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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
54ab50b1-b9ef-47f5-8bcc-320cf527f74a | lb-simtsc-an-efficient-similarity-aware-graph | 2301.04838 | null | https://arxiv.org/abs/2301.04838v2 | https://arxiv.org/pdf/2301.04838v2.pdf | LB-SimTSC: An Efficient Similarity-Aware Graph Neural Network for Semi-Supervised Time Series Classification | Time series classification is an important data mining task that has received a lot of interest in the past two decades. Due to the label scarcity in practice, semi-supervised time series classification with only a few labeled samples has become popular. Recently, Similarity-aware Time Series Classification (SimTSC) is proposed to address this problem by using a graph neural network classification model on the graph generated from pairwise Dynamic Time Warping (DTW) distance of batch data. It shows excellent accuracy and outperforms state-of-the-art deep learning models in several few-label settings. However, since SimTSC relies on pairwise DTW distances, the quadratic complexity of DTW limits its usability to only reasonably sized datasets. To address this challenge, we propose a new efficient semi-supervised time series classification technique, LB-SimTSC, with a new graph construction module. Instead of using DTW, we propose to utilize a lower bound of DTW, LB_Keogh, to approximate the dissimilarity between instances in linear time, while retaining the relative proximity relationships one would have obtained via computing DTW. We construct the pairwise distance matrix using LB_Keogh and build a graph for the graph neural network. We apply this approach to the ten largest datasets from the well-known UCR time series classification archive. The results demonstrate that this approach can be up to 104x faster than SimTSC when constructing the graph on large datasets without significantly decreasing classification accuracy. | ['Jessica Lin', 'Li Zhang', 'Arnav Jain', 'Wenjie Xi'] | 2023-01-12 | null | null | null | null | ['semi-supervised-time-series-classification', 'dynamic-time-warping'] | ['time-series', 'time-series'] | [ 7.08870739e-02 -3.10472369e-01 -1.86335310e-01 -5.24613619e-01
-4.48361844e-01 -7.18145907e-01 3.92708629e-01 7.68884122e-01
-5.68552732e-01 5.06956398e-01 -3.31544161e-01 -5.51571369e-01
-6.11147046e-01 -8.59017789e-01 -4.78063166e-01 -7.13188052e-01
-9.53124702e-01 4.97077614e-01 -5.18280501e-03 -1.88399807e-01
-3.89860384e-03 6.02284253e-01 -1.44359720e+00 4.01624525e-03
6.55502975e-01 1.32773721e+00 -3.98181140e-01 2.89881945e-01
-1.41092673e-01 5.73571205e-01 -4.79705304e-01 -1.89623162e-01
5.72151363e-01 -4.53201771e-01 -6.39282405e-01 -1.39047533e-01
4.77352679e-01 6.65505975e-02 -5.49790382e-01 8.82339954e-01
4.04917181e-01 4.93591934e-01 4.73441184e-01 -1.84471750e+00
-4.53111440e-01 6.58302844e-01 -6.90098882e-01 4.29941565e-01
5.01454100e-02 -2.52989173e-01 1.29985976e+00 -4.13351566e-01
7.21689939e-01 7.86887705e-01 1.11431050e+00 1.59286529e-01
-1.51571500e+00 -7.57288158e-01 3.77054997e-02 5.63895404e-01
