
5.37497461e-01 5.45660615e-01 4.95466352e-01 -6.86212122e-01
1.07074037e-01 -1.21607494e+00 -6.41089439e-01 4.22133118e-01
8.89252424e-02 -1.02628827e+00 -1.27673566e-01 -4.04927552e-01]
|
[14.192084312438965, -2.0177011489868164]
|
557928d5-6c84-4537-9569-4747bd92b98e
|
layoutgan-generating-graphic-layouts-with
|
1901.06767
| null |
http://arxiv.org/abs/1901.06767v1
|
http://arxiv.org/pdf/1901.06767v1.pdf
|
LayoutGAN: Generating Graphic Layouts with Wireframe Discriminators
|
Layout is important for graphic design and scene generation. We propose a
novel Generative Adversarial Network, called LayoutGAN, that synthesizes
layouts by modeling geometric relations of different types of 2D elements. The
generator of LayoutGAN takes as input a set of randomly-placed 2D graphic
elements and uses self-attention modules to refine their labels and geometric
parameters jointly to produce a realistic layout. Accurate alignment is
critical for good layouts. We thus propose a novel differentiable wireframe
rendering layer that maps the generated layout to a wireframe image, upon which
a CNN-based discriminator is used to optimize the layouts in image space. We
validate the effectiveness of LayoutGAN in various experiments including MNIST
digit generation, document layout generation, clipart abstract scene generation
and tangram graphic design.
|
['Jianming Zhang', 'Aaron Hertzmann', 'Tingfa Xu', 'Jimei Yang', 'Jianan Li']
|
2019-01-21
| null | null | null | null |
['scene-generation']
|
['computer-vision']
|
[ 4.12361145e-01 2.42382854e-01 2.86149472e-01 -2.17720658e-01
-6.17002308e-01 -1.01305687e+00 7.42278755e-01 -5.20315826e-01
2.99871564e-01 4.97279346e-01 1.31134391e-01 -6.20523095e-01
2.30849072e-01 -1.23678529e+00 -1.14212573e+00 -2.76160389e-01
3.20256233e-01 4.66386229e-01 -4.06171024e-01 -2.57080466e-01
3.34559977e-01 8.32202792e-01 -8.54413688e-01 3.59772742e-01
9.44336236e-01 5.25814414e-01 2.76764512e-01 1.20938885e+00
-2.18840927e-01 6.38690293e-01 -1.36048532e+00 -6.08087540e-01
5.61567903e-01 -7.88591087e-01 -2.83882946e-01 2.78653860e-01
5.90355992e-01 -4.04847085e-01 -5.03798842e-01 1.00398707e+00
5.37728190e-01 3.68316174e-02 1.10374427e+00 -1.29371333e+00
-1.50585914e+00 7.30698526e-01 -5.78312039e-01 -4.26181614e-01
5.06954849e-01 4.83744711e-01 6.82346940e-01 -5.29639423e-01
7.54492640e-01 1.51436925e+00 5.09526908e-01 5.78904152e-01
-1.51216209e+00 -7.22460628e-01 6.69681430e-02 -5.80671608e-01
-1.34251952e+00 8.57308954e-02 1.13374400e+00 -4.61809695e-01
5.80214143e-01 5.45865655e-01 8.26122463e-01 1.35650814e+00
4.32707727e-01 5.85355103e-01 8.62911224e-01 -3.55188310e-01
1.30270913e-01 -2.04664662e-01 -5.49615562e-01 7.73571789e-01
2.21095920e-01 -4.10629362e-02 1.29554793e-01 1.95511371e-01
1.65611720e+00 -2.15251774e-01 1.01145640e-01 -1.78680271e-01
-1.01440191e+00 8.12467933e-01 7.72030532e-01 -1.58817619e-01
-1.81344539e-01 4.23257440e-01 -8.48103613e-02 1.77232862e-01
1.45372942e-01 1.18668294e+00 2.97680467e-01 2.43748575e-01
-7.28383839e-01 6.34165823e-01 6.68388546e-01 1.41259933e+00
3.41635436e-01 4.01507050e-01 -6.33898139e-01 5.34864902e-01
1.56785905e-01 9.42077756e-01 1.70403481e-01 -8.14603448e-01
7.98645616e-01 6.35912538e-01 -6.66787848e-02 -1.47084355e+00
3.76687832e-02 -1.20484211e-01 -1.26391947e+00 4.42519099e-01
-1.75852373e-01 -3.04212421e-01 -1.27896011e+00 1.27874207e+00
-1.18111074e-01 1.36880949e-01 -4.10815217e-02 7.14670300e-01
8.73906791e-01 8.34851623e-01 -1.60479158e-01 6.60190701e-01
1.09024942e+00 -1.10513973e+00 -4.74723488e-01 -1.19074523e-01
9.06291753e-02 -9.42278087e-01 1.21854627e+00 1.63028687e-01
-1.38191569e+00 -9.13732708e-01 -1.36160874e+00 -2.58208692e-01
-3.72212231e-01 4.23726559e-01 5.70214331e-01 7.68134832e-01
-9.92953897e-01 4.84028518e-01 -3.69279772e-01 2.51944602e-01
6.40929639e-01 1.38374105e-01 -1.91455651e-02 2.54824072e-01
-8.57994080e-01 4.93923187e-01 2.46318653e-01 2.04818875e-01
-9.15319741e-01 -7.00530052e-01 -9.70024586e-01 7.96286836e-02
-1.64977863e-01 -1.17345893e+00 1.01649261e+00 -7.13677049e-01
-1.75172973e+00 7.80051529e-01 4.67644513e-01 -3.30224514e-01
8.31248760e-01 4.12640162e-02 -2.45974541e-01 -2.21384004e-01
-2.98453532e-02 9.70585048e-01 1.11334574e+00 -1.53063583e+00
-1.64189488e-01 3.01189423e-01 3.40689272e-01 3.12289238e-01
3.09313595e-01 -5.16230941e-01 -5.48323691e-01 -1.11785150e+00
-1.39321849e-01 -8.32752883e-01 -4.52416271e-01 -2.99284458e-01
-1.49882245e+00 4.42773432e-01 8.89363110e-01 -4.94382918e-01
1.03076625e+00 -1.76322591e+00 2.33669281e-01 5.10814071e-01
1.20336056e-01 8.83005112e-02 -4.94996369e-01 4.93658870e-01
-2.83404052e-01 6.57812119e-01 1.19842682e-02 -3.40380579e-01
2.12574095e-01 -5.25069349e-02 -6.21669531e-01 -4.56193164e-02
6.14917815e-01 1.60737479e+00 -7.68205523e-01 -4.14419584e-02
5.74891210e-01 4.64609772e-01 -7.35349059e-01 6.70879185e-01
-7.40799725e-01 5.90352356e-01 -5.92918098e-01 2.38541603e-01
8.91839206e-01 -1.41332984e-01 1.25426352e-01 -3.22213709e-01
1.22226715e-01 7.72169828e-02 -9.40211892e-01 1.76117194e+00
-6.17729425e-01 7.58566618e-01 -6.98696852e-01 -4.31441456e-01
1.25659311e+00 -2.87868202e-01 -1.06910944e-01 -7.53195643e-01
3.32566142e-01 -3.07258695e-01 -3.51307660e-01 -2.68252790e-01
7.47635603e-01 3.34546328e-01 -6.70431614e-01 5.15861869e-01
-5.82530677e-01 -9.42306519e-01 -1.18667245e-01 2.83450991e-01
1.06739986e+00 2.51651406e-01 -1.19886659e-01 -5.19919675e-03
1.71138793e-01 -1.69498369e-01 -2.42221162e-01 8.13655555e-01
8.86484444e-01 1.14505732e+00 9.81783509e-01 -4.42819774e-01
-1.78107572e+00 -1.26890445e+00 4.91009951e-01 3.43458861e-01
1.44016713e-01 -3.43079746e-01 -1.25516319e+00 -5.73407054e-01
-1.63682342e-01 8.92067313e-01 -8.99163902e-01 -3.08212876e-01
-9.42123115e-01 -2.05395505e-01 7.09192157e-01 5.48086226e-01
5.06416202e-01 -1.26489472e+00 -5.81719756e-01 -9.25766602e-02
2.67560303e-01 -8.33208680e-01 -9.56697285e-01 -1.23481058e-01
-3.87551397e-01 -9.30375934e-01 -8.96250248e-01 -9.72879052e-01
1.39644432e+00 -1.07167125e-01 1.24107063e+00 7.64154643e-02
-3.47402185e-01 -1.32568866e-01 -9.53087807e-02 -3.22354138e-01
-9.08865750e-01 3.97832394e-01 -5.75472474e-01 -2.49188736e-01
-6.84276700e-01 -5.50455272e-01 -8.04997742e-01 6.40932843e-02
-1.07800603e+00 8.75037193e-01 5.97332954e-01 7.56229520e-01
6.38242602e-01 2.01586574e-01 -9.82147409e-04 -1.40958524e+00
1.10547996e+00 -2.11281419e-01 -8.16973329e-01 2.29333207e-01
-1.88714974e-02 3.39783043e-01 1.08435738e+00 -5.75980008e-01
-7.80053198e-01 -1.18678428e-01 -9.58956778e-02 -6.80130482e-01
-1.53458685e-01 1.63767323e-01 -6.10610664e-01 1.16428278e-01
5.54126620e-01 -1.19381701e-03 -3.60597074e-01 -2.95340717e-01
9.32134986e-01 2.41265386e-01 9.36937630e-01 -7.26081014e-01
1.40022779e+00 -3.57746817e-02 1.29447669e-01 -4.76227283e-01
-3.69328201e-01 8.16417336e-01 -6.69611573e-01 1.30037025e-01
1.04406095e+00 -5.16658902e-01 -6.12766862e-01 5.66427529e-01
-1.60347939e+00 -8.35482419e-01 -4.30240005e-01 -4.93337154e-01
-6.74484432e-01 -1.94576263e-01 -2.93852866e-01 -4.23329175e-01
-3.73054504e-01 -1.40991652e+00 1.38050652e+00 3.80387306e-01
-2.95403183e-01 -8.22156429e-01 -1.93866491e-02 -2.89189756e-01
1.76262587e-01 1.11997700e+00 1.24535418e+00 3.85873616e-02
-1.22748661e+00 -3.62512052e-01 -3.37109536e-01 1.43541008e-01
3.71249199e-01 4.53076929e-01 -5.46558201e-01 7.80358464e-02
-6.81495786e-01 2.00971097e-01 4.38345909e-01 3.73896331e-01
1.51499701e+00 -5.93670249e-01 -3.15006196e-01 1.27140307e+00
1.38766265e+00 7.42620349e-01 1.07754254e+00 1.73296317e-01
1.37168860e+00 -1.10302563e-03 -2.31274627e-02 2.96876013e-01
2.69722164e-01 4.97959137e-01 3.11234713e-01 -4.39137280e-01
-5.05327523e-01 -1.02189517e+00 -1.20647147e-01 5.99855483e-01
3.00122410e-01 -1.07740951e+00 -4.71687615e-01 -5.14838696e-02
-1.29378951e+00 -8.81114006e-01 1.41682193e-01 1.75026774e+00
4.70586300e-01 3.43802899e-01 -1.24209508e-01 -1.05666853e-01
8.43960106e-01 2.15960592e-01 -6.30103588e-01 -7.28566349e-01
-1.85097039e-01 5.61530948e-01 5.44777870e-01 4.42820072e-01
-8.66899490e-01 1.18527865e+00 6.42344522e+00 7.05161452e-01
-1.02987587e+00 -5.66176057e-01 1.06948209e+00 2.76628554e-01
-9.99043465e-01 -2.81861782e-01 -4.71230894e-01 6.77705646e-01
3.26086432e-01 -7.34234303e-02 6.91883802e-01 7.04662919e-01
4.54419442e-02 4.02340025e-01 -9.91584182e-01 1.03799605e+00
9.14650559e-02 -1.81091905e+00 6.45764947e-01 -1.49720442e-02
1.20336866e+00 -9.34842467e-01 7.40589857e-01 4.23150882e-02
8.40051055e-01 -1.60568619e+00 1.04012561e+00 7.18333483e-01
1.47181344e+00 -1.09041929e+00 3.58136110e-02 -1.57126579e-02
-9.40300643e-01 3.00303429e-01 -2.95993745e-01 3.07881236e-01
3.27168077e-01 1.29372790e-01 -1.01962113e+00 3.99757773e-01
-6.48104921e-02 3.15826774e-01 -5.96423805e-01 7.85387814e-01
-7.33195186e-01 3.20924073e-02 5.49579151e-02 -1.45625055e-01
4.30010498e-01 -4.15432930e-01 3.25380266e-01 1.07305408e+00
6.36095524e-01 -3.29271778e-02 -1.84921958e-02 1.57624149e+00
-4.86664861e-01 -2.51368105e-01 -8.72522950e-01 -3.68011326e-01
6.83564305e-01 1.14922237e+00 -9.77818370e-01 -3.18413407e-01
3.47823769e-01 1.42829227e+00 1.09001085e-01 4.10580009e-01
-1.28517842e+00 -8.01049054e-01 5.25722027e-01 4.33825850e-02
2.96594381e-01 -4.19194937e-01 -6.96170926e-01 -7.96433985e-01
-3.11780334e-01 -9.86466706e-01 -2.66321480e-01 -1.25524390e+00
-9.40765202e-01 1.08607459e+00 -2.02584222e-01 -1.15334165e+00
-2.01588541e-01 -4.97345537e-01 -1.06054282e+00 1.20823026e+00
-7.65825868e-01 -1.40945184e+00 -4.46242422e-01 7.50654787e-02
5.66791892e-01 -3.65116179e-01 6.90852880e-01 1.00452779e-02
-5.82819581e-01 9.80449855e-01 -2.26923078e-01 4.28102732e-01
2.82218844e-01 -1.47955871e+00 1.84485841e+00 7.64307678e-01
3.11591744e-01 6.21716201e-01 4.96548086e-01 -1.03321970e+00
-1.44673932e+00 -1.39586413e+00 2.44820505e-01 -6.09060585e-01
1.42471984e-01 -8.27216148e-01 -3.39303315e-01 6.80167973e-01
3.98351133e-01 -5.45019269e-01 2.21990794e-01 -6.39713168e-01
-3.49499702e-01 3.97113562e-01 -1.13214529e+00 1.21835232e+00
1.33612227e+00 -3.71089965e-01 -1.77080393e-01 2.59646535e-01
1.14006782e+00 -1.24466455e+00 -4.37297732e-01 2.76815947e-02
6.39657140e-01 -8.10725808e-01 1.32749724e+00 -5.78960896e-01
1.12317348e+00 -3.37247789e-01 1.53273374e-01 -1.65443170e+00
-4.13700283e-01 -9.91321206e-01 2.02876404e-01 1.11400068e+00
2.83116311e-01 -2.14204222e-01 9.31444168e-01 3.93141657e-01
-3.42914343e-01 -4.95122343e-01 -4.13975269e-02 -4.28547472e-01
-8.97868425e-02 -2.68912166e-01 1.52896500e+00 5.71400702e-01
-7.95083642e-01 4.04904693e-01 -5.24430633e-01 1.73566073e-01
4.41454649e-01 2.64370859e-01 1.20084250e+00 -5.48427522e-01
-4.19246584e-01 -7.06997871e-01 -3.69066030e-01 -1.28056610e+00
1.18988648e-01 -7.83101201e-01 -5.91633618e-02 -1.58199930e+00
-3.41873735e-01 -6.89858437e-01 3.66936147e-01 -7.06158131e-02
-1.06999360e-01 5.03720164e-01 4.28570330e-01 -3.71043295e-01
-1.19942382e-01 4.04357821e-01 2.03045630e+00 -3.98377717e-01
-2.91320264e-01 1.72895826e-02 -9.89804924e-01 4.04214531e-01
5.79156518e-01 -8.00614432e-02 -6.79604769e-01 -8.11803639e-01
3.86266232e-01 2.34020323e-01 4.15783584e-01 -9.77691293e-01
-1.28631249e-01 -1.33326918e-01 9.73265827e-01 -8.56654525e-01
1.84719235e-01 -4.46183294e-01 7.51551032e-01 2.51557082e-01
-7.40159035e-01 4.94225442e-01 2.11052507e-01 1.75325349e-01
3.56800526e-01 -9.64293852e-02 5.36648273e-01 -2.03062445e-01
-2.27100492e-01 4.63941246e-01 -3.18855830e-02 -2.43759118e-02
8.20139885e-01 -2.48775527e-01 -4.75104719e-01 -4.57192034e-01
-5.14193892e-01 -2.20013097e-01 7.46001184e-01 6.46104217e-01
7.47428477e-01 -1.85991287e+00 -4.96083975e-01 6.67915702e-01
-3.36215556e-01 4.23662275e-01 3.05935800e-01 -4.18850809e-01
-1.32740331e+00 4.00768220e-01 -4.49961603e-01 -1.57005742e-01
-7.84903526e-01 7.15259194e-01 1.87709063e-01 -5.98197170e-02
-7.11550832e-01 9.35594380e-01 6.64685130e-01 -2.11844295e-01
5.83980680e-02 -7.39821315e-01 2.27357343e-01 -4.72278506e-01
5.03988564e-01 1.91295400e-01 -3.55299145e-01 -1.85683310e-01
2.17463464e-01 5.75468183e-01 1.29907489e-01 -1.92095399e-01
1.04859138e+00 2.43746474e-01 1.44058898e-01 -8.82065147e-02
1.35753953e+00 1.07155547e-01 -1.62982368e+00 5.87910235e-01
-5.08332372e-01 -5.18798470e-01 -4.15889829e-01 -7.59513617e-01
-1.17503989e+00 7.28990912e-01 2.62077540e-01 2.79161274e-01
8.58110249e-01 -4.05333698e-01 9.61149216e-01 -6.46338612e-03
1.22641802e-01 -4.55914021e-01 6.20428741e-01 3.60770881e-01
1.28519773e+00 -4.53590006e-01 -3.24175596e-01 -3.91206175e-01
-7.00555205e-01 1.05664945e+00 7.25298882e-01 -8.26645553e-01
5.64510748e-03 8.71131718e-01 9.31476057e-02 -1.58393711e-01
-3.57574224e-01 2.41527721e-01 7.15952575e-01 8.69162738e-01
1.63849562e-01 2.27837443e-01 3.37075859e-01 4.85806406e-01
-9.50711250e-01 -4.29783821e-01 3.46342266e-01 4.16729748e-01
1.60511211e-01 -1.32751405e+00 -3.48031342e-01 2.39493534e-01
5.92844933e-03 -1.83321685e-01 -5.34790993e-01 6.92175090e-01
2.69814283e-01 2.49992579e-01 4.44299698e-01 -7.05272377e-01
6.20330215e-01 -4.23271507e-01 7.15340078e-01 -6.32337749e-01
-5.53860188e-01 -3.70580936e-03 -1.59595698e-01 -3.02915305e-01
3.48206371e-01 -1.79268196e-01 -8.65747750e-01 -4.08953846e-01
2.31082395e-01 -1.82674810e-01 6.09295368e-01 2.96726644e-01
4.56569612e-01 1.22908437e+00 9.34577763e-01 -9.70930696e-01
-8.28462467e-02 -4.91853803e-01 -2.73593336e-01 4.23883110e-01
7.02319741e-02 -2.17326730e-01 2.37111092e-01 3.07732970e-01]
|
[11.609295845031738, -0.4226371943950653]
|
03bcfa35-4013-400b-913c-69a96e84cb45
|
normality-guided-distributional-reinforcement
|
2208.13125
| null |
https://arxiv.org/abs/2208.13125v2
|
https://arxiv.org/pdf/2208.13125v2.pdf
|
Normality-Guided Distributional Reinforcement Learning for Continuous Control
|
Learning a predictive model of the mean return, or value function, plays a critical role in many reinforcement learning algorithms. Distributional reinforcement learning (DRL) methods instead model the value distribution, which has been shown to improve performance in many settings. In this paper, we model the value distribution as approximately normal using the Markov Chain central limit theorem. We analytically compute quantile bars to provide a new DRL target that is informed by the decrease in standard deviation that occurs over the course of an episode. In addition, we propose a policy update strategy based on uncertainty as measured by structural characteristics of the value distribution not present in the standard value function. The approach we outline is compatible with many DRL structures. We use two representative on-policy algorithms, PPO and TRPO, as testbeds and show that our methods produce performance improvements in continuous control tasks.
|
['Andrew Perrault', 'Ju-Seung Byun']
|
2022-08-28
| null | null | null | null |
['distributional-reinforcement-learning']
|
['methodology']
|
[-7.54398331e-02 2.23216206e-01 -6.28599644e-01 -1.85477793e-01
-7.15379119e-01 -6.59092307e-01 6.70984626e-01 4.99875039e-01
-8.77010286e-01 1.41469169e+00 4.85651456e-02 -4.82667267e-01
-4.65337008e-01 -7.10688829e-01 -7.83663452e-01 -6.84279084e-01
-3.58965963e-01 6.30796671e-01 3.04301232e-01 -3.24491560e-02
5.30288458e-01 3.59991640e-01 -1.24464297e+00 -1.44285053e-01
7.34155357e-01 9.76696074e-01 -2.21786369e-02 6.57949567e-01
-3.15808319e-02 7.22652614e-01 -9.51389074e-01 6.79510236e-02
2.64340848e-01 -3.60089988e-01 -3.57982427e-01 -2.30674431e-01
-1.83878988e-01 -4.13977504e-01 -3.62467542e-02 1.01201272e+00
3.57235134e-01 5.03443122e-01 9.45500314e-01 -1.49039471e+00
-8.57014135e-02 9.61826384e-01 -7.03317106e-01 2.54291654e-01
1.53514907e-01 2.69432485e-01 7.13346601e-01 1.20121978e-01
2.64359921e-01 1.31679296e+00 5.48670411e-01 4.98529136e-01
-1.69840539e+00 -3.96450222e-01 3.38410616e-01 -1.80056170e-01
-7.50193238e-01 -6.93441331e-02 3.45404357e-01 -4.43050563e-01
7.29568183e-01 -3.32796663e-01 8.06390524e-01 9.74634945e-01
8.77153695e-01 6.75354064e-01 1.42696571e+00 -4.03814495e-01
9.01890159e-01 -4.68535423e-02 -2.62992471e-01 2.45870411e-01
4.82131422e-01 7.68622279e-01 -2.80001670e-01 -3.75959098e-01
8.41782749e-01 -1.96797609e-01 3.96246985e-02 -8.52873504e-01
-6.75917447e-01 1.06942117e+00 -2.78632771e-02 -1.96264356e-01
-6.56380475e-01 8.48472536e-01 4.01172727e-01 4.31785971e-01
2.97325432e-01 4.20791805e-01 -4.18956846e-01 -6.24571145e-01
-6.26384556e-01 7.41066217e-01 1.08252835e+00 7.09751129e-01
3.79633695e-01 2.26941749e-01 -6.23610020e-01 4.20252502e-01
2.33888224e-01 6.89749837e-01 3.82129461e-01 -1.52363706e+00
3.38612974e-01 -1.33508697e-01 9.59293187e-01 -2.35747069e-01
-1.94680318e-01 -3.65845650e-01 -4.26550433e-02 7.02090442e-01
8.72617066e-01 -6.18693948e-01 -6.93825483e-01 1.96067262e+00
2.69281119e-01 -1.02331690e-01 1.52236804e-01 3.84259790e-01
-5.85337818e-01 5.55541933e-01 2.51159340e-01 -5.16876996e-01
6.43754721e-01 -3.30978811e-01 -7.05388069e-01 1.37096465e-01
3.95694643e-01 -3.20423722e-01 1.06154644e+00 5.78249037e-01
-9.67198849e-01 -2.03861907e-01 -7.74805844e-01 8.82982314e-01
3.20204124e-02 -5.12391627e-01 3.51985782e-01 7.49877393e-01
-7.14717150e-01 1.10935473e+00 -1.03087628e+00 -8.88008252e-02
3.50893825e-01 1.26970485e-01 3.96401256e-01 3.16450179e-01
-8.84505987e-01 8.49886954e-01 6.04649901e-01 -5.55322409e-01
-1.32189536e+00 -7.29865253e-01 -5.70926130e-01 1.65924281e-01
7.71378279e-01 -2.35002577e-01 2.04271388e+00 -7.52520382e-01
-2.01151896e+00 -4.29874696e-02 4.54824626e-01 -1.02428913e+00
7.81262040e-01 -2.24345759e-01 1.40138626e-01 1.06665291e-01
-1.54401273e-01 4.42733049e-01 1.09721696e+00 -1.21109772e+00
-8.22352290e-01 -5.43958321e-02 -8.02332442e-03 1.08166672e-01
1.75207943e-01 -4.52408642e-01 5.27455091e-01 -4.19566840e-01
-5.78803360e-01 -8.33192945e-01 -4.91048634e-01 -4.13753062e-01
-9.21883285e-02 -3.89534533e-01 2.40469337e-01 -2.06708640e-01
1.25748646e+00 -1.74854457e+00 -3.53823870e-01 5.34437895e-01
-2.79963344e-01 3.35682333e-02 2.32466161e-02 7.66272366e-01
2.50869721e-01 2.28576157e-02 -1.47360116e-01 2.23267496e-01
5.72781503e-01 4.74179775e-01 -7.76956618e-01 5.07362187e-01
-2.33668745e-01 4.08600330e-01 -1.05082417e+00 -7.98281729e-02
8.57368112e-02 -7.94172585e-02 -6.28210962e-01 2.40612611e-01
-8.15832973e-01 2.56383151e-01 -5.68431556e-01 3.81134264e-02
2.43703261e-01 3.17271262e-01 2.97350764e-01 4.70763981e-01
-2.81379521e-01 3.68457176e-02 -9.71010864e-01 1.06159317e+00
-3.06666255e-01 1.21072821e-01 -2.29681522e-01 -1.10939467e+00
9.57628846e-01 9.53025818e-02 6.19603634e-01 -5.75567484e-01
1.98783115e-01 -1.11996792e-02 1.90364122e-01 -1.88056394e-01
3.55954319e-01 -4.28228289e-01 -1.98285699e-01 6.88545644e-01
-1.05031788e-01 -4.29072082e-01 3.85249108e-01 -2.05130111e-02
9.57482815e-01 6.37650907e-01 5.09460747e-01 -5.03939211e-01
-1.22150011e-01 2.00358382e-03 6.17169142e-01 1.24044502e+00
-5.27113199e-01 -1.16573602e-01 1.39045966e+00 -5.44399545e-02
-1.18791318e+00 -1.43361378e+00 -8.02986324e-02 1.00224197e+00
-1.23238429e-01 -2.25634426e-01 -8.32109571e-01 -5.89679360e-01
5.19753993e-01 1.27427757e+00 -7.15365291e-01 -3.08304131e-01
-1.57099351e-01 -3.55421364e-01 2.68482178e-01 5.21368861e-01
1.72909692e-01 -1.00914907e+00 -1.05338407e+00 6.27699733e-01
4.80022967e-01 -5.96569359e-01 -4.39448863e-01 4.24752653e-01
-8.40640068e-01 -1.00809884e+00 -5.30326366e-01 6.11164309e-02
1.99351028e-01 -5.47358453e-01 9.54659104e-01 -6.84499681e-01
1.00587923e-02 1.08583462e+00 -2.16013923e-01 -8.24870288e-01
-4.80276316e-01 -1.47221327e-01 2.29502782e-01 -2.56455719e-01
2.86902003e-02 -4.50540066e-01 -5.38160563e-01 3.15938368e-02
-7.53355443e-01 -5.42658567e-01 2.21002132e-01 7.69868374e-01
7.34339595e-01 1.04956448e-01 8.66085947e-01 -8.51859272e-01
1.21434629e+00 -4.76861566e-01 -1.28029954e+00 1.74691454e-01
-1.01656604e+00 5.32064319e-01 5.72358310e-01 -6.34035945e-01
-1.09389877e+00 -1.27770007e-01 2.75878876e-01 -3.95832628e-01
-1.84335634e-01 4.49562609e-01 3.44139367e-01 4.58952785e-01
5.04032373e-01 5.95398657e-02 5.64554632e-01 -3.37898225e-01
2.87355810e-01 2.95189470e-01 3.01982880e-01 -1.29075766e+00
5.67197859e-01 -1.92909467e-03 2.66939968e-01 -5.27388811e-01
-7.98234284e-01 4.31168228e-02 -1.66794345e-01 -3.40180755e-01
4.71086979e-01 -5.12149036e-01 -1.32755578e+00 9.16849002e-02
-6.18549526e-01 -1.04570687e+00 -9.52141702e-01 7.69076705e-01
-1.35277796e+00 -1.27595484e-01 -4.17466640e-01 -1.46010268e+00
1.07855266e-02 -8.03841531e-01 4.89074230e-01 5.29847324e-01
2.68843807e-02 -1.10126066e+00 5.22715271e-01 -6.31807745e-01
4.39956307e-01 4.86919045e-01 9.70242143e-01 -6.81837559e-01
-1.59457713e-01 1.49055040e-02 3.01306158e-01 3.67690444e-01
-1.66332304e-01 2.04472672e-02 -4.18718815e-01 -5.31265497e-01
-1.58612236e-01 -4.74616617e-01 6.95814192e-01 8.46407712e-01
1.14420855e+00 -2.77388901e-01 -4.79864292e-02 2.47132424e-02
1.32678807e+00 6.86392605e-01 2.98861891e-01 4.34247762e-01
-7.12739006e-02 4.70024914e-01 9.92876291e-01 1.18145919e+00
-5.21994755e-03 3.50309491e-01 4.47001249e-01 7.23947406e-01
5.94309628e-01 -6.83330357e-01 8.05286646e-01 -1.28114864e-01
1.37076572e-01 -6.66510314e-02 -8.07596266e-01 3.67458135e-01
-1.94046867e+00 -1.30547345e+00 4.37866390e-01 2.61038685e+00
9.91853297e-01 5.74996889e-01 8.22274327e-01 -3.10286283e-01
5.88466644e-01 -1.62710086e-01 -1.02067161e+00 -8.22655618e-01
5.35839796e-01 2.62371033e-01 8.80821645e-01 6.93893075e-01
-6.13647759e-01 6.32294774e-01 7.30182171e+00 8.00333798e-01
-6.61288798e-01 -2.90351272e-01 5.43670714e-01 -1.75135970e-01
-1.49166390e-01 -1.56159792e-02 -8.84603083e-01 6.17434800e-01
1.18319237e+00 -7.50156164e-01 5.77031732e-01 1.01810634e+00
5.09800375e-01 -6.56167865e-01 -1.17215526e+00 4.61269766e-01
-6.20858371e-01 -8.60885441e-01 -3.40006948e-01 3.00066978e-01
7.20373094e-01 -2.52473235e-01 3.32507074e-01 7.43983090e-01
1.03938150e+00 -8.81936014e-01 8.14475775e-01 7.43390739e-01
5.38459122e-01 -1.35147214e+00 6.23138130e-01 5.24532855e-01
-7.76158273e-01 -3.63799989e-01 -4.82410163e-01 -1.37701526e-01
-3.48071903e-02 4.60058928e-01 -8.63587856e-01 6.23423122e-02
3.79318416e-01 3.35116446e-01 -7.14532882e-02 1.29228103e+00
-3.38678837e-01 1.06037021e+00 -4.59448218e-01 -2.92708099e-01
3.66496921e-01 -4.54672039e-01 5.16703606e-01 6.31612659e-01
3.37069899e-01 -3.47899020e-01 5.77521920e-01 8.67079556e-01
2.79375553e-01 -1.21323340e-01 -6.11315429e-01 -3.40400159e-01
5.82109153e-01 6.46897435e-01 -6.15101218e-01 -1.60040036e-01
4.90424782e-02 1.58044294e-01 1.34340227e-01 4.83815759e-01
-8.69420171e-01 -5.09091854e-01 7.34484732e-01 -1.33971453e-01
4.57826585e-01 -1.87790290e-01 1.02166876e-01 -5.62451065e-01
-4.05544847e-01 -7.75478303e-01 3.96951646e-01 -2.09127247e-01
-1.36527717e+00 5.44687361e-03 6.52115345e-01 -9.98666525e-01
-8.69136274e-01 -4.12123829e-01 -5.48699200e-01 7.76859879e-01
-1.29477429e+00 -1.04107410e-01 5.47284424e-01 3.92122805e-01
3.62956196e-01 -9.73753631e-02 4.04166102e-01 -4.51773167e-01
-4.29500133e-01 2.82188267e-01 7.29190886e-01 -2.12239206e-01
5.56211412e-01 -1.70837474e+00 -9.81575847e-02 2.29585335e-01
-3.07069808e-01 3.17378640e-01 1.20494020e+00 -7.92243183e-01
-1.04418659e+00 -7.45153010e-01 -1.92370430e-01 -1.92143574e-01
9.38427389e-01 7.82786012e-02 -6.79096639e-01 7.21158564e-01
1.40388593e-01 -2.36165687e-01 3.89793843e-01 6.96004927e-02
-4.87968810e-02 -2.13176236e-01 -1.24120450e+00 4.83195007e-01
4.99365062e-01 -1.32834017e-01 -5.68501949e-01 -4.79720570e-02
6.68779314e-01 -3.34088415e-01 -8.56667340e-01 1.88983098e-01
6.62555337e-01 -9.32838023e-01 4.37067240e-01 -9.17986572e-01
1.10110871e-01 1.22087859e-02 -1.77633002e-01 -1.97281384e+00
-4.18655528e-03 -1.06149793e+00 -3.24672759e-01 1.09776425e+00
4.79333997e-02 -9.08296108e-01 6.81468546e-01 5.28919041e-01
1.28635257e-01 -8.46784294e-01 -1.01040840e+00 -1.33231831e+00
5.16984999e-01 -3.45074952e-01 5.34032226e-01 1.52964264e-01
1.32659853e-01 -6.66480139e-02 -2.01228082e-01 -1.77250028e-01
1.17984700e+00 1.72840461e-01 5.96541762e-01 -9.73018229e-01
-6.26127481e-01 -5.63706219e-01 9.89333987e-02 -9.06605363e-01
2.68444419e-01 -4.03536767e-01 2.79183954e-01 -1.11336327e+00
-3.90623659e-02 -3.77038062e-01 -4.95802850e-01 2.58848667e-01
1.87903076e-01 -5.41705012e-01 2.77156800e-01 -2.18106821e-01
-6.19025052e-01 9.20807123e-01 1.11634207e+00 2.94973105e-01
-5.34819245e-01 3.31192940e-01 -3.73201847e-01 7.47183859e-01
1.27794456e+00 -6.92815721e-01 -7.45363832e-01 3.16342205e-01
1.24788851e-01 5.34011722e-01 -1.69313904e-02 -9.53147531e-01
-1.91964701e-01 -7.69644737e-01 4.02183801e-01 -4.60839927e-01
-8.98370817e-02 -7.20553041e-01 -2.09485605e-01 8.51312876e-01
-8.81711543e-01 1.98306456e-01 1.52156368e-01 9.85062897e-01
2.62324810e-01 -3.65826339e-01 9.84715104e-01 4.92024757e-02
-2.56803393e-01 1.62393570e-01 -8.34764242e-01 4.38865960e-01
1.11392474e+00 1.96132466e-01 -1.49535939e-01 -6.53924286e-01
-6.63987637e-01 5.43404102e-01 2.55183637e-01 -2.42534410e-02
3.50937724e-01 -1.11756718e+00 -3.14782888e-01 -1.78759187e-01
-1.65765122e-01 -4.57378179e-01 -1.46028668e-01 5.82421720e-01
-1.21092223e-01 2.22685397e-01 -3.94776046e-01 -2.85102814e-01
-6.16229951e-01 6.37090564e-01 5.22499502e-01 -4.35580492e-01
-4.39150870e-01 8.12569186e-02 -2.81131864e-01 -8.54075477e-02
5.39266288e-01 -4.00134832e-01 -4.39850390e-02 1.15527079e-01
5.73493540e-01 4.60849762e-01 -5.63993454e-01 1.99890539e-01
2.08707497e-01 1.18087932e-01 1.56569052e-02 -7.07672715e-01
1.32645571e+00 -7.47132972e-02 3.92386943e-01 9.26160514e-01
4.33887929e-01 -3.53585809e-01 -2.18659997e+00 -1.38724998e-01
4.49807763e-01 -5.06120563e-01 -7.13231489e-02 -8.94441128e-01
-4.67197865e-01 6.58786535e-01 6.38014197e-01 4.56349403e-01
6.39284372e-01 -3.87277693e-01 2.82207847e-01 4.39082921e-01
6.56971157e-01 -1.52749205e+00 1.81113318e-01 5.93429625e-01
6.40338898e-01 -7.64162660e-01 5.77059947e-03 4.25515413e-01
-8.33801448e-01 1.05645382e+00 6.27487004e-01 -3.57120842e-01
5.76243818e-01 4.97548670e-01 -2.50734031e-01 4.07217145e-01
-1.07887471e+00 -3.58216763e-01 -3.39924216e-01 9.43274856e-01
5.20267487e-02 3.41482729e-01 -3.91588092e-01 2.91669279e-01
-1.41721100e-01 9.80392173e-02 7.68519044e-01 1.12741601e+00
-9.88546848e-01 -1.33161139e+00 -5.77031434e-01 5.82255900e-01
-4.67427999e-01 3.20226818e-01 1.31572440e-01 8.66835833e-01
-4.70132947e-01 8.95264626e-01 3.09613198e-01 9.86068621e-02
2.91092485e-01 2.44119734e-01 8.84942710e-01 -3.96956414e-01
-1.90478370e-01 2.14228645e-01 -4.97587398e-02 -6.32036090e-01
5.03284894e-02 -6.93070292e-01 -1.29752684e+00 -3.88523668e-01
1.68248400e-01 5.63855529e-01 5.73018789e-01 8.24795902e-01
-6.36398867e-02 4.77330863e-01 6.38807714e-01 -7.16698468e-01
-1.73021889e+00 -7.57635593e-01 -1.07663286e+00 5.63843846e-02
2.88594037e-01 -1.02506542e+00 -3.56381029e-01 -3.34441721e-01]
|
[4.196836948394775, 2.4997730255126953]
|
afdad2db-efc2-4fad-8362-5d0956c565c1
|
monocular-3d-object-reconstruction-with-gan
|
2207.10061
| null |
https://arxiv.org/abs/2207.10061v1
|
https://arxiv.org/pdf/2207.10061v1.pdf
|
Monocular 3D Object Reconstruction with GAN Inversion
|
Recovering a textured 3D mesh from a monocular image is highly challenging, particularly for in-the-wild objects that lack 3D ground truths. In this work, we present MeshInversion, a novel framework to improve the reconstruction by exploiting the generative prior of a 3D GAN pre-trained for 3D textured mesh synthesis. Reconstruction is achieved by searching for a latent space in the 3D GAN that best resembles the target mesh in accordance with the single view observation. Since the pre-trained GAN encapsulates rich 3D semantics in terms of mesh geometry and texture, searching within the GAN manifold thus naturally regularizes the realness and fidelity of the reconstruction. Importantly, such regularization is directly applied in the 3D space, providing crucial guidance of mesh parts that are unobserved in the 2D space. Experiments on standard benchmarks show that our framework obtains faithful 3D reconstructions with consistent geometry and texture across both observed and unobserved parts. Moreover, it generalizes well to meshes that are less commonly seen, such as the extended articulation of deformable objects. Code is released at https://github.com/junzhezhang/mesh-inversion
|
['Chen Change Loy', 'Bo Dai', 'Chai Kiat Yeo', 'Zhongang Cai', 'Daxuan Ren', 'Junzhe Zhang']
|
2022-07-20
| null | null | null | null |
['3d-object-reconstruction', 'object-reconstruction']
|
['computer-vision', 'computer-vision']
|
[ 1.43789768e-01 5.20459354e-01 -2.79490463e-02 -1.62337527e-01
-8.11854124e-01 -6.33828878e-01 5.54578125e-01 -4.25660133e-01
5.05798459e-01 4.81784046e-01 2.57719129e-01 2.41328135e-01
1.27409011e-01 -1.13403380e+00 -1.24053776e+00 -8.49918544e-01
5.41359186e-01 9.27425623e-01 -1.81406513e-01 -7.06083477e-02
-2.22451195e-01 7.31028318e-01 -1.52211988e+00 1.82550892e-01
4.61361855e-01 9.15072441e-01 3.52275997e-01 4.22492653e-01
5.21218143e-02 4.02147084e-01 4.74011600e-02 -2.05185294e-01
5.97960830e-01 -5.42468965e-01 -7.09972322e-01 7.13513970e-01
8.11879396e-01 -4.30153996e-01 -4.23822731e-01 9.33824182e-01
-2.07414087e-02 -2.07599670e-01 8.22764695e-01 -7.76483297e-01
-6.13185585e-01 1.29422219e-02 -4.76789534e-01 -6.81751549e-01
5.58708370e-01 1.66512206e-01 9.40181255e-01 -1.11889184e+00
1.09616137e+00 1.37292552e+00 5.69796503e-01 5.75946212e-01
-1.52694535e+00 -3.11483711e-01 -1.92419626e-03 -4.40345198e-01
-1.36637938e+00 -2.83977538e-01 1.22729445e+00 -6.85752392e-01
2.90697455e-01 2.97157615e-01 8.05465937e-01 1.37142718e+00
1.54460132e-01 5.79632044e-01 1.04341280e+00 -3.97543818e-01
8.75991285e-02 -1.38071075e-01 -6.65448606e-01 8.80078256e-01
-9.07874331e-02 2.09552035e-01 -5.04621029e-01 -7.71140158e-02
1.46618545e+00 3.40480119e-01 -4.13896322e-01 -1.02377856e+00
-1.38751781e+00 6.44067526e-01 4.36380267e-01 2.29189470e-02
-6.34974420e-01 2.82669812e-01 -1.41858876e-01 1.58883065e-01
9.65669990e-01 1.87739596e-01 -4.27406996e-01 4.84275073e-02
-7.96804011e-01 2.24661589e-01 4.61350232e-01 1.16588831e+00
1.18109822e+00 2.74191558e-01 1.27571777e-01 6.79183841e-01
4.19776440e-01 8.35865080e-01 -2.06301004e-01 -1.39219391e+00
2.10169435e-01 7.05753744e-01 3.06959357e-02 -7.64883578e-01
1.31370708e-01 -2.87798405e-01 -1.09537625e+00 4.70126241e-01
2.40028352e-01 3.16642106e-01 -1.05404961e+00 1.67584276e+00
7.15158403e-01 1.47700965e-01 -2.68823683e-01 9.20096278e-01
8.11059773e-01 3.76678944e-01 -5.92246294e-01 7.00771734e-02
1.12534356e+00 -7.12350488e-01 -4.06178713e-01 -1.06729329e-01
1.06961854e-01 -8.81574631e-01 1.17129922e+00 1.85900718e-01
-1.45369613e+00 -4.82096463e-01 -6.73276484e-01 -2.54227042e-01
9.90380868e-02 -7.70152211e-02 2.70424783e-01 1.67023435e-01
-8.76738608e-01 5.94668090e-01 -9.52996314e-01 -2.60313272e-01
6.08949780e-01 -1.07203089e-01 -6.22574449e-01 -3.69617283e-01
-6.71670198e-01 5.44568896e-01 1.76621731e-02 7.90887102e-02
-1.14525318e+00 -8.91341388e-01 -1.07223928e+00 -2.66505808e-01
4.62770849e-01 -1.25731623e+00 8.71465147e-01 -7.62517095e-01
-1.66440928e+00 1.37537861e+00 -1.81481495e-01 1.57043964e-01
8.35681975e-01 -3.01394742e-02 2.54528105e-01 1.73256829e-01
2.11544096e-01 5.60349286e-01 1.22167635e+00 -1.83002019e+00
1.12339519e-02 -5.43959200e-01 8.89626443e-02 1.71069324e-01
4.46018189e-01 -9.19854581e-01 -5.25848150e-01 -9.10716236e-01
5.05488396e-01 -9.47082698e-01 -6.33965507e-02 4.32371169e-01
-6.81781650e-01 2.21930519e-01 6.90357983e-01 -6.72904849e-01
4.42853063e-01 -2.06959629e+00 7.92087257e-01 3.49378884e-01
3.13379347e-01 -4.28159684e-01 3.73671427e-02 4.23736393e-01
-4.28595953e-02 -1.52122110e-01 -4.26031023e-01 -8.17677736e-01
4.92779613e-02 5.05214930e-01 -2.77923971e-01 7.17846274e-01
2.80206889e-01 1.17137349e+00 -7.60630488e-01 -1.95471719e-01
6.46139681e-01 7.15872705e-01 -8.34147871e-01 5.86360037e-01
-5.35041213e-01 1.35917628e+00 -5.72297573e-01 9.04234827e-01
8.31926227e-01 -5.47376752e-01 -3.02021764e-02 -4.12117511e-01
2.11442653e-02 5.76664768e-02 -1.04271376e+00 2.29890203e+00
-6.49782717e-01 1.25053629e-01 3.47636640e-01 -8.41884971e-01
7.79209793e-01 3.46538633e-01 5.75078189e-01 -4.11398828e-01
1.16255380e-01 3.04473758e-01 -6.84368551e-01 -3.67847741e-01
1.21051192e-01 -3.98414671e-01 1.57633483e-01 2.22310051e-01
5.22868186e-02 -8.42214882e-01 -5.83482265e-01 1.68901024e-04
8.45916986e-01 5.58418036e-01 8.99887234e-02 -2.38309592e-01
2.81021357e-01 -2.54382342e-01 2.41987064e-01 3.99857908e-01
7.25598574e-01 1.22016931e+00 1.78126097e-01 -3.51639241e-01
-1.42684531e+00 -1.41146111e+00 -2.25233361e-01 1.97310761e-01
2.45109141e-01 -2.74642825e-01 -6.82601273e-01 -6.46409810e-01
7.94191733e-02 3.82229894e-01 -1.00593019e+00 -6.09948672e-02
-5.17985165e-01 3.72982472e-02 -6.63649365e-02 2.09846497e-01
2.91741848e-01 -9.47893798e-01 -4.30890471e-01 4.24977317e-02
-2.97103703e-01 -1.23918271e+00 -3.70924592e-01 -1.42747713e-02
-1.01857567e+00 -1.07983184e+00 -8.74989092e-01 -6.24768436e-01
1.02410114e+00 1.22427762e-01 1.44573104e+00 3.18966955e-01
-1.49877176e-01 7.07183361e-01 -2.86924988e-01 1.00416735e-01
-5.92118323e-01 -1.11996047e-01 -2.61075467e-01 3.21184486e-01
-4.94796991e-01 -9.92988884e-01 -6.44131243e-01 3.90038818e-01
-8.94170105e-01 5.89194059e-01 3.54999512e-01 1.02791417e+00
1.09867406e+00 -1.16154984e-01 -4.19873483e-02 -9.55705404e-01
-4.17115480e-01 -5.00234485e-01 -4.70520884e-01 -1.03308661e-02
-2.58959115e-01 4.36697714e-02 3.31720859e-01 -2.28786230e-01
-1.05756688e+00 2.32928097e-01 -3.30961019e-01 -9.73365724e-01
-3.84422630e-01 4.79362309e-02 -4.69602495e-01 4.59432155e-02
3.64809513e-01 3.14313442e-01 2.47950271e-01 -8.75856519e-01
2.91406006e-01 6.82970881e-02 3.81693810e-01 -7.84625292e-01
1.07335126e+00 1.12399650e+00 3.00922364e-01 -8.52297246e-01
-1.04501188e+00 -1.86413616e-01 -7.82592893e-01 -2.70398796e-01
9.90934372e-01 -1.02859056e+00 -4.85663593e-01 4.08300281e-01
-1.04995406e+00 -8.13712716e-01 -8.29709113e-01 2.79638022e-01
-1.17273152e+00 2.77632564e-01 -4.43289012e-01 -4.96398747e-01
-1.23916447e-01 -1.24183488e+00 1.88746452e+00 -3.74585569e-01
-1.35532051e-01 -1.16673756e+00 6.39990345e-03 4.38664526e-01
8.35521594e-02 6.88667238e-01 1.06270695e+00 5.20766795e-01
-1.04266608e+00 5.68263978e-02 1.15995362e-01 4.62821692e-01
3.60250711e-01 -1.18096069e-01 -1.09577549e+00 -1.52546853e-01
1.78118095e-01 -3.47553283e-01 6.15958333e-01 5.88115036e-01
1.24446678e+00 -1.74877003e-01 -5.17778322e-02 9.82607841e-01
1.42508948e+00 -4.18460995e-01 6.34835243e-01 -3.89752984e-02
1.26743376e+00 6.78591073e-01 3.06313694e-01 3.83259863e-01
3.95319849e-01 9.28395808e-01 9.34416533e-01 -3.08087498e-01
-5.34658015e-01 -7.55878627e-01 1.58474132e-01 7.78728783e-01
-2.42172793e-01 -2.69717306e-01 -5.94498694e-01 3.43074501e-01
-1.61308956e+00 -6.03246391e-01 -1.48438811e-01 2.21187806e+00
7.96599984e-01 -1.85479045e-01 -2.06776530e-01 1.23715056e-02
4.99222487e-01 5.84896207e-02 -6.11137986e-01 9.87417921e-02
-3.31544518e-01 3.79681885e-01 1.88699380e-01 8.18920076e-01
-5.91307878e-01 8.37213337e-01 5.13016605e+00 8.94761264e-01
-1.05572331e+00 1.35583028e-01 5.52722991e-01 1.43373132e-01
-8.56039822e-01 1.19880386e-01 -3.67434770e-01 2.87515342e-01
1.31848723e-01 3.31994802e-01 5.99940360e-01 5.85989296e-01
1.05865523e-01 1.61438614e-01 -1.36545479e+00 1.18585432e+00
-8.96170139e-02 -1.62786090e+00 3.35444123e-01 5.23705542e-01
1.18900621e+00 -5.36855422e-02 1.03126019e-01 -2.39975452e-01
1.55286163e-01 -1.01118064e+00 1.16373491e+00 8.54732215e-01
1.29035115e+00 -4.05635595e-01 2.85655379e-01 4.79157269e-01
-9.55865800e-01 6.48318112e-01 -9.43163708e-02 1.88283369e-01
3.31826717e-01 8.55720460e-01 -4.93578434e-01 9.00263608e-01
6.40897989e-01 1.11122167e+00 -9.10932571e-02 2.99714983e-01
-4.03027266e-01 3.21338236e-01 -2.86157817e-01 9.47296679e-01
3.12161744e-02 -5.44145346e-01 7.86812842e-01 5.36908269e-01
6.15329802e-01 2.60162931e-02 1.63563609e-01 1.25717771e+00
-2.80015826e-01 -7.22490549e-02 -8.76842380e-01 3.19685787e-01
1.47990972e-01 1.00413525e+00 -6.72418952e-01 -2.62589842e-01
-1.46022961e-01 1.22175193e+00 2.35669300e-01 4.04418707e-01
-6.42859697e-01 5.92215955e-01 5.80327630e-01 6.55810952e-01
5.14311254e-01 -1.96120888e-01 -3.69786322e-01 -1.36660719e+00
3.00139427e-01 -7.70061612e-01 -1.04462154e-01 -1.06781816e+00
-1.52895272e+00 4.56487685e-01 -9.44624096e-02 -1.52447045e+00
-1.54250965e-01 -3.90997648e-01 -3.39930326e-01 8.69895101e-01
-1.24602616e+00 -1.43468416e+00 -5.92715025e-01 7.29699612e-01
6.27528191e-01 2.34843478e-01 9.26836848e-01 8.61398876e-02
2.82291919e-02 1.49969175e-01 -9.45284888e-02 -2.74645865e-01
5.38545251e-01 -1.18606877e+00 3.54171932e-01 4.50101644e-01
1.80827588e-01 3.45433682e-01 5.41892350e-01 -7.72916496e-01
-1.74978507e+00 -1.14377165e+00 2.63800770e-01 -6.74457133e-01
1.99639648e-01 -4.84261841e-01 -8.53853166e-01 8.61549318e-01
-6.67627230e-02 2.14767322e-01 1.30912915e-01 -3.36786807e-01
-3.83937716e-01 2.55285442e-01 -1.32420850e+00 5.04962862e-01
1.43066192e+00 -6.98267043e-01 -1.57732889e-01 4.37014371e-01
4.75083143e-01 -9.34549689e-01 -1.14700842e+00 5.39382815e-01
5.45000434e-01 -9.98167753e-01 1.19217277e+00 -2.61819571e-01
8.75451982e-01 -2.94975549e-01 -6.15341127e-01 -1.35936034e+00
-9.55274701e-02 -6.48942590e-01 -3.24332416e-01 9.60835099e-01
-7.89146125e-02 -3.95026594e-01 9.47378874e-01 1.19976282e-01
-3.15349013e-01 -8.41521680e-01 -8.64881635e-01 -7.56694078e-01
1.16763048e-01 -4.18023705e-01 7.15838909e-01 1.06475639e+00
-8.88276517e-01 1.87653720e-01 -5.40547073e-01 1.51595876e-01
9.94083047e-01 5.01591861e-01 1.11610687e+00 -1.26923740e+00
-5.04461884e-01 -1.28990412e-01 -4.50928926e-01 -1.50000978e+00
2.98007876e-01 -1.11697185e+00 -2.19008960e-02 -1.34532738e+00
-2.92868819e-03 -5.82332075e-01 5.30962348e-01 1.94774792e-01
2.27405548e-01 6.10420763e-01 -5.35439998e-02 3.05424452e-01
-1.12081528e-01 1.08598363e+00 1.90551889e+00 5.20031014e-03
1.81906596e-01 -8.05392563e-02 -4.14501429e-01 9.52999175e-01
3.68792057e-01 -4.18748051e-01 -2.30873138e-01 -7.10665107e-01
2.48705238e-01 3.07600975e-01 8.77515376e-01 -4.83846813e-01
-3.39003414e-01 -1.99815840e-01 4.66625124e-01 -6.23919725e-01
7.85932481e-01 -1.15456569e+00 9.09875989e-01 1.30129205e-02
-1.47829100e-01 -2.62691826e-01 -1.89533174e-01 7.33064294e-01
-6.30284175e-02 1.37122288e-01 8.35656047e-01 -3.29614460e-01
-1.08532839e-01 8.38309050e-01 2.48984501e-01 2.53114998e-01
7.19599485e-01 -3.35175216e-01 2.25290939e-01 -3.59318256e-01
-1.03792727e+00 -1.50652722e-01 1.38001847e+00 2.11149514e-01
8.60930204e-01 -1.75469804e+00 -7.23964930e-01 6.08185709e-01
2.80987889e-01 8.26633751e-01 5.45580208e-01 6.79349542e-01
-5.54153800e-01 -1.15540206e-01 -1.70885578e-01 -1.08189845e+00
-8.93987834e-01 3.00992876e-01 4.50481266e-01 -9.39321145e-02
-1.23736966e+00 6.90941811e-01 8.70680809e-01 -7.87244856e-01
1.03009129e-02 -3.95668685e-01 5.93221784e-01 -3.84546608e-01
9.11034867e-02 1.56738609e-01 1.86672017e-01 -8.60507369e-01
-8.25573951e-02 1.09936678e+00 4.83878344e-01 -2.32245270e-02
1.60202467e+00 -2.34961018e-01 -2.36614779e-01 6.73970997e-01
1.37149739e+00 3.70128989e-01 -1.67426920e+00 -3.67414713e-01
-5.80933630e-01 -8.88864994e-01 5.53401783e-02 -3.35052222e-01
-1.45818543e+00 7.44241357e-01 1.54957518e-01 5.66890137e-03
8.23753953e-01 5.25388658e-01 6.95640266e-01 -1.61581561e-01
7.37707317e-01 -3.74897242e-01 1.68866411e-01 3.15483332e-01
1.34883773e+00 -1.11157036e+00 3.17253657e-02 -7.72439718e-01
-2.50896931e-01 1.09016860e+00 2.85170674e-01 -4.94977564e-01
9.05101597e-01 2.06595380e-03 -2.70870119e-01 -8.27256918e-01
-4.19927150e-01 5.83569892e-02 4.53443319e-01 5.54478288e-01
5.99584691e-02 2.53282666e-01 4.58999991e-01 2.97575798e-02
-3.01219344e-01 -3.12307984e-01 3.37296993e-01 7.39443481e-01
2.17765644e-01 -1.08103490e+00 -3.48869383e-01 2.42929876e-01
-1.78066283e-01 8.60372335e-02 -2.93824643e-01 6.21840477e-01
3.04738253e-01 3.78448814e-01 1.25206575e-01 -1.22124359e-01
5.23624659e-01 -1.67960793e-01 1.00424290e+00 -9.75013971e-01
-1.73212923e-02 4.01327699e-01 -2.05822244e-01 -8.94073844e-01
-5.16822278e-01 -6.09151781e-01 -9.76650596e-01 -2.59256303e-01
-6.66770488e-02 -2.37419799e-01 5.12493789e-01 7.88915932e-01
3.73321086e-01 3.75661910e-01 7.53875196e-01 -1.35848761e+00
-2.09814668e-01 -6.08723164e-01 -8.03143740e-01 7.14385629e-01
5.68497241e-01 -1.06875587e+00 -5.63981891e-01 2.69911051e-01]
|
[8.89989948272705, -3.285189628601074]
|
53b71516-01ed-4b73-84ec-7e30b24c0eda
|
knowledge-graph-question-answering
|
2201.08174
| null |
https://arxiv.org/abs/2201.08174v1
|
https://arxiv.org/pdf/2201.08174v1.pdf
|
Knowledge Graph Question Answering Leaderboard: A Community Resource to Prevent a Replication Crisis
|
Data-driven systems need to be evaluated to establish trust in the scientific approach and its applicability. In particular, this is true for Knowledge Graph (KG) Question Answering (QA), where complex data structures are made accessible via natural-language interfaces. Evaluating the capabilities of these systems has been a driver for the community for more than ten years while establishing different KGQA benchmark datasets. However, comparing different approaches is cumbersome. The lack of existing and curated leaderboards leads to a missing global view over the research field and could inject mistrust into the results. In particular, the latest and most-used datasets in the KGQA community, LC-QuAD and QALD, miss providing central and up-to-date points of trust. In this paper, we survey and analyze a wide range of evaluation results with significant coverage of 100 publications and 98 systems from the last decade. We provide a new central and open leaderboard for any KGQA benchmark dataset as a focal point for the community - https://kgqa.github.io/leaderboard. Our analysis highlights existing problems during the evaluation of KGQA systems. Thus, we will point to possible improvements for future evaluations.
|
['Ricardo Usbeck', 'Andreas Both', 'Longquan Jiang', 'Liubov Kovriguina', 'Xi Yan', 'Aleksandr Perevalov']
|
2022-01-20
| null |
https://aclanthology.org/2022.lrec-1.321
|
https://aclanthology.org/2022.lrec-1.321.pdf
|
lrec-2022-6
|
['graph-question-answering']
|
['graphs']
|
[-5.59594512e-01 2.92568922e-01 3.92007781e-03 -4.52456832e-01
-7.75721133e-01 -9.51112688e-01 5.74334860e-01 5.69844246e-01
-1.18556947e-01 6.82371914e-01 2.57309049e-01 -5.08559823e-01
-5.17862201e-01 -9.69679713e-01 -6.74768865e-01 -2.28035212e-01
6.36914819e-02 7.86092520e-01 4.05066967e-01 -6.78875327e-01
3.15883011e-01 3.62729490e-01 -1.54512405e+00 3.73550236e-01
1.02502644e+00 7.39680707e-01 -1.77799970e-01 5.46616971e-01
-3.00234109e-01 1.14547479e+00 -6.97458923e-01 -8.84130061e-01
-4.34470363e-02 2.98012793e-02 -1.47648716e+00 -6.56409860e-01
7.04994142e-01 4.80867773e-02 3.75856715e-03 8.33492756e-01
5.83560348e-01 -3.29761982e-01 1.15335949e-01 -1.88036144e+00
-8.54703724e-01 8.17397356e-01 1.20928828e-02 1.16002364e-02
5.96288562e-01 1.34507775e-01 1.25817156e+00 -6.22816145e-01
1.00274837e+00 1.11495197e+00 5.31284809e-01 3.45751762e-01
-8.20102811e-01 -5.36296248e-01 -2.49047708e-02 7.63804674e-01
-1.26641297e+00 -4.93089944e-01 5.97467661e-01 -4.50806916e-01
9.81935084e-01 4.44446385e-01 5.82013369e-01 8.28281462e-01
-3.89628373e-02 4.83349353e-01 1.30124724e+00 -3.98583144e-01
2.97022730e-01 3.59866351e-01 5.69711328e-01 8.12987149e-01
7.29579806e-01 -3.19159299e-01 -9.17721272e-01 -4.65030402e-01
1.36507303e-01 -4.29748833e-01 -4.20132756e-01 -7.17098355e-01
-1.09856331e+00 5.60555935e-01 3.56858581e-01 4.49779391e-01
-1.56893864e-01 4.89923730e-02 3.20511401e-01 7.66849160e-01
8.75251889e-02 8.91066611e-01 -5.76260686e-01 -4.96906012e-01
-7.31761873e-01 5.81967711e-01 1.39889050e+00 9.82002676e-01
8.47913682e-01 -5.20539939e-01 -4.91673592e-03 2.47778207e-01
4.00931776e-01 4.11342084e-01 -1.63341016e-01 -1.18954647e+00
2.72013366e-01 1.13958669e+00 1.88806683e-01 -1.18221605e+00
-3.01891088e-01 -5.15086532e-01 -4.67592567e-01 6.53020814e-02
6.18569613e-01 1.74404919e-01 -5.57710528e-01 1.31797457e+00
5.05219877e-01 -3.55886430e-01 2.22126439e-01 9.59665418e-01
1.28921270e+00 1.86332479e-01 -3.03516816e-02 2.38403916e-01
1.35222578e+00 -8.41096759e-01 -7.87865520e-01 1.58587113e-01
9.20215249e-01 -7.08586335e-01 1.15716374e+00 6.35993659e-01
-9.86315250e-01 -2.31904998e-01 -8.82899344e-01 -2.53112584e-01
-8.56987298e-01 -3.07792157e-01 6.87923670e-01 7.25065649e-01
-1.40873706e+00 3.76362592e-01 -7.74312735e-01 -7.24701762e-01
3.11944038e-01 6.94363117e-02 -5.10015905e-01 -3.29017013e-01
-1.46554577e+00 1.21659851e+00 1.81850672e-01 1.88735612e-02
-1.03980684e+00 -1.03224361e+00 -3.97038907e-01 -1.53091460e-01
6.58810854e-01 -9.74164307e-01 1.18007898e+00 -3.76772821e-01
-1.01936889e+00 8.76243114e-01 1.23734489e-01 -4.25947785e-01
4.87622529e-01 -2.39875123e-01 -7.20780671e-01 6.86811581e-02
6.30868077e-02 2.22698450e-01 2.54625440e-01 -1.16681111e+00
-3.53426397e-01 -5.08380532e-01 5.60612202e-01 1.53825685e-01
-1.76569924e-01 1.18349962e-01 -4.30757970e-01 5.39276712e-02
-2.91063458e-01 -5.22644639e-01 8.17822590e-02 -1.22638591e-01
-1.52910382e-01 -4.87675667e-01 8.19274664e-01 -5.71163118e-01
1.45375931e+00 -1.80981421e+00 -1.37012708e-03 8.59334692e-02
6.46737516e-01 1.69244617e-01 1.72089338e-01 1.22104704e+00
2.80393392e-01 4.17491645e-01 -6.52850373e-03 7.19317794e-02
9.76810977e-02 2.98429519e-01 -3.74295622e-01 2.50888526e-01
1.61389470e-01 1.06001604e+00 -1.10423529e+00 -4.66243505e-01
-2.18417495e-03 2.55463302e-01 -2.36784831e-01 2.19552621e-01
-5.34735799e-01 2.16877759e-01 -5.25647283e-01 9.56119895e-01
5.27242243e-01 -5.95967412e-01 1.45725846e-01 -3.86704028e-01
-1.78221360e-01 5.63247323e-01 -1.04761064e+00 1.99980426e+00
-4.72731702e-02 4.57288384e-01 1.84010297e-01 -6.45086825e-01
9.71381187e-01 4.01062131e-01 2.95332879e-01 -6.57263100e-01
-2.86203355e-01 5.57037532e-01 4.57334938e-03 -4.76870149e-01
6.32636368e-01 2.05858469e-01 5.19437380e-02 5.34829855e-01
1.25609385e-02 -4.15449500e-01 3.93978834e-01 8.20167243e-01
1.34206843e+00 -3.89924124e-02 -6.14662804e-02 -4.57545012e-01
3.89659315e-01 6.14795446e-01 2.07969368e-01 6.79029286e-01
-3.75681251e-01 4.10794765e-01 6.68268859e-01 -4.82217073e-01
-6.38054430e-01 -8.72811258e-01 -1.02904201e-01 9.49297130e-01
-3.06744669e-02 -1.02159297e+00 -6.08273625e-01 -8.79628539e-01
1.04055107e-01 8.43215883e-01 -6.32168770e-01 -6.64796382e-02
-1.47465408e-01 -3.11169654e-01 9.85755622e-01 2.80510008e-01
4.85789061e-01 -1.02390778e+00 -6.40675426e-01 7.24902302e-02
-3.37490171e-01 -9.81084824e-01 3.37904155e-01 -2.42224902e-01
-6.88815594e-01 -1.58253074e+00 -1.49596304e-01 -2.93874234e-01
3.57600123e-01 1.42277658e-01 1.92397821e+00 3.58591765e-01
-3.04815508e-02 9.06966686e-01 -7.34395564e-01 -6.71613693e-01
-5.56456327e-01 2.01836944e-01 -1.76228330e-01 -5.15070856e-01
6.04767084e-01 -2.86627740e-01 -7.48374760e-01 4.59914118e-01
-9.99993443e-01 -1.52622253e-01 4.57564622e-01 3.55022579e-01
4.42798018e-01 -1.95667848e-01 7.44325221e-01 -1.10008347e+00
9.04795885e-01 -5.40841699e-01 -5.57489216e-01 6.86736465e-01
-1.27620912e+00 -5.53317927e-02 2.31256753e-01 5.30498564e-01
-8.32383990e-01 -6.44330561e-01 -8.54993165e-02 1.47771705e-02
-1.05122268e-01 1.02571428e+00 -1.66487783e-01 -1.35213122e-01
1.01029813e+00 -2.51220345e-01 4.23442610e-02 -5.70475698e-01
7.40641057e-01 7.91184306e-01 4.48404759e-01 -9.10649240e-01
6.26563430e-01 4.89063680e-01 -2.54321694e-01 -5.13616502e-01
-7.06062853e-01 -5.31974792e-01 -2.97946334e-01 -2.20428869e-01
3.52558345e-01 -7.96305954e-01 -7.80067682e-01 3.27904195e-01
-1.04028547e+00 -1.59965739e-01 -4.87700790e-01 2.89552659e-02
-1.60995707e-01 4.22960490e-01 -2.78870165e-01 -5.06405532e-01
-7.79065490e-01 -1.01233983e+00 8.08867157e-01 1.70507059e-01
-2.42406592e-01 -9.38354254e-01 4.05427337e-01 9.71195698e-01
7.22264230e-01 2.10238382e-01 8.18114877e-01 -6.23270929e-01
-6.10188603e-01 -2.25546703e-01 -2.27324292e-01 3.16705585e-01
2.99861841e-02 3.94769043e-01 -1.07896423e+00 -4.03378844e-01
-3.65528136e-01 -6.59876287e-01 4.53227341e-01 -2.34728530e-01
8.81699920e-01 -3.91213708e-02 -1.79226100e-01 -2.79767588e-02
1.32682967e+00 -3.16128224e-01 5.65741241e-01 4.92269248e-01
5.91367304e-01 7.55384326e-01 7.47398734e-01 5.44970520e-02
1.08428037e+00 4.43588883e-01 7.05222547e-01 1.83075756e-01
-1.85020849e-01 -1.82560220e-01 2.24705830e-01 1.11044455e+00
-2.37632155e-01 -2.19316095e-01 -1.46595800e+00 8.51821661e-01
-1.84453547e+00 -5.87694943e-01 -6.69781148e-01 2.25615907e+00
9.31170642e-01 -7.67370388e-02 -1.06770974e-02 2.89843250e-02
2.18245283e-01 -4.21887301e-02 -3.91507357e-01 -3.78923386e-01
-2.92730093e-01 1.72614634e-01 1.17145374e-01 2.91618407e-01
-5.11448443e-01 8.39821696e-01 6.54207897e+00 4.38097417e-01
-8.16421390e-01 7.18833432e-02 2.14752659e-01 1.27625212e-01
-6.15223348e-01 4.92014468e-01 -6.10197127e-01 9.61130336e-02
1.24873996e+00 -6.12625241e-01 3.00355673e-01 6.73791051e-01
-8.54218751e-02 -1.25157341e-01 -1.13363743e+00 5.90773225e-01
-8.58550891e-02 -1.65696478e+00 4.78905253e-02 -1.31850109e-01
5.66686630e-01 5.71982920e-01 -2.57734805e-01 3.96820933e-01
7.18610823e-01 -1.10431826e+00 4.41453338e-01 7.69883335e-01
5.88736475e-01 -4.19296145e-01 9.14862096e-01 1.49469301e-01
-7.45791316e-01 3.14529240e-01 -3.78682464e-01 -1.99889481e-01
-1.16396599e-01 8.84994745e-01 -9.17545617e-01 1.26179123e+00
1.13381469e+00 5.94066381e-01 -9.04014766e-01 9.80380297e-01
-2.47828051e-01 9.17444766e-01 -3.34871739e-01 2.28787563e-03
-1.55672210e-03 1.75195802e-02 3.33967865e-01 1.14102471e+00
-1.95251722e-02 -1.42124057e-01 -2.77382806e-02 7.69324839e-01
-1.78227961e-01 2.06028014e-01 -6.56894147e-01 -4.70381349e-01
5.81564486e-01 1.41833150e+00 -1.45395353e-01 -2.48049110e-01
-5.02302349e-01 4.68739033e-01 5.86320519e-01 2.19597355e-01
-4.02253062e-01 -3.00678581e-01 6.45885587e-01 2.79899687e-01
-2.13141501e-01 -2.02851340e-01 -3.94136794e-02 -1.09903812e+00
1.46860987e-01 -1.36759293e+00 9.35130656e-01 -9.31860328e-01
-1.49752879e+00 4.80712861e-01 1.93487923e-03 -8.17308247e-01
-1.90543920e-01 -3.61444056e-01 -8.68472457e-02 8.87302935e-01
-1.77441895e+00 -1.17442346e+00 -7.56841898e-01 6.97480619e-01
-2.58801699e-01 1.13016732e-01 1.11051655e+00 3.58472079e-01
-2.55100429e-01 3.64909708e-01 -1.47069603e-01 3.79511085e-03
1.11935604e+00 -1.41663420e+00 4.34343904e-01 4.04532611e-01
3.78378034e-01 8.49032640e-01 6.80695891e-01 -6.22561753e-01
-2.04297471e+00 -8.67216587e-01 9.82999086e-01 -1.36301076e+00
1.04454207e+00 -2.20684916e-01 -1.24519265e+00 6.37224674e-01
5.31590462e-01 -5.56963570e-02 6.95363402e-01 3.97610128e-01
-6.08656943e-01 -3.35680395e-01 -9.49141800e-01 2.30019718e-01
8.85899484e-01 -6.93751395e-01 -6.47668242e-01 3.71621817e-01
6.79051340e-01 -3.61998886e-01 -1.30123377e+00 4.93767142e-01
4.67904091e-01 -1.17717099e+00 4.77693439e-01 -7.51165628e-01
1.90277010e-01 -6.61253273e-01 1.01396069e-02 -1.27353406e+00
-1.45730935e-03 -5.15054345e-01 -2.60617077e-01 1.25440037e+00
5.81134856e-01 -8.18307877e-01 7.92681813e-01 8.64570796e-01
-1.50626019e-01 -7.66342521e-01 -6.93966508e-01 -5.63917339e-01
1.00880258e-01 -6.66856587e-01 6.19148314e-01 1.28218281e+00
1.41083866e-01 2.25261331e-01 1.79520458e-01 2.33311087e-01
4.05031681e-01 2.81187505e-01 1.13697624e+00 -1.35499942e+00
-9.52953994e-02 -2.77020574e-01 -5.36801577e-01 -4.89101946e-01
-5.15549958e-01 -9.73079264e-01 -5.95150590e-01 -2.16406441e+00
1.76279873e-01 -5.02161026e-01 -3.73655975e-01 7.61351824e-01
-9.00927782e-02 1.06074642e-02 6.82782941e-03 3.76485467e-01
-9.46852267e-01 3.10689896e-01 1.28046238e+00 -1.27494857e-01
1.53819814e-01 -4.17484492e-01 -1.07452083e+00 2.15781391e-01
7.92361736e-01 -2.57446498e-01 -5.96518576e-01 -5.63453138e-01
9.76593852e-01 -3.74331772e-01 5.74035168e-01 -9.25442874e-01
5.49584627e-01 6.40167072e-02 -3.57511073e-01 -5.16573012e-01
-1.34857697e-02 -7.83278286e-01 4.20855641e-01 3.14285785e-01
-4.44918312e-02 2.58486718e-01 3.39201450e-01 2.24083751e-01
-5.43328583e-01 -2.00442430e-02 8.62433687e-02 -1.85759470e-01
-8.13769042e-01 9.83836800e-02 2.93436289e-01 3.80027473e-01
8.36178899e-01 -2.00195815e-02 -1.17980289e+00 -4.55695063e-01
-5.14716446e-01 7.94106245e-01 6.30593181e-01 5.39352357e-01
3.86000723e-01 -9.07667696e-01 -1.00078738e+00 -8.48915502e-02
6.39132857e-01 1.15607105e-01 1.92293450e-01 9.16492105e-01
-8.03696871e-01 5.89467347e-01 -2.60965884e-01 -3.80000442e-01
-9.87963259e-01 3.12259197e-01 3.83683562e-01 -3.43478620e-01
-2.95286715e-01 6.03767335e-01 -4.75347579e-01 -6.45587325e-01
1.26172155e-01 -2.02324718e-01 -2.19956324e-01 7.13092834e-02
3.55862409e-01 6.22390091e-01 5.40920973e-01 -2.30041832e-01
-5.58948159e-01 1.09744564e-01 -2.33430162e-01 4.92242016e-02
1.31690490e+00 2.12234203e-02 -6.54664099e-01 6.32230699e-01
6.11602306e-01 1.66147187e-01 -4.78512704e-01 -1.41818941e-01
1.57383785e-01 -3.37467402e-01 5.89428060e-02 -1.48176527e+00
-9.23692942e-01 6.97748721e-01 4.36825812e-01 6.45849288e-01
8.91402602e-01 2.50177771e-01 2.02545226e-01 6.42161727e-01
8.03897262e-01 -8.85321140e-01 -3.44475836e-01 2.10790798e-01
1.28109217e+00 -1.17273474e+00 1.52274072e-01 -4.41577137e-01
-4.99479800e-01 9.22328055e-01 6.91646934e-01 4.56341892e-01
5.62347651e-01 -1.03668235e-02 5.24093449e-01 -1.05808818e+00
-1.05822957e+00 -7.20636919e-02 1.99605897e-01 5.50864041e-01
6.99334323e-01 1.09015398e-01 -5.56852102e-01 4.17774320e-01
-4.67194617e-01 3.91740978e-01 6.53303325e-01 1.15999818e+00
-1.36072308e-01 -1.41285467e+00 -2.17403695e-01 4.95658308e-01
-3.33741307e-01 -6.16384931e-02 -9.40716505e-01 9.69168961e-01
-2.12989792e-01 1.35265148e+00 -3.91034871e-01 -3.08925748e-01
7.41789460e-01 1.97219342e-01 4.02149171e-01 -5.05926430e-01
-8.74202192e-01 -7.96725690e-01 5.79369545e-01 -7.95049965e-01
-5.10789275e-01 -3.83944392e-01 -1.05935073e+00 -6.43478155e-01
-3.45452458e-01 7.09452808e-01 7.31298923e-01 5.05096257e-01
9.02148664e-01 1.61043003e-01 -4.47419100e-02 3.61980021e-01
-4.93812561e-01 -9.06449378e-01 -3.63201380e-01 4.67845321e-01
-1.32898122e-01 -4.28600043e-01 -1.70718178e-01 -9.63437185e-02]
|
[10.11627197265625, 7.954028129577637]
|
cdb652cd-dcfe-43a7-8a3d-be4ce2efd339
|
achieving-stable-training-of-reinforcement
|
2307.00923
| null |
https://arxiv.org/abs/2307.00923v1
|
https://arxiv.org/pdf/2307.00923v1.pdf
|
Achieving Stable Training of Reinforcement Learning Agents in Bimodal Environments through Batch Learning
|
Bimodal, stochastic environments present a challenge to typical Reinforcement Learning problems. This problem is one that is surprisingly common in real world applications, being particularly applicable to pricing problems. In this paper we present a novel learning approach to the tabular Q-learning algorithm, tailored to tackling these specific challenges by using batch updates. A simulation of pricing problem is used as a testbed to compare a typically updated agent with a batch learning agent. The batch learning agents are shown to be both more effective than the typically-trained agents, and to be more resilient to the fluctuations in a large stochastic environment. This work has a significant potential to enable practical, industrial deployment of Reinforcement Learning in the context of pricing and others.
|
['G. Cevora', 'N. Peace', 'E. Hurwitz']
|
2023-07-03
| null | null | null | null |
['q-learning']
|
['methodology']
|
[-4.11352962e-02 -1.24289557e-01 -2.63826877e-01 -8.55549797e-02
-9.19071198e-01 -4.30166513e-01 5.03122389e-01 1.08976439e-01
-5.69158375e-01 1.34621942e+00 -2.49663934e-01 -4.50362682e-01
-3.88120085e-01 -8.26607227e-01 -6.01925850e-01 -8.54106605e-01
-7.34119058e-01 8.80767047e-01 3.86671275e-01 -6.16648376e-01
4.43430990e-01 1.92202836e-01 -1.69387221e+00 -1.20045081e-01
5.95767438e-01 9.41775143e-01 2.43023634e-01 7.43892670e-01
3.44814025e-02 8.78647685e-01 -9.20204222e-01 -4.06531543e-02
6.18495822e-01 -4.51946378e-01 -2.86020190e-01 2.17118412e-01
1.42472267e-01 -4.38608080e-01 1.83655456e-01 8.16090941e-01
7.88461208e-01 3.81530404e-01 3.51026684e-01 -1.28554082e+00
6.07039705e-02 4.52911586e-01 -6.54759347e-01 3.71301979e-01
4.36446279e-01 4.79909837e-01 9.62737203e-01 -4.51475419e-02
3.30067188e-01 1.44754541e+00 5.00397146e-01 4.25922990e-01
-1.25539291e+00 -5.43704689e-01 2.67533362e-01 6.45073354e-02
-5.04030287e-01 -1.80584863e-01 3.91082972e-01 -7.79467002e-02
1.09744370e+00 4.85415347e-02 6.86195195e-01 8.21109772e-01
7.71813154e-01 6.62439167e-01 1.59322357e+00 -4.13130701e-01
8.16307247e-01 1.17796168e-01 -4.72525060e-01 2.43171379e-01
2.18090326e-01 7.69695461e-01 -2.99519151e-01 -4.71320093e-01
6.11668110e-01 -1.05775841e-01 2.43735418e-01 -8.70357633e-01
-6.47687376e-01 1.12291515e+00 -2.12976396e-01 1.02090333e-02
-7.48582006e-01 5.04956543e-01 6.22147799e-01 9.27470684e-01
4.14464682e-01 6.20976031e-01 -8.46267641e-01 -6.49924815e-01
-5.63880026e-01 9.83349442e-01 1.19193423e+00 5.07670820e-01
3.50436509e-01 6.09617352e-01 -1.45384237e-01 8.28190923e-01
3.97035718e-01 6.29999340e-01 5.47607183e-01 -1.32883227e+00
2.48147517e-01 -1.36587664e-01 5.91580272e-01 -3.18484604e-01
-5.02327502e-01 -3.59533519e-01 -1.66125223e-01 9.72100079e-01
5.18319547e-01 -5.32775581e-01 -4.62391555e-01 1.56541407e+00
3.79471719e-01 2.19963551e-01 1.57046929e-01 3.05605173e-01
-1.82601064e-01 6.62346601e-01 -1.83380675e-02 -6.80598497e-01
8.34683478e-01 -6.40914381e-01 -5.94660640e-01 -1.74051940e-01
3.06269825e-01 -8.47100973e-01 7.61202931e-01 6.63872480e-01
-1.38632011e+00 -2.97815710e-01 -8.43404591e-01 1.19806206e+00
-3.14320415e-01 -9.10321832e-01 7.48549163e-01 1.01810479e+00
-1.19215143e+00 8.20528686e-01 -5.36512494e-01 -4.27593023e-01
8.26323479e-02 5.66350341e-01 4.45049405e-01 1.34200037e-01
-1.34119308e+00 1.14474130e+00 3.67667615e-01 -2.94140369e-01
-9.60429370e-01 -4.00237948e-01 -7.29934871e-01 3.90052982e-02
8.68418038e-01 -2.97390997e-01 2.13292098e+00 -1.02422559e+00
-2.01091719e+00 5.59622347e-02 5.00282705e-01 -8.78419399e-01
6.61521137e-01 1.54545039e-01 -2.32089967e-01 -2.29478315e-01
1.11300759e-01 1.96960524e-01 1.06162202e+00 -1.17741776e+00
-9.78137970e-01 -9.23331305e-02 5.70914112e-02 5.48105419e-01
1.01287283e-01 -3.25696394e-02 4.24253494e-01 -5.48038960e-01
-6.83842361e-01 -8.08082402e-01 -7.13462472e-01 -8.09731603e-01
3.81608486e-01 -4.22717482e-01 6.84709132e-01 -1.21276230e-01
9.49768722e-01 -1.58941007e+00 -2.97481894e-01 4.41048950e-01
-3.43709052e-01 5.69243208e-02 -2.49417767e-01 9.19840991e-01
3.05965357e-02 -1.67122886e-01 1.92875899e-02 1.33195221e-01
2.89062679e-01 6.53223217e-01 -2.42877990e-01 3.42840821e-01
1.04335621e-01 4.22567248e-01 -1.10386407e+00 -1.60310224e-01
2.32969642e-01 -3.21234375e-01 -4.80812013e-01 2.35539496e-01
-7.06488311e-01 1.54102758e-01 -5.79476118e-01 5.99927902e-01
3.87089252e-01 2.55303323e-01 3.50280225e-01 7.99119890e-01
-3.07742715e-01 1.80477614e-03 -1.61044252e+00 8.16834092e-01
-5.29609859e-01 1.05560809e-01 2.78741568e-01 -1.25941503e+00
8.28442156e-01 2.13694856e-01 9.87359345e-01 -1.14174628e+00
6.26561642e-02 1.92963213e-01 1.80784538e-01 -3.95861328e-01
5.07938623e-01 -5.01677513e-01 -2.13416487e-01 8.32171082e-01
2.16047298e-02 -6.11165285e-01 6.71058476e-01 2.33761240e-02
1.06394327e+00 3.61134678e-01 4.17872459e-01 -3.51298392e-01
1.45160973e-01 4.90384325e-02 7.23449469e-01 1.28251600e+00
-4.73426938e-01 -2.43862256e-01 6.53535128e-01 -4.82309848e-01
-9.88136530e-01 -9.41157401e-01 -3.45588252e-02 1.33666670e+00
1.04778238e-01 -3.64586636e-02 -4.03416544e-01 -3.96599978e-01
5.36953628e-01 7.83308566e-01 -4.13690209e-01 3.50147282e-04
-3.14426392e-01 -9.00013089e-01 -1.92939192e-01 2.51381218e-01
2.07239613e-01 -1.42284143e+00 -1.06317353e+00 9.66476679e-01
6.79355085e-01 -6.11999214e-01 -2.68668413e-01 5.07316828e-01
-8.32049489e-01 -1.07160985e+00 -4.90198731e-01 -4.12279695e-01
-1.22381918e-01 1.32177353e-01 1.37923980e+00 -6.86487108e-02
-1.55388892e-01 1.01535571e+00 -3.84260535e-01 -9.69873488e-01
-7.84389019e-01 -1.04665220e-01 2.17541382e-01 -3.16883355e-01
1.48278356e-01 -3.40421736e-01 -4.04071689e-01 3.96071374e-01
-1.10405242e+00 -7.05245554e-01 3.85917544e-01 1.43994880e+00
3.84577006e-01 5.68745673e-01 9.68073249e-01 -1.06407475e+00
1.17519319e+00 -6.17443144e-01 -1.24209154e+00 8.29036757e-02
-1.08786547e+00 5.60921207e-02 8.63037050e-01 -4.10000414e-01
-9.68920827e-01 -1.26251746e-02 1.63284674e-01 2.57981032e-01
3.01481970e-03 4.55353469e-01 3.69668514e-01 -1.49617419e-01
5.19643307e-01 7.18100416e-03 4.79152322e-01 -1.33959487e-01
5.46218827e-02 4.35445070e-01 -8.86600167e-02 -8.75574589e-01
8.82847309e-01 2.23597488e-03 1.09819032e-01 -6.27510846e-01
-1.49425074e-01 -4.00376797e-01 5.84800690e-02 -3.58115673e-01
2.21334428e-01 -7.36652672e-01 -8.49445701e-01 4.30625796e-01
-3.55721474e-01 -1.02630639e+00 -6.10038519e-01 2.60834128e-01
-1.31429029e+00 8.12990516e-02 -4.43941802e-01 -1.05082929e+00
5.31242304e-02 -1.07032156e+00 6.11815631e-01 6.44602060e-01
2.27316603e-01 -1.31726897e+00 7.67655432e-01 -4.92620841e-02
7.18171895e-01 2.49665082e-01 6.56836987e-01 -6.12581134e-01
-1.86291412e-01 -6.39700964e-02 3.97522867e-01 2.11272821e-01
7.76540413e-02 1.27082318e-01 -6.04214728e-01 -8.49261165e-01
4.63874340e-02 -7.56514728e-01 5.47889113e-01 5.20403266e-01
7.03900516e-01 -1.32746428e-01 1.95084602e-01 -2.60680556e-01
1.59401643e+00 7.28412867e-01 3.73828828e-01 9.20047581e-01
-2.63805777e-01 4.86194104e-01 1.12390494e+00 1.06122792e+00
1.83363214e-01 6.40532196e-01 6.24865651e-01 1.62654426e-02
5.87395310e-01 3.71333957e-02 5.93410909e-01 3.78938735e-01
1.58791900e-01 6.87118769e-02 -9.48114574e-01 2.48111457e-01
-2.06846166e+00 -1.45288599e+00 4.28822309e-01 2.23530936e+00
7.47315347e-01 5.09269536e-01 6.39655411e-01 -9.56706181e-02
5.15637696e-01 1.43542988e-02 -9.17389333e-01 -9.29220319e-01
7.68839866e-02 4.35870826e-01 7.98327148e-01 3.03599179e-01
-1.00462437e+00 6.13309860e-01 7.54166269e+00 7.54161179e-01
-6.28270388e-01 -8.34967345e-02 7.05689430e-01 -1.36449113e-01
-1.72167141e-02 -2.27988064e-02 -5.02257049e-01 6.13491654e-01
1.34771752e+00 -3.78806621e-01 8.80780101e-01 9.92736161e-01
5.94533622e-01 -7.11115777e-01 -8.77620935e-01 5.96117973e-01
-2.67922461e-01 -9.51030195e-01 -4.38278586e-01 4.75020260e-02
1.00632143e+00 7.93228149e-02 3.46351534e-01 9.39160764e-01
1.05262029e+00 -9.76112187e-01 4.36299503e-01 1.71839848e-01
2.12323576e-01 -1.16012275e+00 1.00127232e+00 3.60858530e-01
-9.28304613e-01 -5.15756428e-01 -4.72859532e-01 -4.24069583e-01
-7.88194910e-02 3.42878066e-02 -7.65655696e-01 3.46512139e-01
8.16822946e-01 3.60411495e-01 -2.61386842e-01 1.47523975e+00
3.06088895e-01 5.63560784e-01 -2.80340105e-01 -4.63118553e-01
6.31173790e-01 -2.52865821e-01 2.71253526e-01 9.00688231e-01
7.47737586e-02 -2.86749840e-01 7.54338741e-01 1.94625124e-01
3.79062265e-01 2.45690241e-01 -7.80870438e-01 1.12909958e-01
3.44041139e-01 8.77972364e-01 -6.94365263e-01 -2.55507410e-01
-4.56090063e-01 4.12700772e-01 2.07069471e-01 3.70251596e-01
-5.92214584e-01 -3.40762913e-01 6.71615601e-01 -1.12389877e-01
4.70857114e-01 -5.51700667e-02 1.38150424e-01 -6.63048744e-01
-3.73950392e-01 -1.47547400e+00 5.19718766e-01 -3.10361266e-01
-1.62136257e+00 2.77546197e-02 1.65732875e-01 -1.16675448e+00
-8.25477123e-01 -7.76096284e-01 -7.42750704e-01 6.43096209e-01
-1.68039048e+00 -1.92521229e-01 1.50296777e-01 6.35337174e-01
8.48124802e-01 -5.86483061e-01 7.39074588e-01 -2.28798181e-01
-4.15493608e-01 1.42344609e-01 9.61678147e-01 -5.82840621e-01
8.34156036e-01 -1.82006896e+00 5.96438125e-02 3.46568227e-01
-1.36596203e-01 1.84624031e-01 1.07682729e+00 -4.74925011e-01
-1.52424872e+00 -8.08887482e-01 -1.07768908e-01 -1.34307057e-01
1.06590831e+00 2.82954462e-02 -5.55847406e-01 2.26971760e-01
4.91530746e-01 -4.26690370e-01 5.52028239e-01 2.69269720e-02
1.85569987e-01 -3.86420161e-01 -1.27960205e+00 5.64501405e-01
3.11231017e-01 -1.24643311e-01 -4.74070519e-01 4.69011813e-01
2.16810346e-01 -5.25459528e-01 -8.34255517e-01 6.37127459e-02
4.52639490e-01 -9.27861989e-01 5.87334812e-01 -6.06167436e-01
-6.09869510e-02 1.18541047e-01 1.85143091e-02 -1.91608524e+00
-2.04892099e-01 -1.21886599e+00 -1.47753075e-01 8.90986145e-01
3.21814001e-01 -1.06775403e+00 8.08311701e-01 4.89944607e-01
3.80971730e-01 -7.92104363e-01 -1.27538180e+00 -1.20659578e+00
5.68807960e-01 -1.35267735e-01 4.60785717e-01 5.71521461e-01
8.12810063e-02 -2.95463242e-02 -4.81735289e-01 -2.44512811e-01
9.37912941e-01 3.95708568e-02 7.78948665e-01 -1.07112741e+00
-8.86041939e-01 -5.79029500e-01 -2.43253663e-01 -5.56239605e-01
-2.26911362e-02 -8.46224427e-02 2.58652270e-01 -1.22255933e+00
-4.33258247e-03 -5.96956611e-01 -6.85527921e-01 1.53335944e-01
-7.80327916e-02 -1.06961645e-01 3.24130446e-01 -1.89563721e-01
-7.64855683e-01 4.98726636e-01 1.04486334e+00 -1.28856689e-01
-3.15963805e-01 5.33507884e-01 -4.31190938e-01 3.36539418e-01
1.42264271e+00 -5.86099982e-01 -5.25723755e-01 1.88133717e-01
1.80230975e-01 4.52238977e-01 -8.75019059e-02 -7.46447146e-01
1.15018487e-01 -7.59811044e-01 8.04297030e-02 -1.99946001e-01
-7.74328038e-02 -8.18569183e-01 -1.08426511e-01 8.74694049e-01
-2.27371648e-01 5.10926962e-01 4.14032310e-01 7.25528300e-01
-6.39776513e-02 -5.15005350e-01 8.28781247e-01 -4.91132200e-01
-7.91450083e-01 2.28144884e-01 -8.67376268e-01 3.67996186e-01
1.36671793e+00 -1.48140445e-01 -3.59564573e-02 -8.68295312e-01
-5.82920551e-01 8.29520941e-01 3.21355134e-01 1.35659054e-01
3.36292297e-01 -9.75488186e-01 -5.78041017e-01 -1.03100918e-01
-1.17639765e-01 -5.10451972e-01 -5.54921962e-02 4.78359520e-01
-3.23021561e-01 2.59997010e-01 -5.08297980e-01 -2.95565426e-01
-1.01774526e+00 5.44377863e-01 6.54794514e-01 -6.87790632e-01
-2.63878047e-01 2.96205997e-01 -4.77245778e-01 -3.24729025e-01
5.77118576e-01 -3.85906771e-02 3.40610631e-02 9.86148790e-02
3.78810376e-01 3.88057202e-01 -1.45073965e-01 4.85984832e-02
9.45493132e-02 1.88104644e-01 -1.84276789e-01 -3.99200112e-01
1.44333291e+00 5.72541505e-02 2.55722523e-01 6.87451422e-01
3.74252051e-01 -3.96152198e-01 -1.74486434e+00 -3.25087607e-01
4.29244995e-01 -3.54211450e-01 -4.31718417e-02 -8.47902656e-01
-7.66148388e-01 3.08080912e-01 9.24106836e-01 8.37433457e-01
9.38837767e-01 -5.74973047e-01 4.30566818e-01 5.78572214e-01
9.27835107e-01 -1.79528558e+00 2.87283808e-01 6.19692147e-01
5.18029273e-01 -1.32677412e+00 1.05683737e-01 4.43560660e-01
-8.40528369e-01 1.23745883e+00 5.74624121e-01 -4.34302211e-01
4.92266148e-01 5.17242849e-01 1.85926229e-01 1.31532952e-01
-1.31925702e+00 -3.92958641e-01 -6.49110138e-01 8.48267615e-01
3.79601493e-02 1.02099173e-01 -3.47776353e-01 -2.67845899e-01
1.30703256e-01 -5.80075532e-02 8.64091814e-01 1.27279365e+00
-7.25069821e-01 -1.52320850e+00 -5.95673978e-01 6.38177752e-01
-5.03778517e-01 2.77909786e-01 -9.09074098e-02 9.97808814e-01
-3.82557176e-02 9.65049803e-01 1.07008629e-02 1.79440826e-01
4.63440329e-01 -7.45132193e-02 5.14915287e-01 -5.93710721e-01
-9.73156273e-01 3.94115269e-01 8.92433822e-02 -7.36021101e-01
-3.95150423e-01 -1.14307690e+00 -9.48075950e-01 -3.40562582e-01
-7.99568966e-02 4.36026156e-01 5.48581719e-01 5.74876964e-01
-2.96011269e-01 5.25549591e-01 1.19294453e+00 -8.32848787e-01
-1.55640221e+00 -6.83076859e-01 -1.11407351e+00 2.42754638e-01
2.24365547e-01 -9.58832622e-01 -3.09787780e-01 -5.16087651e-01]
|
[4.150516986846924, 2.5090363025665283]
|
d15dd5cb-901f-4cd2-826c-691e57991286
|
speech-enhanced-and-noise-aware-networks-for
|
2203.13696
| null |
https://arxiv.org/abs/2203.13696v3
|
https://arxiv.org/pdf/2203.13696v3.pdf
|
Speech-enhanced and Noise-aware Networks for Robust Speech Recognition
|
Compensation for channel mismatch and noise interference is essential for robust automatic speech recognition. Enhanced speech has been introduced into the multi-condition training of acoustic models to improve their generalization ability. In this paper, a noise-aware training framework based on two cascaded neural structures is proposed to jointly optimize speech enhancement and speech recognition. The feature enhancement module is composed of a multi-task autoencoder, where noisy speech is decomposed into clean speech and noise. By concatenating its enhanced, noise-aware, and noisy features for each frame, the acoustic-modeling module maps each feature-augmented frame into a triphone state by optimizing the lattice-free maximum mutual information and cross entropy between the predicted and actual state sequences. On top of the factorized time delay neural network (TDNN-F) and its convolutional variant (CNN-TDNNF), both with SpecAug, the two proposed systems achieve word error rate (WER) of 3.90% and 3.55%, respectively, on the Aurora-4 task. Compared with the best existing systems that use bigram and trigram language models for decoding, the proposed CNN-TDNNF-based system achieves a relative WER reduction of 15.20% and 33.53%, respectively. In addition, the proposed CNN-TDNNF-based system also outperforms the baseline CNN-TDNNF system on the AMI task.
|
['Hsin-Min Wang', 'Yao-Fei Cheng', 'Yu Tsao', 'Pin-Yuan Chen', 'Hung-Shin Lee']
|
2022-03-25
| null | null | null | null |
['robust-speech-recognition']
|
['speech']
|
[ 1.85338810e-01 -1.63111612e-01 3.32994044e-01 -4.65369374e-01
-1.09151518e+00 8.80195424e-02 3.10881555e-01 -3.61527383e-01
-6.98965907e-01 3.86376649e-01 4.42683727e-01 -5.21927059e-01
2.03431234e-01 -3.07108402e-01 -5.87874115e-01 -8.95755649e-01
6.47119954e-02 -2.89083928e-01 3.33112814e-02 -1.54727325e-01
-3.39036494e-01 2.08736703e-01 -1.49760818e+00 2.74721652e-01
6.87874913e-01 1.35178602e+00 6.59672379e-01 1.06838799e+00
1.47112131e-01 5.11695862e-01 -7.62557209e-01 -2.61813104e-01
8.85885134e-02 -2.70762473e-01 -1.28682092e-01 -1.41122252e-01
8.65275189e-02 -4.39319670e-01 -1.00072098e+00 1.07484961e+00
1.06074154e+00 5.65190554e-01 2.66465455e-01 -6.68556035e-01
-4.01065320e-01 7.56448209e-01 -2.29303017e-01 3.26824725e-01
-2.45011613e-01 -2.38753427e-02 7.97706187e-01 -1.21738744e+00
-2.38120869e-01 1.39970696e+00 5.12212396e-01 5.49524903e-01
-6.61577225e-01 -9.58083689e-01 6.50253380e-03 5.43449402e-01
-1.31828868e+00 -9.36879456e-01 5.66412687e-01 6.02547899e-02
1.71234536e+00 2.79324263e-01 3.25575024e-01 1.03709304e+00
1.69320405e-01 7.29590118e-01 8.64774048e-01 -6.24307215e-01
4.11520489e-02 -2.65395164e-01 1.23352267e-01 2.97367424e-01
-3.41794610e-01 6.72622144e-01 -7.16225207e-01 1.65042639e-01
3.67699265e-01 -3.86740386e-01 -2.92157441e-01 8.04887295e-01
-1.00754082e+00 3.56161803e-01 1.31733656e-01 3.08011055e-01
-4.05200988e-01 2.21605375e-01 5.56017160e-01 2.04234943e-01
4.84753340e-01 -2.66167372e-02 -6.53069735e-01 -5.26914060e-01
-9.53022897e-01 -1.00952916e-01 6.08531594e-01 8.51526618e-01
5.70750535e-01 9.51306403e-01 -2.85406530e-01 1.18188691e+00
6.52069926e-01 8.93053055e-01 6.65429473e-01 -6.21795356e-01
5.65200210e-01 -1.83687687e-01 -3.19461077e-01 -5.67305863e-01
-3.34450364e-01 -1.06019127e+00 -1.12571311e+00 -2.51549929e-01
-2.95421749e-01 -4.50417757e-01 -1.17635632e+00 1.76689839e+00
1.44862965e-01 6.08025968e-01 5.49988091e-01 6.97896183e-01
7.56957471e-01 1.21317172e+00 -1.18401414e-02 -4.73813504e-01
1.12913942e+00 -1.43367982e+00 -1.40420163e+00 -4.19561088e-01
3.56230915e-01 -1.15296853e+00 5.76456308e-01 4.67095912e-01
-1.14922857e+00 -9.64564025e-01 -1.28103209e+00 1.82299137e-01
-1.42415702e-01 4.94829595e-01 -7.41812587e-02 9.52905774e-01
-1.08665121e+00 4.03973013e-01 -7.78460741e-01 2.90114969e-01
-1.02460146e-01 4.62573051e-01 -7.55954012e-02 7.43891075e-02
-1.60317624e+00 8.07888567e-01 5.16050398e-01 5.57516277e-01
-1.03257668e+00 -4.30967659e-01 -9.47359562e-01 3.60341460e-01
1.07446194e-01 -3.13041389e-01 1.65697920e+00 -5.90088546e-01
-2.03061891e+00 -8.91833007e-02 -6.30052328e-01 -7.50187278e-01
-1.81522444e-01 -3.62433165e-01 -1.25691605e+00 -1.65537298e-01
-5.10283589e-01 2.62744844e-01 1.04238892e+00 -8.30736279e-01
-6.33128643e-01 -8.72980282e-02 -5.38269281e-01 4.80000228e-01
-5.18525541e-01 3.45346063e-01 -6.22446597e-01 -9.56535876e-01
1.70485109e-01 -6.43114328e-01 -3.83318782e-01 -7.34054565e-01
-3.87822688e-01 7.59436488e-02 1.12235761e+00 -1.20792937e+00
1.68672073e+00 -2.39016771e+00 -1.46453068e-01 2.22799063e-01
-1.91496089e-01 9.36335385e-01 -2.32371941e-01 5.30584417e-02
-6.24731369e-02 -8.53166282e-02 -1.34972990e-01 -7.98713386e-01
-3.85675952e-02 2.64768988e-01 -1.10171447e-02 1.72583461e-01
1.31758079e-01 5.39553225e-01 -4.69770610e-01 -7.65996426e-02
5.23001969e-01 1.00283098e+00 -4.23600912e-01 3.56378794e-01
2.45945036e-01 2.56318659e-01 5.94954826e-02 3.32697481e-01
8.37490916e-01 3.77845019e-01 -1.92468371e-02 -3.21928442e-01
-2.44217157e-01 7.17033148e-01 -1.26079488e+00 1.53236949e+00
-8.41242909e-01 7.32173920e-01 3.29761028e-01 -7.98795462e-01
1.09477150e+00 8.55713844e-01 3.02534569e-02 -8.96559477e-01
2.34353513e-01 4.76559490e-01 3.52825761e-01 -2.67997742e-01
5.40741205e-01 -1.38869867e-01 1.85169429e-01 -4.59961733e-03
4.69616950e-01 1.45311624e-01 -3.21989805e-01 -1.19546622e-01
9.58411872e-01 -3.99965018e-01 7.36296177e-02 7.35280067e-02
8.55074883e-01 -1.00589442e+00 7.20977783e-01 6.46776736e-01
-2.39622191e-01 4.85891968e-01 -3.03984851e-01 6.91081658e-02
-1.13345337e+00 -1.02532697e+00 -1.74036205e-01 1.00409734e+00
-1.70306280e-01 -4.85048652e-01 -8.36679697e-01 -9.90268514e-02
-4.17294979e-01 7.80379474e-01 -8.28997046e-03 -3.94051343e-01
-6.94767296e-01 -7.11854339e-01 1.00191319e+00 6.07480943e-01
7.82410324e-01 -7.46026933e-01 1.41332969e-01 5.97854614e-01
-3.42082560e-01 -1.40803564e+00 -7.74426699e-01 6.11871660e-01
-3.61562997e-01 -1.76177636e-01 -5.36639273e-01 -8.24087024e-01
3.92874703e-02 1.93847582e-01 4.48105723e-01 -1.44282460e-01
2.64662355e-01 3.61609040e-03 -5.16170442e-01 -4.88451153e-01
-6.91676259e-01 -1.22032464e-01 4.00096476e-01 3.22634965e-01
1.95648223e-01 -5.07253647e-01 -4.06097978e-01 3.96161318e-01
-7.38836884e-01 -1.21501222e-01 7.96639025e-01 1.18517828e+00
5.37841499e-01 8.77408534e-02 8.82289946e-01 -7.82157257e-02
5.30534565e-01 -2.05340520e-01 -4.96369809e-01 8.21305960e-02
-6.05561018e-01 -2.66278330e-02 5.90665281e-01 -5.16829908e-01
-1.58363283e+00 5.77075295e-02 -9.96027648e-01 -5.45312107e-01
-1.38325080e-01 5.98862886e-01 -5.05808294e-01 6.26097769e-02
3.75908852e-01 5.47227681e-01 -1.42978698e-01 -5.60638428e-01
1.06328793e-01 1.42066956e+00 8.16851735e-01 -5.13003506e-02
5.78076303e-01 -2.19112858e-01 -4.54518765e-01 -1.30595994e+00
-5.65802157e-01 -7.22069204e-01 -2.71331429e-01 -2.70495176e-01
7.86965787e-01 -1.32319725e+00 -4.35054868e-01 9.88294125e-01
-1.47940254e+00 -1.59275129e-01 9.08268336e-03 1.08782291e+00
-3.58453721e-01 5.31753361e-01 -9.00145710e-01 -1.21237874e+00
-6.85162425e-01 -1.49692714e+00 9.17775989e-01 2.08200514e-01
3.11511278e-01 -7.67029762e-01 -2.35860690e-01 2.96347708e-01
9.74620998e-01 -5.65677047e-01 5.53945780e-01 -8.95433366e-01
-1.63065344e-01 -2.15684623e-01 3.64767127e-02 1.24325609e+00
2.80662980e-02 -2.89247036e-01 -1.58640969e+00 -2.39234596e-01
3.31476480e-01 3.47026736e-02 8.92580152e-01 6.84471846e-01
1.13982785e+00 -2.90810496e-01 2.21820809e-02 5.69023907e-01
9.63823795e-01 8.82931888e-01 9.10463572e-01 1.03196800e-02
4.18452471e-01 -1.02651395e-01 4.23491389e-01 4.82709855e-01
1.16815418e-01 7.68411815e-01 2.38200545e-01 -1.09895431e-01
-5.15429974e-01 3.97300757e-02 7.89555490e-01 1.77013326e+00
2.03850061e-01 -5.19209087e-01 -6.46195531e-01 2.65565485e-01
-1.55699718e+00 -8.42916071e-01 -1.20702580e-01 2.08754539e+00
7.32416213e-01 1.74186245e-01 -3.88713509e-01 4.82586443e-01
8.91825616e-01 3.10330540e-01 -4.62163240e-01 -6.24373913e-01
-3.87788773e-01 5.97686768e-01 3.53315771e-01 7.51185238e-01
-1.17880332e+00 8.11857224e-01 5.39010048e+00 1.27775943e+00
-1.17031622e+00 4.24041361e-01 5.13952672e-01 5.51229529e-02
1.38598517e-01 -3.24611813e-01 -9.53670442e-01 2.44501039e-01
1.78287017e+00 -8.23446363e-03 5.67983627e-01 6.30283475e-01
5.19208610e-01 3.44996482e-01 -6.03187978e-01 1.07437956e+00
5.04623801e-02 -9.12396729e-01 -2.44195461e-01 -7.62128010e-02
4.79276091e-01 2.86552489e-01 2.64984190e-01 6.62743032e-01
-1.61078364e-01 -9.59813952e-01 8.29398572e-01 3.78186196e-01
8.79047453e-01 -9.76813734e-01 9.45973575e-01 3.08697283e-01
-1.41619956e+00 -2.62136310e-01 -2.61548012e-01 8.01928993e-03
4.22246695e-01 9.12327170e-01 -8.75134826e-01 6.03297055e-01
6.76737607e-01 3.56470674e-01 8.32522288e-02 1.03109312e+00
-1.16568528e-01 1.13217139e+00 -3.90044838e-01 -6.34110942e-02
2.19699919e-01 1.46180943e-01 7.30565548e-01 1.62423217e+00
5.70062101e-01 2.53857803e-02 -9.20792595e-02 2.65039682e-01
-1.24793172e-01 -2.66310945e-02 -7.29605556e-02 6.65382221e-02
6.96881890e-01 1.17149675e+00 1.16576195e-01 -4.43797380e-01
-5.30019283e-01 7.76967168e-01 -8.99711549e-02 5.81740558e-01
-9.21646774e-01 -7.61086702e-01 7.83679187e-01 -5.24603426e-01
5.03979683e-01 -4.14381117e-01 -1.55515969e-01 -9.34648454e-01
9.79667436e-03 -1.00214279e+00 -1.92251742e-01 -7.34943092e-01
-7.41015553e-01 1.11179936e+00 -4.93341476e-01 -1.06186259e+00
-3.85097682e-01 -6.39681935e-01 -5.46387076e-01 1.36452389e+00
-1.65748012e+00 -8.87613952e-01 4.81025651e-02 5.60396731e-01
8.86924446e-01 -4.32281286e-01 9.95242178e-01 8.46355796e-01
-8.31766248e-01 9.73962069e-01 3.50643575e-01 -2.09762082e-02
5.07425129e-01 -8.81666481e-01 7.78824151e-01 1.26624691e+00
-9.00553353e-03 3.97998810e-01 4.81686324e-01 -5.29136658e-01
-1.12402177e+00 -1.47581315e+00 9.99441981e-01 4.12676007e-01
4.54202801e-01 -4.15823430e-01 -1.03399789e+00 3.97190392e-01
4.39259112e-01 -3.32302712e-02 6.00331426e-01 -1.46190152e-01
-3.28464866e-01 -2.02531248e-01 -7.22980440e-01 4.20348793e-01
7.52255738e-01 -8.19893539e-01 -3.01283568e-01 8.14695507e-02
1.27907789e+00 -6.37204885e-01 -8.07079613e-01 5.52898884e-01
4.35965478e-01 -7.32612073e-01 8.08990657e-01 -1.91088662e-01
-1.73187092e-01 -3.64108622e-01 -7.90875077e-01 -1.55938864e+00
-1.71926975e-01 -8.27827990e-01 -3.69190186e-01 1.19734418e+00
8.67955983e-01 -5.12274742e-01 2.30269521e-01 1.28644090e-02
-1.03926265e+00 -6.31196558e-01 -1.21180809e+00 -9.91562963e-01
-2.22599596e-01 -1.11676240e+00 4.99430478e-01 2.29411140e-01
-2.97190696e-01 3.41288745e-01 -6.89892292e-01 6.30027235e-01
2.58021176e-01 -7.76230633e-01 2.83834308e-01 -6.26097918e-01
-5.60764730e-01 -1.06951334e-01 -1.55974865e-01 -1.33395982e+00
1.45261049e-01 -6.64085388e-01 4.91097629e-01 -1.11012030e+00
-2.91609943e-01 3.14550810e-02 -6.05393350e-01 3.11592221e-01
-2.03084305e-01 -2.05375388e-01 1.29842356e-01 -3.60824704e-01
-3.10559213e-01 9.13017273e-01 9.02806461e-01 -1.37367323e-01
-2.22025290e-01 3.58160973e-01 -1.97200224e-01 5.66527247e-01
9.06563044e-01 -2.84807682e-01 -7.88761675e-02 -4.55259174e-01
-3.16065609e-01 3.44855040e-01 5.74284419e-02 -1.31536543e+00
5.24402678e-01 3.87220681e-01 8.77924487e-02 -7.80414283e-01
9.26864386e-01 -6.36436343e-01 7.42323846e-02 4.64674711e-01
-2.78208017e-01 -1.57265082e-01 5.02048969e-01 6.33218408e-01
-6.83528841e-01 -1.10976115e-01 8.58455300e-01 3.25511009e-01
-5.87865889e-01 2.02466965e-01 -8.49071860e-01 -4.81368423e-01
4.75947529e-01 1.42612591e-01 -1.44227743e-01 -5.82106292e-01
-7.96474218e-01 -8.90493914e-02 -6.62475049e-01 4.33135241e-01
9.86270070e-01 -1.40076268e+00 -7.27427065e-01 4.94876474e-01
-2.80457854e-01 -1.25945479e-01 8.32024992e-01 7.33304977e-01
1.22629538e-01 5.84932625e-01 4.04843867e-01 -4.25960273e-01
-1.30193508e+00 2.44358554e-02 6.80717885e-01 -2.60427117e-01
-9.59548652e-02 1.00117350e+00 -4.84817214e-02 -4.98933613e-01
6.60640419e-01 -4.19117361e-01 4.26424034e-02 -4.65100616e-01
7.35555947e-01 6.04695499e-01 6.92822635e-01 -8.82152915e-01
-2.00534761e-01 1.45899981e-01 -7.69034848e-02 -4.53208506e-01
1.19113481e+00 -3.08586627e-01 1.46926671e-01 1.79592162e-01
1.53748178e+00 -9.91556570e-02 -9.83667493e-01 -4.58262712e-01
-2.75573850e-01 5.95330112e-02 7.07751930e-01 -1.00374639e+00
-1.02856362e+00 1.16344154e+00 9.22710478e-01 -1.01787157e-01
1.57877624e+00 -3.22641402e-01 1.16013503e+00 3.31134021e-01
-1.17538236e-01 -1.18186963e+00 1.10822273e-02 1.08435762e+00
8.98261786e-01 -9.38501298e-01 -6.37385428e-01 -1.74888477e-01
-4.25793111e-01 1.19380558e+00 5.52133501e-01 3.69426459e-01
9.37823653e-01 6.89913452e-01 1.45770162e-01 3.64404768e-01
-8.25376570e-01 -2.86537498e-01 4.20496672e-01 5.68028629e-01
2.54189372e-01 1.77961215e-01 1.52073056e-01 1.07329214e+00
-3.55106473e-01 -3.61824632e-01 2.27975547e-01 5.30351222e-01
-7.01175630e-01 -1.01984239e+00 -5.06726563e-01 3.71661186e-01
-4.96859491e-01 -5.37999451e-01 2.35667005e-01 1.88948944e-01
-3.86239626e-02 1.46415091e+00 7.40144625e-02 -1.06598628e+00
5.91135204e-01 2.28516802e-01 -1.61415964e-01 -3.50692451e-01
-7.53691554e-01 7.79531240e-01 3.10444057e-01 -2.92483121e-01
-2.04750150e-01 -5.82759142e-01 -1.28113818e+00 1.18395416e-02
-8.97072136e-01 1.62346333e-01 1.14183342e+00 1.07585430e+00
2.89121985e-01 1.09901536e+00 8.95889163e-01 -8.28736246e-01
-6.77445292e-01 -1.31822515e+00 -5.61821699e-01 -3.63458991e-01
7.18872190e-01 -2.28351116e-01 -5.19727349e-01 -8.14125612e-02]
|
[14.843406677246094, 6.01363468170166]
|
2d8ba437-a203-4f04-aaf5-e3eba7f44530
|
aspect-sentiment-quad-prediction-as
|
2110.00796
| null |
https://arxiv.org/abs/2110.00796v1
|
https://arxiv.org/pdf/2110.00796v1.pdf
|
Aspect Sentiment Quad Prediction as Paraphrase Generation
|
Aspect-based sentiment analysis (ABSA) has been extensively studied in recent years, which typically involves four fundamental sentiment elements, including the aspect category, aspect term, opinion term, and sentiment polarity. Existing studies usually consider the detection of partial sentiment elements, instead of predicting the four elements in one shot. In this work, we introduce the Aspect Sentiment Quad Prediction (ASQP) task, aiming to jointly detect all sentiment elements in quads for a given opinionated sentence, which can reveal a more comprehensive and complete aspect-level sentiment structure. We further propose a novel \textsc{Paraphrase} modeling paradigm to cast the ASQP task to a paraphrase generation process. On one hand, the generation formulation allows solving ASQP in an end-to-end manner, alleviating the potential error propagation in the pipeline solution. On the other hand, the semantics of the sentiment elements can be fully exploited by learning to generate them in the natural language form. Extensive experiments on benchmark datasets show the superiority of our proposed method and the capacity of cross-task transfer with the proposed unified \textsc{Paraphrase} modeling framework.
|
['Wai Lam', 'Lidong Bing', 'Yifei Yuan', 'Xin Li', 'Yang Deng', 'Wenxuan Zhang']
|
2021-10-02
| null |
https://aclanthology.org/2021.emnlp-main.726
|
https://aclanthology.org/2021.emnlp-main.726.pdf
|
emnlp-2021-11
|
['paraphrase-generation', 'paraphrase-generation']
|
['computer-code', 'natural-language-processing']
|
[ 3.48227233e-01 -3.46537568e-02 2.81718597e-02 -6.86642289e-01
-1.03153968e+00 -7.05087900e-01 5.78842342e-01 3.68440658e-01
-8.74041691e-02 3.78740102e-01 4.64413136e-01 -2.20799059e-01
2.10406423e-01 -7.40514934e-01 -6.83445811e-01 -5.57786524e-01
7.33203232e-01 3.14775914e-01 -2.26849914e-01 -4.74611968e-01
3.81628215e-01 -2.33400568e-01 -1.36012042e+00 7.41364717e-01
9.75872695e-01 1.30118454e+00 4.98230308e-02 3.11752379e-01
-4.48656201e-01 7.95753241e-01 -5.42743206e-01 -1.09178972e+00
-1.18604846e-01 -3.65086854e-01 -6.15021825e-01 2.30249271e-01
2.11123765e-01 1.37347072e-01 4.16314542e-01 1.15683722e+00
3.61564606e-01 -1.41821012e-01 5.55144072e-01 -1.08337355e+00
-5.22213101e-01 5.55028677e-01 -8.19428504e-01 -3.09004873e-01
4.39661980e-01 8.37182850e-02 1.73467076e+00 -1.21233678e+00
2.95822799e-01 1.15281498e+00 5.55542648e-01 2.87666351e-01
-9.32774127e-01 -5.75921476e-01 4.77830112e-01 1.68300465e-01
-9.10025954e-01 -1.95938140e-01 1.04858887e+00 -3.55712682e-01
1.07458317e+00 1.86070442e-01 8.10317874e-01 7.76268244e-01
2.65869886e-01 1.35962021e+00 1.02174807e+00 -1.68073267e-01
1.55289873e-01 3.79186422e-01 4.26123559e-01 4.93709356e-01
1.74042851e-01 -6.71356320e-01 -7.73170650e-01 -1.03155542e-02
-1.49332210e-01 -1.90931512e-03 -6.46046028e-02 -1.97076470e-01
-9.02701139e-01 7.90332615e-01 3.44512880e-01 7.38314493e-03
-5.10817647e-01 -3.02236468e-01 7.45056391e-01 2.28425473e-01
8.55112016e-01 4.50993121e-01 -7.83720911e-01 6.65798709e-02
-7.99337983e-01 4.11942780e-01 8.75920594e-01 9.97421265e-01
1.05296421e+00 -1.01916477e-01 -4.10601318e-01 8.99198890e-01
4.25548464e-01 5.84793508e-01 6.76511884e-01 -3.80538046e-01
8.29369128e-01 1.15295434e+00 8.05796236e-02 -1.06202304e+00
-3.66611242e-01 -5.69088876e-01 -7.36249387e-01 -3.45987856e-01
-2.27883816e-01 -2.23352492e-01 -6.59134328e-01 1.59718049e+00
3.90663505e-01 -9.58089903e-02 3.48291516e-01 6.87986612e-01
8.88702631e-01 7.59609461e-01 6.06518276e-02 -6.97199032e-02
1.84542310e+00 -1.32674325e+00 -5.00367880e-01 -8.37115765e-01
6.97568774e-01 -9.73531783e-01 1.21904790e+00 2.87741482e-01
-8.76705289e-01 -4.51288640e-01 -1.08067203e+00 -1.77103341e-01
-2.85433680e-01 2.86387026e-01 5.79410970e-01 4.64984119e-01
-6.64506435e-01 2.04732507e-01 -4.84884143e-01 1.07121304e-01
2.71478117e-01 7.64972344e-02 -1.35046184e-01 -8.13825279e-02
-1.09258628e+00 5.07078409e-01 2.82425046e-01 3.65889460e-01
-4.28608149e-01 -8.12325239e-01 -1.16036332e+00 2.53438234e-01
4.63472962e-01 -9.73101199e-01 1.40096521e+00 -1.16006708e+00
-1.28139651e+00 7.43144631e-01 -7.66427815e-01 -1.77189305e-01
1.14371315e-01 -4.54931796e-01 -2.51757294e-01 -5.69442138e-02
4.96939868e-01 2.74250597e-01 8.22837174e-01 -1.07832551e+00
-8.25100482e-01 -7.08556414e-01 4.12847310e-01 5.32293499e-01
-6.19439542e-01 -5.00011118e-03 -6.45238280e-01 -7.07901478e-01
-8.50042179e-02 -9.67499197e-01 -1.41265213e-01 -4.50486809e-01
-5.13146520e-01 -4.32318062e-01 3.64131600e-01 -5.34288585e-01
1.10852230e+00 -1.97035503e+00 2.61730999e-01 -7.66052026e-03
-8.49682689e-02 1.42031610e-01 -2.19804108e-01 4.06913757e-01
-5.81283905e-02 -8.33796635e-02 -4.63361144e-01 -8.11750174e-01
1.05935685e-01 -9.15668905e-02 -7.20295012e-01 -1.33857921e-01
4.70908672e-01 1.05611348e+00 -8.77969563e-01 -3.59483212e-01
-2.20095813e-01 2.48824179e-01 -6.14668071e-01 2.15031847e-01
-4.23369050e-01 2.61364430e-01 -6.76448405e-01 5.71954906e-01
6.65949464e-01 -2.75164336e-01 -3.51990922e-03 -4.18544710e-01
1.47213832e-01 5.69662631e-01 -7.44507492e-01 1.55905282e+00
-1.01934290e+00 2.60559857e-01 -9.27896723e-02 -8.56107354e-01
9.84536707e-01 2.57406384e-01 3.50385875e-01 -7.16494083e-01
1.31822079e-01 3.23296100e-01 -1.80560142e-01 -3.37463289e-01
8.48520100e-01 -6.00892365e-01 -4.56961989e-01 6.78054035e-01
-2.89082830e-03 -3.67208242e-01 4.77480829e-01 2.22436756e-01
5.96214712e-01 1.35082111e-01 2.97436446e-01 -9.99393314e-02
9.94148672e-01 2.51744926e-01 5.51082134e-01 3.04722041e-01
2.32632950e-01 5.60544431e-01 8.54471385e-01 -2.13840291e-01
-6.70882940e-01 -5.00306427e-01 2.33123675e-01 1.04133558e+00
1.62835747e-01 -7.47382641e-01 -6.40138865e-01 -8.06741118e-01
-2.42758185e-01 8.02972615e-01 -6.33436978e-01 -1.97797716e-01
-4.37976003e-01 -7.97070742e-01 9.19380486e-02 5.57906210e-01
3.23252022e-01 -1.20482266e+00 -2.69963354e-01 1.48714423e-01
-4.39729899e-01 -1.21662104e+00 -6.12820566e-01 -7.31920674e-02
-6.48943186e-01 -8.44482958e-01 -6.90450072e-01 -8.64145279e-01
5.74070215e-01 3.94646287e-01 1.24161053e+00 -2.41954446e-01
1.79337084e-01 8.83129239e-02 -6.46616042e-01 -5.30789852e-01
-1.65049732e-01 2.40942165e-01 -3.00903767e-01 5.11752903e-01
7.11728036e-01 -4.96972203e-01 -8.42020750e-01 1.07079912e-02
-8.81487370e-01 3.71497184e-01 7.95175493e-01 7.41249561e-01
8.17057371e-01 -1.67404294e-01 7.17415273e-01 -1.48095465e+00
8.96511495e-01 -6.58406377e-01 -2.49659419e-01 2.74793714e-01
-6.97110534e-01 -5.20628169e-02 1.07182217e+00 8.45472217e-02
-1.35134709e+00 1.84713621e-02 -4.21117187e-01 -1.35241792e-01
1.94098786e-01 1.00069058e+00 -3.63800019e-01 4.89404738e-01
1.36314660e-01 7.65220702e-01 -1.85951486e-01 -3.43766481e-01
4.15446103e-01 7.19643533e-01 1.43372715e-01 -5.10642350e-01
4.60422635e-01 5.30740082e-01 -1.90944821e-01 -5.74230433e-01
-1.67743957e+00 -7.56520212e-01 -3.60331953e-01 -5.66195175e-02
4.63325828e-01 -1.30175495e+00 -4.01560873e-01 5.08021891e-01
-1.30511689e+00 2.93737054e-01 -2.35902473e-01 -5.85468635e-02
-4.41116065e-01 4.45465833e-01 -4.26101297e-01 -5.78293920e-01
-1.09778404e+00 -1.23920894e+00 1.56054974e+00 3.04211438e-01
-3.74726593e-01 -8.55295599e-01 2.25069612e-01 7.50300586e-01
1.37021422e-01 -1.26937926e-01 1.07936740e+00 -7.83624113e-01
-1.96211308e-01 -4.60486501e-01 -2.71690726e-01 5.91589987e-01
1.35689095e-01 -3.52718621e-01 -9.30753350e-01 -2.11281314e-01
3.60105187e-01 -3.87697935e-01 9.19317842e-01 1.64319724e-02
8.13277781e-01 -3.77670884e-01 9.70002711e-02 3.88398260e-01
1.34511697e+00 -5.48729561e-02 2.44722739e-01 3.56030554e-01
8.44886720e-01 8.56820822e-01 1.05244672e+00 3.91150653e-01
8.39488626e-01 4.28603798e-01 1.67293340e-01 7.98549056e-02
1.02662578e-01 -2.10269570e-01 6.18583858e-01 1.22425222e+00
4.76160765e-01 -1.57196805e-01 -6.49457097e-01 7.30439007e-01
-1.90603471e+00 -5.95766544e-01 -8.62575024e-02 1.59291685e+00
8.52425218e-01 2.38717780e-01 -1.61893144e-01 -7.01128170e-02
4.73519504e-01 5.28242230e-01 -7.60375857e-01 -6.42277122e-01
-8.55142027e-02 3.67184505e-02 -2.56802469e-01 2.27688044e-01
-9.92153704e-01 9.45778668e-01 4.21274233e+00 8.38188708e-01
-1.10561764e+00 -1.52255641e-03 6.25381649e-01 -2.89830938e-02
-8.01859260e-01 1.93653688e-01 -9.25211012e-01 4.14238095e-01
5.10687888e-01 -5.63537002e-01 -2.88766641e-02 1.15098965e+00
1.63267106e-01 6.73789755e-02 -1.02893054e+00 7.34222770e-01
5.42173803e-01 -9.72734511e-01 7.19764650e-01 -3.96109223e-01
9.10132527e-01 -1.12637207e-01 1.32726312e-01 5.63077211e-01
-1.55154720e-01 -4.93093908e-01 8.56439769e-01 3.09590429e-01
5.02987325e-01 -9.01705265e-01 9.23200607e-01 2.88902938e-01
-1.46553349e+00 -5.66320401e-03 -3.65687370e-01 2.36830115e-02
4.28551584e-01 7.14510024e-01 -5.56531250e-01 9.23897982e-01
4.21207488e-01 1.23161364e+00 -7.02489495e-01 6.94171131e-01
-5.48031688e-01 4.18390632e-01 1.43419325e-01 -3.03766191e-01
4.07016873e-01 -4.59839523e-01 5.26980400e-01 1.23364842e+00
2.22598404e-01 -2.09805071e-01 -1.84190981e-02 5.97576261e-01
-2.14391053e-01 5.59999704e-01 -3.11589003e-01 -1.05760738e-01
5.76647557e-02 1.67126310e+00 -6.03304267e-01 -3.86328876e-01
-6.68992996e-01 1.00008345e+00 3.41543287e-01 1.94091514e-01
-4.90383804e-01 -5.04012048e-01 5.71872652e-01 -2.92977035e-01
5.84985614e-01 2.90971309e-01 -5.69458008e-01 -1.60330224e+00
6.37507915e-01 -1.12455976e+00 2.90612042e-01 -8.52242172e-01
-1.37862098e+00 7.95130849e-01 -3.69672716e-01 -1.41061938e+00
-3.54840487e-01 -5.75280309e-01 -9.47610378e-01 9.95565474e-01
-1.81122434e+00 -1.60035157e+00 -1.26589611e-01 4.64413881e-01
1.00828767e+00 -9.41669196e-02 6.92119360e-01 1.32673323e-01
-6.46585524e-01 4.82098758e-01 -1.14792526e-01 -1.06044523e-01
6.56121314e-01 -1.30610538e+00 5.73421597e-01 9.69848514e-01
1.67418998e-02 8.03884685e-01 6.49171889e-01 -5.30763924e-01
-1.39253402e+00 -1.34075153e+00 1.14546752e+00 -3.33837181e-01
9.39371049e-01 -3.59132707e-01 -8.76582980e-01 5.42047620e-01
2.08298802e-01 -6.63903594e-01 9.52052653e-01 3.64301592e-01
-5.80131114e-01 -3.87109816e-01 -4.94802892e-01 7.36480832e-01
4.93961394e-01 -7.35887289e-01 -6.76864207e-01 2.26729140e-01
9.55531597e-01 -1.16790406e-01 -6.01187527e-01 4.89211768e-01
6.03921294e-01 -8.81434262e-01 5.71747482e-01 -4.28325206e-01
1.16081548e+00 -2.65192866e-01 -4.21414198e-03 -1.47707081e+00
4.26126942e-02 -2.45372713e-01 -1.22673996e-01 1.56289673e+00
7.98103452e-01 -3.79747659e-01 7.68546820e-01 4.48014557e-01
-2.34682307e-01 -1.15886962e+00 -6.56262517e-01 -1.04728244e-01
-3.71062197e-02 -6.91335618e-01 7.07078993e-01 5.01286089e-01
1.97611615e-01 1.24165297e+00 -4.01084989e-01 -6.60491502e-03
3.68108660e-01 1.07878780e+00 7.64639914e-01 -7.99222231e-01
-4.89427179e-01 -4.90260631e-01 4.81691137e-02 -1.21618021e+00
3.20838600e-01 -9.50609922e-01 1.96570456e-01 -1.70406365e+00
5.22134244e-01 -2.35473603e-01 -3.24508339e-01 3.48041952e-01
-7.28744507e-01 1.70271531e-01 1.93748772e-01 1.66009858e-01
-8.73003662e-01 8.59618068e-01 1.22999120e+00 -4.07495469e-01
4.71657924e-02 3.42433870e-01 -1.40260386e+00 9.06620681e-01
6.22853756e-01 -4.10023123e-01 -5.60829520e-01 -6.30199194e-01
1.01847923e+00 -1.34603083e-02 -5.92706352e-02 -4.51689631e-01
2.91760862e-01 1.61462486e-01 -9.51433331e-02 -8.07529092e-01
4.31975871e-01 -6.54418528e-01 -4.55458939e-01 1.66566581e-01
-2.46486709e-01 9.77810994e-02 2.52410918e-01 7.24960327e-01
-7.15853930e-01 -3.40258658e-01 3.85532618e-01 -7.59952143e-02
-4.95457143e-01 2.35216498e-01 -1.15464389e-01 1.26317188e-01
7.69052029e-01 1.33432999e-01 -1.97952300e-01 -3.91709030e-01
-3.13984364e-01 5.34350038e-01 2.72993147e-01 6.03733718e-01
6.23763382e-01 -9.85336185e-01 -7.83955812e-01 1.78923443e-01
5.27431071e-01 2.23728329e-01 5.99344075e-01 7.38246262e-01
-1.03045769e-01 4.55399692e-01 8.20480734e-02 -3.65563422e-01
-1.12163591e+00 4.43232000e-01 -2.44436972e-02 -7.83394933e-01
-1.89967096e-01 7.37545013e-01 5.58727086e-01 -5.31403422e-01
-2.56737828e-01 8.67613405e-02 -7.26040304e-01 5.10785520e-01
5.88337839e-01 -9.85604972e-02 2.46068805e-01 -8.07310164e-01
-2.77371496e-01 7.84910142e-01 -5.82855761e-01 2.44185887e-02
1.34586120e+00 -2.77734041e-01 -4.99874055e-01 4.55083251e-01
1.18705034e+00 1.33925050e-01 -7.90938497e-01 -3.86194766e-01
1.77532006e-02 -4.95849028e-02 -2.54948914e-01 -6.50049090e-01
-9.91577148e-01 1.06368363e+00 -1.27836064e-01 1.93568081e-01
1.32774937e+00 -4.37133089e-02 1.22921097e+00 4.59771395e-01
1.22852311e-01 -7.83271313e-01 8.14340636e-02 6.68571234e-01
8.45161200e-01 -1.21773458e+00 -1.20330483e-01 -4.20481473e-01
-1.24819732e+00 8.97927761e-01 6.65928781e-01 -1.68869883e-01
5.63503087e-01 -1.34986654e-01 1.18562236e-01 -3.07365865e-01
-9.52112198e-01 -7.52840042e-02 3.62807989e-01 2.61433329e-02
4.78339761e-01 1.42764952e-02 -5.00257313e-01 1.26922822e+00
-4.62903589e-01 -2.82133818e-01 3.86226237e-01 7.75714397e-01
-2.08412901e-01 -1.09310937e+00 2.51818419e-01 5.68682432e-01
-5.94938874e-01 -5.36675215e-01 -4.15320426e-01 4.61827219e-02
-1.24015950e-01 9.41568196e-01 -1.06080711e-01 -3.86909276e-01
7.31340647e-01 1.07812852e-01 -9.50232819e-02 -9.45143878e-01
-8.66451025e-01 6.07542247e-02 1.51704684e-01 -3.77305657e-01
-4.76164848e-01 -5.52865386e-01 -9.54805255e-01 1.67247742e-01
-2.62306660e-01 4.37440097e-01 6.88127875e-01 1.23352861e+00
6.02338076e-01 5.43373704e-01 9.18587625e-01 -5.79333425e-01
-4.14380074e-01 -9.96755540e-01 -4.43674862e-01 6.94401979e-01
2.00630993e-01 -8.22851360e-02 -1.93857193e-01 1.05552658e-01]
|
[11.482150077819824, 6.6427812576293945]
|
2d427244-9951-4bef-8046-2937c817f6ab
|
neural-document-embeddings-for-intensive-care
|
1612.00467
| null |
http://arxiv.org/abs/1612.00467v1
|
http://arxiv.org/pdf/1612.00467v1.pdf
|
Neural Document Embeddings for Intensive Care Patient Mortality Prediction
|
We present an automatic mortality prediction scheme based on the unstructured
textual content of clinical notes. Proposing a convolutional document embedding
approach, our empirical investigation using the MIMIC-III intensive care
database shows significant performance gains compared to previously employed
methods such as latent topic distributions or generic doc2vec embeddings. These
improvements are especially pronounced for the difficult problem of
post-discharge mortality prediction.
|
['Florian Schmidt', 'Stephanie L. Hyland', 'Paulina Grnarova', 'Carsten Eickhoff']
|
2016-12-01
| null | null | null | null |
['document-embedding']
|
['methodology']
|
[-1.27771690e-01 3.19557816e-01 -1.41338944e-01 -4.33364175e-02
-9.85601783e-01 -6.69579506e-02 4.77038831e-01 1.11886966e+00
-7.06171274e-01 7.33313262e-01 1.15423071e+00 -4.18361664e-01
-6.56007588e-01 -5.95619500e-01 3.22328240e-01 -7.68556356e-01
-7.57477760e-01 8.86668026e-01 -3.98258895e-01 1.78885698e-01
1.41182184e-01 4.20778632e-01 -7.46420979e-01 3.20007265e-01
3.39499950e-01 9.20080960e-01 -5.27451396e-01 8.74395967e-01
-5.03687486e-02 1.02350283e+00 -5.49551666e-01 -4.78882134e-01
-9.83614773e-02 -6.69956207e-02 -8.67637634e-01 -1.92794144e-01
2.82512587e-02 -5.10789514e-01 -7.88107514e-01 3.53839785e-01
1.12627375e+00 1.25845186e-02 1.23949993e+00 -6.61623955e-01
-5.06130099e-01 5.54679871e-01 1.46670446e-01 5.11900723e-01
-2.43651737e-02 -1.57568112e-01 1.13957405e+00 -7.37463653e-01
5.71748614e-01 8.91465247e-01 9.64468479e-01 4.93155599e-01
-1.20720840e+00 -5.40418178e-02 -4.36273366e-01 1.49694011e-01
-1.33055341e+00 -7.25094602e-02 5.46781898e-01 -7.43910372e-01
1.14029849e+00 1.74205616e-01 2.98537344e-01 1.56693435e+00
9.40136135e-01 4.58684713e-01 4.98521447e-01 -3.16115141e-01
1.39416650e-01 2.37432972e-01 3.26404303e-01 7.10330606e-01
4.78540778e-01 -9.29117054e-02 -2.14304924e-01 -1.09373069e+00
3.58849436e-01 5.89991212e-01 -4.73399639e-01 -4.21837360e-01
-1.39059019e+00 1.30183804e+00 8.63128901e-02 2.79574275e-01
-7.37954915e-01 1.00089004e-02 1.16736543e+00 1.95573106e-01
1.00117838e+00 5.03770471e-01 -7.02707112e-01 -2.85997033e-01
-1.02789366e+00 -2.54090019e-02 1.08494318e+00 4.21976268e-01
-3.40491205e-01 -1.24826305e-01 -5.56611657e-01 7.75503635e-01
3.28084081e-01 -6.09834790e-02 7.31526256e-01 -5.45893788e-01
5.49688101e-01 2.88879842e-01 -9.91546884e-02 -1.14044213e+00
-3.79969984e-01 -4.93130565e-01 -1.15893018e+00 -1.70231104e-01
-5.07825054e-02 -5.17229199e-01 -5.68632603e-01 9.95341241e-01
-2.21122373e-02 1.07843526e-01 5.87128639e-01 2.52850652e-01
9.27283227e-01 4.65296358e-01 5.83600640e-01 -5.04574239e-01
1.27742743e+00 -8.25130284e-01 -9.85382497e-01 4.71143246e-01
1.07206750e+00 -4.79867935e-01 2.04583719e-01 2.52534926e-01
-6.45206273e-01 -4.96372022e-02 -5.10862172e-01 6.01399988e-02
-4.10422623e-01 3.93167853e-01 5.56677699e-01 4.31725949e-01
-1.28064167e+00 6.02042198e-01 -9.07619059e-01 -4.72231984e-01
7.10630774e-01 3.99395853e-01 -5.69510639e-01 7.76770618e-03
-1.07315636e+00 7.93025255e-01 2.96133071e-01 -3.33041251e-01
-7.62303889e-01 -9.32652533e-01 -9.80208099e-01 6.51124775e-01
-3.11863989e-01 -1.02736151e+00 9.31663334e-01 -8.84289518e-02
-1.27680886e+00 6.32407963e-01 2.70798594e-01 -7.00404823e-01
3.90976280e-01 -3.48588467e-01 -2.58326769e-01 4.96138901e-01
-3.36693853e-01 2.81488299e-01 3.45770389e-01 -6.64864659e-01
-2.33215854e-01 -1.25462413e-01 -3.70676875e-01 -1.26045458e-02
-1.07972145e+00 9.29012150e-02 1.14566445e-01 -9.17077363e-01
-5.51585019e-01 -6.71784461e-01 -6.90366507e-01 4.99679223e-02
-1.96402863e-01 -5.51206946e-01 5.79118192e-01 -9.27567244e-01
1.43952441e+00 -2.25848222e+00 2.35346258e-01 -1.16704658e-01
7.40560532e-01 1.64757192e-01 1.66884571e-01 1.05142581e+00
-2.67412812e-01 1.20839223e-01 -1.85110509e-01 -5.34005642e-01
-1.14901952e-01 2.72123367e-02 -3.38268071e-01 5.62390149e-01
1.59482434e-01 6.65679455e-01 -7.31266677e-01 -8.24315429e-01
2.41988048e-01 9.54356849e-01 -8.67265522e-01 4.08214808e-01
3.59118700e-01 -5.67619540e-02 -4.70720232e-01 1.89583734e-01
5.17560095e-02 -4.70680535e-01 4.35917377e-01 1.47746518e-01
1.04378201e-01 3.09760690e-01 -2.80820996e-01 1.60069144e+00
-4.41232890e-01 8.67283404e-01 -4.98341382e-01 -9.62297201e-01
6.75484657e-01 1.13994443e+00 1.16185176e+00 2.14708477e-01
4.47642982e-01 -1.78422287e-01 -2.26793960e-01 -5.65589607e-01
2.49440655e-01 -3.83803010e-01 -1.35372683e-01 1.94957569e-01
3.59636158e-01 2.93195575e-01 -3.91198039e-01 4.93183285e-01
1.56614542e+00 -5.27699172e-01 6.68040395e-01 -5.25924563e-01
2.27891490e-01 3.61743243e-03 3.11536551e-01 6.26317501e-01
-3.97253573e-01 9.44574177e-01 8.90301883e-01 -7.75847852e-01
-6.75870836e-01 -8.34094465e-01 -6.75714731e-01 5.23996711e-01
-7.23085284e-01 -8.48215818e-01 -3.42937052e-01 -1.14212072e+00
1.46144271e-01 3.69055152e-01 -1.25257087e+00 -1.05005808e-01
-1.90074682e-01 -9.89202738e-01 5.79533577e-01 8.36976409e-01
-4.40782905e-01 -1.10630012e+00 -6.54195607e-01 6.47417843e-01
1.39766589e-01 -7.74329901e-01 -3.00825328e-01 4.39278990e-01
-1.34939516e+00 -1.08695436e+00 -1.03624821e+00 -8.48459423e-01
3.32762331e-01 -5.92594743e-01 1.11390591e+00 -1.35989517e-01
-9.09815192e-01 7.67332017e-01 -4.41431195e-01 -3.86480689e-01
-2.02606454e-01 5.59725240e-02 2.34756514e-01 -2.50913113e-01
6.07212484e-01 -5.13184845e-01 -1.00956166e+00 -5.81065118e-01
-9.33140755e-01 -4.52346087e-01 4.91893560e-01 1.20136726e+00
2.74017990e-01 -1.59896374e-01 5.18361390e-01 -1.19310927e+00
1.03013408e+00 -1.12953973e+00 2.44262233e-01 -1.53604150e-02
-1.27355337e+00 -1.00191504e-01 5.02544582e-01 -1.98619068e-01
-5.76469541e-01 -3.47595245e-01 -1.88912436e-01 -5.16245067e-01
-4.38520074e-01 7.57566392e-01 5.84394872e-01 6.00424647e-01
5.17733276e-01 9.98446792e-02 -2.53154971e-02 -6.06836975e-01
-8.62793159e-03 1.03811336e+00 2.49792282e-02 -1.16152890e-01
1.52492806e-01 2.69716233e-01 -3.79645675e-02 -7.64382958e-01
-3.38162124e-01 -8.72933686e-01 -7.04516351e-01 4.32836890e-01
1.34228432e+00 -8.48470032e-01 -5.14799476e-01 -1.62735358e-01
-1.24451792e+00 1.39902964e-01 -5.05811691e-01 8.90862346e-01
-5.73743105e-01 5.16821086e-01 -1.13100541e+00 -4.29715127e-01
-7.67791629e-01 -1.01898193e+00 8.93064022e-01 -4.36129898e-01
-5.44755995e-01 -1.67053092e+00 8.45509350e-01 -1.53305471e-01
5.27775764e-01 3.91311824e-01 1.57138681e+00 -1.11634994e+00
2.36396834e-01 -5.85142732e-01 -1.86370999e-01 3.40195477e-01
4.69967723e-01 -3.19548361e-02 -7.57191002e-01 -3.22765976e-01
-3.05201054e-01 -1.49227113e-01 1.11057484e+00 4.48611140e-01
1.31574893e+00 -4.11910683e-01 -5.09468615e-01 6.49350941e-01
1.62480080e+00 8.04286730e-03 5.03199577e-01 2.76146501e-01
5.24094760e-01 3.90923887e-01 -9.16848481e-02 9.98088658e-01
3.23459417e-01 2.92469829e-01 1.34176046e-01 -1.00105211e-01
1.11520380e-01 1.92015231e-01 -1.12405084e-01 1.33232641e+00
-7.44064599e-02 -5.63524723e-01 -1.34388614e+00 9.75599110e-01
-1.65085340e+00 -7.76505947e-01 8.77536163e-02 1.62238967e+00
9.74065840e-01 -1.33065090e-01 -3.83886456e-01 1.65946439e-01
-2.64482312e-02 2.67770857e-01 9.62474942e-02 -8.90550852e-01
3.28119844e-01 6.61429107e-01 3.76953602e-01 9.63306502e-02
-1.40508938e+00 3.45529109e-01 7.42547369e+00 4.74470198e-01
-7.86828101e-01 3.21522087e-01 6.18084550e-01 -8.30911398e-02
-8.07726383e-02 -4.91478175e-01 -3.67333651e-01 3.95093858e-01
1.41519296e+00 -1.08301148e-01 -7.10972130e-01 8.37077916e-01
4.85346951e-02 3.87938410e-01 -1.07130635e+00 1.01312566e+00
2.90248871e-01 -1.51290298e+00 2.67776698e-01 1.68772727e-01
5.89172661e-01 3.84495370e-02 2.63866693e-01 3.74341249e-01
1.38234451e-01 -1.10444903e+00 -2.61917025e-01 6.80605710e-01
1.08335066e+00 -7.20302165e-01 1.42987657e+00 -2.45795488e-01
-5.50234854e-01 -5.98380454e-02 -3.98729652e-01 2.87340492e-01
-3.45891863e-02 4.74462450e-01 -1.14247572e+00 6.27441108e-01
6.61792457e-01 9.26825464e-01 -4.36824173e-01 1.16485250e+00
3.91591400e-01 1.13863313e+00 2.26108626e-01 3.78690630e-01
2.85369396e-01 2.07685605e-01 3.84998798e-01 1.82665932e+00
1.78545713e-01 2.72808373e-01 -1.07095547e-01 1.34691387e-01
-1.67865962e-01 6.83551252e-01 -1.11261868e+00 -4.08441365e-01
-4.20119047e-01 1.07913399e+00 -5.36372721e-01 -5.85117757e-01
-4.30897027e-01 1.01073039e+00 2.78695583e-01 2.12331459e-01
-3.98462504e-01 -4.91235167e-01 8.19004536e-01 1.00835994e-01
4.96395826e-01 -1.66147903e-01 -1.18081108e-01 -1.11812186e+00
-6.98137939e-01 -6.83150232e-01 7.61136353e-01 -1.41282797e-01
-1.52232540e+00 9.88885820e-01 -1.38771996e-01 -1.28522182e+00
-2.51569271e-01 -7.85890639e-01 -6.11240983e-01 5.93331635e-01
-1.36114419e+00 -6.24011159e-01 1.11460015e-01 2.65026361e-01
6.74049377e-01 -5.66491187e-01 1.85146070e+00 5.07623911e-01
-3.83972973e-01 7.59942472e-01 7.71997929e-01 3.78338933e-01
7.74497807e-01 -1.41951966e+00 -1.25640377e-01 -6.05421141e-02
-1.46300077e-01 7.38263309e-01 5.59653640e-01 -4.71837372e-01
-8.56170118e-01 -1.17280734e+00 1.35808945e+00 -8.93781304e-01
5.74762940e-01 1.08271338e-01 -7.12984085e-01 5.16538382e-01
5.90899169e-01 -1.02056831e-01 1.57731974e+00 1.63473949e-01
-2.20497832e-01 3.42907131e-01 -8.70071232e-01 3.93357217e-01
4.82018024e-01 -5.06505907e-01 -8.79563391e-01 5.64095557e-01
6.60442948e-01 1.59005597e-01 -1.45270979e+00 6.20948911e-01
4.12030131e-01 -3.08881164e-01 9.97294843e-01 -1.24893045e+00
1.13421488e+00 5.21372080e-01 -1.38100132e-01 -1.24291682e+00
-5.55571735e-01 -2.97544390e-01 -2.96395451e-01 7.46472001e-01
3.40293735e-01 -4.93008733e-01 5.36305308e-01 3.78830194e-01
1.01391869e-02 -1.05421734e+00 -8.52783382e-01 -7.97884911e-02
3.29700440e-01 1.60269439e-02 -4.65651415e-02 1.36568964e+00
1.76238731e-01 9.39814653e-03 -3.78729522e-01 5.87497093e-02
4.28902686e-01 -1.63445443e-01 1.94506973e-01 -1.35436833e+00
-2.08020255e-01 -2.89995193e-01 -8.53533447e-01 -3.36479306e-01
1.60054326e-01 -9.13012624e-01 -1.91884711e-01 -2.01054668e+00
5.42489767e-01 -2.32519895e-01 -1.26040077e+00 2.99116015e-01
-2.39977092e-01 1.56444460e-01 -3.13726276e-01 2.01836377e-01
-7.11445689e-01 8.83484304e-01 6.61202431e-01 -2.72236019e-01
3.93795855e-02 -2.80070275e-01 -3.82555783e-01 4.09313917e-01
7.47502983e-01 -1.04262793e+00 -1.16784431e-01 -3.94757658e-01
-1.56550512e-01 5.20495117e-01 5.80911189e-02 -9.09002125e-01
-5.17608859e-02 2.45339260e-01 2.30350956e-01 -4.12937373e-01
1.80054873e-01 -8.75732839e-01 -2.58297741e-01 6.41208768e-01
-7.30471611e-01 2.72753626e-01 4.95961577e-01 9.56175327e-01
-6.36481583e-01 3.51801105e-02 8.43354315e-02 1.30555585e-01
-6.41198158e-02 5.77008426e-01 -1.06904519e+00 -6.53289929e-02
8.12203765e-01 1.67730898e-01 -7.64058381e-02 -2.86971688e-01
-1.08367348e+00 -1.53406411e-02 -1.01969488e-01 3.38979810e-01
9.24530566e-01 -1.22210085e+00 -9.75345492e-01 1.12068847e-01
4.58509654e-01 -3.84945840e-01 2.83964783e-01 1.19990087e+00
-8.12235057e-01 7.98992157e-01 -4.21306565e-02 -2.13310704e-01
-1.41839254e+00 7.12890387e-01 9.45036486e-02 -6.27752483e-01
-1.17493558e+00 8.60714614e-01 4.08020288e-01 3.97513732e-02
4.53154892e-01 -3.22242677e-01 -7.85729051e-01 3.69261831e-01
3.80479455e-01 1.09177023e-01 1.09557128e-02 -3.16534281e-01
-4.55751777e-01 1.08516656e-01 -2.09245190e-01 1.33130223e-01
1.80558777e+00 2.32558295e-01 -9.77023542e-02 5.50108373e-01
1.62204242e+00 1.53410230e-02 -5.54735363e-01 -1.38486907e-01
9.59422216e-02 -9.51866508e-02 3.62567037e-01 -4.64914382e-01
-8.74228179e-01 1.08401632e+00 1.02761328e+00 1.56947449e-01
7.74891496e-01 -1.27666384e-01 7.27732122e-01 3.61390024e-01
-9.14161652e-02 -5.23311496e-01 6.48654485e-03 5.18977642e-01
6.71685994e-01 -1.06987906e+00 -6.27748519e-02 2.00308993e-01
-8.47531319e-01 1.38745022e+00 -9.25584286e-02 -1.68761507e-01
1.33346343e+00 1.05578892e-01 3.73136818e-01 -5.32343090e-01
-1.14991307e+00 1.50957361e-01 3.79382104e-01 2.85984039e-01
8.71009827e-01 2.44695976e-01 -6.11154914e-01 6.82957470e-01
2.74340361e-01 3.17876786e-02 4.84675229e-01 9.40448046e-01
-6.15454875e-02 -1.11453390e+00 3.18374276e-01 8.26191723e-01
-1.31560576e+00 -6.03818059e-01 -1.39054894e-01 4.40229207e-01
-5.59172966e-02 5.68038762e-01 -2.82041151e-02 -3.12208652e-01
7.59459799e-05 6.56430006e-01 1.01022847e-01 -1.02333641e+00
-7.77300835e-01 -1.39467835e-01 8.26731920e-02 -4.14694935e-01
-2.92906970e-01 -5.73903561e-01 -8.51794124e-01 -9.68735963e-02
-1.52625158e-01 5.71481526e-01 3.91217679e-01 4.43969637e-01
6.13759935e-01 9.88990784e-01 4.71755147e-01 -3.29383671e-01
-7.38491237e-01 -1.25207460e+00 -5.25775313e-01 5.02470851e-01
5.56845367e-01 -4.20279413e-01 -5.69488406e-01 3.57324898e-01]
|
[7.974947452545166, 6.8873114585876465]
|
be6a8e4e-e65f-4f94-896e-0c16d3e1c498
|
commonsense-aware-prompting-for-controllable
|
2302.01441
| null |
https://arxiv.org/abs/2302.01441v1
|
https://arxiv.org/pdf/2302.01441v1.pdf
|
Commonsense-Aware Prompting for Controllable Empathetic Dialogue Generation
|
Improving the emotional awareness of pre-trained language models is an emerging important problem for dialogue generation tasks. Although prior studies have introduced methods to improve empathetic dialogue generation, few have discussed how to incorporate commonsense knowledge into pre-trained language models for controllable dialogue generation. In this study, we propose a novel framework that improves empathetic dialogue generation using pre-trained language models by 1) incorporating commonsense knowledge through prompt verbalization, and 2) controlling dialogue generation using a strategy-driven future discriminator. We conducted experiments to reveal that both the incorporation of social commonsense knowledge and enforcement of control over generation help to improve generation performance. Finally, we discuss the implications of our study for future research.
|
['Halil Kilicoglu', 'Yiren Liu']
|
2023-02-02
| null | null | null | null |
['dialogue-generation', 'dialogue-generation']
|
['natural-language-processing', 'speech']
|
[ 1.03453584e-01 9.19822156e-01 -9.18055773e-02 -3.54830712e-01
-3.53533834e-01 -4.82901126e-01 1.03484654e+00 2.45552063e-02
-2.56144524e-01 1.06598568e+00 9.72550750e-01 -3.46864536e-02
4.23850745e-01 -9.15360808e-01 -4.01022956e-02 -8.96626860e-02
3.77758831e-01 4.49640632e-01 -3.23123246e-01 -9.61909771e-01
2.90321678e-01 1.66246623e-01 -8.33013535e-01 5.97044826e-01
1.24327612e+00 3.66095275e-01 -1.86256632e-01 7.75035441e-01
-1.39176160e-01 1.76389134e+00 -1.13863242e+00 -7.26819098e-01
-1.48054153e-01 -1.23494017e+00 -1.29580426e+00 -1.50888368e-01
-3.26001972e-01 -4.76013929e-01 -1.31213618e-02 7.68633127e-01
6.85234368e-01 6.00495994e-01 7.35702455e-01 -1.20754302e+00
-1.09268677e+00 1.40844417e+00 1.11942992e-01 -2.10628752e-02
8.12862635e-01 5.33455372e-01 9.20764327e-01 -4.00618047e-01
7.92572975e-01 1.65785587e+00 5.94584167e-01 1.35401225e+00
-1.22563708e+00 -8.18203986e-01 1.82434171e-02 9.01511461e-02
-6.69326007e-01 -6.91260457e-01 1.20128763e+00 -4.11071658e-01
1.04768968e+00 4.70139123e-02 8.66937935e-01 1.62041640e+00
1.31828368e-01 6.55394733e-01 1.21920538e+00 -6.29694045e-01
2.20303461e-01 6.10323548e-01 1.51929900e-01 5.64890981e-01
-2.03360245e-01 2.37409666e-01 -7.74570882e-01 -4.19305086e-01
5.94692945e-01 -8.98518443e-01 -1.86604813e-01 2.24047437e-01
-1.16914368e+00 1.61966002e+00 1.99218780e-01 3.15707237e-01
-4.70206141e-01 -2.87046470e-02 6.69502497e-01 3.19884807e-01
6.68064237e-01 1.47909462e+00 4.55154218e-02 -5.83148479e-01
-4.43392158e-01 2.67886043e-01 1.20490324e+00 6.59997940e-01
3.65864664e-01 4.62625086e-01 -6.95103586e-01 1.10396636e+00
1.52672365e-01 2.30373412e-01 7.53222644e-01 -1.35968888e+00
2.10060790e-01 5.67025244e-01 3.13674569e-01 -9.87634599e-01
-4.53208774e-01 1.01866461e-01 -2.18010500e-01 -1.05618991e-01
2.19994694e-01 -9.13845897e-01 -4.33657207e-02 2.27696109e+00
2.40813136e-01 -2.98521310e-01 7.22462893e-01 8.93933475e-01
8.73600900e-01 7.18392193e-01 5.60217440e-01 -4.08770651e-01
1.12549078e+00 -1.16780341e+00 -9.59195852e-01 -2.44669795e-01
9.09634352e-01 -7.55981326e-01 1.24963105e+00 7.97891542e-02
-1.33604848e+00 -3.44839901e-01 -7.37485230e-01 -1.68933123e-01
1.59391277e-02 -2.42694598e-02 9.74060714e-01 5.65367937e-01
-9.98413622e-01 3.70950133e-01 -4.05827790e-01 -5.28243661e-01
-2.99906805e-02 1.15800453e-02 1.73682999e-02 6.29541695e-01
-1.93211341e+00 1.45122695e+00 4.00069237e-01 -1.53806686e-01
-4.42591071e-01 -3.61551642e-01 -1.01206040e+00 -1.04340993e-01
-8.91286805e-02 -1.00839877e+00 1.60449028e+00 -1.38983095e+00
-2.40272164e+00 6.97261870e-01 1.08166508e-01 -5.25064051e-01
4.37825590e-01 -1.99281752e-01 -5.82280308e-02 2.65482605e-01
2.22824514e-02 1.10624671e+00 4.11145449e-01 -1.24782288e+00
-1.54024184e-01 1.91298351e-01 6.07825577e-01 7.07185447e-01
-6.56574309e-01 1.19373970e-01 7.59342551e-01 -7.21234620e-01
-6.32075012e-01 -9.28673565e-01 -3.15349460e-01 -2.39492580e-01
-3.50456715e-01 -5.22217095e-01 2.85383284e-01 -5.65959871e-01
1.14302087e+00 -1.62114215e+00 1.55754924e-01 -2.31408834e-01
1.05595738e-01 2.61752695e-01 -2.81742543e-01 8.31821918e-01
5.95203303e-02 2.45914564e-01 1.34057999e-01 -3.82590503e-01
2.53143907e-01 -2.60470668e-03 -7.07475245e-01 -2.10558042e-01
2.02708662e-01 1.02374279e+00 -1.16110802e+00 -6.44599259e-01
2.75799066e-01 4.69088227e-01 -7.50278533e-01 6.21524453e-01
-4.46477026e-01 3.99054110e-01 -4.70852047e-01 -4.58932482e-02
8.05562213e-02 -4.39471342e-02 1.22758381e-01 2.34513670e-01
3.10619250e-02 7.83056855e-01 -5.85743248e-01 1.36937749e+00
-8.09414446e-01 5.02172291e-01 -1.09945871e-01 -3.28759432e-01
1.14873540e+00 5.77787817e-01 4.17619273e-02 -4.71127212e-01
3.88372570e-01 -2.32291147e-01 2.88719594e-01 -5.34104884e-01
9.78300691e-01 -9.00913537e-01 -3.46575290e-01 1.17399764e+00
-1.34080917e-01 -5.45063257e-01 6.08231826e-03 4.68654484e-01
6.01213813e-01 -1.62458494e-01 5.38962305e-01 -3.84266153e-02
5.72156727e-01 2.78110892e-01 3.46777231e-01 6.79701507e-01
-5.81016302e-01 1.82943624e-02 5.85592985e-01 -9.90938917e-02
-5.66472352e-01 -6.91458464e-01 3.74732763e-01 1.37858093e+00
1.03791719e-02 -2.81081945e-01 -9.66016948e-01 -4.70486373e-01
-4.10771757e-01 1.58478510e+00 -4.95120198e-01 -6.30504012e-01
-5.41377842e-01 -4.50721741e-01 1.08743155e+00 4.93696988e-01
5.18009484e-01 -1.69075334e+00 -7.54322112e-01 3.78077090e-01
-7.37581491e-01 -1.00901496e+00 -3.74687612e-01 -1.84542909e-01
-6.28759444e-01 -6.23253644e-01 -2.54748344e-01 -6.91695333e-01
2.77977794e-01 6.41285703e-02 9.99093473e-01 1.34978563e-01
2.23503888e-01 6.24857485e-01 -5.31772792e-01 -4.68442678e-01
-1.22705579e+00 1.47716895e-01 -1.62872225e-02 -4.66289729e-01
2.54190445e-01 -3.79740357e-01 -3.23427081e-01 4.83511575e-02
-4.94526058e-01 5.80579519e-01 -4.77290479e-04 1.02218676e+00
-5.27331293e-01 -5.56909502e-01 1.13015223e+00 -8.43202591e-01
1.82168710e+00 -4.96718943e-01 2.64142096e-01 6.93003312e-02
-4.88408595e-01 1.26419097e-01 8.82721364e-01 -7.41125107e-01
-1.71416533e+00 -3.15490991e-01 -9.80917066e-02 6.10728115e-02
-1.36329994e-01 4.03141439e-01 2.31806740e-01 9.58267003e-02
1.08001494e+00 -2.06492785e-02 4.22361255e-01 2.20348105e-01
7.70123661e-01 7.62271047e-01 2.89294660e-01 -1.19396889e+00
3.92035037e-01 7.91008919e-02 -5.62687755e-01 -6.23995900e-01
-9.32148695e-01 4.85097505e-02 -2.43288845e-01 -4.64031488e-01
1.06449962e+00 -9.82329488e-01 -7.36237884e-01 2.56953597e-01
-1.52493477e+00 -9.08288360e-01 -3.88979107e-01 3.61480534e-01
-8.46238315e-01 3.93659204e-01 -1.14859998e+00 -1.11473572e+00
-7.21448064e-01 -7.40623593e-01 6.25105202e-01 3.78517032e-01
-1.27700543e+00 -1.43452549e+00 4.34102058e-01 8.39323759e-01
6.55151308e-01 1.66764557e-01 8.89859259e-01 -8.62425625e-01
1.54954270e-01 1.64790407e-01 1.15979291e-01 1.46806225e-01
1.83489606e-01 -5.29622547e-02 -9.19596910e-01 2.87430257e-01
3.88945878e-01 -1.10595727e+00 3.24297935e-01 -1.71590328e-01
4.47640687e-01 -8.58774185e-01 1.81852938e-05 5.52418791e-02
4.54679459e-01 1.99981928e-01 4.75887895e-01 1.73501879e-01
2.90463179e-01 1.00116658e+00 8.20666969e-01 8.48889589e-01
9.49241102e-01 4.63714480e-01 -2.37053216e-01 1.73142016e-01
-2.62226723e-02 -5.52051485e-01 8.19890082e-01 8.46839130e-01
-6.65030405e-02 -2.09026352e-01 -6.63344920e-01 4.41562444e-01
-1.91140342e+00 -1.41214561e+00 1.28191233e-01 1.29384887e+00
1.58968902e+00 -2.11612821e-01 1.03916854e-01 -2.94914395e-01
7.64608920e-01 4.64152604e-01 -3.25241297e-01 -1.11245203e+00
8.65385309e-03 1.94387436e-01 -4.50343162e-01 1.07627404e+00
-6.35394156e-01 1.50101924e+00 6.47605896e+00 3.40115786e-01
-1.15616775e+00 -5.23903733e-03 4.83235210e-01 3.34995240e-03
-5.61327815e-01 4.35811616e-02 -3.77755582e-01 2.26375952e-01
8.33319247e-01 -7.54888415e-01 4.05426115e-01 9.11472619e-01
5.34613431e-01 4.92996834e-02 -1.16356111e+00 4.74813610e-01
3.87782902e-01 -1.01835310e+00 2.58285344e-01 -2.54443169e-01
6.00078285e-01 -7.83333659e-01 -1.68100461e-01 7.67167509e-01
8.30949366e-01 -1.01300681e+00 6.94097400e-01 3.92790288e-01
2.69396663e-01 -6.79164767e-01 4.78659779e-01 5.83413541e-01
-4.69780236e-01 6.47481531e-02 -1.64601192e-01 -6.85857832e-01
4.05200154e-01 -1.21790901e-01 -1.32982171e+00 -6.32484928e-02
-2.56149024e-01 4.31927234e-01 -2.07633197e-01 1.63058583e-02
-7.22550452e-01 6.60303712e-01 1.50845066e-01 -4.66286570e-01
1.48879230e-01 -1.38798729e-01 5.49087524e-01 1.23733842e+00
-1.12936184e-01 5.35094440e-01 2.18534395e-01 1.29960620e+00
7.17304042e-03 1.91687003e-01 -6.64461613e-01 -3.72567445e-01
6.56797707e-01 1.22158456e+00 -3.14726770e-01 -3.54274899e-01
1.34956434e-01 9.66334820e-01 5.28742373e-01 1.70383319e-01
-9.07358706e-01 -1.61062554e-01 7.32290566e-01 -1.04875363e-01
-4.14273590e-01 -1.55953065e-01 -5.02545416e-01 -1.19405305e+00
-6.79649830e-01 -1.03665781e+00 2.84289271e-01 -1.02954662e+00
-1.52236962e+00 5.67868650e-01 -1.59376264e-02 -5.44269621e-01
-9.01012897e-01 -2.03034744e-01 -1.01543248e+00 6.97750926e-01
-1.19769716e+00 -1.22278535e+00 -1.54519051e-01 4.19589043e-01
6.74099565e-01 -1.86712265e-01 1.31416082e+00 -4.26357478e-01
-4.25303191e-01 7.42781103e-01 -8.57445300e-01 9.23380777e-02
1.14153540e+00 -1.01776552e+00 7.93345273e-02 4.35761690e-01
-4.37566519e-01 9.44634140e-01 9.29566085e-01 -6.88418925e-01
-7.95663893e-01 -8.19284201e-01 1.22430503e+00 -5.15581131e-01
7.99439847e-01 -7.82545283e-02 -8.18002164e-01 6.91173196e-01
9.33303058e-01 -7.74263382e-01 1.24039626e+00 1.92334220e-01
-2.44257972e-01 6.30524278e-01 -1.33772326e+00 1.02616096e+00
9.56320643e-01 -6.40202165e-01 -1.18597817e+00 3.05099875e-01
8.24665368e-01 -3.83362830e-01 -9.26997304e-01 -7.67764524e-02
3.43101501e-01 -8.74976337e-01 7.05799937e-01 -7.93155134e-01
9.50780392e-01 2.01988414e-01 1.93717673e-01 -1.78593707e+00
-1.69067308e-01 -1.06105518e+00 -6.27156794e-02 1.28349686e+00
2.91743904e-01 -6.78619683e-01 4.80292737e-01 1.21466708e+00
-1.02774873e-01 -3.93760741e-01 -4.20428276e-01 -3.40385020e-01
5.54060996e-01 -9.45275575e-02 3.93884361e-01 1.34426498e+00
1.14358521e+00 1.00960648e+00 -6.22670233e-01 -4.16107088e-01
3.03983781e-02 -5.03209932e-03 9.20385897e-01 -9.44882035e-01
-3.87271136e-01 -6.78826094e-01 2.03090072e-01 -9.04433370e-01
9.97266293e-01 -7.86233902e-01 2.47684076e-01 -1.57679427e+00
1.46911278e-01 -1.87574208e-01 3.90050977e-01 5.79737902e-01
-5.72366178e-01 -1.43146887e-01 4.75598067e-01 -1.29183292e-01
-5.19097626e-01 1.06652534e+00 1.39820004e+00 2.54847139e-01
-5.06783545e-01 -2.98023075e-01 -1.30549264e+00 7.33890235e-01
1.15919793e+00 -2.42505759e-01 -7.22684383e-01 -1.40505075e-01
1.07557669e-01 3.21878016e-01 2.50863403e-01 -5.64051330e-01
1.94846377e-01 -7.03449190e-01 -1.88229028e-02 1.84837267e-01
6.40557647e-01 -5.86763248e-02 -4.00341660e-01 4.28320199e-01
-1.07122028e+00 6.07165545e-02 1.98629722e-01 8.06231871e-02
-9.48294550e-02 -3.97060096e-01 9.03604329e-01 -3.39189112e-01
-3.39644730e-01 -5.68486631e-01 -1.08000743e+00 4.38072801e-01
1.07959497e+00 -1.54151499e-01 -5.48479140e-01 -1.18353570e+00
-5.16700387e-01 3.50646138e-01 5.03014445e-01 5.27303457e-01
6.04959726e-01 -1.26490879e+00 -7.19342709e-01 -3.03318143e-01
-4.76424657e-02 -4.76094425e-01 1.54442182e-02 2.88850904e-01
-2.48647347e-01 2.19845548e-01 -3.87513161e-01 7.42104203e-02
-1.24274504e+00 1.30498126e-01 4.52187389e-01 -3.17948192e-01
-2.80321658e-01 9.16000664e-01 6.23177961e-02 -8.18941593e-01
8.80319402e-02 1.26854647e-02 -4.09263760e-01 1.59230471e-01
4.25878763e-01 3.23467076e-01 -7.48118520e-01 -6.20853901e-01
5.30618764e-02 -1.50787234e-01 -8.60667005e-02 -6.62131488e-01
9.18162704e-01 -1.73898444e-01 -2.24501312e-01 5.70520282e-01
5.25290072e-01 1.07394055e-01 -8.36636305e-01 1.57856330e-01
-1.85783178e-01 -4.31043319e-02 -3.89769047e-01 -1.23211193e+00
-2.82666475e-01 7.52579808e-01 -2.57511795e-01 1.08291015e-01
6.44255102e-01 -2.18335211e-01 8.74516308e-01 6.49893165e-01
2.60248452e-01 -1.28847551e+00 6.83128476e-01 1.06172693e+00
1.13855433e+00 -1.12361312e+00 -2.71291643e-01 -2.14394525e-01
-1.58345592e+00 1.18337214e+00 1.25956297e+00 1.52671570e-02
1.43303499e-01 1.49238005e-01 5.50880075e-01 4.87262616e-03
-1.39657485e+00 1.44716442e-01 -1.84229851e-01 6.31627977e-01
1.02299333e+00 1.17596842e-01 -6.50353789e-01 1.07079947e+00
-9.90791559e-01 -3.67079414e-02 9.86427009e-01 6.55760765e-01
-4.10221756e-01 -1.10441208e+00 -2.13645339e-01 7.05359578e-02
-1.22171916e-01 -1.58118933e-01 -1.33722842e+00 5.71496427e-01
-2.07679927e-01 1.43301737e+00 -1.55836344e-01 -4.29813892e-01
2.60005265e-01 4.01628464e-01 4.05979037e-01 -9.59997475e-01
-1.31975007e+00 -5.31008244e-01 8.08453381e-01 -2.15278089e-01
-3.29779625e-01 -6.09978318e-01 -1.51925254e+00 -4.70897406e-01
-4.14527535e-01 4.67763275e-01 8.95837620e-02 9.89883721e-01
3.59632164e-01 2.48383597e-01 6.82387769e-01 -5.63165665e-01
-1.00297451e+00 -1.32949543e+00 -3.51676811e-03 4.56368983e-01
-2.19887644e-02 -4.92070645e-01 -4.97725755e-01 4.61685471e-02]
|
[13.053296089172363, 7.737351894378662]
|
c1f39625-9a1d-4711-97d6-b2890ca20393
|
continuous-online-extrinsic-calibration-of
|
2306.13240
| null |
https://arxiv.org/abs/2306.13240v1
|
https://arxiv.org/pdf/2306.13240v1.pdf
|
Continuous Online Extrinsic Calibration of Fisheye Camera and LiDAR
|
Automated driving systems use multi-modal sensor suites to ensure the reliable, redundant and robust perception of the operating domain, for example camera and LiDAR. An accurate extrinsic calibration is required to fuse the camera and LiDAR data into a common spatial reference frame required by high-level perception functions. Over the life of the vehicle the value of the extrinsic calibration can change due physical disturbances, introducing an error into the high-level perception functions. Therefore there is a need for continuous online extrinsic calibration algorithms which can automatically update the value of the camera-LiDAR calibration during the life of the vehicle using only sensor data. We propose using mutual information between the camera image's depth estimate, provided by commonly available monocular depth estimation networks, and the LiDAR pointcloud's geometric distance as a optimization metric for extrinsic calibration. Our method requires no calibration target, no ground truth training data and no expensive offline optimization. We demonstrate our algorithm's accuracy, precision, speed and self-diagnosis capability on the KITTI-360 data set.
|
['Stefan Milz', 'Florian Ölsner', 'Jeremy Tschirner', 'Jack Borer']
|
2023-06-22
| null | null | null | null |
['depth-estimation', 'monocular-depth-estimation']
|
['computer-vision', 'computer-vision']
|
[ 1.72388386e-02 -8.14064145e-02 2.67193206e-02 -9.10314620e-01
-5.90738058e-01 -6.26343012e-01 3.23311776e-01 3.37012261e-02
-5.40115237e-01 4.53094184e-01 -6.33823872e-01 -7.21638184e-03
2.99058985e-02 -8.36581767e-01 -8.38416517e-01 -5.43683708e-01
5.73947787e-01 5.77866375e-01 5.02207339e-01 -7.15800226e-02
4.34542418e-01 7.42949247e-01 -1.84704113e+00 -5.41135490e-01
8.50593925e-01 1.31781423e+00 4.92390454e-01 7.91911006e-01
8.44329894e-02 2.90459245e-01 -4.02895838e-01 -1.35578245e-01
4.48975354e-01 3.59178185e-01 -9.82218795e-03 7.95221701e-02
5.67856491e-01 -4.40599531e-01 -3.34945202e-01 1.21069753e+00
2.74715126e-01 -3.72158661e-02 3.78510863e-01 -1.67495680e+00
8.02182630e-02 -1.56233519e-01 -4.07752633e-01 -3.79671186e-01
1.12604991e-01 3.37287962e-01 3.67474884e-01 -9.45842862e-01
5.35782218e-01 9.31192577e-01 9.12135243e-01 1.98687509e-01
-1.04757833e+00 -8.12634945e-01 -2.23968491e-01 3.79584312e-01
-1.67475724e+00 -6.43653989e-01 7.73976505e-01 -5.29956698e-01
7.61748433e-01 -1.77659765e-01 5.51839292e-01 4.73341405e-01
4.63198066e-01 -3.13921720e-01 6.60121679e-01 -2.84106135e-01
3.52997601e-01 4.75389063e-01 -1.77980706e-01 4.89130825e-01
5.96734345e-01 4.61214304e-01 -7.84667075e-01 2.68174022e-01
4.21859235e-01 -1.96681693e-01 -1.25394473e-02 -8.75142813e-01
-8.53506207e-01 5.74298978e-01 3.28801721e-01 -3.21561605e-01
-1.70028985e-01 3.62036616e-01 -2.64190108e-01 1.14244454e-01
-1.17499270e-01 1.18866347e-01 -4.13609654e-01 -2.65270054e-01
-6.46704793e-01 -1.90429822e-01 4.14836019e-01 1.41601300e+00
1.53330362e+00 -1.28320947e-01 6.90723836e-01 2.77396500e-01
6.43922031e-01 9.95765448e-01 1.42208740e-01 -1.45781028e+00
6.18757665e-01 7.99679220e-01 2.27415845e-01 -1.19279015e+00
-2.16490805e-01 -1.89276874e-01 -3.66021007e-01 6.74475193e-01
1.62387133e-01 -9.92036462e-02 -7.04234838e-01 1.51509523e+00
5.25235176e-01 3.19651932e-01 2.52102762e-01 5.86507261e-01
4.37213123e-01 1.65864173e-02 -4.89791811e-01 -6.07651062e-02
7.21098006e-01 -2.60982186e-01 -6.81284606e-01 -6.21346414e-01
6.39559746e-01 -9.46374834e-01 6.26591384e-01 3.25989813e-01
-8.07016015e-01 -8.38758469e-01 -1.78250718e+00 -1.37341559e-01
-4.71035868e-01 1.11826502e-01 2.26947382e-01 5.91159582e-01
-9.67755079e-01 3.20931971e-01 -8.02472234e-01 -2.97972888e-01
-4.83766682e-02 6.26658916e-01 -5.60285866e-01 -2.40525290e-01
-9.85373259e-01 1.39345753e+00 3.05907190e-01 1.73322678e-01
-7.19881296e-01 -7.12579072e-01 -9.65902746e-01 -4.52127248e-01
2.87999004e-01 -4.38477516e-01 1.17998123e+00 -6.48849189e-01
-1.49107432e+00 7.59430230e-01 -1.01943471e-01 -4.68503535e-01
5.10673881e-01 -1.74095407e-01 -4.98148441e-01 9.14842710e-02
2.56745249e-01 8.86576653e-01 8.83381069e-01 -1.49874878e+00
-9.60523903e-01 -6.74140990e-01 -1.37201220e-01 3.28172714e-01
2.07240313e-01 -8.05155456e-01 -5.23320377e-01 2.60697484e-01
7.47390866e-01 -8.90320420e-01 -1.74803764e-01 3.41590524e-01
4.40716706e-02 3.88544768e-01 1.25425243e+00 -1.95293903e-01
5.54475605e-01 -2.37977839e+00 -3.22081894e-01 3.27292442e-01
1.00908332e-01 -9.19361115e-02 1.16902359e-01 -8.75455067e-02
1.55384494e-02 -5.16529500e-01 -8.37434903e-02 -4.61857468e-01
-3.09181720e-01 5.51063418e-01 -1.45318732e-01 8.50748003e-01
1.29721582e-01 5.53080797e-01 -7.30935574e-01 -4.58663315e-01
7.54582703e-01 6.29120588e-01 -2.49211565e-01 4.14297909e-01
1.12065993e-01 2.42828384e-01 -9.42742154e-02 4.72100675e-01
9.62541759e-01 3.01484376e-01 -1.79440990e-01 -5.58647573e-01
-3.10939163e-01 2.05627345e-02 -1.38012040e+00 1.75615883e+00
-4.48544413e-01 7.57073700e-01 2.81085491e-01 -5.58351636e-01
1.29991436e+00 5.19691817e-02 5.81476629e-01 -8.09865952e-01
2.80702770e-01 3.18928093e-01 -2.68755227e-01 -2.50037938e-01
8.17779958e-01 7.58078843e-02 -1.83262695e-02 1.04956031e-01
-1.29237697e-01 -8.31892312e-01 -2.78298616e-01 1.44516677e-01
8.15030277e-01 1.96049854e-01 7.78609440e-02 1.26245588e-01
5.78042686e-01 2.89339334e-01 7.50273645e-01 2.47058615e-01
-1.97146758e-01 5.08996308e-01 -7.95560926e-02 -7.89650455e-02
-1.12680197e+00 -1.10004711e+00 -3.31948727e-01 8.42715800e-02
6.69795334e-01 5.80925634e-03 -5.76782525e-01 -1.32874295e-01
3.61432910e-01 8.39566112e-01 -3.59227657e-01 -5.81504107e-01
-4.40286458e-01 -3.96153480e-02 4.16183919e-01 6.34557843e-01
6.64289176e-01 -7.90218785e-02 -9.82403278e-01 2.63362855e-01
7.29085505e-02 -1.54040766e+00 -2.50006676e-01 3.39858115e-01
-8.42688680e-01 -1.24804664e+00 4.14214581e-01 -2.68027335e-01
8.30175817e-01 6.46611631e-01 8.43918681e-01 -8.33308548e-02
-9.65227857e-02 7.09362745e-01 2.19242007e-01 -5.14721215e-01
-3.92348737e-01 -3.97027552e-01 2.26913348e-01 -9.90060419e-02
4.90477413e-01 -5.71570933e-01 -3.63805443e-01 6.25448942e-01
-3.95389110e-01 -9.29258391e-02 1.30128145e-01 2.41873950e-01
8.69950831e-01 1.47189334e-01 1.67503461e-01 -2.61456966e-01
7.60437250e-02 -2.08886713e-01 -1.53370774e+00 -2.48479903e-01
-9.81451154e-01 -1.11025825e-01 1.32262409e-01 -1.02379762e-01
-9.13800359e-01 6.43560290e-01 4.68539625e-01 -7.97139704e-01
-4.09431495e-02 2.54372925e-01 -5.75061917e-01 -2.79874295e-01
7.67053843e-01 -1.73274323e-01 3.15649599e-01 -1.60718739e-01
3.37578267e-01 7.92022288e-01 1.32695675e+00 -2.97426820e-01
1.13998711e+00 8.16889703e-01 4.49962407e-01 -7.18367577e-01
-7.27125764e-01 -6.85093999e-01 -1.19536817e+00 -4.80113715e-01
8.63385081e-01 -1.21836412e+00 -7.92868435e-01 5.22719204e-01
-1.21880746e+00 -4.51378226e-02 -3.04844856e-01 5.73201716e-01
-6.62762463e-01 2.44754538e-01 2.30106026e-01 -8.69732618e-01
9.21303183e-02 -1.27777576e+00 1.00374258e+00 2.43640438e-01
8.25467184e-02 -7.48591423e-01 -3.29625942e-02 3.42778116e-01
6.75798431e-02 5.26018798e-01 2.98002601e-01 4.18773055e-01
-1.19381189e+00 -3.85784924e-01 -1.40607730e-01 5.60722888e-01
3.19477648e-01 2.42494285e-01 -1.24304557e+00 -2.24221274e-01
2.82543242e-01 1.82599470e-01 4.02298868e-01 2.57894278e-01
6.64175093e-01 2.06451267e-01 -3.92938644e-01 1.03371668e+00
1.71029961e+00 3.28471392e-01 6.00504458e-01 3.21359247e-01
6.69191241e-01 5.97844183e-01 1.24663806e+00 1.28423586e-01
8.23679090e-01 6.61559463e-01 8.98876309e-01 1.83604524e-01
9.86796692e-02 -2.73016483e-01 2.56876230e-01 7.17158854e-01
4.66715485e-01 2.90537417e-01 -8.98168623e-01 4.51290011e-01
-1.70406425e+00 -6.56935394e-01 -3.50645483e-01 2.87215161e+00
3.94793034e-01 4.34328318e-01 -4.85393703e-01 2.56809950e-01
7.92545080e-01 -4.30729389e-01 -1.00521302e+00 -3.27229291e-01
5.15119024e-02 -2.37439051e-01 1.29153061e+00 9.06694114e-01
-7.20429182e-01 7.49187589e-01 6.21001911e+00 -6.82674646e-02
-1.25126100e+00 -1.98353566e-02 -1.61034614e-02 5.51845618e-02
-1.41641572e-01 2.77893007e-01 -9.65689063e-01 3.14938605e-01
1.08342886e+00 -1.77484542e-01 5.24640799e-01 9.63687837e-01
2.79397488e-01 -6.73121333e-01 -1.20205224e+00 1.32805789e+00
1.26658916e-01 -1.21771491e+00 -7.56957769e-01 1.96177587e-01
4.75659311e-01 4.72545654e-01 -1.15536056e-01 -1.89022914e-01
2.08099425e-01 -5.88914156e-01 9.17305589e-01 6.90458894e-01
1.06855309e+00 -9.24692452e-01 6.52280688e-01 3.94316733e-01
-1.33084583e+00 -8.24071094e-02 -4.40669447e-01 2.96018329e-02
1.99887797e-01 5.94151795e-01 -1.12155402e+00 3.74176174e-01
6.42573237e-01 7.23334610e-01 -6.96473122e-01 7.59526253e-01
-2.27092415e-01 -4.71091494e-02 -4.85755861e-01 6.48208916e-01
-1.83171511e-01 -3.63484859e-01 5.70413470e-01 3.85108799e-01
4.23806131e-01 -6.44934848e-02 7.15289637e-02 8.15935075e-01
2.68799454e-01 -5.02534032e-01 -8.33176851e-01 2.46070310e-01
8.44283760e-01 1.15016592e+00 -2.16413751e-01 -1.86172187e-01
-2.38383576e-01 3.36297125e-01 -9.36626792e-02 1.83570355e-01
-9.11379457e-01 -2.69617110e-01 1.06852376e+00 1.54748246e-01
8.22655633e-02 -7.01271415e-01 -7.14776218e-01 -6.69893682e-01
1.76582590e-01 -2.26646528e-01 -7.29345754e-02 -1.19424272e+00
-7.15662658e-01 2.02316046e-01 -2.48113368e-02 -1.56833661e+00
-4.73356336e-01 -5.72799444e-01 -8.33803415e-02 1.01661980e+00
-1.61088145e+00 -9.78313208e-01 -1.00134909e+00 7.15086102e-01
1.35303408e-01 -2.15612501e-01 4.68913913e-01 2.90380239e-01
-2.17762247e-01 1.80562153e-01 -1.54128090e-01 -2.51765043e-01
9.58171308e-01 -1.04988170e+00 2.17802137e-01 8.87971520e-01
-4.70798612e-01 2.37010792e-01 9.69285667e-01 -7.07510173e-01
-1.81420302e+00 -1.13517475e+00 5.88902473e-01 -8.27563763e-01
5.01408637e-01 -2.82476962e-01 -6.41376257e-01 7.32242346e-01
-3.12920064e-01 1.81730781e-02 2.02756271e-01 -4.43888575e-01
-2.37892970e-01 -8.65909040e-01 -1.42160892e+00 1.65066481e-01
7.29421377e-01 -6.55984819e-01 -4.18393612e-01 6.89479038e-02
8.47100139e-01 -8.29512239e-01 -7.84186661e-01 5.01366198e-01
5.69144607e-01 -9.56773102e-01 1.02222145e+00 2.34875560e-01
-1.65408567e-01 -8.74791980e-01 -6.83379114e-01 -8.81724060e-01
1.93716064e-01 -2.82049865e-01 2.19338909e-02 1.12276006e+00
1.16547614e-01 -1.02152324e+00 7.31981814e-01 9.33596194e-01
-2.99966335e-01 1.05026166e-03 -1.24829078e+00 -4.96444404e-01
-4.07801956e-01 -1.02177751e+00 8.79343152e-01 6.97191715e-01
-5.73913455e-01 3.41189146e-01 4.60245647e-02 9.50444221e-01
1.12607503e+00 -1.84797823e-01 1.38005674e+00 -1.51997840e+00
1.68456748e-01 8.03873390e-02 -8.69766355e-01 -6.78242981e-01
6.09020628e-02 -3.48202080e-01 3.34548444e-01 -1.27624738e+00
-4.75196838e-01 -6.29358172e-01 1.06074229e-01 9.37686041e-02
4.21646923e-01 6.57160878e-02 -7.80211715e-03 9.77774933e-02
-6.81497231e-02 4.76773351e-01 8.90742421e-01 -8.12133476e-02
-1.44031540e-01 -5.07493764e-02 -3.10035050e-01 7.34764934e-01
6.60207808e-01 -6.23774111e-01 -6.85569644e-01 -6.99087262e-01
4.76462573e-01 7.96471387e-02 3.69769186e-01 -1.38286674e+00
7.40338504e-01 -3.22100371e-01 3.86465043e-01 -1.22313166e+00
9.06597018e-01 -1.45849693e+00 3.93968463e-01 4.09722269e-01
3.46303850e-01 4.33036745e-01 1.03655785e-01 4.76701498e-01
-1.44423842e-01 -1.96394935e-01 1.04258978e+00 2.14313731e-01
-8.42275202e-01 3.54671746e-01 1.88163314e-02 -3.72631013e-01
1.18132639e+00 -9.72029328e-01 -1.26479939e-01 -3.91624808e-01
-3.52199256e-01 3.27178329e-01 1.04661000e+00 5.93045056e-01
7.38163352e-01 -1.33402824e+00 -2.15329736e-01 6.89081609e-01
3.67723852e-01 5.35960555e-01 -1.19539909e-01 5.88517427e-01
-4.73733187e-01 3.16976756e-01 -1.07271932e-01 -1.29253483e+00
-1.09321332e+00 4.13656116e-01 7.90786624e-01 5.78663886e-01
-2.17416629e-01 3.09085548e-01 -1.79977193e-01 -5.60205162e-01
4.71787378e-02 -3.75579715e-01 2.38631234e-01 -1.58114687e-01
2.71744281e-01 4.02976334e-01 2.88496435e-01 -1.14921582e+00
-4.35243934e-01 8.98042083e-01 6.06150150e-01 -3.53780061e-01
9.29921150e-01 -7.68735707e-01 -7.30524212e-02 7.41567671e-01
1.10997081e+00 -2.24255621e-01 -1.57396543e+00 -5.16054966e-02
-8.34884197e-02 -7.75301874e-01 4.67858851e-01 -2.54935980e-01
-1.04634929e+00 9.88568068e-01 1.15262127e+00 -3.52143139e-01
9.10695076e-01 -3.56643081e-01 3.95777464e-01 4.60615963e-01
7.80756652e-01 -1.59819663e+00 -2.06259891e-01 4.08684343e-01
6.90708816e-01 -1.42026353e+00 2.32414618e-01 -2.85263807e-01
-3.29892427e-01 9.91399109e-01 8.53326619e-01 -4.10940684e-03
6.21981144e-01 3.86278480e-01 3.65826935e-01 -7.88385496e-02
-6.40580237e-01 -2.13547144e-02 1.14205601e-02 8.66631866e-01
-3.53798807e-01 9.07575116e-02 4.74242121e-01 -3.40571135e-01
-4.90045935e-01 -1.05001062e-01 6.06421769e-01 8.28473568e-01
-7.46844530e-01 -1.07088506e+00 -6.57849371e-01 1.45888343e-01
5.06962359e-01 4.43209231e-01 -4.88499627e-02 8.53134036e-01
4.37802792e-01 1.13913083e+00 3.65529686e-01 -3.98867190e-01
5.04822671e-01 -2.04643160e-01 4.87216294e-01 -5.68455517e-01
1.04401849e-01 -2.55804330e-01 -9.03670415e-02 -8.60662222e-01
-3.54144633e-01 -8.66298020e-01 -1.35413682e+00 -2.25471348e-01
-4.99462485e-01 -2.65004218e-01 1.44112563e+00 1.04286838e+00
2.97523737e-01 1.34971976e-01 8.43193412e-01 -8.04005325e-01
-4.33137476e-01 -5.82408369e-01 -1.94893271e-01 -9.41178054e-02
5.53905308e-01 -8.75690579e-01 -4.79804128e-01 1.12132192e-01]
|
[7.601121425628662, -2.232048273086548]
|
7f6ed10f-dc9e-4847-b813-a488d8e7d6cc
|
kosign-sign-language-translation-project
| null | null |
https://aclanthology.org/2022.sltat-1.9
|
https://aclanthology.org/2022.sltat-1.9.pdf
|
KoSign Sign Language Translation Project: Introducing The NIASL2021 Dataset
|
We introduce a new sign language production (SLP) and sign language translation (SLT) dataset, NIASL2021, consisting of 201,026 Korean-KSL data pairs. KSL translations of Korean source texts are represented in three formats: video recordings, keypoint position data, and time-aligned gloss annotations for each hand (using a 7,989 sign vocabulary) and for eight different non-manual signals (NMS). We evaluated our sign language elicitation methodology and found that text-based prompting had a negative effect on translation quality in terms of naturalness and comprehension. We recommend distilling text into a visual medium before translating into sign language or adding a prompt-blind review step to text-based translation methodologies.
|
['Jun Woo Lee', 'Kang Suk Byun', 'Hye Jin Myung', 'Du Hui Lee', 'Mathew Huerta-Enochian']
| null | null | null | null |
sltat-lrec-2022-6
|
['sign-language-translation', 'sign-language-production']
|
['computer-vision', 'natural-language-processing']
|
[ 4.50494200e-01 -8.52550790e-02 -2.44181335e-01 -3.21007073e-01
-1.38835549e+00 -1.04486322e+00 6.99411035e-01 -7.07753241e-01
-5.64648807e-01 7.00119853e-01 1.03658307e+00 -4.09805924e-01
1.88384622e-01 -7.85569921e-02 -6.10318065e-01 -2.30180651e-01
6.15123451e-01 4.16719049e-01 1.95810422e-01 -2.31479146e-02
1.50283247e-01 3.87637436e-01 -1.06231809e+00 5.60479224e-01
8.74368668e-01 7.04814672e-01 -2.03796513e-02 9.36721027e-01
-1.65175155e-01 7.00567245e-01 -7.11510837e-01 -4.98495102e-01
5.56296766e-01 -8.04063737e-01 -6.91540778e-01 -2.49530479e-01
1.15639138e+00 -7.71946549e-01 -3.13621163e-01 7.79746890e-01
1.06056046e+00 -2.97556072e-01 6.52477324e-01 -1.31709456e+00
-1.00179815e+00 4.46167827e-01 -8.00673068e-02 -3.62387121e-01
8.90164256e-01 6.13755047e-01 8.41387928e-01 -9.83308852e-01
1.35940135e+00 1.16022611e+00 5.68529010e-01 8.24374855e-01
-1.00571668e+00 -8.51940155e-01 -2.20323101e-01 1.92143574e-01
-9.52798843e-01 -7.55142093e-01 5.16998410e-01 -5.35361409e-01
9.03525293e-01 2.97187924e-01 1.02302742e+00 1.67610025e+00
-4.31434587e-02 9.35163140e-01 1.50050890e+00 -7.27867186e-01
-1.32533208e-01 -2.68405229e-01 -9.47012603e-02 5.17197192e-01
3.57393563e-01 3.63125473e-01 -1.34568810e+00 -1.17868157e-02
8.58586848e-01 -7.77859271e-01 -4.80049372e-01 5.71246780e-02
-1.85811901e+00 1.47794738e-01 -2.91619480e-01 2.38427237e-01
-3.49732518e-01 1.87242493e-01 4.23123598e-01 8.65226805e-01
-9.50720087e-02 3.48576099e-01 -6.26996636e-01 -7.17456341e-01
-7.95128703e-01 9.14532021e-02 6.89581513e-01 1.46832740e+00
-1.43239856e-01 1.20303147e-01 -6.01957262e-01 6.38432682e-01
4.21952635e-01 1.47773826e+00 6.72859907e-01 -1.09195590e+00
8.62298369e-01 4.55404162e-01 3.45784217e-01 -5.73081017e-01
-2.34330818e-01 4.72599685e-01 -1.22460432e-01 3.04712743e-01
8.67534637e-01 -2.86155194e-01 -1.33567548e+00 1.67370057e+00
-1.74897611e-01 -5.50146520e-01 3.83540690e-02 1.14355886e+00
1.18356097e+00 2.50105739e-01 2.49124542e-01 -2.08151340e-01
1.30595016e+00 -9.00668561e-01 -1.12924278e+00 5.94887473e-02
5.77912450e-01 -1.17596352e+00 1.60339117e+00 4.99788016e-01
-1.16737354e+00 -2.15016082e-01 -6.83555961e-01 -4.41482455e-01
-2.51795530e-01 7.78367281e-01 8.77609849e-02 3.33147883e-01
-1.17579579e+00 -9.34339985e-02 -6.96926951e-01 -6.93855703e-01
7.77486190e-02 1.21491425e-01 -7.25383818e-01 -9.06163529e-02
-9.51494098e-01 1.14177597e+00 -9.04843062e-02 1.48554310e-01
-4.12846476e-01 -2.46027246e-01 -8.96549940e-01 -8.68123293e-01
2.67810702e-01 -6.46266043e-01 1.65002990e+00 -9.83524621e-01
-2.04135990e+00 1.19386041e+00 -4.23794419e-01 1.16749018e-01
1.00606418e+00 -4.87546563e-01 -7.92981744e-01 4.19631362e-01
9.86689120e-04 6.78379536e-01 9.78492320e-01 -1.07481170e+00
-5.77549994e-01 -7.73573741e-02 -3.08506429e-01 2.68996358e-01
2.62130201e-01 5.21532953e-01 -4.79501307e-01 -1.06869388e+00
2.92784840e-01 -1.00670159e+00 4.06557411e-01 6.24774098e-01
-4.31621045e-01 6.29650354e-02 5.89687526e-01 -1.67735994e+00
1.14876890e+00 -1.83089876e+00 3.06960762e-01 1.73807740e-01
4.83305193e-02 4.29407835e-01 -6.55104816e-01 4.42751348e-01
3.49086970e-01 -1.35330567e-02 -8.47088769e-02 -1.96001276e-01
2.72478819e-01 3.15232217e-01 -2.90997595e-01 3.47404778e-01
-7.90350661e-02 1.58330607e+00 -9.66470420e-01 -7.14029551e-01
2.71744937e-01 3.27444524e-01 -3.40526909e-01 -1.96478907e-02
-1.93647891e-01 6.91254914e-01 3.36091183e-02 1.22339272e+00
1.49554253e-01 1.69282481e-01 1.35768533e-01 -3.99836719e-01
-1.56436235e-01 4.33251321e-01 -7.52991974e-01 1.81432045e+00
-5.26314080e-01 1.16252840e+00 -6.23637103e-02 1.54424191e-01
5.65086246e-01 7.09523261e-01 2.69837320e-01 -9.69649255e-01
2.83863723e-01 6.62473619e-01 6.79129316e-03 -9.87799466e-01
3.69482815e-01 9.82604027e-02 -3.13256308e-02 5.80269992e-01
1.11597292e-01 -4.99462306e-01 2.93405682e-01 -3.50811407e-02
9.06467557e-01 3.84374738e-01 2.97207147e-01 1.71287939e-01
7.27239549e-02 3.01985204e-01 2.80599028e-01 4.93770391e-01
-3.18505853e-01 6.77806139e-01 5.45102656e-02 4.12181430e-02
-9.86305177e-01 -1.35146284e+00 3.77716064e-01 7.43169844e-01
-3.69888246e-01 -4.94848251e-01 -5.02337098e-01 -5.62172711e-01
-2.23499686e-01 8.48615885e-01 -3.10638249e-01 1.84786871e-01
-8.70593488e-01 2.87774920e-01 1.06928790e+00 8.26840818e-01
3.81155729e-01 -1.42652333e+00 -8.94839406e-01 -7.40937740e-02
-6.09783351e-01 -1.28955102e+00 -1.34344780e+00 -4.69023675e-01
-5.88918090e-01 -1.24176800e+00 -1.35367918e+00 -1.10154986e+00
8.36619318e-01 -2.10879967e-01 8.05060089e-01 -4.65666652e-01
-2.86885217e-04 9.43138659e-01 -7.33843386e-01 -3.33992213e-01
-8.20446074e-01 -3.98647785e-01 3.34611177e-01 -3.45702767e-01
5.99946678e-01 -2.90035456e-01 -8.61490071e-02 3.66467863e-01
-5.48426688e-01 5.13843417e-01 8.55117023e-01 8.43553305e-01
3.41336846e-01 -1.24437118e+00 1.96952950e-02 1.19945794e-01
9.61848080e-01 6.57337487e-01 -4.99148518e-01 5.79422414e-01
-5.67213714e-01 -1.06603250e-01 2.27467895e-01 -1.02950931e+00
-8.71397376e-01 4.14668210e-03 1.85647011e-01 -3.23823065e-01
-3.39251235e-02 4.02269840e-01 -1.39128923e-01 -1.93992808e-01
6.99864030e-01 6.20449007e-01 1.86310083e-01 -3.78716171e-01
6.10712707e-01 1.00722575e+00 1.02388918e+00 -6.15198553e-01
9.02179182e-01 1.78859964e-01 -3.42465818e-01 -7.65668035e-01
-4.53213155e-02 -2.51132607e-01 -7.94568956e-01 -5.50134718e-01
7.30021060e-01 -6.66652977e-01 -6.75181925e-01 9.31967556e-01
-1.35234582e+00 -6.56805396e-01 -5.90791762e-01 9.26861584e-01
-9.47711766e-01 2.76608020e-01 -4.02660012e-01 -5.44483840e-01
-3.44958365e-01 -9.93106961e-01 1.25394583e+00 -3.49742435e-02
-7.92703450e-01 -1.54689103e-01 2.56896079e-01 4.31668401e-01
3.96516263e-01 4.84393165e-02 5.67566037e-01 -2.85923660e-01
-3.46033990e-01 -3.12625736e-01 -4.44459498e-01 5.19635618e-01
4.49294835e-01 1.41462404e-02 -5.13843536e-01 -4.89239246e-02
-6.65905297e-01 -5.50300658e-01 3.15818399e-01 2.95204937e-01
3.00979614e-02 -6.00939333e-01 1.80682555e-01 4.49174851e-01
8.92101824e-01 3.28477561e-01 6.40464664e-01 -6.47987276e-02
6.81455374e-01 4.55604315e-01 5.71349204e-01 -6.07383773e-02
3.38128179e-01 8.54571462e-01 -5.68894863e-01 1.26804009e-01
-8.83489907e-01 -8.72046173e-01 8.60736012e-01 1.25234890e+00
-5.47478080e-01 -1.95765138e-01 -9.98683393e-01 7.23089457e-01
-1.60937536e+00 -7.32481241e-01 -1.75025553e-01 1.87983692e+00
1.06323791e+00 -4.46778834e-01 1.55612230e-01 4.91484180e-02
1.92703143e-01 -1.65967837e-01 -4.73917007e-01 -2.44200677e-01
-6.02549613e-01 6.63712844e-02 7.13152826e-01 7.10333586e-01
-4.93110716e-01 1.21027029e+00 7.40209961e+00 3.39334637e-01
-1.35924232e+00 8.24177265e-02 -4.63745803e-01 -3.78767043e-01
-3.71651888e-01 -2.83961385e-01 -4.69646931e-01 3.90615165e-01
6.15545571e-01 -2.99896002e-01 5.29419303e-01 2.73768753e-01
7.29839683e-01 -2.18576342e-01 -1.07933748e+00 1.32937503e+00
3.92277390e-01 -1.21373928e+00 3.65265965e-01 -1.64255843e-01
6.57196820e-01 1.57086432e-01 -2.29812548e-01 1.31975207e-02
2.63625115e-01 -7.64782786e-01 1.14039588e+00 8.04112375e-01
1.77470267e+00 2.51567394e-01 3.20779920e-01 -5.87795489e-02
-1.14656746e+00 2.89920628e-01 5.28781056e-01 9.43640620e-02
7.55973279e-01 -4.09973472e-01 -7.46366024e-01 2.67012715e-01
4.14133430e-01 5.97706020e-01 -5.01215816e-01 6.95997834e-01
-8.15859854e-01 9.42203641e-01 -6.15791023e-01 -1.88328326e-01
-1.08725198e-01 -2.66683269e-02 9.84885633e-01 1.17590868e+00
3.63062322e-01 1.03163168e-01 -2.91695267e-01 4.98357654e-01
1.78361118e-01 4.27586854e-01 -6.14146888e-01 -6.36620343e-01
3.58987302e-01 3.83359015e-01 -4.82222438e-01 -4.61615890e-01
-4.72017854e-01 1.45237982e+00 -4.45699096e-01 8.84777367e-01
-5.14461815e-01 -3.30072045e-01 5.19727588e-01 4.59050164e-02
-5.27114496e-02 -6.14671648e-01 -4.36501831e-01 -1.30948722e+00
6.44664645e-01 -1.35694563e+00 -6.25923723e-02 -1.26148868e+00
-1.17061329e+00 3.79491031e-01 1.48294434e-01 -1.78050768e+00
-6.56777442e-01 -9.06603634e-01 7.17325509e-02 9.90805149e-01
-1.17189729e+00 -1.65171802e+00 -4.00533974e-01 7.39410281e-01
5.08539617e-01 -2.03255862e-01 8.14545512e-01 3.86231691e-01
3.81935149e-01 9.38309133e-01 -2.38366678e-01 4.67711776e-01
1.26301730e+00 -7.89741099e-01 4.64879185e-01 7.78629243e-01
3.42905045e-01 6.46483541e-01 6.57308280e-01 -9.24717844e-01
-1.30056977e+00 -5.91745913e-01 1.53025353e+00 -8.80716264e-01
6.09969199e-01 -1.30501404e-01 -1.04470909e-01 7.94902384e-01
1.16662644e-01 -3.80860358e-01 3.70967865e-01 -6.32333577e-01
-3.56424540e-01 3.88339669e-01 -1.07145989e+00 1.17577410e+00
1.68228507e+00 -9.75075781e-01 -9.38681543e-01 1.01413913e-01
7.14848697e-01 -7.21144736e-01 -5.89066029e-01 2.68179089e-01
1.38857436e+00 -9.18158144e-02 5.14120162e-01 -7.29526639e-01
3.89748871e-01 -5.68744302e-01 -4.01748031e-01 -1.14371789e+00
1.59622297e-01 -9.97779310e-01 -1.51029125e-01 8.06888402e-01
4.99908149e-01 -6.11974299e-01 3.41613829e-01 6.83449447e-01
-5.44179976e-02 -2.00684622e-01 -1.07652736e+00 -9.58442271e-01
-2.52566576e-01 -5.77995181e-01 2.34041035e-01 7.73755610e-01
7.99391195e-02 2.72725374e-02 -6.50162220e-01 -1.76454306e-01
3.45804900e-01 4.64028708e-04 1.06976914e+00 -7.12029636e-01
-1.24562262e-02 -5.21279931e-01 -3.85281920e-01 -1.17684460e+00
-1.00336164e-01 -7.33627439e-01 2.28301957e-01 -1.88228488e+00
-2.00568244e-01 3.44109923e-01 4.27859485e-01 9.59722221e-01
3.56675088e-01 4.06549841e-01 4.91073549e-01 3.95871311e-01
-7.58580342e-02 3.85283977e-01 1.68549252e+00 -2.44923845e-01
-3.66321057e-01 -2.42953792e-01 -2.00733230e-01 6.76588833e-01
3.98386270e-01 -1.70970649e-01 -1.07518919e-01 -6.65027201e-01
6.78357184e-02 -3.64701413e-02 5.61501801e-01 -5.81336617e-01
3.36895317e-01 -2.94523388e-01 -5.76398037e-02 -7.50940979e-01
1.65596649e-01 -8.20532203e-01 1.53188467e-01 2.98540115e-01
-4.55573350e-01 6.29527122e-02 2.85897225e-01 6.80131912e-02
-9.15154219e-02 2.54721373e-01 2.90807903e-01 1.39895469e-01
-7.24044681e-01 -1.14373520e-01 -6.04959011e-01 3.67124975e-01
5.46798825e-01 -7.35562801e-01 -3.57704937e-01 -7.72081316e-01
-6.28394663e-01 8.79810378e-02 6.53001428e-01 8.11667919e-01
6.71705663e-01 -1.61934483e+00 -8.11451018e-01 3.94835502e-01
4.57094908e-01 -4.75679547e-01 -1.60918251e-01 1.01358020e+00
-6.59555614e-01 6.21133029e-01 -4.09149170e-01 -3.35156858e-01
-1.55000806e+00 -3.17722201e-01 1.02185898e-01 2.69675374e-01
-8.56245458e-01 6.61429644e-01 -5.34736514e-01 -5.92701137e-01
4.51170802e-01 -9.80182350e-01 2.99947172e-01 -4.33613025e-02
4.42308873e-01 3.54972839e-01 -2.89951444e-01 -9.35681522e-01
-5.20507514e-01 9.87861991e-01 4.52093363e-01 -9.86697853e-01
7.04109311e-01 -3.58984023e-02 2.41912737e-01 7.13783801e-01
7.23291516e-01 5.21371186e-01 -8.71214330e-01 -3.06815594e-01
-1.15957163e-01 -4.27525282e-01 -4.31797743e-01 -1.67198610e+00
-5.52003384e-01 3.75873446e-01 5.54089189e-01 -9.54769015e-01
1.08731079e+00 8.18795264e-02 9.59996402e-01 6.79957151e-01
6.23449683e-01 -1.40478945e+00 -7.42030367e-02 5.99912703e-01
1.64250159e+00 -1.28578877e+00 -2.92670697e-01 -1.33878827e-01
-7.72109091e-01 9.02346313e-01 2.23038107e-01 3.90398234e-01
3.47744405e-01 4.24925864e-01 9.79276776e-01 8.13803822e-03
-3.06787848e-01 -8.32393169e-02 8.20975661e-01 8.60127032e-01
4.24881190e-01 2.27763310e-01 -8.85174274e-01 6.89186454e-01
-6.12551868e-01 5.92688680e-01 2.34258741e-01 1.20598269e+00
2.44513631e-01 -9.66860592e-01 -4.03997749e-01 3.40245456e-01
2.36936897e-01 -2.54893571e-01 -9.63863075e-01 7.73980200e-01
-1.50952032e-02 7.32183218e-01 -4.91396397e-01 -4.53082263e-01
6.88074410e-01 4.60141838e-01 9.49844778e-01 -2.90725231e-01
-3.70159358e-01 4.48626699e-03 5.35304606e-01 -8.15588951e-01
-6.34015918e-01 -9.58053112e-01 -1.15660119e+00 1.09187894e-01
-4.45937552e-02 -5.53996921e-01 6.31620347e-01 1.13984966e+00
1.92379177e-01 2.11667657e-01 -4.38158840e-01 -6.91243529e-01
-4.65578139e-01 -1.33181989e+00 -3.12032431e-01 3.90986562e-01
4.15623963e-01 -3.74712467e-01 -3.81386578e-01 7.07158327e-01]
|
[9.179977416992188, -6.507266044616699]
|
7e2688fd-4054-4f0d-b9e6-4ab525f5c2a5
|
learning-facial-liveness-representation-for
|
2208.07828
| null |
https://arxiv.org/abs/2208.07828v1
|
https://arxiv.org/pdf/2208.07828v1.pdf
|
Learning Facial Liveness Representation for Domain Generalized Face Anti-spoofing
|
Face anti-spoofing (FAS) aims at distinguishing face spoof attacks from the authentic ones, which is typically approached by learning proper models for performing the associated classification task. In practice, one would expect such models to be generalized to FAS in different image domains. Moreover, it is not practical to assume that the type of spoof attacks would be known in advance. In this paper, we propose a deep learning model for addressing the aforementioned domain-generalized face anti-spoofing task. In particular, our proposed network is able to disentangle facial liveness representation from the irrelevant ones (i.e., facial content and image domain features). The resulting liveness representation exhibits sufficient domain invariant properties, and thus it can be applied for performing domain-generalized FAS. In our experiments, we conduct experiments on five benchmark datasets with various settings, and we verify that our model performs favorably against state-of-the-art approaches in identifying novel types of spoof attacks in unseen image domains.