-1.57358682e+00 2.07696542e-01 1.17456460e+00 -2.35287085e-01
1.21751499e+00 3.10384095e-01 7.75629818e-01 8.55076492e-01
4.49591875e-03 3.93391967e-01 1.10962296e+00 -3.02021950e-01
5.24135113e-01 -4.84509408e-01 4.85359371e-01 7.35871911e-01
3.99020612e-02 -1.85884833e-01 -2.90272921e-01 -4.41398591e-01
3.53308618e-01 3.01859826e-01 -1.71265692e-01 -6.12874568e-01
-1.13127875e+00 1.00998855e+00 5.90765059e-01 5.44767797e-01
-1.46801114e-01 1.51407436e-01 1.02662122e+00 9.07487750e-01
1.09147775e+00 3.44391704e-01 -3.95789832e-01 -1.69108048e-01
-9.93806362e-01 -3.64157488e-03 9.17514801e-01 6.72463357e-01
5.72200596e-01 2.89524738e-02 4.16838765e-01 9.39864218e-01
-1.75206974e-01 1.40905514e-01 7.05759883e-01 -6.83004439e-01
5.54334462e-01 7.74088979e-01 -5.86407363e-01 -1.40197194e+00
-7.47532308e-01 -5.43914855e-01 -1.18461847e+00 -1.45910129e-01
5.75361252e-01 6.05319887e-02 -7.18556106e-01 1.80463207e+00
4.82573241e-01 2.83958524e-01 -4.42891091e-01 7.73730695e-01
3.91381264e-01 7.41465330e-01 -3.41049552e-01 -5.51938117e-01
9.58796978e-01 -6.55632257e-01 -4.73788083e-01 2.04820380e-01
1.22864604e+00 -2.14324832e-01 8.74571919e-01 3.55163664e-01
-3.85581672e-01 -3.15213472e-01 -1.20244968e+00 4.96639423e-02
-7.02791691e-01 -2.93407768e-01 7.79722571e-01 4.93214935e-01
-1.17657340e+00 1.22433281e+00 -1.03713655e+00 -6.44605517e-01
4.07904685e-01 4.35751975e-01 -5.77881396e-01 -1.43528983e-01
-1.20511699e+00 4.68440711e-01 4.77141619e-01 -6.76528737e-02
-3.29836220e-01 -6.83874369e-01 -7.45651722e-01 -9.89134088e-02
4.41541940e-01 -1.35797575e-01 6.17145061e-01 -9.31730449e-01
-1.02777934e+00 8.09963703e-01 3.01442772e-01 -8.21908891e-01
5.08727551e-01 3.61463815e-01 -5.40850282e-01 2.03667849e-01
2.30663386e-03 2.89454728e-01 7.62026727e-01 -6.83412492e-01
-2.11128876e-01 -6.69773042e-01 -1.68045714e-01 -1.59475774e-01
-7.74401188e-01 -2.03828886e-01 -1.14563011e-01 -8.87678862e-01
2.46764153e-01 -1.24335599e+00 -2.82885373e-01 1.74361199e-01
-2.45796159e-01 -6.24520481e-01 1.02644944e+00 -4.00014639e-01
1.29196608e+00 -2.16937208e+00 1.14843912e-01 5.91243565e-01
6.03804290e-01 1.50884360e-01 -4.75152522e-01 7.80054569e-01
-4.37421352e-01 -1.34257734e-01 -5.86578369e-01 -4.15417314e-01
-4.47167866e-02 3.65658373e-01 -3.11360329e-01 8.61272097e-01
-1.73232630e-01 6.44576967e-01 -1.06274068e+00 -2.89751470e-01
1.83816254e-02 2.63381273e-01 -2.72930503e-01 -1.16556235e-01
-3.97793017e-02 -1.28402948e-01 -1.22699484e-01 4.31894064e-01
5.96547008e-01 -5.76205552e-01 4.99076456e-01 -2.62026191e-01
8.54831412e-02 2.46890523e-02 -8.58914793e-01 1.81496406e+00
-2.00639725e-01 8.24386418e-01 -5.58029115e-01 -1.81985521e+00
1.15958726e+00 4.86348346e-02 1.16346121e+00 -7.13311493e-01
2.59736210e-01 3.47669005e-01 1.47677287e-01 -1.15629539e-01
1.38413459e-01 2.56098751e-02 -4.35009785e-02 9.05341029e-01
5.08048274e-02 1.58034518e-01 4.86613870e-01 3.14925879e-01
1.49789047e+00 -1.41119316e-01 2.83920765e-01 -3.05719286e-01
1.79899499e-01 1.67555839e-01 4.74720746e-01 3.67063016e-01