|
['Yu-Chiang Frank Wang', 'Shang-Fu Chen', 'Chin-Lun Fu', 'Lin-Hsi Tsao', 'Zih-Ching Chen']
|
2022-08-16
| null | null | null | null |
['face-anti-spoofing']
|
['computer-vision']
|
[ 4.50863391e-01 -2.57625699e-01 -2.88868695e-01 -2.19042644e-01
-3.78023684e-01 -6.69012368e-01 8.09701383e-01 -1.24256477e-01
-3.87353115e-02 4.70217437e-01 -2.48492900e-02 -2.79630214e-01
8.90897214e-02 -7.15150118e-01 -5.81595242e-01 -9.84938323e-01
-7.18576014e-02 2.72988409e-01 4.83472757e-02 -3.59653592e-01
1.56262070e-01 7.14831352e-01 -1.47103310e+00 3.03436249e-01
6.11346841e-01 1.10525274e+00 -4.68223631e-01 2.64945090e-01
2.24206716e-01 7.54198253e-01 -7.83166289e-01 -6.60006344e-01
1.79305360e-01 -3.52221936e-01 -6.82448924e-01 9.10405070e-02
6.92561805e-01 -4.59148079e-01 -5.39169669e-01 1.43148434e+00
3.16358775e-01 -2.13892549e-01 7.79083550e-01 -1.56025910e+00
-6.16064489e-01 2.02657327e-01 -7.39290833e-01 2.42800102e-01
4.18404609e-01 1.00025192e-01 7.25445569e-01 -8.35296512e-01
3.98436397e-01 1.76576507e+00 3.04958403e-01 7.53413081e-01
-1.02195656e+00 -1.30962193e+00 -4.01990712e-02 2.31655419e-01
-1.29640114e+00 -8.99685502e-01 1.20640779e+00 -4.33896273e-01
1.14536166e-01 -2.15498954e-02 1.66414067e-01 1.72189593e+00
1.39901668e-01 7.50099242e-01 1.19906247e+00 -2.05154032e-01
7.19313025e-02 1.31000131e-01 -1.89776331e-01 6.11314654e-01
6.71066344e-01 4.27430660e-01 -5.11441648e-01 -3.14010501e-01
6.48741364e-01 -6.83068708e-02 -4.27576452e-01 -6.44207597e-01
-9.64002550e-01 1.01858246e+00 3.26803595e-01 1.65507779e-01
-6.86940998e-02 -1.95923075e-01 7.46711075e-01 4.50949848e-01
4.15002197e-01 1.52228206e-01 -2.97399431e-01 4.89171267e-01
-8.24052751e-01 2.53184468e-01 6.85771704e-01 5.85549593e-01
5.80291748e-01 1.06149681e-01 1.66601792e-01 6.09879971e-01
3.06748003e-01 7.58169949e-01 3.57044309e-01 -5.19461691e-01
3.49713445e-01 1.27950847e-01 1.66715682e-01 -1.59492624e+00
-2.65802622e-01 -2.36281812e-01 -1.02193427e+00 4.61235382e-02
5.97317457e-01 -6.20413618e-03 -8.37141454e-01 1.99701846e+00
4.15189832e-01 6.06399834e-01 9.95928049e-02 7.36678064e-01
7.88892806e-01 3.10768813e-01 9.03257579e-02 -3.26642722e-01
1.50491726e+00 -5.61219275e-01 -7.28583455e-01 -3.21071446e-01
4.45850819e-01 -7.71215618e-01 8.22941005e-01 4.00948137e-01
-5.01299679e-01 -5.35870254e-01 -1.07123530e+00 3.10447067e-01
-3.02246124e-01 -5.51183037e-02 6.23656213e-01 1.10207212e+00
-6.86540961e-01 4.28629994e-01 -6.06585860e-01 -3.03049624e-01
7.90012300e-01 5.56675255e-01 -8.06213021e-01 -2.10522294e-01
-1.51290262e+00 5.49311221e-01 5.29182017e-01 3.00468807e-03
-1.48700428e+00 -3.02536488e-01 -8.68892550e-01 3.30091715e-02
2.86393821e-01 -4.07625884e-01 8.79679501e-01 -1.19366443e+00
-1.27937698e+00 1.22146618e+00 -1.69227123e-01 -2.88000792e-01
4.82032806e-01 4.29631695e-02 -1.01753223e+00 4.43224490e-01
6.72573298e-02 3.12674999e-01 1.75013208e+00 -1.33468235e+00
-3.62023443e-01 -6.39652908e-01 2.28467852e-01 -4.95536357e-01
-8.73172939e-01 3.34379405e-01 6.92076311e-02 -7.08693147e-01
-3.62983570e-02 -9.95516419e-01 3.82462442e-01 2.70350605e-01
-3.07465166e-01 -1.41556039e-01 1.27833772e+00 -6.98215961e-01
8.77372324e-01 -2.34018087e+00 -1.35013089e-01 1.39429986e-01
3.46834958e-01 7.74365842e-01 -3.41946185e-01 2.06836015e-01
-4.36542720e-01 2.91757673e-01 -1.43320262e-01 -2.62875289e-01
-1.68211699e-01 -1.94435492e-02 -5.40608227e-01 1.09545755e+00
4.91580695e-01 4.26490813e-01 -1.00476491e+00 -6.96799159e-01
-8.24376661e-03 7.07940578e-01 -6.09242380e-01 2.09912077e-01
-1.00659542e-01 7.67144680e-01 -5.53021133e-01 7.75153220e-01
1.41439009e+00 -2.00553417e-01 3.56499493e-01 -1.64197698e-01
6.32740319e-01 3.81572932e-01 -9.21658635e-01 1.21496344e+00
-3.83332968e-01 5.79327285e-01 3.08057308e-01 -1.24412632e+00
9.94724393e-01 5.01937926e-01 4.18986946e-01 -4.62943822e-01
3.18344176e-01 2.79631615e-01 8.12605098e-02 -5.15590012e-01
2.17328430e-03 -1.38419718e-01 7.60654062e-02 4.72225010e-01
2.31833115e-01 1.88565537e-01 -1.18653968e-01 -5.29056042e-02
8.67716789e-01 -3.25524211e-01 2.70301312e-01 -3.52946162e-01
1.13859141e+00 -5.79577506e-01 6.33914649e-01 4.93162155e-01
-8.09347987e-01 1.71767861e-01 6.04225636e-01 -5.34043014e-01
-7.51028180e-01 -1.04363644e+00 -2.26907238e-01 1.04898047e+00
3.49055737e-01 -1.68540224e-01 -8.66855502e-01 -1.28188694e+00
1.29812032e-01 1.22972406e-01 -6.56819940e-01 -4.85516012e-01
-6.73815787e-01 -4.96620536e-01 8.85514498e-01 6.85767643e-03
6.41614079e-01 -9.57744956e-01 -1.25204235e-01 -1.89144269e-01
-2.99273044e-01 -1.40031815e+00 -2.53122956e-01 -3.57116133e-01
-4.78414267e-01 -1.45414710e+00 -6.14886582e-01 -9.53478634e-01
4.63177919e-01 6.40306413e-01 8.97694886e-01 2.65852451e-01
1.73553139e-01 -1.08642317e-01 -2.97462553e-01 -3.09500813e-01
-6.72903538e-01 -2.18364969e-02 4.32848990e-01 6.05600238e-01
5.88937879e-01 -5.38541913e-01 -6.81195378e-01 5.28915346e-01
-1.15467036e+00 -4.51796114e-01 2.88700163e-01 9.57221329e-01
-1.21640041e-01 3.75803292e-01 7.66082466e-01 -8.36168170e-01
4.96734679e-01 -5.33650994e-01 -3.61342430e-01 2.79319018e-01
-1.98552802e-01 -4.62099798e-02 7.81850159e-01 -6.56785727e-01
-9.01094973e-01 -1.02298699e-01 -1.70907766e-01 -7.20023215e-01
-3.81350636e-01 -3.61845046e-02 -6.42024100e-01 -5.20358384e-01
6.18637800e-01 2.39593446e-01 6.46452680e-02 -3.29554081e-01
9.31134224e-02 7.76323378e-01 4.39275950e-01 -5.92099071e-01
1.27591169e+00 9.41881835e-01 1.63231194e-01 -9.65688586e-01
-6.66660905e-01 -3.11053514e-01 -4.73506749e-01 -2.19678693e-02
4.99306321e-01 -8.46397638e-01 -8.80780041e-01 8.76761615e-01
-1.35981226e+00 3.89441162e-01 5.30278623e-01 1.44491956e-01
-4.70195234e-01 8.86960387e-01 -5.28271675e-01 -7.91297674e-01
-1.19195774e-01 -1.31594157e+00 1.10493433e+00 5.01180179e-02
3.11387908e-02 -1.04790223e+00 -6.77179918e-02 3.74208838e-01
6.61718100e-02 3.99357110e-01 8.85931253e-01 -1.03492117e+00
-1.22941852e-01 -2.49160454e-01 -3.69343311e-01 4.38018292e-01
4.28856343e-01 -2.94655580e-02 -1.37215304e+00 -7.92117596e-01
4.04351205e-01 -5.79403222e-01 1.03063643e+00 4.66323607e-02
1.26950443e+00 -5.31526089e-01 -4.18070555e-01 8.07149649e-01
1.03169096e+00 2.17480082e-02 6.09433472e-01 1.13643728e-01
6.44152641e-01 8.59651864e-01 6.80773914e-01 4.65731621e-01
-5.26088625e-02 7.62426674e-01 8.02283227e-01 6.23691082e-02
1.71749100e-01 -3.80152643e-01 4.56738949e-01 1.46438226e-01
3.56308460e-01 -5.24663150e-01 -7.19493330e-01 4.42490578e-01
-1.38027966e+00 -1.00580084e+00 3.55079621e-01 2.07920790e+00
4.89990175e-01 4.00961302e-02 2.52211958e-01 4.05192494e-01
1.11747015e+00 7.17397869e-01 -6.47860825e-01 -3.31454158e-01
-1.45328596e-01 2.02185154e-01 2.74517536e-01 1.20783500e-01
-1.68632269e+00 1.11824226e+00 5.43525457e+00 9.43717480e-01
-1.38552868e+00 1.50441900e-01 6.16152465e-01 3.26352835e-01
-2.01512218e-01 -1.43859610e-01 -6.72263205e-01 5.88504255e-01
6.27616823e-01 -1.15428828e-02 4.88486469e-01 7.92040765e-01
-2.98723043e-03 6.02285147e-01 -1.04406130e+00 1.18206799e+00
2.50253648e-01 -1.00906706e+00 2.02081501e-01 2.30141282e-01
5.61105967e-01 -4.24856901e-01 4.32228565e-01 -5.11016399e-02
2.62749851e-01 -1.12747741e+00 4.63846266e-01 -1.96014002e-01
1.00476348e+00 -8.61338675e-01 5.35887659e-01 1.67787820e-01
-1.01834476e+00 -4.53472704e-01 -4.58354115e-01 3.39452177e-01
-2.22056359e-01 3.81829858e-01 -7.86326170e-01 4.87817526e-01
5.58987200e-01 7.41191208e-01 -3.43284428e-01 4.91441488e-01
-2.84727424e-01 5.90432405e-01 -7.14202672e-02 3.94633979e-01
-5.56346821e-03 2.38656342e-01 4.68642354e-01 1.01588774e+00
2.63970613e-01 -3.21738362e-01 -1.86781511e-02 6.21095359e-01
-3.68959367e-01 -3.97352837e-02 -9.27968264e-01 -4.35348988e-01
5.86354196e-01 1.05189872e+00 -4.14896905e-01 2.67912447e-03
-3.03857505e-01 9.91430879e-01 1.83465198e-01 2.62560725e-01
-7.72613406e-01 -3.18694293e-01 1.31947660e+00 1.40817925e-01
3.39237720e-01 1.09531432e-02 1.40730783e-01 -1.47980547e+00
-1.84341818e-01 -1.39458442e+00 5.96836448e-01 -2.22674698e-01
-1.57985532e+00 6.68656290e-01 -3.87281030e-01 -1.17100263e+00
-2.61606932e-01 -8.13565850e-01 -4.55716133e-01 6.26342893e-01
-1.87759852e+00 -1.35547101e+00 -2.11711675e-01 9.97980297e-01
1.75602347e-01 -3.35761815e-01 7.03673720e-01 3.70755196e-01
-6.26083374e-01 9.98303354e-01 -3.70643325e-02 6.79369152e-01
9.90175307e-01 -4.32232440e-01 4.31785911e-01 9.42448854e-01
8.40145275e-02 7.76789725e-01 7.76781917e-01 -4.83920336e-01
-1.37393486e+00 -9.96396363e-01 7.23609507e-01 -1.81695521e-01
8.22983623e-01 -4.07093912e-01 -9.51751590e-01 4.64713067e-01
-1.35862872e-01 4.99474794e-01 6.12242043e-01 -2.90547639e-01
-1.04582393e+00 -2.28552207e-01 -1.53489757e+00 2.72795588e-01
1.11092198e+00 -9.20746148e-01 -1.73543960e-01 4.96852845e-01
5.66151977e-01 1.10632991e-02 -5.42385042e-01 5.67988157e-01
7.79395401e-01 -1.13159359e+00 1.28375947e+00 -9.63720083e-01
5.06529748e-01 -9.12051741e-03 -1.98121592e-01 -1.03820872e+00
-2.47271195e-01 -7.75609851e-01 -3.17912608e-01 1.24311423e+00
-2.65660793e-01 -8.66133928e-01 8.15654516e-01 -1.61344662e-01
5.14152706e-01 -3.25297475e-01 -1.02731133e+00 -9.40072715e-01
3.33469629e-01 -6.03923053e-02 1.10748899e+00 1.24441981e+00
-2.81757146e-01 -3.89120989e-02 -7.12907553e-01 6.00263715e-01
9.18794513e-01 4.09755930e-02 8.14531982e-01 -1.46506751e+00
-1.68597937e-01 -3.69128585e-01 -5.66393912e-01 -1.06880021e+00
8.77694368e-01 -5.89303374e-01 -3.54146242e-01 -5.73146105e-01
8.49505737e-02 -4.33937073e-01 -5.71505368e-01 1.51321188e-01
-1.54305547e-01 4.08420742e-01 1.40856549e-01 3.07225794e-01
-1.60569727e-01 4.58757401e-01 1.38174129e+00 -4.45489258e-01
2.49664441e-01 1.96196496e-01 -8.52169335e-01 7.38482535e-01
7.83794880e-01 -6.44657969e-01 -4.43826735e-01 -2.94191480e-01
-1.25493214e-01 9.34303105e-02 6.07104659e-01 -9.08575296e-01
-1.26331702e-01 -4.36341792e-01 2.43080512e-01 -1.44226313e-01
3.80219549e-01 -9.41629171e-01 -3.91421199e-01 6.46113753e-01
-2.47456685e-01 -3.83886158e-01 -4.91404049e-02 6.34945810e-01
-2.16495603e-01 5.34871817e-02 1.25486517e+00 9.44684353e-03
-4.56756592e-01 7.00805902e-01 7.90550113e-02 1.02895528e-01
9.14763987e-01 -1.02574907e-01 -6.23136759e-01 -5.48117876e-01
-4.40334082e-01 -2.43098065e-01 6.37411535e-01 5.86485803e-01
7.47338712e-01 -1.27869630e+00 -7.79589951e-01 7.08987117e-01
3.40906262e-01 -5.47283888e-01 1.14923373e-01 4.92628187e-01
-4.87642288e-01 4.98302132e-01 -3.68595541e-01 -5.14024794e-01
-1.41765368e+00 1.25160110e+00 2.99806863e-01 -2.33542308e-01
-5.94716556e-02 6.55585408e-01 6.82077646e-01 -2.56441414e-01
5.13946414e-02 3.04124236e-01 -2.92611301e-01 2.57977098e-02
8.39482725e-01 1.38523191e-01 -4.02986370e-02 -1.19319713e+00
-6.48703992e-01 4.06983852e-01 -1.80150270e-01 2.64370710e-01
1.05283201e+00 -5.40362373e-02 -1.32025942e-01 -3.01302850e-01
1.60841608e+00 3.21172029e-02 -1.13403571e+00 -4.06055331e-01
-4.26803045e-02 -8.90734971e-01 -2.58380622e-01 -2.82831609e-01
-1.39972126e+00 1.25626338e+00 6.53378963e-01 2.26157695e-01
1.38565290e+00 -3.61045823e-02 9.73590076e-01 1.88174218e-01
4.92675424e-01 -5.04572272e-01 2.53388345e-01 2.10914701e-01
6.69173837e-01 -1.48116112e+00 -2.14263767e-01 -6.77089155e-01
-2.33669460e-01 1.08442307e+00 4.40744102e-01 -1.96835980e-01
8.39612424e-01 -6.61893114e-02 -6.66867569e-02 -1.33946374e-01
-2.84823209e-01 8.55923742e-02 8.90499055e-02 1.11479032e+00
1.03238165e-01 6.18539527e-02 -9.05585848e-03 2.44406000e-01
-9.30417031e-02 -1.44157559e-01 2.35675871e-01 6.39513910e-01
-2.03003228e-01 -1.31613791e+00 -5.86646020e-01 1.76584590e-02
-7.88979232e-01 2.75104225e-01 -5.73111594e-01 5.68909764e-01
2.15805739e-01 1.29743171e+00 -2.02437431e-01 -6.45034254e-01
-2.04939649e-01 -1.86574414e-01 4.69185710e-01 -3.25645030e-01
-1.75241649e-01 -1.48647010e-01 -1.03163421e-01 -4.18758452e-01
-4.47459668e-01 -4.25062954e-01 -5.75889468e-01 -7.64433622e-01
-3.31837773e-01 -8.24051816e-03 4.28331763e-01 9.03865278e-01
2.21450642e-01 -5.63736334e-02 1.30977345e+00 -7.06072986e-01
-8.55543256e-01 -6.25723124e-01 -5.85861862e-01 9.18839335e-01
8.62103641e-01 -1.14419091e+00 -6.42853975e-01 -6.48380220e-02]
|
[13.051520347595215, 1.1777169704437256]
|
da3fce41-084b-46ae-972d-7f0560e2947b
|
feathers-dataset-for-fine-grained-visual
|
2004.08606
| null |
https://arxiv.org/abs/2004.08606v1
|
https://arxiv.org/pdf/2004.08606v1.pdf
|
Feathers dataset for Fine-Grained Visual Categorization
|
This paper introduces a novel dataset FeatherV1, containing 28,272 images of feathers categorized by 595 bird species. It was created to perform taxonomic identification of bird species by a single feather, which can be applied in amateur and professional ornithology. FeatherV1 is the first publicly available bird's plumage dataset for machine learning, and it can raise interest for a new task in fine-grained visual recognition domain. The latest version of the dataset can be downloaded at https://github.com/feathers-dataset/feathersv1-dataset. We also present feathers classification task results. We selected several deep learning architectures (DenseNet based) for categorical crossentropy values comparison on the provided dataset.
|
['Konstantin Dobratulin', 'Andrey Kuznetsov', 'Alina Belko']
|
2020-04-18
| null | null | null | null |
['fine-grained-visual-recognition', 'fine-grained-visual-categorization']
|
['computer-vision', 'computer-vision']
|
[-2.46656716e-01 -4.81305957e-01 1.37414753e-01 -5.58305025e-01
2.85983980e-01 -7.73031533e-01 5.38497090e-01 1.66577861e-01
-5.47277927e-01 3.82569909e-01 1.18964590e-01 2.29282290e-01
-3.03726465e-01 -9.54561114e-01 -4.99871224e-01 -5.29266179e-01
-3.72342736e-01 2.29294926e-01 1.45707903e-02 -2.82206714e-01
2.06961960e-01 4.66182679e-01 -1.91039884e+00 2.84126818e-01
4.32545006e-01 1.00515151e+00 2.42966890e-01 9.94806767e-01
4.73912477e-01 4.24815685e-01 -4.30175275e-01 -6.73034966e-01
5.22489607e-01 -3.34565252e-01 -7.19811499e-01 -5.37825346e-01
7.52685428e-01 -4.23263162e-01 2.54488051e-01 1.33853650e+00
4.15756613e-01 -6.64038807e-02 1.06077027e+00 -1.13297558e+00
-1.16183186e+00 7.59500384e-01 -5.54501653e-01 4.38777864e-01
-1.65983304e-01 1.99699759e-01 1.37601328e+00 -6.28362656e-01
4.20345843e-01 1.04373538e+00 9.82637942e-01 4.95747060e-01
-1.15953374e+00 -6.91716969e-01 -2.44659945e-01 3.75954300e-01
-1.53627789e+00 -3.16985883e-02 6.38679624e-01 -7.75404274e-01
8.00059319e-01 6.40747607e-01 1.27584648e+00 8.83668184e-01
8.32391158e-02 3.87944758e-01 1.39342237e+00 -9.63469744e-02
-1.75991133e-01 1.50689229e-01 4.42696869e-01 1.13598394e+00
2.82821178e-01 6.71527982e-01 -2.14578688e-01 4.98788990e-02
7.61541247e-01 2.08429649e-01 4.87588421e-02 -2.59957343e-01
-1.21989727e+00 1.03794539e+00 1.21861291e+00 1.11004464e-01
8.01068842e-02 -5.98906986e-02 3.39613140e-01 6.28337622e-01
5.13181925e-01 5.22136688e-01 -1.89399883e-01 1.54790580e-01
-7.45365262e-01 4.98316288e-01 7.19780743e-01 5.81136763e-01
5.83943188e-01 4.36682105e-02 7.88806528e-02 1.35386896e+00
6.24571502e-01 6.64443254e-01 4.45618719e-01 -4.43132728e-01
-4.57476854e-01 5.34442365e-01 -2.10632935e-01 -9.43581343e-01
-7.72014797e-01 -3.17319304e-01 -9.49901462e-01 5.26741564e-01
5.21230102e-01 4.20932425e-03 -6.64872885e-01 1.28659511e+00
3.33389789e-01 -1.47789702e-01 -3.86336207e-01 1.39557981e+00
1.67902339e+00 4.63499069e-01 -2.08550677e-01 5.78508556e-01
1.65132642e+00 -1.12517917e+00 -6.99234381e-02 2.43540272e-01
-1.30295590e-01 -7.62079537e-01 1.04075801e+00 1.69139877e-01
-4.17300373e-01 -7.21926093e-01 -1.30808210e+00 -4.38438766e-02
-8.65574181e-01 4.43156511e-01 5.61411083e-01 6.30757749e-01
-1.32990789e+00 5.13639629e-01 -5.21466434e-01 -7.12487221e-01
5.95910311e-01 1.97181404e-01 -3.16159964e-01 5.35454988e-01
-8.24887395e-01 7.53026128e-01 2.47713327e-01 3.20324093e-01
-1.47785103e+00 -6.44483149e-01 -7.46008515e-01 -1.66889325e-01
-1.51228577e-01 -4.83550638e-01 1.04667342e+00 -9.34350967e-01
-1.19680500e+00 1.45935392e+00 6.77504838e-01 -7.77749002e-01
1.94140047e-01 -2.00299304e-02 -4.33353215e-01 -7.52975270e-02
-1.26750380e-01 8.11441541e-01 1.25675619e+00 -8.79903615e-01
-4.99148965e-01 -2.58792996e-01 5.55140153e-02 -3.89676057e-02
-5.71949258e-02 3.78191024e-01 4.61073995e-01 -1.03507495e+00
-7.20373869e-01 -9.98056352e-01 1.77202985e-01 4.03072327e-01
-2.71441519e-01 -3.61489475e-01 3.96796376e-01 -4.89037067e-01
8.83626759e-01 -2.06758308e+00 1.26569688e-01 -8.32351595e-02
5.21354973e-01 3.73921454e-01 -3.59043211e-01 5.16560853e-01
-7.27168098e-02 9.12760049e-02 -3.54515135e-01 1.19180784e-01
2.36025527e-01 -4.87065278e-02 -3.69635746e-02 8.54058027e-01
2.64971107e-01 8.13218176e-01 -7.69306064e-01 -3.28211457e-01
4.52788174e-01 3.86218995e-01 -4.08290654e-01 3.52524996e-01
1.31641813e-02 5.18394522e-02 1.25104800e-01 8.75460386e-01
8.46831918e-01 -6.69540912e-02 -2.10866690e-01 -1.99378356e-01
-3.61029178e-01 -2.12446764e-01 -7.28872061e-01 1.25827742e+00
-1.95992529e-01 8.31508517e-01 1.84516981e-01 -5.21741390e-01
9.79438245e-01 -4.87177283e-01 1.56233549e-01 -7.36393183e-02
5.32523632e-01 -8.41981471e-02 5.20248294e-01 -1.71810359e-01
4.80175793e-01 -2.32779190e-01 -1.24071054e-01 4.03041720e-01
4.06548262e-01 -2.09704235e-01 3.47638935e-01 -2.30829582e-01
2.60841995e-01 4.30115648e-02 5.67416728e-01 -9.81812716e-01
3.21699351e-01 2.15181299e-02 1.23387344e-01 4.64624554e-01
-6.47344887e-01 5.49265504e-01 1.91012591e-01 -1.26057231e+00
-9.74660575e-01 -1.31177711e+00 -8.25035393e-01 1.57957411e+00
1.56490773e-01 -3.34390968e-01 -9.05560613e-01 -5.68609774e-01
2.95325726e-01 -5.79653727e-03 -1.19474804e+00 -5.26074022e-02
2.77215391e-02 -1.18374181e+00 8.17273736e-01 4.04363394e-01
6.82460845e-01 -1.48785913e+00 -7.59331048e-01 -4.47124004e-01
3.47148716e-01 -2.37622872e-01 -6.94686055e-01 1.88835591e-01
-2.37017140e-01 -1.50805950e+00 -6.92706823e-01 -1.08203030e+00
4.65445578e-01 2.78461166e-02 1.27998710e+00 1.45533651e-01
-9.69045639e-01 -3.24209966e-02 -5.65373063e-01 -5.13117611e-01
-2.66361028e-01 9.02336761e-02 2.03444541e-01 2.29985062e-02
7.07752347e-01 -2.75956184e-01 -6.41605616e-01 3.61988932e-01
-4.97130781e-01 -5.43063059e-02 4.62585121e-01 1.14600956e+00
6.67776644e-01 -3.07585709e-02 4.36088622e-01 -6.62284911e-01
5.52898109e-01 -4.87284184e-01 -9.57048178e-01 -7.05801770e-02
-3.02218258e-01 -7.87030756e-02 8.79778445e-01 -1.53543830e-01
-6.34418428e-01 -1.19640447e-01 -6.82846546e-01 -6.02742936e-03
-5.75772405e-01 1.14354819e-01 4.47383195e-01 -3.26200724e-01
8.96107018e-01 5.42323291e-02 5.70053793e-02 -8.65104258e-01
2.56211787e-01 7.21155345e-01 4.74259883e-01 -1.49663001e-01
6.37407243e-01 2.95933902e-01 -1.90654233e-01 -1.20843244e+00
-7.02063680e-01 -4.39821720e-01 -9.70758796e-01 -4.28173602e-01
1.05145323e+00 -9.25736129e-01 -1.02347529e+00 9.70737696e-01
-4.54467595e-01 -5.60752988e-01 -3.37477595e-01 2.17099994e-01
-3.83586735e-01 -4.69636172e-03 -4.88188237e-01 -4.74351525e-01
-6.54609978e-01 -6.40359998e-01 1.06090093e+00 4.04160202e-01
-5.62281124e-02 -9.53152537e-01 5.54442525e-01 1.39760643e-01
2.37089887e-01 5.20492196e-01 4.91507024e-01 -5.21450222e-01
-8.14702883e-02 7.08524436e-02 -4.03375179e-01 6.01319790e-01
6.77574724e-02 4.08506662e-01 -1.02492297e+00 -6.10006988e-01
-3.80465865e-01 -5.69432139e-01 1.39404929e+00 2.95496196e-01
1.13814139e+00 -4.24037158e-01 1.92343161e-01 1.13972425e+00
1.56558800e+00 -3.15619819e-02 -8.92715305e-02 9.90210921e-02
8.68070900e-01 5.16419113e-01 3.77303153e-01 4.97279435e-01
5.71126282e-01 3.93526644e-01 8.27185512e-01 -1.42269000e-01
-3.95820796e-01 -2.81924158e-01 1.11608222e-01 7.11524189e-01
-4.71317500e-01 7.20389634e-02 -8.76793981e-01 8.37242246e-01
-1.35793889e+00 -1.08303308e+00 -1.90533549e-01 1.84404922e+00
6.67690992e-01 -5.39347470e-01 8.33794177e-01 -1.79882318e-01
6.71703994e-01 4.41128403e-01 -7.00847387e-01 -7.19983041e-01
-2.78770346e-02 2.15282798e-01 5.24587631e-01 3.13920975e-01
-1.61929679e+00 8.40397000e-01 6.81052065e+00 5.59508324e-01
-9.33599532e-01 3.66997048e-02 2.22853169e-01 -3.95691246e-01
2.59249508e-01 -6.43435359e-01 -8.54408145e-01 6.28099024e-01
8.79823864e-01 -2.14031655e-02 7.64956951e-01 9.61108983e-01
-4.11104321e-01 1.18700089e-02 -8.06284547e-01 1.05678177e+00
2.14474916e-01 -1.03512406e+00 -1.20424666e-01 -1.05980307e-01
6.08786345e-01 5.98137319e-01 3.02722752e-01 8.42458159e-02
6.04314029e-01 -1.26453221e+00 6.06372654e-01 3.88905793e-01
1.00856650e+00 -8.33293498e-01 8.36698353e-01 2.76277568e-02
-1.57069850e+00 -8.82849991e-02 -9.54986453e-01 -1.87646672e-01
-5.56235135e-01 7.88201615e-02 -7.40317404e-01 1.63908973e-01
1.61346805e+00 1.22651148e+00 -1.29874504e+00 1.19036806e+00
8.54877904e-02 6.03669882e-01 -3.21630090e-01 -7.49444008e-01
-1.47070354e-02 -3.14644367e-01 4.33041096e-01 1.31363201e+00
-3.04847378e-02 -4.39156055e-01 4.18351851e-02 9.52351391e-01
-4.74320352e-02 2.44827181e-01 -6.90470695e-01 -2.49022871e-01
1.21149965e-01 1.70736754e+00 -7.58990169e-01 -5.99269792e-02
-4.72176671e-01 7.49791145e-01 5.21884918e-01 -1.49965182e-01
-8.21727157e-01 -6.36736989e-01 1.25630403e+00 1.38571680e-01
3.72156382e-01 9.45646986e-02 1.60195574e-01 -1.07438087e+00
-5.04051208e-01 -7.64975250e-01 6.46012008e-01 -5.73931694e-01
-1.96057165e+00 8.82642210e-01 1.97677985e-02 -1.29974186e+00
1.38609903e-02 -1.20386076e+00 -2.68272549e-01 6.65064454e-01
-1.47932184e+00 -1.50062168e+00 -5.15173912e-01 6.08749390e-01
2.17861429e-01 -5.06213963e-01 8.98119569e-01 1.39612913e-01
-2.03308970e-01 5.11657357e-01 3.92058194e-01 4.87407327e-01
8.05125237e-01 -1.76636457e+00 7.02805340e-01 4.87778097e-01
4.63319272e-01 4.24549967e-01 3.73535752e-01 -3.23494881e-01
-1.02902687e+00 -1.39450812e+00 1.64033055e-01 -6.96067274e-01
8.52297902e-01 -5.48705637e-01 -4.25561637e-01 5.36128581e-01
5.19112527e-01 1.00996286e-01 1.16419375e+00 7.43227964e-03
-3.62630248e-01 -3.45918685e-01 -1.33275032e+00 2.54314333e-01
1.04525030e+00 -5.57358861e-01 -6.99772298e-01 3.91649157e-01
2.53516287e-01 -1.88446239e-01 -1.25489473e+00 1.83587879e-01
9.40319419e-01 -1.12746716e+00 9.72618222e-01 -4.14391398e-01
4.57808882e-01 -5.96541107e-01 -8.74435231e-02 -1.59494078e+00
-6.86066389e-01 1.94457974e-02 7.88109601e-02 7.88007855e-01
4.23773713e-02 -6.55597925e-01 2.67331362e-01 -4.94845897e-01
-1.62393674e-01 -4.82318163e-01 -6.39082015e-01 -7.00681210e-01
2.52984792e-01 5.83982058e-02 9.68242824e-01 6.42420590e-01
-4.24032003e-01 3.57399940e-01 -6.60666883e-01 4.17524166e-02
9.29505348e-01 8.80423784e-01 6.11523092e-01 -1.68715179e+00
-1.30116522e-01 -7.05136895e-01 -7.41892517e-01 -5.59907198e-01
-4.55340669e-02 -1.34214497e+00 2.33078793e-01 -1.39679813e+00
2.18195304e-01 -4.33812998e-02 -3.81679088e-01 6.33875549e-01
3.72037925e-02 1.16190660e+00 1.31486669e-01 -9.69748944e-02
-3.47091258e-01 4.50354040e-01 1.21473718e+00 -2.74775624e-01
1.93765178e-01 1.01191841e-01 -6.89717948e-01 7.28312850e-01
1.07743776e+00 -8.90097171e-02 -1.62027225e-01 -3.29704672e-01
2.31199712e-02 -7.37007082e-01 8.23173821e-01 -1.04331398e+00
-2.56440997e-01 6.93338811e-02 5.67396581e-01 -6.40343130e-01
2.29239479e-01 -8.85669529e-01 3.99978817e-01 7.65704215e-01
-2.57567883e-01 9.22320113e-02 2.01500311e-01 3.65281701e-01
-1.61633372e-01 -2.28387237e-01 1.43098700e+00 -2.69222736e-01
-1.05223823e+00 5.66648126e-01 -4.61962849e-01 1.40950933e-01
1.08700681e+00 1.95453078e-01 -7.38358200e-01 2.99826503e-01
-6.74181819e-01 1.26367509e-02 8.09255004e-01 6.65631711e-01
6.59870267e-01 -1.35979462e+00 -1.11813188e+00 8.67831647e-01
4.47711319e-01 -6.27126575e-01 4.20057327e-01 4.57679033e-01
-1.18999588e+00 1.97629869e-01 -1.14189327e+00 -6.37492180e-01
-1.71051836e+00 6.98251724e-01 5.81204355e-01 4.23608035e-01
-6.83223963e-01 1.31486964e+00 3.35760891e-01 -8.16613674e-01
-2.07492456e-01 -4.38975483e-01 -7.36727774e-01 2.43551388e-01
8.15113544e-01 2.75119156e-01 -2.09411010e-01 -1.05178154e+00
-5.96419752e-01 7.78252482e-01 7.95437247e-02 5.09381056e-01
1.63776791e+00 3.74706872e-02 -5.11768818e-01 5.33562899e-01
1.16527271e+00 -1.81897521e-01 -1.09923398e+00 2.84046799e-01
-4.12282377e-01 -6.75845861e-01 -1.55683950e-01 -9.64968324e-01
-1.01174998e+00 8.80162716e-01 1.11451519e+00 6.84659123e-01
1.16264844e+00 2.50659361e-02 3.82450014e-01 3.33020777e-01
3.10081542e-01 -7.77272046e-01 -2.41009504e-01 3.26148033e-01
1.23061323e+00 -1.52021039e+00 7.88908526e-02 1.46350622e-01
-7.52668202e-01 8.55206251e-01 6.11079991e-01 -5.98007023e-01
1.13276219e+00 3.48714769e-01 2.79185712e-01 -3.56963575e-01
-6.36620343e-01 -5.13832331e-01 8.17026079e-01 1.04578185e+00
4.55600441e-01 5.90330064e-01 -1.39272541e-01 8.37695897e-01
-8.47866118e-01 -4.70091879e-01 3.18116039e-01 1.92463741e-01
-3.90780807e-01 -5.65355659e-01 -1.33258089e-01 8.50274026e-01
-3.33563864e-01 -3.91635954e-01 -8.76595318e-01 5.82247257e-01
3.69837761e-01 6.37815416e-01 2.74963856e-01 -6.06227160e-01
1.29190236e-01 -3.82025927e-01 4.90416169e-01 -6.73591435e-01
-1.05504191e+00 -3.65976721e-01 1.11355416e-01 -3.23489338e-01
-7.08225846e-01 -4.74280089e-01 -6.08232200e-01 -7.12323964e-01
8.21341276e-02 8.33512396e-02 3.89259100e-01 6.63804188e-02
-1.89508140e-01 3.45745087e-01 5.35664737e-01 -8.72162342e-01
-2.45287046e-01 -1.13332462e+00 -1.22291470e+00 2.61043429e-01
7.41141975e-01 -8.14566970e-01 -3.39029044e-01 2.13257402e-01]
|
[9.790682792663574, 2.2142117023468018]
|
68434a00-e67d-4be7-8678-86eab0fb2ae6
|
type-prediction-systems
|
2104.01207
| null |
https://arxiv.org/abs/2104.01207v1
|
https://arxiv.org/pdf/2104.01207v1.pdf
|
Type Prediction Systems
|
Inferring semantic types for entity mentions within text documents is an important asset for many downstream NLP tasks, such as Semantic Role Labelling, Entity Disambiguation, Knowledge Base Question Answering, etc. Prior works have mostly focused on supervised solutions that generally operate on relatively small-to-medium-sized type systems. In this work, we describe two systems aimed at predicting type information for the following two tasks, namely, a TypeSuggest module, an unsupervised system designed to predict types for a set of user-entered query terms, and an Answer Type prediction module, that provides a solution for the task of determining the correct type of the answer expected to a given query. Our systems generalize to arbitrary type systems of any sizes, thereby making it a highly appealing solution to extract type information at any granularity.
|
['Mustafa Canim', 'Alfio Gliozzo', 'Nandana Mihindukulasooriya', 'Sarthak Dash']
|
2021-04-02
| null | null | null | null |
['type-prediction', 'knowledge-base-question-answering']
|
['computer-code', 'natural-language-processing']
|
[ 1.04717147e-02 6.82632148e-01 -4.81588334e-01 -4.78588283e-01
-5.91921151e-01 -8.11862171e-01 6.76979721e-01 7.91182637e-01
-5.40295482e-01 1.05930793e+00 6.37341151e-03 -5.61184168e-01
-1.62556529e-01 -1.19243228e+00 -5.11798561e-01 -3.57704580e-01
6.52786419e-02 9.89309371e-01 6.26577795e-01 -4.23968285e-01
3.18773389e-01 2.62179703e-01 -1.91196585e+00 5.87836981e-01
8.06886852e-01 1.15008831e+00 9.13801342e-02 4.99855191e-01
-8.29214156e-01 1.08731139e+00 -6.23767078e-01 -5.52776158e-01
-3.45303863e-01 6.61530346e-02 -1.41676223e+00 -4.97683555e-01
2.53351569e-01 3.04914355e-01 1.53670758e-01 1.02966785e+00
3.25504363e-01 2.46111572e-01 9.22304809e-01 -1.33764195e+00
-1.70641616e-01 8.03726494e-01 1.78294405e-01 1.38865605e-01
9.59524333e-01 -4.16059256e-01 1.62009859e+00 -5.87096632e-01
9.47691083e-01 1.37392128e+00 4.64813530e-01 6.96357012e-01
-9.51384008e-01 -1.53821766e-01 -8.49474370e-02 4.88869607e-01
-1.13707089e+00 -3.93299282e-01 3.16100687e-01 -4.38737810e-01
1.23827291e+00 6.74007356e-01 1.14668295e-01 4.69777167e-01
-3.41633052e-01 8.20590556e-01 8.77544165e-01 -5.18466234e-01
5.41388988e-01 6.65695310e-01 6.14820480e-01 4.35865700e-01
3.69086057e-01 -5.12792468e-01 -2.26172104e-01 -6.14564598e-01
3.03627729e-01 -5.82952499e-01 -2.82118358e-02 -5.90991862e-02
-8.40116143e-01 6.74850345e-01 4.87855434e-01 6.26620531e-01
-3.81789267e-01 -7.76572749e-02 3.52588832e-01 3.79954875e-01
4.22397405e-01 1.04145706e+00 -1.26107287e+00 1.21942014e-01
-3.64047527e-01 6.53054297e-01 1.53573012e+00 1.20000935e+00
1.04492176e+00 -8.33509147e-01 -4.42403466e-01 8.64357591e-01
3.43861252e-01 2.90441096e-01 5.23769796e-01 -6.35710835e-01
4.35782045e-01 1.23034871e+00 4.27875251e-01 -6.77561939e-01
-5.74295640e-01 9.64176562e-03 -1.92609698e-01 -4.78244841e-01
6.06179774e-01 -2.81194061e-01 -6.16910160e-01 1.52986836e+00
8.55913818e-01 -2.55530700e-02 3.06357861e-01 7.22542167e-01
1.37586296e+00 3.70348454e-01 3.60324770e-01 -8.65494907e-02
1.93851793e+00 -5.16429543e-01 -5.38604140e-01 -2.74096698e-01
1.18054175e+00 -5.59064090e-01 7.39908159e-01 -1.40966162e-01
-7.07132757e-01 -6.12505153e-02 -3.58930230e-01 -4.07247275e-01
-1.11291325e+00 -9.81038958e-02 7.32633591e-01 4.68179345e-01
-8.65940630e-01 3.58112991e-01 -6.98135719e-02 -3.14084083e-01
-5.24807498e-02 4.26355869e-01 -3.09595048e-01 3.65528874e-02
-1.68212759e+00 9.09992695e-01 7.07538068e-01 -1.56629741e-01
-9.84662548e-02 -8.41115236e-01 -9.14201736e-01 3.02561015e-01
7.17251301e-01 -1.11699629e+00 1.33663106e+00 -6.70373023e-01
-8.64860296e-01 1.15085924e+00 -6.95321739e-01 -5.39001167e-01
-4.27456535e-02 8.55273232e-02 -5.25590479e-01 -1.51899084e-02
4.92584080e-01 4.60304201e-01 5.28545916e-01 -1.07871735e+00
-1.14156497e+00 -6.42371118e-01 5.06017566e-01 3.48941058e-01
-6.24338649e-02 1.60945043e-01 -3.20337266e-01 -3.49763125e-01
2.27531287e-04 -6.41275167e-01 -2.55318433e-01 -5.02556980e-01
-5.36800027e-01 -1.02091777e+00 5.06357491e-01 -4.48515564e-01
1.29738545e+00 -1.69780016e+00 -2.78880261e-02 3.16534191e-01
2.66716182e-01 1.35672167e-01 3.50903600e-01 2.46603921e-01
-5.96380010e-02 3.79766673e-01 -2.42097229e-02 1.36954188e-01
2.90110677e-01 3.96266907e-01 -3.74728382e-01 -3.11368018e-01
8.91774073e-02 8.09776366e-01 -1.07265472e+00 -6.45086706e-01
-2.74309576e-01 -2.86738634e-01 -5.85903406e-01 5.26731670e-01
-1.10841358e+00 1.36411171e-02 -8.70537221e-01 5.69081366e-01
2.37455353e-01 -1.74642429e-01 4.85597253e-01 -6.32208660e-02
-1.26937717e-01 8.48959625e-01 -1.28023887e+00 9.82637107e-01
-5.49775779e-01 5.48062101e-02 1.96411103e-01 -1.09064257e+00
6.75647497e-01 4.98607755e-01 4.19223815e-01 -3.59829396e-01
1.35266529e-02 6.33576274e-01 -1.36794060e-01 -9.15473580e-01
9.04632270e-01 -9.73509476e-02 -4.67435271e-01 2.19185069e-01
2.44684629e-02 4.50895540e-02 6.59687757e-01 1.71150640e-01
1.16761374e+00 -1.69445142e-01 4.66946334e-01 -3.13154846e-01
8.63994360e-01 4.41014081e-01 5.05771220e-01 7.56764948e-01
2.71547347e-01 -1.32926321e-02 7.27676570e-01 -2.19621345e-01
-5.65750182e-01 -5.31831324e-01 -2.51270175e-01 1.23824215e+00
3.74225974e-01 -3.86510491e-01 -6.02801800e-01 -9.44293857e-01
2.67953992e-01 7.30298698e-01 -3.71336401e-01 1.23105787e-01
-5.89090407e-01 -6.10447645e-01 5.72388053e-01 3.69661331e-01
3.36326718e-01 -1.26213229e+00 -1.82028279e-01 2.08366826e-01
-5.72644651e-01 -1.07711625e+00 1.44317048e-02 2.22482383e-01
-6.20501220e-01 -1.37568784e+00 -3.22808653e-01 -1.14550412e+00
5.34227192e-01 -2.55615920e-01 1.31609070e+00 2.97090620e-01
3.03624794e-02 1.19862124e-01 -4.10173029e-01 -5.11470735e-01
-3.42066467e-01 4.94616598e-01 -1.56528622e-01 4.14505526e-02
8.34127665e-01 -1.62568823e-01 -3.49930167e-01 2.74464518e-01
-7.82446384e-01 -4.53380406e-01 3.88981491e-01 5.34020543e-01
3.87144357e-01 1.74973205e-01 6.89479113e-01 -1.66528857e+00
6.20670438e-01 -9.09214973e-01 -3.60693634e-01 7.23632038e-01
-7.82576144e-01 5.55725574e-01 7.28259683e-01 -1.07253484e-01
-1.31707537e+00 -1.48358002e-01 -6.77734911e-01 4.40129906e-01
-5.79996407e-01 6.93623304e-01 -6.87855482e-01 3.04718971e-01
7.87001431e-01 1.17149130e-01 -4.58591521e-01 -6.25233233e-01
5.29462159e-01 9.11524892e-01 4.02367830e-01 -8.17970216e-01
5.98601699e-01 8.91773850e-02 -8.57264027e-02 -8.81835639e-01
-1.11641240e+00 -1.44258118e+00 -5.38505435e-01 1.88427821e-01
7.09976971e-01 -6.64645672e-01 -7.72650719e-01 3.32016408e-01
-1.00979972e+00 -2.83387363e-01 -2.42798656e-01 4.26787697e-02
-4.82351661e-01 2.14696571e-01 -4.38113183e-01 -6.46328449e-01
-2.99195141e-01 -5.59792161e-01 1.35112906e+00 2.46362343e-01
-3.87738407e-01 -1.37874949e+00 4.72971573e-02 5.71025848e-01
1.63224593e-01 -2.22932965e-01 1.41608703e+00 -1.33841097e+00
-4.72409308e-01 -1.69254541e-01 -2.57283330e-01 -2.32655406e-01
-7.90949985e-02 -5.11677146e-01 -9.48103070e-01 3.64053518e-01
-1.22637995e-01 -1.60202548e-01 7.86835968e-01 4.53297459e-02
9.38078344e-01 -8.24855447e-01 -6.78949594e-01 1.06094196e-01
1.16235960e+00 1.00307487e-01 7.08081067e-01 4.48802024e-01
5.07520974e-01 8.96549225e-01 1.06931567e+00 1.72033966e-01
6.18152738e-01 7.90063977e-01 1.99558288e-01 1.10151127e-01
2.12673828e-01 -2.35271856e-01 -1.99039102e-01 -7.97644928e-02
1.45625949e-01 -3.55602205e-01 -7.36437321e-01 6.78753853e-01
-1.97669256e+00 -1.01235330e+00 -5.19792378e-01 2.01411819e+00
1.27863622e+00 -5.85249178e-02 1.63348660e-01 1.44415691e-01
7.44822800e-01 -1.71055421e-01 -3.68901491e-01 -1.09832376e-01
2.27012560e-01 1.04056433e-01 3.72719198e-01 4.92488980e-01
-1.14798486e+00 1.11365819e+00 5.69905472e+00 9.28157628e-01
-6.92876279e-01 -1.03827603e-01 2.27792084e-01 5.65597653e-01
-6.45519316e-01 3.62034678e-01 -1.34877205e+00 3.70504290e-01
8.00779462e-01 -2.69663781e-01 6.87890276e-02 9.92956340e-01
-1.21851765e-01 -3.24773908e-01 -1.15598273e+00 5.05981684e-01
-1.43027633e-01 -1.25232744e+00 9.45949927e-02 -9.26669836e-02
1.89359918e-01 -5.72707295e-01 -3.98382753e-01 6.39171839e-01
1.88849837e-01 -6.41860843e-01 5.72638869e-01 4.09939170e-01
5.38363457e-01 -4.54537481e-01 8.65419209e-01 6.68336213e-01
-1.26584578e+00 -1.39001757e-01 -2.87631065e-01 -3.74725298e-03
5.78316934e-02 9.03123200e-01 -1.24513614e+00 5.30549049e-01
5.15142024e-01 4.33223188e-01 -4.02672321e-01 1.01449203e+00
-5.09355247e-01 4.09248263e-01 -3.68175387e-01 -5.01177907e-01
-8.47510472e-02 2.45807484e-01 5.82924962e-01 1.17552567e+00
-1.19237036e-01 2.68787682e-01 5.11204749e-02 6.21396661e-01
-2.28850916e-01 3.31390172e-01 -4.96821195e-01 6.70408756e-02
7.33414829e-01 1.38067245e+00 -6.67662799e-01 -8.74248445e-01
-2.20576882e-01 5.51765084e-01 4.55587208e-01 1.52465671e-01
-3.59729558e-01 -6.79710269e-01 6.23844743e-01 1.86934352e-01
1.83045030e-01 1.78092748e-01 -3.12007427e-01 -1.30517697e+00
1.07949026e-01 -6.83505118e-01 1.05579126e+00 -5.62195182e-01
-1.18245721e+00 2.41275117e-01 -2.19956581e-02 -7.80646265e-01
-7.28910148e-01 -8.49615276e-01 -3.30980390e-01 8.99346590e-01
-1.68120670e+00 -8.37330341e-01 7.91966245e-02 5.57232976e-01
2.91458637e-01 8.54455605e-02 9.27043438e-01 4.25140560e-01
-5.12716234e-01 5.06773770e-01 -2.72532403e-01 2.80914247e-01
6.76259279e-01 -1.68605411e+00 2.46929675e-02 5.99043727e-01
6.28536865e-02 9.20029521e-01 8.15461338e-01 -6.32725894e-01
-1.36994922e+00 -1.18582916e+00 1.81018448e+00 -7.23364294e-01
6.96210980e-01 -1.76452056e-01 -1.03665185e+00 5.56082487e-01
-3.90725315e-01 -3.03751051e-01 6.72385395e-01 7.28158772e-01
-4.08907861e-01 1.30001277e-01 -1.32509661e+00 3.31508815e-01
8.89679790e-01 -5.11115968e-01 -1.05384660e+00 5.00148296e-01
6.63980603e-01 -4.85816628e-01 -1.08814406e+00 2.16821939e-01
2.78313369e-01 -4.93258923e-01 1.03253376e+00 -8.41134191e-01
1.82545200e-01 -4.33239043e-01 1.02559604e-01 -1.17247128e+00
-5.05474620e-02 -2.22130671e-01 -4.08095181e-01 1.40962493e+00
1.04481709e+00 -8.55362296e-01 7.67749786e-01 1.16312420e+00
7.38053164e-03 -5.87963104e-01 -7.15010285e-01 -4.72136945e-01
-3.63213062e-01 -2.02524215e-01 8.02904904e-01 9.58392382e-01
3.88226479e-01 8.87473047e-01 4.00717318e-01 3.65338147e-01
2.02669114e-01 6.18513107e-01 5.49822867e-01 -1.42450655e+00
-4.00663942e-01 -5.11794150e-01 -2.74437696e-01 -1.06597817e+00
4.90775734e-01 -1.06342101e+00 2.26031855e-01 -1.56454396e+00
-5.15403971e-02 -1.07491076e+00 1.89190954e-01 6.80370748e-01
-5.40643156e-01 -1.13859594e-01 -4.43300515e-01 1.69923201e-01
-6.30061567e-01 7.96903670e-02 6.40347421e-01 -1.33823350e-01
-1.43523425e-01 5.21576345e-01 -9.62268353e-01 7.91537285e-01
6.16674125e-01 -5.83625793e-01 -3.81630003e-01 -2.99861073e-01
6.87418520e-01 9.16285068e-02 2.45876119e-01 -3.18124771e-01
2.98493177e-01 -3.75779808e-01 -2.07500607e-01 -3.08725953e-01
-5.85552678e-03 -6.78281903e-01 -1.41736478e-01 1.93400130e-01
-6.46173060e-01 -2.93520868e-01 -2.34317228e-01 3.47857922e-01
-4.17790383e-01 -9.46685195e-01 2.87456036e-01 -3.50549161e-01
-1.19245744e+00 2.52761841e-01 -2.14229032e-01 4.77190107e-01
8.37638497e-01 2.03627124e-01 -4.99325722e-01 1.30655142e-02
-8.17389190e-01 4.97822642e-01 2.33659878e-01 4.16612327e-01
2.00051442e-01 -9.60106730e-01 -4.09835994e-01 -2.09243625e-01
3.64917517e-01 2.86862701e-01 -1.43762335e-01 5.53467453e-01
-2.75945693e-01 6.90378845e-01 4.03371245e-01 -1.15631126e-01
-1.34632897e+00 6.13159537e-01 3.25558543e-01 -6.14722371e-01
-8.10170323e-02 9.41637039e-01 7.05671906e-02 -6.72314644e-01
9.33306664e-02 -2.26989418e-01 -9.36947167e-01 4.80789989e-01
6.18912935e-01 1.48032531e-01 4.96895790e-01 -5.32784343e-01
-5.10298252e-01 6.25425428e-02 1.24318879e-02 1.12314068e-01
1.06979990e+00 3.24541703e-02 -6.34553492e-01 2.90690511e-01
1.01865435e+00 1.19658292e-03 -1.82142317e-01 -3.13833803e-01
6.84704900e-01 -4.20174785e-02 -3.80277246e-01 -6.81528091e-01
-5.12019277e-01 4.81541306e-01 -1.72875747e-01 8.90663683e-01
8.10416579e-01 5.76597214e-01 7.46628165e-01 7.73425937e-01
5.90436757e-01 -1.06726193e+00 -5.30850887e-01 9.64567363e-01
5.23174286e-01 -1.03422427e+00 -2.84401476e-01 -8.31649959e-01
-3.13152909e-01 8.03491354e-01 5.92860997e-01 4.83286649e-01
6.35118961e-01 1.90415829e-01 -1.14078961e-01 -4.44358200e-01
-1.02887404e+00 -8.33034158e-01 4.75998640e-01 5.82545280e-01
6.89050972e-01 1.35755107e-01 -5.87133586e-01 5.56510270e-01
-3.65989655e-01 -1.84092119e-01 2.33747065e-01 9.03974473e-01
-6.92201674e-01 -1.34519017e+00 -2.66913801e-01 8.86934638e-01
-5.40779352e-01 -2.78397113e-01 -6.81804895e-01 4.46728528e-01
1.90162510e-01 1.12521899e+00 -8.75674635e-02 -6.86301440e-02
4.17367309e-01 3.48553658e-01 2.54569918e-01 -1.25899637e+00
-7.46122539e-01 -5.51922441e-01 8.81355882e-01 -2.69734889e-01
-5.47774076e-01 -6.44690037e-01 -1.48767352e+00 7.72126913e-02
-4.57383722e-01 6.75083637e-01 3.63555342e-01 1.39890659e+00
2.86533982e-01 8.94797966e-02 4.68233198e-01 -6.13623559e-02
-3.59870285e-01 -1.01267707e+00 -6.05831146e-01 7.37102807e-01
8.00803155e-02 -6.21214688e-01 -2.03888223e-01 3.40614319e-02]
|
[9.659411430358887, 8.75721549987793]
|
d1176801-b1cb-4532-8d89-37c1bd5b53f1
|
faceforensics-a-large-scale-video-dataset-for
|
1803.09179
| null |
http://arxiv.org/abs/1803.09179v1
|
http://arxiv.org/pdf/1803.09179v1.pdf
|
FaceForensics: A Large-scale Video Dataset for Forgery Detection in Human Faces
|
With recent advances in computer vision and graphics, it is now possible to
generate videos with extremely realistic synthetic faces, even in real time.
Countless applications are possible, some of which raise a legitimate alarm,
calling for reliable detectors of fake videos. In fact, distinguishing between
original and manipulated video can be a challenge for humans and computers
alike, especially when the videos are compressed or have low resolution, as it
often happens on social networks. Research on the detection of face
manipulations has been seriously hampered by the lack of adequate datasets. To
this end, we introduce a novel face manipulation dataset of about half a
million edited images (from over 1000 videos). The manipulations have been
generated with a state-of-the-art face editing approach. It exceeds all
existing video manipulation datasets by at least an order of magnitude. Using
our new dataset, we introduce benchmarks for classical image forensic tasks,
including classification and segmentation, considering videos compressed at
various quality levels. In addition, we introduce a benchmark evaluation for
creating indistinguishable forgeries with known ground truth; for instance with
generative refinement models.
|
['Matthias Nießner', 'Christian Riess', 'Andreas Rössler', 'Luisa Verdoliva', 'Justus Thies', 'Davide Cozzolino']
|
2018-03-24
| null | null | null | null |
['image-manipulation-detection']
|
['computer-vision']
|
[ 6.23513520e-01 -7.18902722e-02 2.36602336e-01 -1.93442330e-01
-5.35661817e-01 -7.40281940e-01 7.18296528e-01 -1.57885566e-01
-1.87812328e-01 6.81440115e-01 -2.88826942e-01 8.65982920e-02
2.07264721e-01 -6.89773679e-01 -9.52558160e-01 -5.93466759e-01
-1.42722219e-01 3.72052372e-01 2.30558261e-01 -1.87731609e-01
3.75375420e-01 6.68209076e-01 -2.06129479e+00 4.73050833e-01
5.15631795e-01 1.01947260e+00 -2.18124598e-01 7.63959348e-01
2.14701816e-01 6.67413175e-01 -9.47576880e-01 -1.22934031e+00
4.72198755e-01 -5.37111104e-01 -5.96592009e-01 4.32786554e-01
9.47395146e-01 -7.96760499e-01 -5.35765469e-01 1.47782147e+00
3.53538781e-01 -1.14431530e-01 3.98166418e-01 -1.70466137e+00
-5.45902967e-01 3.84784847e-01 -6.73543096e-01 2.53333241e-01
8.27885747e-01 3.96944344e-01 3.56720328e-01 -6.93297029e-01
9.01418030e-01 1.61711156e+00 4.81166601e-01 6.63446307e-01
-1.19127035e+00 -8.78497720e-01 -4.54060346e-01 3.29916894e-01
-1.37507594e+00 -9.04372692e-01 7.31793046e-01 -5.30274212e-01
3.39357048e-01 3.35369051e-01 6.01710796e-01 1.54230690e+00
-1.27237126e-01 5.68687797e-01 1.00492263e+00 -2.87863225e-01
2.35625617e-02 1.28200099e-01 -4.43738222e-01 7.21045613e-01
5.73535681e-01 2.72177402e-02 -6.28188908e-01 -2.70639390e-01
6.43536985e-01 -1.91384792e-01 -6.73706055e-01 -1.39497161e-01
-9.41553473e-01 7.53470182e-01 -2.83750474e-01 3.24609458e-01
-1.07696630e-01 -2.20447918e-03 5.18364489e-01 3.71911436e-01
2.74444968e-01 4.25051361e-01 2.36764267e-01 -2.18535379e-01
-1.45216727e+00 5.28359950e-01 8.39528680e-01 9.60712016e-01
4.15569991e-01 2.04033628e-01 7.32599348e-02 4.53758985e-01
-5.32234088e-02 5.52432775e-01 2.32309327e-01 -1.17736185e+00
2.99506217e-01 1.95243582e-01 1.58315390e-01 -1.75248861e+00
3.34691107e-01 2.50103801e-01 -8.92633617e-01 2.32097372e-01
7.29495525e-01 2.39036143e-01 -5.76741099e-01 1.42841101e+00
3.53391290e-01 5.75985610e-01 -3.61604869e-01 8.62981319e-01
6.14383042e-01 3.34957480e-01 -2.72350013e-01 -4.51336712e-01
1.26155925e+00 -5.65833330e-01 -8.91823351e-01 9.11410600e-02
2.05117404e-01 -9.83382702e-01 7.98924923e-01 9.77617443e-01
-1.07918298e+00 -3.25553417e-01 -8.91801655e-01 8.77492279e-02
-3.08090150e-01 -7.05346242e-02 4.07698184e-01 1.36156511e+00
-8.75228167e-01 9.04040158e-01 -5.27863741e-01 -1.48560748e-01
9.43548024e-01 1.76995784e-01 -7.41957724e-01 -2.94605076e-01
-9.41752851e-01 5.31879485e-01 1.49352789e-01 1.37866139e-01
-1.21652734e+00 -6.47213340e-01 -7.65207231e-01 -2.48792663e-01
4.76815790e-01 -1.12548567e-01 9.54539597e-01 -1.27102780e+00
-1.28244317e+00 1.24322617e+00 1.67406276e-01 -4.33644772e-01
1.14239478e+00 -2.22767666e-01 -6.55418694e-01 5.25318444e-01
-2.41036955e-02 5.17066061e-01 1.64735103e+00 -1.36747420e+00
-1.75687626e-01 -4.04698223e-01 9.87686515e-02 -4.17543620e-01
-3.61527383e-01 5.11950076e-01 -4.37822193e-01 -8.13630760e-01
-4.58929271e-01 -7.12407231e-01 2.53419518e-01 3.87057751e-01
-5.92640102e-01 2.66147584e-01 1.15998781e+00 -1.03032684e+00
1.07398820e+00 -2.18604326e+00 1.38061970e-01 -4.39544432e-02
2.75117099e-01 5.46186268e-01 -1.27661601e-01 1.67566419e-01
-1.02979548e-01 3.80349904e-01 -4.66741383e-01 -4.95866358e-01
-2.27026984e-01 1.12255894e-01 -4.73792225e-01 9.30767417e-01
2.70766854e-01 4.41932261e-01 -9.63884056e-01 -6.78340733e-01
2.49317020e-01 5.59840620e-01 -5.45394063e-01 2.23700777e-01
-2.53264517e-01 5.70904672e-01 -9.27103162e-02 9.51747358e-01
9.21328723e-01 1.08196981e-01 8.68349224e-02 -1.83906972e-01
3.83344442e-01 -4.49463367e-01 -1.30711520e+00 1.25722063e+00
1.91287756e-01 9.69466388e-01 2.73703963e-01 -8.77769768e-01
5.44565797e-01 3.28374624e-01 4.08553660e-01 -3.13780487e-01
4.10304934e-01 2.00808421e-01 -2.68005520e-01 -7.90930748e-01
7.17814803e-01 1.20067835e-01 2.89977461e-01 4.04092789e-01
-2.31713243e-02 -2.88877696e-01 5.17239630e-01 2.15581730e-01
1.15571916e+00 2.28000693e-02 -8.82401243e-02 2.23718777e-01
4.96332228e-01 -3.19735676e-01 2.64388174e-01 5.56495190e-01
-2.42338389e-01 7.33409643e-01 7.41237581e-01 -2.87966251e-01
-1.05259430e+00 -7.82420516e-01 -5.05357422e-02 5.21381617e-01
1.36696950e-01 -4.50820655e-01 -1.27229321e+00 -5.55245757e-01
4.66757640e-02 3.91977102e-01 -5.16665697e-01 -7.78615996e-02
-6.24676943e-01 -4.10924435e-01 1.13472438e+00 -1.18362479e-01
6.47632718e-01 -9.92977023e-01 -5.89264452e-01 -1.90136254e-01
-2.38150015e-01 -1.75193679e+00 -3.50641310e-01 -9.38504159e-01
-4.99949455e-01 -1.62561333e+00 -7.23708093e-01 -4.56733853e-01
6.69602454e-01 4.52824056e-01 1.17108953e+00 6.30867243e-01
-7.09826946e-01 4.59748060e-01 -3.59995991e-01 -1.04445464e-03
-8.87787223e-01 -6.26783669e-01 1.63389698e-01 4.62169617e-01
2.98051424e-02 -4.35123146e-01 -3.44475240e-01 3.90679002e-01
-1.37696207e+00 -2.29508981e-01 1.03084929e-01 5.31640291e-01
2.19612479e-01 2.55908608e-01 1.84697703e-01 -9.45768178e-01
4.90348369e-01 -5.28032303e-01 -6.60349727e-01 3.08146864e-01
-3.87157984e-02 -2.92285562e-01 6.65022373e-01 -5.89847088e-01
-8.67944717e-01 -6.96810037e-02 1.02705128e-01 -9.73792732e-01
-4.01852697e-01 -2.12795123e-01 -2.36224949e-01 -4.62273896e-01
6.63619101e-01 1.35736585e-01 3.15166861e-02 -2.63917565e-01
4.29179192e-01 6.10635102e-01 8.20593536e-01 -4.88142818e-01
1.07540774e+00 7.31571257e-01 1.08839892e-01 -1.30977714e+00
-2.69292861e-01 7.27167428e-02 -5.22776484e-01 -6.46505475e-01
6.10713840e-01 -5.14854193e-01 -6.87662125e-01 9.73981500e-01
-1.34085631e+00 -1.25155328e-02 6.50689155e-02 2.07326472e-01
-5.58344603e-01 1.12258816e+00 -6.03950977e-01 -7.40258276e-01
6.12376779e-02 -1.36249912e+00 1.15930867e+00 3.19524705e-02
-2.72823256e-02 -5.49456000e-01 -4.03439701e-01 6.16489768e-01
2.56601542e-01 7.99681485e-01 4.10447568e-01 -3.22457820e-01
-7.36354113e-01 -2.51241595e-01 -1.17456533e-01 5.24699986e-01
8.89560133e-02 6.27390981e-01 -1.04183280e+00 -4.23266381e-01
1.73191294e-01 -4.15906608e-01 5.96789539e-01 -8.16196054e-02
1.43120658e+00 -4.14215833e-01 -1.64987966e-01 5.22635162e-01
1.17418206e+00 -6.93206787e-02 9.56759334e-01 -1.30047679e-01
5.75258553e-01 7.58010387e-01 5.75582564e-01 6.01966143e-01
-1.99461699e-01 7.58089423e-01 7.31753945e-01 4.70574439e-01
-1.41232669e-01 -8.07574540e-02 4.26821738e-01 2.99254745e-01
-3.29192489e-01 -4.42456007e-01 -5.32153845e-01 3.04212332e-01
-1.22283912e+00 -1.40407097e+00 -2.18556881e-01 2.49417901e+00
6.12532258e-01 -1.02135064e-02 1.25605166e-01 5.93907535e-01
1.14742398e+00 2.09939867e-01 -2.39247173e-01 -1.31757945e-01
-1.95530444e-01 2.53601223e-01 2.51117945e-01 8.69220719e-02
-1.18939614e+00 7.62416601e-01 5.83944607e+00 1.03877223e+00
-1.12371194e+00 1.13185823e-01 8.40051532e-01 -3.58008258e-02
2.30840128e-02 -2.36249581e-01 -6.40444100e-01 9.45457935e-01
7.38859415e-01 3.11876144e-02 6.57531857e-01 7.76971698e-01
1.68414295e-01 -3.20578188e-01 -1.14750552e+00 1.50788343e+00
8.34778249e-01 -1.28364682e+00 1.60380885e-01 1.29914522e-01
6.13242507e-01 -7.11820066e-01 1.46205351e-01 -1.04578614e-01
-2.34509423e-01 -1.18803501e+00 8.86175215e-01 2.53367424e-01
9.64010894e-01 -6.90150976e-01 4.39811081e-01 1.60466358e-01
-8.85765433e-01 2.12570950e-01 -3.20347995e-01 3.51666421e-01
3.03565711e-01 5.83210111e-01 -4.58971947e-01 3.60762984e-01
7.22857237e-01 5.56675911e-01 -7.92450666e-01 1.09263027e+00
-1.34058729e-01 4.56947088e-01 -3.45541209e-01 3.71356428e-01
-2.95249492e-01 -3.12882036e-01 7.26567328e-01 1.01598263e+00
5.17469645e-01 -6.94914088e-02 -2.79500544e-01 7.41573989e-01
-5.69790363e-01 -4.08832580e-02 -9.44584906e-01 -4.37741131e-01
4.38302755e-01 1.23006141e+00 -9.16162968e-01 -2.20742837e-01
-2.71550447e-01 1.35562253e+00 -7.29664415e-02 -2.44705770e-02
-1.25319314e+00 -2.59273916e-01 6.09206855e-01 4.27236378e-01
1.42370403e-01 -1.77942321e-01 4.27600771e-01 -1.32669854e+00
2.23977879e-01 -1.40557957e+00 4.29896601e-02 -6.90294206e-01
-1.16026115e+00 5.61493039e-01 2.97220051e-03 -1.10452378e+00
-2.70934641e-01 -5.88848710e-01 -3.02869678e-01 1.63057864e-01
-1.17463148e+00 -9.18747127e-01 -6.23679876e-01 7.29644239e-01
5.78395426e-01 -2.02067271e-01 4.96828765e-01 8.02705288e-01
-4.90656525e-01 6.36685133e-01 -1.80550367e-01 3.53884101e-01
8.84190619e-01 -6.32668972e-01 4.39737797e-01 1.02761614e+00
3.60639811e-01 2.13644803e-01 8.80684137e-01 -6.00854933e-01
-1.65024972e+00 -9.73716795e-01 3.20230067e-01 -4.56116945e-01
4.83218759e-01 -4.47950095e-01 -1.03755260e+00 4.66927260e-01
4.04422730e-02 2.60871977e-01 3.73423278e-01 -8.62682223e-01
-4.52493936e-01 2.03730658e-01 -1.62081540e+00 4.50005382e-01
1.08164668e+00 -5.16379595e-01 -2.56411463e-01 6.32431984e-01
2.86728114e-01 -3.76750201e-01 -6.71977043e-01 2.46387079e-01
6.19818389e-01 -1.45457363e+00 1.00139904e+00 -5.36344528e-01
6.35914505e-01 -2.10617259e-01 -7.31937364e-02 -9.28813517e-01
4.87911344e-01 -1.10934830e+00 -3.17422569e-01 1.43597496e+00
-2.52143562e-01 -2.37179130e-01 8.41474712e-01 4.90535885e-01
3.71361285e-01 -3.26600343e-01 -9.10141349e-01 -9.39592600e-01
-4.12269354e-01 -4.54546005e-01 6.26934052e-01 1.17523479e+00
-5.31934083e-01 -3.57344002e-01 -7.88741767e-01 4.97480370e-02
9.55443621e-01 -2.84358442e-01 1.05069351e+00 -9.89906847e-01
-3.69565427e-01 -4.58913267e-01 -1.07228816e+00 -6.58180058e-01
3.11712921e-01 -3.89234245e-01 -2.81627566e-01 -6.28033042e-01
8.90340582e-02 -7.62467682e-02 7.22041607e-01 4.90518361e-02
7.54356757e-02 8.49523067e-01 2.70376772e-01 1.33490399e-01
-2.78815866e-01 2.04885975e-01 1.18759537e+00 -1.39582351e-01
4.30858493e-01 -1.42229691e-01 -2.33194590e-01 9.05704021e-01
7.13465214e-01 -5.84835351e-01 -1.24482580e-01 -2.65163839e-01
2.68343836e-01 5.43418564e-02 6.52558863e-01 -1.26543689e+00
1.33599835e-02 -9.03870240e-02 1.93684593e-01 -2.15271547e-01
5.86423278e-01 -6.16844535e-01 4.89231497e-01 3.29883754e-01
-3.51529568e-02 -2.05775984e-02 -6.19126409e-02 7.44753599e-01
-2.74918169e-01 -4.51479822e-01 1.19323456e+00 -1.59110472e-01
-4.64362681e-01 2.76070744e-01 -2.10693911e-01 2.54468411e-01
1.34575737e+00 -3.31193954e-01 -4.94255275e-01 -5.90279162e-01
-3.92174214e-01 -3.05634230e-01 7.95626581e-01 5.14034390e-01
8.26916754e-01 -1.07712555e+00 -6.40040636e-01 2.97624558e-01
-1.91630557e-01 -3.43821734e-01 2.99157053e-01 7.20243454e-01
-1.06132519e+00 -2.66360998e-01 -4.61659372e-01 -4.95572597e-01
-1.71499670e+00 8.47714245e-01 1.48475960e-01 2.58937418e-01
-2.95701742e-01 6.79469705e-01 -2.29117930e-01 1.77343234e-01
1.93297207e-01 -2.29664817e-02 6.31472170e-02 8.56632590e-02
9.61502671e-01 7.62548089e-01 1.25402406e-01 -1.13132119e+00
-9.29611549e-02 3.53381813e-01 1.94856614e-01 2.10685372e-01
1.17654657e+00 1.42415449e-01 -3.21367711e-01 -1.24339037e-01
1.08898795e+00 1.91018805e-01 -9.90831077e-01 2.68750489e-01
-1.69309482e-01 -1.25471199e+00 -3.10687095e-01 -9.57433060e-02
-1.39634597e+00 9.26280856e-01 4.55990016e-01 4.37516540e-01
9.42081928e-01 -1.91684425e-01 8.96132946e-01 2.57443845e-01
6.96835637e-01 -8.44230831e-01 3.41099322e-01 -9.87105742e-02
9.64406133e-01 -1.33220315e+00 2.04175025e-01 -7.56697237e-01
-3.85695159e-01 1.20847201e+00 2.54371554e-01 -1.95334166e-01
4.32767570e-01 3.63962799e-01 -4.52325284e-01 -6.75703064e-02
-1.59088761e-01 3.44394535e-01 -1.44209057e-01 6.07161999e-01
2.13115606e-02 -7.10216388e-02 -9.74709690e-02 1.55392438e-01
-2.30138034e-01 4.64673750e-02 9.33390915e-01 8.10092986e-01
-2.14209437e-01 -1.20510495e+00 -8.55177522e-01 4.91331398e-01
-8.46173584e-01 1.96887165e-01 -5.39813638e-01 7.31670797e-01
1.05844781e-01 1.12876952e+00 -4.25600670e-02 -1.61569610e-01
-3.47762555e-02 -1.77137449e-01 8.92572403e-01 -2.93799102e-01
-4.95186538e-01 -3.87772501e-01 6.09255992e-02 -8.13315272e-01
-6.52675629e-01 -7.25472569e-01 -6.15970433e-01 -8.29020619e-01
-3.61142337e-01 -1.80987433e-01 6.69297874e-01 6.27607822e-01
1.94069698e-01 -1.01926729e-01 6.34702981e-01 -1.23703158e+00
-4.83666331e-01 -6.61621809e-01 -6.16397619e-01 8.21551204e-01
8.90631527e-02 -8.59342039e-01 -6.82388783e-01 5.45898616e-01]
|
[12.548041343688965, 1.0790483951568604]
|
901d7966-0513-45d1-907f-a96f6fb9f2c5
|
a-comparison-of-multi-view-learning
|
2105.04984
| null |
https://arxiv.org/abs/2105.04984v1
|
https://arxiv.org/pdf/2105.04984v1.pdf
|
A Comparison of Multi-View Learning Strategies for Satellite Image-Based Real Estate Appraisal
|
In the house credit process, banks and lenders rely on a fast and accurate estimation of a real estate price to determine the maximum loan value. Real estate appraisal is often based on relational data, capturing the hard facts of the property. Yet, models benefit strongly from including image data, capturing additional soft factors. The combination of the different data types requires a multi-view learning method. Therefore, the question arises which strengths and weaknesses different multi-view learning strategies have. In our study, we test multi-kernel learning, multi-view concatenation and multi-view neural networks on real estate data and satellite images from Asheville, NC. Our results suggest that multi-view learning increases the predictive performance up to 13% in MAE. Multi-view neural networks perform best, however result in intransparent black-box models. For users seeking interpretability, hybrid multi-view neural networks or a boosting strategy are a suitable alternative.
|
['Oliver Müller', 'Jan-Peter Kucklick']
|
2021-05-11
| null | null | null | null |
['multi-view-learning']
|
['computer-vision']
|
[-4.40219730e-01 -1.52444094e-01 -4.21523660e-01 -4.64753449e-01
-7.36399889e-01 -4.41629946e-01 3.88380587e-01 1.37007341e-01
-2.52863824e-01 3.94396126e-01 5.05773246e-01 -4.48070496e-01
-1.97212398e-01 -1.19629085e+00 -2.72911876e-01 -5.68441093e-01
4.86868620e-02 4.24274892e-01 -3.35281938e-01 -4.04026538e-01
1.25091627e-01 5.44253826e-01 -1.38486981e+00 4.64511424e-01
5.53685009e-01 9.06965017e-01 1.32758752e-01 4.89851713e-01
1.20568082e-01 9.04450476e-01 -1.20945558e-01 -8.43987286e-01
4.02829021e-01 3.82070333e-01 -3.49812984e-01 -7.33416677e-02
4.67876881e-01 -9.73056793e-01 1.85935393e-01 4.72903579e-01
8.43844116e-01 -3.87852043e-01 8.68167162e-01 -9.59346056e-01
-6.16421938e-01 6.61436141e-01 -8.38045597e-01 3.23425382e-01
3.09697211e-01 -4.96211182e-03 1.36860633e+00 -1.49863374e+00
2.17401698e-01 8.91993821e-01 1.04915357e+00 -2.79528260e-01
-1.43268371e+00 -4.42204028e-01 3.83443832e-02 5.42781830e-01
-1.27508831e+00 -5.14747202e-01 7.77862728e-01 -5.81327975e-01
1.40421569e+00 3.01953524e-01 8.67630661e-01 6.81032896e-01
1.85412318e-01 6.12434685e-01 1.67302239e+00 -4.21504229e-01
1.65384896e-02 5.30859411e-01 -9.06548053e-02 2.67868400e-01
2.45470881e-01 2.10456952e-01 -7.02464700e-01 -4.04316276e-01
6.36966467e-01 3.59610081e-01 -2.33429357e-01 -3.96271676e-01
-7.16374815e-01 1.09710932e+00 3.99909019e-01 1.73138842e-01
-5.48722446e-01 -4.35084671e-01 4.15118724e-01 1.78566933e-01
8.81829381e-01 2.61137843e-01 -6.43491626e-01 -1.38338089e-01
-1.23545873e+00 8.69111270e-02 9.11381364e-01 1.74031809e-01
7.32540190e-01 2.80002747e-02 5.60150325e-01 1.10912478e+00
3.37627828e-01 4.57469165e-01 6.26769215e-02 -6.05972528e-01
8.22944701e-01 9.57707882e-01 -1.11492388e-02 -1.21393764e+00
-5.89023590e-01 -3.00709546e-01 -9.29863393e-01 8.87695372e-01
5.02514720e-01 -7.28149042e-02 -5.69653153e-01 8.56018066e-01
-8.18462670e-02 -4.15168643e-01 1.12077951e-01 7.13399351e-01
6.42893910e-01 3.44643474e-01 1.55353054e-01 2.98490208e-02
1.45652735e+00 -6.65978134e-01 -5.05058885e-01 -1.48241386e-01
6.28959417e-01 -6.12105429e-01 1.03229558e+00 7.81155884e-01
-1.21565282e+00 -4.21789289e-01 -1.30841124e+00 5.66792265e-02
-7.17702329e-01 4.63874370e-01 5.81438005e-01 8.63545299e-01
-6.66705310e-01 4.79295135e-01 -7.75667667e-01 -9.42362770e-02
6.19985998e-01 1.79893672e-01 -5.38387775e-01 -1.69833213e-01
-1.09933949e+00 1.29608583e+00 3.13247800e-01 3.49790335e-01
-1.72060058e-01 -7.89213181e-01 -5.83408475e-01 3.09659153e-01
1.20870449e-01 -4.00659025e-01 7.95852959e-01 -9.96669292e-01
-8.14212501e-01 6.11517489e-01 3.73645693e-01 -2.29181215e-01
6.92473114e-01 -2.57321656e-01 -3.24635297e-01 6.50940686e-02
6.50664791e-03 3.41611385e-01 7.63755262e-01 -1.28858948e+00
-7.49356151e-01 -7.56636620e-01 1.97573882e-02 2.52520204e-01
-5.60948074e-01 -1.71052683e-02 1.46599323e-01 -7.77849615e-01
1.72140017e-01 -7.63867974e-01 -1.46575822e-02 -1.75619960e-01
2.49849975e-01 -7.45601580e-02 7.62973964e-01 -1.27497005e+00
1.53434217e+00 -1.85034430e+00 -2.26766750e-01 2.29475766e-01
3.19688171e-01 1.40599862e-01 4.45594907e-01 4.93888617e-01
-3.58414054e-01 6.70509711e-02 -5.76139018e-02 1.61181152e-01
-9.20768678e-02 7.34839812e-02 -1.12318426e-01 2.65206397e-01
1.39404729e-01 8.25912893e-01 -4.45234239e-01 -4.20945972e-01
4.03841063e-02 3.96788031e-01 -3.71033669e-01 -7.32036158e-02
3.73010844e-01 -5.19248784e-01 -2.28610858e-01 8.24172318e-01
8.84730458e-01 -6.13277256e-01 1.72120854e-01 -2.50507325e-01
8.07360038e-02 5.58847971e-02 -1.30793238e+00 1.15279245e+00
-7.67129958e-01 4.58566964e-01 -2.44854074e-02 -9.13000643e-01
8.86150479e-01 4.23764259e-01 3.07791740e-01 -5.12035489e-01
-4.57294405e-01 2.74714679e-01 -1.21931262e-01 -6.36824906e-01
4.15573984e-01 -5.93446195e-01 3.48109543e-01 8.51298213e-01
-3.87726068e-01 2.41002142e-01 -2.19130710e-01 1.20435722e-01
6.78213716e-01 3.67458791e-01 5.96788585e-01 5.25931828e-02
2.39561304e-01 -1.76277086e-01 3.92024070e-01 2.12260693e-01
1.95109189e-01 7.43210733e-01 6.27190530e-01 -1.08544624e+00
-1.20158434e+00 -9.64131534e-01 -3.58464718e-01 1.01241815e+00
-8.01074624e-01 -2.63343424e-01 -3.71360928e-01 -8.32894027e-01
4.02774990e-01 7.17481673e-01 -4.40484643e-01 3.11694235e-01
-3.64168733e-01 -1.19355321e+00 2.03142419e-01 1.04720771e+00
5.80288410e-01 -8.15995395e-01 -1.08911169e+00 3.86099935e-01
-1.30953342e-01 -9.16081309e-01 9.48034972e-02 5.42380273e-01
-1.05035245e+00 -8.64263594e-01 -8.91175747e-01 -1.28028467e-01
4.35471445e-01 2.96342432e-01 1.35052776e+00 8.43804255e-02
4.93816249e-02 2.45938241e-01 -3.49788010e-01 -4.63974714e-01
-1.94688171e-01 3.00424844e-01 -1.98288888e-01 -6.80949986e-02
6.81029797e-01 -9.46175277e-01 -3.41276348e-01 2.09950238e-01
-6.06216252e-01 1.77720308e-01 7.98493683e-01 9.02465641e-01
2.88210452e-01 2.28702519e-02 8.67109776e-01 -8.58078241e-01
5.24327993e-01 -4.96589750e-01 -4.66566712e-01 5.41030526e-01
-1.38253796e+00 -1.61951199e-01 2.19361976e-01 -2.19077244e-02
-1.28982747e+00 6.85203224e-02 2.36240223e-01 -3.16420704e-01
-1.85827598e-01 1.07269049e+00 -2.55934596e-01 1.12918504e-01
8.53149474e-01 -3.51658128e-02 4.40419763e-02 -4.49276388e-01
1.31760359e-01 8.29397976e-01 2.75606588e-02 -3.61139357e-01
5.60881138e-01 5.19348443e-01 -1.87762395e-01 -6.96255744e-01
-5.81634939e-01 -3.69053453e-01 -1.07895303e+00 -3.52216154e-01
8.29843223e-01 -1.29169941e+00 -3.26591730e-01 7.46105671e-01
-8.27158511e-01 -1.70003578e-01 1.22712888e-01 5.02563417e-01
-5.19089699e-01 1.40417367e-01 -3.46858531e-01 -9.50817645e-01
-3.72286826e-01 -8.19934785e-01 7.67701328e-01 7.02754557e-02
-1.96202397e-01 -1.04764211e+00 -8.79048482e-02 9.50181067e-01
3.15440685e-01 2.14486808e-01 1.02093685e+00 -8.55453253e-01
-5.08557856e-01 -2.77910113e-01 -4.91697729e-01 3.34557295e-01
3.10753465e-01 -1.32291928e-01 -1.34737349e+00 -1.38610825e-01
-2.85313949e-02 -4.51423049e-01 7.73908615e-01 3.66836458e-01
5.66411257e-01 -3.43100667e-01 -2.78789718e-02 3.83254856e-01
1.72090828e+00 2.25607194e-02 6.54065728e-01 7.77682006e-01
6.26819551e-01 8.40515912e-01 4.30982649e-01 8.34150136e-01
7.18590081e-01 5.76966822e-01 4.00561988e-01 -2.03710333e-01
5.44185281e-01 -4.19798195e-02 4.24570441e-01 5.48431098e-01
-6.03355706e-01 2.19239891e-02 -1.39525199e+00 4.49279934e-01
-1.83569038e+00 -1.31366789e+00 -1.86252356e-01 1.90217364e+00
3.72133076e-01 3.59289318e-01 4.02321070e-01 3.02814156e-01
2.29554519e-01 2.53621846e-01 -4.56092358e-01 -3.81588250e-01
-2.67633289e-01 -1.29566357e-01 6.76599801e-01 1.80741414e-01
-1.18335438e+00 3.47486168e-01 6.25894642e+00 6.32053554e-01
-1.11779213e+00 -1.50191203e-01 1.22921562e+00 -3.37106049e-01
-3.24011385e-01 -5.15712015e-02 -7.21989155e-01 1.38575688e-01
9.18925047e-01 2.24405259e-01 2.68705249e-01 9.74778414e-01
4.33771014e-01 -6.25933528e-01 -9.57788944e-01 9.17564094e-01
2.39654377e-01 -1.26528668e+00 -1.88865468e-01 6.02754295e-01
5.16607583e-01 4.85444404e-02 1.80839989e-02 -3.74778034e-03
1.39357880e-01 -1.01764715e+00 5.49438238e-01 6.53252780e-01
4.14223433e-01 -8.70077729e-01 9.99160051e-01 4.49454665e-01
-1.12733531e+00 -2.92934239e-01 -4.24826518e-02 -3.62078607e-01
2.30036393e-01 4.17830795e-01 -7.48921990e-01 6.54148161e-01
1.02071798e+00 9.54302430e-01 -9.71276343e-01 5.09565234e-01
4.71189618e-02 3.95052940e-01 -2.93014616e-01 2.85802603e-01
1.66537631e-02 -5.09890020e-01 -2.26514950e-01 9.06896532e-01
3.73584360e-01 4.81598049e-01 -1.05271786e-01 6.70501292e-01
4.57868040e-01 1.65305346e-01 -8.28301370e-01 2.46839836e-01
7.67422691e-02 1.46001506e+00 -6.68335736e-01 -3.00116569e-01
-1.19154859e+00 4.63196516e-01 4.30746347e-01 2.76419520e-01
-4.57445681e-01 2.50293780e-02 3.27086538e-01 4.43434954e-01
5.86901665e-01 -1.11765400e-01 -9.32967842e-01 -1.11057281e+00
2.46379346e-01 -1.02865243e+00 4.11414653e-01 -1.06542516e+00
-1.40947962e+00 1.01803206e-01 1.37485340e-01 -1.33580935e+00
-3.84118319e-01 -6.84206963e-01 -4.54652786e-01 9.36499655e-01
-1.62048388e+00 -1.63695014e+00 -1.49154991e-01 3.09088945e-01
3.83909374e-01 -5.43041587e-01 8.63011241e-01 3.19348007e-01
-2.46309787e-01 3.50016057e-01 3.20632309e-02 3.14534068e-01
7.40226090e-01 -1.48277009e+00 2.12923557e-01 3.04769516e-01
2.45824307e-01 2.21757084e-01 2.16808379e-01 -7.19282031e-01
-9.68155801e-01 -4.65328157e-01 1.12028062e+00 -4.86515343e-01
6.03394330e-01 2.47962754e-02 -1.08758307e+00 6.94736600e-01
3.27480942e-01 -6.88803568e-02 1.16175902e+00 4.12934065e-01
-4.46633518e-01 -3.50059986e-01 -1.00497830e+00 1.52975813e-01
3.31823707e-01 -6.57566071e-01 -4.59463149e-01 -1.28405362e-01
-2.53847718e-01 -8.05154443e-02 -1.50754941e+00 1.85345948e-01
1.05443859e+00 -1.56603777e+00 1.13047326e+00 -4.23783064e-01
9.12164509e-01 1.78651944e-01 -3.94666672e-01 -1.16365802e+00
-3.48445684e-01 2.14610547e-01 1.82026863e-01 1.28330266e+00
8.45894992e-01 -5.96438706e-01 1.16400599e+00 1.05870759e+00
4.31399047e-01 -9.93372917e-01 -9.61825252e-01 -4.88822430e-01
2.15509623e-01 -5.49023390e-01 6.42760575e-01 1.18926501e+00
1.81352153e-01 5.03436804e-01 -3.97983819e-01 1.33907214e-01
4.92087305e-01 1.38584152e-01 6.30209029e-01 -1.48624635e+00
-2.54109055e-01 -3.73077393e-01 -2.45055631e-01 -2.35730499e-01
4.88733388e-02 -6.44246340e-01 -8.86518300e-01 -1.50882256e+00
5.26387453e-01 -5.67436397e-01 -1.30744517e-01 5.28860509e-01
-1.86897799e-01 -1.51763698e-02 3.85625422e-01 2.74556667e-01
3.80494833e-01 2.43373528e-01 8.45796525e-01 -2.35527605e-01
-2.31128812e-01 2.64474332e-01 -5.44037580e-01 1.01944339e+00
8.87742460e-01 -8.48710015e-02 -2.82373399e-01 -4.31390107e-01
8.00036371e-01 2.88755178e-01 6.55395567e-01 -5.97595751e-01
1.82549760e-01 -4.99766469e-02 8.07869077e-01 -9.18156803e-01
4.49578166e-01 -1.22017038e+00 3.20128232e-01 3.79429549e-01
5.15992343e-02 6.17740154e-01 -1.91347697e-03 4.26687419e-01
-2.39222229e-01 -2.69099057e-01 3.75469029e-01 -5.79582930e-01
-4.59045619e-01 -1.41570613e-01 -3.19458842e-01 -3.06894600e-01
6.91875696e-01 -7.99200892e-01 -2.97211744e-02 -6.32912517e-01
-1.10536170e+00 3.29539478e-02 3.48098516e-01 3.23449552e-01
6.30543709e-01 -1.49747074e+00 -9.61228848e-01 1.71169654e-01
7.32281506e-02 -1.71401292e-01 3.53933387e-02 8.02116692e-01
-5.06893992e-01 1.53365850e-01 -4.18181211e-01 -2.89453834e-01
-1.32039297e+00 2.61168808e-01 2.63205320e-01 -5.41713774e-01
-3.86982173e-01 4.16513294e-01 -2.55903006e-01 -1.82517231e-01
-1.98651060e-01 -1.30442400e-02 -6.81320429e-01 1.12574422e+00
3.37771118e-01 7.61171818e-01 3.86572629e-01 -7.84238160e-01
-3.02829202e-02 5.61997116e-01 -2.45227113e-01 -2.94605881e-01
2.03562737e+00 8.53412077e-02 -8.72893184e-02 5.73805392e-01
1.04493117e+00 -3.32944304e-01 -1.04297113e+00 -9.96899828e-02
3.18112731e-01 -5.14346838e-01 3.70978385e-01 -8.57553959e-01
-1.26824260e+00 1.04127610e+00 7.48256028e-01 2.18430087e-01
1.35468483e+00 -2.50818729e-01 4.11227673e-01 1.98848709e-01
2.60245323e-01 -1.42650282e+00 -2.32898295e-02 -4.02200073e-02
9.70439255e-01 -1.56299293e+00 7.97052503e-01 8.49210620e-02
-8.18372369e-01 1.57619536e+00 5.99807203e-01 1.44247457e-01
1.13922870e+00 2.40803361e-01 2.63188213e-01 -2.71910936e-01
-8.11540723e-01 1.28493577e-01 -5.01013137e-02 6.18453026e-01
4.34050381e-01 1.32825717e-01 -1.58240292e-02 5.43812513e-01
-1.90507740e-01 1.49000688e-02 5.66044688e-01 9.19913888e-01
-3.23546857e-01 -1.07703614e+00 -6.27305806e-01 8.95391166e-01
-5.22929907e-01 -2.34727576e-01 -1.96606159e-01 1.00407135e+00
1.05435856e-01 7.48043537e-01 4.79557589e-02 -2.70570427e-01
4.33159322e-01 3.77764761e-01 3.06649860e-02 -2.03988642e-01
-1.04960084e+00 4.10844088e-01 3.02424371e-01 -1.52011588e-01
-7.07916498e-01 -1.36181676e+00 -5.51856220e-01 -3.17000717e-01
-6.69571400e-01 -3.86345744e-01 4.41230416e-01 8.41685295e-01
1.19125366e-01 -5.30455038e-02 8.92235100e-01 -8.24752629e-01
-8.87800276e-01 -8.83053660e-01 -8.41175675e-01 1.14432111e-01
2.62518346e-01 -4.09262985e-01 -2.57569373e-01 3.21400762e-01]
|
[7.096523761749268, 2.3849501609802246]
|
b26e551d-a964-43d7-adae-b6a56eff179e
|
multi-slice-net-a-novel-light-weight
|
2108.03786
| null |
https://arxiv.org/abs/2108.03786v1
|
https://arxiv.org/pdf/2108.03786v1.pdf
|
Multi-Slice Net: A novel light weight framework for COVID-19 Diagnosis
|
This paper presents a novel lightweight COVID-19 diagnosis framework using CT scans. Our system utilises a novel two-stage approach to generate robust and efficient diagnoses across heterogeneous patient level inputs. We use a powerful backbone network as a feature extractor to capture discriminative slice-level features. These features are aggregated by a lightweight network to obtain a patient level diagnosis. The aggregation network is carefully designed to have a small number of trainable parameters while also possessing sufficient capacity to generalise to diverse variations within different CT volumes and to adapt to noise introduced during the data acquisition. We achieve a significant performance increase over the baselines when benchmarked on the SPGC COVID-19 Radiomics Dataset, despite having only 2.5 million trainable parameters and requiring only 0.623 seconds on average to process a single patient's CT volume using an Nvidia-GeForce RTX 2080 GPU.
|
['Clinton Fookes', 'Simon Denman', 'Sridha Sridharan', 'Tharindu Fernando', 'Harshala Gammulle']
|
2021-08-09
| null | null | null | null |
['covid-19-detection']
|
['medical']
|
[ 2.70914346e-01 1.71211705e-01 -6.06196485e-02 -6.21191680e-01
-1.26686919e+00 -4.66147721e-01 1.25531510e-01 2.21731871e-01
-6.02229834e-01 4.38311338e-01 1.21848881e-01 -5.30679643e-01
-3.59839112e-01 -7.45302558e-01 -2.65155166e-01 -7.20762014e-01
-3.26877773e-01 1.04480016e+00 3.55220467e-01 1.52593896e-01
-3.65865797e-01 7.21381783e-01 -1.17376709e+00 3.04311991e-01
4.27184075e-01 1.03515327e+00 1.55658111e-01 1.08737445e+00
5.17521560e-01 8.90349865e-01 -6.21256530e-01 9.69516411e-02
3.45454305e-01 -8.13381523e-02 -7.76462615e-01 3.71120684e-02
4.64401692e-01 -5.61544597e-01 -3.09703887e-01 4.79355037e-01
1.03341937e+00 -6.30951524e-02 5.97416997e-01 -7.84588218e-01
2.28164658e-01 4.24303144e-01 -2.55439401e-01 7.00162292e-01
1.26906678e-01 4.74129230e-01 6.69387341e-01 -5.34267902e-01
7.49589264e-01 7.92250335e-01 9.66858804e-01 6.39865875e-01
-9.33649302e-01 -5.02002835e-01 -4.42761213e-01 -3.32930267e-01
-1.14199913e+00 -5.96219860e-02 -1.59989551e-01 -4.64462370e-01
1.08343422e+00 4.98529106e-01 7.61394501e-01 1.19096041e+00
7.46249020e-01 4.12695348e-01 9.02592480e-01 1.20186701e-01
2.96009868e-01 -2.18285397e-01 3.21435854e-02 9.69803870e-01
2.17375845e-01 1.38949588e-01 -1.50121808e-01 -5.95062315e-01
1.15100539e+00 5.54448403e-02 -2.48723865e-01 -1.85968116e-01
-1.35952842e+00 9.78128672e-01 6.52848363e-01 7.23401010e-02
-5.68134189e-01 2.96103925e-01 9.17602062e-01 -5.59402145e-02
1.38608947e-01 4.20927048e-01 -4.64781672e-01 -1.53424069e-01
-8.51882935e-01 2.42202953e-01 6.79517150e-01 8.29286516e-01
9.14328620e-02 8.32953528e-02 -5.49136698e-01 6.46697283e-01
5.71071580e-02 6.50988400e-01 8.02832782e-01 -7.55811512e-01
3.64261001e-01 3.62463117e-01 -3.79684240e-01 -6.02409840e-01
-1.10564721e+00 -8.48484218e-01 -1.18375397e+00 1.41555667e-01
1.02898791e-01 -4.72515345e-01 -1.48364627e+00 1.27112365e+00
5.43233633e-01 2.72126287e-01 -2.64027894e-01 8.50664079e-01
1.12001252e+00 2.39013180e-01 8.79983902e-02 1.58403441e-01
1.58127093e+00 -8.23013902e-01 -8.40522125e-02 5.69578782e-02
9.99051988e-01 -5.68596601e-01 9.73728836e-01 5.11952460e-01
-1.35171866e+00 -3.10959160e-01 -1.00305259e+00 1.31380886e-01
1.59191526e-02 -7.55326971e-02 8.12657058e-01 7.84401774e-01
-1.38268411e+00 4.76502001e-01 -1.31583178e+00 -1.25824794e-01
5.84443212e-01 9.66445744e-01 -2.22595900e-01 -1.35678813e-01
-7.76359081e-01 8.43342662e-01 5.00122428e-01 -1.51985481e-01
-1.25601649e+00 -1.15211618e+00 -7.68750310e-01 -5.68744205e-02
3.57880712e-01 -1.34297144e+00 1.51008403e+00 -5.80691397e-01
-1.33446729e+00 8.11201513e-01 3.39513987e-01 -4.93232518e-01
7.36844242e-01 3.97703677e-01 -3.85774791e-01 5.40202618e-01
1.47624686e-01 5.75178564e-01 6.46586120e-01 -6.81288898e-01
-7.90200353e-01 -1.34400159e-01 -3.25858772e-01 2.97004044e-01
9.10739601e-02 8.45559500e-03 -6.12004459e-01 -7.16256201e-01
9.03476849e-02 -1.22150469e+00 -6.01716161e-01 -6.62108976e-03
-4.24622089e-01 2.31306911e-01 5.70696294e-01 -3.40709180e-01
9.15673614e-01 -1.70389724e+00 -1.41813114e-01 6.93483829e-01
6.57128036e-01 2.55065769e-01 1.02150450e-02 -1.62476927e-01
-2.17720777e-01 -1.28836066e-01 -3.11925322e-01 -2.57477582e-01
-4.95063603e-01 3.99918556e-01 2.71637946e-01 4.60505575e-01
-2.17459649e-02 9.39305425e-01 -1.01740444e+00 -5.61572134e-01
2.83362806e-01 6.51079118e-01 -7.44281590e-01 1.56273752e-01
2.91560799e-01 5.56317568e-01 -7.31756806e-01 6.65311754e-01
5.49611390e-01 -7.31283605e-01 3.89743410e-02 -6.52315170e-02
3.76552135e-01 1.91764370e-01 -8.18491876e-01 1.69608700e+00
-7.20778346e-01 1.68951154e-01 2.68935591e-01 -4.84777868e-01
3.75974625e-01 5.22248328e-01 9.48113322e-01 -3.31707239e-01
6.11763716e-01 4.27910358e-01 3.14762414e-01 -3.94208461e-01
2.99140155e-01 -3.59704554e-01 -2.69827276e-01 6.93233967e-01
7.00935796e-02 -3.59486043e-01 -1.21540967e-02 2.58087486e-01
1.80426562e+00 -5.30977190e-01 1.93976998e-01 -4.10497904e-01
3.70056391e-01 2.15534329e-01 4.20500517e-01 8.77070725e-01
-1.59354910e-01 8.87290597e-01 2.02979654e-01 -8.03602278e-01
-8.60696018e-01 -1.37941027e+00 -5.50582170e-01 7.99248099e-01
-3.28955412e-01 -1.80075869e-01 -5.30974925e-01 -9.20576513e-01
-1.24969900e-01 3.40728253e-01 -7.50727057e-01 3.90541628e-02
-6.32313192e-01 -1.13042784e+00 7.53131866e-01 8.87679398e-01
2.46924728e-01 -1.18129742e+00 -1.15702260e+00 4.95531648e-01
1.58830255e-01 -1.05061030e+00 -3.59325081e-01 4.40901130e-01
-1.10963106e+00 -1.03562081e+00 -5.43323159e-01 -6.86254561e-01
7.35721588e-01 -2.16989201e-02 1.46952736e+00 3.47918838e-01
-8.64350259e-01 2.47208133e-01 -1.53021067e-01 -3.16881746e-01
-4.51851219e-01 4.52736199e-01 2.17499752e-02 -7.26049781e-01
-1.69512015e-02 -4.39625412e-01 -7.68203080e-01 -3.82522568e-02
-1.01398623e+00 1.75948411e-01 5.94680846e-01 1.01903081e+00
7.89225698e-01 -2.29827732e-01 1.45692363e-01 -1.17413139e+00
4.90772784e-01 -6.22528732e-01 -3.57300639e-01 3.04713435e-02
-3.55492145e-01 -1.11929625e-01 6.62064552e-01 -1.07330248e-01
-8.13181937e-01 2.79482961e-01 -3.85931015e-01 -3.98257285e-01
-8.26553851e-02 4.45910454e-01 5.99138737e-01 -3.14925045e-01
7.85424769e-01 5.18357344e-02 9.76226032e-02 -3.61905806e-02
-2.59340685e-02 4.87557650e-01 7.11155415e-01 -4.76956755e-01
6.92550063e-01 6.32987797e-01 3.24278623e-01 -4.15309727e-01
-7.41476834e-01 -5.56329310e-01 -6.26914203e-01 4.79653943e-03
8.84864628e-01 -9.04856145e-01 -8.28682423e-01 4.46373552e-01
-6.31099224e-01 -4.34204340e-01 -3.53800267e-01 4.84723061e-01
-5.18674672e-01 -1.01059414e-01 -1.02322102e+00 1.49898440e-01
-1.14433658e+00 -1.59061277e+00 1.27274418e+00 1.05483800e-01
-3.28827262e-01 -1.06234372e+00 1.71255231e-01 1.70699377e-02
5.36841750e-01 7.67209828e-01 8.21873367e-01 -5.56542754e-01
-3.92967463e-01 -2.92055994e-01 -2.84829587e-01 -2.33575888e-02
2.84334362e-01 -1.75341710e-01 -7.81249523e-01 -6.25663936e-01
7.02717900e-02 -3.99222553e-01 6.29062712e-01 5.82551897e-01
1.61836207e+00 2.45167166e-02 -3.90411377e-01 1.20385957e+00
1.59839094e+00 -4.03124355e-02 3.10011357e-01 1.80005774e-01
9.43734109e-01 5.19340448e-02 2.09493846e-01 4.53175128e-01
3.25871766e-01 3.50686640e-01 4.55629587e-01 -5.09117782e-01
1.79151502e-02 4.39158827e-01 -3.89021128e-01 8.65020990e-01
-1.93782419e-01 -1.55777305e-01 -1.40750277e+00 3.83564413e-01
-1.40724730e+00 -5.21442354e-01 -2.10257739e-01 1.92634881e+00
6.70968354e-01 1.73937097e-01 2.49241348e-02 -2.43704304e-01
2.36664042e-01 -1.49075896e-01 -5.65155625e-01 -4.66466129e-01
3.12790155e-01 8.63053203e-01 8.71701300e-01 1.76593095e-01
-9.58905756e-01 5.33317626e-01 6.98566389e+00 6.83518350e-01
-1.46804249e+00 2.68079221e-01 7.42138326e-01 -6.47680521e-01
1.66802686e-02 -7.02502787e-01 -5.12419403e-01 2.86551476e-01
1.15485728e+00 7.73059502e-02 -8.76299217e-02 8.36567760e-01
1.10666849e-01 -1.93498656e-01 -1.01635206e+00 9.25229609e-01
-1.20701604e-01 -1.53000867e+00 -1.15584962e-01 3.27704042e-01
6.60914600e-01 6.83656693e-01 1.75267339e-01 3.18335354e-01
6.78270757e-01 -1.37713850e+00 2.08495781e-01 5.30394167e-02
1.16631949e+00 -9.37967479e-01 8.98013711e-01 1.30670950e-01
-1.07716680e+00 2.01728091e-01 -1.76223844e-01 4.16310132e-01
2.43848532e-01 3.90447378e-01 -1.49082553e+00 6.17002964e-01
7.72611558e-01 3.67104352e-01 -6.05967700e-01 9.55561757e-01
7.58280978e-02 5.84046602e-01 -6.02823853e-01 4.73921150e-01
6.10312283e-01 4.38164502e-01 3.36116463e-01 1.32252026e+00
3.69452506e-01 2.57423937e-01 3.29781711e-01 7.13870749e-02
1.69009920e-02 -1.01904515e-02 -2.33369708e-01 6.02347076e-01
2.05697730e-01 1.65177310e+00 -1.05570877e+00 -7.61544943e-01
-1.58926293e-01 8.11459482e-01 9.87441242e-02 -3.24001461e-01
-1.04779124e+00 6.62412122e-02 3.90647560e-01 2.61307023e-02
2.91928798e-01 1.57783642e-01 -4.88467842e-01 -1.02494681e+00
-2.68444628e-01 -1.02534866e+00 8.05789292e-01 -5.78769326e-01
-1.37272692e+00 1.08235884e+00 -8.17373618e-02 -1.27426839e+00
-5.88882506e-01 -7.38400996e-01 -7.31954992e-01 8.83356035e-01
-1.16593003e+00 -1.10501099e+00 -7.36588180e-01 8.59667301e-01
3.38905752e-01 9.68649145e-03 1.15186548e+00 1.96858674e-01
-4.43271816e-01 7.29268432e-01 -7.28074759e-02 1.04961000e-01
5.59130192e-01 -1.42802608e+00 4.55804199e-01 3.55676830e-01
-5.04259944e-01 4.72210556e-01 3.52464914e-01 -6.50692225e-01
-1.32664800e+00 -1.38979328e+00 4.28242892e-01 -5.58419287e-01
4.70741451e-01 -1.32301629e-01 -8.19712579e-01 7.09673703e-01
2.96282824e-02 5.56269288e-01 9.03645217e-01 -1.39018491e-01
4.01977412e-02 -1.25702219e-02 -1.56916559e+00 3.57780129e-01
9.55291152e-01 -1.17052659e-01 -3.30277681e-01 5.34960389e-01
4.55666989e-01 -1.32467008e+00 -1.37472248e+00 5.89552641e-01
4.29859132e-01 -7.79806733e-01 1.05560267e+00 -3.80980402e-01
3.34116429e-01 8.58128071e-04 2.99779743e-01 -1.26280320e+00
-5.01237452e-01 -2.94444799e-01 1.06890179e-01 3.74872535e-01
4.72341835e-01 -8.04310739e-01 1.00033796e+00 4.42493677e-01
-3.48897606e-01 -1.23621154e+00 -9.54908967e-01 -5.36273897e-01
1.11294188e-01 -4.76262867e-01 7.11162746e-01 7.88637817e-01
-3.30725908e-01 -2.20947117e-02 1.35103706e-02 1.92051932e-01
5.78265488e-01 -2.73921549e-01 4.76324499e-01 -1.03755236e+00
-6.03374302e-01 -2.57703781e-01 -4.92294401e-01 -2.88850904e-01
-3.73799115e-01 -1.07075167e+00 1.17977463e-01 -1.35637951e+00
3.68550420e-01 -8.30228150e-01 -4.81917083e-01 5.45633793e-01
-1.38729095e-01 6.52373493e-01 3.61938290e-02 1.45873338e-01
-3.06711078e-01 -4.48604003e-02 1.52916646e+00 1.75643295e-01
-7.73071796e-02 1.36252090e-01 -4.54322815e-01 7.55405247e-01
8.68337274e-01 -6.99012160e-01 -6.05380356e-01 -5.59353530e-01
-4.78321053e-02 4.12762642e-01 3.07799965e-01 -1.32545924e+00
1.59963965e-01 6.77691549e-02 7.21280634e-01 -7.25539505e-01
3.25091749e-01 -7.11187303e-01 2.12226227e-01 9.01141882e-01
3.79410610e-02 5.13321996e-01 4.81251508e-01 2.25856945e-01
6.70183599e-02 3.67726199e-02 9.12134945e-01 -4.36039686e-01
-3.85562599e-01 7.75627911e-01 -6.23761952e-01 1.22887820e-01
1.16485250e+00 -1.53787091e-01 -1.68595508e-01 8.84724502e-03
-6.97410226e-01 3.53731692e-01 4.28472131e-01 1.88799366e-01
5.12028515e-01 -1.13100088e+00 -7.82049417e-01 2.05254391e-01
-4.80927080e-02 5.46167254e-01 7.28853583e-01 8.74234974e-01
-1.17415345e+00 3.20311576e-01 -2.63782650e-01 -1.16529930e+00
-1.25696766e+00 1.89282835e-01 6.41653299e-01 -1.00783288e+00
-1.06711388e+00 1.03619850e+00 1.23799190e-01 -5.49992681e-01
-1.21378951e-01 -5.28199613e-01 1.56219617e-01 -2.13141054e-01
6.30027294e-01 2.32180342e-01 6.82712495e-01 -5.94665051e-01
-6.08859122e-01 3.08726668e-01 -4.10093427e-01 -6.52162358e-02
1.51412630e+00 3.44499141e-01 2.61969596e-01 -1.67823344e-01
1.25042188e+00 -4.27107632e-01 -1.02270412e+00 -8.77180845e-02
-2.27878079e-01 -1.20691881e-01 4.39502209e-01 -9.92280364e-01
-1.37023294e+00 4.80032474e-01 8.21683407e-01 -2.29163811e-01
1.23725498e+00 -8.31971839e-02 9.84062135e-01 1.01558551e-01
4.56534594e-01 -8.18524182e-01 -1.80539750e-02 4.52200174e-01
5.67494750e-01 -1.12444055e+00 3.03089917e-01 -3.98579389e-01
-6.88880026e-01 1.22460294e+00 6.02411449e-01 -3.98885429e-01
5.58446944e-01 7.92143464e-01 4.52859968e-01 -6.20012283e-01
-7.66352355e-01 1.26942664e-01 3.26784790e-01 5.44641316e-01
3.33225995e-01 3.49885523e-01 1.71585441e-01 1.16594899e-02
-5.34612060e-01 1.89438986e-03 6.35122895e-01 1.23429716e+00
-1.69067219e-01 -9.48778927e-01 -3.01175594e-01 1.00803590e+00
-8.20873916e-01 -1.31823912e-01 3.63766789e-01 1.03113246e+00
9.61344317e-02 3.36320132e-01 2.77207136e-01 -7.38989487e-02
2.69036055e-01 -4.58261490e-01 4.76085871e-01 -1.20740449e+00
-1.40045762e+00 -2.86522619e-02 -1.06830329e-01 -8.63885939e-01
-1.72450006e-01 -6.28360868e-01 -1.30497158e+00 -2.73375958e-01
4.87048253e-02 -3.83895129e-01 5.85759342e-01 7.94487596e-01
2.50915945e-01 1.08648610e+00 4.84625936e-01 -7.39838541e-01
-6.31243408e-01 -7.66415060e-01 -4.26979393e-01 3.99225987e-02
3.41056168e-01 -4.96632129e-01 -2.73475386e-02 -3.04761708e-01]
|
[14.85074520111084, -2.2794744968414307]
|
c865034a-0660-423b-a4b5-3f2ac77b5415
|
semeval-2022-task-4-patronizing-and
| null | null |
https://aclanthology.org/2022.semeval-1.38
|
https://aclanthology.org/2022.semeval-1.38.pdf
|
SemEval-2022 Task 4: Patronizing and Condescending Language Detection
|
This paper presents an overview of Task 4 at SemEval-2022, which was focused on detecting Patronizing and Condescending Language (PCL) towards vulnerable communities. Two sub-tasks were considered: a binary classification task, where participants needed to classify a given paragraph as containing PCL or not, and a multi-label classification task, where participants needed to identify which types of PCL are present (if any). The task attracted more than 300 participants, 77 teams and 229 valid submissions. We provide an overview of how the task was organized, discuss the techniques that were employed by the different participants, and summarize the main resulting insights about PCL detection and categorization.
|
['Steven Schockaert', 'Luis Espinosa-Anke', 'Carla Perez-Almendros']
| null | null | null | null |
semeval-naacl-2022-7
|
['semeval-2022-task-4-1-binary-pcl-detection', 'semeval-2022-task-4-1-binary-pcl-detection', 'semeval-2022-task-4-2-multi-label-pcl', 'semeval-2022-task-4-1-binary-pcl-detection']
|
['miscellaneous', 'music', 'natural-language-processing', 'natural-language-processing']
|
[ 9.50709879e-02 -1.89489797e-01 -2.51681745e-01 -1.39241025e-01
-1.02818906e+00 -9.47191060e-01 7.23479033e-01 9.01369095e-01
-5.58382213e-01 6.37221813e-01 3.87703657e-01 -6.71422899e-01
1.63540587e-01 -4.18488920e-01 -8.66460800e-02 -3.68595511e-01
-2.42352098e-01 3.56321126e-01 -3.13360877e-02 2.13464722e-01
8.50361407e-01 3.59815359e-01 -1.13427293e+00 9.31080759e-01
8.00405204e-01 4.17888850e-01 -1.52111381e-01 7.38697171e-01
-2.24997059e-01 8.27447891e-01 -8.06187451e-01 -4.96624112e-01
-1.91699252e-01 -6.67700991e-02 -1.02116811e+00 -1.89861491e-01
8.80013049e-01 1.42929703e-01 1.59149945e-01 1.14034832e+00
4.93609011e-01 -1.53040558e-01 1.13407052e+00 -1.14692187e+00
-5.84269524e-01 8.87202919e-01 -8.05222690e-01 8.13523948e-01
7.03302503e-01 -2.63734162e-01 1.13322318e+00 -1.18527281e+00
5.95663011e-01 1.54513478e+00 1.05430746e+00 4.00609940e-01
-1.50204861e+00 -8.90646517e-01 2.67696261e-01 2.10487604e-01
-1.64951861e+00 -7.90799439e-01 4.34135765e-01 -1.33394599e+00
9.89780545e-01 3.35552365e-01 2.50417322e-01 1.34069526e+00
-1.67278513e-01 5.17697811e-01 1.53215551e+00 -3.88639063e-01
1.28815711e-01 7.36907184e-01 7.80894339e-01 1.69689074e-01
2.50054628e-01 -4.33950335e-01 -4.78382558e-01 -9.60651696e-01
-1.24556720e-01 -5.08792996e-01 -6.04987331e-02 3.48872572e-01
-1.01544476e+00 1.18783259e+00 -1.98842824e-01 6.89146101e-01
-1.30501926e-01 -4.49681312e-01 8.98925185e-01 3.81159633e-01
8.81710529e-01 4.13288146e-01 -1.63622066e-01 -6.57372037e-03
-1.29264402e+00 5.39096951e-01 8.92446697e-01 3.93757463e-01
3.27326477e-01 -6.10476613e-01 -5.98743379e-01 1.20028627e+00
3.89169365e-01 6.19238377e-01 2.00661108e-01 -8.32104266e-01
6.60384953e-01 3.96698207e-01 1.13575511e-01 -1.28595865e+00
-3.80959570e-01 -2.85618812e-01 -4.25908834e-01 4.42827418e-02
5.43648541e-01 -5.05699277e-01 -9.90053341e-02 1.52797651e+00
-8.07115734e-02 -2.03805625e-01 -4.43798840e-01 3.52904141e-01
9.65753615e-01 4.78249252e-01 3.90873492e-01 -2.80408651e-01
1.51585138e+00 -6.45056725e-01 -6.35037124e-01 -3.07718098e-01
7.80712426e-01 -1.07335186e+00 9.10439610e-01 3.34193379e-01
-1.02181828e+00 -2.42357880e-01 -5.90551138e-01 1.96500003e-01
-5.13037026e-01 5.76871745e-02 2.26937935e-01 1.19262326e+00
-1.37002802e+00 3.12470257e-01 -1.98806562e-02 -7.49756753e-01
4.07205909e-01 -1.65421173e-01 -3.71738911e-01 1.86736763e-01
-1.20045364e+00 7.32523799e-01 -1.86842203e-01 -3.47540349e-01
-7.61068940e-01 -7.35732079e-01 -5.00682592e-01 -2.14530364e-01
-1.54537752e-01 2.64087886e-01 9.99609828e-01 -7.39898026e-01
-7.46943653e-01 1.82479179e+00 -2.38787249e-01 5.16208783e-02
5.59079289e-01 3.00038066e-02 -6.87574863e-01 2.54457474e-01
7.60261834e-01 3.49607229e-01 6.63070142e-01 -1.03020835e+00
-7.07749963e-01 -2.41275728e-01 -1.96068197e-01 5.03586791e-03
-6.15859985e-01 1.15731239e+00 2.83009380e-01 -6.96663737e-01
-3.58171940e-01 -5.25261164e-01 1.39907643e-01 -3.24062735e-01
-6.59520030e-01 -7.26107240e-01 4.94335294e-01 -1.00282741e+00
1.43782544e+00 -2.16129541e+00 -2.63933241e-01 2.61598438e-01
7.10169196e-01 1.35091320e-01 -1.79107800e-01 9.70260203e-01
-1.63463309e-01 7.52231061e-01 2.15468146e-02 -6.59919441e-01
2.78882161e-02 -5.89017391e-01 -4.89803702e-01 8.21340501e-01
-3.66516300e-02 3.40605587e-01 -8.83441269e-01 -6.62067950e-01
-4.36056405e-01 2.13025644e-01 -1.39090776e-01 1.81380380e-02
3.22895437e-01 1.48098975e-01 -2.30358183e-01 7.31254458e-01
7.80308962e-01 2.67720222e-02 7.66536295e-02 5.76794446e-01
-7.52929330e-01 5.72777689e-01 -9.33390141e-01 5.80965400e-01
-1.73847273e-01 1.05669487e+00 6.58264577e-01 -7.41679370e-01
1.02128577e+00 3.30837250e-01 1.17702484e-01 -1.96107268e-01
-1.14768542e-01 3.07823479e-01 -2.86041468e-01 -4.88453716e-01
2.68596709e-01 -1.62147015e-01 -3.86128724e-01 9.37154710e-01
-5.65652549e-01 4.28547353e-01 4.12401080e-01 4.18381542e-01
1.36551142e+00 -5.97356319e-01 1.83386728e-01 -7.74447680e-01
1.05292606e+00 -5.23792487e-03 3.68057728e-01 1.16446364e+00
-7.80817986e-01 2.69096226e-01 1.10612857e+00 -3.89496654e-01
-8.24709415e-01 -9.48077619e-01 -4.07139599e-01 1.49920714e+00
-4.59439754e-01 -4.62328404e-01 -7.91357219e-01 -6.48421705e-01
-6.03228575e-03 4.32365149e-01 -6.72603726e-01 3.38696092e-01
-3.44774783e-01 -7.79965162e-01 1.07175267e+00 -2.98003536e-02
2.29868859e-01 -1.33984017e+00 -2.29869291e-01 -9.11879167e-02
-5.76389313e-01 -9.74321961e-01 -6.39586568e-01 -1.55371308e-01
-3.59327853e-01 -1.10511184e+00 -8.89862120e-01 -1.08159959e+00
4.17738229e-01 1.93745852e-01 9.77810681e-01 2.55574435e-01
-3.70693862e-01 1.56270579e-01 -4.50809181e-01 -1.04210414e-01
-6.53664827e-01 3.56382102e-01 1.36983946e-01 1.66032463e-01
6.57678902e-01 -1.82911813e-01 -7.38110170e-02 -1.00128815e-01
-1.88576549e-01 -3.09065163e-01 7.13458508e-02 2.80541599e-01
-3.35301250e-01 -1.23702608e-01 7.27156103e-01 -1.11954677e+00
1.22643721e+00 -8.27892840e-01 6.77388683e-02 3.16271186e-01
-1.54758260e-01 -8.19430351e-01 3.80517364e-01 -5.76829076e-01
-6.41663909e-01 -3.24395627e-01 -3.86798084e-01 4.16273117e-01
-4.61522788e-01 3.78457516e-01 1.71429262e-01 -2.16213360e-01
7.17604041e-01 -1.68456867e-01 -6.77325130e-02 -7.43845046e-01
-1.40767395e-01 1.41226721e+00 2.89987633e-03 -6.30093396e-01
6.33560658e-01 1.03913948e-01 -6.74097598e-01 -1.13367546e+00
-8.56873214e-01 -1.10552180e+00 -6.48878694e-01 -5.12869060e-01
8.86031687e-01 -9.44467604e-01 -6.15978956e-01 9.59493577e-01
-1.34553206e+00 -5.76125503e-01 3.47491980e-01 5.38524613e-02
1.47922903e-01 5.87446570e-01 -9.66429770e-01 -9.53966618e-01
-5.47004580e-01 -8.94995093e-01 7.95577586e-01 -1.43479869e-01
-6.92744911e-01 -1.09095550e+00 4.56470340e-01 7.11359441e-01
4.58554506e-01 1.19417995e-01 1.08575130e+00 -9.91685510e-01
6.82425618e-01 -2.78827190e-01 -5.76279640e-01 -2.18573902e-02
-2.92242050e-01 2.37615854e-01 -1.09028625e+00 -5.17037332e-01
-2.27871746e-01 -7.07891762e-01 6.65624022e-01 -9.10545513e-02
9.10138965e-01 -4.10438925e-01 -6.31277740e-01 -1.71047643e-01
9.26852405e-01 -2.55177587e-01 1.33395135e-01 3.28345329e-01
4.87496287e-01 1.06088519e+00 1.24948613e-01 4.45667803e-01
4.34491903e-01 6.00452542e-01 -5.68138529e-03 1.61100894e-01
-9.03920531e-02 -1.00755922e-01 8.32293272e-01 8.50273013e-01
3.59125346e-01 -8.24875981e-02 -1.46747804e+00 7.93479383e-01
-1.33514738e+00 -1.06631243e+00 -9.03338075e-01 1.84406435e+00
1.00717425e+00 -1.20210955e-02 6.76942289e-01 3.71604562e-01
1.24915707e+00 1.90635949e-01 4.93918657e-02 -7.64988363e-01
-7.73653165e-02 4.85778488e-02 2.38693804e-01 8.76014113e-01
-1.30972683e+00 8.42084587e-01 7.68242788e+00 1.08199370e+00
-9.34088886e-01 4.96640116e-01 7.70738423e-01 3.35110158e-01
-4.48002131e-04 -1.84983984e-01 -1.06570351e+00 5.60671449e-01
1.03308558e+00 -7.28745311e-02 2.08217114e-01 6.05252326e-01
1.78672105e-01 -2.16196060e-01 -7.61509895e-01 4.10716027e-01
4.44545120e-01 -9.09936488e-01 -1.71114653e-01 7.13020563e-02
3.56469005e-01 2.80037075e-01 -6.11339025e-02 5.60718596e-01
1.37186602e-01 -1.02477801e+00 1.01753461e+00 2.71223992e-01
9.68342602e-01 -2.75887847e-01 4.56350923e-01 7.00912654e-01
-9.41342175e-01 -1.92659780e-01 -1.55555874e-01 -7.08114922e-01
-3.09821609e-02 7.02434540e-01 -5.36174536e-01 -1.99560784e-02
9.22463536e-01 8.51934135e-01 -1.01738274e+00 1.24620879e+00
-2.25012109e-01 9.46729302e-01 2.94630900e-02 -3.11391950e-01
2.09234059e-01 6.46848753e-02 7.29314804e-01 1.86158490e+00
7.28590167e-05 -3.21867645e-01 4.51849133e-01 8.06423604e-01
2.84408301e-01 4.82777029e-01 -4.14362818e-01 -1.41559258e-01
6.95891321e-01 1.63142443e+00 -9.34378922e-01 -3.09997082e-01
-2.34742597e-01 6.47551060e-01 7.15598404e-01 3.16239744e-01
-2.98400909e-01 -6.16297066e-01 5.60175657e-01 5.37190318e-01
-1.86296970e-01 -8.57431218e-02 -6.01583183e-01 -9.55760002e-01
-3.48748595e-01 -9.40153301e-01 5.47842979e-01 -2.69073457e-01
-1.76444590e+00 4.82242197e-01 -8.05731714e-02 -4.83177632e-01
4.67772126e-01 -4.83834416e-01 -9.06825900e-01 1.14742935e+00
-1.14385009e+00 -8.89799058e-01 -1.66119725e-01 1.96438864e-01
2.41155788e-01 -2.84482151e-01 7.61632323e-01 7.49126434e-01
-7.73363948e-01 6.77961290e-01 -1.78652838e-01 2.93396562e-01
1.04590166e+00 -1.02464354e+00 1.33066103e-01 5.27150750e-01
-3.60388607e-01 7.17534602e-01 5.81752658e-01 -8.06682289e-01
-3.90785813e-01 -1.04918909e+00 2.16459632e+00 -5.89739799e-01
1.04535913e+00 -9.06736016e-01 -6.66990757e-01 3.55228871e-01
2.35590860e-01 -4.67941076e-01 9.60369527e-01 3.79475832e-01
-5.68485081e-01 4.17527407e-01 -1.18511307e+00 5.69642067e-01
8.56294215e-01 -8.92953157e-01 -5.49761117e-01 7.93748260e-01
3.18844110e-01 2.11676329e-01 -7.13470578e-01 -1.54852033e-01
2.90188700e-01 -7.46151447e-01 8.18125844e-01 -5.04224837e-01
4.72420931e-01 1.98135093e-01 4.93042879e-02 -7.68758595e-01
-7.38965631e-01 -4.51801240e-01 5.22106349e-01 1.60503674e+00
5.34741104e-01 -6.07420802e-01 4.02934700e-01 1.94043055e-01
4.06521782e-02 -3.70858371e-01 -1.05227482e+00 -5.54132819e-01
7.35489428e-01 -2.83716351e-01 1.40091076e-01 1.48589551e+00
3.82275224e-01 5.90313017e-01 -6.32910132e-02 -1.23498850e-01
6.72736406e-01 -2.85250694e-01 3.29654962e-01 -1.47639072e+00
1.53417498e-01 -9.86210763e-01 6.62448555e-02 -3.46935153e-01
6.12758338e-01 -1.33493233e+00 -2.36445263e-01 -1.28984511e+00
7.27002382e-01 -7.04252422e-01 -1.33794099e-02 7.24295795e-01
-3.30189168e-01 5.57171643e-01 1.30658746e-01 4.76169795e-01
-6.67319834e-01 -2.65547216e-01 2.71591902e-01 -2.40680695e-01
-3.39613743e-02 2.69530714e-01 -1.09564149e+00 5.50096750e-01
8.63164425e-01 -6.08358324e-01 3.03252637e-01 -5.22166453e-02
7.22970247e-01 -3.03075194e-01 4.88410681e-01 -8.84540796e-01
-1.82218722e-03 -1.52376011e-01 9.39635038e-02 -5.84141731e-01
-3.57810631e-02 -5.15006222e-02 -8.94504264e-02 6.21858656e-01
-7.88301408e-01 1.10986911e-01 1.89998150e-02 6.15112782e-02
1.38483837e-01 -8.13629508e-01 6.76172078e-01 -7.19599724e-02
-2.61118382e-01 -3.92978862e-02 -1.34364188e+00 3.56747270e-01
1.14886498e+00 1.48972943e-01 -4.72238064e-01 -9.29009765e-02
-7.07386017e-01 2.91263163e-01 3.47649813e-01 3.26120913e-01
2.15859801e-01 -9.72886086e-01 -1.23558831e+00 -9.24068391e-02
1.97251245e-01 -1.12506580e+00 1.47842810e-01 9.57909167e-01
-4.53992963e-01 3.49611282e-01 -1.14747949e-01 -2.02528477e-01
-1.68039048e+00 3.92747015e-01 2.01441035e-01 -4.17712361e-01
-5.07132471e-01 9.06651676e-01 -1.36181628e-02 -8.36138368e-01
4.76927906e-01 4.45889801e-01 -1.01456082e+00 7.22097397e-01
9.01242793e-01 7.53830910e-01 -6.74543018e-03 -9.45941627e-01
-7.43750989e-01 4.12590176e-01 -2.48271018e-01 -2.97076076e-01
9.97111380e-01 -5.52407168e-02 -8.36368144e-01 8.16270471e-01
1.07993710e+00 3.29478115e-01 -3.80926132e-01 -2.29412347e-01
5.19462466e-01 -1.99324101e-01 -8.33357647e-02 -8.69092762e-01
-4.00891662e-01 1.15463531e+00 1.44724056e-01 5.55498958e-01
3.67987812e-01 9.54059511e-03 4.94412839e-01 -2.46266395e-01
3.07674885e-01 -1.15554559e+00 7.88223073e-02 7.66381681e-01
9.98799801e-01 -1.02697706e+00 -1.91737153e-02 -6.34649932e-01
-5.24774790e-01 1.07084858e+00 3.62506211e-01 7.26461112e-02
6.56465232e-01 7.74033964e-02 8.30343515e-02 -3.75628173e-01
-6.90687001e-01 2.07783967e-01 1.03203036e-01 8.30123186e-01
7.01266050e-01 3.47569227e-01 -7.27609038e-01 6.82779133e-01
-1.50551111e-01 -4.24915135e-01 7.24016905e-01 6.28475189e-01
-4.98097897e-01 -1.03816485e+00 -5.89084923e-01 6.62079275e-01
-9.37139213e-01 -3.11224550e-01 -1.10736823e+00 2.38019347e-01
3.61439139e-01 1.44759011e+00 1.21254675e-01 -3.01578134e-01
1.09435000e-01 1.32504225e-01 -8.84453356e-02 -1.12373590e+00
-1.46465480e+00 -1.00571908e-01 5.71537554e-01 6.05543479e-02
-2.10507959e-01 -1.31916857e+00 -6.34965062e-01 -6.21422172e-01
-6.15078658e-02 3.78617764e-01 2.92539269e-01 8.19696784e-01
-1.53995748e-03 -1.37202024e-01 5.84940732e-01 -5.96503139e-01
-4.94121879e-01 -1.27121449e+00 -7.08773732e-01 6.80086836e-02
3.11612099e-01 -4.05232996e-01 -8.00069690e-01 -1.77503452e-01]
|
[8.790444374084473, 10.584943771362305]
|
1d942457-4d32-42ff-8f1b-e97ebacce70d
|
high-fidelity-generalized-emotional-talking
|
2305.02572
| null |
https://arxiv.org/abs/2305.02572v2
|
https://arxiv.org/pdf/2305.02572v2.pdf
|
High-fidelity Generalized Emotional Talking Face Generation with Multi-modal Emotion Space Learning
|
Recently, emotional talking face generation has received considerable attention. However, existing methods only adopt one-hot coding, image, or audio as emotion conditions, thus lacking flexible control in practical applications and failing to handle unseen emotion styles due to limited semantics. They either ignore the one-shot setting or the quality of generated faces. In this paper, we propose a more flexible and generalized framework. Specifically, we supplement the emotion style in text prompts and use an Aligned Multi-modal Emotion encoder to embed the text, image, and audio emotion modality into a unified space, which inherits rich semantic prior from CLIP. Consequently, effective multi-modal emotion space learning helps our method support arbitrary emotion modality during testing and could generalize to unseen emotion styles. Besides, an Emotion-aware Audio-to-3DMM Convertor is proposed to connect the emotion condition and the audio sequence to structural representation. A followed style-based High-fidelity Emotional Face generator is designed to generate arbitrary high-resolution realistic identities. Our texture generator hierarchically learns flow fields and animated faces in a residual manner. Extensive experiments demonstrate the flexibility and generalization of our method in emotion control and the effectiveness of high-quality face synthesis.
|
['Yong liu', 'Zhifeng Xie', 'Chengjie Wang', 'Ying Tai', 'Wenqing Chu', 'Yue Han', 'Jiangning Zhang', 'Junwei Zhu', 'Chao Xu']
|
2023-05-04
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Xu_High-Fidelity_Generalized_Emotional_Talking_Face_Generation_With_Multi-Modal_Emotion_Space_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Xu_High-Fidelity_Generalized_Emotional_Talking_Face_Generation_With_Multi-Modal_Emotion_Space_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['talking-face-generation', 'face-generation']
|
['computer-vision', 'computer-vision']
|
[ 1.89951658e-01 7.21183186e-03 9.36208069e-02 -6.41538084e-01
-5.41837990e-01 -4.05069917e-01 4.13236350e-01 -1.06666780e+00
2.03454390e-01 6.19731009e-01 4.10254240e-01 3.89724344e-01
1.99523076e-01 -6.12136006e-01 -7.15383291e-01 -5.58898389e-01
3.90871257e-01 1.05046995e-01 -3.14710021e-01 -3.79130065e-01
-2.19858304e-01 2.74471223e-01 -1.81012082e+00 5.34695804e-01
8.39259684e-01 1.31620967e+00 9.40909758e-02 7.27172613e-01
-1.73327655e-01 8.43540907e-01 -6.84187770e-01 -7.36848176e-01
1.79372668e-01 -8.43828082e-01 -5.75996876e-01 4.53360766e-01
4.87890810e-01 -5.61101139e-01 -3.33561122e-01 9.21250641e-01
7.86471844e-01 1.84354797e-01 5.16264677e-01 -1.67919743e+00
-9.96253610e-01 1.83909282e-01 -5.24407923e-01 -4.23737377e-01
6.57990456e-01 2.30581611e-01 7.40943909e-01 -1.01101542e+00
8.67951572e-01 1.56273925e+00 4.52561080e-01 9.76205468e-01
-1.07667029e+00 -9.81359422e-01 2.15396985e-01 2.20212996e-01
-1.20673084e+00 -8.11989248e-01 1.23156154e+00 -2.86476701e-01
4.48825091e-01 2.26241127e-01 7.85451293e-01 1.74956298e+00
-5.26010394e-02 6.95270836e-01 9.37662542e-01 -1.46967679e-01
1.45540804e-01 1.47714540e-01 -8.27255487e-01 6.99812233e-01
-5.54239750e-01 1.52355701e-01 -8.19185436e-01 4.68702801e-02
9.54973698e-01 -1.68294515e-02 -3.61207992e-01 -1.71506301e-01
-9.89313483e-01 8.10166478e-01 1.46969602e-01 3.20985578e-02
-2.95377582e-01 1.27735212e-01 5.29329956e-01 3.62174541e-01
5.03315866e-01 2.63675749e-01 -2.95657456e-01 -2.22339377e-01
-9.42904711e-01 4.75710556e-02 5.98454952e-01 1.17768562e+00
6.56690955e-01 6.26438320e-01 -3.71447653e-01 1.07384062e+00
2.07965881e-01 5.56491494e-01 3.65831017e-01 -1.31037033e+00
8.42505917e-02 1.88649118e-01 -1.50997788e-01 -1.14296496e+00
-2.82723876e-03 -1.41640276e-01 -1.08487117e+00 4.15510014e-02
-2.13462412e-01 -4.05289322e-01 -8.59586239e-01 2.03388214e+00
3.91015798e-01 7.02383995e-01 1.04327455e-01 1.09281814e+00
8.00421476e-01 8.37492287e-01 3.01586054e-02 -3.32413167e-01
1.24714148e+00 -1.02457798e+00 -1.13521338e+00 -2.35829386e-03
2.16149893e-02 -9.06870604e-01 1.25897467e+00 3.49292010e-01
-1.25142753e+00 -8.38726640e-01 -7.37479627e-01 -1.32784516e-01
7.32646361e-02 1.60627887e-01 6.03318989e-01 4.43175137e-01
-1.05465114e+00 2.51012295e-01 -4.02620763e-01 -1.81111116e-02
4.34386075e-01 1.64265204e-02 -3.75828892e-01 -1.02034822e-01
-1.43189585e+00 4.35582072e-01 -7.93309733e-02 2.00139344e-01
-9.05356348e-01 -7.21512079e-01 -1.14187384e+00 8.81316885e-03
2.28091091e-01 -8.85478675e-01 1.10639453e+00 -1.56075203e+00
-2.12874985e+00 6.06055796e-01 -2.34289885e-01 1.41787097e-01
4.44717705e-01 -1.54197320e-01 -6.92801178e-01 6.45301223e-01
2.04535276e-02 1.08568537e+00 1.48796558e+00 -1.31996560e+00
-3.28384697e-01 -5.86038791e-02 -1.09027609e-01 2.55756319e-01
-6.25718892e-01 1.71496585e-01 -6.12400293e-01 -1.11476398e+00
-2.70978957e-01 -6.88902378e-01 -3.60067897e-02 3.93026471e-01
-1.56707197e-01 2.94966668e-01 1.15700638e+00 -5.52741647e-01
1.02504694e+00 -2.39974141e+00 3.52522999e-01 -3.37508842e-02
-8.40567611e-03 -1.62896842e-01 -5.50463855e-01 1.14871882e-01
-1.64817348e-01 5.30405752e-02 -2.21135780e-01 -4.20220286e-01
1.44533515e-01 2.20523894e-01 -5.01557171e-01 1.63893849e-01
7.34708190e-01 9.25057352e-01 -8.22209954e-01 -7.31310308e-01
2.24711552e-01 1.05787504e+00 -1.08352172e+00 4.99437928e-01
-2.08704248e-01 7.12336123e-01 -4.72422212e-01 7.83043146e-01
6.34623289e-01 -9.52506363e-02 -6.75964952e-02 -5.01646399e-01
2.62534201e-01 -3.02916855e-01 -1.24886131e+00 2.04062271e+00
-8.44799101e-01 4.59641963e-01 3.93104106e-01 -6.63767099e-01
1.12163687e+00 6.64060891e-01 6.01864696e-01 -6.70778096e-01
1.95747480e-01 1.94022397e-03 -3.26293886e-01 -6.96637511e-01
4.97198939e-01 -5.30209780e-01 -7.72392005e-02 1.66449532e-01
4.56202954e-01 -4.10433739e-01 -1.76718295e-01 -5.21697365e-02
7.03346014e-01 4.43846524e-01 -1.92432866e-01 1.11230977e-01
5.39038718e-01 -6.80453300e-01 8.26134861e-01 -1.23648820e-02
-1.62494436e-01 1.06558597e+00 4.40089047e-01 -1.64750367e-01
-9.61291671e-01 -9.11002278e-01 -8.06940272e-02 1.27615297e+00
2.21302405e-01 -3.98064464e-01 -8.05060208e-01 -5.10000229e-01
-3.21986139e-01 4.28606451e-01 -6.11747980e-01 -4.68103141e-01
-4.80080366e-01 -3.02536100e-01 5.89551210e-01 3.97222459e-01
6.14945412e-01 -1.33069944e+00 -3.82066816e-01 1.77835539e-01
-5.83504200e-01 -1.29580796e+00 -8.77360821e-01 -4.13252771e-01
-3.94506991e-01 -6.65804863e-01 -9.24171329e-01 -8.99188876e-01
4.68283534e-01 -1.26485392e-01 9.44365919e-01 -1.86866075e-01
-3.64256531e-01 5.80510616e-01 -4.72646177e-01 -6.72475770e-02
-3.35843325e-01 -4.11761671e-01 -1.82151347e-02 6.41006827e-01
-1.87050059e-01 -7.76302695e-01 -6.82587385e-01 3.19873422e-01
-1.09106505e+00 1.18179858e-01 3.63675416e-01 1.08799720e+00
4.01788980e-01 -9.00644064e-02 9.40903306e-01 -4.66330886e-01
4.94939357e-01 -4.18937206e-01 -2.27116104e-02 1.49346307e-01
-3.89698558e-02 -5.02892695e-02 9.23022032e-01 -7.84796417e-01
-1.44766057e+00 3.69440466e-02 -4.22783464e-01 -1.13454139e+00
-2.27328375e-01 -8.46442878e-02 -7.17803895e-01 -3.11057102e-02
3.06993842e-01 2.45386481e-01 7.53103793e-02 -6.66673407e-02
6.56476498e-01 5.72615683e-01 7.32100666e-01 -9.23977435e-01
6.96477473e-01 4.31648761e-01 -2.15146139e-01 -7.76898384e-01
-5.46654999e-01 1.08102143e-01 -2.47248650e-01 -5.76204419e-01
9.25677717e-01 -1.18822694e+00 -6.34328127e-01 5.10085106e-01
-1.22449386e+00 -1.91014230e-01 -4.31096226e-01 2.90716380e-01
-9.36732292e-01 3.31423342e-01 -8.19128454e-01 -6.74746931e-01
-2.30744660e-01 -1.29261625e+00 1.51943290e+00 1.65423885e-01
-2.08388120e-01 -7.01238394e-01 -3.04896772e-01 2.99161762e-01
5.32120287e-01 4.79305744e-01 5.68502724e-01 1.44915342e-01
-2.69100547e-01 2.00380579e-01 -2.02568769e-01 3.23651761e-01
1.16940558e-01 1.78197309e-01 -1.09600925e+00 -1.38705835e-01
6.01907782e-02 -8.46954286e-01 5.06649137e-01 1.62869692e-01
1.44206297e+00 -4.37832385e-01 1.00028321e-01 1.05334401e+00
1.10573280e+00 1.42129779e-01 7.10304618e-01 -3.09387952e-01
7.15173483e-01 8.46025288e-01 5.09757936e-01 6.97274089e-01
2.11213753e-01 7.11591363e-01 3.37383568e-01 -2.45677948e-01
-1.99053869e-01 -3.89513433e-01 5.55832803e-01 8.37447107e-01
1.26188040e-01 -2.34614506e-01 -2.63692647e-01 2.56533056e-01
-1.48584688e+00 -1.26284444e+00 5.25056899e-01 1.49815524e+00
1.07205248e+00 -2.99805790e-01 -4.55081500e-02 -5.01192436e-02
8.81355524e-01 2.06525758e-01 -6.41887248e-01 -3.74291480e-01
-2.65861899e-01 3.70910883e-01 -3.74896109e-01 3.73304278e-01
-7.69085824e-01 1.06836152e+00 5.81724882e+00 9.99665916e-01
-1.46146083e+00 1.08898260e-01 7.82577634e-01 -2.92045087e-01
-6.40958309e-01 -3.68591070e-01 -4.52007234e-01 5.73222578e-01
7.16526210e-01 -4.01261598e-02 5.02248526e-01 8.48058879e-01
1.70997351e-01 5.39992392e-01 -9.64161277e-01 1.34503889e+00
3.54945570e-01 -1.11135912e+00 3.17932308e-01 -2.12612584e-01
7.19930947e-01 -8.63786817e-01 2.27165163e-01 4.37388301e-01
-5.92409633e-02 -1.02779555e+00 9.86613512e-01 7.78443873e-01
1.51860416e+00 -7.76607811e-01 1.83849767e-01 -1.32422715e-01
-1.36735022e+00 -1.22526206e-01 -6.04481436e-02 1.75012365e-01
4.45220560e-01 3.08681101e-01 -1.48617357e-01 5.32306373e-01
7.28524387e-01 8.13976586e-01 -1.02951773e-01 4.05311674e-01
-5.39478026e-02 3.63300383e-01 -1.30850568e-01 3.57662439e-01
4.47275080e-02 -1.57389849e-01 4.66510355e-01 1.18182111e+00
5.96837401e-01 2.90105551e-01 2.47353584e-01 9.37095284e-01
-2.78873384e-01 1.17630117e-01 -6.17803335e-01 1.32169183e-02
4.22088563e-01 1.40360725e+00 -3.65445733e-01 -2.15418860e-01
-4.52584565e-01 1.42648423e+00 5.26018851e-02 6.02439642e-01
-1.22596788e+00 -3.86783093e-01 1.03574169e+00 -1.30023718e-01
2.70144612e-01 1.61660552e-01 1.31100789e-02 -1.38308263e+00
3.98227833e-02 -1.07317913e+00 3.27207670e-02 -1.04784298e+00
-1.39094317e+00 9.85162556e-01 -3.59355718e-01 -1.39090431e+00
-4.56967026e-01 -4.25952911e-01 -5.78404427e-01 6.86171889e-01
-1.33515954e+00 -1.28926492e+00 -4.06401932e-01 9.68965113e-01
7.94355333e-01 -2.85790145e-01 7.75379002e-01 5.43790102e-01
-5.63216329e-01 9.05743182e-01 -5.09002745e-01 1.47936806e-01
1.08610213e+00 -8.13271582e-01 -1.44135654e-01 6.60730302e-01
-2.19066277e-01 3.33320796e-01 5.16649961e-01 -4.45019841e-01
-1.42071652e+00 -1.32925093e+00 2.99489826e-01 -9.87023562e-02
4.34428036e-01 -5.55981874e-01 -8.21282446e-01 3.57341707e-01
3.95794332e-01 2.36144349e-01 7.25473940e-01 -3.84260207e-01
-4.02471453e-01 -2.12376162e-01 -1.18451905e+00 8.27679157e-01
1.23612809e+00 -7.34044433e-01 -3.35910410e-01 -1.20278813e-01
1.06170547e+00 -3.91291589e-01 -9.49541330e-01 6.22808635e-01
6.54106617e-01 -1.06559968e+00 9.37577128e-01 -6.06262565e-01
1.00337183e+00 -2.01986879e-01 -2.89898723e-01 -1.30566645e+00
-2.78583705e-01 -9.38667715e-01 -1.38310999e-01 1.60526478e+00
1.38075218e-01 -2.27539077e-01 5.10950983e-01 4.20169383e-01
-2.65042722e-01 -7.56422341e-01 -6.72225773e-01 -4.07233447e-01
-1.94308802e-01 -4.49142069e-01 9.76821840e-01 1.24378538e+00
-1.90722913e-01 4.80008990e-01 -9.12121534e-01 -1.79765362e-03
3.59712064e-01 3.39999646e-01 6.65318072e-01 -8.39142740e-01
-3.24262708e-01 -4.28828329e-01 -2.47891068e-01 -8.78495276e-01
6.69669092e-01 -6.91715777e-01 1.33272141e-01 -1.03112876e+00
-4.28148881e-02 -1.78742796e-01 1.18688932e-02 2.98519880e-01
-1.88241318e-01 4.68676150e-01 2.26651803e-01 -1.78205922e-01
-4.93202239e-01 1.29837263e+00 1.72129714e+00 2.46006362e-02
-2.86383405e-02 -4.93394315e-01 -7.59699583e-01 6.24705434e-01
4.87086177e-01 -2.68723350e-02 -7.53009021e-01 -4.39844757e-01
5.61223142e-02 5.63381672e-01 5.10839164e-01 -9.32717621e-01
6.14734180e-03 -3.28207999e-01 6.78900898e-01 -6.52625933e-02
7.06527591e-01 -9.50379491e-01 3.17281455e-01 -1.23371355e-01
-4.63949025e-01 -9.98827368e-02 1.23472340e-01 5.03351867e-01
-5.80420256e-01 3.24961126e-01 9.63513732e-01 4.75332066e-02
-6.22912109e-01 8.10865045e-01 1.76460799e-02 2.06688643e-01
9.67248082e-01 -3.00042629e-01 -7.57046230e-03 -7.23604798e-01
-8.98652732e-01 1.87380239e-01 4.77585703e-01 7.87289619e-01
8.36809635e-01 -1.77673745e+00 -7.43717730e-01 6.26791894e-01
8.44633859e-03 -5.66844717e-02 7.31486559e-01 4.07572150e-01
-5.53338602e-02 -2.63922721e-01 -5.88002443e-01 -5.38275778e-01
-8.69963586e-01 5.28846383e-01 3.81440669e-01 2.99532562e-01
-6.03250325e-01 8.21302712e-01 5.03721833e-01 -3.58697742e-01
2.28962973e-01 1.93758234e-01 -2.37758160e-02 1.82606906e-01
6.62796915e-01 2.51321867e-02 -3.26585859e-01 -8.17117155e-01
1.77401397e-02 7.56825924e-01 3.58885288e-01 -3.37740004e-01
1.20258117e+00 -2.94219404e-01 1.45523259e-02 3.20088714e-01
1.29708683e+00 -8.67458433e-03 -1.65043449e+00 8.81921947e-02
-6.03295743e-01 -5.78586102e-01 -1.62782729e-01 -5.04742444e-01
-1.49405062e+00 1.02930856e+00 3.90178740e-01 -2.29280755e-01
1.72982836e+00 -2.40855649e-01 1.05306911e+00 -5.44915348e-02
1.96599141e-01 -1.08295751e+00 6.66356683e-01 3.20554495e-01
1.27817452e+00 -9.76204813e-01 -6.59043729e-01 -5.60481727e-01
-1.02225888e+00 1.03982389e+00 1.01143789e+00 9.33202729e-02
5.84230900e-01 5.21925390e-01 1.34875119e-01 5.24761938e-02
-9.88067985e-01 1.09901100e-01 1.67294011e-01 8.00408900e-01
4.07359958e-01 -2.46308535e-01 2.59276479e-01 8.22515428e-01
-4.84573454e-01 1.67631701e-01 3.36872637e-01 4.87518042e-01
-4.05065492e-02 -8.63561213e-01 -4.08023655e-01 1.17882088e-01
-3.45939487e-01 -2.85481084e-02 -1.62250027e-01 4.75974262e-01
3.16989034e-01 8.75021517e-01 1.21098362e-01 -4.99411970e-01
3.23931962e-01 3.46541554e-01 5.27872801e-01 -3.68395567e-01
-3.17082137e-01 3.60398054e-01 -1.73797354e-01 -7.93379426e-01
-3.72026145e-01 -4.00897771e-01 -1.11837769e+00 -2.13673666e-01
-6.30195811e-02 6.93306401e-02 2.22933576e-01 6.60363734e-01
6.86554551e-01 6.87768877e-01 1.10229051e+00 -1.04400623e+00
-1.58123359e-01 -7.99022496e-01 -5.58037996e-01 6.54504299e-01
3.85409832e-01 -7.47020721e-01 -2.72201836e-01 4.82976019e-01]
|
[13.03292179107666, -0.3666633665561676]
|
e333ad71-7efd-471f-a3c0-9047f956a864
|
semi-supervised-object-detection-via-virtual
| null | null |
https://openreview.net/forum?id=HJWD_2bApjI
|
https://openreview.net/pdf?id=HJWD_2bApjI
|
Semi-supervised Object Detection via Virtual Category Learning
|
Due to the lack of large amounts of labelled data to learn rich-expressive features of objects, semi-supervised detectors powered by pseudo labelling techniques usually make a tentative decision for the pseudo labels of confusing samples. When dealing with confusing training samples, neither of the two recently adopted strategies, i.e., discarding or retaining, is optimal. Arbitrarily discarding the valuable confusing samples would compromise the generalisation ability of the model, while using them for model training would exacerbate the confirmation bias issue caused by inevitable mislabelling. To remedy this situation, this paper, for the first time, proposes to make use of these confusing samples without label correction, instead of discarding them. To this end, an alternative virtual category (VC), which is safe for optimisation, is provided for each confusing sample such that they can still contribute to better decision-boundary finding even without a concrete label. It is attributed to a new VC loss formulating the embedding distance between the training sample and the virtual category as the lower bound of the inter-class distance. Moreover, we also modify the localisation loss to allow the high-quality bounding boxes to be used for location regression training. Extensive experimental results demonstrate that the proposed semi-supervised detector underpinned by the VC learning surpasses the current state-of-the-art, when extremely small amounts of annotated labels are available for training (e.g., 0.5\% and 1\% label ratios).
|
['Anonymous']
|
2021-11-25
| null | null | null | null |
['semi-supervised-object-detection']
|
['computer-vision']
|
[ 4.14945036e-01 6.50429904e-01 -3.47251773e-01 -5.42503774e-01
-5.79734087e-01 -4.13677216e-01 7.74341464e-01 6.00589752e-01
-7.77830005e-01 9.19175982e-01 -4.44068640e-01 -1.14233479e-01
-2.99713641e-01 -6.09181106e-01 -5.30253410e-01 -9.96132612e-01
3.49731073e-02 5.73326707e-01 4.89109039e-01 2.41235808e-01
1.90542638e-01 4.66159672e-01 -2.00092006e+00 2.00801954e-01
9.40370560e-01 1.22182727e+00 3.84950757e-01 -4.93500531e-02
-3.11036050e-01 2.74354935e-01 -6.84859753e-01 -3.50908488e-01
3.20410848e-01 -1.97664484e-01 -4.76473123e-01 2.25052580e-01
2.08970279e-01 -8.48922506e-02 4.95111257e-01 1.00242269e+00
2.74630010e-01 1.03898682e-01 8.28076839e-01 -1.12520897e+00
-7.00704604e-02 6.26541078e-01 -5.78272343e-01 -9.24351960e-02
4.98237908e-02 -2.53539402e-02 9.95139539e-01 -9.08859491e-01
6.27796292e-01 8.34631085e-01 6.12401605e-01 5.75591624e-01
-1.36451888e+00 -6.62810206e-01 3.22523743e-01 5.91023266e-02
-1.73619580e+00 -5.43220520e-01 9.05467093e-01 -4.16145235e-01
3.99523377e-01 3.33780915e-01 3.55424106e-01 9.73816514e-01
-2.08021656e-01 6.06417000e-01 1.20920777e+00 -7.40834832e-01
5.34593642e-01 9.33398545e-01 -5.02608065e-03 4.35092181e-01
4.84954923e-01 3.45158167e-02 -2.05908477e-01 -1.22237854e-01
1.99605405e-01 -2.80289501e-01 -2.40763322e-01 -1.01075149e+00
-8.86270940e-01 8.72691333e-01 4.83922035e-01 5.07312536e-01
-1.87658682e-01 -2.32904822e-01 4.30238724e-01 -1.15243077e-01
6.40103400e-01 4.08276796e-01 -2.89952606e-01 2.64868498e-01
-1.17494464e+00 7.84455761e-02 3.92537683e-01 8.52820694e-01
9.07383382e-01 -1.98414862e-01 -1.91006288e-01 8.82175863e-01
4.89912331e-01 5.97217791e-02 3.76274377e-01 -3.98551315e-01
4.64021295e-01 9.63363290e-01 2.10987732e-01 -8.82301629e-01
-4.73674148e-01 -8.33003998e-01 -6.37862980e-01 3.79857570e-01
6.84844792e-01 8.35909918e-02 -7.75803447e-01 1.62124121e+00
6.72183692e-01 8.50282013e-02 -1.00006565e-01 9.13607359e-01
4.53998238e-01 7.23309815e-02 1.60969034e-01 -3.59193563e-01
1.20506871e+00 -6.32951438e-01 -5.73092103e-01 -2.69845784e-01
1.11949956e+00 -4.59750831e-01 9.74533617e-01 4.48010296e-01
-5.17878592e-01 -6.23307168e-01 -1.29248536e+00 4.57077026e-01
-5.53460479e-01 4.42211896e-01 3.94889534e-01 8.47765625e-01
-5.35343707e-01 6.92352891e-01 -5.23860812e-01 -2.01617852e-01
4.82584238e-01 4.01360482e-01 -3.77699375e-01 -7.11965933e-02
-1.07816494e+00 9.52296793e-01 8.20587635e-01 3.59143078e-01
-4.54366863e-01 -4.79147017e-01 -7.37252116e-01 -2.07862854e-01
7.07830250e-01 -4.29716334e-02 8.19612622e-01 -1.05653906e+00
-1.08760333e+00 1.01697671e+00 1.31244525e-01 -6.09250128e-01
9.17307734e-01 6.94547817e-02 -3.87751818e-01 1.08505851e-02
2.35506445e-01 8.09273422e-01 1.19324696e+00 -1.59211588e+00
-8.51065338e-01 -3.44620287e-01 -1.40776649e-01 1.19355358e-01
-5.02838731e-01 -2.41878450e-01 -1.61767602e-01 -4.35502052e-01
1.99087158e-01 -8.46465766e-01 -1.98412329e-01 1.44257113e-01
-5.12520671e-01 -3.08177292e-01 8.03255916e-01 -2.30059385e-01
1.21269226e+00 -2.19994116e+00 -1.93364292e-01 5.13296187e-01
8.41740146e-02 5.44937134e-01 2.19180018e-01 -1.12118185e-01
-6.11558035e-02 -2.15185760e-03 -4.92036313e-01 -5.57697654e-01
-5.55077605e-02 2.35159248e-01 -6.26157150e-02 7.54776776e-01
3.17382395e-01 4.34999526e-01 -9.99291778e-01 -6.85120583e-01
3.96344066e-01 4.14316446e-01 -2.58737355e-01 -7.57146776e-02
-1.72375157e-01 4.01892453e-01 -3.58726025e-01 4.76652026e-01
7.74432719e-01 4.33233790e-02 8.01405981e-02 -5.11551835e-02
-5.86912560e-04 3.76375504e-02 -1.72101879e+00 1.31767464e+00
-4.26735550e-01 2.22358271e-01 9.67786759e-02 -1.00100219e+00
1.22905433e+00 1.14367820e-01 2.45327741e-01 -3.23141605e-01
1.86174184e-01 5.30548930e-01 -1.04875773e-01 -2.12612540e-01
3.16322535e-01 -1.96589485e-01 3.27313729e-02 2.11229682e-01
-5.42782433e-02 1.35515168e-01 2.60670215e-01 -2.12768957e-01
6.99034572e-01 4.22574043e-01 3.86839271e-01 -3.28392118e-01
7.62373388e-01 -2.58860916e-01 6.04111254e-01 5.96654296e-01
-2.67212778e-01 5.36904395e-01 3.21910352e-01 -1.58565968e-01
-8.01646590e-01 -8.02082241e-01 -6.20305717e-01 8.50134194e-01
1.25661358e-01 -2.88660377e-01 -7.35159874e-01 -1.26025426e+00
8.15508962e-02 9.46774006e-01 -6.91339254e-01 -2.68461853e-01
-3.24546099e-01 -7.53597260e-01 5.05316079e-01 3.62596154e-01
3.64730090e-01 -9.00418520e-01 -9.03240740e-01 2.12716028e-01
1.63116455e-02 -8.55699778e-01 9.06650573e-02 7.24834204e-01
-8.65969896e-01 -1.02605581e+00 -6.75434291e-01 -4.61705863e-01
9.12923694e-01 6.84266910e-02 6.61193728e-01 2.74798065e-01
-7.23736063e-02 -2.54936337e-01 -5.85722089e-01 -2.27885336e-01
-5.65437496e-01 1.27276614e-01 6.93666413e-02 3.19097728e-01
3.59961480e-01 -3.00268441e-01 -2.94493318e-01 5.45069039e-01
-7.74472535e-01 -1.76197842e-01 6.50055826e-01 1.05539739e+00
6.81284070e-01 2.09983930e-01 7.50935197e-01 -9.67681408e-01
1.49180442e-01 -3.79572242e-01 -6.30458593e-01 1.95260629e-01
-8.27000856e-01 3.38893056e-01 8.39343786e-01 -7.02519953e-01
-9.47227776e-01 3.23353678e-01 8.19723457e-02 -4.60894138e-01
-5.97888887e-01 1.51489630e-01 -3.13571751e-01 -1.22023880e-01
8.60523164e-01 -7.30834827e-02 -7.57630318e-02 -6.18083358e-01
3.77905399e-01 8.25031877e-01 2.24583164e-01 -2.42690638e-01
7.76691675e-01 4.56748039e-01 1.36341438e-01 -5.69330752e-01
-7.90703118e-01 -7.02709913e-01 -9.71084535e-01 -2.11389780e-01
5.11150837e-01 -4.78578776e-01 -1.85926795e-01 -6.34282082e-02
-8.69693339e-01 7.25883543e-02 -5.40483057e-01 4.58033204e-01
-4.62276489e-01 6.04670525e-01 4.95736040e-02 -1.20767438e+00
1.45435318e-01 -1.02339935e+00 1.08864772e+00 -1.61644027e-01
-2.90942758e-01 -7.72554100e-01 -3.01661253e-01 2.46934086e-01
1.33393076e-03 3.80923182e-01 7.43315935e-01 -9.92652535e-01
-8.80430266e-02 -6.59943759e-01 -1.60500154e-01 4.84042466e-01
4.69571091e-02 -1.86200291e-01 -1.31241727e+00 -2.62154311e-01
1.05070569e-01 -2.24879205e-01 9.47261989e-01 2.99323406e-02
1.03948534e+00 -1.04771443e-01 -6.08426809e-01 2.37076864e-01
1.21471965e+00 8.66210684e-02 3.32800269e-01 4.73052531e-01
4.31710124e-01 1.05888689e+00 1.07090127e+00 4.97953266e-01
-5.49271256e-02 9.40528214e-01 7.89739490e-01 -1.06168340e-03
-7.37602338e-02 -2.41699398e-01 6.90506175e-02 2.49630019e-01
2.04641849e-01 -1.60193205e-01 -7.50369132e-01 4.74633545e-01
-1.75909770e+00 -5.96767366e-01 -2.63558507e-01 2.66918492e+00
6.62901640e-01 6.91879451e-01 4.20717597e-02 8.54769945e-01
8.76176178e-01 5.15577495e-02 -3.09682518e-01 -1.76883757e-01
-9.64378119e-02 -1.41891807e-01 6.69087648e-01 3.34964156e-01
-1.37401128e+00 7.30826914e-01 4.61042309e+00 1.15758991e+00
-9.68073487e-01 1.64312869e-01 5.33745766e-01 6.60711452e-02
-1.49288978e-02 7.18490779e-02 -1.19337404e+00 6.96066380e-01
8.20298016e-01 4.34427798e-01 -2.34361328e-02 9.88533258e-01
1.11399129e-01 -4.89233792e-01 -1.10210919e+00 7.89110661e-01
1.04970492e-01 -7.59914815e-01 -1.76211894e-01 9.42599177e-02
3.47619325e-01 -4.83473361e-01 -2.95180385e-03 2.54625320e-01
-2.38948613e-01 -7.87946761e-01 8.92575324e-01 2.98444122e-01
7.95576870e-01 -8.66056919e-01 1.18227983e+00 8.50158572e-01
-1.00908649e+00 -3.23085636e-01 -4.95118856e-01 1.14574417e-01
2.92813475e-03 9.05884206e-01 -1.08626902e+00 6.34950995e-01
4.13710564e-01 4.08568472e-01 -9.30484354e-01 1.07837141e+00
-3.46455097e-01 4.93843317e-01 -5.16565800e-01 -1.44791054e-02
2.10825652e-01 -4.10447195e-02 5.61904073e-01 1.05589354e+00
3.12441081e-01 -4.42944914e-01 9.32992771e-02 8.99889052e-01
3.22078377e-01 3.06964666e-01 -6.03184819e-01 4.02797252e-01
4.56476241e-01 1.32063520e+00 -1.07411098e+00 -1.73353240e-01
4.41077501e-02 7.34549224e-01 4.51036453e-01 9.01118815e-02
-7.16075003e-01 -3.11135739e-01 -3.79446633e-02 4.43136394e-01
3.49728078e-01 2.13633567e-01 -4.64584351e-01 -7.08804011e-01
3.13200921e-01 -3.23774576e-01 3.00675690e-01 -3.06759328e-01
-1.01936388e+00 6.36075795e-01 2.26542205e-01 -1.51171589e+00
-4.13172811e-01 -4.86932933e-01 -2.56818354e-01 7.23402679e-01
-1.45128894e+00 -1.01452255e+00 -9.97495130e-02 1.37707442e-01
2.22861961e-01 -5.71956448e-02 7.61629939e-01 2.89420992e-01
-4.82434064e-01 8.21869433e-01 3.22317481e-02 -1.70554966e-01
7.21316278e-01 -1.23377228e+00 -3.70519608e-01 6.40699804e-01
2.12361202e-01 3.23903620e-01 9.69077528e-01 -4.93466109e-01
-5.66084504e-01 -1.18280637e+00 1.05034304e+00 -3.59326512e-01
3.66127998e-01 -7.74770319e-01 -1.08723187e+00 1.61270916e-01
-5.96792102e-01 2.86001742e-01 4.74769115e-01 -4.42578793e-02
-2.19262257e-01 -2.48975262e-01 -1.52431595e+00 4.09376085e-01
9.05431986e-01 -2.63964802e-01 -4.56579536e-01 2.08430558e-01
3.80553097e-01 5.06831072e-02 -6.36007488e-01 5.92345476e-01
2.76310652e-01 -8.49292040e-01 7.66705334e-01 -4.80119511e-02
-8.43545571e-02 -6.18722439e-01 -5.08754551e-02 -1.18246269e+00
1.23691922e-02 -1.07392691e-01 -1.74546570e-01 1.54847920e+00
4.32208419e-01 -5.55831075e-01 9.48711753e-01 2.98239529e-01
-5.63193597e-02 -9.13965940e-01 -1.36469352e+00 -9.74152923e-01
-1.44848585e-01 -4.98892516e-01 5.24574399e-01 7.71480620e-01
-1.03488043e-01 1.12377718e-01 -2.19273135e-01 2.56246962e-02
7.49178767e-01 -6.50734082e-02 5.17096460e-01 -1.60947251e+00
-2.44563088e-01 -3.47910672e-01 -6.64511025e-01 -7.57905304e-01
3.30470383e-01 -9.19629574e-01 2.50793159e-01 -1.07729542e+00
-2.18516007e-01 -1.05894530e+00 -3.51109713e-01 4.86839712e-01
3.44155096e-02 5.10853529e-01 6.94294497e-02 2.22047880e-01
-4.32693720e-01 5.15872478e-01 9.42113280e-01 -3.56017649e-02
-3.00220966e-01 2.58011669e-01 -3.63942772e-01 8.28423083e-01
7.18580902e-01 -6.63770437e-01 -3.75906676e-01 2.62173086e-01
5.15294783e-02 -4.41215336e-01 3.17043155e-01 -1.03338945e+00
-4.01855400e-03 8.30288976e-02 4.06698644e-01 -5.85366726e-01
3.46773118e-01 -1.18453002e+00 -1.31634939e-02 3.67245406e-01
-4.82897758e-01 -6.69703960e-01 -2.46169046e-02 7.16606021e-01
-4.13501784e-02 -8.05684686e-01 9.29143190e-01 6.12567738e-02
-6.74402297e-01 -2.85476834e-01 -7.81334490e-02 -2.87901491e-01
1.42474294e+00 -7.10379541e-01 1.17506437e-01 1.49706885e-01
-7.93302357e-01 1.97252527e-01 4.33245957e-01 2.69533932e-01
2.53740877e-01 -1.14947963e+00 -4.53600705e-01 3.49142641e-01
5.35091579e-01 1.94865689e-01 7.18624666e-02 8.50995779e-01
9.78481863e-03 4.26909089e-01 8.90070200e-02 -6.74841225e-01
-1.21865296e+00 8.98732781e-01 2.03863412e-01 -2.70477831e-01
-6.38665080e-01 5.83077550e-01 3.42221819e-02 -4.44426745e-01
5.20992517e-01 -4.30311292e-01 -5.66324472e-01 4.76849437e-01
4.00535017e-01 3.71389717e-01 4.18078184e-01 -8.59111309e-01
-4.62958515e-01 3.90737563e-01 -8.31146315e-02 1.38579115e-01
9.06211734e-01 -1.93035036e-01 2.65947789e-01 6.23718679e-01
9.33583498e-01 -8.37730244e-02 -1.36389983e+00 -1.35963067e-01
4.40852851e-01 -4.93919045e-01 8.08730647e-02 -6.81874037e-01
-7.10422814e-01 8.80355299e-01 8.28778207e-01 1.42645657e-01
8.99294198e-01 3.68924141e-02 1.60204008e-01 2.24127084e-01
5.69692373e-01 -1.34958839e+00 -2.62483060e-01 -2.27235705e-02
6.48822963e-01 -1.26686537e+00 1.15834512e-02 -6.76535964e-01
-4.51886207e-01 9.47364748e-01 5.56932211e-01 -3.49776726e-03
3.89544308e-01 7.03405738e-02 -1.65650666e-01 1.43592000e-01
-4.38629240e-01 -2.86751926e-01 4.01633471e-01 6.90962017e-01
1.61015898e-01 1.31506324e-01 -5.46921909e-01 5.64294577e-01
1.31471485e-01 -2.14009508e-01 2.37745121e-01 8.09738636e-01
-5.92495441e-01 -1.03913879e+00 -6.73329234e-01 5.83040416e-01
-2.18650132e-01 2.29914442e-01 -3.49607676e-01 9.11555350e-01
7.41716504e-01 9.55677390e-01 -3.49527821e-02 -2.58948237e-01
2.86749065e-01 3.27679813e-01 2.78696686e-01 -7.28787184e-01
-5.31024694e-01 1.70325562e-02 6.48251101e-02 -2.10771188e-01
-5.07838964e-01 -4.68815863e-01 -1.20329821e+00 2.67854810e-01
-9.87035453e-01 2.57132113e-01 7.54948258e-01 9.59904909e-01
-4.12681736e-02 2.79051781e-01 5.76191127e-01 -9.26741302e-01
-1.01930523e+00 -9.65567112e-01 -7.75929391e-01 3.81895363e-01
2.00752392e-01 -1.36255670e+00 -6.41115487e-01 -3.45595986e-01]
|
[9.125629425048828, 3.8137247562408447]
|
787b23b4-6cf1-429f-9ea4-7c517f3a9a39
|
revisiting-rumour-stance-classification
| null | null |
https://aclanthology.org/2020.rdsm-1.4
|
https://aclanthology.org/2020.rdsm-1.4.pdf
|
Revisiting Rumour Stance Classification: Dealing with Imbalanced Data
|
Correctly classifying stances of replies can be significantly helpful for the automatic detection and classification of online rumours. One major challenge is that there are considerably more non-relevant replies (comments) than informative ones (supports and denies), making the task highly imbalanced. In this paper we revisit the task of rumour stance classification, aiming to improve the performance over the informative minority classes. We experiment with traditional methods for imbalanced data treatment with feature- and BERT-based classifiers. Our models outperform all systems in RumourEval 2017 shared task and rank second in RumourEval 2019.
|
['Carolina Scarton', 'Yue Li']
| null | null | null | null |
rdsm-coling-2020-12
|
['rumour-detection']
|
['natural-language-processing']
|
[-1.66593015e-01 6.83911666e-02 -8.21340919e-01 -3.87109071e-01
-6.67315423e-01 -3.52012724e-01 7.78652608e-01 7.14652359e-01
-2.65060425e-01 1.05647540e+00 6.72940433e-01 -2.52400637e-01
1.55855238e-01 -7.79622316e-01 -3.35406423e-01 -4.89105105e-01
-3.97197269e-02 8.96896839e-01 3.88876289e-01 -8.01976860e-01
8.13679039e-01 -4.32235561e-02 -1.64833999e+00 1.17459524e+00
7.70048440e-01 1.03759956e+00 -7.85979152e-01 4.89223301e-01
-3.14548165e-01 1.57145798e+00 -1.12014413e+00 -6.15864813e-01
-2.19836980e-01 -4.89714265e-01 -1.16024387e+00 5.09486208e-03
4.47699308e-01 -2.82207787e-01 1.67032685e-02 7.96258330e-01
4.51723844e-01 -2.75749505e-01 8.57099116e-01 -1.47000647e+00
-2.11689234e-01 8.84228408e-01 -8.94174278e-01 7.80280471e-01
6.11690700e-01 -3.84042650e-01 1.23749673e+00 -8.74134362e-01
7.63453841e-01 1.39170730e+00 7.45974660e-01 1.10788949e-01
-9.84958112e-01 -6.80058539e-01 2.40663458e-02 8.61264288e-01
-5.34803808e-01 -4.28992003e-01 8.16939592e-01 -7.70471215e-01
6.12775981e-01 6.71332061e-01 5.24922490e-01 1.38187814e+00
2.28195652e-01 8.82592738e-01 1.47067845e+00 -7.61045814e-02
2.52281666e-01 4.11696345e-01 9.47868943e-01 1.07779749e-01
4.82107133e-01 -3.37997198e-01 -8.33571196e-01 -8.43974292e-01
-3.39240253e-01 -1.16935000e-01 -2.90808380e-01 1.45898372e-01
-7.59096980e-01 1.40830851e+00 5.02228200e-01 1.31531179e-01
-7.14538217e-01 -6.60683990e-01 1.03935468e+00 9.69837725e-01
1.37019038e+00 7.35404134e-01 -3.63548428e-01 -6.95547685e-02
-1.15703905e+00 9.31834817e-01 1.43558002e+00 2.03915402e-01
3.96323472e-01 -1.47304446e-01 -1.32952571e-01 1.12635052e+00
-3.05503845e-01 8.25985149e-02 6.70623958e-01 -7.26836622e-01
5.78957856e-01 6.18419588e-01 3.76781225e-01 -1.24960089e+00
-7.12227106e-01 -6.77030861e-01 -7.66726851e-01 9.21711996e-02
3.23670805e-01 -9.26489606e-02 -3.00152183e-01 9.58486736e-01
3.95533800e-01 -7.15973735e-01 -1.82157949e-01 8.31363142e-01
9.53044713e-01 3.51232618e-01 -3.17586035e-01 -6.68779075e-01
1.58062577e+00 -8.86971474e-01 -1.05640233e+00 -3.72740299e-01
5.43181658e-01 -1.10257840e+00 6.21570289e-01 7.37891197e-01
-1.13590407e+00 2.22293183e-01 -1.13526213e+00 1.70381740e-01
-2.52006918e-01 -5.34097075e-01 3.45556825e-01 5.07667661e-01
-4.02291059e-01 6.53813303e-01 2.69476175e-02 5.08949794e-02
5.72162211e-01 -2.46095657e-01 9.39027313e-03 1.09158747e-01
-1.58565319e+00 1.34358335e+00 2.29884863e-01 -3.53292823e-01
-5.65645993e-01 -7.68800080e-01 -3.47484469e-01 -2.77585536e-01
2.79159099e-01 -2.24687755e-01 1.44752073e+00 -7.99406946e-01
-9.87390280e-01 1.26748192e+00 -1.48386508e-01 -1.04983449e+00
1.12352669e+00 -1.62486538e-01 -2.23057225e-01 8.87336433e-02
9.49308798e-02 -4.47411835e-01 9.27855968e-01 -1.02687359e+00
-6.35058463e-01 -5.19810021e-01 -1.49213970e-01 -9.76446143e-04
-6.19905069e-02 6.61635995e-01 8.59761357e-01 -2.37477526e-01
4.48321849e-01 -6.15538180e-01 2.08571598e-01 -7.17827559e-01
-5.40944159e-01 -4.20610785e-01 7.93751836e-01 -8.57026577e-01
1.16722083e+00 -1.41812491e+00 -1.62150130e-01 -1.56361684e-01
8.92224610e-01 2.07733855e-01 5.36010623e-01 6.19951069e-01
-9.89154950e-02 2.00376004e-01 4.27157044e-01 -1.59712866e-01
-2.45858287e-03 7.96796530e-02 -7.14307129e-01 8.36380601e-01
-1.69720888e-01 5.49037635e-01 -8.27787280e-01 -3.85495216e-01
-3.14239770e-01 -1.67793512e-01 -5.40022135e-01 2.19013765e-01
-1.91104233e-01 1.87972844e-01 -1.76985979e-01 6.29811049e-01
8.80686104e-01 -2.46620476e-01 1.81634113e-01 -3.43625955e-02
-1.22902222e-01 1.29884243e+00 -5.13726950e-01 1.51547879e-01
-3.24326843e-01 9.17896569e-01 2.07317770e-01 -1.02696371e+00
1.03210497e+00 2.45838076e-01 3.48983109e-01 -7.04169631e-01
3.28981608e-01 5.14878333e-01 2.67678071e-02 -5.40752709e-01
7.12768853e-01 -4.75924522e-01 -1.99957520e-01 8.84940922e-01
-4.25469667e-01 -1.17010593e-01 1.61665350e-01 5.16399443e-01
8.50953162e-01 -6.68024719e-01 7.11931288e-01 -4.58389848e-01
2.89926201e-01 1.75390422e-01 4.86477733e-01 7.77715385e-01
-4.72997665e-01 5.33977449e-01 1.38772535e+00 -6.13239467e-01
-1.00862443e+00 -4.89847749e-01 -3.91687900e-01 1.59636116e+00
-2.10334748e-01 -4.38030869e-01 -4.27813113e-01 -8.74237776e-01
2.77225673e-01 7.24584043e-01 -9.63661313e-01 3.31417285e-02
-5.55594087e-01 -1.39773929e+00 4.58536029e-01 -1.71857122e-02
2.22352818e-01 -8.43407750e-01 -3.40339988e-01 2.43020654e-01
-1.03065753e+00 -8.42682719e-01 1.31090403e-01 3.70895326e-01
-5.59267342e-01 -1.37651253e+00 -3.81371439e-01 -2.43649617e-01
-2.89220661e-02 5.03251135e-01 1.61314726e+00 3.46840262e-01
1.35085091e-01 -6.22698188e-01 -6.97549224e-01 -7.52566516e-01
-8.08981657e-01 1.42303914e-01 7.20938444e-02 -1.17854364e-01
6.12306476e-01 -6.14910424e-01 -2.84012020e-01 5.17999768e-01
-3.43688667e-01 -2.35441953e-01 -1.00544207e-01 8.62531245e-01
-2.74554044e-01 -3.35988909e-01 1.17796052e+00 -1.44871569e+00
1.07165790e+00 -1.15726221e+00 2.00575665e-01 -5.21697760e-01
-7.51450181e-01 -5.05366564e-01 3.22625369e-01 -2.52201229e-01
-7.23982811e-01 -1.15759301e+00 -3.17222327e-01 4.18868899e-01
3.23285252e-01 5.21239221e-01 4.47939783e-01 2.81408817e-01
1.40833688e+00 -1.01578698e-01 1.16049029e-01 -7.36377060e-01
-1.52934909e-01 1.27881324e+00 5.85766323e-02 -1.94653243e-01
2.13349596e-01 4.43304747e-01 -4.73295122e-01 -8.13339472e-01
-1.66050041e+00 -8.09663832e-01 -2.85566479e-01 -1.90182671e-01
7.09666405e-03 -8.60607922e-01 -7.04052269e-01 6.33600116e-01
-1.22870302e+00 2.05419302e-01 3.03811152e-02 -4.15072367e-02
-5.34906924e-01 2.52696782e-01 -1.16346097e+00 -9.96839345e-01
-5.68317235e-01 -5.95975578e-01 2.42442891e-01 -3.36434126e-01
-8.70073199e-01 -5.69934070e-01 3.32192004e-01 9.95788634e-01
6.32250905e-01 4.04308617e-01 7.34467268e-01 -1.51875329e+00
1.72827125e-01 -5.46357274e-01 -2.32143372e-01 2.54979670e-01
-2.71539867e-01 -1.85694084e-01 -1.08067203e+00 -1.98500499e-01
5.93321562e-01 -9.11256433e-01 1.33681941e+00 2.18126792e-02
6.75607979e-01 -9.49762106e-01 -3.98338288e-02 -1.68378592e-01
6.68833971e-01 -5.79960823e-01 5.35262287e-01 6.75725460e-01
1.41011458e-02 1.03007567e+00 7.13462114e-01 8.67270112e-01
5.10647476e-01 6.02336347e-01 7.11165905e-01 3.03253829e-01
3.02475896e-02 -6.13684319e-02 4.08940494e-01 8.97209585e-01
3.78038771e-02 -7.66005814e-02 -7.47929871e-01 6.81759894e-01
-1.93815053e+00 -1.23754549e+00 -9.25757885e-01 1.85748422e+00
1.28022468e+00 4.66077954e-01 6.20076358e-01 7.12081552e-01
7.37256229e-01 6.86825573e-01 -3.75779644e-02 -8.88576210e-01
-3.08857083e-01 -4.39091325e-01 1.87541902e-01 6.32285893e-01
-1.11315215e+00 2.39758447e-01 6.79956532e+00 3.90531749e-01
-7.41203308e-01 4.88559067e-01 5.26456892e-01 -4.34201688e-01
-6.53603151e-02 -3.28660578e-01 -8.51631522e-01 6.67347491e-01
1.17081678e+00 -2.50822544e-01 -6.28981069e-02 9.27565932e-01
2.52783060e-01 -3.26571375e-01 -8.52368891e-01 5.25934637e-01
4.53814119e-01 -1.40526879e+00 -1.93164378e-01 -2.78973170e-02
9.06886160e-01 4.03480768e-01 -2.08226129e-01 6.03617668e-01
3.67977381e-01 -7.81733215e-01 7.56928086e-01 2.73190551e-02
-8.57134238e-02 -6.91666484e-01 1.17680037e+00 8.35460126e-01
2.32933015e-02 -1.53966054e-01 -5.54854572e-01 -6.46975338e-01
1.01039454e-01 1.25520277e+00 -1.12640870e+00 2.54791498e-01
6.37776911e-01 6.69483483e-01 -4.09069121e-01 9.90082443e-01
-1.11813210e-01 8.34740758e-01 -1.02292724e-01 -2.64031887e-01
5.71930110e-02 4.22655553e-01 8.24168861e-01 1.30743456e+00
-4.90384936e-01 -3.48589689e-01 3.00754905e-01 3.39288741e-01
-2.94091791e-01 3.40333939e-01 -1.80256188e-01 3.42063487e-01
3.46226275e-01 1.06804287e+00 -2.83634365e-01 -7.00250804e-01
1.43645599e-01 4.91785288e-01 4.75220025e-01 -2.50335336e-01
-6.68946564e-01 2.49982905e-02 5.67754567e-01 6.39241755e-01
-1.52854070e-01 4.12339330e-01 -5.43543279e-01 -1.30598593e+00
-1.58173591e-01 -1.41302979e+00 6.62146449e-01 -2.44899228e-01
-1.73675001e+00 5.58993936e-01 -1.86062783e-01 -1.00905275e+00
-4.85596180e-01 -3.10723871e-01 -6.95569932e-01 6.64166629e-01
-1.90571010e+00 -7.99450696e-01 -4.76629347e-01 1.39975384e-01
5.59209645e-01 1.43853039e-01 5.38291693e-01 3.51912260e-01
-2.61951506e-01 2.48464823e-01 1.26835555e-01 -9.25575907e-04
1.19491911e+00 -1.23201084e+00 2.34912157e-01 -8.53681751e-03
-4.99504745e-01 2.57622808e-01 1.57330859e+00 -6.60016954e-01
-4.88141805e-01 -7.40214586e-01 1.41118324e+00 -7.31239200e-01
1.02455318e+00 -3.02163035e-01 -1.35740316e+00 5.49151003e-01
1.42002657e-01 -3.63873184e-01 7.51433134e-01 6.09478176e-01
-9.12249446e-01 4.34349537e-01 -1.21925974e+00 1.34041637e-01
5.17129064e-01 -3.52519572e-01 -9.83216584e-01 1.06364632e+00
3.30996007e-01 -3.51473778e-01 -5.47559023e-01 1.68108851e-01
5.38663983e-01 -1.51374161e+00 7.38922954e-01 -1.25280619e+00
9.11700547e-01 3.44311237e-01 -5.55462092e-02 -1.47681344e+00
-1.01295538e-01 -3.71474922e-01 -4.04565006e-01 8.49989772e-01
3.13536018e-01 -8.54299068e-01 7.77251780e-01 -1.15569241e-01
2.40900978e-01 -6.95080578e-01 -8.04623425e-01 -5.77788234e-01
5.51532328e-01 -1.12113170e-02 1.97852448e-01 1.27063704e+00
6.20024085e-01 5.78394592e-01 -8.78602386e-01 -5.96526027e-01
7.38837004e-01 5.82562208e-01 6.70777261e-01 -1.68472433e+00
-7.50045255e-02 -6.25004947e-01 -2.51921862e-01 -6.39854550e-01
5.24676777e-02 -8.02415371e-01 -1.16761327e-01 -1.27732360e+00
6.75587177e-01 -1.04260169e-01 -7.97635224e-03 2.40065545e-01
-2.26557449e-01 5.66465437e-01 4.71797511e-02 7.49429882e-01
-6.91557169e-01 3.01514566e-01 8.40279698e-01 -9.86837372e-02
7.46390000e-02 5.10932624e-01 -7.75343299e-01 1.03711224e+00
9.79703248e-01 -7.23594487e-01 1.71750292e-01 3.72490048e-01
7.26158500e-01 1.09158695e-01 3.28744382e-01 -3.64391297e-01
-6.16199635e-02 -1.48137808e-01 -3.22729945e-02 -1.03749037e+00
3.06771547e-01 4.44474109e-02 -4.93033946e-01 5.60229421e-01
-6.42883897e-01 -7.39054605e-02 -3.28680873e-01 4.70265538e-01
-2.15895429e-01 -5.25701463e-01 1.06106746e+00 -2.26242155e-01
1.87885836e-01 -2.80711502e-01 -6.77746296e-01 6.68458998e-01
6.26938820e-01 3.71157825e-01 -1.27161503e+00 -8.17519188e-01
-7.38741159e-01 7.64964297e-02 3.97233337e-01 5.33859313e-01
2.92570561e-01 -8.61684620e-01 -1.56853044e+00 -1.63787425e-01
1.48413852e-01 -7.24136233e-01 2.50416875e-01 1.51395237e+00
-3.18390906e-01 3.23165566e-01 -3.27372849e-01 -2.56638467e-01
-1.42076886e+00 3.89141530e-01 2.73507267e-01 -5.24998844e-01
-4.21951324e-01 5.18435180e-01 -1.42978787e-01 -3.49510431e-01
2.59141084e-02 4.01170701e-01 -6.54456496e-01 1.12173867e+00
1.27132332e+00 9.02199805e-01 6.84708238e-01 -5.53498983e-01
-3.40720564e-01 -5.23460209e-01 -7.32826948e-01 2.50619978e-01
1.42767334e+00 -1.87754467e-01 -7.56045580e-01 7.60733128e-01
1.17272198e+00 2.29083747e-01 -4.06607300e-01 -4.53240216e-01
3.54363322e-01 -5.95960200e-01 1.29177481e-01 -8.67785871e-01
-4.76387233e-01 7.88297951e-01 -2.90085018e-01 1.18101287e+00
2.02195197e-01 -3.47431526e-02 7.36651719e-01 3.48016508e-02
2.57757694e-01 -1.11614025e+00 2.23746166e-01 8.16094935e-01
1.26893914e+00 -1.36403692e+00 5.77646792e-01 -4.83311117e-01
-7.02438354e-01 9.21923578e-01 3.05303574e-01 -2.94663250e-01
4.13919777e-01 5.55138439e-02 1.30457506e-01 -4.69528973e-01
-1.23822749e+00 2.57001609e-01 1.53202727e-01 2.88424671e-01
7.39824355e-01 1.70389056e-01 -1.04933727e+00 7.55563259e-01
-5.02928853e-01 -5.02417386e-01 1.18169963e+00 7.42303610e-01
-1.01940048e+00 -5.51723540e-01 -6.18110538e-01 9.82205927e-01
-1.06093872e+00 -6.94437623e-02 -8.23978662e-01 5.47988892e-01
-2.96414405e-01 1.28119862e+00 -1.09841056e-01 -3.49441081e-01
6.79228157e-02 3.22103053e-01 7.74142519e-02 -5.66902220e-01
-9.94802117e-01 -2.83510417e-01 1.01931643e+00 -5.47655284e-01
-1.53369263e-01 -8.66874218e-01 -6.56400800e-01 -1.11010063e+00
-5.50909519e-01 4.49027896e-01 5.97259879e-01 9.08995032e-01
-1.02649638e-02 5.13456047e-01 1.01922822e+00 -7.65048921e-01
-1.46100676e+00 -1.26399767e+00 -5.79994500e-01 6.45190358e-01
7.90357232e-01 -7.37895727e-01 -9.75207269e-01 -4.33360159e-01]
|
[8.227649688720703, 10.116371154785156]
|
78358021-7c86-48d6-baf6-0d47f995a3d6
|
efficient-uncertainty-estimation-in-spiking
|
2304.10191
| null |
https://arxiv.org/abs/2304.10191v1
|
https://arxiv.org/pdf/2304.10191v1.pdf
|
Efficient Uncertainty Estimation in Spiking Neural Networks via MC-dropout
|
Spiking neural networks (SNNs) have gained attention as models of sparse and event-driven communication of biological neurons, and as such have shown increasing promise for energy-efficient applications in neuromorphic hardware. As with classical artificial neural networks (ANNs), predictive uncertainties are important for decision making in high-stakes applications, such as autonomous vehicles, medical diagnosis, and high frequency trading. Yet, discussion of uncertainty estimation in SNNs is limited, and approaches for uncertainty estimation in artificial neural networks (ANNs) are not directly applicable to SNNs. Here, we propose an efficient Monte Carlo(MC)-dropout based approach for uncertainty estimation in SNNs. Our approach exploits the time-step mechanism of SNNs to enable MC-dropout in a computationally efficient manner, without introducing significant overheads during training and inference while demonstrating high accuracy and uncertainty quality.
|
['Sander Bohte', 'Bojian Yin', 'Tao Sun']
|
2023-04-20
| null | null | null | null |
['medical-diagnosis']
|
['medical']
|
[ 4.26125139e-01 -1.11383848e-01 1.62179962e-01 -3.91921818e-01
-7.58929849e-01 -2.86570638e-01 5.34993649e-01 2.31985196e-01
-8.95367563e-01 1.52397060e+00 -2.27154151e-01 -2.01378480e-01
-1.96696699e-01 -8.45982611e-01 -1.24546564e+00 -6.53457880e-01
-3.16361561e-02 4.90793824e-01 4.38442349e-01 6.39672577e-01
1.83564022e-01 6.18331850e-01 -1.48717010e+00 -1.89953491e-01
8.35664690e-01 1.31020331e+00 -1.71167448e-01 4.43519562e-01
1.43969491e-01 5.43149412e-01 -6.06250226e-01 -2.36230731e-01
-2.52497494e-01 -4.99442250e-01 -2.61121064e-01 -6.95719838e-01
-3.04788262e-01 -2.90371239e-01 -5.94646037e-01 1.20380270e+00
5.43958187e-01 -8.00947472e-03 8.87032211e-01 -1.29367900e+00
2.51312827e-04 9.54214454e-01 -4.23177481e-02 2.18532711e-01
-2.97891289e-01 4.98018682e-01 3.18476021e-01 -6.25804365e-01
2.69077897e-01 9.05300617e-01 7.22354233e-01 8.11038136e-01
-1.45683742e+00 -9.96201813e-01 -2.28368640e-01 -2.88349157e-03
-1.29946470e+00 -7.02604592e-01 3.79679650e-01 -1.54363245e-01
1.24078012e+00 -3.33260298e-01 1.03696966e+00 1.46915734e+00
9.68508124e-01 5.79655230e-01 1.07050896e+00 8.00664574e-02
1.37872708e+00 -3.80784392e-01 2.28145681e-02 1.79861143e-01
7.80305207e-01 3.72425050e-01 -8.81612778e-01 -3.86462629e-01
7.77019620e-01 -2.08160728e-02 6.98087513e-02 4.77097183e-02
-9.28841412e-01 5.93480885e-01 3.15399200e-01 -1.88798800e-01
-3.60031635e-01 1.25573206e+00 3.13617140e-01 -2.74679601e-01
1.43455088e-01 2.94590950e-01 -1.52393490e-01 -6.25248313e-01
-1.00216365e+00 3.77678573e-01 7.78853416e-01 8.42970669e-01
1.83612213e-01 6.09922051e-01 -9.51817706e-02 3.99566770e-01
3.42287332e-01 7.64408529e-01 2.69917578e-01 -1.08631968e+00
-1.13052972e-01 2.76328772e-01 -6.15137666e-02 -6.64781407e-02
-3.26155663e-01 -2.81906396e-01 -1.05174649e+00 4.14806038e-01
3.16835463e-01 -2.54696965e-01 -1.16647077e+00 1.87332404e+00
-2.31812716e-01 4.95352775e-01 2.34351549e-02 4.63293970e-01
5.16803265e-01 4.21411157e-01 2.90626496e-01 -1.72287077e-01
1.16126692e+00 1.24833984e-02 -5.75447917e-01 -4.87885982e-01
1.21820338e-01 6.00310080e-02 5.07207632e-01 3.23968768e-01
-1.20403945e+00 3.60951811e-01 -1.24276173e+00 6.14375509e-02
-1.89280137e-01 -5.55691183e-01 8.60175014e-01 7.77497590e-01
-9.11904216e-01 8.33427310e-01 -1.40005255e+00 -6.71465099e-02
1.37882018e+00 7.64122486e-01 1.21972390e-01 1.35447169e-02
-1.02332544e+00 9.22708929e-01 4.40012574e-01 -7.02272262e-03
-1.32507038e+00 -7.01811671e-01 -7.42679298e-01 1.60896540e-01
7.85812512e-02 -5.70914686e-01 1.28676832e+00 -4.07721490e-01
-1.69913268e+00 1.86806604e-01 -1.77157521e-01 -1.13847864e+00
1.83732376e-01 4.57381099e-01 -2.21650884e-01 -2.24483117e-01
-5.25406599e-01 1.01821172e+00 6.42892957e-01 -6.98354959e-01
-1.59741491e-01 -3.08371782e-01 -3.32726449e-01 -2.80299008e-01
-4.48055416e-02 -2.65976578e-01 7.24245906e-02 -3.79293233e-01
1.18170448e-01 -7.95978546e-01 -3.58687341e-01 3.64011765e-01
-2.06643403e-01 3.88885960e-02 4.50891823e-01 -2.36158952e-01
8.25587869e-01 -1.92269278e+00 -1.16679795e-01 2.84525931e-01
-1.72442831e-02 2.33114362e-02 2.78936327e-01 -1.51438922e-01
6.30241454e-01 -4.79527302e-02 -9.77459848e-01 -1.18860148e-01
3.88499908e-02 4.78873104e-01 -1.87064588e-01 4.16034758e-01
6.87556982e-01 1.30443001e+00 -7.93135285e-01 -4.36662078e-01
1.52994350e-01 7.15601802e-01 -5.23151815e-01 -2.76451916e-01
-8.40985060e-01 2.53809571e-01 -3.07353169e-01 7.46340692e-01
3.16406339e-01 -4.19305295e-01 4.46155667e-02 1.25893578e-01
1.44381136e-01 4.91919458e-01 -9.72185433e-01 1.59048760e+00
-3.89516920e-01 7.25221336e-01 -2.85652637e-01 -6.80746377e-01
7.30285823e-01 1.82431769e-02 9.07847583e-02 -7.70184815e-01
5.28278232e-01 5.46125054e-01 3.44527602e-01 3.29849273e-01
2.34184265e-01 -3.98538522e-02 -1.36654437e-01 4.51080054e-01
4.27695662e-01 -5.54295897e-01 -3.63165885e-02 6.75124675e-02
1.41206336e+00 1.91226453e-01 1.38364643e-01 -3.86816204e-01
-3.93554658e-01 -1.66048989e-01 6.45395517e-01 7.49244452e-01
-2.10900277e-01 4.36493456e-01 5.69984138e-01 1.05762988e-01
-1.14546764e+00 -1.32138109e+00 -5.90048671e-01 -1.19256817e-01
-4.08233181e-02 2.03794226e-01 -9.92288232e-01 1.47081137e-01
1.76842272e-01 9.39431071e-01 -4.00587797e-01 -4.89549756e-01
-1.38633713e-01 -1.05311155e+00 7.77746856e-01 7.12942064e-01
3.69531006e-01 -1.12563288e+00 -1.08035076e+00 6.30690753e-01
4.49382186e-01 -1.06654990e+00 1.21627651e-01 1.00524378e+00
-1.24517012e+00 -5.80386043e-01 -4.33443755e-01 -1.93457812e-01
6.01894021e-01 -7.54485011e-01 9.38878119e-01 -4.75601554e-01
-7.33408868e-01 1.85006782e-01 3.48387927e-01 -8.03843081e-01
-2.93484211e-01 -6.82527944e-02 2.47844040e-01 -4.56226051e-01
4.21900839e-01 -1.01582861e+00 -6.53481066e-01 -9.64974090e-02
-9.82091129e-01 -2.12745756e-01 5.88394582e-01 1.03271234e+00
9.80908453e-01 3.25057991e-02 8.11115146e-01 -6.38464808e-01
4.40770835e-01 -5.76301813e-01 -1.23713529e+00 -3.97857055e-02
-9.15530324e-01 4.57887769e-01 5.65820694e-01 -5.77426732e-01
-7.89180875e-01 6.18890859e-02 -1.88476965e-01 -1.40890390e-01
4.39679511e-02 2.62427002e-01 -5.24808234e-03 -4.61975366e-01
7.20577359e-01 8.43448639e-02 -3.68531346e-02 1.11305878e-01
-2.28786469e-01 3.46901774e-01 5.21028459e-01 -6.00313663e-01
1.04260661e-01 4.17309612e-01 5.08729637e-01 -4.86938506e-01
-1.46496013e-01 5.15790403e-01 2.08682135e-01 -1.60162538e-01
5.02556205e-01 -8.87920141e-01 -1.21031177e+00 7.07755804e-01
-1.12459457e+00 -4.41364825e-01 -6.84831321e-01 6.55963004e-01
-7.64566839e-01 -3.34110111e-01 -6.32574737e-01 -1.24212909e+00
-2.99268901e-01 -1.10830092e+00 6.67417109e-01 8.66541326e-01
-4.03080851e-01 -7.35448480e-01 -3.02168339e-01 -2.96118885e-01
7.72739291e-01 7.75494799e-02 1.11252117e+00 -6.41292036e-01
-1.06307149e+00 -2.68959582e-01 -1.37190983e-01 6.46139234e-02
-3.79409969e-01 3.44226472e-02 -1.31667697e+00 -2.11890694e-02
-3.63941640e-02 -6.06218219e-01 9.77232456e-01 1.03862572e+00
1.47145605e+00 -4.20075469e-02 -4.35597330e-01 4.04920995e-01
1.49486530e+00 3.83455545e-01 7.55107880e-01 -1.91767558e-01
5.20166829e-02 1.04703426e-01 -1.38287589e-01 7.93998241e-01
6.62444308e-02 3.11831869e-02 7.81726718e-01 7.64577866e-01
1.82059214e-01 -2.83105344e-01 2.62752265e-01 5.58260858e-01
4.79518205e-01 -4.80410010e-01 -7.60374904e-01 6.70557439e-01
-1.70302784e+00 -8.35808218e-01 1.38210699e-01 2.59604788e+00
1.10880804e+00 5.54387510e-01 -2.29129285e-01 7.98714533e-02
5.82823038e-01 -2.72164881e-01 -1.61947596e+00 -3.28043044e-01
-4.28995073e-01 5.92334092e-01 7.63131738e-01 7.51447603e-02
-4.56181258e-01 5.10532737e-01 6.64631844e+00 9.97928143e-01
-7.78210580e-01 1.03280582e-01 1.00306726e+00 -5.99923611e-01
-7.43757665e-01 6.07181117e-02 -1.04013658e+00 8.89792621e-01
1.56041074e+00 -4.13140059e-01 8.01590562e-01 5.24012625e-01
-9.67963338e-02 -8.53688598e-01 -1.30119205e+00 1.10103786e+00
-5.98743141e-01 -1.77821922e+00 -1.62641868e-01 -1.92891452e-02
7.76105583e-01 2.45415568e-01 2.59685010e-01 4.96948995e-02
6.71304941e-01 -1.39185679e+00 5.32901645e-01 6.25328422e-01
8.28234553e-01 -1.01308596e+00 6.61948383e-01 1.64108828e-01
-5.22851646e-01 7.09502771e-02 -5.72292447e-01 4.34949249e-02
2.53118187e-01 1.32927656e+00 -3.38969439e-01 -1.89890146e-01
7.43389785e-01 1.39493510e-01 4.73896079e-02 1.32491875e+00
1.68175742e-01 8.73326540e-01 -1.00784016e+00 -8.37071776e-01
-7.71418139e-02 3.18799051e-03 4.16550994e-01 6.97784066e-01
5.71129918e-01 1.14426255e-01 -8.71097565e-01 1.62024117e+00
-5.77155590e-01 -8.22015285e-01 -5.24462223e-01 -6.57807171e-01
1.08926964e+00 8.77144694e-01 -9.01896477e-01 2.68241242e-02
5.13374247e-02 5.35137355e-01 1.68281913e-01 2.37107113e-01
-7.48545110e-01 -4.23798323e-01 6.17804706e-01 6.81169378e-03
1.52958363e-01 -2.22402260e-01 -7.97589064e-01 -6.82919383e-01
8.38512182e-03 -5.00034571e-01 -1.74115822e-01 -6.23701751e-01
-1.06754243e+00 3.65975499e-01 -1.16581462e-01 -7.28697956e-01
-6.29715383e-01 -7.00111985e-01 -4.36991513e-01 7.86101222e-01
-1.25438702e+00 -3.53638351e-01 7.32050091e-02 2.49384180e-01
1.27904460e-01 -2.41334829e-02 9.36120510e-01 -7.16251731e-02
-5.44519663e-01 4.91822600e-01 5.63889682e-01 -1.69778839e-01
1.26502633e-01 -7.45272577e-01 5.60572684e-01 7.61050880e-01
-1.63841352e-01 3.70105714e-01 7.11773753e-01 -7.25413144e-01
-1.68636465e+00 -1.08708477e+00 4.45245296e-01 -1.29370123e-01
5.83158374e-01 -5.02754927e-01 -7.88174689e-01 3.16862315e-01
-1.64025366e-01 2.40793884e-01 4.25866276e-01 -2.80240238e-01
-2.15397015e-01 -1.11956082e-01 -1.57250977e+00 7.92493880e-01
1.02189076e+00 -5.83930016e-01 -8.20383057e-02 -5.09889536e-02
5.31034231e-01 -3.77451271e-01 -7.82504201e-01 5.50075114e-01
6.78541780e-01 -6.95989072e-01 6.25702620e-01 -1.88261732e-01
3.86911035e-01 -1.28533974e-01 -3.07233185e-01 -1.27894020e+00
2.71188736e-01 -6.37862325e-01 -6.81038439e-01 1.05052376e+00
5.20257771e-01 -9.01181281e-01 1.00916660e+00 1.05733907e+00
-1.98720582e-02 -7.30757177e-01 -1.71622252e+00 -9.06338394e-01
1.34153128e-01 -6.25349522e-01 4.41131860e-01 -4.57239002e-02
-1.32304356e-01 -2.55074292e-01 1.25096157e-01 4.10766713e-02
1.17339528e+00 -5.14043570e-01 -1.49348348e-01 -1.28875208e+00
-2.83640504e-01 -6.39124095e-01 -7.31135964e-01 -4.39577490e-01
2.05949396e-01 -5.25411069e-01 5.04607201e-01 -1.31106853e+00
3.27968538e-01 -4.27887946e-01 -4.90264058e-01 2.38276199e-01
7.14273676e-02 2.71368027e-01 -4.16195899e-01 -1.79334104e-01
-3.71494710e-01 7.66658545e-01 2.68456310e-01 -1.88499302e-01
3.90772045e-01 -6.51996257e-03 -2.60817945e-01 5.90471029e-01
8.08718085e-01 -9.67200279e-01 -4.86080796e-01 -3.43457848e-01
3.76935840e-01 1.79203171e-02 6.69181168e-01 -1.51567090e+00
6.83250487e-01 -8.26525241e-02 6.67144418e-01 -3.83224487e-01
5.92575192e-01 -7.30035305e-01 6.00570917e-01 7.26808786e-01
-2.03893363e-01 -2.56292164e-01 5.98801911e-01 9.18664098e-01
2.64733862e-02 -5.26786149e-01 9.90854084e-01 2.19242740e-02
-2.97189653e-01 5.54833710e-01 -8.42038393e-01 3.74827802e-01
9.40382421e-01 -2.23161072e-01 -4.61154312e-01 -2.48651147e-01
-1.38353914e-01 6.94191232e-02 5.49595654e-01 -1.89197168e-01
8.80798697e-01 -1.11779153e+00 -2.23546326e-01 4.04471196e-02
3.51384170e-02 4.47416186e-01 3.64169121e-01 4.22367871e-01
-2.08265752e-01 2.81820923e-01 -2.74440646e-01 -6.82903051e-01
-3.08416307e-01 -1.21637315e-01 4.38141972e-01 9.74976346e-02
-7.66999125e-02 1.10952497e+00 -2.99884558e-01 -4.54439856e-02
4.49776173e-01 -4.94520605e-01 4.91366714e-01 -4.93625224e-01
3.60412300e-01 3.49352062e-01 1.45102620e-01 4.15673673e-01
-6.22130215e-01 -6.84093609e-02 5.50163835e-02 -6.75209105e-01
1.35774279e+00 3.51586759e-01 -3.34501974e-02 9.56189454e-01
7.28938878e-01 -8.39768946e-01 -1.68128538e+00 -1.38095677e-01
2.00298369e-01 2.04260275e-01 3.23289633e-01 -8.86070132e-01
-7.32767463e-01 1.08491790e+00 6.56830251e-01 -4.69445556e-01
9.25188303e-01 -1.57205552e-01 6.82044625e-01 5.41669726e-01
8.98193359e-01 -1.17020690e+00 -2.85188854e-01 4.33837265e-01
3.12599391e-01 -1.12047362e+00 8.31061881e-03 3.36282015e-01
-2.25967079e-01 8.90110373e-01 4.52507943e-01 -1.40717059e-01
7.98280239e-01 1.19089961e+00 -7.80108154e-01 2.38890007e-01
-1.26472461e+00 2.89388329e-01 -1.93129927e-01 6.80303991e-01
-1.02251293e-02 1.24911249e-01 7.63907433e-02 8.33173633e-01
3.88163388e-01 5.85235238e-01 6.36680841e-01 9.98774171e-01
-4.47851777e-01 -7.37030506e-01 1.46401107e-01 1.26528156e+00
-4.19558495e-01 -5.14698207e-01 1.46072745e-01 2.18683407e-01
-3.28056157e-01 6.31332219e-01 4.19003427e-01 -2.25086898e-01
-6.16336130e-02 2.53111899e-01 8.12595546e-01 -3.12257856e-01
-3.80624682e-01 -4.23211336e-01 8.86642784e-02 -4.45380032e-01
2.87103821e-02 -6.79446995e-01 -1.59497631e+00 -4.87863928e-01
-3.17100734e-01 -7.60076120e-02 1.28945947e+00 9.81353939e-01
7.31972337e-01 7.40284979e-01 2.11712256e-01 -8.37460577e-01
-7.83994615e-01 -6.19981647e-01 -7.11995900e-01 -4.65463132e-01
1.46802604e-01 -7.50080466e-01 -4.50544864e-01 -3.82725626e-01]
|
[8.207540512084961, 2.4647674560546875]
|
a987a948-5486-48a8-aa44-2c9f11a404ad
|
graph-exploration-for-effective-multi-agent-q
|
2304.09547
| null |
https://arxiv.org/abs/2304.09547v1
|
https://arxiv.org/pdf/2304.09547v1.pdf
|
Graph Exploration for Effective Multi-agent Q-Learning
|
This paper proposes an exploration technique for multi-agent reinforcement learning (MARL) with graph-based communication among agents. We assume the individual rewards received by the agents are independent of the actions by the other agents, while their policies are coupled. In the proposed framework, neighbouring agents collaborate to estimate the uncertainty about the state-action space in order to execute more efficient explorative behaviour. Different from existing works, the proposed algorithm does not require counting mechanisms and can be applied to continuous-state environments without requiring complex conversion techniques. Moreover, the proposed scheme allows agents to communicate in a fully decentralized manner with minimal information exchange. And for continuous-state scenarios, each agent needs to exchange only a single parameter vector. The performance of the algorithm is verified with theoretical results for discrete-state scenarios and with experiments for continuous ones.
|
['Ali H. Sayed', 'Ainur Zhaikhan']
|
2023-04-19
| null | null | null | null |
['q-learning']
|
['methodology']
|
[-2.44367212e-01 3.76693964e-01 -2.87642390e-01 1.61686748e-01
-1.59474149e-01 -5.45616150e-01 8.18827808e-01 5.48851728e-01
-9.11059618e-01 1.42473626e+00 -4.16486859e-01 -1.46578565e-01
-4.91529167e-01 -9.97548878e-01 -4.33547795e-01 -9.42304671e-01
-5.83276749e-01 8.19811165e-01 3.24679762e-01 -2.44080871e-01
1.42738625e-01 4.15378332e-01 -1.22049510e+00 -6.07706189e-01
7.94024348e-01 6.48091316e-01 3.76701593e-01 7.44729578e-01
1.63765624e-01 9.80960608e-01 -8.31483781e-01 3.10396224e-01
4.27188873e-01 -5.38275003e-01 -5.66547513e-01 3.62328082e-01
-7.06080198e-01 -5.53929865e-01 -2.21524209e-01 1.14147663e+00
4.01123405e-01 3.44470531e-01 4.13823634e-01 -1.39705253e+00
1.60005286e-01 8.38423193e-01 -7.07290292e-01 -3.28043103e-02
1.27710491e-01 2.36781433e-01 6.24996066e-01 -6.24675415e-02
6.09747946e-01 1.27094185e+00 -4.82626408e-02 4.17400599e-01
-1.13790178e+00 -5.01930237e-01 6.15705669e-01 2.22597599e-01
-1.06581259e+00 -1.22686297e-01 6.92295134e-01 1.22985676e-01
8.24561238e-01 1.22296631e-01 1.01509297e+00 3.92454594e-01
4.43237215e-01 5.01807034e-01 1.62286377e+00 -4.30639714e-01
8.88847828e-01 1.65427521e-01 -3.51421684e-01 7.09878683e-01
7.59421825e-01 3.04627120e-01 -2.37753093e-01 -3.50101382e-01
8.16542447e-01 -5.30289579e-03 6.22338690e-02 -7.89887190e-01
-1.13376367e+00 8.67403686e-01 1.58104971e-01 2.85788834e-01
-8.53431523e-01 4.98093843e-01 2.38194942e-01 8.60592186e-01
1.48097515e-01 2.49514893e-01 -7.18336925e-02 -1.27883866e-01
-3.04057807e-01 2.92564243e-01 1.01388526e+00 8.43729615e-01
8.19121957e-01 3.90038371e-01 3.51795793e-01 3.00159186e-01
6.00931168e-01 6.36831284e-01 1.38517514e-01 -1.11182547e+00
5.17348468e-01 4.60596889e-01 6.54183447e-01 -7.35202372e-01
-6.26391292e-01 -2.85281688e-01 -6.91680253e-01 8.48128796e-01
4.99998897e-01 -7.77848125e-01 -2.42671341e-01 1.54254997e+00
5.98133743e-01 -1.48906037e-01 4.32381243e-01 6.98845029e-01
-1.76292628e-01 6.24960721e-01 -1.31504133e-01 -8.46754789e-01
9.67974186e-01 -7.65900850e-01 -8.01531672e-01 -2.19543293e-01
3.87278080e-01 -3.07021230e-01 3.82021725e-01 3.82610530e-01
-1.28166664e+00 1.73518304e-02 -1.02460861e+00 9.97507274e-01
-1.95727289e-01 -7.62348771e-02 5.56806087e-01 5.60240865e-01
-1.08530188e+00 4.99752134e-01 -1.07307255e+00 -3.51835132e-01
-7.57373348e-02 6.36141181e-01 9.40504745e-02 4.70651180e-01
-1.03357410e+00 9.59637940e-01 8.11946630e-01 1.58099793e-02
-1.11389756e+00 1.09459586e-01 -5.62755346e-01 -1.66945364e-02
9.37084138e-01 -5.60437143e-01 1.32604444e+00 -8.00021648e-01
-2.10046673e+00 -2.81373650e-01 2.66128719e-01 -6.49785757e-01
7.82823145e-01 2.47791335e-01 3.10168304e-02 3.16614389e-01
-1.77954450e-01 1.68922767e-01 6.23013437e-01 -1.36224663e+00
-8.29871297e-01 -1.68353036e-01 5.36825299e-01 7.55745351e-01
-3.20311159e-01 -3.44622463e-01 3.05110253e-02 -2.01287672e-01
-2.06428245e-01 -1.14386129e+00 -9.05355334e-01 -2.98704714e-01
-4.05511707e-02 -1.13878734e-01 7.40672350e-01 1.85105890e-01
8.41721892e-01 -1.60777175e+00 2.83150375e-01 5.87782681e-01
1.23858541e-01 -1.43810809e-02 -1.00212961e-01 1.24707353e+00
6.88776791e-01 -2.65715599e-01 3.44674364e-02 -2.19243929e-01
1.66590333e-01 4.33709234e-01 1.43484697e-01 6.54701948e-01
-2.66649693e-01 4.90995616e-01 -1.20486796e+00 -5.45494139e-01
4.16624695e-01 7.64443874e-02 -2.99251825e-01 1.05771236e-01
-3.08349431e-01 5.59861243e-01 -1.16386557e+00 1.16025649e-01
4.02486145e-01 -2.61122137e-01 9.03371632e-01 6.43615544e-01
-4.12399471e-01 3.54953371e-02 -1.71667075e+00 1.15848243e+00
-6.88727617e-01 1.04759321e-01 6.62977576e-01 -1.06144893e+00
9.04735327e-01 4.66818303e-01 8.22301269e-01 -5.61180413e-01
2.89374560e-01 1.20529130e-01 2.10800856e-01 4.53727618e-02
4.65160787e-01 2.34875008e-01 -2.52141148e-01 9.75288510e-01
-1.60492763e-01 -2.19315007e-01 6.23899579e-01 2.26068065e-01
1.08816898e+00 -1.95586830e-01 8.63297820e-01 -1.72686458e-01
4.51141030e-01 7.20548481e-02 6.54954314e-01 1.02896941e+00
-1.86169863e-01 -8.29990804e-01 4.58373100e-01 -3.10334116e-01
-7.51969755e-01 -7.91374683e-01 4.60241288e-01 6.42104745e-01
7.48487711e-01 -1.91694573e-01 -4.90744889e-01 -5.66972315e-01
2.66976133e-02 4.71236765e-01 -5.08869231e-01 2.29944199e-01
-4.65942383e-01 -6.18978441e-01 -6.09166436e-02 4.46645729e-02
7.68200636e-01 -1.06008685e+00 -1.32352865e+00 6.98681772e-01
3.07486922e-01 -7.86287129e-01 -1.69943303e-01 1.01089664e-01
-8.24762821e-01 -1.30809426e+00 -4.22904640e-01 -4.55072016e-01
8.63224983e-01 4.96017747e-02 5.08822262e-01 -6.08480535e-02
7.05000088e-02 7.99468160e-01 -1.83179021e-01 -3.67359102e-01
-6.80944979e-01 -1.18062973e-01 1.88199043e-01 -3.12688947e-02
-2.07766250e-01 -5.00527084e-01 -6.37174189e-01 3.23002905e-01
-7.75376856e-01 -1.44101912e-02 7.07337916e-01 9.10866261e-01
4.51287061e-01 5.17379582e-01 9.70141470e-01 -5.95529139e-01
1.06324792e+00 -3.50172430e-01 -1.31602073e+00 3.00687850e-01
-7.77554870e-01 2.15421289e-01 9.52126026e-01 -5.27542830e-01
-1.17210066e+00 1.32870242e-01 6.86907947e-01 2.43629180e-02
-9.49685574e-02 2.93247163e-01 1.05382852e-01 -3.40608358e-01
1.64721861e-01 3.11484337e-01 2.83094555e-01 -1.30009666e-01
4.37939346e-01 4.25117314e-01 -6.69775158e-02 -6.66781664e-01
7.32357502e-01 3.33838075e-01 4.78489071e-01 -6.98147297e-01
2.01232389e-01 -4.73736525e-02 -3.80007029e-01 -5.65295339e-01
3.21926475e-01 -7.17295408e-01 -1.56134343e+00 4.54069674e-01
-8.87873292e-01 -4.11311001e-01 -4.05694246e-01 7.70808518e-01
-8.37559700e-01 3.95962685e-01 -5.59101701e-01 -1.50810683e+00
-1.97993711e-01 -1.06980765e+00 3.11384648e-01 5.45206785e-01
1.91608056e-01 -1.17305303e+00 3.12239170e-01 -1.98502824e-01
5.51322460e-01 1.73664466e-01 4.16476250e-01 -6.64180875e-01
-8.91661108e-01 1.91784892e-02 3.41194719e-01 -2.98297465e-01
3.36493075e-01 -4.45641488e-01 -2.44592384e-01 -8.38638604e-01
-9.40849483e-02 -4.79185075e-01 2.03469515e-01 3.47131521e-01
3.52132499e-01 -8.26277554e-01 -4.83344108e-01 -2.88487345e-01
1.42299950e+00 7.52821147e-01 -5.46450075e-03 5.14035702e-01
3.94594632e-02 4.33887720e-01 9.21825469e-01 1.14445198e+00
5.85999131e-01 4.94736165e-01 9.07695711e-01 8.16098154e-02
4.61166471e-01 5.81739843e-02 5.44478655e-01 7.44542122e-01
-1.68619648e-01 -3.58766884e-01 -4.60610092e-01 4.91521358e-01
-2.18388176e+00 -9.96675074e-01 3.20946902e-01 2.26676083e+00
7.19436347e-01 3.98369171e-02 2.92801470e-01 -1.98600292e-02
6.64312482e-01 3.11781168e-01 -8.90083194e-01 -5.85783720e-01
9.33294073e-02 -2.22602338e-01 8.29113066e-01 8.20998013e-01
-7.53457665e-01 6.59341693e-01 6.28881550e+00 4.97057080e-01
-7.22596705e-01 1.88772138e-02 3.56601447e-01 -2.65569478e-01
-8.42070282e-02 2.61145264e-01 -4.35231298e-01 3.41333568e-01
9.27265942e-01 -6.36670768e-01 8.55388939e-01 6.64375186e-01
6.73579633e-01 -7.32764602e-01 -6.15969718e-01 4.47584271e-01
-4.89380807e-01 -1.16671038e+00 -4.03067499e-01 2.44364545e-01
8.24259579e-01 -3.61418724e-03 -2.79585034e-01 9.28406343e-02
1.04174387e+00 -4.61277932e-01 4.00686562e-01 4.26388621e-01
2.19630912e-01 -1.26974761e+00 6.47038758e-01 7.64827192e-01
-1.36531556e+00 -5.67668021e-01 -2.75243908e-01 -4.78231668e-01
2.25986853e-01 2.15580195e-01 -1.06117451e+00 8.17617774e-01
2.55419165e-01 2.29693398e-01 1.09114563e-02 9.57534075e-01
-2.13178918e-01 3.30251038e-01 -4.59088534e-01 -9.16288435e-01
5.10077477e-01 -6.63934112e-01 7.47442722e-01 6.06146991e-01
1.98524311e-01 7.19895214e-02 8.75845373e-01 4.86575663e-01
2.74214566e-01 1.12921074e-01 -6.25104725e-01 -1.63255379e-01
7.70641565e-01 1.15944517e+00 -8.48895073e-01 -4.55361962e-01
-2.59451687e-01 6.31386638e-01 2.76720583e-01 5.35168648e-01
-5.89052856e-01 -4.93530184e-01 4.62308496e-01 -3.49291205e-01
1.91446558e-01 -4.98766482e-01 2.51422018e-01 -8.40224504e-01
-1.98351577e-01 -6.81295037e-01 3.71988356e-01 -5.08248433e-02
-8.46096098e-01 4.74326760e-01 2.38881633e-01 -1.26698482e+00
-9.33989286e-01 6.70055673e-02 -5.85675299e-01 5.15976131e-01
-1.53450155e+00 -6.68250442e-01 2.25726575e-01 7.39824712e-01
4.00919616e-01 -4.38104838e-01 7.36778557e-01 -3.45776945e-01
-4.13320154e-01 -8.12157914e-02 5.14213920e-01 -1.81697965e-01
1.19896799e-01 -1.33945489e+00 -1.75934032e-01 6.14819407e-01
-8.22251067e-02 2.18609452e-01 8.99096191e-01 -7.34106123e-01
-1.68243384e+00 -8.86817634e-01 1.90474391e-01 5.77205956e-01
9.53092456e-01 -2.57301051e-02 -3.39625716e-01 3.61833960e-01
7.74805486e-01 -1.49196252e-01 2.23690033e-01 -2.91832894e-01
4.63585556e-01 -1.31966963e-01 -1.19985843e+00 7.00370669e-01
5.33749700e-01 1.59576570e-03 -1.48698181e-01 2.23553181e-01
2.16481149e-01 -2.58200258e-01 -8.89558196e-01 -2.99563669e-02
1.63993105e-01 -6.10684931e-01 4.06010956e-01 -3.48360211e-01
-5.52356958e-01 -5.28593838e-01 2.59917855e-01 -1.90231383e+00
-9.45273191e-02 -1.03307962e+00 -2.66771495e-01 7.70830452e-01
2.26485267e-01 -1.21977222e+00 7.26952195e-01 2.48114824e-01
3.44021589e-01 -4.74415779e-01 -1.44444835e+00 -9.64576721e-01
-1.02242388e-01 2.02787578e-01 5.55881619e-01 5.42074203e-01
5.08420110e-01 4.39314730e-02 -6.43803596e-01 4.74897265e-01
1.06261098e+00 3.13004881e-01 8.64073515e-01 -9.19708252e-01
-8.23891997e-01 -2.56226510e-01 -9.54303890e-02 -7.37537980e-01
4.36285138e-02 -3.55760604e-01 -8.26939866e-02 -1.68412173e+00
1.23658450e-02 -5.71187198e-01 -3.09808880e-01 5.22496164e-01
2.96561897e-01 -3.74284089e-01 5.22824645e-01 9.24079642e-02
-9.78052437e-01 8.41662467e-01 1.29459798e+00 -9.87343937e-02
-5.04097760e-01 3.05791408e-01 -1.32980824e-01 6.54594004e-01
1.23248231e+00 -4.36537236e-01 -9.24304187e-01 -1.03899930e-03
-1.17598539e-02 8.20088744e-01 1.02028258e-01 -9.35104072e-01
4.16493565e-01 -7.84661233e-01 -1.04184255e-01 -3.88068646e-01
4.11753684e-01 -1.07836294e+00 3.20621401e-01 1.13338411e+00
-3.53928685e-01 2.91895002e-01 -1.18227243e-01 1.02989423e+00
-2.11063717e-02 -2.74349004e-01 7.01673210e-01 -1.55172780e-01
-4.24818933e-01 6.80788755e-02 -8.96994889e-01 -3.00452620e-01
1.60077643e+00 1.10098980e-01 -3.62037510e-01 -8.61817241e-01
-6.73400521e-01 7.29762197e-01 4.94282782e-01 1.28606921e-02
4.14568692e-01 -1.03977990e+00 -4.43585604e-01 -1.96876869e-01
-3.35881233e-01 -3.07345003e-01 3.08280960e-02 6.81665123e-01
-1.29336208e-01 3.79028738e-01 -4.22837973e-01 -2.20986798e-01
-1.10581934e+00 5.54233074e-01 4.01830375e-01 -6.65848792e-01
-5.06565213e-01 -1.36908159e-01 -3.15245807e-01 -1.90715864e-01
2.04816565e-01 -1.66715100e-01 -2.22522691e-01 -8.14703405e-02
4.44250554e-01 5.67855120e-01 -4.27326798e-01 -2.22897932e-01
-1.85772181e-01 1.97726905e-01 -5.93721382e-02 -6.29467130e-01
1.33983397e+00 -4.10343617e-01 -1.53380446e-02 5.42508602e-01
4.69765157e-01 9.60695650e-03 -1.41819024e+00 -3.31156433e-01
-1.18405759e-01 -4.52150494e-01 2.45560631e-02 -6.37530208e-01
-1.06952703e+00 2.59775400e-01 3.82629156e-01 5.19831002e-01
1.01534867e+00 -2.80532032e-01 2.39807963e-02 7.34008372e-01
1.12305498e+00 -1.52484977e+00 4.52346876e-02 3.18292767e-01
5.72619677e-01 -9.22091007e-01 1.23999409e-01 1.00430390e-02
-7.85587013e-01 1.04112673e+00 6.06601000e-01 -1.26475453e-01
4.29139644e-01 4.48569804e-01 -1.49791554e-01 5.23461848e-02
-1.20720327e+00 -3.23883712e-01 -5.08301437e-01 6.78842604e-01
-2.14067981e-01 2.33196586e-01 -6.72858059e-01 -2.89501637e-01
3.53668630e-01 -1.18400022e-01 1.06916201e+00 1.23397839e+00
-7.53317595e-01 -1.42655206e+00 -5.20562887e-01 2.75253952e-01
-1.28529789e-02 5.16208351e-01 -1.57864451e-01 1.01825535e+00
-4.93805587e-01 1.34027779e+00 3.21785366e-04 9.91647318e-02
1.74739808e-01 -3.90023977e-01 3.72554928e-01 -2.45627210e-01
-5.08415222e-01 3.17475230e-01 2.20278978e-01 -6.64259195e-01
-5.79363465e-01 -7.23272204e-01 -1.54579461e+00 -1.93955392e-01
-2.46569067e-01 7.47939765e-01 6.30055189e-01 7.20596194e-01
3.12831700e-01 4.88678575e-01 1.23727548e+00 -6.37331486e-01
-1.03090179e+00 -7.73421526e-01 -7.47827232e-01 -2.74674058e-01
2.76809126e-01 -7.89405644e-01 -2.00489387e-01 -6.51737154e-01]
|
[3.9602549076080322, 2.0619235038757324]
|
4fe0c60d-d795-4d3f-ba16-b093a8157e95
|
harnessing-the-power-of-infinitely-wide-deep-1
|
1910.01663
| null |
https://arxiv.org/abs/1910.01663v3
|
https://arxiv.org/pdf/1910.01663v3.pdf
|
Harnessing the Power of Infinitely Wide Deep Nets on Small-data Tasks
|
Recent research shows that the following two models are equivalent: (a) infinitely wide neural networks (NNs) trained under l2 loss by gradient descent with infinitesimally small learning rate (b) kernel regression with respect to so-called Neural Tangent Kernels (NTKs) (Jacot et al., 2018). An efficient algorithm to compute the NTK, as well as its convolutional counterparts, appears in Arora et al. (2019a), which allowed studying performance of infinitely wide nets on datasets like CIFAR-10. However, super-quadratic running time of kernel methods makes them best suited for small-data tasks. We report results suggesting neural tangent kernels perform strongly on low-data tasks. 1. On a standard testbed of classification/regression tasks from the UCI database, NTK SVM beats the previous gold standard, Random Forests (RF), and also the corresponding finite nets. 2. On CIFAR-10 with 10 - 640 training samples, Convolutional NTK consistently beats ResNet-34 by 1% - 3%. 3. On VOC07 testbed for few-shot image classification tasks on ImageNet with transfer learning (Goyal et al., 2019), replacing the linear SVM currently used with a Convolutional NTK SVM consistently improves performance. 4. Comparing the performance of NTK with the finite-width net it was derived from, NTK behavior starts at lower net widths than suggested by theoretical analysis(Arora et al., 2019a). NTK's efficacy may trace to lower variance of output.