-3.10816407e-01 4.46759939e-01 5.61254799e-01 -9.90220726e-01
-9.24597979e-01 -7.53755212e-01 1.25381425e-01 1.15832376e+00
-2.37294778e-01 -7.42360353e-01 -7.16693223e-01 -9.85495031e-01
2.90748626e-01 4.02772337e-01 -8.10868979e-01 -2.84581810e-01
-5.61013460e-01 -8.34923089e-01 6.61048412e-01 4.69677866e-01
2.65477240e-01 -7.31856763e-01 -2.72805274e-01 2.46839702e-01
-3.17138396e-02 -1.03230762e+00 -6.00868344e-01 4.14698452e-01
-1.17312467e+00 -1.05450773e+00 -6.26629949e-01 -6.47213280e-01
6.37397408e-01 4.33285892e-01 8.90785098e-01 1.46353513e-01
-4.41516191e-01 2.44055182e-01 -6.03934050e-01 -3.37978676e-02
-1.54347092e-01 2.57841796e-01 2.50330448e-01 2.70628303e-01
5.93194187e-01 -1.08477080e+00 -4.36160207e-01 1.91190720e-01
-9.18346286e-01 -5.53100035e-02 -1.00805350e-01 7.39540398e-01
6.59848273e-01 2.89608926e-01 7.62999058e-01 -7.01107919e-01
6.69120193e-01 -5.74758410e-01 -7.68043280e-01 1.15416944e-01
-8.66474271e-01 -9.96945128e-02 1.09498131e+00 -6.80261672e-01
1.42824452e-03 2.92680170e-02 3.11844379e-01 -8.63624930e-01
2.35898957e-01 9.37812865e-01 4.96144503e-01 -2.40347400e-01
6.40891016e-01 2.28798613e-01 2.87759036e-01 -4.44100529e-01
2.36302152e-01 4.98098880e-01 2.28479013e-01 -3.13843787e-01
7.11793184e-01 6.25153065e-01 4.20731485e-01 -6.64887726e-01
-8.28737199e-01 -4.60424602e-01 -7.62052715e-01 -3.25141937e-01
5.13502836e-01 -6.17796838e-01 -7.33743250e-01 4.28432971e-01
-7.18190074e-01 -4.56708848e-01 -2.45610446e-01 6.10599220e-01
-3.72169465e-01 4.63611484e-01 -5.93835592e-01 -5.39055288e-01
-4.58185732e-01 -6.06160879e-01 7.29748666e-01 -5.41275620e-01
-2.12599263e-01 -1.32654345e+00 1.80430338e-01 1.13887347e-01
3.11383158e-01 9.10843849e-01 1.10496914e+00 -1.17046511e+00
-1.12746626e-01 -4.96474922e-01 -2.23255113e-01 3.39414001e-01
3.38518023e-01 -2.27358520e-01 -6.25232935e-01 -8.02027404e-01
-4.61205319e-02 -2.77370006e-01 7.61560738e-01 1.56322211e-01
1.57323217e+00 -8.78954828e-02 -1.61418691e-01 7.30121851e-01
1.40565121e+00 1.88619629e-01 3.49573456e-02 3.89562428e-01
1.02684891e+00 5.64457536e-01 4.52537805e-01 6.14656150e-01
3.68413538e-01 5.59145510e-01 2.75381684e-01 5.74324131e-02
9.87223983e-02 -2.55192742e-02 2.01790005e-01 1.45945919e+00
-2.94377238e-01 -1.76028028e-01 -1.25816965e+00 3.93792599e-01
-2.00784993e+00 -8.25021267e-01 -2.09949195e-01 2.29231620e+00
6.55103087e-01 1.82159260e-01 1.95844039e-01 6.11255765e-01
6.90918207e-01 4.22225058e-01 -7.77144730e-01 -3.15461457e-01
-3.27405706e-02 2.96795756e-01 6.87036872e-01 -3.67893279e-02
-1.12925518e+00 4.56951320e-01 5.33016491e+00 9.32494283e-01
-1.44626582e+00 1.14871375e-01 6.08545363e-01 -1.81647435e-01
3.46355577e-04 -1.32117629e-01 -6.13278858e-02 3.93334091e-01
1.18627334e+00 -5.26380837e-01 7.32228935e-01 7.59965897e-01
2.03795075e-01 3.67491812e-01 -1.41178632e+00 1.28569567e+00
1.63943246e-01 -1.20288217e+00 -6.28729761e-02 1.51509300e-01
5.46772420e-01 3.79617304e-01 -1.44631311e-01 3.78786087e-01
1.78811193e-01 -6.89940572e-01 3.81445020e-01 6.97469935e-02
8.68115544e-01 -9.33520675e-01 4.53190207e-01 3.08409423e-01
-1.59536099e+00 -1.69616282e-01 -2.49785170e-01 -5.86758628e-02
-1.21297508e-01 9.36341345e-01 -7.11762786e-01 8.18504155e-01
7.12752581e-01 1.22447610e+00 -6.33172214e-01 8.14622343e-01
4.20248270e-01 7.05231547e-01 -5.38852811e-01 -4.85633090e-02
4.26085651e-01 -4.51439738e-01 3.73727202e-01 8.83902311e-01
5.69925249e-01 -7.25126192e-02 4.55945373e-01 4.09792483e-01
-3.00021291e-01 3.44571143e-01 -8.11011136e-01 -4.23968464e-01
5.83919346e-01 1.30011332e+00 -1.19137239e+00 -3.58135968e-01
-4.08632129e-01 9.23832119e-01 6.46478236e-01 6.55219182e-02
-9.08650100e-01 -6.69381738e-01 2.60280222e-01 -5.90386577e-02
5.92699908e-02 -5.50991297e-01 3.67161669e-02 -1.08651149e+00
9.13415104e-02 -9.94139373e-01 8.47078383e-01 -4.60004508e-01
-1.64976919e+00 7.57808387e-01 1.18056968e-01 -1.68251932e+00
-2.49090686e-01 -5.30019701e-01 -4.14505839e-01 4.01149601e-01
-1.15413058e+00 -8.72501373e-01 -2.81287313e-01 7.61495531e-01
3.72085303e-01 -1.99345037e-01 7.81082213e-01 5.81209421e-01
-6.70726120e-01 6.91699862e-01 4.12824571e-01 2.43223816e-01
6.21796906e-01 -1.22989082e+00 6.60519004e-01 4.86682326e-01
3.84964824e-01 4.14016515e-01 4.54082549e-01 -5.02731681e-01
-1.61186767e+00 -1.38461602e+00 8.00547600e-01 3.32005247e-02
1.18065107e+00 -6.16755247e-01 -1.19085920e+00 4.76260722e-01
-1.11268938e-01 5.25028050e-01 8.85441661e-01 7.73236975e-02
-8.36880803e-01 -4.98001575e-01 -1.05735767e+00 4.18517441e-01
1.43644214e+00 -8.99103761e-01 -2.79424012e-01 6.21329486e-01
8.10253620e-01 -6.49021268e-02 -1.17361951e+00 2.33668670e-01
3.78211856e-01 -7.22146273e-01 7.04785883e-01 -5.81174910e-01
6.13224991e-02 -1.77245304e-01 -1.33657500e-01 -1.54395664e+00
-1.00168966e-01 -6.94519043e-01 -1.20338157e-01 8.10425520e-01
2.80353904e-01 -9.89305615e-01 6.48419201e-01 4.55509067e-01
-3.20968740e-02 -8.04743111e-01 -1.13626444e+00 -1.35146594e+00
-9.98364836e-02 -5.73003769e-01 6.95019245e-01 1.75444841e+00
1.79066762e-01 1.29880786e-01 -2.17443004e-01 -1.68941408e-01
8.49396408e-01 4.86218184e-01 5.29582620e-01 -1.65580511e+00
-1.15398705e-01 -3.19560558e-01 -7.16966927e-01 -3.86646837e-01
3.68427634e-01 -1.53529656e+00 -3.46814603e-01 -1.24593818e+00
-1.34989381e-01 -4.56837416e-01 -6.88903689e-01 6.99090123e-01
3.64352077e-01 3.19499373e-01 2.73073111e-02 3.38745654e-01
-4.89386141e-01 4.45622057e-01 9.61493492e-01 -2.41566718e-01
-2.17944369e-01 -3.57112139e-01 -9.66852829e-02 4.47315454e-01
8.90664160e-01 -6.82787418e-01 -6.37718916e-01 -3.59007597e-01
1.65123969e-01 7.21012577e-02 1.50589824e-01 -9.90151942e-01
3.95287603e-01 8.21138173e-02 8.50264803e-02 -5.88921309e-01
1.89022109e-01 -8.97583842e-01 2.60847390e-01 5.01749337e-01
-5.49059927e-01 6.03898048e-01 4.48884722e-03 7.38158643e-01
-2.59799212e-01 2.45091528e-01 5.32983005e-01 2.65072554e-01
-5.14035404e-01 7.61767268e-01 -7.56477043e-02 -9.25329328e-02
1.02517009e+00 2.22237478e-03 -2.38166839e-01 -2.77534723e-01
-5.71819067e-01 1.69717342e-01 1.99637190e-01 5.09395659e-01
5.12561202e-01 -1.74012673e+00 -4.28522587e-01 1.70371205e-01