|
['Dingli Yu', 'Ruslan Salakhutdinov', 'Sanjeev Arora', 'Ruosong Wang', 'Zhiyuan Li', 'Simon S. Du']
|
2019-10-03
| null |
https://openreview.net/forum?id=rkl8sJBYvH
|
https://openreview.net/pdf?id=rkl8sJBYvH
|
iclr-2020-1
|
['small-data']
|
['computer-vision']
|
[ 5.42932637e-02 1.13510847e-01 -4.55494702e-01 -3.49177837e-01
-5.47697842e-01 -3.71790677e-01 6.36152446e-01 -3.31808150e-01
-9.21580732e-01 7.35620022e-01 -7.55270049e-02 -7.97650933e-01
-1.97054073e-01 -7.89546847e-01 -8.69312048e-01 -6.13655150e-01
-2.51616567e-01 -2.21313626e-01 4.41853434e-01 1.33418590e-01
1.59985237e-02 4.38883245e-01 -1.34731066e+00 2.15478301e-01
6.82379246e-01 1.38064206e+00 -7.21362010e-02 9.19979870e-01
1.99267238e-01 1.03154719e+00 -1.45173475e-01 -5.77054679e-01
5.89003682e-01 7.53013184e-03 -8.83647740e-01 -6.39004648e-01
1.13005507e+00 -3.09182853e-01 -4.72352713e-01 9.46829081e-01
2.95803696e-01 1.91325769e-01 1.03462625e+00 -1.25422680e+00
-8.52361619e-01 7.13840067e-01 -2.24798366e-01 1.89628214e-01
-5.58744967e-01 2.78241545e-01 8.59799385e-01 -9.14285600e-01
3.06902707e-01 7.77247429e-01 1.46010923e+00 5.65199971e-01
-1.46349370e+00 -6.68560803e-01 -3.28627437e-01 1.19873419e-01
-1.35395110e+00 -2.44357571e-01 1.53328821e-01 -5.62239945e-01
1.27897680e+00 1.92993253e-01 4.44391131e-01 1.36642241e+00
2.20199734e-01 5.54521799e-01 1.44089592e+00 -4.44481820e-01
2.86234647e-01 3.88958335e-01 4.41512436e-01 6.26037240e-01
1.33295476e-01 1.79497391e-01 -1.89264163e-01 -2.32500345e-01
9.19972003e-01 -1.43561080e-01 -8.75853151e-02 -2.55082220e-01
-1.01420200e+00 9.20474827e-01 7.42411733e-01 1.74454570e-01
-2.27807403e-01 4.42083240e-01 7.97671974e-01 5.84172130e-01
7.60115981e-01 3.31957370e-01 -6.93566322e-01 -5.95129840e-02
-1.05227828e+00 1.44657627e-01 8.33072484e-01 8.34912539e-01
7.84384489e-01 3.63141358e-01 -1.56024143e-01 1.09009373e+00
-1.04895145e-01 2.46967286e-01 5.97383916e-01 -9.22234416e-01
3.99106443e-01 4.18958925e-02 -1.78506598e-01 -2.85607517e-01
-4.73007828e-01 -6.32642329e-01 -8.83742392e-01 6.30159974e-01
1.07354665e+00 -1.62352577e-01 -1.04693890e+00 1.52113104e+00
-1.99824959e-01 3.19238812e-01 2.26921633e-01 5.65382898e-01
5.62276542e-01 4.87320364e-01 2.83797115e-01 2.49408111e-01
1.24507916e+00 -1.11882472e+00 4.52157855e-02 -1.08787514e-01
9.81344163e-01 -4.29055810e-01 1.42429340e+00 4.04813290e-01
-6.12378359e-01 -6.76105082e-01 -1.01517379e+00 -8.88566151e-02
-5.61866760e-01 3.57741892e-01 8.37095618e-01 8.51921439e-01
-1.24928939e+00 9.47871745e-01 -7.18710244e-01 -5.83387911e-01
6.84325695e-01 2.61861235e-01 -2.21343666e-01 2.69726831e-02
-1.20920014e+00 1.19927764e+00 4.32236254e-01 -6.94349781e-02
-7.70046055e-01 -1.07964230e+00 -6.65903389e-01 -9.37303603e-02
5.39927632e-02 -3.92965823e-01 1.38468266e+00 -1.15401876e+00
-1.47215378e+00 7.76058674e-01 4.24532145e-01 -1.27450395e+00
7.17968762e-01 -5.25854528e-01 -1.96865633e-01 -6.33886689e-03
-1.53276667e-01 8.24606121e-01 8.27519536e-01 -7.75729954e-01
-3.55505764e-01 -1.45309076e-01 8.84226263e-02 -2.72660911e-01
-6.03180110e-01 -3.12910900e-02 3.64271671e-01 -5.55280924e-01
-4.56781417e-01 -9.75949466e-01 -1.56435207e-01 -1.44957227e-03
-3.91289413e-01 -3.36985826e-01 7.27605641e-01 -6.90829515e-01
9.12544966e-01 -1.93340015e+00 -6.17103875e-01 6.27989545e-02
1.08178139e-01 5.29610097e-01 -8.32226798e-02 1.49516165e-01
-2.73984343e-01 1.00810178e-01 -3.36591989e-01 -4.65564616e-03
4.72714938e-02 4.40119915e-02 -5.35884440e-01 6.71871960e-01
4.83952761e-02 1.10558832e+00 -4.47761655e-01 -2.14478761e-01
2.03452617e-01 5.54210484e-01 -2.10020423e-01 -3.39036316e-01
2.33682916e-02 -2.69796968e-01 1.32023439e-01 5.31702101e-01
6.02826476e-01 -2.38031656e-01 -6.31796181e-01 -3.97657245e-01
-3.60911578e-01 -5.89098558e-02 -5.27009368e-01 1.33192694e+00
-6.82151496e-01 1.14919877e+00 -2.55706370e-01 -1.16692293e+00
8.59485030e-01 2.39297375e-01 2.42622141e-02 -4.02850211e-01
-1.67226061e-01 4.43485647e-01 1.19479358e-01 -1.61585867e-01
2.16206655e-01 -1.27406254e-01 2.44138405e-01 4.41657230e-02
5.59184313e-01 1.94583133e-01 -8.95682573e-02 9.86775383e-03
1.22900391e+00 1.91483065e-01 1.14905879e-01 -6.38019800e-01
1.94462582e-01 5.14190942e-02 1.19884074e-01 1.04041731e+00
-3.25610191e-01 4.85406190e-01 4.83075768e-01 -7.15236783e-01
-1.41161513e+00 -1.28836548e+00 -7.50106454e-01 1.38645542e+00
-5.30605435e-01 -1.64762855e-01 -1.02346253e+00 -7.48963058e-01
1.64371699e-01 1.00119877e+00 -1.01980424e+00 -1.77839294e-01
-3.44252288e-01 -8.66980195e-01 1.34019482e+00 7.88416207e-01
8.22278619e-01 -1.05707550e+00 -6.33239865e-01 -8.85751657e-03
3.67325842e-01 -1.21886384e+00 -1.14386976e-01 8.08700144e-01
-1.07587707e+00 -7.87724733e-01 -1.17160141e+00 -5.69117367e-01
3.37704539e-01 -2.33047642e-02 8.81392598e-01 -4.24270004e-01
-4.43693161e-01 3.75391006e-01 -2.90800780e-01 -3.85642469e-01
-3.07913691e-01 2.76480556e-01 1.92541018e-01 -2.32599840e-01
2.60974884e-01 -6.08299792e-01 -5.48696756e-01 3.18488091e-01
-6.24879181e-01 2.73230616e-02 8.78133237e-01 8.58143210e-01
2.76649982e-01 -1.45037353e-01 7.01317787e-01 -9.54064071e-01
3.47818196e-01 -4.07981575e-01 -5.78825474e-01 2.74517566e-01
-1.02234828e+00 1.16769172e-01 1.10020602e+00 -8.16487491e-01
-8.88896108e-01 -2.85950601e-01 -1.26150712e-01 -7.77287304e-01
-3.01346838e-01 1.95095912e-01 5.19615412e-01 -4.34503525e-01
1.27685571e+00 3.12319458e-01 -1.28550544e-01 -4.78594452e-01
3.31098318e-01 5.53670645e-01 4.77151752e-01 -6.04599535e-01
7.89718270e-01 3.19874614e-01 -6.64024726e-02 -1.04522717e+00
-9.30476189e-01 -1.81057200e-01 -5.70127606e-01 -3.17399949e-02
9.71525371e-01 -8.18882108e-01 -5.23003936e-01 7.34719336e-01
-7.21898496e-01 -1.06224036e+00 -6.27080560e-01 8.31839442e-01
-8.10432374e-01 -8.50682482e-02 -1.00985110e+00 -8.87438953e-01
-5.47189474e-01 -7.57686198e-01 5.23815870e-01 6.39194474e-02
-6.23411238e-02 -1.20740223e+00 2.16292478e-02 1.16194159e-01
7.36961365e-01 2.39683986e-02 1.00151956e+00 -9.52410579e-01
-8.25457945e-02 -1.95889711e-01 -5.98478913e-01 1.08352304e+00
-2.16897011e-01 1.01410776e-01 -1.45728958e+00 -1.93590954e-01
-2.68771321e-01 -8.22308660e-01 1.28664696e+00 6.20640278e-01
1.26474786e+00 -3.64996493e-01 1.12686716e-01 9.58785474e-01
1.61870396e+00 -1.81546763e-01 7.28013158e-01 7.22794294e-01
6.47740841e-01 5.72130084e-01 1.75988480e-01 -1.00424834e-01
-8.67046639e-02 3.15886647e-01 2.74810821e-01 -1.05100498e-01
-1.66418374e-01 -2.68934101e-01 5.68554521e-01 5.20765364e-01
-5.42466938e-01 3.55891466e-01 -8.78705323e-01 2.18956202e-01
-1.67590594e+00 -1.04621756e+00 -2.91554481e-01 2.37991309e+00
6.69913650e-01 5.49735665e-01 2.56985128e-01 -1.24252461e-01
5.16144395e-01 1.72322486e-02 -7.34593570e-01 -6.47210121e-01
-1.86636791e-01 4.68227863e-01 1.27539182e+00 1.85187519e-01
-1.20667326e+00 9.59502339e-01 6.30559635e+00 1.37845886e+00
-1.22750580e+00 2.43037358e-01 9.06908572e-01 -9.07335505e-02
9.81828719e-02 3.78728323e-02 -9.37305331e-01 3.73257488e-01
1.50622022e+00 -5.23755066e-02 4.63197082e-01 1.32096851e+00
-2.28692174e-01 -2.00309351e-01 -9.99157548e-01 7.63441443e-01
-3.20855767e-01 -1.37552440e+00 -1.90093160e-01 2.11785972e-01
6.66918695e-01 6.27111614e-01 3.55858564e-01 9.74646389e-01
3.95625025e-01 -1.42454827e+00 7.97714114e-01 3.81810337e-01
1.09256637e+00 -7.64730990e-01 6.97130144e-01 5.18148124e-01
-9.86072898e-01 -5.31690456e-02 -8.93233061e-01 2.58903746e-02
-4.88693476e-01 7.03381121e-01 -8.59769821e-01 1.56193227e-01
8.96171153e-01 4.78145868e-01 -7.85123050e-01 9.57209587e-01
-2.03777310e-02 1.28974152e+00 -3.67995828e-01 -4.34326380e-02
5.08038342e-01 -1.88391820e-01 1.48863107e-01 1.67573225e+00
2.32233539e-01 -3.53012234e-01 -2.20845208e-01 8.54673564e-01
-9.40247104e-02 1.10623375e-01 -7.98259497e-01 5.44328615e-02
2.36412257e-01 1.38398838e+00 -7.92455792e-01 -3.05356979e-01
-6.74262822e-01 1.05935931e+00 6.01947665e-01 6.70778871e-01
-8.62052619e-01 -6.33227110e-01 5.60765266e-01 1.19311772e-01
2.79053509e-01 -4.72914353e-02 -4.33371216e-01 -9.60783958e-01
-8.95467922e-02 -4.19887573e-01 1.69022307e-01 -6.96773410e-01
-1.45049667e+00 6.17595017e-01 1.02078944e-01 -1.07477236e+00
-3.01984344e-02 -1.11110210e+00 -7.63236463e-01 1.03306389e+00
-1.32507741e+00 -1.25062466e+00 -7.40426476e-04 6.11526906e-01
4.65511590e-01 -1.11260310e-01 8.66063058e-01 7.71302870e-03
-3.05574447e-01 7.77391493e-01 4.53858525e-01 5.24030507e-01
6.65457606e-01 -1.42197406e+00 6.77506089e-01 4.29278225e-01
1.94772273e-01 5.99725008e-01 3.90439928e-01 -2.70563155e-01
-8.88025820e-01 -1.29917824e+00 3.49667192e-01 -3.01399767e-01
9.28007007e-01 -3.15984040e-01 -1.07561016e+00 8.35227191e-01
1.06209271e-01 5.47694206e-01 5.68536520e-01 1.78693071e-01
-9.08765912e-01 -1.74523950e-01 -1.15018296e+00 5.28057158e-01
7.65374541e-01 -7.55580485e-01 -3.18491250e-01 2.78321594e-01
6.70829535e-01 2.21646447e-02 -1.01828587e+00 6.46760643e-01
9.17789459e-01 -1.14293444e+00 9.80543673e-01 -7.73020625e-01
4.93077874e-01 3.13875079e-01 -3.21815491e-01 -1.26913249e+00
-2.66175896e-01 -1.93053320e-01 -5.35487719e-02 8.91559541e-01
5.49458683e-01 -9.34267223e-01 8.09738517e-01 5.00327289e-01
-1.83590651e-01 -1.14680004e+00 -9.61740434e-01 -1.51661241e+00
6.60263479e-01 -8.30908597e-01 -2.53501028e-01 8.56678903e-01
-4.07559663e-01 8.69057849e-02 -2.84080207e-01 -2.73574382e-01
7.68517256e-01 -5.03440976e-01 5.86796463e-01 -1.22172058e+00
-2.56166846e-01 -6.50385916e-01 -5.88667274e-01 -6.53740406e-01
3.24123979e-01 -1.17796361e+00 -2.73431689e-01 -1.15110052e+00
5.91119453e-02 -6.07219458e-01 -5.37935734e-01 6.33388102e-01
1.91940323e-01 4.04933244e-01 2.73052510e-02 2.28303492e-01
-1.80835605e-01 1.58024296e-01 7.96847463e-01 8.94403383e-02
6.94548897e-03 2.80762047e-01 -3.09638828e-01 1.11201632e+00
1.00796640e+00 -2.83719689e-01 -4.07845467e-01 7.90498704e-02
2.41814200e-02 -2.78845847e-01 8.67467046e-01 -1.39667261e+00
2.59589970e-01 4.54357639e-03 6.84449315e-01 -8.76968205e-02
2.50609607e-01 -5.26081920e-01 -1.50192350e-01 5.18103242e-01
-5.57752371e-01 -2.34975934e-01 2.41458669e-01 4.20745760e-01
1.62318453e-01 -5.96682906e-01 1.08964336e+00 -6.52709380e-02
-6.95969164e-01 2.93921769e-01 -3.49097311e-01 2.90822178e-01
8.68064165e-01 -6.09754324e-01 -4.74422961e-01 -2.17316821e-01
-6.26874089e-01 -3.61111522e-01 2.94751137e-01 4.30292785e-02
4.34932709e-01 -1.15688527e+00 -6.61545396e-01 4.68340190e-03
3.02171353e-02 -2.39028171e-01 8.11175331e-02 1.08449757e+00
-5.92565656e-01 3.86234760e-01 -2.66931981e-01 -3.92677784e-01
-9.18882251e-01 1.98696285e-01 4.73722488e-01 -2.36619815e-01
-7.64115870e-01 1.16520512e+00 2.41454795e-01 -5.34592271e-01
2.74546057e-01 -5.74775457e-01 1.90632135e-01 -6.18994646e-02
4.45559233e-01 6.52878881e-01 2.33829945e-01 -1.76064208e-01
-2.01728791e-01 2.61330724e-01 -1.96420252e-01 -2.27548704e-02
1.38675797e+00 5.26512563e-01 2.31879547e-01 8.61406982e-01
1.59109044e+00 -2.11284056e-01 -1.47004306e+00 -2.07522646e-01
1.34068951e-01 -5.50513417e-02 8.01384356e-03 -8.22192073e-01
-7.11629689e-01 9.76993084e-01 6.91781282e-01 2.37596899e-01
7.12117851e-01 -1.47202417e-01 5.92807531e-01 6.97281718e-01
3.07901084e-01 -1.09235024e+00 -2.08072290e-01 7.12600529e-01
5.89547396e-01 -1.08358204e+00 -1.54416308e-01 -1.95574248e-03
-5.95991135e-01 1.52586365e+00 7.83847630e-01 -4.22549486e-01
7.04720914e-01 3.27726930e-01 -2.88494881e-02 1.77579179e-01
-7.79775500e-01 -6.61588684e-02 2.39928141e-01 6.14363134e-01
3.11773479e-01 1.49777606e-01 1.53186679e-01 5.16439557e-01
-4.32408065e-01 1.03308029e-01 2.66970158e-01 5.49553871e-01
-5.42911649e-01 -3.91081452e-01 -2.10780486e-01 9.81411934e-01
-4.69002426e-01 -5.53840160e-01 -9.01196301e-02 1.08180583e+00
2.78585497e-03 4.85411555e-01 3.41155678e-02 -5.23163497e-01
1.56133428e-01 5.10828316e-01 5.38148284e-01 -2.93584794e-01
-6.99168921e-01 -4.67050642e-01 4.16305624e-02 -4.22379971e-01
-1.07665181e-01 -2.14493901e-01 -9.76088643e-01 -2.54472584e-01
-4.22976315e-01 -1.19416388e-02 8.22139263e-01 8.19423497e-01
-1.58580408e-01 2.62681544e-01 2.15724528e-01 -7.81048596e-01
-1.25859165e+00 -1.24653411e+00 -9.00847137e-01 1.13207705e-01
2.39455894e-01 -3.64950716e-01 -7.93734014e-01 -1.85138118e-02]
|
[8.945683479309082, 2.888660430908203]
|
b51966f7-57e6-450f-be2d-e1c287c35e9e
|
learning-video-stabilization-using-optical
| null | null |
http://openaccess.thecvf.com/content_CVPR_2020/html/Yu_Learning_Video_Stabilization_Using_Optical_Flow_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Yu_Learning_Video_Stabilization_Using_Optical_Flow_CVPR_2020_paper.pdf
|
Learning Video Stabilization Using Optical Flow
|
We propose a novel neural network that infers the per-pixel warp fields for video stabilization from the optical flow fields of the input video. While previous learning based video stabilization methods attempt to implicitly learn frame motions from color videos, our method resorts to optical flow for motion analysis and directly learns the stabilization using the optical flow. We also propose a pipeline that uses optical flow principal components for motion inpainting and warp field smoothing, making our method robust to moving objects, occlusion and optical flow inaccuracy, which is challenging for other video stabilization methods. Our method achieves quantitatively and visually better results than the state-of-the-art optimization based and deep learning based video stabilization methods. Our method also gives a 3x speed improvement compared to the optimization based methods.
|
[' Ravi Ramamoorthi', 'Jiyang Yu']
|
2020-06-01
| null | null | null |
cvpr-2020-6
|
['video-stabilization']
|
['computer-vision']
|
[-3.25181752e-01 -4.51089591e-01 -4.38480258e-01 -2.60907244e-02
-4.93225247e-01 -4.34132874e-01 4.02900815e-01 -2.78520077e-01
-2.81780452e-01 7.94089258e-01 5.55625379e-01 -1.43555313e-01
4.70349550e-01 -1.91213548e-01 -1.05397570e+00 -7.27944314e-01
-2.26768269e-03 -1.61444739e-01 3.64424139e-01 -9.04017612e-02
4.00479376e-01 4.66485083e-01 -1.12407207e+00 1.50060564e-01
5.03542185e-01 7.63826668e-01 -8.86930078e-02 1.00662732e+00
2.17360228e-01 1.31425560e+00 -7.99806193e-02 2.06356823e-01
5.20618498e-01 -5.04407048e-01 -9.51815367e-01 3.77103865e-01
1.35213125e+00 -9.33854520e-01 -8.72078657e-01 8.78396213e-01
2.71282941e-01 6.49120569e-01 3.06091309e-01 -9.85535443e-01
-6.98815286e-01 2.33092248e-01 -8.42000663e-01 4.46780294e-01
4.78072524e-01 4.69444036e-01 6.69641316e-01 -7.01951265e-01
9.17011261e-01 1.51817334e+00 5.37728250e-01 7.81550109e-01
-1.42690694e+00 -3.24970961e-01 2.55527049e-01 3.56007516e-01
-8.31405759e-01 -7.02315867e-01 7.84463525e-01 -6.10152364e-01
7.38677621e-01 -1.20149598e-01 6.68984115e-01 7.84345269e-01
4.23465788e-01 8.36946547e-01 4.93665516e-01 -3.70379120e-01
2.54131500e-02 -7.15182066e-01 -2.16601312e-01 1.12632215e+00
-1.46460146e-01 1.61558598e-01 -6.86672151e-01 1.69806376e-01
1.34476626e+00 -4.52770293e-02 -6.86758578e-01 -4.42711353e-01
-1.60959101e+00 5.04561186e-01 4.44100022e-01 -1.65893450e-01
-2.91239649e-01 7.70374537e-01 2.41313577e-01 1.02123573e-01
5.35316765e-01 3.86926442e-01 -3.91578734e-01 -8.63890052e-02
-1.38692522e+00 4.24748898e-01 5.58748424e-01 7.32891083e-01
1.04611337e+00 7.38754690e-01 -2.95138180e-01 2.98893392e-01
4.63682443e-01 2.99585044e-01 3.37849259e-01 -1.97055793e+00
2.49692634e-01 1.55133262e-01 5.60075462e-01 -1.15729403e+00
-1.33669570e-01 3.21630269e-01 -6.61986768e-01 6.30357802e-01
6.27035379e-01 -3.10886413e-01 -9.98845577e-01 1.63163841e+00
3.58898789e-01 7.92375088e-01 -2.97675371e-01 1.21137857e+00
4.86165196e-01 7.02899277e-01 -3.08471769e-01 -5.48378646e-01
6.51772738e-01 -1.57649410e+00 -1.05773246e+00 -9.57978219e-02
3.22418809e-01 -9.86465931e-01 6.78203940e-01 1.93576768e-01
-1.35854447e+00 -5.73815286e-01 -9.68795478e-01 -4.29791451e-01
5.73213637e-01 -8.99870843e-02 4.28726673e-01 1.68843970e-01
-1.48138773e+00 9.57188487e-01 -1.51490963e+00 -2.66305715e-01
3.40233117e-01 5.12607872e-01 -5.09096146e-01 1.93587393e-01
-6.42700672e-01 7.43850827e-01 1.56113571e-02 9.61397514e-02
-1.26369441e+00 -8.79346788e-01 -1.09302568e+00 -2.62729563e-02
1.71495497e-01 -8.42011094e-01 1.13158774e+00 -1.45370829e+00
-2.06710243e+00 4.78038818e-01 -6.58388495e-01 -4.20880258e-01
5.52489698e-01 -6.93099618e-01 2.02622876e-01 3.00858825e-01
-2.09854558e-01 8.93670917e-01 1.30283070e+00 -1.01445973e+00
-4.91939873e-01 2.47826695e-01 6.52075335e-02 7.63679221e-02
-1.71497807e-01 2.54551202e-01 -5.74305654e-01 -7.85853684e-01
-1.63597614e-01 -1.01689243e+00 -3.95516634e-01 5.87308466e-01
-1.27277479e-01 3.19706619e-01 1.33690703e+00 -8.80672574e-01
1.17538512e+00 -1.80167365e+00 4.73869324e-01 -3.68951559e-01
3.42170715e-01 4.21316743e-01 -2.97348917e-01 -5.47860079e-02
-1.00353681e-01 -2.82910049e-01 -4.12193723e-02 -6.21589720e-01
-4.59020138e-01 1.45410091e-01 -4.73794013e-01 9.36956406e-01
2.45050654e-01 8.38210404e-01 -1.02392578e+00 -4.94808227e-01
5.85222065e-01 7.97371328e-01 -1.00147343e+00 3.85642558e-01
-3.46518695e-01 9.75810528e-01 -9.30965617e-02 6.50676250e-01
4.94217783e-01 -5.86874895e-02 6.86496422e-02 -4.40665573e-01
-3.13473135e-01 -4.09087874e-02 -1.06898952e+00 2.14338064e+00
2.65895240e-02 1.18662524e+00 4.01950777e-01 -7.95420647e-01
4.48679239e-01 3.16701889e-01 9.66970086e-01 4.52099741e-02
4.02994812e-01 -8.49876031e-02 -2.72401154e-01 -5.36609411e-01
6.48147166e-01 2.33218804e-01 8.69367480e-01 2.36218005e-01
2.82424539e-01 -5.12729306e-03 3.19963068e-01 1.38874620e-01
1.01274860e+00 9.68165874e-01 -1.13281272e-01 -5.03922224e-01
7.37954915e-01 -2.46614173e-01 9.70320463e-01 2.63146281e-01
-6.76668108e-01 9.58166599e-01 4.73538697e-01 -1.03803730e+00
-1.13148260e+00 -8.70358050e-01 3.30462068e-01 7.77758479e-01
3.31380993e-01 -5.62182605e-01 -9.36255217e-01 -3.47138941e-01
-2.16047645e-01 -5.67719201e-03 -4.72993106e-01 1.97371677e-01
-1.14826095e+00 -3.44751209e-01 1.48349985e-01 5.20847023e-01
3.23477060e-01 -6.48341358e-01 -3.89737546e-01 3.48666996e-01
-4.68210220e-01 -1.34985578e+00 -1.20560098e+00 -5.67765594e-01
-1.17753565e+00 -1.13563251e+00 -6.39309168e-01 -9.10731494e-01
6.95704520e-01 3.66312742e-01 9.05261219e-01 3.20295513e-01
-9.89772379e-02 3.78833652e-01 -5.50039485e-03 1.80178180e-01
-3.33731920e-01 -3.00985307e-01 4.55142319e-01 3.45675051e-01
-1.23205610e-01 -5.53123355e-01 -9.79721725e-01 2.32058823e-01
-8.87972713e-01 -5.17629646e-02 -1.54391021e-01 8.24153543e-01
6.35671258e-01 -5.73009253e-01 -4.16219711e-01 -2.47105300e-01
1.34010553e-01 7.33368248e-02 -1.00043726e+00 -7.77951404e-02
-3.13828886e-01 3.93243432e-01 5.29289842e-01 -6.02652013e-01
-9.54435885e-01 5.68541348e-01 2.22183883e-01 -1.15723217e+00
9.16686580e-02 -1.43098161e-02 2.02167973e-01 -5.99321008e-01
6.90841675e-01 -2.66477704e-01 3.86941046e-01 -2.12801576e-01
4.43613827e-01 -1.65223628e-01 8.32444310e-01 -5.10390818e-01
9.00613010e-01 9.67850983e-01 2.46305913e-01 -9.27630007e-01
-7.80019104e-01 -3.59505951e-01 -8.79574120e-01 -4.65292662e-01
1.25370288e+00 -1.06654036e+00 -9.30890143e-01 6.48082256e-01
-1.51671970e+00 -5.10467649e-01 7.59846792e-02 8.33844781e-01
-6.80106580e-01 8.62218857e-01 -1.14600134e+00 -3.17793667e-01
-2.87208259e-01 -1.54637170e+00 1.04564881e+00 3.50284100e-01
-1.58192113e-01 -1.30969846e+00 4.17524695e-01 1.16122000e-01
3.69643778e-01 5.93235433e-01 2.46603191e-01 5.17182350e-01
-1.17345083e+00 1.68825179e-01 4.23249528e-02 4.01454061e-01
3.18264872e-01 6.74647927e-01 -7.20888257e-01 -5.88753462e-01
-3.23270150e-02 -1.65916234e-01 1.01586306e+00 1.09104478e+00
8.82965386e-01 -4.14141357e-01 -1.53303728e-01 1.27976787e+00
1.44087434e+00 -2.44188339e-01 7.89474368e-01 4.01637137e-01
1.24935782e+00 3.41853589e-01 3.08199316e-01 2.83777773e-01
1.03897773e-01 7.09756434e-01 6.19838595e-01 -1.89368501e-01
-2.63713777e-01 8.24609548e-02 9.41495538e-01 6.21387184e-01
-5.59092402e-01 3.03793531e-02 -6.00655735e-01 5.65884948e-01
-2.28167033e+00 -1.20591414e+00 -2.26261333e-01 2.03700829e+00
1.02402174e+00 -1.99597254e-01 4.22101729e-02 -2.63971895e-01
7.45720327e-01 6.15718782e-01 -4.22220320e-01 -2.24767461e-01
-8.73134583e-02 -1.66921429e-02 6.24780953e-01 1.31057453e+00
-1.42928922e+00 1.27983809e+00 6.88683939e+00 1.25492483e-01
-1.56484449e+00 -1.05509432e-02 5.23482740e-01 -5.28207660e-01
-7.39909559e-02 2.30722412e-01 -6.48532331e-01 2.45976776e-01
6.24085546e-01 2.70305991e-01 6.31914198e-01 4.48018402e-01
8.43564570e-01 -1.10358022e-01 -1.14544070e+00 1.00815988e+00
1.64652213e-01 -1.84891427e+00 9.23248753e-03 -1.56503603e-01
1.21303999e+00 1.54494882e-01 -5.38735315e-02 -5.36027908e-01
3.62659186e-01 -7.29759872e-01 7.92026222e-01 6.56132698e-01
4.39088464e-01 -4.01891440e-01 2.89248049e-01 -1.51516154e-01
-1.02361667e+00 1.43543348e-01 -2.28347287e-01 -3.06107491e-01
5.68050504e-01 4.37110305e-01 -2.10734785e-01 5.43017909e-02
7.07689404e-01 1.36607790e+00 -2.27248758e-01 1.01275599e+00
-3.46639931e-01 6.65503561e-01 -1.18912704e-01 5.57744801e-01
3.20200920e-01 -4.86662060e-01 8.03344131e-01 9.80004847e-01
6.85743839e-02 -1.03150681e-01 3.51328850e-01 6.24128282e-01
-1.64574325e-01 -1.94481000e-01 -2.50549763e-01 1.14126436e-01
-3.11275590e-02 1.21050370e+00 -5.51340580e-01 -3.03101987e-01
-5.16526461e-01 1.13942432e+00 1.69365704e-01 7.75356650e-01
-8.77707779e-01 -1.17820069e-01 1.25364447e+00 2.60772258e-01
3.34308147e-01 -8.08210313e-01 7.18525052e-02 -1.88483346e+00
-1.26583591e-01 -6.17771745e-01 -9.20558870e-02 -7.83066809e-01
-7.86886156e-01 3.69901925e-01 -3.41772735e-01 -1.25165451e+00
-3.83892983e-01 -6.88853562e-01 -7.99402058e-01 6.99482918e-01
-1.66847026e+00 -1.17672551e+00 -2.86023617e-01 7.81252325e-01
7.11109996e-01 -2.16178633e-02 3.23038131e-01 2.53292441e-01
-5.80943525e-01 1.84202462e-01 2.64520347e-01 3.66181940e-01
1.38183546e+00 -1.09846091e+00 3.92390609e-01 1.51924849e+00
-3.94193679e-02 5.39891422e-01 9.53419864e-01 -4.64062065e-01
-1.58633459e+00 -1.07803774e+00 6.54591382e-01 -5.86217463e-01
8.03948939e-01 1.56575860e-03 -8.91972661e-01 9.93250191e-01
7.69739509e-01 5.69755793e-01 1.93050846e-01 -5.17054081e-01
-1.81709409e-01 -2.94986069e-01 -6.58826649e-01 7.30954826e-01
6.87959850e-01 -4.32804078e-01 -2.89130569e-01 2.51085460e-01
6.73806131e-01 -7.95488715e-01 -6.19307339e-01 4.52152491e-02
5.70482612e-01 -8.79825056e-01 9.60259497e-01 -8.90540063e-01
8.34616601e-01 -7.91932583e-01 1.31361961e-01 -1.14126503e+00
-5.94145715e-01 -1.42335284e+00 -6.03324473e-01 9.46106434e-01
-1.43649364e-02 6.27034232e-02 9.31616724e-01 8.23743880e-01
-1.51968405e-01 -1.82219312e-01 -6.69858694e-01 -3.14916134e-01
2.84603536e-02 4.22219373e-03 -1.26202360e-01 1.16239274e+00
-1.86910346e-01 8.45048651e-02 -9.00285661e-01 2.34728187e-01
8.22010398e-01 -9.11699682e-02 8.52321744e-01 -7.06099570e-01
-2.06475183e-01 -3.13728243e-01 -4.76596177e-01 -1.23567128e+00
6.04361773e-01 -2.89532214e-01 2.86502630e-01 -1.22551131e+00
-2.16990665e-01 4.54158843e-01 -1.14702627e-01 4.18174684e-01
-3.49019378e-01 3.20691049e-01 3.44934195e-01 3.52562994e-01
-5.25039017e-01 4.69893336e-01 1.43155098e+00 -2.69882739e-01
-4.19522434e-01 -4.25833344e-01 -5.02306260e-02 8.55728269e-01
4.25689012e-01 -1.35907903e-01 -1.58884704e-01 -1.06399417e+00
-1.98495626e-01 9.43761170e-02 2.95927346e-01 -9.65178072e-01
3.77170891e-01 -5.55671275e-01 4.28913862e-01 -2.36223400e-01
1.88531011e-01 -5.27813077e-01 -1.85537457e-01 5.22719264e-01
-2.44339034e-01 3.78007621e-01 2.29872689e-01 4.87481773e-01
-3.32039267e-01 2.36777335e-01 1.19574738e+00 -1.04551755e-01
-7.08325028e-01 6.79916084e-01 -6.60435557e-01 2.82473713e-02
6.71754658e-01 -9.12431106e-02 -2.11742133e-01 -5.74101448e-01
-5.84851384e-01 6.72997236e-02 8.03113699e-01 4.34974134e-01
5.95571458e-01 -1.37105453e+00 -7.72670925e-01 1.28116950e-01
-6.82129979e-01 4.65422012e-02 -1.34437149e-02 1.06456578e+00
-1.34641325e+00 3.51777226e-02 -5.12501895e-01 -7.85068989e-01
-1.31464064e+00 6.51908934e-01 5.84651113e-01 1.25020206e-01
-4.55850691e-01 8.48839700e-01 1.13272913e-01 2.50612259e-01
2.53207177e-01 -5.30750275e-01 1.46369142e-02 -3.10055584e-01
7.22584009e-01 6.13556862e-01 -3.25759739e-01 -8.40074062e-01
-2.39397541e-01 8.43642116e-01 1.97310317e-02 -6.12830073e-02
1.25711298e+00 -3.44805330e-01 -5.18818796e-01 9.54061896e-02
1.23387372e+00 9.72428620e-02 -2.21376562e+00 -1.16280988e-01
-2.56712854e-01 -7.41943121e-01 3.07224125e-01 1.96187235e-02
-1.52543449e+00 8.87831390e-01 4.16936010e-01 -4.31781024e-01
1.01947153e+00 -6.10492051e-01 1.02691495e+00 2.63978124e-01
-2.12031588e-01 -1.05124962e+00 2.37803489e-01 7.47272193e-01
6.54721618e-01 -1.43563437e+00 3.03960472e-01 -2.09424525e-01
-2.10505590e-01 1.41889739e+00 6.97085440e-01 -5.69323957e-01
6.87454998e-01 3.33754420e-01 3.87667716e-01 4.75660056e-01
-7.80440509e-01 2.32067734e-01 4.27885950e-01 5.62049806e-01
6.05561733e-01 -6.45725131e-01 -1.57657683e-01 -4.63727951e-01
5.53027034e-01 2.14251593e-01 8.64387393e-01 8.64676714e-01
-4.46828574e-01 -1.20815551e+00 -5.49066365e-01 -2.04463601e-01
-6.90820694e-01 -5.49039543e-02 -1.21558934e-01 3.50133508e-01
-2.29520321e-01 7.01863110e-01 1.16841942e-01 -1.61062673e-01
1.17009124e-02 -2.76382059e-01 6.81306303e-01 -1.84890330e-01
-4.65824932e-01 5.53255677e-01 -2.82643616e-01 -1.27537560e+00
-1.15786612e+00 -7.66376078e-01 -1.28277504e+00 -5.75303733e-01
-3.73948254e-02 -2.23028779e-01 1.75536722e-01 8.82860243e-01
3.62604409e-01 3.12002718e-01 5.55223405e-01 -1.63408494e+00
5.76908514e-02 -6.16360009e-01 -1.57150164e-01 5.58412671e-01
1.14395201e+00 -5.45325816e-01 -3.40122402e-01 9.27789211e-01]
|
[10.619760513305664, -1.4273602962493896]
|
6aacaa3d-783f-4ff5-a4b8-9dfe0d58c922
|
dial2vec-self-guided-contrastive-learning-of
|
2210.15332
| null |
https://arxiv.org/abs/2210.15332v1
|
https://arxiv.org/pdf/2210.15332v1.pdf
|
Dial2vec: Self-Guided Contrastive Learning of Unsupervised Dialogue Embeddings
|
In this paper, we introduce the task of learning unsupervised dialogue embeddings. Trivial approaches such as combining pre-trained word or sentence embeddings and encoding through pre-trained language models (PLMs) have been shown to be feasible for this task. However, these approaches typically ignore the conversational interactions between interlocutors, resulting in poor performance. To address this issue, we proposed a self-guided contrastive learning approach named dial2vec. Dial2vec considers a dialogue as an information exchange process. It captures the conversational interaction patterns between interlocutors and leverages them to guide the learning of the embeddings corresponding to each interlocutor. The dialogue embedding is obtained by an aggregation of the embeddings from all interlocutors. To verify our approach, we establish a comprehensive benchmark consisting of six widely-used dialogue datasets. We consider three evaluation tasks: domain categorization, semantic relatedness, and dialogue retrieval. Dial2vec achieves on average 8.7, 9.0, and 13.8 points absolute improvements in terms of purity, Spearman's correlation, and mean average precision (MAP) over the strongest baseline on the three tasks respectively. Further analysis shows that dial2vec obtains informative and discriminative embeddings for both interlocutors under the guidance of the conversational interactions and achieves the best performance when aggregating them through the interlocutor-level pooling strategy. All codes and data are publicly available at https://github.com/AlibabaResearch/DAMO-ConvAI/tree/main/dial2vec.
|
['Fei Huang', 'Yongbin Li', 'Junfeng Jiang', 'Rui Wang', 'Che Liu']
|
2022-10-27
| null | null | null | null |
['sentence-embeddings', 'sentence-embeddings']
|
['methodology', 'natural-language-processing']
|
[-4.51766044e-01 2.48473004e-01 -3.65188755e-02 -3.69082242e-01
-6.50133014e-01 -8.09726894e-01 1.09269965e+00 4.50298905e-01
-4.25987840e-01 6.12108767e-01 8.51716578e-01 -4.84007858e-02
8.10990483e-02 -4.95945990e-01 -6.74925372e-02 -5.89975953e-01
-2.74904817e-02 6.58720434e-01 -7.38524124e-02 -6.85489595e-01
3.52102876e-01 4.89994437e-02 -9.95124698e-01 2.40409628e-01
9.81728375e-01 9.10475910e-01 -1.03336088e-01 7.49523222e-01
-4.83025759e-01 7.11488962e-01 -6.85201645e-01 -7.31207550e-01
1.02785751e-02 -3.98028910e-01 -1.13129508e+00 -9.48253199e-02
-1.07358642e-01 -4.84924018e-01 -7.37479866e-01 5.94557762e-01
8.05494726e-01 4.14744437e-01 8.11517954e-01 -1.25294948e+00
-9.23732460e-01 6.15340531e-01 -2.84579456e-01 1.00640923e-01
6.12571716e-01 5.60111403e-02 1.50569832e+00 -1.00733435e+00
5.12107968e-01 1.43537068e+00 4.54592496e-01 6.06164217e-01
-1.12443471e+00 -2.49106914e-01 -1.49963483e-01 7.60294199e-02
-9.59174454e-01 -5.43511391e-01 8.35661292e-01 -5.27542830e-01
1.00962806e+00 1.51186109e-01 1.60856277e-01 1.27850187e+00
-4.40818854e-02 1.03371203e+00 7.37165213e-01 -3.07064205e-01
-5.32958582e-02 6.86883926e-01 7.72043169e-01 3.71411771e-01
-2.20753223e-01 -3.50253344e-01 -5.92824280e-01 -5.03779411e-01
3.94467652e-01 -2.52247393e-01 -4.51024771e-01 -3.11933160e-01
-1.07319295e+00 1.22353899e+00 4.18415129e-01 3.97849739e-01
-1.89377353e-01 -3.16183835e-01 9.84614909e-01 5.06511033e-01
7.63423264e-01 7.82562554e-01 -4.53014374e-01 -4.51362759e-01
-1.50389880e-01 1.05253063e-01 1.24529517e+00 7.45426476e-01
5.49926341e-01 -4.07506496e-01 -4.61402148e-01 1.54587495e+00
1.94365039e-01 -9.21855345e-02 7.00132489e-01 -8.42045307e-01
5.64708531e-01 7.41080523e-01 7.83973858e-02 -1.14736879e+00
-2.49285698e-01 4.72711660e-02 -6.46987498e-01 -5.09356916e-01
4.10664141e-01 -4.55891281e-01 1.55204730e-02 1.60301232e+00
3.46135348e-01 -2.87875652e-01 4.56372768e-01 8.63744080e-01
1.24676704e+00 8.15592527e-01 -1.50171921e-01 2.63690241e-02
1.25796628e+00 -1.37301207e+00 -8.69226217e-01 8.07860345e-02
8.73426974e-01 -8.47430408e-01 1.10327268e+00 3.49488109e-02
-8.41344118e-01 -5.10795295e-01 -8.68357062e-01 -3.87713373e-01
-5.40562928e-01 7.52926804e-03 6.72073007e-01 3.74690086e-01
-9.67974722e-01 3.77035648e-01 -3.66019070e-01 -5.67338884e-01
1.22879490e-01 2.59667367e-01 -5.63095570e-01 2.08522659e-02
-1.46149898e+00 1.09796393e+00 8.21645632e-02 -1.42579526e-01
-6.93024218e-01 -5.04636467e-01 -9.69962597e-01 1.22292265e-01
-8.56129974e-02 -2.03705207e-01 1.09212804e+00 -6.23576045e-01
-1.65274870e+00 9.68535602e-01 -1.24633469e-01 -4.20459419e-01
3.87850791e-01 -4.61488575e-01 -2.06730455e-01 4.76704761e-02
-8.84568784e-03 6.88237488e-01 1.63475275e-01 -1.07991838e+00
-3.28480572e-01 -2.30603829e-01 3.77173007e-01 4.75956559e-01
-8.21471393e-01 -1.78791676e-02 -5.08573115e-01 -2.84181178e-01
-2.87388474e-01 -7.42249429e-01 1.93306550e-01 -1.90153971e-01
-4.74691927e-01 -8.75512123e-01 7.67354012e-01 -7.70497561e-01
1.29407954e+00 -2.31262255e+00 3.59784305e-01 -4.28577036e-01
3.96165222e-01 4.81124282e-01 -3.65068942e-01 1.02561975e+00
2.02512771e-01 8.68987814e-02 -7.89036527e-02 -5.64439833e-01
3.13841879e-01 1.53521642e-01 -2.00591072e-01 4.44502354e-01
2.97160864e-01 8.16216409e-01 -9.02905703e-01 -2.11963043e-01
2.33686298e-01 6.21313274e-01 -5.96581876e-01 9.29033041e-01
1.10224504e-02 3.13372195e-01 -2.22878575e-01 1.14144787e-01
5.10124207e-01 -3.91963832e-02 4.24439549e-01 -1.00163989e-01
1.80419013e-01 7.95799315e-01 -7.06915259e-01 1.67504573e+00
-6.59333944e-01 1.00381446e+00 1.11684136e-01 -1.02228332e+00
1.19970059e+00 4.31306303e-01 3.41981560e-01 -4.70935494e-01
3.48920077e-01 -7.91894495e-02 -1.76100526e-02 -6.30252182e-01
6.54656112e-01 2.43624046e-01 -4.74983215e-01 7.89495111e-01
5.57898819e-01 1.30217567e-01 7.53914118e-02 5.25150418e-01
1.02838099e+00 -3.27453256e-01 3.31429064e-01 -1.90118611e-01
6.42449260e-01 -2.85671830e-01 3.00635427e-01 3.99045825e-01
-4.18214560e-01 4.15999889e-01 1.05294824e+00 -1.99422732e-01
-8.16024184e-01 -1.01686978e+00 -1.69756725e-01 1.53837466e+00
1.41854256e-01 -4.93426353e-01 -5.56806445e-01 -8.95089149e-01
2.54940033e-01 7.06859231e-01 -8.03362250e-01 -2.52287537e-01
-3.06724101e-01 -6.06950760e-01 6.60700738e-01 3.10738295e-01
4.67201024e-01 -8.67963910e-01 7.45959282e-02 1.74999624e-01
-2.75098801e-01 -9.78384554e-01 -6.57514155e-01 8.73357814e-04
-3.27422261e-01 -1.11174142e+00 -6.36380434e-01 -8.17234159e-01
3.84681374e-01 1.94287449e-01 1.00473845e+00 -1.27729282e-01
-1.18098222e-01 4.99851316e-01 -6.33951485e-01 1.07946873e-01
-3.17979425e-01 1.48053750e-01 1.63945511e-01 9.54242870e-02
7.56601572e-01 -4.35918033e-01 -4.52763587e-01 3.11871409e-01
-5.22767484e-01 -4.25969332e-01 -3.94035503e-02 1.35928786e+00
-2.50775844e-01 -4.93379265e-01 9.54225600e-01 -8.35582912e-01
1.47786355e+00 -8.02423418e-01 -6.19015396e-02 2.56580204e-01
-2.67463177e-01 2.34179311e-02 5.75230300e-01 -2.76093274e-01
-8.59881103e-01 -5.53833127e-01 -1.49508506e-01 6.82158768e-02
-1.35519385e-01 2.62592375e-01 -2.28929624e-01 3.21040213e-01
4.75747973e-01 1.75451636e-01 1.91263229e-01 -4.63911206e-01
6.67559981e-01 1.42607570e+00 8.70974883e-02 -6.51462138e-01
1.88328430e-01 -4.74792272e-02 -9.96850014e-01 -9.76526141e-01
-6.46322250e-01 -7.63762355e-01 -6.37927353e-01 -7.76704028e-02
9.51144278e-01 -7.82860935e-01 -8.70690048e-01 2.99249589e-01
-1.51611233e+00 -2.15731159e-01 9.53079686e-02 4.20912325e-01
-2.97864318e-01 5.15015900e-01 -8.34293544e-01 -7.12044120e-01
-4.66554850e-01 -1.05031121e+00 5.89193702e-01 1.39784768e-01
-6.99013412e-01 -1.50204062e+00 3.10367078e-01 5.43428481e-01
5.41365087e-01 -8.81723836e-02 9.16016877e-01 -1.62757444e+00
1.39657736e-01 -1.55825317e-01 -3.52718562e-01 7.93576658e-01
4.50740039e-01 -1.81469500e-01 -1.25496662e+00 -2.01581463e-01
-1.91158369e-01 -6.67348981e-01 7.21746981e-01 -4.53885645e-02
7.57776320e-01 -3.11617732e-01 -1.22928768e-01 1.32469639e-01
9.37293410e-01 8.43270719e-02 2.54304260e-01 5.20042963e-02
5.54796457e-01 1.01325643e+00 5.42464554e-01 6.44862533e-01
6.89001143e-01 6.31060362e-01 1.53386652e-01 2.00847328e-01
6.53080922e-03 -1.66596219e-01 5.09372532e-01 1.25061059e+00
3.24141860e-01 -4.91169184e-01 -8.53577137e-01 8.39937150e-01
-1.85170627e+00 -7.40729153e-01 3.76580544e-02 1.90708518e+00
9.83637094e-01 -1.63969964e-01 1.43443078e-01 -1.21126026e-01
6.20047271e-01 3.99108052e-01 -3.43049645e-01 -9.72630441e-01
3.65259796e-02 -9.30806398e-02 4.03165780e-02 8.54217410e-01
-1.00316918e+00 9.54276919e-01 5.25159597e+00 6.31303430e-01
-8.36848438e-01 3.36567312e-01 5.88238537e-01 1.27847135e-01
-3.40527415e-01 -3.33812773e-01 -7.45232344e-01 4.89641249e-01
8.57969046e-01 -4.50862199e-01 2.64937222e-01 7.58645058e-01
3.15490961e-02 8.25420246e-02 -1.28140604e+00 8.89167786e-01
2.76111066e-01 -1.13387501e+00 -1.01344235e-01 -1.51955694e-01
5.49175858e-01 -7.94015601e-02 -1.01757213e-01 6.87472641e-01
5.45387983e-01 -1.07569063e+00 -2.50054032e-01 9.76544917e-02
2.97140509e-01 -7.81829715e-01 1.20652092e+00 1.87846899e-01
-7.28317261e-01 4.63293642e-02 -3.76432121e-01 -4.59549762e-02
3.23254168e-02 4.19743329e-01 -1.16070187e+00 5.68881035e-01
3.43340933e-01 8.03850234e-01 -2.04841360e-01 5.37205875e-01
-1.99114427e-01 5.74594676e-01 -3.24919075e-02 -2.64201134e-01
4.35561150e-01 -2.22223967e-01 4.23181146e-01 1.52970731e+00
-1.61213785e-01 -1.49208680e-01 -3.14767249e-02 6.00066245e-01
-4.15092230e-01 4.11291182e-01 -7.61288583e-01 -3.23947757e-01
7.63207257e-01 1.28153479e+00 4.97618653e-02 -2.86201686e-01
-4.20798123e-01 1.18249464e+00 7.12164640e-01 2.81829894e-01
-7.37457037e-01 -7.85022318e-01 1.17979729e+00 -3.00967395e-01
9.38613862e-02 -2.63607860e-01 -1.24121174e-01 -1.01046932e+00
-1.18856445e-01 -7.45263219e-01 3.64190251e-01 -2.81830132e-01
-1.76963592e+00 7.56258130e-01 -2.94025779e-01 -8.79327536e-01
-3.57593060e-01 -6.06957734e-01 -8.11537147e-01 1.00834954e+00
-1.27389741e+00 -7.75385261e-01 -2.97464490e-01 3.40961665e-01
7.84025729e-01 -5.19716918e-01 1.20089364e+00 2.69293308e-01
-6.94292724e-01 9.68562305e-01 3.09660971e-01 6.37299657e-01
1.13151574e+00 -1.22037077e+00 1.94650553e-02 -2.16558333e-02
-1.52798682e-01 7.68241465e-01 4.42376852e-01 -5.93778603e-02
-1.27785552e+00 -8.52753282e-01 1.25005794e+00 -5.46152234e-01
9.39999700e-01 -6.61120474e-01 -1.08930957e+00 5.28710365e-01
9.53515172e-01 -3.19361180e-01 1.24861765e+00 4.83037174e-01
-4.60267723e-01 -1.42412679e-02 -1.10959852e+00 5.20489514e-01
5.66318035e-01 -7.55762637e-01 -9.39943373e-01 4.25996721e-01
8.77261221e-01 -1.24783702e-01 -1.21762764e+00 1.02551570e-02
4.24859941e-01 -9.33887005e-01 7.98119962e-01 -8.35816085e-01
7.08737612e-01 2.79655069e-01 -3.42028230e-01 -1.55466795e+00
5.31280115e-02 -5.64102292e-01 4.24573347e-02 1.77611911e+00
3.46108913e-01 -8.21815908e-01 2.22096667e-01 5.38884699e-01
-4.25468497e-02 -8.33081961e-01 -6.96797431e-01 -5.27704298e-01
3.49767268e-01 -1.28335981e-02 3.40690136e-01 1.36897159e+00
6.57065690e-01 7.36235917e-01 -3.68308544e-01 -1.46719441e-01
1.83785751e-01 4.44157777e-04 1.02500296e+00 -8.79433274e-01
-2.34510079e-01 -4.63583738e-01 -3.16363186e-01 -1.44698668e+00
6.51133001e-01 -1.01425290e+00 4.45552804e-02 -1.60655081e+00
1.03064895e-01 -4.69349205e-01 -1.59205407e-01 1.19189434e-01
-1.58846200e-01 -1.18065685e-01 -8.76698922e-03 1.65386405e-02
-6.64533436e-01 9.84656096e-01 9.52872038e-01 -2.52514660e-01
-2.79641896e-01 -2.90391624e-01 -8.27311814e-01 4.24401909e-01
1.10037136e+00 -1.83978319e-01 -3.51150304e-01 -5.56298792e-01
-3.77971709e-01 7.33109713e-02 2.25808322e-02 -4.33408529e-01
1.99965969e-01 2.36013860e-01 -1.29880667e-01 -1.60971761e-01
6.35090053e-01 -3.09776247e-01 -7.40687549e-01 7.19938502e-02
-9.36773062e-01 -1.00058630e-01 1.60734862e-01 4.06320542e-01
-5.93816876e-01 -1.76814377e-01 5.49188673e-01 1.29922032e-01
-5.29950261e-01 -5.39626516e-02 -3.67962778e-01 2.96491146e-01
9.16724980e-01 2.95622945e-01 -4.75442737e-01 -6.43003702e-01
-7.37712860e-01 6.19409502e-01 1.58279464e-01 7.05009580e-01
4.92834330e-01 -1.34899437e+00 -8.44837844e-01 -2.01647654e-02
2.57241964e-01 -2.72957265e-01 2.13858381e-01 8.51206183e-01
-1.66801900e-01 6.12798572e-01 -2.68156052e-01 -3.41792583e-01
-1.42399359e+00 1.36121437e-01 2.39934340e-01 -3.80859852e-01
-1.81737527e-01 1.07865632e+00 3.13533068e-01 -1.03197253e+00
4.79795694e-01 1.14025936e-01 -4.93108660e-01 5.70923269e-01
4.37904269e-01 4.21146840e-01 -2.93700039e-01 -6.92161083e-01
-3.72129589e-01 2.26148218e-01 -3.73964429e-01 6.97030276e-02
1.27627194e+00 -2.54315317e-01 -3.00361425e-01 5.74902117e-01
1.82195830e+00 9.58123282e-02 -8.70959580e-01 -3.56006294e-01
3.32749300e-02 -2.92664200e-01 -3.25459301e-01 -5.29705167e-01
-7.86881149e-01 1.28378391e+00 2.19392926e-01 5.08686721e-01
5.22118688e-01 1.71670496e-01 8.72462332e-01 3.58471245e-01
7.87610561e-03 -1.05221617e+00 3.35129380e-01 9.26873207e-01
1.06887531e+00 -1.47243905e+00 -2.99795002e-01 -2.13965371e-01
-1.06426239e+00 1.05713689e+00 7.70817578e-01 -2.84215510e-01
5.86557925e-01 -1.95698977e-01 2.51890510e-01 -8.09533522e-02
-1.03397799e+00 -1.38223963e-02 1.21164434e-01 4.77473974e-01
9.37948644e-01 1.15096532e-01 -5.46302617e-01 7.39646494e-01
-9.76271927e-02 -6.33997202e-01 4.65220720e-01 5.33718050e-01
-2.95454472e-01 -1.23036873e+00 6.99253604e-02 2.57079691e-01
-2.77897328e-01 -5.98801523e-02 -7.93814123e-01 6.32766247e-01
-3.83883268e-01 1.29613960e+00 2.29530632e-01 -6.63852990e-01
3.04293066e-01 2.83572048e-01 1.17958859e-02 -7.41416633e-01
-8.35011005e-01 -2.39123464e-01 5.89980006e-01 -3.50983143e-01
-1.71269804e-01 -3.48807752e-01 -1.12030959e+00 -5.34820497e-01
-2.73863167e-01 6.65857196e-01 3.56374949e-01 6.19279861e-01
5.50597489e-01 4.24593985e-01 9.89375651e-01 -6.22817218e-01
-7.73311913e-01 -1.39210856e+00 -4.20052558e-01 5.62720001e-01
2.18108594e-01 -6.48218930e-01 -6.40182972e-01 -3.55389118e-01]
|
[12.641233444213867, 7.837474822998047]
|
f70f15bb-0b8d-48f1-850c-01514a4bf4de
|
ultra-low-bitrate-video-conferencing-using
|
2012.00346
| null |
https://arxiv.org/abs/2012.00346v1
|
https://arxiv.org/pdf/2012.00346v1.pdf
|
Ultra-low bitrate video conferencing using deep image animation
|
In this work we propose a novel deep learning approach for ultra-low bitrate video compression for video conferencing applications. To address the shortcomings of current video compression paradigms when the available bandwidth is extremely limited, we adopt a model-based approach that employs deep neural networks to encode motion information as keypoint displacement and reconstruct the video signal at the decoder side. The overall system is trained in an end-to-end fashion minimizing a reconstruction error on the encoder output. Objective and subjective quality evaluation experiments demonstrate that the proposed approach provides an average bitrate reduction for the same visual quality of more than 80% compared to HEVC.
|
['Stéphane Lathuilière', 'Giuseppe Valenzise', 'Goluck Konuko']
|
2020-12-01
| null | null | null | null |
['image-animation']
|
['computer-vision']
|
[ 4.45733339e-01 -1.72519051e-02 -4.09747541e-01 -2.40578413e-01
-7.76426017e-01 2.37106040e-01 2.55209327e-01 -2.00019419e-01
-2.43664473e-01 7.33675599e-01 2.46614829e-01 -4.38921720e-01
-1.34209961e-01 -6.79181218e-01 -7.11906850e-01 -3.90001059e-01
-1.63611189e-01 -1.42498106e-01 -7.28767812e-02 -4.11374383e-02
3.49722445e-01 3.12086880e-01 -1.47003233e+00 4.79787737e-01
4.87476468e-01 1.44557226e+00 3.76486540e-01 1.24331188e+00
4.54842865e-01 1.28124058e+00 -4.62416351e-01 -5.16404271e-01
3.56511176e-01 -3.12363327e-01 -5.63887715e-01 6.82293922e-02
5.21016359e-01 -1.31351209e+00 -1.32048631e+00 1.02212381e+00
6.44209385e-01 -1.22934058e-01 4.19335455e-01 -6.03687704e-01
-2.93446779e-01 2.37044126e-01 -4.80537653e-01 4.86287087e-01
3.35756510e-01 7.98573419e-02 7.88739145e-01 -7.37287939e-01
4.05076921e-01 9.75931704e-01 5.27485549e-01 5.61303556e-01
-9.43034172e-01 -5.75243235e-01 -3.65151942e-01 7.72805512e-01
-1.50347304e+00 -1.12359190e+00 7.27124751e-01 -5.54603077e-02
1.12615967e+00 7.91673884e-02 6.00390673e-01 9.59842563e-01
6.18952692e-01 6.62084639e-01 7.71886557e-02 -2.76370019e-01
2.73984075e-01 -6.14678085e-01 -6.61618590e-01 6.36060596e-01
3.70162912e-02 3.38813961e-01 -5.72243512e-01 3.91621709e-01
1.26899207e+00 -1.45099163e-01 -6.15076959e-01 -7.47421011e-02
-9.03117716e-01 5.96841455e-01 2.79245466e-01 1.94186509e-01
-6.38941705e-01 7.40974426e-01 7.49808729e-01 4.76100177e-01
4.83792990e-01 -2.33438835e-01 -1.28919423e-01 -7.34183371e-01
-1.45619166e+00 1.99946657e-01 6.34891570e-01 9.48011696e-01
1.20565429e-01 5.80796957e-01 -1.68782294e-01 8.54357839e-01
5.35943210e-01 1.84832439e-01 4.21827435e-01 -1.53096819e+00
9.26031411e-01 -4.21848685e-01 4.13510464e-02 -9.56973016e-01
1.12106130e-01 -3.22565019e-01 -1.20831692e+00 2.55046248e-01
-2.64317006e-01 -5.17560877e-02 -9.43226814e-01 1.36700833e+00
-3.97681385e-01 6.18753374e-01 2.40890473e-01 1.06017220e+00
7.36875653e-01 1.07583737e+00 -1.72568351e-01 -5.68349898e-01
6.87246442e-01 -9.26978111e-01 -1.23427725e+00 3.44585441e-02
2.02509865e-01 -7.32170641e-01 4.36054915e-01 6.14176095e-01
-1.98106182e+00 -6.47224844e-01 -1.57045639e+00 -1.72493398e-01
5.86342454e-01 -1.11342512e-01 1.99644580e-01 5.23347378e-01
-1.26851571e+00 8.45451891e-01 -7.91535854e-01 2.91952789e-01
6.03620231e-01 4.57602203e-01 3.69456559e-02 -2.31033206e-01
-1.05055809e+00 6.13763750e-01 4.44684774e-01 -2.89830640e-02
-1.14811981e+00 -7.47239590e-01 -8.77609432e-01 6.69419885e-01
-1.14770725e-01 -7.13830650e-01 1.56179714e+00 -9.16202962e-01
-1.90122247e+00 4.22417969e-01 -1.26242474e-01 -8.54913652e-01
6.59147441e-01 -2.40802482e-01 -4.67757255e-01 6.69642389e-01
-3.52934539e-01 4.62271869e-01 1.24798048e+00 -9.04938281e-01
-7.90120900e-01 1.67573839e-01 -4.26541977e-02 1.76863909e-01
-3.14655691e-01 -2.18311071e-01 -7.72354364e-01 -7.61268556e-01
5.06623127e-02 -1.97280884e-01 3.05722188e-02 2.77247220e-01
3.46042246e-01 2.60015905e-01 1.12510276e+00 -1.03313804e+00
1.48775530e+00 -2.01946640e+00 2.61418641e-01 -1.20832123e-01
4.11858410e-01 5.70846915e-01 -1.56558543e-01 1.41217813e-01
6.58604428e-02 -2.18441769e-01 -1.01792566e-01 -2.85997808e-01
-4.32915062e-01 3.60950753e-02 -1.92458719e-01 3.92452210e-01
-1.38626724e-01 7.57114589e-01 -6.85008883e-01 -3.45610380e-01
6.52450621e-01 9.00759757e-01 -8.99344742e-01 6.21084511e-01
7.52889961e-02 4.85766307e-02 1.67022552e-02 5.38872898e-01
8.37509573e-01 -1.20993443e-01 3.57297152e-01 -4.19947058e-01
2.50120431e-01 4.27056134e-01 -6.90681159e-01 1.95916092e+00
-8.77700448e-01 1.41547024e+00 3.72877240e-01 -1.18469369e+00
5.97303033e-01 7.91105390e-01 6.09915257e-01 -1.05218887e+00
3.13263625e-01 1.91745102e-01 -1.34432599e-01 -7.71193504e-01
7.84629405e-01 -2.30455473e-01 5.68112731e-01 -2.54400838e-02
9.09738466e-02 6.97488710e-02 1.39214564e-02 3.29883285e-02
1.20100033e+00 -9.38577130e-02 3.33869874e-01 1.85215801e-01
4.28870022e-01 -9.18416977e-01 4.11005765e-01 4.11110520e-01
-2.49348968e-01 7.82614291e-01 2.34540477e-01 -4.86807883e-01
-1.84550488e+00 -1.16118705e+00 -1.03661783e-01 6.42041564e-01
1.72819614e-01 -4.10197049e-01 -6.84814394e-01 -1.01402877e-02
-5.41691124e-01 5.47725379e-01 8.56506154e-02 -2.64856219e-01
-9.06638563e-01 -6.03332929e-02 5.18571734e-01 4.51382309e-01
7.77998805e-01 -7.19625533e-01 -8.44043016e-01 5.52775443e-01
-6.06390297e-01 -1.52267766e+00 -2.40416482e-01 -1.88365161e-01
-1.06460464e+00 -5.49939036e-01 -1.01496589e+00 -9.38437760e-01
1.24218993e-01 5.95643669e-02 9.87375557e-01 7.32580125e-02
-2.10379556e-01 -2.81206742e-02 -2.12434515e-01 2.74007767e-01
-6.30780935e-01 -1.40131876e-01 -1.82027832e-01 -1.21702284e-01
-2.02863812e-02 -7.04278648e-01 -8.81049693e-01 -1.83269083e-01
-9.13405955e-01 1.25700220e-01 5.30389726e-01 6.73026919e-01
2.63495266e-01 1.45706937e-01 5.97626090e-01 -1.35167345e-01
6.04631841e-01 -4.04628187e-01 -5.98347723e-01 -1.20446190e-01
-3.56061965e-01 -1.32096931e-01 7.16767669e-01 -1.36201888e-01
-7.98011363e-01 -3.75489116e-01 -6.10023916e-01 -8.80661607e-01
1.73729807e-01 5.68963587e-01 -1.46354496e-01 -1.48889408e-01
1.95499673e-01 4.35768187e-01 -8.80714506e-02 -3.40799391e-01
-7.54225999e-02 1.08997428e+00 7.66016781e-01 8.20475370e-02
2.89964557e-01 2.60657370e-01 1.22936308e-01 -1.09928322e+00
-2.10447326e-01 -8.17516074e-02 -2.19234824e-01 -4.31984782e-01
7.07263231e-01 -1.25013006e+00 -7.86660254e-01 2.51005828e-01
-1.44495368e+00 -2.77980447e-01 -6.80712909e-02 7.64310122e-01
-1.11420870e+00 6.68690264e-01 -9.30873811e-01 -5.55593669e-01
-5.39732099e-01 -1.21560359e+00 8.71217072e-01 -1.56299204e-01
8.21101665e-02 -7.69361615e-01 -1.67027324e-01 3.42164248e-01
7.34735906e-01 -9.89171118e-02 9.24986839e-01 2.61905283e-01
-8.85278344e-01 -2.18442813e-01 -4.67477202e-01 4.96096045e-01
-5.97031880e-03 -2.52614856e-01 -9.05454874e-01 -5.69573224e-01
1.42191023e-01 -2.32289732e-01 8.57154191e-01 7.43429244e-01
1.69795454e+00 -5.99049985e-01 1.93470687e-01 1.16874993e+00
1.62487411e+00 4.42880899e-01 1.24222267e+00 1.07704028e-01
4.47031170e-01 -2.15588100e-02 2.64981180e-01 8.07365179e-01
5.88315800e-02 7.01455534e-01 6.87383413e-01 2.36059740e-01
-3.43790799e-01 -1.33642077e-01 2.30335310e-01 9.59692299e-01
-3.36817473e-01 -6.93692327e-01 -4.72488970e-01 4.93961900e-01
-1.71210825e+00 -1.32802534e+00 4.20446455e-01 2.10577631e+00
6.43179715e-01 3.04856479e-01 -4.32282448e-01 5.14434576e-01
6.07784212e-01 5.67170084e-01 -3.29886138e-01 -7.84731030e-01
2.87220329e-01 2.78129578e-01 5.31271994e-01 6.40290737e-01
-9.62242901e-01 6.48945212e-01 7.19407701e+00 9.38172817e-01
-1.28940713e+00 -4.34509553e-02 6.74720943e-01 -4.41731125e-01
6.51374599e-03 -5.22683799e-01 -7.43005425e-03 5.01556277e-01
1.53870952e+00 -1.75768703e-01 6.40249848e-01 7.19979823e-01
7.22965002e-01 3.13832015e-01 -1.16102064e+00 1.69040990e+00
1.47094995e-01 -1.84873021e+00 2.18771338e-01 1.65822014e-01
5.55731833e-01 -6.09492436e-02 3.19168031e-01 -5.58591671e-02
-4.84911680e-01 -1.18904245e+00 7.45707810e-01 4.56005961e-01
1.48894620e+00 -1.00419772e+00 6.91161513e-01 1.92575112e-01
-1.13805008e+00 -1.84577316e-01 -5.93789995e-01 -2.51983464e-01
5.87992728e-01 2.32046023e-01 -5.36222994e-01 3.13966393e-01
5.84201455e-01 8.76133621e-01 2.48568952e-01 1.18404841e+00
1.40403241e-01 5.76588273e-01 2.42664993e-01 4.41777825e-01
3.24429125e-01 1.30841106e-01 6.06090128e-01 1.21901000e+00
6.66772425e-01 2.02000111e-01 -3.75436246e-01 3.27910900e-01
-4.88512427e-01 -3.92097950e-01 -5.55184305e-01 1.89427406e-01
3.19383591e-01 5.43781400e-01 -4.65454198e-02 -4.62026328e-01
-4.35666800e-01 1.35388827e+00 -9.86466091e-03 5.21968007e-01
-8.80049944e-01 -6.18062973e-01 7.32888579e-01 2.79487848e-01
7.96442509e-01 -3.07410121e-01 -1.05916299e-01 -1.04676926e+00
1.82896376e-01 -8.00504923e-01 -2.47116029e-01 -8.87313843e-01
-4.55047399e-01 4.02087778e-01 -2.40683720e-01 -1.44295835e+00
-6.50697827e-01 -3.42592001e-01 -4.35313284e-01 5.67014992e-01
-1.71380568e+00 -5.16789317e-01 -2.80482590e-01 4.81002569e-01
1.11001611e+00 -5.03759205e-01 5.93009353e-01 8.64306092e-01
-2.75431842e-01 9.00134385e-01 4.52302903e-01 -7.34298155e-02
1.26084968e-01 -4.86745477e-01 3.40786099e-01 1.09885573e+00
-3.81212443e-01 1.76639706e-01 8.04214597e-01 -2.82757640e-01
-1.49217200e+00 -1.12901151e+00 7.85058975e-01 6.61590219e-01
2.76263714e-01 -5.93730062e-02 -8.17632556e-01 4.70682591e-01
5.48005700e-01 8.53308737e-02 3.42326880e-01 -6.41942024e-01
-3.07850353e-02 -2.49957815e-01 -1.30722666e+00 5.24951994e-01
8.04811597e-01 -4.88989055e-01 -2.15935171e-01 -1.51236892e-01
8.07994008e-01 -4.42498863e-01 -9.27262485e-01 5.32501221e-01
6.97464228e-01 -1.06004345e+00 9.83285129e-01 -1.95574179e-01
1.21033990e+00 1.61276892e-01 -6.00406528e-01 -8.86356831e-01
-4.31346864e-01 -7.87417769e-01 -9.87989545e-01 5.44202328e-01
-6.78224787e-02 1.98481366e-01 1.04318929e+00 1.07666031e-01
-1.58338547e-01 -8.64896297e-01 -1.45938659e+00 -5.29471219e-01
-3.48889738e-01 -5.91445923e-01 3.04104358e-01 3.24446768e-01
7.17401803e-02 2.14117959e-01 -1.03242028e+00 4.58493717e-02
7.05364347e-01 -5.18675327e-01 2.59480685e-01 -5.88736594e-01
-4.51166242e-01 -3.60292614e-01 -8.64149094e-01 -1.68315506e+00
6.61800131e-02 -5.80137372e-01 1.08389653e-01 -1.60091138e+00
8.37827772e-02 8.22147131e-02 -4.22689855e-01 -4.20064837e-01
2.92789847e-01 4.61964756e-01 9.83340517e-02 5.69878295e-02
-5.27796805e-01 8.06193531e-01 1.04469192e+00 -2.90159196e-01
4.85509560e-02 -1.91434156e-02 -2.72266716e-01 6.31207883e-01
6.79499984e-01 3.74739021e-02 -5.61241210e-01 -1.05897200e+00
-3.28340717e-02 9.28032577e-01 2.97921777e-01 -1.35484028e+00
7.85111561e-02 1.69215817e-02 4.45036262e-01 -5.51809788e-01
6.71155155e-01 -1.02848744e+00 -2.61017829e-01 7.20766187e-01
-5.69964826e-01 -1.16929270e-01 7.67684728e-02 6.37859166e-01
-4.23639983e-01 -9.51975062e-02 1.02143514e+00 2.01948136e-01
-7.27353692e-01 4.34419066e-01 -8.43574822e-01 -3.87012929e-01
7.75640845e-01 -3.54076624e-01 8.04673061e-02 -9.62390840e-01
-3.72869074e-01 -3.56133997e-01 1.66301131e-01 3.29971611e-01
1.44594562e+00 -1.33711660e+00 -8.55010867e-01 4.11767542e-01
-2.70528346e-01 -3.53343219e-01 4.43758518e-01 2.84513593e-01
-1.22924352e+00 6.76000774e-01 -4.66483742e-01 -6.05892658e-01
-1.38377881e+00 4.15957958e-01 4.76223141e-01 -1.80240199e-01
-8.85030448e-01 7.06852317e-01 -2.54743546e-01 6.33552134e-01
6.09980702e-01 -9.65012982e-02 -1.31676286e-01 -6.09053731e-01
8.67739141e-01 5.54958284e-01 1.53038919e-01 -6.53949797e-01
5.53633831e-02 3.52430880e-01 -9.39403549e-02 -1.98760644e-01
1.41063225e+00 -4.30324823e-01 3.36830944e-01 -3.48988622e-02
1.76012564e+00 -4.95275617e-01 -1.55154359e+00 2.70407144e-02
-5.52572250e-01 -1.09799993e+00 7.15796769e-01 -3.43269050e-01
-1.36453927e+00 1.03482175e+00 1.03110111e+00 -1.32976338e-01
1.45962262e+00 -5.38566291e-01 1.23734581e+00 2.56175458e-01
1.58701047e-01 -1.10720849e+00 1.35685772e-01 3.01982611e-01
8.00880969e-01 -1.06759226e+00 1.30902812e-01 -6.48925230e-02
-1.86113685e-01 1.31975150e+00 2.59416282e-01 -1.62032396e-01
5.25129914e-01 4.66013074e-01 -2.34855324e-01 2.01331362e-01
-1.04950774e+00 3.91875595e-01 -7.91346945e-04 5.16425252e-01
6.61628306e-01 -2.06423208e-01 -4.46530312e-01 -1.56066403e-01
-2.24084826e-03 4.86556113e-01 6.90952480e-01 8.78728688e-01
-6.21966064e-01 -7.01362789e-01 -5.53234741e-02 5.07632494e-01
-8.77198637e-01 -2.22845659e-01 5.55118620e-01 1.57061785e-01
-2.37724900e-01 9.96155679e-01 4.74769711e-01 -5.82039654e-01
-1.25949988e-02 -4.08463746e-01 7.45104671e-01 -6.64457455e-02
1.59943514e-02 2.66950905e-01 5.83174154e-02 -8.22003126e-01
-3.56348485e-01 -3.83591913e-02 -9.00073230e-01 -7.67094195e-01
1.99000508e-01 -2.71033168e-01 7.70210207e-01 7.45594442e-01
2.69931287e-01 9.40719604e-01 8.65461469e-01 -1.05570471e+00
-3.73834074e-01 -7.85466015e-01 -3.72798860e-01 1.35251448e-01
9.55193222e-01 -8.18777159e-02 2.52176933e-02 2.33888909e-01]
|
[11.384016036987305, -1.6105942726135254]
|
a45c90f6-b62e-49ec-bcc9-742941e5e9c5
|
deep-learning-based-parameter-mapping-for
| null | null |
https://openreview.net/forum?id=wthvY6Y9e
|
https://openreview.net/pdf?id=wthvY6Y9e
|
Deep learning-based parameter mapping for joint relaxation and diffusion tensor MR Fingerprinting
|
Magnetic Resonance Fingerprinting (MRF) enables the simultaneous quantification of multiple properties of biological tissues. It relies on a pseudo-random acquisition and the matching of acquired signal evolutions to a precomputed dictionary. However, the dictionary is not scalable to higher-parametric spaces, limiting MRF to the simultaneous mapping of only a small number of parameters (proton density, T1 and T2 in general). Inspired by diffusion-weighted SSFP imaging, we present a proof-of-concept of a novel MRF sequence with embedded diffusion-encoding gradients along all three axes to efficiently encode orientational diffusion and T1 and T2 relaxation. We take advantage of a convolutional neural network (CNN) to reconstruct multiple quantitative maps from this single, highly undersampled acquisition. We bypass expensive dictionary matching by learning the implicit physical relationships between the spatiotemporal MRF data and the T1, T2 and diffusion tensor parameters. The predicted parameter maps and the derived scalar diffusion metrics agree well with state-of-the-art reference protocols. Orientational diffusion information is captured as seen from the estimated primary diffusion directions. In addition to this, the joint acquisition and reconstruction framework proves capable of preserving tissue abnormalities in multiple sclerosis lesions.
|
['Marion I. Menzel', 'Bjoern H. Menze', 'Derek K. Jones', 'Michela Tosetti', 'Valentina Tomassini', 'Joseph R. Whittaker', 'Alberto Merola', 'Sebastian Endt', 'Jonathan Dannenberg', 'Diana Waldmannstetter', 'Anjany Sekuboyina', 'Miguel Molina-Romero', 'Guido Buonincontri', 'Ilona Lipp', 'Pedro A. Gómez', 'Carolin M. Pirkl']
|
2020-01-25
| null | null | null |
midl-2019-7
|
['magnetic-resonance-fingerprinting']
|
['medical']
|
[ 3.89382780e-01 -7.21540973e-02 -7.06104562e-02 -3.45489293e-01
-6.23044968e-01 -4.50666130e-01 5.35259426e-01 1.69663325e-01
-6.10722423e-01 6.73752427e-01 2.69712001e-01 9.41815972e-02
-6.40011847e-01 -4.42607015e-01 -2.88759589e-01 -9.66790259e-01
-6.99989974e-01 6.21825933e-01 4.71269429e-01 2.40559448e-02
2.34091640e-01 6.52891517e-01 -8.65083277e-01 1.32094637e-01
7.76488364e-01 1.01146460e+00 6.06280804e-01 5.75050533e-01
1.63058355e-01 6.70493901e-01 1.86642241e-02 -9.66126472e-03
5.51207550e-02 -3.80979568e-01 -8.18596721e-01 -3.61401021e-01
5.30767024e-01 -6.54480994e-01 -7.94289589e-01 1.04441488e+00
6.88091218e-01 7.88631886e-02 4.88762528e-01 -6.04671001e-01
-7.54636765e-01 4.78519648e-01 -3.04807484e-01 8.43054593e-01
-2.79799663e-02 2.56706446e-01 4.83217210e-01 -6.83887124e-01
1.19650578e+00 4.02344853e-01 7.78999627e-01 3.38689238e-01
-1.40897620e+00 -1.52418628e-01 -4.24931228e-01 2.82309979e-01
-1.06398523e+00 -3.47652197e-01 7.82780111e-01 -7.79641628e-01
9.94708121e-01 1.04026403e-02 8.06889951e-01 9.19180036e-01
7.62297571e-01 2.41470814e-01 1.55012548e+00 -1.87881812e-01
3.86148952e-02 -4.44007367e-01 4.47033010e-02 6.89493954e-01
1.28169358e-01 2.75843531e-01 -4.04343158e-01 -2.84140557e-01
1.07975245e+00 -3.43091004e-02 -7.04176724e-01 -5.94247937e-01
-2.15013409e+00 4.02846992e-01 2.53737599e-01 8.06828856e-01
-8.37604225e-01 1.68049887e-01 3.74220312e-01 1.55813381e-01
1.67588934e-01 5.05206108e-01 -2.37361476e-01 -1.35178879e-01
-1.19484258e+00 8.65208507e-02 2.55905807e-01 2.19235867e-01
4.53886151e-01 2.02135574e-02 -2.86569983e-01 6.53740227e-01
1.73847094e-01 8.70436788e-01 8.22278261e-01 -1.37244689e+00
-1.88354366e-02 -1.27498671e-01 1.13805167e-01 -1.08755398e+00
-8.92888904e-01 -5.69124818e-01 -9.44639623e-01 -1.45733014e-01
6.47511780e-01 1.61974862e-01 -6.50924325e-01 1.68952727e+00
3.87205720e-01 2.25292176e-01 -5.73805571e-01 1.36450791e+00
2.44703144e-01 -1.40528753e-02 -1.22644484e-01 -4.71139580e-01
1.06425858e+00 -5.35146773e-01 -8.30604970e-01 1.18276276e-01
5.87740481e-01 -4.69915509e-01 5.90113699e-01 2.78836370e-01
-1.16186345e+00 -1.46283329e-01 -9.77775991e-01 7.01194927e-02
4.71133888e-02 -2.69764483e-01 6.48303926e-01 3.54919374e-01
-1.10238159e+00 9.98265326e-01 -1.35762131e+00 -4.29702699e-02
2.48912990e-01 4.50059503e-01 -7.43409812e-01 -2.91621536e-01
-1.33274305e+00 1.21930611e+00 1.56892076e-01 2.66946733e-01
-9.78357315e-01 -1.33194315e+00 -3.61324579e-01 -6.09359384e-01
-2.53577381e-01 -6.24872088e-01 6.82593107e-01 -3.29336882e-01
-1.40799725e+00 8.04580450e-01 1.01006627e-01 -3.63494009e-01
6.49212599e-01 2.16474295e-01 -6.12050891e-01 1.01505339e+00
2.31720001e-01 5.11316001e-01 7.64045000e-01 -6.85583532e-01
3.10148954e-01 -7.13098645e-01 -2.62203127e-01 4.08977345e-02
-8.13111439e-02 -1.10286780e-01 2.65795767e-01 -5.47184169e-01
6.28333092e-01 -7.50155270e-01 -2.31028870e-01 4.30528164e-01
-2.73208559e-01 7.38658190e-01 4.68395591e-01 -1.21691501e+00
9.15496230e-01 -1.80349851e+00 3.17925215e-01 2.15094849e-01
8.19627464e-01 -4.48384471e-02 -1.44839987e-01 4.12924252e-02
-3.66381019e-01 -3.71768445e-01 -6.02680743e-01 2.36749440e-01
-1.92871034e-01 1.65459886e-02 1.01849943e-01 1.23714149e+00
-1.49621457e-01 1.15587163e+00 -1.31612122e+00 -3.24178576e-01
2.39729360e-01 8.49237442e-01 -2.96973586e-01 -3.29031399e-03
2.63033807e-01 1.16837335e+00 -3.49060923e-01 3.48948717e-01
8.33547294e-01 -2.22554713e-01 5.13470650e-01 -7.37413108e-01
-2.71319598e-01 3.29897195e-01 -6.95540249e-01 2.27506614e+00
-1.51382163e-01 4.81135309e-01 3.60018343e-01 -1.32737076e+00
7.59597182e-01 4.84597296e-01 1.34619212e+00 -1.14222646e+00
-2.64928471e-02 7.27189362e-01 3.87374103e-01 -7.12306142e-01
-7.29365274e-02 -5.79350591e-01 5.32847941e-01 8.82625461e-01
4.03812617e-01 -1.98638096e-01 1.29665732e-01 -6.49483502e-02
1.28590512e+00 5.05119115e-02 -1.30426943e-01 -7.76650548e-01
5.04186332e-01 -3.30803484e-01 1.72883004e-01 7.11494923e-01
-6.88492954e-01 6.47420347e-01 4.16181982e-01 -5.63651085e-01
-1.50104642e+00 -1.24191833e+00 -6.45858645e-01 2.67023176e-01
-3.15379985e-02 2.92560849e-02 -6.99929595e-01 -3.67471099e-01
1.34117395e-01 1.53912917e-01 -8.23985696e-01 -7.19474107e-02
-8.21531296e-01 -1.13408685e+00 5.42297781e-01 7.23196492e-02
2.93422490e-01 -7.37717330e-01 -7.67174065e-01 5.51829278e-01
-3.06138128e-01 -1.28688610e+00 -5.07276416e-01 1.27386555e-01
-1.21089458e+00 -1.01721632e+00 -1.16131985e+00 -3.08796167e-01
3.42354596e-01 1.53009042e-01 8.94953310e-01 -1.88848570e-01
-4.11887765e-01 3.87971818e-01 1.50133297e-01 4.88195091e-01
-4.40677136e-01 -4.83981557e-02 1.76722080e-01 1.65478721e-01
-1.68140903e-01 -1.21472561e+00 -1.12137461e+00 2.43684813e-01
-8.52419198e-01 -1.46450358e-03 5.03997922e-01 7.47836232e-01
9.28322494e-01 -3.26381117e-01 2.77373999e-01 -4.55368340e-01
4.26431119e-01 -5.36468506e-01 -1.71822280e-01 2.30609998e-01
-5.30255318e-01 1.55443698e-01 2.83015788e-01 -6.27230763e-01
-6.30753040e-01 -2.56175369e-01 1.33605361e-01 -3.77725720e-01
2.69634314e-02 3.22822094e-01 4.16403532e-01 -5.14324963e-01
6.34933233e-01 7.87283063e-01 4.09688294e-01 -3.01149726e-01
2.51776934e-01 1.54656351e-01 8.27281117e-01 -6.43519104e-01
3.01260501e-01 9.13520455e-01 3.02927107e-01 -8.77296865e-01
-4.45887059e-01 -2.42795318e-01 -1.24082363e+00 -4.66571659e-01
7.77014136e-01 -3.39937001e-01 -6.18078411e-01 7.44791210e-01
-9.88764644e-01 -5.09768128e-01 -3.57880205e-01 9.50376451e-01
-7.97407210e-01 7.87029624e-01 -7.64914632e-01 -1.46615714e-01
-4.31835443e-01 -1.35186970e+00 8.84153962e-01 -4.44917440e-01
-7.82593638e-02 -1.24555409e+00 3.50003690e-01 1.08546376e-01
1.02677298e+00 5.36016226e-01 1.03992140e+00 -1.19409919e-01
-4.53380287e-01 1.67435482e-02 -6.01950772e-02 2.08554551e-01
5.67219518e-02 -5.79719126e-01 -4.87104684e-01 -2.45282680e-01
5.35502315e-01 -7.38269091e-02 7.24059582e-01 9.28458273e-01
7.98353493e-01 1.76551297e-01 -4.52885814e-02 8.37303221e-01
1.20824277e+00 -1.71629759e-03 5.88163853e-01 2.58152932e-01
7.35147417e-01 5.50562203e-01 -3.34718823e-02 3.23289275e-01
3.05828124e-01 7.71111071e-01 1.16929904e-01 7.01396316e-02
-5.89776039e-01 4.34363149e-02 9.41497609e-02 1.16691899e+00
-2.74498582e-01 5.80356002e-01 -1.10657585e+00 6.18550479e-01
-1.24756253e+00 -1.07267153e+00 -3.56771678e-01 2.25870347e+00
9.42941129e-01 -2.15205863e-01 -4.81956154e-02 -6.38310239e-02
6.46527410e-01 3.98868620e-01 -8.65906596e-01 4.00461227e-01
-3.83016706e-01 2.06937090e-01 6.99693263e-01 7.94116497e-01
-7.46869981e-01 1.69041798e-01 7.09069109e+00 2.10436031e-01
-1.70645273e+00 7.63624012e-01 2.64254063e-01 -1.70802072e-01
-7.41119564e-01 -6.93168938e-02 -1.74304238e-03 5.31675100e-01
1.24311471e+00 -1.59844535e-03 8.90593469e-01 1.65291578e-02
2.86925405e-01 -7.43403062e-02 -7.01846004e-01 1.02503681e+00
-1.80668265e-01 -1.55542660e+00 -2.92074323e-01 1.96504638e-01
4.74080205e-01 6.91141546e-01 1.07364841e-01 -5.38997889e-01
-3.85879457e-01 -8.99233937e-01 7.51234949e-01 1.06305981e+00
1.21940601e+00 -1.99160874e-01 3.41639191e-01 2.45623216e-02
-6.99477911e-01 1.84511408e-01 -1.78997412e-01 4.55332935e-01
3.43922824e-01 1.12022364e+00 -5.13118327e-01 3.46551239e-01
4.97806638e-01 8.84464324e-01 -2.98239946e-01 8.73282194e-01
7.33039379e-02 2.48900875e-01 -3.61755133e-01 6.92822695e-01
4.23792154e-02 -4.24085319e-01 6.06028497e-01 9.48226810e-01
4.13711041e-01 1.53580636e-01 -1.96775272e-01 1.06373930e+00
2.85671711e-01 -1.79417208e-01 -2.60020107e-01 -2.18671128e-01
2.54431456e-01 1.25338328e+00 -8.49079370e-01 -1.14185736e-01
-1.65026143e-01 9.57825840e-01 1.87946260e-01 4.61050272e-01
-4.69261408e-01 -1.35409772e-01 3.37094307e-01 6.21150494e-01
1.35568976e-01 -7.74002016e-01 3.96802463e-03 -1.42465580e+00
2.02017680e-01 -5.52342296e-01 -2.01596886e-01 -6.30981803e-01
-1.15717411e+00 6.43155754e-01 -2.05424894e-02 -6.12171710e-01
-2.04225387e-02 -5.62637925e-01 -1.19355567e-01 1.16124618e+00
-1.69732368e+00 -8.04678440e-01 -1.98308095e-01 5.75726986e-01
-3.84670764e-01 1.31707877e-01 8.40918303e-01 5.82318246e-01
-8.92051905e-02 -7.99665675e-02 3.59354347e-01 -7.75388628e-02
7.31339514e-01 -1.32858157e+00 1.01154067e-01 5.27793705e-01
-3.37375075e-01 7.89267421e-01 6.13346875e-01 -6.46902263e-01
-1.69088697e+00 -5.50876021e-01 7.24827170e-01 -2.00270578e-01
1.11225355e+00 -1.28171176e-01 -1.11801767e+00 4.26530778e-01
-1.92566663e-01 8.58108878e-01 5.14217556e-01 -5.76379180e-01
-1.61824510e-01 -1.28907368e-01 -1.34992349e+00 -6.98527619e-02
9.26147819e-01 -1.01884782e+00 -2.84701645e-01 5.11286855e-01
3.94635946e-01 -6.10204160e-01 -1.69393826e+00 1.15101725e-01
8.85526955e-01 -1.10818946e+00 1.20791101e+00 -3.30377191e-01
1.87601507e-01 -1.76003322e-01 -2.57790238e-01 -1.17770517e+00
-4.86824691e-01 -6.07513428e-01 -3.42602015e-01 3.80547851e-01
-1.08432934e-01 -8.09147716e-01 6.29418731e-01 5.57040334e-01
-2.39506364e-01 -6.48762524e-01 -1.30619133e+00 -5.35619497e-01
3.08094919e-01 -3.11182499e-01 3.80044132e-01 1.20203447e+00
-9.60943028e-02 -3.55858892e-01 -2.07047641e-01 1.12825841e-01
1.08547878e+00 6.49896311e-03 -2.81562567e-01 -1.03719449e+00
-3.64205450e-01 -3.49164367e-01 -3.79450887e-01 -8.22597086e-01
5.47460914e-02 -1.42104530e+00 -2.81856805e-01 -1.16041052e+00
1.10361777e-01 -6.95753336e-01 -7.09656894e-01 5.03806174e-02
3.51331085e-01 2.89805859e-01 -1.68313220e-01 6.99401617e-01
-1.65236562e-01 4.69512880e-01 2.06705546e+00 -2.10053608e-01
1.28276974e-01 -7.99373746e-01 -1.76431313e-01 2.51443416e-01
4.95546728e-01 -5.62450111e-01 -4.03700233e-01 -6.21637762e-01
-3.28003466e-02 6.86462581e-01 5.68120420e-01 -1.01006615e+00
2.31100500e-01 -8.96050334e-02 3.78216058e-01 -1.37719408e-01
1.50884181e-01 -6.23339474e-01 4.07622099e-01 7.09250391e-01
-3.91895890e-01 1.06138526e-03 -1.53243408e-01 2.64308453e-01
-7.98813477e-02 -1.01898998e-01 9.65907276e-01 -1.93270281e-01
-4.12010998e-01 7.91499436e-01 -1.67575195e-01 1.53162658e-01
5.72331429e-01 -9.63672251e-02 -2.65273273e-01 7.47579113e-02
-1.09314096e+00 -4.63098139e-01 4.18278784e-01 3.44134867e-02
6.48822546e-01 -1.23950720e+00 -7.10146010e-01 2.91340828e-01
-3.95570338e-01 -4.94161427e-01 9.21424448e-01 1.90076971e+00
-8.55995297e-01 6.24971867e-01 -8.00834954e-01 -7.49057591e-01
-2.84832329e-01 4.40772593e-01 1.19136989e+00 -4.90892678e-01
-9.59150553e-01 4.02840227e-01 -1.36320248e-01 -6.15090966e-01
-4.94789124e-01 -3.34719032e-01 1.65519431e-01 -1.88228011e-01
7.46336520e-01 2.99873382e-01 4.14948940e-01 -8.46797764e-01
-5.88401616e-01 6.32460177e-01 4.31077555e-02 -2.12053940e-01
1.57786202e+00 -4.16054547e-01 -4.16238934e-01 3.03198457e-01
1.32247603e+00 -2.54540265e-01 -1.47451282e+00 -3.89658868e-01
-2.19599064e-02 -4.36169624e-01 7.41647422e-01 -9.91976619e-01
-1.37132394e+00 1.07050848e+00 9.19346154e-01 -8.33011493e-02
7.95670688e-01 -1.80459306e-01 1.05364001e+00 -8.02598223e-02
6.38819337e-01 -5.72266877e-01 -1.45670265e-01 2.38141030e-01
9.47321713e-01 -8.73203039e-01 -3.66047248e-02 -1.23974904e-01
-2.78025120e-01 1.39226031e+00 -1.45592287e-01 -5.11880443e-02
7.55264282e-01 4.70912337e-01 2.17283219e-02 -4.94282037e-01
-3.27779263e-01 5.15312016e-01 2.59429097e-01 1.05413592e+00
5.18370628e-01 7.98148736e-02 -3.41581762e-01 9.10602808e-02
9.72238034e-02 2.75466442e-01 4.23034370e-01 6.30384386e-01
-1.61897793e-01 -1.05465770e+00 -1.65925041e-01 4.68598187e-01
-5.00260770e-01 -5.31390421e-02 3.95599365e-01 4.97876495e-01
7.11259991e-02 1.35823220e-01 1.78533178e-02 4.40339744e-02
1.81276962e-01 -5.69160916e-02 1.18901098e+00 -9.64336395e-02
-2.49968052e-01 -1.03628159e-01 -2.43506372e-01 -9.22414005e-01
-7.24686027e-01 -9.51733947e-01 -1.30598664e+00 -3.04086059e-01
2.87619919e-01 -9.21293497e-02 9.43204284e-01 9.43580091e-01
3.00327957e-01 4.21386898e-01 3.77745718e-01 -1.01905000e+00
-5.51858783e-01 -7.59778917e-01 -1.18382668e+00 4.89653975e-01
5.73481739e-01 -8.94792497e-01 -5.63717522e-02 -9.46292654e-02]
|
[13.542041778564453, -2.398016929626465]
|
fb8ee329-3c32-4a60-9577-c6f68ac78f29
|
convolutional-neural-network-with-pruning
|
2101.05996
| null |
https://arxiv.org/abs/2101.05996v1
|
https://arxiv.org/pdf/2101.05996v1.pdf
|
Convolutional Neural Network with Pruning Method for Handwritten Digit Recognition
|
CNN model is a popular method for imagery analysis, so it could be utilized to recognize handwritten digits based on MNIST datasets. For higher recognition accuracy, various CNN models with different fully connected layer sizes are exploited to figure out the relationship between the CNN fully connected layer size and the recognition accuracy. Inspired by previous pruning work, we performed pruning methods of distinctiveness on CNN models and compared the pruning performance with NN models. For better pruning performances on CNN, the effect of angle threshold on the pruning performance was explored. The evaluation results show that: for the fully connected layer size, there is a threshold, so that when the layer size increases, the recognition accuracy grows if the layer size smaller than the threshold, and falls if the layer size larger than the threshold; the performance of pruning performed on CNN is worse than on NN; as pruning angle threshold increases, the fully connected layer size and the recognition accuracy decreases. This paper also shows that for CNN models trained by the MNIST dataset, they are capable of handwritten digit recognition and achieve the highest recognition accuracy with fully connected layer size 400. In addition, for same dataset MNIST, CNN models work better than big, deep, simple NN models in a published paper.
|
['Mengyu Chen']
|
2021-01-15
| null | null | null | null |
['handwritten-digit-recognition']
|
['computer-vision']
|
[-1.57224059e-01 4.46434096e-02 -2.31137857e-01 -1.49787232e-01
8.26034844e-01 -1.12149335e-01 4.21934612e-02 -1.25834554e-01
-6.20740891e-01 5.97891986e-01 -1.19347580e-01 -2.16227695e-01
-2.90690005e-01 -1.21229374e+00 -3.39794636e-01 -6.64659619e-01
9.22726234e-04 1.32445507e-02 4.97928441e-01 -8.20417255e-02
5.56351364e-01 1.06572127e+00 -1.74497068e+00 4.70798135e-01
5.14612436e-01 1.44147146e+00 1.18709028e-01 5.49629390e-01
-3.28175843e-01 9.77575898e-01 -9.25146699e-01 -2.79839694e-01
6.71615124e-01 -5.38168214e-02 -7.42815256e-01 1.47541910e-02
3.83936077e-01 -1.72662646e-01 -5.70861042e-01 1.05667222e+00
2.63642132e-01 -1.37194768e-01 5.68248451e-01 -1.03051710e+00
-4.36123878e-01 8.36744368e-01 -3.56196135e-01 4.24571723e-01
-3.23227257e-01 1.89395510e-02 2.95932293e-01 -5.81307232e-01
4.86251324e-01 9.51548696e-01 9.15306866e-01 3.20615858e-01
-9.64372873e-01 -8.73168588e-01 6.28916919e-02 1.85566828e-01
-1.54284489e+00 2.37158507e-01 4.65999007e-01 -3.18639725e-01
1.16702807e+00 3.27676266e-01 9.77165043e-01 6.50148451e-01
3.17374259e-01 2.45148033e-01 1.04295063e+00 -5.44252574e-01
2.02125728e-01 1.11379653e-01 6.41830385e-01 6.69425786e-01
8.38746965e-01 5.75032346e-02 -2.45050967e-01 3.56610984e-01
1.02000678e+00 2.56867051e-01 -1.95376247e-01 1.68963730e-01
-7.02263594e-01 7.41742074e-01 5.85806847e-01 6.73609912e-01
-3.20794284e-01 5.26183806e-02 3.62929553e-01 5.50836742e-01
-9.70215350e-02 7.08471119e-01 -4.64806855e-01 -1.08170547e-01
-1.33655417e+00 -2.45689135e-03 9.98592913e-01 7.87092328e-01
8.18351269e-01 3.43003422e-01 -1.71859220e-01 8.71461689e-01
-2.02350199e-01 1.87195837e-01 7.44623303e-01 -5.62375784e-01
3.44807625e-01 1.49361598e+00 -6.08349383e-01 -1.09831727e+00
-5.03631592e-01 -5.91017306e-01 -1.29137099e+00 7.06769586e-01
6.59090340e-01 -1.08347006e-01 -1.33692563e+00 8.12957466e-01
-3.45093757e-01 -3.14594388e-01 2.04036236e-01 9.90610242e-01
1.06251359e+00 4.56270188e-01 -7.54985809e-02 1.19037472e-01
1.50601733e+00 -8.78919721e-01 -4.43715960e-01 -5.16257733e-02
4.24802333e-01 -1.02106571e+00 8.67012262e-01 8.10994744e-01
-8.04122329e-01 -9.80186522e-01 -1.35709739e+00 4.14734215e-01
-6.63539648e-01 6.91004276e-01 6.93775356e-01 8.30488205e-01
-7.45110273e-01 1.02658451e+00 -7.03988671e-01 -6.46251976e-01
5.36007226e-01 4.31934923e-01 -3.04042220e-01 -9.97642204e-02
-7.62814403e-01 9.94299591e-01 1.07146823e+00 3.40334803e-01
-5.97172916e-01 -2.93473661e-01 -3.30408275e-01 3.23618084e-01
-1.57163918e-01 -1.04266331e-01 6.61196291e-01 -1.11514843e+00
-1.43184984e+00 6.32112801e-01 4.24642444e-01 -8.11710060e-01
5.02170086e-01 -8.18267688e-02 -5.01240909e-01 1.91875920e-01
-5.67997754e-01 7.73171246e-01 5.62030733e-01 -7.90676832e-01
-4.31070805e-01 -3.05182576e-01 3.59840989e-01 -1.34946197e-01
-7.56953299e-01 -1.66580141e-01 -1.61108971e-01 -5.08380473e-01
4.69213068e-01 -6.01190567e-01 -2.30165511e-01 1.20784268e-01
-5.20791233e-01 -8.66296291e-02 1.28043461e+00 -3.62821221e-01
1.17323959e+00 -2.16125989e+00 -3.55708510e-01 5.44166744e-01
2.28881985e-01 7.22171187e-01 -9.20358673e-02 1.04588076e-01
4.69218232e-02 4.03424442e-01 -1.80413271e-03 4.41550851e-01
-7.33529508e-01 4.68541980e-01 1.17412411e-01 1.05590940e-01
4.11196917e-01 3.68216604e-01 -2.48367131e-01 -4.81151253e-01
1.04209125e-01 3.97962809e-01 -4.18689162e-01 -1.64926514e-01
1.73780322e-01 -2.08937183e-01 -2.58120239e-01 7.62251854e-01
1.08787882e+00 5.46807051e-02 -5.42929210e-02 -4.70081925e-01
-4.27336186e-01 -3.55437070e-01 -1.40803182e+00 9.72615302e-01
-1.81608185e-01 9.96436477e-01 -2.30955705e-01 -1.02630115e+00
1.84498179e+00 -5.28086498e-02 -9.56223309e-02 -5.69931984e-01
3.54202390e-01 2.43347228e-01 3.68492752e-01 -5.29420078e-01
4.63713229e-01 2.94690818e-01 6.36484623e-01 2.66453587e-02
1.17995165e-01 -1.74290299e-01 4.56969976e-01 -1.50232449e-01
1.00931406e+00 -2.29727685e-01 7.66001120e-02 -4.60774481e-01
6.05781257e-01 3.06396246e-01 3.66580099e-01 7.24121571e-01
9.48434174e-02 6.27737463e-01 7.54922271e-01 -1.01829374e+00
-1.20692968e+00 -4.50369269e-01 -6.00707471e-01 4.25458103e-01
1.66178420e-01 2.19977628e-02 -7.57596195e-01 -3.75958562e-01
8.95990580e-02 1.12512320e-01 -6.28135026e-01 -1.56772256e-01
-6.13378942e-01 -7.91962266e-01 1.10836029e+00 5.69993794e-01
1.52832949e+00 -1.35662687e+00 -9.19393539e-01 3.94890122e-02
2.52677321e-01 -1.09818006e+00 4.49185580e-01 5.01361251e-01
-1.53556037e+00 -1.45878196e+00 -7.44075894e-01 -9.60625470e-01
8.33541811e-01 -1.62482560e-02 8.02332342e-01 6.66073680e-01
-6.69918180e-01 -1.40639558e-01 -5.68754375e-01 -5.92705607e-01
-1.35739833e-01 2.85012126e-01 -3.02023828e-01 -4.20041174e-01
3.75881165e-01 -3.88674766e-01 -5.80704689e-01 3.19147766e-01
-8.78402770e-01 -1.77229509e-01 8.95300269e-01 6.13745630e-01
4.53654349e-01 4.24827248e-01 8.06557983e-02 -5.81992209e-01
5.33030748e-01 8.70325863e-02 -5.56034625e-01 2.26830333e-01
-6.47990286e-01 2.05571011e-01 8.56104910e-01 -7.74701595e-01
-6.12389565e-01 -9.78663713e-02 6.43395036e-02 -4.57954437e-01
-2.57697612e-01 4.70991611e-01 1.25246346e-01 -5.26690364e-01
8.41865361e-01 1.96041778e-01 2.15407535e-02 -5.79389811e-01
-1.74774230e-01 5.92481613e-01 2.52821743e-01 -8.76792073e-02
3.35060269e-01 3.41006011e-01 2.25780964e-01 -1.02947140e+00
-1.65488362e-01 2.45333575e-02 -8.42657387e-01 -3.11450362e-01
7.66471744e-01 -4.16559696e-01 -6.17153883e-01 6.95032418e-01
-1.11527300e+00 -9.27104279e-02 -3.99386376e-01 7.82759726e-01
1.85414702e-01 2.60698289e-01 -6.84492409e-01 -7.33481944e-01
-5.07140160e-01 -9.95818198e-01 1.40598163e-01 5.65723598e-01
-2.31618378e-02 -7.21435308e-01 -3.53777438e-01 -9.66734588e-02
5.45817673e-01 1.76546201e-01 1.07615232e+00 -8.25712025e-01
-5.03202260e-01 -5.01872599e-01 -6.99183524e-01 8.27758431e-01
-2.33150199e-01 4.44747955e-01 -6.63094103e-01 2.30152570e-02
-1.76342756e-01 -1.80046231e-01 1.18169677e+00 3.15642774e-01
1.34452963e+00 -4.86771792e-01 -2.80688375e-01 6.98733449e-01
1.81318843e+00 3.82983983e-01 1.38251674e+00 6.83505595e-01
3.78861696e-01 2.49866739e-01 2.01441407e-01 3.29843074e-01
-4.04823035e-01 1.97796989e-02 6.34316623e-01 -3.23418796e-01
-3.89837414e-01 1.55561477e-01 -8.49696621e-02 5.31532347e-01
-6.54871762e-01 -4.60244790e-02 -9.59928095e-01 1.95553988e-01
-1.23613143e+00 -9.08383131e-01 -2.81859905e-01 2.00673723e+00
4.67546403e-01 4.78577882e-01 -2.36170903e-01 6.92233205e-01
6.62507355e-01 -4.30076942e-02 -4.12919760e-01 -8.23749244e-01
-5.18061578e-01 5.08871555e-01 7.67385185e-01 1.04746027e-02
-8.80333006e-01 8.08847964e-01 6.53457355e+00 7.75693297e-01
-1.72050202e+00 -3.31553251e-01 7.03398705e-01 -1.31641477e-01
4.19916242e-01 -9.30987448e-02 -1.07870543e+00 2.51212269e-01
5.20071089e-01 4.93713856e-01 1.73421845e-01 1.04777420e+00
-3.59344259e-02 -4.77877378e-01 -8.16368639e-01 9.73392129e-01
-1.02597579e-01 -1.44171774e+00 6.02499306e-01 1.53737113e-01
6.53119445e-01 -2.14879274e-01 -3.77976835e-01 2.10729152e-01
-1.23309810e-02 -1.27240539e+00 5.80865443e-01 6.60604894e-01
4.75975364e-01 -9.41713512e-01 1.39813626e+00 3.94280016e-01
-1.10382724e+00 -5.09787202e-01 -1.05676007e+00 -5.49180925e-01
-4.97600287e-01 8.13096762e-01 -5.25599301e-01 3.90959203e-01
9.40892577e-01 3.84149909e-01 -8.22124839e-01 1.36273015e+00
-2.27836251e-01 4.95228589e-01 -3.00335705e-01 -3.48599643e-01
3.58954728e-01 -1.50763094e-01 2.47865483e-01 1.44277966e+00
3.13701570e-01 2.16105953e-01 -1.09333977e-01 9.36548829e-01
-1.42711420e-02 1.65955499e-01 -6.05325520e-01 -1.75885007e-01
6.00404203e-01 1.24521661e+00 -1.22733486e+00 -4.15704638e-01
-6.81410134e-02 6.61525846e-01 1.67927757e-01 2.84223616e-01
-4.87665027e-01 -8.82225871e-01 3.95234168e-01 1.79632649e-01
5.77624500e-01 -1.12934239e-01 -9.46057498e-01 -6.77580118e-01
8.06783289e-02 -6.23626828e-01 1.09153397e-01 -5.97509682e-01
-6.71070039e-01 9.34973419e-01 -2.23554090e-01 -1.21459270e+00
5.23118138e-01 -1.14275658e+00 -9.04373407e-01 7.32641518e-01
-1.00072253e+00 -9.06989753e-01 -6.38287485e-01 4.62025255e-01
4.11034465e-01 -5.19981623e-01 8.39051008e-01 1.76212206e-01
-8.62860203e-01 6.78579092e-01 -1.40922949e-01 6.39145315e-01
-9.39893797e-02 -3.79038781e-01 -1.64517447e-01 8.44005764e-01
-1.03222884e-01 5.79203963e-01 2.21541986e-01 -5.74340224e-01
-9.40679133e-01 -9.94860470e-01 5.61637938e-01 2.84083188e-01
5.85794486e-02 -1.33404573e-02 -8.84259045e-01 2.42224246e-01
4.13155071e-02 3.43536325e-02 3.08815360e-01 -2.71247923e-02
-3.92379940e-01 -3.36763650e-01 -1.60387886e+00 3.84229213e-01
7.42297947e-01 -6.50652125e-02 -3.49799991e-01 -5.53324781e-02
2.69773245e-01 -1.40461877e-01 -9.98480558e-01 5.41444480e-01
9.41867530e-01 -1.17638814e+00 7.65611947e-01 -4.64927793e-01
7.02058196e-01 -1.22161537e-01 5.00444733e-02 -6.90406680e-01
-3.56662512e-01 3.72672766e-01 -6.97540417e-02 9.89291310e-01
6.08721077e-01 -6.83232844e-01 1.16569936e+00 2.64470190e-01
6.01701885e-02 -1.17346680e+00 -8.07045758e-01 -1.17422426e+00
-9.04659033e-02 -1.60008281e-01 6.81886554e-01 6.62757277e-01
-4.36937183e-01 -2.89704412e-01 1.92955181e-01 1.48114964e-01
2.18725696e-01 -1.70190856e-01 3.79114419e-01 -1.59704983e+00
1.97281331e-01 -7.54141390e-01 -1.19075906e+00 -8.60832036e-01
-5.40083647e-01 -5.23926437e-01 -4.74409431e-01 -1.66403508e+00
7.99361169e-02 -5.36446750e-01 -2.00483933e-01 7.20873117e-01
5.34522533e-01 6.63704634e-01 4.51863587e-01 4.58754182e-01
1.02602802e-01 -1.94685057e-01 1.25712049e+00 -3.45346004e-01
-3.57697159e-01 -9.46184769e-02 -2.36949027e-01 9.88254070e-01
8.67393732e-01 -4.74703819e-01 -1.28017977e-01 -4.24032271e-01
-1.35177955e-01 -4.35152501e-01 3.63073170e-01 -1.76820540e+00
4.29052055e-01 6.57384023e-02 1.09207916e+00 -8.01815927e-01
9.00183544e-02 -9.49573398e-01 1.48419226e-02 1.19864047e+00
-1.19126819e-01 -1.45545825e-01 4.30152237e-01 -4.14393581e-02
-3.66615921e-01 -8.87675047e-01 9.25239980e-01 -3.64592344e-01
-8.46324325e-01 2.09883638e-02 -6.31892860e-01 -3.04760844e-01
9.81939256e-01 -1.32459319e+00 -1.22592233e-01 1.48964033e-01
-7.38556564e-01 -2.15138391e-01 1.73347324e-01 2.26590663e-01
9.26884055e-01 -1.17565882e+00 -5.21238506e-01 4.19568747e-01
-2.19960272e-01 3.29000093e-02 -6.68652654e-02 5.90422392e-01
-1.17737877e+00 5.22693932e-01 -8.35160792e-01 -4.77064252e-01
-1.47187543e+00 2.01158330e-01 7.06787169e-01 -3.13872844e-01
-6.11937404e-01 9.54197466e-01 -4.05513078e-01 -3.15625399e-01
6.02199972e-01 -8.28358948e-01 -7.37177372e-01 5.15526608e-02
6.42988801e-01 8.78666282e-01 3.38216633e-01 -1.09456070e-01
-2.13772520e-01 8.96668196e-01 -3.08571034e-03 3.68967474e-01
1.34512079e+00 5.19846678e-01 -1.87264219e-01 1.02299385e-01
9.34532940e-01 -3.96192163e-01 -1.00899005e+00 1.78254515e-01
-1.00640111e-01 -4.06350583e-01 7.81912655e-02 -8.50716412e-01
-1.47048008e+00 8.38831067e-01 9.46470559e-01 2.81067640e-01
1.29666591e+00 -5.23666799e-01 3.70091915e-01 7.87602961e-01
3.79874259e-01 -1.21144474e+00 1.55510604e-01 9.00561869e-01
1.00545704e+00 -7.96824098e-01 4.16189343e-01 -4.18060988e-01
-5.26371241e-01 1.92750633e+00 9.97981250e-01 -5.63787818e-01
8.35529506e-01 5.18288791e-01 -1.15484633e-01 -3.56420904e-01
-5.05032651e-02 3.35338935e-02 1.55613199e-01 5.44656217e-01
3.44772130e-01 -1.06227137e-01 -4.76965040e-01 6.46768451e-01
-5.80061555e-01 2.58839607e-01 4.70287234e-01 7.86241531e-01
-7.83909738e-01 -8.31021905e-01 -5.05888164e-01 7.79567838e-01
-2.28336781e-01 -2.08322868e-01 -6.34517312e-01 1.15648937e+00
7.57058799e-01 7.02572584e-01 5.99627435e-01 -5.91268420e-01
6.88536346e-01 -4.09439914e-02 5.71972668e-01 -3.36987793e-01
-9.68916953e-01 -7.02546060e-01 -3.24051529e-02 -3.80645663e-01
-2.21065283e-01 1.13615915e-01 -1.30482960e+00 -7.45512068e-01
-7.23703206e-01 1.37172252e-01 8.54038954e-01 8.05602372e-01
3.11507024e-02 5.39386809e-01 2.93438673e-01 -6.75307691e-01
-3.57154340e-01 -1.24753964e+00 -7.08635032e-01 3.15851271e-02
-1.57341421e-01 -2.69341350e-01 -3.03337425e-01 -7.10899383e-02]
|
[8.602825164794922, 2.917962074279785]
|
011f1ceb-b483-475c-a99c-2ba24eb7d35b
|
precise-facial-landmark-detection-by
|
2303.07840
| null |
https://arxiv.org/abs/2303.07840v1
|
https://arxiv.org/pdf/2303.07840v1.pdf
|
Precise Facial Landmark Detection by Reference Heatmap Transformer
|
Most facial landmark detection methods predict landmarks by mapping the input facial appearance features to landmark heatmaps and have achieved promising results. However, when the face image is suffering from large poses, heavy occlusions and complicated illuminations, they cannot learn discriminative feature representations and effective facial shape constraints, nor can they accurately predict the value of each element in the landmark heatmap, limiting their detection accuracy. To address this problem, we propose a novel Reference Heatmap Transformer (RHT) by introducing reference heatmap information for more precise facial landmark detection. The proposed RHT consists of a Soft Transformation Module (STM) and a Hard Transformation Module (HTM), which can cooperate with each other to encourage the accurate transformation of the reference heatmap information and facial shape constraints. Then, a Multi-Scale Feature Fusion Module (MSFFM) is proposed to fuse the transformed heatmap features and the semantic features learned from the original face images to enhance feature representations for producing more accurate target heatmaps. To the best of our knowledge, this is the first study to explore how to enhance facial landmark detection by transforming the reference heatmap information. The experimental results from challenging benchmark datasets demonstrate that our proposed method outperforms the state-of-the-art methods in the literature.
|
['Wenwen Min', 'Ping Xiong', 'Hang Sun', 'Linlin Shen', 'Zhihui Lai', 'Jie zhou', 'Jun Liu', 'Jun Wan']
|
2023-03-14
| null | null | null | null |
['facial-landmark-detection']
|
['computer-vision']
|
[ 1.08845443e-01 5.65169845e-03 -8.38154405e-02 -7.38625467e-01
-6.44219398e-01 8.89943819e-03 5.54350615e-01 -3.46134394e-01
-1.03637442e-01 2.12375537e-01 -3.91021296e-02 3.92509729e-01
-8.97179693e-02 -6.93276823e-01 -4.87234622e-01 -9.07624483e-01
3.76702249e-01 3.05387020e-01 1.23826303e-01 -7.11692497e-02
3.40259880e-01 5.57478726e-01 -1.61851108e+00 -3.19211185e-02
8.74702990e-01 1.36183298e+00 9.00226459e-02 -3.37005049e-01
-2.94480234e-01 3.63834262e-01 -2.36852691e-01 -2.34781757e-01
3.31711382e-01 -3.30466121e-01 -4.21396762e-01 1.46783795e-02
5.68172812e-01 -2.51427144e-01 -1.53049618e-01 1.21977580e+00
5.80966771e-01 1.30499274e-01 5.39339185e-01 -1.55372822e+00
-6.15478277e-01 1.52701959e-02 -1.04404819e+00 -1.52092129e-01
4.56521660e-01 -1.80485070e-01 5.45258641e-01 -1.39172208e+00
6.24243498e-01 1.59216225e+00 4.92839724e-01 5.31398416e-01
-9.30074632e-01 -1.25542617e+00 2.01927111e-01 3.66519451e-01
-1.91095614e+00 -6.63022995e-01 1.17334926e+00 -1.59805790e-01
3.36231440e-01 3.07355165e-01 5.93427956e-01 5.51254809e-01
-5.11790700e-02 6.71393633e-01 1.02999032e+00 -3.05123597e-01
1.04772791e-01 1.60165682e-01 -4.08921361e-01 1.24184251e+00
-4.70453547e-03 -5.24470657e-02 -5.53669870e-01 -1.89567626e-01
8.71783495e-01 3.93585920e-01 -3.83664340e-01 -5.40707946e-01
-6.88784957e-01 6.26209319e-01 9.42967653e-01 2.25031927e-01
-4.19317663e-01 -1.05062596e-01 -1.73499957e-02 -5.86719252e-02
3.78652126e-01 -6.04191199e-02 -2.32493997e-01 2.67170072e-01
-9.06324863e-01 -1.77796513e-01 3.23794335e-01 9.19151664e-01
1.33729112e+00 -2.28846043e-01 -1.78704113e-01 9.17714715e-01
8.72364938e-01 6.24751687e-01 4.85738069e-01 -6.70306265e-01
4.27770346e-01 1.12593162e+00 -1.03825182e-01 -1.53981888e+00
-5.37260234e-01 -2.42536739e-02 -6.70257807e-01 3.88425253e-02
2.35259071e-01 1.22264683e-01 -1.07796025e+00 1.63452041e+00
7.29700506e-01 5.01002491e-01 -3.24836582e-01 1.12716305e+00
9.42055583e-01 7.35103250e-01 1.02222569e-01 -2.05547303e-01
1.25596189e+00 -8.22172284e-01 -7.09837019e-01 -2.57046819e-01
3.86386812e-01 -9.47854102e-01 8.51996303e-01 -3.00733168e-02
-7.58415461e-01 -6.11660063e-01 -9.43720758e-01 -1.16726689e-01
-3.20434421e-01 5.10494113e-01 4.62958902e-01 4.28476572e-01
-9.83534634e-01 3.61447409e-02 -8.77160907e-01 -3.06082219e-01
7.63087392e-01 5.36035657e-01 -5.87004483e-01 -3.67536604e-01
-9.21313465e-01 7.85204351e-01 5.68483323e-02 6.64764345e-01
-6.50983751e-01 -6.83128774e-01 -1.06020927e+00 -7.98440278e-02
4.37357455e-01 -2.84063011e-01 7.66147614e-01 -9.03691053e-01
-1.44309556e+00 7.32932389e-01 -4.59534645e-01 4.86760765e-01
2.78134584e-01 1.25217158e-03 -3.70383382e-01 1.26991108e-01
4.40178663e-02 7.61398435e-01 1.15666091e+00 -1.22797930e+00
-6.13997936e-01 -8.21296453e-01 -3.90609324e-01 2.17671379e-01
-4.84793603e-01 5.88325113e-02 -6.75134182e-01 -3.71569216e-01
5.36591649e-01 -7.48954952e-01 -6.00092225e-02 3.19089413e-01
-2.79699117e-01 -4.36478198e-01 1.14780426e+00 -5.49991965e-01
1.06599569e+00 -2.20581055e+00 2.59109944e-01 6.31718576e-01
8.99213776e-02 3.28147203e-01 -3.04259002e-01 5.31322174e-02
2.13822499e-02 -7.82851800e-02 -5.92700206e-02 -4.93438214e-01
-7.32048750e-02 2.11324230e-01 -1.11592188e-01 6.75234079e-01
3.39292645e-01 9.89710510e-01 -8.68751585e-01 -8.55821669e-01
4.66567010e-01 9.33060110e-01 -3.56147617e-01 2.16765583e-01
1.91608727e-01 3.50806504e-01 -8.29028249e-01 1.07194412e+00
1.01673996e+00 -4.00360813e-03 -3.67139727e-02 -5.15194535e-01
-1.24804704e-02 -2.92645097e-01 -1.22399974e+00 1.73662603e+00
-4.31782693e-01 1.66985512e-01 2.36854516e-02 -6.84422314e-01
1.47670805e+00 1.80449009e-01 5.87057054e-01 -8.88952136e-01
3.32797050e-01 2.86727220e-01 -3.61808866e-01 -2.71784514e-01
1.57895759e-01 2.46318243e-02 2.78786838e-01 2.12178692e-01
5.46033420e-02 3.53193022e-02 -3.23853731e-01 -9.55114514e-02
3.55597973e-01 2.60214835e-01 1.14456400e-01 -1.97898731e-01
9.70106184e-01 -4.84025300e-01 9.30914640e-01 -7.14687854e-02
-3.04661274e-01 5.22065520e-01 3.00498217e-01 -7.31276691e-01
-6.62643075e-01 -8.02364349e-01 -2.92972594e-01 1.07558954e+00
4.65263277e-01 -5.38363636e-01 -8.88549089e-01 -8.90892148e-01
8.33111256e-03 -1.99958365e-02 -9.81974304e-01 -3.28483850e-01
-5.56679189e-01 -7.13389099e-01 1.34686813e-01 4.73611146e-01
7.43253231e-01 -8.41680884e-01 -3.35697979e-01 -7.09319413e-02
-9.41447914e-02 -8.74907792e-01 -8.38955283e-01 -4.85308886e-01
-4.31322873e-01 -1.19216692e+00 -6.22319281e-01 -1.05538261e+00
1.41880476e+00 2.96057582e-01 4.11271065e-01 4.37041759e-01
-4.41615731e-01 1.15121596e-01 -2.57542968e-01 -2.30986491e-01
1.45189956e-01 -2.20233023e-01 -3.83897908e-02 6.97063386e-01
5.12319446e-01 -3.05571795e-01 -7.89277434e-01 7.47467041e-01
-6.66950285e-01 1.49152562e-01 7.58380115e-01 9.15835142e-01
7.42634594e-01 -3.25584598e-02 2.60119796e-01 -5.58331251e-01
9.37866345e-02 -2.99124569e-01 -6.42215908e-01 4.33673799e-01
-5.77569723e-01 -4.51095812e-02 5.00275135e-01 -2.83073932e-01
-1.03668857e+00 4.18103755e-01 -7.08702579e-02 -5.55345595e-01
1.04137406e-01 2.44228885e-01 -4.35436428e-01 -6.10618770e-01
2.44629055e-01 4.18580443e-01 2.43973702e-01 -5.18190026e-01
3.39945465e-01 4.77634251e-01 4.69310880e-01 -4.89656568e-01
1.09328473e+00 5.65114677e-01 3.39170277e-01 -4.75332946e-01
-6.36786461e-01 -3.57096702e-01 -8.29137206e-01 -3.25830489e-01
6.55393720e-01 -7.98717558e-01 -6.35916293e-01 4.53463912e-01
-8.80525947e-01 2.08345607e-01 2.21710056e-01 1.82444245e-01
-3.09877843e-01 2.61879593e-01 -1.72818199e-01 -7.12341547e-01
-3.63967836e-01 -1.30151665e+00 1.54693294e+00 7.70768762e-01
2.99246937e-01 -6.98330402e-01 -2.38582566e-01 1.24920890e-01
5.07956803e-01 3.99152666e-01 5.42373717e-01 -1.65694714e-01
-6.51638627e-01 -4.27711993e-01 -4.66240555e-01 -1.28789283e-02
5.00781298e-01 1.36690289e-01 -9.47297335e-01 -3.35089982e-01
-1.92551419e-01 -1.98234022e-01 7.25403905e-01 1.41098604e-01
1.13417280e+00 -4.18646544e-01 -5.13490438e-01 9.08934712e-01
1.23569798e+00 1.86508313e-01 5.97885549e-01 2.06155285e-01
9.37657773e-01 5.63500047e-01 9.18808818e-01 4.96969491e-01
6.06499553e-01 1.01752639e+00 4.91404027e-01 -2.95673609e-01
-1.04599409e-01 -5.65135241e-01 3.17543596e-01 5.67060888e-01
-1.46123543e-01 4.73483860e-01 -7.21387327e-01 1.53673261e-01
-1.84791434e+00 -5.98678052e-01 2.65855014e-01 2.08689594e+00
7.76433229e-01 -3.56682122e-01 -1.26578614e-01 5.28164729e-02
7.77871907e-01 1.80423662e-01 -5.07983744e-01 4.83587384e-02
2.13087916e-01 5.19866347e-02 2.13008150e-02 4.26260024e-01
-1.08013368e+00 1.03918934e+00 5.15548754e+00 9.44083095e-01
-1.34136438e+00 1.08760558e-01 5.75955212e-01 1.73025817e-01
-2.06231564e-01 -3.36257331e-02 -8.99914265e-01 5.64302027e-01
1.46557823e-01 -3.46622765e-02 4.37453747e-01 9.89259958e-01
-6.07989803e-02 1.35771021e-01 -7.67520785e-01 1.26208866e+00
4.02114451e-01 -8.70179832e-01 1.66493252e-01 2.65400559e-02
6.23988807e-01 -4.12966132e-01 2.05624089e-01 1.48094654e-01
-1.52213275e-01 -1.03510511e+00 4.26238835e-01 7.76824415e-01
7.88924754e-01 -7.91026533e-01 7.31528163e-01 -7.10328072e-02
-1.67421114e+00 -1.38994560e-01 -6.68002725e-01 3.47995400e-01
-1.44579694e-01 2.51296759e-01 -9.33297515e-01 5.91571212e-01
5.53982794e-01 9.05248463e-01 -9.06214476e-01 1.09787810e+00
-3.82502705e-01 1.03956878e-01 -3.98637235e-01 1.01868786e-01
1.39228955e-01 -2.11687371e-01 1.63994133e-01 7.08591938e-01
4.32654977e-01 6.78983703e-02 3.71269971e-01 8.30585182e-01
1.70280878e-02 5.37014782e-01 -2.47468293e-01 3.70271385e-01
6.32344306e-01 1.71616471e+00 -5.81983805e-01 -1.20097563e-01
-4.20333266e-01 7.99459219e-01 3.42078000e-01 3.66855651e-01
-6.79630280e-01 -3.70826155e-01 7.40980327e-01 8.57549310e-02
-7.38157928e-02 3.12883332e-02 -5.83672710e-02 -9.85533834e-01
1.39160126e-01 -2.80167341e-01 4.34846759e-01 -6.44821942e-01
-1.03046668e+00 6.73481166e-01 -2.10086718e-01 -1.20270061e+00
-2.97929924e-02 -5.26282132e-01 -7.08599627e-01 9.29123163e-01
-1.82007289e+00 -1.56139064e+00 -7.64747441e-01 8.28363717e-01
2.85780966e-01 -1.52832121e-01 6.80125594e-01 3.19286793e-01
-9.20175970e-01 8.99016142e-01 -1.99922234e-01 2.63730884e-01
7.20457196e-01 -8.27248871e-01 5.49053103e-02 5.87498546e-01
3.39924619e-02 6.39376223e-01 1.86900884e-01 -5.13074100e-01
-1.79344368e+00 -1.34636927e+00 5.58541059e-01 -1.59757301e-01
1.90894321e-01 -3.73651415e-01 -9.49319780e-01 4.99131948e-01
-4.05421525e-01 6.25940740e-01 4.23233300e-01 -1.96319908e-01
-4.40963924e-01 -6.00024343e-01 -1.29373193e+00 3.31037670e-01
9.57670629e-01 -4.59912658e-01 -3.12903017e-01 2.19273791e-01
1.96059927e-01 -4.68478680e-01 -8.81315589e-01 6.87122881e-01
8.25514138e-01 -6.80132508e-01 9.97721255e-01 -1.16306618e-01
1.34177664e-02 -6.71029866e-01 -3.15658376e-02 -1.23246121e+00
-3.47955197e-01 -3.53486717e-01 3.22744474e-02 1.38916206e+00
-6.91910386e-02 -7.30793417e-01 6.82119787e-01 6.81589544e-01
-4.07412425e-02 -9.83085215e-01 -1.21847665e+00 -4.01712775e-01
-4.48425472e-01 1.10769406e-01 1.15751421e+00 8.67741883e-01
-2.69039557e-03 -2.07270473e-01 -3.08793336e-01 2.06477195e-01
6.86288416e-01 2.94313312e-01 7.29200065e-01 -1.33880365e+00
5.61826944e-01 -5.87005973e-01 -7.86986828e-01 -7.36899376e-01
2.21755609e-01 -8.93520832e-01 8.25872496e-02 -1.28645003e+00
2.77707309e-01 -7.12956429e-01 -4.96901512e-01 9.63857353e-01
-3.71951401e-01 6.93643987e-01 2.26752385e-01 1.59256965e-01
-5.60916543e-01 9.13644493e-01 1.46883476e+00 -1.11569420e-01
-1.94079220e-01 -2.73555130e-01 -7.83929527e-01 7.12434888e-01
6.86916471e-01 -2.67790437e-01 -3.59566897e-01 -2.87404180e-01
-1.71219915e-01 -3.36150557e-01 3.56363028e-01 -8.59821379e-01
5.31107068e-01 -3.19323540e-01 8.13503861e-01 -4.94626939e-01
4.57631558e-01 -9.35349166e-01 8.03204253e-02 2.72454709e-01
4.38645594e-02 -1.88791707e-01 1.12457164e-01 3.50698799e-01
-2.76357025e-01 3.10874671e-01 8.05548608e-01 1.66038007e-01
-8.97463083e-01 7.91155219e-01 4.79611635e-01 -3.58831555e-01
1.24485338e+00 -3.09584081e-01 -3.34100187e-01 -6.50058836e-02
-3.62389535e-01 3.82420361e-01 5.55619836e-01 8.79712641e-01
1.06545079e+00 -1.77062273e+00 -6.82803154e-01 8.47797692e-01
1.91011339e-01 6.11101165e-02 3.32033634e-01 1.02082360e+00
-1.99610785e-01 2.32447848e-01 -3.96897376e-01 -6.37751341e-01
-1.28895891e+00 5.10140240e-01 3.84804875e-01 3.39765728e-01
-4.16368425e-01 7.75756240e-01 4.61595088e-01 -3.25406194e-01
1.86311334e-01 -1.09934367e-01 -4.01047826e-01 1.17921859e-01
8.82213056e-01 1.88030973e-01 -8.73143598e-03 -1.26504135e+00
-7.12006271e-01 1.39454007e+00 -9.44353119e-02 4.62026507e-01
1.19667292e+00 -2.14957118e-01 -3.61271769e-01 -1.50063396e-01
1.46114576e+00 -1.99374065e-01 -1.24050510e+00 -4.86010969e-01
-2.03470945e-01 -1.05002058e+00 1.49325684e-01 -6.01210713e-01
-1.60727131e+00 8.24548244e-01 8.23365271e-01 -4.12794828e-01
1.53280795e+00 1.92340929e-02 6.83138251e-01 1.97992437e-02
6.26516223e-01 -9.19498742e-01 1.71015427e-01 6.15193248e-02
1.13215089e+00 -1.13105726e+00 -2.58990861e-02 -7.62032270e-01
-5.06453812e-01 1.18187690e+00 9.63818729e-01 1.49462046e-02
8.08943093e-01 -3.84639669e-03 1.53414041e-01 -2.43261427e-01
-2.75056601e-01 -1.42074898e-01 6.42892480e-01 5.22984624e-01
1.64784137e-02 -7.91178569e-02 -1.49838701e-01 5.62704444e-01
-4.68453169e-02 1.48828691e-02 -1.97302908e-01 7.36110806e-01
-6.02676332e-01 -1.14803040e+00 -6.21868074e-01 3.52536082e-01
4.38836291e-02 2.04027161e-01 -2.29093790e-01 5.26354313e-01
3.10739040e-01 7.33633041e-01 -3.45412157e-02 -6.75432563e-01
1.83396086e-01 2.97268145e-02 4.32133734e-01 -3.42144668e-01
8.10715556e-02 9.64148566e-02 -6.16952479e-01 -8.68116140e-01
-3.93624693e-01 -5.69720328e-01 -1.43771672e+00 -8.29645768e-02
-3.51068735e-01 1.42048731e-01 6.70466423e-01 7.83356547e-01
4.19276178e-01 1.26417369e-01 1.06755972e+00 -1.17689979e+00
-2.16697171e-01 -9.53793347e-01 -5.18482566e-01 4.56953108e-01
2.26261139e-01 -1.31128097e+00 -1.23340532e-01 -2.02041984e-01]
|
[13.515851974487305, 0.4081331789493561]
|
5ac82749-5791-47d4-b692-caf7821e1e77
|
fake-news-detection-and-behavioral-analysis
|
2305.16057
| null |
https://arxiv.org/abs/2305.16057v1
|
https://arxiv.org/pdf/2305.16057v1.pdf
|
Fake News Detection and Behavioral Analysis: Case of COVID-19
|
While the world has been combating COVID-19 for over three years, an ongoing "Infodemic" due to the spread of fake news regarding the pandemic has also been a global issue. The existence of the fake news impact different aspect of our daily lives, including politics, public health, economic activities, etc. Readers could mistake fake news for real news, and consequently have less access to authentic information. This phenomenon will likely cause confusion of citizens and conflicts in society. Currently, there are major challenges in fake news research. It is challenging to accurately identify fake news data in social media posts. In-time human identification is infeasible as the amount of the fake news data is overwhelming. Besides, topics discussed in fake news are hard to identify due to their similarity to real news. The goal of this paper is to identify fake news on social media to help stop the spread. We present Deep Learning approaches and an ensemble approach for fake news detection. Our detection models achieved higher accuracy than previous studies. The ensemble approach further improved the detection performance. We discovered feature differences between fake news and real news items. When we added them into the sentence embeddings, we found that they affected the model performance. We applied a hybrid method and built models for recognizing topics from posts. We found half of the identified topics were overlapping in fake news and real news, which could increase confusion in the population.
|
['James Geller', 'Soon Ae Chun', 'Navya Martin Kollapally', 'Chih-Yuan Li']
|
2023-05-25
| null | null | null | null |
['sentence-embeddings', 'fake-news-detection', 'sentence-embeddings']
|
['methodology', 'natural-language-processing', 'natural-language-processing']
|
[-4.33813542e-01 1.36847660e-01 -1.18268549e-01 -7.52981901e-02
-4.25061166e-01 -5.71029246e-01 9.01007533e-01 4.15112495e-01
-2.39302784e-01 8.28211784e-01 5.98449528e-01 -2.30265081e-01
6.67038858e-01 -1.15107417e+00 -6.56537116e-01 -2.87321717e-01
3.50423992e-01 2.59662747e-01 2.71010160e-01 -8.65449905e-01
4.61901516e-01 -1.35741131e-02 -1.09656620e+00 8.15470576e-01
9.05833423e-01 4.45037484e-01 -1.97670788e-01 1.32337868e-01
-3.20785135e-01 8.12606573e-01 -1.39594698e+00 -7.76860237e-01
9.86743793e-02 -5.27998745e-01 -5.71708977e-01 -1.56788886e-01
3.70001972e-01 -3.90143663e-01 -7.59906530e-01 1.38496149e+00
3.90349060e-01 -2.89129257e-01 4.63474274e-01 -1.26891446e+00
-9.88374889e-01 5.93885839e-01 -5.01212776e-01 3.13834935e-01
2.13688791e-01 -2.83137858e-01 4.45000112e-01 -9.70910907e-01
7.63186753e-01 1.43495929e+00 9.15760815e-01 4.80069876e-01
-3.75056237e-01 -9.81121957e-01 -2.64051352e-02 -1.57211744e-03
-1.13760841e+00 -1.57443523e-01 6.15031302e-01 -6.59614444e-01
4.93415505e-01 2.75241643e-01 9.03323233e-01 1.71381271e+00
6.65583014e-01 6.95944190e-01 1.11369801e+00 3.64083089e-02
-1.41696855e-01 6.46250188e-01 1.57080412e-01 5.64130545e-01
7.77526557e-01 1.72093496e-01 -3.53179514e-01 -7.68368781e-01
3.27953398e-01 2.46577591e-01 -3.68363053e-01 5.73476434e-01
-1.45484030e+00 1.35127211e+00 5.98319590e-01 4.66124743e-01
-2.05020741e-01 -1.52803078e-01 5.68364024e-01 6.17633939e-01
1.06055665e+00 9.23835218e-01 -2.57543266e-01 3.45807262e-02
-6.14026189e-01 3.41758192e-01 9.75512981e-01 4.08645570e-01
3.14915448e-01 -1.48127198e-01 5.12088928e-03 6.57654643e-01
8.90853703e-02 1.01400983e+00 8.38321626e-01 -2.30312701e-02
5.96879244e-01 6.20464087e-01 6.61590755e-01 -2.35822368e+00
-3.75686109e-01 -6.28835380e-01 -8.36486876e-01 -5.20347476e-01
4.08360422e-01 -3.71421129e-01 -6.66189194e-01 1.12939322e+00
2.77440757e-01 2.46978968e-01 -5.62686995e-02 1.12332428e+00
9.81360137e-01 1.00513554e+00 -1.35778800e-01 6.71556965e-02
1.50939584e+00 -9.11789656e-01 -1.28804398e+00 -4.66901064e-01
8.79538536e-01 -1.11075687e+00 5.65888286e-01 -1.43383155e-02
-1.90819547e-01 -7.74596259e-02 -9.81518447e-01 1.52799770e-01
-1.04917717e+00 1.56327099e-01 6.22223198e-01 7.44420528e-01
-2.83083677e-01 4.87690032e-01 -5.15746236e-01 -3.00032169e-01
5.66080511e-01 -2.06912085e-01 -3.68392617e-01 9.81918499e-02
-2.07896018e+00 1.30790746e+00 2.80912697e-01 -2.82036010e-02
-4.19274747e-01 -1.54021367e-01 -5.55389285e-01 -3.67724746e-01
2.55727887e-01 -3.28744292e-01 9.00115132e-01 -1.20182180e+00
-8.69785428e-01 8.20299923e-01 7.74095654e-02 -4.97954756e-01
6.72017992e-01 -2.88538158e-01 -1.07266772e+00 -9.17104334e-02
4.22408909e-01 1.73682570e-01 1.11735666e+00 -1.03423202e+00
-5.05314946e-01 -2.77252853e-01 -9.86243859e-02 -4.17770296e-02
-4.04329777e-01 1.07850835e-01 3.52881074e-01 -1.06364059e+00
1.43432930e-01 -8.92465234e-01 1.71641320e-01 -3.23329985e-01
-4.99387980e-01 1.25436252e-02 1.35290527e+00 -9.75907922e-01
1.16268301e+00 -1.93658495e+00 -4.62972969e-01 -2.51682431e-01
5.61907828e-01 4.68222201e-01 8.97098407e-02 6.07302964e-01
2.44877100e-01 5.23194075e-01 1.51923954e-01 6.69894144e-02
-4.16115582e-01 -1.63087189e-01 -9.82805014e-01 9.46948349e-01
1.23864084e-01 7.92534530e-01 -1.28210342e+00 -4.89841141e-02
-3.30094904e-01 3.71547848e-01 -2.25986958e-01 -1.51816040e-01
-8.53980705e-02 3.39390367e-01 -5.33504605e-01 6.42116368e-01
8.38842928e-01 -2.18132392e-01 -1.89742103e-01 -2.35764459e-01
5.44848554e-02 7.45325506e-01 -5.11426866e-01 5.65787852e-01
-2.50853062e-01 1.19674897e+00 -3.61723572e-01 -9.57376540e-01
8.10112715e-01 3.05466056e-01 1.71935976e-01 -5.34819722e-01
4.59276915e-01 3.62657219e-01 -1.10163555e-01 -6.37119710e-01
9.76941466e-01 -2.36280337e-01 -3.52792352e-01 6.46945298e-01
-4.72274572e-01 -1.28919020e-01 -5.62037349e-01 2.62913704e-01
8.01854730e-01 -6.86917603e-01 3.29661369e-01 -2.79704928e-02
2.67399475e-02 5.20955384e-01 3.33764195e-01 6.84421360e-01
-3.68081033e-01 2.86482960e-01 4.43152964e-01 -8.16472411e-01
-9.01042283e-01 -6.05647862e-01 -1.27453238e-01 6.03878617e-01
5.54750562e-01 -2.48981640e-01 -4.33650047e-01 -1.08957636e+00
5.79888336e-02 7.98383355e-01 -7.06652820e-01 -4.09916937e-01
-4.26331252e-01 -1.26981556e+00 7.72827327e-01 -3.31876755e-01
7.14338541e-01 -6.94098771e-01 -2.43597209e-01 4.38635617e-01
-8.41306865e-01 -1.15515149e+00 -6.11724019e-01 -5.21464229e-01
-4.08471942e-01 -1.03328884e+00 -5.13777971e-01 -5.81019878e-01
7.23168910e-01 8.78574193e-01 7.92411387e-01 1.68758288e-01
-1.60194002e-02 -2.38147601e-01 -6.26353860e-01 -8.77723575e-01
-8.78530920e-01 -1.84307039e-01 2.92808056e-01 -3.62862134e-03
5.97041488e-01 2.52949864e-01 -3.69579732e-01 5.44630468e-01
-8.50008965e-01 -2.38984786e-02 2.54538119e-01 1.13808823e+00
-2.40546554e-01 3.58019769e-01 8.90107453e-01 -1.11810637e+00
9.09411609e-01 -1.09059370e+00 -3.29438776e-01 -1.88799888e-01
-1.24500535e-01 -4.53913391e-01 4.95239884e-01 -9.18153107e-01
-6.00689292e-01 -6.69260681e-01 1.74693510e-01 9.36534852e-02
1.99665248e-01 6.14133775e-01 4.33388829e-01 7.74460062e-02
8.20857525e-01 5.06661609e-02 7.94794038e-02 -1.40523046e-01
-2.06080731e-02 1.30903196e+00 -2.24364787e-01 2.13252902e-01
7.85897732e-01 8.23324084e-01 -7.77839422e-01 -1.18589556e+00
-1.41401029e+00 -4.20410037e-01 2.79518783e-01 -1.03519723e-01
5.64021707e-01 -1.10381782e+00 -1.64234772e-01 8.61643553e-01
-1.96823704e+00 4.51688796e-01 3.31741661e-01 6.63044989e-01
4.86529946e-01 5.34194708e-01 -7.53168344e-01 -6.56915665e-01
-2.81673130e-02 -8.85461032e-01 7.41701007e-01 -1.89750642e-01
-2.28735149e-01 -9.81055081e-01 1.19859591e-01 6.57338798e-01
7.24530756e-01 2.87440270e-01 3.94161999e-01 -1.07613778e+00
-1.97295830e-01 -5.81505895e-01 -4.74839091e-01 2.89912045e-01
4.70162183e-01 -2.14185953e-01 -9.39034224e-01 -3.18212688e-01
4.59918916e-01 -9.13479179e-02 8.71283650e-01 -1.87686667e-01
6.01833105e-01 -1.06420100e+00 -6.15828037e-01 -1.03668325e-01
8.16710591e-01 -1.56758893e-02 5.02024233e-01 4.33445811e-01
6.77198708e-01 8.13490033e-01 7.14462936e-01 2.79391438e-01
3.30994010e-01 6.22832656e-01 4.00752395e-01 -1.40767425e-01
1.53695866e-01 -3.56417567e-01 8.11583638e-01 1.01863945e+00
3.79208237e-01 -7.30723083e-01 -9.32347357e-01 6.29349053e-01
-1.56436050e+00 -1.33852947e+00 -5.41432500e-01 1.92198622e+00
5.86755395e-01 4.23336886e-02 -1.14038385e-01 -1.94851398e-01
1.20316195e+00 2.61033416e-01 -1.28146604e-01 -2.61741906e-01
-3.43852013e-01 -5.43869734e-01 8.88519704e-01 3.61346513e-01
-1.54631150e+00 1.06161582e+00 5.86518383e+00 7.49556124e-01
-1.67983627e+00 6.15176439e-01 5.54709315e-01 2.59510040e-01
-2.58495152e-01 -2.74266720e-01 -7.25709796e-01 1.00060236e+00
7.40262210e-01 -5.27893305e-02 1.44436769e-02 7.81766713e-01
5.31147122e-01 1.81509688e-01 -2.31723681e-01 8.74689281e-01
4.78402138e-01 -1.67208600e+00 1.53596401e-01 1.84650078e-01
9.58807826e-01 3.87113124e-01 -1.32862069e-02 2.26267308e-01
2.32305583e-02 -1.03888071e+00 6.27849698e-01 1.62172064e-01
2.71545887e-01 -5.57423532e-01 1.36640787e+00 6.07161641e-01
-2.61331677e-01 1.21842451e-01 -5.85061014e-01 -3.59214246e-01
2.38555178e-01 1.15611374e+00 -1.10337615e+00 8.11994597e-02
3.64284158e-01 7.90720999e-01 -4.24352467e-01 7.96652794e-01
-1.37772784e-01 6.00736439e-01 -2.43751630e-01 -6.12690866e-01
4.39013690e-01 3.33134197e-02 9.28317130e-01 1.13403821e+00
3.73874933e-01 -1.80357501e-01 1.43178448e-01 6.12624824e-01
-2.45113552e-01 1.18485205e-01 -1.05953670e+00 -6.83522284e-01
4.15093690e-01 7.69413888e-01 -7.07147896e-01 -4.59424525e-01
-3.90693665e-01 1.12581003e+00 4.71206009e-02 -6.28867885e-03
-1.14845538e+00 -3.76733959e-01 5.77535450e-01 4.14783895e-01
-1.73151121e-01 -1.34370280e-02 2.25237031e-02 -1.71566415e+00
-1.68289945e-01 -1.13794863e+00 -2.27991752e-02 -3.89870435e-01
-1.75143790e+00 7.36699879e-01 -3.86509627e-01 -1.51961255e+00
3.41136575e-01 -3.36356074e-01 -4.33124959e-01 3.84599715e-01
-1.13128710e+00 -8.80711496e-01 -1.18914999e-01 -9.89128184e-03
4.12839770e-01 -2.21877471e-01 7.86758721e-01 4.71423149e-01
-2.84962863e-01 4.43204373e-01 2.57093072e-01 4.71229345e-01
1.01278436e+00 -6.05024874e-01 6.58409119e-01 6.66451573e-01
7.30990544e-02 5.72824538e-01 9.98447597e-01 -1.13670456e+00
-9.61964667e-01 -1.17793572e+00 1.34625459e+00 -7.52265394e-01
1.09849763e+00 -3.72444570e-01 -9.05364692e-01 4.59398508e-01
6.20788485e-02 -2.89748043e-01 7.03112662e-01 -3.07435244e-01
-5.63778698e-01 5.69616318e-01 -1.45405424e+00 6.23434067e-01
5.65631628e-01 -5.83873689e-01 -8.87886167e-01 9.65419948e-01
1.00700021e+00 -3.92758161e-01 -1.58263400e-01 -1.29404128e-01
4.98375267e-01 -7.60505259e-01 6.78439498e-01 -9.50053930e-01
4.92024869e-01 -2.26478100e-01 2.19868079e-01 -1.72368288e+00
-4.52142693e-02 -3.44156474e-01 3.67763415e-02 7.28980124e-01
4.94901210e-01 -1.22885001e+00 5.27241409e-01 -4.99442667e-02
9.90467295e-02 -4.01724249e-01 -7.94553041e-01 -5.93122005e-01
4.77183759e-02 -5.62101379e-02 4.93289053e-01 1.80524731e+00
6.41220957e-02 3.76165688e-01 -8.61445427e-01 4.76812392e-01
2.83199281e-01 1.38834389e-02 6.96735620e-01 -1.02862799e+00
-7.66927898e-02 -1.80857033e-01 -5.90355456e-01 -9.13899064e-01
-9.10517573e-02 -5.92909098e-01 -2.88951248e-01 -1.23503339e+00
-5.30810049e-03 -3.16991776e-01 3.07625890e-01 2.00852752e-01
-1.20256387e-01 4.63579059e-01 -6.79173991e-02 4.20439541e-01
-1.66616142e-01 6.11843467e-01 1.54832840e+00 -5.38640797e-01
-3.17717716e-02 1.43454507e-01 -7.40309596e-01 9.72588122e-01
9.93462145e-01 -1.13112700e+00 5.69595434e-02 -4.08421487e-01
6.82600796e-01 -7.41107091e-02 5.23035467e-01 -6.13743842e-01
-4.70337197e-02 -1.74956590e-01 8.19414854e-02 -6.45454109e-01
3.55685443e-01 -5.72753370e-01 -1.64842963e-01 8.97553444e-01
5.50830923e-03 1.12373270e-01 6.55181855e-02 8.57441247e-01
-4.23565865e-01 -4.92323423e-03 7.78613210e-01 -2.27177665e-01
-2.29433760e-01 -8.73159543e-02 -9.75324869e-01 1.33914217e-01
9.10985529e-01 2.04385951e-01 -1.03610802e+00 -9.00127769e-01
-2.94717610e-01 -1.27619416e-01 4.23906833e-01 7.49625266e-01
6.32488430e-01 -1.15984941e+00 -9.77933049e-01 -1.05985075e-01
1.70918629e-01 -5.27484298e-01 6.48481771e-02 8.68018866e-01
-6.79104686e-01 3.22459072e-01 -1.01857241e-02 7.47619420e-02
-1.04259539e+00 4.35346723e-01 2.57807314e-01 -1.05531029e-01
-3.36316913e-01 6.32673085e-01 -1.47932008e-01 -6.03689432e-01
-2.07477227e-01 -1.45803928e-01 -2.22836584e-01 4.05589104e-01
8.15608442e-01 3.62869799e-01 -4.31513004e-02 -1.18354082e+00
-4.31380093e-01 -1.78032350e-02 -3.37745756e-01 3.09060514e-01
1.04048729e+00 -4.27467562e-02 -3.45478654e-01 3.12396467e-01
1.31105268e+00 5.51336527e-01 -9.34480801e-02 5.71030118e-02
-6.01540916e-02 -7.07441330e-01 3.46580923e-01 -9.19533730e-01
-8.01028013e-01 5.20991385e-01 2.78619736e-01 7.78024733e-01
1.81512564e-01 -7.97871053e-02 1.34439373e+00 5.73351800e-01
4.28794414e-01 -9.86858845e-01 1.78390235e-01 7.13870823e-01
9.38490689e-01 -1.75300157e+00 7.76646733e-02 -6.50595903e-01
-7.08401859e-01 1.06293952e+00 2.70005912e-01 -1.98064730e-01
7.35815883e-01 -1.83414638e-01 3.73797745e-01 -5.95818520e-01
-2.62118578e-01 4.77689624e-01 2.35732764e-01 3.22845876e-01
4.17828977e-01 2.94208199e-01 -7.53665030e-01 5.25713027e-01
-3.96772265e-01 -5.02647817e-01 1.18172634e+00 6.84501350e-01
-5.81892788e-01 -7.56577909e-01 -6.42529249e-01 6.79605365e-01
-8.31322193e-01 -1.01671524e-01 -7.56195366e-01 6.47583663e-01
3.14230949e-01 1.21147263e+00 -3.18465941e-02 -6.24749184e-01
-1.57553002e-01 -3.72213274e-01 -2.28456214e-01 -6.83004797e-01
-6.13998175e-01 -4.47436243e-01 5.14196634e-01 -3.28606606e-01
-2.24981233e-01 -2.89973229e-01 -8.43196511e-01 -6.79233730e-01
-8.76100004e-01 3.99886161e-01 9.20213580e-01 1.10724962e+00
5.74624777e-01 9.62315053e-02 7.38401592e-01 -3.93173844e-01
-6.87612236e-01 -1.14884925e+00 -2.03292936e-01 5.07173955e-01
4.80861306e-01 -7.93602705e-01 -8.02563429e-01 -4.02107805e-01]
|
[8.148965835571289, 10.259276390075684]
|
7daeef49-9b05-45e7-b79c-35a70bfb1aff
|
diffstyler-controllable-dual-diffusion-for
|
2211.10682
| null |
https://arxiv.org/abs/2211.10682v1
|
https://arxiv.org/pdf/2211.10682v1.pdf
|
DiffStyler: Controllable Dual Diffusion for Text-Driven Image Stylization
|
Despite the impressive results of arbitrary image-guided style transfer methods, text-driven image stylization has recently been proposed for transferring a natural image into the stylized one according to textual descriptions of the target style provided by the user. Unlike previous image-to-image transfer approaches, text-guided stylization progress provides users with a more precise and intuitive way to express the desired style. However, the huge discrepancy between cross-modal inputs/outputs makes it challenging to conduct text-driven image stylization in a typical feed-forward CNN pipeline. In this paper, we present DiffStyler on the basis of diffusion models. The cross-modal style information can be easily integrated as guidance during the diffusion progress step-by-step. In particular, we use a dual diffusion processing architecture to control the balance between the content and style of the diffused results. Furthermore, we propose a content image-based learnable noise on which the reverse denoising process is based, enabling the stylization results to better preserve the structure information of the content image. We validate the proposed DiffStyler beyond the baseline methods through extensive qualitative and quantitative experiments.
|
['Changsheng Xu', 'WeiMing Dong', 'Yong Zhang', 'Haibin Huang', 'Chongyang Ma', 'Fan Tang', 'Yuxin Zhang', 'Nisha Huang']
|
2022-11-19
| null | null | null | null |
['image-stylization']
|
['computer-vision']
|
[ 2.51880437e-01 1.12507999e-01 -4.02832031e-03 -6.45656526e-01
-4.80057955e-01 -5.69613218e-01 8.66371930e-01 -1.50945917e-01
-3.60930055e-01 3.95174176e-01 4.19513822e-01 6.07494591e-03
2.55889714e-01 -8.06042850e-01 -7.67060816e-01 -7.43088782e-01
7.50983477e-01 3.00554454e-01 5.08761108e-02 -2.95254171e-01
3.49570304e-01 3.19901437e-01 -1.01311445e+00 4.08926219e-01
1.15900469e+00 9.18657303e-01 5.28267324e-01 5.14086366e-01
-5.01225054e-01 7.46434093e-01 -4.18645799e-01 -5.47395289e-01
1.94683522e-01 -8.32704365e-01 -6.88822925e-01 4.96209621e-01
5.52795887e-01 -4.73858923e-01 -2.26801187e-01 1.22889173e+00
3.44860047e-01 2.16136705e-02 8.19838405e-01 -1.07806337e+00
-1.40259588e+00 5.73637128e-01 -7.27286935e-01 -2.44568422e-01
2.18153507e-01 5.42132080e-01 8.83756936e-01 -7.23104537e-01
9.32047486e-01 1.44719505e+00 1.03712104e-01 7.23476410e-01
-1.49649191e+00 -5.57747364e-01 4.69770133e-01 -9.58569795e-02
-8.89173806e-01 -2.31482089e-01 1.08683217e+00 -5.37856460e-01
1.70602594e-02 -9.94535834e-02 7.65431166e-01 1.32640004e+00
8.01924020e-02 9.29780543e-01 1.22999215e+00 -2.55387008e-01
1.41946211e-01 3.07954609e-01 -4.29980516e-01 6.67939484e-01
-1.57075807e-01 -4.88620065e-02 -6.36101127e-01 3.95688057e-01
1.15503287e+00 -2.34015044e-02 -3.45151007e-01 -4.21233118e-01
-1.10013843e+00 6.96284473e-01 7.02265739e-01 2.89691955e-01
-4.32402343e-01 2.97446430e-01 3.16902518e-01 2.91797757e-01
7.70139575e-01 4.99797016e-01 -1.75033540e-01 1.70961544e-02
-1.11708331e+00 2.12277457e-01 3.24093819e-01 1.01400995e+00
9.50862110e-01 3.43847215e-01 -7.34217882e-01 6.64080501e-01
2.32061222e-01 4.92042631e-01 4.49081987e-01 -1.04948914e+00
2.50555277e-01 6.23209596e-01 1.92752287e-01 -8.77848446e-01
2.52995044e-01 -3.92264843e-01 -1.10028601e+00 5.07335424e-01
5.06171584e-01 -1.50632665e-01 -1.02816784e+00 1.70403004e+00
8.06490108e-02 -1.69395253e-01 -2.10646316e-01 1.03821027e+00
6.63776636e-01 6.33726537e-01 2.68620908e-01 2.16059700e-01
1.26644266e+00 -1.13089299e+00 -9.11453068e-01 -2.00646177e-01
4.16978300e-01 -7.53295660e-01 1.65960455e+00 2.14495987e-01
-1.24673104e+00 -5.32861292e-01 -8.10874999e-01 -3.51516187e-01
-8.69460404e-02 1.60104021e-01 2.60556757e-01 4.29536730e-01
-1.15741122e+00 5.90834200e-01 -5.81493378e-01 -3.43106806e-01
6.01515591e-01 -1.21703073e-01 -2.49348640e-01 1.41267464e-01
-9.79184210e-01 6.98113501e-01 1.15305327e-01 7.53127187e-02
-8.71053517e-01 -9.24224317e-01 -7.99667597e-01 -4.90692072e-02
8.68717209e-02 -1.02779973e+00 1.12589049e+00 -1.64199424e+00
-1.87367940e+00 1.05962586e+00 -2.27253407e-01 -2.30637357e-01
9.92471039e-01 -3.87137324e-01 1.56452894e-01 1.77753463e-01
2.35862404e-01 1.14847696e+00 1.22754395e+00 -1.57859182e+00
-4.86156672e-01 -1.67858243e-01 8.46023783e-02 4.05736387e-01
-5.10806143e-01 -1.29549414e-01 -8.17338049e-01 -1.07269406e+00
-2.16651753e-01 -6.39325619e-01 -1.51685193e-01 5.90603232e-01
-5.65456271e-01 1.28393859e-01 7.28699148e-01 -6.55646682e-01
9.70639586e-01 -2.21584725e+00 5.80851018e-01 -6.13568642e-04
2.29174435e-01 7.10814446e-02 -4.60550338e-01 1.92918062e-01
6.22877851e-02 1.53729066e-01 -3.06410104e-01 -7.70501792e-01
-1.26509443e-01 1.16338179e-01 -3.94804031e-01 1.65158957e-01
5.00626802e-01 1.15769923e+00 -1.00452471e+00 -4.03207004e-01
2.14603320e-01 5.69812119e-01 -6.46378219e-01 5.82429826e-01
-4.35389757e-01 1.00435817e+00 -5.58157682e-01 2.44285211e-01
6.69460058e-01 -3.83825712e-02 -3.03951889e-01 -4.86169010e-01
8.10062289e-02 5.86625468e-03 -7.46821105e-01 2.00165939e+00
-7.28941977e-01 6.77958071e-01 2.34505951e-01 -5.92955351e-01
9.46025729e-01 5.69293648e-02 1.00666665e-01 -7.67891169e-01
1.38326094e-01 -2.10629925e-02 -3.67563426e-01 -3.53313535e-01
6.93309844e-01 -3.09370816e-01 1.98311225e-01 6.59632862e-01
-1.04619488e-01 -6.21850431e-01 1.05724588e-01 4.29577082e-01
4.65326428e-01 5.41457653e-01 -1.82062030e-01 -3.50575000e-01
5.42023599e-01 -2.43026495e-01 2.33649448e-01 6.02462292e-01
-1.03180772e-02 1.10722935e+00 5.04664838e-01 -2.66296446e-01
-1.12768745e+00 -9.25097585e-01 2.84314424e-01 1.05858648e+00
1.65855244e-01 -1.37450621e-01 -1.20579958e+00 -6.79325819e-01
-2.72189558e-01 6.96662307e-01 -8.80790234e-01 -2.42988557e-01
-3.52219373e-01 -1.53992131e-01 4.38949674e-01 4.00684655e-01
8.81386995e-01 -1.38270879e+00 -1.26987919e-01 1.05826713e-01
-2.79907793e-01 -1.11657083e+00 -1.24132347e+00 -1.54815048e-01
-8.25815797e-01 -5.38301229e-01 -1.31153941e+00 -7.78154135e-01
9.87623930e-01 1.95060402e-01 1.00551331e+00 8.91946852e-02
6.83607385e-02 2.91311204e-01 -3.00308555e-01 -1.30658239e-01
-6.53988123e-01 1.68484151e-01 -2.59605348e-01 4.12687480e-01
-2.37476885e-01 -6.05426133e-01 -9.14646387e-01 1.30454734e-01
-1.36376655e+00 5.49798846e-01 4.83110458e-01 8.17168117e-01
4.59360540e-01 -1.84501991e-01 3.16411138e-01 -9.84678149e-01
9.27199066e-01 5.53482436e-02 -3.36278111e-01 1.59419820e-01
-5.91741562e-01 4.07269537e-01 6.47591650e-01 -6.87897861e-01
-1.44619095e+00 -5.21643497e-02 -1.52495921e-01 -5.01861393e-01
-1.24998108e-01 1.84781343e-01 -3.37609649e-01 5.25512658e-02
6.16165698e-01 4.88771230e-01 1.41401112e-01 -3.15629929e-01
9.98631775e-01 4.32140440e-01 5.21431446e-01 -7.76616633e-01
1.03591037e+00 6.50934696e-01 -4.47016895e-01 -5.79707921e-01
-9.28103387e-01 9.03521180e-02 -7.66042411e-01 -2.12514088e-01
1.08942568e+00 -8.73871088e-01 -3.70742500e-01 8.24560285e-01
-1.32380462e+00 -8.79926682e-01 -6.33093774e-01 -1.95258155e-01
-6.26955807e-01 3.28638881e-01 -7.78165698e-01 -4.83022273e-01
-4.69615072e-01 -1.37777293e+00 1.22065365e+00 8.92231539e-02
-3.66864026e-01 -1.20063484e+00 -1.84954733e-01 3.17932934e-01
6.11393154e-01 5.09636961e-02 1.03854382e+00 1.00566722e-01
-6.69238746e-01 1.28067270e-01 -5.81951141e-01 4.81860161e-01
3.86008471e-01 8.26953202e-02 -8.04283082e-01 -1.22688763e-01
-9.07977819e-02 -2.41184890e-01 9.14836586e-01 4.22159851e-01
1.22563422e+00 -2.25031897e-01 1.62273943e-01 8.46248448e-01
1.19736421e+00 -2.03196600e-01 8.33411872e-01 4.47739750e-01
1.05920494e+00 7.92128026e-01 4.07995433e-01 7.23100826e-02
3.22078526e-01 4.96727496e-01 2.20748201e-01 -6.00905716e-01
-5.30599594e-01 -5.71156740e-01 4.18466479e-01 4.41702098e-01
2.70156488e-02 -3.32080156e-01 -3.28053445e-01 4.25717562e-01
-1.58694577e+00 -7.39642680e-01 1.09545533e-02 1.74155295e+00
1.12409270e+00 1.71542510e-01 1.47320069e-02 -2.83228874e-01
5.28384030e-01 2.83522755e-01 -5.51830053e-01 -3.59716892e-01
-1.09428436e-01 -1.85505494e-01 2.02859402e-01 6.52572095e-01
-6.65849388e-01 1.37455201e+00 5.67364836e+00 1.05562770e+00
-1.51482964e+00 -5.26897907e-02 1.08350372e+00 5.50940586e-03
-8.12009573e-01 -1.80175439e-01 -4.28222716e-01 4.28217113e-01
3.09773684e-01 -6.25541806e-02 4.14651275e-01 4.78963524e-01
6.49664581e-01 -1.03483409e-01 -1.10628557e+00 9.30992842e-01
-1.14358924e-01 -1.36603236e+00 7.34947383e-01 -1.64146975e-01
8.92333210e-01 -4.59363371e-01 5.31786561e-01 -5.29803224e-02
3.38105708e-01 -8.09019327e-01 1.18972707e+00 6.40062511e-01
1.11000967e+00 -5.58040082e-01 1.74092278e-01 1.19837783e-01
-9.35536325e-01 1.94696099e-01 -1.88583937e-02 1.34613633e-01
3.48072499e-01 6.97854877e-01 -1.97280824e-01 4.50771749e-01
6.29996598e-01 8.72063100e-01 -5.50550282e-01 4.58460867e-01
-6.04190350e-01 4.58434761e-01 2.29793385e-01 2.25579634e-01
4.05945152e-01 -7.34087169e-01 4.67735350e-01 1.31833827e+00
3.02084655e-01 -1.06886365e-01 -2.06486583e-01 1.47712553e+00
-2.48192221e-01 1.76532373e-01 -5.67965150e-01 -1.07900023e-01
2.86328141e-02 1.28896570e+00 -7.13086188e-01 -3.75866860e-01
-1.27757594e-01 1.60874951e+00 4.44367468e-01 8.19780350e-01
-4.57096398e-01 -1.98122576e-01 5.54418266e-01 4.07092750e-01
1.41017079e-01 -2.04116344e-01 -6.45288527e-01 -1.26552474e+00
-4.42838185e-02 -9.46390986e-01 -2.40834758e-01 -1.18572009e+00
-1.37136567e+00 7.89565802e-01 -1.60573438e-01 -9.66508090e-01
5.90730160e-02 -3.94765049e-01 -1.00706375e+00 1.15344930e+00
-1.43559349e+00 -1.40571618e+00 -4.84632969e-01 4.94263291e-01
7.96001196e-01 -3.46856453e-02 4.50075567e-01 2.13711932e-01
-4.01611060e-01 5.12568712e-01 -4.53441180e-02 1.79780707e-01
1.15029109e+00 -1.37320077e+00 4.72267449e-01 8.16410661e-01
-1.03759348e-01 5.65450370e-01 9.40408647e-01 -7.39263952e-01
-9.92130756e-01 -1.19046319e+00 4.17523772e-01 -2.76640326e-01
6.68339729e-01 -3.56435180e-01 -1.05134332e+00 4.81866121e-01
7.75569439e-01 -3.98209631e-01 1.49864435e-01 -1.33196309e-01
-4.61008370e-01 -2.31779009e-01 -9.50467646e-01 1.03781402e+00
9.45242345e-01 -6.47526979e-01 -1.69003621e-01 -1.37452139e-02
7.66266525e-01 -2.27983281e-01 -5.41336417e-01 7.14352867e-03
3.80771965e-01 -9.08034384e-01 7.50684619e-01 -3.60543400e-01
9.55153167e-01 -3.41147333e-01 2.68091887e-01 -1.69883204e+00
-3.73230636e-01 -6.74162447e-01 4.06321049e-01 1.58007491e+00
2.73821861e-01 -2.18318388e-01 7.76308179e-01 5.44304311e-01
1.05392411e-01 -5.27768791e-01 -2.90754259e-01 -3.47430408e-01
3.18570077e-01 -2.70836562e-01 5.46486676e-01 6.74002171e-01
-3.46535236e-01 3.81414741e-01 -5.55302262e-01 -3.62029105e-01
6.03681564e-01 1.10709377e-01 9.22067583e-01 -7.61071205e-01
-2.19408661e-01 -7.31472075e-01 1.17413467e-02 -1.49370086e+00
1.50403619e-01 -7.81929374e-01 1.05298623e-01 -1.55851436e+00
2.11457729e-01 -3.00509602e-01 7.65860379e-02 1.65348634e-01
-4.43169802e-01 3.99697691e-01 3.03448528e-01 1.99415177e-01
-3.39323223e-01 9.95798826e-01 2.15333033e+00 -3.52534086e-01
-2.25693002e-01 -2.24101767e-01 -9.60752308e-01 6.23423994e-01
6.48223758e-01 -2.12124154e-01 -5.70571721e-01 -8.07628274e-01
1.52902767e-01 -7.82565866e-03 2.97745287e-01 -5.49084067e-01
7.75871798e-02 -2.08680794e-01 2.91553468e-01 -3.25826377e-01
2.24710256e-01 -6.18492782e-01 -1.24658599e-01 2.93459415e-01
-7.20517516e-01 -1.68316707e-01 4.42708954e-02 6.86941743e-01
-2.92215675e-01 -9.73221064e-02 9.82414186e-01 -1.23231344e-01
-4.02978748e-01 5.76568782e-01 -4.42268193e-01 4.60091345e-02
6.31126642e-01 -2.87670344e-01 7.01653212e-02 -9.02772486e-01
-7.03816354e-01 1.25350446e-01 7.90228546e-01 3.93828273e-01
6.20979667e-01 -1.34801471e+00 -7.01267004e-01 2.04996660e-01
1.54463217e-01 1.20835379e-01 3.02558303e-01 6.61484897e-01
-4.89694983e-01 -1.19071208e-01 -3.67918193e-01 -5.66745937e-01
-9.61952627e-01 3.87692213e-01 4.63496834e-01 -1.84056967e-01
-6.98082685e-01 6.33846223e-01 8.74687374e-01 -2.95062482e-01
2.36129314e-01 -3.55763584e-01 4.74161021e-02 3.88090941e-03
5.29212356e-01 -1.24750510e-01 -3.05139333e-01 -4.45539445e-01
3.37782681e-01 7.08744228e-01 -3.22607756e-01 -4.87628043e-01
1.23840833e+00 -5.84121883e-01 -1.60400853e-01 3.19001108e-01
9.34390247e-01 -1.13268398e-01 -1.91831017e+00 -4.31400627e-01
-2.90283978e-01 -5.02424121e-01 1.04330435e-01 -6.66054249e-01
-1.35810339e+00 1.11302447e+00 4.82490838e-01 -1.51503295e-01
1.19069147e+00 -1.72120467e-01 8.03268611e-01 7.45553672e-02
5.38281128e-02 -8.92813742e-01 5.85674822e-01 3.18039030e-01
1.28618526e+00 -1.27198529e+00 -2.65561014e-01 -3.96536142e-01
-9.47018862e-01 9.35131013e-01 6.55337691e-01 -2.00576767e-01
5.09677052e-01 8.04235637e-02 4.71505821e-01 -1.39988586e-02
-3.65125835e-01 1.48337325e-02 4.74783272e-01 6.91440344e-01
5.45935035e-01 -1.84853226e-01 -3.03762723e-02 3.53397608e-01
-5.72133623e-02 9.29986760e-02 4.52293396e-01 3.83954167e-01
-2.54307002e-01 -1.27718782e+00 -2.84669548e-01 1.15667649e-01
-2.40158379e-01 -2.51873672e-01 -4.51021433e-01 4.07081395e-01
-1.24880120e-01 6.55123651e-01 -1.51675595e-02 -4.96140905e-02
3.09381545e-01 -1.51657656e-01 5.27900040e-01 -5.67311287e-01
-6.64860487e-01 2.62634486e-01 -3.63537461e-01 -4.58808631e-01
-3.44332486e-01 -3.33031207e-01 -1.07043862e+00 -3.22408915e-01
7.92412534e-02 -1.56550914e-01 5.21337092e-01 9.16035891e-01
2.78861672e-01 7.40588963e-01 5.60900807e-01 -1.17690039e+00
-2.02554688e-01 -9.93289411e-01 -6.46061659e-01 8.56048465e-01
2.23261431e-01 -3.43871593e-01 -2.25142583e-01 5.12720406e-01]
|
[11.421856880187988, -0.4735671579837799]
|
ac24ab07-6634-4123-9b95-d67ffbec3a18
|
stack-sentence-ordering-with-temporal
|
2109.02247
| null |
https://arxiv.org/abs/2109.02247v1
|
https://arxiv.org/pdf/2109.02247v1.pdf
|
STaCK: Sentence Ordering with Temporal Commonsense Knowledge
|
Sentence order prediction is the task of finding the correct order of sentences in a randomly ordered document. Correctly ordering the sentences requires an understanding of coherence with respect to the chronological sequence of events described in the text. Document-level contextual understanding and commonsense knowledge centered around these events are often essential in uncovering this coherence and predicting the exact chronological order. In this paper, we introduce STaCK -- a framework based on graph neural networks and temporal commonsense knowledge to model global information and predict the relative order of sentences. Our graph network accumulates temporal evidence using knowledge of `past' and `future' and formulates sentence ordering as a constrained edge classification problem. We report results on five different datasets, and empirically show that the proposed method is naturally suitable for order prediction. The implementation of this work is publicly available at: https://github.com/declare-lab/sentence-ordering.
|
['Soujanya Poria', 'Rada Mihalcea', 'Navonil Majumder', 'Deepanway Ghosal']
|
2021-09-06
| null |
https://aclanthology.org/2021.emnlp-main.683
|
https://aclanthology.org/2021.emnlp-main.683.pdf
|
emnlp-2021-11
|
['sentence-ordering']
|
['natural-language-processing']
|
[ 2.17061579e-01 1.73441678e-01 -4.07236338e-01 -7.27391779e-01
2.70168334e-01 -5.27000248e-01 7.24355698e-01 8.14313710e-01
-1.58003658e-01 5.69375098e-01 7.90637791e-01 -6.18438900e-01
-3.03276777e-01 -8.87850702e-01 -6.02506459e-01 1.45753980e-01
-4.89612550e-01 5.15698075e-01 2.10751057e-01 -3.78358096e-01
7.53266037e-01 2.05641806e-01 -1.23466456e+00 7.85644472e-01
5.90435266e-01 8.05338979e-01 3.71238381e-01 1.06456840e+00
-6.73289895e-02 1.72335148e+00 -2.02781767e-01 -3.65980804e-01
-2.98422426e-01 -7.76479304e-01 -1.34610105e+00 5.35775386e-02
1.55562624e-01 -1.15977548e-01 -6.95173323e-01 9.76253450e-01
-2.54554689e-01 2.96664894e-01 3.51773292e-01 -1.10006392e+00
-8.99284363e-01 1.21177304e+00 1.12477437e-01 8.91228974e-01
8.88857126e-01 -1.48608044e-01 1.76441073e+00 -4.60204333e-01
1.07478106e+00 8.08902085e-01 5.97519577e-01 3.07754815e-01
-9.53199625e-01 9.95179042e-02 2.94491798e-01 9.91783619e-01
-1.15087485e+00 -4.48578298e-01 1.30334377e+00 -5.25864184e-01
1.53842163e+00 5.93774736e-01 9.55823481e-01 1.01703477e+00
7.73676932e-01 7.10254490e-01 8.84740770e-01 -6.59773648e-01
2.13044748e-01 -3.86372238e-01 9.07935619e-01 8.16955566e-01
-1.64735261e-02 -9.45459157e-02 -1.08973873e+00 1.60628006e-01
9.94788781e-02 -7.23396540e-02 -2.93497473e-01 2.03613549e-01
-1.06261516e+00 5.17624199e-01 4.78245944e-01 6.71609640e-01
-4.49337929e-01 1.90613165e-01 6.49828970e-01 4.53582197e-01
8.63861501e-01 4.13848460e-01 -4.29775745e-01 -1.60750568e-01
-9.43536580e-01 1.59240775e-02 1.13447642e+00 5.98616242e-01
5.63949227e-01 -5.98779678e-01 -3.00624501e-02 2.53355980e-01
3.75118792e-01 -2.68686652e-01 2.37916112e-01 -8.33256125e-01
5.65724850e-01 7.46319890e-01 8.88253562e-03 -1.67715836e+00
-7.12785959e-01 -2.95740157e-01 -6.44898355e-01 -7.41275370e-01
2.07689092e-01 2.25112662e-01 -4.71179426e-01 1.56815791e+00
4.55623046e-02 6.40833497e-01 -1.45050749e-01 6.76724255e-01
7.34124064e-01 8.80350947e-01 -1.78051904e-01 -7.16962874e-01
1.35508132e+00 -5.94776988e-01 -1.02719617e+00 -4.26028877e-01
7.80974329e-01 -3.07936758e-01 9.31508422e-01 8.16125572e-02
-9.87788975e-01 -2.08331198e-01 -9.91918623e-01 -3.31020862e-01
-2.77072370e-01 -1.66860595e-01 7.26072133e-01 -5.10629788e-02
-1.37925088e+00 1.03471422e+00 -6.70790374e-01 -4.29984123e-01
1.37704894e-01 1.07579373e-01 -9.38832685e-02 1.70626745e-01
-1.65506816e+00 8.33167851e-01 8.41222703e-01 5.27455509e-01
-3.23646426e-01 -4.26241755e-01 -8.75647843e-01 9.36978403e-03
4.06003296e-01 -8.24317038e-01 1.23590457e+00 -4.70731318e-01
-1.10893309e+00 1.06788433e+00 -5.98851681e-01 -9.28041697e-01
1.38702035e-01 -3.77625227e-02 -6.10651851e-01 2.04684198e-01
1.79850414e-01 1.29735470e-01 3.69298935e-01 -6.94667816e-01
-4.55696732e-01 -4.65939492e-01 1.24361120e-01 3.02893847e-01
-2.54374355e-01 1.18470088e-01 -3.05917054e-01 -2.46791989e-01
3.65181595e-01 -7.81399190e-01 -1.04814924e-01 -8.24521661e-01
-8.79229486e-01 -5.77319741e-01 3.59045684e-01 -9.16984141e-01
1.96160090e+00 -1.77854025e+00 1.18011020e-01 -7.96194896e-02
4.38261002e-01 -7.30205178e-01 3.10009092e-01 9.60563838e-01
-1.88497871e-01 3.63572240e-01 -1.84441686e-01 -1.83762908e-01
2.18890905e-02 2.73917377e-01 -4.64303881e-01 3.98345947e-01
-2.13022791e-02 1.00014281e+00 -1.24979103e+00 -4.94991392e-01
-1.36720791e-01 -6.00685291e-02 -2.99229920e-01 -8.96022934e-03
-7.12540090e-01 3.41734707e-01 -2.06550747e-01 2.03045681e-01
-1.04959890e-01 -8.10284138e-01 7.43555903e-01 -1.69267908e-01
7.48040387e-03 1.01943958e+00 -6.40648901e-01 1.42807734e+00
-2.26908475e-01 1.04503798e+00 -7.84361780e-01 -8.99581492e-01
6.52041197e-01 2.07708523e-01 2.49713898e-01 -8.00138414e-01
5.70269860e-02 -1.91145569e-01 9.72073004e-02 -8.47126484e-01
8.58599484e-01 -1.56540498e-01 -3.18944305e-01 6.97219074e-01
-2.54263908e-01 3.81603949e-02 7.57132173e-01 7.61545777e-01
1.38879049e+00 -2.42308453e-01 6.13427341e-01 -2.28714317e-01
3.92358124e-01 1.26759261e-01 5.61220646e-01 7.49004126e-01
-8.20200220e-02 6.24586642e-02 9.12389278e-01 -1.02945864e+00
-1.11371696e+00 -7.29080975e-01 2.01946720e-01 9.71385777e-01
1.44323707e-01 -9.33849990e-01 -2.33378157e-01 -5.76285005e-01
-3.61929059e-01 1.49925566e+00 -7.93267787e-01 3.48154418e-02
-8.51869524e-01 -4.27589417e-01 1.54003471e-01 3.91156316e-01
1.33974507e-01 -1.40202546e+00 -5.18293798e-01 1.35184482e-01
-7.31156826e-01 -1.29205561e+00 -6.60345972e-01 1.21817537e-01
-9.54029977e-01 -1.21542037e+00 5.34408212e-01 -7.44508982e-01
7.33913422e-01 -1.15203172e-01 1.48868823e+00 2.80468613e-01
-1.53350919e-01 3.49951357e-01 -4.93311226e-01 -1.08313531e-01
-6.60782278e-01 -2.35220000e-01 -3.01596187e-02 -3.46758850e-02
3.73349637e-01 -7.91406870e-01 -6.04528248e-01 -3.52363169e-01
-7.57507145e-01 3.26240748e-01 -1.32531583e-01 6.70190752e-01
3.24542135e-01 5.13046443e-01 1.06681116e-01 -1.18159819e+00
1.12789965e+00 -5.76610982e-01 -3.43296617e-01 3.18439782e-01
-7.25599051e-01 6.64501637e-02 6.57841861e-01 5.75002581e-02
-8.88025045e-01 -2.88983047e-01 2.74017751e-01 1.99166968e-01
-4.98321019e-02 1.12763596e+00 2.80609846e-01 8.47958088e-01
4.96675402e-01 4.74481821e-01 -5.20408332e-01 7.68223330e-02
3.97879988e-01 1.11221239e-01 4.79373276e-01 -3.52787107e-01
1.13089226e-01 4.26110178e-01 1.37729585e-01 -5.69465697e-01
-1.23206675e+00 -5.30750811e-01 -9.39895809e-01 -6.05002940e-01
7.48802185e-01 -4.39106017e-01 -7.70071268e-01 1.08050592e-01
-1.69346333e+00 -4.43356276e-01 -1.16225623e-01 7.33768865e-02
-5.55339813e-01 5.76650798e-01 -9.22397852e-01 -8.12310040e-01
-3.04857075e-01 -4.64363545e-01 6.37202203e-01 -3.84068815e-03
-7.76049733e-01 -1.56792486e+00 2.39667982e-01 3.36411715e-01
-2.38984644e-01 2.20537856e-01 9.34193134e-01 -7.24593937e-01
-6.22776806e-01 -2.79624939e-01 1.29532963e-01 -2.08208725e-01
2.47695133e-01 1.62009165e-01 -4.20269221e-01 2.23952010e-01
1.91324979e-01 9.34100896e-02 9.27926421e-01 5.07284224e-01
9.78781641e-01 -9.21990335e-01 -3.91474336e-01 2.70100497e-02
1.28479373e+00 2.18245044e-01 4.27836865e-01 3.86093706e-01
5.75668097e-01 7.35158026e-01 4.26133931e-01 6.70591056e-01
8.87395799e-01 3.47153634e-01 4.10379082e-01 8.02692592e-01
1.77045483e-02 -4.42372352e-01 2.46115267e-01 1.32899261e+00
5.14074415e-02 -5.92173338e-01 -1.38123572e+00 8.81933212e-01
-2.07304859e+00 -1.52019024e+00 -4.70815748e-01 1.59388041e+00
9.68539178e-01 5.21058440e-01 -1.10868804e-01 2.01884255e-01
8.52249205e-01 6.66401148e-01 -1.50230616e-01 -6.37407660e-01
-2.38831919e-02 -4.59506899e-01 -2.49580089e-02 1.00454390e+00
-1.05683124e+00 8.08153391e-01 6.00181580e+00 2.47462124e-01
-8.02766383e-01 -6.94255754e-02 8.24436188e-01 1.35429054e-01
-5.41384995e-01 4.35712576e-01 -6.86457455e-01 5.58691502e-01
1.29421234e+00 -5.89899778e-01 6.70908332e-01 3.40054423e-01
5.60710907e-01 -1.90126419e-01 -1.35653424e+00 5.89326501e-01
1.48689151e-01 -1.94099247e+00 -2.06493422e-01 -2.81326711e-01
3.91094863e-01 -1.18218074e-02 -3.52004051e-01 -2.04864085e-01
3.81351441e-01 -6.46997869e-01 8.53533864e-01 9.21272218e-01
2.35406086e-01 -4.10583735e-01 5.20980418e-01 6.00001514e-01
-1.14755380e+00 -9.58036408e-02 -4.76552034e-03 -7.56005168e-01
3.74339551e-01 1.13150871e+00 -1.32654917e+00 7.14369237e-01
5.79659998e-01 1.29200125e+00 -6.52510703e-01 4.93554741e-01
-7.46898830e-01 7.26500630e-01 -8.73930287e-03 -5.68109989e-01
-2.11149052e-01 -1.65812328e-01 7.89524496e-01 1.21943760e+00
-1.62526183e-02 2.96247572e-01 3.12150363e-02 6.12182796e-01
-2.84584463e-01 -1.01200163e-01 -9.41029012e-01 -2.67003834e-01
5.03233314e-01 9.35002685e-01 -1.24541056e+00 -2.32967868e-01
-1.69226453e-02 1.11967432e+00 6.74474359e-01 7.45792389e-02
-6.81137323e-01 1.44236851e-02 -7.54792010e-03 2.27507364e-04
1.25746876e-01 -5.11251688e-01 -4.98712957e-01 -1.21930683e+00
3.42165649e-01 -2.37125874e-01 6.92616761e-01 -9.13892508e-01
-1.32784641e+00 8.09820414e-01 -1.31335825e-01 -9.11105931e-01
-3.62865359e-01 -4.27982241e-01 -9.39213276e-01 3.27345639e-01
-1.12886000e+00 -7.86623001e-01 2.82460712e-02 1.51570588e-01
6.24854147e-01 3.82541418e-01 5.67938268e-01 -3.41212690e-01
-4.95579183e-01 -9.41295028e-02 -2.70468831e-01 2.02794090e-01
1.13121860e-01 -1.45770347e+00 5.74848354e-01 1.07196474e+00
5.96424162e-01 7.24995792e-01 1.20691776e+00 -1.05186129e+00
-1.26259434e+00 -8.69622290e-01 2.14229631e+00 -7.04792440e-01
1.46071398e+00 -2.65748352e-01 -6.74357057e-01 1.00416338e+00
6.14537716e-01 -2.15434328e-01 8.09097946e-01 5.48217177e-01
-3.22036564e-01 -6.74000755e-02 -3.82350683e-01 8.03585291e-01
1.46104145e+00 -9.44482744e-01 -9.18960333e-01 8.06820154e-01
1.07806718e+00 -3.89750034e-01 -8.28552067e-01 7.07320720e-02
2.02606484e-01 -1.05436146e+00 6.23957932e-01 -8.26260090e-01
1.00346196e+00 -1.41836092e-01 -1.68160066e-01 -1.13090503e+00
-5.29509068e-01 -5.67908168e-01 -5.92256188e-01 9.67635751e-01
5.89842021e-01 -3.96700084e-01 7.49303043e-01 3.99955094e-01
-1.14660263e-01 -7.37432420e-01 -8.91838908e-01 -6.77770555e-01
-5.24517655e-01 -8.51717889e-01 3.16861778e-01 1.18631649e+00
6.17371500e-01 8.23832452e-01 -3.72202873e-01 2.54215956e-01
4.52534705e-01 4.97502714e-01 -5.98936789e-02 -1.17078030e+00
-3.31364185e-01 -4.64111656e-01 -3.69346857e-01 -7.72391796e-01
6.09033763e-01 -1.29387569e+00 8.70202109e-02 -1.91685319e+00
3.56710106e-01 3.25319290e-01 -2.73817867e-01 3.12843591e-01
-6.05006926e-02 -2.63473064e-01 4.91785333e-02 3.20148528e-01
-1.16073191e+00 3.27183008e-01 1.10000622e+00 -3.38576406e-01
-2.55351275e-01 -8.20496008e-02 -5.32653451e-01 6.12532139e-01
1.08204472e+00 -4.98290777e-01 -5.91353059e-01 -3.96044195e-01
1.11899149e+00 4.29101229e-01 4.63217229e-01 -5.58055103e-01
9.23063934e-01 -3.35273713e-01 5.60824163e-02 -8.13241184e-01
-2.43108198e-02 -5.78345478e-01 1.62053734e-01 4.82304603e-01
-8.40309381e-01 4.43769366e-01 -2.15907171e-02 8.75299454e-01
-2.01221183e-01 -1.77319333e-01 4.11363989e-02 -1.96874529e-01
-1.15235388e+00 9.09323096e-02 -1.83910310e-01 2.69094910e-02
7.48414934e-01 -4.77139242e-02 -5.66941381e-01 -6.76520705e-01
-1.12231755e+00 1.43597707e-01 2.04842150e-01 4.08013135e-01
9.33852017e-01 -1.11001337e+00 -6.49515748e-01 -4.22586650e-01
-1.18609723e-02 -4.69118297e-01 2.32972533e-01 9.08001184e-01
-6.81687236e-01 7.03184545e-01 2.52766550e-01 -3.87626320e-01
-1.24134541e+00 7.16763556e-01 1.08551234e-01 -6.90311372e-01
-7.19033778e-01 8.28185856e-01 -3.47759277e-01 -1.33893549e-01
-1.90021262e-01 -4.76941824e-01 -4.32969630e-01 7.97895864e-02
3.66321564e-01 1.68055147e-01 -1.12692460e-01 -5.04199803e-01
-6.74928308e-01 5.63717112e-02 -1.80917114e-01 -2.45770160e-02
1.57233715e+00 -4.44870502e-01 -1.05594051e+00 1.09094048e+00
1.00254703e+00 -1.08500563e-01 -7.07922220e-01 -1.05691776e-01
8.52011383e-01 -1.80961326e-01 5.21836383e-03 -6.48049235e-01
-2.78584033e-01 4.32993472e-01 -4.21437711e-01 1.13705862e+00
1.11754894e+00 3.83459657e-01 5.74908435e-01 3.70345831e-01
2.03510001e-01 -9.94906068e-01 3.06131572e-01 1.02522862e+00
1.07589328e+00 -1.00915837e+00 2.34503493e-01 -6.12101853e-01
-5.21818638e-01 1.33944988e+00 4.08227712e-01 -1.00369923e-01
8.09783757e-01 -1.21632107e-01 -4.79666054e-01 -5.57558417e-01
-1.27767718e+00 7.48622492e-02 5.55987477e-01 -2.62185130e-02
6.72309399e-01 1.98480695e-01 -4.53529119e-01 4.29057449e-01
-3.99138004e-01 -1.66858751e-02 7.19026089e-01 7.86732495e-01
-3.72450799e-01 -9.22367930e-01 -9.38212350e-02 5.26234508e-01
-1.40975013e-01 -4.97730434e-01 -7.47879088e-01 2.49730110e-01
-1.10835359e-01 1.15301776e+00 2.82508850e-01 -6.36381149e-01
8.40919744e-03 2.54268616e-01 5.05774140e-01 -6.35386765e-01
-1.96799442e-01 -3.03899437e-01 6.85288250e-01 -5.31987309e-01
-5.22195458e-01 -8.46526027e-01 -1.41785800e+00 -4.48792845e-01
6.27364665e-02 1.48656562e-01 2.34940305e-01 1.10586298e+00
4.40676898e-01 7.75089860e-01 6.39717102e-01 -3.02094847e-01
1.49340937e-02 -7.36247003e-01 -5.03908694e-01 4.74508584e-01
2.86546171e-01 -4.30197939e-02 -3.67815763e-01 4.46341693e-01]
|
[11.587079048156738, 9.145392417907715]
|
4cacc6dd-6a7a-467a-97fe-1c51dbac0a96
|
blind-room-parameter-estimation-using
|
2107.13832
| null |
https://arxiv.org/abs/2107.13832v1
|
https://arxiv.org/pdf/2107.13832v1.pdf
|
Blind Room Parameter Estimation Using Multiple-Multichannel Speech Recordings
|
Knowing the geometrical and acoustical parameters of a room may benefit applications such as audio augmented reality, speech dereverberation or audio forensics. In this paper, we study the problem of jointly estimating the total surface area, the volume, as well as the frequency-dependent reverberation time and mean surface absorption of a room in a blind fashion, based on two-channel noisy speech recordings from multiple, unknown source-receiver positions. A novel convolutional neural network architecture leveraging both single- and inter-channel cues is proposed and trained on a large, realistic simulated dataset. Results on both simulated and real data show that using multiple observations in one room significantly reduces estimation errors and variances on all target quantities, and that using two channels helps the estimation of surface and volume. The proposed model outperforms a recently proposed blind volume estimation method on the considered datasets.
|
['Emmanuel Vincent', 'Antoine Deleforge', 'Prerak Srivastava']
|
2021-07-29
| null | null | null | null |
['speech-dereverberation']
|
['speech']
|
[ 2.13228568e-01 -2.09425926e-01 1.00009203e+00 -2.06684977e-01
-1.26074982e+00 -5.25156736e-01 2.93569863e-01 3.41315091e-01
-3.13092679e-01 5.83157957e-01 4.52838004e-01 -2.33074129e-01
-2.43068561e-01 -5.07921517e-01 -6.93431497e-01 -9.73245800e-01
-4.59772289e-01 7.53906071e-02 -3.75564694e-01 1.64701015e-01
-2.61013880e-02 4.81332779e-01 -1.74821234e+00 -2.35851690e-01
6.20804906e-01 1.47550678e+00 9.39144939e-02 1.02706873e+00
2.29820475e-01 4.21671867e-01 -8.84562790e-01 7.82758184e-03
2.83636451e-01 4.81129140e-02 -9.52902585e-02 6.65684715e-02
7.85474539e-01 -3.51497024e-01 -1.65084481e-01 6.62281394e-01
1.08671141e+00 4.30072069e-01 7.12147176e-01 -4.05862540e-01
-1.43418610e-01 1.70866176e-01 -2.73990095e-01 3.05422768e-02
4.61724877e-01 -2.47660559e-02 6.26890481e-01 -7.30560482e-01
-3.38946044e-01 8.70385945e-01 9.92110252e-01 4.97120693e-02
-9.40783203e-01 -6.66981936e-01 -2.54128873e-01 -1.21006660e-01
-1.52314365e+00 -8.47650766e-01 8.83165479e-01 -5.35203695e-01
6.32926166e-01 4.83860791e-01 3.73410255e-01 9.07242358e-01
-1.83818534e-01 1.51759684e-01 1.22332692e+00 -5.59393764e-01
3.67261887e-01 1.49702311e-01 -1.95362553e-01 6.06159806e-01
3.04623455e-01 2.59932846e-01 -4.86257732e-01 -5.10393322e-01
5.05642056e-01 -3.75758439e-01 -8.08010101e-01 -2.98439741e-01
-1.24143994e+00 4.55673128e-01 2.84490585e-01 4.68654156e-01
-4.28527743e-01 2.34290347e-01 1.41545460e-01 6.41505048e-02
6.05588675e-01 4.92165267e-01 -2.73980200e-01 -3.92709933e-02
-9.41007078e-01 -5.71517274e-02 9.01026189e-01 3.92215937e-01
4.88897294e-01 3.61338615e-01 -2.20485047e-01 8.25253546e-01
7.42550254e-01 1.24636137e+00 1.13080420e-01 -6.68528855e-01
5.15703738e-01 -2.19122097e-01 5.41782022e-01 -1.07535863e+00
-7.98007309e-01 -7.16214776e-01 -7.56361425e-01 2.65028980e-03
5.89773893e-01 -4.54397649e-01 -1.05368292e+00 1.61249697e+00
4.94737446e-01 4.80827868e-01 5.73358573e-02 1.13391399e+00
1.03713953e+00 5.34412920e-01 -3.04611832e-01 -2.65959233e-01
1.25475585e+00 -6.04417145e-01 -7.88426399e-01 -5.93622088e-01
1.27476096e-01 -9.83179152e-01 5.36848843e-01 7.57176101e-01
-8.30886245e-01 -5.50173163e-01 -1.04885232e+00 3.25253010e-01
-1.18493669e-01 5.53041875e-01 2.09553361e-01 1.32500112e+00
-8.51472080e-01 1.44617349e-01 -7.15056956e-01 1.36770844e-01
7.49731734e-02 1.65113121e-01 -2.50934631e-01 -1.67560756e-01
-8.10017943e-01 3.53979439e-01 -5.09192884e-01 8.21988344e-01
-1.10524917e+00 -7.63368726e-01 -9.82800007e-01 2.43646037e-02
1.81884989e-01 -5.92646360e-01 1.02791119e+00 -5.21734893e-01
-1.39448011e+00 2.25362077e-01 -2.51466393e-01 -3.81215364e-01
5.57572246e-01 -4.63491619e-01 -5.64261913e-01 1.08464040e-01
-2.05330223e-01 -2.56437272e-01 1.34368718e+00 -1.68699074e+00
9.67283994e-02 -5.55376291e-01 -3.07395488e-01 8.42628404e-02
-2.28648812e-01 -4.75499243e-01 -5.26341936e-03 -4.09275830e-01
4.02808785e-01 -5.53294182e-01 -1.30736858e-01 -4.34341803e-02
-6.46994114e-01 6.28945410e-01 1.34478435e-01 -1.39136159e+00
7.94752538e-01 -2.26104832e+00 -1.55889124e-01 4.12019908e-01
1.91542372e-01 1.29893394e-02 7.92914182e-02 1.72013849e-01
6.52349591e-02 -4.51351583e-01 -6.44868374e-01 -8.41890216e-01
-2.02911988e-01 -4.71054614e-01 -3.00606862e-02 1.05745733e+00
-4.80857879e-01 2.37459391e-01 -5.63252151e-01 -8.29217583e-02
4.77890670e-01 1.08059943e+00 -1.32512525e-01 5.31463861e-01
4.02894437e-01 9.27359879e-01 -6.34243563e-02 5.74898958e-01
9.98665154e-01 1.93969116e-01 -8.64823684e-02 -1.22036159e-01
9.27469507e-03 1.97909012e-01 -1.48387790e+00 1.52572513e+00
-1.52002621e+00 7.49049723e-01 8.43075097e-01 -7.12709486e-01
1.20183110e+00 6.56894028e-01 2.29170114e-01 -7.73208201e-01
3.91508430e-01 3.79474968e-01 -2.29110539e-01 -5.79499662e-01
2.36674160e-01 -3.69838685e-01 1.39002234e-01 4.93745029e-01
-6.43561631e-02 -1.25482917e-01 -6.56587124e-01 -2.39367932e-01
1.02671230e+00 -6.66460693e-01 1.91840038e-01 3.19017246e-02
8.17630470e-01 -9.48291302e-01 -8.78014341e-02 7.20091283e-01
6.55839741e-02 8.53351474e-01 -1.44578531e-01 7.60250315e-02
-7.81320035e-01 -1.06870949e+00 -1.20474920e-01 6.79868340e-01
-1.94861237e-02 1.55297488e-01 -7.38599420e-01 -1.02291822e-01
1.79300502e-01 7.12836325e-01 -6.59164846e-01 1.51605308e-01
-4.65165198e-01 -6.15044594e-01 4.53160912e-01 2.32551202e-01
2.16029212e-01 -6.41100168e-01 -4.35136288e-01 8.02052543e-02
-3.89429778e-01 -1.42723942e+00 -2.03811556e-01 2.00624779e-01
-3.99423331e-01 -8.64586949e-01 -8.19098175e-01 -3.72530460e-01
4.88958627e-01 3.80701393e-01 7.20975816e-01 -3.60963345e-02
-4.40768808e-01 9.95632470e-01 -2.44489640e-01 -5.90405047e-01
-3.28631729e-01 -4.50153232e-01 1.69896498e-01 6.35524571e-01
-3.47758085e-01 -7.26564348e-01 -6.68913782e-01 3.89298558e-01
-4.04293865e-01 -5.05924225e-01 2.49498621e-01 4.31290984e-01
1.11460976e-01 3.20284031e-02 4.33674902e-01 -2.02542856e-01
6.20496631e-01 -3.03305328e-01 -8.41072559e-01 6.10265583e-02
-3.69824499e-01 -1.15956865e-01 5.47346592e-01 -1.53292000e-01
-1.20573449e+00 5.91756180e-02 -1.67610124e-01 -2.20102608e-01
-5.18087685e-01 2.12766856e-01 -5.07237852e-01 -3.37550461e-01
8.05027544e-01 2.17204809e-01 -1.88976437e-01 -6.70751035e-01
2.98994267e-03 9.61611807e-01 5.05458474e-01 -2.98044711e-01
9.51994538e-01 7.39946485e-01 2.30651408e-01 -1.34593213e+00
-3.53927642e-01 -9.78864312e-01 -3.56843531e-01 -2.08311573e-01
5.35862803e-01 -1.18494380e+00 -8.11003387e-01 6.47311628e-01
-1.26603973e+00 -3.88246030e-01 -6.29684031e-02 8.56882513e-01
-3.47845405e-01 5.13113141e-01 7.05462992e-02 -1.60903597e+00
-5.11134267e-01 -9.57072556e-01 1.25316584e+00 -1.16010226e-01
4.12511528e-02 -9.90990281e-01 4.40083742e-02 7.18759298e-01
7.43818223e-01 1.27101704e-01 2.68184036e-01 -3.63730729e-01
-4.11612004e-01 -5.58291972e-01 -6.51651099e-02 4.78009790e-01
2.58680314e-01 -7.42480695e-01 -1.77044463e+00 -3.57388586e-01
2.98490018e-01 -7.73551837e-02 7.60023355e-01 9.01720524e-01
1.01948273e+00 -3.06652755e-01 -8.19033682e-02 7.17921257e-01
1.43431008e+00 -1.56192064e-01 4.86961067e-01 9.32226330e-03
8.40338886e-01 6.25340641e-01 1.94514349e-01 8.57661068e-01
1.01395003e-01 4.84059453e-01 8.55667353e-01 -9.10740495e-02
-1.61778614e-01 1.75113991e-01 1.71531782e-01 6.72444165e-01
-1.04566477e-01 -6.32426620e-01 -7.26359904e-01 7.70944238e-01
-9.00946558e-01 -4.94593263e-01 -2.05343992e-01 2.77164555e+00
2.16205820e-01 -3.29652607e-01 -1.88634813e-01 4.53730226e-01
4.98228043e-01 3.10250878e-01 -1.69112131e-01 -2.85556316e-01
4.21028724e-03 4.26377475e-01 9.58866894e-01 1.00408149e+00
-1.03179777e+00 -1.60210744e-01 6.13143778e+00 2.57322907e-01
-1.18694758e+00 2.68644273e-01 3.66489977e-01 -7.71650597e-02
-4.09139872e-01 -5.72882295e-01 -4.30124760e-01 1.47829711e-01
1.08879781e+00 3.52737248e-01 7.22581685e-01 4.29085851e-01
3.25299621e-01 -2.74881274e-01 -8.45597565e-01 1.20120585e+00
5.19696295e-01 -8.14989150e-01 -5.46116948e-01 8.18941221e-02
3.52220058e-01 3.41362134e-03 2.91731864e-01 -3.17220002e-01
-3.04461807e-01 -1.21043110e+00 6.59229696e-01 7.21404910e-01
7.16723204e-01 -5.09831131e-01 8.96226764e-01 1.96648702e-01
-1.28166592e+00 -1.54039398e-01 2.18488947e-01 2.00944364e-01
2.62226462e-01 9.08714890e-01 -1.10671449e+00 5.23549318e-01
6.18512213e-01 8.55221897e-02 -4.10597742e-01 1.53475392e+00
-2.58086264e-01 8.76647770e-01 -5.32509923e-01 1.21902861e-01
-1.27203733e-01 1.41335756e-01 7.33290792e-01 1.07245767e+00
6.91240966e-01 8.66037458e-02 -2.95495957e-01 7.54366219e-01
2.52539273e-02 4.92238291e-02 -3.67858291e-01 3.41612697e-01
3.14385265e-01 1.35391021e+00 -4.75373715e-01 1.34530917e-01
-1.70113847e-01 6.01470232e-01 -5.17401397e-01 7.49857545e-01
-6.60432875e-01 -3.83553922e-01 4.42085475e-01 2.88361043e-01
5.55551708e-01 -5.27192652e-01 -3.88086826e-01 -5.08300543e-01
3.29616755e-01 -6.61859989e-01 -2.38180086e-01 -7.37285554e-01
-1.01228833e+00 6.32637799e-01 -2.42723674e-01 -1.23004365e+00
-3.37917283e-02 -7.05506086e-01 -6.25919938e-01 1.22860277e+00
-1.63486087e+00 -9.71328497e-01 -6.43870115e-01 5.28185606e-01
1.44098133e-01 -3.02662794e-02 8.77126157e-01 5.32400250e-01
-3.52526397e-01 6.33600414e-01 3.87879163e-01 1.08747117e-01
6.53237343e-01 -1.02788866e+00 4.31368530e-01 7.25290358e-01
6.74148425e-02 3.51899564e-01 1.20903063e+00 -1.79474324e-01
-1.41801298e+00 -9.26174283e-01 4.23614293e-01 -2.36501485e-01
3.12027931e-01 -5.61974943e-01 -8.32194924e-01 2.03122571e-01
-1.96847931e-01 4.11680818e-01 5.78770041e-01 1.20270394e-01
-3.23874712e-01 -1.89339221e-01 -1.23958445e+00 -1.51780427e-01
5.00704050e-01 -6.81614101e-01 -5.76118082e-02 4.41320121e-01
5.16753554e-01 -5.40304840e-01 -6.00197077e-01 7.97239169e-02
5.55157721e-01 -1.00820351e+00 1.19658816e+00 3.52531463e-01
-1.57706171e-01 -1.37896031e-01 -5.21438420e-01 -1.51344252e+00
1.63753927e-01 -6.02385402e-01 -1.16961211e-01 1.22258425e+00
4.41978723e-01 -7.29721487e-01 6.12573683e-01 3.12637240e-01
-2.12032557e-01 -3.20334524e-01 -1.34624767e+00 -6.71420693e-01
-3.64604145e-01 -8.74588966e-01 5.92334807e-01 5.08341372e-01
-7.05441952e-01 1.52445078e-01 -6.16917133e-01 1.00619197e+00
9.35214877e-01 -1.14630543e-01 8.32189679e-01 -1.38025486e+00
-4.64940369e-01 1.19607054e-01 -3.42142671e-01 -8.97414625e-01
-7.27267051e-03 -3.08675975e-01 5.04793644e-01 -1.68235970e+00
-5.25505006e-01 -5.54144442e-01 -5.91060221e-02 -1.01666272e-01
9.56180915e-02 1.56392008e-01 -3.00582707e-01 -4.56881851e-01
-3.79027203e-02 5.77632070e-01 1.00200355e+00 -2.08268806e-01
-2.31203556e-01 4.33719248e-01 -3.46415550e-01 6.37043178e-01
5.64445257e-01 -3.67159903e-01 -1.21339962e-01 -5.20624697e-01
1.75621912e-01 3.74651432e-01 8.12535703e-01 -1.43538094e+00
9.22573954e-02 3.39799255e-01 4.06608701e-01 -2.26845711e-01
9.41645086e-01 -1.07734954e+00 -1.31902963e-01 8.57821759e-03
-1.09319188e-01 -5.87520897e-01 5.99382937e-01 9.13424373e-01
-4.37479839e-02 3.71638499e-02 6.18434250e-01 3.61021794e-02
-5.96145354e-02 -4.10017185e-02 -1.28202647e-01 -2.29019910e-01
4.04136688e-01 -1.61249727e-01 8.00104253e-03 -8.68415713e-01
-4.85800803e-01 -3.29111397e-01 -8.82634893e-02 1.13247201e-01
7.04690933e-01 -9.45770621e-01 -8.34628403e-01 3.85612518e-01
1.21558204e-01 3.72907110e-02 8.26045930e-01 1.05533373e+00
-4.09098625e-01 5.17871141e-01 2.64795572e-01 -5.67571223e-01
-1.43355227e+00 2.55982250e-01 8.48769546e-01 2.23481283e-01
-3.41694891e-01 1.30301738e+00 1.13266274e-01 -6.51022434e-01
5.65620661e-01 -3.65628272e-01 -2.65531600e-01 -3.84681039e-02
6.37282968e-01 5.47394216e-01 6.25298858e-01 -9.81352270e-01
-3.72892261e-01 7.81014860e-01 6.71088398e-01 -3.13825995e-01
1.13096046e+00 -5.54270327e-01 4.58094254e-02 7.80794501e-01
1.38167131e+00 6.20624304e-01 -9.87458169e-01 -2.50421494e-01
-5.70505321e-01 -5.75256050e-01 6.34377599e-01 -9.44367707e-01
-1.01012385e+00 1.21895707e+00 1.06904328e+00 2.31922343e-01
1.20483708e+00 -1.90857291e-01 7.22388029e-01 2.79620737e-01
2.96470553e-01 -6.14944696e-01 -1.56781578e-03 8.86611119e-02
9.21265721e-01 -1.20288277e+00 7.32730776e-02 -3.62918139e-01
-1.79299250e-01 9.38502312e-01 1.60121217e-01 2.56972909e-01
9.30774689e-01 1.69409364e-01 3.08702111e-01 -1.03040509e-01
-9.41628665e-02 -1.59390904e-02 5.79065681e-01 8.20033967e-01
4.41587776e-01 2.30409041e-01 4.65409815e-01 4.54630077e-01
-3.80053043e-01 -4.84015852e-01 4.74678218e-01 5.60426950e-01
-4.81783658e-01 -3.99107397e-01 -1.33290362e+00 3.54962021e-01
-5.18918276e-01 -1.58744633e-01 -1.09752387e-01 2.83923894e-01
1.35291159e-01 1.30665445e+00 1.58595026e-01 -2.58422226e-01
4.82964844e-01 -1.67106465e-01 4.88551497e-01 -3.72239381e-01
-5.63008308e-01 2.33868465e-01 2.54559904e-01 -3.83078963e-01
-4.20047134e-01 -5.80912709e-01 -9.54176188e-01 -8.20265114e-02
-7.10335970e-01 2.40930676e-01 1.33881855e+00 9.20901060e-01
-1.06498655e-02 8.75427783e-01 8.96345973e-01 -1.15830779e+00
-4.10343856e-01 -1.19850993e+00 -1.10393929e+00 -2.39016622e-01
1.30257654e+00 -7.69967675e-01 -1.06526089e+00 -3.11537445e-01]
|
[15.152462005615234, 5.771498680114746]
|
356bc62e-2ce1-4fda-9f3c-9c09cdd47a5a
|
refu-refine-and-fuse-the-unobserved-view-for
|
2211.04753
| null |
https://arxiv.org/abs/2211.04753v1
|
https://arxiv.org/pdf/2211.04753v1.pdf
|
ReFu: Refine and Fuse the Unobserved View for Detail-Preserving Single-Image 3D Human Reconstruction
|
Single-image 3D human reconstruction aims to reconstruct the 3D textured surface of the human body given a single image. While implicit function-based methods recently achieved reasonable reconstruction performance, they still bear limitations showing degraded quality in both surface geometry and texture from an unobserved view. In response, to generate a realistic textured surface, we propose ReFu, a coarse-to-fine approach that refines the projected backside view image and fuses the refined image to predict the final human body. To suppress the diffused occupancy that causes noise in projection images and reconstructed meshes, we propose to train occupancy probability by simultaneously utilizing 2D and 3D supervisions with occupancy-based volume rendering. We also introduce a refinement architecture that generates detail-preserving backside-view images with front-to-back warping. Extensive experiments demonstrate that our method achieves state-of-the-art performance in 3D human reconstruction from a single image, showing enhanced geometry and texture quality from an unobserved view.