3.68756175e-01 -2.57155180e-01 1.06414899e-01 9.73800480e-01
-4.71377969e-01 1.23580188e-01 -2.50340372e-01 -6.69661701e-01
-1.36945176e+00 8.41330409e-01 1.14128366e-01 -3.66285324e-01
-9.21348512e-01 3.56060356e-01 -4.31767106e-01 -4.58460331e-01
1.95923924e-01 -4.05618727e-01 -7.12153614e-02 4.66782898e-01
2.37309664e-01 5.41975975e-01 4.06652838e-01 -5.50088823e-01
-3.92623484e-01 5.77471972e-01 -3.70710716e-02 1.54052049e-01
1.61576462e+00 1.36536524e-01 -3.37753206e-01 6.87906682e-01
1.95055866e+00 -5.52978694e-01 -8.21088910e-01 -4.38384384e-01
3.20782840e-01 -3.70619565e-01 1.93417117e-01 -1.46642730e-01
-1.39197600e+00 7.72496998e-01 6.64081812e-01 9.45803881e-01
1.30135119e+00 -3.59432995e-01 9.84775662e-01 8.28719199e-01
5.19846737e-01 -1.20300400e+00 3.06743011e-02 4.93102193e-01
7.39204764e-01 -1.09117782e+00 1.14870541e-01 -1.43825307e-01
-2.47418240e-01 1.23347521e+00 2.73883760e-01 -4.28764343e-01
9.28855836e-01 -2.23164633e-02 -7.99976662e-02 -3.63179266e-01
-8.61521840e-01 -5.08158356e-02 3.64816576e-01 3.70027393e-01
3.90494287e-01 1.55825779e-01 -4.02574807e-01 -2.31072977e-02
-1.77669406e-01 -1.72162667e-01 2.23105207e-01 8.79579484e-01
-1.55294109e-02 -1.06117845e+00 1.47210017e-01 8.25751722e-01
-2.72146672e-01 9.82894674e-02 -4.75869030e-01 8.33051622e-01
-1.85618341e-01 8.42578411e-01 2.97681004e-01 -7.46887088e-01
1.27780750e-01 5.70116788e-02 2.44075477e-01 -1.85154915e-01
-5.79071879e-01 -1.61020800e-01 1.04243830e-01 -7.42889524e-01
-5.97879708e-01 -5.03467321e-01 -1.22002387e+00 -4.99178439e-01
-2.17791960e-01 1.73973322e-01 6.94157481e-01 8.60396504e-01
4.66580063e-01 2.49911636e-01 1.16513479e+00 -8.23795795e-01
-5.71567237e-01 -7.48971045e-01 -8.21064830e-01 7.61921763e-01
2.71093875e-01 -5.95552862e-01 -6.07441902e-01 -2.73414731e-01] | [7.312307357788086, 3.432619094848633] |
35c43d49-1573-4671-970f-42dbd8c1b76f | pointresnet-residual-network-for-3d-point | 2211.11040 | null | https://arxiv.org/abs/2211.11040v1 | https://arxiv.org/pdf/2211.11040v1.pdf | PointResNet: Residual Network for 3D Point Cloud Segmentation and Classification | Point cloud segmentation and classification are some of the primary tasks in 3D computer vision with applications ranging from augmented reality to robotics. However, processing point clouds using deep learning-based algorithms is quite challenging due to the irregular point formats. Voxelization or 3D grid-based representation are different ways of applying deep neural networks to this problem. In this paper, we propose PointResNet, a residual block-based approach. Our model directly processes the 3D points, using a deep neural network for the segmentation and classification tasks. The main components of the architecture are: 1) residual blocks and 2) multi-layered perceptron (MLP). We show that it preserves profound features and structural information, which are useful for segmentation and classification tasks. The experimental evaluations demonstrate that the proposed model produces the best results for segmentation and comparable results for classification in comparison to the conventional baselines. | ['Shanmuganathan Raman', 'Seema Kumari', 'Saagar Parikh', 'Aadesh Desai'] | 2022-11-20 | null | null | null | null | ['point-cloud-segmentation'] | ['computer-vision'] | [ 2.99755111e-03 -1.73949525e-01 4.44057281e-04 -3.84015977e-01