|
['Jaegul Choo', 'Minsoo Lee', 'Gyumin Shim']
|
2022-11-09
| null | null | null | null |
['3d-human-reconstruction']
|
['computer-vision']
|
[ 7.74000466e-01 4.19686258e-01 4.64643717e-01 -3.11780602e-01
-9.16827083e-01 -4.19924408e-03 4.87791985e-01 -5.10197699e-01
-8.79645068e-03 4.77374852e-01 4.52107430e-01 3.14435303e-01
3.11545521e-01 -8.90907764e-01 -8.43760133e-01 -4.58729535e-01
3.55634391e-01 1.06226885e+00 4.53586102e-01 -1.23412773e-01
-1.99303776e-01 3.56685936e-01 -1.92466605e+00 6.04476273e-01
3.90972525e-01 6.57894194e-01 1.28156051e-01 7.54002333e-01
3.16375494e-01 3.22572678e-01 -1.96475927e-02 -6.65338635e-02
5.48484623e-01 -5.68038881e-01 -5.31213939e-01 6.61038041e-01
8.20099771e-01 -5.47646463e-01 -2.45387331e-01 5.56538343e-01
4.71090943e-01 5.75578623e-02 6.19653523e-01 -5.63541353e-01
-1.46965146e-01 -3.41147393e-01 -8.75717640e-01 -3.43212306e-01
8.53522122e-01 2.03569993e-01 4.92169261e-01 -1.20494509e+00
1.05987537e+00 1.46990299e+00 7.82280147e-01 7.89116144e-01
-1.80462074e+00 -3.87294888e-01 3.41902897e-02 -6.97626412e-01
-1.19871736e+00 -3.92062247e-01 7.17649877e-01 -6.08316422e-01
6.98489070e-01 5.11063218e-01 1.22259593e+00 9.46904778e-01
5.17845452e-01 4.50049311e-01 1.58162725e+00 -3.70086700e-01
1.40398704e-02 -2.60871798e-01 -4.56560731e-01 1.12798071e+00
1.41401635e-03 4.11488056e-01 -9.24702406e-01 -4.03233945e-01
1.55773997e+00 1.31812751e-01 -3.76032680e-01 -7.96158075e-01
-1.36337149e+00 3.47507030e-01 8.42783898e-02 -4.64597374e-01
-5.08862138e-01 2.03445226e-01 -9.77983028e-02 -2.49721985e-02
8.71669531e-01 1.73860893e-01 -1.41730055e-01 1.31059214e-01
-9.92627144e-01 6.24206364e-01 4.93639052e-01 7.23901868e-01
8.20603728e-01 -1.60241112e-01 -1.76648051e-02 5.72311223e-01
4.90098208e-01 8.48474801e-01 -3.54666531e-01 -1.45933485e+00
1.42020553e-01 4.55720544e-01 2.40176186e-01 -6.53921008e-01
-3.42222422e-01 -1.76392883e-01 -9.04751182e-01 8.14845800e-01
2.40876600e-01 2.06952110e-01 -1.40500212e+00 1.31585026e+00
9.23358142e-01 1.20485397e-02 -5.05663514e-01 1.38272429e+00
5.14722943e-01 4.84766066e-01 -1.94607347e-01 -3.33792390e-03
1.35020709e+00 -8.39294493e-01 -4.91361111e-01 -2.80418098e-01
-1.72465920e-01 -5.88027835e-01 1.17417896e+00 5.65882742e-01
-1.63212323e+00 -5.10784566e-01 -8.20891440e-01 -4.19680953e-01
5.65510273e-01 -4.75702405e-01 9.96152684e-02 4.09842998e-01
-9.70054686e-01 7.55463839e-01 -1.17501163e+00 -9.34120938e-02
2.23836318e-01 3.06786329e-01 -5.89578688e-01 -3.50600451e-01
-7.03470588e-01 6.52171731e-01 -3.29744399e-01 -1.85273275e-01
-1.03022921e+00 -1.06641936e+00 -9.83500063e-01 -3.42673272e-01
3.65071148e-01 -1.61801326e+00 1.01249170e+00 -6.07978225e-01
-1.60425222e+00 1.19148862e+00 -2.10199311e-01 7.04282597e-02
7.66876996e-01 -2.95862168e-01 2.18839735e-01 2.27363214e-01
2.62650818e-01 6.81491494e-01 1.02585900e+00 -1.62945426e+00
-3.84342462e-01 -7.61705458e-01 -4.08158392e-01 6.51940882e-01
5.75945199e-01 -4.56391275e-01 -7.06643760e-01 -6.42987192e-01
7.32070148e-01 -1.01618898e+00 -4.36651111e-01 4.67426956e-01
-3.49224478e-01 6.17395401e-01 6.25361145e-01 -9.16789114e-01
7.46522963e-01 -1.82191539e+00 4.26317185e-01 4.77279067e-01
3.06077600e-01 -5.52593350e-01 1.72208160e-01 1.09578460e-01
1.75357133e-01 -1.23301953e-01 -4.80468363e-01 -9.36453164e-01
-2.08160549e-01 2.42567122e-01 2.46045794e-02 7.36547947e-01
-5.96520640e-02 6.60051525e-01 -8.12142074e-01 -4.68338728e-01
6.28310800e-01 1.08687913e+00 -9.59963679e-01 3.85369480e-01
-2.36543715e-01 1.22322249e+00 -3.64180624e-01 4.98763949e-01
5.93761325e-01 -4.48560745e-01 3.57863784e-01 -2.19640091e-01
-5.57190627e-02 1.45604923e-01 -1.08749056e+00 2.17190313e+00
-3.31278592e-01 -1.14739664e-01 5.44236422e-01 -1.95152506e-01
6.84150338e-01 2.99268812e-01 7.62319744e-01 -7.94070244e-01
-1.35333091e-01 1.10071495e-01 -6.12825274e-01 -4.85882163e-02
4.37141955e-01 -7.39757657e-01 -1.26376763e-01 4.06631559e-01
-3.92522216e-01 -5.94161868e-01 -6.89129770e-01 -6.51678666e-02
9.51697230e-01 7.28298247e-01 -1.00498445e-01 -3.42708707e-01
2.19710395e-01 -1.63085852e-02 3.79499048e-01 4.44854051e-01
3.96584034e-01 1.62623465e+00 4.34013680e-02 -7.73846269e-01
-1.58993769e+00 -1.51264906e+00 -7.02162012e-02 6.77816570e-01
2.89204687e-01 -3.23355347e-01 -8.55055153e-01 -1.82366982e-01
1.74774006e-02 2.68202126e-01 -8.31294000e-01 2.14869931e-01
-9.20285344e-01 -3.13757032e-01 -8.54472257e-03 1.91858351e-01
1.97760180e-01 -8.84081841e-01 -1.09254599e+00 2.62200534e-01
-3.58921438e-01 -7.52231538e-01 -5.56129217e-01 -5.62764145e-02
-1.01512039e+00 -1.03210199e+00 -1.14905763e+00 -5.36803067e-01
8.92103434e-01 7.40276277e-02 1.56542242e+00 2.00196326e-01
-4.98684406e-01 5.36872506e-01 9.49716568e-02 1.76310793e-01
-3.35594416e-01 -4.30777401e-01 1.11647949e-01 -5.70618883e-02
-6.21534824e-01 -8.03039372e-01 -1.03617346e+00 5.35245717e-01
-5.41262865e-01 8.44147921e-01 1.33979827e-01 1.02077925e+00
1.32051373e+00 -2.22055659e-01 -4.85737860e-01 -9.48661864e-01
-4.71260995e-02 2.17385292e-02 -4.12107706e-01 -1.14015266e-01
-3.75267565e-01 2.87465230e-02 1.46078184e-01 -7.28814339e-04
-1.38305140e+00 4.64183062e-01 -1.82508051e-01 -7.60225177e-01
-1.59787774e-01 -1.41434252e-01 3.61457020e-02 2.25913510e-01
5.61338246e-01 2.39069179e-01 2.88092554e-01 -7.66647279e-01
1.43200949e-01 1.50239170e-02 6.81176901e-01 -7.60155857e-01
5.57621419e-01 1.10079467e+00 1.35637984e-01 -7.72712946e-01
-8.26336086e-01 -2.45026529e-01 -7.59954393e-01 -4.52106148e-01
1.15428698e+00 -1.20969212e+00 -6.73378646e-01 3.45693082e-01
-9.68432605e-01 -6.79001033e-01 -5.65251470e-01 2.46642455e-01
-9.56522584e-01 2.22363740e-01 -7.30786026e-01 -8.76620471e-01
-4.39034373e-01 -1.14476764e+00 2.00059414e+00 -2.65671760e-01
-4.72088039e-01 -5.56823432e-01 2.01713607e-01 7.66333938e-01
6.24801107e-02 7.03076422e-01 7.00091422e-01 6.56222939e-01
-8.33946288e-01 3.69225830e-01 2.18720764e-01 -3.07804905e-02
-1.12939171e-01 -3.90639693e-01 -9.69227552e-01 -3.21893811e-01
2.25199331e-02 -2.72069722e-01 6.26216829e-01 6.59347832e-01
8.85090232e-01 -5.24989776e-02 -3.08114648e-01 8.15325379e-01
1.21560025e+00 -5.03126621e-01 7.76712000e-01 -1.63619462e-02
9.59134877e-01 7.28872478e-01 6.76647842e-01 7.02979982e-01
3.27282995e-01 8.87442172e-01 2.97173977e-01 -4.90597725e-01
-6.44723594e-01 -7.07912922e-01 1.38134032e-01 6.93237484e-01
-4.96292114e-01 2.56660059e-02 -8.80700827e-01 3.09952646e-01
-1.57548964e+00 -7.42934406e-01 -1.99798629e-01 2.50283265e+00
6.54857159e-01 4.01417650e-02 1.38021380e-01 -1.44217893e-01
4.43818569e-01 8.96769315e-02 -5.27307630e-01 1.36767566e-01
7.91866630e-02 4.28558350e-01 3.05784315e-01 8.59205186e-01
-5.90039372e-01 7.46495783e-01 6.74536610e+00 4.65922266e-01
-9.06683981e-01 2.79452562e-01 7.16180325e-01 -3.99459571e-01
-7.96147108e-01 -1.28558308e-01 -4.38368917e-01 1.84483558e-01
5.22980094e-01 5.42833447e-01 3.45889926e-01 5.31128526e-01
1.49580494e-01 -3.87953281e-01 -1.01553333e+00 1.22510064e+00
1.02414578e-01 -1.40664065e+00 1.17960624e-01 4.78334188e-01
9.54928577e-01 -3.62141207e-02 -5.86636215e-02 -2.37928629e-01
2.08511010e-01 -1.21879160e+00 1.05971074e+00 9.21449840e-01
1.25869930e+00 -5.94966710e-01 2.34976217e-01 5.62315762e-01
-1.28446853e+00 6.91645026e-01 -8.08664188e-02 -1.46958575e-01
7.22420871e-01 6.10784352e-01 -5.10541558e-01 3.57923716e-01
9.21026528e-01 3.80158514e-01 -8.20927024e-02 3.05365354e-01
6.33286387e-02 9.52564552e-02 -5.14504135e-01 7.86090791e-01
-3.08298737e-01 -3.50374579e-01 5.97995043e-01 6.44404173e-01
2.91094571e-01 5.34877002e-01 4.37229604e-01 1.12967157e+00
2.92869657e-01 -2.74160922e-01 -5.96419930e-01 6.25375330e-01
3.11871469e-02 8.52593899e-01 -7.40662098e-01 -4.44815010e-01
-1.09115399e-01 1.45787323e+00 1.72382936e-01 1.21001139e-01
-6.76044524e-01 6.87528312e-01 5.22458076e-01 1.19208515e+00
1.88246205e-01 -1.94872469e-01 -5.30999303e-01 -1.11284542e+00
5.72803766e-02 -7.34014928e-01 8.64602849e-02 -9.16979373e-01
-1.08392406e+00 4.09575760e-01 5.41796163e-02 -1.08138072e+00
-8.16399679e-02 -2.45817691e-01 1.05687737e-01 1.11838055e+00
-9.36238229e-01 -1.07657373e+00 -4.80594069e-01 7.14880168e-01
6.45405293e-01 4.45400327e-01 8.53883505e-01 1.72497362e-01
2.00480759e-01 3.67859341e-02 -2.57787347e-01 -3.12707365e-01
5.75688064e-01 -1.08218765e+00 7.59524524e-01 4.27984267e-01
-1.48850217e-01 3.67171705e-01 6.87844634e-01 -1.23913682e+00
-1.62958038e+00 -8.98595750e-01 4.08313632e-01 -7.30351686e-01
-3.77892852e-01 -4.26976562e-01 -6.99651122e-01 6.50597155e-01
-1.52261183e-01 3.38708013e-01 2.45679349e-01 -2.25676168e-02
-1.75834253e-01 3.66527677e-01 -1.38195109e+00 6.33172452e-01
1.41421688e+00 -5.20236731e-01 -4.15539205e-01 -2.10592989e-02
3.63273919e-01 -1.01647449e+00 -1.13711548e+00 6.03949606e-01
1.15577316e+00 -1.42553163e+00 1.24936259e+00 1.98965166e-02
5.60896695e-01 -4.90389615e-01 -2.76238441e-01 -9.90889430e-01
-4.79159653e-01 -6.70142949e-01 -1.53254136e-01 3.35553676e-01
1.33492410e-01 -3.16195458e-01 1.31267560e+00 8.45927238e-01
-2.72948891e-01 -8.62066209e-01 -1.01062965e+00 -3.88289362e-01
-1.99154884e-01 -2.17573896e-01 2.30351746e-01 5.59270322e-01
-2.96312124e-01 2.58109957e-01 -8.31531942e-01 1.15453899e-01
1.13527584e+00 2.72035956e-01 9.92238939e-01 -1.22410548e+00
-4.72269088e-01 4.12091196e-01 -4.21048813e-02 -1.40076149e+00
-3.72249424e-01 -5.79948604e-01 3.17791134e-01 -1.50829446e+00
3.36156309e-01 -4.91466194e-01 3.75791937e-01 8.98953974e-02
2.36374512e-02 9.36553180e-01 7.11199194e-02 3.40875149e-01
-3.06687206e-01 6.79267347e-01 1.85255659e+00 2.65164196e-01
-1.24178186e-01 -4.11900550e-01 -2.66104251e-01 9.68281984e-01
1.76201269e-01 -4.50294822e-01 -3.19602221e-01 -5.42918563e-01
1.00754343e-01 6.84357047e-01 7.20769584e-01 -1.09943175e+00
-1.40615895e-01 -7.54294172e-02 8.71952951e-01 -9.36920762e-01
1.09490561e+00 -7.72550285e-01 1.03526354e+00 5.03447175e-01
-1.27350464e-01 7.57909641e-02 -2.22720489e-01 7.56548464e-01
2.73157030e-01 6.31306589e-01 1.07362473e+00 -6.35751307e-01
-1.19566895e-01 6.14099026e-01 -1.32049084e-01 1.82352647e-01
5.32434225e-01 -5.87192178e-01 2.43641198e-01 -3.12100977e-01
-1.11970580e+00 -1.00967124e-01 1.41620505e+00 1.50836080e-01
9.67148602e-01 -1.20237958e+00 -8.52531075e-01 7.35558212e-01
-1.96955532e-01 4.95060384e-01 6.17989898e-01 7.09627569e-01
-8.09180558e-01 3.46937291e-02 -2.25467160e-01 -1.11076295e+00
-1.44672990e+00 1.34729311e-01 4.32222724e-01 -3.13533306e-01
-1.59727895e+00 8.14781129e-01 9.07413304e-01 -7.38261998e-01
-5.79571575e-02 -1.67212009e-01 5.95526874e-01 -7.00135469e-01
4.06765252e-01 2.71282434e-01 3.85239944e-02 -8.50950181e-01
-2.18935147e-01 9.85180855e-01 3.84503216e-01 -7.62112141e-01
1.23032188e+00 -3.58321756e-01 2.20293596e-01 4.21946675e-01
9.18300390e-01 1.68986693e-01 -1.85374486e+00 7.61822145e-03
-7.56241143e-01 -8.65784824e-01 -5.65508101e-03 -7.70528793e-01
-9.92174685e-01 7.91277587e-01 6.03397846e-01 -4.61204827e-01
7.64099121e-01 9.44385901e-02 1.04378378e+00 -3.33666950e-01
8.62273991e-01 -8.85722101e-01 4.29677367e-02 3.19446713e-01
1.11821079e+00 -9.17662501e-01 4.88837898e-01 -9.76091146e-01
-4.25606489e-01 7.91672945e-01 5.97027123e-01 -3.59365016e-01
6.93223000e-01 4.67479974e-01 -9.86742377e-02 -6.73329592e-01
-6.74989760e-01 1.57320827e-01 4.90633070e-01 6.98001683e-01
4.06068623e-01 1.06542096e-01 1.83467925e-01 -2.01301766e-03
-2.74656177e-01 -7.92892948e-02 2.07914129e-01 9.54230070e-01
-2.15672031e-01 -9.18511271e-01 -7.96112061e-01 3.85411322e-01
-3.19201171e-01 -2.26889029e-02 -4.68744338e-02 5.55844128e-01
1.41312599e-01 4.10473824e-01 2.51252502e-01 -2.32571483e-01
6.40392065e-01 -9.63297263e-02 1.03356755e+00 -1.00942338e+00
-4.90452141e-01 6.17001534e-01 2.91905195e-01 -1.10243547e+00
-3.70843649e-01 -6.74842536e-01 -1.20420432e+00 -2.51146227e-01
2.18297001e-02 -2.43763894e-01 3.03979188e-01 5.79761147e-01
4.61995453e-01 5.45947373e-01 2.36273468e-01 -1.61200619e+00
-4.51110862e-02 -4.41058397e-01 -9.29359496e-01 6.56026483e-01
4.19120669e-01 -8.34458292e-01 -2.11838290e-01 2.24312931e-01]
|
[7.21945858001709, -1.3019931316375732]
|
50b2ba86-6ed9-4a03-babb-b0a14bbaf811
|
rssod-bench-a-large-scale-benchmark-dataset
|
2306.02351
| null |
https://arxiv.org/abs/2306.02351v1
|
https://arxiv.org/pdf/2306.02351v1.pdf
|
RSSOD-Bench: A large-scale benchmark dataset for Salient Object Detection in Optical Remote Sensing Imagery
|
We present the RSSOD-Bench dataset for salient object detection (SOD) in optical remote sensing imagery. While SOD has achieved success in natural scene images with deep learning, research in SOD for remote sensing imagery (RSSOD) is still in its early stages. Existing RSSOD datasets have limitations in terms of scale, and scene categories, which make them misaligned with real-world applications. To address these shortcomings, we construct the RSSOD-Bench dataset, which contains images from four different cities in the USA. The dataset provides annotations for various salient object categories, such as buildings, lakes, rivers, highways, bridges, aircraft, ships, athletic fields, and more. The salient objects in RSSOD-Bench exhibit large-scale variations, cluttered backgrounds, and different seasons. Unlike existing datasets, RSSOD-Bench offers uniform distribution across scene categories. We benchmark 23 different state-of-the-art approaches from both the computer vision and remote sensing communities. Experimental results demonstrate that more research efforts are required for the RSSOD task.
|
['Xiao Xiang Zhu', 'Qi Wang', 'Yanfeng Liu', 'Zhitong Xiong']
|
2023-06-04
| null | null | null | null |
['salient-object-detection-1']
|
['computer-vision']
|
[ 2.55248755e-01 -2.02537075e-01 -5.31760566e-02 -1.75866261e-01
-4.52398032e-01 -2.51993507e-01 4.30752963e-01 1.29170820e-01
-9.98823643e-02 5.05599797e-01 3.56642723e-01 -1.67469174e-01
-8.72786045e-02 -8.59262884e-01 -2.05583200e-01 -7.42259920e-01
-2.35678419e-01 -1.23861626e-01 6.95611358e-01 -6.76252306e-01
1.56362921e-01 7.88528740e-01 -1.55972850e+00 1.22773387e-01
7.16203094e-01 7.01034129e-01 6.92748427e-01 4.66325074e-01
9.22028720e-02 8.41226339e-01 -2.86778718e-01 9.18193981e-02
6.19683504e-01 -4.97936532e-02 -7.96108365e-01 3.60341519e-01
1.01286411e+00 -4.77331936e-01 -4.00117636e-01 1.27264261e+00
6.89098239e-01 7.71235228e-02 5.12260914e-01 -1.16136229e+00
-1.07022214e+00 6.26705587e-01 -1.15438068e+00 8.03596556e-01
-2.63361722e-01 1.92603935e-02 1.09262645e+00 -1.16568434e+00
4.33992893e-01 1.28750587e+00 9.65754628e-01 7.44990036e-02
-8.16594183e-01 -5.42045116e-01 4.65104729e-01 2.74545606e-02
-1.59241951e+00 -1.24654762e-01 5.93526542e-01 -6.07778549e-01
5.80472350e-01 3.95499200e-01 5.89122295e-01 4.21645612e-01
-1.35333255e-01 1.03259659e+00 8.87945294e-01 -1.71580896e-01
2.55210906e-01 -8.79908502e-02 2.79970109e-01 4.35261250e-01
5.81154585e-01 -1.53048083e-01 -2.52773285e-01 -3.73811536e-02
1.02993858e+00 5.57296157e-01 -3.26998770e-01 -3.34347755e-01
-1.29994583e+00 8.34182501e-01 1.06868756e+00 3.73682268e-02
-4.64772880e-01 -1.92026898e-01 1.86414167e-01 -9.09832194e-02
5.65080106e-01 1.48863658e-01 -3.81312251e-01 6.64358974e-01
-1.01635325e+00 4.32200819e-01 1.10705100e-01 8.52670848e-01
7.77541637e-01 2.97700822e-01 3.13793309e-02 9.73456383e-01
3.23647052e-01 7.71681905e-01 5.01871184e-02 -4.91647303e-01
4.53968644e-01 7.59395182e-01 3.76476794e-01 -1.26688361e+00
-6.91017210e-01 -5.41232884e-01 -9.32129502e-01 2.89315224e-01
5.39704822e-02 -1.61607623e-01 -1.06320846e+00 1.22167504e+00
5.47515035e-01 3.03237200e-01 1.45072743e-01 1.35566139e+00
1.45217776e+00 6.58851504e-01 2.78289557e-01 3.77186745e-01
1.30727684e+00 -1.04171884e+00 -1.98727116e-01 -8.49434853e-01
8.61429200e-02 -6.27465785e-01 1.25451541e+00 -2.12344587e-01
-5.77343881e-01 -4.82933968e-01 -9.20054734e-01 -9.51765925e-02
-4.23130035e-01 3.89328420e-01 7.30596781e-01 1.21651702e-01
-9.25506353e-01 8.49595815e-02 -4.10185724e-01 -8.83096755e-01
6.41433120e-01 -4.24159527e-01 5.89709245e-02 2.28771362e-02
-9.96832311e-01 7.95383334e-01 3.33997935e-01 5.39328814e-01
-1.05134654e+00 -9.03900862e-01 -1.05693734e+00 6.37070015e-02
1.21334083e-01 -2.53963351e-01 1.05866003e+00 -8.83396924e-01
-5.30598819e-01 1.23352790e+00 3.61533642e-01 -2.52934575e-01
2.52558947e-01 -2.41286337e-01 -8.41248989e-01 1.32099763e-01
6.20782852e-01 7.43862092e-01 6.66549981e-01 -1.34284616e+00
-1.00727892e+00 -1.87059850e-01 3.71964216e-01 3.74978691e-01
1.11408338e-01 4.26032424e-01 -3.67618576e-02 -7.93447495e-01
4.65041161e-01 -8.98988545e-01 -5.11959612e-01 3.34549844e-01
-5.34560204e-01 1.14474393e-01 1.07791948e+00 -4.33647543e-01
8.91447425e-01 -2.29841447e+00 -4.92484510e-01 -1.47207364e-01
1.95028007e-01 3.17175597e-01 -2.43479654e-01 4.04104322e-01
-7.31590092e-02 1.50040641e-01 -4.05161232e-01 2.98232436e-01
-1.46786705e-01 1.73643872e-01 -6.40971839e-01 6.01142287e-01
3.53886187e-01 7.64842808e-01 -1.04507446e+00 -6.36160851e-01
3.19444597e-01 2.31177568e-01 5.90311131e-03 -1.93796679e-01
-1.44620001e-01 8.94494131e-02 -7.99972415e-01 1.21417391e+00
1.17767739e+00 -3.41459572e-01 -2.56881654e-01 9.02714953e-02
-4.77596402e-01 1.40824065e-01 -1.48872221e+00 1.17488360e+00
1.72714531e-01 8.86552870e-01 2.57980645e-01 -7.70683527e-01
9.87431467e-01 -2.39357471e-01 3.45609784e-01 -6.09124720e-01
-2.39314914e-01 1.19454913e-01 -3.68240058e-01 -5.73638856e-01
1.11363745e+00 -5.31382822e-02 8.58240053e-02 1.02258272e-01
-7.43189454e-01 -2.09949240e-01 -9.98792052e-02 1.73247650e-01
4.08218175e-01 1.85643304e-02 4.33340490e-01 -6.90873682e-01
1.27833560e-01 6.81540251e-01 6.35183752e-01 9.35223579e-01
-6.27701104e-01 1.14121926e+00 -1.08137667e-01 -9.44166601e-01
-9.24222410e-01 -1.16336954e+00 -4.41511929e-01 1.43936503e+00
6.79262459e-01 3.82896990e-01 -1.98583230e-01 -3.97847295e-01
1.37622997e-01 1.91922754e-01 -6.11877024e-01 3.07277471e-01
-2.10221320e-01 -1.02600014e+00 4.11295444e-01 5.25033891e-01
1.02457118e+00 -1.28676283e+00 -8.99381280e-01 1.88703522e-01
-3.04969370e-01 -1.08897817e+00 -3.26200187e-01 -5.27035035e-02
-7.92980134e-01 -1.05478537e+00 -7.95011580e-01 -1.05788875e+00
5.42847276e-01 1.63402796e+00 1.54639781e+00 8.94819275e-02
-5.32791555e-01 3.47011127e-02 -3.44996333e-01 -8.99343491e-01
1.94127515e-01 -1.35422722e-01 -3.15702140e-01 -1.51905909e-01
3.13868642e-01 -1.93111062e-01 -9.45828021e-01 6.48744404e-01
-1.05388856e+00 -4.57982756e-02 6.57798350e-01 5.43492794e-01
5.46253145e-01 6.17876984e-02 5.61204076e-01 -6.57827735e-01
-1.76619336e-01 -8.95468593e-01 -4.29915130e-01 2.19929069e-01
-6.33434132e-02 -6.96660101e-01 -7.33314604e-02 -5.73624223e-02
-1.12831926e+00 1.56397745e-01 2.64610440e-01 1.36054292e-01
-4.83061701e-01 7.31073916e-01 9.31705013e-02 7.67372847e-02
1.12189257e+00 2.78484762e-01 -4.06436741e-01 -5.45105278e-01
1.11824632e-01 1.05352128e+00 7.56485045e-01 -1.98644906e-01
9.65229809e-01 9.08680856e-01 -3.37877423e-01 -1.39560127e+00
-1.47105491e+00 -9.39986706e-01 -4.69213963e-01 5.37979268e-02
6.49670899e-01 -1.72949803e+00 3.09194326e-01 6.17549121e-01
-6.00782931e-01 -4.55115378e-01 -3.30271155e-01 3.85436654e-01
8.60738233e-02 3.11924875e-01 -2.86543965e-01 -7.66343653e-01
-6.20263100e-01 -5.51317632e-01 1.65638995e+00 6.72136366e-01
2.06664115e-01 -7.50207901e-01 -2.39255026e-01 2.76492566e-01
5.12485445e-01 5.69867849e-01 3.91299069e-01 1.87110119e-02
-7.24753737e-01 1.89428419e-01 -5.81880808e-01 -4.51253448e-03
2.31575713e-01 2.82472640e-01 -8.94671798e-01 -3.51192474e-01
-4.60344404e-01 -2.41355345e-01 1.17241275e+00 6.81144953e-01
8.57811332e-01 -4.66921739e-02 -2.99002677e-01 5.36816180e-01
1.56077111e+00 -6.02385253e-02 7.42412269e-01 9.13426757e-01
7.10945189e-01 5.09677052e-01 1.12440920e+00 5.56277573e-01
5.89494288e-01 3.97360891e-01 1.06482148e+00 -9.56712782e-01
-2.43465066e-01 1.39346004e-01 1.89648375e-01 -9.66616124e-02
-1.14185192e-01 9.48171318e-02 -9.90219235e-01 1.25920808e+00
-1.78101969e+00 -1.28188515e+00 -6.98072076e-01 1.74606526e+00
4.23205942e-01 -3.07957023e-01 1.78214639e-01 -1.97027415e-01
8.63445103e-01 5.88769436e-01 -6.75527990e-01 2.29923576e-01
-6.36156857e-01 -3.80314261e-01 7.11765051e-01 8.53531063e-02
-1.74736953e+00 9.95374441e-01 6.62985373e+00 3.71968746e-01
-1.44490921e+00 -1.47191614e-01 4.08719540e-01 1.56744838e-01
-1.27405107e-01 -1.42375872e-01 -1.02693784e+00 1.40425071e-01
-1.03617743e-01 2.17281953e-01 -1.29282743e-01 1.22395790e+00
4.93501246e-01 -2.40332112e-01 -1.32086501e-01 8.78544748e-01
-1.05051465e-01 -1.38034689e+00 -6.56347275e-02 -2.92222381e-01
1.30215693e+00 8.31842422e-01 1.26091436e-01 1.02534562e-01
5.09376764e-01 -7.72452652e-01 1.01641738e+00 1.41389012e-01
3.79241556e-01 -3.24293643e-01 6.52808368e-01 2.07512632e-01
-1.42001975e+00 -1.42843381e-01 -1.03962934e+00 -2.47478589e-01
-1.82966799e-01 8.00801694e-01 -7.85231411e-01 4.73858774e-01
1.40255666e+00 1.20444536e+00 -8.42597723e-01 1.33112109e+00
-1.91543475e-01 6.45242333e-01 -2.76001334e-01 1.54817343e-01
7.45627701e-01 -2.93680251e-01 5.91798604e-01 1.24627531e+00
1.93830788e-01 1.22339455e-02 5.88424027e-01 7.71210194e-01
2.36430407e-01 -1.42196298e-01 -7.38688588e-01 1.58252895e-01
5.16699851e-01 1.61475825e+00 -7.71929324e-01 -3.29540730e-01
-3.86320233e-01 4.44495738e-01 -1.27435133e-01 4.84930366e-01
-9.43832040e-01 -1.54491767e-01 1.12809527e+00 3.10735196e-01
6.03868425e-01 -3.37747097e-01 -3.18971723e-01 -1.06360638e+00
-5.58976047e-02 -6.84307337e-01 6.74717546e-01 -1.22141695e+00
-1.12112010e+00 1.96450531e-01 -1.16800126e-02 -1.66708028e+00
6.75214052e-01 -2.43732423e-01 -8.37480068e-01 6.26279116e-01
-2.35885501e+00 -1.56307328e+00 -1.04236400e+00 6.10077202e-01
7.28756368e-01 -1.59915298e-01 4.16287601e-01 1.65814444e-01
-6.31682158e-01 -2.79205292e-01 3.76710862e-01 3.42466265e-01
3.78872603e-01 -1.20106733e+00 7.89694369e-01 1.37310529e+00
1.14733383e-01 3.09277743e-01 5.51114261e-01 -3.83194387e-01
-1.03829849e+00 -1.74390817e+00 6.45798147e-01 -2.24142745e-02
6.72214627e-01 1.17535405e-01 -7.79029787e-01 7.89639711e-01
-2.19075358e-03 3.89092922e-01 3.55606973e-01 1.03510208e-02
-2.14347139e-01 -2.28708193e-01 -1.04860878e+00 8.42092454e-01
1.03690410e+00 -3.60615492e-01 -7.62757301e-01 5.90915084e-01
5.31790912e-01 -4.07692254e-01 -4.55259740e-01 2.85430044e-01
1.26692891e-01 -9.55553234e-01 1.38290095e+00 -5.27531147e-01
5.91300130e-01 -8.14872265e-01 -2.49602780e-01 -1.21496642e+00
-7.74798095e-01 -1.43526793e-01 5.17871618e-01 1.11790836e+00
1.45807117e-01 -3.94590080e-01 5.68987429e-01 1.17642678e-01
-6.16499841e-01 -1.22549936e-01 -5.08983850e-01 -6.74647510e-01
-1.86629966e-01 -1.38407916e-01 8.13643277e-01 1.22422302e+00
-5.69345474e-01 2.66792864e-01 -4.76738572e-01 8.27153862e-01
6.92979097e-01 9.39150870e-01 8.63573372e-01 -1.49663818e+00
2.48976454e-01 -3.91685635e-01 -3.84125441e-01 -9.61923242e-01
-4.46793824e-01 -5.65566897e-01 2.08070844e-01 -2.00761771e+00
3.94476712e-01 -7.05449641e-01 -2.46184841e-01 9.68082249e-01
-6.47112668e-01 5.68069279e-01 7.32968152e-02 4.62004036e-01
-6.45553827e-01 4.59344029e-01 1.21909034e+00 -5.69441497e-01
-1.71107709e-01 -1.72711328e-01 -1.17892444e+00 9.94891167e-01
1.04381549e+00 -3.51834089e-01 -1.85652405e-01 -8.31027448e-01
1.30621701e-01 -5.82082570e-01 7.62370884e-01 -1.09326518e+00
-9.12798643e-02 -7.26288557e-01 3.83359611e-01 -1.05874908e+00
-1.14102624e-01 -7.43394911e-01 1.45466179e-01 4.90858644e-01
5.40132150e-02 -3.47240448e-01 2.74840027e-01 4.97816771e-01
-3.13608199e-01 -5.73195294e-02 1.15239215e+00 -2.65674889e-01
-1.61669695e+00 4.37828749e-01 -3.50864887e-01 1.48872450e-01
9.24878895e-01 -4.72526550e-01 -8.91220748e-01 -7.51714632e-02
-3.78809422e-01 3.01054746e-01 5.21648288e-01 5.13964593e-01
6.20996714e-01 -1.08863413e+00 -1.23718286e+00 8.24113265e-02
6.54666841e-01 5.41258693e-01 4.76782054e-01 7.31948018e-01
-9.00817752e-01 -4.48160022e-02 -3.74956429e-01 -7.31532097e-01
-1.15626073e+00 2.81008393e-01 4.05612469e-01 2.12134302e-01
-1.00486350e+00 6.46971643e-01 3.41516525e-01 -5.40037334e-01
-1.90233976e-01 -5.39405227e-01 -4.22562629e-01 1.68614641e-01
7.22324312e-01 3.63697171e-01 1.01726577e-01 -7.96346366e-01
-5.59179366e-01 6.41651571e-01 2.69443929e-01 6.07932568e-01
1.65146017e+00 -4.52849090e-01 -2.43047103e-02 7.71829486e-02
4.55874413e-01 -7.64429122e-02 -1.41702724e+00 -6.20250940e-01
-1.14440210e-01 -6.81369245e-01 3.09381753e-01 -5.94772458e-01
-1.28267705e+00 6.46440804e-01 7.39077866e-01 2.64131397e-01
1.09081948e+00 1.20333053e-01 6.18107677e-01 5.23872614e-01
3.70963126e-01 -1.10482430e+00 -1.24866165e-01 8.51883650e-01
1.01759017e+00 -1.67891574e+00 4.28123295e-01 -5.05554855e-01
-1.01512384e+00 8.32371294e-01 2.83204168e-01 -5.32449372e-02
5.80956221e-01 -1.25560939e-01 5.04538238e-01 -5.33498943e-01
-7.61217475e-02 -9.04659569e-01 2.55492091e-01 7.72086620e-01
1.01162530e-01 3.62857342e-01 1.18750177e-01 -5.70894731e-03
4.58764210e-02 -2.38021240e-01 8.10902119e-01 1.08168113e+00
-1.08903575e+00 -7.67690837e-02 -6.79386556e-01 3.89060557e-01
-3.12075168e-01 -4.54513222e-01 -4.21389431e-01 8.77293646e-01
2.65292935e-02 1.01288760e+00 1.16872594e-01 8.26979503e-02
2.32101083e-01 -6.42717183e-01 -3.25875312e-01 -7.10046172e-01
-4.52156216e-01 -9.09376442e-02 6.10756688e-02 -1.93762496e-01
-8.49279404e-01 -8.76187384e-01 -1.29100192e+00 -1.95717245e-01
-1.49620622e-01 -1.05921410e-01 4.25649941e-01 5.20266593e-01
3.57170612e-01 3.30691159e-01 7.15704799e-01 -1.05641222e+00
-3.14772129e-01 -9.34319019e-01 -1.29752517e+00 3.27975392e-01
5.36145985e-01 -8.35253716e-01 -2.33308733e-01 7.89082423e-02]
|
[9.119061470031738, -0.8726456761360168]
|
f94b1dad-6162-427b-a098-3c702d131c02
|
acceptance-of-covid-19-vaccine-and-its
|
2103.15206
| null |
https://arxiv.org/abs/2103.15206v2
|
https://arxiv.org/pdf/2103.15206v2.pdf
|
Knowledge, beliefs, attitudes and perceived risk about COVID-19 vaccine and determinants of COVID-19 vaccine acceptance in Bangladesh
|
A total of 605 eligible respondents took part in this survey (population size 1630046161 and required sample size 591) with an age range of 18 to 100. A large proportion of the respondents are aged less than 50 (82%) and male (62.15%). The majority of the respondents live in urban areas (60.83%). A total of 61.16% (370/605) of the respondents were willing to accept/take the COVID-19 vaccine. Among the accepted group, only 35.14% showed the willingness to take the COVID-19 vaccine immediately, while 64.86% would delay the vaccination until they are confirmed about the vaccine s efficacy and safety or COVID-19 becomes deadlier in Bangladesh. The regression results showed age, gender, location (urban/rural), level of education, income, perceived risk of being infected with COVID-19 in the future, perceived severity of infection, having previous vaccination experience after age 18, having higher knowledge about COVID-19 and vaccination were significantly associated with the acceptance of COVID-19 vaccines. The research reported a high prevalence of COVID-19 vaccine refusal and hesitancy in Bangladesh.
|
['Miah Akib Zaman', 'Ashraf Uddin Mian', 'Ijaz Ahmed Khan', 'Md. Mohsin', 'Sultan Mahmud']
|
2021-03-28
| null | null | null | null |
['misconceptions']
|
['miscellaneous']
|
[-2.35581622e-01 5.60753932e-03 -5.09432495e-01 -4.22685981e-01
-1.09118335e-02 -6.86739504e-01 -4.22327258e-02 6.45408392e-01
-6.66134238e-01 7.11858332e-01 3.45172852e-01 -8.76569390e-01
1.32323101e-01 -9.80394006e-01 -7.15737581e-01 -6.26387477e-01
1.27197415e-01 1.54297292e-01 -1.34726912e-01 -3.16383421e-01
2.25419268e-01 8.41827244e-02 -1.04017925e+00 -4.75316346e-01
1.37597680e+00 2.01824829e-01 3.94768417e-01 3.74956101e-01
4.48911905e-01 1.54496990e-02 -6.29110157e-01 -2.49915689e-01
6.51500607e-03 -2.21165806e-01 -2.36283794e-01 -5.60157537e-01
1.53934389e-01 -9.77891386e-01 3.46837372e-01 6.57665312e-01
4.54215109e-01 -2.41955668e-01 9.32897806e-01 -1.25604832e+00
-6.87116206e-01 3.05954605e-01 -8.27901781e-01 2.06714403e-02
4.21983987e-01 6.24408983e-02 -2.72721555e-02 -8.91141653e-01
7.27140069e-01 1.17530775e+00 1.01159453e+00 4.71628010e-01
-1.08784759e+00 -1.35717583e+00 -3.35448921e-01 -4.90909815e-01
-1.35498154e+00 -1.17400534e-01 -1.09694628e-02 -7.99976587e-01
8.79805565e-01 3.25722583e-02 8.76517534e-01 3.26495945e-01
9.75281239e-01 -2.98024684e-01 1.71959412e+00 -1.01509411e-02
1.38025254e-01 4.57202017e-01 3.77818286e-01 -5.98842539e-02
1.24439228e+00 1.71746343e-01 2.16303051e-01 -4.99803752e-01
6.80557847e-01 8.09339225e-01 -2.98315827e-02 5.87493414e-03
-9.47023451e-01 1.17045486e+00 2.36272871e-01 6.61591738e-02
-1.07919407e+00 -5.86735547e-01 2.91267157e-01 6.26459181e-01
1.67065993e-01 -2.88976640e-01 -6.79002881e-01 1.95196211e-01
-3.04206312e-01 4.63063478e-01 4.78422135e-01 3.63401204e-01
6.70234919e-01 2.48386800e-01 2.09743902e-02 8.76044780e-02
5.22756398e-01 1.59949768e+00 -3.68604690e-01 -5.03232837e-01
1.37838125e-01 3.19562227e-01 5.62517583e-01 -1.13497925e+00
-3.34181428e-01 -3.81021887e-01 -7.92357206e-01 1.11880109e-01
4.55118924e-01 -8.78616273e-01 -9.77281988e-01 1.88686597e+00
1.94851562e-01 -3.58921647e-01 1.25020981e-01 1.05186963e+00
3.03175598e-01 1.10804045e+00 6.04623914e-01 -6.86137855e-01
1.75113010e+00 -4.78855930e-02 -8.68672669e-01 -4.62818742e-02
-2.89918538e-02 -1.01359761e+00 3.70033085e-01 5.24554811e-02
-1.00125396e+00 -6.60422206e-01 -8.42863023e-01 1.10358894e+00
-3.99170041e-01 -3.20329219e-01 4.77793753e-01 1.31843102e+00
-1.16645038e+00 -3.72408777e-01 -5.81085920e-01 -1.11768234e+00
-1.89012378e-01 6.49423420e-01 -3.68229032e-01 -4.37420905e-01
-1.06083953e+00 1.03029335e+00 1.22198753e-01 -1.21773489e-01
-6.16034925e-01 -8.26362789e-01 -7.30553269e-01 -4.34721380e-01
-1.09311104e-01 -6.85000002e-01 5.99539399e-01 -6.71214163e-01
-4.52336431e-01 5.53351879e-01 -4.93168652e-01 -1.63222328e-01
-1.71123277e-02 -2.49894321e-01 -6.70754075e-01 1.01997711e-01
3.67187262e-01 8.06329846e-01 5.28799713e-01 -6.50209486e-01
-6.42488360e-01 -8.58978808e-01 2.97180172e-02 -3.90977692e-03
2.87170172e-01 6.83976293e-01 8.18474054e-01 -5.19253790e-01
-1.61868110e-01 -8.77224028e-01 -2.38538578e-01 -6.33171916e-01
3.03454876e-01 -4.58221674e-01 9.79071975e-01 -1.03655958e+00
8.45182896e-01 -1.68271363e+00 -9.85157490e-01 1.18895561e-01
7.07159042e-02 5.58574796e-01 2.05468293e-02 1.07002270e+00
1.67537063e-01 1.94493949e-01 3.30974162e-01 1.06270385e+00
-1.66905355e-02 1.30489215e-01 -1.52124077e-01 5.96175432e-01
2.74615020e-01 3.93454045e-01 -8.26317132e-01 -1.30767435e-01
2.69340694e-01 7.95147955e-01 -3.83955449e-01 4.89099115e-01
4.21460301e-01 2.53989279e-01 -5.67785084e-01 5.35674691e-01
1.94141078e+00 2.32720733e-01 -1.28240973e-01 2.06855685e-01
-5.73187530e-01 4.47794944e-02 -9.08997059e-01 2.73021668e-01
4.13246863e-02 1.66001126e-01 5.01474857e-01 -6.41041517e-01
9.30996180e-01 7.79803097e-01 1.71345726e-01 -5.13494253e-01
1.36544213e-01 2.68036485e-01 3.86641026e-01 -4.76907700e-01
3.66247743e-01 -2.22550586e-01 6.41495213e-02 6.82892144e-01
-5.67077100e-01 2.67749876e-01 1.50356412e-01 1.58727258e-01
3.84656638e-01 -3.05256873e-01 5.10017931e-01 -6.03148401e-01
2.94000626e-01 4.96403463e-02 7.22389221e-01 7.69195735e-01
-4.42620844e-01 -2.42290124e-01 -1.59337208e-01 5.62579324e-03
-7.42179036e-01 -1.29517162e+00 -4.50057954e-01 9.81154919e-01
-1.64535344e-02 4.48624492e-01 -5.28151155e-01 -1.02296844e-01
9.43284258e-02 4.46825922e-01 -2.20155105e-01 4.18961287e-01
-2.86357701e-01 -3.63920063e-01 -1.17228217e-01 4.32588041e-01
6.96538746e-01 -5.56218505e-01 -6.85859203e-01 1.52306482e-01
4.27328199e-02 -4.17297006e-01 -3.54402214e-01 -3.00958186e-01
-1.06473494e+00 -7.00169742e-01 -1.02413404e+00 -1.00470471e+00
1.20939279e+00 7.75024772e-01 3.89994234e-01 4.62776929e-01
2.37050250e-01 8.67739171e-02 -9.07020345e-02 -8.92089307e-01
-3.32222015e-01 -5.16218543e-01 3.48490447e-01 -7.45375574e-01
8.31372797e-01 2.23702136e-02 -1.06046116e+00 2.79849142e-01
-6.91059530e-01 -2.13906825e-01 5.09117603e-01 3.25900048e-01
-4.96285111e-02 -3.06678824e-02 1.29338491e+00 -5.60093462e-01
4.76169705e-01 -8.44946504e-01 -3.61390471e-01 3.98286767e-02
-8.33241165e-01 -1.02790368e+00 2.66372085e-01 -3.09525371e-01
-1.00638580e+00 -6.09764159e-01 -4.09509152e-01 9.13150966e-01
-5.33152461e-01 7.44932353e-01 1.51128754e-01 2.26043388e-01
9.07354872e-04 -9.12198871e-02 6.35985017e-01 -1.70496166e-01
-4.69991118e-01 9.54786718e-01 1.38657257e-01 -4.89032604e-02
6.45370007e-01 -1.12416036e-01 -4.98127043e-01 -7.38848209e-01
-3.76613997e-02 -4.39072013e-01 7.53146484e-02 -1.72336161e-01
1.20634556e+00 -1.30457926e+00 -1.22739327e+00 1.09556973e+00
-8.59735191e-01 3.34156603e-02 1.13936532e+00 1.45020092e+00
5.31327128e-01 -1.98885456e-01 -7.14620531e-01 -1.30176795e+00
-6.11976206e-01 -9.23044741e-01 3.33169997e-01 7.22460330e-01
-3.49035650e-01 -9.62981164e-01 -9.39015970e-02 3.93874943e-01
1.07072031e+00 6.29185617e-01 9.99689996e-01 -5.48928678e-01
-2.38639742e-01 -2.61583984e-01 -4.78724152e-01 -1.85126126e-01
5.65673947e-01 -1.43187597e-01 -1.92920625e-01 -9.10089433e-01
-1.27486195e-02 -4.38288867e-01 2.37793326e-01 9.28083956e-01
-1.85774907e-01 -5.23248851e-01 -3.13056290e-01 -3.47109705e-01
1.86775720e+00 9.05101061e-01 4.06036317e-01 -3.18062484e-01
-5.97972721e-02 6.81727827e-01 9.39583540e-01 5.10332584e-01
7.74657130e-01 1.40809104e-01 3.80802602e-01 -2.60720700e-01
4.85261112e-01 -4.33937997e-01 7.50347793e-01 7.84836531e-01
-3.48998427e-01 1.21625938e-01 -9.39945698e-01 6.85983479e-01
-9.46096599e-01 -7.94479728e-01 -7.09467947e-01 2.17839551e+00
7.20493197e-01 -2.08031148e-01 6.02415144e-01 -1.77236915e-01
9.44279432e-01 -4.53364998e-01 -3.36427420e-01 -8.27247262e-01
1.28607765e-01 1.82902619e-01 6.53644621e-01 7.28322268e-01
-7.70439386e-01 2.48835802e-01 6.45065641e+00 2.63436854e-01
-1.16088676e+00 -1.35652617e-01 8.89251113e-01 8.90302539e-01
-5.45451701e-01 1.74954221e-01 -9.39126253e-01 3.30518872e-01
1.19163585e+00 -2.89564550e-01 -1.13689996e-01 1.95560157e-01
5.59334099e-01 -7.05224395e-01 -1.25805512e-01 1.16166323e-01
-1.08181618e-01 -1.00424612e+00 -3.30057770e-01 1.69186756e-01
9.48031485e-01 -3.54112983e-01 9.40370858e-02 3.45163018e-01
6.55250624e-02 -1.03496230e+00 2.03758866e-01 1.28178582e-01
6.03298903e-01 -9.21889305e-01 1.29216611e+00 6.10749602e-01
-9.99557436e-01 3.67316425e-01 -4.78645474e-01 -9.25350845e-01
3.81052732e-01 2.66889364e-01 -9.75283384e-01 2.64313012e-01
7.35369802e-01 -5.90207577e-02 -2.71091819e-01 5.24414122e-01
2.14249477e-01 1.09279501e+00 -5.32788396e-01 -4.00031239e-01
1.06374696e-01 -5.66806614e-01 -4.59729694e-02 8.06621730e-01
5.16989589e-01 3.48370016e-01 -1.03361964e-01 2.90483266e-01
8.37624490e-01 1.20788209e-01 -6.74401522e-01 -2.12141529e-01
8.13588202e-01 3.76870453e-01 -6.50874794e-01 -2.57662177e-01
-5.96408069e-01 -1.56354997e-02 -4.97854143e-01 7.72276044e-01
-6.05324984e-01 -4.66996461e-01 5.86811006e-01 3.04748058e-01
4.55929637e-01 -5.80617562e-02 -3.44057679e-01 -5.87457269e-02
-2.25377306e-01 -9.23011839e-01 3.24192166e-01 -5.40022016e-01
-9.47269917e-01 1.43698096e-01 3.02954048e-01 -4.17551368e-01
-2.15130150e-01 -3.95297348e-01 -6.87797427e-01 1.44053686e+00
-9.32474434e-01 -9.58904564e-01 -1.03054039e-01 1.50569156e-01
1.57611623e-01 8.64600837e-02 9.29421663e-01 3.07159089e-02
-1.38209149e-01 3.08752865e-01 -1.76277116e-01 -2.86535949e-01
5.16570270e-01 -3.87815416e-01 -1.65220171e-01 5.42611182e-01
-1.73202372e+00 1.63219213e+00 8.95851970e-01 -1.35102284e+00
-1.74208605e+00 -9.11339641e-01 1.76075959e+00 6.72720224e-02
2.30469316e-01 -2.76989341e-01 -5.31759381e-01 6.28588378e-01
9.05590296e-01 -9.13942754e-01 8.93616021e-01 -1.07320875e-01
6.13652505e-02 -4.87123169e-02 -1.72053516e+00 7.26496637e-01
3.94208223e-01 -2.43474752e-01 -6.50351882e-01 -1.48496449e-01
6.40573025e-01 2.01997906e-01 -1.23461652e+00 6.46860659e-01
8.91377509e-01 -9.03461576e-01 8.58987331e-01 -3.31241190e-01
-1.55128539e-01 -2.80119747e-01 5.83790801e-02 -6.71789348e-01
-2.80619353e-01 -5.30730367e-01 6.98341846e-01 1.23879504e+00
1.68503135e-01 -1.26940501e+00 3.87328088e-01 7.76652157e-01
3.27385247e-01 -6.45492554e-01 -5.42637050e-01 -3.97012413e-01
3.52010339e-01 2.82703638e-01 7.92306840e-01 9.16349947e-01
-1.38619095e-01 -1.04556210e-01 -2.87085712e-01 2.10013375e-01
2.74056762e-01 5.29925302e-02 7.08272040e-01 -7.25408137e-01
1.81666777e-01 1.61664411e-01 -3.52648646e-02 -7.68300951e-01
-5.63060939e-01 -2.06030548e-01 -6.28869057e-01 -2.01444173e+00
6.37609243e-01 -4.06763107e-01 2.25180052e-02 4.75701481e-01
-6.37122929e-01 -2.02634156e-01 4.03688513e-02 -2.28921950e-01
3.22187215e-01 -6.05210513e-02 1.14588714e+00 1.05084263e-01
-2.00028911e-01 4.78848070e-01 -6.49289250e-01 1.73046276e-01
1.19567001e+00 -1.81428239e-01 -6.41889930e-01 -7.86407143e-02
3.48745644e-01 6.83600903e-01 2.26288080e-01 -1.46664381e-01
-3.26674581e-01 -8.80816400e-01 5.95576227e-01 -1.40564740e+00
-1.38214797e-01 -9.53724802e-01 8.20846438e-01 1.48273504e+00
3.48189831e-01 8.88621390e-01 1.06327467e-01 -7.91030452e-02
1.23500459e-01 -2.36142904e-01 1.76754460e-01 5.76862752e-01
-8.90799314e-02 2.20041685e-02 -1.11408854e+00 -1.47283766e-02
1.06035221e+00 -4.50740308e-01 -5.18853188e-01 -4.59832817e-01
-7.12829903e-02 2.36336455e-01 7.27620304e-01 3.51035208e-01
8.27265859e-01 -1.24188113e+00 -9.76314247e-01 4.21479911e-01
-1.57567888e-01 -7.22589552e-01 4.25262749e-01 1.47020757e+00
-7.74685800e-01 8.83281946e-01 -7.30206907e-01 -2.04215795e-01
-1.04074967e+00 4.87451255e-01 -4.49056357e-01 4.96823192e-01
-6.53781444e-02 3.94701034e-01 3.19137990e-01 -2.76511401e-01
1.15213037e-01 -4.95009646e-02 -2.16983542e-01 -2.77864188e-01
4.83170778e-01 7.36992359e-01 -4.44866478e-01 -7.30762661e-01
-8.97655547e-01 6.41076267e-01 -1.61281273e-01 4.58788574e-02
1.17935956e+00 -1.54485792e-01 -3.80564123e-01 1.88778356e-01
1.02199852e+00 4.47121441e-01 -8.81722927e-01 3.41447800e-01
-5.98932445e-01 -7.64128447e-01 -7.61132479e-01 -8.07750702e-01
-4.94095385e-01 2.96612084e-01 9.50033784e-01 -4.79302779e-02
1.03685868e+00 -1.52840480e-01 9.67462957e-01 -2.15174705e-01
6.96992934e-01 -7.22057819e-01 -7.01179385e-01 4.46245551e-01
6.04914963e-01 -1.30636787e+00 -1.49080187e-01 -2.71029443e-01
-3.61432403e-01 5.65043271e-01 5.74443996e-01 -1.99757174e-01
9.51565504e-01 3.10699530e-02 4.15385038e-01 -1.90161899e-01
-5.49631953e-01 3.74506652e-01 -1.01033725e-01 1.45695913e+00
9.45566297e-01 6.06587946e-01 -1.44438612e+00 3.66863668e-01
-1.75396547e-01 2.77783349e-02 6.93938375e-01 9.02460873e-01
-8.25048506e-01 -8.86577666e-01 -7.41580427e-01 6.34912670e-01
-6.09741867e-01 2.14673817e-01 8.56844184e-04 8.66348207e-01
4.82973546e-01 1.50237846e+00 3.53338629e-01 1.13472693e-01
-1.22203164e-01 -1.64307311e-01 2.88871109e-01 -6.77788779e-02
-6.82584107e-01 1.89961597e-01 2.17126429e-01 4.12054211e-01
-6.58064246e-01 -9.49899435e-01 -1.30288291e+00 -1.16972244e+00
-1.64665312e-01 4.87822771e-01 8.08703780e-01 6.85171008e-01
1.25112042e-01 -4.85442400e-01 6.80273652e-01 2.16775626e-01
-4.82359707e-01 -1.02594340e+00 -3.96199346e-01 -8.98056328e-02
4.80881840e-01 -4.22779858e-01 3.70559609e-03 -2.08509028e-01]
|
[5.723084926605225, 4.718062400817871]
|
2cf46bbb-9c87-4563-8dcf-76ac65900516
|
deep-integro-difference-equation-models-for
|
1910.13524
| null |
https://arxiv.org/abs/1910.13524v3
|
https://arxiv.org/pdf/1910.13524v3.pdf
|
Deep Integro-Difference Equation Models for Spatio-Temporal Forecasting
|
Integro-difference equation (IDE) models describe the conditional dependence between the spatial process at a future time point and the process at the present time point through an integral operator. Nonlinearity or temporal dependence in the dynamics is often captured by allowing the operator parameters to vary temporally, or by re-fitting a model with a temporally-invariant linear operator in a sliding window. Both procedures tend to be excellent for prediction purposes over small time horizons, but are generally time-consuming and, crucially, do not provide a global prior model for the temporally-varying dynamics that is realistic. Here, we tackle these two issues by using a deep convolution neural network (CNN) in a hierarchical statistical IDE framework, where the CNN is designed to extract process dynamics from the process' most recent behaviour. Once the CNN is fitted, probabilistic forecasting can be done extremely quickly online using an ensemble Kalman filter with no requirement for repeated parameter estimation. We conduct an experiment where we train the model using 13 years of daily sea-surface temperature data in the North Atlantic Ocean. Forecasts are seen to be accurate and calibrated. A key advantage of our approach is that the CNN provides a global prior model for the dynamics that is realistic, interpretable, and computationally efficient. We show the versatility of the approach by successfully producing 10-minute nowcasts of weather radar reflectivities in Sydney using the same model that was trained on daily sea-surface temperature data in the North Atlantic Ocean.
|
['Andrew Zammit-Mangion', 'Christopher K. Wikle']
|
2019-10-29
| null | null | null | null |
['spatio-temporal-forecasting']
|
['time-series']
|
[-1.09452799e-01 -4.01346087e-01 5.10337651e-01 -2.53565967e-01
-4.76264387e-01 -6.35746717e-01 8.60719979e-01 -6.35879859e-02
-3.65178257e-01 7.83463776e-01 1.72355935e-01 -6.46305323e-01
-4.25556451e-01 -7.99504101e-01 -5.53628385e-01 -8.84867013e-01
-5.30125260e-01 5.12217820e-01 -1.71770621e-02 -7.38462433e-02
1.73553377e-02 7.48361170e-01 -1.46727347e+00 -2.36839131e-01
7.63226211e-01 9.71105278e-01 5.66177350e-03 9.61265028e-01
1.34281069e-01 5.01770437e-01 -2.61484504e-01 1.49146810e-01
3.65080208e-01 -3.44742805e-01 -3.37967187e-01 -5.70505932e-02
1.64437354e-01 -2.98600435e-01 -1.58844858e-01 8.38087916e-01
1.66720122e-01 3.93473923e-01 7.90504754e-01 -7.09235847e-01
7.70162493e-02 1.09659143e-01 -2.11343884e-01 4.73409116e-01
-1.00045569e-01 2.93323040e-01 7.65067935e-01 -5.54782271e-01
3.03198844e-01 1.28145552e+00 1.23296690e+00 1.97963454e-02
-1.75181067e+00 -5.10565221e-01 5.97657897e-02 -4.51335222e-01
-1.21784866e+00 -3.48441392e-01 4.64035183e-01 -9.10417318e-01
9.18125927e-01 2.52590746e-01 8.79580140e-01 7.53728032e-01
9.95666921e-01 2.06829429e-01 1.25901985e+00 -8.14795792e-02
4.14163530e-01 -2.54300117e-01 -1.93247229e-01 3.64171028e-01
-1.91851497e-01 7.66107857e-01 -1.69903457e-01 -3.88112605e-01
9.30564582e-01 2.28018925e-01 -2.83325464e-01 1.27763941e-03
-1.00054038e+00 7.75166154e-01 1.43470898e-01 2.58808672e-01
-7.78610945e-01 4.04727817e-01 1.63736060e-01 4.49320436e-01
1.11880589e+00 3.25775862e-01 -8.99317682e-01 -2.73934722e-01
-1.54491520e+00 5.29642940e-01 1.18296540e+00 1.47411600e-01
7.46869862e-01 2.79801577e-01 4.75491397e-02 1.79698050e-01
4.99204338e-01 8.58265877e-01 1.13865763e-01 -1.01942945e+00
-8.74216110e-02 -1.36641145e-01 5.89871526e-01 -9.55332279e-01
-4.67535943e-01 -4.62541759e-01 -1.00535810e+00 7.43914127e-01
7.25003839e-01 -8.35745811e-01 -1.17797518e+00 1.59918785e+00
9.07470658e-02 5.90927005e-01 9.27876234e-02 6.03182077e-01
-7.40463436e-02 1.24366176e+00 6.20549656e-02 -3.83771598e-01
1.25583279e+00 -4.21104014e-01 -9.40719843e-01 -1.85399413e-01
3.25254083e-01 -7.26461828e-01 2.55508602e-01 2.26679221e-01
-9.61330235e-01 -5.24181068e-01 -7.71072984e-01 5.75552166e-01
-4.16508526e-01 1.02827987e-02 4.44413632e-01 3.15567881e-01
-1.30334210e+00 1.24177289e+00 -1.48096776e+00 -2.00504676e-01
-1.94032043e-01 1.74116522e-01 -1.36807993e-01 4.72287416e-01
-1.41217017e+00 1.17282999e+00 2.28201404e-01 6.45644784e-01
-1.06250751e+00 -9.17675257e-01 -7.63215303e-01 3.69905084e-01
-2.15890512e-01 -5.30983210e-01 1.41320324e+00 -1.17277503e+00
-1.73503375e+00 7.48551786e-02 -4.39316064e-01 -7.01386333e-01
6.30429924e-01 -1.56628281e-01 -5.63974380e-01 -2.52303869e-01
-1.63149506e-01 1.72426686e-01 9.99751270e-01 -9.42175031e-01
-7.07403183e-01 -7.10272864e-02 -3.43858272e-01 8.91467407e-02
5.00707448e-01 1.49781900e-02 1.34745747e-01 -6.07035220e-01
2.46406347e-01 -1.01432300e+00 -5.33100903e-01 3.62463519e-02
1.55823499e-01 7.68229812e-02 7.64965653e-01 -1.11476421e+00
1.10142815e+00 -2.09960389e+00 -9.85559151e-02 3.55762899e-01
-1.83792248e-01 6.35132417e-02 2.41151109e-01 6.80066228e-01
-1.84606895e-01 1.39915079e-01 -7.05895305e-01 -3.80787373e-01
-7.44463280e-02 4.66036797e-01 -8.09192896e-01 6.70545161e-01
4.11666095e-01 4.71897811e-01 -8.49251449e-01 1.47117540e-01
2.95947850e-01 6.80507898e-01 -1.36543617e-01 1.52078852e-01
-3.72006297e-01 8.31582785e-01 -3.01609654e-02 -2.66336668e-02
7.40293145e-01 1.21108368e-02 6.61377087e-02 2.69074947e-01
-6.28722906e-01 3.58181983e-01 -1.48265243e+00 1.02563167e+00
-7.42217600e-01 8.89237583e-01 4.31507111e-01 -7.15458095e-01
8.25712979e-01 7.18612015e-01 3.53285372e-01 -2.90428936e-01
-1.82763606e-01 2.89483070e-01 8.62021074e-02 -2.37801671e-01
3.32357734e-01 -8.47769320e-01 2.14552298e-01 5.79766989e-01
-1.75695315e-01 -4.22703266e-01 -1.22904725e-01 -3.33084852e-01
7.72789240e-01 4.89185065e-01 2.25382507e-01 -6.63316965e-01
2.38225177e-01 1.66399423e-02 7.38924026e-01 6.97124124e-01
3.60617116e-02 4.06289935e-01 4.74344224e-01 -9.11710978e-01
-1.09559369e+00 -1.04947388e+00 -4.93882388e-01 6.77394509e-01
-5.11068046e-01 -3.02041788e-02 -2.04973876e-01 -6.15843758e-02
2.37106998e-02 8.91616583e-01 -8.32120419e-01 2.16610894e-01
-5.45342863e-01 -9.06781554e-01 2.77490318e-01 5.37361264e-01
3.12321037e-01 -9.11798239e-01 -7.05305636e-01 7.59423018e-01
1.46748394e-01 -5.93648970e-01 -1.19999297e-01 4.77250189e-01
-1.27200019e+00 -5.06698251e-01 -5.42578995e-01 -2.67235428e-01
4.58208799e-01 -4.35523897e-01 1.09997213e+00 -4.58238572e-01
1.10866599e-01 2.02603176e-01 3.07335019e-01 -5.82247615e-01
-4.34294671e-01 -3.75731051e-01 1.37644619e-01 1.06067158e-01
8.27693567e-02 -7.61767268e-01 -4.65213209e-01 1.54593244e-01
-7.84082651e-01 -8.24514553e-02 1.06148027e-01 9.26935494e-01
2.44910926e-01 5.01950800e-01 2.01822251e-01 -4.60470527e-01
5.36529005e-01 -5.04219532e-01 -1.28719068e+00 -8.45087990e-02
-4.15287793e-01 3.12350661e-01 5.69284081e-01 -2.57473558e-01
-1.49928916e+00 2.37760812e-01 1.96576975e-02 -2.84224480e-01
-2.11843491e-01 1.01294053e+00 4.56973463e-01 3.43189240e-01
4.69606072e-01 2.08508343e-01 2.51118511e-01 -5.87734103e-01
1.85024053e-01 2.39668489e-01 5.20600736e-01 -5.49378157e-01
6.92466319e-01 8.81383002e-01 2.35105857e-01 -7.12421358e-01
-6.56520844e-01 -3.92397106e-01 -6.85941339e-01 -2.82451749e-01
8.39942575e-01 -1.09081817e+00 -5.62205970e-01 7.14427590e-01
-1.23354602e+00 -5.61264157e-01 -2.63764381e-01 8.39432538e-01
-4.29146439e-01 -1.04321927e-01 -6.45689964e-01 -1.28857744e+00
1.47916190e-03 -7.78083205e-01 8.61632884e-01 6.34712279e-02
-2.63000607e-01 -1.79552221e+00 5.33053339e-01 -6.87735260e-01
8.04205477e-01 5.26664793e-01 6.42981648e-01 -4.09231573e-01
-4.66640830e-01 -3.62797350e-01 4.35827449e-02 5.13107896e-01
1.62031963e-01 4.80626881e-01 -1.04039025e+00 -2.92054117e-01
4.38616812e-01 4.01781708e-01 9.78386998e-01 9.24373865e-01
5.63896835e-01 -3.15217733e-01 -3.33262712e-01 5.69144368e-01
1.43789375e+00 2.31111348e-01 3.81070346e-01 3.55132064e-03
3.22060108e-01 7.41202354e-01 2.51446605e-01 4.94091451e-01
3.13708335e-02 1.31489754e-01 2.83482343e-01 -8.52085650e-02
5.56935608e-01 6.96551725e-02 5.70883572e-01 7.53858924e-01
-4.03803498e-01 -2.07318012e-02 -1.31325936e+00 7.26393342e-01
-1.81047809e+00 -1.27422869e+00 -2.50427037e-01 2.28619266e+00
6.67265177e-01 4.25433777e-02 -3.84199589e-01 -2.35449672e-01
4.89199072e-01 3.33890796e-01 -1.80548415e-01 -6.07439041e-01
1.61839083e-01 3.33019376e-01 8.47961009e-01 1.02166665e+00
-1.42829084e+00 5.03054142e-01 7.03235435e+00 3.27673614e-01
-1.38988507e+00 -1.15304126e-03 5.22590160e-01 1.37861356e-01
-1.36408791e-01 4.01760250e-01 -6.76988125e-01 3.91632229e-01
1.68518114e+00 -7.63656124e-02 4.17668968e-01 4.27620679e-01
8.78140271e-01 -3.27885836e-01 -9.68296409e-01 3.53279054e-01
-6.11890912e-01 -1.30065596e+00 -4.55352902e-01 1.73081830e-01
8.80737066e-01 1.90725818e-01 -4.85930666e-02 1.97322533e-01
7.42076278e-01 -1.19333160e+00 5.86406052e-01 1.29395711e+00
3.58874947e-01 -7.58122504e-01 9.21285748e-01 5.79430282e-01
-1.23641717e+00 6.96460754e-02 -1.30403787e-01 -6.69442773e-01
5.21745265e-01 8.56745183e-01 -5.84949553e-01 3.67122799e-01
8.53810310e-01 7.71121621e-01 1.93027228e-01 1.04312992e+00
-3.90478410e-02 9.65990961e-01 -8.54741216e-01 3.41930866e-01
5.26324153e-01 -5.11758447e-01 8.21587026e-01 1.09901309e+00
7.32945144e-01 -2.68234555e-02 -2.36528702e-02 8.39190662e-01
6.29813850e-01 -4.69697505e-01 -6.72437489e-01 9.78575572e-02
9.86503735e-02 1.02853847e+00 -5.49407303e-01 -3.19348633e-01
-2.76474774e-01 4.94913906e-01 -2.06853554e-01 6.77275360e-01
-5.73492944e-01 -6.67996705e-02 9.67862487e-01 4.18492220e-02
6.92910373e-01 -6.73222482e-01 -1.19091049e-01 -7.95965314e-01
-2.27464706e-01 -4.36309040e-01 8.95328522e-02 -7.89977312e-01
-1.42586029e+00 3.77229571e-01 1.85283348e-01 -1.14841115e+00
-6.65237427e-01 -7.45850205e-01 -9.10247803e-01 1.67766011e+00
-1.30487192e+00 -6.47656620e-01 1.58893064e-01 1.06193610e-01
1.16587967e-01 3.65917057e-01 9.36432004e-01 -1.55196249e-01
-2.10923597e-01 -5.62137008e-01 8.17120671e-01 -1.24289706e-01
5.45360208e-01 -1.49777770e+00 7.05149114e-01 8.59035313e-01
-2.31001168e-01 6.26727998e-01 1.22008944e+00 -8.24388742e-01
-9.46388543e-01 -1.11619091e+00 1.14277565e+00 -5.21674454e-01
1.10216856e+00 -2.30362251e-01 -1.24509490e+00 9.15461302e-01
3.31099361e-01 2.35874042e-01 4.07063901e-01 -2.20086053e-02
1.98448122e-01 -1.33453980e-01 -7.67818689e-01 3.70517910e-01
1.05026782e-01 -6.63297653e-01 -7.31783092e-01 2.57646561e-01
2.30823204e-01 -5.01743078e-01 -9.27765608e-01 4.67274606e-01
6.49951220e-01 -7.91632533e-01 5.72225153e-01 -4.71854836e-01
1.23779349e-01 -4.86986965e-01 1.40667586e-02 -1.67340159e+00
-2.65940845e-01 -9.14886057e-01 -3.35510187e-02 8.97478402e-01
4.36120242e-01 -1.02347994e+00 2.43615836e-01 7.48107612e-01
7.27162659e-02 -4.53644425e-01 -1.14597607e+00 -8.10554385e-01
3.32735866e-01 -7.11317062e-01 5.13879061e-01 8.00318658e-01
-5.72117865e-01 -1.73690662e-01 -4.67641860e-01 7.21335292e-01
4.39653933e-01 7.67696053e-02 4.13069397e-01 -1.78851271e+00
-2.84842372e-01 -3.12899411e-01 -2.84533966e-02 -9.99554992e-01
1.20371580e-01 -3.55437309e-01 5.29578865e-01 -1.38328934e+00
-3.91361088e-01 -2.04084083e-01 -1.61598444e-01 1.21946037e-01
-6.05085923e-04 -2.72093415e-01 -1.17447786e-01 3.51810306e-01
3.59084725e-01 5.57514548e-01 1.08300650e+00 2.76164562e-01
-2.99961090e-01 2.70041734e-01 1.69769004e-01 9.45424318e-01
7.23172605e-01 -5.78434467e-01 -2.00045854e-01 -1.49170861e-01
3.51245254e-01 3.91243726e-01 6.54435694e-01 -1.13024569e+00
3.25478435e-01 -1.85630217e-01 4.71696258e-01 -7.86181331e-01
3.66610229e-01 -1.01352823e+00 7.28341818e-01 4.74533767e-01
-3.26013491e-02 4.00029600e-01 6.54435039e-01 8.36207151e-01
-3.84778202e-01 -2.22038124e-02 6.71042740e-01 -1.91846132e-01
-5.33424497e-01 2.46642798e-01 -1.04686081e+00 -2.22523004e-01
6.65134192e-01 8.68032593e-03 2.29366779e-01 -7.77786374e-01
-1.12656593e+00 1.90238535e-01 3.01726997e-01 -7.47791976e-02
3.36857326e-02 -8.57045650e-01 -7.81316578e-01 4.03887957e-01
-4.43156362e-01 -2.66000107e-02 3.22738677e-01 9.67176974e-01
-7.30619967e-01 3.46626699e-01 2.17265505e-02 -7.27502823e-01
-6.36165679e-01 1.67640299e-01 9.48231339e-01 -4.62271750e-01
-6.50510371e-01 6.21675670e-01 1.12697102e-01 -5.33861399e-01
-2.44398087e-01 -7.63503671e-01 2.74088047e-02 3.16823930e-01
4.21690524e-01 1.15195289e-01 -1.06687889e-01 -5.39633214e-01
-1.06820114e-01 4.78190064e-01 4.16654617e-01 -6.65814281e-01
1.31384909e+00 -1.59190223e-01 -3.30901533e-01 1.12087071e+00
8.90601397e-01 -2.12332174e-01 -2.09347701e+00 -1.71315163e-01
4.03841361e-02 -4.02396694e-02 5.50674796e-01 -6.70060933e-01
-7.94533014e-01 9.40677047e-01 5.83106935e-01 5.39888382e-01
8.61154854e-01 -4.95803267e-01 4.89979327e-01 1.86333984e-01
-7.60376006e-02 -9.15583372e-01 -7.90544033e-01 1.01810944e+00
9.04598713e-01 -8.19463313e-01 6.42431295e-03 2.15313211e-01
-3.19040030e-01 1.43390751e+00 -1.29672259e-01 -3.79152238e-01
1.48843408e+00 4.85824525e-01 3.29121470e-01 -1.61799848e-01
-9.20405984e-01 2.74905693e-02 3.34044039e-01 1.26515999e-01
4.80830252e-01 1.67847231e-01 9.75769609e-02 -7.57374316e-02
-1.71523467e-01 1.80433393e-02 3.25990647e-01 9.35045302e-01
-2.49148116e-01 -6.01867437e-01 -6.84937894e-01 3.23086977e-01
-5.95738947e-01 -2.69722372e-01 4.67012703e-01 7.32736647e-01
-1.51837543e-01 6.56930506e-01 7.51540482e-01 2.25294143e-01
5.59673570e-02 5.20389378e-01 -4.26494405e-02 -3.15749973e-01
-5.03357053e-01 3.02446127e-01 1.73677772e-01 -4.73604441e-01
-4.80020195e-01 -1.09856200e+00 -9.16755557e-01 -3.92101973e-01
-1.86502978e-01 2.04015076e-01 7.60449708e-01 1.22727835e+00
1.73838243e-01 6.11853480e-01 5.44166923e-01 -1.35462558e+00
-5.86700976e-01 -1.13576245e+00 -7.69813180e-01 -1.49126500e-01
7.66085088e-01 -6.10274315e-01 -9.46454287e-01 2.22513437e-01]
|
[6.565573215484619, 3.2887582778930664]
|
e932902d-2d75-45f6-88d1-ffedf1a17fca
|
micro-video-tagging-via-jointly-modeling
|
2303.08318
| null |
https://arxiv.org/abs/2303.08318v1
|
https://arxiv.org/pdf/2303.08318v1.pdf
|
Micro-video Tagging via Jointly Modeling Social Influence and Tag Relation
|
The last decade has witnessed the proliferation of micro-videos on various user-generated content platforms. According to our statistics, around 85.7\% of micro-videos lack annotation. In this paper, we focus on annotating micro-videos with tags. Existing methods mostly focus on analyzing video content, neglecting users' social influence and tag relation. Meanwhile, existing tag relation construction methods suffer from either deficient performance or low tag coverage. To jointly model social influence and tag relation, we formulate micro-video tagging as a link prediction problem in a constructed heterogeneous network. Specifically, the tag relation (represented by tag ontology) is constructed in a semi-supervised manner. Then, we combine tag relation, video-tag annotation, and user-follow relation to build the network. Afterward, a better video and tag representation are derived through Behavior Spread modeling and visual and linguistic knowledge aggregation. Finally, the semantic similarity between each micro-video and all candidate tags is calculated in this video-tag network. Extensive experiments on industrial datasets of three verticals verify the superiority of our model compared with several state-of-the-art baselines.
|
['Liqiang Nie', 'Dai Meng', 'Jianlong Wu', 'Yinwei Wei', 'Tian Gan', 'Xiao Wang']
|
2023-03-15
| null | null | null | null |
['semantic-textual-similarity']
|
['natural-language-processing']
|
[-1.67917207e-01 1.53139725e-01 -6.40150368e-01 -2.07681611e-01
-2.55799890e-01 -4.07891572e-01 5.88385940e-01 3.50234389e-01
-1.23422906e-01 5.45529246e-01 5.69543064e-01 2.52772003e-01
-2.10174397e-01 -6.93320692e-01 -3.54037076e-01 -5.64325094e-01
-6.06889278e-02 1.28917858e-01 7.87920475e-01 8.92502517e-02
-1.28919519e-02 -3.94691169e-01 -1.53564501e+00 1.56191394e-01
8.12385380e-01 1.20513725e+00 3.19072306e-01 -4.15237658e-02
-3.89860898e-01 1.27393091e+00 -2.05902100e-01 -6.35452032e-01
-3.65841448e-01 -2.65312672e-01 -7.93450832e-01 3.67617518e-01
1.39078528e-01 3.76924090e-02 -5.13891459e-01 1.25701416e+00
2.19333902e-01 1.29903749e-01 4.28038895e-01 -1.47824287e+00
-5.11078775e-01 1.16137314e+00 -4.25949633e-01 1.49669632e-01
4.95424092e-01 -8.56077135e-01 1.16756952e+00 -4.98985678e-01
8.93346369e-01 1.04613233e+00 6.75745070e-01 8.40024352e-02
-7.92434871e-01 -5.07198572e-01 2.93389529e-01 1.69348121e-01
-1.74941826e+00 -8.23445991e-02 7.61406660e-01 -6.68333113e-01
4.54066575e-01 9.52947140e-02 7.86062777e-01 7.87739575e-01
-1.70078240e-02 8.12819123e-01 4.86912757e-01 -2.07231313e-01
-3.05322260e-01 1.91638425e-01 5.42227887e-02 1.03819644e+00
2.01324791e-01 -6.68461621e-01 -7.30604827e-01 -1.27792031e-01
4.83325094e-01 3.25135529e-01 -2.69171089e-01 -5.96048892e-01
-1.25434339e+00 5.52686930e-01 2.34915659e-01 6.81214333e-01
-2.56882280e-01 3.72939147e-02 5.21464467e-01 -1.40450105e-01
1.04377484e+00 -3.64626609e-02 -1.51422054e-01 -1.67499781e-01
-7.51524925e-01 -2.42992342e-01 6.80204153e-01 1.38488376e+00
8.21581244e-01 -3.20868403e-01 -9.91973206e-02 1.02383852e+00
6.07104242e-01 2.31021017e-01 4.33340639e-01 -8.94500077e-01
4.17444915e-01 8.79623890e-01 7.44669745e-03 -1.80412543e+00
-4.11271632e-01 -6.89396858e-01 -5.27703106e-01 -9.41559017e-01
-1.03737988e-01 -1.56011671e-01 -2.54756540e-01 1.41985250e+00
3.96358430e-01 6.97422981e-01 -2.41366804e-01 7.49563336e-01
9.87684965e-01 6.17754459e-01 5.47655225e-01 -5.90309083e-01
1.51459324e+00 -1.04084027e+00 -1.05878627e+00 2.03504100e-01
1.01324093e+00 -8.97686064e-01 5.11646092e-01 3.51081118e-02
-7.24783301e-01 -5.08820713e-01 -7.95885026e-01 3.68056327e-01
-3.22602749e-01 1.54106900e-01 5.41345835e-01 2.10870460e-01
-8.18404615e-01 2.47864172e-01 -4.37420964e-01 -8.38849962e-01
4.42946851e-01 2.67340422e-01 -2.23356202e-01 -9.30280760e-02
-1.22641420e+00 2.24018067e-01 5.83418369e-01 -2.48163074e-01
-4.75827336e-01 -5.56281447e-01 -6.44934058e-01 -1.74155325e-01
9.30518270e-01 -6.59658670e-01 8.83675635e-01 -9.33709443e-01
-8.59281600e-01 8.23534191e-01 -5.93762808e-02 -1.07951373e-01
-6.38194606e-02 -2.24902242e-01 -1.04924631e+00 2.25858569e-01
4.06148821e-01 4.05643284e-01 4.14649427e-01 -1.53111637e+00
-1.42500961e+00 -1.43248767e-01 3.61416548e-01 2.44236410e-01
-9.12806988e-01 1.34785771e-01 -1.21798551e+00 -7.46361136e-01
2.34278783e-01 -8.46841156e-01 1.21357359e-01 -6.51487410e-01
-5.15687406e-01 -4.59021389e-01 9.64838803e-01 -3.86628985e-01
2.04028177e+00 -2.26715660e+00 3.49209532e-02 4.14778024e-01
6.19633138e-01 3.78326252e-02 1.46145657e-01 4.22071934e-01
3.01022083e-01 2.49110132e-01 4.19698924e-01 -2.35269591e-01
1.15749732e-01 1.59409165e-01 -3.00638359e-02 2.63440251e-01
-6.95941746e-01 7.49369323e-01 -1.32634330e+00 -1.30401969e+00
-7.09355101e-02 3.53335261e-01 -5.37221730e-01 1.35800257e-01
-1.15219221e-01 2.10891396e-01 -9.09254253e-01 7.44532168e-01
1.60539240e-01 -5.06844223e-01 6.19486928e-01 -7.13192999e-01
-1.37559488e-01 -8.41268301e-02 -7.19998419e-01 1.87023425e+00
-2.49090046e-01 4.00974900e-01 -1.08325131e-01 -9.26678836e-01
7.70765424e-01 4.17105436e-01 1.30770552e+00 -2.88337588e-01
3.65689754e-01 9.23456624e-02 -3.77810061e-01 -8.52467299e-01
6.33013010e-01 3.16617429e-01 -1.49915636e-01 3.66411835e-01
9.64553133e-02 8.34528506e-01 4.02529091e-01 6.11749649e-01
8.99789989e-01 4.23820883e-01 1.18885033e-01 -3.29346895e-01
6.21971786e-01 -3.60655077e-02 6.29992008e-01 3.52476001e-01
-5.02569042e-02 1.22864731e-01 4.42797929e-01 -1.55803010e-01
-4.13078219e-01 -7.88532555e-01 1.43688589e-01 1.46155262e+00
7.09033847e-01 -1.18363702e+00 -9.05648351e-01 -8.46199274e-01
-2.90780395e-01 2.42617682e-01 -4.53968823e-01 -1.11600846e-01
-2.94552922e-01 -6.66400611e-01 3.68151128e-01 2.20360219e-01
6.56289279e-01 -7.91510582e-01 1.86312541e-01 3.93510908e-01
-7.63298750e-01 -1.45768011e+00 -5.69847524e-01 -4.66726482e-01
-6.00731492e-01 -1.19665563e+00 -3.08797091e-01 -1.18433475e+00
7.93144941e-01 6.51729286e-01 1.22536719e+00 3.94775718e-01
3.40140253e-01 5.77406764e-01 -9.24353123e-01 9.85563025e-02
8.67877603e-02 2.81243175e-01 2.56798685e-01 5.16163349e-01
7.04280078e-01 -4.77674395e-01 -6.44017577e-01 8.19065452e-01
-7.38983214e-01 8.93959552e-02 2.38716915e-01 2.82132119e-01
6.71926618e-01 6.21261716e-01 2.09122196e-01 -1.14019668e+00
2.39805311e-01 -9.42797720e-01 -1.40410483e-01 2.82156974e-01
-7.36103296e-01 -5.31948566e-01 1.28083900e-01 -3.91316414e-01
-1.17949772e+00 -3.98106948e-02 4.07763392e-01 -2.42290094e-01
1.53315425e-01 8.79154503e-01 -3.60948771e-01 1.58125043e-01
6.01342544e-02 -2.02555861e-02 -3.33325624e-01 -4.36195999e-01
1.56337038e-01 7.18474150e-01 2.71561831e-01 -5.15020132e-01
7.45370865e-01 5.42704344e-01 -1.29255608e-01 -4.16867077e-01
-1.45310628e+00 -9.80425358e-01 -5.59920669e-01 -8.32112849e-01
1.24062371e+00 -1.08980083e+00 -7.56654918e-01 1.94892898e-01
-8.06659400e-01 9.14457887e-02 2.39207491e-01 7.57065356e-01
-2.92660296e-01 4.81594205e-01 -5.94059050e-01 -5.12227654e-01
1.70300692e-01 -8.48497987e-01 8.50392997e-01 6.93584755e-02
-1.45623580e-01 -1.34358084e+00 -9.39603522e-02 7.65512347e-01
-1.14785237e-02 7.53904283e-02 6.08976305e-01 -5.98317444e-01
-5.73861182e-01 -3.94008249e-01 -5.15940249e-01 3.52427736e-02
4.15277243e-01 -1.21284719e-03 -5.82313001e-01 -5.08193225e-02
-5.25923371e-01 1.40637934e-01 5.43028474e-01 2.92100459e-01
1.09292877e+00 -3.56200278e-01 -1.01320732e+00 2.37806633e-01
1.22097969e+00 2.86563963e-01 3.49230230e-01 3.12329203e-01
1.35861897e+00 5.86300910e-01 1.29189670e+00 6.77501798e-01
8.29849303e-01 9.35394466e-01 4.53036219e-01 1.06594503e-01
1.02829926e-01 -5.99434555e-01 2.92550236e-01 1.56882143e+00
-5.99997878e-01 -7.67633617e-01 -7.62698352e-01 3.95370424e-01
-2.34129477e+00 -1.08263421e+00 -5.03756762e-01 1.71981835e+00
3.90006751e-01 1.13807149e-01 3.70123029e-01 -1.56487245e-02
1.13953841e+00 1.96483284e-01 1.46502569e-01 5.95322788e-01
-6.65364116e-02 -8.30801487e-01 5.51332295e-01 2.65899837e-01
-1.29262853e+00 1.17683089e+00 4.86578083e+00 1.28863502e+00
-4.22075957e-01 5.69476783e-01 3.90898675e-01 5.63016720e-02
-2.79952407e-01 1.17234350e-03 -8.25585127e-01 7.12106168e-01
6.02508664e-01 -1.72463357e-01 1.32344425e-01 6.58478320e-01
2.81419635e-01 -4.38383929e-02 -6.22915983e-01 1.01007450e+00
4.51401621e-01 -1.28082705e+00 6.52383864e-02 2.09848389e-01
7.70721674e-01 -3.25470060e-01 -2.51068950e-01 1.92147866e-01
2.57459819e-01 -2.99465269e-01 8.38507235e-01 7.37961650e-01
5.14251292e-01 -6.18987620e-01 1.09969342e+00 7.03575313e-02
-1.86903238e+00 -6.31626844e-02 -1.90400332e-01 2.34369427e-01
3.77423435e-01 6.21001244e-01 -2.61169165e-01 9.07502651e-01
8.46332431e-01 1.54193544e+00 -3.79068136e-01 8.81965816e-01
-3.06130294e-02 7.28486180e-01 8.21707323e-02 -4.30852138e-02
2.74989426e-01 -3.66006881e-01 2.93236494e-01 1.19258666e+00
6.41370237e-01 2.72115499e-01 5.39127529e-01 6.75123706e-02
-2.43971959e-01 5.39645731e-01 -4.70925540e-01 -1.15265824e-01
7.09546864e-01 1.35950541e+00 -1.04409099e+00 -2.07539394e-01
-8.67749929e-01 7.59561300e-01 1.66745350e-01 3.16542774e-01
-1.21465456e+00 -8.68171006e-02 4.89087194e-01 4.28771973e-01
1.19875364e-01 -8.64900835e-03 3.37398320e-01 -1.23865592e+00
-1.42545402e-01 -1.35675535e-01 6.05967402e-01 -8.65471482e-01
-1.21597481e+00 6.04159415e-01 8.77958909e-02 -1.71620321e+00
8.31234753e-02 -1.68156311e-01 -4.38136235e-02 -7.36124143e-02
-1.20484102e+00 -1.42093241e+00 -5.36413014e-01 6.29377365e-01
2.13321075e-01 -3.60776246e-01 5.21802545e-01 1.06270087e+00
-5.29842198e-01 3.84443730e-01 -1.78153753e-01 2.37523481e-01
6.19310915e-01 -7.55402029e-01 -5.58993518e-01 5.73321819e-01
1.96909532e-01 4.25615728e-01 4.45399433e-01 -9.27429914e-01
-9.44439709e-01 -1.40051115e+00 1.20956790e+00 -4.39168513e-01
1.07330215e+00 -2.11786851e-01 -5.11563122e-01 7.05727756e-01
2.12740198e-01 1.79531857e-01 1.13716507e+00 2.13169366e-01
-2.39936069e-01 -3.67983282e-01 -6.43966675e-01 5.23928225e-01
1.79434204e+00 -5.67238808e-01 -2.23918810e-01 4.80652839e-01
7.99454391e-01 4.56429534e-02 -1.37621486e+00 2.09321097e-01
4.60058957e-01 -7.42328405e-01 8.97886395e-01 -2.40540981e-01
3.52075398e-01 -5.98921418e-01 -3.58944625e-01 -1.10326982e+00
-6.32753074e-01 -5.23204565e-01 -2.80522943e-01 1.95327783e+00
2.06783786e-01 -2.33492374e-01 7.51797438e-01 1.34356812e-01
-3.17055434e-01 -3.98799032e-01 -5.24206102e-01 -7.03376949e-01
-9.23384845e-01 -3.29688340e-01 5.91367066e-01 1.29401362e+00
4.75559175e-01 4.37654853e-01 -6.86963618e-01 1.57042190e-01
7.23494172e-01 1.64765455e-02 4.33305889e-01 -1.53166831e+00
7.07765073e-02 -2.95092642e-01 -6.35530353e-01 -1.06705737e+00
3.53920579e-01 -7.99720228e-01 -8.95319209e-02 -1.66504800e+00
6.48826420e-01 -7.58492768e-01 -6.37902915e-01 5.19525468e-01
4.16976139e-02 6.71068728e-01 -4.80414443e-02 6.49110854e-01
-1.63215411e+00 3.00649285e-01 1.24386096e+00 -1.18864067e-01
4.16372754e-02 -3.16366136e-01 -5.24627030e-01 9.09421742e-01
5.74376762e-01 -5.62006354e-01 -7.41413236e-01 -4.14102316e-01
6.75216794e-01 -3.79591398e-02 1.19794033e-01 -9.60743189e-01
3.18027794e-01 -8.96952078e-02 -2.74832129e-01 -4.03656214e-01
2.20134705e-01 -1.22978830e+00 5.29817939e-01 1.02064714e-01
-2.77070075e-01 -2.87432164e-01 -4.61723059e-01 8.98317575e-01
-5.60295165e-01 -6.52485043e-02 1.99621320e-02 5.99354692e-02
-9.87879753e-01 6.46716714e-01 -4.48217452e-01 3.96975651e-02
1.14631796e+00 -2.67898172e-01 -3.95987809e-01 -3.80433053e-01
-9.37661529e-01 2.83155829e-01 3.74677926e-01 6.80353999e-01
2.23382756e-01 -1.79066133e+00 -4.15580481e-01 -4.63859379e-01
5.89289784e-01 -4.83017296e-01 4.38202173e-01 1.12039518e+00
-1.40208140e-01 3.64029139e-01 1.59560233e-01 -4.08700973e-01
-1.52711236e+00 5.24315953e-01 -2.06627369e-01 -2.37691458e-02
-3.14072013e-01 6.78352535e-01 1.90463886e-01 1.22078918e-01
2.16057181e-01 9.43506956e-02 -8.91251802e-01 5.99660039e-01
1.44402444e-01 3.68359506e-01 -3.49595100e-01 -1.31535268e+00
-4.19219375e-01 7.55719960e-01 3.15346956e-01 4.07540441e-01
1.27008033e+00 -7.57563174e-01 -2.35808045e-01 4.37305868e-01
1.23733413e+00 2.41645187e-01 -7.80821800e-01 -5.13274014e-01
1.52698189e-01 -4.09030348e-01 2.54081070e-01 -2.28957847e-01
-1.58184814e+00 2.38039792e-01 2.87337065e-01 5.91471672e-01
9.24576998e-01 4.63928014e-01 9.10744905e-01 4.99354489e-02
6.47511065e-01 -1.26191008e+00 1.62537530e-01 1.68219283e-01
1.80169880e-01 -9.63694036e-01 1.49087846e-01 -1.24034810e+00
-6.18232131e-01 6.52298748e-01 6.51046515e-01 3.50708604e-01
1.17173111e+00 -9.32706445e-02 1.18212970e-02 -4.69123840e-01
-4.96321172e-01 -5.36208570e-01 5.02452374e-01 4.65029031e-01
5.68848073e-01 1.10437058e-01 -3.74389648e-01 7.42364109e-01
6.98537976e-02 1.87047526e-01 2.16692284e-01 8.87744665e-01
-5.54770947e-01 -1.03746283e+00 1.38562799e-01 5.17892659e-01
-6.66066766e-01 1.64815392e-02 7.35767558e-02 5.39846897e-01
7.05032110e-01 1.31487954e+00 2.81565696e-01 -9.82528806e-01
-8.69054124e-02 -2.43520215e-01 5.24790138e-02 -6.45735979e-01
-3.24033320e-01 5.16852438e-01 5.04780829e-01 -6.21058643e-01
-1.24739516e+00 -8.34237874e-01 -1.25550508e+00 -3.69300634e-01
-4.11070943e-01 5.49066186e-01 1.58960700e-01 9.18074012e-01
3.44237477e-01 7.12641835e-01 4.55942571e-01 -3.75216275e-01
5.08117259e-01 -7.26420403e-01 -8.41567874e-01 6.14600778e-01
-4.76955801e-01 -1.06877959e+00 -1.93664908e-01 4.96368468e-01]
|
[9.836834907531738, 0.8725608587265015]
|
e4956b70-2e1e-402d-ae53-8a8ee29ec691
|
context-aware-6d-pose-estimation-of-known
|
2212.05560
| null |
https://arxiv.org/abs/2212.05560v1
|
https://arxiv.org/pdf/2212.05560v1.pdf
|
Context-aware 6D Pose Estimation of Known Objects using RGB-D data
|
6D object pose estimation has been a research topic in the field of computer vision and robotics. Many modern world applications like robot grasping, manipulation, autonomous navigation etc, require the correct pose of objects present in a scene to perform their specific task. It becomes even harder when the objects are placed in a cluttered scene and the level of occlusion is high. Prior works have tried to overcome this problem but could not achieve accuracy that can be considered reliable in real-world applications. In this paper, we present an architecture that, unlike prior work, is context-aware. It utilizes the context information available to us about the objects. Our proposed architecture treats the objects separately according to their types i.e; symmetric and non-symmetric. A deeper estimator and refiner network pair is used for non-symmetric objects as compared to symmetric due to their intrinsic differences. Our experiments show an enhancement in the accuracy of about 3.2% over the LineMOD dataset, which is considered a benchmark for pose estimation in the occluded and cluttered scenes, against the prior state-of-the-art DenseFusion. Our results also show that the inference time we got is sufficient for real-time usage.
|
['G. C. Nandi', 'Vandana Kushwaha', 'Priya Shukla', 'Ankit Kumar']
|
2022-12-11
| null | null | null | null |
['6d-pose-estimation-1', '6d-pose-estimation']
|
['computer-vision', 'computer-vision']
|
[ 1.02731414e-01 -5.72465323e-02 1.41850933e-01 -5.03771305e-01
-3.99635524e-01 -2.91699678e-01 4.78779316e-01 2.33573839e-01
-6.84698462e-01 6.44782603e-01 -1.74816787e-01 1.38558358e-01
-9.59279835e-02 -5.86181343e-01 -8.48047137e-01 -6.53460741e-01
-4.43288311e-02 1.03884494e+00 7.12224782e-01 4.43560667e-02
2.35351905e-01 6.68398857e-01 -1.79854679e+00 1.93904072e-01
6.04296565e-01 1.11864042e+00 8.96767557e-01 2.49182686e-01
-6.27088323e-02 2.65786588e-01 -4.34307963e-01 -7.42930621e-02
4.96912658e-01 7.90424645e-03 -5.82337677e-01 1.86254993e-01
6.15002394e-01 -3.88728023e-01 1.84100773e-02 1.08894467e+00
3.52994859e-01 -1.01999426e-03 6.59587801e-01 -1.03316510e+00
1.02026090e-01 2.85097063e-01 -7.83389747e-01 1.86094409e-03
1.73929811e-01 -5.73109686e-02 6.43258631e-01 -9.68995512e-01
7.74943471e-01 1.35150707e+00 3.59341443e-01 3.51471305e-01
-9.83172655e-01 -5.18358052e-01 2.78798282e-01 4.47778612e-01
-1.11587679e+00 -2.05054253e-01 8.19596291e-01 -4.56938863e-01
6.18002176e-01 -3.66110094e-02 4.68005866e-01 9.03120577e-01
2.44991735e-01 7.28445351e-01 1.14076197e+00 -3.18701625e-01
3.41983348e-01 2.15236574e-01 -4.66941558e-02 2.43180528e-01
5.72363734e-01 -1.45997077e-01 -3.55917513e-01 2.82571197e-01
7.03939795e-01 1.14056498e-01 -2.75790781e-01 -1.11274886e+00
-1.31893778e+00 6.91193640e-01 7.52583563e-01 3.90554816e-01
-6.68582141e-01 1.60033599e-01 2.66820759e-01 -1.86093468e-02
2.87305564e-01 4.23220634e-01 -5.94204724e-01 -3.83330621e-02
-8.04653585e-01 3.62524003e-01 7.15059280e-01 1.06359732e+00
6.07222557e-01 -2.20548272e-01 2.07582518e-01 5.73107302e-01
4.59702373e-01 3.83831412e-01 1.47638813e-01 -6.97492659e-01
4.15102303e-01 5.08671582e-01 3.55627656e-01 -9.72871780e-01
-5.87333441e-01 -6.61082268e-01 -6.18970156e-01 4.14706886e-01
6.68524384e-01 1.30095229e-01 -9.45082009e-01 1.40838373e+00
6.77188218e-01 3.04654893e-02 -1.58140108e-01 1.28421211e+00
8.94045651e-01 4.08479512e-01 -2.19029382e-01 -1.89263541e-02
1.63098764e+00 -9.30949628e-01 -7.29160011e-01 -5.94398916e-01
1.04879946e-01 -1.11099720e+00 6.73861325e-01 6.85493410e-01
-8.44373167e-01 -5.43565631e-01 -1.06432199e+00 2.64801849e-02
-3.15582961e-01 2.05538034e-01 6.47076726e-01 2.63494104e-01
-5.44332683e-01 4.32687908e-01 -8.93693864e-01 -6.39239550e-01
4.65599030e-01 4.70501900e-01 -6.95124269e-01 -2.43331596e-01
-6.18936360e-01 1.35281873e+00 5.20556748e-01 3.50376427e-01
-8.19317579e-01 -3.58980864e-01 -6.50296807e-01 -9.37237889e-02
9.02617753e-01 -3.83307874e-01 1.13658857e+00 -5.15416503e-01
-1.19867957e+00 7.97615886e-01 4.77655120e-02 -3.80826414e-01
7.48148441e-01 -6.88736856e-01 2.03472853e-01 6.44035116e-02
-1.69859305e-02 6.59499049e-01 7.55603611e-01 -1.43605733e+00
-6.67442381e-01 -7.42256284e-01 2.83010513e-01 2.83085346e-01
1.91297144e-01 -1.90022558e-01 -6.20303452e-01 -2.90402442e-01
5.89221060e-01 -1.19871175e+00 -1.63835257e-01 4.26708579e-01
-2.19434604e-01 -1.30773231e-01 1.33394659e+00 -5.11793792e-01
3.59488726e-01 -1.85036182e+00 3.91460985e-01 -2.66292915e-02
-3.40939239e-02 2.59540021e-01 1.88725471e-01 3.91816884e-01
1.30141512e-01 -4.77725536e-01 -1.52100369e-01 -4.84072238e-01
-1.08265683e-01 3.60859036e-01 3.79145667e-02 8.63746524e-01
5.36524355e-02 4.87596452e-01 -7.70310163e-01 -4.66439664e-01
5.79243720e-01 6.34190083e-01 -4.60621595e-01 3.01237494e-01
-3.04991066e-01 6.72866046e-01 -4.35935378e-01 5.66939533e-01
9.73668337e-01 4.74707298e-02 3.92516106e-02 -5.62818348e-01
-1.33033782e-01 1.15051620e-01 -1.63476098e+00 1.95622385e+00
-5.75457931e-01 5.19800842e-01 3.20752233e-01 -8.47938359e-01
9.42797124e-01 1.76659837e-01 4.56590444e-01 -4.08705592e-01
3.11091065e-01 3.90540063e-01 3.03509444e-01 -4.79364067e-01
4.50419635e-01 4.11367156e-02 1.91843346e-01 -4.89122421e-02
-1.01857763e-02 -3.79100949e-01 3.41980219e-01 -1.37622178e-01
8.20740104e-01 4.99680191e-01 2.40878850e-01 -3.78509015e-01
4.40685093e-01 -1.57974660e-01 5.10798633e-01 4.54240561e-01
-6.59297481e-02 7.25766361e-01 2.44020864e-01 -4.20800060e-01
-9.23673868e-01 -8.95339727e-01 -4.62786764e-01 5.38266599e-01
5.09436548e-01 -1.60776362e-01 -4.03116286e-01 -5.76730072e-01
7.86689222e-02 4.84493136e-01 -4.20084268e-01 1.63990170e-01
-6.39972806e-01 -6.01701021e-01 -2.61800528e-01 4.70662415e-01
6.97403789e-01 -1.14418280e+00 -1.34365642e+00 1.83994845e-01
-1.26059189e-01 -1.43924570e+00 1.56231597e-01 2.49332011e-01
-8.99000943e-01 -1.15479100e+00 -7.32794881e-01 -6.24727130e-01
7.01487839e-01 2.93986887e-01 9.48027074e-01 -2.51275420e-01
-4.44130749e-01 7.01653436e-02 -4.58470196e-01 -6.35093093e-01
-4.24098317e-03 1.33037847e-02 -4.58612572e-03 -2.02330016e-03
1.22675337e-01 -5.07270277e-01 -7.81741261e-01 5.18156886e-01
-9.10899460e-01 1.88060459e-02 8.23174238e-01 5.90868413e-01
4.84705687e-01 -2.99709588e-02 2.40378961e-01 -7.96999753e-01
-8.50769430e-02 -2.59993374e-01 -8.11900556e-01 4.68339399e-02
-1.35312155e-01 1.33886367e-01 1.49576217e-01 -4.17976081e-01
-9.31665599e-01 6.09384000e-01 -9.06514078e-02 -4.52852935e-01
-3.07745218e-01 3.34583640e-01 -3.13426018e-01 -9.96574014e-02
3.38045180e-01 -2.54802823e-01 -9.35352594e-02 -7.52705097e-01
1.24890357e-02 4.88029689e-01 3.90262455e-01 -4.73166198e-01
6.84730232e-01 5.84594250e-01 4.44494545e-01 -7.73142338e-01
-7.80507088e-01 -7.57798910e-01 -8.34597707e-01 -2.63941824e-01
8.70724499e-01 -7.48800814e-01 -8.18832874e-01 4.36459690e-01
-1.43476641e+00 3.59374993e-02 2.42589880e-02 7.10909665e-01
-4.83761162e-01 3.20539683e-01 -1.14620604e-01 -9.22326207e-01
-7.66531900e-02 -1.50491500e+00 1.28653026e+00 1.11112215e-01
6.94801360e-02 -5.52983046e-01 -3.58533919e-01 2.63340533e-01
4.79392081e-01 5.14050364e-01 4.54117268e-01 -4.55250353e-01
-8.75228703e-01 -2.85903186e-01 -2.84742862e-01 1.67591333e-01
1.77871376e-01 -1.36925444e-01 -9.78248954e-01 -3.34735483e-01
1.44694790e-01 -2.70784855e-01 7.01757669e-01 2.98388273e-01
1.10109246e+00 1.63175926e-01 -4.00931954e-01 2.10410267e-01
1.57351804e+00 9.01880190e-02 5.02570271e-01 3.04882795e-01
5.56397259e-01 9.51294601e-01 1.07441652e+00 4.11287814e-01
4.74024564e-02 1.22671628e+00 1.11244464e+00 -8.59261863e-03
-5.52286096e-02 2.02892408e-01 -5.73606491e-02 3.36899638e-01
-1.40514269e-01 -2.43506029e-01 -9.21841264e-01 4.96620476e-01
-2.00914836e+00 -6.24666333e-01 -4.03417856e-01 2.34811759e+00
3.67185801e-01 3.62412006e-01 -1.95314586e-01 1.78317219e-01
5.80579221e-01 -9.25986096e-02 -4.77899700e-01 -5.11312559e-02
1.48311317e-01 4.83446606e-02 6.12826169e-01 3.36498082e-01
-1.11475670e+00 6.96905255e-01 4.61471891e+00 5.28709829e-01
-1.16189253e+00 1.48862116e-02 1.65123552e-01 3.82719040e-02
2.83045679e-01 -2.55068112e-03 -8.68020475e-01 4.18171734e-01
3.38153213e-01 3.78977418e-01 -9.69126225e-02 1.03012407e+00
-3.86357568e-02 -6.99445546e-01 -1.30465817e+00 1.14341950e+00
1.18879683e-01 -8.19024920e-01 -3.87775034e-01 6.38797134e-02
4.26271111e-01 1.14160635e-01 -1.08806394e-01 7.71449357e-02
-1.88622639e-01 -8.04440081e-01 9.12618756e-01 4.13969219e-01
2.68121690e-01 -6.38256967e-01 1.19965160e+00 6.83115363e-01
-1.06564009e+00 4.19446304e-02 -4.82501388e-01 -8.37374926e-02
3.51352245e-01 8.62681448e-01 -1.10633743e+00 5.98554194e-01
8.48557115e-01 4.11445856e-01 -5.49154341e-01 1.42339957e+00
-2.32020587e-01 1.05061330e-01 -6.94909871e-01 1.11547848e-02
1.65628701e-01 -7.76720420e-02 6.77105129e-01 8.70251656e-01
2.61836469e-01 -1.31124541e-01 3.33158433e-01 5.37852287e-01
2.26857141e-01 1.69359837e-02 -4.29986626e-01 4.40149933e-01
1.71327740e-01 1.40971279e+00 -1.14887559e+00 -1.87556162e-01
-2.20707878e-01 9.23848510e-01 1.93512052e-01 -8.84227529e-02
-7.36863792e-01 -1.61609232e-01 3.80977064e-01 3.10197473e-01
6.77761495e-01 -5.57442188e-01 -2.37495769e-02 -1.10999918e+00
3.76934707e-01 -5.14656186e-01 1.40451128e-02 -7.56073773e-01
-9.85993207e-01 6.50124311e-01 4.51140612e-01 -1.19593251e+00
-2.24126950e-01 -9.43940282e-01 -1.87502310e-01 6.33681953e-01
-1.49973416e+00 -1.08744931e+00 -6.80077016e-01 2.05202073e-01
8.50275815e-01 3.17570716e-01 6.05979800e-01 3.32768410e-01
-2.48448655e-01 -1.78893041e-02 -1.43316880e-01 -6.19099699e-02
7.52109766e-01 -1.08015895e+00 -5.15340827e-02 7.49684095e-01
2.33906299e-01 4.01057631e-01 1.09680641e+00 -5.82053065e-01
-1.54819393e+00 -7.00025022e-01 7.07126677e-01 -3.25526476e-01
1.97874039e-01 -7.24401593e-01 -8.79587352e-01 4.42143679e-01
3.89393307e-02 3.69562715e-01 2.32960805e-02 4.93046967e-03
-4.22217101e-02 -3.10753703e-01 -1.28462958e+00 2.90607482e-01
1.02872467e+00 1.38020724e-01 -5.96307814e-01 4.27306056e-01
4.16512460e-01 -7.86016047e-01 -6.28212333e-01 8.10880899e-01
6.91451132e-01 -1.20929337e+00 8.94688308e-01 -1.71290576e-01
2.23875478e-01 -5.89892566e-01 -2.13044196e-01 -1.01878834e+00
4.09267694e-02 -1.23433340e-02 -7.42751407e-03 9.17193472e-01
-5.25647588e-02 -4.49097365e-01 9.84491110e-01 3.48461539e-01
-1.16180263e-01 -7.87071466e-01 -1.07032359e+00 -7.57772267e-01
-3.96852136e-01 -4.10015225e-01 3.57462883e-01 4.23560739e-01
-6.46428168e-01 3.38182986e-01 -2.85046697e-01 3.25073123e-01
7.24331439e-01 4.50192422e-01 1.02468145e+00 -1.48261321e+00
-2.07449377e-01 -1.51094213e-01 -9.16835606e-01 -1.04948306e+00
-7.96883628e-02 -4.53911930e-01 3.28974873e-01 -1.79284966e+00
-3.27241346e-02 -5.85240901e-01 6.62708208e-02 3.51079464e-01
2.04769403e-01 4.33040828e-01 4.53860193e-01 1.26150072e-01
-4.52909291e-01 4.92848188e-01 1.33767855e+00 4.27064411e-02
9.24592614e-02 2.38411874e-01 -4.53259908e-02 8.92191648e-01
6.62636518e-01 -4.63017136e-01 -3.29089969e-01 -4.29558098e-01
1.18821315e-01 -1.56888023e-01 4.73740250e-01 -1.35338271e+00
1.68230325e-01 -1.48542356e-02 3.84325713e-01 -1.05058455e+00
9.24248457e-01 -1.60878015e+00 2.27312684e-01 6.89798415e-01
9.04907286e-02 -6.42596036e-02 1.90058365e-01 4.38832760e-01
-1.10935703e-01 -4.90809083e-01 7.86046028e-01 -2.02207804e-01
-9.07275975e-01 2.24857301e-01 6.70231953e-02 -3.13226968e-01
1.27042294e+00 -1.39280111e-01 -1.20599493e-02 -3.94786268e-01
-5.82396686e-01 1.18423417e-01 4.62928087e-01 5.07172287e-01
6.20312214e-01 -9.38690901e-01 -8.24195623e-01 1.15178730e-02
2.06718177e-01 5.49793303e-01 1.74911469e-01 9.35897529e-01
-6.77109003e-01 3.16200316e-01 -3.94971400e-01 -1.13918293e+00
-1.43866158e+00 5.60286760e-01 3.67601290e-02 -1.39139548e-01
-5.83254099e-01 6.56885684e-01 3.58538628e-01 -2.30964944e-01
4.26873863e-01 -6.03520453e-01 -2.10041285e-01 1.98830701e-02
2.48292908e-01 2.31881574e-01 3.62806350e-01 -6.35431349e-01
-4.89592373e-01 8.98394048e-01 -1.30653426e-01 4.48644795e-02
1.52010441e+00 7.64022768e-02 -1.18877172e-01 5.89762270e-01
9.73749220e-01 -1.02540404e-01 -1.31406701e+00 -2.67980069e-01
-2.35045534e-02 -7.61418104e-01 4.45751026e-02 -6.51720285e-01
-1.11226106e+00 1.02945948e+00 7.17144728e-01 2.83748321e-02
8.98681223e-01 1.68772221e-01 4.06866938e-01 5.15246809e-01
8.24545681e-01 -8.70211601e-01 -2.96366345e-02 3.76826793e-01
1.23759651e+00 -1.44393969e+00 4.66256082e-01 -8.20050240e-01
-4.20595348e-01 1.07779324e+00 8.29868853e-01 -3.43772024e-01
5.58607817e-01 2.96595544e-01 -1.02798961e-01 -1.04253538e-01
-4.01904285e-01 -3.41774434e-01 4.22606796e-01 5.15344977e-01
2.08191261e-01 -2.26896003e-01 -3.42286706e-01 -5.33410907e-02
-7.37467110e-02 -3.11467201e-01 2.98727781e-01 1.16442192e+00
-5.37018955e-01 -1.05431437e+00 -6.48162723e-01 2.55015224e-01
-3.36203545e-01 4.34671879e-01 -1.03552759e-01 1.09581387e+00
4.16425467e-01 6.90481186e-01 2.09081188e-01 1.31163940e-01
4.67458695e-01 -2.18535379e-01 8.90477359e-01 -6.65514052e-01
-3.75934750e-01 2.36288473e-01 -8.57695099e-03 -6.15219116e-01
-8.46418679e-01 -6.47353053e-01 -1.17806768e+00 2.17762083e-01
-6.17237389e-01 -1.09890796e-01 1.33285725e+00 8.51025820e-01
5.35846427e-02 4.82605368e-01 2.08472967e-01 -1.42682648e+00
-4.30579990e-01 -1.09205163e+00 -3.99582267e-01 4.38108325e-01
2.07036197e-01 -1.25463009e+00 -1.26770094e-01 -2.26608053e-01]
|
[7.403671741485596, -2.398400068283081]
|
e425735a-9fe2-4349-84e0-8553929d98ee
|
icolorit-towards-propagating-local-hint-to
|
2207.06831
| null |
https://arxiv.org/abs/2207.06831v4
|
https://arxiv.org/pdf/2207.06831v4.pdf
|
iColoriT: Towards Propagating Local Hint to the Right Region in Interactive Colorization by Leveraging Vision Transformer
|
Point-interactive image colorization aims to colorize grayscale images when a user provides the colors for specific locations. It is essential for point-interactive colorization methods to appropriately propagate user-provided colors (i.e., user hints) in the entire image to obtain a reasonably colorized image with minimal user effort. However, existing approaches often produce partially colorized results due to the inefficient design of stacking convolutional layers to propagate hints to distant relevant regions. To address this problem, we present iColoriT, a novel point-interactive colorization Vision Transformer capable of propagating user hints to relevant regions, leveraging the global receptive field of Transformers. The self-attention mechanism of Transformers enables iColoriT to selectively colorize relevant regions with only a few local hints. Our approach colorizes images in real-time by utilizing pixel shuffling, an efficient upsampling technique that replaces the decoder architecture. Also, in order to mitigate the artifacts caused by pixel shuffling with large upsampling ratios, we present the local stabilizing layer. Extensive quantitative and qualitative results demonstrate that our approach highly outperforms existing methods for point-interactive colorization, producing accurately colorized images with a user's minimal effort. Official codes are available at https://pmh9960.github.io/research/iColoriT
|
['Jaegul Choo', 'Minho Park', 'Jooyeol Yun', 'Sanghyeon Lee']
|
2022-07-14
| null | null | null | null |
['colorization', 'point-interactive-image-colorization']
|
['computer-vision', 'computer-vision']
|
[ 1.72157630e-01 -2.84137309e-01 -3.43939774e-02 -2.28278801e-01
-7.63152301e-01 -7.69189537e-01 2.01792374e-01 -1.37975588e-01
-3.58265340e-01 5.19054890e-01 5.79896159e-02 -4.45806026e-01
4.60677743e-01 -6.73373878e-01 -8.37018788e-01 -5.56783020e-01
4.13073897e-01 -2.85214245e-01 3.31588477e-01 -3.05963252e-02
4.86328363e-01 5.75907826e-01 -1.30546355e+00 4.51448739e-01
1.04826021e+00 7.21491873e-01 4.73976642e-01 8.49752188e-01
-2.13484004e-01 4.93686467e-01 -5.48690736e-01 -2.51006126e-01
4.05989289e-01 -5.16092420e-01 -5.62770963e-01 4.91188243e-02
7.84258246e-01 -6.64170682e-01 -1.20635524e-01 1.19878662e+00
2.81055301e-01 -6.67034090e-02 2.50742763e-01 -1.20270693e+00
-1.24936736e+00 5.00233829e-01 -1.08650672e+00 1.15018331e-01
1.37665942e-01 3.34009916e-01 8.20266485e-01 -1.05672264e+00
3.15644771e-01 1.03476405e+00 3.73702854e-01 6.17212951e-01
-1.45570111e+00 -9.06243384e-01 3.84762585e-01 1.97400078e-01
-1.38727808e+00 -2.96070695e-01 8.29245865e-01 -1.39624998e-01
5.38406909e-01 5.60433030e-01 7.71957636e-01 4.90644306e-01
7.48630017e-02 7.94533730e-01 1.17893291e+00 -3.23466897e-01
2.71601498e-01 -3.30030955e-02 -2.59775035e-02 8.20420802e-01
9.66452584e-02 -4.55638170e-02 -6.34415030e-01 8.55190232e-02
1.34718037e+00 2.77209908e-01 -4.85076457e-01 -3.74273658e-01
-1.36944616e+00 3.65182281e-01 9.39225674e-01 4.19545844e-02
-4.44264293e-01 7.30157256e-01 4.12616245e-02 -5.76565303e-02
2.98079580e-01 5.24420142e-01 -2.67796218e-01 -1.26427755e-01
-9.92262900e-01 -1.97539881e-01 1.39872700e-01 1.05732667e+00
1.16323376e+00 1.20898917e-01 -3.17681104e-01 8.48981917e-01
6.10447628e-03 6.47135317e-01 1.13722004e-01 -1.25116742e+00
3.42276752e-01 6.26700044e-01 3.31849575e-01 -8.14169884e-01
-7.15465024e-02 -1.31320775e-01 -7.87603319e-01 7.41894245e-01
5.30815065e-01 -2.68126994e-01 -1.14342248e+00 1.68600559e+00
1.18352830e-01 7.04719871e-02 -3.58289510e-01 1.21028054e+00
4.13235843e-01 7.90097654e-01 1.96310654e-01 3.47241670e-01
1.55666375e+00 -1.02497101e+00 -4.41973716e-01 -2.32794002e-01
6.98469654e-02 -8.59036267e-01 1.80627894e+00 2.93905854e-01
-1.20909536e+00 -4.43466365e-01 -1.07448101e+00 -3.22997481e-01
-4.02498126e-01 4.91836190e-01 6.57689929e-01 6.27908468e-01
-1.40393186e+00 2.57047266e-01 -8.56253207e-01 -1.02762841e-01
4.65144902e-01 2.86618054e-01 -1.55564442e-01 4.62225080e-03
-8.76301229e-01 3.76282722e-01 -1.04821622e-01 1.43260926e-01
-6.43817544e-01 -1.00170302e+00 -5.17589569e-01 2.32923001e-01
1.79802895e-01 -4.70623463e-01 1.18809402e+00 -1.34050369e+00
-1.54148400e+00 4.63862211e-01 -3.65518570e-01 -5.33252880e-02
4.68294740e-01 -3.01137865e-01 -4.45764065e-02 5.22583306e-01
-1.16498610e-02 1.06138730e+00 9.77580130e-01 -1.45933628e+00
-7.88662434e-01 -9.65282768e-02 2.06375778e-01 2.49634907e-01
-4.88198072e-01 -1.27710074e-01 -1.25012600e+00 -7.27067888e-01
8.88181105e-02 -8.47399831e-01 -2.55428791e-01 4.16709632e-01
-6.83979988e-01 3.32472771e-01 8.61676574e-01 -5.35756648e-01
1.02772152e+00 -2.29079270e+00 -1.98144183e-01 1.28955126e-01
4.59224731e-01 1.45445704e-01 -2.04491466e-01 2.21658558e-01
-1.95699483e-01 1.11893073e-01 -1.83668673e-01 -1.62171707e-01
-1.37741283e-01 -1.83172166e-01 -2.86532015e-01 2.86702603e-01
9.51888040e-02 9.54906285e-01 -9.51665342e-01 -2.97424227e-01
5.26328444e-01 7.33524501e-01 -7.52585053e-01 -6.21727407e-02
-1.13784127e-01 1.94843858e-01 -2.50322342e-01 6.63725674e-01
1.11067712e+00 -3.68112266e-01 4.70145307e-02 -5.86319745e-01
-3.18281561e-01 -1.30942896e-01 -9.29825842e-01 1.56986487e+00
-5.34583867e-01 8.74311507e-01 5.40860966e-02 -2.73729533e-01
6.14360929e-01 4.87813377e-04 2.53067732e-01 -9.19966102e-01
-1.15646303e-01 2.39708461e-02 -4.77549583e-01 7.67894685e-02
7.86589205e-01 3.12720448e-01 -4.26551066e-02 6.00535572e-01
-4.84522551e-01 -3.39555927e-02 8.96446705e-02 5.23629069e-01
8.15468669e-01 -1.01234661e-02 -1.56163201e-01 -1.83769032e-01
1.84306890e-01 -8.41823667e-02 3.58974159e-01 6.32096291e-01
-1.66562423e-01 1.03336442e+00 6.13691151e-01 -2.33167812e-01
-1.13842535e+00 -1.23155928e+00 2.43960589e-01 1.21880388e+00
5.05180180e-01 -2.57093400e-01 -9.17995095e-01 -5.76289117e-01
-1.39470041e-01 7.58934498e-01 -7.89941788e-01 -7.10807592e-02
-4.86692727e-01 -4.31014359e-01 2.80481726e-01 6.21238649e-01
7.27611661e-01 -9.32697237e-01 -8.68159711e-01 -4.50058132e-02
-1.60278752e-01 -6.97718799e-01 -1.18770373e+00 1.23914719e-01
-7.28721261e-01 -9.97387886e-01 -1.26933575e+00 -7.50005305e-01
1.29159200e+00 8.58184695e-01 9.00605142e-01 1.70526445e-01
-3.34289253e-01 2.64316440e-01 -3.06263983e-01 1.43315047e-01
2.59495061e-02 -1.04732119e-01 -4.67463851e-01 1.22532500e-02
1.77024186e-01 -3.31179857e-01 -1.30193019e+00 3.67238134e-01
-9.72355545e-01 7.31073260e-01 6.38068616e-01 7.01248944e-01
5.93068480e-01 -2.97749579e-01 7.51845986e-02 -9.43097711e-01
5.80511689e-01 1.16518512e-01 -8.46881151e-01 2.97580153e-01
-3.98601860e-01 2.19123453e-01 8.65418375e-01 -3.77598763e-01
-1.12004101e+00 1.39357299e-01 2.34428659e-01 -4.47703928e-01
3.03501040e-02 -6.75055683e-02 -2.38744151e-02 -2.56110579e-01
5.44586837e-01 3.22130620e-01 -2.65389353e-01 -2.95594931e-01
8.92551601e-01 4.42668080e-01 6.85475230e-01 -4.84161019e-01
7.61463761e-01 5.96875131e-01 -5.62023938e-01 -4.60493505e-01
-2.41244718e-01 -9.85669568e-02 -3.58504087e-01 -3.55454862e-01
8.68896365e-01 -7.16845393e-01 -8.10366452e-01 4.92037773e-01
-9.02110100e-01 -7.64826000e-01 -7.90745020e-02 6.43698499e-02
-2.62877584e-01 2.53360361e-01 -5.94285727e-01 -5.23996770e-01
-3.65031421e-01 -1.20947444e+00 1.05887663e+00 5.88839769e-01
-7.37887323e-02 -5.38665533e-01 -4.96721029e-01 -6.33658245e-02
5.06503344e-01 2.17545591e-02 9.55131352e-01 3.33352387e-01
-9.78262782e-01 -6.01305850e-02 -9.02654588e-01 9.13395882e-02
4.80167747e-01 2.48141319e-01 -8.65849197e-01 -1.51112065e-01
-7.94275165e-01 -3.33643518e-02 8.51749599e-01 4.09910172e-01
1.50932539e+00 -2.55553871e-01 -1.22241184e-01 8.08957756e-01
1.55224979e+00 3.65218639e-01 7.00078011e-01 3.17625761e-01
8.57929647e-01 9.70532969e-02 3.94719899e-01 4.84423637e-01
2.18458250e-01 4.25416231e-01 3.80041212e-01 -8.08332622e-01
-5.72084248e-01 -4.83055294e-01 3.23534571e-02 1.92096144e-01
5.90313673e-02 -3.74876428e-03 -6.22529685e-01 5.16127288e-01
-1.42344129e+00 -6.71878755e-01 2.03816220e-02 2.30935550e+00
9.98928428e-01 -1.53744102e-01 -6.56511709e-02 -1.18723854e-01
8.55893254e-01 9.69393030e-02 -7.19331086e-01 -3.02726805e-01
7.90894553e-02 1.74509332e-01 9.56337750e-01 7.12057352e-01
-7.62014568e-01 1.12266636e+00 6.18025541e+00 6.43995941e-01
-1.50203991e+00 -2.86369417e-02 9.40000057e-01 -3.69245321e-01
-7.02321172e-01 -2.42250431e-02 -3.71254891e-01 6.71001911e-01
2.93748021e-01 8.21212605e-02 7.77403355e-01 6.78168416e-01
5.27825713e-01 -4.19841558e-01 -7.00892031e-01 1.10662341e+00
-1.39493823e-01 -1.37369251e+00 9.18727964e-02 -1.72483236e-01
7.43958354e-01 -1.69215590e-01 6.93412125e-01 -2.28909314e-01
5.80442727e-01 -5.72693467e-01 9.50180471e-01 4.10585076e-01
1.27583265e+00 -7.79160440e-01 -1.11275777e-01 -4.58404958e-01
-1.20972824e+00 5.30313924e-02 -2.96266109e-01 3.59478652e-01
1.24626502e-01 5.74870706e-01 -7.53155112e-01 -6.12334199e-02
8.86533141e-01 3.22697163e-01 -8.08245063e-01 1.06516898e+00
-4.37065244e-01 5.13991058e-01 -3.01027507e-01 -1.02643318e-01
2.61193246e-01 -1.85414135e-01 -8.24934244e-03 1.16493440e+00
3.91277164e-01 -2.14939360e-02 -4.52419609e-01 1.16941774e+00
-5.09665050e-02 -1.56042174e-01 -8.02562684e-02 7.94240311e-02
6.64166331e-01 1.29686606e+00 -9.63263929e-01 -4.17753071e-01
-3.88870746e-01 1.63783944e+00 2.39687383e-01 9.56476271e-01
-1.11092246e+00 -9.12114859e-01 6.39821887e-01 1.11634858e-01
3.76871854e-01 -3.94450456e-01 -6.74112618e-01 -9.57404315e-01
-1.91479191e-01 -7.03905642e-01 1.86034031e-02 -1.31168723e+00
-6.91518903e-01 5.63772798e-01 -1.97816923e-01 -1.31318557e+00
2.95331568e-01 -4.81089383e-01 -6.61450148e-01 1.01974630e+00
-1.48680580e+00 -1.05497527e+00 -6.85766220e-01 7.74411976e-01
3.64824295e-01 3.12653720e-01 4.13860589e-01 3.81353676e-01
-5.61032295e-01 8.65327895e-01 1.40539303e-01 2.18019560e-02
1.00680292e+00 -1.41872048e+00 7.25260913e-01 1.11102915e+00
9.80174541e-03 7.90435851e-01 5.83988190e-01 -4.63580757e-01
-1.49767911e+00 -9.70838189e-01 3.98964792e-01 -1.18132783e-02
3.08344215e-01 -4.65044469e-01 -8.23796093e-01 4.33725864e-01
4.31090444e-01 -2.18595117e-02 3.75612527e-01 -1.87341198e-01
-5.64174116e-01 -3.99225920e-01 -9.58334386e-01 1.22610843e+00
6.42429590e-01 -5.98728836e-01 1.03755355e-01 -2.75776554e-02
5.41399837e-01 -2.67911971e-01 -2.46248007e-01 -3.37762326e-01
6.15641832e-01 -1.02066135e+00 1.01645422e+00 1.76588818e-01
4.27322388e-01 -7.58248448e-01 1.70923799e-01 -1.31327462e+00
-5.23070931e-01 -6.51607513e-01 3.92343521e-01 1.03956664e+00
5.84571064e-01 -6.91638649e-01 9.47678983e-01 1.06366527e+00
-2.56796386e-02 -3.66999000e-01 -5.11573732e-01 -2.11432323e-01
-1.79291740e-01 -3.16841006e-01 7.61151969e-01 5.91295600e-01
9.19278637e-02 -2.48654306e-01 -4.24403131e-01 2.99305916e-01
6.89217746e-01 3.43151808e-01 6.67742491e-01 -3.23663682e-01
-1.52823478e-01 -5.72808862e-01 -7.31383311e-03 -1.37728167e+00
-5.14731705e-01 -4.48078245e-01 1.57189161e-01 -1.56598771e+00
2.11512595e-01 -6.04895413e-01 -3.95264924e-01 7.89994240e-01
-5.23478448e-01 8.79166424e-01 4.59671021e-01 1.32216454e-01
-5.18827975e-01 2.74636984e-01 1.61489439e+00 -1.10338636e-01
-3.05817574e-01 -4.11604762e-01 -1.09661317e+00 4.31616992e-01
8.74631166e-01 -4.27100249e-02 -5.88734806e-01 -8.36893439e-01
1.32909134e-01 -1.44534066e-01 5.00717759e-01 -8.66355538e-01
2.83910066e-01 -3.45024049e-01 8.02771211e-01 -6.31579757e-01
2.74320781e-01 -7.70854771e-01 5.63876368e-02 4.24129486e-01
-3.19875270e-01 1.06915534e-01 3.80522072e-01 3.41057867e-01
-1.22469701e-02 1.01210311e-01 1.07061350e+00 -6.53096884e-02
-1.02437150e+00 2.70476937e-01 -5.57607114e-01 -2.42955208e-01
9.81885195e-01 -4.00669754e-01 -4.22208697e-01 -4.52138841e-01
-3.32926184e-01 2.48237513e-02 8.92528772e-01 2.42850333e-01
8.29311132e-01 -1.33140695e+00 -2.42910713e-01 4.80759948e-01
7.36579448e-02 -2.70525068e-01 6.34459794e-01 5.47366619e-01
-1.06165552e+00 2.21162260e-01 -3.89098287e-01 -3.45584333e-01
-1.21442342e+00 5.29661536e-01 3.20348144e-01 3.99180353e-01
-8.23751271e-01 1.00653028e+00 7.28735805e-01 1.45737290e-01
1.03753641e-01 -6.52087808e-01 2.03371525e-01 -3.05369109e-01
7.05014825e-01 1.36020407e-01 -1.50766864e-01 -1.04833089e-01
-2.35573784e-01 6.61388099e-01 -2.43623465e-01 -4.39468205e-01
8.78247917e-01 -3.77400786e-01 5.90550229e-02 -6.52334318e-02
1.20327377e+00 2.42026731e-01 -1.93567491e+00 1.22210216e-02
-6.13485813e-01 -8.95065546e-01 1.52986079e-01 -1.02036369e+00
-1.56012559e+00 8.85902822e-01 7.67665207e-01 2.85351444e-02
1.59142685e+00 -2.61732459e-01 8.23890746e-01 -1.15646265e-01
2.58523136e-01 -9.24491167e-01 1.88378379e-01 6.91230297e-02
7.79815793e-01 -9.01678443e-01 -1.99683949e-01 -3.93898606e-01
-8.41228724e-01 1.06652451e+00 8.26925635e-01 -1.59480467e-01
3.31409633e-01 4.37213063e-01 4.18537647e-01 7.76209217e-03
-4.27885175e-01 -4.90625240e-02 1.60104379e-01 6.25641465e-01
5.66618025e-01 2.28469282e-01 2.60372534e-02 1.39338776e-01
-9.32116620e-03 -1.21476769e-01 5.64182937e-01 6.44190848e-01
-3.94643396e-01 -9.50055003e-01 -5.81931591e-01 2.62321413e-01
-2.94168919e-01 -4.34852481e-01 -3.44602764e-01 5.36258340e-01
-2.42104098e-01 8.39035869e-01 2.19585195e-01 -2.78540790e-01
1.65547505e-01 -4.15941536e-01 2.53529996e-01 -2.44588926e-01
-4.36163068e-01 2.95980245e-01 -4.07770932e-01 -8.39655340e-01
8.97997096e-02 -1.68867871e-01 -1.41568458e+00 -5.40051341e-01
3.72304320e-02 5.70769496e-02 6.79595351e-01 1.72595233e-01
5.95511436e-01 7.75855601e-01 5.99717617e-01 -1.03226900e+00
1.70794144e-01 -4.78174835e-01 -6.61056995e-01 3.99879962e-01
4.64591473e-01 -1.71390906e-01 -1.89124405e-01 2.37150177e-01]
|
[11.386664390563965, -0.9991265535354614]
|
70dfdc7f-01db-4ae3-a550-1449800c8572
|
in2i-unsupervised-multi-image-to-image
|
1711.09334
| null |
http://arxiv.org/abs/1711.09334v1
|
http://arxiv.org/pdf/1711.09334v1.pdf
|
In2I : Unsupervised Multi-Image-to-Image Translation Using Generative Adversarial Networks
|
In unsupervised image-to-image translation, the goal is to learn the mapping
between an input image and an output image using a set of unpaired training
images. In this paper, we propose an extension of the unsupervised
image-to-image translation problem to multiple input setting. Given a set of
paired images from multiple modalities, a transformation is learned to
translate the input into a specified domain. For this purpose, we introduce a
Generative Adversarial Network (GAN) based framework along with a multi-modal
generator structure and a new loss term, latent consistency loss. Through
various experiments we show that leveraging multiple inputs generally improves
the visual quality of the translated images. Moreover, we show that the
proposed method outperforms current state-of-the-art unsupervised
image-to-image translation methods.
|
['Pramuditha Perera', 'Vishal M. Patel', 'Mahdi Abavisani']
|
2017-11-26
| null | null | null | null |
['multimodal-unsupervised-image-to-image']
|
['computer-vision']
|
[ 9.91326034e-01 2.49851197e-01 -1.97971910e-01 -5.39891124e-01
-1.22677183e+00 -7.33382761e-01 8.27859700e-01 -4.99899626e-01
-2.59438038e-01 7.41001844e-01 6.42236918e-02 1.93685293e-02
3.60852063e-01 -7.36588836e-01 -1.12406445e+00 -8.29424083e-01
7.15898395e-01 4.71894443e-01 -2.50635922e-01 1.78845376e-01
-7.46626407e-02 1.49650156e-01 -9.38685536e-01 3.41546208e-01
7.99702942e-01 7.35150933e-01 3.41686904e-02 5.29060185e-01
1.38241976e-01 6.38425946e-01 -4.15614426e-01 -5.65845966e-01
4.21288967e-01 -1.15664709e+00 -8.23452592e-01 5.85611165e-01
6.51970267e-01 -3.77278417e-01 -1.48944303e-01 1.17109656e+00
3.81572515e-01 4.01647501e-02 8.53422761e-01 -1.52778411e+00
-1.16421771e+00 3.41343939e-01 -5.47861814e-01 -3.61977398e-01
2.91507423e-01 1.82644248e-01 6.65183544e-01 -9.82000709e-01
9.65986073e-01 1.08905673e+00 1.85343057e-01 6.94229424e-01
-1.72915339e+00 -4.69340712e-01 -2.59963214e-01 -2.86528289e-01
-1.26279724e+00 -5.52088559e-01 9.36742783e-01 -4.88070637e-01
3.52620304e-01 1.64576317e-03 2.92477936e-01 1.22874200e+00
6.11082017e-02 5.36801696e-01 1.50548553e+00 -8.26128125e-01
1.03794359e-01 3.32069546e-01 -8.31447005e-01 6.85295939e-01
-1.43131822e-01 3.46618146e-01 -5.66012263e-01 1.03345141e-02
9.92951572e-01 1.42739471e-02 -3.86836268e-02 -5.54696858e-01
-1.28139853e+00 8.46794426e-01 6.06958389e-01 2.11343676e-01
-2.92727053e-01 2.13437319e-01 7.24059492e-02 4.26843733e-01
4.45442975e-01 2.87693769e-01 5.55837750e-02 2.63743013e-01
-9.91918743e-01 1.79484971e-02 3.18587571e-01 9.43766415e-01
8.62542510e-01 2.10978344e-01 -3.04500043e-01 5.82703114e-01
2.74324179e-01 8.76753807e-01 3.53660166e-01 -1.10828698e+00
5.64349473e-01 3.38175267e-01 8.26899782e-02 -8.04558158e-01
3.53494912e-01 -1.78158313e-01 -9.69758630e-01 3.83682907e-01
1.27814069e-01 3.20619680e-02 -1.09629595e+00 2.17936397e+00
2.00556681e-01 8.70321244e-02 2.66034424e-01 8.24102402e-01
5.39036751e-01 6.45176291e-01 2.27773599e-02 -2.25760028e-01
8.84733796e-01 -1.20601249e+00 -7.74322689e-01 -3.15219790e-01
-3.72779854e-02 -8.97516429e-01 1.20597529e+00 -4.34176251e-02
-1.45111215e+00 -6.95455372e-01 -9.62580502e-01 -1.29341543e-01
-2.03415543e-01 2.40977913e-01 6.69696480e-02 4.84831631e-01
-1.02389073e+00 1.25280216e-01 -6.23094857e-01 -3.74322712e-01
4.14389998e-01 2.82703042e-01 -7.40824342e-01 -3.75274479e-01
-8.72681618e-01 8.00406337e-01 3.61972004e-01 -2.65174925e-01
-1.22856116e+00 -4.39086199e-01 -9.71819043e-01 -1.83432668e-01
7.96756968e-02 -1.19546580e+00 1.05926728e+00 -1.66811717e+00
-1.79899406e+00 1.32343078e+00 -1.94787428e-01 -2.63446182e-01
6.31622672e-01 1.05207548e-01 -2.01574519e-01 4.16419119e-01
3.15596074e-01 1.00530624e+00 1.46058357e+00 -1.78716755e+00
-2.10425049e-01 -1.57357052e-01 -1.44767091e-01 1.69907913e-01
-2.07325548e-01 4.89027128e-02 -5.68364143e-01 -8.75979722e-01
-1.14499018e-01 -1.08162713e+00 -3.64039242e-02 7.91667551e-02
-6.31412148e-01 5.52440763e-01 7.81712174e-01 -6.76434517e-01
4.53196913e-01 -2.19194865e+00 7.27004349e-01 2.12826893e-01
6.24013767e-02 -8.70734230e-02 -4.76566613e-01 4.44524765e-01
-7.98519403e-02 -2.51179039e-02 -6.48731709e-01 -7.99185097e-01
-8.16240162e-02 4.23543453e-01 -3.81607771e-01 5.02487302e-01
3.36745411e-01 1.27061117e+00 -8.85473907e-01 -6.13410115e-01
1.91404015e-01 6.97465599e-01 -4.72144902e-01 5.76532960e-01
-9.79537815e-02 1.19646573e+00 -1.79090112e-01 4.87115204e-01
6.80038452e-01 -2.60094941e-01 1.69301573e-02 -1.42414927e-01
4.45977747e-01 -1.98221847e-01 -5.41749299e-01 2.04290962e+00
-5.81072509e-01 5.10424435e-01 -1.47005588e-01 -1.05006194e+00
7.88892269e-01 4.94721085e-01 3.51854593e-01 -6.78507328e-01
2.11681291e-01 1.32152990e-01 -1.77438617e-01 -2.00489596e-01
1.49845541e-01 -6.20614529e-01 -2.47196957e-01 6.61861420e-01
3.82768333e-01 -4.59762633e-01 1.10339578e-02 2.96625793e-01
7.41413176e-01 3.44725966e-01 -1.98801253e-02 1.87009141e-01
4.54787821e-01 -1.67973638e-01 3.15260410e-01 6.17176175e-01
9.49684903e-02 1.05524766e+00 2.65397906e-01 4.64502256e-03
-1.51861918e+00 -1.42335844e+00 1.77713543e-01 7.05200553e-01
-1.21546872e-02 8.04014355e-02 -1.00188017e+00 -7.53381073e-01
-3.10240120e-01 4.98399168e-01 -6.61604106e-01 -2.40685940e-01
-3.42604965e-01 -2.05119133e-01 5.81166685e-01 3.96438658e-01
6.86150849e-01 -1.02693868e+00 -1.57142729e-01 -4.32645231e-02
-4.63899612e-01 -1.51224029e+00 -8.87274265e-01 -1.80919260e-01
-8.51923645e-01 -6.99240386e-01 -9.84007835e-01 -1.16622007e+00
1.46852517e+00 6.42914623e-02 1.15687156e+00 -2.07080081e-01
9.91699100e-03 5.63127398e-01 -2.52019912e-01 -8.01165700e-02
-8.88117433e-01 -1.12421356e-01 8.28024074e-02 5.13609767e-01
-3.50481421e-01 -7.04481900e-01 -5.18027604e-01 2.99969614e-01
-1.52840734e+00 5.26247919e-01 7.24650741e-01 1.13534510e+00
9.75478530e-01 -1.73520267e-01 3.96223217e-01 -9.90884185e-01
4.38159943e-01 -1.33978963e-01 -4.37368333e-01 3.54434758e-01
-5.11195779e-01 2.29255125e-01 7.27093339e-01 -6.42507195e-01
-1.16400111e+00 4.39521968e-01 3.29755157e-01 -7.25042641e-01
-7.22867856e-03 4.04671013e-01 -4.51469243e-01 -2.95350790e-01
5.67786038e-01 4.88113016e-01 1.92834184e-01 -1.51160076e-01
9.05449092e-01 3.86327177e-01 1.03348899e+00 -6.62962317e-01
1.35547888e+00 5.73455691e-01 -3.82186286e-02 -1.72190666e-01
-6.25754476e-01 -2.38654837e-02 -8.43936324e-01 -8.39137584e-02
9.89939690e-01 -9.46356893e-01 -1.07172504e-01 5.01938820e-01
-1.14082289e+00 -5.39579153e-01 -3.54552120e-01 3.30723912e-01
-1.07608247e+00 3.73981008e-03 -4.17715162e-01 -3.19449216e-01
-3.15983146e-01 -1.24058282e+00 1.33964849e+00 1.64308414e-01
1.20043062e-01 -1.06383622e+00 2.26044804e-01 5.77347398e-01
4.09900278e-01 5.12305260e-01 7.34818101e-01 -1.85941532e-01
-7.08051026e-01 -2.19203308e-01 -1.58908784e-01 7.28821158e-01
4.44711030e-01 -2.58469731e-01 -5.95438421e-01 -4.70750630e-01
6.68110047e-03 -6.15395427e-01 6.85186327e-01 2.90657818e-01
8.87680650e-01 -4.90375519e-01 -1.47371337e-01 7.86676586e-01
1.57109630e+00 -8.16115760e-04 9.07578468e-01 1.04936361e-01
7.73534119e-01 2.82491624e-01 3.96885037e-01 -5.51075339e-02
9.87917632e-02 8.27573419e-01 1.98424876e-01 -5.71061373e-01
-3.11671972e-01 -6.74563229e-01 5.05936146e-01 6.55855119e-01
2.68249866e-02 -3.12981069e-01 -4.55474555e-01 5.70717931e-01
-1.65818238e+00 -8.70755970e-01 5.20493925e-01 2.26802135e+00
1.03068066e+00 -1.99405864e-01 -7.94035010e-03 -2.51760721e-01
8.11304867e-01 -6.10572472e-02 -5.99870086e-01 -2.19825745e-01
-2.28593111e-01 3.97861004e-01 3.70785773e-01 6.24807715e-01
-9.41640735e-01 1.05456126e+00 6.63640070e+00 5.98423839e-01
-1.26402915e+00 3.82986486e-01 8.12456906e-01 9.93132815e-02
-5.07638574e-01 6.05792515e-02 -1.49832457e-01 3.30646574e-01
7.19907045e-01 -2.15600014e-01 6.57734990e-01 4.84299272e-01
1.65388227e-01 1.42778486e-01 -1.17166233e+00 9.13382053e-01
4.51498121e-01 -1.19111943e+00 3.08818996e-01 1.04721628e-01
1.32023084e+00 -3.06535542e-01 5.19831777e-01 -1.82906821e-01
2.76930571e-01 -1.24274945e+00 7.74008751e-01 5.37958264e-01
1.29034448e+00 -6.05549276e-01 4.30571854e-01 7.25944946e-03
-7.80029178e-01 4.92978960e-01 2.12842021e-02 3.65930915e-01
4.00982380e-01 3.01784605e-01 -6.09403968e-01 6.32229805e-01
3.21386129e-01 4.67827648e-01 -4.71561342e-01 5.79606831e-01
-7.42557406e-01 4.43361998e-01 -7.11052865e-02 7.73046315e-01
1.33740813e-01 -3.77193749e-01 4.15831029e-01 7.78946161e-01
4.03729260e-01 -1.50225535e-01 1.97812274e-01 1.27692986e+00
-5.10346293e-01 1.77822895e-02 -9.96050954e-01 -8.96299109e-02
1.93244129e-01 1.18414974e+00 -5.97930789e-01 -4.51424688e-01
-3.40003043e-01 1.79722714e+00 1.04991831e-01 6.05058014e-01
-8.55941176e-01 -1.38638362e-01 1.41782671e-01 -5.48357852e-02
2.95237005e-01 -4.97699007e-02 -1.84329614e-01 -1.27123857e+00
1.68202132e-01 -9.82913375e-01 1.54854106e-02 -1.16142535e+00
-1.27582943e+00 6.28206313e-01 -1.82508864e-02 -1.48429894e+00
-5.54635346e-01 -1.19550750e-01 -6.30021691e-01 9.62292075e-01
-1.32184255e+00 -1.80620575e+00 -2.73550957e-01 7.52076864e-01
2.99982727e-01 -3.15508604e-01 8.88614297e-01 1.67541236e-01
-9.49170962e-02 8.19973946e-01 3.61384213e-01 3.72061312e-01
9.70175445e-01 -1.11227429e+00 3.22082520e-01 1.08633053e+00
4.93690968e-01 4.33315128e-01 6.37953460e-01 -4.61640835e-01
-1.35406685e+00 -1.25189149e+00 5.85465789e-01 -4.00803864e-01
2.71081388e-01 -2.56945550e-01 -5.71717858e-01 9.99457181e-01
7.91340172e-01 3.82286787e-01 6.74777687e-01 -6.60768688e-01
-4.66117084e-01 -2.11663637e-02 -1.43785059e+00 5.31363606e-01
7.24850237e-01 -8.60514998e-01 -2.93126196e-01 2.68541306e-01
6.81503773e-01 -5.52353501e-01 -9.33853626e-01 2.25358725e-01
3.99925768e-01 -5.50227225e-01 1.15427756e+00 -6.18589282e-01
8.86175513e-01 -4.35361385e-01 -2.36357853e-01 -1.52277517e+00
-6.97920471e-02 -7.09312260e-01 2.28384659e-01 1.32536340e+00
4.34103400e-01 -4.52351570e-01 5.48309803e-01 4.57696974e-01
1.15577556e-01 -2.81254292e-01 -8.13899636e-01 -7.51031518e-01
2.55014658e-01 8.74842182e-02 2.61206865e-01 8.54327857e-01
-3.37482750e-01 5.38434327e-01 -8.89109612e-01 1.32296175e-01
9.47856307e-01 2.91799366e-01 9.85263169e-01 -4.24502671e-01
-5.64330220e-01 -1.08405866e-01 -3.69040877e-01 -8.34606290e-01
3.09092075e-01 -9.71965194e-01 1.86326414e-01 -1.31398606e+00
5.85522950e-01 2.30531935e-02 -2.54790127e-01 5.46494126e-01
-1.12810373e-01 9.50979233e-01 3.91700357e-01 3.19010377e-01
-4.30904210e-01 7.64326513e-01 1.54059482e+00 -4.70765740e-01
2.64464016e-03 -1.76409513e-01 -5.97119331e-01 2.09487721e-01
7.45895565e-01 -7.63425052e-01 -4.98074621e-01 -6.00830019e-01
-7.09346086e-02 1.97207049e-01 4.90814060e-01 -6.54376030e-01
-6.43202290e-02 -2.31483892e-01 4.15152848e-01 -9.73652974e-02
3.31713736e-01 -7.49992728e-01 4.26268667e-01 3.10030073e-01
-5.79900026e-01 3.44589655e-03 1.07672270e-02 5.04565597e-01
-5.74218154e-01 2.22160872e-02 9.71273422e-01 -7.07923993e-02
-1.98215425e-01 3.62395823e-01 2.30435841e-02 4.06735539e-02
9.30208385e-01 9.84184742e-02 -2.07004309e-01 -6.53455019e-01
-7.65239418e-01 -2.23638684e-01 8.71641517e-01 4.35023695e-01
7.80536950e-01 -1.95038092e+00 -8.35790277e-01 3.80377740e-01
3.60043466e-01 -1.67892709e-01 -7.44307190e-02 8.74847114e-01
-3.33228350e-01 1.08848311e-01 -4.93515104e-01 -6.32422924e-01
-1.22820282e+00 4.98590469e-01 2.00325489e-01 -2.50842482e-01
-4.57967669e-01 5.37990987e-01 3.88096452e-01 -6.31527126e-01
-1.04658388e-01 1.35912849e-02 4.60805237e-01 -5.61577260e-01
2.84070671e-01 -2.96470672e-01 -2.40676478e-01 -1.05100572e+00
-4.31544371e-02 7.08952069e-01 1.69240385e-01 -7.62575746e-01
1.15355945e+00 -2.44148210e-01 -2.00265840e-01 2.66636550e-01
1.46172023e+00 -1.29243433e-02 -1.32062364e+00 -4.48497713e-01
-6.24915600e-01 -7.06216037e-01 -6.69562221e-02 -8.59389663e-01
-1.30057549e+00 7.06829250e-01 8.00976396e-01 -2.15886444e-01
1.44039774e+00 7.13644102e-02 7.60699034e-01 9.17365868e-03
2.72608340e-01 -6.75911129e-01 4.47744846e-01 7.76877254e-02
9.25802052e-01 -1.51665843e+00 -1.87685296e-01 -1.62478715e-01
-7.99518704e-01 7.04789817e-01 3.60184282e-01 -9.00899246e-02
2.69408166e-01 1.18473314e-01 2.74589151e-01 5.90954535e-02
-4.08466488e-01 -1.37659267e-01 4.55208957e-01 8.43019366e-01
1.83590680e-01 2.67579798e-02 -5.41427620e-02 7.10450262e-02
3.15535143e-02 2.17380047e-01 2.87494123e-01 7.88103104e-01
1.81429595e-01 -1.75527990e+00 -5.28193295e-01 -7.34307393e-02
-5.02882838e-01 -1.63346127e-01 -5.02635360e-01 4.89956707e-01
4.50729132e-02 7.82910705e-01 -8.17576200e-02 -2.66675770e-01
5.89703023e-02 5.55626266e-02 9.76071000e-01 -6.08460784e-01
-2.68827617e-01 1.31423026e-01 -4.70396459e-01 -4.08805668e-01
-9.49283242e-01 -4.88196611e-01 -9.79116321e-01 -4.50359471e-02
-1.21202268e-01 -7.76316077e-02 7.10670948e-01 9.00282383e-01
4.37893003e-01 4.74280179e-01 8.91837120e-01 -9.47358489e-01
-2.68379599e-01 -8.75862122e-01 -3.36206198e-01 9.58316505e-01
3.01647276e-01 -4.21810627e-01 -1.66815102e-01 9.03380990e-01]
|
[11.66956901550293, -0.3979678750038147]
|
72238fb8-b280-430a-99e5-d64bab8f9c0e
|
relative-positional-encoding-for-transformers
|
2105.08399
| null |
https://arxiv.org/abs/2105.08399v2
|
https://arxiv.org/pdf/2105.08399v2.pdf
|
Relative Positional Encoding for Transformers with Linear Complexity
|
Recent advances in Transformer models allow for unprecedented sequence lengths, due to linear space and time complexity. In the meantime, relative positional encoding (RPE) was proposed as beneficial for classical Transformers and consists in exploiting lags instead of absolute positions for inference. Still, RPE is not available for the recent linear-variants of the Transformer, because it requires the explicit computation of the attention matrix, which is precisely what is avoided by such methods. In this paper, we bridge this gap and present Stochastic Positional Encoding as a way to generate PE that can be used as a replacement to the classical additive (sinusoidal) PE and provably behaves like RPE. The main theoretical contribution is to make a connection between positional encoding and cross-covariance structures of correlated Gaussian processes. We illustrate the performance of our approach on the Long-Range Arena benchmark and on music generation.
|
['Gaël Richard', 'Yi-Hsuan Yang', 'Umut Şimşekli', 'Shih-Lun Wu', 'Ondřej Cífka', 'Antoine Liutkus']
|
2021-05-18
| null | null | null | null |
['music-generation', 'music-generation']
|
['audio', 'music']
|
[ 3.12517315e-01 7.17456937e-02 3.40155542e-01 2.55370587e-01
-6.82260871e-01 -7.13355124e-01 7.85597324e-01 1.78888872e-01
-3.69425684e-01 9.32380795e-01 2.02026054e-01 -3.14720511e-01
-6.55904055e-01 -7.21543431e-01 -7.24542618e-01 -8.67399931e-01
-2.53394842e-01 4.87718016e-01 1.39381364e-01 -3.09738606e-01
2.21304774e-01 2.97405005e-01 -1.64371407e+00 -9.39180478e-02
7.71852911e-01 8.32701564e-01 3.14272791e-01 8.29331338e-01
2.51047760e-02 5.81141651e-01 -4.97176617e-01 -7.61482716e-01
2.61843413e-01 -7.70701766e-01 -5.05442500e-01 -1.66934431e-01
1.30843088e-01 1.02306075e-01 -1.01457946e-01 9.15442884e-01
5.80619931e-01 1.19866177e-01 7.73882508e-01 -1.08528042e+00
-6.03672206e-01 9.76463616e-01 -3.18865210e-01 7.18043223e-02
2.75000483e-01 -4.81753647e-02 1.29730749e+00 -5.76229393e-01
4.12144065e-01 9.62346315e-01 9.64229822e-01 2.01350152e-01
-1.56830907e+00 -2.58161515e-01 -1.06428184e-01 3.59414935e-01
-1.36404371e+00 -2.28969082e-01 6.19651020e-01 -4.62647110e-01
7.47226357e-01 4.05440032e-01 8.33707809e-01 1.45766890e+00
8.51740986e-02 7.41148293e-01 1.15381944e+00 -5.02283156e-01
3.66443276e-01 -1.75268680e-01 1.09719120e-01 2.87857205e-01
1.69377476e-01 4.81553614e-01 -6.15479410e-01 -2.26592347e-01
6.32670224e-01 -4.10783172e-01 -4.17402029e-01 -4.98157054e-01
-1.14616966e+00 6.08071566e-01 8.98005515e-02 4.07472819e-01
-3.42354059e-01 3.76844078e-01 1.98277175e-01 3.96239698e-01
3.38097394e-01 6.29926801e-01 -3.05359900e-01 -7.65842915e-01
-1.03073871e+00 5.44614971e-01 9.98814583e-01 7.48212755e-01
3.98466974e-01 2.77341902e-01 -4.21656877e-01 5.26463687e-01
2.74101291e-02 6.07562244e-01 3.55870694e-01 -9.47255194e-01
1.93484560e-01 1.49042467e-02 1.33667395e-01 -9.99993384e-01
-3.07572752e-01 -1.01111841e+00 -1.08448994e+00 7.88956732e-02
8.16128790e-01 -4.41753604e-02 -3.29548210e-01 1.92820573e+00
2.12345235e-02 6.20308638e-01 -1.41208116e-02 5.64943671e-01
-1.10098504e-01 5.20646632e-01 -3.40330899e-01 -2.77754158e-01
1.18448257e+00 -6.16176665e-01 -6.76909149e-01 6.41337484e-02
3.81313145e-01 -7.21033633e-01 1.08454823e+00 7.23269999e-01
-1.23587990e+00 -3.72984856e-01 -9.58382368e-01 1.70792326e-01
-1.15548976e-01 1.03838870e-03 5.48542917e-01 9.82363343e-01
-1.20486724e+00 1.03488016e+00 -5.45688868e-01 6.25838665e-03
8.68193656e-02 8.05233344e-02 -5.25072552e-02 4.11863744e-01
-1.32520282e+00 5.83557010e-01 9.97626483e-02 3.88704240e-01
-4.45212364e-01 -8.66382897e-01 -5.53592920e-01 4.54997927e-01
4.89590466e-01 -8.69342566e-01 1.19259715e+00 -7.32261360e-01
-1.96071076e+00 4.18891400e-01 -2.61468943e-02 -9.20750856e-01
8.17009628e-01 -4.06520844e-01 1.46196023e-01 -2.03919441e-01
-3.60855669e-01 1.22129999e-01 9.75272059e-01 -8.49574983e-01
-2.40584746e-01 -1.77369878e-01 -3.37440111e-02 -9.15165097e-02
-7.02059502e-03 -3.32770318e-01 -1.67769402e-01 -9.18217838e-01
-2.03087687e-01 -1.24582839e+00 -2.76903480e-01 -5.32850623e-01
-4.43925470e-01 -1.48640508e-02 8.62780288e-02 -5.20637214e-01
1.30375898e+00 -2.06238079e+00 5.47702134e-01 2.78276116e-01
-6.35107458e-02 2.24990889e-01 -1.35544434e-01 7.06118703e-01
-2.10808560e-01 1.81497801e-02 -4.36668336e-01 -4.62418854e-01
4.83598858e-01 1.41277447e-01 -8.02328229e-01 3.23392272e-01
3.25982682e-02 9.93812561e-01 -8.97893131e-01 3.80310742e-03
2.94943452e-02 5.70314944e-01 -7.05014229e-01 -1.18928418e-01
-3.21958393e-01 5.18788934e-01 -8.10382441e-02 -1.02067299e-01
4.65482563e-01 -8.25939998e-02 9.96989384e-02 7.77662471e-02
-2.00005591e-01 4.51791614e-01 -1.14946902e+00 1.67555583e+00
-4.46888685e-01 4.81715113e-01 -2.81146765e-01 -7.77744353e-01
8.39489043e-01 2.91649342e-01 3.67791712e-01 -5.61361849e-01
8.72890726e-02 3.40208977e-01 -5.52826468e-03 3.56701836e-02
5.90475082e-01 6.94761705e-03 -7.52383992e-02 3.88437063e-01
-2.64047384e-01 -7.03045493e-03 4.64734524e-01 -4.40445989e-02
1.14077306e+00 4.92895186e-01 2.48699352e-01 -2.89768338e-01
5.99956691e-01 -4.97838825e-01 3.28092903e-01 9.91354525e-01
4.60701257e-01 6.38911009e-01 1.00468349e+00 -5.25001623e-02
-1.03854954e+00 -1.22533584e+00 6.10281490e-02 6.29866660e-01
-3.51358801e-01 -6.86005890e-01 -6.75178051e-01 -2.76452273e-01
-2.49888584e-01 8.05203319e-01 -6.62272394e-01 -1.78990439e-01
-5.67879379e-01 -6.62669480e-01 6.98315620e-01 4.17353630e-01
6.97324723e-02 -7.47180641e-01 -7.89249182e-01 2.23757014e-01
-1.03615582e-01 -8.73301089e-01 -2.37317741e-01 2.67482072e-01
-7.49718189e-01 -7.14838564e-01 -9.61361706e-01 -3.91133465e-02
-1.60550373e-03 -2.11873621e-01 1.08022976e+00 -4.25187945e-01
-4.82048132e-02 1.90915719e-01 -3.68431866e-01 -3.18948150e-01
-3.14169794e-01 3.26793432e-01 -8.82983208e-04 9.57467929e-02
-1.95873573e-01 -1.09572935e+00 -4.67805356e-01 -1.35640232e-02
-8.91541898e-01 6.42082468e-02 7.61720896e-01 1.02301550e+00
3.84048283e-01 -1.16742641e-01 4.05412287e-01 -8.10884893e-01
7.44226694e-01 -2.57869393e-01 -8.90629768e-01 9.26857218e-02
-6.13638997e-01 5.89832008e-01 6.50645494e-01 -3.72605592e-01
-7.82119095e-01 -1.01561464e-01 -4.42265004e-01 -4.39699680e-01
2.25324973e-01 5.87557018e-01 -3.94207500e-02 2.57942349e-01
3.44781160e-01 5.00601172e-01 -1.95416659e-01 -6.50194585e-01
4.68742281e-01 2.54502088e-01 5.33388019e-01 -7.24164844e-01
7.27212429e-01 2.37671554e-01 5.41908562e-01 -7.00241148e-01
-6.62013590e-01 4.42184992e-02 -4.69425023e-01 2.36441404e-01
6.27721429e-01 -4.63822126e-01 -1.01828110e+00 3.18052530e-01
-1.14812231e+00 -5.54452121e-01 -7.08846390e-01 3.87344897e-01
-9.47591603e-01 4.50663626e-01 -6.31661654e-01 -1.06232905e+00
-9.96389017e-02 -1.00954390e+00 1.01316762e+00 -2.18562275e-01
-2.37027228e-01 -9.46954191e-01 3.83648962e-01 -2.58253366e-01
6.32222414e-01 1.21380322e-01 9.43215072e-01 -4.37306941e-01
-5.62464356e-01 -8.54829792e-03 -8.42624344e-03 2.61139005e-01
-2.54501790e-01 -3.80424261e-02 -8.63865554e-01 -6.77163601e-02
1.47592500e-01 3.42745960e-01 6.93647742e-01 3.48144472e-01
1.05746531e+00 -1.38466492e-01 1.02775864e-01 6.48184061e-01
1.24787664e+00 1.29290745e-02 8.69755089e-01 8.72357860e-02
4.45350826e-01 5.71851671e-01 2.36369118e-01 5.48949897e-01
1.06066123e-01 1.16809058e+00 1.32359236e-01 4.65816081e-01
-6.14812188e-02 -4.21350718e-01 4.10843939e-01 9.71701145e-01
-2.72905558e-01 -2.38554657e-01 -8.26354265e-01 3.09614778e-01
-1.82057095e+00 -1.22625947e+00 -4.08148110e-01 2.60696363e+00
6.48899376e-01 6.87735975e-02 1.31436288e-01 3.89532655e-01
3.06104988e-01 -3.70649993e-02 -5.55631481e-02 -3.03035021e-01
-3.02129269e-01 5.96988738e-01 4.99786019e-01 6.07883930e-01
-8.14647019e-01 4.85089213e-01 6.59918594e+00 1.07168949e+00
-9.05196846e-01 1.97162136e-01 2.40259424e-01 -1.15180477e-01
-2.35472873e-01 1.00230195e-01 -6.96542025e-01 7.11595178e-01
1.17643499e+00 -2.93409973e-01 5.32872856e-01 4.99280065e-01
8.06123763e-02 -2.69257743e-02 -1.18217158e+00 1.15741050e+00
2.03021970e-02 -1.12128377e+00 -1.18409283e-01 2.90075362e-01
5.26504397e-01 -3.49337071e-01 4.55066919e-01 2.89258271e-01
1.51306108e-01 -9.85329330e-01 8.37129772e-01 8.16395283e-01
4.27271575e-01 -8.56225669e-01 7.01599598e-01 3.46424878e-01
-9.95157599e-01 -8.71686786e-02 -2.35653400e-01 -2.48495996e-01
3.90201718e-01 7.91666746e-01 -5.43464780e-01 8.58735204e-01
2.61543900e-01 5.00205159e-01 -4.09157157e-01 1.13846362e+00
-1.98315024e-01 6.96172178e-01 -3.92647386e-01 2.09407769e-02
2.30163619e-01 -6.23578668e-01 1.04813397e+00 1.02812123e+00
1.00490665e+00 -5.24591267e-01 -2.48494893e-01 8.54843616e-01
2.42354199e-01 6.94587652e-04 -5.43825090e-01 -6.81420788e-02
2.67661631e-01 8.71873677e-01 -5.55485904e-01 -8.25119298e-03
-9.28577706e-02 1.07280672e+00 1.59821928e-01 2.50784636e-01
-9.25173223e-01 -1.91331819e-01 4.76106495e-01 9.45710987e-02
6.67381406e-01 -4.14058954e-01 -9.65771824e-02 -9.23130333e-01
-7.88988471e-02 -8.30017984e-01 -2.48325653e-02 -6.59689426e-01
-1.21067917e+00 5.12551129e-01 -1.04922466e-02 -1.26040804e+00
-6.17088854e-01 -4.71609324e-01 -1.41609266e-01 9.19123411e-01
-1.21554029e+00 -7.72615135e-01 1.69208258e-01 2.98874587e-01
2.13902950e-01 1.67643711e-01 9.36459959e-01 4.25671399e-01
-2.65701473e-01 6.10915124e-01 3.42987925e-01 -4.21355695e-01
5.92021883e-01 -1.63991785e+00 5.38889170e-01 7.45635033e-01
5.66534400e-01 6.59549475e-01 1.19227707e+00 -3.27290118e-01
-1.37939811e+00 -7.47284889e-01 9.80615258e-01 -7.23705709e-01
1.01939809e+00 -4.73104328e-01 -6.97457433e-01 4.91567284e-01
2.20604882e-01 -5.18893421e-01 6.25559270e-01 2.28324577e-01
-2.50366807e-01 -1.32957622e-01 -4.54851478e-01 8.23279262e-01
1.18734360e+00 -4.20804054e-01 -4.35280800e-01 -1.37809934e-02
6.95712090e-01 -3.71643245e-01 -6.41263366e-01 1.81093052e-01
6.65767729e-01 -1.15832722e+00 1.04620898e+00 -3.79187651e-02
4.30974722e-01 -3.99084657e-01 -1.41643614e-01 -1.50913751e+00
-1.66563079e-01 -1.16104186e+00 -3.01703721e-01 1.18047071e+00
3.43790352e-01 -7.67198563e-01 4.70746130e-01 -1.23489842e-01
-1.27654329e-01 -7.45320261e-01 -1.09443760e+00 -1.15810513e+00
1.57522842e-01 -8.21217954e-01 7.00714290e-01 4.62088048e-01
-2.26039574e-01 4.57474828e-01 -8.03954005e-01 -8.13972056e-02
5.89919686e-01 1.83247685e-01 7.71809936e-01 -1.31245291e+00
-1.13000286e+00 -7.00504005e-01 -5.19372940e-01 -1.30523622e+00
-5.19748032e-03 -8.12221944e-01 -4.41007502e-02 -8.85745406e-01
-9.94323045e-02 -5.09813726e-01 -2.03178808e-01 -9.37836915e-02
-5.65201156e-02 3.01932186e-01 4.83907819e-01 -6.36830628e-02
-3.20733249e-01 6.74329340e-01 9.16038036e-01 1.87721252e-01
-6.47815317e-02 3.15344006e-01 -6.39722943e-01 4.42786723e-01
6.01387322e-01 -4.46163416e-01 -5.88944733e-01 -2.44275078e-01
1.08555722e+00 2.42974892e-01 5.12330711e-01 -1.22158849e+00
1.48394153e-01 4.49905246e-01 -7.85902664e-02 -5.82234800e-01
4.97517407e-01 -8.14086556e-01 6.98776126e-01 4.31684613e-01
-4.85039443e-01 2.12398365e-01 1.01299991e-03 8.14194441e-01
-1.23890318e-01 -2.90379465e-01 2.55154371e-01 2.52602577e-01
-6.44023642e-02 3.82806733e-02 -4.29061741e-01 -8.37603882e-02
7.49636233e-01 -4.65502813e-02 -1.34071214e-02 -4.30862695e-01
-8.59777808e-01 -2.34546393e-01 3.79410923e-01 2.28581980e-01
1.74503118e-01 -1.10626817e+00 -7.16039240e-01 -3.37317511e-02
-2.76979268e-01 -4.03311789e-01 5.09889305e-01 1.38407958e+00
-2.92398989e-01 5.45305312e-01 -6.34462293e-03 -5.93753159e-01
-1.25622368e+00 7.58065462e-01 6.88919052e-02 -8.36080551e-01
-5.97867310e-01 5.67185819e-01 1.15852274e-01 -8.34811255e-02
8.31998512e-02 -3.91618788e-01 -4.98578139e-03 8.06920454e-02
6.26905739e-01 4.42539722e-01 4.99551594e-02 -2.64400989e-01
-1.29962638e-01 5.97304761e-01 4.81291026e-01 -5.17288566e-01
1.10451031e+00 -2.95836806e-01 -1.08923584e-01 1.13518715e+00
7.78137565e-01 5.16851723e-01 -1.17331922e+00 -8.19740593e-02
1.32715449e-01 -3.07817400e-01 -3.48378748e-01 -5.84815264e-01
-6.56668186e-01 1.14771473e+00 4.84775960e-01 3.66971999e-01
1.14502192e+00 -2.72391051e-01 6.71389341e-01 3.00039411e-01
6.54729664e-01 -6.17253542e-01 -2.58259714e-01 7.64836907e-01
8.09102476e-01 -3.81396890e-01 -3.20292890e-01 -3.61686110e-01
-4.02831227e-01 9.79256749e-01 -2.38675013e-01 -1.90494180e-01
4.55615491e-01 5.10504603e-01 -5.73285758e-01 2.11766079e-01
-8.30196083e-01 -4.90672022e-01 1.51749983e-01 4.92548764e-01
4.08387125e-01 6.18084222e-02 -5.97475469e-01 6.43870592e-01
-7.10710287e-01 -1.95800830e-02 3.46500158e-01 5.32020271e-01
-1.09198920e-01 -1.51098704e+00 -2.22537205e-01 1.12877861e-01
-6.13309979e-01 -4.13961262e-01 -1.44603908e-01 6.56931281e-01
8.79101604e-02 6.83620930e-01 9.43541601e-02 -2.67602772e-01
2.49117196e-01 2.88194627e-01 7.29217350e-01 -1.81111723e-01
-7.15816021e-01 5.48538305e-02 -4.17448729e-02 -4.50490445e-01
-1.34699628e-01 -7.07219362e-01 -5.34612775e-01 -3.08580190e-01
-6.45285174e-02 4.17667538e-01 5.15644372e-01 7.85688162e-01
3.95361245e-01 8.58929515e-01 3.13795328e-01 -9.26924050e-01
-8.15444410e-01 -8.50316823e-01 -4.62998837e-01 2.14987323e-01
1.99795261e-01 -5.92499793e-01 -3.58057618e-01 -1.27897337e-01]
|
[15.608009338378906, 5.5897955894470215]
|
94c28084-fb24-4c46-b913-ac236eabfd27
|
wasserstein-cnn-learning-invariant-features
|
1708.02412
| null |
http://arxiv.org/abs/1708.02412v1
|
http://arxiv.org/pdf/1708.02412v1.pdf
|
Wasserstein CNN: Learning Invariant Features for NIR-VIS Face Recognition
|
Heterogeneous face recognition (HFR) aims to match facial images acquired
from different sensing modalities with mission-critical applications in
forensics, security and commercial sectors. However, HFR is a much more
challenging problem than traditional face recognition because of large
intra-class variations of heterogeneous face images and limited training
samples of cross-modality face image pairs. This paper proposes a novel
approach namely Wasserstein CNN (convolutional neural networks, or WCNN for
short) to learn invariant features between near-infrared and visual face images
(i.e. NIR-VIS face recognition). The low-level layers of WCNN are trained with
widely available face images in visual spectrum. The high-level layer is
divided into three parts, i.e., NIR layer, VIS layer and NIR-VIS shared layer.
The first two layers aims to learn modality-specific features and NIR-VIS
shared layer is designed to learn modality-invariant feature subspace.
Wasserstein distance is introduced into NIR-VIS shared layer to measure the
dissimilarity between heterogeneous feature distributions. So W-CNN learning
aims to achieve the minimization of Wasserstein distance between NIR
distribution and VIS distribution for invariant deep feature representation of
heterogeneous face images. To avoid the over-fitting problem on small-scale
heterogeneous face data, a correlation prior is introduced on the
fully-connected layers of WCNN network to reduce parameter space. This prior is
implemented by a low-rank constraint in an end-to-end network. The joint
formulation leads to an alternating minimization for deep feature
representation at training stage and an efficient computation for heterogeneous
data at testing stage. Extensive experiments on three challenging NIR-VIS face
recognition databases demonstrate the significant superiority of Wasserstein
CNN over state-of-the-art methods.
|
['Xiang Wu', 'Zhenan Sun', 'Tieniu Tan', 'Ran He']
|
2017-08-08
| null | null | null | null |
['heterogeneous-face-recognition']
|
['computer-vision']
|
[ 1.68918625e-01 -2.41316780e-01 1.02122188e-01 -6.38005555e-01
-7.69286752e-01 -3.49739902e-02 2.40952507e-01 -7.97521412e-01
-2.34249592e-01 3.50519270e-01 8.08674395e-02 1.19042955e-01
-4.16115582e-01 -7.75922775e-01 -4.79665786e-01 -1.13723147e+00
1.77084118e-01 7.88533986e-02 -5.41428745e-01 -1.19635716e-01
-1.98371366e-01 1.05040121e+00 -1.69587231e+00 3.72707337e-01
2.77871966e-01 1.65138793e+00 -7.60116708e-03 2.51015455e-01
6.83020847e-03 6.53237164e-01 -1.19427942e-01 -2.49007523e-01
6.63231254e-01 -5.68755746e-01 -3.49265218e-01 2.19656676e-01
7.08308458e-01 -2.60389268e-01 -6.78347886e-01 1.15684497e+00
9.52155113e-01 3.23837578e-01 6.53578639e-01 -1.41165829e+00
-1.07579482e+00 6.90848678e-02 -9.21685755e-01 2.51226556e-02
3.53631191e-02 1.68088853e-01 5.12972176e-01 -1.15075004e+00
5.16330063e-01 1.35407412e+00 7.14032590e-01 7.46122718e-01
-9.45351899e-01 -9.19811189e-01 -1.82221919e-01 3.83588493e-01
-1.75381219e+00 -6.84301615e-01 1.14937425e+00 -3.20757538e-01
6.78555846e-01 3.43812883e-01 4.48762119e-01 9.62088764e-01
1.01635680e-01 2.67249316e-01 1.06963825e+00 -8.76794755e-02
-1.13154195e-01 -1.83555737e-01 -1.80399716e-01 9.47386146e-01
3.30837779e-02 2.38568708e-01 -7.64960647e-01 -1.47147343e-01
6.39895976e-01 4.93644089e-01 -2.90337235e-01 -2.78534472e-01
-7.87880540e-01 7.05984116e-01 5.57786286e-01 2.42691517e-01
-5.61545253e-01 -1.39911890e-01 3.34749997e-01 4.31259513e-01
5.00503600e-01 -3.69338363e-01 -3.23397458e-01 5.36449671e-01
-7.94334352e-01 -2.01186940e-01 2.98632532e-01 6.95766389e-01
9.19688165e-01 3.05156976e-01 -3.76172632e-01 1.34183288e+00
7.16223538e-01 8.25452089e-01 3.73159945e-01 -6.67636693e-01
4.28336650e-01 3.54732811e-01 -4.53610569e-01 -1.11077023e+00
-3.90984893e-01 -4.32692349e-01 -1.25749457e+00 2.67336428e-01
1.64859101e-01 -1.02178551e-01 -8.76909077e-01 1.65965223e+00
5.25252879e-01 2.54695714e-01 1.32050321e-01 1.12468290e+00
1.14585125e+00 4.80379820e-01 -1.67115927e-01 -2.14170322e-01
1.18304396e+00 -6.71100676e-01 -5.45960963e-01 9.81468782e-02
1.13705933e-01 -8.88729393e-01 6.68133974e-01 -1.17533498e-01
-8.77442479e-01 -6.70045197e-01 -8.74341786e-01 -1.10569835e-01
-3.07820290e-01 2.90333778e-01 2.27815866e-01 5.40180802e-01
-9.52495337e-01 4.24475104e-01 -4.01562482e-01 -2.68165231e-01
9.17244136e-01 5.26875019e-01 -9.09055889e-01 -5.37660718e-01
-9.81398821e-01 5.79323530e-01 9.59559456e-02 7.43706703e-01
-1.08237016e+00 -8.16766500e-01 -1.07275569e+00 6.22086599e-03
1.34824887e-01 -5.20085514e-01 4.11795288e-01 -1.15750635e+00
-1.60220885e+00 1.11256719e+00 -1.73702519e-02 4.16003793e-01
3.93445998e-01 3.75615090e-01 -6.28757536e-01 5.90012819e-02
-1.70566380e-01 1.82344019e-01 1.19217134e+00 -1.22634184e+00
-5.83352223e-02 -9.59967971e-01 -4.60922956e-01 2.81365253e-02
-5.06220877e-01 2.60357171e-01 -4.42275286e-01 -4.92844671e-01
4.40191388e-01 -8.95181656e-01 3.67834687e-01 2.56025761e-01
-3.84685725e-01 -1.65222853e-01 1.14761174e+00 -8.80842209e-01
3.68704796e-01 -2.24097323e+00 1.49615919e-02 5.30052304e-01
1.93377018e-01 2.89300859e-01 -8.29283535e-01 1.58012167e-01
-5.87568283e-01 -4.34300691e-01 -1.51265308e-01 -2.32710585e-01
1.73282344e-02 -1.53252929e-01 4.71691154e-02 1.26882076e+00
3.46250415e-01 9.19443786e-01 -5.68151355e-01 -2.57025152e-01
3.37179244e-01 1.13084888e+00 -2.00468659e-01 3.72546613e-01
2.62610674e-01 6.22133613e-01 -2.92099446e-01 1.16033936e+00
1.26469219e+00 1.18210390e-01 -8.35100040e-02 -6.43290222e-01
9.24722850e-02 -5.49090981e-01 -9.69161630e-01 1.71422672e+00
-4.99849707e-01 5.38681090e-01 2.20474660e-01 -1.16290486e+00
1.09523404e+00 3.00659090e-01 6.68293476e-01 -9.43037689e-01
5.56978703e-01 1.30259320e-01 -4.01983522e-02 -8.18382502e-01
-9.42983702e-02 -4.86404240e-01 5.16368985e-01 2.40122169e-01
3.20263773e-01 3.19782645e-01 -4.64803278e-01 -4.38067824e-01
6.97973073e-01 1.09123744e-01 -2.03330025e-01 -2.54937619e-01
8.55049551e-01 -6.89612448e-01 7.00096905e-01 3.06849808e-01
-3.27052802e-01 8.29685152e-01 9.36221629e-02 -5.91396451e-01
-6.78852499e-01 -9.50605214e-01 -2.61961222e-01 9.61916506e-01
1.12676278e-01 2.82111138e-01 -5.90764523e-01 -5.34395874e-01
1.73282430e-01 -5.57071604e-02 -1.00631106e+00 -3.82228494e-01
-3.75045568e-01 -6.55323327e-01 5.12283742e-01 4.03541744e-01
9.07997072e-01 -1.06981015e+00 -6.46163225e-02 -4.54744548e-01
1.38919100e-01 -1.04449773e+00 -5.31977594e-01 -9.16464925e-02
-4.69286263e-01 -1.26817107e+00 -9.22832727e-01 -8.94472837e-01
7.49603987e-01 5.13639271e-01 6.69905186e-01 6.30259067e-02
-7.62226343e-01 4.91190672e-01 -3.37396711e-02 -1.82450920e-01
1.39539197e-01 -4.30578738e-01 4.00837250e-02 7.49986470e-01
5.53209186e-01 -3.37966442e-01 -6.75540984e-01 3.04623932e-01
-1.03181076e+00 -4.81444389e-01 5.97137213e-01 1.22127140e+00
5.94802558e-01 -1.69583425e-01 2.88163841e-01 -5.69853902e-01
1.35396570e-01 -5.47125518e-01 -4.55815107e-01 5.95084369e-01
1.48091223e-02 -2.35419214e-01 6.32901490e-01 -3.81384134e-01
-1.00465691e+00 8.65681842e-02 -1.30061597e-01 -9.15406048e-01
-1.41113847e-01 4.36115056e-01 -6.69676661e-01 -6.77571476e-01
5.05444765e-01 1.69191092e-01 3.21407676e-01 -1.98122993e-01
1.56008676e-01 8.00359249e-01 4.21059847e-01 -4.71809685e-01
1.12627697e+00 6.41805470e-01 3.25399101e-01 -1.12846076e+00
-8.35269630e-01 -4.20437425e-01 -5.25985777e-01 -4.21995580e-01
8.21002185e-01 -1.14031398e+00 -1.04151988e+00 5.71285129e-01
-9.15486634e-01 -1.19582556e-01 -1.02677047e-01 6.12178981e-01
-4.02622610e-01 1.59333155e-01 -3.78701955e-01 -6.82892263e-01
-4.23745960e-01 -1.02876997e+00 1.10025406e+00 3.79131615e-01
6.06261790e-01 -8.48982513e-01 -6.73856512e-02 5.90310276e-01
6.02832854e-01 4.43609297e-01 5.85349083e-01 -3.35319698e-01
-2.31479287e-01 -3.79938006e-01 -5.48122823e-01 7.38224447e-01
3.13773811e-01 -1.04541518e-01 -1.56143117e+00 -5.77083170e-01
1.62250161e-01 -5.27708888e-01 9.20607805e-01 3.37270796e-01
1.52619743e+00 -3.17885401e-03 2.43706815e-02 1.13533282e+00
1.54006577e+00 5.42784780e-02 6.58477664e-01 -2.55660955e-02
1.02705610e+00 9.51779306e-01 3.88253659e-01 4.01108623e-01
2.33568139e-02 5.83866179e-01 5.47622144e-01 -2.73619771e-01
-2.63242304e-01 1.13968387e-01 4.33771670e-01 5.61453640e-01
-1.70397952e-01 1.04188807e-01 -8.12819183e-01 1.26703635e-01
-1.71076632e+00 -1.24874532e+00 3.77033889e-01 2.32907343e+00
2.45047584e-01 -7.87470341e-01 -1.99512497e-01 -1.86161876e-01
9.07581031e-01 2.95630008e-01 -8.62438858e-01 1.20316662e-01
-4.43754315e-01 1.66425228e-01 2.76777655e-01 2.32559592e-01
-9.06152546e-01 6.21578515e-01 4.75847769e+00 8.52993488e-01
-1.37033856e+00 3.48114520e-01 7.83998311e-01 -3.36400211e-01
-2.51542360e-01 -4.52448189e-01 -4.68474597e-01 1.98883578e-01
5.30032039e-01 2.87594348e-01 6.44136906e-01 5.77196181e-01
3.56849246e-02 5.04653931e-01 -9.41009879e-01 1.72777236e+00
5.24937034e-01 -1.09067035e+00 -2.97329128e-02 5.83748035e-02
7.63552368e-01 2.37999186e-01 4.00824487e-01 9.00251642e-02
-1.98307872e-01 -1.30320668e+00 6.22725129e-01 7.55331397e-01
1.06943989e+00 -1.02225208e+00 8.05292130e-01 -1.13004364e-01
-1.36324358e+00 -2.39099860e-01 -9.22678888e-01 4.73988414e-01
-2.84629643e-01 4.29874897e-01 2.96067428e-02 7.78413951e-01
8.13473403e-01 9.53964174e-01 -3.09105963e-01 5.49708068e-01
3.16525877e-01 -5.95892556e-02 -4.70873937e-02 5.16605914e-01
9.90597978e-02 -4.98411864e-01 1.72472864e-01 8.55104387e-01
4.16574806e-01 1.92576423e-01 1.22554086e-01 9.02413666e-01
-4.91459548e-01 -5.95241599e-03 -8.44062984e-01 -1.68188829e-02
1.67392254e-01 1.73767352e+00 -2.29576126e-01 3.98261696e-02
-5.54697096e-01 9.47181106e-01 2.40388140e-01 6.00579083e-01
-7.58812547e-01 -3.61161441e-01 8.31589997e-01 -1.64689735e-01
1.42582625e-01 3.55059467e-02 1.24801725e-01 -1.14512169e+00
1.42460302e-01 -8.67882550e-01 3.57110292e-01 -5.78977227e-01
-1.62091589e+00 8.77481282e-01 -3.59330803e-01 -1.11405134e+00
2.48872146e-01 -8.25267851e-01 -5.95546067e-01 1.42811334e+00
-2.00802445e+00 -1.59257364e+00 -8.71965528e-01 1.22729969e+00
8.42441767e-02 -7.48753846e-01 6.41118169e-01 4.84308451e-01
-9.40523505e-01 8.47448528e-01 3.01798731e-01 4.63386327e-01
5.97740114e-01 -4.33712304e-01 -3.13185066e-01 6.23086333e-01
-2.31278867e-01 6.74479187e-01 9.73574966e-02 -3.86682481e-01
-1.88163316e+00 -1.33803058e+00 4.49217796e-01 7.20027760e-02
3.18291157e-01 -2.09653944e-01 -6.91041410e-01 4.07140791e-01
1.95616931e-02 1.06426167e+00 1.10694170e+00 -1.45092025e-01
-9.23720241e-01 -6.44935966e-01 -1.51296890e+00 1.88200191e-01
1.02318633e+00 -1.22566378e+00 8.13809931e-02 5.32129824e-01
1.28928125e-01 -1.08535774e-01 -1.19958615e+00 5.06449163e-01
9.20784533e-01 -1.00803924e+00 1.06081617e+00 -6.90047622e-01
1.80364043e-01 -3.94465387e-01 -5.86778581e-01 -9.79705572e-01
-2.55922109e-01 -4.85117346e-01 1.42713025e-01 1.23492026e+00
3.51446457e-02 -7.85734534e-01 5.95439970e-01 3.68235022e-01
-9.11399499e-02 -6.53160930e-01 -1.05817688e+00 -8.47405732e-01
-8.53466839e-02 -1.47803411e-01 7.06299901e-01 1.13703239e+00
-6.07308090e-01 -2.18845904e-02 -2.65612602e-01 3.91967416e-01
1.03529692e+00 1.02433369e-01 4.05499965e-01 -1.14384878e+00
2.38274001e-02 -2.50904918e-01 -5.83211422e-01 -2.56848663e-01
7.98597991e-01 -1.08915746e+00 1.28333315e-01 -1.10639834e+00
6.29655063e-01 -2.86009461e-01 -5.84098816e-01 5.97968280e-01
1.24150224e-01 8.66058767e-01 2.36908600e-01 1.90866888e-01
-2.91085154e-01 1.02835035e+00 1.24956298e+00 -5.49410224e-01
1.10091947e-01 -4.10022944e-01 -4.57900256e-01 4.61833924e-01
7.09697306e-01 -3.02694499e-01 -3.43943298e-01 -5.01166880e-01
-1.29385352e-01 1.06132627e-01 6.53522313e-01 -9.53827262e-01
2.75128156e-01 -2.80100763e-01 9.16612446e-01 -2.08488792e-01
3.71762782e-01 -1.07360494e+00 3.57244253e-01 2.65748858e-01
-2.61162013e-01 -1.42091736e-01 1.03174374e-01 3.66172254e-01
-3.69173557e-01 1.38027385e-01 1.36334312e+00 5.27105331e-02
-4.24381137e-01 1.13104510e+00 2.39376202e-01 -2.76121974e-01
1.06078577e+00 -1.80698335e-01 -4.52755690e-01 7.46874558e-03
-5.71724713e-01 -1.59188285e-01 2.77432770e-01 6.68612123e-01
1.05921161e+00 -1.72694325e+00 -8.27689052e-01 6.77365780e-01
4.02460486e-01 -3.07526886e-01 8.32367659e-01 9.24226582e-01
-5.15039802e-01 8.54631066e-02 -5.71325898e-01 -5.32341003e-01
-1.39035046e+00 3.41259450e-01 7.89013505e-01 3.53232473e-01
-3.87324363e-01 9.74346042e-01 4.00453135e-02 -6.82025671e-01
2.49453619e-01 3.44829828e-01 -1.82705343e-01 1.53923690e-01
8.22594285e-01 2.94006854e-01 1.80731624e-01 -1.40890586e+00
-6.65248811e-01 1.06007826e+00 1.33275181e-01 2.15949744e-01
1.55987501e+00 3.23573723e-02 -6.54187024e-01 6.51538894e-02
2.01241922e+00 -4.64305878e-01 -1.33442652e+00 -4.49381858e-01
-4.58370477e-01 -6.33059859e-01 3.70503396e-01 -4.59285945e-01
-1.83097064e+00 1.03875017e+00 1.26374340e+00 -2.97974437e-01
1.29161477e+00 -1.49896145e-01 6.14811480e-01 4.00203794e-01
1.00611381e-01 -9.96589720e-01 1.12694293e-01 3.20463151e-01
1.11811388e+00 -1.54261553e+00 -1.30203024e-01 -1.13244079e-01
-3.62266064e-01 1.34983242e+00 6.01731062e-01 6.86566234e-02
1.03477728e+00 -7.95350149e-02 3.22915912e-01 -3.51060420e-01
-2.34920070e-01 -2.56533861e-01 6.50356948e-01 7.83518672e-01
3.90115142e-01 -1.62548631e-01 3.30793053e-01 1.80658534e-01
2.36624688e-01 -1.38445303e-01 -2.87198424e-01 7.09323943e-01
-1.43657550e-01 -6.91337407e-01 -4.95039880e-01 3.88310403e-01
-3.22878510e-01 1.65078416e-01 -1.61874041e-01 4.91629243e-01
3.34476024e-01 1.01742685e+00 8.06698725e-02 -6.71388388e-01
2.77820110e-01 -1.06689848e-01 6.84516668e-01 -3.61546129e-01
-5.33664584e-01 -7.88535923e-02 -5.56173146e-01 -8.43875766e-01
-5.85380852e-01 -4.99727279e-01 -8.94452512e-01 -6.25937641e-01
-1.99309200e-01 -9.59619358e-02 8.00601304e-01 9.51513529e-01
2.88283914e-01 4.50559586e-01 1.16792262e+00 -1.04526091e+00
-5.61233699e-01 -9.65875208e-01 -1.06895363e+00 4.69841719e-01
6.19390488e-01 -6.25062168e-01 -2.98263937e-01 -1.87901706e-01]
|
[13.14852237701416, 0.4679422080516815]
|
f10bc8f8-4922-43bd-9c3a-790587cf4082
|
joint-action-loss-for-proximal-policy
|
2301.10919
| null |
https://arxiv.org/abs/2301.10919v1
|
https://arxiv.org/pdf/2301.10919v1.pdf
|
Joint action loss for proximal policy optimization
|
PPO (Proximal Policy Optimization) is a state-of-the-art policy gradient algorithm that has been successfully applied to complex computer games such as Dota 2 and Honor of Kings. In these environments, an agent makes compound actions consisting of multiple sub-actions. PPO uses clipping to restrict policy updates. Although clipping is simple and effective, it is not efficient in its sample use. For compound actions, most PPO implementations consider the joint probability (density) of sub-actions, which means that if the ratio of a sample (state compound-action pair) exceeds the range, the gradient the sample produces is zero. Instead, for each sub-action we calculate the loss separately, which is less prone to clipping during updates thereby making better use of samples. Further, we propose a multi-action mixed loss that combines joint and separate probabilities. We perform experiments in Gym-$\mu$RTS and MuJoCo. Our hybrid model improves performance by more than 50\% in different MuJoCo environments compared to OpenAI's PPO benchmark results. And in Gym-$\mu$RTS, we find the sub-action loss outperforms the standard PPO approach, especially when the clip range is large. Our findings suggest this method can better balance the use-efficiency and quality of samples.
|
['Simon Lucas', 'Greg Slabaugh', 'Yizhao Jin', 'Xiulei Song']
|
2023-01-26
| null | null | null | null |
['dota-2']
|
['playing-games']
|
[-3.11289966e-01 -1.84241176e-01 -7.12505400e-01 5.55640385e-02
-6.76052094e-01 -3.39596242e-01 4.49209511e-01 7.66534135e-02
-1.14451408e+00 1.29994202e+00 1.32200763e-01 -4.67224389e-01
-9.72143933e-02 -6.50580704e-01 -7.48265386e-01 -6.15646899e-01
-4.28451210e-01 6.03879154e-01 6.78168297e-01 -3.60379934e-01
2.71403342e-01 2.39090279e-01 -1.40118396e+00 -1.23182498e-01
9.52352166e-01 9.33314562e-01 2.56334305e-01 7.95890689e-01
3.22169885e-02 9.61490095e-01 -8.51037502e-01 -2.97868401e-01
6.91528976e-01 -5.10062099e-01 -4.67619061e-01 -3.80197406e-01
2.78799325e-01 -7.38670886e-01 -2.93088853e-01 1.02572620e+00
8.57640684e-01 5.25780976e-01 3.97608340e-01 -1.40850246e+00
8.81327614e-02 5.94259918e-01 -7.55536079e-01 4.65223879e-01
3.36397558e-01 5.32473207e-01 1.04391253e+00 -1.84759140e-01
5.13010979e-01 1.24659634e+00 5.22270203e-01 5.96601546e-01
-1.22096431e+00 -7.23173797e-01 5.01548827e-01 9.20327678e-02
-8.44571888e-01 -1.80801317e-01 2.94013888e-01 1.09934863e-02
1.06162882e+00 2.12536231e-01 8.60577166e-01 7.80776262e-01
3.22025806e-01 1.23932087e+00 1.23165452e+00 -2.57596940e-01
5.29923081e-01 -7.71652907e-02 -2.24673927e-01 5.62454462e-01
-9.20968428e-02 4.23815459e-01 -5.69505394e-01 -5.06898880e-01
7.31267631e-01 -1.21005997e-01 -1.45198703e-01 -2.95781702e-01
-8.39695811e-01 7.80962169e-01 1.61943510e-01 -1.34006798e-01
-6.58090472e-01 7.67186105e-01 4.34056520e-01 3.34622264e-01
4.53282148e-01 3.69594157e-01 -2.62615144e-01 -9.60787952e-01
-1.05914140e+00 9.01746035e-01 8.42149019e-01 7.76156068e-01
4.45836276e-01 7.76259378e-02 -6.34474874e-01 8.12992096e-01
3.33249122e-02 2.98396647e-01 3.76289070e-01 -1.49333048e+00
6.96007967e-01 -1.24474950e-02 7.05363452e-01 -5.13314307e-01
-2.15530992e-01 -2.15339288e-01 -1.40584216e-01 8.40075970e-01
8.37395608e-01 -4.99494880e-01 -8.09531391e-01 2.09377575e+00
3.51129770e-01 1.73741668e-01 -2.40943089e-01 9.35038328e-01
-1.43026963e-01 6.90970004e-01 2.35224202e-01 -3.36840421e-01
1.08202434e+00 -1.03370309e+00 -5.83715260e-01 -4.84669358e-01
3.83425534e-01 -6.21080220e-01 1.29986072e+00 4.60363865e-01
-1.34244025e+00 -1.11588247e-01 -8.94338548e-01 3.51110280e-01
-4.21317481e-03 -2.57821769e-01 7.34990418e-01 4.98155892e-01
-8.48916173e-01 1.01779914e+00 -8.94084156e-01 -1.73888370e-01
5.39531231e-01 4.20933634e-01 1.96828201e-01 5.94570190e-02
-1.09611511e+00 8.98511648e-01 3.64336401e-01 -6.12006843e-01
-1.14282894e+00 -8.61762404e-01 -3.98579866e-01 3.28646630e-01
9.40171242e-01 -4.33218122e-01 1.60836625e+00 -7.88430214e-01
-2.04370832e+00 1.63123950e-01 8.36675614e-02 -8.62586081e-01
1.04272151e+00 -4.71280783e-01 1.62050560e-01 2.28510816e-02
4.03603911e-02 7.21342087e-01 9.47184443e-01 -7.91190684e-01
-1.17364454e+00 -6.52460828e-02 5.19413948e-01 8.16341043e-01
-1.61184102e-01 1.04641668e-01 -4.81868893e-01 -5.32562494e-01
-5.88872969e-01 -9.02322471e-01 -4.56264019e-01 4.49526235e-02
-3.78242917e-02 -3.41643572e-01 4.40615326e-01 -3.50423604e-01
1.52621341e+00 -2.05166101e+00 -1.30770981e-01 1.39658943e-01
3.95389870e-02 4.50755537e-01 6.65507987e-02 2.83638716e-01
4.07470822e-01 -7.53689706e-02 -6.84873462e-02 -3.73478979e-01
2.66508102e-01 3.62233460e-01 -1.49964228e-01 4.29736942e-01
-4.79939491e-01 4.64020371e-01 -1.18440914e+00 -3.87653917e-01
1.39016315e-01 2.36867685e-02 -9.14744854e-01 -4.08967249e-02
-5.18997610e-01 9.23788249e-02 -5.16649187e-01 3.07758182e-01
4.07914788e-01 2.07866967e-01 5.25053963e-02 5.42699635e-01
-2.44694397e-01 3.90345007e-01 -1.27276814e+00 1.39393282e+00
-3.97902638e-01 3.73343110e-01 2.39615217e-01 -7.38884091e-01
4.44483578e-01 1.79016158e-01 6.98974311e-01 -6.64830267e-01
2.83368528e-02 2.97915071e-01 8.27981532e-02 -1.97613128e-02
7.85967410e-01 1.44603765e-02 8.16670135e-02 5.12836337e-01
-3.69619936e-01 -1.88941449e-01 6.60887480e-01 3.59253198e-01
1.32992387e+00 4.92456704e-01 2.64672130e-01 -4.18464392e-01
5.89357577e-02 -5.21338880e-02 8.35113049e-01 1.17372489e+00
-7.97653258e-01 2.25305468e-01 8.37560356e-01 -1.71417877e-01
-9.28785205e-01 -9.45630610e-01 2.38508269e-01 1.58977199e+00
2.00899631e-01 -4.59793359e-01 -5.95020592e-01 -8.48414242e-01
3.84162039e-01 7.82949090e-01 -4.70105201e-01 -5.64309359e-02
-6.38211966e-01 -7.58822143e-01 4.76748914e-01 2.85113007e-01
6.64977551e-01 -1.27064228e+00 -9.42307174e-01 5.63249230e-01
-9.09484252e-02 -6.30097926e-01 -1.08698952e+00 1.07232518e-01
-7.44100928e-01 -9.50655460e-01 -9.10191715e-01 -1.55592635e-01
2.49836653e-01 6.00729249e-02 1.04608750e+00 -3.51299822e-01
2.43491353e-03 4.81788546e-01 -3.13074410e-01 -5.52144408e-01
-2.15201706e-01 -8.59549865e-02 2.69482821e-01 -2.61913449e-01
1.48116112e-01 -6.04768038e-01 -7.96854436e-01 2.59984374e-01
-5.04745066e-01 -2.67148554e-01 3.63213301e-01 8.45358968e-01
5.35765707e-01 -2.83933878e-01 3.94911230e-01 -5.16808450e-01
1.19017875e+00 -3.59492004e-01 -8.72714996e-01 2.77089886e-02
-6.74978673e-01 1.46480218e-01 7.26788580e-01 -7.76266575e-01
-1.00754130e+00 -3.71072680e-01 9.29655507e-02 -5.03734052e-01
3.14681858e-01 1.74841762e-01 3.02205086e-01 1.63962483e-01
6.86969280e-01 8.82489458e-02 1.86060190e-01 -3.64958018e-01
2.56929904e-01 4.05734867e-01 1.80071592e-01 -9.52043176e-01
5.07472940e-02 3.65203530e-01 -1.94396704e-01 -6.18790209e-01
-3.46936941e-01 -3.34638983e-01 5.06292641e-01 -3.36515963e-01
6.86049461e-01 -6.48205876e-01 -1.17268968e+00 3.97393703e-01
-6.14871860e-01 -1.00305116e+00 -6.24024332e-01 8.06662798e-01
-8.24375272e-01 3.41175586e-01 -6.95487142e-01 -1.03102410e+00
-2.38151684e-01 -1.36301589e+00 5.08784592e-01 5.14491439e-01
-7.49239028e-02 -6.49943233e-01 1.88265532e-01 3.77026051e-02
4.33724552e-01 -1.31061479e-01 2.98885226e-01 -4.22086984e-01
-3.24096829e-01 5.45306541e-02 1.15340464e-01 4.08424586e-01
-1.63504258e-01 -3.33773047e-01 -3.74507666e-01 -6.94584429e-01
-2.76084483e-01 -5.09729862e-01 7.71334052e-01 6.71521783e-01
1.04572618e+00 -2.99474031e-01 -2.54112035e-01 2.90073305e-01
1.25640392e+00 4.94194448e-01 6.68396175e-01 6.01768434e-01
1.47191092e-01 1.52506813e-01 9.19537604e-01 8.80252242e-01
2.41811752e-01 8.95042002e-01 5.10138810e-01 2.13321045e-01
5.37971556e-02 -2.09651038e-01 8.53990972e-01 3.41101736e-02
-2.99679697e-01 -2.03872651e-01 -4.68649209e-01 5.64912438e-01
-2.12477708e+00 -1.15971613e+00 4.59396034e-01 2.60332656e+00
1.11953628e+00 6.62054062e-01 5.68550467e-01 -2.64323682e-01
5.57577074e-01 3.45049262e-01 -8.14314723e-01 -7.05185115e-01
2.27711335e-01 3.81165802e-01 1.10699868e+00 7.48110652e-01
-7.98352540e-01 1.03278244e+00 6.18077230e+00 1.50083268e+00
-7.77821124e-01 2.91211784e-01 5.38882673e-01 -8.82048905e-01
1.40002161e-01 1.24668419e-01 -9.29354191e-01 9.63673830e-01
6.64153039e-01 -1.76507473e-01 8.42486620e-01 9.95692611e-01
4.26284999e-01 -8.18633437e-01 -8.40969324e-01 7.34422922e-01
-3.65903467e-01 -1.11240613e+00 -4.46540982e-01 2.91313827e-01
6.69564009e-01 2.61923850e-01 -1.82590540e-02 7.64455318e-01
1.02412212e+00 -5.65051854e-01 9.12072241e-01 2.73216516e-01
4.47346091e-01 -1.00567114e+00 4.98043597e-01 4.63630229e-01
-1.04423153e+00 -1.91665426e-01 -2.38930672e-01 -3.17318380e-01
2.86036760e-01 2.69491762e-01 -5.80288529e-01 3.31041873e-01
9.39662695e-01 2.75485963e-01 6.69678822e-02 1.16177022e+00
-3.22266966e-01 6.60298705e-01 -8.00262988e-01 -3.58734995e-01
6.88660622e-01 -3.89262468e-01 9.18021023e-01 7.71057844e-01
1.21799754e-02 -2.00808272e-02 7.24669158e-01 6.02161646e-01
5.09619936e-02 -5.31084277e-02 -1.86749354e-01 1.67902261e-01
6.13746405e-01 7.54615068e-01 -4.32630301e-01 -5.62949419e-01
-3.33635330e-01 8.88991475e-01 5.26464105e-01 4.58273470e-01
-1.01663375e+00 -4.75484818e-01 1.14105344e+00 -2.63884012e-02
3.70722026e-01 -1.59179643e-02 -2.56353412e-02 -8.49203169e-01
1.13636963e-01 -9.08880174e-01 3.86524051e-01 -3.59680563e-01
-9.16608512e-01 7.09332004e-02 4.10006970e-01 -1.10853338e+00
-3.25354636e-01 -1.69534609e-01 -7.17718601e-01 7.92554736e-01
-1.29328609e+00 -4.22634512e-01 3.77669454e-01 4.76648092e-01
6.62273645e-01 6.47603869e-02 2.57163197e-01 4.49617386e-01
-5.84251940e-01 7.29330778e-01 3.65422279e-01 -2.15038061e-01
6.67380989e-01 -1.43652534e+00 -2.42739054e-03 9.21554744e-01
-3.35718632e-01 3.27552676e-01 1.00183761e+00 -7.36108541e-01
-8.94188583e-01 -5.68140745e-01 3.52546036e-01 6.38629496e-02
4.79510099e-01 -1.08382804e-03 -5.23377180e-01 5.35176754e-01
1.81842238e-01 -1.70948133e-02 2.63648659e-01 -1.67019032e-02
2.21099079e-01 -1.87827870e-01 -1.27978587e+00 9.97183025e-01
1.04216421e+00 -1.36532724e-01 -4.00048107e-01 1.81007117e-01
4.48470533e-01 -5.23600876e-01 -7.46152639e-01 1.47688076e-01
5.80247402e-01 -1.26734257e+00 7.66563177e-01 -7.01355577e-01
1.98005307e-02 -1.73446730e-01 4.30320576e-02 -1.53217876e+00
-1.11811407e-01 -1.15606916e+00 -3.41603309e-01 9.05282676e-01
3.15616161e-01 -6.80285215e-01 1.06254244e+00 7.27444708e-01
-1.80772588e-01 -1.04867816e+00 -1.24063933e+00 -1.11021352e+00
1.29310623e-01 -5.53856969e-01 4.61925596e-01 5.06934166e-01
2.60107964e-01 -1.66969717e-01 -6.88868940e-01 -3.25427294e-01
6.24933124e-01 -2.71378517e-01 7.52107382e-01 -5.08730531e-01
-7.92276859e-01 -8.19634259e-01 1.98225692e-01 -1.38697720e+00
-1.91029847e-01 -5.34674406e-01 6.35850355e-02 -1.50281119e+00
-1.06895186e-01 -8.64645123e-01 -3.34622145e-01 6.31750464e-01
-4.82368886e-01 -8.17581713e-02 7.15202093e-01 1.86068505e-01
-7.06918836e-01 7.78307617e-01 1.32178044e+00 1.01677135e-01
-7.74564505e-01 1.69438452e-01 -2.23879516e-01 7.25589097e-01
9.84130979e-01 -6.44859374e-01 -3.75562787e-01 -1.65935069e-01
4.16287370e-02 2.14029342e-01 -9.40063968e-02 -1.03015983e+00
1.10176958e-01 -4.80438679e-01 -5.14533110e-02 -2.95594662e-01
4.66801137e-01 -5.35359263e-01 -2.28627026e-01 8.56104910e-01
-3.02152008e-01 -4.38895598e-02 2.20951766e-01 6.56483233e-01
1.66615948e-01 -3.13795835e-01 8.78790021e-01 -3.66689086e-01
-5.19675672e-01 4.32094902e-01 -7.11270452e-01 3.33596796e-01
1.22120249e+00 -1.56763941e-01 -9.75560993e-02 -7.66871929e-01
-5.67953050e-01 7.70837069e-01 3.35752606e-01 5.07235453e-02
3.33770066e-01 -1.18086457e+00 -4.89303321e-01 -3.26935589e-01
-3.45489055e-01 -2.31761903e-01 -2.89735552e-02 9.15538669e-01
-5.77277839e-01 4.18189447e-03 -2.84058958e-01 -1.95313111e-01
-1.16479671e+00 3.57370794e-01 4.00089949e-01 -8.47050250e-01
-5.17409444e-01 7.98188627e-01 -1.85578912e-01 -2.01330557e-01
6.06007397e-01 -2.57272094e-01 5.68663068e-02 -1.82066619e-01
5.93658447e-01 8.74742925e-01 -5.09019375e-01 -6.69173896e-03
-2.24276587e-01 -1.32026440e-02 -1.77366272e-01 -6.78400934e-01
1.13729262e+00 1.36373460e-01 2.31144816e-01 -1.54862385e-02
5.81697345e-01 4.42823134e-02 -1.95328760e+00 -9.74567756e-02
-2.80817419e-01 -7.60802090e-01 7.74224624e-02 -8.01622808e-01
-9.61997151e-01 3.73959601e-01 6.16693914e-01 4.86007929e-02
9.54696715e-01 -3.44393224e-01 7.83382356e-01 2.25798637e-01
7.48254776e-01 -1.55681336e+00 7.65856057e-02 5.68112195e-01
5.25517106e-01 -1.03632331e+00 3.51269603e-01 1.24408878e-01
-8.26449513e-01 5.52673697e-01 7.63329685e-01 -3.71035874e-01
4.38957185e-01 8.56312588e-02 -3.24449599e-01 1.78590804e-01
-7.78821766e-01 -4.55395162e-01 -2.55258024e-01 4.42340970e-01
-1.49045169e-01 2.62254477e-01 -1.00866139e+00 2.41197959e-01
-9.45480838e-02 1.89692006e-01 5.42270660e-01 1.39127457e+00
-5.89076281e-01 -1.36641383e+00 -3.97199273e-01 7.26420164e-01
-8.53247523e-01 -9.41166729e-02 3.92430425e-02 8.38481605e-01
-1.28339101e-02 8.39313447e-01 8.17661062e-02 -1.02401830e-01
3.45778495e-01 -9.40176323e-02 5.28270543e-01 -2.28906304e-01
-7.71205783e-01 3.10506433e-01 3.57351631e-01 -1.05983567e+00
-2.22424179e-01 -6.73347473e-01 -1.26691234e+00 -7.04712749e-01
-1.29444227e-01 3.47167850e-01 4.37446058e-01 7.54824340e-01
3.50101024e-01 5.05725563e-01 4.37984973e-01 -9.91968155e-01
-1.06261373e+00 -9.60868001e-01 -7.14334965e-01 4.25699890e-01
1.71888441e-01 -1.03425193e+00 -3.92409861e-01 -7.10392237e-01]
|
[4.061579704284668, 2.3138651847839355]
|
a7886223-2c6f-4168-9bf0-40ae025b6c62
|
tncr-table-net-detection-and-classification
|
2106.15322
| null |
https://arxiv.org/abs/2106.15322v1
|
https://arxiv.org/pdf/2106.15322v1.pdf
|
TNCR: Table Net Detection and Classification Dataset
|
We present TNCR, a new table dataset with varying image quality collected from free websites. The TNCR dataset can be used for table detection in scanned document images and their classification into 5 different classes. TNCR contains 9428 high-quality labeled images. In this paper, we have implemented state-of-the-art deep learning-based methods for table detection to create several strong baselines. Cascade Mask R-CNN with ResNeXt-101-64x4d Backbone Network achieves the highest performance compared to other methods with a precision of 79.7%, recall of 89.8%, and f1 score of 84.4% on the TNCR dataset. We have made TNCR open source in the hope of encouraging more deep learning approaches to table detection, classification, and structure recognition. The dataset and trained model checkpoints are available at https://github.com/abdoelsayed2016/TNCR_Dataset.
|
['Daniyar Nurseitov', 'Islam Nuradin', 'Alexander Berendeyev', 'Abdelrahman Abdallah']
|
2021-06-19
| null | null | null | null |
['table-detection']
|
['miscellaneous']
|
[ 1.32950008e-01 -2.39433795e-01 -4.67221320e-01 -1.45935416e-01
-1.36031163e+00 -7.76153386e-01 2.61053443e-01 3.74848425e-01
-5.70542291e-02 3.07247847e-01 3.10805976e-01 -1.55655727e-01
1.66953593e-01 -1.09037173e+00 -7.13810027e-01 -3.78184289e-01
3.23206410e-02 5.21045566e-01 2.34502122e-01 -3.88587832e-01
4.33254808e-01 6.67026639e-01 -1.07075047e+00 1.15365183e+00
1.93910196e-01 1.31705868e+00 -1.36039615e-01 1.08374894e+00
6.60615265e-02 1.38287973e+00 -8.95488441e-01 -8.01160634e-01
3.06422144e-01 -1.06152706e-01 -1.04617655e+00 1.24163955e-01
6.25899732e-01 -6.10332072e-01 -7.10924864e-01 6.46642745e-01
6.09589159e-01 -2.45770931e-01 6.32984161e-01 -9.54288721e-01
-1.25454557e+00 9.62972105e-01 -1.10816371e+00 3.57657969e-01
4.82458293e-01 -4.50979173e-02 1.28770971e+00 -8.94212663e-01
9.26152468e-01 1.23012674e+00 6.12859666e-01 4.64106530e-01
-9.20585692e-01 -7.94727147e-01 -3.03508222e-01 3.07013839e-01
-1.55918205e+00 -4.75384563e-01 2.69642055e-01 -2.13088602e-01
1.03423405e+00 2.25905269e-01 2.04672560e-01 1.06334782e+00
2.74409950e-01 1.19864202e+00 6.76626384e-01 -3.32846731e-01
-9.68988910e-02 -9.02012661e-02 -1.30602112e-02 1.03826845e+00
3.65926921e-01 -7.81223595e-01 -8.87101352e-01 8.52330700e-02
6.08893275e-01 -8.97340570e-03 1.26907796e-01 -4.94423583e-02
-1.19416678e+00 6.66659296e-01 6.40884578e-01 4.62484926e-01
-1.38841256e-01 8.05261508e-02 6.60729587e-01 2.06219509e-01
3.16112965e-01 3.57961237e-01 -5.38609207e-01 6.72866628e-02
-4.91530836e-01 2.74222404e-01 4.73298758e-01 1.19659436e+00
2.26575285e-01 -1.75777763e-01 -4.58064497e-01 1.01829076e+00
-8.33568498e-02 5.72508037e-01 1.95800453e-01 -8.84379983e-01
8.14636409e-01 9.82033908e-01 -3.47173326e-02 -1.17347729e+00
-3.13791066e-01 -3.82268608e-01 -1.06349683e+00 -6.65729716e-02
5.23440659e-01 1.63836226e-01 -9.97887611e-01 8.19292963e-01
-6.70212656e-02 -7.26962507e-01 -5.14533296e-02 5.49849391e-01
1.31530929e+00 7.03513503e-01 -5.04387200e-01 5.49270630e-01
1.53113508e+00 -9.66222644e-01 -6.85879529e-01 1.51297793e-01
7.69671500e-01 -1.11826766e+00 1.23766577e+00 7.78424859e-01
-1.00349677e+00 -4.01517600e-01 -1.25406492e+00 -5.33016741e-01
-8.54110658e-01 6.56293869e-01 4.29383337e-01 7.23342836e-01
-1.04412043e+00 2.83527911e-01 -6.60269976e-01 -3.11285645e-01
9.71848071e-01 3.01599234e-01 -5.13249457e-01 -3.12058091e-01
-7.23576725e-01 4.60514665e-01 1.23971269e-01 6.42884895e-02
-9.68495309e-01 -4.04673308e-01 -3.53636563e-01 -1.48942024e-02
5.68359852e-01 6.15812689e-02 1.30459070e+00 -2.58888960e-01
-9.29554880e-01 1.52865183e+00 1.14841036e-01 -5.05541980e-01
5.76223314e-01 -2.44039580e-01 -4.29245561e-01 3.16956520e-01
2.03255013e-01 5.11543870e-01 3.80977482e-01 -9.66208935e-01
-5.15122831e-01 -5.70205152e-01 -4.59833369e-02 -6.74533620e-02
-1.47224560e-01 2.94399679e-01 -8.23690474e-01 -6.19732797e-01
3.76299769e-02 -7.01792002e-01 1.27960593e-01 3.21915373e-02
-9.22807395e-01 -1.52482033e-01 5.77718675e-01 -8.33071709e-01
9.38016772e-01 -1.93031335e+00 -4.67071772e-01 -1.99057281e-01
4.53235537e-01 3.42724174e-01 -2.21707299e-01 6.00346863e-01
-1.30264251e-03 5.48367321e-01 8.18456188e-02 -1.33487821e-01
-1.27824485e-01 -6.05066717e-01 -1.57860070e-01 5.32441199e-01
4.29711938e-01 1.00041878e+00 -4.65175092e-01 -5.34777284e-01
1.38217553e-01 6.85729504e-01 -1.91656083e-01 -6.05751155e-03
-7.91351795e-02 -2.89421920e-02 -3.58603776e-01 1.16699517e+00
8.85131299e-01 -6.65733516e-01 2.80008137e-01 -2.88932532e-01
3.35094869e-01 2.95883715e-01 -1.17788231e+00 1.49851239e+00
-9.43559594e-03 8.01535189e-01 -8.62144157e-02 -5.15693486e-01
1.11930132e+00 -1.64589167e-01 3.83345395e-01 -9.21445429e-01
3.31616670e-01 1.31835192e-01 -1.64605856e-01 -9.83366817e-02
7.44906962e-01 6.27899945e-01 -2.08707109e-01 2.06112817e-01
-5.29493243e-02 1.49614155e-01 4.56239223e-01 5.71489573e-01
1.53770542e+00 -2.58579582e-01 1.76432148e-01 -9.35539305e-02
5.62270045e-01 2.55827606e-01 1.04423724e-01 9.32898819e-01
-1.55828446e-01 9.68151629e-01 5.49844921e-01 -7.38580525e-01
-1.03596795e+00 -8.79524350e-01 -2.00910971e-01 1.07591021e+00
-2.01364949e-01 -7.36071587e-01 -7.09298849e-01 -6.25413597e-01
1.12175673e-01 1.41925797e-01 -9.51785147e-01 1.27617851e-01
-7.31525540e-01 -7.83031940e-01 7.96019733e-01 7.53513694e-01
9.34697509e-01 -1.19234085e+00 -2.82530814e-01 1.02017559e-01
-3.01014870e-01 -1.30741048e+00 -4.30206180e-01 3.43306184e-01
-5.35587549e-01 -1.36161280e+00 -5.93438387e-01 -8.18643630e-01
2.49264717e-01 3.13851267e-01 1.50196671e+00 3.81045043e-01
-6.59393311e-01 3.14845541e-03 -4.38279718e-01 -3.01142812e-01
-3.12233806e-01 5.10001421e-01 -4.57135558e-01 -2.93372810e-01
5.90500176e-01 1.95665315e-01 -6.86544299e-01 2.86289901e-01
-7.17369735e-01 -1.62811011e-01 6.64864182e-01 5.33116579e-01
8.29477370e-01 8.61652941e-02 3.14532995e-01 -1.14676070e+00
5.21131516e-01 -5.86905405e-02 -6.71993017e-01 3.15396100e-01
-5.43113410e-01 -1.38580993e-01 4.76217121e-01 3.80494818e-02
-9.43339407e-01 1.11834839e-01 -4.55567271e-01 4.77570407e-02
-1.17547721e-01 1.22212157e-01 -1.65245369e-01 2.55778402e-01
8.18789899e-01 1.85852230e-01 -4.51085061e-01 -6.12468183e-01
2.29591548e-01 9.99308527e-01 5.71190059e-01 -1.09651901e-01
6.09563053e-01 6.56969249e-01 -2.12430656e-01 -4.94945198e-01
-1.16868162e+00 -5.15148103e-01 -6.12110972e-01 -1.32914126e-01
1.13138235e+00 -1.13972342e+00 -8.75179052e-01 7.91230977e-01
-7.48971224e-01 -4.62045312e-01 9.06420499e-02 -2.45397642e-01
-2.41992116e-01 -3.10835205e-02 -1.24272680e+00 -5.38955390e-01
-7.17269540e-01 -9.99428213e-01 1.45167923e+00 2.93670166e-02
-2.28276178e-02 -5.57314575e-01 -2.59457350e-01 8.11869502e-01
3.69288325e-01 4.88009721e-01 6.41221941e-01 -8.12438369e-01
-8.48177969e-01 -4.44511116e-01 -5.45430660e-01 5.68465889e-02
1.29339427e-01 3.59504670e-01 -9.93530571e-01 -1.47685006e-01
-5.93446553e-01 -7.79398263e-01 1.20239282e+00 3.16796809e-01
1.69023764e+00 -4.50527258e-02 -3.11281532e-01 4.33932245e-01
1.56154728e+00 4.11078453e-01 1.00562131e+00 8.36140573e-01
7.51554549e-01 1.26917303e-01 6.30728066e-01 4.85880196e-01
2.79283047e-01 4.17281091e-01 3.68027270e-01 -2.20065996e-01
-3.14107180e-01 -9.27130058e-02 9.08796564e-02 4.30196077e-01
1.78109631e-01 -7.05421269e-01 -1.27201080e+00 2.78603226e-01
-1.42751145e+00 -6.76622093e-01 -3.92419100e-01 1.66176677e+00
8.28165412e-01 5.73137283e-01 2.59714067e-01 4.05719042e-01
8.95263672e-01 2.97715306e-01 -5.14063537e-01 -3.19903761e-01
-2.93663681e-01 1.90318912e-01 7.46102214e-01 9.27682295e-02
-1.33238316e+00 9.22149301e-01 6.03387070e+00 1.04834914e+00
-8.34144413e-01 -2.06343874e-01 1.35364640e+00 -2.24525668e-03
2.98118591e-01 -6.59010470e-01 -1.30590832e+00 1.65266432e-02
1.02442682e+00 1.15719967e-01 3.16153705e-01 8.59917939e-01
-3.82264912e-01 -1.09175064e-01 -8.21776748e-01 1.03953969e+00
2.14896962e-01 -1.85667455e+00 -2.00034995e-02 7.26075098e-02
5.59224367e-01 2.24772811e-01 3.24158490e-01 3.27886045e-01
5.24669230e-01 -1.34409595e+00 6.56660974e-01 5.64464591e-02
1.12321460e+00 -9.49628413e-01 8.87230575e-01 -2.41360158e-01
-1.33947504e+00 7.59607479e-02 -4.54859942e-01 3.42698455e-01
-6.92715883e-01 4.20976758e-01 -8.90489936e-01 2.84046710e-01
1.42239988e+00 9.64012980e-01 -1.18397868e+00 5.64030111e-01
4.78365347e-02 5.55864155e-01 -1.35069668e-01 -1.21786855e-01
4.95749451e-02 2.63430744e-01 -1.82851300e-01 1.24875283e+00
-2.90454000e-01 -2.21683949e-01 -1.91266518e-02 5.97662151e-01
-9.63881910e-01 2.17453316e-01 -5.57031751e-01 -2.10526690e-01
4.11651224e-01 1.55384636e+00 -1.51934659e+00 -4.23857898e-01
-3.05615723e-01 6.95264637e-01 1.51849285e-01 -2.38584802e-01
-7.10355520e-01 -6.72473550e-01 1.78888306e-01 1.09805226e-01
6.57911837e-01 1.45429045e-01 -4.67895806e-01 -1.28469896e+00
2.41018027e-01 -1.33501101e+00 7.44006217e-01 -7.87073672e-01
-1.31443703e+00 8.86249125e-01 -4.75833565e-01 -8.92603159e-01
2.96383083e-01 -1.02818692e+00 1.96583774e-02 5.30894995e-01
-1.16573727e+00 -1.16499686e+00 -4.72581118e-01 6.58098936e-01
6.39387786e-01 -3.67951959e-01 9.20320392e-01 2.33561993e-01
-7.68422842e-01 9.67108846e-01 4.54548240e-01 9.28966165e-01
8.54343295e-01 -1.54994476e+00 9.05207276e-01 7.30378568e-01
2.99057011e-02 3.84409249e-01 3.16600561e-01 -6.00337625e-01
-1.58552372e+00 -1.25808036e+00 4.96246338e-01 -8.93970013e-01
6.01047635e-01 -9.64534163e-01 -8.38764191e-01 7.95148909e-01
4.07958210e-01 1.03730291e-01 6.18757904e-01 6.64539114e-02
-8.19295406e-01 -4.69033867e-01 -1.40158141e+00 4.74372596e-01
9.30923820e-01 -5.94812870e-01 1.64620727e-01 7.93837011e-01
4.71128285e-01 -7.27055132e-01 -1.18349683e+00 1.83952287e-01
7.57528961e-01 -1.01654923e+00 1.01487660e+00 -3.63350287e-02
8.39724004e-01 -1.17335558e-01 -3.50838989e-01 -4.71025497e-01
-4.42377478e-01 -1.03159718e-01 -7.57348984e-02 1.18207061e+00
7.59633780e-01 -1.60598025e-01 1.35031009e+00 -7.83294514e-02
2.37557232e-01 -9.32736158e-01 -3.57034743e-01 -6.03708327e-01
3.91488910e-01 -2.19677225e-01 6.68847263e-01 6.60525143e-01
-5.88752627e-01 6.05889618e-01 -2.81654656e-01 -8.54713917e-02
4.15137172e-01 1.84523880e-01 8.21867645e-01 -9.20231283e-01
-1.50069315e-02 -2.45530874e-01 -1.96219563e-01 -7.41998434e-01
-3.11743557e-01 -8.79725277e-01 -2.89701134e-01 -1.81353080e+00
5.72287261e-01 -4.91179489e-02 -2.43877724e-01 7.59734094e-01
1.77079663e-01 6.71231985e-01 2.11939350e-01 3.25655252e-01
-1.09697330e+00 2.16563661e-02 1.25526047e+00 -5.34343183e-01
1.86359465e-01 -3.17179114e-01 -9.87001061e-01 2.54323691e-01
1.40096319e+00 -5.99048913e-01 4.92520705e-02 -3.74610037e-01
4.23039228e-01 -1.36056347e-02 -2.06017971e-01 -1.09674704e+00
-1.62338927e-01 2.13851750e-01 1.29717696e+00 -1.33898115e+00
2.54573137e-01 -4.16139781e-01 -2.59852797e-01 6.29441559e-01
-7.92435408e-01 3.58433753e-01 2.41938010e-01 2.99758375e-01
2.65855598e-03 7.32852444e-02 8.56189251e-01 -4.70684409e-01
-5.19335389e-01 1.57551154e-01 -7.14777052e-01 1.38058096e-01
5.86339712e-01 5.61304204e-02 -7.87043333e-01 -3.79562914e-01
-3.36810350e-01 1.08154453e-01 4.61886048e-01 5.80417454e-01
7.47443318e-01 -1.04974413e+00 -9.37982082e-01 -5.48620336e-02
5.10316968e-01 -1.50814936e-01 -4.76357294e-03 2.15452313e-01
-9.02579546e-01 6.75832808e-01 -3.07512850e-01 -6.69788897e-01
-1.39059246e+00 5.29374957e-01 3.70899111e-01 -6.80112720e-01
-6.64263368e-01 9.72949505e-01 -3.99941355e-01 -6.79879308e-01
3.86349827e-01 -3.33922565e-01 -2.25497574e-01 5.57962758e-03
8.23930383e-01 3.96493345e-01 8.07274699e-01 -3.82043064e-01
-6.42632782e-01 3.92497420e-01 -8.54678690e-01 3.25399011e-01
1.40888274e+00 1.67637438e-01 9.89773721e-02 2.27333277e-01
1.30127108e+00 1.78802148e-01 -6.07920825e-01 -2.07480956e-02
1.57356769e-01 -4.33289319e-01 -4.73150350e-02 -1.39931560e+00
-1.37183309e+00 6.16050124e-01 9.67078686e-01 1.67541847e-01
1.21837854e+00 2.69602821e-03 8.34076643e-01 7.79714227e-01
1.37921557e-01 -9.20466006e-01 5.61684251e-01 3.92432034e-01
9.51046109e-01 -1.54302609e+00 2.92679131e-01 -4.86282587e-01
-5.50402105e-01 1.22926772e+00 6.99766219e-01 -2.66682178e-01
4.97982115e-01 6.28328443e-01 2.23383367e-01 -4.56309140e-01
-1.04011285e+00 -2.36838475e-01 1.09232172e-01 4.70305771e-01
1.00745189e+00 2.66624577e-02 1.05494581e-01 2.29717404e-01
-3.02092373e-01 -1.38517991e-01 8.56494665e-01 1.38547134e+00
-1.17531851e-01 -1.05110025e+00 -3.38830948e-01 8.47519696e-01
-1.11827719e+00 -9.09312665e-02 -9.26871300e-01 9.90296364e-01
-2.31215954e-01 1.11362147e+00 -1.53325750e-02 -5.64548850e-01
4.41350341e-01 -3.59214664e-01 4.06945080e-01 -5.15280843e-01
-1.01684034e+00 7.19793141e-02 3.91020298e-01 -6.56648755e-01
-1.05614655e-01 -5.05373478e-01 -1.11370099e+00 -8.00476909e-01
-2.77631283e-01 -2.05858022e-01 3.90250295e-01 2.13570416e-01
4.68206018e-01 5.94295681e-01 4.02775824e-01 -2.07018793e-01
-2.90213466e-01 -9.40753579e-01 -6.29072070e-01 1.55015007e-01
2.59019047e-01 -2.77348191e-01 -3.59128825e-02 2.06249565e-01]
|
[11.69363784790039, 3.006596088409424]
|
1d0ed87f-b212-449c-9e4c-5ed5cd56e9d6
|
extreme-multi-label-classification-from
|
2004.00198
| null |
https://arxiv.org/abs/2004.00198v1
|
https://arxiv.org/pdf/2004.00198v1.pdf
|
Extreme Multi-label Classification from Aggregated Labels
|
Extreme multi-label classification (XMC) is the problem of finding the relevant labels for an input, from a very large universe of possible labels. We consider XMC in the setting where labels are available only for groups of samples - but not for individual ones. Current XMC approaches are not built for such multi-instance multi-label (MIML) training data, and MIML approaches do not scale to XMC sizes. We develop a new and scalable algorithm to impute individual-sample labels from the group labels; this can be paired with any existing XMC method to solve the aggregated label problem. We characterize the statistical properties of our algorithm under mild assumptions, and provide a new end-to-end framework for MIML as an extension. Experiments on both aggregated label XMC and MIML tasks show the advantages over existing approaches.
|
['Hsiang-Fu Yu', 'Yanyao Shen', 'Sujay Sanghavi', 'Inderjit Dhillon']
|
2020-04-01
| null |
https://proceedings.icml.cc/static/paper_files/icml/2020/1388-Paper.pdf
|
https://proceedings.icml.cc/static/paper_files/icml/2020/1388-Paper.pdf
|
icml-2020-1
|
['extreme-multi-label-classification']
|
['methodology']
|
[ 6.80315435e-01 7.62011409e-02 -3.97496790e-01 -8.42544556e-01
-1.60567629e+00 -6.68845594e-01 2.72112101e-01 2.51356333e-01
-3.19143951e-01 8.29200923e-01 -2.71081746e-01 -2.43942767e-01
-4.22562093e-01 -1.53114006e-01 -5.84930837e-01 -7.63153255e-01
1.74408495e-01 9.88456011e-01 -2.28095949e-01 5.67279398e-01
1.69733629e-01 2.28961688e-02 -1.77665138e+00 7.16038346e-01
6.13026023e-01 8.87253046e-01 -1.30377918e-01 7.58334875e-01
-1.73118964e-01 8.57320726e-01 -4.74415362e-01 -3.64634812e-01
2.54281700e-01 -4.72610205e-01 -1.24633765e+00 3.91825736e-01
8.81612539e-01 1.10180199e-01 7.84400761e-01 8.71606171e-01
5.78043103e-01 1.12947538e-01 1.04518867e+00 -1.94386029e+00
-4.87573415e-01 8.06454122e-01 -8.41432989e-01 -4.66248661e-01
2.35863864e-01 -3.14194620e-01 1.34509408e+00 -8.51780713e-01
6.50595307e-01 1.48927665e+00 9.73632455e-01 8.90770912e-01
-1.53084242e+00 -6.24437571e-01 2.50737339e-01 -7.53932968e-02
-1.13347208e+00 -1.80060178e-01 3.87826532e-01 -4.28972125e-01
6.38876379e-01 4.89403129e-01 -4.11202729e-01 1.02330351e+00
-4.87600446e-01 1.00006676e+00 1.63964772e+00 -8.55418921e-01
4.66634095e-01 2.07681984e-01 7.13920951e-01 4.37186211e-01
-4.05540727e-02 -3.50190818e-01 -2.72802770e-01 -7.60653675e-01
-3.81662734e-02 1.02534570e-01 1.95643365e-01 -3.67719680e-02
-1.25025582e+00 8.67741406e-01 -1.16942830e-01 -8.49472061e-02
-1.19979493e-02 4.46752936e-01 5.71739256e-01 4.50533569e-01
1.02099609e+00 3.13297659e-01 -9.58659887e-01 2.12953165e-01
-9.39931095e-01 3.49620432e-01 7.95875549e-01 1.27318525e+00
7.96285033e-01 -8.93638611e-01 -2.97744870e-01 1.13530016e+00
2.73021460e-01 4.74262774e-01 4.53307301e-01 -1.47777307e+00
5.12521327e-01 3.63505304e-01 1.13742381e-01 -2.70946532e-01
-7.70562291e-01 -2.21995264e-01 -4.44369614e-01 1.75745830e-01
6.15140975e-01 -1.66292325e-01 -6.37710869e-01 1.99951792e+00
7.42415786e-01 2.89814383e-01 -1.13404319e-01 4.15525734e-01
6.05754673e-01 4.52539831e-01 4.02506441e-01 -4.55056518e-01
1.35710132e+00 -1.19488013e+00 -5.40997267e-01 -2.34416828e-01
1.45983171e+00 -6.01769924e-01 1.16109931e+00 2.22451881e-01
-8.54442239e-01 -4.20498163e-01 -5.05951464e-01 -2.43415162e-01
-4.58980680e-01 1.52016103e-01 6.04109287e-01 7.50223100e-01
-1.09668076e+00 6.03006601e-01 -2.04687133e-01 -1.63868681e-01
3.33052278e-01 3.71273011e-01 -1.83857620e-01 -2.93838561e-01
-9.14258897e-01 6.27323091e-01 5.99263310e-01 -3.03109378e-01
-4.03023839e-01 -7.31399536e-01 -6.62372172e-01 -1.20701611e-01
6.12489939e-01 -3.60135227e-01 1.56483638e+00 -1.00352287e+00
-9.31623936e-01 1.36622393e+00 -3.78116637e-01 -8.97340290e-03
4.61914778e-01 2.23642394e-01 -2.22197950e-01 -8.00734758e-02
7.17059612e-01 1.02515745e+00 5.15146911e-01 -1.68591273e+00
-1.30145013e+00 -2.17327580e-01 -2.67999154e-02 1.36372283e-01
-6.54034019e-02 3.64748627e-01 -8.12267587e-02 -4.01973099e-01
9.30664614e-02 -1.24550128e+00 -3.26949298e-01 -1.75953478e-01
-5.24310708e-01 -8.47088099e-01 6.65421844e-01 -3.53005826e-01
9.52209651e-01 -2.07098150e+00 -2.94674158e-01 1.77667718e-02
2.68315732e-01 -1.27746686e-01 -3.78121138e-01 2.22452924e-01
-3.82668078e-01 3.44068229e-01 -2.79514372e-01 -1.29126918e+00
5.03436804e-01 9.84986871e-02 -1.53496832e-01 4.72298145e-01
9.96372849e-03 8.96028757e-01 -1.12531471e+00 -9.22472596e-01
-2.41310254e-01 -8.17314014e-02 -3.75655413e-01 -3.23112272e-02
-6.68724000e-01 3.08601111e-01 -1.92812115e-01 9.51856911e-01
7.62764871e-01 -7.87158370e-01 2.40562528e-01 3.60979199e-01
3.17037433e-01 -1.67895094e-01 -1.23578787e+00 1.32169461e+00
-6.31834984e-01 1.76674709e-01 -1.24413423e-01 -9.13620532e-01
4.15404737e-01 5.64137399e-01 8.16539168e-01 -1.59498543e-01
3.45503353e-02 4.96223718e-01 -6.86816096e-01 -3.69885832e-01
1.03832945e-01 -5.70282757e-01 -5.79041839e-01 1.20894516e+00
6.93957582e-02 3.88088048e-01 3.94280553e-01 1.57250211e-01
8.37755620e-01 -2.10797992e-02 5.70639670e-01 -2.14698166e-01
3.22555751e-01 -1.80396616e-01 6.02392912e-01 9.93287921e-01
-3.48654896e-01 6.53031111e-01 4.98300344e-01 -2.33247086e-01
-8.22562635e-01 -6.54085279e-01 -3.97982121e-01 1.76790631e+00
-1.93734035e-01 -2.97409624e-01 -7.87081957e-01 -1.43355882e+00
2.74273098e-01 8.63202274e-01 -6.69741094e-01 2.08971262e-01
-2.12298021e-01 -9.76143956e-01 3.16481978e-01 3.69002998e-01
2.01324746e-02 -1.16135335e+00 -1.47284567e-01 2.26377249e-01
-4.42858398e-01 -1.27790987e+00 -8.15930188e-01 4.08049375e-01
-6.25506699e-01 -1.03565395e+00 -3.23227406e-01 -1.03916919e+00
7.86377132e-01 1.12865113e-01 1.38129580e+00 5.66076674e-03
-1.58409834e-01 3.51524979e-01 -3.97845775e-01 -4.06222314e-01
-7.54019976e-01 8.83026347e-02 -2.56694499e-02 3.16885829e-01
4.73199815e-01 -3.28341484e-01 -1.38415575e-01 3.39606404e-01
-7.42437065e-01 2.03897387e-01 2.58660108e-01 7.85748839e-01
8.48492980e-01 7.40038380e-02 1.22233176e+00 -1.91719282e+00
4.12000686e-01 -8.11118305e-01 -2.44349748e-01 6.82363570e-01
-1.07604277e+00 8.59050453e-03 7.93302655e-01 -6.94174528e-01
-7.73046792e-01 3.29413623e-01 -2.30943579e-02 -2.29711577e-01
-5.06953478e-01 2.96924055e-01 -3.04213315e-01 9.85971279e-03
5.33898711e-01 -2.85918981e-01 -1.15141869e-01 -7.09148049e-01
6.72100306e-01 1.11443627e+00 3.29901308e-01 -9.86215174e-01
1.06431514e-01 2.40964979e-01 2.36469254e-01 2.44765673e-02
-1.59923816e+00 -8.97784770e-01 -8.79838467e-01 -2.74846464e-01
5.64895034e-01 -8.41174483e-01 -9.60774601e-01 2.75649726e-01
-9.56074178e-01 -5.26583552e-01 -4.34792668e-01 1.07937969e-01
-7.94758737e-01 3.47260863e-01 -7.13732839e-01 -9.17554379e-01
-1.86398506e-01 -1.15363181e+00 1.67198622e+00 -7.42359683e-02
-2.93709904e-01 -1.33493984e+00 5.07940315e-02 6.86779737e-01
-1.29499272e-01 4.01533037e-01 1.12212121e+00 -9.52309728e-01
-1.80235371e-01 -3.96742344e-01 -2.28016943e-01 2.82655716e-01
-2.63628270e-02 -3.47489774e-01 -1.18404472e+00 -5.66215217e-01
-1.29964426e-01 -9.38816845e-01 8.78296673e-01 3.60076487e-01
1.38487399e+00 -2.31509358e-01 -5.33684850e-01 2.73492336e-01
1.66163170e+00 -1.08129188e-01 -6.30032718e-02 1.33612007e-01
7.46263266e-01 7.23482668e-01 8.40105355e-01 5.13104439e-01
5.06224632e-01 6.27473176e-01 2.67235070e-01 -7.07452372e-02
-1.13982908e-01 1.43213689e-01 4.97532114e-02 6.03958905e-01
4.12700564e-01 -3.81920397e-01 -8.33239913e-01 4.21276003e-01
-2.10437942e+00 -7.52044082e-01 -3.75828594e-01 2.10526109e+00
1.18203080e+00 -1.68019891e-01 2.41116017e-01 9.16864872e-02
1.08824730e+00 -2.21883371e-01 -7.08410680e-01 -2.56911427e-01
-1.06783979e-01 1.30298719e-01 4.82491791e-01 5.00660300e-01
-1.55972946e+00 6.07763350e-01 7.09237719e+00 1.10904455e+00
-4.02505964e-01 6.66586339e-01 9.79064584e-01 -4.60404634e-01
-1.28205225e-01 1.74983591e-01 -1.25981224e+00 7.22060084e-01
1.28411877e+00 1.28400207e-01 2.72889882e-01 9.48727071e-01
-2.71908104e-01 3.52047831e-02 -1.58833706e+00 1.01126504e+00
9.06492844e-02 -8.22867036e-01 -4.09554332e-01 2.47591570e-01
1.29259348e+00 -2.35567734e-01 2.00382635e-01 6.13271475e-01
8.03726733e-01 -6.74435854e-01 7.35595822e-01 2.10176960e-01
1.23561108e+00 -6.58736765e-01 5.98857820e-01 7.66141415e-01
-9.25741315e-01 -4.25655812e-01 -3.81856173e-01 6.23494089e-02
1.30012229e-01 8.98219347e-01 -7.06613779e-01 2.97168076e-01
1.80409193e-01 5.39342642e-01 -9.52968538e-01 6.82328880e-01
1.15565479e-01 7.26929247e-01 -3.21207076e-01 7.89771751e-02
2.51666129e-01 8.62562656e-02 -1.27286926e-01 1.27280414e+00
1.42864734e-01 -2.40388781e-01 6.69882774e-01 5.79340160e-01
-3.38816434e-01 3.93212199e-01 -2.36552656e-01 4.13384378e-01
7.19097555e-01 1.47374642e+00 -1.00338912e+00 -5.68889022e-01
-6.85914040e-01 9.86262381e-01 8.03685725e-01 2.61816770e-01
-8.61144125e-01 1.94640178e-02 2.41895497e-01 -4.24588948e-01
-6.09203726e-02 6.15129471e-01 -5.06234169e-01 -8.98572981e-01
2.21371949e-02 -8.84880781e-01 8.96550357e-01 -4.32945997e-01
-1.83467281e+00 -2.06943527e-02 1.04834460e-01 -1.14989102e+00
-4.61046100e-01 -5.99485815e-01 -1.99193060e-01 8.83414447e-01
-1.32141113e+00 -1.41681027e+00 1.28896773e-01 2.52129167e-01
6.20498538e-01 1.80082873e-01 1.08604431e+00 2.87437648e-01
-5.37831485e-01 9.04875755e-01 7.54250288e-01 -3.60335648e-01
1.09285045e+00 -1.78694963e+00 2.53314346e-01 4.76775050e-01
1.20570295e-01 1.91015229e-01 3.46281439e-01 -4.68272805e-01
-4.73441243e-01 -1.67828333e+00 1.43111658e+00 -8.85922790e-01
2.88511395e-01 -6.41536772e-01 -6.22186542e-01 1.01715970e+00
-1.95133403e-01 3.10644239e-01 1.29809773e+00 6.31922722e-01
-5.24474919e-01 1.23686455e-01 -1.24487710e+00 3.19991499e-01
1.05479169e+00 -6.57114506e-01 1.98193118e-02 1.03270817e+00
9.05063391e-01 -8.92642513e-02 -1.03389776e+00 2.63477325e-01
2.80760944e-01 -5.41371882e-01 9.25704956e-01 -7.04642594e-01
4.79165018e-01 -1.32853836e-01 -2.36910671e-01 -1.17433953e+00
-5.19107819e-01 -2.45637223e-01 -6.21399432e-02 1.61021101e+00
7.57377088e-01 -6.41517401e-01 7.54302800e-01 9.62851226e-01
9.14412886e-02 -7.47129738e-01 -7.79471457e-01 -8.44785035e-01
3.21455538e-01 -5.33500493e-01 6.65786564e-01 1.20899141e+00
9.78196636e-02 4.83877212e-01 -5.00451028e-01 1.26945585e-01
1.19763875e+00 6.27410710e-01 3.46484482e-01 -1.57507312e+00
-5.10458827e-01 -1.92004383e-01 7.77577162e-02 -5.23782969e-01
8.85150909e-01 -1.53339934e+00 2.62972146e-01 -1.32935119e+00
8.21722507e-01 -9.78173077e-01 -4.57200974e-01 8.31436515e-01
-7.65570104e-01 4.77142215e-01 2.55852491e-01 4.42372531e-01
-1.31583905e+00 -1.73269689e-01 8.63426626e-01 -6.80942461e-02
2.48954192e-01 2.93856233e-01 -9.27802503e-01 4.20176834e-01
7.67299473e-01 -9.81129050e-01 -5.19119442e-01 5.48709221e-02
2.06908360e-01 3.98891643e-02 3.96251418e-02 -7.85074532e-01
1.56854540e-01 -2.50894994e-01 7.56514594e-02 -5.68701863e-01
-1.09532490e-01 -7.77345896e-01 2.55287826e-01 -7.57991672e-02
-1.16953647e+00 -4.24372196e-01 -4.25956845e-01 6.13836586e-01
1.66368082e-01 -8.09524000e-01 9.03684020e-01 -2.49544963e-01
-5.05526304e-01 3.21840674e-01 -1.13772657e-02 4.67639297e-01
1.23139870e+00 1.80027008e-01 -2.90447891e-01 -3.39506660e-03
-9.99041200e-01 5.47815979e-01 4.19903517e-01 -3.17545943e-02
8.93179402e-02 -1.66129017e+00 -8.82658839e-01 -7.54251704e-02
4.67475265e-01 1.55350938e-01 8.58764499e-02 5.75075209e-01
5.60649261e-02 3.15665275e-01 4.37849432e-01 -4.31076050e-01
-1.54634845e+00 1.14513540e+00 8.15625414e-02 -6.86502457e-01
-3.08731914e-01 8.11328471e-01 3.12447578e-01 -1.07795560e+00
3.23890120e-01 -3.80714647e-02 -4.83059287e-02 1.28267929e-01
5.20398498e-01 4.15063113e-01 -7.75969699e-02 -5.52930415e-01
-1.28289849e-01 5.38426518e-01 -1.66665569e-01 -8.74521658e-02
9.94619966e-01 -4.33239907e-01 -2.72907078e-01 1.04953158e+00
1.74397850e+00 -5.07839859e-01 -1.09543967e+00 -6.63328826e-01
5.27365625e-01 -1.75433204e-01 -4.15311009e-01 -9.58705127e-01
-7.52246857e-01 6.11051321e-01 3.54270369e-01 2.74741560e-01
1.07739270e+00 3.69300961e-01 4.88596708e-01 2.41979703e-01
5.28154790e-01 -1.29733562e+00 -7.07403123e-02 3.23244929e-01
3.32687438e-01 -1.51470649e+00 -1.90785810e-01 -5.59832096e-01
-6.08978868e-01 8.32667410e-01 4.92922306e-01 4.39891994e-01
7.49151528e-01 3.45700353e-01 1.83181956e-01 -1.44759938e-01
-1.20197082e+00 -2.14785978e-01 -8.94142464e-02 4.42500859e-01
4.75113541e-01 3.73973578e-01 -3.13277036e-01 5.47749817e-01
2.92640954e-01 -1.42792568e-01 2.92429686e-01 9.67143059e-01
-3.24490875e-01 -1.32584286e+00 -4.52769667e-01 7.52352536e-01
-6.29474282e-01 -4.19136658e-02 -3.09851468e-01 4.07880902e-01
5.83502114e-01 1.11606121e+00 -1.51447296e-01 -1.31273583e-01
6.94523752e-02 7.15825975e-01 2.34680414e-01 -7.97338784e-01
-4.06302780e-01 -1.37317274e-02 1.31757200e-01 -3.00429463e-01
-7.13551879e-01 -1.14483237e+00 -1.34530163e+00 -1.28591135e-01
-6.39832914e-01 1.37605578e-01 4.95735914e-01 1.03371537e+00
1.65025160e-01 1.39600888e-01 1.11013687e+00 -6.98571920e-01
-8.13292563e-01 -1.05571783e+00 -1.08082950e+00 1.01128423e+00
2.80260742e-01 -5.27025163e-01 -5.72369039e-01 3.80285382e-01]
|
[9.470165252685547, 4.34206485748291]
|
c2ccc2de-99de-4430-b0be-ad42fd5db523
|
docred-fe-a-document-level-fine-grained
|
2303.11141
| null |
https://arxiv.org/abs/2303.11141v2
|
https://arxiv.org/pdf/2303.11141v2.pdf
|
DocRED-FE: A Document-Level Fine-Grained Entity And Relation Extraction Dataset
|
Joint entity and relation extraction (JERE) is one of the most important tasks in information extraction. However, most existing works focus on sentence-level coarse-grained JERE, which have limitations in real-world scenarios. In this paper, we construct a large-scale document-level fine-grained JERE dataset DocRED-FE, which improves DocRED with Fine-Grained Entity Type. Specifically, we redesign a hierarchical entity type schema including 11 coarse-grained types and 119 fine-grained types, and then re-annotate DocRED manually according to this schema. Through comprehensive experiments we find that: (1) DocRED-FE is challenging to existing JERE models; (2) Our fine-grained entity types promote relation classification. We make DocRED-FE with instruction and the code for our baselines publicly available at https://github.com/PKU-TANGENT/DOCRED-FE.
|
['Sujian Li', 'Yu Xia', 'Dawei Zhu', 'YiFan Song', 'Weimin Xiong', 'Hongbo Wang']
|
2023-03-20
| null | null | null | null |
['relation-classification', 'joint-entity-and-relation-extraction']
|
['natural-language-processing', 'natural-language-processing']
|
[-4.34748411e-01 1.43449321e-01 -5.10747313e-01 -3.83676082e-01
-9.67389405e-01 -6.84958696e-01 6.07115746e-01 3.83753538e-01
-4.64044660e-01 1.04699695e+00 6.00651920e-01 -2.57336855e-01
-1.56423375e-01 -1.01583302e+00 -6.20404959e-01 1.58980843e-02
1.06023103e-01 6.10276222e-01 1.86542511e-01 -3.40903342e-01
-1.22537777e-01 2.36399308e-01 -1.06346560e+00 4.96462673e-01
8.94799590e-01 7.21792758e-01 -1.45184770e-01 6.26680970e-01
-1.26277730e-01 6.84394300e-01 -6.13525689e-01 -7.04863548e-01
7.81496242e-02 1.24367975e-01 -1.43266821e+00 -3.89340073e-01
1.72137409e-01 -1.92525521e-01 -4.91547376e-01 9.66929495e-01
2.62347639e-01 -3.27220969e-02 7.39592969e-01 -1.28721118e+00
-9.85844314e-01 1.12283337e+00 -4.76070136e-01 8.21105763e-02
3.91914457e-01 -1.85238063e-01 1.41947079e+00 -9.38184738e-01
1.02432001e+00 1.20868409e+00 9.32701707e-01 3.76577199e-01
-1.04413486e+00 -6.96241617e-01 2.19843522e-01 2.06183404e-01
-1.66682363e+00 -3.61520141e-01 3.86301935e-01 -4.41469908e-01
1.30324614e+00 2.81826168e-01 2.03573093e-01 1.06840789e+00
5.10419793e-02 8.29782486e-01 1.08963537e+00 -3.48170310e-01
-1.21780664e-01 -1.52723894e-01 7.98662841e-01 5.08580506e-01
7.57363379e-01 -3.60940874e-01 -2.61597693e-01 -7.95017853e-02
4.42285031e-01 -1.61116764e-01 -1.09026976e-01 1.95986584e-01
-1.14784133e+00 4.94353175e-01 3.41483325e-01 5.51443040e-01
-4.44213778e-01 -2.04525784e-01 3.59862536e-01 1.32182658e-01
5.61483920e-01 8.53583932e-01 -1.10676622e+00 -3.14431518e-01
-7.35000014e-01 5.29948533e-01 1.09579980e+00 1.38002455e+00
7.06714213e-01 -6.58367813e-01 -6.10304654e-01 1.01838481e+00
-9.91944596e-03 1.95072219e-01 3.08336079e-01 -7.17578292e-01
8.74913692e-01 8.19267452e-01 1.80861622e-01 -8.28888476e-01
-6.57334328e-01 -2.80925274e-01 -1.02231836e+00 -6.10414088e-01
2.52574176e-01 -5.62469542e-01 -8.95952523e-01 1.46022785e+00
2.58456379e-01 -1.31659463e-01 2.36144170e-01 6.42060637e-01
1.50914562e+00 5.04029930e-01 3.73048067e-01 1.12486981e-01
1.79988158e+00 -1.05068636e+00 -9.48963821e-01 -3.15712065e-01
9.28415596e-01 -5.61376929e-01 8.38241816e-01 -2.71259733e-02
-8.11066210e-01 -2.39044800e-01 -8.96042168e-01 -5.78440368e-01
-8.65429282e-01 3.31299871e-01 8.49499762e-01 2.81015605e-01
-5.76662123e-01 3.86388034e-01 -8.07701468e-01 -2.05652609e-01
3.18500131e-01 1.07818902e-01 -6.68292403e-01 -6.30255789e-02
-1.75381505e+00 9.80378270e-01 7.53560841e-01 4.35673818e-02
-2.37666965e-01 -8.83653402e-01 -1.06396592e+00 1.61393210e-01
8.04218590e-01 -7.97063053e-01 1.42319453e+00 2.56435066e-01
-8.13901663e-01 7.54981279e-01 -4.01393145e-01 -3.81091654e-01
-1.60359610e-02 -5.51960826e-01 -5.61937869e-01 -2.56829143e-01
3.44463348e-01 4.09012139e-01 -6.93991855e-02 -1.13489497e+00
-7.73070037e-01 -1.91884488e-01 2.70028353e-01 9.96108875e-02
-2.31186301e-01 2.01343790e-01 -7.52666414e-01 -9.24414575e-01
-2.72297412e-01 -6.67458534e-01 -1.72352210e-01 -7.45535314e-01
-8.90935123e-01 -7.95068681e-01 4.96870667e-01 -6.83389843e-01
1.97166121e+00 -1.96062374e+00 -2.01138481e-01 -2.03644842e-01
6.22596920e-01 4.65391368e-01 -1.33624151e-01 5.72523355e-01
-1.82579756e-01 5.60632885e-01 -2.57314816e-02 -3.24400455e-01
3.48396689e-01 1.15022451e-01 -2.35965885e-02 -2.02040553e-01
3.00623834e-01 1.33811569e+00 -1.01956582e+00 -6.08467877e-01
-3.66259485e-01 1.97552457e-01 -4.03742135e-01 2.55217463e-01
-1.76239103e-01 -2.06979156e-01 -7.98004925e-01 7.68773615e-01
6.69979513e-01 -3.73535931e-01 1.94477454e-01 -6.53605402e-01
-6.65790029e-03 8.66772711e-01 -1.05552411e+00 1.34523630e+00
-3.17215085e-01 4.08871979e-01 -5.49265742e-02 -5.20998657e-01
7.15417504e-01 2.68061012e-01 4.96069968e-01 -3.37490112e-01
6.73341677e-02 1.76119119e-01 -1.64368600e-01 -3.68809789e-01
1.17292833e+00 1.47751316e-01 -6.16904020e-01 2.24173442e-01
1.48113891e-01 -1.83450833e-01 5.63672602e-01 5.91062427e-01
1.53936553e+00 3.45705561e-02 7.49469936e-01 -6.52704462e-02
-3.19680534e-02 4.18260470e-02 9.22458053e-01 6.29721403e-01
3.43502499e-02 5.54742813e-01 5.28727889e-01 -1.02685094e-01
-8.30500185e-01 -6.16984069e-01 -3.42181593e-01 8.84022593e-01
1.15901986e-02 -1.18658912e+00 -7.19904125e-01 -1.02174854e+00
2.32867122e-01 6.95513964e-01 -6.24049306e-01 1.83901235e-01
-4.48545665e-01 -1.00768602e+00 9.10805106e-01 6.42482221e-01
6.39066875e-01 -1.03375196e+00 8.58447030e-02 4.07049268e-01
-6.21765673e-01 -1.40387762e+00 -4.95145947e-01 3.26788843e-01
-1.55244455e-01 -1.07918751e+00 -3.37939739e-01 -4.59117711e-01
3.54369283e-01 4.42691781e-02 1.69096744e+00 6.75108470e-03
-4.08239365e-02 -8.91247690e-02 -8.04961681e-01 -3.86137456e-01
-5.17758392e-02 7.03284144e-01 -2.09677845e-01 -5.89595199e-01
9.55041051e-01 -2.11946502e-01 -3.09391260e-01 1.97803080e-01
-6.81558609e-01 1.39801160e-01 6.68493807e-01 7.30200171e-01
5.06856024e-01 4.18446898e-01 6.42435610e-01 -1.70854402e+00
8.30767632e-01 -7.29828119e-01 -3.24563920e-01 4.22177315e-01
-8.52036774e-01 1.19859524e-01 4.84915465e-01 -7.79566402e-03
-1.05788267e+00 -3.79779011e-01 -5.32430947e-01 2.05027953e-01
-3.44255090e-01 9.33251023e-01 -3.94440055e-01 5.80843985e-01
5.22320032e-01 -3.53585869e-01 -8.88714433e-01 -7.55127907e-01
5.40376127e-01 1.14373744e+00 6.18218839e-01 -1.03629589e+00
7.76920676e-01 -2.18391225e-01 -3.06428105e-01 -3.24159026e-01
-1.46599233e+00 -5.60204327e-01 -7.39746392e-01 6.10204279e-01
8.22464764e-01 -1.14990962e+00 -4.21529710e-01 5.45708656e-01
-1.17095459e+00 -4.73742515e-01 -2.99711138e-01 2.51651019e-01
4.23481390e-02 1.51257619e-01 -1.18164396e+00 -3.38910818e-01
-4.67102349e-01 -7.72145152e-01 1.34496450e+00 2.37807617e-01
-5.15382230e-01 -9.10814166e-01 1.18266732e-01 4.44285691e-01
3.02169640e-02 1.60931677e-01 9.46335018e-01 -6.73050404e-01
-2.53569007e-01 -1.32699981e-01 -5.85915565e-01 -5.00740670e-02
3.33542794e-01 1.69002600e-02 -5.39976656e-01 1.25881910e-01
-7.11788595e-01 -2.65973866e-01 8.75082910e-01 7.42224008e-02
1.13624239e+00 -5.65041840e-01 -7.04309583e-01 5.08945346e-01
1.12386715e+00 9.63845849e-02 5.69644213e-01 6.05821311e-01
9.36145484e-01 3.75583351e-01 1.01084197e+00 3.59967768e-01
1.15422058e+00 7.54981935e-01 -2.04991043e-01 -2.13006452e-01
-2.08135620e-01 -3.17092717e-01 -2.14065462e-02 8.00187409e-01
-1.06082231e-01 -5.24832189e-01 -9.71047044e-01 8.89418960e-01
-1.98728406e+00 -8.16631377e-01 -2.46324956e-01 1.40659797e+00
1.64212894e+00 1.43708140e-01 5.31057194e-02 -2.58490779e-02
6.42570257e-01 3.05205453e-02 -1.21265978e-01 -2.57326156e-01
-2.32371539e-01 3.37017566e-01 5.98419964e-01 4.18499649e-01
-1.47189546e+00 1.29367006e+00 5.46642876e+00 9.85026121e-01
-5.48700333e-01 6.28807992e-02 3.03810269e-01 2.34458879e-01
-3.32998335e-01 2.90912855e-02 -1.58263266e+00 4.93084371e-01
8.86019468e-01 -3.00838619e-01 9.34381932e-02 6.40964389e-01
-2.40022868e-01 1.32015705e-01 -1.15189981e+00 6.17622375e-01
-4.09072876e-01 -1.31908619e+00 -1.87693015e-01 2.28343029e-02
5.79115033e-01 1.25034871e-02 -4.73116338e-01 8.73251736e-01
1.03945827e+00 -1.01974928e+00 4.87810761e-01 2.64335424e-01
1.02906072e+00 -4.72182602e-01 1.02453804e+00 1.70892194e-01
-1.52837491e+00 2.82073587e-01 -2.20059142e-01 8.18091333e-02
1.95324808e-01 9.44213331e-01 -7.33483076e-01 9.65715528e-01
8.30441356e-01 9.30488765e-01 -8.14905941e-01 6.87243521e-01
-4.27735925e-01 7.38183975e-01 -3.27005029e-01 6.29766732e-02
1.51116122e-02 1.44859746e-01 3.41863304e-01 1.81977403e+00
1.20658509e-01 4.99860585e-01 1.04828939e-01 5.64481974e-01
-6.19755089e-01 -2.50043243e-01 -2.50511527e-01 -2.67610341e-01
1.00610685e+00 1.60024703e+00 -2.16080591e-01 -5.31488955e-01
-5.87260842e-01 6.67034566e-01 6.96278632e-01 2.62977839e-01
-5.19894481e-01 -8.26283336e-01 7.15119481e-01 -1.74332093e-02
2.80935973e-01 -1.16763800e-01 -3.85244727e-01 -1.58340764e+00
1.22468835e-02 -7.58105755e-01 7.88299382e-01 -7.07707465e-01
-1.51760161e+00 8.09208512e-01 3.20300788e-01 -8.14472854e-01
-4.08845007e-01 -5.20387650e-01 -8.97735544e-03 8.00681055e-01
-1.51297212e+00 -1.35773385e+00 -2.16015249e-01 3.04922134e-01
4.03908044e-01 1.51307315e-01 1.09902322e+00 6.13164902e-01
-8.92986596e-01 9.43957031e-01 -1.66296855e-01 8.26525271e-01
8.86770070e-01 -1.65410364e+00 8.70941341e-01 9.73045647e-01
8.46849531e-02 9.82653677e-01 5.52985966e-01 -9.41614628e-01
-1.01035404e+00 -1.51619959e+00 1.51827967e+00 -8.03638339e-01
8.45227659e-01 -5.62626481e-01 -1.08039844e+00 1.11683881e+00
2.90271103e-01 1.76647499e-01 5.62634468e-01 9.58160937e-01
-6.46696150e-01 1.02179497e-01 -1.13312459e+00 5.18541098e-01
1.12479305e+00 -4.65127468e-01 -9.26289499e-01 2.19507620e-01
8.89452696e-01 -6.27609432e-01 -1.60010386e+00 6.34907663e-01
4.19103920e-01 -2.64471024e-01 7.02287257e-01 -7.33518600e-01
4.61142212e-01 -3.31322610e-01 2.02349294e-02 -1.60540605e+00
-6.51054144e-01 -4.61551845e-01 -6.13214731e-01 1.79784989e+00
8.82559597e-01 -5.37218630e-01 5.70215702e-01 7.93656349e-01
-2.82016516e-01 -1.00813591e+00 -3.19207042e-01 -8.27714086e-01
2.57881552e-01 -1.53997272e-01 9.21311438e-01 1.13711274e+00
2.18899235e-01 7.80455172e-01 -1.49542779e-01 1.76389188e-01
2.51768529e-01 3.77270490e-01 6.76105082e-01 -1.13794529e+00
-3.26818258e-01 -2.59870350e-01 -3.22817452e-02 -1.03033006e+00
1.90029874e-01 -7.75637984e-01 5.92790954e-02 -1.97264099e+00
5.02568126e-01 -9.33069229e-01 -2.94138074e-01 1.25026023e+00
-6.76110923e-01 1.87269151e-01 5.30298315e-02 2.01145083e-01
-8.34599018e-01 3.58477354e-01 1.01269245e+00 -2.89196242e-02
-3.95914204e-02 -2.60632336e-01 -1.18421292e+00 4.96665031e-01
6.59926832e-01 -5.86076438e-01 4.84656803e-02 -4.14935321e-01
3.81882399e-01 -1.22426204e-01 -3.68491374e-02 -5.48009872e-01
1.85691595e-01 -3.20142776e-01 2.65806377e-01 -6.30499303e-01
1.71353772e-01 -5.04657865e-01 1.25585184e-01 -1.12611696e-01
-2.60768116e-01 -1.15424670e-01 1.41076177e-01 -8.17925762e-03
-4.78888303e-01 -2.75254130e-01 3.45910549e-01 8.63090083e-02
-8.84514928e-01 4.75560248e-01 -3.34404670e-02 6.19883358e-01
6.79242194e-01 3.28912824e-01 -9.30867136e-01 5.14719523e-02
-5.61637461e-01 5.02426326e-01 3.04329008e-01 5.55395246e-01
1.27180174e-01 -1.31976914e+00 -8.87027800e-01 -2.87316591e-01
4.36811239e-01 4.63872850e-01 3.47501338e-02 5.10346413e-01
-1.62287802e-01 7.30854869e-01 1.37959912e-01 1.96066886e-01
-1.33760548e+00 4.57285911e-01 4.28274609e-02 -9.33863759e-01
-5.38077831e-01 8.59747708e-01 7.29849786e-02 -6.85130477e-01
-2.19721302e-01 -4.05058205e-01 -4.60823327e-01 6.51929108e-03
3.66910517e-01 1.14969909e-01 3.59219432e-01 -4.38037723e-01
-5.72107017e-01 3.50913227e-01 -5.50594628e-01 1.93987921e-01
1.43811595e+00 -1.19100973e-01 -2.57614255e-01 6.74748868e-02
9.76485133e-01 3.80781353e-01 -8.24480295e-01 -4.76472229e-01
3.65975887e-01 -7.23939687e-02 -1.65906139e-02 -1.13049364e+00
-7.74995744e-01 3.29811841e-01 -2.92354703e-01 2.50549495e-01
1.06111097e+00 3.07647467e-01 1.11246443e+00 5.06636679e-01
4.46740359e-01 -9.09449875e-01 -7.15061069e-01 9.21758711e-01
7.55276620e-01 -9.39678133e-01 2.64369845e-01 -9.85726416e-01
-6.09217763e-01 5.22289455e-01 7.53459275e-01 1.77738383e-01
8.56084526e-01 8.15774918e-01 -1.30971923e-01 -2.52132833e-01
-1.07437277e+00 -5.38760424e-01 5.63824832e-01 4.18102831e-01
9.57971871e-01 3.60676169e-01 -6.09097540e-01 1.56260681e+00
-5.32903373e-01 1.84991241e-01 3.93798321e-01 9.84362066e-01
4.58219880e-03 -1.44005787e+00 1.95062965e-01 7.64863312e-01
-7.08583534e-01 -5.47292709e-01 -5.01797676e-01 8.42982411e-01
2.06688076e-01 9.21924174e-01 -2.78563589e-01 -4.85867321e-01
5.67000747e-01 -1.54336557e-01 2.91298956e-01 -1.10726058e+00
-6.87660575e-01 -3.47821951e-01 7.95430362e-01 -3.85061175e-01
-1.14305362e-01 -6.13236785e-01 -1.42869306e+00 -3.45679313e-01
-2.44648859e-01 5.25465131e-01 1.56829745e-01 1.02986026e+00
7.67621577e-01 6.67479157e-01 1.51642025e-01 -2.05809817e-01
-1.19093671e-01 -1.36003435e+00 -5.82734823e-01 5.31827271e-01
1.41888514e-01 -8.20017099e-01 -7.33692665e-03 1.31672528e-02]
|
[9.402658462524414, 8.691110610961914]
|
89c8ecbd-5bfd-4201-bcaf-a268f1bf9ada
|
advpicker-effectively-leveraging-unlabeled
|
2106.02300
| null |
https://arxiv.org/abs/2106.02300v2
|
https://arxiv.org/pdf/2106.02300v2.pdf
|
AdvPicker: Effectively Leveraging Unlabeled Data via Adversarial Discriminator for Cross-Lingual NER
|
Neural methods have been shown to achieve high performance in Named Entity Recognition (NER), but rely on costly high-quality labeled data for training, which is not always available across languages. While previous works have shown that unlabeled data in a target language can be used to improve cross-lingual model performance, we propose a novel adversarial approach (AdvPicker) to better leverage such data and further improve results. We design an adversarial learning framework in which an encoder learns entity domain knowledge from labeled source-language data and better shared features are captured via adversarial training - where a discriminator selects less language-dependent target-language data via similarity to the source language. Experimental results on standard benchmark datasets well demonstrate that the proposed method benefits strongly from this data selection process and outperforms existing state-of-the-art methods; without requiring any additional external resources (e.g., gazetteers or via machine translation). The code is available at https://aka.ms/AdvPicker
|
['Yi Guan', 'Börje F. Karlsson', 'Qianhui Wu', 'Huiqiang Jiang', 'WEILE CHEN']
|
2021-06-04
| null |
https://aclanthology.org/2021.acl-long.61
|
https://aclanthology.org/2021.acl-long.61.pdf
|
acl-2021-5
|
['cross-lingual-ner']
|
['natural-language-processing']
|
[-4.52110805e-02 -4.14062515e-02 -3.97440463e-01 -6.71135843e-01
-1.11579096e+00 -9.55969810e-01 7.24027753e-01 -2.78923452e-01
-8.31845343e-01 9.18399274e-01 1.87966675e-01 -2.88171202e-01
5.00737786e-01 -7.20134377e-01 -8.66565347e-01 -3.28952521e-01
1.84172988e-01 4.80733037e-01 -2.34376073e-01 -3.41674387e-01
-3.00275803e-01 3.46870363e-01 -7.07810044e-01 1.06080823e-01
1.20588851e+00 6.21875525e-01 -1.13834880e-01 2.40137771e-01
-3.80335838e-01 8.05788815e-01 -6.46461904e-01 -9.37950432e-01
5.19395351e-01 -4.42072868e-01 -7.80897856e-01 -4.33125347e-01
4.75905478e-01 -1.85797900e-01 -4.32268679e-01 1.12539673e+00
7.41625130e-01 4.93744276e-02 5.06902516e-01 -1.13274276e+00
-1.33675230e+00 9.26421404e-01 -1.54443204e-01 2.82765087e-02
4.47626114e-02 2.16330215e-01 9.20670569e-01 -1.16651618e+00
7.40411043e-01 9.33666348e-01 7.84419358e-01 1.00575054e+00
-1.09132826e+00 -1.15486681e+00 2.83366684e-02 -1.25949085e-01
-1.57534432e+00 -7.57716715e-01 8.83423388e-01 -1.41728461e-01
7.02976584e-01 -7.83092529e-02 -1.36569351e-01 1.70147455e+00
-2.16022640e-01 9.64657247e-01 1.21462846e+00 -4.67003822e-01
1.73837885e-01 4.34009522e-01 -8.91197622e-02 5.68744421e-01
1.68373168e-01 2.66799033e-01 -4.70625341e-01 -1.27960473e-01
3.72696668e-01 -1.02803871e-01 -3.82126629e-01 -3.20129067e-01
-1.14592254e+00 8.48596573e-01 5.75471461e-01 5.14390349e-01
-3.90899509e-01 -1.94738239e-01 4.53088552e-01 6.08865917e-01
6.38543785e-01 6.65781975e-01 -8.57153296e-01 1.69519894e-02
-9.28950787e-01 -1.63922325e-01 9.14646089e-01 1.10061622e+00
8.62641573e-01 3.21656466e-01 -7.47972578e-02 9.73353088e-01
9.18384790e-02 8.82240176e-01 7.04191685e-01 -5.36236346e-01
9.03676748e-01 6.21138930e-01 5.62142022e-02 -5.28443754e-01
-5.25920503e-02 -2.39943296e-01 -8.71932387e-01 1.24377254e-02
4.35869157e-01 -6.50469303e-01 -9.70714033e-01 2.09060216e+00
2.32023984e-01 3.53811920e-01 5.49163222e-01 6.64165258e-01
8.79284382e-01 5.16088903e-01 3.70904416e-01 2.49728337e-01
1.01894343e+00 -1.08971560e+00 -5.71065128e-01 -5.17797410e-01
7.74923742e-01 -5.78394771e-01 1.10807824e+00 -2.23298799e-02
-7.70405531e-01 -4.23189104e-01 -9.25111473e-01 -2.30356887e-01
-7.94580638e-01 2.58583218e-01 5.28498650e-01 7.28521645e-01
-9.08996046e-01 4.60692763e-01 -8.65768850e-01 -2.80555606e-01
5.60422122e-01 2.79816180e-01 -8.63219082e-01 -2.33314037e-01
-1.66484070e+00 9.88266706e-01 5.06034911e-01 5.97571135e-02
-9.35438871e-01 -8.10465991e-01 -1.07706213e+00 -2.56740361e-01
2.92236835e-01 -5.15976310e-01 1.22660840e+00 -1.34984660e+00
-1.55416012e+00 9.89542246e-01 1.01272799e-01 -4.66676652e-01
6.22452676e-01 -4.62035328e-01 -7.38985538e-01 -9.50439200e-02
1.90333351e-01 6.04668617e-01 6.08022034e-01 -1.16209829e+00
-2.18581408e-01 -2.38800585e-01 1.72030658e-01 9.97654423e-02
-6.92262590e-01 2.44501576e-01 -4.41919953e-01 -8.54618788e-01
-6.31805062e-01 -9.89596248e-01 -2.97645062e-01 -1.92165166e-01
-6.50200427e-01 -2.75101513e-01 5.52970529e-01 -8.10036838e-01
9.94965196e-01 -1.97341537e+00 -9.68436301e-02 -5.74261285e-02
-6.69527948e-02 8.68245125e-01 -3.36056173e-01 4.11685288e-01
-1.15412354e-01 4.70590025e-01 -4.11181122e-01 -5.47143161e-01
3.02753579e-02 3.33302617e-02 -2.74665982e-01 4.05621856e-01
5.34403324e-01 1.01807332e+00 -9.97689784e-01 -3.14816326e-01
2.13600323e-02 7.28386045e-01 -3.04543942e-01 5.15939534e-01
-2.83886250e-02 5.76859117e-01 -6.25052750e-01 6.01181149e-01
6.74984694e-01 -1.74887389e-01 1.85590431e-01 -7.68370926e-02
1.46001071e-01 5.06589949e-01 -1.04527533e+00 1.97065127e+00
-8.61517906e-01 3.67066920e-01 1.49942832e-02 -8.47129464e-01
1.06988251e+00 4.79447931e-01 9.64757130e-02 -6.69012129e-01
9.87272933e-02 4.01312530e-01 -2.55842537e-01 -1.79066405e-01
3.89682829e-01 -1.22125812e-01 -4.92103636e-01 4.17770326e-01
4.06587422e-01 2.29370564e-01 -1.43740568e-02 1.87313259e-01
1.10339034e+00 1.70771390e-01 3.63539696e-01 -5.17317764e-02
5.37849963e-01 1.23907693e-01 9.73471642e-01 6.50464058e-01
-5.40156484e-01 3.69149148e-01 -8.03191364e-02 -1.79471284e-01
-1.09960556e+00 -9.84236240e-01 -9.63956714e-02 1.26694822e+00
5.59698679e-02 -1.26923442e-01 -8.56437683e-01 -1.26621199e+00
2.55285092e-02 9.21836853e-01 -6.91385090e-01 -2.23792478e-01
-6.82922006e-01 -4.57036883e-01 1.25824845e+00 7.51898289e-01
5.53096116e-01 -1.28049910e+00 1.35070011e-01 9.75602642e-02
-4.36680466e-02 -1.20680773e+00 -6.86532676e-01 3.70486498e-01
-4.91364777e-01 -7.24010408e-01 -8.45394075e-01 -8.47575068e-01
6.49576604e-01 -1.26509607e-01 1.42836463e+00 -2.62352943e-01
8.60199034e-02 3.41581076e-01 -4.20275748e-01 -3.80330652e-01
-7.57527649e-01 4.87861574e-01 2.33837530e-01 1.57098453e-02
8.15541565e-01 -3.85507494e-01 -3.68029207e-01 2.20428243e-01
-8.09115410e-01 -1.36519939e-01 7.66205251e-01 1.08039272e+00
6.09743357e-01 -3.47135425e-01 8.91806602e-01 -1.42362046e+00
3.96572471e-01 -8.13102067e-01 -4.55093265e-01 4.13118154e-01
-5.97930193e-01 1.88514471e-01 1.10199654e+00 -5.15226543e-01
-1.23400497e+00 -3.68998013e-02 -1.89099655e-01 -5.87246120e-01
-4.63659078e-01 4.12587285e-01 -5.77014685e-01 -4.65937927e-02
8.21441770e-01 1.52278587e-01 -4.21778381e-01 -6.58616662e-01
6.31619513e-01 9.35602605e-01 5.21628976e-01 -5.34047186e-01
1.07711649e+00 1.49333745e-01 -6.36941552e-01 -2.36468479e-01
-1.03107488e+00 -3.52595419e-01 -7.94767141e-01 3.12907428e-01
7.95626283e-01 -1.33759344e+00 6.42878413e-02 4.50146139e-01
-1.08000147e+00 -3.78263265e-01 -1.36961162e-01 5.28477073e-01
-1.81941077e-01 -8.33946243e-02 -6.02138400e-01 -3.83465677e-01
-6.99038029e-01 -8.60247731e-01 8.29320371e-01 3.90866250e-01
6.04077764e-02 -1.28806770e+00 2.71836758e-01 3.52173209e-01
5.26792228e-01 2.39449084e-01 4.14443970e-01 -1.49394977e+00
-2.91771233e-01 -2.67362922e-01 1.51479691e-01 7.48844862e-01
2.37083882e-01 -3.00067693e-01 -1.04772818e+00 -3.94367546e-01
-2.52654254e-01 -7.52138436e-01 5.90299904e-01 -3.41050237e-01
8.66204262e-01 -3.28929037e-01 -3.14287215e-01 8.11457872e-01
1.51484203e+00 -4.87824194e-02 3.41120631e-01 4.30497289e-01
9.48949397e-01 3.53452951e-01 4.91509438e-01 3.99651565e-02
4.25625056e-01 5.69078088e-01 3.40670198e-02 -4.52243179e-01
-2.58420765e-01 -6.05750978e-01 5.90012729e-01 9.64152634e-01
1.77079543e-01 -4.11184728e-01 -1.18430388e+00 8.39710295e-01
-1.40375233e+00 -8.90356481e-01 3.49471152e-01 2.02156830e+00
1.38242793e+00 -1.47655904e-01 -2.59514928e-01 -4.66010481e-01
8.11477661e-01 9.76076201e-02 -9.84367371e-01 -1.84092149e-01
-2.75816083e-01 5.41897774e-01 8.13196778e-01 2.47427300e-01
-1.37809396e+00 1.36732531e+00 5.63171482e+00 8.26548457e-01
-1.31662118e+00 4.29943591e-01 4.09509450e-01 8.74985233e-02
-3.43687892e-01 -2.12158799e-01 -9.75219429e-01 5.08501828e-01
1.19281006e+00 -4.88351852e-01 3.91188979e-01 1.17134655e+00
-2.35074788e-01 7.91510284e-01 -1.11074626e+00 7.11862087e-01
5.58716767e-02 -1.01029539e+00 -4.28095311e-02 -9.00449678e-02
9.01112676e-01 6.54189408e-01 -9.39924084e-03 7.49209344e-01
1.08250368e+00 -1.06227219e+00 4.26876068e-01 2.84130216e-01
1.05563343e+00 -7.26860881e-01 8.71179819e-01 3.98972303e-01
-1.01922929e+00 3.01506668e-01 -3.86486709e-01 5.43334126e-01
9.48702469e-02 4.33278859e-01 -9.49647605e-01 8.24906945e-01
5.92648447e-01 6.73243463e-01 -6.25427663e-01 5.55127323e-01
-6.58049941e-01 9.73672509e-01 -1.74738839e-01 8.22703019e-02
1.58777222e-01 5.13762906e-02 3.71454984e-01 1.38799191e+00
1.19987905e-01 -2.08805308e-01 3.15565765e-01 8.14339340e-01
-9.00916040e-01 4.59949464e-01 -8.94740880e-01 -4.01700705e-01
8.09508562e-01 1.19556165e+00 -1.38322292e-02 -3.10739905e-01
-8.62296700e-01 1.35349309e+00 7.74756193e-01 3.88120323e-01
-8.55191886e-01 -6.07996166e-01 8.09454918e-01 -2.83774048e-01
3.84526968e-01 -1.83172692e-02 -1.28014311e-01 -1.68447936e+00
-4.12035398e-02 -1.15654898e+00 4.24980819e-01 -4.33384866e-01
-1.86026442e+00 1.00558770e+00 -5.61621308e-01 -1.27485299e+00
-3.58228803e-01 -6.18527234e-01 -4.70111430e-01 1.18969154e+00
-1.78562760e+00 -1.50428104e+00 5.06841987e-02 9.30920422e-01
3.96363318e-01 -5.86816370e-01 1.20323479e+00 5.26889503e-01
-6.46751940e-01 1.26829338e+00 5.15779197e-01 9.69971120e-01
1.22181880e+00 -1.38206673e+00 6.98663235e-01 1.10642886e+00
4.83023196e-01 9.14823413e-01 1.57411069e-01 -5.54423749e-01
-1.35269761e+00 -1.51621175e+00 1.10879135e+00 -6.76695466e-01
7.21354783e-01 -5.10730267e-01 -1.00510240e+00 9.37329769e-01
4.60524797e-01 4.69765663e-01 1.24834490e+00 2.63564706e-01
-7.93592811e-01 9.30103939e-03 -1.30635238e+00 4.37514305e-01
1.03742528e+00 -8.37130189e-01 -8.10425699e-01 1.93216830e-01
7.97283590e-01 -3.65472138e-01 -1.09212840e+00 3.63458276e-01
1.26939088e-01 -4.95584577e-01 8.77591491e-01 -1.04487503e+00
3.10251027e-01 -3.79659414e-01 -6.82215393e-02 -1.60932815e+00
-7.57208616e-02 -4.78679240e-01 4.53994907e-02 1.74062729e+00
7.13054895e-01 -7.35922873e-01 4.70618755e-01 7.99194455e-01
-6.42916337e-02 -4.72666770e-01 -7.72132277e-01 -1.05505443e+00
6.19113207e-01 -2.32904226e-01 7.40669191e-01 1.61993051e+00
-4.20052320e-01 4.67433006e-01 -4.77177322e-01 4.56554234e-01
5.31851411e-01 2.95243226e-03 8.01257372e-01 -9.27127361e-01
-1.36832893e-01 -5.00127561e-02 -2.02507392e-01 -8.09123814e-01
8.60626519e-01 -1.33562875e+00 1.09553717e-01 -1.22015858e+00
-1.40978783e-01 -7.90691912e-01 -8.20402861e-01 9.82423186e-01
-4.88388360e-01 3.12316895e-01 2.04492599e-01 2.74089605e-01
-6.48581922e-01 6.96664810e-01 9.30948913e-01 -1.90276206e-01
2.04827823e-02 -2.65463263e-01 -8.93625498e-01 6.66938841e-01
8.00115585e-01 -8.22858393e-01 -2.00478598e-01 -8.03359568e-01
-1.37958512e-01 -3.05311620e-01 2.22788174e-02 -8.26211452e-01
1.43253401e-01 -6.37730435e-02 4.14762020e-01 4.92042042e-02
-4.52802330e-02 -7.70944893e-01 -9.03210491e-02 4.13519517e-02
-5.40632963e-01 -1.41206365e-02 2.04407364e-01 4.71144080e-01
-4.10381794e-01 -3.22847784e-01 7.98466265e-01 -6.12575188e-02
-9.46570396e-01 5.70001245e-01 2.75645733e-01 5.59841990e-01
9.68959153e-01 3.85903955e-01 -3.09612304e-01 -1.46476358e-01
-4.85936493e-01 1.94349915e-01 7.01363802e-01 7.20306993e-01
1.40139926e-02 -1.61810482e+00 -9.96515214e-01 2.38494035e-02
2.95574665e-01 -1.73602670e-01 -1.12348758e-01 1.77343398e-01
-1.82909608e-01 2.09017307e-01 -7.04085678e-02 -1.22764438e-01
-9.22573149e-01 5.83004177e-01 4.26795810e-01 -4.89234805e-01
-2.96277165e-01 9.92612660e-01 6.19063042e-02 -1.19730425e+00
-5.20920195e-02 2.23749027e-01 -4.81224880e-02 -1.95512533e-01
5.69385588e-01 -8.16864148e-02 9.02792811e-02 -7.88869202e-01
-5.97735524e-01 2.58232832e-01 -3.14557970e-01 -1.77538581e-02
1.37904489e+00 -1.67653356e-02 3.78327221e-01 3.20929259e-01
1.36372030e+00 5.58112085e-01 -1.10061347e+00 -7.12870121e-01
7.90575668e-02 -2.11595669e-01 -1.43755451e-02 -1.12830329e+00
-1.24689317e+00 8.11144054e-01 5.40407777e-01 -1.25655085e-01
1.12845469e+00 -9.05609801e-02 1.01702380e+00 5.19452751e-01
5.32754600e-01 -9.57300723e-01 -2.55746603e-01 6.01154447e-01
6.43109322e-01 -1.70315504e+00 -4.82005507e-01 -2.13281959e-01
-9.32621419e-01 7.24090278e-01 7.34662056e-01 -1.97921380e-01
6.15890920e-01 4.33999062e-01 7.13478446e-01 2.89862961e-01
-5.18906355e-01 -2.90827513e-01 2.23609000e-01 7.72019625e-01
7.45037496e-01 1.40661433e-01 -8.78780708e-02 1.03385985e+00
-1.87754318e-01 -1.15761617e-02 1.48443431e-01 8.39286923e-01
2.41504565e-01 -1.71235120e+00 -5.78981973e-02 9.53915939e-02
-9.04375076e-01 -6.18443966e-01 -5.21702945e-01 7.72864342e-01
2.86630660e-01 7.84029305e-01 -2.11618513e-01 -2.64601260e-01
4.23563063e-01 3.93328547e-01 -3.46041806e-02 -7.32426226e-01
-9.40889835e-01 -6.48905218e-01 2.13607386e-01 -5.24551094e-01
-4.48861837e-01 -5.00037551e-01 -1.10210466e+00 -1.89565420e-01
-2.98652947e-01 4.14792925e-01 6.85747445e-01 7.20752239e-01
7.90368974e-01 2.41242170e-01 7.50356972e-01 -3.27087075e-01
-6.90296590e-01 -1.05566263e+00 -2.40069643e-01 7.68301070e-01
2.32172638e-01 -4.52394545e-01 -3.48006785e-01 3.11841398e-01]
|
[10.022905349731445, 9.653237342834473]
|
48da4b32-4991-453a-bb99-6a9825bebfad
|
compound-prototype-matching-for-few-shot
|
2207.05515
| null |
https://arxiv.org/abs/2207.05515v5
|
https://arxiv.org/pdf/2207.05515v5.pdf
|
Compound Prototype Matching for Few-shot Action Recognition
|
Few-shot action recognition aims to recognize novel action classes using only a small number of labeled training samples. In this work, we propose a novel approach that first summarizes each video into compound prototypes consisting of a group of global prototypes and a group of focused prototypes, and then compares video similarity based on the prototypes. Each global prototype is encouraged to summarize a specific aspect from the entire video, for example, the start/evolution of the action. Since no clear annotation is provided for the global prototypes, we use a group of focused prototypes to focus on certain timestamps in the video. We compare video similarity by matching the compound prototypes between the support and query videos. The global prototypes are directly matched to compare videos from the same perspective, for example, to compare whether two actions start similarly. For the focused prototypes, since actions have various temporal variations in the videos, we apply bipartite matching to allow the comparison of actions with different temporal positions and shifts. Experiments demonstrate that our proposed method achieves state-of-the-art results on multiple benchmarks.
|
['Lijin Yang', 'Yifei HUANG', 'Yoichi Sato']
|
2022-07-12
| null | null | null | null |
['few-shot-action-recognition', 'video-similarity']
|
['computer-vision', 'computer-vision']
|
[ 1.68258280e-01 -3.83406490e-01 -4.37582046e-01 -3.94062191e-01
-5.32338381e-01 -5.04607737e-01 5.51449418e-01 3.35597456e-01
-2.59433717e-01 5.25158525e-01 3.44467461e-01 3.63858819e-01
-8.41447040e-02 -3.98018718e-01 -5.23013949e-01 -7.58920014e-01
-2.62185037e-01 2.53542334e-01 8.67626727e-01 2.03672975e-01
4.16709542e-01 5.96884549e-01 -1.70521355e+00 7.21422851e-01
2.61716872e-01 1.04224217e+00 3.28205705e-01 5.31766534e-01
-1.38585284e-01 1.02636170e+00 -5.74132621e-01 -1.08227413e-02
3.05642784e-01 -7.73041546e-01 -7.62101412e-01 5.32375276e-01
7.76476204e-01 -4.42363918e-01 -5.78809261e-01 1.06532741e+00
1.25132859e-01 6.71525240e-01 3.34116697e-01 -1.50922954e+00
-1.50099143e-01 3.79627615e-01 -5.60000598e-01 5.43709993e-01
7.47800410e-01 1.69064291e-02 9.20161784e-01 -7.89076030e-01
1.06033063e+00 9.74023879e-01 3.52391183e-01 3.74468088e-01
-9.16289270e-01 -5.59852540e-01 6.45632088e-01 9.74039435e-01
-1.31777370e+00 -6.35362625e-01 7.86518633e-01 -5.41509032e-01
7.18166947e-01 2.12741449e-01 7.55459309e-01 7.44386137e-01
-5.22900708e-02 7.73185194e-01 6.58359349e-01 -1.86699271e-01
6.02119863e-01 -1.20686702e-01 2.96476811e-01 6.17481768e-01
-9.60591659e-02 -2.86100060e-01 -5.09256959e-01 -2.03729793e-01
2.96950817e-01 6.08166814e-01 -4.44396913e-01 -6.67510867e-01
-1.52560496e+00 5.11228204e-01 1.67090356e-01 4.94849026e-01
-4.89898503e-01 6.51267171e-02 6.68774545e-01 2.48430356e-01
-1.34637654e-01 1.67742103e-01 -2.57703990e-01 -2.27566734e-01
-1.16486311e+00 -2.78536789e-02 6.50632381e-01 1.03363287e+00
1.11518145e+00 -3.91063571e-01 -4.91548449e-01 5.29059410e-01
-6.55540973e-02 9.55008939e-02 5.81699908e-01 -1.17869282e+00
4.66427028e-01 6.33418620e-01 3.15324694e-01 -1.27305496e+00
-5.23210429e-02 2.47572124e-01 -4.98509884e-01 1.79600713e-04
3.95010144e-01 2.72771388e-01 -7.70149469e-01 1.48078597e+00
4.94380623e-01 6.59091890e-01 -7.20537603e-02 9.15356159e-01
7.91637599e-01 7.96682477e-01 -7.12496415e-02 -6.54610038e-01
1.13644540e+00 -1.35024691e+00 -5.77033639e-01 -4.68799053e-03
6.72108412e-01 -6.82662904e-01 5.86909711e-01 1.02049798e-01
-1.02756345e+00 -8.69491458e-01 -8.65209520e-01 3.65500599e-01
-1.88323215e-01 2.32797235e-01 1.00791074e-01 7.55537227e-02
-6.67126596e-01 8.62060547e-01 -9.87009823e-01 -7.56170928e-01
1.09964535e-01 -2.69534159e-02 -5.21860838e-01 -3.27103287e-01
-8.82804632e-01 5.09532213e-01 5.69626510e-01 -3.28629851e-01
-9.10240114e-01 -3.27940524e-01 -8.74710739e-01 2.74039786e-02
6.76679075e-01 -2.22278923e-01 1.02891397e+00 -1.24033070e+00
-1.20861375e+00 7.14250028e-01 -3.85361731e-01 -3.21700364e-01
3.95450920e-01 1.74752306e-02 -6.79834127e-01 7.27674961e-01
2.64240295e-01 5.77456653e-01 9.57920790e-01 -8.38239908e-01
-1.28691721e+00 -1.58145502e-01 2.59913206e-01 1.74769536e-01
-3.17802697e-01 3.46787125e-01 -9.16620612e-01 -5.30475318e-01
2.00581357e-01 -9.57507908e-01 3.46589722e-02 2.17400908e-01
-7.17286021e-02 -1.14787832e-01 1.09445572e+00 -5.44915676e-01
1.43190706e+00 -2.38504982e+00 1.39275044e-01 1.01477616e-01
-3.74858975e-02 1.92540720e-01 -1.79576099e-01 6.25623882e-01
-2.22151369e-01 -2.62345344e-01 -1.46494526e-02 -6.08464144e-02
-3.74739498e-01 2.11191759e-01 -2.04983994e-01 5.61974943e-01
-1.49176300e-01 4.47703987e-01 -1.27488315e+00 -7.74771392e-01
1.77894011e-01 3.03500220e-02 -2.95699209e-01 3.54424536e-01
-5.83146065e-02 5.28031290e-01 -4.30312961e-01 6.38047814e-01
4.14651722e-01 -5.81431538e-02 4.24196601e-01 -4.46876526e-01
-2.54058540e-01 5.55842258e-02 -1.54727018e+00 1.67351210e+00
8.83443728e-02 5.91761053e-01 -2.47547641e-01 -1.23798788e+00
7.77831137e-01 3.41118991e-01 7.63501287e-01 -3.42614055e-01
-2.22847208e-01 1.08919758e-02 -7.36114830e-02 -7.49793768e-01
4.07302141e-01 2.23493993e-01 1.83867320e-01 6.51635528e-01
8.55865851e-02 3.89400512e-01 8.93694639e-01 1.97876334e-01
1.26501226e+00 1.61797911e-01 5.21692574e-01 2.38460183e-01
7.96764135e-01 -5.57126813e-02 8.45315397e-01 6.76662147e-01
-5.05606532e-01 5.96662402e-01 3.68065774e-01 -5.19989312e-01
-8.00672174e-01 -8.48887682e-01 4.35066670e-01 1.28504670e+00
4.58747566e-01 -7.18420625e-01 -3.89860839e-01 -9.91760910e-01
-1.30385101e-01 3.50131452e-01 -6.88156128e-01 -1.04673021e-01
-7.63283372e-01 1.19332895e-01 3.18078287e-02 7.02901363e-01
4.86112356e-01 -1.13073552e+00 -1.07950699e+00 2.03841865e-01
-2.29586497e-01 -9.69726264e-01 -9.21575785e-01 -1.58437759e-01
-8.62454534e-01 -1.30115187e+00 -8.65039647e-01 -9.12708700e-01
6.80271864e-01 8.13343465e-01 7.21945107e-01 4.38759662e-02
-6.16301857e-02 6.28281951e-01 -7.13731050e-01 3.61949384e-01
-3.71525109e-01 -4.55815703e-01 1.55477971e-01 4.78666455e-01
2.16982394e-01 -4.58970487e-01 -6.18849516e-01 8.63444030e-01
-7.86165714e-01 -1.74538314e-01 3.13601971e-01 6.83759868e-01
8.07315350e-01 4.27961349e-02 1.67100906e-01 -3.57325166e-01
7.16778785e-02 -4.65983868e-01 -2.98082232e-01 7.31713057e-01
1.00368680e-02 -2.24893212e-01 6.68958366e-01 -8.51407170e-01
-7.99585521e-01 2.90284842e-01 6.02042437e-01 -9.69960809e-01
-4.06949878e-01 4.95597064e-01 -9.94187370e-02 2.10215837e-01
3.58604044e-01 3.08176756e-01 -2.42431641e-01 -3.13158959e-01
4.56353992e-01 5.31976819e-01 6.11171484e-01 -2.29867756e-01
4.37258482e-01 6.16252661e-01 -2.07404509e-01 -7.97607839e-01
-4.84310985e-01 -1.06480038e+00 -8.89924049e-01 -6.27055168e-01
7.93318808e-01 -6.45169675e-01 -3.32494587e-01 3.12786877e-01
-9.18427169e-01 -6.97919056e-02 -4.82019961e-01 8.20612729e-01
-6.89726591e-01 8.67530942e-01 -3.80457997e-01 -5.42246819e-01
2.77919453e-02 -1.08946431e+00 7.48011410e-01 3.45824182e-01
-3.16562623e-01 -6.02017045e-01 2.95030326e-01 8.39537084e-02
-2.00496003e-01 8.94277692e-02 5.67849755e-01 -9.67937052e-01
-5.77198327e-01 -3.32180351e-01 7.84713626e-02 2.97387213e-01
5.74377060e-01 3.43308032e-01 -4.26309526e-01 -5.02059519e-01
-7.75425583e-02 -9.15581137e-02 7.34816253e-01 1.59903213e-01
8.40762019e-01 -2.28628397e-01 -7.81974018e-01 2.68326998e-01
9.97590840e-01 8.24251592e-01 6.25902832e-01 1.29884958e-01
5.83970845e-01 6.01592660e-01 1.17666841e+00 5.28194308e-01
-7.36862049e-02 8.26278985e-01 7.93559030e-02 4.83289659e-01
6.94245892e-03 -4.45743278e-02 6.53555036e-01 7.41126657e-01
-1.49610072e-01 1.15357351e-03 -5.49581230e-01 5.53396404e-01
-2.17831278e+00 -1.59670317e+00 1.95685506e-01 2.52730703e+00
3.27179164e-01 2.53078416e-02 3.76924574e-01 4.95822309e-03
1.31136906e+00 5.36642373e-01 -7.27257133e-01 -2.87604742e-02
2.03699932e-01 -1.38448894e-01 3.57855745e-02 -6.50297552e-02
-1.20727849e+00 5.92502177e-01 5.74458694e+00 7.32562244e-01
-1.04621696e+00 -1.92966685e-02 2.09178776e-01 -3.02589893e-01
2.82861233e-01 2.21705422e-01 -6.76590204e-01 6.46937907e-01
5.53403139e-01 -3.58894497e-01 2.34231696e-01 8.83164167e-01
2.35434830e-01 -4.55820948e-01 -1.47900116e+00 1.05616546e+00
5.05552173e-01 -1.30994141e+00 3.48452479e-02 -3.12965721e-01
8.85610759e-01 -2.60608077e-01 -4.51977968e-01 2.15232342e-01
-3.09154183e-01 -2.46219501e-01 7.65389740e-01 7.30386555e-01
3.90749365e-01 -6.15025103e-01 3.79549891e-01 2.80074626e-01
-1.66288316e+00 -2.68451512e-01 -4.68177289e-01 4.18894626e-02
1.51115745e-01 2.00497806e-01 -5.08148968e-01 6.37148023e-01
8.21924686e-01 1.23031878e+00 -4.50653464e-01 1.38112879e+00
-3.98963578e-02 2.86246598e-01 -5.45187891e-02 1.20135896e-01
2.09478900e-01 -4.38080132e-01 6.32580101e-01 1.06981564e+00
4.72818643e-01 2.68546849e-01 6.23885393e-01 3.21653247e-01
1.49412230e-01 2.48733193e-01 -4.91351068e-01 -1.66232139e-01
5.16845644e-01 1.23296046e+00 -8.94133747e-01 -8.81533742e-01
-7.95286179e-01 1.20144248e+00 1.72376409e-01 2.91746706e-01
-8.27407539e-01 -4.49929327e-01 7.42232978e-01 -1.49930390e-02
7.95835674e-01 -1.15439385e-01 7.52176106e-01 -1.17934763e+00
1.21196911e-01 -7.19034255e-01 9.00701463e-01 -9.20588911e-01
-1.02868712e+00 4.05773193e-01 3.12282056e-01 -1.92408335e+00
-3.36116165e-01 -8.93057957e-02 -8.52000475e-01 3.07265073e-01
-9.43561137e-01 -6.57212198e-01 -5.22672474e-01 8.93135488e-01
7.38610506e-01 -9.82046574e-02 4.41919029e-01 2.41351351e-01
-4.84104544e-01 2.71692991e-01 3.18772733e-01 1.61958918e-01
1.03186429e+00 -7.02609241e-01 8.31364095e-02 8.65361512e-01
3.50750238e-01 5.35520911e-01 6.25378549e-01 -8.47751200e-01
-1.07202554e+00 -1.07616866e+00 7.43548930e-01 2.77455281e-02
7.39792287e-01 1.67282432e-01 -1.04170573e+00 7.12756395e-01
2.24134363e-02 1.70942828e-01 7.80466735e-01 -3.78532112e-01
-3.52899462e-01 -2.76396006e-01 -8.79021645e-01 6.32093608e-01
1.14738655e+00 -4.53818470e-01 -9.81329262e-01 3.73252869e-01
4.67904449e-01 -6.77806810e-02 -7.42819190e-01 3.18497300e-01
7.20356822e-01 -1.21399832e+00 6.67916954e-01 -8.52403700e-01
2.00981975e-01 -7.84279644e-01 -2.53911316e-01 -1.17460823e+00
-5.04458427e-01 -4.18923050e-01 -2.47493684e-01 1.21052086e+00
-1.69543579e-01 -3.26536030e-01 8.40196431e-01 3.12209487e-01
-1.51985034e-01 -5.57404041e-01 -9.16289032e-01 -1.14807367e+00
-8.02510738e-01 -1.02286130e-01 3.83888632e-01 9.66294050e-01
3.68084908e-01 -1.25945807e-01 -5.61875165e-01 -1.74364708e-02
2.03112200e-01 7.30065167e-01 7.93447673e-01 -9.77771521e-01
-3.60399097e-01 -2.88776547e-01 -9.69991207e-01 -1.09105229e+00
1.26457483e-01 -6.71493232e-01 9.97666046e-02 -1.39255142e+00
5.05492687e-01 1.14260837e-01 -5.79677820e-01 4.46979672e-01
-7.36023784e-02 1.85000435e-01 3.90428752e-01 4.74161923e-01
-1.08153415e+00 3.55747610e-01 8.73380959e-01 -3.27705175e-01
-3.66490096e-01 5.53661361e-02 -2.30279155e-02 6.79006934e-01
5.36497474e-01 -4.40148681e-01 -4.69586104e-01 6.92594722e-02
-3.51888090e-01 2.86637336e-01 4.53333296e-02 -1.46447432e+00
4.71884310e-01 -4.92008746e-01 2.47953072e-01 -6.95291817e-01
2.47249410e-01 -9.38783765e-01 4.73312318e-01 6.44584417e-01
-3.95644546e-01 -3.91053408e-02 -1.66317597e-01 8.53031218e-01
-4.11558896e-01 -3.99671435e-01 7.39834011e-01 -2.10370660e-01
-1.22303033e+00 3.92560899e-01 -3.39687854e-01 -1.48015663e-01
1.57908010e+00 -7.64474750e-01 -2.06116170e-01 -4.69503939e-01
-1.05325270e+00 2.18372777e-01 7.11449921e-01 4.68189746e-01
7.29924619e-01 -1.53605843e+00 -2.96279311e-01 -6.58630729e-02
5.75291097e-01 -6.74030542e-01 4.57784384e-01 9.80378747e-01
-1.78800449e-01 1.29829735e-01 -5.04093468e-01 -6.32370591e-01
-1.80355310e+00 1.17907405e+00 1.43729150e-01 5.34348451e-02
-7.31115639e-01 4.42105174e-01 2.95286238e-01 2.85195082e-01
3.71577084e-01 -4.16904420e-01 -3.90104741e-01 5.36848366e-01
8.88502419e-01 6.72111392e-01 -1.88626856e-01 -9.77686405e-01
-5.75052440e-01 7.22161889e-01 -7.56502151e-02 8.67226571e-02
1.01546597e+00 -1.16583705e-01 2.26290338e-02 7.84568489e-01
1.34070575e+00 -1.76074579e-01 -1.45352328e+00 -3.63504887e-01
-1.43618239e-02 -8.98253202e-01 -5.37993848e-01 -3.35923284e-01
-1.17032647e+00 5.12468517e-01 5.83404183e-01 -3.05914320e-03
1.24622750e+00 1.75398588e-01 6.76180422e-01 5.73834002e-01
5.06926477e-01 -1.18484902e+00 3.27399582e-01 4.29918557e-01
6.79576337e-01 -9.21929181e-01 1.17683209e-01 -1.87637135e-01
-8.28134000e-01 1.28269625e+00 4.10224408e-01 2.70679519e-02
4.31754380e-01 -3.24211925e-01 -1.74208432e-01 -1.09852016e-01
-5.93033016e-01 -2.40367204e-01 2.86060959e-01 3.81632060e-01
-2.48692762e-02 -1.83759540e-01 -4.90153551e-01 9.21242833e-02
5.65866649e-01 7.60906786e-02 4.64641035e-01 1.28188515e+00
-6.92041159e-01 -9.70009267e-01 -3.58106643e-01 4.39719439e-01
8.58231634e-03 2.60684371e-01 -3.83860528e-01 6.01282835e-01
1.03512540e-01 8.93020511e-01 4.97364700e-01 -5.40384650e-01
4.46362019e-01 2.84327418e-01 5.07704675e-01 -6.69213414e-01
-3.30178022e-01 1.32879943e-01 5.26327407e-03 -8.86850774e-01
-8.28970432e-01 -1.03467715e+00 -1.26406622e+00 3.50663774e-02
-1.91623539e-01 2.95381069e-01 2.19184697e-01 9.19405937e-01
3.70629877e-01 9.40088630e-02 8.65482748e-01 -9.58142698e-01
-1.89950034e-01 -7.53497481e-01 -8.10972691e-01 7.96133161e-01
3.23796757e-02 -8.01489234e-01 -3.52851480e-01 4.68571842e-01]
|
[8.533966064453125, 0.6552910208702087]
|
64e3796b-6e31-43d5-b5e8-b292151304c4
|
unsupervised-learning-of-fine-structure-1
| null | null |
http://openaccess.thecvf.com//content/ICCV2021/html/Chen_Unsupervised_Learning_of_Fine_Structure_Generation_for_3D_Point_Clouds_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Chen_Unsupervised_Learning_of_Fine_Structure_Generation_for_3D_Point_Clouds_ICCV_2021_paper.pdf
|
Unsupervised Learning of Fine Structure Generation for 3D Point Clouds by 2D Projections Matching
|
Learning to generate 3D point clouds without 3D supervision is an important but challenging problem. Current solutions leverage various differentiable renderers to project the generated 3D point clouds onto a 2D image plane, and train deep neural networks using the per-pixel difference with 2D ground truth images. However, these solutions are still struggling to fully recover fine structures of 3D shapes, such as thin tubes or planes. To resolve this issue, we propose an unsupervised approach for 3D point cloud generation with fine structures. Specifically, we cast 3D point cloud learning as a 2D projection matching problem. Rather than using entire 2D silhouette images as a regular pixel supervision, we introduce structure adaptive sampling to randomly sample 2D points within the silhouettes as an irregular point supervision, which alleviates the consistency issue of sampling from different view angles. Our method pushes the neural network to generate a 3D point cloud whose 2D projections match the irregular point supervision from different view angles. Our 2D projection matching approach enables the neural network to learn more accurate structure information than using the per-pixel difference, especially for fine and thin 3D structures. Our method can recover fine 3D structures from 2D silhouette images at different resolutions, and is robust to different sampling methods and point number in irregular point supervision. Our method outperforms others under widely used benchmarks. Our code, data and models are available at https://github.com/chenchao15/2D_projection_matching.
|
['Matthias Zwicker', 'Yu-Shen Liu', 'Zhizhong Han', 'Chao Chen']
|
2021-01-01
| null | null | null |
iccv-2021-1
|
['point-cloud-generation']
|
['computer-vision']
|
[-2.27558389e-02 1.69240370e-01 2.57405676e-02 -1.80135533e-01
-8.22767794e-01 -6.78530633e-01 6.71007693e-01 -3.55156153e-01
7.66514018e-02 3.11732680e-01 -1.35140836e-01 -3.13171089e-01
3.69230598e-01 -1.11406326e+00 -1.22361732e+00 -4.59243029e-01
3.14002037e-01 9.97162879e-01 1.24069884e-01 -1.61360174e-01
3.24722379e-01 1.12172496e+00 -1.58206534e+00 3.32991295e-02
9.08511102e-01 8.99034858e-01 3.40264551e-02 4.16629732e-01
-5.20215213e-01 -2.25522280e-01 -4.48303163e-01 -1.92413628e-01
1.02338040e+00 1.77687809e-01 -5.06176710e-01 3.80143523e-01
1.16710341e+00 -6.69900477e-01 1.23260029e-01 1.04027355e+00
3.43722761e-01 -1.80023149e-01 8.31069231e-01 -1.10514367e+00
-6.17076814e-01 -1.08677514e-01 -9.75113928e-01 -4.65156913e-01
4.41382557e-01 2.24413484e-01 6.70473278e-01 -1.28807330e+00
6.54614925e-01 1.43159938e+00 9.91437256e-01 5.49764574e-01
-1.31317127e+00 -7.40565956e-01 5.33619002e-02 -7.84997642e-01
-1.24459159e+00 -5.96321523e-02 1.19825947e+00 -7.68362880e-01
8.90555263e-01 4.58968319e-02 7.96355426e-01 8.97255242e-01
1.25064449e-02 5.20640373e-01 1.04756200e+00 -2.31701389e-01
5.19173779e-03 -2.22168535e-01 -1.23219751e-01 7.02383220e-01
3.02661955e-01 2.79623955e-01 -2.31202066e-01 -2.90291846e-01
1.77857852e+00 4.83010143e-01 -3.35601270e-01 -7.88380802e-01
-1.31700575e+00 6.72618985e-01 6.96544290e-01 -1.76730022e-01
-2.86936402e-01 1.65047809e-01 -1.12053461e-01 6.64333180e-02
8.04196298e-01 5.27419746e-01 -5.39570987e-01 1.93020880e-01
-7.69798696e-01 5.01369238e-01 5.13347089e-01 1.35463917e+00
1.22907805e+00 5.39432988e-02 2.89611816e-01 5.32121837e-01
3.58139992e-01 8.43524694e-01 3.55329476e-02 -1.48687756e+00
5.95420122e-01 9.17313039e-01 1.51515409e-01 -1.00592232e+00
-1.38002962e-01 -1.86354831e-01 -9.79917645e-01 8.10594261e-01
3.05238843e-01 2.32025869e-02 -1.26285934e+00 1.14945745e+00
6.69429719e-01 1.08908147e-01 -3.10414612e-01 9.81429458e-01
7.86356449e-01 5.91637254e-01 -8.23914349e-01 3.72968614e-01
1.07140994e+00 -7.18607306e-01 6.72060102e-02 -4.22612093e-02
3.48742783e-01 -8.30474973e-01 1.11342907e+00 1.44489750e-01
-1.38345158e+00 -5.14052927e-01 -9.14497852e-01 -3.61762553e-01
1.60706639e-02 -1.89910337e-01 3.21087778e-01 1.39501497e-01
-1.03467488e+00 7.75219977e-01 -8.88999164e-01 7.72137642e-02
7.44340718e-01 3.06806654e-01 -3.90028507e-01 4.39200364e-03
-5.70150316e-01 4.23738271e-01 7.40479901e-02 -1.76488176e-01
-6.47293210e-01 -1.21055329e+00 -8.38476181e-01 -1.74087301e-01
1.51993170e-01 -1.15350664e+00 1.13482368e+00 -5.94096661e-01
-1.39870322e+00 1.40177262e+00 -2.09318131e-01 -1.96215764e-01
6.97699606e-01 -1.99963242e-01 3.93460363e-01 1.25945672e-01
2.58723021e-01 8.72501612e-01 9.20316935e-01 -1.79241490e+00
-2.75765628e-01 -6.29953086e-01 -6.73931465e-02 3.79240632e-01
4.63181734e-01 -5.72642148e-01 -2.72118568e-01 -4.01286483e-01
6.90208137e-01 -6.85821831e-01 -2.84707129e-01 4.04386520e-01
-5.46250701e-01 -1.10394664e-01 9.82196391e-01 -1.88221529e-01
2.24896893e-01 -2.04894376e+00 -1.21075124e-01 2.36772314e-01
5.19655824e-01 3.62253077e-02 2.91938186e-02 6.64988235e-02
-5.16101867e-02 3.32901716e-01 -3.98485988e-01 -6.70059204e-01
-1.84056573e-02 1.64484769e-01 -5.36730707e-01 2.74973452e-01
4.43398803e-01 8.59892726e-01 -7.32166111e-01 -1.06091514e-01
4.59688246e-01 7.58055329e-01 -6.36069357e-01 1.97926372e-01
-5.21961570e-01 5.63766360e-01 -5.82896531e-01 6.53947055e-01
1.16756666e+00 -5.34569502e-01 -5.38672507e-01 -2.38001689e-01
-1.54523000e-01 3.54564875e-01 -1.10863817e+00 1.87125897e+00
-4.00459290e-01 3.38508129e-01 1.61328241e-01 -5.45738459e-01
1.36923826e+00 2.46339232e-01 5.83232641e-01 -2.97428727e-01
-4.01942320e-02 4.27052349e-01 -5.08753300e-01 5.58377057e-02
3.80453974e-01 -2.08611652e-01 2.36910790e-01 3.68746161e-01
-2.87584037e-01 -1.03327680e+00 -4.47830200e-01 2.84190550e-02
7.50680387e-01 4.80835050e-01 2.08800156e-02 2.79955920e-02
5.08786626e-02 1.22606665e-01 5.03498614e-01 3.42439085e-01
2.93822169e-01 1.29975963e+00 4.45103645e-01 -8.70600343e-01
-1.66739106e+00 -1.41889346e+00 -3.37438703e-01 2.25467817e-03
2.82779306e-01 -1.99670538e-01 -5.76935828e-01 -5.98071456e-01
2.76458949e-01 3.55978727e-01 -3.24760228e-01 3.11113894e-01
-8.19337726e-01 -1.63388506e-01 1.36756063e-01 3.78299147e-01
4.54328030e-01 -8.14757109e-01 -6.79085791e-01 -1.50639325e-01
8.16239789e-02 -1.01132083e+00 -5.27401924e-01 1.42762721e-01
-1.31678057e+00 -1.10296798e+00 -9.19154584e-01 -6.15141749e-01
1.20524681e+00 5.47730505e-01 1.56904936e+00 1.34460583e-01
5.10316156e-02 3.14385712e-01 -4.30768169e-02 -4.21421289e-01
-3.36312383e-01 4.41716015e-02 -4.67608869e-03 -3.54262233e-01
2.35111788e-01 -9.67649400e-01 -6.70843840e-01 5.16034663e-01
-7.94427752e-01 5.57694376e-01 2.87047029e-01 7.17074096e-01
1.21923172e+00 -2.90817410e-01 -1.29378572e-01 -8.72147977e-01
2.77970016e-01 -2.23278195e-01 -9.89813268e-01 -3.13906133e-01
-3.02905530e-01 8.56099799e-02 4.85406458e-01 -1.65249676e-01
-6.46008968e-01 1.99507505e-01 -2.44462743e-01 -1.23650134e+00
-4.71852005e-01 -1.35110945e-01 -4.21270467e-02 -1.92871653e-02
8.39577734e-01 1.86677113e-01 2.73871481e-01 -6.73027813e-01
1.29785314e-01 1.83210626e-01 2.39263088e-01 -8.64042580e-01
1.13385844e+00 9.12525058e-01 2.12067664e-01 -6.33914411e-01
-8.92090023e-01 -2.64182836e-01 -9.78842974e-01 4.30182990e-04
7.58444071e-01 -9.89801407e-01 -5.80321252e-01 4.72205222e-01
-1.56614935e+00 -5.16680360e-01 -5.15217543e-01 1.33096322e-01
-7.08931565e-01 2.78114617e-01 -5.45147598e-01 -4.33492482e-01
-6.06428027e-01 -1.36997974e+00 1.84959686e+00 1.18424362e-02
-2.19252836e-02 -7.67632246e-01 4.47497368e-02 2.54635453e-01
-1.71338357e-02 7.15204120e-01 7.69227266e-01 8.93938616e-02
-1.20082152e+00 -1.34067431e-01 -2.24872693e-01 3.15403670e-01
3.14247966e-01 2.10823908e-01 -8.85393202e-01 -1.09824494e-01
2.18243986e-01 -2.55379975e-01 4.36107278e-01 4.68374193e-01
1.31036508e+00 -2.07722351e-01 -2.93005675e-01 1.30703366e+00
1.31103170e+00 -2.32386604e-01 4.69033539e-01 2.69051850e-01
1.07494640e+00 4.20591623e-01 3.23976368e-01 2.34870568e-01
3.58472228e-01 5.52258551e-01 8.36505473e-01 -1.89251587e-01
-1.21802799e-01 -5.95339537e-01 -3.07803191e-02 6.20691299e-01
-3.31410199e-01 2.63204634e-01 -1.09409523e+00 2.28277206e-01
-1.42969501e+00 -6.75154746e-01 -4.13301080e-01 2.31998158e+00
7.08904922e-01 1.58297718e-01 -2.37472318e-02 -1.21446550e-01
5.98210096e-01 1.30080730e-01 -8.05554867e-01 -9.70611051e-02
-3.91413011e-02 2.34827012e-01 4.71593797e-01 6.58464551e-01
-6.87234342e-01 9.28095639e-01 5.26464128e+00 5.72007239e-01
-1.28849816e+00 -1.97858810e-01 5.86028755e-01 -1.68443501e-01
-8.35116386e-01 -5.32985432e-03 -1.08866405e+00 4.87623841e-01
1.22881405e-01 2.02080742e-01 2.73654431e-01 8.95687938e-01
9.57876742e-02 3.05774689e-01 -1.13992786e+00 1.29630661e+00
-1.01743124e-01 -1.73137176e+00 4.36764807e-01 4.02387798e-01
1.15390909e+00 4.82085347e-01 1.08047500e-01 -1.58069074e-01
5.05354643e-01 -1.08222997e+00 6.46982789e-01 4.77907658e-01
1.15525913e+00 -5.39412379e-01 2.34667227e-01 7.43856430e-01
-9.58193481e-01 6.68691635e-01 -4.90410447e-01 -1.24199232e-02
2.57980555e-01 9.91992295e-01 -9.87844110e-01 5.70971489e-01
8.40854585e-01 9.07937169e-01 -1.52245209e-01 7.84644783e-01
-2.11134732e-01 4.27631214e-02 -7.77466714e-01 4.07693595e-01
2.28746220e-01 -6.07615173e-01 7.33337581e-01 5.22171617e-01
7.22211123e-01 -4.93630916e-02 8.38645250e-02 1.40697527e+00
-1.63926437e-01 -2.41144001e-01 -1.09930658e+00 5.36117196e-01
6.62590742e-01 1.06407857e+00 -6.45306587e-01 -2.93923557e-01
-1.46902412e-01 7.21795142e-01 4.46754217e-01 3.23709577e-01
-4.43769097e-01 1.36720920e-02 7.86902606e-01 6.55520856e-01
2.42634907e-01 -5.59362650e-01 -7.28057384e-01 -1.30152225e+00
3.73124331e-01 -6.20425045e-01 -2.08860099e-01 -1.20513773e+00
-1.36872625e+00 5.72954535e-01 1.03444360e-01 -1.69513988e+00
-1.00177720e-01 -7.76399016e-01 -6.68611109e-01 1.16176927e+00
-1.51832139e+00 -1.01192033e+00 -5.28989792e-01 4.36300933e-01
4.49640632e-01 1.05130173e-01 6.27645910e-01 -9.81051847e-02
9.09313485e-02 2.33507723e-01 -2.10280746e-01 7.74803460e-02
5.34364700e-01 -1.30021799e+00 1.00867784e+00 4.44497436e-01
1.95794374e-01 5.85132837e-01 2.80193925e-01 -7.11169720e-01
-1.27943122e+00 -1.09637725e+00 3.78585547e-01 -8.80734801e-01
7.95517415e-02 -3.86569768e-01 -1.13172507e+00 7.59786308e-01
-1.12365797e-01 2.82345563e-01 1.75345451e-01 -1.44733459e-01
-4.91427630e-01 8.03106353e-02 -1.27158785e+00 6.21541739e-01
1.33272219e+00 -3.48715782e-01 -4.81807798e-01 4.04381156e-01
9.20271635e-01 -1.10083210e+00 -9.35708106e-01 5.60712337e-01
2.86301762e-01 -1.13202810e+00 1.28264856e+00 -1.81119159e-01
8.25262666e-01 -5.97219467e-01 -1.13940239e-03 -1.42471051e+00
4.23881458e-03 -6.62474334e-01 -1.40171617e-01 7.01894164e-01
2.90760159e-01 -7.68914223e-01 1.23804641e+00 4.32832778e-01
-5.24688721e-01 -1.09968293e+00 -8.29811037e-01 -5.58916688e-01
4.44890618e-01 -3.46529692e-01 1.08826947e+00 9.98749554e-01
-8.64541471e-01 2.53431171e-01 1.14613891e-01 4.16548401e-01
9.07254398e-01 5.55061698e-01 1.29176068e+00 -1.44933021e+00
-4.28727977e-02 -5.43625414e-01 -2.90104806e-01 -1.59972584e+00
4.84023727e-02 -1.09846473e+00 -1.55930340e-01 -1.55247366e+00
-3.61342192e-01 -8.74271929e-01 5.76481998e-01 2.97674805e-01
1.15734942e-01 2.49001190e-01 1.71955004e-01 5.39470434e-01
1.55936152e-01 6.26982749e-01 1.86725354e+00 9.30008367e-02
-2.39104018e-01 1.48236141e-01 -5.45714438e-01 1.09719443e+00
7.81184375e-01 -3.06088567e-01 -2.63915509e-01 -1.08469796e+00
4.12289709e-01 3.43605429e-02 6.52447224e-01 -8.76157761e-01
-3.36352810e-02 -1.60086721e-01 7.45985687e-01 -1.28673220e+00
6.43944860e-01 -1.01488185e+00 2.45096013e-01 1.88004076e-02
5.22709899e-02 1.56814903e-01 7.84600005e-02 2.52083659e-01
1.38246492e-01 -1.06896438e-01 7.79024482e-01 -6.37141168e-01
-3.79650183e-02 9.57723439e-01 3.93844992e-01 6.31857067e-02
8.00655425e-01 -6.55303836e-01 -2.22225040e-01 -2.06507314e-02
-4.52014089e-01 1.94160357e-01 1.35386467e+00 2.00492367e-01
9.35648739e-01 -1.55026364e+00 -6.46805227e-01 7.13817298e-01
-7.38218352e-02 1.54055095e+00 6.08799607e-02 2.66746193e-01
-9.67388570e-01 3.92317884e-02 -1.23664968e-01 -1.33717203e+00
-8.94005597e-01 3.01122159e-01 6.92669868e-01 4.61282618e-02
-1.05475664e+00 7.46861696e-01 5.12125194e-01 -1.10882020e+00
3.03351344e-03 -6.77284956e-01 3.72045100e-01 -5.73486805e-01
1.68000177e-01 1.84999965e-02 1.54783875e-01 -4.49365795e-01
1.27228647e-01 1.27869833e+00 -5.20851389e-02 1.33998156e-01
1.34873033e+00 3.16295534e-01 4.70643910e-03 4.29634333e-01
1.28763437e+00 1.35464966e-01 -1.88682818e+00 -2.16152787e-01
-5.48642516e-01 -7.72937298e-01 -2.03791261e-01 -1.94146082e-01
-1.19565153e+00 1.09621918e+00 1.63211510e-01 2.14385942e-01
5.48226178e-01 7.77542368e-02 8.90515208e-01 2.52998292e-01
4.83427763e-01 -5.62069774e-01 -8.38089455e-03 7.80407369e-01
1.10656226e+00 -1.30087817e+00 2.98593380e-02 -6.69519424e-01
-2.41364419e-01 1.11468947e+00 8.14386070e-01 -7.76343465e-01
5.99127054e-01 3.12350988e-01 1.41472116e-01 -5.54729879e-01
-5.08055449e-01 3.07119280e-01 2.85625964e-01 6.81212366e-01
2.60824531e-01 6.13502711e-02 4.72977817e-01 -1.85148865e-01
-6.82656467e-01 -1.82698652e-01 3.42908829e-01 6.31376088e-01
-2.10813999e-01 -1.02158487e+00 -6.91644967e-01 5.07978082e-01
-2.29245294e-02 6.34480119e-02 -2.46587351e-01 5.97111166e-01
2.21078359e-02 1.56773645e-02 6.91732824e-01 -2.23381445e-01
5.25299609e-01 -2.20066145e-01 5.85117757e-01 -9.20245707e-01
-1.70922399e-01 1.30191758e-01 -4.16307032e-01 -5.39763331e-01
-3.03226799e-01 -5.19667089e-01 -1.43820989e+00 -3.24657977e-01
7.79107809e-02 -1.29914597e-01 6.64190769e-01 5.13251185e-01
7.77720690e-01 6.84642345e-02 9.39121366e-01 -1.65662181e+00
-4.85505551e-01 -5.65580070e-01 -2.83369303e-01 5.86205721e-01
5.61195970e-01 -6.58880949e-01 -6.77743733e-01 -5.27345240e-02]
|
[8.532735824584961, -3.5607972145080566]
|
7aacac16-4659-47db-8b77-802418e44022
|
a-novel-two-stream-decision-level-fusion-of
|
2306.15765
| null |
https://arxiv.org/abs/2306.15765v1
|
https://arxiv.org/pdf/2306.15765v1.pdf
|
A Novel Two Stream Decision Level Fusion of Vision and Inertial Sensors Data for Automatic Multimodal Human Activity Recognition System
|
This paper presents a novel multimodal human activity recognition system. It uses a two-stream decision level fusion of vision and inertial sensors. In the first stream, raw RGB frames are passed to a part affinity field-based pose estimation network to detect the keypoints of the user. These keypoints are then pre-processed and inputted in a sliding window fashion to a specially designed convolutional neural network for the spatial feature extraction followed by regularized LSTMs to calculate the temporal features. The outputs of LSTM networks are then inputted to fully connected layers for classification. In the second stream, data obtained from inertial sensors are pre-processed and inputted to regularized LSTMs for the feature extraction followed by fully connected layers for the classification. At this stage, the SoftMax scores of two streams are then fused using the decision level fusion which gives the final prediction. Extensive experiments are conducted to evaluate the performance. Four multimodal standard benchmark datasets (UP-Fall detection, UTD-MHAD, Berkeley-MHAD, and C-MHAD) are used for experimentations. The accuracies obtained by the proposed system are 96.9 %, 97.6 %, 98.7 %, and 95.9 % respectively on the UP-Fall Detection, UTDMHAD, Berkeley-MHAD, and C-MHAD datasets. These results are far superior than the current state-of-the-art methods.
|
['Shaik Ali Akbara', 'Hari Mohan Pandey', 'Kamlesh Tiwari', 'Egna Praneeth Gummana', 'Muhtashim Rafiqi', 'Santosh Kumar Yadav']
|
2023-06-27
| null | null | null | null |
['pose-estimation', 'activity-recognition', 'human-activity-recognition', 'human-activity-recognition']
|
['computer-vision', 'computer-vision', 'computer-vision', 'time-series']
|
[ 2.31639087e-01 -4.58644748e-01 -1.19852118e-01 -3.70178163e-01
-7.25375116e-01 1.56062126e-01 3.82198930e-01 1.12979412e-01
-9.40474391e-01 7.93734968e-01 1.29502803e-01 1.62806436e-02
1.18275456e-01 -6.90322638e-01 -6.68344438e-01 -7.24792063e-01
-4.31868434e-01 -2.54148338e-03 2.79098630e-01 -6.02221899e-02
7.84439370e-02 3.98601741e-01 -1.89576769e+00 5.36281824e-01
4.98386383e-01 1.68782377e+00 -1.27654284e-01 9.54859436e-01
2.18962967e-01 7.00155258e-01 -2.55098462e-01 1.60198182e-01
2.44447906e-02 -1.46706149e-01 -5.00453949e-01 -4.54145223e-02
4.12457407e-01 -5.53783000e-01 -4.42791820e-01 7.10259020e-01
7.45855033e-01 6.25950098e-01 3.86031777e-01 -1.13052368e+00
-2.17194691e-01 7.23742694e-02 -6.50797665e-01 4.65761513e-01
7.34293163e-01 -6.83748396e-03 3.80795807e-01 -1.07009792e+00
1.77261293e-01 1.27605879e+00 6.58554554e-01 5.08707285e-01
-5.90487719e-01 -4.53969419e-01 4.63993959e-02 5.73771358e-01
-1.30309880e+00 -3.18291038e-01 5.12278199e-01 -4.94948119e-01
1.32193696e+00 2.56687284e-01 7.61748195e-01 1.09522951e+00
5.54577172e-01 8.44991386e-01 7.82606721e-01 -3.42451692e-01
4.32677358e-01 -3.14706743e-01 4.69302297e-01 1.05864859e+00
2.88167689e-02 4.19406183e-02 -1.00099838e+00 -1.32308021e-01
6.45725489e-01 6.95150435e-01 -3.10545057e-01 3.35451812e-02
-1.36222184e+00 5.05395353e-01 9.24007416e-01 2.26464957e-01
-8.83696139e-01 -8.42872486e-02 4.42550808e-01 2.69785430e-02
1.88116491e-01 -4.51760173e-01 -3.27739388e-01 -1.94123134e-01
-9.00778234e-01 1.88444614e-01 3.16538185e-01 5.35640001e-01
4.74314362e-01 -2.12839544e-01 -2.31576741e-01 6.92212760e-01
8.11871827e-01 7.15479612e-01 9.59225357e-01 -4.72242087e-01
8.36807609e-01 8.47892463e-01 3.86952490e-01 -1.06413329e+00
-7.32107580e-01 -1.18891068e-01 -9.31280434e-01 3.91076595e-01
1.64600909e-01 -4.78406787e-01 -1.27452040e+00 1.22169459e+00
3.23529601e-01 2.27909252e-01 2.78458863e-01 1.29800737e+00
8.10667753e-01 7.46291816e-01 2.48218849e-01 -1.02845743e-01
1.51160431e+00 -7.00368166e-01 -5.71924031e-01 -3.87730241e-01
3.89504522e-01 -5.05492628e-01 9.57149088e-01 4.89245683e-01
-8.04522812e-01 -9.64774430e-01 -1.53993261e+00 1.27849624e-01
-3.69156569e-01 4.99788076e-01 2.74565727e-01 3.76570612e-01
-8.33162904e-01 4.23435181e-01 -1.50129509e+00 -5.41393816e-01
5.36362350e-01 6.57384992e-01 -5.07413447e-01 1.30016744e-01
-1.11522448e+00 8.44884455e-01 5.53599775e-01 5.88483274e-01
-7.67621458e-01 1.64217100e-01 -1.16899157e+00 -2.73756385e-01
-2.24056348e-01 -7.20607877e-01 1.29924822e+00 -6.99221611e-01
-1.29796767e+00 5.83998919e-01 -3.62296104e-01 -5.61184168e-01
4.48206067e-01 -5.53623497e-01 -5.47735870e-01 8.79674032e-02
3.49207968e-02 4.25308377e-01 6.41022444e-01 -7.78975189e-01
-1.27232146e+00 -8.57244730e-01 -2.44759619e-01 6.08849466e-01
-2.25716382e-01 -1.16216451e-01 -4.37881351e-01 -1.95446551e-01
5.69203079e-01 -8.29550862e-01 5.76021522e-02 -1.80255592e-01
-4.73894536e-01 -1.72986194e-01 1.10494947e+00 -7.45568812e-01
1.13799882e+00 -2.11903977e+00 2.38975927e-01 5.18594533e-02
-1.26563251e-01 3.08944762e-01 3.97334933e-01 6.32804260e-02
8.07098448e-02 -6.07687116e-01 -5.49730882e-02 -7.46004224e-01
-2.31013462e-01 1.44129217e-01 1.04765207e-01 6.38553381e-01
6.91430718e-02 5.55786371e-01 -6.06292903e-01 -4.37214434e-01
6.94691420e-01 6.57658041e-01 -2.98101828e-02 3.30318123e-01
2.08309948e-01 2.24081099e-01 -4.18504566e-01 7.88586378e-01
3.74494970e-01 2.23105431e-01 -3.54950070e-01 -2.80810416e-01
-2.34372675e-01 5.57573466e-03 -1.29853582e+00 1.67405939e+00
-1.95096314e-01 5.46456575e-01 -2.70667285e-01 -8.38250577e-01
8.23548257e-01 3.37629378e-01 3.56261820e-01 -7.86892772e-01
4.58640933e-01 5.61740920e-02 -3.43144298e-01 -8.92577291e-01
4.87950474e-01 -8.48330855e-02 -2.76185930e-01 4.02646437e-02
1.05752908e-01 8.38212192e-01 -8.82204343e-03 -2.52191275e-01
8.83552790e-01 3.04518789e-01 7.97378346e-02 3.06261897e-01
7.17535734e-01 -5.13531417e-02 4.73004460e-01 4.59048480e-01
-4.53623354e-01 5.14503777e-01 -1.87206030e-01 -6.57309175e-01
-6.23240411e-01 -1.14709902e+00 1.23645291e-01 1.17350423e+00
2.81971782e-01 -1.25830248e-01 -7.33328819e-01 -5.69434345e-01
-6.62449896e-02 2.88784027e-01 -7.34581947e-01 -3.93666327e-01
-4.52677131e-01 -8.03251505e-01 5.44920146e-01 1.03684127e+00
1.13196468e+00 -1.12167990e+00 -1.27931070e+00 2.13642493e-01
-2.26030186e-01 -8.61719012e-01 -5.03757522e-02 2.62917548e-01
-9.26653624e-01 -1.07660365e+00 -9.22513545e-01 -9.25216556e-01
3.85365993e-01 3.49259302e-02 2.19838858e-01 -3.05053800e-01
-1.30752325e-01 1.92729637e-01 -1.47325844e-01 -4.73437726e-01
4.04983878e-01 -1.35640264e-01 4.10745978e-01 4.42921609e-01
6.17447138e-01 -3.50861192e-01 -8.20846856e-01 1.53941661e-01
-5.19541860e-01 -1.71415091e-01 6.20839000e-01 5.25671184e-01
7.94395447e-01 -2.05075443e-01 5.28304130e-02 -7.06893951e-02
5.89669168e-01 -2.74828762e-01 -3.54101896e-01 -7.87851692e-04
-1.10133328e-01 -8.56077522e-02 3.66959512e-01 -4.89377350e-01
-1.12320673e+00 4.47409213e-01 -1.82906806e-01 -3.39729220e-01
-5.00581861e-01 6.32512331e-01 -1.86634749e-01 1.46014616e-01
7.37894475e-01 2.38908231e-01 -2.65093595e-01 -4.31494713e-01
4.97660302e-02 1.15135145e+00 9.75686193e-01 -2.08623335e-02
2.64662623e-01 4.37023312e-01 -1.86395437e-01 -9.34651911e-01
-4.82644409e-01 -5.17151296e-01 -6.96138144e-01 -4.44845825e-01
1.33123922e+00 -1.10679305e+00 -6.49434745e-01 1.14134789e+00
-9.40435827e-01 -4.53696176e-02 2.06998885e-01 9.17314529e-01
-4.71089900e-01 2.07890585e-01 -7.92300940e-01 -1.14293385e+00
-6.97555542e-01 -1.04340053e+00 1.22048986e+00 8.38684380e-01
-2.16985680e-02 -5.54445446e-01 -1.29333287e-01 3.76289546e-01
1.83694124e-01 6.23157382e-01 3.04474473e-01 -4.64114308e-01
4.96594384e-02 -5.86904883e-01 1.45984339e-02 2.65500665e-01
1.59854591e-01 -1.71942309e-01 -1.02830517e+00 -1.25753582e-01
-1.29873948e-02 -2.90679991e-01 8.14445734e-01 6.03874922e-01
5.15729070e-01 -6.71778917e-02 -5.93247354e-01 4.80811447e-01
1.23254466e+00 6.81910396e-01 5.27619302e-01 6.07213736e-01
7.24122167e-01 -1.16964407e-01 8.58109295e-01 4.55680013e-01
6.92976117e-01 4.09211308e-01 5.71280062e-01 -1.42775804e-01
3.30338687e-01 1.05391908e-02 5.48095167e-01 3.97090048e-01
-3.13107282e-01 2.26926971e-02 -9.64467168e-01 3.17952842e-01
-2.23146987e+00 -8.74149144e-01 -1.91780671e-01 2.09114552e+00
3.03218246e-01 3.95687193e-01 2.40057364e-01 7.72149920e-01
6.57244742e-01 3.01815607e-02 -4.20073122e-01 -2.72363096e-01
1.02707878e-01 1.13272905e-01 4.61037755e-01 3.59498590e-01
-1.81953645e+00 6.06289327e-01 4.49905062e+00 -1.58104859e-02
-1.39546824e+00 1.14820272e-01 4.44498181e-01 -2.28042737e-01
8.39282155e-01 -6.03062034e-01 -7.55362213e-01 5.84327877e-01
1.06509054e+00 4.42168057e-01 8.96959975e-02 8.06759715e-01
4.30961877e-01 -7.12336779e-01 -9.66144800e-01 1.41592491e+00
1.12236939e-01 -6.64704084e-01 -2.25821227e-01 -2.27972507e-01
3.89890224e-01 2.33714923e-01 -1.44125074e-01 2.65105873e-01
-2.07743898e-01 -8.54026198e-01 6.51067972e-01 1.03731453e+00
3.72308850e-01 -1.12522268e+00 1.09167171e+00 5.17131090e-01
-1.36156154e+00 -3.30984622e-01 -2.66327530e-01 -2.65994281e-01
3.31880450e-01 2.17395201e-01 -4.94968563e-01 5.71655989e-01
1.14182329e+00 5.47317266e-01 -4.66613442e-01 1.16375816e+00
-9.85464901e-02 3.78517002e-01 -5.41448355e-01 -1.54358923e-01
4.15891379e-01 1.53462276e-01 2.16607511e-01 1.10012686e+00
4.29129124e-01 1.43175900e-01 2.54661083e-01 1.10713571e-01
1.59117877e-01 -2.98029482e-01 -4.09311593e-01 2.94901758e-01
-8.87876898e-02 1.06832886e+00 -4.25306946e-01 -5.93686342e-01
-3.33542824e-01 1.22413826e+00 5.17280810e-02 4.69130307e-01
-9.58048284e-01 -7.19742477e-01 5.36679864e-01 -2.88358554e-02
2.31780797e-01 -4.08857405e-01 -1.92026109e-01 -1.11518478e+00
2.33892173e-01 -4.45902199e-01 8.77080977e-01 -9.13625717e-01
-9.54740286e-01 6.60714209e-01 -6.77728280e-02 -1.17790449e+00
-3.98048133e-01 -7.12232351e-01 -5.62433481e-01 1.08250856e+00
-1.03657246e+00 -1.07658148e+00 -7.74789691e-01 1.04712164e+00
4.27679121e-01 -1.09460607e-01 7.15552747e-01 5.16355038e-01
-7.36270487e-01 3.79478514e-01 -2.19528243e-01 3.93510073e-01
4.21239167e-01 -1.02733040e+00 1.98382452e-01 1.07855403e+00
-1.96592391e-01 5.24290025e-01 3.63510489e-01 -6.05207145e-01
-1.43474269e+00 -1.24340296e+00 9.90591466e-01 -1.37995377e-01
1.90443471e-02 -1.67661488e-01 -7.34463930e-01 7.36376524e-01
-6.56966195e-02 1.77854881e-01 5.53982556e-01 -3.90113235e-01
2.45624334e-01 -1.80316746e-01 -1.25636959e+00 2.84916133e-01
7.71827579e-01 -4.80139107e-01 -9.53797102e-01 -8.88943300e-03
1.59682557e-01 -7.72162259e-01 -9.14912462e-01 4.74711418e-01
8.04330111e-01 -8.92772675e-01 8.03457141e-01 -5.95710337e-01
3.65125611e-02 -5.03431737e-01 -5.27545393e-01 -1.02511704e+00
-1.55050680e-01 7.01225176e-02 -1.69431761e-01 7.57795334e-01
2.10608780e-01 -5.19078493e-01 7.23463178e-01 4.93771285e-01
-1.90198720e-01 -7.66972780e-01 -1.16904640e+00 -2.25732744e-01
-6.62443578e-01 -6.86324000e-01 2.65927672e-01 4.50603068e-01
1.70730084e-01 4.03220087e-01 -3.18463206e-01 4.46286470e-01
5.95119238e-01 -1.66767955e-01 3.91176254e-01 -1.05483305e+00
1.80358380e-01 -1.10404082e-01 -8.33420217e-01 -8.42657864e-01
-4.42819208e-01 -3.93147320e-01 1.73902437e-01 -1.81786346e+00
-2.30450213e-01 2.78434068e-01 -5.59292316e-01 7.72739172e-01
-1.14375673e-01 4.41148162e-01 -1.12642817e-01 -5.59898168e-02
-5.12095153e-01 5.31612396e-01 6.90591574e-01 -1.04236215e-01
-6.11074150e-01 3.46116453e-01 -9.75930020e-02 8.30495834e-01
8.63210142e-01 -4.12704200e-02 -2.86401391e-01 -4.21475947e-01
-4.45929050e-01 1.44012630e-01 6.39716566e-01 -1.82596123e+00
3.04036975e-01 1.14518695e-01 1.28389668e+00 -9.85879242e-01
5.68055868e-01 -6.45835638e-01 -2.05727100e-01 8.85005236e-01
-1.15256403e-02 2.74523258e-01 3.54738712e-01 4.56929624e-01
-1.93137750e-01 3.98439348e-01 6.67305112e-01 -3.01389955e-02
-1.06677437e+00 2.92950720e-01 -6.75132990e-01 -5.46651304e-01
1.05591619e+00 -4.80542392e-01 6.03707358e-02 -2.91112542e-01
-1.17022133e+00 1.88778728e-01 -5.72832301e-02 6.61220670e-01
9.96374249e-01 -1.62255538e+00 -2.36642540e-01 5.72243154e-01
1.00624017e-01 1.79416746e-01 3.01445305e-01 9.58042383e-01
-4.28010792e-01 3.85349154e-01 -4.86114621e-01 -9.70048249e-01
-1.40598238e+00 2.15011463e-01 5.97273052e-01 2.25304395e-01
-4.17032123e-01 7.70287395e-01 -5.78886688e-01 -1.06675737e-01
6.13378823e-01 -8.14510107e-01 -3.14780653e-01 1.55504629e-01
7.65535116e-01 4.65262055e-01 3.84911895e-01 -1.08799493e+00
-8.25968504e-01 4.76411760e-01 2.96467990e-01 -3.52749854e-01
1.19553053e+00 -8.51418525e-02 3.02835733e-01 6.04492843e-01
1.03781068e+00 -7.88104177e-01 -1.20442998e+00 -1.79147422e-01
-5.89053566e-03 -1.13961592e-01 2.88870670e-02 -8.02326858e-01
-9.40575063e-01 1.10050154e+00 1.49322820e+00 -1.67283326e-01
1.36704957e+00 -2.68874913e-01 9.33567047e-01 4.05472875e-01
3.53139669e-01 -1.01675773e+00 -1.14056244e-01 5.44728935e-01
6.30636930e-01 -9.74939346e-01 -3.57116848e-01 4.76699203e-01
-5.90686798e-01 1.01233697e+00 8.01973581e-01 -2.05957055e-01
5.89038074e-01 2.15771720e-01 2.96102852e-01 5.68066537e-03
-4.19736117e-01 -2.63119370e-01 4.58057612e-01 5.13892353e-01
3.86173785e-01 7.91691430e-03 -1.06168471e-01 6.74879253e-01
-1.39930397e-01 3.97383749e-01 -1.11171873e-02 1.32905948e+00
-8.08954120e-01 -4.39933807e-01 -7.22337842e-01 3.82506013e-01
-3.48631650e-01 3.73080611e-01 -3.67225051e-01 6.06674910e-01
5.06470501e-01 1.05234551e+00 1.61666021e-01 -8.81294549e-01
7.39662409e-01 4.14938509e-01 4.43507046e-01 -2.13197842e-01
-5.75992584e-01 1.19470753e-01 1.12819105e-01 -7.42183864e-01
-3.53113472e-01 -5.65723777e-01 -1.82430315e+00 -5.99359069e-03
-5.07469624e-02 -2.88900658e-02 5.83579063e-01 1.19513547e+00
3.01873147e-01 5.15860438e-01 1.40764654e-01 -1.45429134e+00
-3.13159138e-01 -1.17369127e+00 -2.60400862e-01 3.85339767e-01
6.76903486e-01 -8.34694564e-01 7.35195875e-02 1.10852003e-01]
|
[7.822396278381348, 0.48544570803642273]
|
363da4ba-6222-4158-b213-e8f5b17fa89a
|
vspw-a-large-scale-dataset-for-video-scene
| null | null |
http://openaccess.thecvf.com//content/CVPR2021/html/Miao_VSPW_A_Large-scale_Dataset_for_Video_Scene_Parsing_in_the_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Miao_VSPW_A_Large-scale_Dataset_for_Video_Scene_Parsing_in_the_CVPR_2021_paper.pdf
|
VSPW: A Large-scale Dataset for Video Scene Parsing in the Wild
|
In this paper, we present a new dataset with the target of advancing the scene parsing task from images to videos. Our dataset aims to perform Video Scene Parsing in the Wild (VSPW), which covers a wide range of real-world scenarios and categories. To be specific, our VSPW is featured from the following aspects: 1) Well-trimmed long-temporal clips. Each video contains a complete shot, lasting around 5 seconds on average. 2) Dense annotation. The pixel-level annotations are provided at a high frame rate of 15 f/s. 3) High resolution. Over 96% of the captured videos are with high spatial resolutions from 720P to 4K. We totally annotate 3,337 videos, including 239,934 frames from 124 categories. To the best of our knowledge, our VSPW is the first attempt to tackle the challenging video scene parsing task in the wild by considering diverse scenarios. Based on VSPW, we design a generic Temporal Context Blending (TCB) network, which can effectively harness long-range contextual information from the past frames to help segment the current one. Extensive experiments show that our TCB network improves both the segmentation performance and temporal stability comparing with image-/video-based state-of-the-art methods. We hope that the scale, diversity, long-temporal, and high frame rate of our VSPW can significantly advance the research of video scene parsing and beyond.
|
['Yi Yang', 'Guangrui Li', 'Chen Liang', 'Yu Wu', 'Yunchao Wei', 'Jiaxu Miao']
|
2021-06-19
| null | null | null |
cvpr-2021-1
|
['scene-parsing']
|
['computer-vision']
|
[ 4.93252665e-01 -1.57182008e-01 -2.89380819e-01 -2.76139051e-01
-7.24027038e-01 -4.97296154e-01 4.01136488e-01 -2.29851082e-01
-4.36009288e-01 3.60909522e-01 -3.94179457e-04 -4.82879654e-02
1.84468716e-01 -6.05556786e-01 -9.70849395e-01 -6.24689996e-01
-1.18371025e-01 -1.50643274e-01 9.94504750e-01 6.04082793e-02
-2.13225409e-02 3.31835926e-01 -1.72019148e+00 4.46294904e-01
7.78647959e-01 1.35444248e+00 6.09450936e-01 6.76440835e-01
-7.42854774e-02 9.75340962e-01 -2.23632395e-01 -4.01514918e-01
3.00990582e-01 -2.34843343e-01 -8.53183448e-01 2.78635412e-01
6.64070308e-01 -5.81196845e-01 -4.26135242e-01 9.32239175e-01
1.73883393e-01 1.45284131e-01 -1.15921795e-01 -1.24255371e+00
-1.04684025e-01 5.09407759e-01 -6.15694225e-01 3.41051370e-01
4.52837974e-01 4.39686924e-01 7.73814499e-01 -7.20130324e-01
1.01601839e+00 1.03665447e+00 4.86275196e-01 4.02132154e-01
-7.19164133e-01 -6.33188426e-01 5.70768654e-01 3.85455579e-01
-1.09342945e+00 -4.55357432e-01 7.75503516e-01 -4.53358680e-01
7.11022258e-01 1.44013688e-01 8.91736567e-01 1.26099050e+00
-2.12088093e-01 9.91994560e-01 7.12651789e-01 -2.41621867e-01
1.97636098e-01 -5.96607089e-01 -4.84916940e-02 6.12258554e-01
3.24028954e-02 -1.24633245e-01 -7.52107143e-01 3.99326980e-01
7.49919295e-01 1.03573509e-01 -3.86123598e-01 -2.19365090e-01
-1.46154952e+00 4.19972479e-01 2.17558011e-01 3.85016501e-01
-3.53879601e-01 2.49361694e-01 5.90487599e-01 -1.10023320e-01
2.93746740e-01 1.37606554e-03 -5.37023127e-01 -4.19945240e-01
-1.07988739e+00 9.81470942e-02 4.93644416e-01 1.29491878e+00
5.51990151e-01 1.10239886e-01 -2.38520041e-01 7.41617203e-01
-3.77378948e-02 6.44974291e-01 1.35869026e-01 -1.54632652e+00
6.55536890e-01 1.81893751e-01 3.81858461e-02 -1.01772916e+00
-2.59165138e-01 1.12931952e-01 -6.20029271e-01 -2.98666269e-01
5.16755521e-01 -1.44722030e-01 -9.13958430e-01 1.61784041e+00
4.60275322e-01 5.93145728e-01 -1.60065830e-01 1.03769779e+00
9.31012988e-01 9.38653648e-01 2.76141137e-01 -4.91056323e-01
1.51595640e+00 -1.27471256e+00 -7.44811296e-01 -3.31104577e-01
3.00760508e-01 -7.41128564e-01 1.03402960e+00 5.01533985e-01
-1.11498857e+00 -7.68660903e-01 -8.46020222e-01 -6.43277392e-02
-1.67892519e-02 2.27787662e-02 5.57140827e-01 2.98417926e-01
-8.39066327e-01 4.07312334e-01 -8.77049983e-01 -4.22386616e-01
6.05210602e-01 -6.49655387e-02 -4.79945391e-01 -4.23277825e-01
-1.02938008e+00 1.43740103e-01 4.70768094e-01 1.46671891e-01
-8.84607732e-01 -8.04957390e-01 -8.93295050e-01 -1.21283136e-01
9.99092221e-01 -4.11046058e-01 1.26895332e+00 -1.05344343e+00
-1.32008195e+00 6.86966300e-01 -3.61548215e-01 -4.22966331e-01
5.87326229e-01 -5.25375426e-01 -2.76641786e-01 7.40364015e-01
2.72656083e-01 8.58301997e-01 6.90456092e-01 -1.07124913e+00
-9.02705848e-01 -9.65026021e-02 8.27787593e-02 9.57514271e-02
-3.43084037e-01 1.79284468e-01 -1.37941968e+00 -7.98717737e-01
-6.23366535e-02 -8.93371582e-01 -2.65443418e-02 4.85568009e-02
-3.10025960e-01 1.26962125e-01 1.24225283e+00 -7.93701530e-01
1.12618661e+00 -2.27977037e+00 1.02590449e-01 -5.29256880e-01
9.17133093e-02 5.18035710e-01 -2.10120425e-01 2.30344504e-01
1.80872992e-01 4.83490750e-02 -3.35957021e-01 -3.35058182e-01
-3.43782932e-01 4.09199148e-01 -4.12796289e-01 1.79884985e-01
1.01224877e-01 9.93124127e-01 -1.01940775e+00 -8.74391317e-01
4.94734228e-01 3.78077984e-01 -5.79370141e-01 4.33072001e-01
-4.03008968e-01 5.01398146e-01 -4.04708177e-01 8.56994748e-01
7.59247780e-01 -3.24021094e-02 1.70201659e-01 -4.00803894e-01
-3.11766297e-01 -1.65676907e-01 -1.12959254e+00 2.09997725e+00
-3.19891460e-02 8.58632207e-01 2.43237197e-01 -8.97297263e-01
4.79347616e-01 9.94068161e-02 7.97243893e-01 -8.94888341e-01
1.97530556e-02 9.15449262e-02 -4.34360296e-01 -8.99119318e-01
5.93295753e-01 3.56312633e-01 -5.70502914e-02 -1.17368214e-01
1.90554634e-01 -1.75560210e-02 8.07327747e-01 3.06147903e-01
1.07054257e+00 5.07893384e-01 -3.10475826e-02 4.30904068e-02
4.19647694e-01 -5.35188355e-02 1.17385995e+00 6.82321548e-01
-5.80338120e-01 7.91306913e-01 5.97537994e-01 -4.20661032e-01
-9.85778391e-01 -9.52450275e-01 -4.68126461e-02 1.09163344e+00
5.31776249e-01 -5.22207022e-01 -9.76240575e-01 -6.11064911e-01
-4.10813570e-01 1.78312555e-01 -3.35011303e-01 3.18945616e-01
-9.78109717e-01 -2.57848352e-01 4.74178404e-01 7.58529484e-01
1.06052315e+00 -1.15773439e+00 -8.25588107e-01 2.47909799e-01
-5.89184225e-01 -2.08868647e+00 -5.72741270e-01 -2.45777190e-01
-7.89298177e-01 -1.28064299e+00 -7.73968995e-01 -8.46942842e-01
1.87630028e-01 6.57588542e-01 1.13226163e+00 -7.19918311e-02
-3.11295480e-01 4.11017209e-01 -8.37693989e-01 2.03215461e-02
-3.97258699e-02 -2.57777572e-01 -2.52822042e-01 1.35693133e-01
-2.83001885e-02 -4.92366374e-01 -7.37605333e-01 5.54177105e-01
-1.06492138e+00 4.92646724e-01 4.26657379e-01 5.21499693e-01
9.41744506e-01 1.05963975e-01 2.14395866e-01 -8.91325951e-01
-2.36368775e-01 -3.19490641e-01 -5.54975510e-01 2.21985921e-01
3.28914188e-02 -5.27465403e-01 5.55696309e-01 -4.99932557e-01
-1.18927848e+00 2.68321246e-01 -1.03664882e-01 -5.99198699e-01
-3.62390280e-01 1.24500677e-01 -2.92247325e-01 7.84710124e-02
1.16136856e-01 1.82774484e-01 -2.73059458e-01 -3.64208162e-01
5.03086805e-01 4.37553078e-01 1.10602236e+00 -6.74536765e-01
5.34875512e-01 5.58485568e-01 -1.00507244e-01 -9.90486920e-01
-9.48252857e-01 -5.59149325e-01 -7.00988591e-01 -5.76430738e-01
1.33784091e+00 -1.15053093e+00 -5.38954020e-01 7.12663949e-01
-1.04794240e+00 -6.92207098e-01 -1.28595173e-01 3.65367502e-01
-6.37006819e-01 7.17604399e-01 -7.88631916e-01 -6.94819391e-01
-2.51240373e-01 -1.30260420e+00 1.28180552e+00 3.54168922e-01
2.13446066e-01 -5.14291525e-01 -3.92909676e-01 6.61895156e-01
1.02463730e-01 5.18916428e-01 3.97228450e-01 -9.07503441e-02
-8.88618469e-01 2.36976966e-01 -4.50073749e-01 3.20084065e-01
-1.86009154e-01 2.81676471e-01 -9.84651327e-01 -2.75391210e-02
-9.91750360e-02 -1.03189424e-01 1.00069332e+00 5.59006572e-01
1.44398463e+00 5.29252440e-02 -1.36591703e-01 7.97909498e-01
1.23942292e+00 4.54872787e-01 7.63517439e-01 2.87987411e-01
1.06098533e+00 5.43767095e-01 1.26128924e+00 3.62935424e-01
3.75696898e-01 8.77106249e-01 6.67409956e-01 -1.72962040e-01
-3.77648652e-01 -2.28172928e-01 4.40760821e-01 8.50657761e-01
-2.46173099e-01 -3.85575324e-01 -8.43313456e-01 5.94305992e-01
-2.02346683e+00 -1.11148190e+00 -2.75637180e-01 1.87474120e+00
4.92027879e-01 1.47127494e-01 2.09326491e-01 7.06347600e-02
9.16254878e-01 5.80749989e-01 -5.81637621e-01 4.79227491e-02
-2.24928305e-01 -1.60918295e-01 4.02129263e-01 2.45075285e-01
-1.40374649e+00 1.00793850e+00 5.35648108e+00 9.75775480e-01
-1.08696592e+00 7.05305263e-02 7.13963807e-01 -2.53091961e-01
1.85337037e-01 1.61989853e-02 -6.32049561e-01 8.37445378e-01
7.30062366e-01 2.03291059e-01 3.87936711e-01 9.78255987e-01
3.45063329e-01 -2.99276471e-01 -8.54014337e-01 1.19546700e+00
8.10422525e-02 -1.45915902e+00 -2.33387679e-01 -6.18231185e-02
7.86324203e-01 7.60213882e-02 -1.92841977e-01 2.23864198e-01
-2.44466886e-01 -7.06224680e-01 1.15323234e+00 4.10864264e-01
9.77130234e-01 -6.29412472e-01 6.01134300e-01 2.31394723e-01
-1.71726668e+00 -2.06856042e-01 -8.72410014e-02 2.22495392e-01
7.57556021e-01 7.63625681e-01 -2.15154946e-01 7.11480558e-01
1.18517637e+00 1.19130921e+00 -5.84980607e-01 9.47379827e-01
-1.03670619e-01 7.38697767e-01 -3.66012782e-01 3.01779360e-01
2.24355459e-01 -2.25953177e-01 3.49674046e-01 1.43588936e+00
3.87785494e-01 3.80933553e-01 4.38869566e-01 4.08032686e-01
-3.40923630e-02 -4.10837531e-02 -5.07160962e-01 -7.16740862e-02
5.41032910e-01 1.27992809e+00 -1.04943597e+00 -4.88776475e-01
-4.54778016e-01 1.00064492e+00 -1.07349396e-01 3.00666213e-01
-1.19187117e+00 -8.33791941e-02 5.29138982e-01 7.46095404e-02
7.14888692e-01 -4.08683747e-01 -3.59338671e-02 -1.38159537e+00
4.38504398e-01 -6.94911897e-01 4.18971181e-01 -1.04896629e+00
-9.42278862e-01 7.12380648e-01 1.27394453e-01 -1.24150801e+00
4.03383598e-02 -4.14425552e-01 -4.58668858e-01 7.80030340e-02
-1.64846838e+00 -1.26790011e+00 -7.58699238e-01 5.97898304e-01
1.06129920e+00 1.70415089e-01 3.00037920e-01 6.04677975e-01
-8.46536040e-01 2.16491133e-01 -2.62367904e-01 2.89864868e-01
6.75230086e-01 -8.76807988e-01 3.67443651e-01 1.35200739e+00
1.09162867e-01 1.08044423e-01 4.54316586e-01 -6.16517901e-01
-1.61833739e+00 -1.30571067e+00 4.31984603e-01 -2.23038286e-01
4.31491822e-01 -3.03392440e-01 -8.31750214e-01 4.34115142e-01
6.98845759e-02 4.51984018e-01 2.66702712e-01 -3.70859593e-01
-2.84350753e-01 -1.97593689e-01 -8.83614659e-01 4.21441168e-01
1.40211117e+00 -4.13897306e-01 -2.59645015e-01 1.69658557e-01
1.23657978e+00 -6.77811086e-01 -8.66012514e-01 6.90058589e-01
4.79598105e-01 -1.29769623e+00 1.11911750e+00 6.66261604e-03
6.91114902e-01 -4.31543767e-01 -3.58960927e-01 -5.16284943e-01
5.13005145e-02 -7.32469380e-01 -2.36218363e-01 1.60420811e+00
-2.56273508e-01 -1.43307090e-01 6.71470642e-01 5.94543517e-01
-3.48218203e-01 -8.54765594e-01 -8.85026932e-01 -6.90604806e-01
-3.64153057e-01 -9.76917326e-01 4.11490470e-01 6.89563274e-01
-4.18946594e-01 -5.82838655e-02 -6.00516081e-01 7.72352237e-03
6.19131982e-01 3.53922725e-01 7.88014710e-01 -9.53294992e-01
-2.34071508e-01 -1.78491831e-01 -5.03522336e-01 -1.35766661e+00
4.70688902e-02 -3.77261043e-01 1.05745956e-01 -1.27679396e+00
2.38756478e-01 -2.73708165e-01 1.11584812e-01 4.52608794e-01
-1.62159339e-01 5.06437182e-01 6.08012795e-01 1.11499205e-01
-1.08809400e+00 3.39845657e-01 1.27748072e+00 3.85850109e-02
3.24268937e-02 -3.83219779e-01 -3.69018346e-01 9.28005815e-01
4.84680086e-01 -1.84505448e-01 -4.71730053e-01 -5.70226967e-01
-2.23215342e-01 3.70556653e-01 3.04034144e-01 -1.01933467e+00
2.12899193e-01 -4.80970025e-01 8.54055509e-02 -6.98461771e-01
5.19524038e-01 -7.22038746e-01 3.07954848e-01 3.06873709e-01
-4.56792675e-03 -1.71094105e-01 1.29391536e-01 8.09786856e-01
-4.03334975e-01 4.71566468e-02 7.62260616e-01 -5.32478653e-02
-1.51740205e+00 4.57719356e-01 -1.13227986e-01 3.39073062e-01
1.35135484e+00 -3.61373812e-01 -4.15834814e-01 -2.04270959e-01
-4.73530293e-01 2.88304955e-01 5.50342381e-01 4.71770525e-01
5.56848586e-01 -1.13158023e+00 -3.96689206e-01 8.67928192e-02
1.05364926e-01 2.92136699e-01 7.19006360e-01 8.82059813e-01
-7.40042090e-01 2.72762120e-01 -1.50957406e-01 -9.62745607e-01
-1.39344287e+00 6.05473638e-01 -1.21160723e-01 4.33726646e-02
-1.10555458e+00 7.20846057e-01 3.56562197e-01 2.25069806e-01
2.95408756e-01 -4.04433489e-01 -1.92231089e-01 1.68953434e-01
7.46061385e-01 3.19542766e-01 -1.75903037e-01 -8.43097031e-01
-2.66840279e-01 8.69494617e-01 1.79172143e-01 -3.21587063e-02
1.43430161e+00 -2.60057598e-01 7.31069446e-02 2.48482063e-01
1.16152215e+00 -1.24023616e-01 -1.89846289e+00 -1.24480478e-01
-2.72188246e-01 -8.80920827e-01 -1.51657000e-01 -4.61835861e-01
-1.45794487e+00 7.27674603e-01 3.90209556e-01 9.89817530e-02
1.55374134e+00 -4.20094002e-03 1.30135787e+00 2.12609079e-02
5.94487190e-01 -1.13727057e+00 2.41603091e-01 4.71940994e-01
5.69996834e-01 -1.17629766e+00 -9.76387337e-02 -8.39465559e-01
-8.72688830e-01 1.15142429e+00 6.53047979e-01 4.74476442e-02
2.33099312e-01 3.99602085e-01 -5.70975617e-02 1.06668785e-01
-6.53086424e-01 -3.53795499e-01 1.09402411e-01 5.95085204e-01
6.17112257e-02 -1.44887403e-01 -1.17958069e-01 4.98355269e-01
1.03608176e-01 5.29643223e-02 6.07178926e-01 9.07691956e-01
-3.94667268e-01 -9.05256033e-01 -2.01539382e-01 5.52345030e-02
-3.86307657e-01 1.87449828e-01 1.10038735e-01 7.29719341e-01
2.79580235e-01 1.21732879e+00 -5.08563034e-02 -2.30498955e-01
3.10950369e-01 -1.80785298e-01 2.65480369e-01 -1.99570954e-01
-2.29120851e-01 2.94210821e-01 1.77590713e-01 -1.25153041e+00
-8.44864309e-01 -7.45270789e-01 -1.34301043e+00 -2.54838765e-01
-2.57732980e-02 -1.57842383e-01 5.92844486e-01 7.96436787e-01
4.24280792e-01 6.41897738e-01 3.92843276e-01 -1.23661828e+00
2.35020295e-01 -6.04063988e-01 -4.10353005e-01 6.22736454e-01
2.12581024e-01 -6.10160410e-01 -1.19868085e-01 6.23531401e-01]
|
[9.264372825622559, -0.004658365622162819]
|
b4d6a7d2-ed80-4e53-8ce5-1f76620a7ae0
|
deep-image-homography-estimation
|
1606.03798
| null |
http://arxiv.org/abs/1606.03798v1
|
http://arxiv.org/pdf/1606.03798v1.pdf
|
Deep Image Homography Estimation
|
We present a deep convolutional neural network for estimating the relative
homography between a pair of images. Our feed-forward network has 10 layers,
takes two stacked grayscale images as input, and produces an 8 degree of
freedom homography which can be used to map the pixels from the first image to
the second. We present two convolutional neural network architectures for
HomographyNet: a regression network which directly estimates the real-valued
homography parameters, and a classification network which produces a
distribution over quantized homographies. We use a 4-point homography
parameterization which maps the four corners from one image into the second
image. Our networks are trained in an end-to-end fashion using warped MS-COCO
images. Our approach works without the need for separate local feature
detection and transformation estimation stages. Our deep models are compared to
a traditional homography estimator based on ORB features and we highlight the
scenarios where HomographyNet outperforms the traditional technique. We also
describe a variety of applications powered by deep homography estimation, thus
showcasing the flexibility of a deep learning approach.
|
['Tomasz Malisiewicz', 'Daniel DeTone', 'Andrew Rabinovich']
|
2016-06-13
| null | null | null | null |
['homography-estimation']
|
['computer-vision']
|
[ 2.72057682e-01 1.11382768e-01 -1.76108345e-01 -3.41722786e-01
-6.39421880e-01 -5.09948075e-01 7.88462162e-01 -6.55596018e-01
-1.74327463e-01 2.21682295e-01 1.90948486e-01 -9.18460116e-02
1.97241887e-01 -8.02238286e-01 -1.25110698e+00 -5.47972262e-01
-2.27445085e-02 6.22444093e-01 6.04831167e-02 -2.25449100e-01
5.38221538e-01 6.06405199e-01 -1.15019333e+00 1.17774688e-01
2.52747357e-01 1.13489568e+00 -1.23043865e-01 1.05877459e+00
5.78085780e-01 5.94847560e-01 -2.72750407e-01 -4.44457471e-01
8.97176623e-01 -3.98398697e-01 -6.81801975e-01 2.27986023e-01
1.27384663e+00 -1.09650910e+00 -9.35736179e-01 9.67625380e-01
1.08245537e-01 -1.91610634e-01 4.74222541e-01 -1.23383713e+00
-8.09464455e-01 2.99152225e-01 -4.78733778e-01 -2.97821403e-01
4.44260269e-01 2.52283782e-01 1.18039453e+00 -7.01311707e-01
1.15172648e+00 1.48785019e+00 8.13823402e-01 -7.87340477e-03
-1.32897425e+00 -5.05240202e-01 -6.52318954e-01 7.30010420e-02
-9.95353401e-01 -3.33988011e-01 7.37962961e-01 -4.38216120e-01
1.30051923e+00 -2.24540606e-01 9.14506495e-01 9.20245469e-01
7.80659139e-01 5.49768507e-01 8.44470859e-01 -4.17586893e-01
-1.58350348e-01 -5.02343953e-01 -3.62670332e-01 8.40783358e-01
4.38729636e-02 5.18193364e-01 -4.84840661e-01 7.27261081e-02
1.61376786e+00 2.13186562e-01 -2.22179294e-01 -9.90233004e-01
-1.59645152e+00 7.52782047e-01 9.59859014e-01 -8.35266635e-02
-2.57150948e-01 8.33660305e-01 -6.34966865e-02 4.20439690e-01
-1.87292337e-01 5.53276420e-01 2.49295123e-02 -1.84308994e-03
-9.64005709e-01 2.53777504e-01 8.94395530e-01 1.12778175e+00
1.27416778e+00 -1.41953245e-01 5.55845737e-01 3.64871174e-01
2.86904484e-01 4.92983192e-01 3.64058673e-01 -1.54992199e+00
2.80747712e-01 3.93948495e-01 -1.08960338e-01 -1.28301895e+00
-2.48825774e-01 -1.06612213e-01 -6.34483993e-01 4.65265512e-01
2.62640446e-01 7.76921539e-03 -1.05509305e+00 1.34141576e+00
-1.53944092e-02 -9.81004909e-03 -2.00038165e-01 1.08213496e+00
3.53592962e-01 4.62215722e-01 -6.98641717e-01 6.14371359e-01
1.07952189e+00 -1.04481292e+00 -3.67984086e-01 -5.10290325e-01
3.42705011e-01 -1.14731467e+00 6.82131648e-01 2.14989930e-01
-1.20491242e+00 -4.38229561e-01 -1.65450573e+00 -9.40478265e-01
-5.41077495e-01 3.15187871e-01 7.22792923e-01 8.40139315e-02
-1.42307425e+00 1.02119565e+00 -1.02936006e+00 -4.91567075e-01
1.04813777e-01 6.78795516e-01 -7.14033842e-01 1.99569285e-01
-9.38087106e-01 9.12449121e-01 4.05931234e-01 4.41473983e-02
-6.46448016e-01 -3.75455350e-01 -1.18562603e+00 3.10179204e-01
-2.41720960e-01 -9.00882125e-01 1.27057934e+00 -1.00318539e+00
-1.67735481e+00 1.14756751e+00 2.09859654e-01 -6.66233718e-01
7.32380509e-01 -2.13482410e-01 6.16702959e-02 3.76793981e-01
2.29164988e-01 1.27692461e+00 9.97337997e-01 -1.08670044e+00
-5.58812261e-01 -1.58270612e-01 4.74936850e-02 3.25213104e-01
2.14772478e-01 -3.88377905e-01 -6.39863729e-01 -3.68618935e-01
7.07939208e-01 -1.18404400e+00 -3.27311009e-02 3.18766177e-01
-5.83944917e-01 4.68265355e-01 9.05203223e-01 -6.70187175e-01
5.45802593e-01 -1.96754777e+00 1.02772601e-01 3.87887627e-01
3.62454146e-01 -2.63226271e-01 -9.91451591e-02 5.78333378e-01
-3.22638214e-01 -3.18841726e-01 -6.16189092e-02 -1.28933266e-01
1.48159832e-01 -3.84385325e-02 -5.99041939e-01 8.14423859e-01
1.88271314e-01 1.08948672e+00 -7.96442747e-01 -4.78984788e-02
6.22923315e-01 7.10879683e-01 -5.84715843e-01 3.38322461e-01
4.84155603e-02 5.79064935e-02 2.67991900e-01 4.11643863e-01
7.75291681e-01 -3.65220159e-01 3.51078421e-01 -6.89142466e-01
-1.32342681e-01 3.83624017e-01 -9.29306507e-01 1.73464942e+00
-5.01995504e-01 1.17349982e+00 -1.56125858e-01 -3.98578078e-01
1.14032078e+00 -6.38833791e-02 4.82283443e-01 -4.97114778e-01
5.65315008e-01 3.72307569e-01 9.03440863e-02 4.79593351e-02
6.45365000e-01 1.62849084e-01 -6.32383153e-02 3.84899944e-01
4.92830902e-01 -5.87846637e-01 -1.96809676e-02 9.87565517e-02
8.82669866e-01 5.42708635e-01 5.82295060e-01 -1.88543797e-01
1.35634005e-01 -9.99679193e-02 6.78296164e-02 4.30984557e-01
-2.01268002e-01 1.05901670e+00 8.80057812e-01 -9.84802186e-01
-1.56379712e+00 -1.14075196e+00 5.72558939e-02 4.65781271e-01
3.29368412e-01 -3.44264627e-01 -5.92007518e-01 -1.54766262e-01
3.62088561e-01 8.19412544e-02 -8.22584152e-01 -1.16407767e-01
-8.11702907e-01 -4.71023668e-04 2.84299850e-01 6.07217431e-01
1.00238478e+00 -6.66215718e-01 -7.96181858e-01 3.88985164e-02
6.99534044e-02 -1.30663466e+00 -7.19737172e-01 3.45217913e-01
-7.83699095e-01 -1.21759307e+00 -5.36030173e-01 -1.04975140e+00
7.27195799e-01 2.11234644e-01 1.03858161e+00 -1.97042331e-01
-1.85117766e-01 2.39690974e-01 2.84133911e-01 2.87071049e-01
-3.50317270e-01 1.13960735e-01 -1.49606332e-01 -3.94757539e-02
4.42847133e-01 -7.92325139e-01 -9.60550249e-01 4.67329323e-01
-6.00792289e-01 2.04957977e-01 7.60816276e-01 9.13283467e-01
5.88651776e-01 -7.98540413e-01 -3.76335561e-01 -6.00771964e-01
2.14987814e-01 -2.52912007e-02 -1.08891010e+00 8.40903148e-02
-2.00043544e-01 2.27525428e-01 6.30929589e-01 -1.37593254e-01
-5.55116296e-01 7.35718071e-01 1.10791393e-01 -8.66291225e-01
1.25363424e-01 -9.87239107e-02 1.14868872e-01 -6.12340689e-01
6.94647372e-01 -2.52770465e-02 1.27660424e-01 -2.10542232e-01
7.50817120e-01 3.11589837e-01 1.28890681e+00 -1.06567219e-01
1.04938817e+00 7.79129267e-01 2.83848077e-01 -5.49381495e-01
-2.19824210e-01 -4.02785152e-01 -1.34361196e+00 8.88793170e-02
9.64459419e-01 -9.80096698e-01 -9.37883794e-01 4.27003026e-01
-1.33412373e+00 -3.57148618e-01 1.39274662e-02 5.35969913e-01
-1.08153319e+00 4.15763855e-01 -7.20265031e-01 2.67370176e-02
-3.14037919e-01 -1.39025581e+00 1.56710672e+00 5.60498163e-02
-2.21265420e-01 -1.00013232e+00 2.14577168e-01 -9.95062441e-02
1.39877439e-01 3.75295401e-01 5.81221998e-01 -1.95453197e-01
-1.26904690e+00 -4.80688930e-01 -5.14487743e-01 1.51836455e-01
-7.90538117e-02 1.84145808e-01 -9.52218294e-01 -2.96046108e-01
-1.12071440e-01 -5.61049700e-01 8.64221871e-01 5.37260175e-01
8.05157125e-01 -1.99085936e-01 -1.57146469e-01 1.49953640e+00
1.45587265e+00 3.43222134e-02 8.93931508e-01 7.28207231e-01
9.88482416e-01 2.40532354e-01 1.74326584e-01 -4.06497670e-03
3.84188384e-01 6.57096267e-01 6.59065306e-01 -2.50006318e-01
-3.17916542e-01 -6.93380654e-01 1.88212618e-01 6.61167026e-01
6.12570532e-02 2.19862714e-01 -7.03091264e-01 3.93721312e-01
-1.70739162e+00 -6.56508029e-01 1.79150760e-01 2.22219706e+00
4.94516373e-01 -2.85436790e-02 -1.95282057e-01 -2.64180243e-01
4.00924861e-01 4.79904026e-01 -6.14712596e-01 -4.88013744e-01
-8.37410390e-02 -6.56174729e-04 1.00360310e+00 8.71083498e-01
-1.36162567e+00 1.24210250e+00 7.32132769e+00 1.41214639e-01
-1.39218879e+00 -5.30687094e-01 3.28283608e-01 2.62910604e-01
-9.12572145e-02 4.34833854e-01 -5.31222224e-01 7.00058192e-02
5.56699276e-01 5.53412884e-02 8.71848166e-01 1.04348409e+00
-3.73651266e-01 -1.29129766e-02 -1.46181285e+00 1.25695980e+00
2.93444842e-01 -1.55285168e+00 9.67153087e-02 3.46772403e-01
1.05300272e+00 4.15435731e-01 1.65807396e-01 -2.14378074e-01
6.65260613e-01 -8.97203684e-01 4.90716428e-01 3.38837296e-01
1.07135808e+00 -4.60032701e-01 4.92538303e-01 -2.46156499e-01
-1.07285643e+00 1.66805446e-01 -5.73556423e-01 1.42298834e-02
1.74191356e-01 -3.44617330e-02 -1.14478517e+00 1.99510068e-01
6.58458948e-01 9.22461271e-01 -4.38920170e-01 8.08294356e-01
-3.00084472e-01 -1.39376014e-01 -4.21621114e-01 4.58095580e-01
5.18460035e-01 -4.78174150e-01 2.66965687e-01 9.54205811e-01
3.92911941e-01 -2.62810051e-01 -1.00930810e-01 1.12576449e+00
-4.83757854e-01 -3.88942778e-01 -1.00886166e+00 7.38085210e-02
3.13122362e-01 1.30987966e+00 -7.66642272e-01 -3.99456561e-01
-3.36159497e-01 1.36631966e+00 1.99127287e-01 3.71089578e-01
-4.39213037e-01 -9.10916686e-01 6.46699250e-01 -8.06980506e-02
4.15907562e-01 -5.07146537e-01 -3.64383817e-01 -1.42936647e+00
-2.45857552e-01 -8.63500953e-01 -1.37264067e-02 -1.09840882e+00
-8.03360105e-01 4.63406265e-01 -1.18467897e-01 -1.27706444e+00
-8.80834341e-01 -1.06993484e+00 -7.13302195e-01 8.16996396e-01
-1.19054902e+00 -1.27884626e+00 -5.83482325e-01 3.08858752e-01
1.98416963e-01 -9.92054641e-02 5.59558690e-01 4.76656258e-02
3.78573723e-02 4.07546729e-01 1.31674007e-01 4.96307254e-01
9.28222120e-01 -1.40046382e+00 1.30523789e+00 8.64788651e-01
9.64356884e-02 8.56653094e-01 6.24739707e-01 -5.62767088e-01
-1.62116313e+00 -6.46492243e-01 8.96487296e-01 -5.83543837e-01
8.37827504e-01 -5.63726544e-01 -4.96235073e-01 1.44819295e+00
4.67060864e-01 9.38166454e-02 -2.88000070e-02 -3.88868719e-01
-6.69286728e-01 -1.41153574e-01 -7.71454334e-01 7.87819624e-01
9.21029568e-01 -9.30731356e-01 -3.12418103e-01 1.90626442e-01
4.16698992e-01 -9.43266451e-01 -8.57512116e-01 3.67152505e-02
1.11026490e+00 -1.39900851e+00 1.17956126e+00 -1.82843208e-01
8.23040426e-01 -2.48839676e-01 -2.44812816e-01 -1.25234747e+00
-3.66865337e-01 -9.10136342e-01 6.96939975e-02 2.47522637e-01
-9.98392468e-04 -6.84085071e-01 8.54150057e-01 3.19508404e-01
2.87625641e-02 -4.78534430e-01 -7.53285587e-01 -5.36393225e-01
7.98457935e-02 1.50286734e-01 6.15731478e-01 8.34918857e-01
-6.68030977e-02 2.88361371e-01 -5.18070281e-01 1.91924840e-01
7.03211308e-01 3.55637878e-01 1.22382081e+00 -6.52940571e-01
-3.53768498e-01 -6.08350515e-01 -1.04900181e+00 -1.57515275e+00
-4.31079604e-02 -7.92288780e-01 6.41894341e-03 -1.08964527e+00
-5.37795052e-02 3.97620499e-01 3.42515916e-01 4.53540295e-01
4.12492096e-01 6.65062666e-01 2.55023897e-01 4.34515893e-01
-5.77036431e-03 3.21082175e-01 1.23031855e+00 -4.23910804e-02
-7.56029040e-02 -4.72956270e-01 -1.43041283e-01 7.03130662e-01
5.30829012e-01 -1.86305977e-02 -2.59846926e-01 -6.34648561e-01
1.19925454e-01 1.86813995e-02 6.06582224e-01 -1.08148730e+00
3.93011600e-01 1.90739617e-01 9.91214573e-01 -7.30281889e-01
4.59274113e-01 -7.97480762e-01 2.01582551e-01 5.12772560e-01
-2.90715009e-01 3.20915967e-01 -1.49289489e-01 1.74907848e-01
-1.74776688e-01 2.28307366e-01 8.77633989e-01 -1.76230058e-01
-6.27077162e-01 4.03383940e-01 1.92412585e-01 -4.73172396e-01
6.63602352e-01 -3.64525318e-01 -4.83896673e-01 -7.70552933e-01
-4.72877055e-01 1.98291782e-02 1.07137191e+00 4.26926672e-01
6.32274687e-01 -1.47595024e+00 -1.08616240e-01 8.04009855e-01
-8.63880366e-02 -2.77282089e-01 -1.60252884e-01 5.37989736e-01
-1.51893520e+00 5.43937981e-01 -8.59913290e-01 -7.89480805e-01
-8.36820602e-01 3.57109338e-01 8.24762046e-01 4.09283526e-02
-7.85785556e-01 2.95804590e-01 4.14874792e-01 -6.06672287e-01
1.27293184e-01 -5.85778654e-01 3.61708909e-01 -4.42178667e-01
1.74225271e-01 4.34313327e-01 -4.41680439e-02 -8.99656653e-01
-9.92903784e-02 9.55591977e-01 -6.20059446e-02 -4.18266177e-01
1.08814692e+00 6.59812381e-03 -2.00483382e-01 7.13328347e-02
1.93959427e+00 -3.81753504e-01 -1.80868101e+00 -1.12707214e-02
-2.17979982e-01 -6.51091576e-01 -3.48508311e-03 -5.50078273e-01
-1.11776042e+00 1.19734871e+00 5.12054205e-01 -1.38053611e-01
7.26852834e-01 -2.82166600e-01 9.30510342e-01 6.86916530e-01
7.70208389e-02 -8.78079534e-01 8.49823430e-02 6.79836452e-01
1.08427119e+00 -1.24183941e+00 2.19294474e-01 -2.05403000e-01
-3.93195122e-01 1.81299758e+00 5.24921596e-01 -9.88414586e-01
5.31235218e-01 2.07744092e-01 4.75388229e-01 -3.28180552e-01
-4.65981632e-01 -6.35044277e-02 3.80575836e-01 5.29202104e-01
3.26265097e-01 -4.59005199e-02 1.71400309e-01 -4.23051119e-01
-8.17728400e-01 -1.59120634e-01 5.64322472e-01 6.98544443e-01
-3.70300144e-01 -1.07725644e+00 -4.74703044e-01 4.63754125e-02
5.23097254e-02 4.78265062e-02 -7.54655957e-01 1.08851922e+00
-2.08290100e-01 1.69922918e-01 4.41834480e-01 -6.32385671e-01
1.58292487e-01 -2.60131091e-01 6.35791123e-01 -1.90381303e-01
-3.33685368e-01 2.83069968e-01 -2.57146716e-01 -1.16799343e+00
-8.50832984e-02 -3.25354367e-01 -9.64316130e-01 -5.71552396e-01
2.00515673e-01 -6.17669225e-01 7.76571631e-01 5.29687583e-01
3.26024592e-01 1.46016842e-02 6.99205935e-01 -1.60268486e+00
-6.24735594e-01 -6.82324350e-01 -3.80303919e-01 3.88006628e-01
7.82257080e-01 -3.58675808e-01 -4.66530174e-01 2.15586066e-01]
|
[8.447640419006348, -2.1880438327789307]
|
71cfb4ab-2caf-481e-91bc-93a10a7c6eee
|
superpixel-meshes-for-fast-edge-preserving
| null | null |
http://openaccess.thecvf.com/content_cvpr_2015/html/Bodis-Szomoru_Superpixel_Meshes_for_2015_CVPR_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2015/papers/Bodis-Szomoru_Superpixel_Meshes_for_2015_CVPR_paper.pdf
|
Superpixel Meshes for Fast Edge-Preserving Surface Reconstruction
|
Multi-View-Stereo (MVS) methods aim for the highest detail possible, however, such detail is often not required. In this work, we propose a novel surface reconstruction method based on image edges, superpixels and second-order smoothness constraints, producing meshes comparable to classic MVS surfaces in quality but orders of magnitudes faster. Our method performs per-view dense depth optimization directly over sparse 3D Ground Control Points (GCPs), hence, removing the need for view pairing, image rectification, and stereo depth estimation, and allowing for full per-image parallelization. We use Structure-from-Motion (SfM) points as GCPs, but the method is not specific to these, e.g.~LiDAR or RGB-D can also be used. The resulting meshes are compact and inherently edge-aligned with image gradients, enabling good-quality lightweight per-face flat renderings. Our experiments demonstrate on a variety of 3D datasets the superiority in speed and competitive surface quality.
|
['Luc van Gool', 'Andras Bodis-Szomoru', 'Hayko Riemenschneider']
|
2015-06-01
| null | null | null |
cvpr-2015-6
|
['stereo-depth-estimation']
|
['computer-vision']
|
[ 4.92067367e-01 8.75484198e-02 1.63713261e-01 -1.94109395e-01
-8.22634339e-01 -4.39536870e-01 4.76496994e-01 5.81667619e-03
-1.88495278e-01 5.69595575e-01 3.64100486e-02 3.34806442e-02
2.15470418e-01 -9.30144370e-01 -8.27390075e-01 -4.25737709e-01
1.78479061e-01 4.93218839e-01 6.22308552e-01 -2.60472707e-02
5.66478968e-01 8.83171439e-01 -1.92988634e+00 3.24480295e-01
8.86678040e-01 1.11814213e+00 2.43132353e-01 7.60726035e-01
-1.29227817e-01 1.87375486e-01 7.69574642e-02 -3.33531797e-01
4.74881679e-01 -6.06375299e-02 -5.87138057e-01 4.45243418e-01
1.17890739e+00 -6.74874365e-01 2.89212763e-02 7.61521101e-01
2.94493735e-01 -7.72545263e-02 3.87001455e-01 -8.69886756e-01
9.63483453e-02 -7.51187503e-01 -9.42511559e-01 -3.33860904e-01
7.92011917e-01 2.31747236e-02 8.70399892e-01 -1.27803028e+00
9.99978244e-01 1.28168023e+00 7.40999699e-01 4.62855071e-01
-1.50809026e+00 -2.84947902e-01 1.35647118e-01 -1.21314034e-01
-1.35565364e+00 -5.66887677e-01 8.97128463e-01 -4.06397671e-01
1.13290775e+00 5.86701334e-01 8.09342027e-01 5.83899796e-01
3.14164132e-01 3.73611897e-01 1.28181112e+00 -2.69990146e-01
2.73281902e-01 -7.99578577e-02 -2.94276685e-01 8.71303499e-01
1.58983335e-01 1.91301033e-02 -8.55895042e-01 -2.56373703e-01
1.35666680e+00 1.38904840e-01 -5.83517373e-01 -7.99151778e-01
-1.19658160e+00 4.83099878e-01 3.37913215e-01 -1.78682908e-01
-2.97552437e-01 2.39858150e-01 8.02515000e-02 5.86910993e-02
8.31108272e-01 2.99325194e-02 -4.44700360e-01 5.53829744e-02
-1.06376839e+00 3.07855487e-01 5.61740696e-01 1.10050511e+00
1.33683348e+00 -1.75513327e-01 5.06554425e-01 6.31478786e-01
2.28301674e-01 6.26188338e-01 -2.78221458e-01 -1.80163431e+00
4.65385765e-01 6.55765593e-01 2.60794818e-01 -1.00117087e+00
-2.40059748e-01 1.05340987e-01 -8.41379225e-01 8.09145868e-01
7.18033984e-02 3.73712420e-01 -9.18269753e-01 1.04115629e+00
7.56160676e-01 9.06282589e-02 -2.89614320e-01 7.61605561e-01
5.76475620e-01 4.43953872e-01 -5.98150969e-01 -2.45915785e-01
1.21255171e+00 -6.83498681e-01 -2.86066115e-01 -3.83767575e-01
3.19081366e-01 -8.07123125e-01 1.05872321e+00 5.80525041e-01
-1.50282276e+00 -2.32356817e-01 -8.53741527e-01 -5.57464957e-01
1.29672378e-01 -3.37821811e-01 6.09539807e-01 3.08740258e-01
-1.30620801e+00 9.38982010e-01 -9.12500978e-01 -1.68127850e-01
3.32141489e-01 3.50075513e-01 -7.45655715e-01 -4.30060536e-01
-3.92913997e-01 4.47383255e-01 -2.56970704e-01 -1.28548741e-01
-5.41332066e-01 -1.06058574e+00 -9.68759239e-01 -1.24126881e-01
1.77771226e-01 -1.10619509e+00 8.42864871e-01 -8.97873759e-01
-1.47599220e+00 1.30553591e+00 -6.07457817e-01 1.09086908e-01
7.04374015e-01 -1.60979971e-01 1.24370888e-01 4.68295783e-01
7.64962137e-02 7.56116748e-01 8.29839706e-01 -1.55801857e+00
-5.63831925e-01 -6.92398012e-01 5.21625243e-02 5.31421363e-01
1.68007296e-02 -4.11018550e-01 -6.99467778e-01 -1.97961718e-01
5.96976280e-01 -6.06988847e-01 -3.25941086e-01 6.63031161e-01
-2.81657159e-01 3.25460494e-01 6.88982010e-01 -7.09785521e-01
8.67483079e-01 -2.23854065e+00 3.25371742e-01 2.88843900e-01
3.11609179e-01 -3.65097910e-01 1.34325728e-01 3.29912663e-01
3.18042397e-01 -1.09150097e-01 -5.33419371e-01 -8.05728793e-01
-3.44572276e-01 3.63281906e-01 2.23064587e-01 6.08913958e-01
6.07057624e-02 5.26359379e-01 -7.64842570e-01 -5.11133015e-01
4.74039376e-01 7.52978086e-01 -9.75214124e-01 1.82211787e-01
-1.90246895e-01 7.78228760e-01 -2.45380163e-01 7.54658163e-01
9.76301193e-01 -2.71380305e-01 3.80118154e-02 -4.36608523e-01
-5.08368313e-01 3.06304008e-01 -1.49056339e+00 2.07087302e+00
-7.97739923e-01 4.97619003e-01 7.32332706e-01 -2.91978627e-01
6.11891508e-01 1.01912357e-01 5.49136519e-01 -5.99478245e-01
-2.49953240e-01 5.38134158e-01 -6.91238821e-01 5.46186790e-02
5.39273322e-01 -6.17204234e-02 4.13047791e-01 -1.64764572e-03
-3.88683587e-01 -6.60571337e-01 -8.35938677e-02 1.89855292e-01
8.12174380e-01 3.64758313e-01 1.70046106e-01 -4.36327964e-01
5.22214413e-01 -1.15896864e-02 6.11784577e-01 4.14347872e-02
4.55722004e-01 1.07150197e+00 2.32165262e-01 -2.83767015e-01
-1.29347670e+00 -1.19847214e+00 -4.93547440e-01 2.77397037e-01
4.49220300e-01 -5.39449990e-01 -6.53379738e-01 -4.20722924e-02
2.29767919e-01 3.90880793e-01 -3.04951221e-01 5.74783802e-01
-6.71996117e-01 -1.79066598e-01 -2.92356431e-01 4.48547661e-01
4.66362476e-01 -4.13465261e-01 -8.20289552e-01 1.65087998e-01
2.42980588e-02 -1.17083192e+00 -5.47906399e-01 -2.09786236e-01
-1.48954487e+00 -1.17479205e+00 -7.87102282e-01 -4.75183696e-01
9.62352157e-01 5.12610555e-01 1.41999388e+00 1.16879195e-01
-3.19660008e-01 7.24246562e-01 2.62410074e-01 3.07815045e-01
-2.18275264e-02 -4.12210166e-01 -2.02390954e-01 2.25901529e-02
-3.24841231e-01 -1.14121509e+00 -1.13324821e+00 4.00288880e-01
-6.64654076e-01 7.28017092e-01 2.14372367e-01 6.57846153e-01
1.11130118e+00 -4.11871344e-01 -3.30248415e-01 -8.73210013e-01
-2.92160064e-01 5.38315102e-02 -9.34783101e-01 -1.67083412e-01
-6.05433583e-01 -2.96764314e-01 3.12113106e-01 1.54276088e-01
-1.07103395e+00 2.15852022e-01 -2.43694842e-01 -6.01834238e-01
-9.82429162e-02 -1.14685282e-01 -2.16604576e-01 -5.24930596e-01
3.51663798e-01 1.44870698e-01 -2.27285586e-02 -6.49881959e-01
2.39806205e-01 3.91868114e-01 3.22353452e-01 -3.22912365e-01
5.32343924e-01 1.20572007e+00 4.30952579e-01 -1.24611509e+00
-5.04386902e-01 -6.00951195e-01 -8.08871388e-01 -3.36423546e-01
9.43534315e-01 -1.01871669e+00 -6.95943713e-01 3.84456605e-01
-1.31610024e+00 -3.87346029e-01 -1.83760449e-01 2.58056015e-01
-7.37117112e-01 6.40840709e-01 -7.53645599e-01 -7.43498623e-01
-3.09017450e-01 -1.14680314e+00 1.71524930e+00 -1.83819413e-01
-5.28351590e-02 -9.89770830e-01 -1.86335906e-01 5.29898763e-01
1.78736791e-01 5.47672331e-01 6.49309635e-01 7.73994684e-01
-1.29141104e+00 1.88861191e-01 -3.18663538e-01 3.00036252e-01
4.47703786e-02 1.02568150e-01 -1.18221402e+00 -3.11947256e-01
1.41334921e-01 -5.13515398e-02 5.81731081e-01 3.95057023e-01
1.02349865e+00 -1.21614270e-01 -3.74301076e-01 1.22981524e+00
1.89376962e+00 -3.24499071e-01 8.46538663e-01 1.98807061e-01
1.26479518e+00 7.46835053e-01 5.88950992e-01 3.60249937e-01
6.14807665e-01 8.76800895e-01 6.75549328e-01 -2.87568182e-01
-2.28933901e-01 -1.32460147e-01 1.67774633e-01 8.37711692e-01
-4.32991594e-01 1.11431122e-01 -7.43861079e-01 3.55642170e-01
-1.48679554e+00 -6.04472995e-01 -8.84683967e-01 2.58199549e+00
5.37981749e-01 -9.55862850e-02 -2.11346298e-01 1.48376971e-01
3.41163486e-01 8.86731371e-02 -5.00793457e-01 -3.38959187e-01
-6.46361038e-02 2.42098108e-01 8.64297926e-01 1.08563209e+00
-6.75899744e-01 6.84451640e-01 5.71698761e+00 7.13463008e-01
-8.49903524e-01 1.75259337e-01 6.34176791e-01 -3.02122205e-01
-8.95923674e-01 1.44749194e-01 -7.83954620e-01 1.65870011e-01
3.52880895e-01 2.64504552e-01 4.73121077e-01 7.41710782e-01
1.20552368e-01 -5.44348657e-01 -1.05195689e+00 1.52333653e+00
3.75197418e-02 -1.41935933e+00 5.53012304e-02 3.85305792e-01
9.87813413e-01 1.57320261e-01 -3.70781809e-01 -6.05977178e-01
-2.10468903e-01 -6.80867434e-01 6.80836320e-01 3.18312109e-01
1.24542046e+00 -7.47204840e-01 4.79928553e-02 3.09119076e-01
-1.36775959e+00 3.96641284e-01 -2.74574310e-01 7.01240897e-02
5.42282164e-01 9.39722180e-01 -2.34834924e-02 7.75833726e-01
7.98321962e-01 7.97054231e-01 -1.06055506e-01 6.50595307e-01
4.12860587e-02 -1.86473459e-01 -5.66608489e-01 5.68686724e-01
-5.05881347e-02 -6.35084212e-01 4.82297122e-01 7.94699311e-01
4.46365684e-01 2.34902173e-01 4.45283465e-02 6.68475807e-01
1.36559367e-01 1.73181370e-01 -7.03939080e-01 5.96129417e-01
1.18247002e-01 1.30227900e+00 -8.89662147e-01 -3.72009456e-01
-7.26311207e-01 1.66894448e+00 3.34303558e-01 1.98506668e-01
-3.46253663e-01 2.44808290e-02 9.80670154e-01 8.50283265e-01
2.03713581e-01 -5.37724435e-01 -5.68372488e-01 -1.39928520e+00
3.22705448e-01 -5.21965086e-01 -9.92825069e-03 -7.39989698e-01
-1.11258316e+00 3.42434257e-01 -2.09140748e-01 -1.20921516e+00
-8.80603865e-02 -6.81975782e-01 -1.91709533e-01 1.08133066e+00
-1.86332595e+00 -1.01738489e+00 -5.95241964e-01 7.32926369e-01
5.39635360e-01 7.35710979e-01 6.44706666e-01 4.83510137e-01
1.03120394e-01 -2.93494519e-02 -3.64050902e-02 -5.75257897e-01
4.29910809e-01 -1.17292702e+00 7.06472576e-01 5.76242208e-01
-1.95394129e-01 4.04240072e-01 2.63440281e-01 -8.16416800e-01
-1.87647402e+00 -7.91413248e-01 7.49513626e-01 -4.45682138e-01
1.19595721e-01 -5.58130562e-01 -8.74302983e-01 4.75696474e-01
-1.83640942e-02 3.00313812e-02 3.26826185e-01 -1.83472902e-01
-1.86757058e-01 -5.16912378e-02 -1.32552147e+00 5.79357743e-01
1.58365357e+00 -5.77335954e-01 -6.24821242e-03 3.32298338e-01
3.42436105e-01 -8.91039610e-01 -9.36558843e-01 4.51159269e-01
6.01920128e-01 -1.61160779e+00 1.22574198e+00 1.91556439e-01
4.36429799e-01 -4.47292954e-01 -2.83931673e-01 -9.80383873e-01
-2.17697844e-01 -5.96878886e-01 -3.83316278e-02 9.55942810e-01
3.35167758e-02 -7.48561084e-01 9.32034731e-01 6.99552536e-01
-4.72635269e-01 -7.97472835e-01 -1.10549974e+00 -5.85276902e-01
-4.25289363e-01 -3.64719212e-01 2.76015967e-01 9.49365973e-01
-5.10631919e-01 1.27847940e-01 -2.18101203e-01 3.59645635e-01
1.17933547e+00 3.86526048e-01 9.17063773e-01 -1.38005459e+00
-4.86132443e-01 -2.82311857e-01 -4.40319479e-01 -1.52510655e+00
-3.19047958e-01 -7.17579782e-01 -1.49043635e-01 -1.80142558e+00
-1.58224642e-01 -4.99981314e-01 5.35459816e-01 -4.39674854e-02
2.82791145e-02 5.83664238e-01 -2.26620033e-01 1.08961172e-01
-7.27702603e-02 4.73547846e-01 1.52978456e+00 3.47467333e-01
-8.00269768e-02 -3.22473139e-01 -1.38711989e-01 9.82532680e-01
3.32471758e-01 1.40233319e-02 -4.33997542e-01 -6.80033445e-01
3.76637429e-01 2.95998394e-01 4.67275918e-01 -8.17956388e-01
3.23771383e-03 -2.24245355e-01 4.05400574e-01 -6.27248228e-01
9.66710627e-01 -9.25010264e-01 5.19809842e-01 2.61792034e-01
5.15179992e-01 2.60562263e-02 7.42395148e-02 5.12853384e-01
-8.29694644e-02 1.12647712e-01 9.90999162e-01 -3.84809315e-01
-6.82983398e-01 6.27053559e-01 2.22819448e-01 -8.02758336e-03
9.14053619e-01 -7.44412482e-01 2.01461822e-01 -1.77121505e-01
-4.91901875e-01 7.41232932e-02 1.51187253e+00 -2.05254748e-01
9.39818442e-01 -1.30937767e+00 -5.24227083e-01 4.98030514e-01
8.08806643e-02 5.52107453e-01 5.08129239e-01 7.94937134e-01
-1.08631277e+00 1.19088747e-01 3.28240730e-02 -1.06290114e+00
-1.41147554e+00 2.43092030e-01 2.47040719e-01 7.65159875e-02
-1.09474158e+00 8.65731657e-01 6.48045659e-01 -3.51112992e-01
-7.80283958e-02 -2.99363583e-01 3.89687240e-01 -2.09647164e-01
4.81874049e-01 5.97150087e-01 3.18280041e-01 -5.15747726e-01
-3.38934332e-01 1.26612663e+00 3.45926732e-01 -2.18007565e-01
1.41868043e+00 -3.00266206e-01 -2.92127669e-01 4.29899812e-01
1.21681643e+00 4.00357425e-01 -1.73055363e+00 2.43758157e-01
-3.08374792e-01 -1.07889390e+00 3.15516055e-01 -8.26926082e-02
-1.09256470e+00 1.04563594e+00 3.37630898e-01 -1.51751891e-01
1.11677921e+00 -1.57989949e-01 9.67059076e-01 -1.80075034e-01
9.83071089e-01 -9.04528201e-01 -2.84353614e-01 1.53173029e-01
8.78402770e-01 -1.12263560e+00 2.58839399e-01 -1.15311468e+00
-1.13639943e-01 1.12481320e+00 4.52383876e-01 -2.88342208e-01
5.72439432e-01 2.72433847e-01 -2.07909673e-01 -4.13196534e-01
-4.58034843e-01 5.30784763e-02 2.56757945e-01 5.36150515e-01
4.30419266e-01 -2.16481835e-01 -1.77334130e-01 -5.21565676e-01
1.68426678e-01 -2.26132423e-01 2.95743883e-01 1.00580990e+00
-3.67752373e-01 -1.22708070e+00 -4.15842235e-01 5.06351113e-01
-1.72780648e-01 -2.56639034e-01 -3.28326458e-03 5.90893567e-01
-1.51266411e-01 6.52780831e-01 4.10647362e-01 -1.70212463e-02
6.04196191e-01 -3.15016955e-01 8.76206875e-01 -7.25536227e-01
-2.85978407e-01 1.90168723e-01 2.23247468e-01 -1.29825902e+00
-3.95952433e-01 -8.27307284e-01 -1.10413182e+00 -5.37189245e-01
-1.03750557e-01 -3.01331729e-01 8.74595582e-01 4.13534164e-01
7.29642391e-01 2.91777961e-02 8.24917853e-01 -1.50338280e+00
1.38127312e-01 -3.51243407e-01 -6.52145982e-01 4.68269378e-01
3.47459286e-01 -6.58706903e-01 -6.56418681e-01 -7.32657835e-02]
|
[9.048332214355469, -2.9267799854278564]
|
7be328a3-99b4-46fa-a407-63655b497345
|
temporal-inductive-logic-reasoning
|
2206.05051
| null |
https://arxiv.org/abs/2206.05051v1
|
https://arxiv.org/pdf/2206.05051v1.pdf
|
Temporal Inductive Logic Reasoning
|
Inductive logic reasoning is one of the fundamental tasks on graphs, which seeks to generalize patterns from the data. This task has been studied extensively for traditional graph datasets such as knowledge graphs (KGs), with representative techniques such as inductive logic programming (ILP). Existing ILP methods typically assume learning from KGs with static facts and binary relations. Beyond KGs, graph structures are widely present in other applications such as video instructions, scene graphs and program executions. While inductive logic reasoning is also beneficial for these applications, applying ILP to the corresponding graphs is nontrivial: they are more complex than KGs, which usually involve timestamps and n-ary relations, effectively a type of hypergraph with temporal events. In this work, we study two of such applications and propose to represent them as hypergraphs with time intervals. To reason on this graph, we propose the multi-start random B-walk that traverses this hypergraph. Combining it with a path-consistency algorithm, we propose an efficient backward-chaining ILP method that learns logic rules by generalizing from both the temporal and the relational data.
|
['Faramarz Fekri', 'James C Kerce', 'Siheng Xiong', 'Yuan Yang']
|
2022-06-09
| null | null | null | null |
['inductive-logic-programming']
|
['methodology']
|
[ 1.14662826e-01 5.27011514e-01 -7.83463538e-01 -5.55653453e-01
1.92675993e-01 -5.74472547e-01 5.02029836e-01 7.62385428e-01
9.14982185e-02 8.25254142e-01 -2.10790619e-01 -9.63513196e-01
-7.21796930e-01 -1.50544882e+00 -1.18699241e+00 -2.28914097e-01
-7.86671221e-01 6.58497036e-01 8.77354145e-01 -1.76971838e-01
1.12576254e-01 4.75414932e-01 -1.54870331e+00 3.48570675e-01
6.69870734e-01 8.70581150e-01 -4.25655067e-01 3.51811141e-01
-4.59518671e-01 1.56108081e+00 -1.33569658e-01 -5.61366022e-01
4.96030748e-02 -1.61549538e-01 -1.25028634e+00 -2.91617274e-01
1.35049835e-01 4.81309332e-02 -5.10338366e-01 1.08361256e+00
-2.40302250e-01 2.22406924e-01 3.22875500e-01 -1.97576904e+00
-4.52807188e-01 1.37565637e+00 -5.79369545e-01 4.37236764e-02
7.05571294e-01 -1.51824817e-01 1.28765750e+00 -2.02590805e-02
7.87971079e-01 1.42476141e+00 5.23165822e-01 1.14974804e-01
-1.27569568e+00 -5.45497537e-01 4.66342211e-01 9.74689662e-01
-1.27122271e+00 5.52639253e-02 7.22226679e-01 -2.49264777e-01
9.05514121e-01 4.10545051e-01 6.68847859e-01 5.84796131e-01
2.27410600e-01 9.08646226e-01 1.13827074e+00 -5.20642102e-01
4.08975184e-01 -1.10060856e-01 6.27023160e-01 9.23742890e-01
6.38941407e-01 -1.88243285e-01 -6.85193181e-01 -1.92138761e-01
5.53861022e-01 9.57573354e-02 -1.96120694e-01 -5.40657640e-01
-1.00913000e+00 6.82639778e-01 4.76261616e-01 2.37299040e-01
-4.14964706e-02 4.91525799e-01 6.37730062e-01 4.00188267e-01
2.58976649e-02 1.66996405e-01 -5.87138951e-01 1.18339188e-01
-5.17001390e-01 3.25345367e-01 1.46361232e+00 1.39474452e+00
8.96947920e-01 -4.03190821e-01 4.41170633e-02 1.23404838e-01
3.47637504e-01 2.25144431e-01 -2.74356026e-02 -5.63467383e-01
3.01828265e-01 1.04617453e+00 -3.38559806e-01 -1.40584314e+00
-3.41593266e-01 2.11952731e-01 -6.38668239e-01 -3.39037895e-01
3.16320926e-01 3.68868381e-01 -8.70239854e-01 1.71404600e+00
4.88062859e-01 6.28376961e-01 -3.42556462e-02 3.22394490e-01
7.79884160e-01 6.72706902e-01 7.05366507e-02 -6.00915313e-01
1.19662011e+00 -5.39132178e-01 -6.91474438e-01 1.56665388e-02
7.45068252e-01 1.92562550e-01 6.72924399e-01 4.02573228e-01
-9.06242669e-01 -1.00356318e-01 -8.95241737e-01 -4.69603203e-03
-7.07764506e-01 -8.67285907e-01 1.13168049e+00 2.70432770e-01
-1.02447677e+00 5.47009766e-01 -8.82721007e-01 -3.16616297e-01
1.61069110e-01 4.65642244e-01 -3.66261333e-01 -4.08247143e-01
-1.53182161e+00 6.73599362e-01 1.20575559e+00 8.99977423e-03
-5.79060555e-01 -6.94462061e-01 -1.14510596e+00 -1.19764833e-02
1.41242611e+00 -5.90502381e-01 1.04542089e+00 -4.54595834e-01
-1.07308269e+00 7.68055797e-01 -2.89046913e-02 -7.02801228e-01
1.84827194e-01 1.10842600e-01 -6.68556988e-01 -1.60219334e-02
-9.82711911e-02 -4.78014313e-02 5.92155397e-01 -1.15246391e+00
-5.57052016e-01 -6.53747857e-01 8.06672633e-01 -3.49533170e-01
2.47085486e-02 -2.87955284e-01 -6.47061646e-01 -3.43248844e-02
3.56761813e-01 -9.97406125e-01 -2.56821632e-01 -5.07900119e-01
-1.02027977e+00 -5.80403984e-01 8.41185570e-01 -2.89859653e-01
1.74233055e+00 -1.83985317e+00 1.23410992e-01 7.82227516e-01
3.41778874e-01 -1.94238365e-01 4.82332408e-01 7.68307567e-01
-1.58066720e-01 2.07171202e-01 -1.22913375e-01 5.36189735e-01
2.66348898e-01 7.90129125e-01 -5.73814631e-01 2.20152840e-01
-5.32690361e-02 1.06421876e+00 -1.18597317e+00 -8.57222259e-01
-1.76394522e-01 -3.26885402e-01 -5.57574391e-01 1.02911808e-01
-1.03110993e+00 9.62010324e-02 -5.25300622e-01 6.31695986e-01
3.61031920e-01 -5.25355577e-01 9.46955562e-01 -2.23957479e-01
8.01199377e-02 1.22312211e-01 -1.35248542e+00 1.62479401e+00
-3.20110649e-01 3.00680339e-01 -4.53801960e-01 -1.20897388e+00
7.59600818e-01 -8.86273477e-03 4.20075268e-01 -5.05925357e-01
-2.50695080e-01 -1.32511377e-01 -1.09672129e-01 -8.79231572e-01
4.16736931e-01 -8.43429416e-02 -2.82568574e-01 4.10738975e-01
-1.45202711e-01 -3.03057760e-01 7.42246807e-01 6.13885939e-01
1.42322266e+00 2.41488755e-01 6.14031255e-01 -2.30290756e-01
7.20773578e-01 1.04888849e-01 8.01330090e-01 7.12032259e-01
4.32439953e-01 -3.30400109e-01 1.21695375e+00 -7.03628182e-01
-2.91216135e-01 -1.05652618e+00 6.54179528e-02 9.01722491e-01
4.23684061e-01 -1.11975050e+00 -2.82282650e-01 -9.95959044e-01
2.73901343e-01 6.89695060e-01 -4.05137151e-01 -3.64672393e-01
-5.98666310e-01 -3.45457852e-01 3.22733611e-01 7.05795348e-01
2.39606351e-01 -8.98409545e-01 -3.40250105e-01 1.49771437e-01
-2.17364922e-01 -1.19264019e+00 1.78804770e-01 2.93671459e-01
-8.00545990e-01 -1.75055563e+00 6.82539344e-01 -6.80108726e-01
6.93061411e-01 -2.59071350e-01 1.53330851e+00 2.81000674e-01
-1.80640608e-01 6.88078642e-01 -3.69282782e-01 -6.83309615e-01
-3.09973359e-01 -4.58343849e-02 2.21565161e-02 -2.61300374e-02
5.68326414e-01 -8.44476700e-01 -8.16204026e-02 1.48482099e-01
-1.29451966e+00 1.77561700e-01 3.75737190e-01 4.45139945e-01
9.66998398e-01 9.22594130e-01 3.55311930e-01 -1.65423071e+00
3.74821723e-01 -7.53370106e-01 -9.62567031e-01 7.45196998e-01
-6.70875430e-01 3.95106971e-01 8.28246117e-01 -3.31146449e-01
-7.66235530e-01 -1.96615472e-01 5.11963367e-01 -3.71441722e-01
-1.20459571e-02 1.20876062e+00 -4.61950570e-01 6.64044498e-03
3.53780210e-01 5.67150442e-03 -4.37720418e-01 2.18323812e-01
5.98256230e-01 6.00431673e-02 5.91880918e-01 -1.19109488e+00
8.54009569e-01 4.30810899e-01 8.09931099e-01 -5.43320298e-01
-7.39241183e-01 -2.68903732e-01 -5.98155558e-01 -1.57790646e-01
5.33417463e-01 -3.37108850e-01 -1.39323962e+00 -1.16138741e-01
-9.33017015e-01 -6.33293033e-01 -3.90261889e-01 3.62463772e-01
-7.64524400e-01 3.18666846e-01 -5.27574599e-01 -8.42797458e-01
1.97234571e-01 -8.45436513e-01 6.80176079e-01 -5.83952814e-02
-2.23815635e-01 -1.36745656e+00 1.53338805e-01 -1.10018045e-01
-1.68541595e-01 7.82097757e-01 1.58323812e+00 -8.11637700e-01
-1.01840699e+00 -6.16083518e-02 -1.34920418e-01 -1.31534696e-01
1.14370793e-01 2.28987068e-01 -3.98382455e-01 5.48234582e-02
-2.68304825e-01 -2.79397607e-01 3.19958687e-01 1.02262706e-01
1.66352355e+00 -6.43948615e-01 -6.58408344e-01 3.04858744e-01
1.62283468e+00 3.82334948e-01 5.52967906e-01 1.57780781e-01
8.95768166e-01 7.49446273e-01 6.70531631e-01 2.95443892e-01
8.14732432e-01 3.48656148e-01 5.90169013e-01 4.11404163e-01
3.19595069e-01 -5.10279536e-01 -3.52291875e-02 7.68812299e-01
-3.04327011e-01 -1.16499782e-01 -1.20679593e+00 4.48740900e-01
-2.30604148e+00 -1.02606642e+00 -5.14115274e-01 2.12236524e+00
1.15651619e+00 2.58709908e-01 -5.97983040e-02 4.20531094e-01
5.75376928e-01 1.80468559e-01 -6.63580000e-01 -3.79081905e-01
5.29068559e-02 2.07834020e-01 5.23975909e-01 4.74857539e-01
-8.76284003e-01 7.67272830e-01 5.54333115e+00 6.41062438e-01
-7.54369676e-01 -4.18362737e-01 2.83484697e-01 2.94452071e-01
-4.94525403e-01 5.15304744e-01 -7.34471142e-01 2.96811491e-01
1.09108543e+00 -5.86250901e-01 6.81693554e-01 9.12870169e-01
-3.47347766e-01 -2.71640807e-01 -1.90504885e+00 9.37891901e-01
-6.53074682e-02 -1.24950182e+00 1.10388771e-01 -1.98965028e-01
7.60744691e-01 -6.82470202e-01 -3.50718439e-01 5.31573415e-01
7.23114908e-01 -1.00045061e+00 5.57676435e-01 6.25734329e-01
5.88434815e-01 -7.01722801e-01 4.06857789e-01 3.34364027e-01
-1.60487080e+00 -1.79377049e-01 -3.63578051e-02 -6.05657324e-03
1.20166004e-01 8.71961713e-01 -8.17409277e-01 1.35359097e+00
6.76487565e-01 1.04534745e+00 -4.07381982e-01 7.10473001e-01
-6.25124753e-01 5.20127833e-01 -3.40223223e-01 1.49344355e-01
-1.24273062e-01 -2.99228817e-01 1.97125778e-01 9.77636278e-01
-3.97186875e-02 3.68475288e-01 3.52845907e-01 5.13055682e-01
-2.91361064e-01 -2.38048851e-01 -1.08722591e+00 -7.91949555e-02
4.19632256e-01 8.99054229e-01 -8.94995272e-01 -5.46312928e-01
-6.55829906e-01 1.01111129e-01 3.58723551e-01 4.38370436e-01
-1.05260813e+00 -2.45721117e-01 3.40921581e-01 3.74026597e-01
7.64955953e-02 -2.39785552e-01 4.13790718e-02 -9.97363985e-01
1.85703233e-01 -7.82706618e-01 9.86194015e-01 -6.74970865e-01
-1.29747522e+00 1.99997723e-01 6.83087587e-01 -8.22943032e-01
-3.07334661e-01 -5.14967382e-01 -4.08722252e-01 2.78323233e-01
-1.52503800e+00 -1.14645636e+00 -4.15346444e-01 1.09834504e+00
-1.32883057e-01 5.57507932e-01 5.58946431e-01 6.47932738e-02
-4.23224390e-01 7.36934394e-02 -6.47115052e-01 -2.32638717e-02
5.04347086e-01 -1.55308473e+00 -1.02628432e-01 8.59581888e-01
2.83939511e-01 8.66255999e-01 6.52372539e-01 -8.20672333e-01
-2.16315341e+00 -1.33981287e+00 7.53146112e-01 -2.91873872e-01
1.17504764e+00 -1.02321520e-01 -1.16363144e+00 1.57378650e+00
-9.57758725e-03 3.91088903e-01 4.94053841e-01 5.17822206e-01
-6.59520686e-01 -5.09421706e-01 -8.10429633e-01 7.32876360e-01
1.52930415e+00 -5.87760150e-01 -7.18426824e-01 5.78897476e-01
8.95154834e-01 -5.89459598e-01 -1.15350580e+00 7.04592824e-01
2.55142540e-01 -7.55084693e-01 9.68463123e-01 -1.04261005e+00
3.79698366e-01 -7.57335901e-01 -2.17282102e-02 -1.06095719e+00
-1.73045903e-01 -6.78019345e-01 -8.04512322e-01 1.03288412e+00
2.67500609e-01 -7.14419842e-01 7.11266577e-01 7.26365864e-01
6.11765273e-02 -7.90032446e-01 -5.54412007e-01 -1.05706358e+00
-5.26446819e-01 -6.60586536e-01 8.81483018e-01 1.18915844e+00
5.63010156e-01 3.95623356e-01 -3.37786414e-02 4.44843054e-01
6.66378617e-01 6.52263641e-01 7.92379618e-01 -1.51614177e+00
-4.99246716e-01 -1.82425648e-01 -6.63621664e-01 -5.99851012e-01
6.24947190e-01 -1.17477131e+00 1.54771423e-02 -1.62164307e+00
2.89896913e-02 -6.63220525e-01 -1.20133750e-01 8.23840976e-01
1.06206678e-01 -4.45273995e-01 -2.27276817e-01 -8.28499570e-02
-1.02384782e+00 5.09986505e-02 1.10609198e+00 -4.47740048e-01
-2.37158656e-01 -5.47075905e-02 -3.93986762e-01 7.39753604e-01
4.78035152e-01 -2.64323503e-01 -1.09417665e+00 -8.83120075e-02
1.03742945e+00 5.47730267e-01 2.49888152e-01 -7.35029519e-01
8.62961411e-01 -8.25859129e-01 -3.86901587e-01 -5.53802013e-01
-1.80625081e-01 -1.14152193e+00 8.03392112e-01 4.25945073e-01
-2.28588000e-01 1.05468400e-01 2.40972802e-01 9.12698805e-01
-5.27784228e-01 -4.64002863e-02 2.57074863e-01 -1.45358384e-01
-1.23619640e+00 6.26026869e-01 2.01717064e-01 1.54333651e-01
1.49658668e+00 6.42127395e-02 -4.80389774e-01 -2.18447179e-01
-7.92786121e-01 5.45430064e-01 3.56993377e-01 1.76379398e-01
5.12020826e-01 -1.24540138e+00 -1.33519441e-01 -9.56456363e-02
4.38522816e-01 7.16630101e-01 1.68167263e-01 1.06173611e+00
-4.62311745e-01 3.79741460e-01 9.55476165e-02 -4.66902554e-01
-1.12825286e+00 1.20904124e+00 -1.23489462e-01 -6.68900788e-01
-5.35618544e-01 3.60675991e-01 2.70157576e-01 -6.33680522e-02
2.67549336e-01 -8.02802026e-01 -2.75004238e-01 -7.42330402e-02
4.41292703e-01 3.84098619e-01 9.60993022e-03 3.02974768e-02
-6.26705885e-01 3.51212949e-01 8.30494333e-03 4.33380216e-01
1.27387297e+00 1.37232438e-01 -8.09894443e-01 8.78128529e-01
6.05890036e-01 6.16417676e-02 -6.21409893e-01 -5.64759791e-01
6.16450667e-01 -3.12168807e-01 -6.16159558e-01 -4.21950489e-01
-9.76909816e-01 3.67641091e-01 -5.78882635e-01 8.61645401e-01
1.27015364e+00 3.04876715e-01 4.43063200e-01 6.95079327e-01
9.62079287e-01 -8.63759339e-01 -9.72303823e-02 6.38801336e-01
5.76749682e-01 -1.03552568e+00 1.95300430e-01 -7.98933148e-01
-2.10073888e-01 1.18487954e+00 6.43849671e-01 1.68695629e-01
7.57977784e-01 4.48765814e-01 -6.92547202e-01 -4.69765425e-01
-8.40294540e-01 -1.75485283e-01 -8.33345130e-02 4.49376971e-01
7.23235831e-02 3.76423985e-01 -3.21393132e-01 5.73588371e-01
-6.61558881e-02 3.90094846e-01 5.98064005e-01 1.35284019e+00
4.22056392e-02 -1.26096416e+00 -1.65986672e-01 5.90963423e-01
-2.28563458e-01 1.78849604e-02 -1.91065386e-01 1.01513839e+00
2.45851781e-02 7.24553466e-01 -4.20971699e-02 -4.24293756e-01
4.46851015e-01 4.19664159e-02 6.87399507e-01 -7.81778812e-01
-1.18569978e-01 -8.98909390e-01 3.17887098e-01 -8.74299288e-01
-6.03877127e-01 -4.05001462e-01 -1.55079138e+00 -5.99360704e-01
-9.50829238e-02 3.38622361e-01 3.68000388e-01 8.99016917e-01
5.08452691e-02 7.19789922e-01 3.77364904e-01 -6.92635914e-03
3.63767557e-02 -3.36893618e-01 -9.31461394e-01 5.14157414e-01
-5.33222407e-02 -6.96940899e-01 -9.10080895e-02 3.06545377e-01]
|
[8.855167388916016, 7.680927276611328]
|
c97042e6-0c81-43a7-98ec-d35a07335e2d
|
nlp-workbench-efficient-and-extensible
|
2303.01410
| null |
https://arxiv.org/abs/2303.01410v1
|
https://arxiv.org/pdf/2303.01410v1.pdf
|
NLP Workbench: Efficient and Extensible Integration of State-of-the-art Text Mining Tools
|
NLP Workbench is a web-based platform for text mining that allows non-expert users to obtain semantic understanding of large-scale corpora using state-of-the-art text mining models. The platform is built upon latest pre-trained models and open source systems from academia that provide semantic analysis functionalities, including but not limited to entity linking, sentiment analysis, semantic parsing, and relation extraction. Its extensible design enables researchers and developers to smoothly replace an existing model or integrate a new one. To improve efficiency, we employ a microservice architecture that facilitates allocation of acceleration hardware and parallelization of computation. This paper presents the architecture of NLP Workbench and discusses the challenges we faced in designing it. We also discuss diverse use cases of NLP Workbench and the benefits of using it over other approaches. The platform is under active development, with its source code released under the MIT license. A website and a short video demonstrating our platform are also available.
|
['Denilson Barbosa', 'Natalie Hervieux', 'Kostyantyn Guzhva', 'Abeer Waheed', 'Matej Kosmajac', 'Peiran Yao']
|
2023-03-02
| null | null | null | null |
['semantic-parsing']
|
['natural-language-processing']
|
[-1.84609473e-01 3.39921832e-01 -3.82708490e-01 -6.17357671e-01
-5.48725188e-01 -6.73003614e-01 3.48020971e-01 3.85359347e-01
-3.48748028e-01 5.55940330e-01 9.50975493e-02 -5.17568707e-01
1.43724948e-01 -8.21392238e-01 -3.82729441e-01 -1.25397697e-01
4.45192724e-01 8.23617756e-01 5.25292695e-01 -3.25202078e-01
4.18154120e-01 2.27404520e-01 -1.75246418e+00 4.74233210e-01
5.67130268e-01 9.41076934e-01 3.56158584e-01 4.01501566e-01
-9.47904170e-01 4.85261023e-01 -4.35341030e-01 -5.63361406e-01
1.75999537e-01 1.46077976e-01 -1.09927201e+00 -4.65932608e-01
1.35360003e-01 1.01371743e-01 2.15553522e-01 1.04565489e+00
3.11054885e-01 -2.02871054e-01 3.21227685e-02 -1.46479344e+00
-3.81231815e-01 7.67910719e-01 -3.33403528e-01 8.32254663e-02
6.55371785e-01 -4.95402813e-01 1.15918052e+00 -6.44335210e-01
8.99483502e-01 8.95818293e-01 6.58137739e-01 4.41099882e-01
-7.12094545e-01 -7.24756062e-01 4.52422313e-02 2.48078287e-01
-1.18017137e+00 -4.31482941e-01 5.24082601e-01 -1.20262340e-01
1.83565760e+00 3.09542954e-01 5.82852066e-01 9.49727714e-01
1.63237050e-01 8.07125688e-01 7.45257020e-01 -7.64974117e-01
2.39914626e-01 5.92802942e-01 6.59027934e-01 8.74853313e-01
4.58211184e-01 -5.68653882e-01 -7.36946523e-01 -4.55472589e-01
1.95108816e-01 -9.83178467e-02 3.00327003e-01 -2.40087405e-01
-8.30132067e-01 5.28213799e-01 -4.01662402e-02 5.90936959e-01
-2.21822694e-01 -2.20952421e-01 5.97575188e-01 2.45762497e-01
4.59711343e-01 3.46233279e-01 -1.05706275e+00 -5.38527429e-01
-9.24920678e-01 2.16505200e-01 1.40583706e+00 1.18457580e+00
7.14343786e-01 -3.89926195e-01 6.51608765e-01 7.86453545e-01
2.72424728e-01 1.03893399e-01 8.19935322e-01 -7.12586701e-01
4.09014374e-01 1.17045605e+00 3.73251773e-02 -6.99720263e-01
-7.28179157e-01 -2.15097032e-02 7.40864724e-02 -1.53369665e-01
1.12153530e-01 -2.38621414e-01 -4.79981869e-01 1.10523546e+00
5.56958795e-01 -9.22391862e-02 2.60147661e-01 5.65107286e-01
1.03563762e+00 4.40971166e-01 2.45021939e-01 6.63324445e-02
1.76534975e+00 -1.09179533e+00 -7.89131463e-01 -3.17602873e-01
8.03077638e-01 -9.55464840e-01 1.13423574e+00 3.63936096e-01
-1.00695753e+00 -1.46241874e-01 -1.03266740e+00 -4.16325659e-01
-1.08897948e+00 8.28747153e-02 1.02030814e+00 8.85565400e-01
-9.54265356e-01 3.87248397e-01 -1.13588572e+00 -1.02260351e+00
2.37706512e-01 4.92730111e-01 -4.34896857e-01 4.51161206e-01
-1.06363964e+00 7.77437031e-01 7.44015753e-01 -5.20512640e-01
-3.66061106e-02 -7.38106072e-01 -7.02520967e-01 1.76880077e-01
5.03623664e-01 -5.95374048e-01 1.48721874e+00 -9.20539439e-01
-1.52343273e+00 1.05659115e+00 -2.71287233e-01 -4.34378207e-01
6.45596385e-02 -4.01324749e-01 -6.45259023e-01 8.04472491e-02
2.30004519e-01 4.45549667e-01 3.35635930e-01 -5.54512382e-01
-9.03299451e-01 -5.47036946e-01 -1.81619495e-01 1.08409032e-01
-9.63687181e-01 5.36056519e-01 -6.90106452e-01 -2.34989464e-01
-2.76004881e-01 -6.80924535e-01 4.20761220e-02 -3.36919814e-01
-2.88164794e-01 -3.46516162e-01 1.06558394e+00 -5.49379528e-01
1.30477035e+00 -1.99135876e+00 -5.70387065e-01 2.30056912e-01
-1.28157809e-01 2.13582277e-01 8.89701173e-02 9.42607462e-01
3.95113192e-02 1.75333574e-01 1.70411006e-01 -1.61599457e-01
1.61582306e-01 1.54002860e-01 -8.36442336e-02 -1.22418121e-01
-1.73392624e-01 7.01620281e-01 -7.16348767e-01 -6.35668933e-01
2.56267756e-01 2.74607569e-01 -1.96305647e-01 -3.45047146e-01
-4.91644263e-01 -8.10804442e-02 -7.61604071e-01 7.09755540e-01
4.56095904e-01 -5.39127767e-01 7.23875999e-01 1.97427962e-02
-3.88900846e-01 6.51450813e-01 -1.10572720e+00 2.01514363e+00
-6.14362657e-01 4.72964525e-01 2.21008763e-01 -8.96459758e-01
1.05030036e+00 3.66176873e-01 5.40289402e-01 -4.77374494e-01
2.51436085e-01 3.02844524e-01 -5.10089099e-01 -6.75865591e-01
7.08369911e-01 2.44293064e-01 2.70628929e-02 7.09742665e-01
2.30021730e-01 6.45441636e-02 3.99030566e-01 1.53773114e-01
1.33062959e+00 4.10256624e-01 6.55513048e-01 -4.37537223e-01
4.98754263e-01 5.96411645e-01 4.10678327e-01 2.98873097e-01
-1.24627277e-02 -1.95638672e-01 3.67708504e-01 -6.51048541e-01
-9.66251612e-01 -8.03144336e-01 -3.03473443e-01 1.30098665e+00
-3.22721787e-02 -1.00581169e+00 -8.20674181e-01 -8.13301504e-01
-1.31437898e-01 6.51943445e-01 -1.51354030e-01 5.32768726e-01
-3.62197578e-01 -6.88006163e-01 4.61505592e-01 5.46693206e-01
6.18051887e-01 -1.06180561e+00 -7.10400164e-01 1.62778676e-01
-2.29354277e-01 -1.46564567e+00 4.64450940e-02 1.65757701e-01
-8.03730011e-01 -1.14063668e+00 1.65008947e-01 -1.27190268e+00
4.97901827e-01 1.40507206e-01 1.12787485e+00 -1.14110380e-01
-4.05819714e-01 3.54546845e-01 -5.55830479e-01 -8.93031061e-01
-2.63425618e-01 5.82421184e-01 -1.81916952e-01 -5.93815506e-01
1.20003986e+00 -4.78802264e-01 -3.84569675e-01 3.29474628e-01
-7.06402659e-01 2.30101794e-01 1.73730463e-01 2.81755656e-01
5.12741566e-01 2.76283026e-01 6.23428345e-01 -1.18370938e+00
6.43591881e-01 -5.92693090e-01 -7.18637407e-01 3.58592749e-01
-9.54086840e-01 -8.86266381e-02 5.04723907e-01 1.09548084e-01
-1.32031441e+00 2.46817604e-01 -4.20420736e-01 2.44453982e-01
-2.93510675e-01 5.91495931e-01 -2.05066189e-01 8.90715644e-02
4.81784016e-01 -3.29166323e-01 -5.36630005e-02 -7.13506579e-01
2.57543802e-01 1.09754682e+00 3.64809841e-01 -5.08394063e-01
1.45151243e-01 6.88537419e-01 -3.78848374e-01 -8.40804338e-01
-7.57399082e-01 -7.75511265e-01 -3.36241186e-01 1.33386657e-01
6.86047137e-01 -8.96760881e-01 -7.51436174e-01 1.96384974e-02
-9.19439793e-01 -1.23413943e-01 -1.25426382e-01 2.01139942e-01
-2.83377439e-01 1.11277834e-01 -6.27125800e-01 -4.11928058e-01
-8.33503723e-01 -7.32089520e-01 1.06759882e+00 4.75920647e-01
-6.58233821e-01 -1.08724225e+00 1.05093963e-01 9.23578382e-01
2.09943086e-01 -2.14432981e-02 8.30772400e-01 -1.28015959e+00
-1.19709544e-01 -4.19147640e-01 -7.19964430e-02 1.22096859e-01
-1.35416500e-02 4.22137417e-02 -9.61637974e-01 8.92762616e-02
-1.43093690e-01 -1.57048911e-01 3.73008817e-01 1.36202812e-01
9.81809974e-01 -1.37937561e-01 -9.31990802e-01 2.68705457e-01
1.36777103e+00 2.44720653e-01 4.45345730e-01 9.78310585e-01
3.92154843e-01 8.77509296e-01 7.85109997e-01 4.32359934e-01
6.04999125e-01 5.58228970e-01 1.40234157e-01 1.47091091e-01
2.88716108e-01 -2.77751952e-01 2.11944699e-01 8.82594466e-01
5.82821518e-02 -4.64080200e-02 -1.28605306e+00 4.55764413e-01
-2.18077254e+00 -6.44830108e-01 -3.95191461e-01 1.77753341e+00
6.99877918e-01 2.04882637e-01 1.10910691e-01 1.24563091e-02
6.12707555e-01 -3.64608854e-01 -2.59190649e-01 -7.86288500e-01
5.15200086e-02 5.74759960e-01 3.74924004e-01 2.70917743e-01
-1.05050457e+00 1.32523203e+00 6.35469532e+00 8.09080303e-01
-1.04574668e+00 3.73118371e-01 1.47440001e-01 -1.49556175e-01
-5.04790731e-02 2.66109079e-01 -1.17755866e+00 2.68445700e-01
1.31960869e+00 -3.92656744e-01 1.92861289e-01 1.28227639e+00
1.92904919e-01 -8.09021667e-02 -8.24953914e-01 6.77704811e-01
-4.50700894e-02 -1.74746287e+00 -3.39679331e-01 -9.06441137e-02
4.19718921e-01 6.23650253e-01 -4.56629157e-01 2.12032646e-01
3.42205197e-01 -5.14192939e-01 4.56791073e-01 1.20538130e-01
3.63627076e-01 -6.08561635e-01 7.63704419e-01 2.42630064e-01
-1.23942471e+00 1.03854530e-01 -1.50643528e-01 -2.14310646e-01
9.07057151e-02 5.21405339e-01 -9.52604651e-01 7.45755434e-01
1.21007848e+00 5.50101221e-01 -5.45570135e-01 4.75526690e-01
-2.50259370e-01 3.67166162e-01 -4.65620935e-01 -2.63225317e-01
-1.52025625e-01 -1.74063891e-01 3.38243097e-01 1.37698889e+00
2.25705639e-01 -2.58022547e-01 1.66412756e-01 4.53232586e-01
1.17743053e-02 6.25376046e-01 -4.48102534e-01 -1.28917232e-01
6.09162152e-01 1.74780440e+00 -1.00965166e+00 -4.24297094e-01
-9.29139435e-01 7.44669139e-01 2.40364894e-01 -6.00559600e-02
-5.81929743e-01 -7.92065918e-01 6.62889421e-01 6.55754805e-02
3.82442951e-01 2.40177941e-03 -5.29777944e-01 -1.27607191e+00
1.51856542e-01 -8.39025021e-01 6.53382123e-01 -8.24077368e-01
-1.04790699e+00 6.63355827e-01 1.98301976e-03 -6.48346901e-01
-2.76773363e-01 -8.60248625e-01 -3.95562410e-01 5.24989605e-01
-1.23511410e+00 -1.32302594e+00 -2.89324105e-01 5.65784395e-01
5.47259092e-01 -2.77277827e-01 1.20551920e+00 2.22309753e-01
-5.86966991e-01 2.14757219e-01 5.20855263e-02 1.24323659e-01
7.21151292e-01 -1.08962214e+00 6.55704081e-01 6.45241857e-01
2.92234063e-01 7.43406177e-01 6.30116463e-01 -7.65878379e-01
-1.46653867e+00 -9.23708916e-01 1.35446978e+00 -4.74444628e-01
1.17974830e+00 -5.40908217e-01 -6.03001833e-01 9.28215563e-01
3.06896925e-01 -2.50955820e-01 1.16930270e+00 4.16730970e-01
-3.11366111e-01 -2.12847188e-01 -1.23750663e+00 4.42278206e-01
8.05803895e-01 -4.11250591e-01 -8.00864577e-01 3.25236171e-01
5.45856714e-01 -2.17040583e-01 -9.37072515e-01 1.31796613e-01
7.03190088e-01 -9.07463133e-01 6.53425336e-01 -1.91182584e-01
1.86277717e-01 -2.33133674e-01 8.97863731e-02 -6.89287782e-01
2.27570742e-01 -8.21662486e-01 -1.25696927e-01 1.31811404e+00
6.97201014e-01 -8.55250895e-01 1.15142882e+00 6.20582759e-01
-7.61603639e-02 -5.48990250e-01 -5.83848476e-01 -6.64616227e-01
-2.00488329e-01 -8.54870319e-01 8.03551316e-01 1.07009292e+00
8.97345722e-01 5.80857515e-01 3.35295618e-01 1.48779646e-01
2.92869598e-01 3.11708272e-01 6.91645861e-01 -1.38962865e+00
-3.64581138e-01 -2.98822999e-01 -2.99517542e-01 -5.60887873e-01
3.20059747e-01 -1.16604018e+00 -5.68674207e-01 -1.68003321e+00
1.20895348e-01 -2.19976187e-01 -9.51272696e-02 1.01912355e+00
2.80030847e-01 1.74870625e-01 -1.01180434e-01 2.49550432e-01
-6.64433300e-01 -1.00451306e-01 6.13972127e-01 1.04797363e-01
-3.71898174e-01 -2.99412787e-01 -9.41825032e-01 8.70387316e-01
1.23824394e+00 -5.29280305e-01 -3.80086869e-01 -2.86530048e-01
6.29165471e-01 -3.84089142e-01 -9.90075096e-02 -8.67209911e-01
3.75935495e-01 6.49207979e-02 1.25080928e-01 -4.14155364e-01
1.55978099e-01 -9.01471794e-01 1.10679783e-01 1.65715978e-01
-1.00175645e-02 1.79587424e-01 4.48507965e-01 7.11708143e-02
-2.34408736e-01 -4.81754571e-01 2.46431708e-01 -2.01205239e-01
-9.53345537e-01 -9.44345593e-02 -3.42595696e-01 -1.99529245e-01
1.23037457e+00 -3.41774784e-02 -6.48158133e-01 3.74426022e-02
-6.66240275e-01 3.79781067e-01 5.25166631e-01 5.85331023e-01
2.02536225e-01 -6.94623053e-01 -1.39404804e-01 1.97372958e-01
2.28852838e-01 -3.38461190e-01 -1.48112759e-01 6.44160867e-01
-8.64558399e-01 5.89272857e-01 -1.33839026e-01 -6.34717569e-02
-1.62217951e+00 5.07393301e-01 -1.30358547e-01 -5.20856202e-01
-6.96708500e-01 4.54311490e-01 -3.36839259e-01 -5.22778511e-01
4.14749831e-02 -1.16032228e-01 -2.00548366e-01 -1.52039185e-01
7.30921209e-01 3.46549898e-01 5.30396104e-01 -7.54912272e-02
-7.20332384e-01 3.10698181e-01 -3.13394278e-01 -8.58341604e-02
1.34086156e+00 -2.17579335e-01 -2.99960941e-01 5.24936259e-01
8.97745430e-01 -1.14516066e-02 -4.79322046e-01 1.26817718e-01
3.81153494e-01 9.56503004e-02 1.62521929e-01 -1.06467116e+00
-6.85730934e-01 4.94845986e-01 2.98025936e-01 4.25762415e-01
1.08778679e+00 1.64292440e-01 1.03494942e+00 7.60472834e-01
5.52823544e-01 -1.74701476e+00 -4.18473542e-01 5.87795436e-01
3.51432562e-01 -1.13923931e+00 3.33400369e-02 -7.18218029e-01
-6.06569707e-01 1.36567271e+00 4.78967667e-01 4.19612229e-02
9.54636395e-01 8.81287098e-01 3.73898417e-01 -4.11988854e-01
-9.14845705e-01 -2.62238026e-01 -2.66862154e-01 5.76812625e-01
7.99790442e-01 3.99239212e-02 -6.30936086e-01 8.10252368e-01
-5.57721734e-01 3.28193814e-01 2.20065296e-01 1.41097772e+00
-3.01595837e-01 -1.78517115e+00 -1.28522351e-01 3.80758137e-01
-8.45756352e-01 -2.70475656e-01 -5.94626665e-01 8.92709672e-01
4.87082750e-02 1.13484609e+00 2.35493362e-01 -2.66946018e-01
1.97547153e-01 4.46208030e-01 3.15998852e-01 -7.41466463e-01
-1.01737094e+00 -4.91977260e-02 7.20055223e-01 -6.91800356e-01
-4.13810819e-01 -5.97648025e-01 -1.72386813e+00 -3.75476748e-01
-2.08480850e-01 4.78918493e-01 1.35424709e+00 8.07009041e-01
8.35624397e-01 1.76640332e-01 1.08909924e-02 -3.11731547e-01
1.26830846e-01 -8.52434874e-01 -3.85923743e-01 1.01543911e-01
-5.01300216e-01 -3.34490567e-01 1.35025857e-02 2.71584660e-01]
|
[9.507044792175293, 8.794059753417969]
|
58fb5467-5851-4033-9956-e0665bc9c64b
|
lbmt-team-at-vlsp2022-abmusu-hybrid-method
|
2304.05205
| null |
https://arxiv.org/abs/2304.05205v1
|
https://arxiv.org/pdf/2304.05205v1.pdf
|
LBMT team at VLSP2022-Abmusu: Hybrid method with text correlation and generative models for Vietnamese multi-document summarization
|
Multi-document summarization is challenging because the summaries should not only describe the most important information from all documents but also provide a coherent interpretation of the documents. This paper proposes a method for multi-document summarization based on cluster similarity. In the extractive method we use hybrid model based on a modified version of the PageRank algorithm and a text correlation considerations mechanism. After generating summaries by selecting the most important sentences from each cluster, we apply BARTpho and ViT5 to construct the abstractive models. Both extractive and abstractive approaches were considered in this study. The proposed method achieves competitive results in VLSP 2022 competition.
|
['Thi-Hai-Yen Vuong', 'Ha-Thanh Nguyen', 'Tam Doan Thanh', 'Hai-Long Nguyen', 'Hoang-Trung Nguyen', 'Thai-Binh Nguyen', 'Tan-Minh Nguyen']
|
2023-04-11
| null | null | null | null |
['multi-document-summarization', 'document-summarization']
|
['natural-language-processing', 'natural-language-processing']
|
[ 2.96898782e-01 2.52257921e-02 -1.64207757e-01 -1.76457852e-01
-1.05453849e+00 -5.64378142e-01 6.76809251e-01 8.33820343e-01
-2.84904420e-01 1.12261605e+00 1.12490869e+00 2.21813560e-01
-4.50978279e-01 -4.70292866e-01 -1.58244491e-01 -5.33688188e-01
4.00656015e-02 6.72579050e-01 3.91963243e-01 -1.34895980e-01
1.11799228e+00 1.27408400e-01 -1.22972167e+00 8.10543001e-01
1.35976696e+00 3.19796264e-01 4.55482006e-01 1.19423664e+00
-4.82655078e-01 6.61118865e-01 -1.24079597e+00 -7.78686181e-02
-1.29818663e-01 -8.71399641e-01 -9.40961242e-01 8.61415789e-02
4.70279455e-01 -1.63798690e-01 8.68688896e-02 8.62416685e-01
7.84178913e-01 3.95001441e-01 1.00940609e+00 -9.16771770e-01
-2.08649606e-01 1.15073931e+00 -1.08563590e+00 4.91295993e-01
6.90031767e-01 -6.76790237e-01 1.27825665e+00 -5.33433378e-01
6.62678301e-01 1.32671106e+00 9.31177214e-02 2.19285086e-01
-7.28933036e-01 -1.49107084e-01 2.03374639e-01 3.86995614e-01
-9.25339103e-01 -3.92217159e-01 6.65147603e-01 4.01974395e-02
1.11879420e+00 8.54187489e-01 6.21808231e-01 4.06426489e-01
5.39159656e-01 1.07950854e+00 6.93086207e-01 -5.71690738e-01
2.39886194e-01 -7.06650913e-02 7.54742026e-01 2.76636332e-01
7.91444302e-01 -1.01351535e+00 -5.04170179e-01 -5.85380018e-01
-1.91770658e-01 -2.18112141e-01 -2.04338372e-01 2.49611855e-01
-1.10546970e+00 7.27134883e-01 -8.06167945e-02 5.54623604e-01
-7.47472286e-01 -4.87508997e-02 7.35037506e-01 1.56174265e-02
4.04698044e-01 5.22395253e-01 -6.30224198e-02 5.51140085e-02
-1.56041932e+00 4.09804881e-01 1.01962006e+00 8.58796358e-01
3.32027674e-01 -1.69873282e-01 -8.89916182e-01 7.42525816e-01
3.65412325e-01 2.49893188e-01 6.53971314e-01 -9.58775401e-01
9.23054397e-01 6.60269737e-01 -6.15525953e-02 -1.14937615e+00
-1.70243636e-01 -4.96996373e-01 -9.93843973e-01 -4.77070481e-01
-6.63309395e-01 -4.41863716e-01 -8.02694619e-01 1.07750893e+00
1.07370242e-01 -1.45031005e-01 4.76091564e-01 3.64367068e-01
1.18720877e+00 1.29417539e+00 -2.31379941e-01 -9.02568996e-01
1.34104776e+00 -1.26741159e+00 -9.15638149e-01 1.56694993e-01
3.19986731e-01 -1.06699133e+00 2.01956257e-01 5.31815290e-01
-1.44443870e+00 -3.15067351e-01 -1.38752711e+00 5.30714877e-02
5.62752690e-03 2.37776279e-01 1.71355665e-01 3.68674815e-01
-1.17823231e+00 6.16444886e-01 -4.54846174e-01 -6.32303655e-01
6.53159618e-02 2.40113661e-01 -9.87173244e-02 2.16108393e-02
-7.49455035e-01 7.57884562e-01 1.13764131e+00 -2.07121477e-01
-2.39676371e-01 -1.53635412e-01 -3.12706947e-01 4.27913964e-01
2.11478382e-01 -1.03435910e+00 1.17932510e+00 -5.04323065e-01
-1.08153379e+00 9.83972624e-02 -4.84198928e-01 -6.71676457e-01
2.98892736e-01 -4.23919886e-01 -6.08208925e-02 8.04005146e-01
2.02299163e-01 2.80679345e-01 3.52533937e-01 -1.31894159e+00
-1.11205912e+00 -2.03173891e-01 -4.46575314e-01 7.85982609e-01
-3.20028275e-01 1.81046829e-01 -6.77514911e-01 -4.95207608e-01
1.92671970e-01 -5.98797619e-01 -2.55474180e-01 -1.18038499e+00
-1.00388646e+00 -4.61115658e-01 9.78190899e-01 -9.19243872e-01
1.72847843e+00 -1.50570679e+00 2.35148683e-01 3.08130890e-01
4.37788457e-01 3.57945651e-01 6.09456562e-03 1.12781215e+00
2.85270631e-01 4.16959584e-01 -7.66237080e-02 -3.18926960e-01
-2.78524160e-01 -3.08060776e-02 -3.07290554e-01 -5.82884178e-02
-2.44656518e-01 4.20105815e-01 -8.48082602e-01 -1.11508465e+00
-7.46100470e-02 -9.43530500e-02 -2.50355244e-01 9.82259586e-02
-1.78185105e-01 -4.45147343e-02 -7.50887692e-01 1.76283598e-01
6.21920705e-01 1.91670969e-01 2.58052915e-01 -2.47274265e-01
-3.01644862e-01 3.13463956e-02 -1.19952369e+00 1.38551724e+00
-1.41945994e-02 6.07250392e-01 -2.87661463e-01 -1.04694009e+00
7.68953204e-01 3.95235062e-01 6.09587431e-01 -1.74183935e-01
9.79045779e-03 1.29990265e-01 -2.14275077e-01 -4.60338384e-01
1.27894700e+00 2.59004474e-01 -1.52005211e-01 5.20896614e-01
-1.04259528e-01 -1.52954072e-01 1.13063920e+00 1.19669104e+00
1.06754363e+00 -2.38946825e-01 7.59155869e-01 -3.92528802e-01
7.43367553e-01 3.32426429e-01 5.38368165e-01 1.01544559e+00
3.45334798e-01 7.17114747e-01 7.61802554e-01 2.16010928e-01
-1.02560210e+00 -8.23489547e-01 4.10158008e-01 7.08827436e-01
1.11261360e-01 -9.47620273e-01 -6.96954548e-01 -6.72308505e-01
-3.47518742e-01 1.01847172e+00 -3.68037969e-01 -7.79666677e-02
-5.41662753e-01 -7.54288554e-01 4.40192580e-01 2.24499032e-01
7.58959174e-01 -7.80937374e-01 -4.26560193e-01 3.53302538e-01
-6.05979979e-01 -6.41965628e-01 -6.42220616e-01 -1.75126478e-01
-8.97619069e-01 -1.00855935e+00 -7.98678517e-01 -7.34660327e-01
6.49333060e-01 6.21623576e-01 7.46949077e-01 1.19589984e-01
1.44820943e-01 1.53908044e-01 -8.92872751e-01 -5.47490716e-01
-6.95844173e-01 4.48930740e-01 -2.01519564e-01 -2.01654419e-01
-1.83554571e-02 -4.06364411e-01 -4.20747578e-01 -3.94246966e-01
-9.39462841e-01 1.46058843e-01 8.65264952e-01 5.85461199e-01
3.56248945e-01 3.06784600e-01 8.91925335e-01 -1.05142403e+00
1.49668419e+00 -4.63599920e-01 8.26078281e-02 7.93444037e-01
-8.35443318e-01 4.36903119e-01 8.55153859e-01 4.14839536e-02
-1.43877161e+00 -2.49521986e-01 3.32338922e-02 1.38189331e-01
1.94422811e-01 7.67530680e-01 3.47154550e-02 7.02782929e-01
3.06751728e-01 5.18482149e-01 -4.39845502e-01 -4.83182698e-01
2.30877578e-01 9.24042284e-01 5.02497792e-01 -4.59547043e-01
5.17495453e-01 1.07030548e-01 -6.46397546e-02 -1.21209502e+00
-6.72291219e-01 -9.23426092e-01 -5.96776664e-01 -2.91324586e-01
8.02473187e-01 -6.08698547e-01 -2.95320392e-01 5.69005050e-02
-1.56786048e+00 7.66759515e-01 1.35669261e-02 6.13077581e-01
-2.67556548e-01 1.07573092e+00 -4.00197625e-01 -7.40240514e-01
-1.19595695e+00 -6.56454504e-01 8.51619601e-01 5.38834274e-01
-4.16062593e-01 -6.45396411e-01 4.20219749e-01 2.98582107e-01
1.46620169e-01 4.71256882e-01 9.42995071e-01 -1.29986382e+00
-1.26891389e-01 -3.07999492e-01 -7.47141466e-02 2.78780669e-01
2.60014325e-01 4.57690060e-01 -2.58816540e-01 -2.49308243e-01
-1.38209343e-01 1.33957744e-01 1.40790856e+00 6.06947541e-01
4.83780950e-01 -7.84524500e-01 -5.45177817e-01 -1.25746414e-01
1.49659264e+00 2.13397592e-01 5.18185556e-01 1.08291760e-01
5.75764000e-01 5.46314359e-01 6.12273932e-01 6.64142013e-01
3.55481565e-01 2.47341961e-01 -1.51088208e-01 3.78611118e-01
-9.67708677e-02 -4.99458984e-02 2.17360646e-01 1.57260919e+00
-8.67477898e-03 -8.65927875e-01 -6.75674438e-01 5.20799041e-01
-2.22725081e+00 -1.26148283e+00 -6.23332322e-01 1.76614833e+00
7.25396574e-01 1.70452923e-01 5.84331229e-02 2.14782551e-01
9.48510945e-01 3.43167633e-01 -1.30386159e-01 -9.61084366e-01
-2.82169312e-01 -6.06706329e-02 3.47684532e-01 5.21585405e-01
-6.87349558e-01 8.32838416e-01 5.93690062e+00 9.73299861e-01
-4.57275152e-01 -1.87339112e-01 3.25796098e-01 -2.98239887e-01
-2.69390166e-01 2.39472911e-01 -9.22763824e-01 3.53543550e-01
8.93939376e-01 -9.49710846e-01 -3.96961838e-01 4.29995269e-01
6.03154361e-01 -7.57305145e-01 -5.89864075e-01 5.30551136e-01
5.62620163e-01 -1.42259467e+00 6.22406960e-01 -9.56594273e-02
1.09390843e+00 -1.98098987e-01 -6.02848113e-01 -1.03374077e-02
3.44423026e-01 -4.08596069e-01 4.85146344e-01 6.20571077e-01
6.45359680e-02 -1.00978172e+00 9.20584023e-01 4.61927742e-01
-1.23958313e+00 1.29374459e-01 -5.34443557e-01 2.35731050e-01
2.64963537e-01 6.14341915e-01 -7.64484286e-01 1.27312589e+00
2.74663717e-01 7.76452303e-01 -6.66027069e-01 1.38967335e+00
-7.34535903e-02 6.67040646e-01 -1.51092917e-01 -5.82321048e-01
3.96596849e-01 -2.40639329e-01 8.88774574e-01 1.55415976e+00
4.36179042e-01 2.25757658e-01 2.29567379e-01 7.09734708e-02
-2.52525322e-02 7.01439917e-01 -3.06233525e-01 1.07170567e-01
5.41063547e-01 1.32505524e+00 -9.99573112e-01 -9.77245033e-01
5.50248288e-02 9.81365144e-01 -1.02303587e-01 4.24684808e-02
-3.30835640e-01 -8.07124436e-01 -3.99811059e-01 -3.57245326e-01
2.34189332e-01 -1.40425831e-01 -2.80352920e-01 -9.86135900e-01
2.76888404e-02 -6.97483540e-01 6.70557261e-01 -6.37314618e-01
-7.28896737e-01 6.26179397e-01 5.19553483e-01 -1.08014297e+00
-3.71897072e-01 3.40026975e-01 -1.29654336e+00 5.91597199e-01
-1.05444789e+00 -8.31181765e-01 -4.71497402e-02 6.03282545e-03
1.06461418e+00 -2.22117752e-01 4.84691590e-01 -2.15779647e-01
-6.18909359e-01 7.17546865e-02 7.35717356e-01 -4.14529383e-01
7.28980958e-01 -1.48267150e+00 1.49698600e-01 1.06339121e+00
1.94851495e-02 5.37994981e-01 1.21578264e+00 -1.04227209e+00
-9.80983734e-01 -8.96280050e-01 1.18122756e+00 1.21005796e-01
1.69611335e-01 2.55868375e-01 -8.12694967e-01 2.28503585e-01
1.29048550e+00 -1.20421350e+00 6.51127875e-01 -1.48538113e-01
2.59135365e-01 -1.38839006e-01 -9.70766544e-01 6.09829068e-01
4.46076691e-01 5.12659609e-01 -1.24677908e+00 3.83109212e-01
6.70568883e-01 4.51437430e-03 -5.72228968e-01 5.98876551e-02
3.72616202e-01 -9.01303589e-01 4.89594847e-01 -3.79250944e-01
6.09823287e-01 -2.36067802e-01 8.13554153e-02 -1.64361095e+00
-3.65384638e-01 -5.78863263e-01 -1.81206852e-01 1.70295513e+00
5.40008307e-01 -1.78998530e-01 5.79701185e-01 5.66296838e-02
-1.91405103e-01 -5.06267548e-01 -4.65065986e-01 -5.32966197e-01
-8.86996984e-02 3.58349770e-01 3.21369648e-01 4.49098229e-01
3.51029694e-01 9.72532868e-01 -3.05080205e-01 -1.19080380e-01
8.16753089e-01 1.35763422e-01 6.20756149e-01 -1.38814688e+00
-4.78706788e-03 -5.83435416e-01 -1.44805328e-03 -7.10624933e-01
-4.23072912e-02 -8.19184065e-01 6.56172261e-02 -2.64891505e+00
8.79740834e-01 4.47517633e-01 -2.34803304e-01 -7.24685192e-02
-4.05983090e-01 -3.07902396e-01 4.15150642e-01 4.51558292e-01
-1.16830969e+00 4.81509656e-01 8.11010182e-01 -2.14500904e-01
-5.15990376e-01 1.51596949e-01 -1.03502393e+00 4.08617407e-01
1.21037281e+00 -6.10057831e-01 -5.25695980e-01 -1.70343533e-01
3.41693014e-02 2.90991247e-01 -4.71155852e-01 -1.16183949e+00
7.50450194e-01 -2.11323112e-01 2.65050560e-01 -1.50214267e+00
-5.94308376e-02 -1.36147305e-01 -1.17844298e-01 6.72671914e-01
-7.66732693e-01 4.64182347e-01 8.34810585e-02 7.15261757e-01
-3.35677207e-01 -6.98736012e-01 4.20629263e-01 -2.80129015e-01
-3.46696019e-01 -1.73247367e-01 -6.82986140e-01 6.41825348e-02
8.30579877e-01 -1.89969271e-01 -5.82555771e-01 -5.16573310e-01
-2.47413233e-01 7.32892931e-01 8.23154673e-02 2.90675312e-01
8.46246600e-01 -1.03776145e+00 -1.35555267e+00 -6.62506044e-01
-1.44070402e-01 -2.24346995e-01 2.19613418e-01 5.78546107e-01
-6.91890597e-01 8.13295841e-01 -2.15663850e-01 -1.33914381e-01
-1.88776994e+00 1.67912915e-01 -4.26586419e-01 -8.03691685e-01
-4.58817780e-01 2.39908889e-01 -2.79565871e-01 3.44819069e-01
6.68212958e-03 -6.26452416e-02 -9.65988219e-01 4.88028258e-01
7.39492834e-01 8.44633579e-01 7.31740966e-02 -5.01319468e-01
-2.46108621e-01 4.27874535e-01 -5.94813585e-01 -5.18950999e-01
1.28828704e+00 -4.00389373e-01 -5.92564702e-01 3.12874377e-01
1.16094124e+00 3.14805299e-01 -3.96492004e-01 -7.75833102e-03
4.73191023e-01 -2.36306078e-04 7.96406344e-02 -7.10424602e-01
-5.22855878e-01 3.99880856e-01 -1.72661692e-01 4.10727531e-01
1.08786166e+00 -2.14567408e-01 8.03241611e-01 6.13531530e-01
-2.68678665e-01 -1.53752482e+00 1.62532866e-01 4.13996667e-01
1.06606638e+00 -7.76467144e-01 8.26173425e-01 -1.23230115e-01
-8.89730036e-01 1.20198929e+00 5.21716356e-01 -5.96037097e-02
1.52409419e-01 -1.20674260e-01 -4.00946796e-01 -1.30247518e-01
-1.13056242e+00 2.63591614e-02 3.72936517e-01 2.31693521e-01
5.13783276e-01 -1.48315400e-01 -1.43102598e+00 5.48219681e-01
-1.66298896e-01 -3.40387791e-01 1.10293663e+00 1.02881968e+00
-1.21344423e+00 -1.03637409e+00 -5.62384307e-01 9.97154772e-01
-6.96353316e-01 -1.03067599e-01 -9.23307896e-01 3.15686435e-01
-4.79967356e-01 1.51888502e+00 -2.06009910e-01 -3.49614769e-01
2.48599797e-01 -7.35851005e-02 3.00921887e-01 -8.32048953e-01
-8.55412602e-01 3.81525904e-01 4.88401026e-01 9.65687633e-02
-4.98094887e-01 -8.38662207e-01 -1.39618325e+00 -3.11432213e-01
-4.81509149e-01 1.03203988e+00 8.04512322e-01 8.26033235e-01
4.31576461e-01 7.55693138e-01 8.64196837e-01 -6.42566442e-01
-5.25527835e-01 -1.35742986e+00 -5.78265250e-01 7.07910359e-02
8.22249576e-02 1.02781355e-01 -1.38721183e-01 -1.17142953e-01]
|
[12.60445499420166, 9.581603050231934]
|
1c281b29-48d4-49bd-8fd2-6945f9205d70
|
community-detection-exact-recovery-in
|
2102.04439
| null |
https://arxiv.org/abs/2102.04439v1
|
https://arxiv.org/pdf/2102.04439v1.pdf
|
Community Detection: Exact Recovery in Weighted Graphs
|
In community detection, the exact recovery of communities (clusters) has been mainly investigated under the general stochastic block model with edges drawn from Bernoulli distributions. This paper considers the exact recovery of communities in a complete graph in which the graph edges are drawn from either a set of Gaussian distributions with community-dependent means and variances, or a set of exponential distributions with community-dependent means. For each case, we introduce a new semi-metric that describes sufficient and necessary conditions of exact recovery. The necessary and sufficient conditions are asymptotically tight. The analysis is also extended to incomplete, fully connected weighted graphs.
|
['Aria Nosratinia', 'Mohammad Esmaeili']
|
2021-02-08
| null | null | null | null |
['stochastic-block-model']
|
['graphs']
|
[ 1.27486408e-01 6.18720278e-02 -9.94012132e-02 6.52569309e-02
-2.82888919e-01 -5.73522151e-01 3.35578203e-01 2.88926184e-01
-1.43195987e-01 7.97030270e-01 -1.35379493e-01 -2.19305754e-01
-4.03586149e-01 -8.19832504e-01 -3.21359158e-01 -1.05340421e+00
-5.96001387e-01 9.53144431e-01 3.57161343e-01 8.72225985e-02
1.98885024e-01 3.31817746e-01 -9.76469278e-01 -4.45834994e-01
7.99371541e-01 2.02177420e-01 8.95075053e-02 1.16380620e+00
2.80137453e-02 6.61792934e-01 -4.77163941e-01 -2.97864974e-01
1.49016231e-01 -3.42150599e-01 -5.73096037e-01 4.19040889e-01
-3.52544218e-01 -2.78500885e-01 -4.98840868e-01 1.47833729e+00
4.01060462e-01 -4.57831830e-01 1.21771693e+00 -1.74363542e+00
-5.44965744e-01 1.00205064e+00 -1.23169041e+00 1.33404359e-01
3.05896193e-01 -3.79724860e-01 8.30466807e-01 -5.48362672e-01
5.59682846e-01 1.16521740e+00 7.43881404e-01 1.47793815e-01
-1.60154688e+00 -7.13705540e-01 -1.05616130e-01 7.98995420e-02
-2.07875538e+00 -2.24125609e-01 4.32413727e-01 -7.71535754e-01
4.59487468e-01 -5.73937967e-03 5.28703570e-01 8.54382753e-01
1.46549463e-01 4.67014641e-01 8.62400889e-01 -5.14265478e-01
4.68814522e-01 4.29760106e-03 3.41906399e-01 2.54590243e-01
1.32476246e+00 3.87633443e-02 8.90273675e-02 -7.88024902e-01
6.04788780e-01 6.58863634e-02 -2.78316021e-01 -8.61309946e-01
-9.46420908e-01 1.13329399e+00 -6.74812719e-02 4.73920107e-01
-4.07771587e-01 1.92894608e-01 1.80464402e-01 2.45363817e-01
2.42391899e-01 -4.78133678e-01 2.02544436e-01 4.45652038e-01
-9.82567668e-01 1.00066856e-01 1.27359819e+00 1.55963385e+00
6.45833790e-01 2.82986537e-02 3.07919923e-02 2.29077935e-01
4.39347565e-01 1.13931334e+00 -4.72302020e-01 -7.30361462e-01
3.52552861e-01 1.40347004e-01 4.19315815e-01 -1.22679222e+00
-1.59458175e-01 -5.04504979e-01 -1.37769544e+00 -7.18032643e-02
6.13597691e-01 -1.44124523e-01 -4.68570858e-01 1.71096468e+00
8.84941220e-02 3.36482823e-01 1.21125486e-02 5.84092200e-01
2.03287870e-01 2.80526400e-01 -1.81466043e-01 -5.60121536e-01
8.49353492e-01 -2.74740726e-01 -8.52674842e-01 1.34149060e-01
1.36653453e-01 -6.04092538e-01 2.74693687e-02 4.26989019e-01
-9.30762827e-01 9.86839309e-02 -9.00764644e-01 6.50123119e-01
1.96619242e-01 -2.51805007e-01 3.22402222e-03 8.89115870e-01
-1.41466296e+00 3.98141116e-01 -6.32610679e-01 -4.86665130e-01
1.08958915e-01 1.74408600e-01 -5.23703098e-01 -4.96671617e-01
-7.10466921e-01 3.15898240e-01 3.57383877e-01 2.58043587e-01
-1.06379271e+00 9.15622711e-02 -7.26630509e-01 1.18982770e-01
1.75815493e-01 -6.06027782e-01 6.32328928e-01 -5.70094109e-01
-4.88099903e-01 7.97333539e-01 -7.10527450e-02 -5.26578248e-01
5.41206419e-01 6.60849631e-01 -1.89095646e-01 6.02533937e-01
2.50894248e-01 -3.37651163e-01 8.98515999e-01 -1.42866850e+00
-2.83116341e-01 -3.56732845e-01 -3.02031994e-01 -2.66298532e-01
-2.35585168e-01 1.52880266e-01 -1.73594415e-01 -4.31142181e-01
2.36240089e-01 -1.09254372e+00 -5.59490621e-01 -1.51839539e-01
-6.12533092e-01 7.04184622e-02 3.60589385e-01 -6.76538765e-01
1.43736434e+00 -2.07915711e+00 3.69708240e-01 7.02964842e-01
6.80365562e-01 -3.17428976e-01 -1.44973993e-01 8.22715521e-01
-1.23785548e-02 3.30145925e-01 -4.94576126e-01 -2.37271711e-01
-8.51510912e-02 2.19912574e-01 2.57323086e-01 1.18918288e+00
-4.30333242e-02 6.80945348e-03 -1.12153327e+00 -6.00291789e-01
-1.44951642e-01 4.94764268e-01 -3.42425883e-01 -1.06936857e-01
5.87483406e-01 1.28573731e-01 -7.28132904e-01 4.05237287e-01
1.26897860e+00 -4.69296604e-01 5.53465307e-01 5.10793984e-01
1.09152026e-01 -5.75334668e-01 -1.69820213e+00 7.75437653e-01
7.97344372e-02 4.47140664e-01 9.38707352e-01 -1.21513450e+00
9.72839952e-01 5.98149598e-01 5.60258269e-01 4.15375829e-01
3.15055609e-01 1.97562724e-01 2.00138852e-01 -2.38428950e-01
2.18959212e-01 -3.34647804e-01 -1.47716314e-01 7.86385238e-01
-1.31556690e-01 2.70760804e-02 4.86130208e-01 7.15655744e-01
1.66003871e+00 -7.15455890e-01 4.46858138e-01 -5.92366755e-01
3.87140125e-01 -2.18251437e-01 5.62106609e-01 1.15170801e+00
-4.24902678e-01 6.79835796e-01 7.36652732e-01 4.35888559e-01
-1.30678153e+00 -1.48291242e+00 -1.01023018e-01 2.63935924e-01
3.27972054e-01 -2.87599325e-01 -6.27085328e-01 -1.75361127e-01
2.30819479e-01 2.37707406e-01 -6.16455853e-01 -1.36496410e-01
3.52376476e-02 -8.79240215e-01 3.63715470e-01 2.67252773e-01
1.79108322e-01 -4.22648966e-01 1.41697541e-01 3.31178457e-01
-4.87676889e-01 -9.58714545e-01 -6.11913145e-01 4.87874411e-02
-8.76092076e-01 -1.61070335e+00 -1.00377262e+00 -8.87849033e-01
9.79285955e-01 7.74048865e-01 1.10521328e+00 3.02450895e-01
-2.59179354e-01 5.30187666e-01 -5.04083037e-01 4.03983938e-03
-5.44618607e-01 -2.31447995e-01 9.48705152e-02 1.38535619e-01
3.82140130e-01 -6.51566565e-01 -4.12579805e-01 4.66817588e-01
-8.34891558e-01 -5.16843736e-01 2.15330318e-01 7.52151370e-01
3.38791996e-01 6.39389038e-01 4.39617127e-01 -5.85820615e-01
6.04257464e-01 -9.55201328e-01 -5.07349491e-01 1.85843065e-01
-4.48292375e-01 -8.63991678e-02 1.97357297e-01 -3.37912500e-01
-7.76832163e-01 -9.46439952e-02 5.32030106e-01 -4.21096116e-01
4.44362313e-02 5.22792459e-01 -1.60363719e-01 -4.11165915e-02
4.72786516e-01 8.67103040e-02 1.12476438e-01 -2.34582603e-01
3.04845750e-01 8.54552388e-01 4.36638832e-01 -5.41109204e-01
1.03726280e+00 8.75292957e-01 1.57205403e-01 -9.59793866e-01
-1.69993155e-02 -1.00995696e+00 -8.93334150e-01 -2.00731203e-01
4.87635821e-01 -9.70105410e-01 -5.22625923e-01 6.09020650e-01
-1.05560637e+00 -1.55633435e-01 1.70617774e-02 5.54794073e-01
-5.87020874e-01 8.58858407e-01 -9.49926496e-01 -1.34993064e+00
-9.84531865e-02 -8.13600123e-01 6.40941143e-01 -8.97330493e-02
7.65109733e-02 -1.23567605e+00 3.43845665e-01 -2.20378548e-01
2.40447134e-01 4.80417848e-01 5.94698668e-01 -6.16372883e-01
-3.57463777e-01 -6.55587971e-01 -5.34634948e-01 2.98666179e-01
5.10188080e-02 3.10504884e-01 -2.74575800e-01 -7.43613720e-01
-1.13149434e-01 3.20565939e-01 6.66261137e-01 7.54686892e-01
5.33563554e-01 -2.54159689e-01 -8.79900992e-01 1.48801833e-01
1.68799829e+00 -1.84876427e-01 3.93953025e-01 -7.70179033e-02
3.08785349e-01 7.22550929e-01 2.43828267e-01 8.17507505e-01
3.11075568e-01 2.29990333e-02 5.83034813e-01 3.56088310e-01
3.76982003e-01 2.99961939e-02 8.99129063e-02 9.86489058e-01
-9.71979797e-02 -5.89790642e-01 -8.84100020e-01 1.07722318e+00
-1.78253174e+00 -1.50707304e+00 -1.08895504e+00 2.49827027e+00
6.47394896e-01 -1.64542906e-03 3.90986860e-01 4.07004684e-01
1.85191488e+00 -2.45379373e-01 -1.44996017e-01 -2.42176326e-03
-4.52843994e-01 -2.88334507e-02 8.45147908e-01 7.10722625e-01
-8.41622531e-01 8.09177384e-02 7.49637508e+00 7.76721299e-01
-1.06956542e-01 2.41045251e-01 1.46760315e-01 4.14940357e-01
-3.25960785e-01 3.30151141e-01 -4.17054474e-01 3.74562860e-01
7.06011534e-01 -7.97408223e-01 2.53989130e-01 5.97640634e-01
2.78182954e-01 -1.59732282e-01 -6.85707271e-01 7.66760707e-01
-1.31148368e-01 -6.48345768e-01 -5.29847980e-01 7.92508960e-01
1.11728740e+00 -1.77654728e-01 -3.52330208e-01 -1.81549191e-01
1.07105863e+00 -7.47418165e-01 5.07360876e-01 5.35408139e-01
8.86146069e-01 -8.00832331e-01 8.52607489e-01 4.86057758e-01
-1.46759462e+00 -7.95548558e-02 -5.25236964e-01 -1.88457966e-01
4.41027045e-01 1.05855429e+00 -7.13259637e-01 6.78655267e-01
4.62330490e-01 8.00677061e-01 -7.94498473e-02 1.60103106e+00
-5.03124744e-02 8.70521247e-01 -5.23927867e-01 6.81605637e-02
-2.31309652e-01 -6.14322364e-01 8.91424537e-01 1.28264391e+00
6.02197945e-01 7.35455826e-02 2.35651329e-01 6.82520211e-01
2.97535181e-01 -6.45599002e-03 -7.32583284e-01 -2.88991965e-02
7.18801856e-01 1.24916124e+00 -1.20524335e+00 -2.88000822e-01
-4.10522133e-01 7.47170448e-01 1.48119405e-01 4.50732321e-01
-5.22477448e-01 -6.04921222e-01 3.33667040e-01 4.89657372e-01
7.22122550e-01 -3.28479588e-01 -6.82138801e-02 -1.10231590e+00
-2.56563991e-01 -3.75941753e-01 4.71727550e-01 -3.03695589e-01
-1.71795833e+00 4.83568072e-01 3.38886470e-01 -1.10746276e+00
-1.69452012e-01 -3.00672263e-01 -8.51882219e-01 8.70103359e-01
-9.24138665e-01 -8.17843139e-01 -2.31502026e-01 7.85780966e-01
-5.08843839e-01 -4.68468219e-02 5.15849292e-01 2.97167212e-01
-5.57833970e-01 1.31767303e-01 6.74176574e-01 2.28884488e-01
4.46747303e-01 -1.53213716e+00 8.19886103e-02 1.02227581e+00
-4.93113041e-01 5.34909964e-01 1.23456585e+00 -8.71529341e-01
-9.01083589e-01 -9.16047096e-01 8.64078760e-01 5.95017187e-02
9.97418463e-01 -3.50249648e-01 -8.66075099e-01 5.41867673e-01
1.70116082e-01 1.69873208e-01 6.49504483e-01 -5.60730919e-02
-3.11068892e-01 4.29169595e-01 -1.44067049e+00 2.33288094e-01
1.11705089e+00 -2.52456188e-01 -3.73174530e-03 3.71576250e-01
1.59557775e-01 4.61800635e-01 -7.82610297e-01 2.58944362e-01
3.24778974e-01 -8.72648656e-01 8.67451608e-01 -4.35904056e-01
-1.11766189e-01 -5.73239803e-01 -3.54326218e-01 -1.12391448e+00
-6.76101565e-01 -6.53486967e-01 2.64674164e-02 1.24314654e+00
-4.10000905e-02 -6.00876212e-01 6.60752535e-01 -1.87097177e-01
5.61203122e-01 -2.94231344e-02 -1.13994777e+00 -9.92129087e-01
1.62825674e-01 -1.19355105e-01 3.35524589e-01 7.16478705e-01
2.11992756e-01 8.51920843e-02 -4.05684441e-01 2.41483480e-01
1.46251583e+00 -1.43998995e-01 4.83362317e-01 -1.66771555e+00
-3.01477373e-01 -4.82547790e-01 -5.55013120e-01 -5.95555007e-01
2.62728930e-01 -7.51328886e-01 3.08518142e-01 -1.67576241e+00
9.88728464e-01 -6.49345100e-01 5.09796850e-02 -2.72034168e-01
-4.80605662e-02 1.11034542e-01 -5.78823164e-02 3.38917822e-01
-5.64496279e-01 3.73655260e-01 8.68259251e-01 -1.19416945e-01
1.66160747e-01 6.12625480e-01 -5.32866836e-01 5.07913530e-01
6.34610713e-01 -7.09136605e-01 -4.13190156e-01 1.00173213e-01
3.89809042e-01 5.44835269e-01 3.21226090e-01 -6.68483138e-01
2.71252215e-01 -4.77424823e-02 -2.80921549e-01 -8.99049580e-01
-9.24283192e-02 -8.03384364e-01 6.46711767e-01 6.89681649e-01
-6.63763136e-02 7.67776519e-02 -4.08846408e-01 1.45777714e+00
-5.05409092e-02 -7.13181615e-01 7.24915683e-01 -3.22885923e-02
1.58583850e-01 3.21042836e-01 -6.49064600e-01 1.71718791e-01
1.30763340e+00 -3.30478072e-01 -9.88956392e-02 -9.55030143e-01
-1.18218184e+00 3.58149707e-01 6.35264993e-01 -3.03490460e-01
4.84230548e-01 -1.40033162e+00 -1.31760705e+00 -3.23909223e-01
8.66893083e-02 -2.00402647e-01 2.46454254e-01 1.08386004e+00
-5.30495822e-01 -7.36030862e-02 1.28844365e-01 -8.32844317e-01
-1.32506168e+00 9.60469186e-01 2.42189169e-01 -2.97747314e-01
-2.05588311e-01 4.24298197e-01 7.47075006e-02 -1.79127604e-01
3.76428515e-02 2.26344392e-01 -6.49827048e-02 2.24739537e-02
5.57484925e-01 7.60436296e-01 -2.91612357e-01 -8.95611107e-01
-3.62640440e-01 2.94228911e-01 1.81722209e-01 -2.80203134e-01
1.14274597e+00 -5.84977984e-01 -5.86111605e-01 2.38526940e-01
8.88483465e-01 1.77751690e-01 -1.04139876e+00 -3.54850262e-01
-1.00216888e-01 -5.73174119e-01 -2.81394511e-01 1.44975990e-01
-1.04483867e+00 7.16468632e-01 1.08023889e-01 7.87998080e-01
8.19157779e-01 2.89783359e-01 1.20682856e-02 -4.35497135e-01
8.28110099e-01 -5.90506852e-01 -2.23267540e-01 2.66697824e-01
5.85088968e-01 -8.30741465e-01 -3.11049935e-03 -7.85308599e-01
-2.14829966e-01 1.00793481e+00 -1.13960905e-02 -5.14001131e-01
1.25910521e+00 6.93264782e-01 -7.07493544e-01 1.72432866e-02
-4.86965925e-01 -4.86717701e-01 -1.96563631e-01 1.37202835e+00
3.25622261e-01 4.01954591e-01 -6.06323183e-01 6.57464921e-01
1.01398878e-01 -1.78016156e-01 1.19120586e+00 7.51282871e-01
-6.73220158e-01 -8.83939028e-01 -7.90150464e-01 4.19438004e-01
-2.59302050e-01 8.78700521e-03 -1.00390740e-01 6.45630002e-01
-3.48642617e-01 1.51247299e+00 5.02672009e-02 -1.55153260e-01
7.97035843e-02 -4.93667603e-01 5.20891964e-01 -7.08249748e-01
4.93543558e-02 3.07283133e-01 -1.06470987e-01 1.03432901e-01
-4.83317256e-01 -1.07157338e+00 -8.22724879e-01 -1.04081106e+00
-9.68374431e-01 4.45195556e-01 1.92370668e-01 6.65810049e-01
-8.99558738e-02 9.84179080e-02 8.07835221e-01 -5.54221630e-01
-8.94841194e-01 -1.17212749e+00 -1.38502145e+00 3.80201489e-02
2.78192580e-01 -5.31087518e-01 -9.60451245e-01 -3.21835913e-02]
|
[6.94626522064209, 5.168193340301514]
|
62f9964c-82a6-4d29-a0fe-15d970b4c906
|
mix-and-reason-reasoning-over-semantic
|
2210.07571
| null |
https://arxiv.org/abs/2210.07571v1
|
https://arxiv.org/pdf/2210.07571v1.pdf
|
Mix and Reason: Reasoning over Semantic Topology with Data Mixing for Domain Generalization
|
Domain generalization (DG) enables generalizing a learning machine from multiple seen source domains to an unseen target one. The general objective of DG methods is to learn semantic representations that are independent of domain labels, which is theoretically sound but empirically challenged due to the complex mixture of common and domain-specific factors. Although disentangling the representations into two disjoint parts has been gaining momentum in DG, the strong presumption over the data limits its efficacy in many real-world scenarios. In this paper, we propose Mix and Reason (\mire), a new DG framework that learns semantic representations via enforcing the structural invariance of semantic topology. \mire\ consists of two key components, namely, Category-aware Data Mixing (CDM) and Adaptive Semantic Topology Refinement (ASTR). CDM mixes two images from different domains in virtue of activation maps generated by two complementary classification losses, making the classifier focus on the representations of semantic objects. ASTR introduces relation graphs to represent semantic topology, which is progressively refined via the interactions between local feature aggregation and global cross-domain relational reasoning. Experiments on multiple DG benchmarks validate the effectiveness and robustness of the proposed \mire.
|
['Yizhou Yu', 'Yue Huang', 'Gangming Zhao', 'Feng Liu', 'Luyao Tang', 'Chaoqi Chen']
|
2022-10-14
| null | null | null | null |
['relational-reasoning']
|
['natural-language-processing']
|
[ 3.88710856e-01 4.87602621e-01 -9.94192958e-02 -5.46793163e-01
-1.88726217e-01 -7.41303802e-01 1.02224123e+00 1.58691034e-01
6.97429404e-02 4.71117914e-01 1.20044261e-01 1.03074469e-01
-5.80619812e-01 -9.42868710e-01 -7.66463339e-01 -8.15075815e-01
1.00185156e-01 6.37807906e-01 4.21966195e-01 -3.47008228e-01
5.01826704e-02 4.52572316e-01 -1.64180338e+00 4.87126201e-01
1.05899096e+00 1.12339437e+00 1.12354793e-01 -6.18378446e-02
-4.61381376e-01 6.65316224e-01 -4.32132334e-01 -3.80743355e-01
3.89009804e-01 -4.57169890e-01 -9.84525561e-01 1.46163493e-01
4.49300140e-01 1.79928273e-01 -7.76377320e-02 1.30560434e+00
6.22296371e-02 2.47648358e-01 8.35455656e-01 -1.64647532e+00
-9.82752442e-01 7.41453528e-01 -4.37546313e-01 8.87897089e-02
1.62006930e-01 -1.55904815e-01 1.11619663e+00 -7.56565392e-01
7.71999180e-01 1.64317167e+00 5.53097129e-01 5.84108353e-01
-1.71236205e+00 -6.25181794e-01 5.48181474e-01 2.21529856e-01
-1.38827336e+00 -1.16833284e-01 1.18827140e+00 -6.00420058e-01
5.65874815e-01 -1.26073584e-01 2.10250244e-01 1.29589128e+00
-5.95158413e-02 6.01992667e-01 1.42452013e+00 -2.73702353e-01
5.04539371e-01 4.03157651e-01 1.33048594e-01 3.38695556e-01
4.55216557e-01 1.38211548e-01 -5.44418573e-01 -3.66564244e-02
7.87481666e-01 -6.73049316e-02 -2.99549192e-01 -1.32873154e+00
-1.02274477e+00 9.51607466e-01 8.27642739e-01 2.93911308e-01
-2.46722341e-01 -2.27592930e-01 4.28025961e-01 5.82430542e-01
3.94025028e-01 5.08066237e-01 -3.71351302e-01 4.84718919e-01
-5.16098261e-01 2.37992972e-01 4.65398908e-01 1.02631354e+00
7.60611415e-01 -7.87143707e-02 2.42350087e-01 9.96179938e-01
2.40539536e-01 2.71110058e-01 5.53330243e-01 -7.32520759e-01
4.31670606e-01 1.16571426e+00 -3.09825480e-01 -1.14010859e+00
-4.12881970e-01 -5.09256184e-01 -8.80870879e-01 4.07069564e-01
5.12758911e-01 2.65135556e-01 -7.50087798e-01 2.32789755e+00
3.35800499e-01 1.84156131e-02 1.54728517e-01 9.99146402e-01
8.05162489e-01 1.60521314e-01 3.04534912e-01 2.80957848e-01
1.13105261e+00 -5.70686638e-01 -3.03112835e-01 -2.77263254e-01
4.32049960e-01 -8.25743452e-02 8.91512036e-01 2.44068265e-01
-8.73024166e-01 -7.64116466e-01 -1.23687696e+00 -1.17800623e-01
-8.77021194e-01 -3.20577264e-01 5.51659346e-01 4.29532290e-01
-9.37286496e-01 4.63811010e-01 -5.41327834e-01 -5.73504925e-01
8.46533656e-01 3.54310542e-01 -6.25413716e-01 -2.17710510e-01
-1.24125791e+00 8.42571616e-01 7.65706599e-01 -1.88977748e-01
-6.94006205e-01 -7.79238760e-01 -8.65425169e-01 1.19571257e-02
5.32899737e-01 -8.44183385e-01 6.51925743e-01 -1.24091363e+00
-1.28294539e+00 1.17028809e+00 3.71417850e-01 -5.18692374e-01
6.73902869e-01 -4.09707101e-03 -5.92548549e-01 2.85995871e-01
2.36104742e-01 8.36070240e-01 1.04603887e+00 -1.63103294e+00
-5.85335612e-01 -7.03858078e-01 1.78137377e-01 3.52766007e-01
-1.18039556e-01 -6.09366000e-01 2.96468325e-02 -8.12461555e-01
4.63905603e-01 -7.39861846e-01 1.48932606e-01 8.53332058e-02
-1.66246206e-01 -2.98219830e-01 8.71362448e-01 -3.80036920e-01
7.20678568e-01 -2.35577488e+00 5.70677638e-01 3.27624142e-01
4.06102985e-01 8.89704749e-02 -2.10557655e-01 1.66472167e-01
-4.55100894e-01 -1.36237964e-01 -4.40919518e-01 -1.00068212e-03
9.47962850e-02 4.22138721e-01 -5.70004702e-01 4.49239016e-01
3.74281049e-01 8.65541577e-01 -1.06443930e+00 -2.55212128e-01
2.60057241e-01 3.66941988e-01 -4.03231412e-01 4.94269701e-03
-3.39042485e-01 6.02891088e-01 -5.32129228e-01 4.50133264e-01
8.74765038e-01 -3.74482751e-01 4.26330447e-01 -3.33244264e-01
4.76607084e-01 2.48386219e-01 -1.44438434e+00 1.97560453e+00
-1.51360288e-01 2.21326277e-01 8.72037411e-02 -1.58739388e+00
1.20179892e+00 -1.58326346e-02 2.80174136e-01 -8.50349724e-01
4.92021702e-02 1.85849711e-01 -1.27457723e-01 -1.11024357e-01
2.26094425e-02 -3.47788721e-01 -1.35339856e-01 3.10024798e-01
4.40799534e-01 -1.18199646e-01 -5.81394397e-02 3.24058622e-01
7.98382998e-01 3.69839191e-01 3.62280875e-01 -7.04365969e-01
7.28829503e-01 -1.13806343e-02 6.42407656e-01 6.15100563e-01
-2.88504839e-01 3.80725831e-01 5.28162181e-01 -2.69083798e-01
-7.14547276e-01 -1.75519574e+00 -3.97681504e-01 1.10466790e+00
8.13772678e-01 9.00347531e-03 -6.53438628e-01 -1.18449974e+00
3.15402985e-01 8.22120667e-01 -8.56161475e-01 -6.21700525e-01
-4.95720744e-01 -5.85908353e-01 4.28742468e-01 6.21709943e-01
6.25264287e-01 -9.93380785e-01 -6.57312274e-01 1.00063346e-01
-7.91229829e-02 -1.19579184e+00 5.95589727e-02 3.03080320e-01
-9.57051933e-01 -1.38784909e+00 -3.87456805e-01 -6.77947760e-01
7.32244134e-01 3.20577860e-01 1.18343759e+00 -1.49284199e-01
-1.79933727e-01 6.47754431e-01 -3.97074223e-01 -1.05304189e-01
-5.05857587e-01 -5.55258393e-02 5.70887141e-02 3.62305194e-01
5.29544353e-01 -8.58745039e-01 -4.25966203e-01 4.51051652e-01
-1.13651192e+00 2.14777179e-02 5.71348369e-01 9.12256837e-01
4.99761373e-01 3.75807613e-01 7.54392207e-01 -1.07330668e+00
4.59204227e-01 -6.94592953e-01 -3.64970565e-01 3.86936456e-01
-8.07460725e-01 3.83998156e-01 6.01383269e-01 -4.37882483e-01
-1.38135684e+00 -1.52610451e-01 4.97209489e-01 -5.02043307e-01
-4.22401696e-01 1.92117482e-01 -6.45535290e-01 4.38165031e-02
7.79259741e-01 1.11803308e-01 2.19064116e-01 -5.06198168e-01
7.93816447e-01 9.14648995e-02 7.03655183e-01 -8.98315191e-01
9.92710292e-01 6.65375531e-01 9.72655937e-02 -6.61417007e-01
-7.40891933e-01 -2.59199053e-01 -9.16051745e-01 7.43859485e-02
8.12067449e-01 -7.71721125e-01 -1.71702608e-01 3.70194614e-01
-6.90432489e-01 -1.30290389e-01 -7.31923163e-01 9.94393155e-02
-6.42209291e-01 2.90911049e-01 -4.71214056e-02 -1.74143329e-01
2.78300732e-01 -9.44307745e-01 8.38722885e-01 5.23234382e-02
-2.84199893e-01 -1.36425650e+00 -1.61886066e-01 3.35880578e-01
1.96689084e-01 5.11142075e-01 1.39011252e+00 -1.06367338e+00
-4.48861897e-01 1.65094733e-01 -5.05202115e-01 4.32504535e-01
3.54376107e-01 -4.74107444e-01 -1.11430359e+00 -2.92757988e-01
-4.42518331e-02 -3.88927221e-01 9.83412266e-01 1.55538052e-01
1.02648747e+00 -2.08263233e-01 -4.21599686e-01 4.65905696e-01
1.61642265e+00 1.13104634e-01 3.41987997e-01 3.24869305e-01
7.06450522e-01 9.47473049e-01 3.73457193e-01 1.19118407e-01
2.70926774e-01 6.64619923e-01 6.10510588e-01 -5.01931347e-02
-4.64199752e-01 -3.45226377e-01 1.19958654e-01 2.42555246e-01
1.04327917e-01 5.00244573e-02 -8.13008785e-01 4.43674743e-01
-1.79338467e+00 -8.79969418e-01 7.36548156e-02 2.10925484e+00
6.66215658e-01 2.41111919e-01 9.25976112e-02 2.27336243e-01
8.26598346e-01 4.00879011e-02 -8.38987529e-01 -1.57500893e-01
-4.66358364e-01 2.70217806e-01 2.06579566e-01 3.10704470e-01
-1.05801380e+00 8.64432693e-01 4.93070507e+00 6.92124724e-01
-9.65216756e-01 7.30238259e-02 2.88440168e-01 2.33434334e-01
-4.51364815e-01 -3.26156169e-02 -5.00568092e-01 2.59356469e-01
2.70074069e-01 -1.40546575e-01 4.25814211e-01 9.99285936e-01
-4.24190044e-01 1.28187507e-01 -1.45607936e+00 8.37494254e-01
7.98985288e-02 -1.25317883e+00 5.38469732e-01 8.13792422e-02
7.42228687e-01 -1.55830413e-01 2.08716780e-01 3.43246073e-01
6.13289177e-01 -8.72417867e-01 9.97664392e-01 4.05650645e-01
6.37426794e-01 -4.48066831e-01 4.53241736e-01 1.96791530e-01
-1.20662653e+00 -3.88912052e-01 -2.17198119e-01 2.58273989e-01
-2.83469498e-01 4.61240172e-01 -6.66707933e-01 9.78487313e-01
7.31438696e-01 9.18783367e-01 -7.40887463e-01 5.29608250e-01
-2.57965475e-01 1.11141853e-01 -1.58174500e-01 6.35070980e-01
1.14411786e-01 -2.48649761e-01 6.25235140e-01 9.51981723e-01
-9.50919241e-02 1.68017913e-02 1.61737695e-01 1.40464759e+00
-5.10865338e-02 -1.73983946e-01 -6.89831138e-01 2.02261209e-01
6.16451919e-01 9.26604569e-01 -9.29720044e-01 -2.10614920e-01
-3.51626903e-01 9.30559039e-01 4.79822546e-01 4.97843832e-01
-6.65735126e-01 -7.00343475e-02 8.66635859e-01 2.43188560e-01
3.23479533e-01 3.43718566e-02 -5.83800554e-01 -1.09344327e+00
1.16976254e-01 -7.79691517e-01 8.03694248e-01 -5.14199793e-01
-1.74463558e+00 4.41863120e-01 3.11233222e-01 -1.21097171e+00
8.65741894e-02 -7.02735305e-01 -3.26659143e-01 8.33186328e-01
-1.70998478e+00 -1.37081206e+00 -4.50249821e-01 8.60700250e-01
5.38583517e-01 -1.70362711e-01 6.94465578e-01 4.03566435e-02
-3.31944793e-01 5.04024446e-01 -1.13841064e-01 -5.93045354e-02
6.60636961e-01 -1.38887703e+00 8.20389241e-02 6.93567634e-01
8.67469087e-02 7.15141475e-01 4.99291003e-01 -5.71759582e-01
-1.00524783e+00 -1.17022121e+00 5.26576877e-01 -5.92214406e-01
6.09863639e-01 -5.48622310e-01 -1.31799698e+00 6.14836216e-01
-2.57154763e-01 1.86877832e-01 3.65416646e-01 3.59230638e-02
-1.11045814e+00 -2.93949872e-01 -1.35246015e+00 3.86743933e-01
1.38253951e+00 -6.09904289e-01 -1.09526300e+00 -1.05760060e-02
6.22673750e-01 8.86509791e-02 -9.48398471e-01 5.20632982e-01
1.91254050e-01 -1.22412646e+00 1.17808342e+00 -6.23047411e-01
1.78064138e-01 -4.89567041e-01 -4.48284984e-01 -1.39374936e+00
-5.10799229e-01 -3.06652170e-02 7.44018555e-02 1.40904117e+00
1.02629952e-01 -8.85528803e-01 5.44346631e-01 4.04715210e-01
-7.55427480e-02 -2.15337828e-01 -1.00512946e+00 -1.14841759e+00
4.57752079e-01 -2.71476567e-01 8.85783374e-01 1.42369497e+00
-4.34536859e-03 5.30922532e-01 2.67664194e-01 2.96223193e-01
7.81386971e-01 3.20003778e-01 4.66341257e-01 -1.90512347e+00
-6.10858761e-02 -9.09548700e-01 -6.20956779e-01 -7.95189083e-01
3.62371176e-01 -1.38316441e+00 -2.21788168e-01 -1.45712149e+00
2.15403456e-02 -6.41577780e-01 -4.53243017e-01 4.05660808e-01
2.67427992e-02 -1.70503154e-01 1.57467853e-02 3.09652388e-01
-4.95251387e-01 7.89532065e-01 1.17305493e+00 -3.29884499e-01
-1.17580384e-01 -2.55880952e-01 -8.84817183e-01 7.62154579e-01
6.57787800e-01 -2.84738243e-01 -9.82446551e-01 -2.44732991e-01
3.20236571e-02 -4.91894752e-01 7.61268139e-01 -9.31736410e-01
-2.54415851e-02 -5.86835854e-02 3.75062078e-01 -5.96855581e-02
1.42676309e-01 -1.10287738e+00 1.00030340e-01 2.34028056e-01
-4.60719317e-01 -3.79731029e-01 8.55686441e-02 8.97566795e-01
-2.62194902e-01 1.87082559e-01 1.20308447e+00 -1.11326978e-01
-1.17522931e+00 4.60432284e-02 3.26062232e-01 4.81967837e-01
1.12113619e+00 -6.04741871e-01 -4.38028485e-01 1.35013670e-01
-9.24178898e-01 2.40397938e-02 4.96940792e-01 7.89919019e-01
5.28453112e-01 -1.35611069e+00 -4.88453716e-01 5.29161513e-01
5.16267836e-01 1.01251282e-01 3.19528073e-01 5.77545404e-01
3.22230123e-02 2.55459756e-01 -4.74578500e-01 -8.25010777e-01
-7.66172111e-01 9.45340991e-01 5.04171431e-01 -1.29807934e-01
-9.81390417e-01 9.48257446e-01 9.42536533e-01 -6.21963799e-01
2.64077634e-01 -3.25136006e-01 -2.64190853e-01 7.70367831e-02
1.31430745e-01 3.99494767e-01 1.59503937e-01 -7.41223514e-01
-3.89499962e-01 5.74871719e-01 -9.36386213e-02 1.59524307e-01
1.17847300e+00 -3.50644737e-01 2.68148780e-02 3.98424804e-01
9.65106428e-01 -4.69536781e-01 -1.50564814e+00 -6.15095198e-01
3.63345534e-01 -3.32403541e-01 -1.26917243e-01 -1.02196372e+00
-9.77728307e-01 8.04038346e-01 6.04749143e-01 2.43843406e-01
1.25868499e+00 3.29901010e-01 3.17456961e-01 1.52461110e-02
4.95184839e-01 -1.12914932e+00 3.04051131e-01 1.86848342e-01
1.06438923e+00 -1.15382600e+00 -2.46318206e-01 -6.08167112e-01
-6.88724458e-01 8.68237555e-01 7.48413444e-01 -2.62278676e-01
6.86031878e-01 -2.19890311e-01 -1.63363904e-01 -3.92308295e-01
-4.08722639e-01 -8.96338969e-02 4.72950846e-01 1.06345129e+00
-1.10323615e-01 5.44591323e-02 1.16471492e-01 8.20379436e-01
-8.10626224e-02 -3.30197096e-01 1.07555203e-01 6.86940849e-01
-2.56309807e-01 -1.03076243e+00 -3.42369348e-01 1.27113938e-01
7.32562020e-02 2.40991607e-01 -6.66582584e-01 1.14176250e+00
5.12604535e-01 8.16935480e-01 9.00992155e-02 -2.52405703e-01
4.99716610e-01 2.83790141e-01 5.04277170e-01 -5.48636913e-01
-2.84698039e-01 -2.11439207e-01 -2.24536970e-01 -6.76903963e-01
-4.54124421e-01 -7.40048468e-01 -1.35005939e+00 1.45130875e-02
1.00028962e-01 -5.82546890e-02 4.14406210e-01 1.00278974e+00
5.08597732e-01 6.44853592e-01 5.15868068e-01 -4.08028305e-01
-7.40034163e-01 -5.42601645e-01 -8.43434870e-01 1.08583128e+00
4.04403299e-01 -1.41379368e+00 -5.11141360e-01 2.03422546e-01]
|
[10.080403327941895, 2.687117099761963]
|
513d69ff-5fd6-4b7d-8320-e020d5cb7abf
|
relay-hindsight-experience-replay-continual
|
2208.00843
| null |
https://arxiv.org/abs/2208.00843v2
|
https://arxiv.org/pdf/2208.00843v2.pdf
|
Relay Hindsight Experience Replay: Self-Guided Continual Reinforcement Learning for Sequential Object Manipulation Tasks with Sparse Rewards
|
Exploration with sparse rewards remains a challenging research problem in reinforcement learning (RL). Especially for sequential object manipulation tasks, the RL agent always receives negative rewards until completing all sub-tasks, which results in low exploration efficiency. To solve these tasks efficiently, we propose a novel self-guided continual RL framework, RelayHER (RHER). RHER first decomposes a sequential task into new sub-tasks with increasing complexity and ensures that the simplest sub-task can be learned quickly by utilizing Hindsight Experience Replay (HER). Secondly, we design a multi-goal & multi-task network to learn these sub-tasks simultaneously. Finally, we propose a Self-Guided Exploration Strategy (SGES). With SGES, the learned sub-task policy will guide the agent to the states that are helpful to learn more complex sub-task with HER. By this self-guided exploration and relay policy learning, RHER can solve these sequential tasks efficiently stage by stage. The experimental results show that RHER significantly outperforms vanilla-HER in sample-efficiency on five singleobject and five complex multi-object manipulation tasks (e.g., Push, Insert, ObstaclePush, Stack, TStack, etc.). The proposed RHER has also been applied to learn a contact-rich push task on a physical robot from scratch, and the success rate reached 10/10 with only 250 episodes.
|
['Bo Song', 'Zhiyong Sun', 'Erkang Cheng', 'Qiang Zhang', 'Kun Dong', 'Yuxin Wang', 'Yongle Luo']
|
2022-08-01
| null | null | null | null |
['robot-manipulation']
|
['robots']
|
[-1.61723092e-01 8.02158266e-02 -2.99461842e-01 1.10276394e-01
-5.91404378e-01 -2.33762130e-01 9.73217636e-02 -3.09169620e-01
-7.63450980e-01 1.22948980e+00 -5.82821369e-02 -1.87354565e-01
-4.55449522e-01 -5.49591482e-01 -7.47839749e-01 -8.10620606e-01
-4.82278764e-01 5.81433058e-01 3.98593694e-01 -6.64638460e-01
4.45862442e-01 2.87098587e-01 -1.48557067e+00 -1.40571907e-01
9.57257152e-01 7.50941038e-01 1.24835706e+00 4.99126971e-01
-9.58261788e-02 9.80804563e-01 -4.84144747e-01 5.49625993e-01
4.73570198e-01 -2.47310326e-01 -9.28305626e-01 2.40799878e-02
-6.38007581e-01 -7.25874484e-01 -3.33535135e-01 8.74551713e-01
5.52125275e-01 6.76358581e-01 2.87987351e-01 -1.36438167e+00
-3.73417705e-01 8.82780790e-01 -8.26788902e-01 4.23384123e-02
3.76122743e-01 4.80228573e-01 6.96854353e-01 -6.37091458e-01
5.37886918e-01 1.59042466e+00 -1.37490675e-01 7.68576384e-01
-6.85284913e-01 -8.51745367e-01 6.19807303e-01 2.00215265e-01
-7.44917989e-01 4.29078490e-02 6.79681242e-01 -3.40512544e-02
1.02827060e+00 -1.41138375e-01 6.58831716e-01 1.01335764e+00
4.39813614e-01 1.28760278e+00 1.46984982e+00 -3.18415016e-02
5.31927824e-01 -1.53825700e-01 -1.58115461e-01 7.84300387e-01
1.09885074e-01 4.97319847e-01 -5.14606237e-01 1.67476386e-01
1.15433073e+00 2.31224567e-01 -5.58622321e-03 -4.83773112e-01
-1.51731277e+00 7.19932139e-01 5.91530144e-01 1.07928529e-01
-6.22264326e-01 4.73530978e-01 4.83625412e-01 8.62347901e-01
-2.63700664e-01 6.05093718e-01 -6.20937049e-01 -2.59821177e-01
-1.00730509e-01 5.69464684e-01 6.94903672e-01 1.27007723e+00
1.00339162e+00 3.46293449e-01 -1.11717194e-01 7.87246704e-01
2.90452451e-01 6.02337718e-01 6.04804754e-01 -1.11848068e+00
8.81450355e-01 3.41513038e-01 6.15629137e-01 -5.28839529e-01
-7.28090942e-01 -3.72592568e-01 -7.71705747e-01 7.40932882e-01
1.00201787e-02 -5.25847852e-01 -8.00414383e-01 1.55066347e+00
6.48600817e-01 -3.10172707e-01 4.44400102e-01 1.03262043e+00
5.41789055e-01 7.32793689e-01 -7.16013685e-02 -5.21200597e-01
1.06219220e+00 -1.55842364e+00 -8.34228516e-01 -5.61383784e-01
4.56303388e-01 -2.43928701e-01 1.22047246e+00 8.04436445e-01
-1.09716940e+00 -6.41007006e-01 -1.00752127e+00 2.71859288e-01
-7.98521042e-02 1.38468832e-01 8.54894876e-01 -5.86019345e-02
-7.74825931e-01 7.28674173e-01 -7.96637535e-01 8.19143876e-02
3.87433559e-01 5.00933230e-01 -2.30951250e-01 -2.05289915e-01
-9.97940481e-01 1.01907015e+00 8.40409935e-01 2.65505195e-01
-1.76735580e+00 -8.92205611e-02 -6.79094493e-01 -1.29470035e-01
1.22583365e+00 -2.65067041e-01 1.52967024e+00 -4.36548710e-01
-2.11141300e+00 1.10204540e-01 3.99337001e-02 -1.19051903e-01
4.91952896e-01 -5.38222551e-01 2.17966679e-02 -7.98134655e-02
3.97873878e-01 8.14910889e-01 9.46485400e-01 -1.49426353e+00
-9.25139546e-01 -1.25106514e-01 3.16081166e-01 7.77328372e-01
1.97130218e-02 -2.30216295e-01 -7.53352493e-02 -3.56271774e-01
1.11758284e-01 -9.70968962e-01 -6.57948494e-01 -3.10972750e-01
-2.52847999e-01 -7.10025430e-01 6.85685515e-01 -2.63181686e-01
8.39428782e-01 -1.97502136e+00 7.93625116e-01 -1.39081508e-01
1.41737506e-01 -4.58294563e-02 -4.87110019e-01 5.21652460e-01
2.43489236e-01 -3.41280162e-01 -5.86401150e-02 -2.29363620e-01
-1.08572647e-01 6.42012775e-01 -3.26592565e-01 9.95887369e-02
-1.59535676e-01 1.13306177e+00 -1.44591355e+00 -3.39952379e-01
5.42277172e-02 -3.98517013e-01 -5.67591727e-01 5.11841238e-01
-5.65192461e-01 6.67845190e-01 -9.62875247e-01 7.98379421e-01
4.68980670e-01 -1.32275656e-01 -1.05526477e-01 5.11858702e-01
-3.34013164e-01 -1.78148076e-02 -1.40158868e+00 2.03204894e+00
-6.46626115e-01 -2.99835876e-02 3.25137883e-01 -9.08137441e-01
1.12567401e+00 1.82682294e-02 4.36460763e-01 -9.65818107e-01
7.62578622e-02 3.46381038e-01 4.42977808e-02 -7.79974997e-01
6.05622113e-01 1.15490027e-01 -2.37087891e-01 6.65693521e-01
-6.07611611e-02 -3.33808362e-01 1.40605628e-01 -3.09802629e-02
1.18310356e+00 6.46746874e-01 3.27096581e-01 -1.28655374e-01
3.51017654e-01 2.30797231e-01 6.64413214e-01 1.10222411e+00
-3.02568227e-01 -8.16294625e-02 3.93643439e-01 -3.50202680e-01
-6.12380326e-01 -9.71466601e-01 6.31739616e-01 1.46849227e+00
6.98127329e-01 8.71564671e-02 -1.11802936e-01 -7.49113500e-01
8.48025978e-02 6.92278445e-01 -5.54293215e-01 -1.18152700e-01
-9.89991724e-01 -4.18215156e-01 -1.46085128e-01 3.24168414e-01
9.10771251e-01 -2.19734097e+00 -1.23549509e+00 5.58766723e-01
-3.20287142e-03 -6.63350761e-01 -3.34291130e-01 7.06773400e-01
-9.83574331e-01 -9.98065829e-01 -8.81244898e-01 -1.13855112e+00
6.41382337e-01 4.92282182e-01 5.03808498e-01 6.67245360e-03
-1.86590433e-01 2.72720605e-01 -6.18708849e-01 -2.96937138e-01
-2.43050661e-02 3.92876565e-02 2.61465818e-01 -6.46023035e-01
-1.69814542e-01 -5.56138098e-01 -5.47070503e-01 5.51235497e-01
-5.56884170e-01 1.89534262e-01 1.02523530e+00 9.84814882e-01
6.36566937e-01 1.73147082e-01 1.06700516e+00 -3.62332195e-01
1.08934605e+00 -6.46424711e-01 -7.05030680e-01 1.73075914e-01
-5.26743770e-01 2.38951951e-01 7.84260750e-01 -9.79995966e-01
-1.19753408e+00 -4.30102088e-02 1.74576461e-01 -4.04288411e-01
2.47830711e-02 4.47851062e-01 1.85312435e-01 -7.49856085e-02
4.80737269e-01 5.98667979e-01 7.55231455e-02 -4.27289367e-01
2.14900836e-01 5.19033968e-01 2.48034894e-01 -9.19475317e-01
6.89347506e-01 6.35198206e-02 6.23532608e-02 -3.58847737e-01
-6.77069068e-01 -3.15443575e-01 -1.42766848e-01 -3.18509758e-01
5.93792439e-01 -8.44187856e-01 -1.43973017e+00 6.65196180e-01
-9.20901179e-01 -1.10318542e+00 -3.39952141e-01 5.46421587e-01
-9.74051774e-01 2.30361402e-01 -5.51315963e-01 -1.05807555e+00
-2.95922250e-01 -1.42434144e+00 8.07707071e-01 5.69753766e-01
2.77558208e-01 -4.19732392e-01 -1.18005194e-01 1.15310706e-01
5.09566426e-01 8.56279954e-02 6.36337936e-01 -3.19775701e-01
-8.75305176e-01 4.74453092e-01 -2.73696985e-02 -1.26289055e-01
1.74523786e-01 -8.46463442e-01 -1.56847581e-01 -7.53841698e-01
9.21384245e-02 -1.27989793e+00 7.43487418e-01 2.16003060e-01
1.12844253e+00 -3.54091436e-01 -4.34018791e-01 3.44679564e-01
1.22439659e+00 7.32682109e-01 4.36432987e-01 7.29668498e-01
4.21288848e-01 5.71270406e-01 1.60421848e+00 6.95626616e-01
4.99022633e-01 3.82727921e-01 8.82870615e-01 3.12927783e-01
3.20896417e-01 -3.00724834e-01 6.56702101e-01 6.87407732e-01
-3.56585272e-02 -3.27334329e-02 -4.31747437e-01 5.21544933e-01
-2.22459173e+00 -6.64762199e-01 2.15619504e-01 1.82101190e+00
8.53937805e-01 3.43421191e-01 1.77633271e-01 -1.68463334e-01
3.38939220e-01 5.31241819e-02 -1.22137225e+00 -2.85891265e-01
2.92799950e-01 1.65909249e-02 3.74574840e-01 5.29637933e-01
-6.04550064e-01 1.39983940e+00 5.55787468e+00 9.38890994e-01
-8.51159811e-01 5.79958744e-02 7.00593293e-02 -2.16638565e-01
4.26271334e-02 -3.45068723e-02 -9.58354115e-01 3.48458081e-01
8.48400071e-02 -1.00513712e-01 1.15721977e+00 1.20641494e+00
1.02377526e-01 -8.34978402e-01 -1.00150919e+00 9.08505201e-01
-3.14263463e-01 -8.50822389e-01 -3.69362086e-01 -1.51548281e-01
6.75792456e-01 8.26583803e-02 2.38467753e-02 1.13333285e+00
9.78821039e-01 -9.42682683e-01 5.47225416e-01 2.36317217e-01
5.96349835e-01 -7.86124110e-01 4.29763675e-01 9.13390279e-01
-1.23429239e+00 -7.08950698e-01 -5.80153584e-01 -2.13491201e-01
2.88893700e-01 -5.76977804e-02 -6.36375129e-01 6.21155143e-01
7.90562332e-01 5.19420922e-01 1.38044387e-01 8.89414787e-01
-5.55291951e-01 -1.84805408e-01 -2.10270822e-01 -7.18298435e-01
7.05738783e-01 -2.81911731e-01 6.73896611e-01 4.65259969e-01
1.57344192e-01 3.70809227e-01 8.12437236e-01 7.62400389e-01
3.96978498e-01 -1.85164928e-01 -3.88933361e-01 6.27775267e-02
5.65481603e-01 1.27759743e+00 -6.34260058e-01 -1.27731755e-01
1.87563479e-01 9.52261388e-01 8.33718836e-01 4.47086215e-01
-6.58381760e-01 -6.69568002e-01 3.13181311e-01 -4.58969712e-01
3.70498538e-01 -4.67077136e-01 1.89357281e-01 -7.78091967e-01
1.50073186e-01 -1.02589500e+00 1.17254891e-01 -8.88759613e-01
-7.62515724e-01 4.71424371e-01 1.81576073e-01 -1.14776099e+00
-2.86286533e-01 -4.11378950e-01 -5.62597036e-01 6.97070301e-01
-1.98052764e+00 -6.55880988e-01 -3.63575339e-01 7.82950938e-01
1.03214312e+00 -5.28584719e-01 5.63013077e-01 -3.19523633e-01
-5.48972130e-01 2.51823485e-01 -6.75552562e-02 -2.88895726e-01
3.43495518e-01 -1.25845659e+00 7.23835528e-02 1.83579847e-01
-6.53044581e-01 4.41372067e-01 4.56613809e-01 -1.04622459e+00
-1.78376329e+00 -7.65329778e-01 -6.52405992e-02 2.77796000e-01
5.54699481e-01 -2.00475395e-01 -6.38977468e-01 6.39258087e-01
-4.01126817e-02 -2.87960302e-02 -1.60826981e-01 -8.60030856e-03
3.62555295e-01 6.48205280e-02 -1.10952234e+00 7.26084888e-01
1.33917129e+00 1.44982710e-01 -7.91058719e-01 4.07623172e-01
1.04267454e+00 -8.90228510e-01 -5.39582551e-01 4.05135810e-01
2.87251264e-01 -7.56094813e-01 7.84860492e-01 -5.68769991e-01
4.84184146e-01 -3.35704863e-01 1.65300444e-01 -1.69205952e+00
-4.48711753e-01 -9.98140454e-01 -3.95863175e-01 5.57542086e-01
3.63633968e-02 -9.03073311e-01 8.20785522e-01 -9.45353657e-02
-3.32709521e-01 -1.17986202e+00 -9.39973593e-01 -1.01274061e+00
-1.04200304e-01 5.51219545e-02 4.67599034e-01 4.77533966e-01
1.66263714e-01 1.25380769e-01 -7.96562016e-01 4.09881398e-02
6.51239932e-01 3.25908780e-01 1.00417686e+00 -8.44623804e-01
-5.18519700e-01 -9.86417830e-02 5.49284458e-01 -1.70893729e+00
1.30654439e-01 -7.23435223e-01 4.82099116e-01 -1.79612303e+00
2.26013102e-02 -1.03293848e+00 -2.39071369e-01 7.65249312e-01
-1.78769663e-01 -8.44587207e-01 3.19049984e-01 3.18748713e-01
-1.09688258e+00 1.02107477e+00 2.15754080e+00 6.90957084e-02
-7.84231842e-01 9.21670869e-02 -5.34394920e-01 4.72500682e-01
1.00652456e+00 -5.51588178e-01 -5.23592651e-01 -3.72421801e-01
1.92616850e-01 9.35828090e-01 2.09909435e-02 -9.08352137e-01
3.30651581e-01 -7.55752265e-01 1.55270711e-01 -7.96543360e-01
3.70907366e-01 -7.47127056e-01 -3.27400923e-01 1.13660681e+00
-4.13750023e-01 1.16573796e-01 1.23321801e-01 7.88441181e-01
1.68200389e-01 -4.88778263e-01 3.58326793e-01 -4.79790092e-01
-8.94894898e-01 3.54549944e-01 -4.11170125e-01 1.42749876e-01
1.44424939e+00 -6.32733256e-02 -2.55626738e-01 -1.65662214e-01
-8.84005964e-01 1.34165812e+00 -1.37078688e-01 6.71735525e-01
9.69852448e-01 -1.23358977e+00 -4.84904706e-01 -3.25593539e-02
-1.72240332e-01 6.37998998e-01 2.76067048e-01 5.84187746e-01
-1.47731647e-01 1.95068330e-01 -5.77603757e-01 -4.02635485e-01
-9.62303996e-01 6.83022022e-01 4.20077853e-02 -7.55550802e-01
-7.27400064e-01 8.38121295e-01 2.68348679e-02 -7.10392118e-01
5.85629761e-01 -3.71994197e-01 -2.86183149e-01 -2.23271430e-01
2.97156483e-01 6.82359636e-01 -6.60093606e-01 3.30319405e-01
-1.18946232e-01 3.90273929e-01 -3.77767235e-01 -1.93729579e-01
1.56694770e+00 -1.03268012e-01 1.06623955e-02 4.37849671e-01
6.76285744e-01 -5.09723663e-01 -1.75574088e+00 -3.24229628e-01
-1.92730129e-01 -4.33252871e-01 -1.68628678e-01 -9.50684667e-01
-7.95778573e-01 5.49529910e-01 2.97167242e-01 -1.83637649e-01
8.51375878e-01 -1.67635068e-01 7.67381191e-01 1.22362578e+00
1.09846592e+00 -1.48210418e+00 1.03024900e+00 8.60632479e-01
1.38538837e+00 -1.40058029e+00 9.95675325e-02 2.54739132e-02
-9.78363693e-01 9.82027709e-01 1.29625034e+00 -4.04215872e-01
2.05915183e-01 1.44339159e-01 -4.17897850e-01 -2.08851978e-01
-9.01098907e-01 -2.26877153e-01 -3.90212953e-01 5.56981981e-01
-4.50036675e-01 9.69486870e-03 -1.92402124e-01 4.08170611e-01
-1.57883480e-01 2.69507676e-01 4.98304486e-01 1.48590314e+00
-1.00314379e+00 -8.90192747e-01 -2.70266950e-01 4.47634161e-01
1.12005554e-01 3.69345784e-01 2.52878606e-01 8.76568437e-01
-1.50212705e-01 8.06545496e-01 -2.95468599e-01 -3.24495256e-01
2.20789790e-01 -5.18666148e-01 6.69455588e-01 -7.06060350e-01
-5.87278247e-01 -1.18415654e-02 -1.12394944e-01 -8.87829065e-01
-4.60985303e-02 -5.56028724e-01 -1.74803567e+00 1.15815811e-01
-2.83286422e-01 2.13843152e-01 5.39815545e-01 9.06138182e-01
8.60024914e-02 8.47547293e-01 9.19264674e-01 -1.07536268e+00
-1.26009512e+00 -8.75165582e-01 -4.90513265e-01 -1.68191567e-01
5.45868993e-01 -1.08250237e+00 -2.50721842e-01 -7.70034075e-01]
|
[4.219226837158203, 1.619389295578003]
|
7e692b4c-ba2d-4ef8-9532-72e5ef6e218c
|
evaluation-of-medium-large-language-models-at
|
2305.11991
| null |
https://arxiv.org/abs/2305.11991v2
|
https://arxiv.org/pdf/2305.11991v2.pdf
|
Evaluation of medium-large Language Models at zero-shot closed book generative question answering
|
Large language models (LLMs) have garnered significant attention, but the definition of "large" lacks clarity. This paper focuses on medium-sized language models (MLMs), defined as having at least six billion parameters but less than 100 billion. The study evaluates MLMs regarding zero-shot generative question answering, which requires models to provide elaborate answers without external document retrieval. The paper introduces an own test dataset and presents results from human evaluation. Results show that combining the best answers from different MLMs yielded an overall correct answer rate of 82.7% which is better than the 60.9% of ChatGPT. The best MLM achieved 71.8% and has 33B parameters, which highlights the importance of using appropriate training data for fine-tuning rather than solely relying on the number of parameters. More fine-grained feedback should be used to further improve the quality of answers. The open source community is quickly closing the gap to the best commercial models.
|
['Johannes Wirth', 'René Peinl']
|
2023-05-19
| null | null | null | null |
['generative-question-answering']
|
['natural-language-processing']
|
[-3.65872175e-01 3.59949350e-01 1.12546086e-01 -2.13233888e-01
-1.45085800e+00 -6.09505892e-01 6.82671428e-01 8.02108645e-02
-6.57112002e-01 7.02577472e-01 3.03281099e-01 -4.88322347e-01
1.61717981e-01 -6.54607236e-01 -4.28799629e-01 -3.04381698e-01
5.35957515e-01 9.60932612e-01 4.27456409e-01 -3.55060875e-01
5.35441995e-01 -1.34906381e-01 -1.20733309e+00 4.85757172e-01
9.52095866e-01 2.86893308e-01 4.29664224e-01 1.02769613e+00
-8.52939844e-01 9.15289700e-01 -1.32683265e+00 -7.74855733e-01
-3.24777722e-01 -5.69098473e-01 -9.30182040e-01 -3.30707133e-01
3.50572318e-01 -1.94703951e-01 8.54785293e-02 4.82229948e-01
8.21638703e-01 1.27675042e-01 4.68819171e-01 -9.23251033e-01
-9.65520144e-01 8.49867702e-01 -3.23811583e-02 5.22605062e-01
4.17554468e-01 3.25016886e-01 1.03525484e+00 -8.53962600e-01
5.50617516e-01 1.41505992e+00 4.74843204e-01 7.00093150e-01
-1.16988933e+00 -6.29327536e-01 -6.57746568e-02 -3.54046258e-03
-1.43448198e+00 -3.20016325e-01 9.27346200e-02 -4.64386672e-01
1.55294657e+00 3.48152697e-01 2.57442296e-01 9.13567781e-01
1.84840485e-01 4.71246541e-01 1.01627159e+00 -6.82904899e-01
2.00206131e-01 5.95837414e-01 1.19494513e-01 4.97898519e-01
2.13265717e-01 -4.73676056e-01 -3.80179018e-01 -5.13478279e-01
4.45477396e-01 -5.04214466e-01 -8.39788765e-02 2.00639516e-01
-1.03621089e+00 1.15502036e+00 -3.14275213e-02 6.19836867e-01
-2.63340175e-01 1.17786765e-01 1.07849061e-01 3.19492519e-01
4.13928688e-01 8.79544437e-01 -3.76514494e-01 -4.64340478e-01
-8.56360674e-01 3.98000360e-01 1.21512687e+00 7.02439487e-01
5.45439839e-01 5.15738763e-02 -3.17648828e-01 1.17358136e+00
2.24898696e-01 5.92906237e-01 7.23923743e-01 -1.28752434e+00
6.92777395e-01 7.20700800e-01 3.37050706e-01 -6.63113534e-01
-1.24689609e-01 -5.70603013e-01 -3.26242536e-01 -3.10483187e-01
4.79766399e-01 -2.08141848e-01 -7.05110312e-01 1.71743095e+00
-1.88987125e-02 -3.30691397e-01 1.23402968e-01 5.95746756e-01
1.03342748e+00 1.10450423e+00 4.47950959e-01 6.27879873e-02
1.07239270e+00 -9.25023496e-01 -6.77483022e-01 -3.57164741e-01
9.27368343e-01 -9.30391252e-01 1.59704471e+00 2.95862079e-01
-1.35063446e+00 -4.93849367e-01 -7.38379061e-01 -1.53497448e-02
-3.36363941e-01 -1.58561081e-01 4.08346206e-01 9.04128730e-01
-1.34562230e+00 -5.05741350e-02 -4.90957648e-01 -6.00877047e-01
-1.94877312e-01 2.95640290e-01 6.41419878e-03 -5.34533225e-02
-1.26885128e+00 1.20041490e+00 9.30914208e-02 -3.35117757e-01
-5.90060532e-01 -5.65138876e-01 -4.68219787e-01 1.70998618e-01
3.21772903e-01 -8.36202502e-01 1.63484597e+00 -3.78346771e-01
-1.26438630e+00 7.68411100e-01 -2.90646344e-01 -4.85253423e-01
3.51663291e-01 -1.52970627e-01 -4.10131842e-01 1.57903761e-01
1.50564045e-01 7.90913165e-01 3.33196223e-01 -1.12323487e+00
-5.22633135e-01 4.48712334e-02 3.37403268e-01 1.55215696e-01
-3.95899773e-01 3.46935064e-01 -6.42346144e-01 -4.44416195e-01
-2.40553185e-01 -7.86822081e-01 -2.26698503e-01 -8.13898981e-01
2.17647664e-02 -6.63273811e-01 2.88332105e-01 -6.98064685e-01
1.79916596e+00 -1.56808937e+00 -2.13836581e-01 -1.78175613e-01
8.08979869e-02 4.36798543e-01 -4.42197531e-01 8.25997889e-01
3.76248777e-01 5.63914180e-01 1.15576126e-01 -3.77407700e-01
1.90259174e-01 1.88044652e-01 -3.65667254e-01 -3.34645957e-01
3.97268720e-02 1.24377894e+00 -6.40826941e-01 -5.65326750e-01
-5.72194420e-02 3.96521479e-01 -6.45836473e-01 2.28348807e-01
-4.82297540e-01 8.96683335e-02 -3.56831551e-01 3.62182736e-01
1.03223868e-01 -6.49588048e-01 -1.50354818e-01 5.30473828e-01
1.82177827e-01 4.51734602e-01 -9.20261562e-01 1.32575691e+00
-7.12396979e-01 3.64244550e-01 -5.00337221e-03 -2.81444907e-01
9.54013050e-01 5.69164813e-01 -7.72693520e-03 -9.80273545e-01
1.90707576e-02 4.95043010e-01 2.04916984e-01 -6.61943138e-01
5.79529643e-01 -1.78256273e-01 -2.04535767e-01 8.21669996e-01
3.13553885e-02 -4.57575202e-01 5.44602573e-01 6.59015417e-01
1.05873930e+00 -3.39403957e-01 1.87521074e-02 -1.06934808e-01
4.89967704e-01 -6.29716441e-02 8.55929106e-02 1.17005622e+00
4.97622937e-02 3.96806717e-01 2.46986791e-01 5.32764308e-02
-8.90060782e-01 -8.57584834e-01 2.71802157e-01 1.26430809e+00
-4.79823917e-01 -7.41340756e-01 -1.01813114e+00 -4.08746570e-01
-1.72561392e-01 1.23730171e+00 -2.64570266e-01 -6.45153448e-02
-6.81323469e-01 -7.53277779e-01 6.72160983e-01 4.75655973e-01
3.10946107e-01 -1.15026164e+00 -4.94921654e-01 3.76005173e-01
-7.38561749e-01 -8.78011286e-01 -3.65159839e-01 -4.97282967e-02
-7.84461796e-01 -6.11624658e-01 -1.00698423e+00 -6.51587605e-01
3.11636865e-01 1.98437944e-02 1.55114377e+00 3.47185224e-01
7.52623901e-02 6.80711806e-01 -3.33330482e-01 -4.43947405e-01
-7.97636390e-01 4.49211776e-01 -3.28323603e-01 -4.68666703e-01
5.86767733e-01 -3.80674839e-01 -2.88806677e-01 2.62915313e-01
-8.50808322e-01 -9.99323949e-02 7.01608658e-01 5.06957710e-01
2.52931386e-01 -3.82329315e-01 1.21427190e+00 -9.53490436e-01
1.18867409e+00 -4.75490481e-01 -3.50246191e-01 5.53407371e-01
-8.10485244e-01 2.33859152e-01 4.55227882e-01 -5.70733488e-01
-1.08389676e+00 -7.28452742e-01 -4.88982469e-01 1.60668582e-01
-6.72387704e-02 5.10397077e-01 1.65646538e-01 1.38542712e-01
8.91100407e-01 3.04154195e-02 -3.16716909e-01 -6.23555064e-01
2.71003872e-01 7.19450533e-01 2.51250207e-01 -5.48936903e-01
4.26368445e-01 -1.22018971e-01 -7.34077632e-01 -7.78986037e-01
-7.48665512e-01 -6.12312496e-01 -2.50641666e-02 -8.80970210e-02
7.20687628e-01 -6.86507046e-01 -4.91004825e-01 2.56909251e-01
-1.04770911e+00 -6.51473403e-01 -2.13807628e-01 3.99896830e-01
-4.54805374e-01 2.68385023e-01 -9.73844469e-01 -9.53868449e-01
-5.71596384e-01 -8.96643877e-01 8.12230229e-01 2.20920011e-01
-7.43341625e-01 -1.02949429e+00 2.95883834e-01 1.05630720e+00
9.19533730e-01 -4.68785107e-01 1.17003334e+00 -9.70257640e-01
-5.65792561e-01 -4.20419514e-01 1.49921536e-01 2.55953759e-01
-3.08603913e-01 -2.24818289e-01 -8.71703506e-01 -1.11219116e-01
7.91671276e-02 -6.53512478e-01 7.31044888e-01 1.59432575e-01
7.38213480e-01 -3.82601291e-01 -9.14630201e-03 -2.39248425e-01
1.32969248e+00 2.55724758e-01 4.62952346e-01 3.53607684e-01
2.89176255e-01 6.22478843e-01 3.82853061e-01 3.14172268e-01
7.62300909e-01 4.30158406e-01 -1.61796045e-02 4.20323461e-01
-2.93772370e-01 -3.40097636e-01 3.05021137e-01 1.24988246e+00
1.97143331e-01 -6.96050286e-01 -1.04837251e+00 5.38523376e-01
-1.59601378e+00 -8.88066709e-01 -2.20116153e-01 2.20198822e+00
9.30947065e-01 3.29726994e-01 -2.24112403e-02 -2.44953737e-01
4.24796253e-01 -1.80049036e-02 -3.40332419e-01 -7.06707895e-01
-1.63863257e-01 5.26232839e-01 -2.91693141e-03 1.02187383e+00
-3.54043424e-01 9.82295513e-01 7.04360104e+00 9.86772299e-01
-7.40678668e-01 2.44545832e-01 5.08323371e-01 -1.76099569e-01
-7.17905164e-01 1.62597880e-01 -1.13773263e+00 4.07890975e-01
1.74284196e+00 -3.78701299e-01 2.40816966e-01 5.40737391e-01
2.22317100e-01 -3.98697823e-01 -5.09557068e-01 7.17651665e-01
2.17274636e-01 -1.24627030e+00 2.07786024e-01 -6.46861643e-02
7.52308607e-01 8.34145844e-02 -1.79622591e-01 9.98103619e-01
5.21606863e-01 -1.00298464e+00 6.12041771e-01 8.23648095e-01
2.72311032e-01 -6.92528725e-01 8.25156212e-01 1.01384163e+00
-8.29739988e-01 -9.17815417e-02 -3.42469752e-01 -6.67698532e-02
3.94927144e-01 1.53468043e-01 -1.10600436e+00 1.13464288e-01
6.28051698e-01 -2.00827748e-01 -1.00826633e+00 9.88195121e-01
-7.12143928e-02 1.08083463e+00 -4.15419191e-01 -4.86883283e-01
2.23987699e-01 1.96925595e-01 1.05849311e-01 1.32992601e+00
3.36095065e-01 1.04910329e-01 1.23638846e-01 7.55832911e-01
-3.18402774e-03 4.03491616e-01 -1.32857338e-01 -4.68737721e-01
6.33886456e-01 1.04171920e+00 -4.97095048e-01 -6.04761481e-01
-4.15197700e-01 8.23402464e-01 3.62782687e-01 3.87590766e-01
-7.38183558e-01 -3.10706317e-01 -4.49583270e-02 3.55954319e-01
5.55296652e-02 -1.31269440e-01 -2.22218841e-01 -9.66444135e-01
-1.34676412e-01 -1.10591888e+00 5.28243959e-01 -1.03520262e+00
-1.18608212e+00 5.51512539e-01 4.82772700e-02 -4.32341963e-01
-8.23524177e-01 -2.45305941e-01 -5.88254631e-01 1.25889313e+00
-1.03734136e+00 -9.16446269e-01 -9.51696262e-02 2.08393946e-01
7.06972659e-01 2.46114340e-02 1.14247584e+00 3.53963614e-01
-3.70285273e-01 7.86670327e-01 -7.39196911e-02 -2.06667230e-01
8.02326977e-01 -1.09383225e+00 4.19620454e-01 4.42515433e-01
5.48865162e-02 1.02584779e+00 7.86089897e-01 -6.54124975e-01
-9.92322326e-01 -6.84707284e-01 1.83396304e+00 -1.02285457e+00
6.30210042e-01 -3.16034526e-01 -1.15217447e+00 5.46384752e-01
5.47502041e-01 -8.94326448e-01 8.85805666e-01 1.82630382e-02
-1.64890543e-01 1.33240670e-01 -9.59342301e-01 6.35573447e-01
4.61779177e-01 -5.40427864e-01 -6.60654962e-01 3.61616433e-01
9.02976155e-01 -3.30115482e-02 -7.01799572e-01 3.05778593e-01
1.95744425e-01 -9.10398960e-01 8.64141107e-01 -5.48927903e-01
5.67261018e-02 1.03867359e-01 -1.76837742e-01 -9.60166276e-01
-2.88894355e-01 -3.85798126e-01 -3.08838606e-01 1.38939965e+00
7.46884823e-01 -7.80784965e-01 6.61872268e-01 1.04113305e+00
1.47711695e-03 -9.55953956e-01 -5.00964582e-01 -6.51586711e-01
4.18218493e-01 -6.00591600e-01 3.77752244e-01 5.72273195e-01
-1.66236877e-01 7.98438013e-01 -1.74259305e-01 -2.32132167e-01
7.26417676e-02 -4.85769026e-02 8.70268464e-01 -1.00515330e+00
-6.36124790e-01 -5.23050427e-01 2.30399773e-01 -9.82073128e-01
-4.42546383e-02 -7.81411648e-01 -1.94604233e-01 -1.98152065e+00
4.09556389e-01 -2.93835521e-01 5.52581213e-02 3.98364574e-01
-3.68124276e-01 4.32610720e-01 1.03127286e-01 1.61342040e-01
-8.00277472e-01 3.70980829e-01 1.08649623e+00 9.84275639e-02
-1.07719883e-01 1.35466650e-01 -1.00859916e+00 6.45705581e-01
1.08543754e+00 -4.12760824e-01 -6.11806870e-01 -5.17209053e-01
4.23991859e-01 1.65670410e-01 5.58052696e-02 -9.35665429e-01
3.81020784e-01 -9.66405794e-02 7.74357766e-02 -6.08003139e-01
5.53629220e-01 -2.13176906e-01 1.30043790e-01 4.82560843e-01
-6.63677454e-01 3.73440385e-01 2.14017287e-01 2.75900573e-01
-1.23063281e-01 -6.53113306e-01 4.77682054e-01 -6.27454817e-01
-6.52067959e-01 -1.30834207e-01 -6.51549816e-01 4.16217387e-01
6.83334351e-01 -1.07640192e-01 -3.07387859e-01 -8.50878298e-01
-7.64779031e-01 2.62422979e-01 1.94005355e-01 5.78872085e-01
3.79002631e-01 -9.63799536e-01 -8.78689587e-01 -1.88684434e-01
2.01541454e-01 -4.02449936e-01 3.88934493e-01 5.92498779e-01
-3.95868927e-01 9.82093453e-01 5.44325888e-01 -2.78357625e-01
-1.23222125e+00 2.68581569e-01 1.89712912e-01 -5.80095470e-01
-3.33671600e-01 1.02241683e+00 -1.17533311e-01 -6.32647872e-01
3.59173656e-01 -9.20995027e-02 -2.92143077e-01 2.14304980e-02
6.39434576e-01 4.67618883e-01 1.85622826e-01 -3.05043757e-01
9.57624428e-03 1.59126073e-01 -1.66029930e-01 -6.83880389e-01
9.24009800e-01 -2.21213698e-01 -1.17380783e-01 6.90537632e-01
8.76804829e-01 2.30886564e-01 -6.09831929e-01 -2.91183800e-01
2.63283432e-01 -1.97603852e-01 -3.36323500e-01 -1.19409573e+00
-4.13314164e-01 1.02529299e+00 2.14060947e-01 2.54060835e-01
7.55459487e-01 3.33391577e-01 9.61540639e-01 5.67368090e-01
5.20625234e-01 -1.13400531e+00 5.08697689e-01 8.36250484e-01
8.47509861e-01 -1.01520586e+00 -4.96376038e-01 2.33374357e-01
-6.23295605e-01 7.63494134e-01 7.39602149e-01 1.02717474e-01
2.04147533e-01 1.56449780e-01 3.90370727e-01 -1.61846757e-01
-1.25315952e+00 1.93196838e-03 2.59436846e-01 1.43165514e-01
8.03903580e-01 -1.61900371e-02 -6.56243384e-01 8.40262890e-01
-5.81577420e-01 -1.31796628e-01 3.93294215e-01 8.60012829e-01
-9.14304197e-01 -1.32253647e+00 -3.69564891e-01 5.67056715e-01
-6.60952389e-01 -3.51368517e-01 -5.99358737e-01 6.71229661e-01
-1.42014295e-01 1.47880065e+00 -1.86580122e-01 -1.89442530e-01
2.43332893e-01 6.85508847e-01 3.16505462e-01 -9.60523427e-01
-9.20712173e-01 1.11266365e-02 3.33982378e-01 -2.04729825e-01
-1.99294612e-02 -2.74130881e-01 -1.23296332e+00 -4.75740910e-01
-4.05425489e-01 6.15658760e-01 6.02715790e-01 8.01845670e-01
4.33252305e-01 1.77413285e-01 -4.99128513e-02 -1.92251027e-01
-8.75109792e-01 -1.37535858e+00 -1.07603772e-02 2.55149547e-02
-8.57124403e-02 -6.40945807e-02 -4.20241863e-01 -1.55331209e-01]
|
[11.455549240112305, 8.306624412536621]
|
e4d06a75-a182-4909-aba4-a2f81e313ae0
|
context-aware-mixup-for-domain-adaptive
|
2108.03557
| null |
https://arxiv.org/abs/2108.03557v3
|
https://arxiv.org/pdf/2108.03557v3.pdf
|
Context-Aware Mixup for Domain Adaptive Semantic Segmentation
|
Unsupervised domain adaptation (UDA) aims to adapt a model of the labeled source domain to an unlabeled target domain. Existing UDA-based semantic segmentation approaches always reduce the domain shifts in pixel level, feature level, and output level. However, almost all of them largely neglect the contextual dependency, which is generally shared across different domains, leading to less-desired performance. In this paper, we propose a novel Context-Aware Mixup (CAMix) framework for domain adaptive semantic segmentation, which exploits this important clue of context-dependency as explicit prior knowledge in a fully end-to-end trainable manner for enhancing the adaptability toward the target domain. Firstly, we present a contextual mask generation strategy by leveraging the accumulated spatial distributions and prior contextual relationships. The generated contextual mask is critical in this work and will guide the context-aware domain mixup on three different levels. Besides, provided the context knowledge, we introduce a significance-reweighted consistency loss to penalize the inconsistency between the mixed student prediction and the mixed teacher prediction, which alleviates the negative transfer of the adaptation, e.g., early performance degradation. Extensive experiments and analysis demonstrate the effectiveness of our method against the state-of-the-art approaches on widely-used UDA benchmarks.
|
['Lizhuang Ma', 'Jianping Shi', 'Xuequan Lu', 'Guangliang Cheng', 'Jiangmiao Pang', 'Qiqi Gu', 'Zhengyang Feng', 'Qianyu Zhou']
|
2021-08-08
| null | null | null | null |
['synthetic-to-real-translation']
|
['computer-vision']
|
[ 3.90555322e-01 -4.87517267e-02 -1.49813548e-01 -6.60251260e-01
-7.60022044e-01 -5.29572368e-01 3.45131069e-01 9.93954986e-02
-4.91806090e-01 5.60073972e-01 -3.88840847e-02 -6.72636479e-02
-2.77205277e-02 -8.21579516e-01 -6.52783990e-01 -8.35440934e-01
6.97340906e-01 3.20671499e-01 8.00360382e-01 -7.82932043e-02
1.02403201e-01 1.21470548e-01 -1.23980415e+00 9.48915854e-02
1.56662536e+00 1.11288738e+00 7.45472133e-01 1.26301544e-02
-3.99760038e-01 3.93100113e-01 -5.50595880e-01 -1.69104815e-01
3.44771035e-02 -6.48211718e-01 -7.56808758e-01 3.90432745e-01
2.27193937e-01 -1.44676998e-01 3.75602245e-02 1.28106892e+00
4.70027745e-01 4.06605691e-01 5.98143458e-01 -9.36547875e-01
-5.10096371e-01 4.45552558e-01 -7.83850849e-01 2.15004176e-01
-2.17372239e-01 3.33300680e-01 8.16901147e-01 -8.43323350e-01
6.26623690e-01 1.22212183e+00 4.25167054e-01 4.85632539e-01
-1.19922590e+00 -7.30835795e-01 8.84772599e-01 3.28553498e-01
-1.28719521e+00 -4.96984236e-02 1.16092908e+00 -3.24619740e-01
3.81921142e-01 -1.93311349e-01 4.88854170e-01 8.70070636e-01
-3.34976822e-01 1.05304337e+00 1.18984222e+00 -4.36224759e-01
4.16054785e-01 1.18352108e-01 1.43540353e-01 4.07109588e-01
7.46321902e-02 -2.46224284e-01 -3.53744805e-01 2.15397000e-01
7.33227730e-01 -6.76189661e-02 -1.15616627e-01 -6.93254530e-01
-1.09209096e+00 6.41696632e-01 4.50991482e-01 8.26383010e-02
-3.15566391e-01 -3.92526776e-01 4.03284907e-01 -1.16463393e-01
5.22472978e-01 1.79048881e-01 -6.60967827e-01 1.23190127e-01
-7.19541430e-01 6.87472001e-02 1.70484170e-01 1.04069579e+00
9.32852745e-01 7.86517281e-03 -4.96460676e-01 1.26535189e+00
4.58851665e-01 3.04226369e-01 6.40956581e-01 -7.01567531e-01
5.26003361e-01 9.27356780e-01 -7.96366110e-02 -7.01835632e-01
-2.34201506e-01 -6.15937471e-01 -4.45134163e-01 6.59417957e-02
5.80974936e-01 -2.25637302e-01 -1.38038504e+00 1.91475213e+00
8.07305753e-01 5.68921864e-01 1.12585261e-01 1.14911485e+00
7.46384561e-01 5.87588668e-01 4.69510347e-01 -5.95353208e-02
1.25537050e+00 -1.32551944e+00 -5.82443118e-01 -5.45876086e-01
4.27827358e-01 -7.33093798e-01 1.39505172e+00 1.39084876e-01
-8.73352408e-01 -7.58415163e-01 -1.13287067e+00 2.80333273e-02
-2.34633982e-01 1.53411180e-01 2.41070509e-01 3.82227629e-01
-5.72552443e-01 2.80378431e-01 -7.57040143e-01 -3.68521094e-01
6.33144736e-01 1.59387738e-01 1.05873793e-01 -3.03175509e-01
-1.27985275e+00 5.34381628e-01 7.32439518e-01 -1.51240021e-01
-8.02829146e-01 -6.35990381e-01 -8.92183900e-01 -5.37879067e-03
7.67068982e-01 -4.45678353e-01 1.33157372e+00 -1.14040947e+00
-1.64893258e+00 7.18277574e-01 -9.16686505e-02 -3.35662961e-01
5.47621608e-01 -1.55441940e-01 -3.93980384e-01 9.52187032e-02
3.03582400e-01 9.12909567e-01 8.17726552e-01 -1.39165676e+00
-9.46507514e-01 -4.39234346e-01 -1.27473131e-01 6.29448533e-01
-5.09608746e-01 -3.68993759e-01 -9.69057620e-01 -1.00151944e+00
3.38927329e-01 -7.88881004e-01 -4.70850468e-01 -9.36146379e-02
-3.30947042e-01 -2.45155558e-01 9.54301596e-01 -4.25972730e-01
1.27980256e+00 -2.32277131e+00 1.49857938e-01 1.65928677e-01
-1.03338115e-01 4.37498510e-01 -1.87899187e-01 -1.23272933e-01
8.26797187e-02 -2.58977145e-01 -8.19878519e-01 -1.87884197e-01
-8.26615617e-02 2.99780071e-01 -1.37006283e-01 9.44861323e-02
3.15908968e-01 5.84758878e-01 -1.12883770e+00 -7.89477646e-01
3.40953737e-01 3.66382718e-01 -5.47278762e-01 3.20498139e-01
-6.47357583e-01 7.70942271e-01 -8.21859300e-01 5.95115602e-01
9.10346329e-01 -2.11918414e-01 1.81400180e-01 -1.22288786e-01
5.17142341e-02 1.77670911e-01 -1.17356956e+00 2.05089879e+00
-5.07456422e-01 1.52608350e-01 6.23565577e-02 -1.05958259e+00
1.13567615e+00 -1.17509626e-01 2.71919906e-01 -9.67093110e-01
1.67840406e-01 3.33942056e-01 -4.64358442e-02 -2.36160219e-01
3.52152258e-01 -3.39626484e-02 -1.05089717e-01 1.79969132e-01
6.50120527e-02 -3.13467942e-02 2.16989294e-02 -4.22289744e-02
5.56187272e-01 5.93474090e-01 1.68816581e-01 -3.31427217e-01
6.85099959e-01 1.99932039e-01 1.06682634e+00 4.25804526e-01
-5.21563053e-01 8.47372115e-01 3.67080271e-01 1.27735604e-02
-7.96861887e-01 -1.01282489e+00 -1.44256249e-01 1.30208361e+00
7.91543901e-01 8.69953558e-02 -1.11353838e+00 -1.11945772e+00
-6.78432211e-02 8.34374487e-01 -4.10168260e-01 -3.87487024e-01
-5.33870816e-01 -7.95814276e-01 1.07771412e-01 7.29462087e-01
8.44762981e-01 -1.10972607e+00 -3.78280520e-01 3.41057688e-01
-1.29299372e-01 -1.30051351e+00 -7.09047794e-01 1.81464627e-01
-1.04935682e+00 -7.69920170e-01 -1.06648195e+00 -1.07154715e+00
8.36549759e-01 2.87371010e-01 9.38427091e-01 -3.10261250e-01
1.39772013e-01 1.09250247e-01 -3.51165116e-01 -1.96764514e-01
-2.37228900e-01 1.37811154e-01 -2.53533572e-01 1.67176247e-01
5.27030766e-01 -4.77213740e-01 -8.55911255e-01 5.86369216e-01
-9.90850985e-01 2.05696568e-01 7.04594314e-01 8.88323188e-01
1.11285710e+00 6.69982731e-02 8.42775285e-01 -1.10784411e+00
5.10995209e-01 -5.55611670e-01 -5.45386851e-01 2.57190377e-01
-8.09550703e-01 -1.84135474e-02 6.93233311e-01 -6.75618052e-01
-1.56465542e+00 3.35979849e-01 -1.79758862e-01 -4.81026947e-01
-3.59668791e-01 3.74049604e-01 -8.13653767e-01 3.96229178e-02
5.71842253e-01 3.92318219e-01 -3.16792220e-01 -5.65240145e-01
4.63347644e-01 5.31989217e-01 7.03976870e-01 -8.46214116e-01
5.31146049e-01 3.45769018e-01 -3.77602160e-01 -4.38995093e-01
-9.12400901e-01 -5.82302213e-01 -7.15475857e-01 -1.01540796e-01
9.15300906e-01 -9.53188956e-01 6.96105510e-02 7.32663035e-01
-9.11288619e-01 -6.27584517e-01 -4.13115323e-01 2.98856765e-01
-3.95765901e-01 4.07759547e-01 -2.39089638e-01 -4.08203244e-01
-1.64530098e-01 -1.34573793e+00 9.55452025e-01 7.32929766e-01
1.17691323e-01 -9.49706376e-01 -1.07367799e-01 4.65311021e-01
7.87144750e-02 2.37445265e-01 9.01380181e-01 -7.99764693e-01
-3.50063771e-01 2.89933771e-01 -6.15312397e-01 5.10661781e-01
2.62962312e-01 -4.01569754e-01 -9.40687120e-01 -2.39060819e-02
-2.21962601e-01 -1.68909624e-01 9.44219649e-01 3.92408431e-01
1.33476937e+00 -2.68721301e-02 -3.71148616e-01 5.16626537e-01
1.32028127e+00 4.51729059e-01 3.68401766e-01 3.96945953e-01
1.00688541e+00 6.60255790e-01 1.16556323e+00 3.57593089e-01
5.05049407e-01 7.45975971e-01 3.30745399e-01 -2.46648982e-01
-3.47463995e-01 -3.66221875e-01 8.92170593e-02 6.30867422e-01
4.55741584e-01 -2.10238278e-01 -9.26242530e-01 9.14656401e-01
-1.95240963e+00 -2.09652483e-01 6.26166388e-02 2.09190655e+00
1.14484274e+00 4.18422788e-01 1.90738320e-01 -2.10794970e-01
9.87819672e-01 1.50162816e-01 -1.09412730e+00 -9.81671959e-02
3.46701331e-02 8.03733692e-02 4.78524059e-01 3.57432216e-01
-1.19015431e+00 1.26136887e+00 5.10365582e+00 1.29827809e+00
-1.07863450e+00 2.05071360e-01 7.37868607e-01 1.65885001e-01
-4.63593423e-01 -1.51951700e-01 -6.63015962e-01 7.20564485e-01
3.93087000e-01 -1.94507614e-02 4.76689897e-02 1.17055655e+00
1.91172898e-01 -1.18545450e-01 -7.96283901e-01 6.08065665e-01
-2.03413054e-01 -8.73276472e-01 7.62924924e-02 -2.07221836e-01
1.06118155e+00 -2.48797446e-01 9.34839919e-02 4.35479015e-01
3.76708120e-01 -4.98399019e-01 7.40613818e-01 7.26016164e-02
6.95058048e-01 -8.87886226e-01 6.09449089e-01 2.82264590e-01
-1.23056507e+00 -2.70322859e-02 -2.08949685e-01 3.25540870e-01
4.14546691e-02 6.01245701e-01 -5.96530318e-01 5.13462722e-01
6.09029233e-01 7.13475049e-01 -5.48888326e-01 8.89476597e-01
-4.28775042e-01 8.31808627e-01 -2.24770516e-01 2.14184105e-01
6.60113990e-01 -2.94864148e-01 5.63870668e-01 1.27280807e+00
9.37281623e-02 1.66407287e-01 3.67070496e-01 8.24949801e-01
-6.87102079e-02 2.09218547e-01 -4.10131291e-02 2.72881866e-01
6.83622062e-01 1.09720361e+00 -1.03851056e+00 -3.93637031e-01
-4.56858754e-01 1.16737401e+00 1.95703208e-01 4.72139865e-01
-8.42235506e-01 -3.69226098e-01 6.11840427e-01 1.17653720e-01
5.03393233e-01 9.69240814e-02 -7.23928988e-01 -9.14778113e-01
8.82298872e-02 -6.91150486e-01 6.20521605e-01 -5.43429017e-01
-1.31857133e+00 3.92838955e-01 1.72683328e-01 -1.38087952e+00
-1.35438386e-04 -2.63074726e-01 -7.67758846e-01 8.99688244e-01
-1.80528057e+00 -1.16711879e+00 -4.79089320e-01 5.23426056e-01
9.04506803e-01 -4.34848182e-02 2.52250493e-01 4.24571127e-01
-8.46474946e-01 7.25356281e-01 6.38403147e-02 -2.80034821e-02
9.12012279e-01 -1.27746785e+00 3.59117717e-01 1.04231608e+00
-2.55741715e-01 3.71838897e-01 5.31351507e-01 -6.02929354e-01
-6.16252184e-01 -1.37814140e+00 3.34214538e-01 -5.26649058e-02
4.03322846e-01 -1.42106727e-01 -1.33157670e+00 3.49915743e-01
-5.68263568e-02 1.31272197e-01 4.14481938e-01 -7.44207650e-02
-2.13851392e-01 -3.33783686e-01 -1.25186777e+00 6.62840128e-01
9.77333248e-01 -2.71961331e-01 -5.24029493e-01 1.57503113e-02
9.78577673e-01 -5.88961363e-01 -5.89392662e-01 6.50280237e-01
2.55462110e-01 -8.35931242e-01 9.26449299e-01 -2.50741005e-01
4.71221656e-01 -6.43460631e-01 1.61119625e-01 -1.33250988e+00
-2.91019291e-01 -1.14516012e-01 2.20582765e-02 1.63403332e+00
2.75909394e-01 -5.31504989e-01 8.97734344e-01 4.26791996e-01
-3.33876610e-01 -8.81355166e-01 -7.42529213e-01 -7.29686201e-01
2.45626152e-01 -3.79244715e-01 7.61579394e-01 1.20278525e+00
-4.26872075e-01 3.02671909e-01 8.91315565e-02 2.84901857e-01
5.37103832e-01 2.85795271e-01 5.23230076e-01 -1.03914976e+00
-1.69671357e-01 -6.32342041e-01 -2.01460585e-01 -1.35548329e+00
-2.90417690e-02 -6.83423400e-01 3.89755338e-01 -1.58337855e+00
8.82196352e-02 -7.33167887e-01 -6.91298842e-01 4.91705894e-01
-6.49699509e-01 1.76793467e-02 4.14309613e-02 1.98431030e-01
-7.21190095e-01 6.27409995e-01 1.59929848e+00 -9.06606689e-02
-5.18707514e-01 -2.67436318e-02 -8.31153929e-01 8.63856971e-01
7.63502717e-01 -4.79947418e-01 -8.29841673e-01 -5.21753907e-01
-5.05119026e-01 -2.70651460e-01 2.14780688e-01 -1.07597399e+00
1.04255110e-01 -4.85582739e-01 2.77382076e-01 -5.40680528e-01
-1.08160168e-01 -8.12755585e-01 -4.45401162e-01 8.35808143e-02
-3.17461938e-01 -5.42601347e-01 3.37436259e-01 7.80196667e-01
-3.55709523e-01 -1.93398997e-01 1.20220923e+00 8.82867258e-03
-1.39375556e+00 2.96938449e-01 7.78579712e-02 4.19674128e-01
1.17492533e+00 -3.06207210e-01 -4.02873866e-02 8.96616504e-02
-6.12918496e-01 6.40199006e-01 5.77741623e-01 4.17702496e-01
3.88514161e-01 -1.22188354e+00 -5.08394361e-01 1.36064127e-01
4.49101299e-01 7.53688693e-01 6.08173490e-01 5.49626768e-01
-2.48309851e-01 2.26924438e-02 -2.53356576e-01 -7.38215029e-01
-9.16311741e-01 4.67780262e-01 2.92201042e-01 -3.66744399e-01
-6.00161910e-01 9.48007405e-01 7.87073314e-01 -5.36757708e-01
2.38906443e-01 -3.32113028e-01 -3.62600744e-01 1.37436287e-02
2.93723553e-01 1.41557679e-01 -9.87902060e-02 -5.78764558e-01
-4.20958340e-01 6.07973516e-01 -2.72994846e-01 -1.55200148e-02
9.58929002e-01 -3.57554793e-01 4.34464693e-01 2.69259483e-01
8.49997222e-01 -9.83778015e-02 -2.05815363e+00 -6.21315479e-01
-1.75821129e-02 -2.97374874e-01 1.10028289e-01 -1.05974627e+00
-1.25143647e+00 9.92665589e-01 7.19554961e-01 -2.12679327e-01
1.51475537e+00 -4.90565039e-02 1.15666699e+00 -9.31988955e-02
4.16298881e-02 -1.46364391e+00 1.52468368e-01 3.71962368e-01
3.06173116e-01 -1.28552270e+00 -1.21318489e-01 -6.62639737e-01
-1.09963691e+00 6.65557027e-01 1.17051637e+00 9.31443349e-02
2.86690980e-01 7.93923810e-03 2.69076318e-01 2.65927702e-01
-2.99817115e-01 -4.23203707e-01 3.07500869e-01 8.23825121e-01
6.90806434e-02 -2.14484497e-03 -3.88314128e-01 1.00088000e+00
3.66865039e-01 -2.51794625e-02 8.04324746e-02 9.05922949e-01
-5.63610077e-01 -1.04849458e+00 -3.56738120e-01 8.55362192e-02
-2.35170484e-01 -2.57285801e-03 -1.91896498e-01 6.17236435e-01
4.83083785e-01 7.26543665e-01 1.50759835e-02 -2.37157091e-01
6.24438345e-01 5.10342568e-02 1.07097566e-01 -6.86188459e-01
-2.32268721e-01 3.16397190e-01 -2.63016433e-01 -2.97085375e-01
-3.15643787e-01 -6.29342496e-01 -1.77137721e+00 2.78060377e-01
-2.64819801e-01 -4.96578738e-02 4.21994537e-01 1.07637572e+00
4.21610624e-01 8.22300255e-01 5.16604125e-01 -4.70622867e-01
-2.57963598e-01 -9.12665904e-01 -4.11727488e-01 4.72861767e-01
1.52114823e-01 -8.09272110e-01 1.83785055e-03 2.07099423e-01]
|
[9.681595802307129, 1.4299215078353882]
|
5a6c32ff-7d03-4f2d-a041-aba6b4d5950a
|
protein-secondary-structure-prediction-with
|
1412.7828
| null |
http://arxiv.org/abs/1412.7828v2
|
http://arxiv.org/pdf/1412.7828v2.pdf
|
Protein Secondary Structure Prediction with Long Short Term Memory Networks
|
Prediction of protein secondary structure from the amino acid sequence is a
classical bioinformatics problem. Common methods use feed forward neural
networks or SVMs combined with a sliding window, as these models does not
naturally handle sequential data. Recurrent neural networks are an
generalization of the feed forward neural network that naturally handle
sequential data. We use a bidirectional recurrent neural network with long
short term memory cells for prediction of secondary structure and evaluate
using the CB513 dataset. On the secondary structure 8-class problem we report
better performance (0.674) than state of the art (0.664). Our model includes
feed forward networks between the long short term memory cells, a path that can
be further explored.
|
['Søren Kaae Sønderby', 'Ole Winther']
|
2014-12-25
| null | null | null | null |
['protein-secondary-structure-prediction']
|
['medical']
|
[ 4.17108566e-01 -5.76072112e-02 -3.36509556e-01 -3.70393336e-01
-2.69092977e-01 -2.59470880e-01 2.97533065e-01 2.31697112e-01
-5.54072261e-01 8.36911142e-01 1.99020028e-01 -1.07911432e+00
1.47595406e-01 -7.10343540e-01 -9.35927689e-01 -9.41337228e-01
-2.72465914e-01 4.32233602e-01 4.61195290e-01 -4.42957938e-01
1.66430727e-01 4.44100648e-01 -1.22027183e+00 8.84786725e-01
4.41767812e-01 7.82836437e-01 4.73512143e-01 1.00621557e+00
-5.70867002e-01 1.18087959e+00 -4.28990632e-01 1.52149469e-01
-6.48973435e-02 -4.18151826e-01 -1.21780384e+00 -7.26390183e-01
2.12068837e-02 -8.29685014e-03 3.19726281e-02 4.59446251e-01
5.56180477e-01 -2.57104002e-02 4.36575919e-01 -5.84794700e-01
-5.64355791e-01 5.43229640e-01 -2.86259145e-01 3.95307839e-01
4.16532576e-01 4.98591550e-02 1.07707906e+00 -1.08919132e+00
9.54058290e-01 1.03020716e+00 1.10682881e+00 3.67207259e-01
-1.53031182e+00 -5.06508231e-01 2.03584880e-01 7.17378199e-01
-8.15114677e-01 -6.46053851e-02 3.01216543e-01 -3.25034350e-01
2.07309055e+00 2.35843524e-01 7.11026669e-01 1.14007747e+00
6.16821170e-01 8.12207043e-01 1.04572666e+00 -5.97730279e-01
1.01766124e-01 -3.86179626e-01 9.70762551e-01 6.83959126e-01
-3.49112451e-01 3.83581161e-01 -5.55625021e-01 -6.66419685e-01
3.36999893e-01 4.70266402e-01 -1.58830360e-01 -2.86264330e-01
-9.25264716e-01 1.06630707e+00 5.35913467e-01 2.25571454e-01
-5.41474044e-01 -8.59156623e-02 6.50714815e-01 9.97656643e-01
4.40438241e-01 4.44149464e-01 -1.10210943e+00 -4.27114069e-02
-8.96667182e-01 1.35113701e-01 1.03558052e+00 4.40110296e-01
5.77572644e-01 -1.62498206e-01 1.19758256e-01 9.64918733e-01
2.93283552e-01 1.15049646e-04 1.04012096e+00 -1.74755841e-01
1.06391288e-01 6.19546533e-01 -1.13422915e-01 -4.05348688e-01
-8.75549614e-01 -3.07038158e-01 -8.05254340e-01 3.91043425e-01
5.85856199e-01 -1.48603424e-01 -1.11906111e+00 1.27367580e+00
1.90265439e-02 1.10461406e-01 3.60630333e-01 6.95089161e-01
9.77015853e-01 1.10954046e+00 -1.87331066e-02 -4.92295176e-01
1.07011354e+00 -1.07921517e+00 -6.33905053e-01 4.93286252e-02
1.15645957e+00 -8.90105307e-01 6.65646255e-01 5.01902401e-01
-7.83595622e-01 -2.45870084e-01 -1.06752360e+00 -2.22177178e-01
-7.72050858e-01 -8.54171515e-02 6.25465333e-01 2.59261370e-01
-8.59422624e-01 8.81474912e-01 -1.11614323e+00 -3.98925930e-01
-1.14696287e-01 6.78871810e-01 -2.59808034e-01 4.72411960e-01
-1.43268645e+00 1.12731266e+00 4.22141552e-01 2.90717073e-02
-3.65839273e-01 -6.84244335e-01 -6.37588561e-01 -1.10923730e-01
-3.68332188e-03 -4.65250313e-01 1.24314761e+00 -1.02989674e+00
-1.71264434e+00 6.67286932e-01 -4.66854274e-01 -1.25271928e+00
5.54690585e-02 -1.74870074e-01 -1.33735016e-01 -2.11437285e-01
-5.07971227e-01 5.48296571e-01 2.84402996e-01 -4.15113896e-01
-4.79153126e-01 -2.59132922e-01 -4.36096191e-01 -2.96419561e-02
-4.45479974e-02 1.54692322e-01 5.44652283e-01 -6.88120008e-01
1.13877602e-01 -1.21378601e+00 -5.02233803e-01 -2.70354956e-01
-9.64899883e-02 -3.62717181e-01 7.47013152e-01 -6.85364544e-01
1.20535564e+00 -1.56438720e+00 1.59301102e-01 7.23215491e-02
-4.68635000e-02 5.05577385e-01 -2.22137481e-01 7.12851644e-01
-7.51787841e-01 -3.53890777e-01 -6.17333129e-02 3.66252005e-01
-4.68254894e-01 1.38783827e-01 -6.41480386e-01 3.50781798e-01
1.38080344e-01 1.03334415e+00 -4.40599442e-01 9.62454230e-02
-3.28584500e-02 4.61117804e-01 -3.03790003e-01 3.32260340e-01
-4.82691169e-01 -1.50876552e-01 -9.09298733e-02 3.46471757e-01
2.85250455e-01 -8.02766323e-01 5.32152295e-01 8.54664668e-02
-2.20405847e-01 6.79032445e-01 -4.62852418e-01 1.59818387e+00
-2.70751029e-01 5.75264990e-01 -5.74253142e-01 -1.03193593e+00
1.28857327e+00 5.03444672e-01 3.87104183e-01 -5.13060033e-01
-1.39226824e-01 1.93159014e-01 4.67708081e-01 -3.92745167e-01
1.50745660e-01 -4.65039909e-01 4.52339262e-01 4.68955070e-01
2.28202164e-01 6.26542687e-01 -1.02049401e-02 -1.57459214e-01
1.41296589e+00 3.42696607e-01 5.25273263e-01 -1.34174466e-01
3.96611691e-01 1.50548279e-01 6.12699449e-01 5.34584582e-01
-5.78472987e-02 5.96873939e-01 5.41710913e-01 -1.19152033e+00
-1.08462775e+00 -5.54183602e-01 -1.54041946e-01 1.68822789e+00
-4.43930686e-01 -5.73559880e-01 -3.78279388e-01 -7.16318130e-01
-1.68605775e-01 3.49818319e-01 -7.05338478e-01 -1.50940940e-02
-7.62542307e-01 -1.08743501e+00 4.59081650e-01 5.87659776e-01
-2.34987512e-01 -1.53201115e+00 -6.66280210e-01 5.58037698e-01
-5.04079051e-02 -5.56665599e-01 -8.82220268e-02 1.26657903e+00
-1.18390179e+00 -1.06509721e+00 -1.01564276e+00 -1.15138316e+00
1.84221685e-01 -1.02739401e-01 1.09819365e+00 1.54111549e-01
-1.69842556e-01 -7.21333981e-01 -3.57160866e-01 -3.09021264e-01
-5.08621216e-01 4.67649430e-01 -4.30528969e-02 -4.09495860e-01
7.58818626e-01 -5.70618033e-01 -3.94940972e-01 2.87839800e-01
-4.52738941e-01 3.27455670e-01 7.01654375e-01 1.63080978e+00
6.57849371e-01 -6.68866634e-01 5.61809123e-01 -1.35471070e+00
2.99246222e-01 -5.34002244e-01 -5.06369293e-01 3.35774362e-01
-8.78465235e-01 5.61776042e-01 9.06252027e-01 -5.25910497e-01
-9.60000753e-01 4.46087778e-01 -6.97490275e-01 -1.51878461e-01
-1.24669500e-01 6.99902713e-01 5.30484140e-01 -9.40044597e-02
9.44501519e-01 4.01172489e-01 4.49421227e-01 -7.13159919e-01
1.30133390e-01 6.53547168e-01 4.42490205e-02 2.63436302e-03
-1.07734248e-01 1.08210720e-01 -6.52688816e-02 -7.66922176e-01
-5.14530778e-01 -5.71213186e-01 -5.51834345e-01 3.13316524e-01
5.08676171e-01 -6.29641354e-01 -7.98959434e-01 3.87880921e-01
-1.21645415e+00 -5.76665342e-01 9.24756899e-02 4.95286703e-01
-8.78617525e-01 2.81731993e-01 -1.42552876e+00 -4.75616127e-01
-7.64743984e-01 -1.07325017e+00 6.82589054e-01 -7.13720545e-02
-6.64324999e-01 -9.60121334e-01 4.93886530e-01 -1.18664000e-02
3.22802424e-01 -8.60036835e-02 1.06186664e+00 -1.05600166e+00
-1.32297143e-01 1.77035257e-01 2.92845219e-02 -4.39306088e-02
2.03865729e-02 2.64868569e-02 -7.60786593e-01 -3.74969751e-01
-1.17292637e-02 -4.72797841e-01 1.42608225e+00 6.62618995e-01
8.39532256e-01 -2.66296923e-01 -5.00840902e-01 5.79764068e-01
1.23482895e+00 4.86613780e-01 6.97833180e-01 5.73276281e-01
3.28316212e-01 6.09154344e-01 4.39056545e-01 1.67880550e-01
-3.73269357e-02 4.11172658e-01 2.65829802e-01 -4.20636415e-01
1.46590292e-01 -6.12504967e-02 5.11736929e-01 8.35457861e-01
1.01965331e-01 4.79389355e-02 -1.09636414e+00 2.24905282e-01
-2.05725098e+00 -9.94767725e-01 -4.62014169e-01 1.76544142e+00
1.17385113e+00 1.58102348e-01 4.11501288e-01 1.70344353e-01
5.15688062e-01 -2.64302921e-02 -6.38663769e-01 -1.12363625e+00
-2.64862716e-01 3.86380225e-01 6.40657067e-01 4.65355068e-01
-9.35884535e-01 8.71342361e-01 7.74352598e+00 7.84881890e-01
-1.48120284e+00 -2.40241736e-01 7.89977372e-01 -1.79523975e-01
8.14527944e-02 -1.95416361e-02 -1.08961737e+00 4.73071605e-01
1.68617773e+00 2.77524740e-01 9.06866193e-02 8.00049305e-01
-2.86866110e-02 1.56048432e-01 -9.62776244e-01 5.78671992e-01
-1.88261554e-01 -1.87479961e+00 -1.15967080e-01 3.81235890e-02
3.97854418e-01 6.14107072e-01 -3.07501368e-02 3.25250059e-01
5.72080493e-01 -1.19342077e+00 8.55078641e-03 6.49433672e-01
4.37771857e-01 -7.21673906e-01 8.71026039e-01 5.68509817e-01
-8.14263880e-01 -5.23418970e-02 -5.22784531e-01 -4.45672542e-01
-8.16637278e-03 5.29506862e-01 -1.09516263e+00 5.22355922e-02
8.50436866e-01 9.98225987e-01 -5.17364919e-01 8.47300291e-01
1.25439733e-01 9.74017084e-01 -2.58047938e-01 -5.62752068e-01
2.21761793e-01 -6.27198219e-02 3.18620682e-01 1.50182962e+00
-2.35795289e-01 -2.65484210e-02 1.27206638e-01 3.64024222e-01
2.61481285e-01 3.33141863e-01 -6.11606598e-01 -6.73807114e-02
-1.96995631e-01 9.07834113e-01 -6.50791705e-01 -3.70412797e-01
-4.82934177e-01 8.44699264e-01 6.11305714e-01 3.95497203e-01
-4.88892615e-01 -6.11171722e-01 7.09497809e-01 4.45689484e-02
6.44529104e-01 -2.67012650e-03 -3.12599450e-01 -8.35361779e-01
-4.65281874e-01 -1.03626120e+00 6.12037122e-01 -7.07440317e-01
-1.42518556e+00 7.35220611e-01 -6.85158372e-01 -9.60571051e-01
-6.00110888e-01 -1.02470171e+00 -4.24950480e-01 9.38987434e-01
-1.30825937e+00 -1.08385217e+00 5.69217443e-01 1.16774648e-01
5.87159395e-01 -3.09927225e-01 1.37129068e+00 -1.44820765e-01
-4.13002759e-01 3.55006605e-01 5.09113610e-01 -2.28687841e-02
8.06735218e-01 -1.32511890e+00 7.95572400e-01 2.76047856e-01
1.69533998e-01 1.01227450e+00 8.58168662e-01 -6.09281063e-01
-1.20155573e+00 -9.83125746e-01 1.34067643e+00 -1.47386074e-01
7.32777774e-01 -3.20243090e-01 -1.51842332e+00 8.35254967e-01
2.75439799e-01 2.63236742e-02 1.13211560e+00 6.79038405e-01
-4.04328555e-01 3.41263890e-01 -7.23040223e-01 4.46405113e-01
7.80266345e-01 -5.12073219e-01 -7.60081649e-01 2.85300195e-01
9.83246386e-01 -2.59955227e-01 -8.95095408e-01 4.95966971e-01
9.68704343e-01 -9.53126490e-01 8.77589881e-01 -1.03472495e+00
4.03487027e-01 -1.31093681e-01 1.83716163e-01 -1.21955132e+00
-4.70988065e-01 -5.11388838e-01 -2.95306981e-01 3.55448216e-01
9.47087288e-01 -7.37441361e-01 1.19240332e+00 1.75207313e-02
6.62458539e-02 -1.15103650e+00 -5.99810421e-01 -7.62404919e-01
4.72708732e-01 -8.59917141e-03 2.77514666e-01 7.87178218e-01
5.21524370e-01 7.02574432e-01 -4.61651295e-01 -5.59218764e-01
-8.91785622e-02 5.41608691e-01 2.93319315e-01 -1.02426922e+00
-5.33807635e-01 -3.61242801e-01 -3.91146541e-01 -1.26638031e+00
3.86892378e-01 -9.43023682e-01 1.31663963e-01 -1.13127911e+00
2.98839808e-01 -2.07323418e-03 -6.45556688e-01 9.02523220e-01
-5.73656075e-02 1.31153435e-01 -2.57227331e-01 1.89112663e-01
-4.61622804e-01 3.18672597e-01 6.29803479e-01 -1.75564617e-01
-3.74657750e-01 1.63617015e-01 -1.42304108e-01 6.04300976e-01
1.03700089e+00 -4.86686319e-01 4.83294204e-02 1.68344349e-01
4.30012941e-01 2.03872576e-01 -8.86703357e-02 -6.23754501e-01
2.86600739e-01 -4.61510941e-02 5.96361101e-01 -9.50986922e-01
2.80160427e-01 -5.11543930e-01 1.99712232e-01 1.10472965e+00
-8.42049718e-01 3.00948828e-01 8.74945521e-02 5.66822648e-01
-1.38438374e-01 -5.09455577e-02 6.95702493e-01 -7.01250508e-02
-4.43554431e-01 -1.79400761e-02 -1.13834310e+00 -6.94485366e-01
7.26674318e-01 -2.74871886e-01 -1.97267637e-01 2.47625355e-02
-1.34353423e+00 2.95512937e-02 3.46568227e-01 3.56388718e-01
5.98041892e-01 -9.95499074e-01 -5.13457060e-01 5.60704648e-01
8.82145017e-02 -7.15925097e-01 -8.60238299e-02 7.64358103e-01
-7.05842674e-01 7.79267430e-01 -4.19446290e-01 -6.50465786e-01
-1.94191682e+00 7.74957001e-01 5.05738080e-01 -4.09855783e-01
-6.85310543e-01 9.18650389e-01 -5.72692119e-02 -7.04962671e-01
1.22375675e-01 -4.66029614e-01 -5.29396176e-01 1.19552411e-01
5.68036735e-01 2.72681713e-01 3.76897186e-01 -3.13481659e-01
-4.41418052e-01 1.92240044e-01 -5.41261435e-01 3.13681901e-01
1.59583151e+00 3.06622863e-01 -3.46066415e-01 9.45560992e-01
1.47026312e+00 -7.53893673e-01 -8.78376007e-01 -3.97221714e-01
5.53462982e-01 3.01768631e-01 -2.76246786e-01 -9.83282447e-01
-5.42713940e-01 9.02907670e-01 9.84653771e-01 1.56726301e-01
8.02712858e-01 -3.79934549e-01 1.03100955e+00 8.38681757e-01
7.02060163e-02 -9.83351886e-01 -2.72616178e-01 1.14264739e+00
6.09644055e-01 -1.21943605e+00 -2.38950089e-01 -9.35733095e-02
-3.31280053e-01 1.59603190e+00 2.74633229e-01 -4.44409043e-01
8.78690720e-01 6.83582366e-01 3.01874161e-01 -2.34847851e-02
-1.79683995e+00 -3.29279192e-02 3.80238076e-03 2.52419114e-01
1.23985612e+00 1.06995609e-02 -6.80980682e-01 3.74228895e-01
-2.10194752e-01 2.68929005e-01 3.86176616e-01 1.04211700e+00
-6.02818012e-01 -1.48170543e+00 -2.04919294e-01 6.58985972e-01
-6.88416839e-01 -4.72648293e-01 -7.24801421e-01 -6.28900081e-02
-3.49648505e-01 8.05996656e-01 1.48466257e-02 -3.62333834e-01
-9.70405787e-02 4.66257989e-01 2.70482957e-01 -5.87267160e-01
-1.23318183e+00 1.95873678e-02 1.50358826e-01 -6.23230875e-01
-2.00018302e-01 -4.58980143e-01 -1.51634896e+00 -2.87430733e-01
-6.30232990e-01 2.71016806e-01 6.28968596e-01 7.94423342e-01
5.12155175e-01 4.36668962e-01 3.61820877e-01 -4.32553440e-01
-5.37762880e-01 -1.09908950e+00 -2.98484504e-01 1.98746622e-01
5.50314724e-01 -2.52394557e-01 -2.90377438e-03 1.16674624e-01]
|
[4.717435359954834, 5.618133068084717]
|
8dda4c4b-5992-439b-95a6-737efd488bed
|
on-large-scale-dynamic-topic-modeling-with
|
2001.00631
| null |
https://arxiv.org/abs/2001.00631v2
|
https://arxiv.org/pdf/2001.00631v2.pdf
|
On Large-Scale Dynamic Topic Modeling with Nonnegative CP Tensor Decomposition
|
There is currently an unprecedented demand for large-scale temporal data analysis due to the explosive growth of data. Dynamic topic modeling has been widely used in social and data sciences with the goal of learning latent topics that emerge, evolve, and fade over time. Previous work on dynamic topic modeling primarily employ the method of nonnegative matrix factorization (NMF), where slices of the data tensor are each factorized into the product of lower-dimensional nonnegative matrices. With this approach, however, information contained in the temporal dimension of the data is often neglected or underutilized. To overcome this issue, we propose instead adopting the method of nonnegative CANDECOMP/PARAPAC (CP) tensor decomposition (NNCPD), where the data tensor is directly decomposed into a minimal sum of outer products of nonnegative vectors, thereby preserving the temporal information. The viability of NNCPD is demonstrated through application to both synthetic and real data, where significantly improved results are obtained compared to those of typical NMF-based methods. The advantages of NNCPD over such approaches are studied and discussed. To the best of our knowledge, this is the first time that NNCPD has been utilized for the purpose of dynamic topic modeling, and our findings will be transformative for both applications and further developments.
|
['Deanna Needell', 'Kathryn Leonard', 'Alona Kryshchenko', 'Miju Ahn', 'R. W. M. A. Madushani', 'Chuntian Wang', 'Nicole Eikmeier', 'Jamie Haddock', 'Lara Kassab', 'Elena Sizikova']
|
2020-01-02
| null | null | null | null |
['dynamic-topic-modeling']
|
['natural-language-processing']
|
[-1.45507067e-01 -3.98190409e-01 -1.85820028e-01 1.07785888e-01
9.98444390e-03 -6.44027591e-01 7.87063658e-01 2.89330650e-02
-1.86816528e-01 3.95243376e-01 2.29641825e-01 -3.37899804e-01
-2.63657331e-01 -5.11655867e-01 -1.96921870e-01 -8.43873501e-01
-4.48780477e-01 4.04916078e-01 8.88433978e-02 -1.26680434e-01
8.26529562e-02 2.56707221e-01 -1.57358277e+00 3.02575618e-01
5.23335397e-01 5.70910394e-01 4.61831726e-02 3.49423200e-01
-3.36458981e-01 4.58624840e-01 -4.54534590e-01 -3.28924894e-01
2.78928846e-01 2.69818455e-02 -5.70636392e-01 5.55335701e-01
-3.92553657e-02 3.47829238e-02 -2.81111151e-01 7.72866070e-01
-2.87793321e-03 3.77153248e-01 4.11048859e-01 -1.60405803e+00
-4.16391671e-01 2.91685402e-01 -9.57331359e-01 2.62951255e-01
1.71798036e-01 -3.51881176e-01 1.11278403e+00 -1.11002028e+00
8.85637522e-01 1.15323830e+00 3.30827087e-01 2.12771595e-01
-1.41487074e+00 -5.41129589e-01 5.59727013e-01 1.25293136e-01
-1.31442785e+00 1.65142473e-02 1.07562757e+00 -8.00630391e-01
8.67609024e-01 4.06269640e-01 9.97050405e-01 9.93125796e-01
4.38480556e-01 8.95207465e-01 8.21401119e-01 -3.34779620e-01
2.95850486e-01 4.21909429e-02 3.74147743e-01 1.17059402e-01
2.25406915e-01 -2.92364627e-01 -8.12102973e-01 -5.50879359e-01
3.95614028e-01 4.16466326e-01 -6.83228076e-02 -6.92074239e-01
-1.47779465e+00 8.40514839e-01 -1.49234906e-01 5.63566685e-01
-6.47736669e-01 -1.79233387e-01 4.61015135e-01 3.68525803e-01
1.06770372e+00 1.42424718e-01 -4.17153388e-01 -4.39215243e-01
-1.21947169e+00 4.32675064e-01 7.40637362e-01 7.48491526e-01
4.27485615e-01 -7.70909116e-02 2.74312496e-01 8.16901267e-01
4.41078603e-01 3.38229328e-01 6.92399800e-01 -7.37185717e-01
4.57326949e-01 7.90962815e-01 1.85799181e-01 -1.40866268e+00
-3.81454349e-01 -3.11504722e-01 -8.38654697e-01 -1.40538067e-01
8.53136107e-02 2.62074005e-02 -7.77121186e-01 1.61182559e+00
5.27828276e-01 1.69719324e-01 9.78862047e-02 6.62204266e-01
2.28211164e-01 9.99869466e-01 -4.82941419e-02 -7.30056465e-01
1.21881378e+00 -6.64165437e-01 -9.84285235e-01 1.84907869e-01
3.37513596e-01 -9.16504025e-01 7.24408925e-01 7.84529865e-01
-7.92718410e-01 -6.50383234e-02 -7.21412778e-01 2.58164465e-01
-3.62457633e-01 -1.51839614e-01 1.02035391e+00 5.43704748e-01
-1.10333490e+00 1.18727252e-01 -1.21930945e+00 -5.06525993e-01
-5.00851199e-02 2.04745516e-01 -3.23881119e-01 -1.09324425e-01
-1.13289797e+00 2.57364690e-01 2.28673130e-01 2.72805542e-01
-5.55718482e-01 -8.57837439e-01 -5.13222694e-01 -1.37752876e-01
4.23739552e-01 -5.20721674e-01 1.08107913e+00 -5.27522147e-01
-1.20826733e+00 3.87976795e-01 -4.06741470e-01 -2.60785431e-01
3.18593442e-01 -7.63704032e-02 -5.33485115e-01 4.31103595e-02
1.77571207e-01 3.14781785e-01 9.70165491e-01 -1.16547143e+00
-3.80432129e-01 -4.21399504e-01 1.17818657e-02 6.37492463e-02
-9.28175151e-01 7.48153180e-02 -6.30501747e-01 -9.84935164e-01
7.47362792e-01 -1.38290918e+00 -3.09631050e-01 -3.12272787e-01
-2.88083255e-01 -3.42049748e-01 1.22200906e+00 -6.32700324e-01
1.63629723e+00 -2.19673347e+00 6.58106506e-01 1.88395590e-01
4.67684925e-01 -1.35846525e-01 7.49865398e-02 9.83141780e-01
-3.36953908e-01 2.11955681e-01 -1.66649163e-01 -7.15962052e-01
-3.15110713e-01 3.27211827e-01 -7.90728569e-01 4.44558173e-01
-9.82239172e-02 4.46086615e-01 -8.82284641e-01 -1.43750221e-01
8.83429646e-02 4.77506250e-01 -4.24945325e-01 -2.02967629e-01
-2.30946004e-01 2.09745660e-01 -2.43099675e-01 5.37510335e-01
5.94614685e-01 -3.03985775e-01 4.50705320e-01 -7.83556327e-02
-5.26470542e-01 -1.30576408e-02 -1.25533819e+00 1.46951950e+00
-2.06012711e-01 8.06395292e-01 -3.82722579e-02 -8.00798714e-01
6.21458828e-01 6.75057054e-01 1.23360348e+00 -2.94516414e-01
-2.94862986e-01 -6.81817606e-02 -3.20072472e-02 -3.15126657e-01
1.03430009e+00 -3.38535041e-01 2.79111098e-02 8.91887665e-01
-3.38973626e-02 6.08791187e-02 5.90694606e-01 6.89303815e-01
7.97646105e-01 -2.95842528e-01 5.16710654e-02 -3.26569736e-01
2.83587009e-01 1.38760611e-01 6.04118228e-01 3.28620523e-01
-3.81462947e-02 3.13888907e-01 6.68798268e-01 -4.34425741e-01
-1.15764797e+00 -9.42417681e-01 -2.10213453e-01 1.02320099e+00
-1.09230779e-01 -9.23291504e-01 -3.56700301e-01 -3.46658170e-01
7.02568963e-02 4.53832746e-01 -7.17080355e-01 1.24195702e-01
-3.89676720e-01 -1.17825770e+00 -8.42514858e-02 1.66882202e-01
-1.84848651e-01 -5.95591426e-01 -2.69643039e-01 2.84285307e-01
-2.81815112e-01 -8.77367735e-01 -2.47292161e-01 -1.21413283e-01
-1.32470882e+00 -7.16810524e-01 -7.77592242e-01 -1.17380962e-01
6.75647259e-01 1.03286064e+00 7.22113371e-01 -2.72534996e-01
-1.01058692e-01 6.54202759e-01 -5.36634743e-01 -2.32672870e-01
-6.46119192e-02 2.61353012e-02 5.43959916e-01 4.14401770e-01
2.10840940e-01 -8.44404280e-01 -4.73698616e-01 3.35433304e-01
-1.41527736e+00 2.78685063e-01 1.64730728e-01 7.54714310e-01
4.03436124e-01 6.01214290e-01 3.49946022e-01 -8.73185337e-01
8.47956240e-01 -9.94489014e-01 -4.89217252e-01 5.74881136e-02
-7.90055573e-01 -1.54507086e-01 2.88427204e-01 -7.55670011e-01
-1.02963626e+00 -3.69784802e-01 6.70453727e-01 -8.37519705e-01
3.93287033e-01 1.02079904e+00 2.23166585e-01 3.06308031e-01
2.97478616e-01 2.88585484e-01 -3.41201685e-02 -5.77569544e-01
3.96708935e-01 5.58830678e-01 -3.71812358e-02 -4.24273014e-01
7.17752099e-01 1.02790034e+00 -4.09427509e-02 -1.17888486e+00
-5.34892440e-01 -1.02166069e+00 -6.03592813e-01 -5.13193011e-01
4.76573437e-01 -8.70848179e-01 -4.19859201e-01 5.33260047e-01
-1.02146745e+00 1.53619975e-01 -2.10638896e-01 5.90826929e-01
-2.99988270e-01 5.21343589e-01 -4.95375454e-01 -1.03328669e+00
-1.09519370e-01 -7.59315848e-01 8.57042849e-01 -2.78737903e-01
-3.39197993e-01 -1.09299362e+00 3.79818887e-01 3.36536080e-01
2.79926479e-01 5.68114184e-02 9.55275893e-01 -3.71110380e-01
-4.36804295e-01 -3.71172696e-01 2.18594819e-01 1.66033015e-01
3.02284360e-01 3.97544265e-01 -6.59867525e-01 -6.13209009e-01
3.79201502e-01 2.89273530e-01 8.13381374e-01 2.26172671e-01
5.15844345e-01 -1.64430797e-01 -3.53952050e-01 1.23829097e-01
1.12395799e+00 1.95171744e-01 1.70957133e-01 3.93812686e-01
5.92774630e-01 7.90610015e-01 5.87062359e-01 7.79941380e-01
5.27954161e-01 6.63556516e-01 2.55530685e-01 1.44734278e-01
3.05606633e-01 9.35149416e-02 4.03631806e-01 1.28743207e+00
-2.27375165e-01 -3.57892305e-01 -1.22528780e+00 7.88544655e-01
-2.10026360e+00 -9.39795434e-01 -2.38306329e-01 2.10882330e+00
4.45612997e-01 2.74066441e-02 2.76694983e-01 3.87075216e-01
5.16571879e-01 3.79891187e-01 -3.77347022e-01 -1.98069274e-01
-3.31094377e-02 -4.45840359e-01 9.70517397e-02 1.31736726e-01
-1.08947527e+00 6.96440816e-01 6.32287741e+00 3.96146923e-01
-1.27107978e+00 3.08180183e-01 1.27066448e-01 -3.99659157e-01
-5.26719511e-01 1.41931325e-01 -4.97898072e-01 3.93898934e-01
9.88676429e-01 -4.77242559e-01 2.81792909e-01 7.27901042e-01
5.30952930e-01 -8.26128051e-02 -6.26996100e-01 8.00417721e-01
4.90668342e-02 -1.00842118e+00 6.41946569e-02 3.07796299e-01
9.70672190e-01 5.64404540e-02 3.85918885e-01 6.83608055e-02
-2.79916301e-02 -2.24129200e-01 7.72198081e-01 5.39098322e-01
1.61306977e-01 -6.07120454e-01 4.01975840e-01 4.28241342e-01
-1.19888556e+00 -1.33497000e-01 -3.88695329e-01 -1.32633656e-01
5.19262135e-01 1.07560730e+00 -9.21402812e-01 6.35708034e-01
7.17976272e-01 1.15468216e+00 -2.43825719e-01 9.04591322e-01
1.06300235e-01 9.28634226e-01 -4.60847855e-01 2.62919396e-01
2.30980650e-01 -6.03467107e-01 1.05271721e+00 8.35283637e-01
3.83452833e-01 -1.85559869e-01 1.42843634e-01 4.96165037e-01
3.31172645e-01 2.76360840e-01 -5.41784048e-01 -6.05899036e-01
8.11291263e-02 1.21341372e+00 -1.04031968e+00 -2.04977021e-01
-6.48044527e-01 7.69544005e-01 -7.15729268e-03 4.63423342e-01
-4.67508316e-01 2.72048384e-01 8.55651855e-01 1.33919403e-01
2.75547117e-01 -9.21618462e-01 -1.90113962e-01 -1.63750136e+00
2.73863465e-01 -5.97079098e-01 4.78512555e-01 -3.92125010e-01
-1.32690203e+00 5.94836354e-01 3.59468728e-01 -1.44738936e+00
-2.62156755e-01 -1.60743892e-01 -3.05791855e-01 6.25535905e-01
-1.00332820e+00 -1.18630755e+00 1.46200091e-01 5.98892272e-01
6.49567008e-01 -9.11557227e-02 5.94184577e-01 3.14658314e-01
-7.64554322e-01 -6.56108037e-02 5.60041845e-01 -5.51338434e-01
4.77856845e-01 -1.00654590e+00 3.57904047e-01 9.25628543e-01
2.71494448e-01 1.10863948e+00 8.89127076e-01 -1.02100718e+00
-1.53634584e+00 -9.65945721e-01 9.10208583e-01 -3.01765501e-01
1.30047727e+00 -6.91719115e-01 -9.60176110e-01 6.64121270e-01
-1.74705051e-02 -2.54431397e-01 1.14252067e+00 5.78242302e-01
-4.33520287e-01 5.94184101e-02 -6.49283171e-01 6.91312850e-01
6.06189966e-01 -4.32257622e-01 -4.44677949e-01 5.04524589e-01
9.73058641e-01 9.39721465e-02 -1.06010473e+00 1.09538518e-01
6.71500802e-01 -7.06124663e-01 7.91141808e-01 -7.75062323e-01
1.42923415e-01 -4.03575480e-01 -3.17582756e-01 -1.04693174e+00
-3.03568661e-01 -7.37288415e-01 -5.24181843e-01 1.30490553e+00
3.77106220e-01 -4.19470578e-01 8.98421347e-01 6.29452407e-01
1.94346279e-01 -6.92181468e-01 -9.55055594e-01 -7.52693892e-01
-1.12046190e-01 -7.54989624e-01 2.98968464e-01 1.28652537e+00
2.67577350e-01 1.07841760e-01 -7.68685639e-01 9.62523669e-02
5.68442404e-01 4.20376748e-01 5.38467109e-01 -1.37080836e+00
-1.02124803e-01 -7.41402209e-02 -5.20249009e-01 -7.33559489e-01
-5.57759553e-02 -7.25894570e-01 -3.40004146e-01 -1.06724870e+00
3.57278824e-01 -3.59124511e-01 -2.38198876e-01 2.16050252e-01
3.21822055e-02 4.60855477e-02 4.82424378e-01 8.05340528e-01
-4.96861696e-01 6.67874038e-01 1.00907207e+00 1.04347551e-02
-5.58714449e-01 2.18920380e-01 -4.22881961e-01 4.36819017e-01
6.59863591e-01 -5.27229130e-01 -7.48793662e-01 -1.88308030e-01
6.31013989e-01 1.19395591e-01 -6.10464178e-02 -6.45606518e-01
4.28883642e-01 -4.20904964e-01 -1.26751363e-01 -9.98511255e-01
4.79295760e-01 -9.46858585e-01 6.28238618e-01 2.93977022e-01
1.58895239e-01 4.18350816e-01 2.45276257e-01 9.83817101e-01
-4.92982715e-01 2.45127127e-01 1.63392052e-01 6.67515844e-02
-5.50564587e-01 3.32157314e-01 -8.12102079e-01 -5.29847145e-01
1.13389647e+00 1.19875828e-02 -4.58859745e-03 -4.27138686e-01
-8.67771149e-01 1.73640147e-01 3.80215615e-01 7.02871919e-01
6.02481425e-01 -1.32332933e+00 -4.34964567e-01 1.72523946e-01
7.23378174e-03 -1.71422929e-01 7.23270297e-01 1.05252290e+00
-2.22970787e-02 8.70809793e-01 1.30521998e-01 -6.79904878e-01
-1.22690463e+00 9.81456816e-01 -3.50476295e-01 -4.40545440e-01
-6.72134340e-01 3.87593687e-01 2.28342116e-01 -2.20116422e-01
-1.13054745e-01 -1.33882448e-01 -2.28828952e-01 6.71941102e-01
4.68181998e-01 5.42113781e-01 1.42638445e-01 -8.01315427e-01
-1.57855883e-01 2.82464117e-01 -4.49834466e-01 -3.89328003e-01
1.72844481e+00 -3.94212008e-01 -6.86660826e-01 1.02336717e+00
1.00417054e+00 -2.75607072e-02 -9.25646126e-01 -2.23876819e-01
1.44563898e-01 -4.53566670e-01 1.16063282e-01 -2.63138413e-01
-9.91221905e-01 6.87761545e-01 2.86816776e-01 7.69449830e-01
1.19021201e+00 2.61020605e-02 5.92585385e-01 8.71855542e-02
3.61215979e-01 -7.62785316e-01 1.84844628e-01 3.99560720e-01
8.94960046e-01 -9.31784868e-01 1.40323266e-01 -4.58757371e-01
-5.37683904e-01 1.04294264e+00 1.25115111e-01 1.23264872e-01
1.08213913e+00 1.32627813e-02 -1.55647993e-01 -3.80307645e-01
-1.09875381e+00 2.43354201e-01 2.99591184e-01 1.41547516e-01
3.50787848e-01 2.47935608e-01 -5.53676069e-01 4.51913893e-01
-1.71035975e-01 -1.30796760e-01 6.95527911e-01 1.19623089e+00
-9.32771713e-02 -1.44091392e+00 -5.81480980e-01 3.93470079e-01
-4.54485625e-01 -9.27094445e-02 -3.33151668e-01 6.31848991e-01
-1.61350682e-01 8.63896191e-01 5.43766245e-02 -4.98557776e-01
-1.44369025e-02 1.76200807e-01 -7.14468956e-02 -7.44655073e-01
-2.41987690e-01 4.01271492e-01 -3.57701331e-01 -4.59365666e-01
-6.90003514e-01 -1.33620930e+00 -7.75995255e-01 -2.31021419e-01
-3.52744132e-01 3.76884490e-01 1.02708876e+00 6.62439466e-01
4.20546830e-01 3.45948488e-01 6.65731549e-01 -7.07806945e-01
-1.83232486e-01 -8.56568456e-01 -7.91958928e-01 2.94477999e-01
3.33592415e-01 -8.92532527e-01 -4.37146395e-01 2.61502177e-01]
|
[9.353681564331055, 5.584170341491699]
|
9359a38d-6d5f-419a-8121-20bbaee4b36c
|
do-we-really-need-to-access-the-source-data
|
2002.08546
| null |
https://arxiv.org/abs/2002.08546v6
|
https://arxiv.org/pdf/2002.08546v6.pdf
|
Do We Really Need to Access the Source Data? Source Hypothesis Transfer for Unsupervised Domain Adaptation
|
Unsupervised domain adaptation (UDA) aims to leverage the knowledge learned from a labeled source dataset to solve similar tasks in a new unlabeled domain. Prior UDA methods typically require to access the source data when learning to adapt the model, making them risky and inefficient for decentralized private data. This work tackles a practical setting where only a trained source model is available and investigates how we can effectively utilize such a model without source data to solve UDA problems. We propose a simple yet generic representation learning framework, named \emph{Source HypOthesis Transfer} (SHOT). SHOT freezes the classifier module (hypothesis) of the source model and learns the target-specific feature extraction module by exploiting both information maximization and self-supervised pseudo-labeling to implicitly align representations from the target domains to the source hypothesis. To verify its versatility, we evaluate SHOT in a variety of adaptation cases including closed-set, partial-set, and open-set domain adaptation. Experiments indicate that SHOT yields state-of-the-art results among multiple domain adaptation benchmarks.
|
['Dapeng Hu', 'Jiashi Feng', 'Jian Liang']
|
2020-02-20
| null |
https://proceedings.icml.cc/static/paper_files/icml/2020/194-Paper.pdf
|
https://proceedings.icml.cc/static/paper_files/icml/2020/194-Paper.pdf
|
icml-2020-1
|
['universal-domain-adaptation', 'partial-domain-adaptation']
|
['computer-vision', 'methodology']
|
[ 3.34518403e-01 3.00120831e-01 -7.37210691e-01 -7.30701447e-01
-9.81219411e-01 -8.40953529e-01 7.01305270e-01 -9.88365486e-02
-2.68436581e-01 9.86711860e-01 1.00775279e-01 -1.00877509e-01
1.71242997e-01 -6.00804746e-01 -8.56366038e-01 -7.50644684e-01
2.52602875e-01 8.86918783e-01 -3.06420252e-02 -1.86276406e-01
-3.29326689e-01 1.60351455e-01 -1.18666816e+00 2.48141214e-01
7.98085570e-01 8.18587005e-01 3.44943523e-01 1.20603070e-01
-2.21856266e-01 4.82884616e-01 -5.33085346e-01 -4.53780383e-01
7.53098607e-01 -6.77996457e-01 -1.08159995e+00 2.76232511e-01
1.06916718e-01 -2.41225109e-01 -8.48752931e-02 9.47734654e-01
4.94444221e-01 1.79645777e-01 9.47947264e-01 -1.27479792e+00
-8.17468286e-01 5.70195913e-01 -4.73885030e-01 1.52457759e-01
2.33046219e-01 1.06572762e-01 9.06169415e-01 -8.20786417e-01
1.08483744e+00 8.61655295e-01 4.74600703e-01 9.63392675e-01
-1.37008893e+00 -7.31834710e-01 3.89721483e-01 1.22004203e-01
-1.24659014e+00 -6.47237062e-01 8.52794349e-01 -2.75406748e-01
7.50922740e-01 -2.74178356e-01 1.92723677e-01 1.49344528e+00
-2.96647668e-01 8.49793494e-01 1.20736635e+00 -6.19146287e-01
6.73346341e-01 5.63359261e-01 1.33489177e-01 3.33702594e-01
2.44742990e-01 -4.70226072e-02 -6.80333316e-01 -3.95125985e-01
4.92528856e-01 2.59509701e-02 -1.40463054e-01 -1.16680968e+00
-1.19738245e+00 9.55687821e-01 2.01534122e-01 -4.09714095e-02
-5.22591233e-01 -5.07299066e-01 3.58200133e-01 6.93562150e-01
4.95800942e-01 4.33145881e-01 -1.05389321e+00 1.65862501e-01
-4.59401399e-01 9.63098630e-02 1.01708531e+00 1.34328163e+00
1.15355551e+00 -4.98286299e-02 -9.36399773e-03 9.45629001e-01
2.04217419e-01 7.01176882e-01 8.80035400e-01 -7.77669311e-01
5.49885452e-01 5.45143545e-01 1.68085426e-01 -2.35172540e-01
-1.40227629e-02 -2.63097823e-01 -6.32011831e-01 -7.38865808e-02
6.22825384e-01 -3.77336293e-01 -8.95760357e-01 2.06728673e+00
7.22807109e-01 2.46501520e-01 6.86149359e-01 8.81681442e-01
5.20190060e-01 5.01049101e-01 1.14525110e-01 -2.29274690e-01
1.09234810e+00 -1.03566265e+00 -3.41976196e-01 -4.50615108e-01
6.94042027e-01 -4.99734163e-01 1.15034330e+00 -7.19860476e-03
-6.73479795e-01 -4.07475680e-01 -9.84979987e-01 1.11587226e-01
-5.16865015e-01 -9.54503939e-02 3.73676270e-01 4.94451284e-01
-7.20273256e-01 3.11429292e-01 -5.41149259e-01 -7.85997450e-01
7.59158611e-01 4.38575298e-01 -5.88380516e-01 -4.75379825e-01
-1.20065129e+00 7.41790652e-01 3.96089315e-01 -7.70757377e-01
-1.25980151e+00 -6.27332807e-01 -8.84153962e-01 2.55075432e-02
4.76584911e-01 -9.18778718e-01 1.50236595e+00 -1.38366425e+00
-1.89222026e+00 1.13445914e+00 -2.05749080e-01 -6.45377874e-01
2.44559199e-01 -1.05793834e-01 -3.92148286e-01 -2.25530621e-02
3.26715410e-01 5.31827033e-01 1.20798528e+00 -1.33365142e+00
-7.09993720e-01 -5.34586608e-01 -2.36485694e-02 3.72830302e-01
-7.05226243e-01 -1.72724396e-01 -1.78277910e-01 -6.10756159e-01
-2.19926909e-01 -8.83356452e-01 -2.85487682e-01 -1.86694905e-01
-1.38269901e-01 -2.02003971e-01 9.18326616e-01 -3.59136075e-01
9.65555191e-01 -2.16697574e+00 1.44147143e-01 3.13466400e-01
-8.29051211e-02 3.17592263e-01 -4.58464712e-01 3.71839672e-01
-1.34673297e-01 -3.36745471e-01 -4.72656399e-01 -2.33382300e-01
5.76995872e-02 4.67304975e-01 -6.17663503e-01 3.46432894e-01
8.48940536e-02 6.30110621e-01 -9.78365064e-01 -3.17688435e-01
-1.65423855e-01 1.03147320e-01 -6.74656153e-01 5.76346278e-01
-4.18739706e-01 7.01646864e-01 -7.22345173e-01 6.65879786e-01
5.97188711e-01 -5.97704351e-01 6.26134157e-01 2.48018652e-01
4.56308424e-01 1.90456450e-01 -1.17220855e+00 2.19357586e+00
-5.22684991e-01 6.12018220e-02 5.83353937e-02 -1.42578852e+00
1.08160031e+00 3.30366969e-01 6.00043952e-01 -5.87834954e-01
-9.98714492e-02 2.34032214e-01 -1.56315833e-01 -4.00542498e-01
3.29685062e-02 -1.04185231e-01 -3.59918833e-01 7.71781385e-01
6.80280983e-01 2.70454466e-01 -2.79171795e-01 2.44550750e-01
1.18869686e+00 3.52772206e-01 8.38731349e-01 -2.43404120e-01
3.58271480e-01 1.66396871e-01 7.07041800e-01 8.00377190e-01
-3.66609633e-01 5.41380227e-01 4.11616713e-02 -5.07799387e-01
-1.02960825e+00 -1.20348513e+00 7.54622966e-02 1.57310581e+00
1.48400337e-01 -1.42543197e-01 -7.12023973e-01 -1.42177796e+00
2.09546164e-01 7.54425645e-01 -6.55373693e-01 -4.05876845e-01
-2.83805251e-01 -4.87818062e-01 1.89385638e-01 4.20276880e-01
4.84033167e-01 -9.00470614e-01 -2.99551219e-01 2.57333010e-01
-1.36261076e-01 -1.10650611e+00 -5.76991320e-01 4.32275504e-01
-8.07315052e-01 -1.02314544e+00 -7.87423015e-01 -9.05321121e-01
9.48819995e-01 3.68561238e-01 1.02206910e+00 -6.62505865e-01
2.06717536e-01 6.48898721e-01 -3.86721611e-01 -3.73605162e-01
-5.49952447e-01 5.40050983e-01 3.75046849e-01 3.56196135e-01
8.31523716e-01 -6.26676798e-01 -2.84018099e-01 5.19003391e-01
-7.85494685e-01 -3.37130457e-01 6.71106279e-01 1.04402208e+00
6.55042052e-01 -3.82131130e-01 1.08751285e+00 -1.43042111e+00
4.34860110e-01 -1.04078114e+00 -5.21399260e-01 4.59835023e-01
-8.45427334e-01 1.14931874e-01 9.41975951e-01 -7.56596982e-01
-1.41731203e+00 5.73017001e-01 4.44394588e-01 -6.84724927e-01
-4.81986821e-01 4.01451945e-01 -5.75879753e-01 1.24484308e-01
1.15513897e+00 2.85782963e-01 1.50271371e-01 -7.20678687e-01
6.07953668e-01 9.63160813e-01 5.22216916e-01 -8.62496614e-01
1.02928245e+00 5.31386912e-01 -5.43448150e-01 -4.13430959e-01
-1.15834033e+00 -7.93842494e-01 -9.14386451e-01 2.26944074e-01
4.46377218e-01 -1.27334237e+00 1.75753026e-03 2.09937572e-01
-8.40677142e-01 -5.63071311e-01 -9.46090043e-01 4.59600151e-01
-6.98159337e-01 9.67174917e-02 -5.93886562e-02 -2.79390395e-01
-2.26368070e-01 -7.69936562e-01 8.63635480e-01 3.31019253e-01
-1.80339336e-01 -1.16860878e+00 3.02991748e-01 2.97368884e-01
3.88107687e-01 -5.09838946e-02 7.06537068e-01 -1.45578384e+00
-2.85976112e-01 -1.00055255e-01 8.73192772e-02 3.67366731e-01
4.71206486e-01 -7.50779867e-01 -1.23410594e+00 -5.27101994e-01
5.95593601e-02 -6.52915120e-01 5.95270038e-01 6.37005195e-02
9.85154986e-01 -5.21374226e-01 -4.07108575e-01 7.04704583e-01
1.26323736e+00 -8.66542384e-03 2.43966132e-01 4.57552940e-01
4.72507417e-01 3.96561116e-01 8.39132786e-01 7.29985237e-01
6.08990133e-01 7.23944128e-01 1.13273459e-02 1.49083212e-02
-1.80714101e-01 -5.31970680e-01 6.66866422e-01 7.08385468e-01
2.79403627e-01 -1.28433451e-01 -9.13347781e-01 8.72342348e-01
-1.84136903e+00 -7.33004749e-01 7.20034659e-01 2.30581021e+00
1.21742141e+00 -2.76851594e-01 1.94879979e-01 -5.23331046e-01
8.24651718e-01 -4.45532463e-02 -1.11780334e+00 -1.80226833e-01
-2.25024104e-01 2.79735267e-01 6.05571210e-01 1.32329419e-01
-1.31155849e+00 1.14181638e+00 6.19342136e+00 7.86159635e-01
-9.05916214e-01 4.77449745e-01 3.34813714e-01 1.01833358e-01
-2.83684731e-01 1.88942239e-01 -8.99796844e-01 3.68816763e-01
1.13982296e+00 -4.38061178e-01 4.84777838e-01 1.32299268e+00
-4.06420678e-01 3.12438339e-01 -1.40153444e+00 6.95005894e-01
1.41027912e-01 -1.18394458e+00 1.29165113e-01 1.26155257e-01
1.15011621e+00 2.98807472e-01 8.29590410e-02 6.54064476e-01
9.61137533e-01 -4.07949805e-01 2.33806789e-01 2.11277246e-01
9.76090610e-01 -4.41721261e-01 6.31004810e-01 5.29618084e-01
-8.48342419e-01 -2.33648613e-01 -4.87482369e-01 2.70650797e-02
-1.73448503e-01 1.46900713e-01 -1.37001407e+00 6.68558002e-01
6.59469187e-01 9.64636028e-01 -3.79427880e-01 6.96787894e-01
-2.42675304e-01 8.01938534e-01 -3.80896032e-01 4.11292166e-01
3.88314351e-02 -1.17599837e-01 5.13894260e-01 9.40887451e-01
2.47487724e-01 1.18909828e-01 3.85301054e-01 5.78026354e-01
-5.89517832e-01 2.62693405e-01 -9.98336375e-01 4.44840975e-02
6.58762276e-01 9.84472394e-01 -2.68320143e-01 -5.01367927e-01
-6.43296719e-01 1.18316853e+00 5.43166816e-01 6.46465123e-01
-4.59932595e-01 -7.61554614e-02 8.14638257e-01 1.01223260e-01
5.62304080e-01 2.25674957e-01 -1.16804138e-01 -1.47324228e+00
-1.32312074e-01 -1.03733802e+00 1.01051652e+00 -3.12305748e-01
-1.86997569e+00 2.32837498e-01 1.74247131e-01 -1.53729773e+00
-6.27677739e-01 -2.54190266e-01 -3.61857980e-01 7.72213280e-01
-1.84585524e+00 -1.27827430e+00 -9.31387097e-02 1.25078273e+00
6.46572590e-01 -8.49357009e-01 1.29555941e+00 1.57264382e-01
-4.85091358e-01 9.40508306e-01 6.45083547e-01 1.43640637e-01
1.38727963e+00 -1.05368888e+00 3.16016316e-01 6.70815587e-01
7.96558633e-02 6.12945914e-01 2.85371423e-01 -4.66849446e-01
-1.42471874e+00 -1.27750421e+00 6.69214427e-01 -5.72795331e-01
4.75676686e-01 -3.98475826e-01 -1.00618505e+00 1.02387464e+00
1.86562672e-01 4.07540500e-01 1.13714123e+00 1.19376294e-01
-7.65301228e-01 -1.99791864e-01 -1.52481163e+00 2.37198740e-01
9.46392477e-01 -4.61267740e-01 -7.45437860e-01 3.86888117e-01
6.97517693e-01 -2.31121734e-01 -9.12712216e-01 1.54638171e-01
1.71152994e-01 -3.97448719e-01 9.22038853e-01 -1.15302849e+00
1.95165768e-01 -2.90040523e-01 -3.53018820e-01 -1.53625429e+00
-2.47625887e-01 -5.77867627e-01 -4.30558115e-01 1.34982800e+00
4.12075251e-01 -1.00565720e+00 7.75634170e-01 6.87820435e-01
4.73200046e-02 -2.26288348e-01 -1.01969254e+00 -1.08129489e+00
2.23022282e-01 1.13717847e-01 8.99864912e-01 1.60861957e+00
7.20763355e-02 5.61773896e-01 -2.48351872e-01 2.81519294e-01
7.72530198e-01 4.26822186e-01 1.24950504e+00 -1.25717247e+00
-4.14223105e-01 7.99327865e-02 -7.10040629e-02 -1.02712584e+00
5.90165317e-01 -1.22990906e+00 -1.16551615e-01 -1.07235253e+00
1.75689638e-01 -6.19447231e-01 -6.05978966e-01 9.06658947e-01
-3.29598039e-02 -1.45609751e-01 6.28115982e-02 6.98271632e-01
-8.27703893e-01 6.28565133e-01 9.32485938e-01 -2.05548882e-01
-3.56770724e-01 2.34780267e-01 -1.01544523e+00 6.75228119e-01
8.45086992e-01 -7.25714326e-01 -6.69168711e-01 -4.74194854e-01
-3.94217908e-01 -2.14882731e-01 4.04262915e-02 -7.10515440e-01
2.47829750e-01 -3.74692649e-01 3.72174382e-01 1.20671511e-01
8.89746845e-02 -1.05824780e+00 -1.83824301e-01 9.23109949e-02
-4.94771540e-01 -5.27070403e-01 2.15616319e-02 8.53265703e-01
-2.06258744e-01 -2.04326570e-01 1.00152886e+00 -9.90689024e-02
-1.10774088e+00 3.72549832e-01 7.39722028e-02 4.37937081e-01
1.34169638e+00 -5.58524430e-02 -2.35360175e-01 -4.09129888e-01
-7.81104267e-01 2.55279452e-01 6.88721836e-01 3.50257397e-01
3.62063140e-01 -1.34560096e+00 -7.29176462e-01 4.61240321e-01
6.11973166e-01 2.02657580e-01 5.62065914e-02 3.70260745e-01
1.97402701e-01 8.99660885e-02 -4.00636077e-01 -5.59390008e-01
-7.18528688e-01 8.80150437e-01 1.81876823e-01 -4.44868445e-01
-5.32332778e-01 8.01080227e-01 3.83563608e-01 -9.35907066e-01
1.37418419e-01 2.36163214e-01 -6.24656118e-02 -2.33574770e-02
3.59408230e-01 3.27087454e-02 -1.82256058e-01 -5.24291754e-01
-2.95413285e-01 2.92829156e-01 -4.86378700e-01 -3.86448391e-02
1.33089328e+00 -4.43086475e-01 3.94393861e-01 3.21437240e-01
1.13750255e+00 -2.53814191e-01 -1.60388088e+00 -1.10857344e+00
1.96480192e-02 -4.08418626e-01 -4.81314152e-01 -9.74179506e-01
-9.05491769e-01 7.14139581e-01 6.40485585e-01 -2.72528678e-01
1.11403883e+00 2.12473750e-01 7.49884009e-01 7.78201938e-01
6.64783955e-01 -1.39999306e+00 1.01817921e-01 4.60143596e-01
5.09223878e-01 -1.58837509e+00 -9.51828808e-02 -1.20662652e-01
-1.10913825e+00 7.35109150e-01 8.71296227e-01 1.86218936e-02
7.44570315e-01 -1.22939810e-01 2.43370220e-01 1.35577992e-01
-7.71293700e-01 -2.12537885e-01 -2.50944495e-02 1.10904217e+00
-1.92412898e-01 7.18924925e-02 3.15153360e-01 1.08917665e+00
7.85761923e-02 6.41853139e-02 3.10330868e-01 1.10813367e+00
-1.59306496e-01 -1.50688028e+00 -2.60276318e-01 3.34605068e-01
-2.60521054e-01 1.01727478e-01 -5.52692354e-01 7.19682395e-01
-1.09858297e-01 6.19543016e-01 -2.19598919e-01 -2.10019574e-01
3.27265590e-01 5.29541671e-01 2.17897952e-01 -1.03573215e+00
-4.05652583e-01 -9.35169607e-02 -2.08840832e-01 -4.96235639e-01
-6.25494063e-01 -8.74948800e-01 -1.18702304e+00 2.16143221e-01
-6.86457381e-02 2.46218711e-01 5.06433189e-01 9.07317340e-01
8.21200430e-01 1.64899956e-02 9.04654562e-01 -4.25707459e-01
-1.02690220e+00 -9.23183501e-01 -6.40142798e-01 6.61801815e-01
2.67131388e-01 -7.36685216e-01 -2.16720268e-01 6.56394064e-01]
|
[10.380387306213379, 3.129263162612915]
|
ede4a625-25d4-4c10-b178-0b246102f381
|
when-bert-fails-the-limits-of-ehr
|
2208.10245
| null |
https://arxiv.org/abs/2208.10245v1
|
https://arxiv.org/pdf/2208.10245v1.pdf
|
When BERT Fails -- The Limits of EHR Classification
|
Transformers are powerful text representation learners, useful for all kinds of clinical decision support tasks. Although they outperform baselines on readmission prediction, they are not infallible. Here, we look into one such failure case, and report patterns that lead to inferior predictive performance.
|
['Carsten Eickhoff', 'Augusto Garcia-Agundez']
|
2022-07-26
| null | null | null | null |
['readmission-prediction']
|
['medical']
|
[ 3.00655544e-01 4.02780056e-01 -8.50074232e-01 -3.72922748e-01
-9.67740178e-01 -1.78652167e-01 5.08359015e-01 9.39563453e-01
-5.04576087e-01 7.89602816e-01 9.50109899e-01 -1.36635864e+00
-6.07857704e-01 -4.73574311e-01 -1.77471682e-01 -3.19171250e-01
-2.79513687e-01 7.97005236e-01 -1.69170424e-01 -2.32695490e-01
2.46529147e-01 2.53178746e-01 -1.15569472e+00 8.90408397e-01
7.94438541e-01 4.28914219e-01 -2.88696349e-01 7.16432273e-01
-6.56140223e-02 1.50589395e+00 -5.47337890e-01 -4.37635988e-01
-5.67739844e-01 -3.71860355e-01 -1.00227416e+00 -6.05721951e-01
1.67851523e-01 -2.14581355e-01 -3.96919042e-01 4.07156974e-01
8.77774358e-01 -1.09233320e-01 1.27075577e+00 -5.98256230e-01
-4.04556721e-01 6.54529154e-01 -3.71160299e-01 9.12785411e-01
6.06546700e-01 -2.27617040e-01 1.05150759e+00 -6.29888713e-01
6.22329772e-01 1.00776029e+00 1.01509404e+00 6.85116410e-01
-1.13007438e+00 -4.99949187e-01 -3.19932066e-02 3.50385532e-02
-5.88158965e-01 -7.58187652e-01 1.09353915e-01 -5.38052857e-01
1.31024230e+00 5.30474484e-01 2.58601874e-01 1.59279168e+00
7.36208498e-01 7.40150511e-01 9.03639257e-01 -4.87948209e-01
-1.05994917e-01 9.11184680e-03 7.50639260e-01 5.53104758e-01
7.09715307e-01 1.45569280e-01 -5.18350303e-01 -5.88787079e-01
3.67300481e-01 5.74977279e-01 -4.80235606e-01 2.50150740e-01
-1.31429946e+00 9.54930604e-01 4.07171071e-01 4.97609138e-01
-2.84092665e-01 -1.73138812e-01 6.60275578e-01 6.53981328e-01
5.13505638e-01 6.80156410e-01 -3.92310411e-01 -3.65823001e-01
-7.67437100e-01 1.42261237e-01 7.73783267e-01 5.09752393e-01
-3.40909004e-01 -2.03698605e-01 -5.73336065e-01 9.84356880e-01
-2.12414697e-01 4.86433022e-02 7.99580276e-01 3.09630539e-02
5.94439030e-01 3.70184213e-01 -8.75593349e-02 -7.54208922e-01
-1.26225364e+00 -5.09575665e-01 -9.69431460e-01 -2.52329409e-01
2.21511588e-01 -1.34156436e-01 -7.26284683e-01 1.24998951e+00
-4.41520214e-01 -4.78040613e-02 1.84791058e-01 4.08794940e-01
1.23258758e+00 2.72368848e-01 6.16695702e-01 -4.77655500e-01
1.43656802e+00 -5.46016097e-01 -7.34818935e-01 -5.34305334e-01
1.34672225e+00 -5.46317756e-01 8.70738983e-01 5.29080093e-01
-1.08599091e+00 1.99989472e-02 -9.67122078e-01 6.09931983e-02
-1.87903062e-01 4.72469861e-03 8.30098629e-01 4.95274603e-01
-7.93611884e-01 6.15321159e-01 -8.06289792e-01 -4.66266215e-01
6.21044159e-01 1.69717386e-01 -3.53543758e-01 -2.30299354e-01
-1.01605606e+00 1.25270975e+00 3.50605756e-01 -3.02796096e-01
-5.63482583e-01 -7.16049731e-01 -6.83677793e-01 2.70501943e-03
-1.05813757e-01 -8.51881981e-01 1.41414678e+00 -3.25844973e-01
-7.27804720e-01 1.11206949e+00 7.09501980e-03 -6.89553201e-01
4.20908153e-01 -1.74568251e-01 -4.64655071e-01 -7.92852789e-02
-2.26248816e-01 2.16273330e-02 5.14369130e-01 -7.41407335e-01
-5.87443054e-01 -2.04491630e-01 -3.50324005e-01 9.28422585e-02
-2.46203795e-01 3.29204828e-01 4.30269480e-01 -9.08331454e-01
-1.08749829e-01 -5.06923556e-01 -4.86111820e-01 -4.84540671e-01
-4.37833577e-01 -5.02729237e-01 1.64026007e-01 -3.90080005e-01
1.45607686e+00 -2.14691806e+00 -3.02177638e-01 -2.25249678e-01
4.91508693e-01 2.72672921e-01 9.21386480e-02 5.21801412e-01
-3.35960656e-01 4.54926193e-01 -1.54083287e-02 -1.73900232e-01
-2.10723743e-01 1.12332329e-02 -5.81977844e-01 4.38855708e-01
2.54405081e-01 1.05794990e+00 -9.75089073e-01 -4.43422318e-01
1.53876796e-01 2.00028554e-01 -7.62012422e-01 9.40841809e-02
1.25680923e-01 4.77905534e-02 -4.88924831e-01 4.71690655e-01
1.82978120e-02 -6.19479716e-01 2.70402044e-01 3.19401801e-01
3.27846527e-01 1.06039608e+00 -1.94756538e-01 1.26404190e+00
-3.40287685e-01 5.49253106e-01 -4.40751940e-01 -1.12973738e+00
6.63353086e-01 5.34990609e-01 4.10311520e-01 -4.95134115e-01
4.82189767e-02 2.59915888e-01 4.59913760e-01 -7.67603874e-01
2.19575748e-01 -8.70168805e-01 -6.62323460e-02 6.55645013e-01
-1.61379099e-01 1.96146607e-01 -2.98263103e-01 1.55885801e-01
1.56033814e+00 -5.66382051e-01 8.00599158e-01 -3.48496646e-01
-9.15441886e-02 -3.97842284e-03 3.87400180e-01 1.05927241e+00
2.10087206e-02 9.33554769e-01 6.91466033e-01 -8.79434228e-01
-4.62162316e-01 -9.10774231e-01 -6.46136761e-01 1.11654723e+00
-5.11279225e-01 -7.88375556e-01 3.64480689e-02 -8.25707018e-01
6.91126287e-02 8.77024770e-01 -7.43742943e-01 -4.83200490e-01
-2.80023634e-01 -1.19519591e+00 5.95563591e-01 7.49324322e-01
-3.78127784e-01 -1.30709267e+00 -8.21611285e-01 3.25799882e-01
-5.98498955e-02 -7.27219880e-01 -5.16803004e-02 7.14964926e-01
-1.20133662e+00 -1.28075135e+00 -6.89807296e-01 -7.58620918e-01
3.72004926e-01 -5.44660427e-02 1.59095883e+00 4.28387821e-01
-6.54776931e-01 4.67912525e-01 -3.71096820e-01 -7.96382904e-01
-7.21702278e-01 4.21993583e-01 7.39895329e-02 -7.03719974e-01
6.33879840e-01 -3.67758125e-01 -7.93513417e-01 -1.22560248e-01
-6.59383237e-01 -2.16728926e-01 6.99075818e-01 9.95887220e-01
2.61068314e-01 -5.15594900e-01 1.08374584e+00 -1.44192755e+00
1.00095403e+00 -9.60594535e-01 4.29561585e-01 1.24274798e-01
-9.57482219e-01 -2.31632337e-01 5.52199841e-01 -1.41990378e-01
-5.35681844e-01 -6.19297326e-01 -3.89616013e-01 1.27461880e-01
-3.70567620e-01 9.15381312e-01 7.41167188e-01 5.57123959e-01
1.20452666e+00 6.33581430e-02 -1.18546421e-02 -6.32584631e-01
-3.28650206e-01 7.96304405e-01 2.58215457e-01 -2.26552233e-01
1.20342374e-01 4.99742217e-02 -2.33132705e-01 -7.23667264e-01
-9.26281393e-01 -4.22910452e-01 -2.40193874e-01 2.78693825e-01
7.59865463e-01 -7.55346596e-01 -4.52055305e-01 -2.34436631e-01
-1.09665716e+00 -1.43555909e-01 -3.31753105e-01 6.05278015e-01
-4.25819486e-01 -1.18455216e-01 -8.28980565e-01 -4.91067469e-01
-4.34927166e-01 -8.82010996e-01 6.95590138e-01 -2.72087783e-01
-6.98377907e-01 -1.13943267e+00 2.10590795e-01 7.54352808e-02
4.89959210e-01 5.92989065e-02 1.51317513e+00 -1.29826820e+00
3.34818393e-01 -6.11751199e-01 -1.46227792e-01 -5.49585596e-02
3.09581965e-01 -2.20968127e-01 -8.65189612e-01 -3.87634218e-01
-2.24361777e-01 -6.37328267e-01 1.38131988e+00 5.97480178e-01
1.50879455e+00 -4.39140648e-01 -8.43649745e-01 4.42729682e-01
8.03495705e-01 2.74036437e-01 5.53347051e-01 2.88955927e-01
1.96214810e-01 4.68354553e-01 -1.00810587e-01 4.56730694e-01
5.27535856e-01 3.03109705e-01 -1.12018157e-02 -3.10464203e-01
-7.17090592e-02 -2.74871290e-01 5.55107705e-02 8.20002496e-01
-1.54434428e-01 -3.95804673e-01 -1.38384342e+00 3.55084360e-01
-1.72714603e+00 -8.90228808e-01 -4.28958423e-02 2.14862490e+00
7.61544764e-01 4.30206865e-01 2.26299480e-01 1.48647994e-01
8.38866308e-02 2.28855371e-01 -1.93328485e-01 -8.08514714e-01
6.14743354e-03 3.38232726e-01 8.35210159e-02 2.87914991e-01
-9.74689484e-01 3.29464257e-01 8.57476330e+00 2.04287305e-01
-9.89413917e-01 1.78219229e-01 9.30869102e-01 -1.68144390e-01
-4.79261667e-01 -5.05463243e-01 -3.76864851e-01 2.48029441e-01
1.26093292e+00 -1.35337576e-01 -3.04783523e-01 5.87201953e-01
2.73925122e-02 8.32523927e-02 -1.48879981e+00 1.12068021e+00
1.70584604e-01 -1.50619864e+00 2.43407771e-01 -1.62894413e-01
3.11276436e-01 2.77077883e-01 1.84874199e-02 5.49489141e-01
4.79054481e-01 -1.76030457e+00 -1.09443916e-02 4.90951806e-01
9.52596843e-01 -3.87438923e-01 1.03993380e+00 2.19592407e-01
-1.84113368e-01 -2.82039016e-01 -4.84921455e-01 -1.68895379e-01
-1.35426864e-01 4.53907549e-01 -9.02795732e-01 3.85229439e-01
5.23879945e-01 9.99299705e-01 -7.37548411e-01 1.13998449e+00
1.07783072e-01 1.10250008e+00 4.32199650e-02 -9.83444303e-02
3.37047093e-02 5.10239422e-01 2.07727715e-01 1.66063118e+00
2.55281538e-01 6.75099075e-01 5.71257398e-02 1.92294046e-01
-2.34565616e-01 3.86056721e-01 -1.18512440e+00 -1.37831852e-01
1.01222053e-01 5.95323026e-01 -6.57672703e-01 -2.59474754e-01
-6.19999528e-01 5.47291636e-01 4.64481860e-01 2.71396488e-01
-2.88994640e-01 -2.19542384e-01 4.73990351e-01 3.94387156e-01
-1.62875742e-01 3.94219905e-01 -4.72057015e-01 -1.18504524e+00
-2.05863684e-01 -1.14763558e+00 1.20833480e+00 -4.69943196e-01
-1.48777938e+00 4.99688566e-01 -2.04994649e-01 -1.33818734e+00
-5.25973201e-01 -7.60980606e-01 -6.59848154e-01 4.29660231e-01
-1.53309429e+00 -6.14531398e-01 -1.53539022e-02 3.06844980e-01
3.56210679e-01 -3.72221917e-01 1.36920512e+00 2.84200311e-01
-5.04123330e-01 8.89801323e-01 3.26659009e-02 7.36604556e-02
7.91381359e-01 -1.03335917e+00 1.07947446e-01 3.07666771e-02
2.56054550e-01 6.98434412e-01 6.40366375e-01 -3.58585566e-01
-1.07302582e+00 -8.43740880e-01 1.20596337e+00 -8.06544602e-01
5.01307547e-01 9.06334147e-02 -9.47227359e-01 9.93254244e-01
-2.38131080e-02 -1.68478593e-01 1.19582915e+00 7.74179161e-01
-5.52806795e-01 1.86572060e-01 -7.23192811e-01 4.85349804e-01
9.39486384e-01 -4.22467977e-01 -1.02630913e+00 7.28098869e-01
4.52104479e-01 -2.55462557e-01 -9.49036300e-01 4.52776849e-01
4.91206914e-01 -8.18700790e-01 1.19272745e+00 -1.67411840e+00
1.03396308e+00 4.01771039e-01 1.02262512e-01 -1.32408071e+00
-5.24647415e-01 -5.26521623e-01 6.91848025e-02 5.50382793e-01
9.47293758e-01 -8.30596924e-01 5.01557171e-01 3.14930499e-01
-3.29784125e-01 -9.81088936e-01 -1.03625166e+00 -5.52911758e-01
6.74596786e-01 -4.92227256e-01 2.90092140e-01 1.25639737e+00
1.08572567e+00 6.61246419e-01 -8.90794545e-02 -4.24913734e-01
3.56010169e-01 5.58191128e-02 1.50498152e-01 -1.40882373e+00
-2.31843039e-01 -1.01552391e+00 -5.62304437e-01 -6.19689643e-01
2.98536979e-02 -1.34968162e+00 -1.09506235e-01 -2.12236166e+00
5.84303737e-01 -6.94227874e-01 -9.11785901e-01 7.46340632e-01
-5.95142663e-01 -8.18744488e-03 -4.49679829e-02 1.25945017e-01
-6.62409425e-01 8.22474733e-02 8.12533081e-01 -1.99534178e-01
-3.30182686e-02 4.03537095e-01 -1.13197100e+00 6.26454949e-01
9.23916042e-01 -6.06565058e-01 -5.61872542e-01 -4.84249055e-01
5.26636899e-01 5.65888286e-01 1.90338686e-01 -7.97388911e-01
-8.40728357e-02 -1.67413086e-01 4.54764128e-01 -3.37262839e-01
-1.86118573e-01 -2.12246403e-01 -3.17362934e-01 6.70831621e-01
-7.75524676e-01 3.97632271e-01 3.69374335e-01 5.74426770e-01
-1.46800324e-01 -2.34666064e-01 4.77369517e-01 -3.85276601e-02
4.48728390e-02 3.59676093e-01 -7.99708903e-01 5.12861729e-01
5.77554703e-01 9.15435925e-02 -7.04803765e-01 -3.45242947e-01
-6.97856605e-01 -1.69002097e-02 6.11890629e-02 5.13114333e-01
7.01926351e-01 -1.01175320e+00 -1.08560717e+00 2.25643720e-02
4.39583093e-01 -2.15803117e-01 -9.81114283e-02 1.07637882e+00
-3.41674656e-01 6.80883884e-01 1.39849380e-01 -3.67862195e-01
-1.11409497e+00 6.49439514e-01 4.75989103e-01 -4.33178425e-01
-1.30355704e+00 7.84362257e-01 2.17441872e-01 -2.50518769e-01
4.68829334e-01 -7.88066030e-01 -5.03389478e-01 2.91113704e-01
1.11406672e+00 3.76828909e-02 4.72468108e-01 5.77180944e-02
-6.21879995e-01 7.08980486e-02 -4.61523056e-01 3.02853376e-01
1.71536601e+00 4.32652265e-01 1.57668114e-01 7.19425261e-01
9.21186864e-01 -2.66251564e-01 -2.63467729e-01 -1.51947260e-01
3.58219087e-01 -1.13530718e-01 5.50592467e-02 -1.12892270e+00
-5.56556404e-01 9.66553807e-01 4.45703566e-01 3.10113341e-01
7.83997178e-01 1.38415590e-01 4.17308986e-01 5.79483807e-01
-1.93719994e-02 -4.87995118e-01 -1.31652147e-01 5.25051236e-01
8.73514593e-01 -1.40732515e+00 3.60881060e-01 4.76065837e-02
-5.26080191e-01 1.15987194e+00 2.19591424e-01 7.20736012e-02
7.93219566e-01 4.57959622e-01 1.65849224e-01 -5.24250805e-01
-1.35642731e+00 7.79744014e-02 3.81925106e-01 7.20120788e-01
1.26448607e+00 2.45832607e-01 -4.64394838e-01 8.63851130e-01
-1.05580613e-01 2.60298103e-01 5.37118495e-01 9.09772456e-01
-4.35347080e-01 -6.89359963e-01 -1.03616677e-01 1.50138032e+00
-9.10963655e-01 -3.46323490e-01 -3.73950869e-01 6.27953172e-01
-5.67708552e-01 8.03437650e-01 9.79967937e-02 -4.05047446e-01
5.71263015e-01 6.54848993e-01 3.15032274e-01 -7.71962464e-01
-8.53236496e-01 -4.06887025e-01 5.70438981e-01 -2.93266207e-01
8.55036750e-02 -8.29012096e-01 -1.07251322e+00 -4.38694805e-01
4.01044749e-02 2.60053247e-01 -1.04188271e-01 7.71232963e-01
3.20546657e-01 7.03207195e-01 2.77714938e-01 -4.37874421e-02
-6.31628096e-01 -1.23345721e+00 -4.60290879e-01 4.18949902e-01
8.13384473e-01 -4.43376303e-01 -3.97119910e-01 -1.58178285e-01]
|
[8.046751022338867, 6.635904788970947]
|
173e1285-4003-4dc1-8738-842c2a96c1bd
|
cd2-combined-distances-of-contrast
|
1911.07995
| null |
https://arxiv.org/abs/1911.07995v2
|
https://arxiv.org/pdf/1911.07995v2.pdf
|
CD2 : Combined Distances of Contrast Distributions for the Assessment of Perceptual Quality of Image Processing
|
The quality of visual input is very important for both human and machine perception. Consequently many processing techniques exist that deal with different distortions. Usually image processing is applied freely and lacks redundancy regarding safety. We propose a novel image comparison method called the Combined Distances of Contrast Distributions (CD2) to protect against errors that arise during processing. Based on the distribution of image contrasts a new reduced-reference image quality assessment (IQA) method is introduced. By combining various distance functions excellent performance on IQA benchmarks is achieved with only a small data and computation overhead.
|
['Sascha Xu', 'Jan Bauer', 'Benjamin Axmann']
|
2019-11-18
| null | null | null | null |
['small-data']
|
['computer-vision']
|
[ 2.98438281e-01 -6.07881188e-01 2.71307886e-01 -6.09838784e-01
-4.68562543e-01 -4.92457926e-01 5.89487553e-01 6.48805082e-01
-6.09931827e-01 5.33964157e-01 -1.82527021e-01 -1.26806468e-01
-1.11782206e-02 -8.09122503e-01 -3.35172862e-01 -4.93071169e-01
6.01478852e-02 -4.01198179e-01 6.70548379e-01 -2.08530545e-01
8.81856918e-01 5.88812530e-01 -1.84979689e+00 1.67073324e-01
1.10269916e+00 1.12882733e+00 1.61339656e-01 8.24248493e-01
1.21637709e-01 4.22300875e-01 -8.75219285e-01 -3.31265420e-01
7.86252320e-01 -6.32911801e-01 -7.18219757e-01 1.80735826e-01
3.80938858e-01 -2.73505628e-01 4.96033989e-02 1.49512887e+00
7.76450217e-01 3.45783412e-01 6.51789129e-01 -1.30638182e+00
-5.22663772e-01 -5.43089584e-02 -6.43249631e-01 5.74270904e-01
4.09953266e-01 4.25155222e-01 2.44961783e-01 -5.10378838e-01
3.27023059e-01 1.00656104e+00 3.11570704e-01 2.18231812e-01
-1.38714707e+00 -2.70051509e-01 -2.73804009e-01 5.77346683e-01
-1.50828421e+00 -2.72541344e-01 5.73012292e-01 -2.83757031e-01
8.69108617e-01 5.29569805e-01 2.55275637e-01 3.97291869e-01
5.80465734e-01 6.83650002e-02 1.67294061e+00 -5.97535431e-01
1.94166347e-01 2.92331666e-01 2.15974912e-01 4.39861596e-01
3.56729656e-01 1.00991063e-01 -1.38570234e-01 9.70803127e-02
4.48367178e-01 -3.86076182e-01 -2.92041779e-01 -1.71253175e-01
-1.00848556e+00 3.24845940e-01 3.53024662e-01 3.24233830e-01
-6.90016598e-02 -1.72766134e-01 6.19767785e-01 6.29264951e-01
-9.43838712e-03 4.18086380e-01 4.92050536e-02 1.78211287e-01
-6.24398649e-01 2.03176573e-01 2.51841635e-01 6.65663123e-01
6.02499783e-01 1.51866274e-02 -3.52152824e-01 7.78871000e-01
-3.12915556e-02 6.70529723e-01 4.95934427e-01 -8.53992164e-01
2.29531214e-01 6.10336185e-01 2.09534138e-01 -1.65059435e+00
-4.10781592e-01 -1.45884395e-01 -1.05386233e+00 1.23375928e+00
4.46034998e-01 3.79277647e-01 -8.61258686e-01 1.46539283e+00
1.73336908e-01 -3.50029945e-01 6.72101229e-02 9.41933930e-01
4.78439748e-01 7.63493836e-01 6.94689602e-02 -4.11269337e-01
1.40864551e+00 -5.29311776e-01 -8.93260598e-01 1.96625471e-01
9.40875560e-02 -1.15746307e+00 1.09171367e+00 9.09644186e-01
-1.45070529e+00 -1.08354771e+00 -1.53177476e+00 -2.55346298e-01
-5.00668824e-01 -3.20702374e-01 2.58935206e-02 1.07553017e+00
-1.17213988e+00 7.63104320e-01 -3.15587252e-01 -7.12093413e-02
2.17956260e-01 4.34844553e-01 -4.67786461e-01 1.29202113e-01
-9.22705233e-01 1.27697754e+00 4.02496248e-01 -1.25886962e-01
-7.18996048e-01 -4.56983328e-01 -6.39019549e-01 -2.12239027e-01
9.12399963e-02 -4.26896125e-01 8.42665970e-01 -1.20594883e+00
-1.52290142e+00 9.46933627e-01 9.81781855e-02 -5.46760678e-01
6.31804466e-01 -7.88044184e-02 -6.80794716e-01 1.71539918e-01
-2.62269169e-01 4.92525280e-01 1.07470810e+00 -1.21561980e+00
-5.33685029e-01 -4.07282531e-01 -2.28207797e-01 2.69245893e-01
9.50103402e-02 3.09814513e-01 -1.80614918e-01 -5.91150701e-01
-1.44605845e-01 -4.16607708e-01 -1.59038141e-01 2.76290834e-01
-2.14029804e-01 8.51828903e-02 6.03256345e-01 -6.30761147e-01
1.47753465e+00 -2.09920049e+00 -2.29741693e-01 5.29148817e-01
-8.66132881e-03 6.85451984e-01 -1.96616620e-01 -1.37653307e-03
-1.44042626e-01 1.43991947e-01 -2.40393400e-01 1.42500088e-01
-2.51746148e-01 -1.84721470e-01 3.86801958e-02 5.76663971e-01
1.47028968e-01 2.65292704e-01 -6.67221844e-01 -5.63211739e-01
4.30559307e-01 3.56146872e-01 -3.51411194e-01 5.43021679e-01
2.98838288e-01 3.39936107e-01 1.15550108e-01 3.38961780e-01
1.25131309e+00 2.15934858e-01 -1.44509539e-01 -5.78286350e-01
-3.22626680e-01 -1.47378579e-01 -1.34479547e+00 1.40595901e+00
-2.91137159e-01 5.14976978e-01 -1.00077830e-01 -7.03486443e-01
1.10099161e+00 2.33303949e-01 5.45165800e-02 -1.13715994e+00
5.14365792e-01 9.57560614e-02 2.03735277e-01 -5.19741833e-01
5.38291156e-01 7.12703094e-02 2.71103829e-01 2.24763498e-01
-1.80787578e-01 -3.43714029e-01 2.93809921e-01 -1.59998517e-02
9.08471107e-01 -1.76177099e-01 7.90067792e-01 -5.62222958e-01
1.00212228e+00 -1.93172723e-01 4.65278685e-01 5.73891819e-01
-8.12056601e-01 7.64898837e-01 3.47236902e-01 -4.63929743e-01
-1.39086604e+00 -1.21199489e+00 -2.75556356e-01 4.74475712e-01
5.60359359e-01 -2.08354548e-01 -7.96156466e-01 -4.14716929e-01
-3.38439137e-01 3.49763244e-01 -3.46707821e-01 -1.89316198e-01
-4.19642836e-01 -5.44164002e-01 3.97718012e-01 2.07582086e-01
9.08009887e-01 -8.53159428e-01 -9.83205259e-01 -1.44378737e-01
9.24082845e-02 -8.99098814e-01 -6.33371353e-01 -2.28904456e-01
-7.37566829e-01 -1.07836950e+00 -4.76111263e-01 -5.63414514e-01
6.49672925e-01 4.10419673e-01 1.13742435e+00 3.53753775e-01
-5.25909483e-01 4.81919572e-02 -3.27129066e-01 -3.77513200e-01
-6.28398836e-01 -6.68333471e-01 7.28516653e-02 5.54439202e-02
1.33428946e-01 -5.13349175e-01 -8.47737432e-01 5.14779925e-01
-1.04448009e+00 -1.71180695e-01 4.81274575e-01 6.02119267e-01
5.83745837e-01 5.46646059e-01 4.01334703e-01 -6.00456774e-01
8.17692459e-01 -2.99328770e-02 -9.70120013e-01 2.76186764e-01
-9.71293569e-01 2.39212006e-01 7.91061401e-01 -1.71039253e-01
-1.12140155e+00 -2.49580994e-01 3.69571359e-03 -1.48184285e-01
-3.97941500e-01 -2.07457885e-01 -4.56565589e-01 -4.87622738e-01
8.81893039e-01 -1.30053060e-02 2.11965218e-01 -2.98910141e-01
1.54925004e-01 5.97727597e-01 9.14520800e-01 -2.27590173e-01
6.58489287e-01 3.10198307e-01 3.82257015e-01 -7.57675052e-01
-1.44905299e-01 -3.06509644e-01 -4.01974946e-01 -3.20650578e-01
9.02434886e-01 -4.25357133e-01 -8.49572539e-01 8.67610514e-01
-1.16652930e+00 1.88544154e-01 4.50739153e-02 5.37943304e-01
-4.53373015e-01 7.37085700e-01 -2.49248698e-01 -7.96151698e-01
-4.49503452e-01 -1.33612084e+00 3.49050581e-01 3.12144488e-01
1.52801722e-01 -5.37470222e-01 1.35034295e-02 7.76840448e-02
5.77581227e-01 5.70760071e-01 8.00057948e-01 -1.09438099e-01
-4.67339844e-01 6.96529597e-02 -5.25832593e-01 8.48360479e-01
3.18426907e-01 1.21560387e-01 -8.29799950e-01 -2.24597111e-01
2.31625050e-01 -8.06366950e-02 4.83346909e-01 3.15685749e-01
1.34226930e+00 -1.72314137e-01 2.04541534e-01 5.25330424e-01
1.89862835e+00 6.24019682e-01 1.25524843e+00 4.48697239e-01
2.44905233e-01 6.89935863e-01 7.59002805e-01 2.93712467e-01
-2.87905961e-01 7.19985127e-01 4.05553848e-01 -1.23520240e-01
-7.63545707e-02 3.21068525e-01 1.28870398e-01 6.79656684e-01
-2.47955829e-01 -3.73587787e-01 -7.04094410e-01 3.45689833e-01
-1.37423980e+00 -1.04196858e+00 -3.64689976e-01 2.65569043e+00
8.63932073e-01 4.75103050e-01 1.49126217e-01 7.25402534e-01
1.00859416e+00 -1.21184364e-02 -3.60001534e-01 -1.09315026e+00
-2.31593445e-01 2.07709044e-01 6.81651413e-01 5.02257526e-01
-1.00541770e+00 1.57729968e-01 7.45381403e+00 8.82593989e-01
-8.49683881e-01 -1.93367153e-03 7.17906415e-01 3.04009587e-01
-2.47188676e-02 -2.04519123e-01 -3.39103431e-01 6.47967875e-01
9.68708456e-01 -3.73611152e-01 1.83169618e-01 6.39608681e-01
3.41596335e-01 -6.38066411e-01 -7.08673358e-01 1.34666336e+00
1.92877397e-01 -8.17345083e-01 1.91562042e-01 -3.01763058e-01
4.93066251e-01 -6.51946604e-01 3.23894471e-01 -3.62354219e-01
-3.10967714e-01 -9.07917082e-01 5.84581375e-01 5.04797041e-01
9.62079227e-01 -1.01583874e+00 8.13043952e-01 -1.57446310e-01
-1.03984666e+00 1.06424093e-01 -5.07381737e-01 6.05071243e-03
-4.06361039e-04 4.96817321e-01 -2.21316800e-01 6.55993760e-01
6.99532926e-01 8.85576531e-02 -9.34307218e-01 1.30246270e+00
1.68383252e-02 9.56345275e-02 6.09948151e-02 2.24018499e-01
-1.14968300e-01 -2.75791734e-01 6.47188962e-01 1.06664562e+00
3.61028761e-01 1.91843867e-01 -2.73542613e-01 6.44380033e-01
1.90918609e-01 3.36651593e-01 -6.73661590e-01 4.21794742e-01
3.05800110e-01 1.01463830e+00 -8.88323426e-01 -2.04124883e-01
-3.56852025e-01 1.12925136e+00 -2.70614982e-01 -6.51518255e-02
-6.17781937e-01 -8.14005733e-01 6.05928361e-01 1.37364924e-01
-1.75349876e-01 -3.08230460e-01 -3.78586262e-01 -7.38435209e-01
-1.03692161e-02 -1.10703635e+00 4.04711574e-01 -9.22039449e-01
-1.19518101e+00 8.81489933e-01 2.55374014e-01 -1.72539926e+00
5.68582676e-02 -7.23497331e-01 -4.86973286e-01 1.04000854e+00
-1.43285310e+00 -4.52042252e-01 -6.49061263e-01 8.39789867e-01
4.80453581e-01 -5.49653843e-02 5.19374430e-01 2.92373985e-01
-2.98312128e-01 8.03776205e-01 -1.54706106e-01 -1.36376143e-01
9.61224079e-01 -1.23359048e+00 1.39095590e-01 1.37321448e+00
-3.81393254e-01 3.69256645e-01 9.85471606e-01 -3.16105843e-01
-9.95418251e-01 -6.91131473e-01 5.27617991e-01 -1.29222721e-01
7.60379583e-02 1.65829092e-01 -1.11538029e+00 -1.16357014e-01
8.55207562e-01 8.94462392e-02 5.07391393e-01 -6.03887141e-01
-5.41432023e-01 -6.14178360e-01 -1.64468157e+00 4.37447935e-01
6.86593950e-01 -4.01072174e-01 -6.06601715e-01 -2.44768038e-01
5.69532275e-01 -1.93835631e-01 -7.12101936e-01 4.18743610e-01
2.43521377e-01 -1.57568955e+00 9.21737373e-01 9.30247735e-03
4.35198508e-02 -8.68655264e-01 -1.30346820e-01 -1.22018313e+00
-3.43940407e-01 -6.13769054e-01 4.20919418e-01 1.15172112e+00
6.16875291e-03 -5.55706084e-01 5.98571002e-02 6.39706314e-01
7.13018030e-02 -9.74628255e-02 -5.92622697e-01 -1.09838331e+00
-3.74994010e-01 -2.28674099e-01 5.39466441e-01 6.15600705e-01
-1.17647581e-01 7.53619522e-02 -4.14282054e-01 3.05804104e-01
8.60108793e-01 -2.55420178e-01 5.32251537e-01 -9.18546557e-01
-1.87130377e-01 -4.98609006e-01 -1.17007577e+00 -3.12985480e-01
-5.67049444e-01 -4.25883770e-01 2.14823168e-02 -1.09278405e+00
1.58399209e-01 -1.46914557e-01 -5.64733565e-01 3.26589867e-02
-1.80007368e-01 5.68664253e-01 2.41072759e-01 4.26415280e-02
-4.58643228e-01 2.33544827e-01 1.12464452e+00 -2.21539587e-02
-1.49022684e-01 -2.11896613e-01 -2.10776478e-01 5.18999159e-01
9.78100359e-01 -3.81922573e-01 -5.60724437e-01 -4.30837959e-01
1.21222615e-01 -1.59885690e-01 1.96711913e-01 -1.57992136e+00
1.64606705e-01 -2.23054096e-01 2.85877168e-01 -6.49479687e-01
-1.64601341e-01 -9.76084173e-01 3.20015520e-01 5.34225345e-01
-3.21773469e-01 5.97463906e-01 2.32092127e-01 4.19095516e-01
-4.62653697e-01 -3.41985703e-01 1.52174175e+00 -4.01765779e-02
-8.28115940e-01 -1.60485469e-02 -3.12830687e-01 -1.63298056e-01
1.26897132e+00 -4.22165871e-01 -4.88620967e-01 -9.46093127e-02
-4.29246575e-01 -1.37705207e-01 6.95063770e-01 1.41106620e-01
7.74707735e-01 -1.38437164e+00 -5.71921468e-01 2.82163382e-01
1.94537908e-01 -5.22064090e-01 4.15227026e-01 7.06592083e-01
-9.01264608e-01 1.00298181e-01 -1.09592509e+00 -4.04094189e-01
-1.69726431e+00 1.08416629e+00 3.80962700e-01 -3.31685618e-02
-2.35619068e-01 4.80851382e-01 -7.05334023e-02 3.93869191e-01
8.11703801e-02 -1.91030577e-01 -4.45332527e-01 -2.00579002e-01
1.09878337e+00 8.16139579e-01 2.92783946e-01 -7.61057913e-01
-3.64454776e-01 7.64079869e-01 1.85337529e-01 -2.73333937e-01
7.18602479e-01 -5.16696334e-01 -2.05514804e-01 2.92079031e-01
1.22982013e+00 6.19644001e-02 -1.02102566e+00 1.09856047e-01
-1.21886253e-01 -1.26247275e+00 5.90603910e-02 -8.28272700e-01
-8.90755177e-01 1.01907420e+00 1.42264426e+00 4.13397372e-01
1.97615790e+00 -8.38801265e-01 4.91142273e-01 1.22003127e-02
4.69081193e-01 -1.25999141e+00 2.20937043e-01 -5.40186465e-02
9.88016784e-01 -1.35555816e+00 1.28229320e-01 -2.59689063e-01
-7.12223053e-01 1.05060506e+00 6.36264384e-01 -2.41914049e-01
5.57397425e-01 2.66450495e-01 2.47926682e-01 2.69506216e-01
-4.18638945e-01 -2.66998082e-01 3.99612010e-01 1.00870478e+00
2.07730159e-01 -2.76349992e-01 -1.04102659e+00 1.20120779e-01
3.21636051e-01 -8.65816697e-02 7.06319511e-01 9.03644204e-01
-6.08123958e-01 -1.38511395e+00 -6.97713852e-01 2.23978117e-01
-6.40781581e-01 5.66429421e-02 -1.02990661e-02 6.00874782e-01
4.24699783e-01 1.22853637e+00 3.32945623e-02 -6.12589717e-01
4.60988760e-01 -3.90035361e-01 4.93164271e-01 -1.57240495e-01
-7.61167526e-01 -2.13280097e-01 -2.68329561e-01 -1.00405192e+00
-7.45265007e-01 -1.84953138e-01 -1.07655716e+00 -5.22676110e-01
-1.98989064e-01 9.60867032e-02 7.78961182e-01 4.14427996e-01
2.98231870e-01 3.44997346e-01 9.18741524e-01 -4.74640906e-01
-4.28046376e-01 -5.95342755e-01 -6.75877154e-01 7.27730870e-01
3.55465084e-01 -5.38800836e-01 -4.13235158e-01 2.34519213e-01]
|
[11.739117622375488, -1.968567967414856]
|
1592a8c3-bc95-4179-9be8-df8219be0f8a
|
learning-unsupervised-multilingual-word
| null | null |
https://aclanthology.org/N19-1188
|
https://aclanthology.org/N19-1188.pdf
|
Learning Unsupervised Multilingual Word Embeddings with Incremental Multilingual Hubs
|
Recent research has discovered that a shared bilingual word embedding space can be induced by projecting monolingual word embedding spaces from two languages using a self-learning paradigm without any bilingual supervision. However, it has also been shown that for distant language pairs such fully unsupervised self-learning methods are unstable and often get stuck in poor local optima due to reduced isomorphism between starting monolingual spaces. In this work, we propose a new robust framework for learning unsupervised multilingual word embeddings that mitigates the instability issues. We learn a shared multilingual embedding space for a variable number of languages by incrementally adding new languages one by one to the current multilingual space. Through the gradual language addition the method can leverage the interdependencies between the new language and all other languages in the current multilingual space. We find that it is beneficial to project more distant languages later in the iterative process. Our fully unsupervised multilingual embedding spaces yield results that are on par with the state-of-the-art methods in the bilingual lexicon induction (BLI) task, and simultaneously obtain state-of-the-art scores on two downstream tasks: multilingual document classification and multilingual dependency parsing, outperforming even supervised baselines. This finding also accentuates the need to establish evaluation protocols for cross-lingual word embeddings beyond the omnipresent intrinsic BLI task in future work.
|
['Marie-Francine Moens', "Ivan Vuli{\\'c}", 'Geert Heyman', 'Bregt Verreet']
|
2019-06-01
| null | null | null |
naacl-2019-6
|
['multilingual-word-embeddings']
|
['methodology']
|
[-2.83582211e-01 4.29791696e-02 -6.28502548e-01 -3.71194512e-01
-1.01205778e+00 -9.90656078e-01 7.27730930e-01 1.82297856e-01
-8.29248130e-01 7.89530039e-01 6.14806771e-01 -7.04699814e-01
1.07803851e-01 -4.52428609e-01 -7.29231536e-01 -5.98609149e-01
-1.11328073e-01 7.13542342e-01 -1.49586409e-01 -4.28081989e-01
-2.40384609e-01 1.39652446e-01 -9.76907194e-01 -7.91001916e-02
1.00131428e+00 -1.67084515e-01 2.26495385e-01 4.28101659e-01
-2.61577249e-01 2.50375301e-01 -6.06312826e-02 -5.00026941e-01
2.76667535e-01 -4.63317007e-01 -7.25217879e-01 -3.35389733e-01
4.70330358e-01 9.74326879e-02 -1.76877439e-01 1.03461313e+00
4.27956879e-01 -3.00640792e-01 3.95678878e-01 -9.67996836e-01
-9.09230053e-01 9.96797740e-01 -4.50395763e-01 3.35925519e-01
1.85718000e-01 9.57167074e-02 1.52815640e+00 -1.19555748e+00
1.02984571e+00 1.25125098e+00 7.93766320e-01 2.40387440e-01
-1.63777053e+00 -8.75745177e-01 4.08776581e-01 -1.50953932e-02
-1.24940848e+00 -3.88733089e-01 6.81918502e-01 -4.79049414e-01
1.33521688e+00 -1.15574621e-01 4.01721954e-01 1.16597366e+00
3.17004293e-01 5.15205801e-01 1.45151746e+00 -9.04125988e-01
-2.60033637e-01 5.84362864e-01 3.01390558e-01 7.28226006e-01
4.47523654e-01 2.80340016e-01 -5.68438709e-01 -3.51264030e-02
3.36882591e-01 -3.64170253e-01 8.82825702e-02 -6.91353083e-01
-1.54550123e+00 1.16684854e+00 2.53863871e-01 7.95090973e-01
-1.46028344e-02 -2.10542828e-01 5.23424327e-01 6.27066791e-01
6.77112401e-01 5.70274770e-01 -7.95457900e-01 1.25103951e-01
-6.21516585e-01 -1.22895643e-01 8.06316853e-01 7.13900685e-01
1.15079129e+00 -1.33687988e-01 4.53902274e-01 9.09623384e-01
3.61387730e-01 4.20814216e-01 6.15942776e-01 -3.44294697e-01
6.53941870e-01 4.11326975e-01 -3.68494004e-01 -5.33563972e-01
-4.26352412e-01 -4.19675320e-01 -4.68377143e-01 9.13654342e-02
3.79151195e-01 -3.43010515e-01 -4.97108698e-01 2.17044473e+00
2.72101372e-01 -9.43920612e-02 4.46249276e-01 5.75266242e-01
3.28090966e-01 6.04683280e-01 2.07897559e-01 -1.74049754e-02
1.34364104e+00 -1.07143891e+00 -5.84272027e-01 -5.48216343e-01
1.13829947e+00 -9.60582078e-01 1.16409421e+00 -5.40688746e-02
-6.52244627e-01 -6.01228356e-01 -1.28296196e+00 -1.85880125e-01
-6.45291448e-01 -8.85894522e-02 1.03272545e+00 6.32933557e-01
-1.14249361e+00 2.74506211e-01 -8.61043513e-01 -6.55793786e-01
-1.12703115e-01 4.33770925e-01 -8.75408292e-01 -2.60833919e-01
-1.48243165e+00 1.33379126e+00 5.62207997e-01 -1.28896758e-01
-5.05394518e-01 -7.13043869e-01 -1.23944819e+00 -4.34338927e-01
1.07689932e-01 -2.54687279e-01 6.39175117e-01 -9.58421290e-01
-1.22039282e+00 1.13441467e+00 -1.30139783e-01 -2.08420500e-01
1.19318254e-01 -1.44560605e-01 -5.20213187e-01 -4.43256438e-01
5.70351243e-01 7.22830653e-01 3.03901464e-01 -1.17688107e+00
-4.90429908e-01 -4.95664299e-01 -7.40295798e-02 4.06205565e-01
-5.87659597e-01 2.77944148e-01 -3.50900263e-01 -6.81459069e-01
5.03774174e-02 -1.29212403e+00 -2.52953380e-01 -5.59685171e-01
-6.43413067e-02 -2.56601661e-01 4.33218777e-01 -8.22248101e-01
1.14829338e+00 -2.03279519e+00 4.35946167e-01 -2.70905588e-02
-1.24971382e-01 1.32830262e-01 -5.46322048e-01 6.06174111e-01
-4.64613080e-01 1.00264974e-01 -1.26487225e-01 -4.82391745e-01
-1.24419555e-02 6.09847903e-01 -1.12473965e-01 6.15000665e-01
3.94688606e-01 9.48360920e-01 -1.18820381e+00 -5.26278317e-01
6.51713321e-03 5.24660647e-01 -6.12425566e-01 -6.44768178e-02
2.33499706e-01 6.71210527e-01 2.33553410e-01 4.61903542e-01
5.15232146e-01 1.85171336e-01 8.91300023e-01 -1.41333029e-01
-3.85167748e-01 6.87880635e-01 -1.17256784e+00 2.03118467e+00
-8.97484660e-01 7.12084651e-01 -1.08528358e-03 -9.71750498e-01
7.65924871e-01 4.69108284e-01 2.39522099e-01 -5.41949034e-01
-2.29087546e-01 6.96854174e-01 4.49441940e-01 -2.65534282e-01
2.52986461e-01 -5.35244763e-01 -4.07896698e-01 8.02439094e-01
7.62956798e-01 1.30458161e-01 2.59804249e-01 6.77310079e-02
7.18499422e-01 2.77289689e-01 4.05399740e-01 -7.82540917e-01
5.70230961e-01 -4.89414297e-02 5.67820847e-01 3.98799837e-01
-1.18112125e-01 8.69347602e-02 2.83122778e-01 -5.26377141e-01
-1.21388173e+00 -1.39339471e+00 -5.91356754e-01 1.57065868e+00
-1.28430843e-01 -5.41476190e-01 -3.95930231e-01 -8.95012558e-01
4.64509875e-02 5.78766644e-01 -4.39442694e-01 1.19166169e-02
-8.93816710e-01 -1.06404674e+00 6.24395251e-01 6.03896618e-01
-1.60160750e-01 -1.05489898e+00 9.98818576e-02 3.49216700e-01
-1.49258152e-01 -1.16866255e+00 -6.07827008e-01 7.76298285e-01
-8.16599786e-01 -7.31449544e-01 -3.65529746e-01 -1.42214346e+00
6.89321280e-01 -9.82389748e-02 1.24973035e+00 -3.14873785e-01
-9.21193883e-02 1.26812428e-01 -8.93963799e-02 -6.36795610e-02
-7.14913011e-01 5.85572541e-01 5.70432544e-01 -2.68856972e-01
7.25373507e-01 -4.97747570e-01 -1.03797771e-01 -1.64587609e-02
-8.71638477e-01 -8.38783160e-02 4.42207456e-01 1.11626542e+00
4.46021944e-01 -2.32541829e-01 6.45558059e-01 -9.36620891e-01
7.98017025e-01 -6.31528199e-01 -5.06553292e-01 2.69447803e-01
-8.13145518e-01 7.13584244e-01 5.34990549e-01 -5.08438468e-01
-8.24244797e-01 -9.48398784e-02 -5.03772087e-02 2.60287195e-01
-3.24280486e-02 6.05833530e-01 -2.17254594e-01 1.13912061e-01
5.89092314e-01 -8.41793567e-02 -2.86535975e-02 -5.15605032e-01
8.54890049e-01 7.06992984e-01 3.20735931e-01 -7.23188221e-01
9.22202468e-01 1.45208135e-01 -4.52711016e-01 -6.12934411e-01
-7.42016792e-01 -5.79692602e-01 -1.19047046e+00 2.99290389e-01
1.02470243e+00 -1.29933286e+00 5.16742468e-02 8.62511247e-02
-1.18202579e+00 -2.79418230e-01 -9.89071652e-02 8.05997133e-01
-2.17056960e-01 3.30326825e-01 -6.64276481e-01 -3.02418560e-01
-4.95220236e-02 -1.39573514e+00 8.13133240e-01 -2.57249922e-01
-4.83040601e-01 -1.68495035e+00 8.22924793e-01 1.63585499e-01
1.94346100e-01 -7.66632259e-02 1.15319955e+00 -7.78346598e-01
-2.28189439e-01 -3.29375528e-02 2.40117554e-02 5.68675637e-01
2.92800426e-01 -3.63591969e-01 -7.15874314e-01 -6.40904546e-01
-1.44471392e-01 -4.27024394e-01 7.69191265e-01 -1.19616449e-01
-1.61100291e-02 -8.74257162e-02 -2.37019375e-01 7.36223340e-01
1.65982294e+00 -1.95839971e-01 2.86620334e-02 6.26833081e-01
8.31364751e-01 7.63979614e-01 1.78067759e-01 -3.80767047e-01
7.54520714e-01 5.77324092e-01 -2.24557608e-01 -4.03365642e-01
-1.73932582e-01 -3.63018274e-01 7.94450939e-01 1.64667714e+00
2.78702885e-01 2.61708975e-01 -1.23559523e+00 1.03509653e+00
-1.56860459e+00 -5.37082434e-01 1.13335803e-01 2.37535691e+00
1.32147932e+00 1.52563989e-01 -3.03112753e-02 -2.05971062e-01
4.22363549e-01 2.36569211e-01 -2.10457742e-01 -7.47429311e-01
-4.10999745e-01 6.57106042e-01 8.37866902e-01 1.18860471e+00
-1.13114536e+00 1.51417291e+00 6.23022938e+00 2.69039452e-01
-1.09678733e+00 6.99727416e-01 2.80914664e-01 1.63296655e-01
-6.12150431e-01 4.78079438e-01 -1.01970708e+00 1.36256948e-01
1.11311638e+00 -1.51358292e-01 5.92167318e-01 4.01123971e-01
-2.92012453e-01 1.23343520e-01 -1.35947859e+00 5.99516928e-01
2.32198685e-01 -8.85574758e-01 -2.51532048e-01 1.74992934e-01
1.02693844e+00 7.78623283e-01 9.85853095e-03 5.08298874e-01
7.07138360e-01 -1.05249488e+00 2.86560565e-01 -1.12894319e-01
9.77450311e-01 -7.59243786e-01 6.53871715e-01 2.23394766e-01
-1.18255186e+00 5.49632385e-02 -2.42454231e-01 -7.54462928e-02
2.19785258e-01 3.08314830e-01 -8.82672310e-01 4.33879614e-01
4.58580762e-01 6.45232439e-01 -7.61941612e-01 3.36777955e-01
-4.69157726e-01 6.07615173e-01 -2.61342704e-01 3.11311334e-01
5.33755898e-01 -4.48789895e-01 5.08931756e-01 1.49624443e+00
9.21215713e-02 -6.45083666e-01 2.97558099e-01 5.29607296e-01
-7.98844621e-02 6.12325191e-01 -1.12724257e+00 -1.91405028e-01
1.72645018e-01 1.03391957e+00 -5.52918494e-01 -2.16397837e-01
-8.31306875e-01 1.01518869e+00 7.92247474e-01 3.47267181e-01
-5.88024318e-01 -2.70889401e-01 8.30116749e-01 -1.74861789e-01
2.08573893e-01 -6.46007299e-01 -2.42996663e-01 -1.52422035e+00
3.91933583e-02 -1.03921080e+00 3.21596205e-01 -5.68338260e-02
-1.38963771e+00 7.05136955e-01 -6.77364245e-02 -9.30381119e-01
-4.37161535e-01 -8.17183673e-01 -3.40170562e-01 1.15323555e+00
-1.71256220e+00 -1.29088414e+00 4.67449844e-01 3.85267824e-01
3.98204356e-01 -3.63428295e-01 1.32812703e+00 3.80672961e-01
-4.80243146e-01 8.20144236e-01 3.82105887e-01 2.53285348e-01
1.11739039e+00 -1.40868735e+00 5.73459446e-01 9.24241722e-01
8.57709169e-01 9.05420005e-01 4.79453295e-01 -5.73528051e-01
-1.27432871e+00 -7.85946250e-01 1.73413420e+00 -7.43438303e-01
1.30215824e+00 -8.70841742e-01 -8.39549661e-01 1.08532917e+00
7.01693177e-01 4.07113228e-03 1.05635965e+00 8.17088306e-01
-7.76258111e-01 -8.92802104e-02 -5.75209200e-01 7.67451286e-01
1.00459182e+00 -1.10210836e+00 -7.06145704e-01 5.87198138e-01
8.01471591e-01 1.57561705e-01 -8.95944297e-01 2.59821445e-01
5.48131466e-01 -6.40836000e-01 9.63777304e-01 -8.08454454e-01
1.91263124e-01 -6.50798082e-02 -1.78353474e-01 -1.63038290e+00
-2.44484469e-01 -4.90167469e-01 4.42379981e-01 1.27327490e+00
8.27281594e-01 -9.13808882e-01 3.10393453e-01 3.59817855e-02
1.69238552e-01 -5.03531575e-01 -1.01294982e+00 -1.03324306e+00
8.72744560e-01 -3.85466546e-01 1.92449197e-01 1.39539886e+00
3.48306954e-01 8.83088827e-01 -1.68116212e-01 2.16613904e-01
7.24250197e-01 -6.30456805e-02 6.12315536e-01 -1.12771416e+00
-3.42581630e-01 -3.85578245e-01 -4.08408493e-01 -6.55265033e-01
8.41870904e-01 -1.72477317e+00 6.96796626e-02 -1.20835805e+00
3.40756625e-01 -5.93952835e-01 -6.58281922e-01 4.54285860e-01
-3.52188677e-01 2.88466364e-01 1.30057767e-01 1.18228979e-01
-3.58116567e-01 2.36686006e-01 6.76461935e-01 -8.74685943e-02
-2.91216791e-01 -6.76256537e-01 -9.27683830e-01 6.94087625e-01
6.99080050e-01 -8.15299630e-01 -2.38467470e-01 -7.95882702e-01
2.46796459e-01 -4.09726620e-01 -1.81224585e-01 -6.48107588e-01
8.33704844e-02 1.58803150e-01 -1.26050804e-02 -1.00984730e-01
-1.32955104e-01 -6.23240948e-01 -1.74487948e-01 3.74561429e-01
-1.91272676e-01 5.04875481e-01 3.42217445e-01 2.12771341e-01
-2.69652277e-01 -2.56849170e-01 5.98797023e-01 -8.66436735e-02
-6.10582948e-01 -3.87039818e-02 -2.42627531e-01 2.96278089e-01
7.71442354e-01 2.06011236e-01 1.48446262e-01 1.04940720e-01
-7.02535033e-01 1.05557173e-01 5.71147323e-01 8.39280903e-01
-1.49816722e-02 -1.58747423e+00 -9.17602599e-01 5.90875268e-01
4.23104852e-01 -5.57534039e-01 -3.86065185e-01 8.41934621e-01
-3.62029672e-01 6.72990143e-01 -1.22005753e-01 -5.58082521e-01
-1.01615167e+00 5.79111218e-01 1.21246286e-01 -7.78716922e-01
-3.46959382e-01 7.52301037e-01 2.72951841e-01 -1.34960949e+00
-1.07076978e-02 -2.20611796e-01 6.47116378e-02 2.09101066e-01
3.15271169e-02 -5.61889373e-02 6.54576421e-02 -1.04048264e+00
-5.19456148e-01 7.09758461e-01 -3.47981691e-01 -4.89058882e-01
1.39954746e+00 -1.17555350e-01 -3.76882970e-01 7.93009222e-01
1.58026576e+00 5.25765240e-01 -7.87360311e-01 -4.46902603e-01
5.11785030e-01 -1.51732117e-01 -8.71701986e-02 -6.21797442e-01
-6.68092906e-01 7.43127108e-01 6.31980062e-01 -2.84371346e-01
6.31544292e-01 3.27782810e-01 6.59634590e-01 2.37490013e-01
3.15134764e-01 -1.17807734e+00 -1.95068479e-01 7.44263649e-01
5.40407479e-01 -1.57873750e+00 -1.49087653e-01 1.89096108e-02
-6.39429510e-01 9.17478323e-01 3.21314454e-01 -1.96986854e-01
7.67662704e-01 5.14084816e-01 5.65057695e-01 -3.98250520e-02
-5.89949846e-01 -2.13611707e-01 2.32554600e-01 5.35746098e-01
9.89281118e-01 3.23320478e-01 -6.11283541e-01 4.49170351e-01
-3.23808700e-01 -5.93540013e-01 7.72091970e-02 6.97376490e-01
1.79804210e-02 -1.94813740e+00 -3.28407496e-01 -1.11867398e-01
-3.47174972e-01 -6.55866265e-01 -1.32645100e-01 9.89806592e-01
4.08822924e-01 7.18168736e-01 1.14139058e-01 -7.47554302e-02
1.33329540e-01 6.06684566e-01 5.62857449e-01 -8.40162456e-01
-5.20233274e-01 5.65798581e-02 1.96767151e-02 -3.48564774e-01
-3.98284912e-01 -9.56551313e-01 -1.01945972e+00 2.06862301e-01
-3.35563481e-01 5.92846349e-02 7.43237138e-01 9.53741133e-01
1.08725145e-01 1.53836459e-01 4.62005168e-01 -6.77317917e-01
-4.41297948e-01 -9.69716549e-01 -2.30735436e-01 5.08326888e-01
2.36688837e-01 -5.19583404e-01 -4.38699931e-01 -7.19861388e-02]
|
[11.010897636413574, 10.068422317504883]
|
8cdf53b9-a54b-4287-8797-58745d42de7d
|
who-should-go-first-a-self-supervised-concept
|
2104.03682
| null |
https://arxiv.org/abs/2104.03682v2
|
https://arxiv.org/pdf/2104.03682v2.pdf
|
Who Should Go First? A Self-Supervised Concept Sorting Model for Improving Taxonomy Expansion
|
Taxonomies have been widely used in various machine learning and text mining systems to organize knowledge and facilitate downstream tasks. One critical challenge is that, as data and business scope grow in real applications, existing taxonomies need to be expanded to incorporate new concepts. Previous works on taxonomy expansion process the new concepts independently and simultaneously, ignoring the potential relationships among them and the appropriate order of inserting operations. However, in reality, the new concepts tend to be mutually correlated and form local hypernym-hyponym structures. In such a scenario, ignoring the dependencies of new concepts and the order of insertion may trigger error propagation. For example, existing taxonomy expansion systems may insert hyponyms to existing taxonomies before their hypernym, leading to sub-optimal expanded taxonomies. To complement existing taxonomy expansion systems, we propose TaxoOrder, a novel self-supervised framework that simultaneously discovers the local hypernym-hyponym structure among new concepts and decides the order of insertion. TaxoOrder can be directly plugged into any taxonomy expansion system and improve the quality of expanded taxonomies. Experiments on the real-world dataset validate the effectiveness of TaxoOrder to enhance taxonomy expansion systems, leading to better-resulting taxonomies with comparison to baselines under various evaluation metrics.
|
['Jiawei Han', 'Jieyu Zhang', 'Jiaming Shen', 'Xiangchen Song']
|
2021-04-08
| null | null | null | null |
['taxonomy-expansion']
|
['natural-language-processing']
|
[ 6.66867495e-02 1.57553442e-02 -4.35321450e-01 -2.43459523e-01
4.57393169e-01 -7.43785203e-01 3.13388646e-01 6.13097787e-01
-4.48695272e-01 5.26193261e-01 1.16622047e-02 -5.19759715e-01
-6.35048687e-01 -1.25539207e+00 -2.21530460e-02 -3.04497153e-01
-1.63698848e-02 1.13024056e+00 1.71334833e-01 -4.17928010e-01
1.79150328e-01 2.28638619e-01 -1.93655884e+00 1.51156560e-01
1.16766453e+00 7.17042208e-01 2.50013202e-01 -1.96586713e-01
-6.30987108e-01 8.61847326e-02 -5.21918774e-01 -4.82011735e-01
6.29069328e-01 7.74336383e-02 -1.04679406e+00 5.16023710e-02
2.78323203e-01 -9.84028578e-02 -9.13318321e-02 1.13931346e+00
-2.10541785e-02 1.84536591e-01 4.01852369e-01 -1.38219666e+00
-1.87066048e-01 8.20116341e-01 -5.31454742e-01 6.89209029e-02
3.06902796e-01 -3.50041598e-01 1.55834937e+00 -7.49620497e-01
8.05188835e-01 1.14747322e+00 3.14102530e-01 1.67740032e-01
-1.02845371e+00 -1.01548445e+00 3.81857276e-01 4.18737084e-01
-1.51750958e+00 2.70032406e-01 4.25398052e-01 -2.33503580e-01
1.07788301e+00 2.85620332e-01 8.03114474e-01 4.48202491e-01
-1.24879144e-01 3.45513135e-01 6.59940064e-01 -4.91121769e-01
3.57987314e-01 2.70856500e-01 3.35047781e-01 3.41621637e-01
7.41466284e-01 -1.43650681e-01 -3.00711423e-01 -1.64052397e-01
4.56243098e-01 3.62738818e-01 2.27186456e-02 -4.49074537e-01
-8.99602950e-01 8.75282109e-01 4.28558707e-01 3.35524201e-01
-2.73781598e-01 -6.48172557e-01 4.72336769e-01 3.47157091e-01
7.81147853e-02 1.09572935e+00 -5.85839689e-01 7.72459060e-02
-8.17820787e-01 4.09877121e-01 7.34230220e-01 1.15776694e+00
9.94975328e-01 -6.79185688e-01 2.14257792e-01 1.04295552e+00
-9.38605741e-02 -2.04657659e-01 9.36873376e-01 -6.87438726e-01
4.93469864e-01 1.66244900e+00 -2.12146968e-01 -8.31490040e-01
-6.07176244e-01 -6.14655614e-01 -6.87533736e-01 -3.07716221e-01
-5.06753102e-02 2.52188712e-01 -7.58536279e-01 1.42657280e+00
5.98371565e-01 1.14920825e-01 -8.06557164e-02 6.56053364e-01
7.62198687e-01 3.56338620e-01 5.95426410e-02 -3.03660274e-01
1.59353149e+00 -6.95726216e-01 -4.59762752e-01 -3.16082358e-01
1.02486181e+00 -6.43766344e-01 1.17156410e+00 4.83722359e-01
-4.14444208e-01 -2.74620920e-01 -8.44879508e-01 1.91626042e-01
-6.82348788e-01 -1.72776461e-01 1.13656628e+00 5.66209555e-01
-4.16988850e-01 5.91993630e-01 -2.53434956e-01 -5.88497162e-01
2.39914224e-01 4.42866713e-01 -3.45183074e-01 -3.23792577e-01
-1.45957887e+00 8.01894844e-01 1.29253876e+00 -3.30522120e-01
-2.56658733e-01 -8.59880447e-01 -6.62503421e-01 2.80980557e-01
1.08539379e+00 -5.70621967e-01 1.09397030e+00 -3.98078829e-01
-6.72867954e-01 5.73025346e-01 6.05727620e-02 -4.97196138e-01
-1.09610692e-01 6.10797182e-02 -7.20651567e-01 -2.07585454e-01
2.21151441e-01 6.75211549e-01 5.91934323e-02 -9.73875344e-01
-1.42897451e+00 -4.11615700e-01 4.58236009e-01 6.39568210e-01
-1.02209663e+00 -2.59149045e-01 -3.99685532e-01 -4.25379962e-01
5.29595554e-01 -8.35137129e-01 -3.57429028e-01 -5.05495131e-01
-1.11898549e-01 -6.17812991e-01 7.94763863e-01 2.24435758e-02
1.61203647e+00 -1.79080176e+00 1.16594017e-01 4.43312317e-01
7.60754645e-01 2.66078442e-01 -4.68229316e-02 5.59898615e-01
-3.40361446e-01 2.24933162e-01 -4.20609675e-03 2.32585415e-01
-1.44814581e-01 6.91344976e-01 -2.79307753e-01 -1.59007564e-01
-2.76765406e-01 7.10753858e-01 -1.12070680e+00 -4.15253788e-01
2.65087932e-01 -3.95403951e-01 -6.18452013e-01 -1.64614338e-02
-2.73160577e-01 -1.58636436e-01 -2.85609998e-02 7.02517271e-01
5.13009667e-01 -2.38932401e-01 5.65671265e-01 -1.32835180e-01
1.50617361e-01 5.61284542e-01 -1.35761201e+00 1.21143544e+00
-6.34291887e-01 1.18798286e-01 -4.06225324e-01 -1.15201986e+00
1.02962005e+00 2.04778746e-01 8.50851536e-01 -5.38619339e-01
3.53501877e-03 4.41378802e-01 4.51781809e-01 -3.95709515e-01
7.26586342e-01 -2.84618616e-01 1.79893132e-02 5.13548851e-01
-9.10689309e-03 -4.89396974e-02 9.01087999e-01 4.97213125e-01
1.04278898e+00 -4.12850201e-01 8.34045470e-01 -1.90170109e-01
4.05212939e-01 1.64648846e-01 8.44275713e-01 4.38979059e-01
3.26895595e-01 -4.72593233e-02 2.62278557e-01 -4.23954248e-01
-8.68246675e-01 -8.54551196e-01 -2.36878335e-01 1.14054894e+00
4.04490322e-01 -1.14286530e+00 -2.14832708e-01 -8.14650834e-01
1.75511122e-01 7.64891982e-01 -1.79604128e-01 -2.61831701e-01
-2.40892932e-01 -6.49131775e-01 4.31865126e-01 4.83790517e-01
1.36228710e-01 -9.85610962e-01 -6.86862230e-01 5.10283828e-01
-3.76150936e-01 -1.32406533e+00 -1.23408914e-01 3.58310878e-01
-9.63133633e-01 -1.21225441e+00 6.52954802e-02 -5.52659929e-01
5.58275521e-01 4.12167102e-01 1.16891599e+00 2.30214208e-01
-2.21141160e-01 -1.34489641e-01 -7.92747021e-01 -4.70677286e-01
-2.38970414e-01 5.26229858e-01 1.76035389e-01 -5.10787368e-01
9.42402124e-01 -8.95383239e-01 -2.25769073e-01 5.35679221e-01
-1.22359788e+00 -1.22776031e-02 4.49124604e-01 8.91218483e-01
2.57799476e-01 9.56053674e-01 6.17662728e-01 -1.20236266e+00
7.53647089e-01 -6.01186574e-01 -7.42475092e-01 4.59216028e-01
-1.38003063e+00 3.20358872e-02 7.19596982e-01 -4.87437457e-01
-8.68109643e-01 -1.03523500e-01 1.59463167e-01 -2.07962751e-01
-1.26553997e-01 1.14320278e+00 -4.14037406e-01 2.71470666e-01
5.81523180e-01 -1.70006216e-01 -4.52155083e-01 -5.53779960e-01
4.78166580e-01 8.35727334e-01 6.97936416e-01 -7.49985218e-01
8.21320176e-01 3.76515031e-01 1.31568134e-01 -4.86086637e-01
-6.06444061e-01 -1.20637178e+00 -8.89059484e-01 1.46914348e-01
-3.18824612e-02 -7.85943151e-01 -5.79801083e-01 -8.35659280e-02
-8.94271553e-01 3.57195973e-01 -4.54166830e-01 3.32933336e-01
1.17211223e-01 6.32343054e-01 -2.42396593e-01 -4.28319305e-01
-3.63068789e-01 -8.41663420e-01 8.08301151e-01 2.44244322e-01
-6.42408073e-01 -8.71143997e-01 -1.22221746e-01 3.61928374e-01
-6.95319474e-02 -3.03976893e-01 1.56129992e+00 -1.15445757e+00
-3.18482161e-01 -4.45646286e-01 -1.76427618e-01 -2.43790057e-02
5.94084561e-01 -1.34526148e-01 -3.38137537e-01 -2.27923110e-01
-2.89461553e-01 -3.08449268e-02 6.68628097e-01 -2.29445219e-01
1.07158315e+00 -4.14939851e-01 -6.02072537e-01 5.06478548e-01
1.21438670e+00 4.23346162e-01 3.53999883e-01 5.45162976e-01
5.44765532e-01 9.68010902e-01 9.14656460e-01 7.36780882e-01
4.94278312e-01 7.60455251e-01 3.22606415e-01 1.49810806e-01
3.16029698e-01 -2.54152358e-01 -9.96995941e-02 7.47436523e-01
2.38713756e-01 -2.61416703e-01 -1.07825243e+00 6.80336952e-01
-1.77841878e+00 -6.03368819e-01 -1.34212628e-01 2.43961883e+00
1.06780195e+00 2.06013128e-01 2.32409731e-01 4.92663771e-01
5.53862572e-01 -4.54457372e-01 -6.22298837e-01 -1.36395693e-01
-1.17062174e-01 3.79339904e-01 4.59278941e-01 1.88858971e-01
-7.73933947e-01 1.32572401e+00 5.18716335e+00 8.61556053e-01
-8.68516564e-01 -1.37498766e-01 -1.28001720e-01 -2.47425824e-01
-4.84852523e-01 5.44723392e-01 -1.07690728e+00 1.50845066e-01
4.43921924e-01 -7.98585594e-01 3.61767292e-01 8.81281555e-01
-1.57191738e-01 -9.98208448e-02 -1.22538316e+00 8.12321901e-01
-2.54809767e-01 -1.14420199e+00 4.09160167e-01 3.31598997e-01
6.82346880e-01 -2.87400991e-01 -3.02346379e-01 4.32038218e-01
6.56984091e-01 -9.18660522e-01 1.12288319e-01 -3.46941710e-01
6.90396726e-01 -7.95194447e-01 7.96602190e-01 3.82083267e-01
-1.56053114e+00 -5.23865581e-01 -5.44153214e-01 -4.57311571e-01
9.71327871e-02 9.00934339e-01 -1.43407309e+00 9.03945446e-01
6.90593600e-01 6.46222532e-01 -6.95041776e-01 1.24914289e+00
-4.07267094e-01 3.37029755e-01 -6.18842959e-01 1.15281843e-01
1.94885835e-01 -2.83833176e-01 5.18270969e-01 7.92046547e-01
2.26813152e-01 6.80618957e-02 3.95054787e-01 6.16400003e-01
-1.10429630e-01 4.12787914e-01 -2.99056262e-01 -1.87086180e-01
1.19671297e+00 1.39964449e+00 -8.66347134e-01 -5.90560675e-01
-3.17999095e-01 4.24158841e-01 2.12331176e-01 2.41313696e-01
-4.92011219e-01 -5.08251667e-01 8.20696652e-01 2.95235634e-01
5.50238937e-02 4.81237322e-02 -4.76723194e-01 -1.14359295e+00
1.03883430e-01 -1.02402520e+00 1.08144760e+00 -3.00832927e-01
-1.34568191e+00 4.51905131e-01 4.09750044e-01 -1.23068559e+00
-3.13460797e-01 -4.92530346e-01 -2.97363341e-01 5.44435382e-01
-9.98726845e-01 -7.93474317e-01 -3.90680492e-01 2.69773155e-01
4.58634466e-01 -2.97367454e-01 8.05741727e-01 3.95960599e-01
-5.67348540e-01 6.61369622e-01 -3.64242904e-02 -1.96191698e-01
6.87429965e-01 -1.17521644e+00 5.30224085e-01 8.89012396e-01
4.20667350e-01 1.04542339e+00 5.52090406e-01 -1.00483191e+00
-6.49573565e-01 -1.07575274e+00 1.00697947e+00 -6.46028668e-02
7.11987674e-01 -3.90998214e-01 -1.18676722e+00 5.88455796e-01
-1.35226503e-01 -4.36967850e-01 8.13744485e-01 7.27411032e-01
-7.90088058e-01 -2.95587450e-01 -9.00130033e-01 7.30739474e-01
1.38055718e+00 -9.53140482e-02 -8.74086320e-01 4.20478821e-01
7.74103343e-01 -5.68987094e-02 -9.07465935e-01 7.42624760e-01
5.36606312e-01 -5.86942136e-01 8.80380094e-01 -6.99645877e-01
3.13511431e-01 -4.24750060e-01 4.34104241e-02 -1.35251474e+00
-3.35959643e-01 -3.53522241e-01 -2.32629418e-01 1.17109370e+00
3.84811968e-01 -7.70253718e-01 8.93689394e-01 5.26265442e-01
1.98030815e-01 -7.48279810e-01 -8.92686188e-01 -1.24587715e+00
-2.38298867e-02 -2.11374059e-01 1.18419087e+00 1.19431269e+00
6.50732219e-01 6.91848278e-01 -1.76420361e-02 -3.49943601e-02
3.25562954e-01 4.68897641e-01 6.55703664e-01 -1.89438581e+00
-6.40064105e-02 -8.28723788e-01 -5.47680080e-01 -9.74182606e-01
9.15272087e-02 -1.20140707e+00 -1.56640038e-01 -1.55086327e+00
2.11499125e-01 -8.71347368e-01 -2.74828523e-01 8.92496347e-01
-3.19971502e-01 -3.21969688e-01 6.58266246e-02 3.60389948e-01
-3.88284981e-01 4.18254495e-01 8.26126695e-01 -1.33299232e-01
-5.41539073e-01 -5.15977181e-02 -8.34618747e-01 6.76288486e-01
6.21600509e-01 -6.06480241e-01 -7.90455401e-01 -2.19338504e-03
4.70759243e-01 -4.97088403e-01 -9.28713530e-02 -7.94270933e-01
4.56604928e-01 -3.68385673e-01 -2.68181533e-01 -5.56273878e-01
3.88953723e-02 -1.06113505e+00 1.58530146e-01 3.28224123e-01
-2.41114795e-01 1.39457464e-01 1.35236859e-01 3.81606638e-01
-2.47118145e-01 -6.07525826e-01 4.31628317e-01 -1.23736195e-01
-1.01120186e+00 3.01990002e-01 -2.13309988e-01 -1.81850232e-02
1.01622617e+00 -4.90008622e-01 -1.88715488e-01 -1.36202415e-02
-2.71682501e-01 5.54109812e-01 5.31932712e-01 7.90616691e-01
4.77643698e-01 -1.08387470e+00 -4.10399139e-01 5.50608598e-02
7.09534705e-01 3.92137706e-01 -1.82672217e-01 2.84859210e-01
-1.19420126e-01 5.81095815e-01 -2.60990053e-01 -2.89827704e-01
-1.45201147e+00 6.83928192e-01 8.19531232e-02 -5.49600065e-01
-5.12879789e-01 7.15484202e-01 4.10053819e-01 -7.95873463e-01
3.15311193e-01 -3.93535912e-01 -5.56098878e-01 2.23815501e-01
3.06662828e-01 2.56990582e-01 4.55460548e-01 -2.24414945e-01
-3.13368440e-01 1.53332457e-01 -5.57634056e-01 1.62837476e-01
1.10553491e+00 -1.24062166e-01 -4.85212266e-01 1.81224972e-01
6.94253862e-01 -2.79927105e-01 -2.70978391e-01 -6.87443137e-01
6.18599057e-01 -6.04580581e-01 -3.16009134e-01 -8.01428735e-01
-9.54362333e-01 5.86540103e-01 1.84878990e-01 1.63608089e-01
1.37164128e+00 -1.54871112e-02 8.34382713e-01 5.59169114e-01
4.47045177e-01 -1.06519485e+00 -6.85295761e-02 5.44875145e-01
3.75517815e-01 -7.39222527e-01 3.82884264e-01 -9.39787090e-01
-4.24331516e-01 1.02070987e+00 1.08245826e+00 4.98241127e-01
4.47317362e-01 2.63917625e-01 -1.45590633e-01 -3.54821235e-01
-8.49577010e-01 -3.18740815e-01 3.36835563e-01 4.89208519e-01
3.24725181e-01 2.29063198e-01 -7.21754014e-01 6.84740603e-01
-6.28729522e-01 -1.78762451e-01 4.57986653e-01 7.45749652e-01
-4.01958108e-01 -1.72320068e+00 -1.40664130e-01 8.95549893e-01
1.40572056e-01 -4.06174630e-01 -5.87628722e-01 6.19164586e-01
6.99153304e-01 8.69078815e-01 9.86074731e-02 -6.49495304e-01
3.32617939e-01 3.64419669e-02 1.76020488e-01 -1.18146396e+00
-4.88541931e-01 -1.42470812e-02 9.52070579e-02 -1.34562746e-01
-7.51442909e-02 -5.23517489e-01 -1.58687341e+00 -2.59663701e-01
-7.75695920e-01 3.71641636e-01 2.99442470e-01 1.27277684e+00
4.17154312e-01 3.67965817e-01 3.42811823e-01 1.36666194e-01
-5.59223175e-01 -9.47079837e-01 -5.88345706e-01 6.05467916e-01
-5.35424292e-01 -1.01108527e+00 2.12257225e-02 -2.82278478e-01]
|
[9.225075721740723, 7.954920768737793]
|
69bb5692-ea20-4539-9042-e404373e7c18
|
with-measured-words-simple-sentence-selection
|
2101.10096
| null |
https://arxiv.org/abs/2101.10096v1
|
https://arxiv.org/pdf/2101.10096v1.pdf
|
With Measured Words: Simple Sentence Selection for Black-Box Optimization of Sentence Compression Algorithms
|
Sentence Compression is the task of generating a shorter, yet grammatical version of a given sentence, preserving the essence of the original sentence. This paper proposes a Black-Box Optimizer for Compression (B-BOC): given a black-box compression algorithm and assuming not all sentences need be compressed -- find the best candidates for compression in order to maximize both compression rate and quality. Given a required compression ratio, we consider two scenarios: (i) single-sentence compression, and (ii) sentences-sequence compression. In the first scenario, our optimizer is trained to predict how well each sentence could be compressed while meeting the specified ratio requirement. In the latter, the desired compression ratio is applied to a sequence of sentences (e.g., a paragraph) as a whole, rather than on each individual sentence. To achieve that, we use B-BOC to assign an optimal compression ratio to each sentence, then cast it as a Knapsack problem, which we solve using bounded dynamic programming. We evaluate B-BOC on both scenarios on three datasets, demonstrating that our optimizer improves both accuracy and Rouge-F1-score compared to direct application of other compression algorithms.
|
['Oren Tsur', 'Meir Kalech', 'Yotam Shichel']
|
2021-01-25
| null |
https://aclanthology.org/2021.eacl-main.139
|
https://aclanthology.org/2021.eacl-main.139.pdf
|
eacl-2021-2
|
['sentence-compression']
|
['natural-language-processing']
|
[ 7.47934461e-01 3.78585964e-01 -9.28457007e-02 -4.41557020e-01
-9.84330833e-01 -5.25934160e-01 1.87777802e-02 5.89939654e-01
-5.71111739e-01 5.25092483e-01 3.88798535e-01 -4.95368987e-01
-1.21931732e-01 -8.40715528e-01 -8.84215474e-01 -4.47713405e-01
2.65191458e-02 5.57461500e-01 -1.30012274e-01 -1.38978243e-01
6.44552827e-01 1.61968753e-01 -1.41975260e+00 3.84044021e-01
1.10664630e+00 7.17258275e-01 7.48804688e-01 1.22504485e+00
-7.33196214e-02 5.48379660e-01 -6.83513701e-01 -7.94965506e-01
3.56061846e-01 -6.89113975e-01 -1.15294600e+00 2.80097902e-01
1.38262689e-01 -3.69478941e-01 -2.35640019e-01 1.19550526e+00
2.11454302e-01 1.39905244e-01 3.85686159e-01 -7.62480974e-01
-3.37096363e-01 9.31397974e-01 -5.51746845e-01 1.02804467e-01
6.00104094e-01 1.05012029e-01 1.26566100e+00 -3.61617476e-01
5.50141990e-01 1.12405419e+00 2.14853346e-01 5.03626108e-01
-1.38935733e+00 5.90141602e-02 -2.38197342e-01 -1.14944182e-01
-1.12311864e+00 -5.69137156e-01 2.61383653e-01 -4.33756188e-02
1.31664753e+00 6.76119149e-01 6.97830021e-01 3.41608256e-01
4.43251967e-01 6.17843151e-01 4.92685467e-01 -6.74170673e-01
4.83622074e-01 -1.97225492e-02 1.01489156e-01 4.39562052e-01
3.24611843e-01 -3.62534672e-01 -4.27757710e-01 -6.83746636e-02
-3.68635952e-01 -2.44277090e-01 -2.89920121e-01 2.20520973e-01
-8.84204626e-01 8.15885484e-01 1.10147819e-01 1.98098466e-01
-3.62715244e-01 9.41212475e-02 5.44607222e-01 5.27667940e-01
5.63656747e-01 6.82279468e-01 -2.87689149e-01 -5.01790404e-01
-1.43574250e+00 7.23709643e-01 1.18478668e+00 9.17448103e-01
3.74562919e-01 -4.20454293e-01 -2.86470711e-01 8.57131898e-01
9.72220004e-02 4.40953702e-01 4.38348144e-01 -8.99865210e-01
1.26859677e+00 3.13564956e-01 1.68852147e-03 -9.47534502e-01
-6.27568215e-02 -2.65245438e-01 -7.46561170e-01 -3.79085481e-01
8.48166496e-02 -1.10548273e-01 -6.76285386e-01 1.77539790e+00
4.29692976e-02 -3.24292421e-01 9.61669683e-02 8.51855397e-01
1.47441268e-01 1.13984931e+00 -4.06477123e-01 -7.56542623e-01
1.19289744e+00 -9.41606462e-01 -5.74170828e-01 -6.15616739e-01
9.50456321e-01 -7.48287916e-01 1.01646101e+00 2.29274824e-01
-1.71749580e+00 -2.61194468e-01 -1.31033826e+00 -1.47552669e-01
8.15282315e-02 -1.60267130e-01 -1.02827922e-01 7.85949767e-01
-1.17834783e+00 8.64218533e-01 -6.10006392e-01 -1.50611892e-01
1.70575216e-01 3.25361460e-01 -1.78984314e-01 -3.05057079e-01
-8.69242191e-01 7.43162930e-01 7.04236031e-01 -1.87548816e-01
-6.17538810e-01 -5.58891237e-01 -7.50338197e-01 6.93208873e-01
3.29880804e-01 -7.73398459e-01 1.34249651e+00 -5.79608083e-01
-1.14725816e+00 7.27554083e-01 -5.56292295e-01 -9.72762764e-01
3.10604334e-01 -2.09821686e-01 1.07668608e-01 3.53890091e-01
8.08406770e-02 6.43008113e-01 6.50126278e-01 -8.93875003e-01
-6.62730157e-01 -3.24147433e-01 1.58118933e-01 4.21928018e-01
-4.89835858e-01 1.48610910e-02 -6.30212307e-01 -4.90899771e-01
1.22102365e-01 -9.00576353e-01 -2.55691767e-01 -3.59291762e-01
-5.49432278e-01 3.28135267e-02 6.66020930e-01 -1.02460563e+00
1.72139800e+00 -2.03345346e+00 4.31331933e-01 1.30515844e-01
1.48988843e-01 2.23273724e-01 -3.68400455e-01 6.38701022e-01
1.36712462e-01 4.53235507e-01 -7.10788310e-01 -8.17741394e-01
-2.80324072e-01 3.28576386e-01 -3.80582869e-01 2.89028108e-01
1.22873351e-01 8.09030235e-01 -7.91052997e-01 -6.18536830e-01
-1.56508461e-01 -9.56189334e-02 -9.85576093e-01 3.28646690e-01
-4.31741655e-01 -2.24311426e-01 -2.03907788e-01 2.49294996e-01
7.64167726e-01 -1.51867464e-01 3.84692460e-01 1.28855392e-01
3.19531769e-01 4.70018297e-01 -9.73807573e-01 1.49332345e+00
-5.48433423e-01 7.60651469e-01 7.94501007e-02 -1.17062795e+00
8.22500706e-01 1.13263093e-01 4.19031411e-01 -6.11045718e-01
3.90514210e-02 2.31065303e-01 -4.78356965e-02 -6.84028029e-01
1.04889452e+00 -2.56734192e-01 -1.41422197e-01 8.60496879e-01
-1.94239095e-01 -4.34781939e-01 8.68201852e-01 5.71909666e-01
1.20592678e+00 -4.47843403e-01 3.18921387e-01 -1.13878518e-01
4.05198216e-01 -1.79140821e-01 2.49381170e-01 1.01418483e+00
1.21021867e-01 8.03963065e-01 7.68784046e-01 1.22230090e-02
-1.69684136e+00 -5.28056145e-01 8.68063346e-02 8.28831434e-01
-4.18366790e-02 -8.06513667e-01 -1.16960669e+00 -6.21474147e-01
-6.38026297e-02 1.01031256e+00 -2.26790503e-01 -2.78155714e-01
-8.37886453e-01 -6.19235098e-01 3.77985716e-01 8.82484391e-02
2.12328270e-01 -9.48328078e-01 -1.06680954e+00 2.61970788e-01
-6.38871133e-01 -1.09558010e+00 -9.86537099e-01 1.87095851e-01
-9.90286350e-01 -7.81275749e-01 -4.18078899e-01 -7.23208189e-01
6.75954640e-01 3.84170443e-01 1.10976005e+00 4.31100637e-01
-7.79763386e-02 -4.58939150e-02 -6.85022116e-01 -1.16084464e-01
-7.94125974e-01 3.09436381e-01 -3.29939604e-01 -1.88050538e-01
-5.93310501e-03 -3.95575970e-01 -3.52074236e-01 -2.67797559e-01
-1.23915935e+00 1.06621265e-01 5.35424769e-01 7.24321067e-01
6.86126292e-01 3.10798019e-01 2.61684507e-01 -9.31663215e-01
1.08398974e+00 -3.48383486e-01 -4.08460021e-01 5.59320331e-01
-8.26167881e-01 2.76034296e-01 1.01859832e+00 -1.59752760e-02
-5.24258494e-01 -6.56054392e-02 -4.07617480e-01 -2.52531826e-01
2.87608474e-01 8.31783473e-01 -1.00188047e-01 4.20112550e-01
4.57133800e-01 7.02645481e-01 2.62698203e-01 -2.71344364e-01
1.93673000e-01 1.04915702e+00 6.72702074e-01 -4.86846924e-01
4.40933853e-01 -7.44513199e-02 6.17182404e-02 -6.99277759e-01
-8.95831347e-01 -3.57851416e-01 -4.57669377e-01 -2.34962013e-02
7.16015697e-01 -4.34699625e-01 -4.55093324e-01 -5.24205063e-03
-1.26857400e+00 -7.52644315e-02 -4.38701868e-01 1.10490419e-01
-7.89689839e-01 7.04019845e-01 -4.20124620e-01 -9.29602206e-01
-8.23435068e-01 -1.24866784e+00 1.15163219e+00 -2.04502605e-02
-1.81404978e-01 -6.59381986e-01 3.39204632e-02 4.01687801e-01
1.94554001e-01 1.75630227e-02 1.09428430e+00 -6.31444752e-01
-2.92628437e-01 -5.18339992e-01 -2.45953519e-02 4.26695526e-01
-1.50846571e-01 -1.18421465e-01 -3.57521087e-01 -7.42961168e-01
3.93470645e-01 -3.92373204e-01 8.24060917e-01 3.31106782e-01
1.51375842e+00 -1.04627717e+00 -1.07130125e-01 8.00545335e-01
1.48238051e+00 2.87310004e-01 7.73473144e-01 3.08845222e-01
3.23266909e-02 4.97899622e-01 6.02408767e-01 6.24930859e-01
2.94635594e-01 6.71790183e-01 4.18463737e-01 5.86537480e-01
2.12841570e-01 -3.60634416e-01 2.66162574e-01 1.05276740e+00
3.12411338e-01 -7.81561613e-01 -6.41669869e-01 4.83622760e-01
-1.71057117e+00 -1.02582133e+00 1.97266951e-01 2.42047572e+00
9.04473424e-01 3.48217189e-01 6.86752573e-02 4.16425586e-01
7.05563545e-01 3.35613698e-01 -5.02725780e-01 -9.08137918e-01
-1.30803928e-01 1.01505406e-01 4.41083103e-01 7.29409397e-01
-8.04310560e-01 4.58104372e-01 6.35450840e+00 8.27994287e-01
-9.15833354e-01 -1.73499748e-01 9.15491700e-01 -5.93981385e-01
-5.07400453e-01 2.73974299e-01 -6.99261844e-01 8.12824607e-01
1.46787798e+00 -7.76252508e-01 9.98322606e-01 6.02131605e-01
3.85870874e-01 -1.71093807e-01 -1.06462395e+00 7.72027314e-01
2.05714986e-01 -1.43684483e+00 7.87340254e-02 1.96396589e-01
4.99413639e-01 -3.24994504e-01 -3.38922381e-01 2.07454264e-01
-1.57757968e-01 -1.07249880e+00 9.56333578e-01 3.30689222e-01
7.86803842e-01 -7.69736469e-01 7.44215488e-01 8.31994951e-01
-8.49373341e-01 -2.23063707e-01 -5.53608358e-01 1.18078545e-01
6.48481131e-01 8.47164690e-01 -6.79449022e-01 6.10254884e-01
4.85009462e-01 4.63698268e-01 -4.32797134e-01 9.53934610e-01
7.42560402e-02 6.33753419e-01 -3.91058832e-01 -2.49462232e-01
1.28618360e-01 -2.92474657e-01 6.31525576e-01 1.53416479e+00
5.11663616e-01 1.65290728e-01 4.35456708e-02 6.12677157e-01
-3.20830107e-01 2.05194384e-01 -3.86820674e-01 -4.00231093e-01
7.52914846e-01 8.62685740e-01 -6.13499045e-01 -3.86454344e-01
-1.06105432e-01 1.04087126e+00 5.11276305e-01 -9.63958353e-02
-5.78482628e-01 -7.52073884e-01 2.03257576e-01 1.57983769e-02
4.52295572e-01 -5.77680096e-02 -7.44064450e-01 -1.08190179e+00
4.34915751e-01 -1.08913374e+00 4.14699316e-01 -6.55055940e-01
-7.16735423e-01 4.73002017e-01 -1.26896976e-02 -1.07594395e+00
-3.79079849e-01 -1.00165710e-01 -5.92313826e-01 9.28132355e-01
-1.09412372e+00 -4.21712905e-01 2.33835867e-03 4.12535407e-02
8.16389382e-01 2.21693162e-02 5.82368910e-01 1.18925326e-01
-4.35968250e-01 8.96512568e-01 2.88162351e-01 -3.61654252e-01
8.28605220e-02 -1.13352418e+00 4.34492171e-01 1.11148322e+00
2.30716486e-02 5.48081040e-01 9.61776972e-01 -6.06898963e-01
-1.60380006e+00 -9.86073792e-01 1.44648063e+00 1.87880307e-01
1.79156870e-01 -3.36120546e-01 -8.29801023e-01 3.63000214e-01
-3.90072324e-04 -5.03142118e-01 4.09772962e-01 -2.10290954e-01
2.57943533e-02 -1.87538743e-01 -1.36576235e+00 5.59435725e-01
7.89379299e-01 -3.51709068e-01 -5.35865784e-01 6.61597729e-01
1.14500213e+00 -6.22782648e-01 -7.65549004e-01 1.33779526e-01
4.08584833e-01 -8.57091784e-01 6.85111225e-01 -7.02108562e-01
1.27780819e+00 4.03983369e-02 -4.84243453e-01 -1.23937178e+00
-1.54095381e-01 -6.20590687e-01 -4.81777877e-01 8.78457844e-01
5.32699943e-01 -1.59891471e-01 8.49125683e-01 5.98123908e-01
-2.86529660e-01 -1.26752639e+00 -1.02333808e+00 -7.72005618e-01
1.59808517e-01 -4.84488994e-01 9.20745492e-01 2.59039789e-01
3.36260974e-01 4.83640581e-01 -4.78893876e-01 -1.08358696e-01
4.33475196e-01 3.23679745e-01 6.81575894e-01 -3.80228102e-01
-7.87314355e-01 -5.07070899e-01 -4.79562767e-02 -1.49848628e+00
-4.99439240e-03 -1.07963920e+00 3.18666846e-01 -1.55127108e+00
6.60358727e-01 -1.64676949e-01 3.51207972e-01 2.52704829e-01
-4.44667101e-01 -1.43101960e-01 7.03186870e-01 3.49476874e-01
-5.25243163e-01 5.11321723e-01 1.12071347e+00 -2.68395543e-01
-2.36955762e-01 1.34626791e-01 -9.07936335e-01 -5.72776087e-02
7.00484335e-01 -5.57247460e-01 -4.39508855e-01 -7.04546630e-01
3.72482061e-01 8.11904132e-01 -1.50189221e-01 -7.87245989e-01
4.09722477e-01 -2.03952700e-01 -1.79899797e-01 -7.06571639e-01
1.16322592e-01 -5.30863285e-01 -3.29198390e-02 7.36065447e-01
-7.72290528e-01 3.82060319e-01 -1.40139624e-01 5.39487123e-01
-1.85188800e-01 -9.98043895e-01 7.99551249e-01 -8.10467526e-02
8.23210403e-02 8.15061033e-02 -1.99049428e-01 1.07839137e-01
9.34540927e-01 -2.59329498e-01 -3.75401527e-01 -5.62484264e-01
-4.73884344e-01 4.46697414e-01 5.05195677e-01 1.14997216e-01
7.81487942e-01 -8.61948192e-01 -1.01821923e+00 1.21693030e-01
-1.24176309e-01 1.79994836e-01 1.38952523e-01 6.11853659e-01
-7.32831478e-01 7.11599529e-01 2.54338056e-01 -4.07676935e-01
-1.45432186e+00 7.33834267e-01 1.67969629e-01 -8.10998142e-01
-6.70167267e-01 6.94990933e-01 -3.41744781e-01 -2.40525991e-01
1.16833322e-01 -5.01801819e-02 9.37473476e-02 -3.86941820e-01
4.98060882e-01 2.72313595e-01 3.03940684e-01 -3.40891540e-01
-1.71330441e-02 1.76705822e-01 -3.04166436e-01 -2.90283442e-01
1.41499579e+00 -2.67696112e-01 -4.25824702e-01 -7.94832110e-02
1.64099658e+00 -8.73394310e-02 -8.16301525e-01 1.96668040e-02
-5.41049056e-02 -8.47599268e-01 -8.83209631e-02 -5.16220450e-01
-1.02381921e+00 6.07672393e-01 3.56354229e-02 7.34448373e-01
1.43806553e+00 -2.11916536e-01 1.12129414e+00 4.97886300e-01
1.97181523e-01 -9.94166613e-01 -5.73133193e-02 6.41005278e-01
8.07938099e-01 -6.77386999e-01 3.14118505e-01 -1.74092114e-01
-6.72631681e-01 1.27419043e+00 2.64192075e-01 6.07546456e-02
3.35489899e-01 3.13317209e-01 -6.33539021e-01 1.02094203e-01
-1.17589259e+00 2.90051430e-01 5.96345514e-02 1.17324917e-02
3.72876942e-01 7.65410066e-02 -8.77145350e-01 2.92667150e-01
-9.16453600e-01 -5.17480560e-02 6.67995870e-01 1.04691386e+00
-7.57820606e-01 -1.12996650e+00 -1.71592742e-01 1.11300004e+00
-5.39496601e-01 -1.06767744e-01 -1.62869155e-01 2.09543761e-02
-2.01853722e-01 1.21355414e+00 1.87942073e-01 -6.27453685e-01
2.74136513e-01 -1.13914674e-02 2.61281788e-01 -5.90577722e-01
-4.79045033e-01 -1.17592812e-01 1.22567445e-01 -2.58130014e-01
2.75025032e-02 -7.94383883e-01 -8.95244658e-01 -7.85324693e-01
-3.92968863e-01 3.55546206e-01 7.92443573e-01 1.10161626e+00
3.79151016e-01 2.83084929e-01 9.62424755e-01 -4.80358809e-01
-1.08725572e+00 -9.83944178e-01 -4.90679264e-01 4.66001779e-01
3.76821250e-01 3.90224010e-01 -3.86774093e-01 1.35754213e-01]
|
[12.173003196716309, 9.190479278564453]
|
1a134823-5b8e-45ea-a6ae-330daae16fa2
|
focal-loss-for-dense-object-detection
|
1708.02002
| null |
http://arxiv.org/abs/1708.02002v2
|
http://arxiv.org/pdf/1708.02002v2.pdf
|
Focal Loss for Dense Object Detection
|
The highest accuracy object detectors to date are based on a two-stage
approach popularized by R-CNN, where a classifier is applied to a sparse set of
candidate object locations. In contrast, one-stage detectors that are applied
over a regular, dense sampling of possible object locations have the potential
to be faster and simpler, but have trailed the accuracy of two-stage detectors
thus far. In this paper, we investigate why this is the case. We discover that
the extreme foreground-background class imbalance encountered during training
of dense detectors is the central cause. We propose to address this class
imbalance by reshaping the standard cross entropy loss such that it
down-weights the loss assigned to well-classified examples. Our novel Focal
Loss focuses training on a sparse set of hard examples and prevents the vast
number of easy negatives from overwhelming the detector during training. To
evaluate the effectiveness of our loss, we design and train a simple dense
detector we call RetinaNet. Our results show that when trained with the focal
loss, RetinaNet is able to match the speed of previous one-stage detectors
while surpassing the accuracy of all existing state-of-the-art two-stage
detectors. Code is at: https://github.com/facebookresearch/Detectron.
|
['Piotr Dollár', 'Tsung-Yi Lin', 'Priya Goyal', 'Ross Girshick', 'Kaiming He']
|
2017-08-07
|
focal-loss-for-dense-object-detection-1
|
http://openaccess.thecvf.com/content_iccv_2017/html/Lin_Focal_Loss_for_ICCV_2017_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2017/papers/Lin_Focal_Loss_for_ICCV_2017_paper.pdf
|
iccv-2017-10
|
['dense-object-detection']
|
['computer-vision']
|
[ 1.80551827e-01 4.20612469e-02 -2.24547848e-01 -2.84700036e-01
-7.00673759e-01 -2.67953068e-01 6.11318529e-01 1.66595921e-01
-6.62768662e-01 3.66626918e-01 -1.93207204e-01 -2.34187797e-01
2.72507668e-01 -6.54035926e-01 -7.47452676e-01 -4.20514733e-01
2.30971184e-02 4.28905308e-01 7.76907504e-01 1.59406513e-01
1.28164560e-01 5.66533744e-01 -1.78614414e+00 4.29523706e-01
5.63137829e-01 8.90336990e-01 6.19195439e-02 6.39245808e-01
1.57064572e-01 8.39077532e-01 -6.01086378e-01 -4.66042608e-01
4.76381928e-01 -3.41280937e-01 -4.30702925e-01 -7.52075985e-02
8.12086225e-01 -5.75119734e-01 -3.98418993e-01 9.99122977e-01
5.26572406e-01 -2.12306902e-01 4.82107788e-01 -1.14642429e+00
-4.28891212e-01 4.38657671e-01 -1.00592279e+00 7.92428136e-01
5.42478599e-02 3.33562553e-01 1.13340771e+00 -1.26522970e+00
4.50260699e-01 1.16646707e+00 1.08541846e+00 5.80022812e-01
-1.18660343e+00 -6.98580623e-01 3.08680415e-01 9.09141824e-02
-1.58558559e+00 -5.41989863e-01 4.91684824e-01 -4.37496185e-01
1.01336014e+00 1.13275379e-01 5.67496419e-01 8.31487656e-01
8.92203674e-02 9.73087788e-01 8.66373301e-01 -4.94816482e-01
1.13055825e-01 3.67161423e-01 2.15893701e-01 8.30869377e-01
7.87161827e-01 1.86429307e-01 -3.75328034e-01 -1.49046466e-01
5.94374359e-01 1.92235976e-01 -8.01062956e-03 -5.41706264e-01
-7.12398648e-01 9.34737682e-01 7.74417043e-01 2.44911611e-01
-2.84915984e-01 1.99062809e-01 2.15568423e-01 -1.11289471e-02
5.83890378e-01 4.69454318e-01 -2.04514518e-01 1.89718544e-01
-1.17860603e+00 1.90107077e-01 6.15262985e-01 5.66340864e-01
6.61440432e-01 -2.62201279e-01 -2.45761663e-01 5.67312002e-01
3.27053607e-01 1.75625280e-01 3.91536653e-01 -5.48630297e-01
2.60780036e-01 8.62904191e-01 -5.48459999e-02 -7.91056275e-01
-3.47990304e-01 -9.04448390e-01 -4.57182229e-01 4.69662428e-01
6.36679530e-01 -8.86156410e-02 -1.13699591e+00 1.34854364e+00
3.77018034e-01 9.61283594e-02 -4.47163761e-01 8.72765064e-01
4.28717494e-01 3.63234371e-01 1.89859748e-01 1.49847001e-01
1.24252558e+00 -7.69928217e-01 -2.68218249e-01 -5.76791823e-01
6.62688375e-01 -6.00077450e-01 8.74620914e-01 2.66775846e-01
-1.09180105e+00 -5.01365721e-01 -1.17724800e+00 1.20312884e-01
-3.76284242e-01 2.66446322e-01 5.74639678e-01 8.36346745e-01
-9.92120385e-01 6.19184136e-01 -8.89337659e-01 -3.96109045e-01
9.88387942e-01 3.25512886e-01 -8.47249404e-02 -2.47632772e-01
-8.05395544e-01 1.28472590e+00 1.32617638e-01 2.48833112e-02
-9.24616158e-01 -8.15474391e-01 -5.77511966e-01 7.61591122e-02
3.39316219e-01 -5.20969987e-01 1.44015229e+00 -1.11388338e+00
-8.34091544e-01 1.07084036e+00 -2.85432078e-02 -7.48004973e-01
7.36318052e-01 -4.36125070e-01 -1.23776130e-01 1.16414189e-01
1.46647081e-01 7.76037157e-01 7.12554276e-01 -9.87754822e-01
-8.77955437e-01 -3.24930459e-01 -7.81928450e-02 8.31782818e-02
-3.15570027e-01 2.21219823e-01 -3.58296394e-01 -3.63106757e-01
-1.16623089e-01 -6.86883807e-01 -3.88270229e-01 4.67872769e-01
-2.66589642e-01 -1.66618869e-01 7.50702143e-01 -3.10557693e-01
1.15574336e+00 -2.29084229e+00 -4.85745013e-01 2.32362464e-01
5.21559596e-01 6.26912177e-01 -2.65351161e-02 6.41270652e-02
-5.27682342e-02 1.58968702e-01 -1.96475223e-01 -3.29293221e-01
-1.73886970e-01 -1.46667421e-01 -1.79433897e-01 8.43799412e-01
6.75850213e-01 6.71313345e-01 -8.88303757e-01 -3.96925896e-01
1.25307292e-01 4.40582365e-01 -6.78082049e-01 -8.39088857e-02
-9.82287377e-02 -2.29878083e-01 -3.51149559e-01 7.11014569e-01
6.83872700e-01 -5.49789608e-01 -3.98558490e-02 4.16425131e-02
-1.27804279e-01 3.35637301e-01 -1.24177480e+00 1.15833354e+00
-1.37287974e-01 8.36630285e-01 -1.23969384e-01 -7.90161967e-01
7.77362168e-01 -9.76223424e-02 2.67450035e-01 -7.67133951e-01
1.76785842e-01 3.54697049e-01 2.76486874e-01 -8.18937272e-02
1.31972507e-01 -1.32204965e-01 2.25601912e-01 1.21442765e-01
-1.21612422e-01 2.90595293e-01 1.16840228e-01 2.49178022e-01
1.58331716e+00 -7.13323802e-02 3.51500630e-01 -1.07222795e-01
1.19786352e-01 1.59463525e-01 6.31902158e-01 1.16749859e+00
-4.75129783e-01 7.69479215e-01 4.18429047e-01 -5.56552172e-01
-1.06099367e+00 -1.06571519e+00 -3.33452851e-01 1.02139723e+00
1.40428888e-02 -1.67662039e-01 -5.88427246e-01 -7.96478093e-01
4.50107187e-01 5.28123677e-01 -6.59302831e-01 -3.02056432e-01
-4.41203475e-01 -1.01247454e+00 7.20955372e-01 6.81051850e-01
4.76566434e-01 -8.20084870e-01 -9.99968767e-01 9.02438238e-02
3.43639493e-01 -9.40176010e-01 -1.98033467e-01 4.48407501e-01
-7.69462466e-01 -1.20156956e+00 -6.88381016e-01 -5.59701204e-01
8.36656153e-01 3.41018856e-01 1.45036745e+00 3.65678906e-01
-7.87232161e-01 1.43348813e-01 -2.67664462e-01 -7.48036981e-01
-1.32404327e-01 1.31536022e-01 -9.05688554e-02 -5.13346456e-02
7.78430700e-01 -2.19229385e-01 -8.44304204e-01 2.26494431e-01
-6.37994528e-01 -2.19855770e-01 9.03175592e-01 7.64583051e-01
4.13276970e-01 -1.55959532e-01 5.34273684e-01 -1.02098215e+00
2.47196808e-01 -5.70502818e-01 -7.27787554e-01 -4.22137510e-03
-5.84227383e-01 -1.05653457e-01 2.39849716e-01 -5.87318242e-01
-7.42324173e-01 5.10270357e-01 -3.49883251e-02 -6.11388803e-01
3.15757021e-02 6.38708174e-02 1.92677945e-01 -2.42745534e-01
1.07476592e+00 -1.70903444e-01 -8.01341534e-02 -4.54378545e-01
2.13519245e-01 3.55189800e-01 2.82494426e-01 -6.88932240e-02
9.12989676e-01 5.70219219e-01 -1.77755386e-01 -5.73259532e-01
-1.06456900e+00 -7.75515974e-01 -4.31190073e-01 -2.94630200e-01
5.73055744e-01 -1.12873864e+00 -2.92922258e-01 3.45409155e-01
-9.97972488e-01 -3.49586308e-01 -6.60988629e-01 3.17073524e-01
-1.53092012e-01 -4.33898382e-02 -5.01182437e-01 -9.84537125e-01
-4.73381467e-02 -7.52030015e-01 9.15626466e-01 3.07396919e-01
-1.18174843e-01 -6.23500049e-01 1.22664042e-01 -1.70968827e-02
4.96051073e-01 1.97513252e-01 2.59158134e-01 -8.98849785e-01
-6.68800831e-01 -5.99833965e-01 -4.75756705e-01 3.66008580e-01
-1.69242740e-01 1.25227347e-01 -1.25695360e+00 -2.42911249e-01
-2.85323679e-01 -2.81708717e-01 1.46016943e+00 4.77848470e-01
1.01798630e+00 1.53391110e-02 -6.80774868e-01 3.99558276e-01
1.49111938e+00 -1.22369573e-01 5.95187247e-01 4.39017147e-01
5.06541312e-01 3.87433678e-01 4.61810291e-01 3.48732084e-01
1.84843153e-01 5.87138593e-01 4.47215170e-01 -6.06221735e-01
-3.26431036e-01 -2.45265439e-01 2.50324249e-01 -2.26532575e-02
2.94026703e-01 -2.31849283e-01 -1.20837986e+00 8.01815391e-01
-1.76973188e+00 -9.17324126e-01 9.14958800e-05 2.21948695e+00
7.38017380e-01 7.86031306e-01 4.35779184e-01 1.15074269e-01
8.09548795e-01 -2.13487763e-02 -5.75556159e-01 -1.72330678e-01
-1.62159204e-02 2.08786026e-01 7.79467285e-01 2.76768595e-01
-1.25562763e+00 7.66226411e-01 6.61631346e+00 5.87819397e-01
-9.67408955e-01 1.94514677e-01 8.77038360e-01 -5.88637829e-01
1.56760082e-01 -3.00237536e-02 -1.14607084e+00 3.75709862e-01
9.82352376e-01 1.11007743e-01 -6.17357418e-02 1.19995332e+00
-5.84977381e-02 -3.16193670e-01 -1.23294485e+00 7.28623033e-01
1.34339571e-01 -1.37136674e+00 -2.62761921e-01 -2.89705954e-02
5.17885745e-01 3.89581949e-01 1.77649289e-01 4.02019918e-01
3.98757219e-01 -9.71900821e-01 7.87619472e-01 2.83337563e-01
5.76033652e-01 -5.00318646e-01 7.88673222e-01 3.36980492e-01
-1.00687695e+00 -3.59057039e-01 -3.94981682e-01 -2.65882999e-01
-1.12730905e-01 1.01308286e+00 -1.12591434e+00 -1.09176815e-01
8.05757463e-01 4.75630820e-01 -9.32402849e-01 1.72292006e+00
-7.03258719e-03 6.60934746e-01 -5.78622222e-01 -1.57973543e-01
1.86380669e-01 5.38908005e-01 4.49847192e-01 1.50274277e+00
1.08000375e-01 -1.00273162e-01 1.71888471e-01 9.01485503e-01
-1.58746660e-01 -9.58388671e-02 -7.93921232e-01 2.55198359e-01
4.24143046e-01 1.17680371e+00 -9.20965254e-01 -4.35756028e-01
-5.59091568e-01 6.26696229e-01 4.42992389e-01 1.43067795e-03
-8.13115358e-01 -2.92478532e-01 4.57997143e-01 5.80958188e-01
5.57492912e-01 1.98981047e-01 -4.74328250e-01 -8.07128191e-01
2.00082824e-01 -6.09772325e-01 4.64959353e-01 -4.57031161e-01
-1.36597800e+00 3.23155910e-01 -1.81109443e-01 -1.10981333e+00
1.23071633e-01 -7.71654546e-01 -7.78945327e-01 6.47801578e-01
-1.70933843e+00 -7.08565354e-01 -2.71322340e-01 2.59232342e-01
5.00387907e-01 1.67535581e-02 2.53825486e-01 6.17091954e-01
-8.21670771e-01 7.56106794e-01 -8.07078257e-02 2.99983829e-01
5.82881629e-01 -1.23105049e+00 5.58224976e-01 1.06036687e+00
1.99840009e-01 4.69125748e-01 6.64450526e-01 -6.08983815e-01
-9.56285775e-01 -1.15488541e+00 8.05992663e-01 -7.00552106e-01
4.87492472e-01 -4.89252687e-01 -9.52417672e-01 5.77176154e-01
-2.73971140e-01 4.44165170e-01 3.05534780e-01 9.35112163e-02
-5.60408235e-01 -1.33787617e-01 -1.23114884e+00 4.40966487e-01
9.54069555e-01 -2.80087173e-01 -4.52391595e-01 3.93861383e-01
3.51935506e-01 -2.03093126e-01 -1.70133352e-01 4.70808983e-01
5.12273371e-01 -1.09418237e+00 8.83327484e-01 -4.33701515e-01
3.01794142e-01 -3.25199813e-01 1.25765473e-01 -9.02446091e-01
-4.96885329e-01 -2.80300617e-01 -7.80142099e-02 9.41934228e-01
6.87928438e-01 -6.28453434e-01 9.83839154e-01 4.26434666e-01
-4.11340455e-03 -8.31022203e-01 -8.49296093e-01 -7.85500109e-01
-6.06009327e-02 -2.19880819e-01 1.19799323e-01 6.66396439e-01
-3.32908571e-01 2.01726750e-01 -5.47796208e-03 2.29362875e-01
7.08439052e-01 -3.23803842e-01 4.95400518e-01 -1.35581946e+00
-2.48352230e-01 -6.80738389e-01 -6.12179220e-01 -8.55719805e-01
-3.03200811e-01 -7.42034793e-01 2.66171336e-01 -1.34679115e+00
4.96412903e-01 -6.46899760e-01 -4.65710014e-01 5.47349215e-01
-3.36503178e-01 6.01201594e-01 2.47943938e-01 2.73197263e-01
-8.11509848e-01 1.95839316e-01 7.62719512e-01 2.38897745e-03
-1.32633194e-01 6.63800091e-02 -7.81396866e-01 8.84214997e-01
8.09539318e-01 -7.92301238e-01 -7.21363127e-02 -2.00344115e-01
1.92788884e-01 -5.78360438e-01 6.81994379e-01 -1.35726929e+00
3.71416181e-01 1.56833351e-01 7.92575657e-01 -5.33278406e-01
1.90696090e-01 -6.38979435e-01 -2.54517138e-01 7.14526117e-01
-3.72060388e-01 5.24137877e-02 3.75690103e-01 5.55334389e-01
2.01420635e-02 -2.04571739e-01 1.06432140e+00 -1.86781257e-01
-7.11162746e-01 7.85246193e-02 -2.79283613e-01 9.07733515e-02
1.18353653e+00 -4.04688030e-01 -4.08143312e-01 9.23110396e-02
-2.99864888e-01 1.51233315e-01 5.45323431e-01 2.85823554e-01
5.97154617e-01 -1.01593232e+00 -7.57416964e-01 1.54688939e-01
8.57865140e-02 -5.94512932e-02 -1.20272346e-01 8.29521060e-01
-5.11426747e-01 1.92056686e-01 -1.97365835e-01 -7.29513645e-01
-1.12581861e+00 4.48308498e-01 6.33344233e-01 -2.66911030e-01
-8.34115028e-01 1.14829588e+00 3.02327842e-01 -6.09550020e-03
4.28825855e-01 -1.42137632e-01 3.96305732e-02 -9.44857579e-03
7.16621101e-01 2.83114463e-01 2.01037854e-01 -2.39134088e-01
-7.40568519e-01 2.11240843e-01 -4.53249305e-01 1.49817273e-01
1.21128333e+00 3.59063596e-01 3.14764708e-01 2.50875235e-01
9.10682678e-01 -1.17107585e-01 -1.46997249e+00 -1.96506277e-01
1.98581904e-01 -6.14472628e-01 2.63680995e-01 -7.05019653e-01
-1.06283379e+00 7.63246298e-01 9.31497991e-01 2.15388939e-01
1.01491117e+00 1.93539143e-01 3.72926950e-01 1.90417364e-01
-1.81590747e-02 -1.03090751e+00 2.73461133e-01 3.87304336e-01
4.27364320e-01 -1.29687572e+00 2.29100823e-01 -3.18681240e-01
-4.62498754e-01 7.37125158e-01 9.04397368e-01 -5.39201915e-01
5.92745066e-01 5.44741392e-01 -4.62183878e-02 -2.89975554e-01
-9.00214732e-01 -4.33222443e-01 2.37398058e-01 5.29249787e-01
4.23230976e-01 -4.13628444e-02 3.91915925e-02 2.07597822e-01
1.99392438e-01 2.44508144e-02 4.20682818e-01 1.05385041e+00
-7.80982196e-01 -6.15235567e-01 -4.18874919e-01 7.70049095e-01
-5.87400019e-01 -9.12902355e-02 -4.56982940e-01 8.51507425e-01
3.01374733e-01 7.30992615e-01 3.44243407e-01 -2.34979510e-01
5.25758743e-01 -4.27083075e-02 3.18446279e-01 -8.15360963e-01
-7.16021240e-01 -2.41008401e-01 1.31625216e-02 -6.88085020e-01
-1.65539876e-01 -7.31112659e-01 -9.36357498e-01 -8.06717575e-02
-6.52605355e-01 -3.03611934e-01 4.91959184e-01 7.12654054e-01
3.23560685e-01 4.76893157e-01 4.87395763e-01 -9.94872689e-01
-9.33584571e-01 -7.60402501e-01 -3.34221750e-01 2.71088243e-01
6.55771255e-01 -8.65874052e-01 -5.00032485e-01 -2.45906413e-01]
|
[9.176859855651855, 1.1036165952682495]
|
42947ce2-3140-4072-898f-93c8ee86cb3d
|
multiple-kernel-k-means-clustering-using-min
|
1803.02458
| null |
https://arxiv.org/abs/1803.02458v2
|
https://arxiv.org/pdf/1803.02458v2.pdf
|
Robust Multiple Kernel k-means Clustering using Min-Max Optimization
|
Multiple kernel learning is a type of multiview learning that combines different data modalities by capturing view-specific patterns using kernels. Although supervised multiple kernel learning has been extensively studied, until recently, only a few unsupervised approaches have been proposed. In the meanwhile, adversarial learning has recently received much attention. Many works have been proposed to defend against adversarial examples. However, little is known about the effect of adversarial perturbation in the context of multiview learning, and even less in the unsupervised case. In this study, we show that adversarial features added to a view can make the existing approaches with the min-max formulation in multiple kernel clustering yield unfavorable clusters. To address this problem and inspired by recent works in adversarial learning, we propose a multiple kernel clustering method with the min-max framework that aims to be robust to such adversarial perturbation. We evaluate the robustness of our method on simulation data under different types of adversarial perturbations and show that it outperforms several compared existing methods. In the real data analysis, We demonstrate the utility of our method on a real-world problem.
|
['Yao-Liang Yu', 'Wei Wu', 'Seojin Bang']
|
2018-03-06
| null | null | null | null |
['multiview-learning']
|
['computer-vision']
|
[ 1.27048030e-01 -1.64827839e-01 -1.85259044e-01 -2.13774100e-01
-8.64484549e-01 -8.18163037e-01 5.53994536e-01 2.43863225e-01
-1.10477749e-02 4.92644250e-01 8.89447704e-02 1.25443414e-01
-2.30893955e-01 -4.87005681e-01 -8.15724850e-01 -1.02406824e+00
-1.24773189e-01 -4.09958884e-02 3.97020221e-01 -1.08762175e-01
1.62841618e-01 3.90846819e-01 -1.22187841e+00 1.47703782e-01
8.55099738e-01 7.24737704e-01 -3.11361969e-01 4.10408765e-01
4.75124836e-01 8.81609976e-01 -7.29900062e-01 -4.99974221e-01
4.60159510e-01 -3.06002349e-01 -5.24586141e-01 2.31050253e-01
5.25515318e-01 -3.30511369e-02 -2.97955155e-01 1.21055317e+00
6.91340625e-01 2.26537749e-01 8.33904684e-01 -1.67542541e+00
-7.58590221e-01 4.96378064e-01 -7.24158466e-01 1.85366869e-02
1.85463607e-01 1.78900026e-02 5.30974805e-01 -6.33154750e-01
3.54962319e-01 1.32214344e+00 4.57922518e-01 5.69197595e-01
-1.34525919e+00 -6.81406498e-01 3.96641284e-01 2.99377799e-01
-1.22438943e+00 -1.37507960e-01 1.18754542e+00 -4.21465248e-01
2.91799784e-01 2.11345941e-01 1.26373306e-01 1.31362033e+00
3.96120965e-01 9.41083372e-01 1.46282697e+00 -2.26566181e-01
2.16849759e-01 4.59764302e-01 -1.92030430e-01 5.31165600e-01
2.14805678e-02 4.86987419e-02 -2.89561570e-01 -4.66660738e-01
2.29362354e-01 3.72786403e-01 -3.64353001e-01 -8.53119731e-01
-1.03071630e+00 9.10739839e-01 3.39436501e-01 1.65907238e-02
1.10246524e-01 -5.02847694e-03 6.75590158e-01 4.49327350e-01
5.74734151e-01 9.54835638e-02 -2.11291626e-01 4.75679755e-01
-5.86736858e-01 1.20335713e-01 6.76733851e-01 7.87229776e-01
5.76569080e-01 3.70714158e-01 4.49470021e-02 8.41706097e-01
1.41293883e-01 5.74082017e-01 3.82761925e-01 -7.35426724e-01
5.15604079e-01 4.61425304e-01 1.17503842e-02 -1.37380183e+00
-1.23806156e-01 -7.94587135e-02 -1.22430766e+00 6.51307702e-01
3.25589389e-01 -2.10274428e-01 -3.54762018e-01 1.80823648e+00
4.64551806e-01 5.83341360e-01 3.72346699e-01 8.28152478e-01
5.24670243e-01 5.31110287e-01 -8.57317299e-02 -4.69800174e-01
7.76381671e-01 -8.05080116e-01 -7.78489888e-01 1.11839816e-01
3.02030027e-01 -9.12691236e-01 8.73761237e-01 5.30773759e-01
-6.97397947e-01 -4.96228427e-01 -1.17261720e+00 6.79401577e-01
-5.59234560e-01 -3.88030767e-01 2.60162830e-01 8.70520473e-01
-5.27056277e-01 4.34576631e-01 -5.07843673e-01 -3.78727883e-01
1.96308210e-01 2.55798757e-01 -6.24580622e-01 -1.08316690e-01
-1.29258287e+00 8.10122490e-01 3.88783365e-01 -1.68405756e-01
-1.10060179e+00 -5.12543559e-01 -7.74694026e-01 -3.27987671e-01
8.94844234e-01 -5.22122860e-01 6.88269794e-01 -1.17946398e+00
-1.40449870e+00 5.35506845e-01 1.86599955e-01 -3.32831025e-01
5.67969501e-01 -3.56292367e-01 -6.82218075e-01 1.54174984e-01
-1.14285685e-01 -7.26449117e-02 1.42309046e+00 -1.77072084e+00
-1.39650613e-01 -4.99231696e-01 3.24830145e-01 2.50936955e-01
-4.42975551e-01 7.71584213e-02 -1.77680328e-01 -1.11044562e+00
-2.26934806e-01 -1.10043430e+00 -3.57402503e-01 -3.07954520e-01
-5.51252067e-01 -8.39615688e-02 1.36061370e+00 -2.47959837e-01
1.06815696e+00 -2.35017967e+00 4.36051100e-01 2.47180879e-01
-3.45423967e-02 3.11046064e-01 3.14230025e-01 7.08797812e-01
-3.60417902e-01 3.31010818e-01 -2.62748688e-01 -3.14657271e-01
-3.57366540e-02 2.54302382e-01 -6.00728273e-01 9.27769363e-01
-1.25728995e-01 4.50955540e-01 -7.85371423e-01 -6.99114382e-01
4.27644223e-01 2.70292580e-01 -3.16327721e-01 5.11707664e-01
1.87313050e-01 4.19480354e-01 -3.49622279e-01 7.86510170e-01
7.31538236e-01 4.04245481e-02 8.04557353e-02 -3.20244521e-01
3.19901079e-01 -1.10356510e+00 -1.62870646e+00 1.28945708e+00
-1.45316839e-01 3.14685017e-01 2.48554289e-01 -1.37633157e+00
4.98510033e-01 5.17431378e-01 5.52368224e-01 -1.41865447e-01
9.34633426e-03 -1.01336949e-01 -2.88954705e-01 -3.88329148e-01
1.13893971e-01 -3.99700940e-01 -3.83252919e-01 3.75752479e-01
1.16816118e-01 -1.08985491e-01 -3.78793448e-01 4.26853329e-01
9.29710150e-01 -9.24290493e-02 5.58140278e-01 -3.75288203e-02
7.19259858e-01 -3.59111220e-01 6.74014866e-01 8.02076995e-01
-4.18217927e-01 6.79616153e-01 3.59753609e-01 -3.29156578e-01
-8.85404527e-01 -1.28856170e+00 2.18345840e-02 8.44918430e-01
4.24762547e-01 -4.18217629e-01 -6.63671672e-01 -1.06036282e+00
1.93850964e-01 3.40418488e-01 -7.61272192e-01 -4.28775549e-01
-3.16086441e-01 -8.32207203e-01 7.03763485e-01 4.05761272e-01
4.35775489e-01 -8.45539153e-01 -9.17453319e-02 -6.56795651e-02
2.21521296e-02 -1.22229183e+00 -4.81305748e-01 6.32917881e-02
-5.96821606e-01 -1.26195931e+00 -5.26283383e-01 -5.52646697e-01
6.29466832e-01 2.07570598e-01 6.53715372e-01 -2.56512761e-01
-1.79671541e-01 9.92112517e-01 -5.08444309e-01 -4.90501016e-01
-6.50166631e-01 -1.86970130e-01 6.60688579e-01 5.95034063e-01
-2.90724218e-01 -6.69896245e-01 -2.82829404e-01 5.47795773e-01
-1.45597029e+00 -4.06330794e-01 4.90808576e-01 7.63414860e-01
4.30293322e-01 2.25555941e-01 8.80925000e-01 -1.16644597e+00
6.37646198e-01 -6.08539999e-01 -4.74510878e-01 4.61003304e-01
-6.08602941e-01 2.36985460e-03 1.29444480e+00 -7.43946791e-01
-9.30438876e-01 9.13543329e-02 3.77824485e-01 -1.05996060e+00
-5.07370889e-01 2.13352263e-01 -6.52443409e-01 -3.26423585e-01
6.16105616e-01 2.57649183e-01 -8.45279992e-02 -1.87209114e-01
6.10177815e-01 3.84557813e-01 4.20280010e-01 -5.65609872e-01
1.46735287e+00 7.82227695e-01 1.47789106e-01 -7.93086231e-01
-8.90603065e-01 -4.52493131e-01 -6.87700331e-01 -5.23400903e-01
9.35393453e-01 -7.19249725e-01 -8.01171958e-01 5.83675206e-01
-6.76299512e-01 9.96452495e-02 1.41915623e-02 3.72580618e-01
-8.85125279e-01 8.87508154e-01 -2.16920972e-01 -7.83991635e-01
-4.75743748e-02 -1.18184733e+00 4.92486000e-01 1.08096398e-01
1.74783036e-01 -1.23934591e+00 1.29564330e-01 3.71867478e-01
1.77646637e-01 8.91831517e-01 9.36413646e-01 -9.36863184e-01
-3.92511964e-01 -2.80888408e-01 3.57294172e-01 6.56248868e-01
3.22324336e-01 -1.28077388e-01 -9.42983687e-01 -6.51134908e-01
3.12978864e-01 -6.56822741e-01 6.76773071e-01 3.07907034e-02
1.23548174e+00 -5.23020625e-01 -2.15571597e-01 6.35957301e-01
1.66519737e+00 4.56976891e-03 3.63711208e-01 2.81356782e-01
9.63035941e-01 4.86433864e-01 7.43262470e-01 4.75805253e-01
-3.20995860e-02 5.59163332e-01 8.71869206e-01 8.11797660e-03
5.70348024e-01 3.92133594e-02 5.74276447e-01 7.04421699e-01
-9.39087272e-02 -2.83180207e-01 -6.10840738e-01 3.47491682e-01
-2.15416050e+00 -1.35422599e+00 3.18146586e-01 2.30403852e+00
5.68090379e-01 1.23343937e-01 5.07040992e-02 6.56870827e-02
7.96925306e-01 6.71959579e-01 -6.70301795e-01 -2.95755535e-01
-3.56701702e-01 -2.12353647e-01 5.21215916e-01 3.07496727e-01
-1.60454905e+00 8.16211224e-01 5.89817762e+00 9.40373659e-01
-9.12150681e-01 1.51912810e-03 3.50292802e-01 2.76852604e-02
-1.99866537e-02 5.42797968e-02 -4.68462527e-01 3.21686745e-01
4.78979856e-01 -2.02634230e-01 2.77475595e-01 1.15912545e+00
5.51495478e-02 -1.74754839e-02 -9.73853946e-01 9.83141601e-01
6.38787687e-01 -8.73342276e-01 2.53495544e-01 -5.60823679e-02
9.15121019e-01 -4.35565203e-01 3.15625250e-01 2.47224316e-01
4.13777888e-01 -1.00326836e+00 2.74122238e-01 6.40442967e-01
2.90781826e-01 -1.29300225e+00 7.08519757e-01 7.16259778e-01
-1.14660418e+00 -2.10195258e-01 -3.81935149e-01 4.65259194e-01
6.28674701e-02 4.34752643e-01 -5.45879126e-01 1.17359006e+00
6.61585212e-01 7.52856374e-01 -7.93346226e-01 7.40175724e-01
-3.81011609e-03 7.35247910e-01 -1.51919443e-02 4.94894534e-01
7.04260543e-02 -1.20981470e-01 8.58808398e-01 9.25377429e-01
-1.73792467e-01 -6.33150414e-02 6.27334476e-01 3.63815516e-01
3.63816656e-02 3.18649560e-01 -1.23200119e+00 3.17076027e-01
1.94513023e-01 1.20868587e+00 -4.88187611e-01 -2.51450270e-01
-6.73442245e-01 1.21156991e+00 4.28401619e-01 4.27466601e-01
-1.04547632e+00 -1.27725989e-01 6.07357383e-01 -1.82805628e-01
2.53014565e-01 -1.09608099e-01 1.40893191e-01 -1.35050488e+00
1.07264919e-02 -1.17619550e+00 6.68869078e-01 -3.55698258e-01
-1.75866628e+00 1.25754297e-01 2.59812176e-01 -1.75348699e+00
-1.37510642e-01 -4.07475859e-01 -7.99898803e-01 5.63001096e-01
-1.03585255e+00 -1.35267568e+00 -9.97500494e-02 1.14043903e+00
4.15418863e-01 -4.41940308e-01 9.13064897e-01 2.99765635e-02
-5.18792331e-01 7.91691244e-01 5.33405662e-01 2.58664519e-01
1.44406390e+00 -1.32181096e+00 -1.91004336e-01 9.47309256e-01
8.20197389e-02 5.85295081e-01 7.43114591e-01 -5.46384394e-01
-1.44307232e+00 -1.31289232e+00 -1.30961508e-01 -5.73052645e-01
5.89538097e-01 -2.70020187e-01 -1.01176405e+00 8.48738134e-01
5.42243898e-01 3.13747317e-01 1.06049883e+00 -1.90725446e-01
-6.93036377e-01 -2.85744995e-01 -1.26603007e+00 6.19422674e-01
5.30409038e-01 -4.52779889e-01 -5.82665801e-01 1.35491058e-01
6.33846939e-01 -1.28975883e-01 -1.15598464e+00 6.51248097e-01
3.86349678e-01 -1.13127756e+00 1.25046790e+00 -7.55280375e-01
1.17958158e-01 -6.23769701e-01 -4.06166703e-01 -1.73874664e+00
-2.65337497e-01 -4.36171174e-01 -1.69644326e-01 1.41258669e+00
-5.73699549e-02 -6.05398357e-01 4.40123260e-01 1.75251871e-01
9.37598795e-02 -6.26937985e-01 -9.68239903e-01 -1.06629384e+00
2.37673059e-01 -3.11760843e-01 1.38603121e-01 1.39979863e+00
-1.66856170e-01 8.10642838e-02 -1.00624204e+00 7.69961834e-01
1.05628765e+00 1.18792541e-01 1.12606144e+00 -9.49844480e-01
-3.05506378e-01 2.69519910e-03 -7.27351248e-01 -2.65706390e-01
5.43369651e-01 -7.80630887e-01 -3.57370794e-01 -1.02998841e+00
1.73935756e-01 -1.30066825e-02 -4.26150948e-01 2.43472785e-01
-4.93231148e-01 1.92340612e-02 4.35089290e-01 2.12473035e-01
-7.96900094e-01 5.05502105e-01 9.36365128e-01 -4.39237833e-01
2.26476759e-01 1.39632821e-01 -5.63470900e-01 9.99030352e-01
8.88125718e-01 -5.06677628e-01 -6.42226338e-01 1.75156087e-01
-4.64811884e-02 3.64371575e-02 4.58308727e-01 -1.22332680e+00
2.88466841e-01 -4.36261177e-01 4.58985060e-01 -5.73857546e-01
2.74543822e-01 -1.31266022e+00 1.57292545e-01 2.20666185e-01
-1.04946397e-01 1.57221615e-01 -1.28580607e-03 1.25905955e+00
-3.64396095e-01 1.34039400e-02 1.15337873e+00 -2.37985551e-01
-7.14636981e-01 3.33291799e-01 -3.29844713e-01 2.54516423e-01
1.59334338e+00 -3.88350561e-02 -2.72305548e-01 -4.46066022e-01
-9.19079185e-01 2.72881329e-01 6.24970853e-01 5.54992676e-01
7.29994535e-01 -1.59065139e+00 -5.69357157e-01 -4.56886031e-02
3.77581447e-01 -1.90145418e-01 5.41362107e-01 7.51867235e-01
1.27581134e-01 -1.38913929e-01 -1.40883699e-01 -4.73235935e-01
-1.56478667e+00 1.27961111e+00 2.74796605e-01 -3.18223983e-01
-3.76852125e-01 2.44096071e-01 3.84196311e-01 -4.32052970e-01
4.21201020e-01 1.77888647e-01 -4.04769152e-01 1.87731162e-01
3.29526454e-01 4.96264815e-01 -3.59457314e-01 -8.08336794e-01
-3.37169468e-01 7.64680922e-01 -6.26252741e-02 2.76615024e-02
1.03549016e+00 -8.94717872e-02 -1.91779472e-02 9.27223861e-01
1.29434550e+00 2.89882034e-01 -9.86137211e-01 -3.11379761e-01
-2.33881563e-01 -4.17104363e-01 -6.44569993e-01 -4.61428314e-01
-9.81358647e-01 7.47494340e-01 7.14190960e-01 4.83941287e-01
1.18116033e+00 -9.85213146e-02 3.97638023e-01 3.57989073e-01
3.11307728e-01 -1.20102167e+00 5.85655451e-01 3.07805508e-01
8.87309968e-01 -1.63894951e+00 9.35559049e-02 -4.52256083e-01
-9.29665625e-01 9.49783564e-01 7.80220747e-01 -3.53182614e-01
9.05388117e-01 1.55492216e-01 2.75032550e-01 6.95942119e-02
-3.96892250e-01 5.27907126e-02 2.06362471e-01 7.48856246e-01
-1.52480274e-01 7.48390406e-02 6.87623024e-02 4.88506436e-01
2.68515706e-01 -4.82804239e-01 4.99130040e-01 1.01816356e+00
-2.12369546e-01 -1.18803036e+00 -7.42440701e-01 1.95664391e-01
-8.56348455e-01 2.46216267e-01 -5.31184316e-01 8.35369706e-01
5.57416640e-02 9.79003787e-01 -6.25999928e-01 -6.17479324e-01
4.87383485e-01 2.43992776e-01 1.76053613e-01 -4.49756563e-01
-7.59321094e-01 -8.94697085e-02 -3.35203171e-01 -4.87444878e-01
-8.59865010e-01 -6.38088346e-01 -8.39532673e-01 -1.71359748e-01
-2.95376718e-01 1.71318322e-01 7.99993053e-02 8.80371928e-01
5.93662113e-02 3.44535351e-01 1.15616357e+00 -7.05355287e-01
-8.02067995e-01 -7.37114608e-01 -8.23083699e-01 9.09643173e-01
4.36534226e-01 -7.45240152e-01 -8.28593791e-01 2.26555049e-01]
|
[5.65727424621582, 7.814080715179443]
|
b081431f-2e7a-4166-9393-643b26795043
|
transductive-few-shot-learning-with-prototype
|
2304.11598
| null |
https://arxiv.org/abs/2304.11598v1
|
https://arxiv.org/pdf/2304.11598v1.pdf
|
Transductive Few-shot Learning with Prototype-based Label Propagation by Iterative Graph Refinement
|
Few-shot learning (FSL) is popular due to its ability to adapt to novel classes. Compared with inductive few-shot learning, transductive models typically perform better as they leverage all samples of the query set. The two existing classes of methods, prototype-based and graph-based, have the disadvantages of inaccurate prototype estimation and sub-optimal graph construction with kernel functions, respectively. In this paper, we propose a novel prototype-based label propagation to solve these issues. Specifically, our graph construction is based on the relation between prototypes and samples rather than between samples. As prototypes are being updated, the graph changes. We also estimate the label of each prototype instead of considering a prototype be the class centre. On mini-ImageNet, tiered-ImageNet, CIFAR-FS and CUB datasets, we show the proposed method outperforms other state-of-the-art methods in transductive FSL and semi-supervised FSL when some unlabeled data accompanies the novel few-shot task.
|
['Piotr Koniusz', 'Hao Zhu']
|
2023-04-23
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Zhu_Transductive_Few-Shot_Learning_With_Prototype-Based_Label_Propagation_by_Iterative_Graph_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Zhu_Transductive_Few-Shot_Learning_With_Prototype-Based_Label_Propagation_by_Iterative_Graph_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['graph-construction']
|
['graphs']
|
[-2.93528233e-02 2.24834308e-01 -3.26955527e-01 -7.01551259e-01
-5.27150035e-01 -2.43209258e-01 5.49329996e-01 4.66729939e-01
-3.93305659e-01 8.48890543e-01 -4.65273969e-02 3.26635182e-01
-1.49088949e-01 -9.18312371e-01 -7.51545787e-01 -5.50279558e-01
-5.86757474e-02 8.83009911e-01 7.14314938e-01 -5.11723645e-02
4.69237454e-02 2.36911476e-01 -1.62796974e+00 4.95861098e-02
8.40333104e-01 8.54363084e-01 4.73988801e-02 4.26975578e-01
-3.41759920e-01 8.54903996e-01 -1.62499279e-01 -2.72887468e-01
8.49449411e-02 -5.58087766e-01 -7.70459116e-01 3.95538747e-01
4.71134067e-01 -7.91319087e-02 -3.36309552e-01 1.14376318e+00
4.27863955e-01 6.17544413e-01 8.81125033e-01 -1.40693760e+00
-8.04584861e-01 7.52499282e-01 -5.33513606e-01 1.67272225e-01
1.91046625e-01 -1.48646474e-01 9.76754725e-01 -1.26211619e+00
9.27373648e-01 1.20258141e+00 6.97028160e-01 6.36092544e-01
-1.10113871e+00 -5.41731775e-01 4.41316277e-01 7.21496582e-01
-1.49966264e+00 -5.54868340e-01 9.57261920e-01 -4.53867882e-01
6.48896933e-01 -2.23824590e-01 8.61469507e-01 7.82512367e-01
-4.10666823e-01 7.51098812e-01 1.08437181e+00 -6.58853352e-01
8.40320647e-01 4.26450461e-01 5.51892817e-01 1.03842354e+00
2.46519014e-01 -4.86996882e-02 -4.14796263e-01 -3.03851455e-01
3.82722825e-01 3.48220438e-01 -1.60899237e-01 -9.40208197e-01
-7.20258951e-01 9.98365521e-01 6.48171067e-01 2.83050656e-01
-3.05846930e-01 -1.00850789e-02 3.92444491e-01 2.23808661e-01
6.07724607e-01 1.54252574e-01 -7.55947307e-02 2.48551562e-01
-9.35068309e-01 -1.45732149e-01 8.74185622e-01 1.30699635e+00
1.31499255e+00 1.13102570e-01 -2.52564520e-01 1.03414786e+00
5.30369341e-01 -1.42538231e-02 7.08719730e-01 -3.65307838e-01
5.22017181e-02 8.85305524e-01 -3.09877485e-01 -6.65101230e-01
-3.13386947e-01 -3.40794295e-01 -5.57119727e-01 -2.19996691e-01
-2.51087211e-02 -1.44698232e-01 -1.49431765e+00 1.48930120e+00
6.70706809e-01 7.91000724e-01 -5.59868515e-02 6.91318870e-01
1.12213099e+00 5.62217236e-01 2.07087904e-01 -5.86972356e-01
8.48516047e-01 -1.18598926e+00 -6.69305265e-01 -2.48515546e-01
7.98065186e-01 -2.68508136e-01 9.58562970e-01 -2.06625946e-02
-4.29176718e-01 -5.97221851e-01 -1.13634121e+00 2.52693892e-01
-7.15408921e-01 -4.52893168e-01 7.37187803e-01 5.91337264e-01
-1.07915604e+00 5.29121637e-01 -5.64689577e-01 -9.09910083e-01
6.45398319e-01 1.62102014e-01 -2.28470877e-01 -5.72098017e-01
-1.08247268e+00 6.50442183e-01 6.89021230e-01 -2.79824227e-01
-9.31027293e-01 -6.03412926e-01 -1.06150019e+00 1.38253361e-01
6.17619693e-01 -3.80010337e-01 1.03949845e+00 -9.05064762e-01
-1.32651997e+00 6.83363974e-01 1.42945334e-01 -5.15470147e-01
1.92284197e-01 3.06734502e-01 -4.77414668e-01 2.68933207e-01
6.46969723e-03 6.90069616e-01 8.51240993e-01 -1.29564595e+00
-6.13142192e-01 -3.05133402e-01 -5.54955378e-02 2.93021411e-01
-3.76744986e-01 -5.50706506e-01 -3.82853925e-01 -3.03901196e-01
9.20329913e-02 -8.53489161e-01 -2.10600123e-01 1.99738480e-02
-4.15902555e-01 -5.85852027e-01 9.46610749e-01 2.55563885e-01
1.06409252e+00 -2.04829860e+00 -1.29242942e-01 7.55388513e-02
3.29057157e-01 5.75560391e-01 -1.52752087e-01 4.59864080e-01
1.08793154e-01 -2.05161467e-01 -1.67400584e-01 -2.05592722e-01
-9.06536430e-02 2.74596274e-01 -3.39266285e-02 4.83470321e-01
3.74978632e-02 9.68673110e-01 -1.31373167e+00 -8.63628209e-01
3.42599452e-01 2.41842493e-01 -3.02529246e-01 1.61070019e-01
-1.51055127e-01 -8.32922533e-02 -3.17449689e-01 8.48484397e-01
4.51408654e-01 -3.93041581e-01 9.35189947e-02 -2.71439433e-01
2.42750585e-01 -4.52842802e-01 -1.06419241e+00 1.76540720e+00
-1.42404497e-01 3.24704498e-01 -4.33830827e-01 -1.31242895e+00
1.10837901e+00 2.87271112e-01 2.91947901e-01 -3.14376056e-01
2.17072651e-01 1.06626702e-02 -7.94980675e-02 -3.62406790e-01
6.21747747e-02 -4.41754401e-01 2.08580509e-01 3.75895917e-01
8.32551539e-01 -9.75093246e-02 5.14476299e-01 4.83915091e-01
1.05778849e+00 -3.94694582e-02 7.34343112e-01 -1.63447917e-01
1.98133782e-01 3.52196470e-02 6.47633970e-01 1.01901829e+00
-5.38641274e-01 4.88163441e-01 8.61895010e-02 -3.80256623e-01
-6.33843660e-01 -1.16828167e+00 7.40827667e-03 1.21317399e+00
2.42068201e-01 -3.27416658e-01 -5.55503011e-01 -9.47872996e-01
1.06467977e-02 9.70015287e-01 -7.95327365e-01 -5.90331435e-01
-2.93474691e-03 -4.82558966e-01 5.61623788e-03 3.19418788e-01
4.80170310e-01 -9.88912463e-01 -2.29995504e-01 4.77842748e-01
2.21173525e-01 -8.37625146e-01 -4.44727421e-01 1.72296822e-01
-9.68684196e-01 -1.31581211e+00 -6.89799130e-01 -1.00282085e+00
1.05736470e+00 4.68031645e-01 8.38692546e-01 -1.27209365e-01
-4.72433865e-01 7.48427391e-01 -6.69331729e-01 -3.09396088e-01
-1.38858169e-01 -8.74378383e-02 5.44558875e-02 4.10969466e-01
6.68110132e-01 -5.70755243e-01 -4.63230044e-01 1.44986600e-01
-7.66040444e-01 -1.08464107e-01 4.87501413e-01 9.23716545e-01
8.54310751e-01 -1.59910291e-01 9.28621411e-01 -1.64936018e+00
6.34577215e-01 -7.90017068e-01 -2.25663617e-01 7.83269703e-01
-1.02891207e+00 -2.90515665e-02 6.23637021e-01 -8.31787705e-01
-1.12640285e+00 2.95907140e-01 4.37542260e-01 -8.00271332e-01
-1.74601719e-01 5.74680150e-01 2.03245610e-01 -2.69363701e-01
9.90693629e-01 1.13273263e-01 -2.84896463e-01 -3.59315693e-01
7.91906297e-01 5.62006176e-01 1.20162830e-01 -2.33929724e-01
6.85012996e-01 4.56533402e-01 -1.62135854e-01 -9.75058913e-01
-1.05835009e+00 -9.18778658e-01 -8.46378088e-01 -5.38477540e-01
4.92889285e-01 -8.32951665e-01 -1.84841171e-01 3.19846392e-01
-6.78377986e-01 -1.64638296e-01 -7.44069099e-01 7.33664393e-01
-5.01974642e-01 2.83399105e-01 -5.69655716e-01 -7.60609925e-01
-4.20406938e-01 -6.93462670e-01 6.03619456e-01 4.87917125e-01
1.30067796e-01 -1.17529881e+00 4.15093988e-01 -8.62720609e-02
2.47307420e-01 1.65176004e-01 8.57696891e-01 -1.00530195e+00
-2.18882948e-01 -6.37088537e-01 -2.34537944e-01 8.89941603e-02
2.70204931e-01 -5.93179315e-02 -1.05069685e+00 -4.91792560e-01
-1.67806134e-01 -7.91678786e-01 1.08053815e+00 3.39774340e-01
5.93709767e-01 -2.21260667e-01 -5.10950744e-01 4.31127280e-01
1.58938491e+00 1.08527377e-01 2.72635967e-01 -2.75006711e-01
8.96379352e-01 4.55520093e-01 8.42781246e-01 5.41022718e-01
4.26780105e-01 2.28205666e-01 1.16909102e-01 9.16921198e-02
-2.62271315e-01 -3.34351301e-01 -3.30977626e-02 1.16699159e+00
1.10163040e-01 -1.30447641e-01 -7.94134080e-01 5.51603675e-01
-2.19846749e+00 -8.11633706e-01 3.07775855e-01 2.12662220e+00
7.54982591e-01 1.91398248e-01 4.33458984e-02 -2.35155255e-01
1.08020222e+00 7.67147020e-02 -8.86673748e-01 -1.32205546e-01
1.64207309e-01 1.00312337e-01 3.46937746e-01 3.60988081e-01
-9.99067545e-01 1.23858011e+00 5.99722719e+00 8.57260108e-01
-9.72475290e-01 3.38237435e-01 5.29597402e-01 -5.43423779e-02
-5.55637591e-02 2.05798671e-01 -8.81591201e-01 6.62793443e-02
7.28577256e-01 -3.60449135e-01 3.72657150e-01 1.13038146e+00
-3.87281328e-01 -1.45494878e-01 -1.27534735e+00 1.06739688e+00
5.58333099e-01 -1.24471545e+00 5.41466996e-02 -3.89528871e-01
8.84237528e-01 2.42982194e-01 -3.56336325e-01 8.77608359e-01
5.12086511e-01 -4.84344631e-01 3.85995239e-01 7.08724737e-01
6.72719002e-01 -5.88402510e-01 5.10750353e-01 3.78798902e-01
-1.36910236e+00 -9.54274461e-03 -7.74116516e-01 3.46294306e-02
5.97587898e-02 6.63044751e-01 -1.23684359e+00 1.94055066e-01
3.71821523e-01 1.01656878e+00 -8.65426183e-01 1.35179532e+00
-4.10334378e-01 7.83001184e-01 -1.32280335e-01 -3.61993879e-01
2.60500222e-01 -1.72453225e-01 3.69293064e-01 9.77142334e-01
8.37847292e-02 1.73246101e-01 5.46799183e-01 8.31534386e-01
-2.58225709e-01 3.85081083e-01 -8.81820321e-01 -1.43753991e-01
6.94121838e-01 1.42317724e+00 -9.47361469e-01 -8.87464106e-01
-5.57064116e-01 7.38855600e-01 9.07309115e-01 5.12473524e-01
-5.48284054e-01 -5.05655587e-01 -2.38810703e-02 6.38436973e-02
3.62276495e-01 1.05439514e-01 3.81808668e-01 -1.13152599e+00
-4.56575006e-01 -1.60365835e-01 7.07037091e-01 -6.34476900e-01
-1.52183950e+00 5.66745758e-01 1.21369004e-01 -1.20346177e+00
-2.06295207e-01 -2.61752699e-02 -7.07466662e-01 1.81876644e-01
-1.45837283e+00 -1.16633570e+00 -3.82978112e-01 6.73654437e-01
7.43380666e-01 -1.84919804e-01 8.47568512e-01 2.23642081e-01
-4.56383526e-01 4.67834294e-01 1.66362002e-01 9.75908265e-02
8.14891636e-01 -1.09646368e+00 1.96164474e-01 6.67942286e-01
4.04585063e-01 3.67098987e-01 4.71474737e-01 -8.65687966e-01
-1.13180876e+00 -1.39654303e+00 5.79487741e-01 8.71303007e-02
7.08905101e-01 -4.93972242e-01 -1.05420661e+00 6.92558765e-01
-1.46109521e-01 7.55814433e-01 8.12846959e-01 1.21114247e-01
-4.47899997e-01 -2.34662145e-01 -1.23982549e+00 4.12336379e-01
1.18546605e+00 -5.03401339e-01 -7.03569472e-01 6.11209571e-01
9.07973170e-01 2.23748595e-01 -7.29366601e-01 3.30152541e-01
2.96362072e-01 -8.22384357e-01 7.08786964e-01 -6.79653108e-01
-3.56265634e-01 -1.92655027e-01 -6.39907047e-02 -1.73089612e+00
-7.68138528e-01 -1.68407857e-01 -3.86224568e-01 1.24235368e+00
3.14875573e-01 -6.75651312e-01 7.78525114e-01 4.53935623e-01
-1.75371543e-01 -6.89460754e-01 -8.72339308e-01 -1.01182806e+00
-5.26883006e-01 -3.05729862e-02 3.03648323e-01 1.19657719e+00
2.84171104e-01 8.52067590e-01 -3.85040402e-01 -3.08002204e-01
1.05759192e+00 9.99212041e-02 5.46855509e-01 -1.54347801e+00
-1.99254617e-01 9.52569302e-03 -7.95844555e-01 -3.92647862e-01
1.75187767e-01 -9.88571942e-01 2.87922084e-01 -1.71776819e+00
2.98403472e-01 -5.82092285e-01 -6.01183057e-01 6.26680493e-01
-1.33092925e-01 1.52952954e-01 5.71424924e-02 1.99206412e-01
-1.25832403e+00 7.09359467e-01 1.17117703e+00 -3.93517584e-01
-4.70069319e-01 -6.75200671e-02 -2.81304985e-01 6.56688333e-01
7.39320338e-01 -6.60459936e-01 -8.69227111e-01 1.77938998e-01
-3.48309167e-02 -1.67948708e-01 -5.12492470e-02 -1.12778580e+00
6.14300907e-01 -1.66204005e-01 2.52166212e-01 -5.17525971e-01
3.63087982e-01 -7.24514842e-01 3.51787582e-02 5.73520601e-01
-3.78401071e-01 -7.13003099e-01 -2.75996834e-01 1.22766650e+00
-5.91171868e-02 -4.96051937e-01 1.08754706e+00 -3.59415025e-01
-1.14958394e+00 8.62907708e-01 -1.81266144e-01 3.41797352e-01
1.42415309e+00 -3.55439991e-01 -4.16074991e-01 -1.66037336e-01
-7.95641005e-01 3.10206443e-01 3.26464117e-01 4.14609820e-01
9.29008305e-01 -1.51239944e+00 -5.60417116e-01 -2.37434283e-02
7.68215120e-01 1.91876933e-03 2.51499593e-01 9.43926990e-01
-1.99294835e-01 9.90482047e-02 -4.38597538e-02 -4.74862933e-01
-1.00823200e+00 1.08291078e+00 8.69832486e-02 8.14815983e-02
-6.13981605e-01 1.02284706e+00 2.32747674e-01 -5.44392705e-01
3.79775703e-01 3.51726338e-02 -4.64025825e-01 3.62633675e-01
4.49785769e-01 4.90611941e-01 -1.23097204e-01 -5.39831281e-01
-3.22812617e-01 3.44396144e-01 -3.67897630e-01 1.78925455e-01
1.33056247e+00 -9.76554528e-02 5.21437898e-02 1.17182863e+00
1.26770449e+00 -7.16279209e-01 -1.03440547e+00 -9.01305377e-01
2.39261780e-02 -2.78702259e-01 1.48823075e-02 -4.26321626e-01
-8.58459175e-01 7.55860329e-01 8.38746309e-01 1.48500830e-01
6.10335171e-01 2.35169768e-01 4.67157990e-01 6.34192348e-01
5.55529118e-01 -1.52003086e+00 1.58571258e-01 2.80770540e-01
3.81028235e-01 -1.54575753e+00 2.49240085e-01 -4.13726360e-01
-5.24200499e-01 1.02541041e+00 7.22915471e-01 -2.69221216e-01
1.19588625e+00 -4.09938425e-01 -1.54619262e-01 -4.76740360e-01
-8.34459364e-01 -5.19947708e-01 4.59838867e-01 7.18592167e-01
2.74573173e-02 2.03906864e-01 -8.15349445e-02 2.44342610e-01
4.17075366e-01 7.03436062e-02 3.47985297e-01 1.05960429e+00
-1.02543271e+00 -6.37592375e-01 4.13253717e-02 1.05203843e+00
3.08440268e-01 -1.40469968e-01 -4.83513802e-01 6.02003157e-01
-9.95336249e-02 1.09114730e+00 -1.45328278e-03 -4.01838154e-01
1.60164103e-01 2.50933766e-01 6.11041546e-01 -1.31826556e+00
-7.53302351e-02 -1.49294615e-01 -3.53859253e-02 -2.42551446e-01
-6.17860973e-01 -3.17345411e-01 -1.37144399e+00 4.70577776e-02
-9.53167617e-01 3.81498605e-01 4.72671032e-01 1.05485058e+00
3.02238554e-01 2.44807109e-01 7.88325787e-01 -6.81809604e-01
-3.95979464e-01 -1.21551359e+00 -1.07606220e+00 4.62044388e-01
-1.47506624e-01 -1.01300180e+00 -4.94363666e-01 -8.16875044e-03]
|
[9.980371475219727, 3.048858404159546]
|
2b6cdfe7-31d3-4b4a-980d-cf79effe8157
|
drag-your-gan-interactive-point-based
|
2305.10973
| null |
https://arxiv.org/abs/2305.10973v1
|
https://arxiv.org/pdf/2305.10973v1.pdf
|
Drag Your GAN: Interactive Point-based Manipulation on the Generative Image Manifold
|
Synthesizing visual content that meets users' needs often requires flexible and precise controllability of the pose, shape, expression, and layout of the generated objects. Existing approaches gain controllability of generative adversarial networks (GANs) via manually annotated training data or a prior 3D model, which often lack flexibility, precision, and generality. In this work, we study a powerful yet much less explored way of controlling GANs, that is, to "drag" any points of the image to precisely reach target points in a user-interactive manner, as shown in Fig.1. To achieve this, we propose DragGAN, which consists of two main components: 1) a feature-based motion supervision that drives the handle point to move towards the target position, and 2) a new point tracking approach that leverages the discriminative generator features to keep localizing the position of the handle points. Through DragGAN, anyone can deform an image with precise control over where pixels go, thus manipulating the pose, shape, expression, and layout of diverse categories such as animals, cars, humans, landscapes, etc. As these manipulations are performed on the learned generative image manifold of a GAN, they tend to produce realistic outputs even for challenging scenarios such as hallucinating occluded content and deforming shapes that consistently follow the object's rigidity. Both qualitative and quantitative comparisons demonstrate the advantage of DragGAN over prior approaches in the tasks of image manipulation and point tracking. We also showcase the manipulation of real images through GAN inversion.
|
['Christian Theobalt', 'Abhimitra Meka', 'Lingjie Liu', 'Thomas Leimkühler', 'Ayush Tewari', 'Xingang Pan']
|
2023-05-18
| null | null | null | null |
['image-manipulation']
|
['computer-vision']
|
[ 2.44113013e-01 3.61251235e-01 1.62559628e-01 1.02844298e-01
-3.30622613e-01 -1.16037834e+00 8.14836919e-01 -4.59522486e-01
3.79517078e-02 4.76253510e-01 1.08209170e-01 1.09167598e-01
2.98541963e-01 -8.98550928e-01 -1.09090245e+00 -7.90440738e-01
3.44771534e-01 4.78308916e-01 8.48262906e-02 -5.32630026e-01
1.10826269e-01 9.17357802e-01 -1.37402081e+00 -1.00727543e-01
6.85476780e-01 6.04268253e-01 -3.51771787e-02 7.11501181e-01
2.60162115e-01 4.02000219e-01 -5.70093274e-01 -3.46291631e-01
4.09839451e-01 -4.34628725e-01 -3.37617099e-01 4.38949883e-01
5.40021658e-01 -3.30045462e-01 -2.46120200e-01 9.04500484e-01
2.77511001e-01 1.44525185e-01 7.40291417e-01 -1.29199159e+00
-9.42334712e-01 1.90651745e-01 -6.24413729e-01 -5.43013930e-01
4.28464025e-01 6.30231559e-01 7.22808838e-01 -3.99594247e-01
9.22917843e-01 1.26391518e+00 4.93694156e-01 8.66296470e-01
-1.48067701e+00 -5.75517654e-01 1.84223428e-01 -5.06708860e-01
-1.21209109e+00 -3.61692071e-01 9.53701437e-01 -7.04008877e-01
1.55475974e-01 4.57034022e-01 7.95385778e-01 1.26964450e+00
6.15943111e-02 5.57576239e-01 8.07625592e-01 -4.25065845e-01
1.87317297e-01 1.85458109e-01 -5.82658470e-01 7.57497787e-01
-1.15730070e-01 2.49449000e-01 -1.34971812e-01 -2.52690669e-02
1.42462802e+00 1.42529458e-01 -4.52037096e-01 -9.68529463e-01
-1.33174157e+00 7.54677951e-01 6.64222240e-01 4.13173512e-02
-2.09193408e-01 4.38102216e-01 -1.51384071e-01 -3.04267220e-02
1.64762720e-01 8.79395306e-01 -2.17142358e-01 1.02388494e-01
-5.91477752e-01 5.92907548e-01 5.76110661e-01 1.21186125e+00
6.50587082e-01 1.79242969e-01 -4.60879982e-01 2.71242112e-01
8.25622678e-02 7.45871186e-01 1.48044959e-01 -1.18300474e+00
1.21083260e-01 6.30679667e-01 3.59638721e-01 -1.06832087e+00
-4.37368266e-02 -1.84348375e-01 -7.09492803e-01 6.63737655e-01
4.11868989e-01 -2.75245756e-01 -1.27718115e+00 2.05674648e+00
6.04966640e-01 -1.80121243e-01 -2.15602681e-01 9.82405424e-01
4.33391929e-01 6.55720532e-01 4.24819104e-02 3.19815665e-01
1.07353342e+00 -8.19527864e-01 -4.12178606e-01 -9.88176912e-02
1.50161520e-01 -6.03559911e-01 1.44143915e+00 9.12701413e-02
-1.39304435e+00 -4.21358019e-01 -8.05876553e-01 -2.00171605e-01
-2.85684943e-01 1.16269492e-01 4.40662026e-01 3.80048454e-01
-9.20506299e-01 4.58531529e-01 -9.41697180e-01 -2.43111867e-02
4.64164019e-01 2.89513260e-01 -4.40750331e-01 1.09866872e-01
-7.54954815e-01 6.78062439e-01 -1.13263272e-01 -3.54758501e-02
-1.04036427e+00 -9.01976049e-01 -9.46798801e-01 8.37035030e-02
3.61590087e-01 -9.79768455e-01 9.57433224e-01 -1.20744944e+00
-1.74653184e+00 9.54048336e-01 3.02703947e-01 5.16644344e-02
8.75164032e-01 -2.16638595e-01 3.36272687e-01 -4.18860279e-02
-2.34537479e-02 1.07725346e+00 1.24105096e+00 -1.67944622e+00
-5.59346415e-02 -2.95202941e-01 2.79681474e-01 1.92548752e-01
-8.60715657e-02 -4.69353020e-01 -5.45903802e-01 -9.38373744e-01
-2.08833814e-01 -1.28645635e+00 -2.88623422e-01 5.25813520e-01
-8.16181839e-01 2.26489902e-01 1.16251612e+00 -4.93346363e-01
6.27374470e-01 -2.33588481e+00 6.98420346e-01 3.90474021e-01
1.34387136e-01 2.20355034e-01 -1.41816005e-01 4.08485204e-01
4.75341873e-03 2.07711369e-01 -1.61639944e-01 -3.86629969e-01
9.90746021e-02 1.30276635e-01 -4.16471988e-01 3.48944306e-01
3.05174381e-01 1.30023289e+00 -8.42285454e-01 -3.90862636e-02
3.46152246e-01 7.68103361e-01 -8.92941952e-01 3.54229122e-01
-6.13062739e-01 9.70381081e-01 -5.44294715e-01 4.59472448e-01
4.16666865e-01 -1.44727573e-01 7.80550838e-02 -3.48746866e-01
-7.55439000e-03 -1.73863515e-01 -9.93005633e-01 1.61304653e+00
-5.19319117e-01 4.10470665e-01 2.52811491e-01 -4.52179968e-01
7.54921675e-01 1.03016891e-01 3.03833336e-01 -2.46411026e-01
6.16669394e-02 -2.50548273e-01 -1.80582449e-01 -3.27517599e-01
2.34253049e-01 -8.25197175e-02 -2.04320461e-01 3.14938992e-01
-3.69302958e-01 -7.84556866e-01 -2.22078249e-01 8.36223662e-02
9.06495154e-01 4.37347800e-01 -3.81423458e-02 -3.06422301e-02
7.23762438e-02 -7.18452185e-02 1.51153728e-01 4.41498995e-01
3.25129122e-01 1.02817440e+00 5.57884336e-01 -2.28845701e-01
-1.34087884e+00 -1.26461291e+00 2.66936749e-01 7.13461816e-01
2.29940236e-01 -3.14973993e-03 -8.48189175e-01 -5.32660007e-01
1.32243633e-01 7.01297998e-01 -7.90412009e-01 -2.75383383e-01
-7.46828735e-01 -6.08215993e-03 2.29285434e-01 5.55765569e-01
3.77965480e-01 -1.14417493e+00 -8.36984277e-01 -1.10123158e-01
1.21576712e-01 -9.91654456e-01 -9.04129148e-01 -1.46396562e-01
-3.65959316e-01 -8.96786869e-01 -6.85919344e-01 -7.57787049e-01
1.08266962e+00 -9.43941250e-02 9.10498559e-01 5.96439093e-02
-3.19056630e-01 4.78783906e-01 -2.80993164e-01 -1.61935657e-01
-4.07109976e-01 1.97439995e-02 -2.56340742e-01 5.46831824e-02
-6.18918002e-01 -8.71977985e-01 -6.96393907e-01 4.73060817e-01
-1.14319646e+00 3.62659484e-01 4.74544883e-01 6.33135140e-01
5.73743045e-01 -3.51753794e-02 8.19869712e-02 -8.31503808e-01
3.63126844e-01 -9.63518843e-02 -7.04163671e-01 8.90132859e-02
-5.50476387e-02 6.13653548e-02 5.88751137e-01 -9.32605326e-01
-6.36270702e-01 4.60946947e-01 5.61175086e-02 -7.91358471e-01
-7.80062154e-02 -2.17270598e-01 -5.79164028e-01 -2.49826074e-01
6.05069816e-01 7.60265812e-02 1.42078996e-01 -2.20101759e-01
8.23233247e-01 3.55836041e-02 7.35208392e-01 -7.99156845e-01
1.33966637e+00 6.57223821e-01 7.19175041e-02 -5.08490145e-01
-4.83527750e-01 2.34561473e-01 -5.53352296e-01 -3.28887999e-02
1.03192377e+00 -5.51764071e-01 -7.88748264e-01 3.95559669e-01
-1.02514112e+00 -8.34427714e-01 -5.59375823e-01 -2.18793415e-02
-7.86204278e-01 -1.96424171e-01 -4.58689243e-01 -3.61151934e-01
-2.04567477e-01 -1.20014369e+00 1.52961290e+00 1.49331108e-01
-4.00881827e-01 -9.83451843e-01 -2.12236375e-01 1.07153475e-01
5.08422434e-01 1.02510917e+00 1.03489935e+00 5.48669286e-02
-9.05797362e-01 -3.69625777e-01 1.53665140e-01 2.42522165e-01
2.99771309e-01 3.16975117e-01 -5.31763554e-01 -3.80249083e-01
-3.18514735e-01 -2.05154642e-01 2.27138996e-01 2.18282342e-01
1.19853055e+00 -7.17589378e-01 -4.44940001e-01 9.15445626e-01
1.11967373e+00 1.91008851e-01 8.36567223e-01 6.40085414e-02
1.07776606e+00 5.39889038e-01 2.37396464e-01 1.20135181e-01
7.71820098e-02 1.01714170e+00 6.40914440e-01 -2.31067002e-01
-2.94176668e-01 -6.41829431e-01 1.95072189e-01 5.93654104e-02
-1.03177525e-01 -4.48358893e-01 -6.12208486e-01 2.06979901e-01
-1.53436697e+00 -9.22795773e-01 3.26473713e-01 2.03361344e+00
8.39446843e-01 -2.34311093e-02 4.09620702e-02 -1.12004779e-01
6.35650218e-01 -4.95968899e-03 -6.41308546e-01 -2.07055598e-01
2.15503886e-01 2.11605251e-01 4.95564461e-01 4.22206581e-01
-8.88725758e-01 9.56934869e-01 5.82982111e+00 5.65474689e-01
-1.54132462e+00 -1.01583220e-01 4.78769392e-01 -2.65635520e-01
-6.82232082e-01 -1.30265008e-03 -4.65280443e-01 5.27543545e-01
-2.76301652e-02 1.83195800e-01 6.44787610e-01 6.47862256e-01
2.92848557e-01 3.59768361e-01 -1.08796442e+00 7.79013515e-01
-1.47027345e-02 -1.46534491e+00 3.11881959e-01 2.43231222e-01
9.12632287e-01 -6.35322988e-01 4.74737406e-01 -1.22295618e-02
5.79038560e-01 -1.12157285e+00 1.08442509e+00 6.97850764e-01
1.14983332e+00 -6.36033893e-01 -3.78162302e-02 3.99688572e-01
-5.83833873e-01 1.31979510e-01 1.85944632e-01 2.61022270e-01
3.40898007e-01 2.44708315e-01 -6.15261734e-01 -1.09500056e-02
4.20503259e-01 2.94916421e-01 -2.44140431e-01 4.56864715e-01
-6.82428181e-01 2.66679198e-01 -2.26349428e-01 2.20023260e-01
2.49808386e-01 -2.17234313e-01 6.76590562e-01 7.77103245e-01
2.57317156e-01 8.74806419e-02 1.29752696e-01 1.39376032e+00
-2.00636372e-01 -4.24275160e-01 -8.01657617e-01 -5.86371608e-02
3.70024472e-01 1.32266581e+00 -5.57969570e-01 1.35733649e-01
1.92250773e-01 1.12279594e+00 1.69898346e-01 4.21232611e-01
-1.06641352e+00 -2.91662216e-01 7.67488360e-01 7.22370446e-01
5.66615641e-01 -5.19706011e-01 -2.34999761e-01 -9.80451167e-01
1.45938948e-01 -9.75182056e-01 -3.50706667e-01 -1.18092453e+00
-9.88733590e-01 4.14153665e-01 -1.67811662e-02 -1.01808453e+00
-3.00305605e-01 -3.90670717e-01 -8.70542824e-01 7.24786282e-01
-8.33271623e-01 -1.64445639e+00 -5.93333304e-01 5.99642098e-01
2.90823817e-01 1.40998602e-01 6.68593705e-01 -8.58802721e-03
-3.10095847e-01 6.64196491e-01 -2.55436808e-01 2.61141390e-01
4.36098099e-01 -1.16662562e+00 5.33462763e-01 6.38263941e-01
-4.20610607e-02 6.00497484e-01 7.48301148e-01 -4.90899950e-01
-1.54974484e+00 -1.12762547e+00 -1.44101130e-02 -7.96767950e-01
3.10983628e-01 -6.65715814e-01 -5.36502600e-01 9.96639371e-01
-2.56332196e-03 3.43280807e-02 9.52235609e-03 -3.74578565e-01
-2.72232860e-01 -2.60142945e-02 -1.30352628e+00 9.48459625e-01
1.14876544e+00 -3.80099565e-01 -1.50750786e-01 3.39197427e-01
7.70752966e-01 -7.88840771e-01 -5.76310694e-01 3.03498477e-01
5.66607833e-01 -7.77895927e-01 1.18796873e+00 -6.14909232e-01
6.09413564e-01 -5.34829497e-01 4.72241789e-02 -1.56722486e+00
-2.61560529e-01 -9.82037365e-01 1.07122697e-01 1.17948329e+00
1.15426116e-01 -3.61162096e-01 8.72110069e-01 7.32003391e-01
-1.70634553e-01 -6.85044646e-01 -5.13209879e-01 -5.43552876e-01
2.17550382e-01 6.31824955e-02 7.07887411e-01 8.86092961e-01
-3.67201149e-01 1.96486935e-01 -3.86297256e-01 2.61400849e-01
3.66723448e-01 1.12232737e-01 1.22696114e+00 -6.35560215e-01
-4.43134248e-01 -4.49013144e-01 -5.90444922e-01 -1.08934987e+00
1.31605819e-01 -5.67914367e-01 5.93498126e-02 -1.22028649e+00
-1.09987311e-01 -7.16470003e-01 6.49908841e-01 6.05866730e-01
-5.25149032e-02 2.72543103e-01 3.33165407e-01 1.99515313e-01
-6.50646836e-02 5.42634785e-01 1.95404589e+00 -1.05922714e-01
-3.68824452e-01 2.28513847e-03 -7.01923788e-01 6.85575545e-01
5.51796734e-01 -1.29861444e-01 -5.45631826e-01 -5.82479596e-01
8.98377523e-02 6.12683184e-02 8.84514868e-01 -8.05544078e-01
-7.86145702e-02 -3.67634028e-01 6.03761852e-01 1.43901734e-02
5.39475083e-01 -8.72035980e-01 6.02153599e-01 4.00633007e-01
-4.04056937e-01 1.40687659e-01 5.07462621e-02 4.05648232e-01
1.93821758e-01 1.41868800e-01 9.83933091e-01 -2.63208181e-01
-2.11627364e-01 4.69932646e-01 2.88641099e-02 1.37950620e-02
1.37637460e+00 -8.47198069e-02 -2.19608113e-01 -6.71623349e-01
-8.17908049e-01 8.14742818e-02 1.18399894e+00 5.02075911e-01
3.99320066e-01 -1.38873637e+00 -3.74881864e-01 4.89401072e-01
-1.15542524e-01 5.68274498e-01 -4.37811688e-02 5.61704040e-01
-6.73637867e-01 -1.54674008e-01 -2.40194917e-01 -7.22342789e-01
-8.74750733e-01 8.46855044e-01 5.83272219e-01 1.15269408e-01
-7.57329047e-01 5.26924908e-01 5.84475338e-01 -3.87376428e-01
5.91827184e-02 -2.92045861e-01 2.40359008e-01 -3.87409359e-01
1.28030539e-01 2.69730501e-02 -2.91202992e-01 -3.97662044e-01
-3.78484395e-03 7.41950214e-01 1.94755748e-01 -1.19711600e-01
1.16605544e+00 9.10395682e-02 7.11962655e-02 4.15591374e-02
9.90544379e-01 5.00870347e-01 -1.88398099e+00 2.80803323e-01
-6.47390068e-01 -5.88770986e-01 -3.39941621e-01 -7.60746598e-01
-1.36748397e+00 6.02356970e-01 3.00224602e-01 2.15608165e-01
8.86680007e-01 1.05991460e-01 7.43083656e-01 -9.18681398e-02
3.10642749e-01 -5.67028344e-01 4.19236392e-01 1.32007167e-01
1.12545717e+00 -7.40610123e-01 -4.05653030e-01 -4.99796361e-01
-6.76710069e-01 8.43941748e-01 6.42973661e-01 -3.83673877e-01
3.76625866e-01 4.92552012e-01 4.31480668e-02 -2.04373106e-01
-3.32652301e-01 1.65390238e-01 3.52373362e-01 8.38744342e-01
8.24746862e-02 1.15309462e-01 2.29972556e-01 -1.70660131e-02
-4.71981764e-01 -3.07343543e-01 3.09276462e-01 8.70519936e-01
-1.44685477e-01 -9.80783343e-01 -3.63813221e-01 1.78286824e-02
-4.99327630e-02 1.78904071e-01 -4.31236565e-01 1.02119827e+00
3.32199097e-01 4.26767141e-01 1.50083929e-01 -2.01980188e-01
5.54303110e-01 -2.85943151e-01 7.02764213e-01 -7.78281212e-01
-3.38638783e-01 -9.59526822e-02 -3.79717827e-01 -6.69852436e-01
-1.70168713e-01 -6.69686317e-01 -1.15780461e+00 -3.19619238e-01
2.27914695e-02 -1.90776467e-01 6.21515453e-01 6.84550464e-01
4.86843407e-01 4.80727345e-01 6.99100614e-01 -1.34778416e+00
-6.38584912e-01 -4.84703928e-01 -3.43196064e-01 8.28503430e-01
4.66980696e-01 -7.35740900e-01 -3.29766423e-01 3.05596739e-01]
|
[11.864034652709961, -0.4648391604423523]
|
d9bacc63-a8bf-4b20-89e6-dcd88f981309
|
probabilistic-dag-search
|
2106.08717
| null |
https://arxiv.org/abs/2106.08717v1
|
https://arxiv.org/pdf/2106.08717v1.pdf
|
Probabilistic DAG Search
|
Exciting contemporary machine learning problems have recently been phrased in the classic formalism of tree search -- most famously, the game of Go. Interestingly, the state-space underlying these sequential decision-making problems often posses a more general latent structure than can be captured by a tree. In this work, we develop a probabilistic framework to exploit a search space's latent structure and thereby share information across the search tree. The method is based on a combination of approximate inference in jointly Gaussian models for the explored part of the problem, and an abstraction for the unexplored part that imposes a reduction of complexity ad hoc. We empirically find our algorithm to compare favorably to existing non-probabilistic alternatives in Tic-Tac-Toe and a feature selection application.
|
['Philipp Hennig', 'Cheng Zhang', 'Julia Grosse']
|
2021-06-16
| null | null | null | null |
['game-of-go']
|
['playing-games']
|
[ 3.33585590e-01 2.41598010e-01 -5.20679295e-01 -2.30098844e-01
-8.53450000e-01 -6.54768467e-01 7.51716316e-01 3.86167038e-03
-3.93786967e-01 8.04667413e-01 3.03061884e-02 -6.60666049e-01
-6.72743618e-01 -7.06832230e-01 -4.40020077e-02 -7.66141415e-01
-1.82069227e-01 9.22988772e-01 5.54531753e-01 -1.75700158e-01
3.53679597e-01 3.34398627e-01 -1.50214100e+00 9.22124162e-02
5.98257840e-01 9.67113912e-01 1.69569537e-01 5.44526517e-01
-2.15559110e-01 6.69844151e-01 -2.10084051e-01 -4.45781916e-01
2.62978762e-01 -8.78322497e-02 -1.22565699e+00 -2.76412554e-02
-1.27155975e-01 6.26318306e-02 -3.58564377e-01 1.04772389e+00
1.34358481e-01 2.49445558e-01 8.17522466e-01 -1.22814500e+00
1.95971489e-01 7.39843667e-01 -4.03534174e-01 2.52118677e-01
4.00310755e-01 7.22847730e-02 1.40604961e+00 -3.16457391e-01
6.66934729e-01 1.25317001e+00 5.93252063e-01 2.36757129e-01
-1.75582623e+00 -3.24855447e-01 1.99264571e-01 3.60450655e-01
-1.51987433e+00 -1.36626244e-01 7.86288798e-01 -4.95166838e-01
8.05255532e-01 3.12207460e-01 6.32948875e-01 1.18503630e+00
5.70998013e-01 1.13913012e+00 1.42206693e+00 -5.60013413e-01
8.30283642e-01 -2.59128232e-02 2.30517685e-01 6.40608013e-01
1.21531948e-01 5.62552929e-01 -7.71768570e-01 -6.78670049e-01
6.26375735e-01 -1.08839840e-01 2.42557123e-01 -9.81010437e-01
-5.83606601e-01 1.19718063e+00 -1.43344447e-01 3.69830996e-01
-3.36375594e-01 1.63210094e-01 1.20744005e-01 2.19404027e-01
3.26663077e-01 3.10048014e-01 -5.74721515e-01 -5.80026507e-01
-1.33764410e+00 5.18651485e-01 1.34424174e+00 8.35450113e-01
8.45980883e-01 -3.56374055e-01 -2.05144227e-01 2.96715409e-01
4.43740189e-01 -3.40991259e-01 4.88163501e-01 -9.55811143e-01
2.20596671e-01 3.08762103e-01 3.31364453e-01 -5.01058757e-01
-2.95881271e-01 -6.45312965e-01 -4.56142038e-01 4.69359457e-01
5.71664929e-01 -3.30737941e-02 -8.69710803e-01 1.69878232e+00
4.33348149e-01 -1.59346387e-01 -8.79850984e-02 3.76077563e-01
-1.96061209e-01 2.36745313e-01 4.06778045e-02 -2.60594934e-01
1.26894820e+00 -7.90931284e-01 -5.68199754e-01 -3.91851574e-01
5.22001684e-01 -3.77450049e-01 6.13400996e-01 7.51801312e-01
-8.53073895e-01 -2.70743370e-01 -9.44866776e-01 9.04797092e-02
-1.89926550e-01 -3.15629303e-01 1.17886519e+00 8.89173627e-01
-1.11449671e+00 9.20291185e-01 -9.67431724e-01 -3.84470254e-01
3.36187929e-01 3.74402404e-01 -1.50734782e-01 1.93529576e-02
-1.05255473e+00 1.00600541e+00 8.29434454e-01 -1.05545633e-01
-8.89843702e-01 -1.50893658e-01 -7.01450288e-01 1.05590343e-01
1.03444970e+00 -7.04674482e-01 1.65469980e+00 -3.62591833e-01
-1.72942507e+00 6.09231174e-01 -2.65027106e-01 -6.20905697e-01
6.23027563e-01 -2.74746537e-01 -6.88577816e-03 -3.02239269e-01
2.09995672e-01 2.36916900e-01 8.68964136e-01 -6.32658482e-01
-8.00070226e-01 -4.92731094e-01 1.43885598e-01 1.93106204e-01
2.71156013e-01 1.28193378e-01 -3.99604052e-01 -4.99589205e-01
4.26841587e-01 -1.05302334e+00 -9.69442666e-01 -2.96290070e-01
-5.32140136e-01 -5.61519027e-01 2.01973021e-01 -1.21035077e-01
1.60654652e+00 -1.80416310e+00 4.48858857e-01 3.39149892e-01
2.22416073e-01 -4.30415004e-01 2.83422232e-01 8.13087821e-01
-1.32078335e-01 1.87800437e-01 -2.59994835e-01 -1.95012465e-01
3.58565032e-01 4.61853534e-01 -3.52018923e-01 5.88751018e-01
-7.88491368e-02 7.34192073e-01 -9.05096531e-01 -5.08095562e-01
-1.10533834e-02 -2.52809286e-01 -5.84758878e-01 -1.45715684e-01
-5.45903087e-01 -8.64011720e-02 -1.10112488e+00 6.03060246e-01
3.02362829e-01 -2.60252267e-01 4.15070474e-01 3.49709779e-01
-3.97074610e-01 5.78500032e-01 -1.42563903e+00 2.05411124e+00
-1.01140887e-01 3.72537374e-01 -7.20507512e-03 -9.44700658e-01
4.88820970e-01 -4.52414934e-05 5.12270987e-01 5.03332168e-02
6.04958273e-02 5.56163974e-02 3.51586640e-02 -9.57332551e-02
5.06105065e-01 -3.04751337e-01 -4.19664145e-01 5.68572104e-01
2.55092889e-01 -4.16707128e-01 9.12965760e-02 -8.21223632e-02
1.43441689e+00 4.74976599e-01 8.69318902e-01 -5.15399337e-01
2.73390878e-02 2.28065535e-01 4.58267123e-01 1.46673250e+00
-1.32253051e-01 4.42088768e-03 6.49979889e-01 -4.54693437e-01
-5.85441887e-01 -1.19468570e+00 -4.54128653e-01 1.07390118e+00
-1.30297601e-01 -8.24419856e-01 -5.49545467e-01 -6.56388819e-01
-5.76335518e-03 1.00265205e+00 -6.80415809e-01 -1.15687393e-01
6.27584383e-02 -5.93225002e-01 3.65565091e-01 3.11345100e-01
6.89009055e-02 -8.21718514e-01 -9.34321940e-01 5.02953589e-01
2.48703763e-01 -7.65266836e-01 -2.84431670e-02 1.03724039e+00
-1.11306751e+00 -7.92479873e-01 -1.24609672e-01 -1.47575900e-01
-1.21860646e-01 -3.32019627e-01 1.21138310e+00 -6.16381884e-01
-4.62327093e-01 3.94034684e-01 -2.37053588e-01 -4.10398394e-01
-2.39255905e-01 3.54204565e-01 2.33629778e-01 -2.59836465e-01
7.45654464e-01 -7.64349401e-01 -1.29068822e-01 1.29305929e-01
-5.96762657e-01 1.81379676e-01 7.04885006e-01 9.05041754e-01
4.27401155e-01 6.14954114e-01 -2.77269185e-01 -7.06861377e-01
6.18551672e-01 -6.93872869e-01 -8.57918561e-01 1.66780517e-01
-7.85214365e-01 6.62811697e-01 -9.75917950e-02 -4.70002323e-01
-9.69451487e-01 3.44558328e-01 2.12529004e-01 -3.09128553e-01
-4.20330763e-01 8.63227546e-01 -6.09100349e-02 1.65853083e-01
4.95012224e-01 2.46889114e-01 -3.27229172e-01 -8.14766645e-01
4.66802537e-01 3.90929222e-01 2.22091481e-01 -9.63842392e-01
8.79006624e-01 3.62047493e-01 3.71436179e-01 -6.49323642e-01
-7.96702683e-01 -5.28141737e-01 -7.26376295e-01 1.48120582e-01
6.89780056e-01 -5.55477738e-01 -5.91721833e-01 1.85385808e-01
-9.36123013e-01 -1.89231992e-01 -5.77257037e-01 5.06747425e-01
-1.00156808e+00 2.54683912e-01 -4.10167933e-01 -1.44375229e+00
2.72207916e-01 -1.14182985e+00 1.09322011e+00 1.85151145e-01
-5.87938786e-01 -9.46529567e-01 5.03964365e-01 6.32849857e-02
2.18954995e-01 -2.40519885e-02 9.76897180e-01 -1.01462054e+00
-7.56847978e-01 -3.01289052e-01 1.51338607e-01 -3.35398763e-01
1.26464054e-01 -2.38432631e-01 -7.85348177e-01 -1.35454491e-01
3.09422582e-01 -1.31315470e-01 8.80404651e-01 4.56135392e-01
8.88330162e-01 -1.93645954e-01 -6.97532654e-01 3.79308462e-01
1.38818073e+00 2.51963228e-01 2.47800976e-01 5.09949982e-01
4.67111059e-02 8.93719912e-01 6.37950659e-01 6.40789509e-01
2.12352574e-01 9.72969770e-01 3.68061006e-01 4.29735184e-01
6.78051293e-01 -4.83474225e-01 1.91468522e-01 2.39069134e-01
2.14597043e-02 -1.00676760e-01 -1.12074888e+00 4.01510447e-01
-2.05673695e+00 -9.62426245e-01 2.75464743e-01 2.08550954e+00
9.02515590e-01 5.73475897e-01 1.33973926e-01 3.51676308e-02
4.10350889e-01 2.77997673e-01 -5.43993890e-01 -4.65748638e-01
1.58392280e-01 5.89521587e-01 4.59738970e-01 5.28630853e-01
-1.14381301e+00 9.99949276e-01 7.72333336e+00 1.30047274e+00
-4.20175254e-01 5.49672544e-02 3.67211550e-01 -1.28560036e-01
-3.04081827e-01 5.97482204e-01 -9.74051416e-01 2.79903501e-01
9.29159760e-01 -3.79860550e-01 3.09082389e-01 1.13469732e+00
-1.00960329e-01 -5.61529756e-01 -1.44034064e+00 8.32991064e-01
-4.83809531e-01 -1.09987855e+00 -1.99846342e-01 5.21846294e-01
3.79573524e-01 5.50554432e-02 9.41768512e-02 4.23816919e-01
1.06634533e+00 -1.20110965e+00 7.66512394e-01 4.93446648e-01
4.56688434e-01 -5.15725791e-01 3.61958832e-01 7.57342696e-01
-1.07320440e+00 -2.88081259e-01 -6.07582033e-02 -3.33031893e-01
1.77113578e-01 5.63641489e-01 -5.21326363e-01 7.11810648e-01
6.48464739e-01 3.24569523e-01 -3.33248943e-01 1.15854156e+00
-2.70639390e-01 5.86845636e-01 -7.00881004e-01 -2.47339249e-01
5.80218136e-01 -3.94780427e-01 8.91751945e-01 9.09236729e-01
2.04591945e-01 6.21852279e-03 3.91979605e-01 1.01214516e+00
5.65842450e-01 -2.13310689e-01 -6.88739836e-01 -1.79193661e-01
2.43312329e-01 8.57939899e-01 -6.44371152e-01 -6.32907748e-02
-2.69505769e-01 9.33202863e-01 3.00820440e-01 2.65962392e-01
-5.08557498e-01 4.35163602e-02 6.12657964e-01 -1.05978839e-01
4.39304113e-01 -2.62572676e-01 -2.16412291e-01 -1.22379744e+00
-2.71521080e-02 -8.24455559e-01 5.79876006e-01 -4.87618297e-01
-1.15864384e+00 5.53521693e-01 4.66993392e-01 -8.91268492e-01
-9.15687144e-01 -6.62809432e-01 -4.90915239e-01 1.11885297e+00
-1.01314509e+00 -8.91261637e-01 4.67280537e-01 4.14724201e-01
6.91530049e-01 -1.54998735e-01 9.47581232e-01 -4.98312056e-01
-4.04636830e-01 1.94119811e-01 1.39287040e-01 -4.77945566e-01
2.08370890e-02 -1.54565370e+00 4.00420427e-01 7.20172107e-01
5.32651544e-01 6.73333943e-01 1.17411506e+00 -6.25607967e-01
-1.34898651e+00 -3.89522344e-01 7.88344979e-01 -6.90602124e-01
1.08182752e+00 -2.93261975e-01 -5.81677258e-01 7.35703290e-01
-2.87064947e-02 -4.41193223e-01 7.27764308e-01 7.70930946e-01
-3.37344736e-01 3.80453020e-01 -8.38098645e-01 6.41036153e-01
1.05681789e+00 -7.98497677e-01 -1.01505244e+00 -5.86069711e-02
3.49288076e-01 -1.92940161e-01 -5.36342621e-01 3.93150985e-01
7.85035133e-01 -9.76024449e-01 8.27476680e-01 -1.02780437e+00
1.05222389e-01 -1.09736003e-01 -3.91273588e-01 -1.03716779e+00
-5.29318750e-01 -1.23019028e+00 -4.15761381e-01 8.33915591e-01
2.52669692e-01 -4.00754392e-01 1.24585867e+00 7.40719438e-01
3.17715049e-01 -9.83127713e-01 -1.47348213e+00 -8.19722116e-01
1.35744825e-01 -8.28844666e-01 3.34844828e-01 5.82942188e-01
1.28932923e-01 2.53455073e-01 -1.95318878e-01 -7.68722035e-03
9.99687314e-01 2.71542758e-01 4.92844075e-01 -1.68954825e+00
-7.76915014e-01 -7.03646481e-01 -5.08812428e-01 -1.23933089e+00
2.10024759e-01 -6.88630641e-01 3.47812437e-02 -1.07624841e+00
2.84852117e-01 -4.73074257e-01 -5.16878963e-01 3.88930649e-01
1.42686293e-01 -6.60205781e-01 -8.74528363e-02 1.79169774e-01
-7.60204256e-01 5.37782848e-01 5.28280556e-01 1.28352135e-01
-2.92026222e-01 4.88558829e-01 -6.20981932e-01 9.89359021e-01
4.99559313e-01 -8.59285235e-01 -3.62917721e-01 1.55349135e-01
3.72422248e-01 5.00541210e-01 8.82586986e-02 -8.25496554e-01
5.35501361e-01 -4.80486512e-01 -3.18974815e-02 -7.64726460e-01
4.49627876e-01 -8.14919710e-01 4.47782427e-01 6.00682735e-01
-3.70771408e-01 -1.31232634e-01 1.48638740e-01 9.57941532e-01
-1.67861938e-01 -6.25037372e-01 3.69375557e-01 -1.17401272e-01
-7.75085151e-01 3.22942466e-01 -8.47211838e-01 -1.55076772e-01
8.61033976e-01 -3.05815369e-01 3.67089510e-01 -3.49257171e-01
-9.71042812e-01 1.64223388e-01 4.25330251e-01 3.08645844e-01
1.25120640e-01 -1.01185429e+00 -3.90404344e-01 -5.15864752e-02
8.47253017e-03 -1.33954167e-01 1.00507595e-01 6.22511804e-01
-6.52034506e-02 6.80755377e-01 -7.53677636e-02 -5.59835672e-01
-1.04759169e+00 6.75120175e-01 2.59426981e-01 -1.06442225e+00
-3.40225279e-01 8.71368706e-01 2.06412390e-01 -9.93685126e-02
1.48845300e-01 -2.36614421e-01 8.33104178e-03 1.93119243e-01
1.83669284e-01 2.63197213e-01 -4.00553465e-01 -1.69190973e-01
-4.84586656e-01 2.12073028e-01 -1.78813860e-01 -5.89366794e-01
1.18892241e+00 -1.54610336e-01 -3.50280963e-02 8.33669662e-01
6.04485691e-01 -2.27366954e-01 -1.12336004e+00 -6.02774382e-01
7.21585333e-01 -3.50749582e-01 3.76985759e-01 -6.65895641e-01
-2.65932769e-01 6.38874769e-01 4.90810961e-01 4.33856100e-01
8.34900022e-01 2.74121851e-01 -3.89427990e-02 6.42049193e-01
1.02130055e+00 -9.91012275e-01 -4.51237440e-01 5.83110034e-01
5.19297063e-01 -9.39669549e-01 6.35348260e-02 -3.52052301e-01
-3.23497504e-01 1.08802903e+00 9.66446027e-02 8.60575065e-02
8.11668575e-01 4.58371073e-01 -6.01621747e-01 -3.02747220e-01
-1.30259693e+00 -5.06639898e-01 2.70973861e-01 4.15789813e-01
5.63774742e-02 5.86853772e-02 -1.63119912e-01 8.72206032e-01
-2.89301515e-01 1.74566552e-01 8.95744935e-02 1.28674686e+00
-3.97490054e-01 -1.40532625e+00 -1.71300769e-01 4.63348091e-01
-5.62318563e-01 -2.21952036e-01 -3.28248739e-01 8.08225334e-01
-5.31374104e-02 8.22178185e-01 -1.16143614e-01 -4.02983606e-01
-3.99058759e-02 4.21944171e-01 5.85092962e-01 -8.06946218e-01
-2.36545712e-01 2.74497241e-01 2.80404329e-01 -9.69678462e-01
-1.63314566e-01 -1.05496514e+00 -4.70134974e-01 5.03487652e-03
-5.21602511e-01 6.06770456e-01 5.89627922e-01 1.22419572e+00
-4.84439507e-02 2.75711775e-01 2.65305012e-01 -7.56732345e-01
-1.14664018e+00 -8.47063482e-01 -1.07529056e+00 -3.20930600e-01
6.17602356e-02 -1.07920992e+00 -4.32746589e-01 -4.80750740e-01]
|
[4.10031795501709, 1.8857680559158325]
|
c8b9cf9c-4741-41ba-9e4c-c7542a885d0a
|
comprehensive-benchmark-datasets-for-amharic
|
2203.12165
| null |
https://arxiv.org/abs/2203.12165v1
|
https://arxiv.org/pdf/2203.12165v1.pdf
|
Comprehensive Benchmark Datasets for Amharic Scene Text Detection and Recognition
|
Ethiopic/Amharic script is one of the oldest African writing systems, which serves at least 23 languages (e.g., Amharic, Tigrinya) in East Africa for more than 120 million people. The Amharic writing system, Abugida, has 282 syllables, 15 punctuation marks, and 20 numerals. The Amharic syllabic matrix is derived from 34 base graphemes/consonants by adding up to 12 appropriate diacritics or vocalic markers to the characters. The syllables with a common consonant or vocalic markers are likely to be visually similar and challenge text recognition tasks. In this work, we presented the first comprehensive public datasets named HUST-ART, HUST-AST, ABE, and Tana for Amharic script detection and recognition in the natural scene. We have also conducted extensive experiments to evaluate the performance of the state of art methods in detecting and recognizing Amharic scene text on our datasets. The evaluation results demonstrate the robustness of our datasets for benchmarking and its potential of promoting the development of robust Amharic script detection and recognition algorithms. Consequently, the outcome will benefit people in East Africa, including diplomats from several countries and international communities.
|
['Xiang Bai', 'Minghui Liao', 'Dingkang Liang', 'Wondimu Dikubab']
|
2022-03-23
| null | null | null | null |
['scene-text-detection']
|
['computer-vision']
|
[-5.93745708e-02 -7.03139603e-01 2.73469388e-01 -9.50398110e-03
-4.72474098e-01 -7.24823952e-01 8.58402550e-01 -2.28001294e-03
-3.70435476e-01 5.25121272e-01 2.30575547e-01 -2.85985887e-01
3.26906532e-01 -6.90177202e-01 -1.80991098e-01 -8.42076600e-01
2.62320161e-01 6.46944344e-01 3.12712610e-01 -3.95557225e-01
6.23801172e-01 8.01223338e-01 -1.36779249e+00 2.80441701e-01
8.52977991e-01 2.50638872e-01 3.49916697e-01 1.12872815e+00
-5.65490648e-02 4.45334345e-01 -8.93044233e-01 -7.69123733e-01
8.68465304e-02 -5.53995490e-01 -4.97939765e-01 1.25096470e-01
6.27833426e-01 -3.70595634e-01 -3.05760682e-01 7.82449305e-01
4.57613498e-01 -1.57641798e-01 1.07868266e+00 -5.07247627e-01
-8.06586802e-01 4.40676898e-01 -9.12201107e-01 1.69411868e-01
2.65881568e-01 2.70692736e-01 9.65665817e-01 -1.47775877e+00
7.24361658e-01 1.26065993e+00 7.50243425e-01 1.94046259e-01
-5.72189093e-01 -4.38196421e-01 -4.00232673e-01 2.30379730e-01
-1.60437500e+00 -4.17397231e-01 3.31662625e-01 -7.10796237e-01
8.67465794e-01 4.92362589e-01 4.78850484e-01 6.98456585e-01
-2.27582897e-03 8.05555582e-01 1.31081736e+00 -7.38571942e-01
-4.61427495e-03 8.72526988e-02 -6.87286481e-02 9.56419945e-01
2.08149195e-01 -6.36982620e-01 -6.18256867e-01 -2.77801100e-02
4.43691701e-01 -1.95983872e-01 6.93745241e-02 4.95477825e-01
-1.15832543e+00 9.01884079e-01 -5.07537387e-02 3.23903918e-01
-9.31667089e-02 -2.96319455e-01 4.78933215e-01 -1.84124202e-01
1.85522959e-01 1.39405429e-01 2.58445114e-01 -4.55906153e-01
-1.17879200e+00 1.45049784e-02 5.31537294e-01 6.68713629e-01
4.19837326e-01 3.82868230e-01 4.02886197e-02 1.56597781e+00
3.23417574e-01 1.05541968e+00 2.48671100e-01 -3.74967128e-01
7.28791296e-01 4.70826149e-01 -7.77791888e-02 -9.90582883e-01
-6.83196187e-02 1.81404680e-01 -6.67139947e-01 -8.54320452e-02
4.00808275e-01 -2.77126133e-01 -1.08695745e+00 8.70660484e-01
8.77022520e-02 -3.54687870e-01 -1.36636809e-01 9.81425226e-01
4.80328560e-01 1.09008694e+00 -2.78886646e-01 1.76842630e-01
1.46091318e+00 -9.48192120e-01 -6.12915158e-01 -1.75534427e-01
1.00356355e-01 -1.61077273e+00 1.23560524e+00 3.44342798e-01
-8.63527536e-01 -2.24520624e-01 -1.04309857e+00 2.24582389e-01
-3.52383763e-01 6.06374979e-01 3.36568117e-01 1.13519442e+00
-7.16321409e-01 1.41364455e-01 -8.27827334e-01 -9.32456374e-01
2.40775391e-01 -3.64304967e-02 -4.08164769e-01 -1.41851343e-02
-8.49153996e-01 9.81359541e-01 2.82472461e-01 3.75971019e-01
-6.12373292e-01 -7.49210119e-02 -5.70703328e-01 -2.50213206e-01
-3.91658599e-04 2.59841740e-01 7.41058230e-01 -7.03075230e-01
-1.46975362e+00 1.23759401e+00 -1.10224017e-03 2.40376107e-02
6.22028053e-01 -2.70099282e-01 -7.87719250e-01 5.00313163e-01
8.80497247e-02 2.67108202e-01 8.62038195e-01 -1.02901912e+00
-6.26704156e-01 -2.37351373e-01 -7.46084929e-01 2.70438284e-01
-4.47599858e-01 9.00223970e-01 -5.76126277e-01 -1.05147433e+00
1.70344710e-02 -9.62205648e-01 3.93975407e-01 -3.26459706e-01
-5.93318045e-01 4.02181894e-02 1.14490402e+00 -1.46584153e+00
1.20695066e+00 -2.08330464e+00 -2.08421454e-01 4.33863193e-01
-2.01028809e-01 5.29471874e-01 -2.33912226e-02 9.53076243e-01
3.46947879e-01 1.17274232e-01 -4.64650333e-01 2.09150389e-01
7.63672143e-02 1.69269249e-01 -3.63576204e-01 7.09987283e-01
4.03669953e-01 4.15298223e-01 -6.70675159e-01 -6.59082711e-01
2.55533665e-01 4.44552004e-01 6.55665100e-02 4.08431292e-02
1.71789348e-01 -1.65466428e-01 -4.63822708e-02 1.01683021e+00
7.71599591e-01 2.34325930e-01 2.07896993e-01 3.50130111e-01
-4.65482384e-01 2.21649066e-01 -1.20867801e+00 6.75738990e-01
-3.87092046e-02 1.06787741e+00 5.79360537e-02 -3.09133500e-01
1.17488062e+00 6.92708343e-02 -6.76089972e-02 -3.68512511e-01
-4.84522525e-03 5.75099468e-01 1.91373959e-01 -4.94303495e-01
9.94532824e-01 1.68737844e-01 4.50797901e-02 4.47728693e-01
-2.58236021e-01 -3.08321059e-01 5.87484360e-01 2.65364945e-01
5.77974379e-01 -1.31720692e-01 2.46571332e-01 -3.35952163e-01
6.84704542e-01 2.90117115e-02 3.98662269e-01 7.84406960e-01
-2.55819052e-01 9.39110219e-01 4.31756318e-01 -1.56439796e-01
-1.49830258e+00 -1.21982217e+00 -3.79350364e-01 1.08396721e+00
-5.09135067e-01 -4.13488150e-01 -9.93329048e-01 -2.23685771e-01
-2.34407261e-01 4.41308379e-01 -4.33932930e-01 4.12402242e-01
-9.28767979e-01 -7.15086699e-01 1.17357421e+00 3.42443466e-01
8.86692941e-01 -1.18195307e+00 -4.13716644e-01 -2.00581551e-02
-5.65018877e-02 -1.02666056e+00 -8.27102125e-01 -2.00782701e-01
-3.20643783e-01 -1.03337753e+00 -1.15706575e+00 -9.00572777e-01
4.75203216e-01 2.53224075e-01 6.23254359e-01 3.28718722e-02
-6.02101922e-01 9.16054025e-02 -4.23847139e-01 -2.14747488e-01
-8.04230392e-01 -2.60338098e-01 3.99084277e-02 8.81182402e-02
2.55423933e-01 1.63962975e-01 -3.61003041e-01 3.23153615e-01
-6.86443090e-01 7.72979036e-02 4.47758734e-01 6.49239540e-01
8.59604776e-02 -2.94462264e-01 1.88618839e-01 -9.41681623e-01
3.81596416e-01 -3.02411586e-01 -4.45012510e-01 4.53765571e-01
-1.76798880e-01 -5.20389855e-01 7.67033339e-01 -1.30586684e-01
-1.15068400e+00 -3.66901346e-02 -2.50500381e-01 1.63471475e-01
-1.24794520e-01 5.63116312e-01 4.68122736e-02 1.40879363e-01
5.52809179e-01 5.22878647e-01 -3.30864042e-01 -3.18361908e-01
2.51820713e-01 1.46511233e+00 8.21529150e-01 -6.25399470e-01
8.67671967e-01 3.53733242e-01 -2.61705071e-01 -1.76778162e+00
-1.52205095e-01 -5.87504387e-01 -7.28831351e-01 -5.22758543e-01
1.00377691e+00 -8.00183535e-01 -3.37018728e-01 1.28022790e+00
-1.06923223e+00 -6.29866822e-03 3.23587358e-01 2.13945791e-01
5.10432292e-03 1.00165951e+00 -9.80887651e-01 -1.09942377e+00
-4.28355873e-01 -8.93324316e-01 8.89519870e-01 4.18420613e-01
-3.81531924e-01 -7.70738482e-01 3.54844064e-01 5.47771335e-01
4.28643078e-02 1.57393083e-01 9.99536932e-01 -5.41449130e-01
-1.54538929e-01 -1.50689716e-02 -3.55219364e-01 4.14672554e-01
3.39133650e-01 9.22567368e-01 -9.23345745e-01 -1.59231752e-01
-4.67275470e-01 -1.81489542e-01 8.83736491e-01 -5.79276383e-02
4.03205693e-01 -1.47300169e-01 3.08815151e-01 2.30930597e-01
1.10802329e+00 4.51997668e-01 6.66189551e-01 3.50915492e-01
8.26569200e-01 4.62865949e-01 4.42657173e-01 5.88145614e-01
1.50886714e-01 4.72658128e-01 -1.47232950e-01 -4.08576615e-02
-4.05865461e-01 -9.74745452e-02 8.38729739e-01 1.23372042e+00
-2.49513254e-01 -3.06282975e-02 -1.47595119e+00 7.71443367e-01
-1.36157227e+00 -9.22657669e-01 -6.61129177e-01 2.10629177e+00
9.63243961e-01 -2.61719882e-01 4.26655263e-01 2.36348510e-01
1.19107461e+00 1.86455011e-01 -8.33329633e-02 -8.47582161e-01
-6.11080587e-01 5.83094001e-01 4.24163878e-01 2.55553067e-01
-1.12625825e+00 1.31562841e+00 6.07275963e+00 8.48843217e-01
-1.16755939e+00 -2.22981215e-01 4.88912046e-01 2.82438993e-01
2.31827676e-01 -2.26161525e-01 -7.79679954e-01 6.71595991e-01
8.07873368e-01 5.44438511e-02 4.59204197e-01 4.64797914e-01
4.83027279e-01 -3.72961760e-01 -4.34738129e-01 8.44747245e-01
4.21795875e-01 -1.39941156e+00 5.63226528e-02 -2.84042537e-01
1.11197722e+00 2.81174660e-01 1.53347880e-01 -1.32501736e-01
3.53035778e-01 -1.02694750e+00 8.82178664e-01 1.39828026e-01
8.46320748e-01 -7.23092079e-01 5.92495501e-01 -3.99836972e-02
-1.11151624e+00 2.13174924e-01 -3.78868014e-01 1.18159130e-01
-2.69365579e-01 2.33059615e-01 -1.05837059e+00 2.71854311e-01
6.57281935e-01 5.36991596e-01 -1.07710505e+00 9.49148834e-01
-4.53289241e-01 1.22914243e+00 -4.24275279e-01 -5.20686209e-01
5.18829405e-01 -5.95732868e-01 5.32417238e-01 1.93679655e+00
3.63846123e-01 -3.00296366e-01 -6.51895925e-02 4.29495275e-01
2.72245593e-02 5.73759794e-01 -3.52479219e-01 -4.26301390e-01
4.93727475e-01 1.15932047e+00 -1.30718732e+00 -2.54080504e-01
-4.29045498e-01 1.15051401e+00 -2.21266989e-02 3.01315367e-01
-5.98314226e-01 -8.05441856e-01 4.41529721e-01 -8.11158195e-02
2.85840780e-01 -5.31906545e-01 -4.72782195e-01 -9.96787667e-01
1.94929745e-02 -1.25204170e+00 6.65150523e-01 -6.33447409e-01
-1.24816215e+00 2.96372443e-01 -4.44963664e-01 -7.69701242e-01
1.08519286e-01 -1.04958665e+00 -9.75561798e-01 1.19018435e+00
-1.02244425e+00 -1.42959845e+00 -1.89030506e-02 4.56592351e-01
7.56373584e-01 -6.24513149e-01 8.09460044e-01 9.62186530e-02
-9.26985204e-01 3.88852209e-01 7.67194271e-01 6.95497513e-01
8.81268680e-01 -1.31140637e+00 6.16610348e-01 1.28731966e+00
2.78340995e-01 6.05373561e-01 5.63106894e-01 -9.12648737e-01
-1.28951323e+00 -8.91977549e-01 1.04459262e+00 -3.81130815e-01
9.65224504e-01 -6.57721698e-01 -8.02855670e-01 5.05030453e-01
4.73047495e-01 -7.80943036e-01 5.39401233e-01 -1.76365301e-01
-4.99767840e-01 2.58012861e-01 -8.62484515e-01 9.33358490e-01
1.82247296e-01 -6.24273837e-01 -4.96730238e-01 5.68771660e-01
-2.02494875e-01 -1.67819450e-03 -5.59100091e-01 -3.91668022e-01
7.01545417e-01 -1.01141024e+00 6.14154100e-01 -1.97506830e-01
6.50114536e-01 -3.74127001e-01 -1.86341017e-01 -9.61321175e-01
-1.13203734e-01 -7.83011317e-01 2.01439694e-01 1.50283492e+00
3.28494817e-01 -5.23560286e-01 7.63505340e-01 4.17776518e-02
-1.41709253e-01 -5.41086011e-02 -7.84012318e-01 -7.67813504e-01
2.30658546e-01 -8.53789151e-02 2.51500756e-01 9.67917621e-01
-3.12013656e-01 2.02472836e-01 -5.87761223e-01 7.31123984e-02
3.64591509e-01 -7.61014074e-02 5.57477415e-01 -6.12700224e-01
-7.86240548e-02 -7.50642180e-01 -2.33772248e-01 -2.25275353e-01
-2.36411348e-01 -8.64703715e-01 6.32231906e-02 -1.43118393e+00
3.51393759e-01 -2.20806107e-01 2.75016725e-01 4.88039911e-01
-4.57639873e-01 5.92607081e-01 4.78363454e-01 4.30442005e-01
-1.20759502e-01 1.55547872e-01 9.51800466e-01 -2.60144621e-01
-8.66609116e-05 -3.05104136e-01 -1.28064185e-01 7.47788846e-01
9.81333375e-01 -1.95255265e-01 2.58646965e-01 -2.94801265e-01
1.13203578e-01 -4.18012440e-01 -2.50951666e-03 -7.28088140e-01
1.48555515e-02 -4.59857464e-01 4.76541519e-01 -7.22021520e-01
2.02725470e-01 -3.12516958e-01 3.92566808e-02 7.01636970e-01
8.36727321e-02 3.67309362e-01 7.49698579e-02 -1.96409240e-01
-8.63798857e-02 -5.03828824e-01 1.18788719e+00 5.94113618e-02
-8.63752723e-01 -2.43008882e-01 -1.05789852e+00 -5.80263697e-02
9.11394000e-01 -3.49268764e-01 -6.29083693e-01 -1.68792754e-01
6.62501603e-02 -1.14047147e-01 5.86723387e-01 3.31369042e-01
7.87298262e-01 -9.45598960e-01 -1.32743776e+00 1.68312356e-01
1.75048500e-01 -7.66235709e-01 4.88295332e-02 6.22579396e-01
-1.57806778e+00 3.31992745e-01 -5.05427480e-01 -5.44100165e-01
-1.49234307e+00 -1.62864298e-01 -2.06635483e-02 9.09249187e-02
-5.40787816e-01 6.18582726e-01 -2.85439789e-01 -5.55985987e-01
-5.10199778e-02 3.26884449e-01 -2.60789096e-01 1.25514984e-01
3.85521501e-01 6.65848970e-01 1.03393905e-01 -1.07189727e+00
-4.82016802e-01 5.31781554e-01 -2.77234167e-01 -3.63706648e-01
1.01899695e+00 2.71979302e-01 -5.00905037e-01 4.79338348e-01
5.97600639e-01 7.26079881e-01 -8.60747993e-01 3.21278989e-01
1.95385903e-01 -6.30119979e-01 -3.55279624e-01 -9.89909291e-01
-4.91472155e-01 9.61729109e-01 4.41061199e-01 -1.86657496e-02
8.68878603e-01 -3.13007116e-01 6.01072669e-01 4.77845043e-01
-1.32439956e-01 -1.58809304e+00 6.55420199e-02 8.90326202e-01
9.24425244e-01 -7.68161416e-01 -2.94839442e-02 -2.51651973e-01
-8.52587879e-01 1.38851893e+00 3.25772226e-01 2.76283436e-02
-7.07400683e-03 2.18170524e-01 4.94469106e-01 1.57305539e-01
-2.51226872e-01 -2.24645078e-01 1.27618968e-01 6.50552154e-01
7.33605742e-01 2.57329255e-01 -5.02681434e-01 -2.75889002e-02
-4.18081939e-01 -8.01995575e-01 9.38380778e-01 8.35543871e-01
-5.27865708e-01 -9.80578542e-01 -9.83219683e-01 5.82565129e-01
-4.48545367e-01 -3.67898345e-01 -9.58586216e-01 1.10609424e+00
-1.91003934e-01 9.90584433e-01 1.09884754e-01 -9.29077491e-02
7.91284516e-02 1.24673620e-01 4.35495496e-01 -5.19539297e-01
-9.37838197e-01 4.67078626e-01 2.63226718e-01 4.71220344e-01
-2.06824783e-02 -1.04169929e+00 -1.19427562e+00 -7.85937965e-01
-1.05421059e-01 1.79746822e-01 8.68272483e-01 5.81822217e-01
-2.41835356e-01 -9.67900455e-02 6.74672365e-01 -3.33307534e-01
-2.53343642e-01 -1.08975911e+00 -7.14344025e-01 3.10832262e-01
-2.93661535e-01 -1.86355919e-01 8.48060101e-02 3.32799673e-01]
|
[11.85122299194336, 2.5887222290039062]
|
37d95336-8980-4ec1-8f9c-3ca672ecf99f
|
markov-switching-model-for-driver-behavior
|
2108.12801
| null |
https://arxiv.org/abs/2108.12801v1
|
https://arxiv.org/pdf/2108.12801v1.pdf
|
Markov Switching Model for Driver Behavior Prediction: Use cases on Smartphones
|
Several intelligent transportation systems focus on studying the various driver behaviors for numerous objectives. This includes the ability to analyze driver actions, sensitivity, distraction, and response time. As the data collection is one of the major concerns for learning and validating different driving situations, we present a driver behavior switching model validated by a low-cost data collection solution using smartphones. The proposed model is validated using a real dataset to predict the driver behavior in short duration periods. A literature survey on motion detection (specifically driving behavior detection using smartphones) is presented. Multiple Markov Switching Variable Auto-Regression (MSVAR) models are implemented to achieve a sophisticated fitting with the collected driver behavior data. This yields more accurate predictions not only for driver behavior but also for the entire driving situation. The performance of the presented models together with a suitable model selection criteria is also presented. The proposed driver behavior prediction framework can potentially be used in accident prediction and driver safety systems.
|
['Walid Gomaa', 'Mohamed A. Khamis', 'Ahmed B. Zaky']
|
2021-08-29
| null | null | null | null |
['motion-detection']
|
['computer-vision']
|
[-2.21477170e-02 -4.08714145e-01 -3.82490695e-01 -6.75212681e-01
-6.48177743e-01 -1.94096133e-01 3.59146923e-01 2.12962076e-01
-5.59002578e-01 4.95860279e-01 -1.25407711e-01 -7.39834964e-01
-4.73714501e-01 -5.24580359e-01 -2.69548029e-01 -7.83015966e-01
4.76508498e-01 3.01349878e-01 4.24546152e-01 -3.44112247e-01
3.22069585e-01 8.16872597e-01 -2.49072981e+00 -7.16754720e-02
8.54433239e-01 7.12111831e-01 6.20560408e-01 1.02555346e+00
1.88564882e-01 7.19772637e-01 -4.31714445e-01 -2.50048074e-03
-8.59267861e-02 1.10889554e-01 -1.06231973e-01 1.70150101e-02
3.14646326e-02 -1.57412350e-01 -2.39309266e-01 4.18921381e-01
3.53974432e-01 6.72713637e-01 4.55491543e-01 -1.83335149e+00
1.67244181e-01 -7.05031231e-02 1.39529100e-02 5.44023156e-01
1.56669572e-01 4.86858010e-01 -2.77588703e-02 -3.50584179e-01
2.36331602e-03 9.74223435e-01 3.62647682e-01 2.97778994e-01
-8.18306565e-01 -7.71566212e-01 -1.43053129e-01 8.52215052e-01
-1.44364381e+00 -4.89240438e-01 6.99090183e-01 -7.85393715e-01
9.61991012e-01 6.65217280e-01 6.00902259e-01 9.32721257e-01
8.58078182e-01 6.37767553e-01 1.21394479e+00 -1.70268893e-01
5.65154664e-02 8.71081233e-01 1.16287756e+00 3.04520428e-01
2.03246266e-01 4.96879548e-01 -1.05543338e-01 1.27835423e-01
-1.20434225e-01 3.73247117e-01 4.98534501e-01 1.28087670e-01
-6.45480037e-01 6.26828969e-01 -4.24260944e-01 2.35236455e-02
-7.44070172e-01 -5.95409721e-02 2.46054664e-01 1.95837408e-01
2.03948528e-01 -3.03480864e-01 -3.73800606e-01 -6.26381814e-01
-7.41775215e-01 3.57165545e-01 4.42578137e-01 1.10283256e+00
1.11435688e+00 2.13049218e-01 -6.59521222e-01 6.66150868e-01
2.54098773e-01 8.20708632e-01 3.71600568e-01 -8.41585577e-01
2.15904027e-01 6.02834821e-01 2.70371616e-01 -9.00746584e-01
-7.89211035e-01 -3.23483735e-01 -4.12981927e-01 1.83748290e-01
6.91526309e-02 -3.69003385e-01 -6.08383834e-01 1.29565108e+00
2.96829313e-01 3.75188053e-01 3.99005450e-02 4.19788957e-01
6.13300920e-01 7.34196603e-01 4.55414623e-01 -3.60414207e-01
1.36490452e+00 -6.18780494e-01 -1.07053244e+00 -1.81652874e-01
9.27906632e-01 -6.33960545e-01 7.98908949e-01 4.12849098e-01
-8.88910234e-01 -1.11563003e+00 -5.15090644e-01 9.28188413e-02
-6.00739837e-01 2.92266995e-01 3.51582140e-01 1.11754060e+00
-7.25437582e-01 -2.87400872e-01 -8.98570061e-01 -3.89112771e-01
-1.25250772e-01 5.69773793e-01 -1.16862528e-01 -1.55829627e-03
-1.05999434e+00 1.30720854e+00 -3.79004180e-02 -2.26446316e-02
-9.21476066e-01 -7.05930352e-01 -8.44286561e-01 -1.53741568e-01
4.22047406e-01 -5.30923009e-01 1.20390952e+00 -3.87700826e-01
-1.45966375e+00 4.65481132e-01 -8.60323668e-01 -6.78526700e-01
4.44938779e-01 -4.47599888e-02 -1.22075355e+00 -5.58680773e-01
-2.20223819e-03 1.58444136e-01 6.06661618e-01 -8.63742173e-01
-1.11319482e+00 -4.01160717e-01 -3.66898596e-01 -7.66345114e-02
1.36473835e-01 1.70021683e-01 -2.68337905e-01 -3.84924263e-02
-8.37383270e-01 -1.28272498e+00 -1.66059002e-01 -1.02851593e+00
-1.54533356e-01 -4.46146041e-01 1.09538269e+00 -6.27718985e-01
1.73243451e+00 -2.11968374e+00 -5.71880400e-01 4.32160795e-01
-2.56503016e-01 3.43439251e-01 1.50931790e-01 3.15097958e-01
2.06560530e-02 -6.04272246e-01 1.34131655e-01 -3.66559952e-01
-2.23997042e-01 3.37823063e-01 1.15289263e-01 4.62004691e-01
4.49764356e-02 8.35924029e-01 -3.75338942e-01 -3.64030242e-01
1.20729840e+00 5.29979825e-01 -2.28200257e-01 4.15523231e-01
4.98010039e-01 5.48746824e-01 -2.58829653e-01 1.65765405e-01
6.85811400e-01 6.48358762e-01 -6.73254669e-01 9.73668024e-02
-7.60862172e-01 -4.57523428e-02 -1.15849733e+00 6.98851407e-01
-8.98876905e-01 8.92522514e-01 -3.68428051e-01 -9.78934646e-01
1.07818222e+00 2.12724924e-01 6.77661002e-01 -7.73806155e-01
4.11714584e-01 -3.23848099e-01 1.90502778e-01 -1.16868258e+00
5.76002479e-01 2.71680266e-01 -1.43422261e-01 1.33737743e-01
-2.93761581e-01 2.64160573e-01 3.87451977e-01 -2.97269255e-01
6.97873890e-01 -3.77922833e-01 2.54578769e-01 -2.39505559e-01
1.04103434e+00 3.44377846e-01 4.86941427e-01 5.79513550e-01
-6.23844206e-01 -2.50646144e-01 -1.00800078e-02 -2.91953653e-01
-5.28364539e-01 -6.60583913e-01 -1.06928326e-01 1.05100608e+00
1.88893542e-01 1.48984760e-01 -8.00960362e-01 3.45930979e-02
1.28545947e-02 1.56778395e+00 -4.94686782e-01 -7.25411773e-01
-2.25748181e-01 -5.40534019e-01 1.20353296e-01 3.79072219e-01
2.69899786e-01 -8.19531977e-01 -8.38206351e-01 2.21447363e-01
-1.01537434e-02 -1.10443079e+00 -1.68295473e-01 -1.65451884e-01
-4.29236054e-01 -9.32310283e-01 -1.16839938e-01 -4.21946943e-01
2.19465762e-01 8.09703112e-01 5.99064827e-01 -1.85795724e-02
-3.17926437e-01 8.33283603e-01 1.33858100e-01 -1.15760124e+00
-7.85240591e-01 6.84159473e-02 1.40658140e-01 4.77193534e-01
1.15267301e+00 2.10133091e-01 -5.63895226e-01 7.45244563e-01
-6.64177656e-01 -1.64433643e-01 6.07202947e-01 1.67527974e-01
4.40716594e-01 3.94698411e-01 6.52170062e-01 -5.97395778e-01
8.72697532e-01 -7.75578916e-01 -8.45415890e-01 3.09770435e-01
-9.98068452e-01 -2.43280470e-01 2.72617042e-01 -3.41096342e-01
-1.37709236e+00 1.77755728e-01 -2.16032922e-01 -1.98762938e-01
-1.05506051e+00 1.06284894e-01 -1.25742078e-01 1.11398883e-01
3.12149346e-01 3.53553176e-01 4.99804646e-01 -1.65772542e-01
-7.37575516e-02 9.42752182e-01 3.55923444e-01 3.05606514e-01
2.47620121e-01 1.67095974e-01 2.18563408e-01 -1.20833898e+00
-2.00384796e-01 -1.13803625e+00 -5.46715319e-01 -9.53316867e-01
9.40119982e-01 -7.68595755e-01 -1.46073592e+00 4.31603044e-01
-6.87313676e-01 -1.31507948e-01 7.86930621e-02 9.94654238e-01
-6.83373809e-01 -1.18292756e-01 -6.68538269e-03 -1.70204842e+00
2.32912496e-01 -1.59458566e+00 6.94640875e-01 1.73528552e-01
-3.83879274e-01 -1.04300523e+00 8.11974928e-02 8.48974645e-01
4.57718551e-01 -8.70028883e-02 5.95224738e-01 -5.37765205e-01
-4.72795367e-01 -7.12703884e-01 2.36580640e-01 2.54191935e-01
1.12481117e-01 3.22987348e-01 -8.94495368e-01 2.61373427e-02
-7.15172663e-02 6.18525505e-01 5.31804681e-01 9.89917696e-01
1.15847492e+00 1.31331265e-01 -5.80607474e-01 -4.82980274e-02
1.10234618e+00 7.12811112e-01 7.69139528e-01 2.65468508e-01
4.74224687e-01 1.17830884e+00 1.17484534e+00 3.25563133e-01
7.47015834e-01 7.55961537e-01 2.26513341e-01 -6.25016615e-02
-8.61623604e-03 5.00504114e-02 5.81931829e-01 4.99481529e-01
-3.05655468e-02 -2.72558242e-01 -1.04563868e+00 6.16655946e-01
-1.94707143e+00 -1.28628647e+00 -9.68221128e-01 2.36085463e+00
-1.29171014e-01 6.85781538e-02 8.50155175e-01 5.08204341e-01
6.50175691e-01 -5.53847194e-01 -4.50379014e-01 -8.85162771e-01
2.04437137e-01 -2.11416900e-01 1.08165503e+00 8.86310279e-01
-7.16300488e-01 6.97221041e-01 6.77231550e+00 7.36094236e-01
-9.93383050e-01 1.90314725e-01 4.59488600e-01 -9.64706093e-02
3.62625793e-02 -1.80022344e-01 -1.24612796e+00 7.27614939e-01
1.87579381e+00 -3.87325436e-01 1.55579418e-01 8.64695132e-01
1.27638328e+00 -5.69001198e-01 -5.26357591e-01 1.05727875e+00
-1.54367626e-01 -8.65672350e-01 -3.09125096e-01 1.05901167e-01
3.31521034e-01 -2.11968243e-01 3.45694363e-01 5.80250025e-01
-1.79557428e-01 -7.49999344e-01 3.36604536e-01 1.16160810e+00
1.05443403e-01 -1.25600553e+00 9.32531178e-01 7.12898254e-01
-8.72989893e-01 -5.13540804e-01 -2.48207785e-02 -1.94230095e-01
4.15184349e-01 5.18669605e-01 -9.77546096e-01 2.73130506e-01
4.46918845e-01 6.65387392e-01 -7.40000308e-01 1.07578623e+00
5.42611361e-01 9.24797058e-01 -3.89178880e-02 -3.17856133e-01
4.73959297e-02 -4.02086943e-01 5.10316730e-01 1.42160916e+00
5.51520705e-01 -2.13315636e-01 -1.26312554e-01 5.71866393e-01
8.54933083e-01 1.83293715e-01 -9.20681000e-01 3.73684347e-01
2.26738989e-01 1.17291725e+00 -2.04837278e-01 -2.14004233e-01
-7.06335783e-01 1.85498908e-01 -3.89203727e-01 3.87153536e-01
-1.09809148e+00 -2.77176708e-01 1.14337456e+00 4.62443292e-01
-2.09801689e-01 -4.00214553e-01 -3.60661596e-01 -2.26122573e-01
-4.03069347e-01 -2.20938742e-01 2.48629838e-01 -6.64619625e-01
-5.84937155e-01 3.26718628e-01 6.13360822e-01 -1.32333720e+00
-4.25332159e-01 -5.75152040e-01 -5.48714161e-01 8.58044267e-01
-1.53818107e+00 -6.84193611e-01 -5.62774837e-01 9.74082232e-01
8.15679669e-01 -5.19913197e-01 2.36596897e-01 6.49007022e-01
-1.06844592e+00 5.59726477e-01 -9.46239680e-02 -6.31453395e-01
4.32133645e-01 -6.10858023e-01 -6.71795569e-03 7.60702729e-01
-5.39943933e-01 4.59248245e-01 1.05509782e+00 -5.95924079e-01
-1.40649283e+00 -1.26862061e+00 8.25664580e-01 -8.44942391e-01
1.04884125e-01 -5.78865483e-02 -8.93567681e-01 5.15528262e-01
4.33279872e-02 -6.51295781e-01 9.89155114e-01 -2.27945551e-01
6.95327520e-01 -5.36770582e-01 -9.37193215e-01 4.45829093e-01
4.16357696e-01 -3.39511722e-01 -3.34124357e-01 7.77582377e-02
2.02335864e-01 -2.47059874e-02 -5.22159636e-01 1.28688887e-01
4.98388410e-01 -9.83913183e-01 6.60781026e-01 -6.93607152e-01
-3.92967135e-01 -2.63521522e-01 2.05654189e-01 -1.12006497e+00
-4.67465758e-01 -6.20301783e-01 5.06733917e-02 9.51994896e-01
5.37062168e-01 -8.01537871e-01 4.89095718e-01 1.30519676e+00
-1.97898149e-01 -3.80321085e-01 -7.90527999e-01 -7.86946654e-01
-4.66323614e-01 -1.27783453e+00 7.08469152e-01 2.10853860e-01
-3.80766034e-01 1.80640325e-01 -5.77248693e-01 3.94134849e-01
4.39774543e-01 -2.95161903e-01 1.18032849e+00 -1.24257815e+00
4.13969845e-01 -3.67256254e-01 -7.48681605e-01 -7.31953502e-01
4.04394567e-01 -4.33444172e-01 8.42970759e-02 -1.23461831e+00
-1.04519531e-01 -1.66022882e-01 -4.58495945e-01 -1.62666991e-01
-1.71222419e-01 -3.96367818e-01 -3.61559212e-01 -2.11709484e-01
-1.11714371e-01 1.95986122e-01 7.24835336e-01 1.16635710e-01
-4.52850103e-01 9.96256709e-01 -3.17584246e-01 2.24043667e-01
1.00062752e+00 -4.56896842e-01 -8.33775282e-01 4.74815257e-02
-1.52760893e-01 1.77441657e-01 5.67975402e-01 -1.18068397e+00
4.68678653e-01 -7.03619301e-01 -2.22011328e-01 -1.09545898e+00
4.74441051e-01 -1.33466637e+00 6.32143617e-01 5.63845396e-01
-4.15200830e-01 4.62249517e-01 5.08672416e-01 7.49752760e-01
-5.52812852e-02 1.10370152e-01 6.03318751e-01 4.99705404e-01
-1.02778542e+00 2.90060580e-01 -1.33340001e+00 -6.97778165e-01
1.87608826e+00 -6.86759710e-01 2.23097540e-02 -6.34101570e-01
-7.91202486e-01 4.33837086e-01 -2.09448755e-01 1.00121450e+00
5.37875414e-01 -1.06812119e+00 -5.01294255e-01 6.06896639e-01
4.01825398e-01 -7.70458400e-01 7.95985222e-01 1.31332636e+00
-5.66132180e-02 9.92581487e-01 -3.69601727e-01 -8.11985075e-01
-1.61354828e+00 9.03003097e-01 3.50148171e-01 3.08782429e-01
-4.18198258e-02 1.92335062e-03 -1.43423423e-01 -4.16014567e-02
1.55710923e-02 -3.64487827e-01 -5.40073335e-01 -1.52716532e-01
5.57581127e-01 1.30458236e+00 3.18107456e-01 -1.15631235e+00
-4.83182460e-01 3.77954453e-01 3.45319539e-01 3.26018520e-02
7.40236998e-01 -7.75596917e-01 4.77609396e-01 8.77717197e-01
1.13166189e+00 -2.49326944e-01 -9.83631551e-01 2.11600304e-01
8.61008912e-02 -4.31436956e-01 4.68854785e-01 -4.62503523e-01
-8.02750468e-01 8.90131056e-01 1.30312347e+00 2.79919684e-01
9.25751865e-01 -3.08676422e-01 8.25177789e-01 2.24405900e-01
1.63870186e-01 -1.39433110e+00 -7.21394658e-01 1.80566892e-01
4.90514785e-01 -1.65820706e+00 -5.47752261e-01 -2.45916858e-01
-1.13066769e+00 7.10212529e-01 5.38901448e-01 1.96071684e-01
1.17911804e+00 2.78478920e-01 2.47602090e-01 -7.36693619e-03
-9.06304479e-01 -6.00943387e-01 5.31809926e-01 8.23781610e-01
2.04251871e-01 4.15269703e-01 -3.99423122e-01 4.13443476e-01
-1.18017927e-01 2.01275364e-01 5.08696318e-01 6.19364738e-01
-4.56459939e-01 -8.84652972e-01 -4.53619689e-01 5.98282933e-01
-9.43846628e-02 5.16033411e-01 4.72982787e-02 6.80866003e-01
1.98179796e-01 1.73969972e+00 1.35360166e-01 -5.61209917e-01
8.79276991e-01 2.71268159e-01 -3.12349707e-01 -2.83072621e-01
-4.96561170e-01 -4.01308417e-01 5.04277907e-02 -5.72765946e-01
-5.16205430e-01 -9.01828110e-01 -8.39029014e-01 -6.25928998e-01
-1.69704959e-01 -1.01390667e-02 1.06022203e+00 1.10913193e+00
7.14316010e-01 7.45265543e-01 9.20482039e-01 -1.93519935e-01
3.69669572e-02 -8.59855413e-01 -3.94931883e-01 2.94185936e-01
5.81116796e-01 -9.64256763e-01 -2.28196651e-01 1.11618079e-02]
|
[5.871513366699219, 1.0351743698120117]
|
03bcd330-dabc-4f9f-a1bb-20c50d19d73c
|
meta-learning-the-step-size-in-policy
| null | null |
https://openreview.net/forum?id=zRn12do9p0
|
https://openreview.net/pdf?id=zRn12do9p0
|
Meta Learning the Step Size in Policy Gradient Methods
|
Policy-based algorithms are among the most widely adopted techniques in model-free RL, thanks to their strong theoretical groundings and good properties in continuous action spaces. Unfortunately, these methods require precise and problem-specific hyperparameter tuning to achieve good performance and, as a consequence, they tend to struggle when asked to accomplish a series of heterogeneous tasks. In particular, the selection of the step size has a crucial impact on the ability to learn a highly performing policy, affecting the speed and the stability of the training process, and often being the main culprit for poor results. In this paper, we tackle these issues with a Meta Reinforcement Learning approach, by introducing a new formulation, known as meta-MDP, that can be used to solve any hyperparameter selection problem in RL with contextual processes. After providing a theoretical Lipschitz bound to the performance in different tasks, we adopt the proposed framework to train a batch RL algorithm to dynamically recommend the most adequate step size for different policies and tasks. In conclusion, we present an experimental campaign to show the advantages of selecting an adaptive learning rate in heterogeneous environments.
|
['Marcello Restelli', 'Francesco Corda', 'Luca Sabbioni']
|
2021-05-20
| null | null | null |
icml-workshop-automl-2021-7
|
['policy-gradient-methods']
|
['methodology']
|
[ 2.01903507e-01 -1.84231594e-01 -3.36167246e-01 7.48072639e-02
-7.59158552e-01 -3.75951380e-01 6.68134749e-01 2.75004655e-01
-8.72940779e-01 1.04426908e+00 -2.47446537e-01 -2.77033877e-02
-5.33616364e-01 -4.50723380e-01 -4.60561395e-01 -1.08494210e+00
2.67508686e-01 5.54942667e-01 2.07673043e-01 -2.46275440e-01
5.44362903e-01 3.76298249e-01 -1.73881960e+00 -4.24396425e-01
9.50426042e-01 9.29105043e-01 6.06050193e-01 4.36092675e-01
-9.53643247e-02 7.43506491e-01 -6.99815154e-01 -6.78227544e-02
1.15898222e-01 -6.01513624e-01 -4.70571548e-01 4.13246602e-02
-2.20495209e-01 2.78101061e-02 4.46403682e-01 8.25984120e-01
8.28868032e-01 6.77309752e-01 7.09440649e-01 -8.94518793e-01
4.31567699e-01 5.31641185e-01 -3.82533818e-01 1.39165297e-01
1.44884154e-01 9.03589353e-02 9.49234068e-01 -1.93983182e-01
4.64532256e-01 1.10968542e+00 1.23841474e-02 6.90989912e-01
-1.27742958e+00 -4.19659317e-01 4.39256340e-01 2.54258633e-01
-1.06223917e+00 -3.86638403e-01 6.36994421e-01 -3.20902228e-01
4.13552463e-01 9.19686332e-02 6.33961499e-01 1.17250037e+00
8.18631425e-02 6.40472829e-01 1.23071837e+00 -5.45368314e-01
8.64875495e-01 2.54166484e-01 -4.25859779e-01 3.13110828e-01
2.22316712e-01 -3.69208120e-02 -3.62509042e-01 -2.11270034e-01
6.20143473e-01 -3.90292525e-01 -2.12887600e-01 -6.09831333e-01
-7.79173970e-01 8.04180801e-01 1.51382402e-01 6.02166831e-01
-6.10528290e-01 2.62213558e-01 3.77862632e-01 2.94028312e-01
1.87184662e-01 7.97628641e-01 -2.94638902e-01 -3.64449382e-01
-6.02533758e-01 5.52292764e-01 6.12417936e-01 2.29332626e-01
3.34796607e-01 -1.22042872e-01 -4.16781783e-01 7.93071032e-01
6.55462369e-02 2.68439353e-01 7.81565368e-01 -9.69768584e-01
5.23296475e-01 2.90049911e-01 7.71692455e-01 -8.53172660e-01
-3.99292916e-01 -7.14998186e-01 -5.03656089e-01 4.02733088e-01
6.31117523e-01 -3.58236283e-01 -1.90605283e-01 1.71135509e+00
6.18193507e-01 2.08095443e-02 5.57358712e-02 9.31741953e-01
-2.49146163e-01 5.36941469e-01 4.39125717e-01 -6.23178840e-01
1.06733143e+00 -6.75317824e-01 -7.74194241e-01 -2.25891665e-01
3.28285366e-01 -4.85809386e-01 1.23134530e+00 5.87341011e-01
-1.05258787e+00 -5.43798625e-01 -9.22544718e-01 7.32833922e-01
-1.11693382e-01 2.12263763e-01 2.32362449e-01 4.23463285e-01
-6.48967147e-01 7.96869099e-01 -7.28715241e-01 -2.25086242e-01
-3.84131223e-02 3.77918929e-01 1.88907444e-01 1.83765620e-01
-1.00559270e+00 1.00176525e+00 7.58774519e-01 1.36135310e-01
-8.48247886e-01 -2.26377413e-01 -1.57088682e-01 1.79541409e-01
9.53012168e-01 -5.41080058e-01 1.35077333e+00 -1.13349044e+00
-2.07164073e+00 1.89769566e-01 1.60651103e-01 -5.92818201e-01
1.06640792e+00 -2.95105934e-01 -3.68247144e-02 1.67397484e-02
-2.69486129e-01 1.51470944e-01 1.39703751e+00 -1.20055103e+00
-7.98211455e-01 -2.90650159e-01 1.92873895e-01 7.04533815e-01
-3.23845297e-01 -2.54998386e-01 -2.77011186e-01 -3.35133880e-01
-2.64908969e-01 -9.92532790e-01 -3.97642910e-01 -7.02664912e-01
-3.74659896e-02 -2.67945647e-01 3.22820514e-01 -8.24550241e-02
1.27787423e+00 -1.95063961e+00 6.56518757e-01 1.80666834e-01
-1.70577407e-01 4.03574616e-01 4.74489704e-02 3.13746691e-01
5.43427825e-01 -1.22488432e-01 -2.20330767e-02 -2.67282754e-01
4.01735306e-02 9.81556997e-02 -9.78751108e-02 2.60764927e-01
-5.36219366e-02 3.65389287e-01 -9.39864039e-01 -4.36246842e-01
2.46257603e-01 3.48081559e-01 -3.61588359e-01 4.05528575e-01
-7.29073584e-01 9.11500335e-01 -1.07829142e+00 -3.16722468e-02
3.83948795e-02 -6.51336461e-02 4.91301864e-01 1.31526798e-01
-2.14057460e-01 -1.53438538e-01 -1.33133984e+00 1.34944332e+00
-9.75792468e-01 3.78163345e-02 1.88714013e-01 -1.06081831e+00
9.73141372e-01 3.31241637e-01 6.03311777e-01 -7.92886019e-01
4.16465968e-01 3.65832895e-01 -3.11605353e-02 -4.43525702e-01
4.51057583e-01 -1.20293848e-01 1.19796097e-01 4.11078662e-01
-3.45883697e-01 -2.32106462e-01 3.49144131e-01 -3.62004459e-01
8.16997230e-01 4.26099926e-01 4.60076749e-01 -3.23892049e-02
1.05162966e+00 -2.83668011e-01 3.79722774e-01 8.10074747e-01
-1.20838732e-01 2.00991817e-02 6.33909464e-01 -2.29972482e-01
-7.50390172e-01 -3.70138913e-01 9.80385616e-02 1.27869260e+00
2.24431138e-02 1.00580774e-01 -7.57889509e-01 -5.37919581e-01
-1.88221633e-01 8.05772245e-01 -4.90458280e-01 -1.61084995e-01
-7.13950753e-01 -7.44030118e-01 -5.00760935e-02 -7.37070292e-02
4.91486043e-01 -1.35691845e+00 -1.13116789e+00 5.01507998e-01
-1.27309754e-01 -9.99236524e-01 -2.68049240e-02 3.66339982e-01
-8.99938762e-01 -1.03413856e+00 -1.06141448e+00 -2.60472298e-01
3.61642808e-01 -1.25482664e-01 8.86609733e-01 -2.16276217e-02
1.35702953e-01 2.39162713e-01 -3.85745585e-01 -3.14026535e-01
-6.34159207e-01 5.76983213e-01 -6.40003309e-02 1.94199368e-01
-3.59872550e-01 -2.87047714e-01 -5.50369680e-01 5.17514765e-01
-8.64493430e-01 -1.49935693e-01 6.98999882e-01 6.87918365e-01
7.21291184e-01 3.36933821e-01 7.65694439e-01 -9.09864366e-01
1.10262442e+00 -2.11276770e-01 -9.85277653e-01 4.66412842e-01
-7.54063487e-01 4.96953666e-01 9.42593575e-01 -5.83617628e-01
-1.11361587e+00 2.34025214e-02 2.69484334e-02 -2.53182560e-01
-1.02948032e-01 2.00771570e-01 -1.39223009e-01 -2.94216834e-02
5.72169065e-01 1.74681976e-01 4.28428836e-02 -3.25207233e-01
2.07019567e-01 3.43409091e-01 -2.43583713e-02 -8.63560557e-01
4.42186981e-01 1.27682269e-01 1.42213464e-01 -6.18121684e-01
-9.92522717e-01 -4.21226650e-01 -3.54570568e-01 -3.75450373e-01
6.17912889e-01 -4.74489093e-01 -8.22073042e-01 2.17349142e-01
-6.33561790e-01 -6.81063414e-01 -2.22970665e-01 4.88383204e-01
-1.08401430e+00 -1.15766367e-02 2.65517961e-02 -1.07116830e+00
-2.27539212e-01 -1.21871734e+00 7.38757014e-01 3.31470966e-01
3.29515338e-02 -9.76647735e-01 2.77556628e-01 3.26216787e-01
5.36574900e-01 2.43883014e-01 9.18314040e-01 -4.85789955e-01
-2.67445892e-01 1.15227059e-01 3.15167874e-01 2.07633033e-01
-4.22941484e-02 -2.50208318e-01 -7.20031321e-01 -3.59784037e-01
1.95497841e-01 -4.51123267e-01 4.96620387e-01 4.85872000e-01
1.03752756e+00 -1.59526139e-01 -8.23537782e-02 1.73823342e-01
1.44425261e+00 3.58868510e-01 2.70945102e-01 8.59907389e-01
6.84016496e-02 3.86369050e-01 1.08017838e+00 8.67810845e-01
-1.18509382e-01 1.07171237e+00 5.34952879e-01 3.95010591e-01
3.23534578e-01 -1.06082432e-01 3.27767462e-01 2.53066957e-01
-2.84552217e-01 -2.74075389e-01 -4.36054885e-01 9.61939469e-02
-2.01565146e+00 -8.17446589e-01 4.50447023e-01 2.63784719e+00
5.97956479e-01 3.47514480e-01 4.79916334e-01 2.60498136e-01
6.69824302e-01 2.10486889e-01 -7.19211757e-01 -3.01337451e-01
2.47117132e-01 4.92207659e-03 4.20720637e-01 3.29043001e-01
-8.78179073e-01 9.23350334e-01 5.10763645e+00 1.05601597e+00
-1.20128977e+00 2.27676835e-02 4.72493023e-01 -2.22931489e-01
-6.22756220e-03 -2.94446737e-01 -7.76689529e-01 4.84143913e-01
1.04840827e+00 1.05248950e-02 6.85780585e-01 7.16299176e-01
7.17961013e-01 -3.89708996e-01 -6.77534938e-01 8.47987533e-01
-2.85009384e-01 -6.81322515e-01 -3.74752432e-01 -8.80379379e-02
7.22392917e-01 -3.81086886e-01 1.66770861e-01 3.68862808e-01
9.22635719e-02 -8.12245369e-01 7.37361968e-01 2.71283656e-01
2.13941202e-01 -1.05065143e+00 7.68489778e-01 6.44164503e-01
-7.35290468e-01 -4.42300230e-01 -4.28944737e-01 2.69535154e-01
-3.27045806e-02 2.81208843e-01 -7.85457551e-01 4.43091303e-01
9.65389237e-02 5.85918240e-02 -1.58159658e-01 1.08056855e+00
-3.73258203e-01 3.41404408e-01 -1.92619443e-01 -5.67522168e-01
4.31767642e-01 -1.94720775e-01 4.16161478e-01 6.19335294e-01
4.14726853e-01 -1.50540635e-01 5.76382354e-02 4.47982550e-01
1.93071172e-01 5.31941473e-01 -2.90851682e-01 -2.91408766e-02
4.01713938e-01 1.15065849e+00 -8.15142393e-01 1.45390462e-02
2.18348801e-01 6.93136275e-01 4.32740003e-01 3.61581683e-01
-8.59205782e-01 2.63212286e-02 3.30034405e-01 4.40858975e-02
3.67559046e-01 -2.87363529e-01 1.60812616e-01 -7.66168416e-01
-2.33512312e-01 -1.03569007e+00 3.35553825e-01 -1.86740741e-01
-7.73787379e-01 5.25716543e-01 6.36045858e-02 -1.16085470e+00
-5.93559146e-01 -2.48633921e-01 -2.31912568e-01 5.27515531e-01
-1.58971596e+00 -3.25921208e-01 -1.70938790e-01 5.56015253e-01
6.29212916e-01 -1.64601743e-01 5.15226245e-01 2.88076308e-02
-6.80551827e-01 2.04141945e-01 4.08780426e-01 -6.52645648e-01
6.28819227e-01 -1.08213782e+00 -3.86737227e-01 2.94249594e-01
-9.44660082e-02 1.02877662e-01 1.01001668e+00 -3.00759614e-01
-1.37875307e+00 -6.49558663e-01 3.58871877e-01 1.18640952e-01
5.33157170e-01 -6.48033712e-03 -5.47355294e-01 5.09382086e-03
-2.18588173e-01 -3.16867620e-01 2.12138072e-01 1.76268473e-01
3.82187426e-01 -2.50261724e-01 -1.08089387e+00 6.73192203e-01
6.74122930e-01 6.59777224e-03 -1.37552455e-01 2.49987289e-01
3.47515285e-01 -4.11050320e-01 -6.59180522e-01 5.73495664e-02
4.35032189e-01 -1.08226252e+00 7.41480589e-01 -3.20198268e-01
-9.98864695e-02 -5.50543629e-02 1.85706303e-01 -1.53600502e+00
-3.12353019e-02 -6.66003585e-01 -1.73089534e-01 9.17164147e-01
2.82127470e-01 -6.18545413e-01 7.25982845e-01 4.20849830e-01
5.14196634e-01 -8.20942223e-01 -8.82808924e-01 -7.17204392e-01
-1.87230259e-01 -1.51356161e-01 4.75650549e-01 3.16853672e-01
-4.27361846e-01 2.37347305e-01 -4.67252016e-01 -1.21914856e-01
5.13082266e-01 9.21582654e-02 7.69135296e-01 -1.20002711e+00
-6.65254593e-01 -6.01732135e-01 1.00307144e-01 -7.37581432e-01
2.85757989e-01 -1.56311065e-01 4.12661642e-01 -1.05302787e+00
-2.50539511e-01 -8.27224672e-01 -5.07440448e-01 -5.01115285e-02
-1.98426381e-01 -4.16925371e-01 4.48874354e-01 3.37477267e-01
-8.24071825e-01 6.64694726e-01 1.49555612e+00 1.04187049e-01
-6.90110147e-01 5.90953469e-01 -3.34061831e-01 4.26316351e-01
1.07921648e+00 -4.39631790e-01 -7.68667519e-01 -1.48652539e-01
5.07005811e-01 4.13270563e-01 -1.13909632e-01 -1.15041828e+00
-5.41599281e-02 -5.00899017e-01 7.63997249e-03 2.89913993e-02
3.91489863e-01 -9.84208405e-01 3.88932787e-02 5.70724666e-01
-7.30518401e-01 -1.70462299e-03 -3.06023974e-02 7.54577637e-01
5.26973270e-02 -6.40026212e-01 8.67149413e-01 -2.00208575e-01
-5.84912121e-01 1.73899382e-02 -3.11696708e-01 9.07820910e-02
1.13979173e+00 2.00222909e-01 1.95898652e-01 -3.92744809e-01
-7.39867210e-01 2.41998553e-01 2.70259500e-01 3.78095418e-01
1.52584553e-01 -5.90930521e-01 -4.09737080e-01 -9.52643678e-02
-1.60281509e-01 -2.23972529e-01 1.05372481e-01 8.57510388e-01
-2.31193691e-01 4.60558861e-01 -2.38295376e-01 -3.38954210e-01
-9.38806415e-01 4.75796729e-01 4.24661458e-01 -7.81784177e-01
-4.84367728e-01 3.51335853e-01 -3.26324642e-01 1.48191020e-01
5.91744184e-01 -1.84325695e-01 -5.96644700e-01 2.71338731e-01
4.21573907e-01 4.47305113e-01 3.13926548e-01 -1.74886137e-01
-3.35953608e-02 5.33909261e-01 1.86542720e-01 -4.40505296e-01
1.03727543e+00 -2.18310043e-01 4.08074349e-01 4.94030893e-01
5.65583527e-01 -9.34897065e-02 -1.44044304e+00 -8.53849351e-02
2.24706009e-01 -3.20343107e-01 1.98553517e-01 -7.51470566e-01
-7.90893614e-01 4.69067663e-01 6.99724495e-01 3.51454943e-01
1.09886539e+00 -3.57016891e-01 3.74727815e-01 5.23160160e-01
6.18812084e-01 -1.44824827e+00 1.56006232e-01 2.48850554e-01
6.27604008e-01 -1.09156823e+00 -9.37627181e-02 1.85742095e-01
-8.93521547e-01 1.05128396e+00 4.44897979e-01 2.87278537e-02
1.60707965e-01 -3.90374772e-02 7.83114880e-02 1.52936503e-01
-8.42357397e-01 -3.72262418e-01 -4.04150747e-02 2.68848062e-01
3.19399178e-01 7.43845627e-02 -8.07674527e-01 -7.94439390e-02
1.89175859e-01 1.01435602e-01 8.71570036e-02 8.28501403e-01
-6.03692532e-01 -1.46183693e+00 -4.75335956e-01 1.90155432e-01
-5.02450466e-01 5.77516139e-01 -1.88168208e-03 6.90749109e-01
-3.22211124e-02 9.12216187e-01 -2.51033783e-01 1.10346302e-01
3.23286176e-01 8.18147734e-02 7.01316237e-01 -3.11972469e-01
-6.89495325e-01 1.68004632e-01 2.82244105e-03 -5.23744583e-01
-5.46544671e-01 -6.46093547e-01 -1.14891350e+00 1.92188229e-02
-2.38028124e-01 4.85329330e-01 9.57190096e-01 1.11499596e+00
1.15577236e-01 5.62298536e-01 7.39075720e-01 -7.10574448e-01
-1.15894818e+00 -8.56606781e-01 -4.60867047e-01 2.65097827e-01
1.01298444e-01 -1.00704968e+00 -2.96732694e-01 -6.36428535e-01]
|
[4.305161476135254, 2.3530526161193848]
|
1a06d11d-d55d-4755-b284-32d04f6f4f1b
|
using-implicit-feedback-to-improve-question
|
2304.13664
| null |
https://arxiv.org/abs/2304.13664v1
|
https://arxiv.org/pdf/2304.13664v1.pdf
|
Using Implicit Feedback to Improve Question Generation
|
Question Generation (QG) is a task of Natural Language Processing (NLP) that aims at automatically generating questions from text. Many applications can benefit from automatically generated questions, but often it is necessary to curate those questions, either by selecting or editing them. This task is informative on its own, but it is typically done post-generation, and, thus, the effort is wasted. In addition, most existing systems cannot incorporate this feedback back into them easily. In this work, we present a system, GEN, that learns from such (implicit) feedback. Following a pattern-based approach, it takes as input a small set of sentence/question pairs and creates patterns which are then applied to new unseen sentences. Each generated question, after being corrected by the user, is used as a new seed in the next iteration, so more patterns are created each time. We also take advantage of the corrections made by the user to score the patterns and therefore rank the generated questions. Results show that GEN is able to improve by learning from both levels of implicit feedback when compared to the version with no learning, considering the top 5, 10, and 20 questions. Improvements go up from 10%, depending on the metric and strategy used.
|
['Luisa Coheur', 'Eric Nyberg', 'Hugo Rodrigues']
|
2023-04-26
| null | null | null | null |
['question-generation']
|
['natural-language-processing']
|
[ 5.91013014e-01 4.05977577e-01 4.34940875e-01 -3.28653157e-01
-9.10667717e-01 -7.79724956e-01 6.99553072e-01 7.48663068e-01
-5.74272871e-01 9.17580605e-01 3.29068750e-01 -3.91402662e-01
7.22380867e-03 -9.08789933e-01 -5.02325118e-01 -3.49687755e-01
3.63457471e-01 7.97757089e-01 6.11477673e-01 -5.27060032e-01
5.30173302e-01 1.21228114e-01 -1.67878222e+00 5.66166878e-01
1.33101606e+00 4.30656463e-01 5.52920699e-01 9.66076851e-01
-7.28789926e-01 9.90281522e-01 -8.39697123e-01 -4.87544656e-01
2.02000551e-02 -1.02953923e+00 -1.04304552e+00 7.51732737e-02
3.15403759e-01 -8.36477876e-02 3.06047648e-01 7.59596050e-01
4.11158741e-01 3.96644235e-01 5.13256967e-01 -8.11136842e-01
-4.71485615e-01 6.88613176e-01 1.23979405e-01 8.23798254e-02
8.87607276e-01 7.41819814e-02 1.10772872e+00 -1.28554106e+00
6.43702090e-01 1.08265793e+00 3.14164609e-01 8.97714496e-01
-1.06119716e+00 -2.20592335e-01 5.63462712e-02 2.17177048e-01
-8.10718358e-01 -2.96329618e-01 5.69823861e-01 -4.11034852e-01
8.42188537e-01 4.98982996e-01 3.38982433e-01 6.80543780e-01
-8.47550482e-02 7.61588812e-01 9.99790668e-01 -9.79828298e-01
4.40273046e-01 6.15227103e-01 3.63435358e-01 4.72748697e-01
-3.44904736e-02 -1.68862328e-01 -4.01216507e-01 -1.66446030e-01
6.03509098e-02 -4.17546540e-01 -3.97534341e-01 1.07294776e-01
-8.24896336e-01 8.43822420e-01 2.57704467e-01 6.51322186e-01
-5.13919652e-01 -4.12652880e-01 -4.84971479e-02 6.62150443e-01
4.47305292e-01 1.05272794e+00 -6.05327666e-01 -2.59099662e-01
-1.09300256e+00 5.53633988e-01 1.15983796e+00 6.40619040e-01
1.07976472e+00 -5.19800782e-01 -7.79977083e-01 9.31118667e-01
1.45325735e-01 2.28295222e-01 7.62912452e-01 -5.99690497e-01
4.31021422e-01 9.94970798e-01 4.59410727e-01 -1.00456750e+00
-1.74357489e-01 -2.16165613e-02 -4.11866516e-01 3.12930852e-01
5.31973779e-01 -3.42746973e-01 -1.00344276e+00 1.53512692e+00
3.23499829e-01 -3.35034817e-01 -4.05511260e-02 5.90997458e-01
7.70800710e-01 8.96605194e-01 -4.20834385e-02 -4.04577225e-01
1.17180943e+00 -1.07339084e+00 -7.86009192e-01 -2.91599721e-01
6.23099685e-01 -9.85846400e-01 1.28528821e+00 5.16194046e-01
-9.86959338e-01 -6.50410235e-01 -6.37061954e-01 4.10744511e-02
-4.47684854e-01 1.19992241e-01 -8.89427885e-02 5.63785434e-01
-1.19106591e+00 7.72246003e-01 -3.67121786e-01 -4.24292982e-01
-2.11838588e-01 2.70855069e-01 6.12908080e-02 -7.47485086e-02
-1.37998521e+00 9.63805914e-01 2.50757903e-01 -2.98057888e-02
-4.03946102e-01 -6.23856246e-01 -6.80810332e-01 3.35670784e-02
4.98618335e-01 -6.77797258e-01 1.61111999e+00 -1.21320450e+00
-1.60772157e+00 3.68188530e-01 -5.41683674e-01 -3.84173125e-01
4.78341699e-01 -2.87829369e-01 -1.51816294e-01 1.80390552e-01
2.46880904e-01 7.45090365e-01 7.73429990e-01 -1.17489016e+00
-6.30135179e-01 -3.80701460e-02 3.53727311e-01 1.92840353e-01
-2.60037750e-01 2.21968263e-01 -2.50461429e-01 -4.75574493e-01
-8.18648860e-02 -8.25480759e-01 -4.33827013e-01 -3.92822087e-01
-2.05821872e-01 -6.95987344e-01 5.50292611e-01 -9.27437305e-01
1.29230750e+00 -1.79761302e+00 3.75496000e-02 2.37263963e-01
4.77505215e-02 5.44757545e-01 -3.32391441e-01 7.10007787e-01
4.71352935e-02 2.17162371e-01 -3.59162837e-01 -3.57897252e-01
-8.23084414e-02 5.38367480e-02 -2.10169449e-01 -6.70876324e-01
5.08594334e-01 7.42121220e-01 -1.29297745e+00 -3.11842650e-01
-3.41211222e-02 -1.98603608e-02 -7.15048850e-01 5.04728377e-01
-7.69079328e-01 2.61406183e-01 -2.98685014e-01 -1.09381676e-01
3.72913003e-01 -2.31946751e-01 3.77882607e-02 3.12654644e-01
-8.68344977e-02 6.83066308e-01 -1.25338686e+00 1.13682795e+00
-6.12394989e-01 4.95628506e-01 -3.72818589e-01 -7.29377806e-01
1.15641689e+00 5.53606212e-01 -1.01124115e-01 -5.39697826e-01
-1.90876976e-01 3.90852720e-01 9.57045406e-02 -7.47914910e-01
7.20985770e-01 -9.98929366e-02 7.03074709e-02 7.71016419e-01
6.11071289e-02 -2.25604862e-01 9.49221671e-01 5.49793243e-01
1.24285638e+00 -3.20306644e-02 4.43401009e-01 1.28659382e-01
9.54403281e-01 1.80028558e-01 2.40683138e-01 1.00174165e+00
3.20793808e-01 7.08732843e-01 5.85252941e-01 -2.36697465e-01
-8.69836450e-01 -7.82348037e-01 4.19555485e-01 1.19134986e+00
-3.81642073e-01 -5.44300139e-01 -8.81094456e-01 -1.06860232e+00
-2.14567527e-01 1.11895394e+00 -6.73361421e-01 -1.01536430e-01
-6.30400300e-01 -2.47824073e-01 -2.47879438e-02 1.87313378e-01
5.10166362e-02 -1.76919019e+00 -4.26526904e-01 6.53471649e-01
-4.84434038e-01 -6.26523554e-01 -4.80487347e-01 -3.50839645e-02
-8.35762739e-01 -9.89147305e-01 -5.98595023e-01 -7.11983204e-01
9.41150188e-01 1.27584726e-01 1.26946223e+00 3.91721934e-01
1.43686965e-01 2.43260309e-01 -9.31670308e-01 -4.73731160e-01
-1.07831120e+00 2.70154089e-01 -3.26584518e-01 2.39284351e-01
3.31248343e-01 -3.59248340e-01 -2.13769421e-01 1.10454120e-01
-1.03545499e+00 8.65858719e-02 7.85577893e-01 8.52699935e-01
3.14791232e-01 -7.60776699e-02 9.71571207e-01 -1.27846456e+00
1.10117197e+00 -3.61708611e-01 -5.10502636e-01 3.87189180e-01
-5.72625458e-01 3.48295182e-01 9.38344896e-01 -1.86337754e-01
-1.14102292e+00 -4.82198112e-02 -3.45604271e-01 6.97748512e-02
-4.31118995e-01 7.27119565e-01 -1.08902551e-01 1.70707822e-01
1.04158700e+00 2.49034569e-01 -1.50389709e-02 -4.39833194e-01
3.78652543e-01 7.55896389e-01 1.09571308e-01 -1.55241713e-01
9.63396430e-01 -3.33096117e-01 -5.53604126e-01 -7.59435534e-01
-1.04744315e+00 -4.56964731e-01 -4.70605463e-01 -4.25727367e-01
6.29220188e-01 -1.44867018e-01 -1.19345058e-02 1.28303394e-01
-1.42064929e+00 -3.52095217e-01 -5.56319475e-01 1.62841797e-01
-1.43004045e-01 2.71136343e-01 -2.35105425e-01 -8.59958708e-01
-4.25921112e-01 -7.82697499e-01 6.00353360e-01 3.88959259e-01
-6.09370649e-01 -9.26434755e-01 1.50625721e-01 5.34618378e-01
5.01902342e-01 -2.80664712e-02 1.03558993e+00 -8.94323707e-01
-4.99884218e-01 -2.49059767e-01 1.60076275e-01 5.65742016e-01
4.56476837e-01 -1.19903587e-01 -7.83447504e-01 -4.18678410e-02
1.27233014e-01 -2.78177053e-01 6.70848310e-01 -2.15321645e-01
8.57598960e-01 -6.63103402e-01 1.54820448e-02 -3.63919407e-01
1.01643264e+00 1.40277192e-01 5.72323561e-01 2.51683056e-01
3.04180682e-01 9.37806189e-01 7.20340312e-01 1.04216054e-01
4.00928855e-01 5.25427699e-01 1.90464437e-01 1.98508427e-01
-8.03175122e-02 -2.77829945e-01 5.32466233e-01 9.33976948e-01
2.17325658e-01 -3.62286627e-01 -7.62743652e-01 7.33748972e-01
-1.62804830e+00 -9.66778100e-01 -3.60389560e-01 2.29995823e+00
1.19541276e+00 1.84333950e-01 9.20089781e-02 5.07514775e-01
6.22625887e-01 -1.51502863e-01 -2.62746125e-01 -6.56319380e-01
1.26818344e-01 5.77374041e-01 -3.00525904e-01 7.60951161e-01
-5.42618036e-01 8.33085656e-01 5.37221479e+00 4.89061475e-01
-9.95993674e-01 -9.70225558e-02 4.71410811e-01 5.95394559e-02
-6.49834037e-01 6.44970089e-02 -7.81313062e-01 4.51982230e-01
1.13201761e+00 -5.83339036e-01 3.42447311e-01 5.57794452e-01
4.41862553e-01 -3.80441755e-01 -1.13862932e+00 3.51181895e-01
3.75837624e-01 -1.02752268e+00 3.60555917e-01 -3.89791965e-01
8.24975848e-01 -3.68388951e-01 -4.03728068e-01 3.99256498e-01
3.27417761e-01 -7.22731471e-01 4.55675930e-01 5.08424819e-01
2.10186183e-01 -7.67991543e-01 1.00863075e+00 8.24319482e-01
-7.28308678e-01 -1.00884251e-01 -3.06933492e-01 -2.91214287e-01
2.74321169e-01 7.46886492e-01 -1.46521914e+00 5.89266479e-01
4.20170993e-01 1.29916921e-01 -9.20834124e-01 1.18116975e+00
-8.58377635e-01 8.78747582e-01 -2.40134120e-01 -6.00936770e-01
1.15343466e-01 3.16820480e-02 5.24701536e-01 1.14952850e+00
4.99048144e-01 2.37906665e-01 1.05411112e-01 8.92488897e-01
-7.63482600e-02 5.63888490e-01 -3.06998312e-01 1.15477994e-01
3.42213601e-01 1.43579495e+00 -4.99669701e-01 -5.89917004e-01
-2.15096474e-01 9.62996006e-01 3.50822210e-01 1.92559093e-01
-2.66670346e-01 -8.02304149e-01 3.93285695e-03 3.09225172e-01
3.69765610e-01 1.46300599e-01 -2.55230758e-02 -9.27280605e-01
2.69802541e-01 -1.09761059e+00 3.56443614e-01 -8.69527936e-01
-1.17700136e+00 7.36644804e-01 -2.99681216e-01 -1.02773607e+00
-8.61501396e-01 -3.84673476e-01 -7.66164839e-01 1.04404294e+00
-1.39797246e+00 -3.32933813e-01 -3.82613003e-01 2.45287389e-01
6.73370779e-01 2.79660612e-01 8.81436884e-01 1.73530802e-01
-1.25578968e-02 5.40154517e-01 -2.59724647e-01 1.55884206e-01
7.48127759e-01 -1.70193720e+00 4.94519562e-01 1.02016532e+00
5.19774914e-01 5.60222447e-01 9.01068032e-01 -5.35334527e-01
-7.94305444e-01 -1.23027647e+00 1.94655108e+00 -5.89771748e-01
4.28636223e-01 -3.57991397e-01 -1.28337920e+00 2.26499036e-01
5.13507009e-01 -5.10653555e-01 6.69181585e-01 2.99143121e-02
3.98067534e-02 -5.00016995e-02 -1.09959757e+00 5.38016319e-01
4.17679012e-01 -2.40695804e-01 -1.16495395e+00 3.73925716e-01
9.34524536e-01 -2.05570340e-01 -3.81882548e-01 -2.33529173e-02
-3.18055376e-02 -9.09554124e-01 2.64515549e-01 -4.53781396e-01
4.58820045e-01 -4.50983137e-01 5.40977836e-01 -1.68663156e+00
-3.00932497e-01 -7.83775270e-01 1.22933701e-01 1.29912305e+00
1.02483046e+00 -5.17738938e-01 7.12952733e-01 5.68778276e-01
-2.60661487e-02 -6.25744998e-01 -6.00611866e-01 -6.12961531e-01
-1.30529702e-01 -2.71155536e-01 5.89417398e-01 6.00091517e-01
1.50104374e-01 7.69678235e-01 -1.74182907e-01 -3.47427838e-02
2.32260555e-01 5.63796051e-02 8.92868340e-01 -1.19683778e+00
-3.57069761e-01 -2.30199844e-01 7.92972296e-02 -1.05277681e+00
-3.55554670e-01 -7.61296570e-01 5.42502105e-01 -1.85791969e+00
-1.58516541e-01 -2.84439623e-01 -9.66525450e-02 5.13935745e-01
-8.70253026e-01 5.63666783e-02 4.77611631e-01 -1.36510730e-01
-5.11699915e-01 3.70592207e-01 1.11237788e+00 1.14066370e-01
-5.21215022e-01 3.04215193e-01 -7.96030939e-01 5.88721335e-01
1.01595223e+00 -5.85665762e-01 -3.11844498e-01 -3.06584001e-01
4.11161870e-01 1.23133119e-02 1.10686824e-01 -1.08232546e+00
3.92514735e-01 1.40334782e-03 1.87543839e-01 -6.16609752e-01
-4.42882180e-02 -4.70238239e-01 -2.11576596e-01 4.61822242e-01
-6.04425311e-01 2.10478947e-01 4.05639224e-02 3.37090373e-01
-3.50897580e-01 -8.97593915e-01 6.15856230e-01 -2.73946732e-01
-3.65865082e-01 2.19042953e-02 -7.30116010e-01 1.83544859e-01
5.59791565e-01 -1.11350633e-01 5.84400147e-02 -7.04634249e-01
-8.71476889e-01 3.65726024e-01 1.65886894e-01 4.66627866e-01
5.78112245e-01 -1.09525013e+00 -9.81898010e-01 1.06236108e-01
2.41293475e-01 4.58757430e-02 -3.69532816e-02 4.72919524e-01
-2.68153369e-01 3.07538539e-01 3.11260998e-01 -2.98929930e-01
-1.38774765e+00 3.40767920e-01 1.46776274e-01 -7.99236536e-01
-1.83441862e-01 7.12996542e-01 -1.83050573e-01 -5.75507760e-01
-3.59657891e-02 -3.61674547e-01 -8.80153239e-01 3.98128361e-01
8.84634495e-01 5.53271472e-02 4.22666550e-01 -1.07919693e-01
5.21223918e-02 2.82502502e-01 -3.43424976e-01 -4.67685640e-01
1.21907353e+00 5.17516807e-02 -2.24708006e-01 5.03007650e-01
6.48296893e-01 2.34814331e-01 -7.00796425e-01 -3.39012802e-01
4.88214046e-01 -4.43267792e-01 -6.33293271e-01 -1.12278748e+00
-5.60010135e-01 8.59684825e-01 1.04080759e-01 7.18595564e-01
1.05316401e+00 1.45101957e-02 7.09786355e-01 6.38540387e-01
2.24525213e-01 -1.05630183e+00 4.23197240e-01 8.78779650e-01
1.10932660e+00 -9.12914097e-01 -5.05727530e-01 -3.69520515e-01
-4.77191001e-01 1.19702852e+00 7.30285764e-01 4.20117453e-02
1.86695516e-01 -2.21137702e-01 3.23209792e-01 9.06012431e-02
-1.09980702e+00 -2.33537063e-01 3.15267593e-01 4.77379948e-01
5.82147479e-01 -1.81591526e-01 -7.25429237e-01 4.00620848e-01
-5.82442522e-01 1.30397007e-01 7.34582067e-01 9.05160964e-01
-9.25273955e-01 -1.75909960e+00 -3.90106738e-01 6.66961789e-01
-2.90939987e-01 -2.72664011e-01 -7.95962453e-01 3.23432714e-01
1.23905785e-01 1.42419612e+00 -2.35383153e-01 -1.77501917e-01
6.21964335e-01 3.00041020e-01 1.38267845e-01 -1.24824989e+00
-9.86325800e-01 -4.16035593e-01 3.00797552e-01 -2.09541678e-01
-1.33047059e-01 -7.83943236e-01 -1.15014315e+00 2.07800850e-01
-4.53399777e-01 8.66407871e-01 4.62286383e-01 1.06786001e+00
3.85947466e-01 4.36916500e-01 9.26767230e-01 -4.10248518e-01
-6.70602381e-01 -1.27406144e+00 7.48264268e-02 6.02649331e-01
1.69466823e-01 -2.03609779e-01 -7.65320837e-01 1.85261860e-01]
|
[11.604025840759277, 8.343402862548828]
|
b5e86f47-155b-4a9b-b7a5-b0aaf1a44b10
|
hierarchical-convolutional-deconvolutional
|
1710.04540
| null |
http://arxiv.org/abs/1710.04540v1
|
http://arxiv.org/pdf/1710.04540v1.pdf
|
Hierarchical Convolutional-Deconvolutional Neural Networks for Automatic Liver and Tumor Segmentation
|
Automatic segmentation of liver and its tumors is an essential step for
extracting quantitative imaging biomarkers for accurate tumor detection,
diagnosis, prognosis and assessment of tumor response to treatment. MICCAI 2017
Liver Tumor Segmentation Challenge (LiTS) provides a common platform for
comparing different automatic algorithms on contrast-enhanced abdominal CT
images in tasks including 1) liver segmentation, 2) liver tumor segmentation,
and 3) tumor burden estimation. We participate this challenge by developing a
hierarchical framework based on deep fully convolutional-deconvolutional neural
networks (CDNN). A simple CDNN model is firstly trained to provide a quick but
coarse segmentation of the liver on the entire CT volume, then another CDNN is
applied to the liver region for fine liver segmentation. At last, the segmented
liver region, which is enhanced by histogram equalization, is employed as an
additional input to the third CDNN for tumor segmentation. Jaccard distance is
used as loss function when training CDNN models to eliminate the need of sample
re-weighting. Our framework is trained using the 130 challenge training cases
provided by LiTS. The evaluation on the 70 challenge testing cases resulted in
a mean Dice Similarity Coefficient (DSC) of 0.963 for liver segmentation, a
mean DSC of 0.657 for tumor segmentation, and a root mean square error (RMSE)
of 0.017 for tumor burden estimation, which ranked our method in the first,
fifth and third place, respectively
|
['Yading Yuan']
|
2017-10-12
| null | null | null | null |
['liver-segmentation', 'automatic-liver-and-tumor-segmentation']
|
['medical', 'medical']
|
[-2.11284488e-01 -1.36136532e-01 -2.14640573e-01 -2.60859281e-01
-7.15388775e-01 -3.82392764e-01 5.11835575e-01 4.26733196e-01
-5.35196781e-01 4.37446803e-01 2.93505579e-01 -4.90072370e-01
1.63836867e-01 -5.41590691e-01 -2.16365665e-01 -1.13915801e+00
-2.54760921e-01 5.71964145e-01 1.33684412e-01 4.41323847e-01
-9.56115425e-02 7.33030856e-01 -4.93892431e-01 4.47333194e-02
1.08792996e+00 1.29457271e+00 2.08550647e-01 5.83200276e-01
-2.67582051e-02 7.04323649e-01 -2.47127756e-01 -5.52574694e-02
2.20307589e-01 -5.36973715e-01 -9.20925677e-01 2.27139860e-01
-1.21066459e-02 -3.90846133e-01 -2.95291573e-01 1.15105581e+00
6.34380698e-01 -1.36403278e-01 8.44346464e-01 -8.11475277e-01
-2.21177101e-01 8.30343962e-01 -5.39639115e-01 3.62550050e-01
-3.74350727e-01 5.05398393e-01 3.90999854e-01 -9.70542550e-01
4.11849648e-01 4.20279592e-01 6.96557343e-01 4.01232243e-01
-9.80416775e-01 -4.85966593e-01 -5.24328172e-01 -7.89337307e-02
-1.37179148e+00 -1.06484443e-01 2.33246908e-01 -7.32645571e-01
3.64809692e-01 3.56615245e-01 9.36257303e-01 3.92478794e-01
4.10526723e-01 7.90567219e-01 1.02517664e+00 -1.08747996e-01
1.50420710e-01 -9.45254192e-02 7.54092410e-02 9.79998946e-01
2.63656259e-01 1.78977907e-01 2.66225159e-01 -3.62483133e-03
7.31600106e-01 8.11844021e-02 -5.09896934e-01 -4.45773035e-01
-1.63975585e+00 9.39003527e-01 1.06468725e+00 5.15333056e-01
-7.58837402e-01 3.19979861e-02 6.74405396e-01 -9.78693664e-02
4.11017239e-01 1.69688299e-01 -3.64758112e-02 4.57775265e-01
-1.17334211e+00 -2.42067188e-01 7.76745141e-01 4.12121713e-01
1.18673928e-01 -3.34743261e-02 -7.45440245e-01 5.76924741e-01
3.77132475e-01 2.86814809e-01 9.69603240e-01 -4.30612594e-01
-1.38003081e-01 6.06702328e-01 -3.18109654e-02 -3.90019596e-01
-7.62481630e-01 -9.82445478e-01 -1.46851897e+00 6.40485510e-02
6.90185845e-01 -1.85393989e-01 -1.16624999e+00 1.34131277e+00
5.12518346e-01 1.77915409e-01 -1.00587197e-01 1.25882399e+00
1.17252922e+00 2.00584546e-01 4.18113083e-01 -2.21505746e-01
1.61463976e+00 -1.27767861e+00 -2.96735376e-01 3.90945017e-01
9.92881238e-01 -8.40329289e-01 6.43707991e-01 2.00450793e-03
-1.04129040e+00 -8.63851383e-02 -7.33201325e-01 2.74470776e-01
-1.81499198e-01 5.08487999e-01 5.48900306e-01 5.88478088e-01
-1.20527256e+00 4.92881745e-01 -1.11850512e+00 -3.43420684e-01
6.79829717e-01 4.95010972e-01 -2.40633383e-01 -1.08698905e-01
-9.42555904e-01 1.00169241e+00 3.95951807e-01 -1.17129795e-02
-1.35685873e+00 -1.22013652e+00 -8.61763537e-01 1.74612686e-01
8.31069127e-02 -9.75561142e-01 9.82582450e-01 -6.55804694e-01
-1.51641989e+00 9.98445630e-01 1.39284536e-01 -8.66314054e-01
8.19081068e-01 4.60702449e-01 3.69746462e-02 2.97823012e-01
6.16052523e-02 8.25384796e-01 4.61039156e-01 -8.25528026e-01
-4.33347076e-01 -1.87563583e-01 -5.11959076e-01 2.95024812e-01
1.09618060e-01 -4.64152023e-02 -3.73749852e-01 -6.96696103e-01
1.89149991e-01 -9.65882421e-01 -3.44282776e-01 1.21268027e-01
-5.95124364e-01 -1.06080934e-01 5.26101232e-01 -1.14986920e+00
9.89747822e-01 -1.90871847e+00 2.80323233e-02 4.21935678e-01
5.97286582e-01 1.03826866e-01 8.70659053e-02 -4.86842304e-01
-2.84378648e-01 1.05008267e-01 -3.82230431e-01 -1.68451682e-01
-1.98629558e-01 -2.89369106e-01 4.37771559e-01 8.55537534e-01
-9.79993939e-02 1.17161274e+00 -9.22716498e-01 -5.69410384e-01
4.11696523e-01 4.85600531e-01 -1.93787947e-01 1.82146102e-01
2.18125239e-01 7.87391901e-01 -7.14031309e-02 7.97644079e-01
6.33424878e-01 -4.91725475e-01 -7.81211108e-02 -3.77801776e-01
-1.10478587e-01 -4.24901536e-03 -6.44775271e-01 1.58105016e+00
-2.69590199e-01 4.33078825e-01 1.76580817e-01 -6.97122633e-01
5.53648174e-01 5.06877005e-01 1.18745303e+00 -5.34411848e-01
5.33470392e-01 3.30741405e-01 3.80413175e-01 -3.29335779e-01
-2.49281585e-01 -2.89398730e-01 1.56343728e-01 5.83861053e-01
-1.52055413e-01 -1.58940941e-01 3.67204994e-01 1.73051894e-01
9.63333726e-01 -4.72006977e-01 2.82741517e-01 -6.95420861e-01
8.31628203e-01 2.17514202e-01 3.26279491e-01 2.66956300e-01
-6.35313213e-01 5.56380033e-01 3.79952997e-01 -7.27333069e-01
-8.23162496e-01 -1.01221621e+00 -3.81698251e-01 6.12990439e-01
3.47304232e-02 1.32999644e-01 -9.79654253e-01 -9.85112011e-01
-1.53692961e-01 4.17263538e-01 -7.47276783e-01 5.58609515e-02
-3.69959116e-01 -1.20503306e+00 4.71449375e-01 3.93406034e-01
7.36216247e-01 -8.68608415e-01 -7.35437691e-01 1.79802060e-01
-4.45616364e-01 -8.37629795e-01 -7.95933187e-01 2.09832892e-01
-8.99573803e-01 -1.39149582e+00 -1.40042150e+00 -7.83889294e-01
1.04660141e+00 6.02424368e-02 1.13140595e+00 4.21472222e-01
-5.69038391e-01 6.66737929e-02 8.67707729e-02 -2.11679682e-01
-5.65856099e-01 8.01577419e-02 -1.09135896e-01 -1.18225031e-01
1.06512912e-01 -6.13078289e-02 -1.07532084e+00 3.43495429e-01
-7.33993530e-01 1.87275827e-01 8.82255077e-01 1.10244894e+00
7.02591300e-01 -1.12818226e-01 1.19679548e-01 -6.02171242e-01
4.59852308e-01 -4.51134980e-01 -8.56728196e-01 2.77641863e-01
-5.42972922e-01 -2.37759814e-01 4.43564922e-01 -3.74269962e-01
-6.85136557e-01 2.88706869e-01 -7.23719597e-02 -2.79488057e-01
-5.79101332e-02 7.38761425e-01 5.08514285e-01 -3.71811271e-01
5.59218228e-01 4.34967786e-01 3.49314958e-01 -1.91342533e-02
-1.90121215e-02 1.60020754e-01 4.29978460e-01 -1.18546382e-01
4.99210685e-01 2.49294534e-01 3.69628489e-01 -3.96200746e-01
-5.82722425e-01 -5.61153948e-01 -7.41575420e-01 -1.18602619e-01
1.06305969e+00 -8.43758047e-01 -5.95924854e-01 6.11924946e-01
-6.53254628e-01 -7.39241123e-01 -3.54702175e-01 8.52354050e-01
-2.33639657e-01 3.79421055e-01 -1.10595977e+00 -6.81560412e-02
-9.89625514e-01 -1.69937611e+00 6.25265539e-01 2.23741427e-01
-4.46267091e-02 -1.26880801e+00 -1.76307797e-01 1.94444314e-01
9.66378570e-01 4.73943532e-01 1.02745843e+00 -8.82927001e-01
-4.57068205e-01 -3.30104500e-01 -5.82737625e-01 2.91628927e-01
1.01729535e-01 -1.46938950e-01 -5.10359228e-01 -4.53670055e-01
-8.26863796e-02 -9.38629806e-02 8.31617177e-01 1.03213835e+00
1.23895884e+00 -1.42499730e-01 -3.22932810e-01 9.34801877e-01
1.43242502e+00 9.62594450e-02 3.81380856e-01 1.11989370e-02
5.67249238e-01 1.23654701e-01 2.18725637e-01 2.68846691e-01
3.03235233e-01 2.78276920e-01 6.40852511e-01 -5.53823352e-01
-5.78231096e-01 3.55963141e-01 -7.39495084e-02 7.45842040e-01
1.32830366e-01 1.08623624e-01 -1.25142837e+00 6.23326004e-01
-1.16301751e+00 -4.67638284e-01 -4.00672823e-01 2.22314811e+00
7.03929722e-01 -2.72664994e-01 1.60134032e-01 -1.82611510e-01
6.20494783e-01 -3.34210575e-01 -3.65716904e-01 1.33760303e-01
2.98776001e-01 1.10495821e-01 6.21361792e-01 4.95225072e-01
-1.42992520e+00 4.98537749e-01 5.80555391e+00 7.91362703e-01
-1.60317385e+00 1.72746405e-01 1.13421488e+00 2.45165572e-01
2.77927220e-01 -2.46067241e-01 -2.96102971e-01 6.06595874e-01
6.39594018e-01 -2.26013929e-01 1.25801489e-01 6.59917355e-01
2.49981657e-01 -4.49615121e-01 -9.20482874e-01 8.39258790e-01
-1.87682107e-01 -1.41357946e+00 -1.24432154e-01 2.38314912e-01
8.59473884e-01 4.59680110e-01 2.98366006e-02 1.79943264e-01
2.54871309e-01 -1.21791780e+00 9.77619812e-02 4.82773930e-01
9.91117358e-01 -5.32800257e-01 1.31552589e+00 3.03076595e-01
-1.06597531e+00 3.19764644e-01 -1.86134636e-01 6.73542082e-01
-1.63944378e-01 8.62963498e-01 -1.44350982e+00 5.22398174e-01
2.99451232e-01 4.40552384e-01 -5.16569436e-01 1.62148273e+00
7.73703353e-03 5.73002517e-01 -3.61562759e-01 2.38273084e-01
3.37702960e-01 -1.97461829e-01 3.97713304e-01 1.34336913e+00
4.45379853e-01 8.01177397e-02 4.59458351e-01 7.82588720e-01
-2.94409305e-01 2.72159338e-01 1.19880199e-01 3.77173960e-01
1.55336305e-01 1.79702711e+00 -1.28716528e+00 -7.52326250e-01
-2.11747568e-02 8.58101428e-01 -3.24614137e-01 6.40451387e-02
-9.15192723e-01 -5.79602458e-02 -6.84285983e-02 5.58771342e-02
-4.83437777e-02 2.07590297e-01 -4.37532425e-01 -1.13143909e+00
-6.76126659e-01 -5.37404478e-01 6.01499975e-01 -3.92520159e-01
-1.16065180e+00 6.80447221e-01 -2.84920603e-01 -1.15794158e+00
-1.08144075e-01 -4.59124207e-01 -9.15359735e-01 1.18192244e+00
-1.60109341e+00 -9.35044467e-01 -9.00221765e-01 5.01352608e-01
3.41652811e-01 1.56493448e-02 7.43279517e-01 3.85421664e-01
-5.73220372e-01 4.40388680e-01 -1.51482582e-01 6.08834386e-01
5.52766204e-01 -1.51618421e+00 -3.06952655e-01 7.47640550e-01
-4.69866812e-01 1.99966237e-01 3.02762330e-01 -5.41932106e-01
-9.81115997e-01 -1.39029241e+00 7.76757538e-01 -6.83656931e-02
5.36810577e-01 2.99582273e-01 -8.21647465e-01 4.52694088e-01
1.58961430e-01 6.63295984e-01 8.67827535e-01 -6.15020990e-01
1.75402224e-01 5.92746399e-02 -1.59581816e+00 3.74242306e-01
1.43352017e-01 3.86365913e-02 -1.98627170e-02 5.77117264e-01
2.51134843e-01 -8.61490726e-01 -1.45993567e+00 5.72010994e-01
4.45001215e-01 -7.47305810e-01 9.57826734e-01 -1.65167868e-01
4.01237071e-01 -3.28058839e-01 2.77494609e-01 -1.43729341e+00
-4.93431479e-01 -1.40439823e-01 1.70472354e-01 4.36994523e-01
3.17787290e-01 -3.52717042e-01 8.74440610e-01 5.83964527e-01
-4.18244720e-01 -1.14254534e+00 -6.63069963e-01 -3.07001501e-01
3.59515578e-01 4.18552160e-02 4.44477499e-01 1.10750103e+00
-2.63878465e-01 -1.79211244e-01 3.40951204e-01 5.90776987e-02
9.01753366e-01 -4.69571315e-02 2.87925750e-01 -1.06498039e+00
2.58747339e-01 -1.11954153e+00 -2.16621488e-01 -7.03819096e-01
-2.14676574e-01 -1.32640624e+00 -1.30122796e-01 -1.69319475e+00
6.34917974e-01 -5.31233013e-01 -4.45001423e-01 4.24483895e-01
-1.91430256e-01 4.90427703e-01 9.04326588e-02 3.64661247e-01
-3.40259194e-01 2.12900966e-01 1.58436823e+00 -3.73130143e-01
-1.51970992e-02 2.78085411e-01 -3.94957691e-01 7.49908209e-01
7.61710882e-01 -3.22182089e-01 1.52864987e-02 -1.20832078e-01
-6.49258196e-01 3.55663091e-01 4.82438356e-01 -1.10133135e+00
4.04298902e-01 1.76222473e-02 9.57978010e-01 -7.06625879e-01
-3.03054184e-01 -8.81452203e-01 1.15629084e-01 1.48519826e+00
-2.62631983e-01 -1.48227364e-01 2.74642445e-02 -8.74637589e-02
-1.79555550e-01 -2.05568865e-01 1.24251854e+00 -4.00124192e-01
-3.23084831e-01 7.83545077e-01 -4.28461313e-01 -1.22170322e-01
1.24229848e+00 9.12161730e-03 2.82055582e-03 -6.26910999e-02
-1.12998474e+00 4.16285127e-01 1.21307157e-01 -2.43923128e-01
5.60342491e-01 -1.27990115e+00 -1.04603088e+00 1.39070019e-01
-1.31715164e-01 2.59263605e-01 1.96920246e-01 1.89928627e+00
-9.88672316e-01 5.01338184e-01 2.49628499e-02 -9.13780928e-01
-1.25555766e+00 3.52559656e-01 9.70242381e-01 -9.73823369e-01
-6.36067510e-01 8.80864143e-01 2.95392126e-01 -2.23054439e-01
2.40858659e-01 -6.43788457e-01 -5.77265769e-02 -9.80135724e-02
3.96894187e-01 2.54948586e-01 2.71437764e-01 -6.61323488e-01
-3.83462459e-01 3.37401241e-01 4.14852314e-02 2.74731010e-01
1.11687374e+00 6.87169377e-03 -4.04156417e-01 -1.82070002e-01
1.07223213e+00 -3.05423588e-01 -1.07309496e+00 -3.71749043e-01
9.22566876e-02 -6.05508685e-02 4.79915828e-01 -1.29443264e+00
-1.52613091e+00 8.95813823e-01 9.44760203e-01 6.42036498e-02
1.21875739e+00 -2.09360525e-01 8.21515441e-01 -2.26395130e-01
-3.03435270e-02 -3.61793131e-01 -2.77998038e-02 5.65569222e-01
8.48655283e-01 -1.49181771e+00 6.86520711e-02 -3.55007857e-01
-7.31671512e-01 1.35872352e+00 3.13999861e-01 -1.03693292e-01
7.31433809e-01 3.34385723e-01 1.67137206e-01 -2.36615270e-01
-3.27058971e-01 -1.10847354e-01 5.32588422e-01 2.27110088e-01
5.77381790e-01 3.75729561e-01 -2.87610322e-01 4.29062486e-01
-7.71572767e-03 1.41469851e-01 2.29125425e-01 3.78552467e-01
-4.39296275e-01 -5.16122878e-01 -2.34967440e-01 6.39334619e-01
-7.28281379e-01 -2.56566465e-01 4.59000207e-02 8.63425255e-01
1.69000495e-02 3.22484046e-01 -2.03376580e-02 1.30602837e-01
-1.30062148e-01 5.30017056e-02 3.02583009e-01 -4.14204031e-01
-1.00947809e+00 3.22777450e-01 -4.74476248e-01 -3.48866731e-01
-7.76128173e-02 -3.56619537e-01 -1.40628886e+00 -2.53380984e-01
-2.89281279e-01 2.76453733e-01 9.41823065e-01 8.02503943e-01
-4.38303128e-02 6.83834076e-01 6.86712027e-01 -7.50726342e-01
-6.32625580e-01 -1.06004584e+00 -4.16945696e-01 4.16115314e-01
3.06649357e-01 -3.41619074e-01 -3.93240958e-01 -3.09171379e-02]
|
[14.480391502380371, -2.7060933113098145]
|
9ed5e42c-0998-4dcd-b735-b9e9395611bb
|
learning-robust-visual-semantic-embedding-for
|
2304.09498
| null |
https://arxiv.org/abs/2304.09498v1
|
https://arxiv.org/pdf/2304.09498v1.pdf
|
Learning Robust Visual-Semantic Embedding for Generalizable Person Re-identification
|
Generalizable person re-identification (Re-ID) is a very hot research topic in machine learning and computer vision, which plays a significant role in realistic scenarios due to its various applications in public security and video surveillance. However, previous methods mainly focus on the visual representation learning, while neglect to explore the potential of semantic features during training, which easily leads to poor generalization capability when adapted to the new domain. In this paper, we propose a Multi-Modal Equivalent Transformer called MMET for more robust visual-semantic embedding learning on visual, textual and visual-textual tasks respectively. To further enhance the robust feature learning in the context of transformer, a dynamic masking mechanism called Masked Multimodal Modeling strategy (MMM) is introduced to mask both the image patches and the text tokens, which can jointly works on multimodal or unimodal data and significantly boost the performance of generalizable person Re-ID. Extensive experiments on benchmark datasets demonstrate the competitive performance of our method over previous approaches. We hope this method could advance the research towards visual-semantic representation learning. Our source code is also publicly available at https://github.com/JeremyXSC/MMET.
|
['Yuzhuo Fu', 'Dahong Qian', 'Ting Liu', 'Chengfeng Zhou', 'Jiacheng Ruan', 'Mengyuan Guan', 'Jingsheng Gao', 'Suncheng Xiang']
|
2023-04-19
| null | null | null | null |
['person-re-identification', 'generalizable-person-re-identification']
|
['computer-vision', 'computer-vision']
|
[ 8.05654377e-02 -2.47245476e-01 -2.26874679e-01 -2.35665455e-01
-5.23141503e-01 -3.72390836e-01 7.73258924e-01 -5.26393540e-02
-3.04868281e-01 3.67091477e-01 5.28926790e-01 5.19921258e-02
1.09951749e-01 -5.67932248e-01 -3.95971119e-01 -7.55136847e-01
4.63490933e-01 6.58149868e-02 7.31537268e-02 -1.63902864e-01
6.98136687e-02 1.55963719e-01 -1.48434675e+00 2.92829871e-01
6.74015522e-01 8.46001685e-01 2.20742211e-01 1.38202906e-01
-2.93903705e-03 3.99033844e-01 -3.20657849e-01 -7.81024516e-01
6.96119368e-02 -4.03825194e-01 -5.35973370e-01 1.87735125e-01
3.01842660e-01 -1.18501231e-01 -7.39569306e-01 1.33165503e+00
7.28912592e-01 2.59185016e-01 6.78982377e-01 -1.45759499e+00
-1.12128496e+00 2.93019056e-01 -8.46295536e-01 2.73530960e-01
4.64147389e-01 1.68121710e-01 7.47508764e-01 -1.02000201e+00
1.63250536e-01 1.49262643e+00 6.27394021e-01 8.60443950e-01
-9.27965999e-01 -8.71185839e-01 3.01941097e-01 7.14149594e-01
-1.51299548e+00 -4.50641155e-01 8.64739120e-01 -3.97468269e-01
4.40728813e-01 3.52029532e-01 4.68281478e-01 1.27370572e+00
-2.66630441e-01 9.64439571e-01 1.00026631e+00 -3.05368423e-01
-2.32050389e-01 4.94184226e-01 5.41359037e-02 5.72176397e-01
1.65339231e-01 2.03186758e-02 -3.53580683e-01 -5.18055074e-02
6.97847784e-01 5.94852865e-01 -4.89244610e-01 -2.23785266e-01
-1.25869000e+00 7.74041295e-01 6.53930664e-01 3.21295381e-01
-7.81371370e-02 -3.35384570e-02 6.01950586e-01 1.28748521e-01
4.27500933e-01 -4.93314788e-02 3.75586711e-02 1.90395892e-01
-5.17319500e-01 1.55913219e-01 6.29193336e-02 6.92112923e-01
6.80516303e-01 -9.39349458e-02 -4.10614848e-01 1.23268223e+00
3.64382267e-01 6.59994543e-01 9.02179897e-01 -5.81246614e-01
5.93856752e-01 8.92829239e-01 -3.09476648e-02 -1.21771264e+00
-2.35403478e-01 -2.34667331e-01 -1.04273880e+00 -1.56017035e-01
1.48081452e-01 9.83824655e-02 -7.75564611e-01 1.64309716e+00
4.13641632e-01 4.65005606e-01 1.29763633e-01 1.11471879e+00
1.16944289e+00 6.80462003e-01 2.57352054e-01 5.46771102e-02
1.77429187e+00 -1.07049561e+00 -6.22647285e-01 -3.50361317e-01
3.84165198e-01 -6.76234484e-01 9.56776679e-01 -5.95375896e-02
-6.98617041e-01 -7.43292451e-01 -9.07442629e-01 -1.28868362e-02
-4.36358303e-01 3.01139861e-01 4.64457989e-01 7.27977216e-01
-7.72734880e-01 -9.11734346e-03 -5.15410066e-01 -6.53801620e-01
5.22989154e-01 2.31646270e-01 -7.02399254e-01 -4.13393557e-01
-1.22017813e+00 7.00665951e-01 4.64986473e-01 3.61807972e-01
-7.90930688e-01 -3.21326882e-01 -9.88203108e-01 1.52160004e-02
3.36141020e-01 -6.28113151e-01 8.37729335e-01 -1.05054772e+00
-1.09457242e+00 9.70007002e-01 -3.92155170e-01 -1.10609137e-01
3.57977629e-01 4.21969630e-02 -6.09920859e-01 1.19006783e-01
1.62716031e-01 6.19276106e-01 1.06888199e+00 -1.33186412e+00
-5.69990098e-01 -6.99517429e-01 -3.40234824e-02 3.13445389e-01
-9.11526203e-01 7.27030262e-02 -7.51287222e-01 -9.40965891e-01
-1.18231215e-01 -9.24380958e-01 -8.40994418e-02 -1.59888864e-01
-3.63406152e-01 -4.86702770e-01 6.44573212e-01 -9.11683917e-01
1.05204523e+00 -2.31455374e+00 3.51618707e-01 8.79041925e-02
2.25380614e-01 3.57678443e-01 -2.11824045e-01 4.96655256e-01
-1.81931660e-01 5.17543107e-02 -1.25509039e-01 -4.67248648e-01
-1.10295732e-02 -1.13271788e-01 -2.04566017e-01 4.97661948e-01
-8.75306651e-02 1.13510787e+00 -6.13734901e-01 -4.20025200e-01
3.32442969e-01 6.29915893e-01 -2.04715997e-01 1.81993529e-01
2.44391918e-01 5.40082157e-01 -5.00013888e-01 6.86720729e-01
8.64206433e-01 -2.72875994e-01 -3.66547368e-02 -5.10424197e-01
1.30069301e-01 -3.91436577e-01 -9.79065776e-01 1.59525597e+00
-2.71549553e-01 3.03830236e-01 -1.96731240e-01 -1.24108911e+00
8.65265608e-01 2.81865835e-01 2.89481193e-01 -9.25471306e-01
1.41467512e-01 -8.85683075e-02 -3.38473499e-01 -7.17336595e-01
3.55212182e-01 -1.00372910e-01 -5.61608654e-03 2.84977794e-01
-1.08863026e-01 7.18273699e-01 -1.59861177e-01 1.78090483e-01
4.17908877e-01 -1.78629041e-01 1.51909024e-01 3.01107634e-02
9.39518332e-01 -2.84688473e-01 6.04496181e-01 3.48470569e-01
-3.38976324e-01 5.49053311e-01 1.17145270e-01 -3.09450120e-01
-9.29822981e-01 -8.19561124e-01 -7.94443563e-02 1.20373559e+00
6.54703200e-01 -4.22303438e-01 -6.66293502e-01 -7.39255905e-01
1.18748005e-02 3.43424737e-01 -7.20846713e-01 -4.15416211e-01
-3.80999863e-01 -9.92156327e-01 5.49302101e-01 6.16480947e-01
8.08717191e-01 -9.41771686e-01 5.96853085e-02 -1.94431171e-01
-4.84438956e-01 -1.08551145e+00 -7.97902703e-01 -5.97717762e-01
-5.81253290e-01 -1.07448232e+00 -1.34765160e+00 -1.11582541e+00
7.15194404e-01 7.35378683e-01 3.66755605e-01 2.25616977e-01
-3.09657812e-01 7.97309697e-01 -6.11264884e-01 -9.77047756e-02
-6.30135462e-02 -1.18850894e-01 1.36437804e-01 6.04734123e-01
6.49592400e-01 -1.92506030e-01 -7.69907713e-01 4.19119000e-01
-9.07178164e-01 1.15686610e-01 4.54643399e-01 1.05027318e+00
2.89823443e-01 6.01345412e-02 5.74272633e-01 -5.48970044e-01
5.55039108e-01 -5.70362985e-01 -1.34252429e-01 4.73347634e-01
-4.26397651e-01 -1.79488927e-01 3.79721969e-01 -6.28160834e-01
-1.11255848e+00 -1.76217511e-01 -1.01315729e-01 -4.89838898e-01
-1.84007347e-01 3.94197166e-01 -5.17635703e-01 -1.43240243e-01
2.11138710e-01 5.73117197e-01 1.48104131e-02 -6.53723717e-01
2.47959360e-01 8.82686317e-01 4.37873423e-01 -3.89972150e-01
1.00340128e+00 5.54335058e-01 -2.61430025e-01 -6.81747258e-01
-4.15238768e-01 -5.40188193e-01 -3.19933087e-01 -1.67529181e-01
9.96847808e-01 -1.16853583e+00 -7.07690656e-01 5.85967481e-01
-1.07101440e+00 2.04713807e-01 3.06255043e-01 4.10048068e-01
-6.02181628e-02 8.12983096e-01 -3.98468077e-01 -7.11266994e-01
-3.26565564e-01 -1.17920196e+00 1.07747197e+00 5.79423726e-01
2.62683094e-01 -1.13013160e+00 -3.98110412e-02 7.32846797e-01
2.75877684e-01 3.23962942e-02 8.12240839e-01 -6.83200359e-01
-3.89479756e-01 -3.15561026e-01 -5.82919240e-01 4.36239570e-01
1.82070255e-01 -6.03763103e-01 -1.07899165e+00 -6.36421919e-01
-2.10963741e-01 -2.56418169e-01 1.10223413e+00 -2.76207412e-03
1.29508734e+00 -3.53762209e-01 -6.53137863e-01 6.11998379e-01
1.34448099e+00 -2.73640337e-03 6.45012617e-01 5.01584351e-01
1.05093539e+00 7.59907186e-01 5.02603114e-01 5.21059632e-01
7.23466098e-01 9.05944109e-01 3.47538233e-01 -2.68677354e-01
-2.10091278e-01 -4.12626833e-01 4.36252415e-01 5.19600809e-01
-2.46334702e-01 -2.03964919e-01 -7.45338082e-01 5.39264560e-01
-1.98896444e+00 -1.14228249e+00 1.38140216e-01 2.11072874e+00
4.10762936e-01 -4.51823413e-01 2.97805130e-01 9.88274887e-02
1.29231977e+00 2.21811578e-01 -4.99687552e-01 1.16360955e-01
-1.54374197e-01 -2.38115236e-01 3.35627347e-01 1.77306116e-01
-1.17217863e+00 9.11648691e-01 4.60229349e+00 1.14642131e+00
-9.80458081e-01 4.13837880e-01 4.96076912e-01 1.66154698e-01
-4.00679976e-01 -2.37219349e-01 -8.34326029e-01 8.03531945e-01
4.07100230e-01 -1.69378102e-01 4.57423210e-01 6.84837520e-01
1.07449807e-01 3.81842434e-01 -8.30210447e-01 1.64873672e+00
5.34857869e-01 -1.07441509e+00 4.45325106e-01 -2.82547586e-02
4.62158918e-01 -3.87178540e-01 4.49905723e-01 2.82247484e-01
-2.38306835e-01 -1.15869629e+00 5.24176419e-01 5.01972437e-01
9.29738283e-01 -7.89961278e-01 8.58737469e-01 1.50510281e-01
-1.51535642e+00 -3.09356064e-01 -5.58063447e-01 2.72572577e-01
4.84634228e-02 3.97274159e-02 -5.03405154e-01 7.11021364e-01
9.32606518e-01 1.10879588e+00 -1.02895701e+00 1.03289008e+00
-3.99760008e-02 3.98614109e-01 1.33262753e-01 5.87051995e-02
-1.29540071e-01 -7.53718168e-02 5.16402543e-01 1.23275292e+00
2.26642549e-01 -4.98735756e-02 2.16218233e-01 6.95170581e-01
2.67143324e-02 2.30396837e-01 -5.21474600e-01 4.68916111e-02
3.59198868e-01 1.19812751e+00 -4.46578920e-01 -1.95775256e-01
-6.59733653e-01 1.43124199e+00 1.85771689e-01 6.85613632e-01
-8.72380495e-01 -3.61373186e-01 5.31125367e-01 3.73025499e-02
2.18550444e-01 -9.15056001e-03 5.31749167e-02 -1.56095302e+00
1.72093868e-01 -9.65382278e-01 6.76000834e-01 -4.85690236e-01
-1.72427762e+00 5.40655851e-01 4.43903496e-03 -1.43602228e+00
1.79668024e-01 -6.41250134e-01 -6.13503218e-01 8.32297742e-01
-1.51069880e+00 -1.73435879e+00 -6.33545280e-01 1.06462908e+00
5.30559719e-01 -5.43070078e-01 5.51974118e-01 5.60768664e-01
-7.94499397e-01 1.08926201e+00 1.55179098e-01 3.38685125e-01
8.52301478e-01 -7.96503127e-01 9.51622874e-02 8.31609845e-01
-4.23310464e-03 6.42081082e-01 3.37709397e-01 -5.15061736e-01
-1.34766448e+00 -1.23371696e+00 5.65838099e-01 -4.05650914e-01
4.30265039e-01 -4.40521359e-01 -9.61168587e-01 7.50162780e-01
2.50102252e-01 -2.20751926e-01 7.94935644e-01 -2.56796200e-02
-5.08372843e-01 -2.49897167e-01 -1.01747048e+00 6.89012706e-01
1.04163516e+00 -6.51333332e-01 -5.76646149e-01 2.99035966e-01
6.00201070e-01 5.87900765e-02 -7.36140966e-01 5.15300870e-01
4.74909484e-01 -8.22235882e-01 1.26708031e+00 -4.75419134e-01
5.71868978e-02 -4.42572474e-01 -1.99030906e-01 -1.04983199e+00
-4.59482193e-01 -1.39870077e-01 5.77566326e-02 1.74368918e+00
-8.04202184e-02 -9.49315846e-01 5.55256128e-01 4.81617600e-01
1.75948873e-01 -4.58365411e-01 -9.58695292e-01 -7.20016718e-01
-2.02464405e-02 -1.38176218e-01 6.74751759e-01 1.01410019e+00
-7.54067823e-02 2.15137213e-01 -7.36972809e-01 2.92080283e-01
6.67629361e-01 4.48687822e-02 7.53539383e-01 -1.01392233e+00
-1.50509670e-01 -4.32806313e-01 -7.08184421e-01 -1.10710585e+00
3.39680642e-01 -1.27843440e+00 -3.86588156e-01 -1.44193411e+00
7.55694270e-01 -3.30521017e-01 -4.81393278e-01 6.21852517e-01
-5.08999705e-01 5.17210901e-01 2.60618418e-01 4.62957382e-01
-7.04756439e-01 1.00707698e+00 1.14740598e+00 -5.41154802e-01
8.01214576e-02 5.13828546e-03 -9.42304850e-01 4.92212653e-01
7.44091749e-01 -1.88647225e-01 -4.18418914e-01 -5.23731947e-01
-2.02645510e-01 -2.49249160e-01 9.09339488e-01 -8.33276927e-01
2.36741945e-01 -5.38236573e-02 6.88900948e-01 -2.26869673e-01
4.34030801e-01 -7.44591653e-01 1.31912008e-01 3.03375483e-01
-1.98252201e-01 1.76530942e-01 1.82349965e-01 8.00703466e-01
-3.73057485e-01 -2.56285489e-01 6.75531685e-01 -1.78864986e-01
-1.03826082e+00 6.30676150e-01 -4.40409370e-02 -1.83127239e-01
1.08659005e+00 -2.85934448e-01 -4.92097050e-01 -3.43420416e-01
-5.67374825e-01 3.45528722e-01 5.96654654e-01 8.76108408e-01
1.01713455e+00 -1.72259593e+00 -7.98101008e-01 2.12678388e-01
5.05742192e-01 -5.65111756e-01 7.53158867e-01 8.33192647e-01
-1.49214923e-01 3.36733460e-01 -2.27089807e-01 -5.16767502e-01
-1.40616810e+00 8.46734166e-01 3.27277005e-01 1.62097886e-01
-7.19311595e-01 7.08401680e-01 6.48809612e-01 -2.78500587e-01
3.51922780e-01 4.40010428e-01 -5.91090143e-01 -8.71597007e-02
8.59502614e-01 4.60767239e-01 -3.82360071e-01 -1.16369236e+00
-5.59701681e-01 8.93266499e-01 -2.41017967e-01 -5.05310297e-02
9.57928479e-01 -4.36800927e-01 -1.51944906e-01 2.65495598e-01
1.27356243e+00 -4.94074747e-02 -9.06218827e-01 -4.98189479e-01
-1.81049138e-01 -5.66292286e-01 -2.13223755e-01 -4.34940010e-01
-1.09484982e+00 1.01551318e+00 1.00735974e+00 -5.87543212e-02
1.08888340e+00 2.07833201e-01 1.00711465e+00 1.75705314e-01
1.59631625e-01 -8.02679777e-01 3.94828051e-01 3.43835689e-02
9.14228618e-01 -1.53487742e+00 -5.98394424e-02 -2.84417897e-01
-9.89667177e-01 9.49360192e-01 6.93996549e-01 5.25297150e-02
4.52949286e-01 -4.31660444e-01 5.20994887e-02 3.28942016e-02
-1.43420726e-01 -3.06157887e-01 4.91052896e-01 8.29159141e-01
1.46213114e-01 9.25205573e-02 -1.17191389e-01 9.55092430e-01
1.51771262e-01 -3.34951460e-01 2.11626682e-02 5.74900925e-01
-3.59662861e-01 -1.20249689e+00 -5.50079703e-01 2.28248864e-01
-2.84563243e-01 -9.08049121e-02 -2.37942338e-01 4.77547050e-01
3.71292830e-01 1.06445122e+00 -1.45278305e-01 -6.45153403e-01
1.91791549e-01 -7.67278746e-02 3.85163337e-01 -4.14771616e-01
-3.24727505e-01 -5.26704602e-02 -3.00366759e-01 -2.93822289e-01
-5.74667633e-01 -6.80728316e-01 -1.00757897e+00 -2.80830324e-01
-9.50027555e-02 1.55255765e-01 3.61482024e-01 8.12712193e-01
3.50902975e-01 1.76786765e-01 5.83418071e-01 -7.97854066e-01
-3.21552902e-01 -9.46551979e-01 -4.53497499e-01 7.61753261e-01
3.67869347e-01 -8.82738054e-01 -2.76161820e-01 9.40313861e-02]
|
[14.67236614227295, 0.9798128604888916]
|
652fcf17-03ab-4bd4-a03d-0e16deff3653
|
few-shot-action-recognition-with-prototype
|
2101.08085
| null |
https://arxiv.org/abs/2101.08085v4
|
https://arxiv.org/pdf/2101.08085v4.pdf
|
Few-shot Action Recognition with Prototype-centered Attentive Learning
|
Few-shot action recognition aims to recognize action classes with few training samples. Most existing methods adopt a meta-learning approach with episodic training. In each episode, the few samples in a meta-training task are split into support and query sets. The former is used to build a classifier, which is then evaluated on the latter using a query-centered loss for model updating. There are however two major limitations: lack of data efficiency due to the query-centered only loss design and inability to deal with the support set outlying samples and inter-class distribution overlapping problems. In this paper, we overcome both limitations by proposing a new Prototype-centered Attentive Learning (PAL) model composed of two novel components. First, a prototype-centered contrastive learning loss is introduced to complement the conventional query-centered learning objective, in order to make full use of the limited training samples in each episode. Second, PAL further integrates a hybrid attentive learning mechanism that can minimize the negative impacts of outliers and promote class separation. Extensive experiments on four standard few-shot action benchmarks show that our method clearly outperforms previous state-of-the-art methods, with the improvement particularly significant (10+\%) on the most challenging fine-grained action recognition benchmark.
|
['Juan-Manuel Perez-Rua', 'Tao Xiang', 'Brais Martinez', 'Li Zhang', 'Antoine Toisoul', 'Xiatian Zhu']
|
2021-01-20
| null | null | null | null |
['few-shot-action-recognition', 'fine-grained-action-recognition']
|
['computer-vision', 'computer-vision']
|
[ 4.61723149e-01 -1.08805902e-01 -7.33064353e-01 -3.68184060e-01
-1.09319687e+00 3.53976756e-01 5.04118919e-01 2.50374556e-01
-6.08500063e-01 8.76701176e-01 2.35486254e-01 4.82750952e-01
-3.65871757e-01 -5.85947216e-01 -5.35320818e-01 -8.76295328e-01
3.06589622e-02 4.11133945e-01 6.46417975e-01 -1.30058587e-01
3.73735696e-01 1.93492606e-01 -1.95811605e+00 5.99496186e-01
1.22456813e+00 1.34366488e+00 4.75851409e-02 3.39132607e-01
-1.38399109e-01 1.23700106e+00 -6.25376523e-01 -1.47437707e-01
1.89861968e-01 -7.04829633e-01 -4.41536903e-01 2.73814350e-01
3.78170729e-01 -4.31266129e-01 -1.55109823e-01 8.20200682e-01
6.20631754e-01 6.76683009e-01 5.10283470e-01 -1.51395249e+00
-3.42321843e-01 2.68620521e-01 -6.33129776e-01 4.12172705e-01
1.18457973e-01 3.33698094e-01 9.60428774e-01 -9.36131060e-01
3.46129090e-01 1.01413906e+00 6.32683873e-01 7.15125501e-01
-1.01400352e+00 -4.61742818e-01 5.28779209e-01 7.03322887e-01
-1.07903671e+00 -5.29312015e-01 8.33095908e-01 -2.62270600e-01
1.14149892e+00 3.14352065e-02 5.98967731e-01 1.13703310e+00
1.57625392e-01 1.14657950e+00 8.02121162e-01 -4.11903739e-01
7.68750846e-01 9.21283104e-03 3.42952579e-01 5.49070299e-01
-7.66862407e-02 8.54934081e-02 -6.00933552e-01 -9.24961045e-02
2.16762990e-01 4.35168445e-01 -5.33675030e-02 -7.50932336e-01
-7.24593997e-01 8.13795924e-01 3.42453539e-01 2.14164466e-01
-6.35168731e-01 -3.02151591e-02 6.32624328e-01 1.06505819e-01
6.67814732e-01 2.18166724e-01 -2.72820592e-01 -5.59808731e-01
-9.23851132e-01 1.81802318e-01 3.53768051e-01 6.63360834e-01
6.06448114e-01 7.12841153e-02 -6.94994926e-01 1.12652659e+00
1.10946529e-01 9.78623778e-02 1.02334070e+00 -7.88854182e-01
5.84055603e-01 9.27109599e-01 -5.11602918e-03 -5.74395716e-01
-1.91426352e-01 -4.46075290e-01 -5.58393538e-01 3.58569920e-01
2.96875477e-01 1.01479352e-01 -9.41036344e-01 1.65323663e+00
4.09194469e-01 4.39513892e-01 1.55523196e-02 6.24624014e-01
6.22437418e-01 4.15242761e-01 3.40369523e-01 -4.14546013e-01
9.35899615e-01 -1.34572840e+00 -6.73469424e-01 -3.23937416e-01
6.53489232e-01 -2.22690582e-01 1.20360899e+00 4.13732886e-01
-1.03488255e+00 -7.54488051e-01 -1.16590536e+00 2.13859677e-01
-3.17860454e-01 1.30087301e-01 3.96816343e-01 4.61937487e-01
-3.93749654e-01 6.68400764e-01 -8.75554264e-01 -2.59137213e-01
9.83255565e-01 1.37414381e-01 -1.17365621e-01 -1.41799703e-01
-9.31544363e-01 7.76849210e-01 5.19865453e-01 -2.46360376e-01
-8.62603605e-01 -8.39930892e-01 -8.20871890e-01 1.10691354e-01
7.43364453e-01 -4.43841189e-01 1.31426525e+00 -1.17833030e+00
-1.50341964e+00 5.25637746e-01 -1.38890028e-01 -7.33789444e-01
7.30485499e-01 -5.33116639e-01 -4.36822623e-01 1.59764335e-01
1.23666435e-01 4.42338049e-01 9.99551415e-01 -9.83823180e-01
-1.02323186e+00 -5.10254622e-01 -1.39606163e-01 3.45781714e-01
-5.05328178e-01 -2.89201021e-01 -4.08014059e-01 -6.72331452e-01
-7.06427470e-02 -5.62695026e-01 -1.23888999e-01 -4.01284583e-02
-1.31963464e-02 -3.57051641e-01 9.38810349e-01 -1.84790313e-01
1.34246099e+00 -2.25982571e+00 -1.86507904e-03 -2.72722065e-01
7.85599276e-03 6.59254909e-01 -3.00809711e-01 3.32792699e-01
-1.25407085e-01 -5.27928174e-01 -2.67978907e-01 -4.92284924e-01
-1.12932742e-01 1.67032465e-01 -3.82847995e-01 2.76069701e-01
3.29489648e-01 8.04562211e-01 -1.06925797e+00 -4.98624295e-01
4.18053687e-01 2.69086033e-01 -4.53736901e-01 2.04089880e-01
-3.98155421e-01 1.38469249e-01 -3.93950075e-01 7.29725182e-01
3.39845538e-01 -9.16927531e-02 -2.28301704e-01 1.54122412e-02
-3.37166600e-02 -2.59553585e-02 -1.23691809e+00 1.56741560e+00
-1.21506929e-01 2.04527840e-01 -3.93029541e-01 -1.25808978e+00
9.24270511e-01 1.19268276e-01 8.16259325e-01 -9.16695178e-01
7.12571368e-02 6.53897822e-02 -1.57809779e-01 -6.14125669e-01
3.57761174e-01 -1.85005605e-01 1.71670541e-01 3.57574463e-01
7.37949759e-02 2.54491419e-01 3.44112426e-01 -7.06527084e-02
1.18029225e+00 3.53142619e-01 4.55691248e-01 2.33362630e-01
4.88091886e-01 -5.58887199e-02 1.07983971e+00 7.70149648e-01
-7.83323884e-01 5.55045366e-01 3.61907035e-01 -4.55032796e-01
-6.43495023e-01 -8.47516894e-01 1.37103528e-01 1.42026699e+00
1.42026767e-01 -3.73275995e-01 -6.06305480e-01 -1.05679011e+00
7.88826644e-02 9.60168779e-01 -7.84288764e-01 -7.48806357e-01
-4.08501923e-01 -8.83208275e-01 1.78246781e-01 9.58630681e-01
6.42737865e-01 -1.30035436e+00 -1.03160131e+00 3.14345479e-01
4.12057377e-02 -6.91563070e-01 -3.68942559e-01 3.19877684e-01
-1.08418572e+00 -1.23388696e+00 -6.86564505e-01 -4.98643279e-01
4.40078378e-01 3.53507310e-01 7.53998280e-01 -2.26632923e-01
-4.42396760e-01 4.71670747e-01 -7.12611794e-01 -5.91350496e-01
-1.71845034e-02 -9.34134051e-02 -2.56551299e-02 4.26192820e-01
7.72263885e-01 -4.29959327e-01 -5.25063455e-01 2.31254727e-01
-7.86337376e-01 -3.30331713e-01 7.40020394e-01 1.16924524e+00
7.40814865e-01 -5.35627594e-03 9.31478798e-01 -7.53581941e-01
3.77946556e-01 -5.85285783e-01 -1.97129920e-01 3.69038850e-01
-7.41151989e-01 -2.52865583e-01 5.53741932e-01 -6.67521656e-01
-1.15475667e+00 1.01408020e-01 1.67276204e-01 -6.01885438e-01
-2.15110153e-01 2.42678791e-01 -2.40514800e-01 2.46198729e-01
5.53310096e-01 4.46597725e-01 2.32973501e-01 -4.09137100e-01
1.96234167e-01 5.49758315e-01 4.26673472e-01 -3.16044837e-01
2.72254705e-01 5.09009957e-01 -2.07049042e-01 -7.92573988e-01
-1.08107555e+00 -6.82222426e-01 -6.97500288e-01 -3.61145496e-01
6.88173294e-01 -8.09058547e-01 -3.96009266e-01 6.64729595e-01
-5.19575715e-01 -4.39713180e-01 -1.03465855e+00 6.94811940e-01
-9.76133347e-01 2.88483620e-01 -3.26279104e-01 -1.05515516e+00
-2.65317351e-01 -8.84447098e-01 1.00316179e+00 3.72045577e-01
-9.16120410e-02 -6.04927778e-01 3.93336922e-01 5.40363729e-01
2.34383777e-01 2.26111129e-01 7.29079366e-01 -8.68539333e-01
-2.41683856e-01 -3.68636906e-01 6.21876307e-02 6.48688316e-01
3.24248224e-01 -1.70795485e-01 -1.04469657e+00 -3.83690357e-01
1.27875194e-01 -8.46912682e-01 1.21932459e+00 3.96498978e-01
1.20362985e+00 7.22333556e-03 -2.04984769e-01 3.47692043e-01
1.26663232e+00 5.99756539e-01 8.26420486e-01 4.09294248e-01
4.05107081e-01 3.21845323e-01 1.22195268e+00 7.93394566e-01
1.19887426e-01 6.18101835e-01 4.54156816e-01 1.10470660e-01
-9.42332074e-02 -2.14881897e-01 3.64164859e-01 1.98012605e-01
-8.23057350e-03 -3.86845320e-02 -5.86855173e-01 5.15825152e-01
-2.33962011e+00 -1.42311060e+00 4.04609501e-01 2.39614916e+00
6.58492684e-01 3.20891738e-01 4.94197577e-01 3.13584179e-01
6.01370156e-01 3.06443810e-01 -1.05843568e+00 3.27610932e-02
-2.15650871e-02 1.21904597e-01 2.97537632e-02 1.46800473e-01
-1.53208673e+00 8.57612073e-01 5.43051720e+00 9.48515534e-01
-9.05690849e-01 8.17488432e-02 6.14899814e-01 -6.18963838e-01
4.28061604e-01 -1.69366196e-01 -8.35331976e-01 6.19021297e-01
7.88645744e-01 -8.24894607e-02 -7.75256902e-02 1.17457628e+00
2.79599249e-01 -3.66448641e-01 -1.12562883e+00 1.01780999e+00
3.62512648e-01 -1.05122793e+00 -4.61146794e-02 -1.76506966e-01
7.61225820e-01 -7.30619654e-02 2.90567484e-02 7.70168006e-01
-4.04214561e-02 -6.46152794e-01 6.41634047e-01 8.16203654e-01
3.41424793e-01 -9.72085059e-01 6.22398078e-01 4.98267770e-01
-1.06541979e+00 -6.63595617e-01 -4.68628705e-01 -1.61376506e-01
-3.89335826e-02 3.98704320e-01 -4.37633663e-01 5.41013837e-01
6.00785136e-01 1.03689051e+00 -6.46768868e-01 1.37380624e+00
1.37923762e-01 4.70576048e-01 1.05521247e-01 -9.08666383e-03
3.26994747e-01 -5.73897138e-02 5.31314611e-01 7.78364420e-01
2.99626477e-02 1.14896193e-01 4.74668056e-01 5.67028940e-01
1.81904286e-01 2.36856043e-01 -3.54753882e-01 5.54146506e-02
3.88342917e-01 8.84628892e-01 -5.02413690e-01 -5.78926086e-01
-4.96380836e-01 1.09827399e+00 4.91506130e-01 9.44708437e-02
-8.58593225e-01 -5.19315481e-01 5.41473866e-01 -9.60629433e-02
6.50423050e-01 2.71995991e-01 -1.38802201e-01 -1.12244880e+00
1.81154013e-01 -1.00759327e+00 7.78850913e-01 -3.99475098e-01
-1.35047174e+00 3.09511214e-01 -9.78699233e-03 -1.62677193e+00
-3.28324854e-01 -2.79549062e-01 -7.77755320e-01 3.91931981e-01
-1.41895878e+00 -9.35442805e-01 -3.99956405e-01 5.55728495e-01
1.01125193e+00 -3.95140946e-01 6.66433275e-01 4.21865970e-01
-9.06131923e-01 7.16678262e-01 2.69597828e-01 -6.35509491e-02
8.50394368e-01 -1.14653337e+00 -2.59479970e-01 7.21733272e-01
-1.17075495e-01 1.69479281e-01 3.43264431e-01 -6.90581560e-01
-1.02337551e+00 -1.29881644e+00 6.58943355e-01 -3.63819540e-01
3.34165454e-01 1.42808584e-02 -1.08413780e+00 4.88386065e-01
-3.13377678e-01 2.38433242e-01 8.85899127e-01 -2.81407102e-03
-3.49180639e-01 -5.53348064e-01 -1.18664217e+00 4.20076072e-01
9.21658337e-01 -1.42506152e-01 -8.72388959e-01 3.31825674e-01
4.21456218e-01 -4.75402698e-02 -6.45132482e-01 5.45949042e-01
5.48246503e-01 -1.27137363e+00 7.56696641e-01 -9.44513738e-01
3.21241915e-01 -1.50735185e-01 -1.47665754e-01 -1.26200974e+00
-3.29292357e-01 -2.61510402e-01 -5.33871830e-01 1.22733557e+00
2.85604317e-02 -5.61255753e-01 1.02452731e+00 5.28901637e-01
-2.63892472e-01 -1.11567259e+00 -1.00906944e+00 -1.19151390e+00
-1.49811253e-01 -3.70963812e-01 3.63818973e-01 6.62332952e-01
1.22958578e-01 3.05702955e-01 -5.38132071e-01 -3.93134952e-01
6.42759681e-01 1.13186784e-01 7.56512225e-01 -1.14617634e+00
-3.95386487e-01 -5.27572215e-01 -4.69980508e-01 -8.08678210e-01
-5.39121553e-02 -3.98297399e-01 3.26855540e-01 -1.20555758e+00
4.52993572e-01 -1.73281476e-01 -7.00099409e-01 6.05622351e-01
-4.80507076e-01 8.58308822e-02 2.70622313e-01 2.79016525e-01
-1.14858818e+00 1.28223598e+00 8.11843336e-01 -2.40369320e-01
-5.83211839e-01 3.82397294e-01 -4.79097724e-01 7.91572988e-01
7.24458992e-01 -4.42577869e-01 -7.63762951e-01 1.12777054e-02
-4.71264482e-01 -2.46580854e-01 2.77382910e-01 -1.47670686e+00
1.64163992e-01 -2.82442600e-01 4.32084113e-01 -6.96600616e-01
5.65861702e-01 -6.82817698e-01 -4.19949770e-01 7.10543871e-01
-4.74147290e-01 -4.58235323e-01 -1.24350442e-02 9.80434537e-01
-2.73729801e-01 -2.45348766e-01 1.11626697e+00 2.18017716e-02
-1.00084531e+00 4.78217185e-01 -6.67989776e-02 2.63634652e-01
1.54632556e+00 -5.49226284e-01 -2.94123232e-01 -1.75714627e-01
-7.04653621e-01 3.25114638e-01 2.60953695e-01 5.15943050e-01
7.25694537e-01 -1.40659904e+00 -5.21917701e-01 2.54429728e-01
6.05634093e-01 -2.46154875e-01 5.08710384e-01 9.79949594e-01
7.25987777e-02 1.90099850e-01 -3.52208704e-01 -4.60487634e-01
-1.28499079e+00 7.13900447e-01 2.97580540e-01 -3.42744112e-01
-7.49973714e-01 7.46374667e-01 -5.54865077e-02 -5.63313551e-02
7.78790534e-01 -1.88137114e-01 -2.85282344e-01 3.70232135e-01
9.73527789e-01 8.50178778e-01 -5.30401133e-02 -5.04215896e-01
-3.90195876e-01 4.00963247e-01 -2.95196533e-01 1.45783246e-01
1.49706662e+00 1.37619630e-01 3.88271660e-01 8.15477729e-01
9.81843412e-01 -5.19984841e-01 -1.73305762e+00 -4.73072201e-01
7.98913538e-02 -6.69228733e-01 -1.30451620e-01 -8.80748212e-01
-8.78738880e-01 7.68893421e-01 9.40981507e-01 -1.31007880e-01
1.34059584e+00 -1.51982546e-01 7.99803138e-01 4.86053973e-01
1.86092868e-01 -1.65968549e+00 5.88330865e-01 4.80797410e-01
8.08342040e-01 -1.42380571e+00 4.33697626e-02 -4.97996174e-02
-8.56184006e-01 8.72095406e-01 9.75079536e-01 -1.94098413e-01
4.71174270e-01 -1.67842478e-01 -1.19745918e-01 -1.20246254e-01
-9.04111147e-01 -3.50885600e-01 2.64395833e-01 6.16451740e-01
2.19159760e-02 -2.80711740e-01 -3.91475767e-01 7.65885115e-01
5.64614356e-01 2.91732281e-01 1.14000119e-01 1.32290697e+00
-8.47715914e-01 -9.58289087e-01 -5.21468092e-03 6.41292930e-01
-1.71437293e-01 2.19146401e-01 -2.93916106e-01 6.29654765e-01
2.81838059e-01 8.98977697e-01 2.81939238e-01 -4.25702959e-01
6.67589784e-01 3.56223226e-01 3.84489328e-01 -7.47442901e-01
-3.66774321e-01 -8.87684077e-02 -1.76334143e-01 -9.38212812e-01
-4.39018339e-01 -7.76746154e-01 -1.14916563e+00 2.49513984e-01
-3.37767363e-01 5.08573204e-02 1.62015066e-01 1.03962457e+00
5.15571058e-01 6.41843081e-01 7.41326630e-01 -7.25089371e-01
-1.07686973e+00 -1.04600298e+00 -6.99162662e-01 6.03466153e-01
2.88387746e-01 -9.77073729e-01 -3.73673856e-01 -1.08456068e-01]
|
[8.496225357055664, 0.8858839273452759]
|
c9c4a376-de73-411f-a18d-785a90e4e4a2
|
automatic-renal-segmentation-in-dce-mri-using
|
1712.07022
| null |
http://arxiv.org/abs/1712.07022v1
|
http://arxiv.org/pdf/1712.07022v1.pdf
|
Automatic Renal Segmentation in DCE-MRI using Convolutional Neural Networks
|
Kidney function evaluation using dynamic contrast-enhanced MRI (DCE-MRI)
images could help in diagnosis and treatment of kidney diseases of children.
Automatic segmentation of renal parenchyma is an important step in this
process. In this paper, we propose a time and memory efficient fully automated
segmentation method which achieves high segmentation accuracy with running time
in the order of seconds in both normal kidneys and kidneys with hydronephrosis.
The proposed method is based on a cascaded application of two 3D convolutional
neural networks that employs spatial and temporal information at the same time
in order to learn the tasks of localization and segmentation of kidneys,
respectively. Segmentation performance is evaluated on both normal and abnormal
kidneys with varying levels of hydronephrosis. We achieved a mean dice
coefficient of 91.4 and 83.6 for normal and abnormal kidneys of pediatric
patients, respectively.
|
['Simon K. Warfield', 'Marzieh Haghighi', 'Sila Kurugol']
|
2017-12-19
| null | null | null | null |
['kidney-function']
|
['medical']
|
[-9.02858227e-02 -2.63252586e-01 2.95112967e-01 -7.13155031e-01
-1.31072044e-01 -6.48477197e-01 2.24497944e-01 2.88339406e-01
-6.68953001e-01 5.22142231e-01 -1.79067224e-01 -2.31870130e-01
-2.36984327e-01 -8.87766421e-01 -2.15144128e-01 -7.30699778e-01
-4.76883799e-01 9.11914766e-01 2.97289610e-01 4.21399921e-01
2.39618599e-01 1.06587863e+00 -1.07349944e+00 -9.87269580e-02
1.30244851e+00 5.13928950e-01 4.33063924e-01 1.09003222e+00
-5.24627984e-01 9.47500408e-01 -9.89192165e-03 -1.86552361e-01
4.47626054e-01 -5.18353462e-01 -6.59834623e-01 1.70922801e-01
6.15611970e-01 -6.87966168e-01 -4.56576258e-01 1.07211924e+00
9.17214274e-01 -8.85692611e-02 5.12452006e-01 -4.98249471e-01
-5.03908753e-01 7.95016110e-01 -6.12335682e-01 6.58088863e-01
-4.35458988e-01 1.79830983e-01 -1.06717482e-01 -5.65239608e-01
5.30753613e-01 7.51755297e-01 5.33559382e-01 5.90116858e-01
-1.00330687e+00 -7.78916299e-01 -5.35930574e-01 1.77926775e-02
-9.51555967e-01 1.98975235e-01 1.27935246e-01 -7.23668516e-01
5.71880460e-01 -4.07285709e-03 9.46172833e-01 -2.36324489e-01
4.01467443e-01 6.63307190e-01 1.42785668e+00 -1.45571902e-01
2.22208366e-01 -6.85889482e-01 3.12133789e-01 5.53509772e-01
5.01378477e-01 6.26173839e-02 2.44937807e-01 3.41105014e-01
1.13520825e+00 2.21186563e-01 -5.05938940e-02 -4.95024502e-01
-1.22676051e+00 6.54497683e-01 5.01270831e-01 5.04704535e-01
-8.92810702e-01 3.75423491e-01 4.95869815e-01 7.93159232e-02
1.40582681e-01 -9.80307534e-03 -2.17239380e-01 -6.77735582e-02
-8.95647585e-01 1.52789816e-01 6.29851401e-01 6.82988465e-01
4.40741442e-02 -4.91050370e-02 -3.40809852e-01 8.93194377e-01
6.02680780e-02 6.67726040e-01 4.77007568e-01 -8.55434537e-01
4.66405638e-02 4.80204582e-01 -2.42023200e-01 -7.92243704e-02
-7.03624785e-01 -3.26731116e-01 -1.02449501e+00 5.36163449e-01
4.01514620e-01 -1.72072381e-01 -1.87993705e+00 1.17692435e+00
6.15036190e-01 3.14905107e-01 -1.25450596e-01 1.05431843e+00
9.67190027e-01 1.18887916e-01 3.75062674e-01 -4.66258712e-02
1.36005998e+00 -1.02806485e+00 -6.94974899e-01 4.77355808e-01
4.39489335e-01 -7.61284709e-01 1.94412529e-01 -1.76128913e-02
-1.47579610e+00 4.20358144e-02 -7.60333240e-01 8.54192451e-02
-9.30775777e-02 4.34159823e-02 7.57996500e-01 5.88926494e-01
-1.08551228e+00 8.04455757e-01 -1.33247340e+00 -1.98271453e-01
6.85936511e-01 7.47179747e-01 -4.61255252e-01 -2.47796178e-01
-6.64686561e-01 1.17837942e+00 3.18433464e-01 2.32019454e-01
-5.63450694e-01 -1.04393077e+00 -4.98281419e-01 9.29242522e-02
-3.23749602e-01 -6.41177595e-01 1.35700715e+00 -2.68104970e-01
-1.39946890e+00 1.12114012e+00 1.61645412e-01 -4.92852420e-01
9.93281484e-01 2.64288206e-02 -2.03262642e-02 6.22056127e-01
6.08267151e-02 5.27102411e-01 3.06258742e-02 -7.58301497e-01
-6.48837805e-01 -8.44457388e-01 -5.89924037e-01 2.96850801e-01
5.55091500e-01 1.33576244e-01 -4.74939764e-01 -3.27479333e-01
7.49478936e-01 -9.26212609e-01 -5.22744358e-01 3.23733568e-01
-5.40120080e-02 3.91837768e-03 6.22028470e-01 -1.24914646e+00
6.71610773e-01 -1.38326859e+00 -4.98693675e-01 4.72901464e-01
5.32153070e-01 5.22642434e-01 1.56888455e-01 -1.69046089e-01
-2.28671327e-01 -1.54937610e-01 -3.79501075e-01 3.60347658e-01
-4.10931140e-01 3.04257661e-01 5.61510861e-01 7.72787273e-01
-1.99898437e-01 1.24272895e+00 -9.23998296e-01 -5.66502273e-01
6.52722597e-01 6.06999993e-01 -7.28261098e-02 1.72990173e-01
3.62600267e-01 8.58286083e-01 -3.68951321e-01 6.46516085e-01
1.16678178e+00 -8.51759762e-02 2.17437834e-01 2.68119782e-01
-4.72635865e-01 -2.58260310e-01 -8.73690009e-01 1.50007427e+00
-2.39708304e-01 3.79050851e-01 3.56558412e-01 -7.65826166e-01
6.41437232e-01 5.35635889e-01 9.17288899e-01 -1.14412069e+00
1.99569210e-01 4.71534640e-01 4.94724423e-01 -7.99110889e-01
-1.74636126e-01 -3.59313369e-01 6.15097106e-01 5.54862976e-01
-1.18863612e-01 -4.05743606e-02 6.88567996e-01 -1.41311899e-01
1.26768851e+00 -8.09325799e-02 5.31445956e-03 -4.86610264e-01
5.38758755e-01 -4.34088521e-02 2.74987847e-01 8.23219836e-01
-6.38357341e-01 8.16417754e-01 2.15090275e-01 -7.00487137e-01
-1.41406226e+00 -1.39125741e+00 -4.98898596e-01 3.79275143e-01
2.63478547e-01 7.20907867e-01 -8.13012183e-01 -4.75034833e-01
2.58255005e-01 2.36614048e-01 -5.77178061e-01 6.59609318e-01
-1.05606616e+00 -6.56316578e-01 6.03450656e-01 8.41892660e-01
7.24552274e-01 -1.08853924e+00 -8.99233103e-01 4.37619865e-01
2.52339393e-01 -8.14141870e-01 -1.64466321e-01 1.39236227e-01
-1.28983974e+00 -1.34141004e+00 -1.38911641e+00 -9.92396832e-01
9.63188648e-01 1.92424636e-02 9.40315127e-01 2.12872699e-01
-8.53900492e-01 2.09815223e-02 3.29005308e-02 -2.58031845e-01
-2.77660638e-01 -1.98796839e-01 -4.96099859e-01 -7.70319581e-01
1.77875131e-01 -7.75464952e-01 -1.23775530e+00 -1.26353472e-01
-8.88724983e-01 1.75834179e-01 8.87172043e-01 7.02001333e-01
5.36740243e-01 -3.20639342e-01 3.07061106e-01 -1.12818301e+00
3.65639240e-01 -3.88595670e-01 -1.09238851e+00 4.30395901e-01
-8.27590764e-01 9.34329405e-02 2.10438430e-01 -2.18500748e-01
-9.05378997e-01 3.26579154e-01 -2.06601739e-01 -3.65866154e-01
-2.65535653e-01 1.98082313e-01 7.63507724e-01 -4.87646312e-01
1.58746600e-01 3.23229104e-01 1.99174270e-01 -6.21071994e-01
2.57005632e-01 3.49236965e-01 1.01561761e+00 -3.74889821e-01
3.74158174e-01 4.92954850e-01 4.32384908e-01 -4.01787490e-01
-1.14536330e-01 -5.89008570e-01 -8.92459810e-01 -2.57868052e-01
1.19144022e+00 -6.53896451e-01 -8.26459467e-01 8.68911028e-01
-8.43941271e-01 -2.40269467e-01 -6.69855028e-02 9.56499696e-01
-2.51739562e-01 4.68401074e-01 -1.24660003e+00 -4.39196229e-01
-1.02750564e+00 -1.11247110e+00 4.03899968e-01 7.42394686e-01
4.00612146e-01 -1.08049011e+00 1.52236149e-01 3.69445175e-01
9.33849394e-01 6.21137142e-01 9.49994862e-01 -9.31888342e-01
-7.98106134e-01 -3.23283404e-01 -5.41587353e-01 1.56535268e-01
-1.78637467e-02 -1.30875617e-01 -5.25147319e-01 -1.06240734e-02
-2.52612174e-01 4.82991561e-02 7.08598256e-01 1.06854248e+00
1.09170353e+00 2.24816710e-01 -2.07709428e-03 6.33951604e-01
1.79204214e+00 6.99776113e-01 6.41287506e-01 1.47556737e-01
5.06859660e-01 5.69230139e-01 2.74993181e-01 3.49875569e-01
2.40664721e-01 -6.89171255e-02 2.83832401e-01 -2.40248695e-01
-3.32790524e-01 6.37374446e-02 -7.73354411e-01 9.40600932e-01
-1.83812633e-01 2.45926753e-01 -1.46248710e+00 9.49271798e-01
-1.45645201e+00 -8.11998069e-01 -6.25770032e-01 2.07938957e+00
6.64494097e-01 -3.60203028e-01 -3.24204534e-01 -3.24428558e-01
1.22459650e+00 -5.04087150e-01 -5.53451180e-01 -3.66143227e-01
3.76404881e-01 1.00276256e+00 9.59288657e-01 4.50375646e-01
-9.90163803e-01 4.47131962e-01 6.21388817e+00 9.10581946e-02
-1.47337317e+00 -8.06258805e-03 5.31171560e-01 1.27869651e-01
-8.81223008e-03 -5.12135169e-03 -2.04125550e-02 5.31662703e-01
4.67073202e-01 -2.51578569e-01 2.43679807e-01 4.81781632e-01
1.57329559e-01 -4.54046279e-01 -7.08586574e-01 6.82083011e-01
-4.97802138e-01 -1.33904803e+00 -5.83793759e-01 -7.02381507e-02
8.87720466e-01 5.65032840e-01 -2.72642732e-01 4.02750038e-02
2.43249685e-01 -1.19353497e+00 -1.78930722e-02 6.17272854e-01
6.28723323e-01 -6.87485933e-01 1.09777629e+00 2.39662841e-01
-1.06471550e+00 3.83415580e-01 2.97942050e-02 3.97228338e-02
2.80229338e-02 6.20742679e-01 -1.12643993e+00 1.73510864e-01
5.73267341e-01 1.36901051e-01 -1.45164266e-01 1.89078999e+00
-2.78660804e-01 3.62060040e-01 -4.85447973e-01 1.63674369e-01
2.83837497e-01 -4.99576032e-01 3.52259316e-02 1.37534690e+00
3.00897777e-01 7.62568593e-01 -4.15300541e-02 8.21023941e-01
-2.26161078e-01 2.58631259e-01 -2.90130824e-01 1.84330434e-01
4.07104909e-01 1.32262647e+00 -1.34900653e+00 -5.63550353e-01
-2.97896624e-01 6.41572058e-01 2.55412459e-02 1.12651199e-01
-5.88791490e-01 -3.84853601e-01 6.56784028e-02 -9.57663804e-02
3.90124053e-01 4.95669991e-02 -7.49446511e-01 -9.67324853e-01
-5.24136191e-03 -2.85332967e-02 3.39597821e-01 -4.15667057e-01
-1.06678867e+00 1.38129383e-01 -2.81550378e-01 -5.86077094e-01
-6.43898696e-02 -2.46519655e-01 -9.86451089e-01 1.15652072e+00
-1.75649631e+00 -9.64818895e-01 -5.90108693e-01 2.30511159e-01
2.93159157e-01 4.26103532e-01 5.44645727e-01 5.78979075e-01
-3.73973161e-01 1.31396800e-01 3.37306470e-01 6.13361597e-01
3.12677801e-01 -1.76240516e+00 1.91912308e-01 1.03113687e+00
-8.09418619e-01 5.29139638e-01 4.89951462e-01 -8.79582405e-01
-1.03836370e+00 -9.83335257e-01 1.05297959e+00 3.03099543e-01
1.37206018e-01 3.76739740e-01 -8.25541437e-01 5.06180108e-01
2.57350504e-01 6.77752912e-01 4.95256126e-01 -5.74405551e-01
2.23149374e-01 1.75537989e-01 -1.71027994e+00 1.88562438e-01
5.49689770e-01 -2.27800407e-03 -5.26943505e-01 1.08474664e-01
6.11374117e-02 -1.12273848e+00 -1.42352569e+00 8.23147953e-01
8.67564857e-01 -9.34193671e-01 7.63606548e-01 -2.64037579e-01
5.20672739e-01 -2.89659262e-01 4.94703233e-01 -8.40937018e-01
-1.39768235e-02 4.09292318e-02 1.49601065e-02 7.64420390e-01
-5.51096424e-02 -4.72635061e-01 1.11685240e+00 1.07403767e+00
-2.03664169e-01 -9.36693609e-01 -7.04418480e-01 -3.64958912e-01
3.89942408e-01 1.21212872e-02 2.01348871e-01 1.01367795e+00
-4.23192382e-01 -3.48924220e-01 2.53607571e-01 2.24500179e-01
1.06000972e+00 4.56061244e-01 1.59540772e-01 -1.15584254e+00
4.61587518e-01 -5.39987266e-01 -7.33668447e-01 -5.73989034e-01
-1.39321432e-01 -1.06285989e+00 -1.02595367e-01 -2.06224227e+00
6.13870203e-01 -5.52764595e-01 -3.24047744e-01 3.39873850e-01
-4.18894403e-02 2.19159916e-01 1.40033886e-01 1.49719402e-01
-1.85641080e-01 5.67898452e-02 1.67984939e+00 1.17931254e-01
-8.80632773e-02 3.30707967e-01 -8.76205100e-04 5.24697125e-01
7.88027585e-01 -4.50449198e-01 -2.21849009e-01 -3.21909010e-01
-6.76623285e-01 4.77582335e-01 1.57274365e-01 -9.19626594e-01
5.36229074e-01 3.18995975e-02 8.89814913e-01 -1.05918658e+00
-3.15113872e-01 -7.88793206e-01 2.21254483e-01 1.37320817e+00
-4.08648491e-01 1.45927057e-01 -4.99628671e-02 9.35917497e-02
-3.57844859e-01 -2.70856261e-01 1.41197968e+00 -6.56370342e-01
-8.03658903e-01 5.99812329e-01 -3.61663669e-01 -1.66764081e-01
1.21034479e+00 -6.80265650e-02 -8.55460763e-02 -3.48885916e-02
-1.04801905e+00 8.05721045e-01 1.57229453e-01 -8.99024755e-02
7.39485145e-01 -1.15044439e+00 -8.90043020e-01 5.89979291e-02
-1.54612109e-01 2.68220216e-01 6.17137313e-01 1.32187438e+00
-1.80049658e+00 4.90032047e-01 -5.82952619e-01 -1.04370892e+00
-1.34740639e+00 2.01854818e-02 6.72759771e-01 -5.40063918e-01
-9.00921643e-01 6.02123916e-01 -2.52592295e-01 -5.72910368e-01
2.43888766e-01 -5.26463509e-01 -1.51314259e-01 -5.07192969e-01
3.97227615e-01 5.33166707e-01 1.36839092e-01 -3.14706296e-01
-2.62055695e-01 5.39713860e-01 -2.28170961e-01 1.41089961e-01
1.43373227e+00 -2.16865242e-01 -4.29272920e-01 -3.01032424e-01
9.76658881e-01 -4.30968434e-01 -1.06817877e+00 -2.25096583e-01
2.79436350e-01 -5.15970826e-01 1.56270564e-01 -1.19415700e+00
-1.40352476e+00 9.51392591e-01 1.34420836e+00 -2.46570244e-01
1.11702073e+00 -2.27616504e-01 1.03851664e+00 -1.64094180e-01
8.58466700e-02 -8.50083411e-01 -6.85057044e-01 3.41693848e-01
1.86042860e-01 -1.30252755e+00 1.16220407e-01 -2.79748976e-01
-4.37638938e-01 1.50763357e+00 6.89615250e-01 -3.79923820e-01
5.68261981e-01 5.65004051e-01 5.66167891e-01 -3.66913974e-01
-4.67273027e-01 -3.59219491e-01 1.37503907e-01 4.29228693e-01
7.78603077e-01 2.55741477e-01 -1.01954591e+00 -1.59002468e-02
3.70924562e-01 3.15341473e-01 4.59538907e-01 1.36063826e+00
-6.43333077e-01 -9.66627240e-01 -1.02932774e-01 7.09135771e-01
-9.72221017e-01 1.81428522e-01 1.84132874e-01 8.87931168e-01
1.26429006e-01 2.31643558e-01 -3.95046594e-03 3.61013770e-01
1.00123554e-01 4.09595445e-02 8.18924785e-01 -3.42139989e-01
-8.98003995e-01 1.81449801e-01 -3.14605147e-01 -5.14062405e-01
-4.73277330e-01 -3.71826231e-01 -1.88704872e+00 -3.92813295e-01
-1.63077742e-01 -1.48613021e-01 1.19809902e+00 9.93779421e-01
-1.37562633e-01 4.41309750e-01 6.25825226e-01 -4.50514704e-01
-5.65561652e-01 -8.54098082e-01 -8.38528991e-01 5.65388739e-01
3.14940810e-01 -1.38136655e-01 2.37941444e-01 -1.24347910e-01]
|
[14.264305114746094, -2.5941662788391113]
|
08f68735-0add-4428-9c13-d1bcbda78b79
|
lung-nodule-classification-using-biomarkers
|
2010.11682
| null |
https://arxiv.org/abs/2010.11682v1
|
https://arxiv.org/pdf/2010.11682v1.pdf
|
Lung Nodule Classification Using Biomarkers, Volumetric Radiomics and 3D CNNs
|
We present a hybrid algorithm to estimate lung nodule malignancy that combines imaging biomarkers from Radiologist's annotation with image classification of CT scans. Our algorithm employs a 3D Convolutional Neural Network (CNN) as well as a Random Forest in order to combine CT imagery with biomarker annotation and volumetric radiomic features. We analyze and compare the performance of the algorithm using only imagery, only biomarkers, combined imagery + biomarkers, combined imagery + volumetric radiomic features and finally the combination of imagery + biomarkers + volumetric features in order to classify the suspicion level of nodule malignancy. The National Cancer Institute (NCI) Lung Image Database Consortium (LIDC) IDRI dataset is used to train and evaluate the classification task. We show that the incorporation of semi-supervised learning by means of K-Nearest-Neighbors (KNN) can increase the available training sample size of the LIDC-IDRI thereby further improving the accuracy of malignancy estimation of most of the models tested although there is no significant improvement with the use of KNN semi-supervised learning if image classification with CNNs and volumetric features are combined with descriptive biomarkers. Unexpectedly, we also show that a model using image biomarkers alone is more accurate than one that combines biomarkers with volumetric radiomics, 3D CNNs, and semi-supervised learning. We discuss the possibility that this result may be influenced by cognitive bias in LIDC-IDRI because malignancy estimates were recorded by the same radiologist panel as biomarkers, as well as future work to incorporate pathology information over a subset of study participants.
|
['David R. Chapman', 'Phuong Nguyen', 'Sumeet Menon', 'Jayalakshmi Mangalagiri', 'Arshita Jain', 'Kushal Mehta']
|
2020-10-19
| null | null | null | null |
['lung-nodule-classification']
|
['medical']
|
[ 3.17989498e-01 7.03043044e-02 -5.56861639e-01 -2.27598831e-01
-9.66370344e-01 -4.64788258e-01 5.36846936e-01 4.37222391e-01
-7.39762664e-01 4.14618522e-01 4.23372924e-01 -8.39358449e-01
-4.32276398e-01 -9.02904809e-01 -6.12652183e-01 -7.40288615e-01
-7.18707889e-02 8.61121118e-01 3.17440242e-01 4.08248991e-01
-2.69692808e-01 6.71313167e-01 -1.12507975e+00 5.31893134e-01
4.64086771e-01 1.30016732e+00 5.07688701e-01 1.02367651e+00
9.04616192e-02 1.13097370e+00 1.75984949e-01 1.13191746e-01
4.32295114e-01 -2.73578048e-01 -8.81166339e-01 2.22752467e-01
5.81662297e-01 -6.03670180e-01 -1.58586070e-01 3.87853295e-01
3.72750849e-01 -5.55300891e-01 1.05037773e+00 -6.33338094e-01
-1.23141646e-01 4.11952913e-01 -2.98066705e-01 2.33152077e-01
-2.14139789e-01 5.77465475e-01 5.36546469e-01 -5.42862892e-01
5.47409594e-01 5.60238779e-01 1.20705378e+00 1.05420172e-01
-9.69432950e-01 -3.79859716e-01 -5.70814908e-01 -9.25740153e-02
-1.29390645e+00 3.00276816e-01 -3.79978307e-02 -7.02183604e-01
8.02276850e-01 5.28376818e-01 1.00286424e+00 5.54817855e-01
2.79262185e-01 1.60825014e-01 1.16940629e+00 -4.63013381e-01
-4.86585200e-02 3.57715666e-01 3.24817537e-03 1.32801974e+00
4.91798431e-01 4.12660211e-01 1.92274719e-01 -5.48612535e-01
9.45110261e-01 2.24385574e-01 -1.17406219e-01 -3.33972216e-01
-1.36649156e+00 8.25557411e-01 9.16675568e-01 4.43169355e-01
-6.00356400e-01 3.99256259e-01 5.73093772e-01 -1.89817280e-01
3.02556872e-01 4.19739127e-01 -2.39333495e-01 4.42129791e-01
-1.25725567e+00 -2.45212242e-01 6.10593617e-01 1.65632546e-01
4.80570465e-01 -2.60833502e-01 -3.40287685e-01 8.85333538e-01
2.80536354e-01 4.11184579e-01 9.67180550e-01 -9.23351884e-01
-1.31550148e-01 8.78764451e-01 -2.87137091e-01 -4.33717340e-01
-8.94950390e-01 -7.37232685e-01 -9.66651499e-01 7.29985014e-02
5.27556658e-01 4.08255309e-02 -1.33320045e+00 1.08336282e+00
1.13550434e-02 9.51483659e-03 -2.53535241e-01 9.26284134e-01
8.94177675e-01 -2.52309263e-01 4.01415646e-01 7.98768625e-02
1.64702952e+00 -7.60119140e-01 2.01400165e-02 1.84595644e-01
1.41541588e+00 -3.17310631e-01 7.29538500e-01 7.87860006e-02
-7.49026895e-01 -4.45978463e-01 -9.34329510e-01 2.22854078e-01
-2.95038998e-01 5.91798067e-01 7.50551403e-01 9.64403868e-01
-1.24827051e+00 4.66158956e-01 -9.75876570e-01 -7.95202732e-01
5.89825153e-01 6.93151116e-01 -4.75297600e-01 -2.09493548e-01
-9.58128691e-01 1.18427229e+00 3.90423834e-01 -2.30488792e-01
-1.10291886e+00 -8.15250814e-01 -6.83516443e-01 -1.96841046e-01
1.87941447e-01 -9.83069777e-01 1.17847431e+00 -9.07461643e-01
-6.82802260e-01 9.75750446e-01 9.28144827e-02 -6.74613118e-01
6.17882311e-01 6.24560118e-01 1.50221601e-01 4.57398981e-01
2.01244131e-01 9.36886668e-01 3.71436954e-01 -8.35617304e-01
-6.55985117e-01 -4.19575870e-01 -3.65203887e-01 1.62479296e-01
-2.17325643e-01 -5.15175641e-01 -1.37948886e-01 -3.07055801e-01
1.41746044e-01 -1.20869613e+00 -5.84634006e-01 1.83029771e-01
-1.34187713e-01 -1.10128112e-01 5.42929769e-01 -5.88738978e-01
7.03848898e-01 -1.81198406e+00 -5.23030877e-01 5.64618289e-01
5.36842346e-01 1.98836312e-01 5.95163517e-02 -2.83165485e-01
-4.64231431e-01 6.95913732e-01 -3.72185744e-02 1.34686351e-01
-5.96098542e-01 1.93489864e-01 8.15677583e-01 6.15442514e-01
3.33339691e-01 1.19331646e+00 -5.86903453e-01 -9.79880512e-01
3.81649166e-01 2.06810355e-01 -3.92967135e-01 -1.08053952e-01
-7.29880994e-03 2.85694093e-01 -2.43045628e-01 6.20806992e-01
3.22718203e-01 -6.35289907e-01 2.11683571e-01 -3.46302509e-01
-1.56922843e-02 1.55963883e-01 -4.90265399e-01 1.30746841e+00
-5.26864648e-01 4.07582760e-01 -2.40389824e-01 -6.85565174e-01
5.55930138e-01 7.17786491e-01 8.03496540e-01 -2.93200761e-01
1.51245221e-01 4.69501585e-01 3.93519104e-01 -5.27752638e-01
-2.26061672e-01 -3.63416761e-01 4.33239430e-01 5.75538278e-01
-2.06418335e-03 -3.27816755e-01 -9.28520188e-02 -2.02159919e-02
1.84796441e+00 -3.44714522e-01 4.60044205e-01 -2.50780076e-01
3.95177007e-01 7.42054462e-01 2.10472997e-02 1.02828407e+00
-1.97664097e-01 7.41293311e-01 3.93243492e-01 -4.14867461e-01
-1.07376623e+00 -1.00388634e+00 -5.34357965e-01 5.65688133e-01
-5.95096469e-01 9.21726823e-02 -2.52424091e-01 -1.04202819e+00
1.02399625e-01 3.29080909e-01 -1.04672587e+00 -6.26661777e-02
-2.83085197e-01 -1.02936542e+00 8.17381144e-01 6.49695516e-01
3.08106154e-01 -6.87183559e-01 -7.54526794e-01 4.96667810e-02
-1.31760299e-01 -7.01700211e-01 1.06428251e-01 8.55887711e-01
-1.33045149e+00 -1.47287893e+00 -1.02208924e+00 -7.02850461e-01
7.98844099e-01 2.16885000e-01 9.68434751e-01 5.62168181e-01
-6.04057550e-01 7.24333525e-01 -3.07347029e-01 -4.02004212e-01
-7.41394818e-01 2.72614419e-01 -4.67431813e-01 -5.74881196e-01
1.85398608e-01 -7.51021951e-02 -6.61797225e-01 1.00092374e-01
-8.98887277e-01 1.81311771e-01 1.31533790e+00 7.43222892e-01
6.06434464e-01 1.49142906e-01 2.05163881e-01 -1.16814649e+00
1.31436536e-04 -5.92973411e-01 -3.15244757e-02 1.22671418e-01
-5.40475070e-01 -1.15205117e-01 1.70620218e-01 -4.18987274e-01
-7.47704268e-01 4.32200223e-01 8.66942704e-02 -3.44221294e-01
-4.22838718e-01 8.79008174e-01 7.01086879e-01 -4.04111445e-01
1.13043535e+00 -1.19150825e-01 4.49807972e-01 7.51460418e-02
-2.79535204e-01 7.00095773e-01 3.11704218e-01 -1.16974398e-01
6.39195085e-01 7.32732236e-01 5.83153784e-01 -6.80886686e-01
-7.75831878e-01 -8.22877944e-01 -7.59243965e-01 -2.52286583e-01
9.85634983e-01 -1.00679922e+00 -3.90887469e-01 8.99860561e-02
-6.12323880e-01 -5.18018603e-01 -3.91693652e-01 1.12132037e+00
-4.92512077e-01 1.12727985e-01 -7.29422152e-01 -4.28238451e-01
-4.40169066e-01 -1.33747482e+00 9.03980613e-01 -1.09808490e-01
-3.17513943e-01 -9.02045608e-01 1.88406426e-02 4.03889000e-01
5.83583713e-01 2.37032801e-01 1.15267861e+00 -9.44246054e-01
-4.90729779e-01 -5.51784456e-01 -5.88353097e-01 2.92779177e-01
3.51598740e-01 -6.58965483e-02 -7.19151258e-01 7.82644823e-02
-4.58090007e-02 -5.00045598e-01 1.19304299e+00 5.82353234e-01
1.24260747e+00 -6.35177717e-02 -5.65574527e-01 3.55699599e-01
1.75700068e+00 -1.84220411e-02 4.90041256e-01 4.34670866e-01
6.86550319e-01 3.79366666e-01 1.59758776e-01 8.27982128e-02
2.66608179e-01 2.47866102e-02 6.69840991e-01 -5.02641380e-01
-4.06308800e-01 2.23241165e-01 -3.13519239e-01 1.80676967e-01
-3.78860623e-01 2.72322372e-02 -1.53551793e+00 6.42786920e-01
-1.17746150e+00 -5.95699906e-01 -5.23870051e-01 1.98181748e+00
5.26681423e-01 1.58062875e-01 3.02771293e-02 1.88929021e-01
7.61586010e-01 -3.59102875e-01 -2.38701984e-01 -5.88786602e-02
2.38566875e-01 3.45737040e-01 1.13112307e+00 2.36419775e-02
-1.09159660e+00 1.48572654e-01 6.47129869e+00 5.99929690e-01
-1.28797174e+00 3.20030719e-01 1.06385863e+00 -1.05124973e-02
6.82304502e-02 -1.04320355e-01 -1.91734642e-01 -5.05438931e-02
1.11669230e+00 4.76793498e-01 -4.64606993e-02 7.41115808e-01
2.85258025e-01 -8.75292838e-01 -9.80458081e-01 4.16647464e-01
5.78198805e-02 -1.40055764e+00 -1.99963152e-01 4.19796944e-01
5.05826116e-01 5.24191380e-01 8.48567635e-02 9.02333036e-02
2.42010817e-01 -1.10960937e+00 1.59240544e-01 5.10106385e-01
1.04615211e+00 -1.40989810e-01 1.28198731e+00 3.37389052e-01
-9.41353679e-01 4.35535572e-02 -3.21622454e-02 1.36147454e-01
-6.04448140e-01 4.90991235e-01 -1.96362162e+00 4.68921661e-01
6.17843986e-01 4.00700033e-01 -1.22837687e+00 1.23919475e+00
3.86786103e-01 1.00499046e+00 -4.29758221e-01 -1.75517827e-01
4.64262396e-01 5.10439992e-01 -1.71197634e-02 1.16705799e+00
3.50093931e-01 1.46847172e-02 1.79882407e-01 5.50588071e-01
3.01259398e-01 1.65754169e-01 -3.46139908e-01 -2.07534447e-01
3.95176038e-02 1.50338733e+00 -1.19539630e+00 -4.65326846e-01
-5.74019551e-01 3.55045646e-01 8.09773579e-02 -1.52848348e-01
-7.09401369e-01 3.03732961e-01 -4.01058316e-01 8.14080477e-01
7.61382058e-02 9.53555629e-02 -4.28316087e-01 -3.43127191e-01
-6.47347331e-01 -3.92559320e-01 5.80539882e-01 -1.00227475e+00
-1.27531123e+00 2.86201537e-01 -1.29015714e-01 -1.10665786e+00
-1.77371264e-01 -7.01666474e-01 -5.36817312e-01 7.46010900e-01
-1.35273266e+00 -1.24762964e+00 -6.91085696e-01 2.83530086e-01
1.39335599e-02 1.02414362e-01 8.72973382e-01 -8.68157595e-02
-8.23066458e-02 2.41268545e-01 -1.47880226e-01 4.84419554e-01
7.89194763e-01 -1.39473796e+00 -6.42327785e-01 -1.11418508e-01
-4.39755976e-01 7.32447803e-02 -1.53783903e-01 -9.70312417e-01
-8.98794353e-01 -1.68380296e+00 4.10837293e-01 -4.95872706e-01
4.98965800e-01 3.47508192e-01 -5.98341465e-01 5.83887696e-01
-1.67434067e-01 3.71087372e-01 9.55193877e-01 -4.12969649e-01
-3.68019119e-02 1.62521541e-01 -1.28159523e+00 2.23612174e-01
4.79017735e-01 -3.71388614e-01 -1.90146208e-01 5.62333465e-01
3.96383286e-01 -2.89149165e-01 -1.27146626e+00 9.83647525e-01
6.05786383e-01 -8.03204060e-01 8.83124769e-01 -1.17869712e-01
4.47725564e-01 -1.68214783e-01 -1.68672130e-01 -7.52389789e-01
-6.14642799e-01 8.34286988e-01 7.57642627e-01 6.13770485e-01
7.42582381e-01 -3.90095919e-01 1.15469515e+00 5.02637446e-01
-5.30614927e-02 -8.22278142e-01 -6.75990224e-01 -5.89998364e-01
1.61881834e-01 -6.05698466e-01 1.67159420e-02 6.39010429e-01
-3.67944837e-01 -2.98453450e-01 4.23686683e-01 1.30821228e-01
3.12252492e-01 -3.50179791e-01 2.45942399e-01 -1.14262867e+00
-3.02698433e-01 -3.57123852e-01 -7.06135571e-01 1.09566428e-01
-2.01489449e-01 -1.34448218e+00 -1.93860512e-02 -1.86317933e+00
8.63311470e-01 -8.22269380e-01 -3.02139133e-01 5.16095042e-01
2.21613366e-02 7.05213904e-01 -3.60136144e-02 5.09439409e-01
-4.50557977e-01 -3.03317308e-01 1.34505248e+00 -2.38116115e-01
-5.34866229e-02 6.24247566e-02 -5.62727273e-01 7.88440764e-01
7.39049315e-01 -7.44551480e-01 -1.16297156e-01 4.85128015e-02
-1.79414824e-01 2.63138831e-01 1.02329171e+00 -1.43811786e+00
2.21396089e-01 -1.32520748e-02 1.32580650e+00 -8.23280275e-01
2.35753775e-01 -9.48832989e-01 4.18760180e-01 1.26039934e+00
-3.77515763e-01 -3.28107804e-01 1.42757013e-01 5.34088850e-01
9.19459313e-02 -4.27328557e-01 7.57848024e-01 -9.63223040e-01
-2.54191339e-01 2.28059098e-01 -1.03659606e+00 -4.62463379e-01
9.33118522e-01 -4.00582075e-01 -2.13486701e-01 -2.46655390e-01
-8.70733142e-01 -1.41460467e-02 4.69692945e-01 -3.14105242e-01
2.32130229e-01 -1.23243427e+00 -7.85424292e-01 -2.53466696e-01
9.23455730e-02 1.76133767e-01 1.70649409e-01 1.49091518e+00
-9.36591566e-01 8.23720276e-01 -5.12103457e-03 -1.04308867e+00
-1.10185564e+00 4.52829748e-01 8.50732982e-01 -7.99085259e-01
-8.31713900e-02 6.30272150e-01 1.38710991e-01 -5.63734889e-01
1.26527116e-01 -3.59803051e-01 -4.61215563e-02 -5.22779254e-03
6.22390099e-02 2.98894227e-01 4.02184844e-01 -4.59110290e-01
-3.78812224e-01 3.09839994e-01 -1.14660442e-01 -4.49711233e-02
1.05014563e+00 1.89282089e-01 -1.74753278e-01 2.89837182e-01
1.30655849e+00 -4.27077711e-01 -4.83475804e-01 -2.12819904e-01
8.86503905e-02 1.05737358e-01 4.09898877e-01 -1.20110941e+00
-9.85532284e-01 5.41396677e-01 1.20129716e+00 -1.38392091e-01
1.09328103e+00 2.98018366e-01 2.03426003e-01 3.16370219e-01
-2.65792422e-02 -4.39871371e-01 1.76808447e-01 1.22748055e-01
5.84531546e-01 -1.46629071e+00 3.33235562e-01 -4.59499627e-01
-6.03982985e-01 1.21965897e+00 7.37657547e-01 -3.95433426e-01
9.04677033e-01 2.30128199e-01 1.77258193e-01 -2.35325068e-01
-7.64767170e-01 -7.30723977e-01 3.42515290e-01 5.87269425e-01
6.61951184e-01 2.70590454e-01 -2.14290708e-01 5.75069308e-01
-7.13812113e-02 4.27697450e-01 4.24917281e-01 9.30782974e-01
-5.53839505e-01 -8.73779893e-01 -5.37058234e-01 1.51369894e+00
-5.37532866e-01 -1.13358974e-01 -5.55533230e-01 1.20101523e+00
5.44955254e-01 6.18320465e-01 3.48912805e-01 -3.36457700e-01
-1.23309880e-01 7.79848918e-02 3.76404732e-01 -9.59140718e-01
-9.97811019e-01 2.72065014e-01 8.15174133e-02 -1.39679521e-01
-6.63297653e-01 -6.29025459e-01 -1.16398787e+00 1.38491541e-01
-6.58170879e-01 -1.05871789e-01 8.14448774e-01 1.03559256e+00
-2.18814790e-01 7.71563888e-01 3.64987195e-01 -6.07384741e-01
-1.56148106e-01 -1.07303977e+00 -5.55767536e-01 -1.24366663e-01
1.67892158e-01 -4.70495909e-01 -3.71834874e-01 -1.22411311e-01]
|
[15.35405158996582, -2.182542324066162]
|
0ea6aa81-6381-4a64-91a2-41b0226ecedc
|
a-physics-informed-neural-network-for-wind
| null | null |
http://www.phmsociety.org/node/2736
|
http://www.phmsociety.org/sites/phmsociety.org/files/phm_submission/2019/ijphm_20_003.pdf
|
A physics-informed neural network for wind turbine main bearing fatigue
|
Unexpected main bearing failure on a wind turbine causes unwanted maintenance and increased operation costs (mainly due to crane, parts, labor, and production loss). Unfortunately, historical data indicates that failure can happen far earlier than the component design lives. Root cause analysis investigations have pointed to problems inherent from manufacturing as the major contributor, as well as issues related to event loads (e.g., startups, shutdowns, and emergency stops), extreme environmental conditions, and maintenance practices, among others. Altogether, the multiple failure modes and contributors make modeling the remaining useful life of main bearings a very daunting task. In this paper, we present a novel physics-informed neural network modeling approach for main bearing fatigue. The proposed approach is fully hybrid and designed to merge physics-informed and data-driven layers within deep neural networks. The result is a cumulative damage model where the physics-informed layers are used model the relatively well-understood physics (L10 fatigue life) and the data-driven layers account for the hard to model components (e.g., contribution due to poor greasing conditions).
|
['Yigit A. Yucesan', 'Felipe A. C. Viana']
|
2020-05-05
| null | null | null |
international-journal-of-prognostics-and
|
['physics-informed-machine-learning', 'graph-regression', 'graph-to-sequence']
|
['graphs', 'graphs', 'natural-language-processing']
|
[-4.37480360e-01 -2.38066792e-01 2.65471935e-01 3.33286732e-01
3.02153151e-03 -6.57930970e-02 1.45024061e-01 3.14190209e-01
2.03990310e-01 7.39083052e-01 -1.17019452e-01 -1.85633332e-01
-7.06165075e-01 -9.62882817e-01 -7.06626534e-01 -7.85032630e-01
-2.35827610e-01 5.91901720e-01 1.32390216e-01 -3.50490868e-01
1.63296372e-01 8.91755521e-01 -1.89524853e+00 -3.07786196e-01
6.44303083e-01 9.95321631e-01 2.98827112e-01 2.46023461e-01
1.71653241e-01 6.17187500e-01 -8.75011325e-01 1.43697143e-01
-2.90416479e-01 5.03360108e-02 -7.07094133e-01 -4.76506501e-02
-3.70160669e-01 -8.46339524e-01 -2.02052742e-01 5.58961034e-01
4.01437074e-01 3.34545046e-01 8.98977101e-01 -1.33114052e+00
-4.93093938e-01 3.48584384e-01 -5.27552128e-01 1.50178775e-01
-2.30156556e-01 1.03869781e-01 4.66731668e-01 -1.01382685e+00
4.65694964e-02 1.02250171e+00 8.07380140e-01 1.74692452e-01
-6.29391432e-01 -1.56077847e-01 -2.51718611e-01 2.69121140e-01
-1.14036655e+00 -9.38145891e-02 1.04833376e+00 -5.38166940e-01
1.40586472e+00 5.44902273e-02 5.48548222e-01 8.22775662e-01
1.23588789e+00 2.41913870e-01 4.99648571e-01 -2.63833165e-01
4.97116953e-01 -3.59123170e-01 1.45625532e-01 2.52038985e-01
5.79240620e-01 7.09317446e-01 -1.99486300e-01 1.59350842e-01
6.27614498e-01 2.73840338e-01 -1.25820011e-01 9.84065160e-02
-1.51839316e-01 3.51876408e-01 3.94006431e-01 2.81653464e-01
-6.33740187e-01 2.82838702e-01 4.57295120e-01 1.02902591e-01
4.75868911e-01 2.20349252e-01 -9.82916832e-01 -2.48248070e-01
-1.04653001e+00 3.04795742e-01 5.56951463e-01 4.82162029e-01
5.83731174e-01 5.97974300e-01 4.90555078e-01 6.75910771e-01
4.82131541e-01 3.30866754e-01 2.00543761e-01 -3.57417077e-01
-1.04937866e-01 2.30684966e-01 1.52183682e-01 -8.66727829e-01
-8.54246557e-01 -9.07320857e-01 -8.87732565e-01 4.75469261e-01
-3.55254486e-02 -2.92961597e-01 -9.47946727e-01 1.38880312e+00
1.39993921e-01 -2.83136964e-02 -4.18998182e-01 8.85344446e-01
7.12698698e-01 5.02671719e-01 2.41155908e-01 -3.19943354e-02
1.18461204e+00 -6.57666743e-01 -8.01343262e-01 -4.20915455e-01
4.84726965e-01 -6.53753519e-01 8.02454174e-01 6.23873889e-01
-1.29718208e+00 -7.28414893e-01 -1.42757893e+00 1.94511279e-01
-5.71306050e-01 2.45332584e-01 6.53970897e-01 4.12054151e-01
-7.54498780e-01 1.12997055e+00 -1.08124411e+00 -4.12166230e-02
7.82564953e-02 2.58108705e-01 4.67347912e-02 1.15392990e-01
-1.38207972e+00 1.32211947e+00 3.33419263e-01 5.91456234e-01
-1.23909509e+00 -8.17305148e-01 -5.65295815e-01 3.60508263e-01
5.09015918e-01 -5.34356117e-01 1.17433155e+00 8.21233913e-02
-1.07655370e+00 -1.59789279e-01 3.22408766e-01 -1.36922836e-01
1.44117577e-02 -6.05098248e-01 -6.17323399e-01 -4.42927182e-02
-4.62913781e-01 6.72403257e-03 6.10258937e-01 -1.28829014e+00
-1.37617856e-01 -1.29071474e-01 1.07879182e-02 -2.09461868e-01
-4.35476899e-01 -9.39533114e-02 1.50660723e-01 -5.85750222e-01
-1.60952538e-01 -7.41929591e-01 -6.79475665e-02 -3.94618213e-01
-2.21521124e-01 -4.00654495e-01 1.11750984e+00 -1.14038765e+00
1.36225772e+00 -1.97781861e+00 1.48154572e-01 -2.38256082e-01
3.25671323e-02 -1.62821747e-02 2.86794722e-01 9.09123778e-01
-4.50639784e-01 1.67868942e-01 -2.21964613e-01 -2.62411654e-01
-1.11331388e-01 4.73968685e-01 -5.45560792e-02 2.72171199e-01
5.92132270e-01 4.51392442e-01 -4.75537002e-01 1.27761215e-01
5.37765682e-01 4.33805078e-01 -1.99739099e-01 -1.36708450e-02
-1.96724325e-01 -3.60773951e-02 -2.58599520e-01 7.63895750e-01
9.20240939e-01 4.59613830e-01 -4.82844621e-01 -3.89134139e-01
-1.15902945e-01 1.16835825e-01 -8.71252537e-01 1.38128495e+00
-9.39941943e-01 2.14852765e-01 1.82026654e-01 -8.30818832e-01
9.38134849e-01 4.43466544e-01 4.92703140e-01 -5.12195528e-01
5.05449235e-01 2.97114283e-01 2.22995691e-02 -5.73248804e-01
8.04138362e-01 -5.35445869e-01 -2.08913386e-02 2.66107976e-01
6.91808164e-02 -2.41124481e-01 -2.54774597e-02 6.47589564e-02
1.20433462e+00 4.23008412e-01 -4.03006941e-01 -4.15463924e-01
1.86502814e-01 1.42788589e-02 7.33897805e-01 -1.18330538e-01
3.12742451e-03 2.54847646e-01 5.26286662e-01 -4.29706901e-01
-1.17922962e+00 -9.05364335e-01 3.82464230e-02 4.14018273e-01
1.75578430e-01 -1.08485214e-01 -6.79491639e-01 -2.29528353e-01
1.99952170e-01 9.93323326e-01 -4.78331298e-01 -9.40974057e-01
-5.64535141e-01 -8.19036305e-01 8.44581202e-02 1.04294980e+00
5.17931320e-02 -1.04152429e+00 -8.05599630e-01 6.50557280e-01
2.87725300e-01 -5.76050103e-01 7.40873963e-02 6.02371633e-01
-1.21121621e+00 -1.09406853e+00 -2.13286802e-01 -3.77128273e-01
4.76036042e-01 7.87212402e-02 1.26327372e+00 6.32303536e-01
-6.82043850e-01 -1.26480922e-01 -1.55023843e-01 -5.63132465e-01
-4.39853579e-01 -9.91648212e-02 4.46586072e-01 -8.13222051e-01
1.51398554e-01 -5.16776204e-01 -5.69629490e-01 4.95899916e-01
-9.50721383e-01 -2.97007829e-01 6.54085696e-01 7.93374479e-01
2.08858788e-01 1.43539453e+00 9.13452983e-01 -2.66542763e-01
6.20121360e-01 -7.57750392e-01 -2.16077536e-01 -1.34788856e-01
-7.72470236e-01 -2.13866144e-01 8.69146645e-01 -1.42449439e-01
-1.26448095e+00 -5.18788636e-01 -3.60409081e-01 -5.62322438e-01
-5.22489488e-01 1.12870502e+00 -3.65333080e-01 2.85591841e-01
1.95072651e-01 -1.74648359e-01 1.42119169e-01 -8.44935596e-01
-1.70665160e-01 4.45456982e-01 4.38138634e-01 -7.85458446e-01
1.05691624e+00 -1.27093662e-02 3.27472627e-01 -9.53159690e-01
-4.17212427e-01 1.00998946e-01 -4.34451789e-01 -7.74292529e-01
3.48660499e-01 -7.75453150e-01 -6.61521971e-01 1.00481546e+00
-8.98039758e-01 -1.84508801e-01 -3.29093069e-01 3.20372790e-01
-1.74412638e-01 2.85990685e-01 -8.93492579e-01 -1.05055428e+00
-4.51663554e-01 -1.24060392e+00 7.43038476e-01 3.19631219e-01
-2.01942474e-01 -1.02211165e+00 -3.90744567e-01 2.03599676e-01
6.38857663e-01 7.54978776e-01 1.20654738e+00 6.95029944e-02
-2.34400287e-01 -4.13192064e-01 -2.08731703e-02 6.62192464e-01
3.65193754e-01 4.67870027e-01 -6.20306075e-01 -4.86043602e-01
3.68134141e-01 -2.64198452e-01 4.43448901e-01 6.18660450e-01
1.04014552e+00 2.49680057e-01 -3.15258712e-01 -2.73659024e-02
1.64859474e+00 4.81073052e-01 6.45264268e-01 1.87844053e-01
4.34834450e-01 7.32275188e-01 8.53422344e-01 5.39036036e-01
-1.24560721e-01 3.25028121e-01 1.02012694e+00 -2.29969904e-01
-1.39937699e-01 1.34563996e-02 2.54367232e-01 1.08286726e+00
-2.04959437e-01 -4.61006701e-01 -8.32093656e-01 8.06127608e-01
-1.35871315e+00 -6.76850200e-01 -5.76508105e-01 2.17783928e+00
4.03304368e-01 6.90780580e-01 -3.14285427e-01 6.03111804e-01
4.08278555e-01 -3.33941430e-01 -6.96563423e-01 -6.87550664e-01
7.12554753e-02 2.59650201e-01 4.76238787e-01 1.74119666e-01
-6.04192257e-01 1.94691136e-01 6.22393131e+00 1.01414609e+00
-7.90782452e-01 1.26557142e-01 5.46220601e-01 -2.83461690e-01
-1.24107867e-01 1.68685704e-01 -4.48347896e-01 5.36961257e-01
1.28153479e+00 1.45555511e-01 1.55986637e-01 5.96210361e-01
5.43160081e-01 -4.53429937e-01 -8.63853455e-01 1.66154727e-01
-3.07054430e-01 -1.08092308e+00 -3.91853869e-01 3.81365150e-01
3.17316234e-01 -1.43858880e-01 -4.00660545e-01 2.52671480e-01
-5.17032966e-02 -6.71491206e-01 1.08148098e+00 6.79526329e-01
5.83795786e-01 -1.21562803e+00 1.24944854e+00 2.61442125e-01
-1.22168970e+00 -3.23889673e-01 -2.99484819e-01 -5.30274928e-01
6.40679359e-01 1.23313022e+00 -2.76125818e-01 1.02080679e+00
9.67393279e-01 4.40760583e-01 -2.35374093e-01 8.94896805e-01
-1.51450172e-01 7.21268356e-01 -6.41363084e-01 1.59660995e-01
-4.57582995e-02 1.55986398e-02 3.76443982e-01 3.27495664e-01
5.33893406e-01 -5.76860130e-01 -4.70525287e-02 8.40521753e-01
3.74925464e-01 -6.02692604e-01 -3.89735132e-01 4.33586426e-02
4.16004270e-01 1.32202315e+00 -8.28729033e-01 2.83231407e-01
-1.64598033e-01 4.78704542e-01 -7.63137564e-02 1.92909166e-01
-1.04657793e+00 -5.69524646e-01 6.70058131e-01 4.94579881e-01
5.37384897e-02 -6.43235922e-01 -4.78916585e-01 -1.92273363e-01
1.89253390e-01 -1.14662215e-01 -4.62204404e-02 -1.09413886e+00
-1.27583170e+00 1.80584118e-01 3.08035105e-01 -1.06136954e+00
7.20556155e-02 -9.23954427e-01 -1.11297464e+00 1.02744794e+00
-1.39142382e+00 -8.67145836e-01 -5.56530319e-02 -1.98882312e-01
6.87785804e-01 -7.07440898e-02 5.11581719e-01 6.57459617e-01
-9.73378062e-01 1.99082211e-01 2.18223780e-01 -4.83428359e-01
1.83132827e-01 -1.03686357e+00 3.43599856e-01 8.10876369e-01
-8.14331174e-01 5.64613223e-01 9.15459275e-01 -1.06652153e+00
-1.66686916e+00 -7.96984196e-01 6.91241145e-01 -2.19758630e-01
8.74916017e-01 -2.28298325e-02 -1.13475025e+00 -6.81437626e-02
2.52684593e-01 -7.20375896e-01 3.35108727e-01 2.09221467e-01
6.02107763e-01 5.61620817e-02 -1.11307108e+00 3.10834646e-01
6.93516254e-01 -2.48067766e-01 -4.99112427e-01 2.83615172e-01
6.05983675e-01 -3.27585340e-01 -1.13613510e+00 4.22500432e-01
3.13571960e-01 -6.78754032e-01 9.41463351e-01 -1.25832260e-01
7.67488003e-01 -3.89178514e-01 4.17254746e-01 -1.25903594e+00
-5.83122194e-01 -1.98346257e-01 -6.41300440e-01 1.52046466e+00
1.65968120e-01 -4.08251733e-01 3.78045470e-01 6.74826860e-01
-9.68307436e-01 -1.19472647e+00 -1.24586177e+00 -9.68892753e-01
1.55312091e-01 -5.15167415e-01 8.14645171e-01 3.86719525e-01
-4.65934336e-01 1.01997934e-01 -2.82143772e-01 4.71932054e-01
3.47334206e-01 -3.33255738e-01 -4.77886107e-03 -1.71037459e+00
-2.38999858e-01 -2.09304571e-01 -1.09227009e-01 -3.01310658e-01
-1.84703052e-01 -5.72133735e-02 2.89426625e-01 -1.82826936e+00
-1.87874258e-01 -5.93525290e-01 -5.07723391e-01 4.59117234e-01
6.73418790e-02 -2.95529723e-01 -1.94848403e-01 1.08258106e-01
5.87126076e-01 8.07529569e-01 1.15596533e+00 -2.27095813e-01
1.48855165e-01 3.08997347e-03 -4.51029688e-01 6.11627102e-01
8.86706650e-01 -4.96720165e-01 -4.33674544e-01 -7.31806636e-01
7.21102536e-01 1.67381719e-01 6.75507665e-01 -1.31170487e+00
3.75474058e-02 -1.30955186e-02 3.04785609e-01 -1.15284252e+00
3.20489049e-01 -1.18428218e+00 6.98067307e-01 7.16343045e-01
6.32738113e-01 1.22131146e-01 7.82849729e-01 5.17719865e-01
-2.29232714e-01 -5.96751213e-01 4.96443659e-01 2.08638266e-01
-2.77044266e-01 1.04838714e-01 -6.03975415e-01 -7.16725171e-01
8.59923244e-01 -5.00146747e-01 -3.81468564e-01 -6.17621886e-03
-7.94158995e-01 3.13655019e-01 3.88931751e-01 7.19781280e-01
5.63486755e-01 -1.07242906e+00 -3.22880834e-01 8.42706487e-02
-4.51846242e-01 2.39919245e-01 9.65481997e-01 6.59565687e-01
-6.52530611e-01 2.02350080e-01 -2.24593475e-01 -7.27557912e-02
-7.21024752e-01 6.10512078e-01 2.66402900e-01 -3.43744606e-01
-5.37452638e-01 9.03371871e-01 -1.45820484e-01 -1.27301002e-02
-2.01799363e-01 -6.39717460e-01 2.83041764e-02 2.89016575e-01
4.71057221e-02 8.88908267e-01 7.38301039e-01 -3.52603406e-01
-4.00461584e-01 5.89392722e-01 -3.78289893e-02 5.51243603e-01
1.40902805e+00 4.47276747e-03 -1.69609189e-01 4.78197932e-01
6.71438694e-01 -5.08683860e-01 -1.27280855e+00 3.45643252e-01
-3.10960680e-01 1.58827931e-01 7.74277747e-01 -1.05134356e+00
-1.39837563e+00 9.66853738e-01 5.33482552e-01 4.89597887e-01
1.39089859e+00 -2.84814119e-01 1.20544505e+00 -7.45982006e-02
5.30026317e-01 -1.56159830e+00 -2.03399301e-01 5.26410580e-01
1.07129085e+00 -5.13935447e-01 2.91072369e-01 -2.30003059e-01
1.21263247e-02 1.18150198e+00 8.12126517e-01 1.36010582e-02
1.14276421e+00 6.02490962e-01 -3.63385111e-01 -4.57568020e-01
-7.10220337e-01 3.70885134e-01 -1.23380899e-01 3.63286227e-01
3.44215155e-01 5.90056628e-02 -2.43108988e-01 9.37271655e-01
8.01108256e-02 1.40476599e-01 5.39931357e-01 1.41336060e+00
-5.07969081e-01 -9.45988297e-01 -5.97123146e-01 6.68932617e-01
-2.85052747e-01 8.67076069e-02 6.10016100e-02 6.63810253e-01
4.67126310e-01 1.11747015e+00 1.36343554e-01 -5.75494826e-01
5.42463362e-01 -3.30073610e-02 1.20379627e-02 -3.53020400e-01
-6.84789002e-01 4.27375287e-02 1.60152450e-01 -2.56402165e-01
-5.73950969e-02 -4.70094472e-01 -1.51697886e+00 -4.64458227e-01
-1.04647839e+00 2.76969522e-02 9.64642465e-01 1.08200336e+00
3.75823021e-01 1.57749772e+00 6.04276538e-01 -1.15095758e+00
-4.02667761e-01 -1.24049914e+00 -1.05817342e+00 1.98121667e-02
6.88501373e-02 -1.50341094e+00 -6.04730368e-01 -1.28046572e-01]
|
[6.757068157196045, 2.493236780166626]
|
8684aec0-6fc9-4c28-a995-298659a32522
|
hierarchical-stochastic-neighbor-embedding-as
|
1910.02696
| null |
https://arxiv.org/abs/1910.02696v1
|
https://arxiv.org/pdf/1910.02696v1.pdf
|
Hierarchical stochastic neighbor embedding as a tool for visualizing the encoding capability of magnetic resonance fingerprinting dictionaries
|
In Magnetic Resonance Fingerprinting (MRF) the quality of the estimated parameter maps depends on the encoding capability of the variable flip angle train. In this work we show how the dimensionality reduction technique Hierarchical Stochastic Neighbor Embedding (HSNE) can be used to obtain insight into the encoding capability of different MRF sequences. Embedding high-dimensional MRF dictionaries into a lower-dimensional space and visualizing them with colors, being a surrogate for location in low-dimensional space, provides a comprehensive overview of particular dictionaries and, in addition, enables comparison of different sequences. Dictionaries for various sequences and sequence lengths were compared to each other, and the effect of transmit field variations on the encoding capability was assessed. Clear differences in encoding capability were observed between different sequences, and HSNE results accurately reflect those obtained from an MRF matching simulation.
|
['Peter Börnert', 'Kirsten Koolstra', 'Oleh Dzyubachyk', 'Boudewijn Lelieveldt', 'Andrew Webb']
|
2019-10-07
| null | null | null | null |
['magnetic-resonance-fingerprinting']
|
['medical']
|
[ 3.00214440e-01 -3.83010685e-01 -1.52869001e-01 -3.76155138e-01
-6.13444209e-01 -6.64779663e-01 3.84194106e-01 1.27949804e-01
-4.35450703e-01 6.53059721e-01 6.21321738e-01 -5.97692840e-02
-5.84698856e-01 -4.40335870e-01 -3.57474416e-01 -1.08254111e+00
-6.31325662e-01 4.18660700e-01 1.31706327e-01 -2.06797004e-01
2.71186829e-01 9.01120484e-01 -1.22685945e+00 3.15176338e-01
3.39458048e-01 5.43747306e-01 3.79481256e-01 8.67876709e-01
1.66432261e-01 4.54297692e-01 -4.79090542e-01 1.06377423e-01
-6.90856352e-02 -3.39838654e-01 -6.12776279e-01 -5.33340752e-01
4.30721551e-01 -2.36597508e-01 -9.74209547e-01 9.91140544e-01
9.33916450e-01 3.24607901e-02 1.01176465e+00 -7.66402125e-01
-4.57629889e-01 5.03985763e-01 -7.38332421e-02 6.95666969e-01
4.07820642e-01 -7.70227909e-02 6.65417969e-01 -6.95877373e-01
9.86868739e-01 9.00250018e-01 8.03938925e-01 4.94497836e-01
-1.64532220e+00 -4.16554809e-01 -6.31404877e-01 3.16420823e-01
-1.37856770e+00 -3.61940086e-01 8.50507379e-01 -8.04109275e-01
6.36080325e-01 4.44956630e-01 6.40912473e-01 9.05604959e-01
7.07179666e-01 2.88186938e-01 1.33309829e+00 -4.80169654e-01
5.40437624e-02 -1.95550751e-02 4.07436907e-01 6.48154736e-01
1.39381945e-01 4.35557961e-01 -4.71652806e-01 -4.87402231e-01
7.28421509e-01 -1.62004501e-01 -5.94129384e-01 -8.20264161e-01
-1.77034867e+00 8.67293239e-01 2.91233301e-01 7.78091431e-01
-1.64411187e-01 2.17300773e-01 4.18337673e-01 3.11694145e-01
-7.16253147e-02 6.31849647e-01 9.72663015e-02 -1.65583506e-01
-9.82546329e-01 8.18952844e-02 4.56035167e-01 3.01871449e-01
5.30030727e-01 1.97825298e-01 -2.12352872e-01 7.37939119e-01
5.78607954e-02 6.10384285e-01 6.21732771e-01 -1.07022345e+00
1.30044907e-01 -2.35996336e-01 4.17832211e-02 -1.26339436e+00
-9.00169015e-01 -2.85660595e-01 -6.63932502e-01 1.44654229e-01
4.53123719e-01 1.71873830e-02 -5.60147345e-01 1.60847676e+00
6.13843314e-02 9.74082798e-02 -2.57343557e-02 1.21965921e+00
6.33260727e-01 4.64260161e-01 -1.88117862e-01 -7.06461200e-04
1.33055055e+00 -3.26502800e-01 -8.23529899e-01 2.87699968e-01
6.67312145e-01 -5.93326926e-01 8.13095212e-01 1.18690401e-01
-7.42064953e-01 -4.47020680e-01 -1.45641506e+00 1.20453857e-01
-1.37437522e-01 -1.84662133e-01 5.37652135e-01 9.57738340e-01
-1.02194059e+00 9.51998591e-01 -9.29219902e-01 4.76923771e-02
5.11487620e-03 3.85427058e-01 -6.28808975e-01 -1.71031922e-01
-1.69676840e+00 9.59726632e-01 2.99491405e-01 -6.84509054e-02
-9.10587847e-01 -9.52105582e-01 -8.07036459e-01 -1.71385795e-01
-4.36822444e-01 -2.65234649e-01 5.90549052e-01 -7.97744468e-02
-1.13551164e+00 7.85738289e-01 7.03264624e-02 -5.37125111e-01
2.34543532e-01 3.92215908e-01 -9.59865212e-01 5.23803830e-01
-9.93872806e-02 4.67958987e-01 8.57520163e-01 -9.56511796e-01
2.44083703e-01 -3.87247086e-01 -1.07174873e-01 3.09021864e-02
-4.06462289e-02 -1.55376762e-01 1.31060451e-01 -7.90974379e-01
2.96450824e-01 -9.62968469e-01 -1.56864673e-01 -5.50105385e-02
-8.52212384e-02 5.64096808e-01 5.61930597e-01 -7.12567747e-01
1.31299973e+00 -2.45663595e+00 3.20231557e-01 4.87181246e-01
5.73910892e-01 -1.54426619e-01 -9.70877036e-02 5.48791230e-01
-5.11128783e-01 -1.97198316e-01 -3.03887516e-01 4.60670292e-01
-3.08931582e-02 1.26007900e-01 -3.85190360e-02 9.81028080e-01
-2.41597131e-01 8.13339770e-01 -1.06318843e+00 -4.39386994e-01
3.87343466e-01 8.39525580e-01 -3.98724794e-01 -1.77220583e-01
4.83263701e-01 6.32942200e-01 -2.90257961e-01 1.40601486e-01
6.11455083e-01 -1.75051004e-01 5.76325834e-01 -6.59097910e-01
1.17753912e-03 -1.81920882e-02 -9.51789141e-01 1.66866577e+00
-3.25187296e-01 1.11339438e+00 -2.01648489e-01 -9.34381008e-01
8.07191849e-01 2.74419308e-01 7.94143140e-01 -9.88368809e-01
-2.58116573e-01 1.69753194e-01 3.90710652e-01 -4.97607827e-01
4.61921006e-01 -3.27348650e-01 -6.66742697e-02 7.49252319e-01
2.24393159e-01 1.81077763e-01 5.93170151e-02 2.41206437e-01
9.31166351e-01 -3.69845331e-01 -7.84809589e-02 -7.20179915e-01
4.46134478e-01 -4.01739568e-01 7.79280290e-02 9.32342350e-01
-3.59524429e-01 7.93646157e-01 6.40993416e-01 -6.29337847e-01
-1.38676834e+00 -1.14893210e+00 -8.77867281e-01 5.87692440e-01
2.30751768e-01 -4.31725591e-01 -5.69463611e-01 -2.57835388e-01
2.08138987e-01 4.37217891e-01 -9.00867820e-01 -3.10301840e-01
-7.82766223e-01 -1.01388037e+00 6.14440203e-01 2.88383693e-01
-1.00348629e-01 -6.87486827e-01 -7.07298756e-01 3.11500311e-01
-1.76609814e-01 -1.06158423e+00 -5.01543939e-01 3.81136835e-01
-7.96360552e-01 -8.50109279e-01 -1.09303725e+00 -5.26360035e-01
3.62232059e-01 -7.84246475e-02 9.73236561e-01 -1.78641766e-01
-5.70951104e-01 5.40943980e-01 -1.17621459e-01 3.30429018e-01
-6.27982497e-01 1.74323171e-02 3.94891977e-01 -1.59171283e-01
-8.33479315e-03 -6.19849503e-01 -9.10282671e-01 4.01817381e-01
-8.81047547e-01 -3.05839688e-01 3.14015538e-01 9.36937988e-01
6.08887732e-01 -2.58876622e-01 2.62784421e-01 -7.75342286e-01
6.58898115e-01 -3.18041027e-01 -3.32821935e-01 1.95238322e-01
-6.50282204e-01 6.38268232e-01 4.19856936e-01 -3.29613954e-01
-3.87037039e-01 -2.79833049e-01 -1.12717584e-01 -3.54115605e-01
9.70030278e-02 2.50254720e-01 2.15042159e-01 -4.59888190e-01
8.44162524e-01 4.83831495e-01 2.04421446e-01 -5.13798296e-01
4.45780069e-01 5.63196898e-01 5.20744979e-01 -4.49047178e-01
2.80884862e-01 4.51404929e-01 1.91157117e-01 -9.83270764e-01
4.14508162e-03 -1.23463571e-01 -6.71352386e-01 -4.16070133e-01
8.31085205e-01 -3.42541337e-01 -7.25400329e-01 1.89754650e-01
-6.01180911e-01 -2.58593380e-01 -1.51071697e-01 8.49010766e-01
-7.13419259e-01 6.20442390e-01 -6.82432175e-01 -3.88015836e-01
-6.53848425e-02 -1.49794626e+00 7.76717246e-01 -1.37626901e-01
-2.22975656e-01 -1.25241315e+00 3.85499507e-01 1.07628785e-01
4.87134755e-01 3.04493994e-01 1.46609890e+00 -2.72085607e-01
-2.56597698e-01 -4.07571755e-02 5.64790443e-02 -1.19522266e-01
5.53085506e-02 -5.26890278e-01 -8.43002856e-01 -5.48538387e-01
-3.01898527e-03 1.52807117e-01 7.35455811e-01 5.87356746e-01
9.80028808e-01 -3.95177081e-02 -4.55605447e-01 6.73967659e-01
1.40151119e+00 1.53269932e-01 7.45531440e-01 4.17615801e-01
6.74908698e-01 2.91710377e-01 1.98795050e-01 1.96991369e-01
-9.64356363e-02 1.09005225e+00 -2.56239057e-01 8.05898104e-04
-3.44543397e-01 -1.51975200e-01 1.89467520e-01 9.99883354e-01
1.86152235e-01 2.10181549e-01 -1.01372468e+00 2.62027949e-01
-1.17060089e+00 -9.98196721e-01 -8.89669135e-02 2.30078459e+00
6.84128404e-01 -1.10763006e-01 1.71724200e-01 3.09984982e-01
7.95124173e-01 5.19828320e-01 -5.00930488e-01 -2.26994067e-01
-3.65673959e-01 1.46919176e-01 7.11194396e-01 7.30193317e-01
-8.06875885e-01 1.58141494e-01 8.15354443e+00 7.29429007e-01
-1.48952997e+00 1.68444529e-01 3.56912851e-01 1.24083669e-03
-6.17820621e-01 -2.53670603e-01 -1.20850220e-01 7.49805927e-01
1.22626555e+00 -8.76115412e-02 6.35862529e-01 2.10182801e-01
-2.23214887e-02 1.01575419e-01 -9.20550823e-01 1.36890161e+00
-9.29523855e-02 -1.64716136e+00 -4.81704026e-02 4.40295078e-02
4.41033304e-01 1.70675352e-01 3.18041623e-01 -1.98483258e-01
-2.78997391e-01 -1.13332915e+00 4.10108387e-01 8.41817737e-01
1.26917815e+00 -7.04769731e-01 5.86960852e-01 -2.11992577e-01
-8.74957323e-01 1.43719852e-01 -2.87186950e-01 4.71777141e-01
3.55665237e-01 6.41824722e-01 -6.40156984e-01 3.61356944e-01
5.64909339e-01 3.62488687e-01 -4.01059777e-01 7.48874128e-01
3.27543855e-01 4.30609077e-01 -2.39928320e-01 5.94647564e-02
1.17471263e-01 -2.79502928e-01 7.29141891e-01 1.30253589e+00
2.44656634e-02 1.17503583e-01 -4.16176200e-01 5.58403790e-01
3.80090833e-01 -1.18593797e-01 -4.72831011e-01 -3.62653673e-01
3.73528123e-01 9.91312683e-01 -7.86394835e-01 -4.05460522e-02
-2.60626972e-01 9.28319216e-01 7.14707151e-02 4.45178896e-01
-6.95193708e-01 -7.19970226e-01 8.15816402e-01 2.88013458e-01
3.22655678e-01 -5.68032086e-01 1.05262503e-01 -1.05558991e+00
-2.38905132e-01 -7.76986539e-01 2.13740557e-01 -7.18104959e-01
-9.20933485e-01 5.81024706e-01 2.68607289e-02 -1.06566203e+00
-3.40377033e-01 -5.89238048e-01 1.09373763e-01 8.61129761e-01
-1.05033910e+00 -1.91938907e-01 2.71264259e-02 4.59957480e-01
-2.57753104e-01 -1.19659126e-01 1.07842052e+00 5.78605115e-01
-1.43713638e-01 7.58506775e-01 8.26358795e-01 1.80839062e-01
4.53430712e-01 -1.23427904e+00 3.02407622e-01 2.20762432e-01
4.37017888e-01 7.54429877e-01 9.32625711e-01 -3.68353695e-01
-1.58559871e+00 -5.02266347e-01 3.98778647e-01 -4.26812768e-01
6.06768310e-01 -4.76772130e-01 -1.07441676e+00 2.56010681e-01
-1.92336246e-01 2.23135382e-01 9.35434818e-01 -2.73491770e-01
-2.01436594e-01 6.93236962e-02 -1.26583111e+00 4.17162031e-01
8.18316519e-01 -1.16350770e+00 -3.67043644e-01 2.18925565e-01
4.38931555e-01 -5.07808149e-01 -1.51301062e+00 1.79798141e-01
1.06115067e+00 -1.04617476e+00 1.20977020e+00 -7.20482528e-01
-1.23368613e-02 -3.90593469e-01 -4.31094527e-01 -1.18864179e+00
-6.93146586e-01 -3.07597965e-01 -7.63374418e-02 3.97939056e-01
2.02216893e-01 -6.87057316e-01 7.56747484e-01 4.03174907e-01
3.15366060e-01 -7.36304045e-01 -1.14187908e+00 -6.79781437e-01
9.75575224e-02 -2.95256257e-01 5.88684559e-01 1.11669290e+00
8.78760442e-02 4.18492779e-02 -3.38554382e-01 1.79777652e-01
8.47185433e-01 1.54895335e-01 -1.36459172e-01 -8.34409237e-01
-3.17616552e-01 -2.46919930e-01 -1.11095715e+00 -7.36853063e-01
1.08337484e-01 -1.28267026e+00 -5.43389201e-01 -9.68003511e-01
1.88515618e-01 -7.44384944e-01 -6.48010314e-01 -9.74444225e-02
3.09807491e-02 4.49335366e-01 2.71570444e-01 3.88439596e-01
-3.72469723e-02 3.41953903e-01 1.10046971e+00 -2.49614894e-01
1.40860483e-01 -5.03634930e-01 -2.13188648e-01 1.07128844e-01
6.07305288e-01 -6.79313660e-01 -2.29775116e-01 -3.59226912e-01
2.00067073e-01 5.69520056e-01 1.56400040e-01 -1.20720840e+00
-8.04078951e-02 4.13533270e-01 6.46118641e-01 -3.18633974e-01
2.20568731e-01 -6.50633514e-01 6.03476644e-01 6.57253981e-01
-5.38200855e-01 1.48723692e-01 1.69238061e-01 4.34029818e-01
-1.66694105e-01 -4.04683441e-01 7.81467140e-01 1.23660430e-01
-7.63808310e-01 8.49635825e-02 -5.90670705e-01 6.59434572e-02
5.57733536e-01 -3.78993034e-01 1.22111812e-01 -3.52390021e-01
-1.15728700e+00 -3.23389828e-01 5.55186152e-01 2.15282395e-01
5.39333582e-01 -1.70375335e+00 -4.58079159e-01 4.43340659e-01
7.78595582e-02 -1.10164702e+00 7.26936996e-01 9.83652472e-01
-7.69713283e-01 5.91262102e-01 -5.49637914e-01 -7.49623418e-01
-1.00438690e+00 6.20695829e-01 7.77802587e-01 -1.45039529e-01
-6.72568798e-01 4.65328038e-01 6.33922638e-03 -4.22865808e-01
-2.62194693e-01 6.91778734e-02 -2.39677876e-01 5.33462130e-02
5.92646480e-01 5.22537112e-01 1.83816507e-01 -7.90384948e-01
-6.22684777e-01 9.29443538e-01 -8.91894624e-02 -3.97523522e-01
1.10447931e+00 -1.87476948e-01 1.40642136e-01 5.97608209e-01
1.70083904e+00 2.88631581e-02 -1.08741236e+00 -8.41428116e-02
-5.36718592e-02 -3.80770743e-01 2.96154648e-01 -5.89459121e-01
-1.02372408e+00 9.10859048e-01 1.45398641e+00 -5.80401309e-02
7.19499707e-01 -8.17171931e-02 6.15501523e-01 7.36403242e-02
6.43617511e-01 -6.19823575e-01 -2.25507870e-01 1.84076697e-01
6.77526951e-01 -7.69718528e-01 -8.28431174e-03 -4.04090956e-02
-4.70698714e-01 1.43874204e+00 -2.15452403e-01 5.59775382e-02
6.00796580e-01 3.55286360e-01 1.65345788e-01 -5.00453293e-01
-4.10154089e-02 4.30078685e-01 2.96481937e-01 8.33359838e-01
5.08425295e-01 2.35754788e-01 -3.48727465e-01 1.66555762e-01
-2.63755649e-01 -1.59737706e-01 4.63622153e-01 7.09288001e-01
-9.38890502e-02 -1.04872227e+00 -3.61664832e-01 5.29607773e-01
-3.83935213e-01 -1.02338202e-01 7.59376138e-02 6.15351379e-01
-1.74483418e-01 2.40777373e-01 8.67837369e-02 -5.09361863e-01
2.56211787e-01 8.76959264e-02 9.47446525e-01 -2.73054335e-02
-1.59707025e-01 -3.68723154e-01 -8.40693247e-03 -7.73697615e-01
-2.00968102e-01 -6.85907066e-01 -1.29170537e+00 -3.11948925e-01
-1.22263886e-01 4.04032171e-01 6.99137628e-01 5.01925409e-01
3.44357163e-01 3.22966099e-01 5.62892795e-01 -7.79076874e-01
-3.48627478e-01 -5.57100952e-01 -1.08685708e+00 4.70544457e-01
7.10446775e-01 -8.52595985e-01 -3.56959999e-01 -3.38808775e-01]
|
[13.491772651672363, -2.3802764415740967]
|
130846e8-3ba0-466c-9cd5-8ab765a1ed64
|
biometric-signature-verification-using
|
2205.02934
| null |
https://arxiv.org/abs/2205.02934v1
|
https://arxiv.org/pdf/2205.02934v1.pdf
|
Biometric Signature Verification Using Recurrent Neural Networks
|
Architectures based on Recurrent Neural Networks (RNNs) have been successfully applied to many different tasks such as speech or handwriting recognition with state-of-the-art results. The main contribution of this work is to analyse the feasibility of RNNs for on-line signature verification in real practical scenarios. We have considered a system based on Long Short-Term Memory (LSTM) with a Siamese architecture whose goal is to learn a similarity metric from pairs of signatures. For the experimental work, the BiosecurID database comprised of 400 users and 4 separated acquisition sessions are considered. Our proposed LSTM RNN system has outperformed the results of recent published works on the BiosecurID benchmark in figures ranging from 17.76% to 28.00% relative verification performance improvement for skilled forgeries.
|
['Javier Ortega-Garcia', 'Julian Fierrez', 'Ruben Vera-Rodriguez', 'Ruben Tolosana']
|
2022-05-03
| null | null | null | null |
['handwriting-recognition']
|
['computer-vision']
|
[ 6.13596797e-01 -3.89848202e-01 3.30728203e-01 -4.79395926e-01
-6.12686276e-01 -7.10154176e-02 6.50431037e-01 -1.64988980e-01
-7.92410254e-01 5.52847445e-01 -1.12771302e-01 -4.58184719e-01
-3.37107062e-01 -2.56724745e-01 -5.19716144e-01 -8.01219523e-01
-2.88830996e-01 3.91623288e-01 -1.74619332e-01 -5.15229225e-01
5.92470348e-01 8.60658228e-01 -1.23346817e+00 4.57750350e-01
3.05414677e-01 9.39287961e-01 -2.41039708e-01 1.00637209e+00
3.59167248e-01 7.68214166e-01 -1.03954816e+00 -4.14777637e-01
2.87120342e-01 -3.42157304e-01 -6.97467029e-01 -4.48090434e-01
4.50522304e-01 -1.92226678e-01 -7.03458548e-01 8.79252195e-01
1.00519466e+00 2.95149267e-01 3.69863570e-01 -8.28960359e-01
-5.75706720e-01 9.29678738e-01 -3.80064934e-01 3.90549898e-01
1.56123012e-01 -1.41115203e-01 6.75115108e-01 -7.92228758e-01
3.83202255e-01 7.99781203e-01 9.47264075e-01 6.00207627e-01
-7.51702905e-01 -7.26988256e-01 -4.63547677e-01 7.00171828e-01
-1.46981514e+00 -5.12427151e-01 4.37130958e-01 -1.56854056e-02
1.29253447e+00 1.96859837e-01 -8.74912962e-02 1.66028643e+00
4.06579047e-01 7.85025001e-01 9.83279169e-01 -4.72273022e-01
-1.89050347e-01 1.36063322e-01 6.02818906e-01 7.09010363e-01
1.09996624e-01 4.13115680e-01 -7.38593102e-01 -6.48487434e-02
4.61353749e-01 -9.04122069e-02 -6.37439936e-02 1.93290129e-01
-1.29009092e+00 5.22043586e-01 2.20151260e-01 1.10439575e+00
-4.91467655e-01 3.05729061e-01 1.06506515e+00 8.07949245e-01
7.19715357e-02 5.39042413e-01 -1.24878764e-01 -5.55626452e-01
-1.36301267e+00 7.30890036e-02 7.83526778e-01 6.86007679e-01
-3.38953465e-01 6.45895183e-01 -3.03115457e-01 9.28087533e-01
1.52119845e-01 4.52610314e-01 1.11946321e+00 -1.55446947e-01
5.32402575e-01 1.06197111e-01 9.49942768e-02 -1.16540337e+00
-4.86311853e-01 -8.07228208e-01 -1.12109220e+00 6.06414815e-03
3.74622285e-01 -2.68586189e-01 -9.70562279e-01 1.23507714e+00
-4.11460310e-01 5.13339102e-01 2.98105597e-01 9.64585543e-01
7.80655742e-01 5.18335819e-01 -3.44322979e-01 3.22795063e-01
1.09673929e+00 -6.81657851e-01 -7.35281348e-01 2.23712623e-01
5.42575896e-01 -8.82983029e-01 3.89938354e-01 7.30903864e-01
-7.47138441e-01 -6.56971812e-01 -1.43370497e+00 2.37445071e-01
-6.27339602e-01 4.27609682e-01 -2.46269200e-02 1.29792774e+00
-1.19628882e+00 1.16650271e+00 -6.94382668e-01 -4.74757612e-01
7.06084967e-02 6.43339217e-01 -2.49596342e-01 3.71686071e-01
-1.57170594e+00 1.13590550e+00 4.62109089e-01 9.15474534e-01
-1.10200417e+00 -2.90481120e-01 -6.01131260e-01 -2.34689023e-02
3.39505784e-02 -1.18609458e-01 1.17296660e+00 -6.83735430e-01
-1.71573579e+00 8.49489868e-01 1.72217309e-01 -1.20789802e+00
6.95634484e-01 -2.25860834e-01 -8.01393747e-01 -3.64464998e-01
-6.31229222e-01 -1.87620342e-01 7.36852586e-01 -6.39937997e-01
-3.30119461e-01 -3.28461856e-01 -6.87229216e-01 -2.45294049e-01
-4.59598333e-01 5.09480119e-01 7.06243813e-02 -8.61644506e-01
-1.49933606e-01 -1.07131445e+00 5.35137244e-02 -7.31574416e-01
-5.14594078e-01 -1.08468860e-01 1.06111288e+00 -1.09641838e+00
1.00986326e+00 -1.97528183e+00 7.81287998e-02 5.84308922e-01
-2.93421417e-01 9.92874146e-01 -2.62354642e-01 4.85053092e-01
-2.91355789e-01 -2.43625432e-01 -8.02454427e-02 -5.62947750e-01
1.89248204e-01 -1.21576898e-01 -6.36252046e-01 8.74862015e-01
-4.48354557e-02 7.78388917e-01 -6.34631038e-01 5.21227671e-03
1.17985606e-01 6.89403832e-01 5.56522191e-01 2.03961641e-01
3.19394708e-01 -3.04701310e-02 -1.30421445e-01 6.50767028e-01
4.35007811e-01 1.03499092e-01 2.82291710e-01 2.22911425e-02
1.02059796e-01 1.30257942e-02 -9.83536482e-01 1.66624498e+00
-4.51038927e-01 1.21888566e+00 -2.19663948e-01 -1.10105467e+00
1.27555954e+00 5.92699945e-01 2.96082109e-01 -7.17165291e-01
4.82322276e-01 4.51023459e-01 3.70156527e-01 -2.58249313e-01
7.75147021e-01 -2.63215691e-01 3.72961350e-02 8.69355559e-01
2.25115672e-01 5.38290858e-01 -2.75463998e-01 -1.27029851e-01
1.16743851e+00 5.62378839e-02 -2.13157728e-01 -8.19858816e-03
8.99401963e-01 -6.05574369e-01 1.82587728e-01 1.05193353e+00
-2.62108237e-01 4.66277152e-01 1.33556217e-01 -1.72909111e-01
-1.15143418e+00 -6.25565112e-01 2.37554833e-02 1.04356003e+00
-5.13420999e-01 8.53473619e-02 -6.49959564e-01 -3.10821027e-01
4.65447269e-02 7.12009549e-01 -6.80418730e-01 -5.85723408e-02
-8.54542315e-01 -7.07404256e-01 1.82074237e+00 6.07568562e-01
6.28935218e-01 -1.29348457e+00 -7.92987227e-01 4.02157307e-01
3.04458886e-01 -1.10515416e+00 -2.76869327e-01 7.75239319e-02
-6.66266680e-01 -7.92025745e-01 -1.16877532e+00 -5.64957619e-01
2.75386691e-01 -1.64512545e-01 4.60295022e-01 1.94215160e-02
-3.82027656e-01 1.36940209e-02 -3.75913829e-01 -5.02909362e-01
-5.55023253e-01 3.81869793e-01 4.51074451e-01 3.13065648e-01
3.69194180e-01 -2.93951750e-01 -1.19920850e-01 4.83748794e-01
-7.94242442e-01 -6.93077266e-01 8.62689972e-01 1.32179940e+00
-1.60454631e-01 -2.88832217e-01 7.30941474e-01 -8.65596890e-01
1.04936039e+00 -1.77596450e-01 -6.37044251e-01 5.89266062e-01
-8.52288604e-01 2.84533501e-01 7.65459657e-01 -9.92408991e-02
-1.17724264e+00 -3.14813316e-01 -1.27430335e-01 -6.86219990e-01
9.75794941e-02 4.73851323e-01 5.10315776e-01 -5.19337058e-01
6.94583297e-01 7.10071981e-01 1.15498014e-01 -4.15683717e-01
-2.08039641e-01 1.12373173e+00 6.23965204e-01 -3.11715722e-01
6.01834297e-01 6.33879006e-02 -1.68827813e-04 -8.93711329e-01
-1.82729617e-01 -5.75874805e-01 -5.74064016e-01 -1.82143882e-01
3.12812686e-01 -2.77728945e-01 -1.18154645e+00 1.21475363e+00
-1.13216555e+00 -1.86113745e-01 2.52753109e-01 4.65167195e-01
-3.39584708e-01 7.62289405e-01 -1.04621327e+00 -1.15233815e+00
-1.04716504e+00 -9.62386966e-01 7.93430507e-01 1.30399108e-01
-1.58781990e-01 -1.02351320e+00 -6.22188905e-03 3.08864027e-01
8.40645552e-01 3.02725911e-01 3.66185784e-01 -1.23367035e+00
-1.16048411e-01 -7.57412255e-01 -7.45740831e-02 6.55863464e-01
4.55782674e-02 -1.85616817e-02 -1.27018309e+00 -5.64556897e-01
1.57606184e-01 -3.08137774e-01 9.26057935e-01 -1.62453707e-02
1.01883221e+00 -3.05555891e-02 -1.21531725e-01 3.91256154e-01
1.07073486e+00 3.31800252e-01 7.19401300e-01 3.30488801e-01
4.79181707e-01 4.99129504e-01 5.44717193e-01 3.20975184e-01
-5.27465463e-01 6.57962739e-01 -1.84722781e-01 1.39564335e-01
2.26494536e-01 1.49796993e-01 6.99116349e-01 8.26352596e-01
-2.28294402e-01 -2.80090630e-01 -1.15905452e+00 4.79830086e-01
-1.93034923e+00 -1.17977130e+00 7.94883072e-02 2.29584885e+00
3.93787920e-01 1.99247748e-01 2.74713691e-02 5.49136341e-01
8.21180046e-01 2.20650434e-01 -3.94351482e-01 -9.25801039e-01
-2.99286935e-02 5.16092718e-01 8.68391752e-01 1.91732600e-01
-9.45465624e-01 6.64260030e-01 6.28411436e+00 8.79122078e-01
-1.43071401e+00 1.07508361e-01 3.32704693e-01 -1.74959555e-01
4.85432297e-01 -7.49943078e-01 -6.54145420e-01 3.61161351e-01
1.87062585e+00 -5.28127216e-02 4.83712628e-02 4.97957736e-01
2.26090863e-01 1.79868445e-01 -1.02548051e+00 1.06211853e+00
5.87578237e-01 -1.18885481e+00 -1.91994846e-01 -6.82761073e-02
4.07004565e-01 3.34178895e-01 4.75695908e-01 2.36059204e-01
-1.85099971e-02 -1.57205498e+00 3.52389783e-01 6.95901036e-01
8.33875000e-01 -8.96171153e-01 1.43194306e+00 2.30820343e-01
-7.34240949e-01 -8.65554437e-02 -2.90934652e-01 2.58129835e-01
9.76468846e-02 2.75618911e-01 -1.54060376e+00 8.67448390e-01
3.48235697e-01 5.85353673e-01 -4.08269435e-01 1.03003764e+00
5.01687936e-02 6.83183610e-01 -6.67224005e-02 -5.28282106e-01
6.60845876e-01 -7.63124004e-02 5.25256097e-01 1.70657158e+00
2.66301453e-01 -2.33857170e-01 -7.03064322e-01 5.70071101e-01
-2.42958795e-02 -1.31831586e-01 -5.48910737e-01 -4.48457509e-01
-1.05865449e-02 9.47658300e-01 -4.05855268e-01 -4.17134225e-01
1.69041932e-01 1.09643853e+00 -1.24914274e-01 1.91262826e-01
-8.31840932e-01 -1.16137016e+00 4.10434246e-01 -3.87471020e-01
3.72522771e-01 -2.06573486e-01 -1.72452495e-01 -9.02240574e-01
1.14740200e-01 -1.11308932e+00 5.09654164e-01 -3.98723453e-01
-1.25014675e+00 9.22902048e-01 -4.02722180e-01 -1.08033025e+00
-5.83926737e-01 -9.33432102e-01 -6.14696622e-01 1.17210984e+00
-1.17794847e+00 -1.10326004e+00 -5.50275184e-02 4.30368870e-01
4.08595920e-01 -8.74987245e-01 9.10766602e-01 5.20999074e-01
-7.89769471e-01 1.29074204e+00 5.95423341e-01 6.09103322e-01
7.00945914e-01 -9.02057827e-01 9.24214661e-01 9.37767804e-01
2.52219409e-01 1.09751034e+00 8.09934855e-01 -5.16228676e-01
-1.46576691e+00 -9.15529013e-01 1.03340387e+00 -2.89169788e-01
7.10791111e-01 -3.39185059e-01 -8.38810682e-01 6.97933495e-01
2.55391032e-01 -1.47600353e-01 6.72142267e-01 -4.32384089e-02
-3.42637271e-01 -1.82969451e-01 -1.31772089e+00 2.74223387e-01
7.87176073e-01 -9.90410089e-01 -5.36314547e-01 1.75712436e-01
-1.01857074e-01 -5.58716118e-01 -9.71016645e-01 2.36535534e-01
9.16420698e-01 -1.02741623e+00 7.47770965e-01 -8.92549634e-01
-9.49884281e-02 -9.26963836e-02 9.22793671e-02 -1.11449361e+00
2.00805038e-01 -9.28557217e-01 -2.92342622e-02 8.68795156e-01
4.63819563e-01 -7.03382730e-01 9.92846608e-01 4.12943602e-01
-1.95367094e-02 -7.59978056e-01 -1.08450830e+00 -9.87014115e-01
-2.44574860e-01 -3.81062031e-01 2.20508546e-01 8.85137320e-01
5.20015135e-02 -5.69326384e-03 -9.30202603e-01 -7.48645188e-03
6.17740452e-01 -1.11211151e-01 4.37218577e-01 -6.91226304e-01
-4.35506880e-01 -5.05385220e-01 -9.38792646e-01 -6.38803840e-01
3.92097712e-01 -8.03160489e-01 6.72325343e-02 -8.99387956e-01
-3.33207846e-02 -9.79753435e-02 -8.61179233e-01 4.62540776e-01
4.19628114e-01 1.42107502e-01 9.88271534e-02 -2.72042036e-01
-2.43784353e-01 2.66120315e-01 2.65395910e-01 -3.12295884e-01
7.42443949e-02 2.77845174e-01 -1.91212982e-01 8.56363326e-02
8.53598773e-01 -3.25575083e-01 -1.58229247e-02 -3.15721035e-01
-2.75333136e-01 7.31461272e-02 2.13099390e-01 -1.08748174e+00
5.16142309e-01 4.82577890e-01 5.10339200e-01 -5.13254762e-01
4.81189787e-01 -5.32654345e-01 1.30908415e-01 1.00425589e+00
-6.23218298e-01 3.12985629e-01 2.58122116e-01 4.77978885e-01
-3.13098133e-01 -4.70807999e-01 6.29377723e-01 1.02654845e-01
-5.89842796e-01 -6.35620207e-02 -5.82702458e-01 -5.64486623e-01
8.65290999e-01 -4.75257456e-01 -2.56116003e-01 -3.94313931e-01
-4.00191784e-01 -1.62416831e-01 -1.23992182e-01 6.65547073e-01
1.03995824e+00 -1.18165636e+00 -9.70225096e-01 1.51125103e-01
-1.80468351e-01 -9.07083333e-01 2.41205961e-01 8.17160010e-01
-5.99799573e-01 1.17280233e+00 -6.09252393e-01 -2.82717347e-01
-1.91956604e+00 9.76542011e-02 7.92198241e-01 -2.52789706e-01
-3.45950097e-01 9.72506642e-01 -1.03707910e+00 -5.14767289e-01
5.65654337e-01 -2.55404294e-01 -1.65555224e-01 2.20347513e-02
5.56586087e-01 8.21061015e-01 7.00365126e-01 -7.65333235e-01
-4.70257461e-01 2.93092340e-01 -4.46282268e-01 -2.48464093e-01
1.47015643e+00 6.61559880e-01 -2.95666959e-02 4.87228245e-01
1.38636518e+00 -3.58290881e-01 -3.67513955e-01 -3.41016263e-01
5.98198831e-01 -3.70514899e-01 -4.33129929e-02 -8.18899274e-01
-9.89553392e-01 9.75291312e-01 1.01953137e+00 -1.44365385e-01
7.85252035e-01 -9.19980586e-01 1.15328336e+00 1.03198051e+00
5.05346835e-01 -1.26860404e+00 -6.04822449e-02 7.76853085e-01
8.73592794e-01 -9.29482937e-01 -2.13673607e-01 6.23593330e-01
-6.15913451e-01 1.48087132e+00 1.32364452e-01 -1.66657716e-01
3.89052659e-01 1.49006799e-01 1.62248373e-01 -2.10188776e-01
-5.00966549e-01 3.37660998e-01 4.48105037e-01 2.60174245e-01
6.28526092e-01 -7.02211931e-02 -2.51303136e-01 2.55151242e-01
-1.12630405e-01 5.69215156e-02 7.39869475e-01 9.22003388e-01
9.31588039e-02 -1.25026524e+00 -4.46369231e-01 4.97496456e-01
-7.42455781e-01 4.68672113e-03 -6.92665815e-01 4.07315016e-01
-3.96782428e-01 8.64966035e-01 -2.74484247e-01 -8.18861246e-01
3.85654241e-01 1.97405472e-01 4.25983131e-01 -2.23307893e-01
-1.36297703e+00 -4.56390470e-01 2.80978262e-01 -5.23889065e-01
-3.66988093e-01 -7.86083996e-01 -1.14145875e+00 -4.05083477e-01
-3.92515242e-01 1.61471322e-01 9.52872932e-01 8.18233252e-01
4.23400611e-01 5.59493542e-01 4.97871250e-01 -5.01183629e-01
-1.25859499e+00 -1.35626733e+00 -8.13480258e-01 2.31895193e-01
4.36012745e-01 -3.55827481e-01 6.22258894e-03 -3.65912735e-01]
|
[11.983776092529297, 2.502734899520874]
|
532ca085-e9c1-4da8-8fbb-3aac09b6f53e
|
bottlenet-an-end-to-end-approach-for-feature
|
1910.14315
| null |
https://arxiv.org/abs/1910.14315v5
|
https://arxiv.org/pdf/1910.14315v5.pdf
|
BottleNet++: An End-to-End Approach for Feature Compression in Device-Edge Co-Inference Systems
|
The emergence of various intelligent mobile applications demands the deployment of powerful deep learning models at resource-constrained mobile devices. The device-edge co-inference framework provides a promising solution by splitting a neural network at a mobile device and an edge computing server. In order to balance the on-device computation and the communication overhead, the splitting point needs to be carefully picked, while the intermediate feature needs to be compressed before transmission. Existing studies decoupled the design of model splitting, feature compression, and communication, which may lead to excessive resource consumption of the mobile device. In this paper, we introduce an end-to-end architecture, named BottleNet++, that consists of an encoder, a non-trainable channel layer, and a decoder for more efficient feature compression and transmission. The encoder and decoder essentially implement joint source-channel coding via convolutional neural networks (CNNs), while explicitly considering the effect of channel noise. By exploiting the strong sparsity and the fault-tolerant property of the intermediate feature in a deep neural network (DNN), BottleNet++ achieves a much higher compression ratio than existing methods. Furthermore, by providing the channel condition to the encoder as an input, our method enjoys a strong generalization ability in different channel conditions. Compared with merely transmitting intermediate data without feature compression, BottleNet++ achieves up to 64x bandwidth reduction over the additive white Gaussian noise channel and up to 256x bit compression ratio in the binary erasure channel, with less than 2% reduction in accuracy. With a higher compression ratio, BottleNet++ enables splitting a DNN at earlier layers, which leads to up to 3x reduction in on-device computation compared with other compression methods.
|
['Jun Zhang', 'Jiawei Shao']
|
2019-10-31
| null | null | null | null |
['feature-compression']
|
['computer-vision']
|
[ 2.29120910e-01 -4.64582182e-02 -4.58297729e-01 -2.71377228e-02
-5.88553548e-01 4.42514047e-02 3.30284536e-02 -4.49978560e-02
-4.37785506e-01 4.17661518e-01 2.55709440e-02 -4.87768143e-01
5.79098053e-02 -9.84762132e-01 -1.09790826e+00 -6.13846779e-01
-1.97556168e-01 1.14482611e-01 -8.13872665e-02 1.89055681e-01
-2.46321693e-01 -1.21536732e-01 -1.33064544e+00 2.18206823e-01
5.76914370e-01 1.83634984e+00 4.80838060e-01 6.90639138e-01
6.30224124e-02 6.97515845e-01 -1.39282122e-01 -5.94298422e-01
3.73638004e-01 -1.97811827e-01 -4.75630701e-01 6.15290664e-02
-7.27113187e-02 -8.20225120e-01 -8.91640902e-01 1.10404444e+00
7.03281403e-01 -4.19654340e-01 2.23984256e-01 -1.25109398e+00
-4.29357231e-01 8.08820605e-01 -3.10822517e-01 -1.36401191e-01
-1.08777747e-01 6.38873801e-02 9.27718878e-01 -7.02702403e-01
2.95491397e-01 6.85663104e-01 8.90359521e-01 4.11518157e-01
-8.49506974e-01 -7.07844198e-01 -2.22878948e-01 4.07492667e-01
-1.69349778e+00 -6.83392644e-01 7.47100413e-01 1.00990266e-01
1.11782467e+00 8.74937922e-02 7.06986606e-01 9.52494264e-01
2.57882625e-01 8.96219432e-01 3.85402083e-01 -2.14557618e-01
6.17497861e-01 -2.92814169e-02 -2.79666513e-01 7.40825355e-01
2.67218202e-01 -1.80859044e-02 -6.08374596e-01 -1.08324036e-01
7.68828571e-01 5.39969921e-01 -2.59344399e-01 7.92114735e-02
-7.49029040e-01 5.00428736e-01 6.57723904e-01 9.23923478e-02
-6.71488166e-01 7.83141553e-01 7.19631433e-01 3.01098585e-01
2.92097121e-01 -3.21082950e-01 -5.88067234e-01 -5.82897305e-01
-1.05075276e+00 -1.60283834e-01 8.27895284e-01 1.30205655e+00
6.90304518e-01 4.21802513e-02 5.11993654e-02 5.84057868e-01
2.33365804e-01 4.49484497e-01 5.88656962e-01 -8.95986855e-01
9.09467399e-01 3.76875937e-01 -4.10174757e-01 -9.22151864e-01
-2.21303493e-01 -8.32190871e-01 -1.42945611e+00 -3.24025333e-01
-1.96622267e-01 -4.57706511e-01 -6.23470187e-01 1.64451909e+00
4.81281988e-02 4.47417051e-01 1.67948514e-01 7.25449204e-01
2.96362072e-01 6.78463876e-01 -2.48737574e-01 -1.55836031e-01
1.34272242e+00 -1.16239142e+00 -6.35356128e-01 -2.26895571e-01
1.04034364e+00 -4.54950631e-01 7.73372114e-01 3.02160591e-01
-1.32468498e+00 -4.79980260e-01 -1.22559714e+00 -2.54857510e-01
2.26844884e-02 2.29025364e-01 6.49080992e-01 6.59584463e-01
-1.03235233e+00 6.41910434e-01 -9.14350510e-01 2.11123303e-01
7.26223767e-01 7.62822449e-01 -3.32661867e-01 -3.76022428e-01
-9.84096527e-01 1.35835290e-01 3.18776190e-01 8.42799768e-02
-6.41180873e-01 -6.45924091e-01 -8.12345564e-01 6.77776635e-01
2.55800396e-01 -9.68105555e-01 1.28819239e+00 -9.83239114e-01
-1.59274423e+00 1.65305734e-01 -2.39873096e-01 -9.09163296e-01
3.20510089e-01 -5.63188381e-02 -4.11737144e-01 1.56931445e-01
-3.87678504e-01 4.47624445e-01 6.78970098e-01 -6.23221993e-01
-7.09254801e-01 -3.48604888e-01 3.62069309e-02 1.27839670e-01
-7.30035007e-01 -4.85216141e-01 -1.04359579e+00 -6.74350381e-01
8.05566162e-02 -1.03990281e+00 -2.18776688e-01 1.90751508e-01
-2.71293759e-01 2.08950058e-01 1.16038370e+00 -6.32232964e-01
1.47831607e+00 -2.51271033e+00 -6.83796629e-02 1.64722368e-01
5.70661545e-01 2.68567443e-01 1.13320118e-02 1.82258368e-01
5.16067743e-01 3.42430957e-02 -1.39214113e-01 -6.73363805e-01
-5.54619282e-02 2.65004009e-01 1.46866599e-02 2.55230725e-01
-3.26019496e-01 1.07878709e+00 -5.77706158e-01 -2.36877799e-01
-1.92980781e-01 7.64872849e-01 -1.02542901e+00 -6.98555037e-02
-5.56580126e-02 -1.79963887e-01 -3.80691230e-01 7.07034349e-01
7.33724833e-01 -5.65413415e-01 2.55709857e-01 -1.33513421e-01
3.73100162e-01 4.62585807e-01 -9.31296051e-01 1.76559901e+00
-9.96226490e-01 7.22779989e-01 3.84957731e-01 -9.46435690e-01
5.31332195e-01 5.40931642e-01 3.91692042e-01 -1.02285421e+00
4.54753458e-01 5.30576050e-01 -1.45413145e-01 -2.87113369e-01
5.13745070e-01 1.42774403e-01 -9.65781957e-02 3.02285820e-01
-4.58952300e-02 5.08420527e-01 -3.05197567e-01 6.00494854e-02
1.32213950e+00 -4.71217334e-01 -1.28443867e-01 8.57358873e-02
9.90432203e-02 -6.18875921e-01 6.98151588e-01 5.99529684e-01
1.48676679e-01 2.25360170e-01 3.02936554e-01 -3.67500067e-01
-1.18260705e+00 -6.17052317e-01 -9.32863113e-05 7.47829437e-01
3.09050173e-01 -8.01262617e-01 -8.80810380e-01 -4.36048418e-01
-5.94493598e-02 2.51557231e-01 -1.36382341e-01 -5.70576072e-01
-2.85217166e-01 -6.29614294e-01 7.81724870e-01 8.08808506e-01
1.22020376e+00 -3.62397671e-01 -7.59722114e-01 2.98182130e-01
-1.02477230e-01 -1.29975390e+00 -6.38448417e-01 4.99954015e-01
-1.12104607e+00 -4.99090075e-01 -4.74371403e-01 -7.81317770e-01
4.90866601e-01 3.12885344e-01 6.74256325e-01 3.02526265e-01
5.53029887e-02 -9.62756202e-02 -3.71684939e-01 -3.92322503e-02
-3.30817178e-02 3.91208112e-01 -7.14353696e-02 -8.88051279e-03
2.87110299e-01 -8.57828856e-01 -7.80994833e-01 -4.40690480e-03
-9.92779970e-01 3.41553628e-01 8.32075715e-01 9.34279919e-01
6.34336293e-01 6.12945139e-01 2.48851582e-01 -6.92761958e-01
1.98598295e-01 -6.52491570e-01 -3.58665735e-01 -5.62100746e-02
-7.23933518e-01 1.02403291e-01 9.38758790e-01 -3.67817283e-01
-6.38636827e-01 2.67502278e-01 -4.34181690e-01 -5.55526376e-01
3.50066513e-01 5.59198737e-01 -5.55077493e-01 -2.02103019e-01
2.26779923e-01 4.03655648e-01 -1.36945888e-01 -4.00193274e-01
1.36097729e-01 1.15583301e+00 3.61343294e-01 -1.42663941e-01
3.48744065e-01 3.76613647e-01 -1.80839747e-02 -6.45447016e-01
-1.53375939e-01 2.35761963e-02 -1.64365500e-01 1.88053876e-01
3.57488871e-01 -1.35421014e+00 -8.63400221e-01 5.59030890e-01
-9.84134972e-01 -5.30761659e-01 -1.60600841e-01 5.72455227e-01
-3.88545126e-01 1.22383498e-01 -1.03673124e+00 -4.59853828e-01
-6.65602326e-01 -1.30779171e+00 1.09089649e+00 4.83408906e-02
3.08132112e-01 -8.03357005e-01 -7.64783502e-01 2.42969394e-01
7.84581840e-01 -1.92825004e-01 9.40396249e-01 -3.08575332e-01
-7.79401660e-01 -4.71285820e-01 -3.59675646e-01 3.57975334e-01
-9.85200405e-02 -6.55371904e-01 -9.74617481e-01 -3.86977434e-01
1.25488788e-01 -1.04498893e-01 9.28038836e-01 3.93318206e-01
1.67033494e+00 -7.45549142e-01 -4.07691717e-01 1.17967188e+00
1.50007451e+00 3.07495236e-01 6.84529841e-01 -1.01708062e-01
7.35311985e-01 -2.69449413e-01 -1.61386222e-01 6.97713852e-01
5.56748390e-01 6.81052923e-01 5.80220759e-01 -9.05141905e-02
-2.40317047e-01 -2.82526791e-01 3.93264115e-01 1.09593976e+00
1.78136051e-01 -3.72422487e-01 -5.11726379e-01 3.15516353e-01
-1.68723059e+00 -7.09823906e-01 2.14509010e-01 2.37298584e+00
7.96306670e-01 4.03044373e-01 -2.06988722e-01 5.46478212e-01
4.61277366e-01 -1.15540382e-02 -7.59372711e-01 -2.68101722e-01
9.85709131e-02 -2.81462504e-04 8.69928896e-01 3.33806396e-01
-7.91620195e-01 5.55271983e-01 5.33559227e+00 1.03238916e+00
-1.51733279e+00 4.42478359e-01 8.95747066e-01 -3.12000483e-01
-2.72982299e-01 -7.52529800e-02 -7.86867380e-01 9.29255903e-01
1.38359106e+00 3.06820329e-02 6.58095658e-01 1.23782897e+00
1.90596636e-02 7.00055137e-02 -1.04874980e+00 1.44800198e+00
-1.35583922e-01 -1.54958737e+00 -1.73714817e-01 4.30420548e-01
4.47481871e-01 3.12310129e-01 6.24337085e-02 2.00796917e-01
-1.45728111e-01 -8.66764009e-01 6.83938980e-01 2.03899041e-01
1.19683766e+00 -9.66107666e-01 9.14335370e-01 4.66978729e-01
-1.30701113e+00 -4.24309820e-01 -4.31095302e-01 -2.85035729e-01
2.18799874e-01 1.09423935e+00 -5.29453158e-01 4.52650875e-01
7.47533321e-01 5.93741000e-01 -1.40258819e-01 8.91531110e-01
2.58928090e-01 6.53960407e-01 -5.51103115e-01 7.73712546e-02
4.05081995e-02 2.03581944e-01 8.59546736e-02 1.18583822e+00
6.70554817e-01 -1.00532353e-01 1.25996754e-01 4.05803978e-01
-7.76563644e-01 -1.78728998e-01 -4.33635563e-01 1.15467906e-01
8.05494726e-01 9.19051349e-01 -4.54701990e-01 -4.59201187e-01
-5.56139827e-01 1.52341115e+00 1.79441318e-01 2.83074170e-01
-9.43512559e-01 -5.80835879e-01 5.87978423e-01 1.45463243e-01
4.28947926e-01 -3.08143437e-01 -5.57877779e-01 -1.09001374e+00
4.21640545e-01 -5.94650805e-01 -1.86439902e-01 -4.03708667e-01
-6.02568448e-01 6.39352620e-01 -5.87808609e-01 -1.16611898e+00
-2.30178963e-02 -3.08446437e-01 -3.16841722e-01 6.24178350e-01
-1.55667865e+00 -8.88911545e-01 -3.39656115e-01 6.36383891e-01
4.07872587e-01 -5.27385846e-02 6.83739066e-01 1.02338803e+00
-6.93164766e-01 1.09915352e+00 3.18486452e-01 1.11011118e-01
6.28027990e-02 -6.48034275e-01 4.06350702e-01 6.87761724e-01
-1.63081333e-01 3.17127556e-01 2.13462368e-01 -5.42385638e-01
-1.91318536e+00 -1.34141600e+00 1.01717222e+00 4.58515346e-01
2.61179656e-01 -5.79082012e-01 -6.83261454e-01 4.87735987e-01
-1.18127493e-02 4.38765824e-01 7.40278542e-01 -2.06853166e-01
-3.82239878e-01 -3.76941949e-01 -9.91895020e-01 5.35663366e-01
1.22840178e+00 -7.97371030e-01 3.92342657e-01 2.86390424e-01
9.53460634e-01 -5.66345096e-01 -7.07467139e-01 2.29121745e-01
5.91991961e-01 -7.65244365e-01 5.86006343e-01 -4.20258306e-02
6.02521598e-01 -7.87349269e-02 -4.47176307e-01 -8.87578249e-01
-2.69893199e-01 -7.63464391e-01 -6.99524045e-01 9.02829468e-01
5.63463748e-01 -3.50285798e-01 1.10109055e+00 5.95742464e-01
-2.96339303e-01 -1.20321572e+00 -1.16225946e+00 -7.47248948e-01
-4.11783516e-01 -8.51770699e-01 7.55626798e-01 4.49884206e-01
-4.05111946e-02 3.27222317e-01 -6.80481434e-01 2.58557469e-01
2.47932836e-01 -2.86872357e-01 5.71957529e-01 -1.02782309e+00
-7.53169894e-01 -2.43480086e-01 -4.94727224e-01 -1.70062482e+00
-9.98278558e-02 -9.13250864e-01 -9.05137882e-02 -1.24066210e+00
1.07748061e-01 -8.11097801e-01 -1.73162684e-01 5.69954813e-01
2.09690154e-01 2.58991510e-01 1.80983856e-01 2.72559524e-01
-6.75532877e-01 7.15685606e-01 8.32985938e-01 -4.32140864e-02
-2.62423515e-01 1.20796695e-01 -8.01179588e-01 5.82102597e-01
6.87609375e-01 -4.30955976e-01 -5.16482830e-01 -1.06186175e+00
4.16164666e-01 2.96228647e-01 2.82214582e-01 -1.23553801e+00
6.92903221e-01 5.27158797e-01 3.15075219e-01 -5.92380874e-02
4.25674886e-01 -1.30171776e+00 2.82594413e-01 7.50367105e-01
-4.89736944e-02 1.75159365e-01 -5.74055202e-02 6.54649138e-01
-1.73408270e-01 2.95357667e-02 4.12863165e-01 2.66366422e-01
-4.28993255e-01 5.95422685e-01 -2.81597257e-01 -5.04771709e-01
8.57687891e-01 -2.25162700e-01 -1.07140444e-01 -4.27233636e-01
-2.98573166e-01 1.01980045e-01 3.90121192e-01 1.93185300e-01
6.28071308e-01 -1.35229337e+00 -2.00006068e-01 6.10217929e-01
-2.70356148e-01 1.89266413e-01 6.74625695e-01 1.00013983e+00
-4.32061255e-01 3.78620327e-01 1.32763252e-01 -3.93540442e-01
-9.75135744e-01 3.68144661e-01 2.11870089e-01 -2.42966428e-01
-7.02563167e-01 7.90885806e-01 -1.61030456e-01 1.88839078e-01
5.15997887e-01 -3.24186981e-01 5.12706220e-01 -4.12336469e-01
7.48110414e-01 4.34524983e-01 3.52354497e-01 -3.46287489e-01
-4.34651263e-02 2.24278525e-01 -3.50820199e-02 2.11218059e-01
1.09031105e+00 -3.24933320e-01 1.19571887e-01 -7.49377832e-02
1.72460580e+00 -1.50492340e-01 -1.49018002e+00 -4.21588063e-01
-3.63766819e-01 -1.53948784e-01 7.16832221e-01 -4.12916571e-01
-1.67767286e+00 7.48881280e-01 6.75094604e-01 1.76014043e-02
1.62870622e+00 -2.35967606e-01 1.70869720e+00 4.38670337e-01
5.53916395e-01 -1.10302126e+00 -2.82018334e-01 5.13163567e-01
1.70066282e-01 -1.00720334e+00 -3.60237092e-01 -4.12995011e-01
-3.49738032e-01 9.89906609e-01 1.47344977e-01 1.20372228e-01
1.07414496e+00 8.27984452e-01 -4.72230345e-01 1.35382250e-01
-8.77046525e-01 1.23191699e-01 -1.36437863e-01 3.19289327e-01
5.48195839e-02 1.97684571e-01 -8.88808519e-02 1.04879975e+00
-1.36345267e-01 4.82151031e-01 1.50598168e-01 9.43680584e-01
-2.46600688e-01 -1.04009295e+00 1.54651627e-01 6.56198859e-01
-5.83524287e-01 -3.99321884e-01 2.50906587e-01 1.87326595e-01
3.62202793e-01 7.98297226e-01 3.00629109e-01 -1.14274025e+00
1.02886586e-02 -1.68605864e-01 -2.73402575e-02 -2.66544938e-01
-4.96458620e-01 1.49459943e-01 -5.69378920e-02 -7.48742104e-01
4.82405052e-02 -1.33140847e-01 -1.38512397e+00 -7.10269034e-01
-4.01943773e-01 -2.81222817e-02 9.35193658e-01 1.04863358e+00
1.10410333e+00 6.24272287e-01 7.42171407e-01 -8.56347263e-01
-3.48894864e-01 -4.95529294e-01 -7.49945700e-01 -8.43627304e-02
4.73920017e-01 -1.52393803e-01 -1.67269170e-01 -3.61887552e-02]
|
[8.445035934448242, 2.878067970275879]
|
764ca703-0b91-4e17-9b2f-424be2dc9811
|
a-large-scale-chinese-short-text-conversation
|
2008.03946
| null |
https://arxiv.org/abs/2008.03946v2
|
https://arxiv.org/pdf/2008.03946v2.pdf
|
A Large-Scale Chinese Short-Text Conversation Dataset
|
The advancements of neural dialogue generation models show promising results on modeling short-text conversations. However, training such models usually needs a large-scale high-quality dialogue corpus, which is hard to access. In this paper, we present a large-scale cleaned Chinese conversation dataset, LCCC, which contains a base version (6.8million dialogues) and a large version (12.0 million dialogues). The quality of our dataset is ensured by a rigorous data cleaning pipeline, which is built based on a set of rules and a classifier that is trained on manually annotated 110K dialogue pairs. We also release pre-training dialogue models which are trained on LCCC-base and LCCC-large respectively. The cleaned dataset and the pre-training models will facilitate the research of short-text conversation modeling. All the models and datasets are available at https://github.com/thu-coai/CDial-GPT.
|
['Yinhe Zheng', 'Minlie Huang', 'Yida Wang', 'Xiaoyan Zhu', 'Yong Jiang', 'Pei Ke', 'Kaili Huang']
|
2020-08-10
| null | null | null | null |
['short-text-conversation']
|
['natural-language-processing']
|
[-1.15555391e-01 4.32570934e-01 1.94776133e-01 -6.01255655e-01
-8.47731531e-01 -4.58208293e-01 7.95920253e-01 -1.47210568e-01
-3.00822198e-01 1.20083737e+00 7.91177869e-01 -3.72541100e-01
5.01544833e-01 -7.40592718e-01 -1.81280926e-01 -4.99416471e-01
9.95115414e-02 7.34642386e-01 -2.06740290e-01 -6.85044467e-01
1.92303121e-01 -5.17787576e-01 -9.59923923e-01 6.70796633e-01
1.20826149e+00 6.19638562e-01 3.21535617e-01 1.09915113e+00
-5.03825665e-01 7.50518024e-01 -1.02751303e+00 -6.78236246e-01
-7.04897940e-02 -9.13311720e-01 -1.33533800e+00 -9.29248706e-02
-2.38741368e-01 -4.64739472e-01 -1.75494120e-01 6.72719300e-01
6.96619153e-01 7.19369873e-02 3.46308202e-01 -1.21920025e+00
-6.13405108e-01 1.10216713e+00 4.25003096e-02 -3.13382328e-01
4.32665348e-01 1.95742726e-01 8.52694273e-01 -6.03117049e-01
5.64017236e-01 1.29324460e+00 5.87349534e-01 1.06884623e+00
-7.63924181e-01 -6.50685191e-01 4.72871326e-02 -1.68793008e-01
-5.99168777e-01 -6.95744574e-01 5.67901969e-01 -2.41594762e-01
1.05598807e+00 3.40833843e-01 6.69720411e-01 1.55663419e+00
-5.66453785e-02 9.60300148e-01 1.05846119e+00 -6.01601362e-01
-2.05489136e-02 2.51295000e-01 5.66276133e-01 4.28667039e-01
-1.60172775e-01 -4.37298775e-01 -4.22982454e-01 -3.06694955e-01
4.96150851e-01 -3.33811611e-01 -3.82788062e-01 3.36480349e-01
-1.10691881e+00 1.09809101e+00 -4.57990281e-02 4.06953067e-01
-9.16027948e-02 -4.33183491e-01 7.15422273e-01 4.89824235e-01
6.30465865e-01 5.58266521e-01 -7.01036751e-01 -8.80538881e-01
-1.86000124e-01 2.18021512e-01 1.50196242e+00 1.38998842e+00
4.94916558e-01 -1.93906412e-01 -2.73935407e-01 1.44605827e+00
1.62223756e-01 1.51437685e-01 6.72169983e-01 -8.10336351e-01
9.44265723e-01 6.16700172e-01 6.18177801e-02 -6.09761715e-01
-4.17528480e-01 1.42901927e-01 -1.14572155e+00 -6.96995735e-01
6.58596516e-01 -8.91013026e-01 -2.97633082e-01 1.81787169e+00
1.98497355e-01 -2.98874557e-01 6.04709029e-01 6.26716673e-01
1.26441073e+00 9.41909492e-01 -1.88561603e-01 -3.35225880e-01
1.32270122e+00 -1.50562227e+00 -1.16087902e+00 -3.97155099e-02
9.40018296e-01 -7.29235411e-01 1.51196492e+00 4.63802248e-01
-1.00987041e+00 -4.57195520e-01 -7.46524870e-01 -3.05488408e-01
-4.37178165e-01 9.09617096e-02 6.67315602e-01 6.96666777e-01
-8.78136814e-01 1.91086739e-01 -3.62031788e-01 -3.95465851e-01
9.09440592e-02 3.31554711e-02 -1.87200338e-01 6.88498840e-02
-1.66353250e+00 8.98815036e-01 4.14051563e-01 3.53259265e-01
-7.08571136e-01 -2.32886150e-01 -8.81812692e-01 -4.43346612e-02
3.95589352e-01 -2.28027657e-01 1.83038735e+00 -6.34648502e-01
-2.22258544e+00 6.11525893e-01 -1.95763767e-01 -3.89148682e-01
6.37737870e-01 -4.04266089e-01 -3.15735519e-01 -2.38612279e-01
-6.50405288e-02 3.84768963e-01 2.05063790e-01 -1.07020772e+00
-6.94067955e-01 -1.91106290e-01 8.71296525e-02 3.08535725e-01
-3.77443075e-01 5.20967729e-02 -5.49637675e-01 -3.66348237e-01
-4.47646081e-01 -8.54708672e-01 -2.77761698e-01 -9.99300241e-01
-8.15971613e-01 -5.07524073e-01 5.24420738e-01 -9.47980702e-01
1.41702116e+00 -1.63559842e+00 -1.08966388e-01 -3.56344044e-01
3.55752222e-02 4.82237548e-01 -2.79051930e-01 1.03094995e+00
3.17876726e-01 2.91996747e-01 -2.31334075e-01 -8.24371040e-01
2.00651482e-01 1.31683201e-01 -1.99781686e-01 -1.90183699e-01
1.92043215e-01 8.54740202e-01 -9.07287776e-01 -3.57082188e-01
1.04759313e-01 2.06007045e-02 -3.91822428e-01 9.03463185e-01
-4.69711602e-01 6.05679393e-01 -3.74208361e-01 1.85792118e-01
6.72524095e-01 -7.90631771e-02 4.15157974e-01 2.28226632e-01
-1.37352705e-01 8.54984701e-01 -5.96439600e-01 1.92685163e+00
-3.93249154e-01 5.31107605e-01 3.55023921e-01 -5.83169639e-01
1.07274556e+00 5.85585535e-01 1.25261188e-01 -4.76103306e-01
3.68781120e-01 -7.31433257e-02 8.19354691e-03 -6.34315014e-01
8.86946321e-01 1.81306094e-01 -5.36584556e-01 9.66559291e-01
1.71833500e-01 -2.35993594e-01 5.46248198e-01 4.22468543e-01
8.36448431e-01 -2.95989066e-01 1.88907415e-01 -9.26073343e-02
6.08655334e-01 1.15090951e-01 7.21450627e-01 6.03376985e-01
-2.94230372e-01 5.47797859e-01 9.95102525e-01 -1.14357643e-01
-9.84195411e-01 -3.08118910e-01 -2.62438562e-02 1.23357129e+00
-2.99759805e-01 -8.25545490e-01 -1.18753219e+00 -7.47660756e-01
-2.84070045e-01 7.89078116e-01 -5.60730338e-01 8.00499022e-02
-6.14610672e-01 -9.25106645e-01 8.63318682e-01 1.52990088e-01
9.07958508e-01 -1.34418499e+00 2.58001447e-01 2.90455550e-01
-7.06950128e-01 -9.34797704e-01 -5.93196392e-01 -8.88510719e-02
-4.47551429e-01 -1.19626629e+00 -4.41960871e-01 -9.69657123e-01
4.11981225e-01 1.53955117e-01 1.25630498e+00 2.16816649e-01
2.07725108e-01 -1.16719872e-01 -7.02627957e-01 -6.69351280e-01
-1.02366936e+00 5.97592294e-01 -1.22496054e-01 -2.53719687e-01
6.39236450e-01 -1.40022829e-01 -1.05591245e-01 3.25911164e-01
-4.33510214e-01 7.01102912e-01 3.94703567e-01 1.30909348e+00
-7.15796649e-02 -2.66778618e-01 1.18869090e+00 -1.29774106e+00
1.46903670e+00 -4.77441579e-01 -3.43152523e-01 3.58749926e-01
-4.70512539e-01 -2.00312659e-01 6.79385424e-01 -2.58688003e-01
-1.56406379e+00 -3.29622895e-01 -6.24315143e-01 3.97729397e-01
-3.99128616e-01 5.81839800e-01 -5.07642508e-01 7.08725572e-01
3.56059611e-01 1.31180540e-01 2.77368069e-01 -6.95931017e-01
4.89395142e-01 1.58308518e+00 2.92040139e-01 -7.18069315e-01
1.35736421e-01 -2.17140436e-01 -1.02089798e+00 -1.02138293e+00
-7.70635962e-01 -3.78085077e-01 -9.04623806e-01 -1.70761228e-01
9.40893948e-01 -7.54541278e-01 -7.30943918e-01 9.52102661e-01
-1.37626266e+00 -8.84621918e-01 2.50533968e-01 2.40952700e-01
-3.17831397e-01 3.45504075e-01 -1.26796687e+00 -9.53160346e-01
-8.86256695e-01 -8.90950680e-01 5.75956702e-01 4.57021743e-01
-3.19131941e-01 -1.01352930e+00 4.36369359e-01 7.52058923e-01
3.49899054e-01 -2.31318772e-01 6.82804048e-01 -1.02991426e+00
-3.70143801e-02 -8.09067935e-02 -2.67449636e-02 5.94900191e-01
2.74932504e-01 8.27165693e-02 -1.13622987e+00 -1.21668458e-01
1.10599115e-01 -8.97064030e-01 4.91321653e-01 1.01238945e-02
9.68770444e-01 -5.07336080e-01 6.92789257e-02 1.56623885e-01
6.69825017e-01 3.79906505e-01 6.58361733e-01 7.22844377e-02
6.45749092e-01 8.87065530e-01 7.59269476e-01 4.13679302e-01
8.87138307e-01 2.77887881e-01 -3.88766266e-02 -1.15556093e-02
3.53270084e-01 -3.34503025e-01 1.97604284e-01 1.75526726e+00
1.40262991e-02 -3.80723447e-01 -9.29289222e-01 5.51883101e-01
-2.08376884e+00 -9.01258051e-01 -3.32852453e-01 1.76418579e+00
1.56391609e+00 4.75662202e-02 2.00651720e-01 -2.38412187e-01
6.82330191e-01 1.39116287e-01 -2.69934475e-01 -6.56522572e-01
-5.16123548e-02 -2.43171185e-01 -2.86060333e-01 8.23527038e-01
-9.49102402e-01 1.03022063e+00 6.14587021e+00 5.82490504e-01
-6.52709961e-01 1.12922020e-01 8.43135953e-01 1.62425544e-02
-8.16372558e-02 -1.08635008e-01 -1.01403534e+00 7.23396540e-01
1.35307205e+00 -2.70061493e-01 3.48155320e-01 7.72558808e-01
3.57304305e-01 -1.81131378e-01 -9.05875802e-01 7.60993600e-01
3.65871787e-02 -1.19933581e+00 -9.30985734e-02 7.52231777e-02
6.25427783e-01 -4.37564738e-02 -5.80032110e-01 9.93770719e-01
7.68979251e-01 -9.82488930e-01 7.88395777e-02 3.75127494e-01
5.35441160e-01 -6.79166555e-01 1.11461067e+00 7.49866366e-01
-7.70209193e-01 1.26540422e-01 -4.71706271e-01 -2.06267655e-01
1.94408759e-01 3.25411916e-01 -1.14206374e+00 5.97427368e-01
4.81550664e-01 7.12302983e-01 -3.60076517e-01 4.57395911e-01
-2.69294351e-01 8.54571879e-01 -3.06337960e-02 -5.36900043e-01
1.68793425e-01 -5.62802732e-01 2.06782259e-02 1.59252656e+00
-2.88937420e-01 3.68975312e-01 2.00174481e-01 5.51312566e-01
-2.90295541e-01 2.10969284e-01 -3.14991295e-01 -1.52602240e-01
7.95842826e-01 1.55059397e+00 7.52105983e-03 -5.33565521e-01
-4.34429914e-01 7.77849376e-01 6.22328997e-01 1.82776988e-01
-5.81259608e-01 -5.94519615e-01 7.46654451e-01 -5.98896682e-01
-3.39576572e-01 -1.65702775e-01 -2.68441826e-01 -1.47518706e+00
-5.68190143e-02 -1.25399923e+00 2.82134324e-01 -5.03336906e-01
-1.58993232e+00 8.62901688e-01 -1.61139116e-01 -8.27831864e-01
-7.28486359e-01 -4.94368941e-01 -9.46284652e-01 9.87300813e-01
-1.11407983e+00 -1.14091694e+00 -5.71355283e-01 3.85969460e-01
1.02040637e+00 -3.03702414e-01 1.30715144e+00 2.09818527e-01
-1.16357255e+00 6.86647654e-01 3.16766232e-01 7.84300208e-01
1.03656125e+00 -1.42928028e+00 6.66832507e-01 3.83397818e-01
-5.54237723e-01 7.53058374e-01 3.55295509e-01 -7.41282940e-01
-1.20065296e+00 -1.01954091e+00 1.06677020e+00 -6.37388229e-01
5.85395753e-01 -9.29155290e-01 -1.11430395e+00 7.76620567e-01
9.17905807e-01 -9.42318559e-01 9.77685094e-01 5.40840030e-01
2.40986839e-01 2.37620860e-01 -9.35274243e-01 6.12840950e-01
8.07567716e-01 -5.39032221e-01 -7.77746737e-01 3.74331862e-01
8.96046162e-01 -5.75510502e-01 -1.13676965e+00 -5.31218983e-02
5.13277471e-01 -8.48284245e-01 2.48182848e-01 -7.96416223e-01
4.43739831e-01 3.52165461e-01 2.35850617e-01 -1.68371367e+00
9.26309079e-02 -1.05340707e+00 1.40413875e-02 1.68306875e+00
7.08691061e-01 -6.71676457e-01 7.01747954e-01 9.05915856e-01
-4.30914521e-01 -7.84960985e-01 -4.79893297e-01 -3.15271616e-01
3.85362715e-01 -2.39750311e-01 9.00830448e-01 1.29055905e+00
7.73479640e-01 1.02672601e+00 -7.04480588e-01 -4.72821027e-01
1.32561803e-01 4.39984637e-04 1.34290504e+00 -9.79171455e-01
-7.00010061e-02 -3.47356170e-01 6.02092683e-01 -1.31061625e+00
6.79197162e-02 -3.92093629e-01 2.88930207e-01 -1.68000805e+00
-1.53705506e-02 -6.04398966e-01 4.29557383e-01 4.18738216e-01
-5.18864036e-01 -3.59558731e-01 -3.99134345e-02 1.08303241e-01
-6.47970617e-01 1.08544338e+00 1.28241694e+00 1.75648332e-01
-5.09266138e-01 1.54137760e-01 -7.47976899e-01 5.22559643e-01
1.36773694e+00 -1.03995159e-01 -4.08491880e-01 -4.56037372e-01
-2.68048227e-01 1.71167180e-01 -3.77211362e-01 -3.75875920e-01
1.67490512e-01 -1.42869502e-01 -2.11210232e-02 -7.13524222e-01
5.93761325e-01 -5.70788197e-02 -3.11889470e-01 1.70780659e-01
-8.08874249e-01 -5.47140390e-02 4.92516048e-02 2.43130773e-01
-2.53103226e-01 -3.41108829e-01 3.99109513e-01 -4.57426161e-01
-2.82520056e-01 -2.29317918e-02 -5.42399466e-01 3.05010945e-01
6.65176213e-01 3.97043586e-01 -8.63986313e-01 -8.77646267e-01
-5.30900002e-01 7.93916464e-01 1.25384942e-01 6.79715395e-01
2.53326029e-01 -1.15099537e+00 -9.65822458e-01 1.70401335e-01
-3.12353428e-02 3.73790175e-01 2.92491645e-01 4.79390234e-01
-4.61521983e-01 7.03144729e-01 -2.63988137e-01 -1.35077223e-01
-1.39716256e+00 9.29189324e-02 3.04510385e-01 -5.06964028e-01
-3.70462209e-01 6.94284022e-01 -1.95694640e-01 -1.19530499e+00
4.78316694e-01 -4.43146378e-01 -4.83586460e-01 9.64052379e-02
7.92367041e-01 2.14112118e-01 -9.90800783e-02 -4.09868836e-01
1.90026164e-01 -2.91358739e-01 -2.83940017e-01 -2.03017309e-01
1.26723778e+00 -4.15810019e-01 -4.09809411e-01 6.53844059e-01
1.00315356e+00 6.98544458e-02 -1.06507027e+00 -2.94125229e-01
-4.58265990e-02 -2.43443847e-01 -5.12425303e-01 -9.62644815e-01
-5.29567897e-01 8.11245143e-01 -1.03671670e-01 5.68083286e-01
6.34122491e-01 -3.20648104e-01 1.11007202e+00 7.96070158e-01
2.24036321e-01 -1.33277261e+00 6.80906400e-02 1.28002417e+00
1.13473988e+00 -1.51264250e+00 -4.37151819e-01 -3.84526879e-01
-1.16376793e+00 9.60797727e-01 1.10312092e+00 2.84751058e-01
5.69994390e-01 1.48844570e-01 5.90432405e-01 -6.11386001e-02
-1.18717694e+00 7.12933904e-03 -1.97493076e-01 5.36372244e-01
7.29292750e-01 1.74872354e-01 -3.64172667e-01 1.29070044e+00
-8.03095758e-01 -2.91058272e-01 8.00244868e-01 8.40212047e-01
-3.15669596e-01 -1.42408121e+00 -9.07310024e-02 5.04919231e-01
-2.17159733e-01 -2.51119703e-01 -9.71633434e-01 6.94316804e-01
-3.38782430e-01 1.60412645e+00 -2.37665325e-02 -6.63877487e-01
3.37937564e-01 4.02164817e-01 -1.61092326e-01 -8.23648632e-01
-9.36822534e-01 -1.17200585e-02 1.08036709e+00 -8.94936919e-02
-2.19427779e-01 -3.84054750e-01 -1.13993609e+00 -7.42511332e-01
-5.44477165e-01 9.09205675e-01 5.80876529e-01 7.14950383e-01
4.45602894e-01 2.96635628e-01 6.80350244e-01 -6.55098975e-01
-5.90738833e-01 -1.95027900e+00 -3.73689324e-01 2.98631191e-01
8.61656889e-02 -1.30975738e-01 -2.38214090e-01 1.63616743e-02]
|
[12.788250923156738, 8.077248573303223]
|
9b84b9d4-8d1c-4bac-84bb-4e7373ca1ce4
|
multi-task-determinantal-point-processes-for
|
1805.09916
| null |
http://arxiv.org/abs/1805.09916v2
|
http://arxiv.org/pdf/1805.09916v2.pdf
|
Multi-Task Determinantal Point Processes for Recommendation
|
Determinantal point processes (DPPs) have received significant attention in
the recent years as an elegant model for a variety of machine learning tasks,
due to their ability to elegantly model set diversity and item quality or
popularity. Recent work has shown that DPPs can be effective models for product
recommendation and basket completion tasks. We present an enhanced DPP model
that is specialized for the task of basket completion, the multi-task DPP. We
view the basket completion problem as a multi-class classification problem, and
leverage ideas from tensor factorization and multi-class classification to
design the multi-task DPP model. We evaluate our model on several real-world
datasets, and find that the multi-task DPP provides significantly better
predictive quality than a number of state-of-the-art models.
|
['Jérémie Mary', 'Mike Gartrell', 'Romain Warlop']
|
2018-05-24
| null | null | null | null |
['product-recommendation']
|
['miscellaneous']
|
[-3.72603923e-01 -7.76060998e-01 -8.98862243e-01 -1.79674625e-01
-8.42586100e-01 -6.37849271e-01 5.84105790e-01 2.80786306e-01
2.08409220e-01 3.89091372e-01 6.64636433e-01 -4.96029735e-01
-3.12503636e-01 -5.72691202e-01 -7.24956572e-01 -4.26592827e-01
-1.66486070e-01 7.02223837e-01 -9.74235088e-02 -3.65047425e-01
3.52095097e-01 -1.01879679e-01 -1.42224836e+00 8.88014019e-01
8.19480598e-01 1.15354860e+00 -1.10076889e-01 5.70057571e-01
-1.49268612e-01 8.23276758e-01 -1.92858595e-02 -8.94518673e-01
3.88201654e-01 2.21286967e-01 -6.25479639e-01 -7.43087456e-02
3.39443713e-01 -8.18786696e-02 -1.23040885e-01 6.40661895e-01
2.27779388e-01 8.16550553e-02 9.92271543e-01 -1.46062779e+00
-1.20939505e+00 5.41512847e-01 -8.76096725e-01 3.19945872e-01
4.35884118e-01 -1.76671386e-01 1.98055255e+00 -1.20066583e+00
2.03286588e-01 1.36332297e+00 9.55862701e-01 2.66712047e-02
-1.82620370e+00 -5.15863180e-01 4.88009691e-01 5.87112783e-03
-9.09030795e-01 1.94052551e-02 6.16946936e-01 -8.32044959e-01
5.90521038e-01 2.99797118e-01 7.20465124e-01 1.14280319e+00
6.24996245e-01 1.54561388e+00 1.05184352e+00 -8.66230726e-02
2.24855304e-01 -2.06026748e-01 5.72378218e-01 3.78527284e-01
4.33935732e-01 -8.61578584e-02 -8.19129586e-01 -8.52464318e-01
7.97569633e-01 6.58391535e-01 1.39709096e-02 -4.17003095e-01
-1.00088215e+00 1.23317707e+00 6.38221726e-02 -1.29919350e-01
-6.28274441e-01 1.74971670e-01 3.79044503e-01 3.20520818e-01
1.10280609e+00 4.37145919e-01 -7.40322471e-01 -3.01345527e-01
-9.60085034e-01 9.02833462e-01 1.11784089e+00 6.70431674e-01
4.35198873e-01 -2.51235664e-01 -4.58395660e-01 9.95962203e-01
5.77580869e-01 5.52837014e-01 1.99655041e-01 -7.05540240e-01
5.25374591e-01 4.33194369e-01 3.59467357e-01 -1.03566241e+00
-2.44851977e-01 -7.12782681e-01 -6.94015682e-01 -2.68604815e-01
3.98499578e-01 2.14796379e-01 -6.63619339e-01 1.47690570e+00
1.73971176e-01 1.34838268e-01 -3.63065302e-01 5.51529408e-01
5.08817434e-01 7.14051843e-01 2.62372941e-01 1.68377571e-02
1.25292242e+00 -1.03309202e+00 -3.04656684e-01 1.63325593e-01
5.57247400e-01 -1.00941443e+00 1.12660718e+00 9.48970675e-01
-1.06392705e+00 -5.05113184e-01 -5.88648915e-01 3.05256154e-02
4.45488002e-03 -2.50488017e-02 1.41391110e+00 5.84694207e-01
-9.21251535e-01 6.65365696e-01 -6.80030823e-01 5.19196950e-02
5.94539404e-01 1.01266786e-01 1.90423355e-02 -4.30846483e-01
-7.66712606e-01 3.15803200e-01 -3.87390316e-01 -2.03925937e-01
-8.55356932e-01 -1.22871780e+00 -4.59495634e-01 2.44167492e-01
9.35312808e-02 -1.07019258e+00 1.40034056e+00 -5.83566666e-01
-1.18843305e+00 3.72764051e-01 -1.28497094e-01 -3.62136513e-01
1.36029720e-01 -4.92901862e-01 -3.59254092e-01 -3.69581342e-01
2.61597246e-01 2.23254532e-01 9.95806336e-01 -9.69693363e-01
-9.40065622e-01 -4.24855202e-01 3.09413999e-01 5.02105244e-02
-2.92283535e-01 -8.72500166e-02 -4.84815329e-01 -1.00114191e+00
7.82136098e-02 -1.16743624e+00 -4.36153710e-01 -2.82740027e-01
-3.26098770e-01 -5.78085601e-01 2.81153321e-01 -4.64273721e-01
1.56537867e+00 -1.95359623e+00 2.15034649e-01 1.54735660e-02
6.25729263e-01 -2.35866029e-02 -3.31624329e-01 5.33200681e-01
1.06488369e-01 3.97964358e-01 4.54056025e-01 -8.84238124e-01
2.05979213e-01 3.61282110e-01 -6.01087809e-01 3.49147916e-01
-2.82517653e-02 1.06355166e+00 -8.13025951e-01 6.64712414e-02
-1.36107743e-01 2.58614212e-01 -9.95568693e-01 -2.02875450e-01
-5.12640059e-01 9.47809890e-02 -5.61734915e-01 7.47013450e-01
6.57107890e-01 -6.48601413e-01 6.49215132e-02 -1.53345600e-01
1.61693275e-01 4.43871796e-01 -1.20869780e+00 1.55747271e+00
-2.85192609e-01 1.43972203e-01 -1.84993505e-01 -6.72617078e-01
4.49011058e-01 1.71798319e-01 1.01348245e+00 -6.49337947e-01
-3.29669386e-01 2.28429556e-01 -1.98234230e-01 -6.54491559e-02
9.18691814e-01 -1.99475288e-01 -1.74186349e-01 7.59215832e-01
2.72945445e-02 3.10663104e-01 3.55746955e-01 3.07384670e-01
1.04475260e+00 3.01208440e-02 -9.17577446e-02 -4.65193838e-01
-2.01702882e-02 -7.39752278e-02 7.20375180e-01 8.55298996e-01
9.25682634e-02 4.64975417e-01 4.57630783e-01 -7.61248350e-01
-1.13624322e+00 -1.11940432e+00 -3.25292833e-02 1.58589351e+00
-1.51695520e-01 -8.54863226e-01 1.25893727e-01 -5.63228190e-01
7.88248837e-01 1.67835459e-01 -7.41335452e-01 2.53236443e-01
-2.27210417e-01 -1.09870410e+00 1.91327393e-01 8.02496195e-01
-1.47411302e-01 -2.93681264e-01 3.24235588e-01 4.40513700e-01
-2.50171810e-01 -7.65015006e-01 -6.17192149e-01 -8.29531997e-03
-1.05234098e+00 -1.11164057e+00 -8.57554257e-01 -4.55637336e-01
7.73867080e-03 7.82125652e-01 1.61384881e+00 -1.39411345e-01
1.34112865e-01 4.98642892e-01 -5.31832635e-01 -5.19478440e-01
-1.70895979e-01 1.60424858e-01 3.82032394e-01 2.17164248e-01
3.60090375e-01 -7.06219435e-01 -7.32519507e-01 4.89289433e-01
-9.71898496e-01 8.63642246e-02 2.84491658e-01 8.41541171e-01
6.82330370e-01 -1.31597549e-01 5.47032535e-01 -1.04259694e+00
1.14426720e+00 -8.83510947e-01 -2.22856522e-01 5.66611588e-02
-8.23599875e-01 8.65720138e-02 5.12906671e-01 -6.30629122e-01
-8.03852201e-01 -6.02772653e-01 -2.16331959e-01 -4.84955758e-01
4.44290251e-01 1.03181279e+00 4.15782541e-01 -3.17350565e-03
5.38804471e-01 -3.59065793e-02 -3.67404878e-01 -1.05285549e+00
6.62204683e-01 3.52064520e-01 -6.61786422e-02 -1.01799071e+00
4.96958524e-01 6.45272136e-01 1.51604772e-01 -4.50482517e-01
-1.28197885e+00 -9.44840431e-01 -2.25359738e-01 2.29546323e-01
3.69884968e-01 -1.33567524e+00 -8.46294463e-01 2.77026027e-01
-9.07236636e-01 -2.66812831e-01 -3.54079485e-01 4.90393639e-01
-3.98653477e-01 2.06874177e-01 -1.11171830e+00 -8.21139634e-01
-4.16799992e-01 -9.22307253e-01 1.23410749e+00 -1.55109450e-01
-7.42609277e-02 -1.28036726e+00 4.67736155e-01 3.39323223e-01
3.34767967e-01 -2.15792414e-02 1.19093823e+00 -5.07852256e-01
-5.24815619e-01 -1.78700447e-01 -2.96071880e-02 3.34785759e-01
-9.03457552e-02 4.24069688e-02 -5.80217659e-01 -2.08990678e-01
-9.41684693e-02 -2.05948293e-01 1.25718868e+00 7.59434640e-01
9.55017447e-01 -1.87211573e-01 -2.74620980e-01 5.39807260e-01
1.32986724e+00 -2.42428139e-01 3.61168087e-01 1.57504566e-02
7.88047910e-01 -4.64025810e-02 5.06885588e-01 6.74482048e-01
9.44620669e-01 7.18712807e-01 1.30879104e-01 -5.61023541e-02
2.74263546e-02 -5.54101348e-01 2.88508803e-01 1.08868599e+00
-2.41780967e-01 -1.89912900e-01 -7.59076178e-01 3.57964039e-01
-2.41192818e+00 -1.07289779e+00 -4.13572878e-01 1.96833742e+00
4.45945561e-01 1.18233189e-01 7.57614255e-01 9.96044353e-02
1.87606066e-01 2.93099701e-01 -3.28984559e-01 -4.23287213e-01
-2.31390357e-01 1.26595318e-01 3.64699155e-01 1.58571914e-01
-1.21156847e+00 4.86136049e-01 7.56840086e+00 9.18029964e-01
-8.11170757e-01 3.73960435e-01 7.70384371e-01 -2.84910947e-01
-6.52725637e-01 -9.32447612e-02 -1.06250787e+00 5.42338610e-01
6.33256137e-01 -1.23796672e-01 4.27213252e-01 9.98497486e-01
1.14765711e-01 1.03419162e-01 -1.45696640e+00 1.20883238e+00
8.36052373e-02 -1.58267152e+00 3.07511806e-01 4.93410051e-01
1.27424765e+00 3.24130476e-01 6.79373443e-01 6.21465862e-01
7.30668187e-01 -6.94849551e-01 7.49260128e-01 5.12660086e-01
4.42271948e-01 -4.69502807e-01 3.00397545e-01 3.37198794e-01
-1.31147468e+00 -5.56877494e-01 -5.30824661e-01 -5.13687730e-01
3.29283714e-01 9.19401526e-01 -1.86812803e-01 5.16392887e-01
6.36276245e-01 1.15707421e+00 -3.43172640e-01 1.34877503e+00
1.72965109e-01 1.03580642e+00 -2.73907810e-01 7.47037008e-02
2.84440249e-01 -3.39421540e-01 4.86061871e-01 1.11653399e+00
2.09832639e-01 -2.33700231e-01 5.47384501e-01 5.39120078e-01
-2.35702798e-01 2.70669580e-01 -2.59317994e-01 -1.98788553e-01
-1.27748907e-01 9.81699288e-01 -3.22198898e-01 -8.48410130e-02
-7.03213513e-01 5.03342628e-01 3.25025380e-01 4.47030604e-01
-5.90714395e-01 5.73515415e-01 1.20987785e+00 2.42784336e-01
4.97162879e-01 -4.59508717e-01 -5.88537931e-01 -1.68160081e+00
-9.65692401e-02 -9.13117528e-01 3.98848623e-01 -4.03235942e-01
-2.10826564e+00 1.12545732e-02 -2.47879028e-01 -1.13144910e+00
-1.10909324e-02 -7.12320864e-01 -4.09440428e-01 7.68872619e-01
-1.56835508e+00 -1.30108535e+00 3.31636846e-01 6.41047537e-01
6.28775477e-01 -8.57983306e-02 7.91633308e-01 7.43432879e-01
-4.54378098e-01 4.24156368e-01 7.35269427e-01 -1.68448001e-01
5.19825757e-01 -1.35198748e+00 5.99190891e-01 4.94500846e-01
5.54699302e-01 7.21799552e-01 6.67991519e-01 -6.72057629e-01
-1.87451482e+00 -8.97819996e-01 7.47130275e-01 -9.48735476e-01
9.20019388e-01 -4.65119332e-01 -4.66390431e-01 7.96667337e-01
-4.28639472e-01 6.29631756e-03 1.29558086e+00 1.17299783e+00
-8.93788695e-01 -1.85542554e-01 -5.76866090e-01 5.41644692e-01
1.00882828e+00 -5.55072427e-01 -3.00098032e-01 5.70800424e-01
7.01338589e-01 -1.47560477e-01 -1.05859721e+00 1.62597269e-01
9.46872473e-01 -7.29707420e-01 1.19175696e+00 -1.20768452e+00
8.88138771e-01 7.79923201e-02 -5.56469500e-01 -1.42746115e+00
-1.17369401e+00 -6.94204092e-01 -6.29607499e-01 9.82817352e-01
5.88839889e-01 -3.85829240e-01 8.94725144e-01 8.08994651e-01
-5.77262305e-02 -1.07794559e+00 -6.68942809e-01 -7.08387733e-01
2.60365784e-01 -7.16001749e-01 6.47108674e-01 6.04227722e-01
2.19253264e-02 7.03779757e-01 -7.35276937e-01 -2.89300352e-01
5.36105990e-01 6.02181435e-01 1.01605153e+00 -1.82715118e+00
-8.50604355e-01 -5.41877151e-01 -1.31378978e-01 -1.48019719e+00
-1.81860358e-01 -1.02917218e+00 -5.24384677e-01 -1.58630061e+00
7.13403165e-01 -1.07835078e+00 -5.62393546e-01 3.36195379e-01
-3.12312454e-01 4.00537997e-01 4.89422172e-01 7.49203861e-01
-1.02816284e+00 4.37885135e-01 1.29317427e+00 -9.59985331e-02
-3.08024347e-01 6.12062395e-01 -1.38286722e+00 5.33167481e-01
3.53377730e-01 -5.43574870e-01 -3.82527709e-01 -5.26019692e-01
1.02890873e+00 -7.81030431e-02 8.24629366e-02 -5.25747359e-01
-1.23169251e-01 -2.22349390e-01 3.75884324e-01 -4.84140724e-01
5.38465798e-01 -4.49249566e-01 1.82501301e-01 9.23142433e-02
-3.39279652e-01 3.78859222e-01 -1.85433663e-02 1.04106009e+00
-3.12040579e-02 2.85297453e-01 8.21927115e-02 -1.63676515e-02
-2.75140256e-01 8.34927499e-01 -7.23019987e-02 2.87289321e-02
5.57148099e-01 1.90863624e-01 -3.85171860e-01 -3.65823686e-01
-7.36230969e-01 1.39264330e-01 1.64553896e-01 5.80084383e-01
3.09142232e-01 -1.55613577e+00 -8.42865348e-01 -3.22024412e-02
1.26662225e-01 -5.06479502e-01 1.63477466e-01 9.23471868e-01
9.42699015e-02 5.08318245e-01 1.06197312e-01 -6.77528381e-01
-9.23630357e-01 7.56000698e-01 -2.70707339e-01 -1.01502800e+00
-3.33848357e-01 8.32316935e-01 3.05838376e-01 -1.42824680e-01
-1.30014205e-02 -5.00440896e-01 -2.47074403e-02 1.53847352e-01
6.93955600e-01 4.91784394e-01 1.85549706e-02 -4.44842815e-01
6.39630808e-03 2.97504127e-01 -5.08176565e-01 7.15910867e-02
1.66987360e+00 -8.60572010e-02 7.23120496e-02 6.67915761e-01
1.13926852e+00 -5.96801303e-02 -1.01779389e+00 -5.19195855e-01
-1.02795810e-01 -7.09800124e-01 5.11200093e-02 -6.37496829e-01
-7.41767943e-01 7.60126650e-01 3.75197008e-02 6.35034323e-01
7.48190105e-01 -7.27304369e-02 9.75815773e-01 6.12258799e-02
7.48010099e-01 -7.91795850e-01 3.48616630e-01 6.61140084e-01
6.88410342e-01 -1.18224311e+00 1.37116775e-01 -4.06897336e-01
-8.34656537e-01 7.33427048e-01 -7.90972263e-02 -4.15917128e-01
1.31134439e+00 7.88001642e-02 -5.15392363e-01 -1.30898654e-01
-1.13664079e+00 7.27770478e-02 6.29326999e-01 3.86365235e-01
5.47063291e-01 5.81580222e-01 -4.13596094e-01 1.08697069e+00
7.84257576e-02 3.73258352e-01 1.40115321e-01 6.94916248e-01
-2.60516047e-01 -1.44698334e+00 -1.26160890e-01 1.10495484e+00
-8.11561584e-01 -3.69308442e-01 1.75316647e-01 1.86583281e-01
-2.90945489e-02 1.04200959e+00 -5.34558669e-02 -7.24366128e-01
1.77832127e-01 -6.22581877e-02 5.04054368e-01 -8.62358212e-01
-8.53865027e-01 3.50144386e-01 1.32892653e-01 -6.27763093e-01
-3.30516875e-01 -9.04119372e-01 -2.69005567e-01 -8.29098642e-01
-2.70354837e-01 1.53728679e-01 6.20406449e-01 9.07153368e-01
5.82787454e-01 1.95563108e-01 6.85290992e-01 -9.06305730e-01
-9.03253376e-01 -7.29488492e-01 -1.07386875e+00 6.82305932e-01
1.41008973e-01 -9.15564120e-01 -2.71684527e-02 -1.53048456e-01]
|
[9.731524467468262, 5.510112762451172]
|
a9c96c24-6b9d-4a0e-8335-7896fc2d8197
|
seamlessgan-self-supervised-synthesis-of
|
2201.05120
| null |
https://arxiv.org/abs/2201.05120v1
|
https://arxiv.org/pdf/2201.05120v1.pdf
|
SeamlessGAN: Self-Supervised Synthesis of Tileable Texture Maps
|
We present SeamlessGAN, a method capable of automatically generating tileable texture maps from a single input exemplar. In contrast to most existing methods, focused solely on solving the synthesis problem, our work tackles both problems, synthesis and tileability, simultaneously. Our key idea is to realize that tiling a latent space within a generative network trained using adversarial expansion techniques produces outputs with continuity at the seam intersection that can be then be turned into tileable images by cropping the central area. Since not every value of the latent space is valid to produce high-quality outputs, we leverage the discriminator as a perceptual error metric capable of identifying artifact-free textures during a sampling process. Further, in contrast to previous work on deep texture synthesis, our model is designed and optimized to work with multi-layered texture representations, enabling textures composed of multiple maps such as albedo, normals, etc. We extensively test our design choices for the network architecture, loss function and sampling parameters. We show qualitatively and quantitatively that our approach outperforms previous methods and works for textures of different types.
|
['Elena Garces', 'Carlos Rodriguez-Pardo']
|
2022-01-13
| null | null | null | null |
['texture-synthesis']
|
['computer-vision']
|
[ 7.60813236e-01 4.83786881e-01 1.29420489e-01 9.02088080e-03
-7.88332641e-01 -8.23844016e-01 8.60283017e-01 -5.04392385e-01
1.79310322e-01 8.44544828e-01 1.99710563e-01 -9.99686718e-02
2.23699287e-01 -1.14717054e+00 -9.42049205e-01 -9.11642849e-01
4.43482101e-02 3.42744619e-01 3.00733838e-02 -2.57594347e-01
8.20616856e-02 5.61223924e-01 -1.67078245e+00 4.29813534e-01
7.32653499e-01 1.10002649e+00 -1.81782171e-01 6.94449663e-01
9.98960435e-02 5.48549235e-01 -7.21587837e-01 -4.25861537e-01
6.98160529e-01 -7.22446680e-01 -6.21547878e-01 2.86140263e-01
7.42118776e-01 -2.04490542e-01 2.06047408e-02 9.71912146e-01
3.79357427e-01 -1.09408922e-01 8.20781469e-01 -1.25972307e+00
-7.66797781e-01 3.95443082e-01 -2.23323241e-01 -5.85901380e-01
1.08634464e-01 2.50387281e-01 8.19475234e-01 -7.16552675e-01
8.71308148e-01 1.27583456e+00 8.05898428e-01 6.21469498e-01
-1.79030859e+00 -5.16232789e-01 -1.84771478e-01 -6.78065181e-01
-1.24983239e+00 -5.57611406e-01 6.94247723e-01 -3.53683442e-01
4.53255743e-01 6.73817158e-01 6.97597027e-01 1.40080380e+00
4.47071522e-01 4.46734995e-01 1.49579167e+00 -6.10337138e-01
3.16609561e-01 1.03129605e-02 -8.96221578e-01 6.31066263e-01
5.40184230e-02 1.76128864e-01 -4.57642853e-01 4.24153171e-03
1.51435220e+00 -2.95535564e-01 -2.41838738e-01 -7.25938380e-01
-1.55454373e+00 7.11581171e-01 3.17983240e-01 3.00018229e-02
-1.43976703e-01 6.48924053e-01 5.48237413e-02 6.59166992e-01
7.18293786e-01 8.89615715e-01 -4.41065431e-02 1.56002305e-02
-1.01166904e+00 3.71499479e-01 7.99052477e-01 9.74825978e-01
9.25589621e-01 3.37704659e-01 -4.44143295e-01 6.61659718e-01
-1.25876412e-01 6.50283456e-01 2.10155010e-01 -1.09012079e+00
1.92961067e-01 4.10321385e-01 2.11404726e-01 -9.12566245e-01
1.46002904e-01 -2.95141667e-01 -8.94960165e-01 7.63458610e-01
5.39456606e-01 -2.53022969e-01 -1.25483322e+00 1.70466805e+00
1.61607295e-01 -6.19847551e-02 2.36266060e-03 7.35197604e-01
3.26405764e-01 4.15966541e-01 -3.18299383e-01 3.26459795e-01
1.20299006e+00 -8.64983559e-01 -4.84346777e-01 1.01231948e-01
1.14262998e-01 -1.12084568e+00 1.37746453e+00 3.54063362e-01
-1.12893653e+00 -4.06963795e-01 -1.08582389e+00 5.53127341e-02
-3.21520358e-01 8.27048421e-02 7.82513618e-01 8.29914927e-01
-1.33801329e+00 7.31510818e-01 -6.62506223e-01 -2.21094042e-01
2.68295527e-01 1.93417609e-01 -3.52566719e-01 1.86691806e-01
-9.23054636e-01 7.04975843e-01 1.00875206e-01 -5.88977635e-02
-9.41583514e-01 -6.98985040e-01 -6.98966384e-01 -1.06078543e-01
4.24489938e-02 -8.87407541e-01 8.86075675e-01 -1.41321576e+00
-2.03940845e+00 8.19254458e-01 5.89124002e-02 -2.93525279e-01
9.09088850e-01 6.66077510e-02 -2.30777353e-01 -7.07838759e-02
-9.75160487e-03 1.00772238e+00 1.25355506e+00 -1.46223497e+00
-2.72370875e-01 2.29004413e-01 -3.35004851e-02 1.10154778e-01
-6.57537505e-02 -4.49473977e-01 -1.64367229e-01 -1.14188671e+00
2.30957970e-01 -9.97105062e-01 -2.34800696e-01 2.37514272e-01
-6.51021898e-01 5.45773029e-01 7.26639032e-01 -3.39781880e-01
6.36372387e-01 -2.02269197e+00 3.40647876e-01 4.45209235e-01
2.19580099e-01 -3.44906449e-01 -1.99264154e-01 5.30342817e-01
3.95594863e-03 4.02189225e-01 -2.74005532e-01 -4.20184404e-01
2.38357365e-01 2.69044548e-01 -7.29489267e-01 3.59488338e-01
4.44321930e-01 9.51967001e-01 -6.54849052e-01 -1.14064135e-01
2.71284997e-01 5.86392343e-01 -6.93168998e-01 9.82286781e-02
-4.62791920e-01 6.62751198e-01 -2.64574081e-01 6.05996609e-01
6.53503358e-01 -1.00486442e-01 1.12773925e-01 1.17165707e-02
-1.27768874e-01 6.38865456e-02 -1.12576628e+00 1.66517854e+00
-7.68533826e-01 7.21288443e-01 -5.71019463e-02 -5.07156432e-01
1.25714910e+00 3.14495414e-01 2.67269135e-01 -6.56471193e-01
-6.17506914e-02 4.59392428e-01 -3.72249991e-01 1.22655211e-02
7.30426133e-01 -1.61823958e-01 -2.04475850e-01 5.68685591e-01
-6.92543015e-02 -4.77597952e-01 -2.40445942e-01 -3.20909679e-01
1.05389953e+00 4.18474615e-01 -1.74305364e-01 -5.63285172e-01
-6.31412268e-02 -6.53960854e-02 2.37199903e-01 9.03672576e-01
4.07692403e-01 1.29579425e+00 8.30991209e-01 -6.05060756e-01
-1.90920448e+00 -1.30133736e+00 -1.80237785e-01 8.06686163e-01
1.38559148e-01 -5.68426438e-02 -9.44552064e-01 -4.30785358e-01
-5.09282388e-02 4.76707071e-01 -1.04799199e+00 4.25462686e-02
-5.67111492e-01 -4.73311573e-01 7.21035957e-01 1.09168693e-01
6.04456067e-01 -1.12448347e+00 -7.06482351e-01 1.53924078e-01
-6.92870244e-02 -6.83442712e-01 -4.36665446e-01 2.32213587e-01
-5.84311426e-01 -7.26742685e-01 -8.54398012e-01 -5.83257139e-01
8.26903164e-01 -1.37167349e-01 1.35565913e+00 3.32106091e-02
-2.78891236e-01 1.31931350e-01 -2.57084370e-01 -2.07643241e-01
-7.04989970e-01 1.45235956e-01 -2.77254879e-01 2.98655957e-01
-3.87880147e-01 -8.00060272e-01 -7.47588575e-01 4.24244940e-01
-1.27583826e+00 5.35331786e-01 4.73351389e-01 9.82495606e-01
6.51374519e-01 3.42885330e-02 2.96729863e-01 -9.66140628e-01
4.82850283e-01 -1.92260519e-01 -6.14610672e-01 2.63326526e-01
-2.44757429e-01 4.32583362e-01 6.16948009e-01 -5.38872719e-01
-8.42764318e-01 -8.98450986e-03 3.42509858e-02 -3.04272294e-01
-1.77720934e-01 -1.51095033e-01 -2.02053472e-01 -2.48910785e-01
8.12302351e-01 2.65844345e-01 1.39387190e-01 -2.25165308e-01
5.43660700e-01 9.10016969e-02 4.41467702e-01 -7.78132141e-01
8.89867067e-01 7.33319402e-01 1.33441299e-01 -5.28246999e-01
-3.25831026e-01 4.15811151e-01 -5.15181720e-01 -1.83843210e-01
7.20333457e-01 -6.61753893e-01 -5.12735009e-01 5.96818268e-01
-8.50619376e-01 -7.25058794e-01 -7.77955234e-01 5.06010000e-03
-9.82301474e-01 -2.63710052e-01 -2.89858788e-01 -5.43727100e-01
-1.23735085e-01 -1.23656917e+00 1.48635375e+00 -2.13893265e-01
-3.62810582e-01 -1.01731908e+00 7.27221370e-02 -3.52397412e-01
8.69701624e-01 9.48318839e-01 9.35922265e-01 1.25518560e-01
-8.48937750e-01 9.13268980e-03 -3.61158811e-02 1.03552543e-01
2.93396413e-01 1.42746523e-01 -1.03369403e+00 -3.95436406e-01
-1.62700519e-01 -4.05704170e-01 9.07090485e-01 2.51720458e-01
1.11780250e+00 -6.80026412e-01 -1.16961449e-01 9.68289673e-01
1.54649341e+00 -1.24547146e-01 1.07617736e+00 4.84675258e-01
6.49564207e-01 4.30392087e-01 2.03152057e-02 2.48930261e-01
-9.19033885e-02 8.44506681e-01 3.94587278e-01 -4.88380879e-01
-4.37470168e-01 -4.90816653e-01 1.71217620e-01 2.15014040e-01
-3.84329706e-02 -4.81295198e-01 -5.48523664e-01 4.20570582e-01
-1.61023784e+00 -9.50133145e-01 4.25160438e-01 2.25642967e+00
8.56832564e-01 1.04219951e-01 -2.79895905e-02 9.85114723e-02
6.48015559e-01 3.61459881e-01 -3.86373729e-01 -5.95388114e-01
-3.90503973e-01 6.54089034e-01 8.33586752e-01 5.32354355e-01
-1.09296048e+00 1.09264898e+00 7.37573814e+00 9.98901367e-01
-1.44832087e+00 -1.61903381e-01 8.76182675e-01 -1.74894944e-01
-1.06658840e+00 -6.01419900e-03 -3.72055501e-01 3.69731545e-01
3.80733132e-01 3.28020900e-01 6.55011714e-01 5.67176938e-01
-4.49193753e-02 -2.91403905e-02 -9.41019297e-01 4.91494834e-01
-1.35545149e-01 -1.64283025e+00 4.08844024e-01 1.04633465e-01
1.21288371e+00 -3.66390705e-01 6.47998571e-01 -1.89882830e-01
7.66443372e-01 -1.59501719e+00 1.14399207e+00 6.95290625e-01
1.42944467e+00 -8.21284592e-01 2.40495339e-01 -1.62054926e-01
-9.31252956e-01 3.62141997e-01 -2.29300186e-01 5.80673739e-02
-1.94293335e-01 6.85551703e-01 -7.21709311e-01 4.24077690e-01
4.99584287e-01 3.04070681e-01 -3.93439919e-01 5.58711410e-01
-2.77999282e-01 4.27585870e-01 -3.98446918e-01 1.75366312e-01
1.69917345e-01 -1.07727990e-01 4.88018095e-01 9.12745535e-01
6.47265553e-01 -4.17642832e-01 1.39300287e-01 1.45324731e+00
2.97080372e-02 -2.67809719e-01 -8.88767660e-01 1.27183750e-01
5.61357081e-01 1.07756686e+00 -1.02314198e+00 -8.51736888e-02
1.30964547e-01 1.07051861e+00 1.55220345e-01 5.53174138e-01
-7.35593200e-01 -4.88999873e-01 7.54110694e-01 1.96424410e-01
4.83901501e-01 -3.10670525e-01 -6.50383770e-01 -1.11618507e+00
1.59968749e-01 -1.00026250e+00 -5.56016147e-01 -6.93798482e-01
-1.02954710e+00 8.84548843e-01 -3.04802507e-01 -1.35543907e+00
-3.22865337e-01 -3.85367721e-01 -5.58752596e-01 1.18118072e+00
-1.12227440e+00 -1.67328477e+00 -2.61917800e-01 3.14296842e-01
2.01222062e-01 -1.64970994e-01 1.05387187e+00 -1.20196812e-01
-1.26074120e-01 7.55735934e-01 2.69677728e-01 -5.02013303e-02
6.26026452e-01 -1.26547706e+00 9.73134995e-01 8.79845858e-01
8.05518106e-02 4.28770304e-01 8.20897520e-01 -5.51029384e-01
-1.25177324e+00 -1.23261988e+00 4.03637022e-01 -5.19256771e-01
4.13754046e-01 -8.24116647e-01 -5.04698217e-01 5.79615057e-01
2.42501438e-01 -1.20418899e-01 1.68798774e-01 -2.35623330e-01
-5.33715248e-01 8.09816867e-02 -1.21349704e+00 9.53722537e-01
1.00949037e+00 -4.34113801e-01 8.85512158e-02 8.14956352e-02
7.21288264e-01 -7.09775209e-01 -9.34592962e-01 4.51942682e-01
8.21484864e-01 -1.20332193e+00 8.20899725e-01 -2.51915038e-01
8.04763317e-01 -3.93019140e-01 -1.54525444e-01 -1.42794418e+00
-2.30743691e-01 -9.24557209e-01 4.36911672e-01 1.08151066e+00
4.45382565e-01 -6.83425367e-01 9.60915923e-01 1.86200023e-01
-6.70955256e-02 -5.99551499e-01 -8.61713588e-01 -7.43183732e-01
2.67122954e-01 5.48771396e-03 1.15010333e+00 8.65425169e-01
-6.34937465e-01 -2.15207100e-01 -7.28567898e-01 -2.48832554e-02
5.32405674e-01 4.12915468e-01 1.04251790e+00 -8.60591888e-01
-3.60456347e-01 -5.73678136e-01 -2.83670187e-01 -8.73156726e-01
-1.53746545e-01 -7.07680464e-01 1.97377905e-01 -1.10460305e+00
-2.69994795e-01 -9.01080489e-01 2.39903286e-01 6.35539591e-01
2.27062121e-01 8.70215833e-01 1.93281788e-02 2.81582385e-01
-4.57680784e-02 4.77668166e-01 1.77375519e+00 7.63279013e-03
-7.22272843e-02 -3.07678908e-01 -5.66872358e-01 3.56795996e-01
7.45034158e-01 -3.11347693e-01 -4.26892847e-01 -5.59584260e-01
2.96615720e-01 -6.07506000e-02 7.32083559e-01 -1.16671073e+00
-2.52895474e-01 -3.47271621e-01 6.05578423e-01 -4.87370528e-02
3.70204270e-01 -6.64657295e-01 7.84157276e-01 2.06664756e-01
-5.16485572e-01 -5.73627874e-02 6.86271563e-02 2.36072987e-01
-1.63114667e-01 1.46442935e-01 7.84618616e-01 -1.96749404e-01
-3.08361709e-01 2.65498638e-01 -4.10960048e-01 -2.05204710e-02
8.25578749e-01 -4.53521729e-01 -3.46251994e-01 -4.24825251e-01
-6.01307094e-01 -2.92890787e-01 1.20677209e+00 4.46796775e-01
3.49546909e-01 -1.75722420e+00 -7.37135589e-01 7.14750290e-01
6.69175535e-02 6.87587634e-02 -7.87924528e-02 2.69195676e-01
-1.00591791e+00 2.33846568e-02 -5.83042145e-01 -6.04441762e-01
-7.11289823e-01 1.96992382e-01 4.76647645e-01 -1.67040378e-01
-6.94689751e-01 6.95654631e-01 2.97898442e-01 -4.78667378e-01
-1.83180312e-03 -4.00737047e-01 3.92327309e-01 -3.13109875e-01
1.60056055e-01 -4.27805223e-02 6.35079145e-02 -2.73107320e-01
2.07543865e-01 6.32317662e-01 4.27242249e-01 -4.88721550e-01
1.16416681e+00 1.09302022e-01 -1.96564063e-01 3.18050295e-01
1.11093092e+00 3.36481720e-01 -1.71214509e+00 7.51958862e-02
-4.68209684e-01 -5.97327471e-01 -3.34952205e-01 -7.40028262e-01
-1.13149130e+00 6.01061344e-01 4.34171319e-01 3.77342969e-01
1.02997744e+00 -3.34052920e-01 5.18162370e-01 9.11749229e-02
5.56548893e-01 -8.62759411e-01 1.88282445e-01 4.38193053e-01
1.01907492e+00 -9.08516407e-01 -2.26242274e-01 -3.89048070e-01
-4.34655875e-01 1.08512390e+00 2.75976002e-01 -6.00375891e-01
2.55678624e-01 6.45272315e-01 2.27016121e-01 -2.60046497e-02
-6.32814288e-01 1.43579036e-01 3.53326470e-01 6.17061555e-01
3.86134803e-01 2.34874070e-01 2.11209357e-01 -2.30832979e-01
-7.90480912e-01 -2.85051852e-01 4.79290247e-01 7.76553333e-01
-3.50140601e-01 -1.36336124e+00 -5.24073541e-01 2.94190764e-01
-3.44345719e-01 -1.50043920e-01 -4.18828338e-01 8.89982760e-01
2.62155056e-01 4.65383589e-01 4.07497108e-01 -4.41620916e-01
1.95444115e-02 -5.16743958e-02 6.53413892e-01 -4.66350377e-01
-5.37820697e-01 9.57354158e-02 4.33959700e-02 -5.93048453e-01
-2.13306129e-01 -4.10688519e-01 -5.52811146e-01 -5.38499236e-01
-3.71936522e-02 -9.92007256e-02 5.30017078e-01 5.78854918e-01
2.99540937e-01 5.94960928e-01 7.59135842e-01 -1.03009915e+00
-1.26559973e-01 -6.72423482e-01 -6.42820537e-01 5.03240407e-01
4.85950381e-01 -6.62155092e-01 -2.43938744e-01 1.93180278e-01]
|
[11.593559265136719, -0.4848795533180237]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.