-5.19827068e-01 -2.18861938e-01 5.74607491e-01 3.03023588e-02
-4.01304364e-01 1.45873874e-01 -4.24134672e-01 -4.97851372e-01
1.89510345e-01 -8.92429471e-01 -8.90683532e-01 -6.09746933e-01
-2.22230971e-01 7.74396718e-01 3.72627556e-01 -9.64373350e-02
5.96467018e-01 1.42366314e+00 -1.74097180e+00 2.43476316e-01
7.33155012e-01 1.35268116e+00 3.44616532e-01 4.88654912e-01
-6.50404513e-01 3.18497479e-01 -4.77315754e-01 9.55610573e-02
7.93600202e-01 4.11163539e-01 -6.89488947e-01 1.78867981e-01
6.54859304e-01 -5.47281265e-01 -9.89512075e-03 1.15688717e+00
3.32647532e-01 2.37821341e-01 7.20066905e-01 -1.10711682e+00
-4.73507583e-01 1.54629737e-01 -9.79777634e-01 -1.87932685e-01
8.74426365e-02 -5.37914596e-02 6.74172223e-01 -1.25924635e+00
2.66272038e-01 1.49018598e+00 1.01020563e+00 3.48012269e-01
-9.52126265e-01 -3.96546006e-01 1.33987293e-01 1.15535647e-01
-1.27083659e+00 4.70606349e-02 9.63857472e-01 -6.62260532e-01
1.16406631e+00 1.78832188e-01 8.36742938e-01 4.01632667e-01
2.19729900e-01 8.97368014e-01 1.12436187e+00 -3.24654669e-01
3.82590652e-01 -1.76582471e-01 5.49489498e-01 4.85790461e-01
3.07167973e-02 1.25203595e-01 1.28305674e-01 3.04863099e-02
1.25064564e+00 4.14836705e-01 -7.23080011e-03 -4.75234121e-01
-1.10384858e+00 7.39443719e-01 1.12232208e+00 1.18380271e-01
-6.74783111e-01 2.64809519e-01 1.64576411e-01 -9.52494796e-03
7.22393095e-01 2.53889263e-01 -5.27788639e-01 3.69500875e-01
-1.03668880e+00 3.93308908e-01 3.80921692e-01 1.13993824e+00
9.35497940e-01 3.58621180e-02 2.11857960e-01 9.20635760e-01
5.65994203e-01 4.00390983e-01 3.18373054e-01 -1.02286947e+00
2.37693787e-01 8.86276066e-01 -7.05844685e-02 -9.06962514e-01
-6.68195128e-01 -2.87873656e-01 -1.11161602e+00 1.03288388e+00
-4.69977409e-02 1.88719526e-01 -1.54165232e+00 8.88141692e-01
4.01422232e-01 4.90377188e-01 -1.12122014e-01 9.49137807e-01
1.25000584e+00 7.55637348e-01 -8.19399655e-02 3.19305003e-01
1.14420474e+00 -8.75287533e-01 -1.37692630e-01 -2.41749510e-01
3.49082083e-01 -6.42833471e-01 8.87133479e-01 3.14620197e-01
-1.33533859e+00 -9.36906040e-01 -1.12691176e+00 -5.09230375e-01
-5.01393497e-01 -1.02064721e-01 6.17945015e-01 3.27526599e-01
-1.19721150e+00 8.01398277e-01 -1.12129843e+00 -1.68350309e-01
8.74664068e-01 6.54869437e-01 -1.23128936e-01 2.02263907e-01
-4.39526498e-01 7.19024956e-01 3.60125780e-01 3.02074879e-01
-2.31626406e-01 -7.13836730e-01 -8.34322453e-01 1.24951161e-01
-1.74748838e-01 -8.12547147e-01 1.40436697e+00 -5.08371532e-01
-1.50569034e+00 1.12708545e+00 -1.18342750e-01 -5.33455849e-01
4.58788246e-01 -3.27641666e-01 4.86457914e-01 7.94179961e-02
9.11964104e-02 1.16149306e+00 7.62548625e-01 -1.58151865e+00
-9.52811241e-01 -7.11388171e-01 -4.43608388e-02 3.00313413e-01
3.55370700e-01 -3.29223722e-01 -6.26206398e-01 -3.61548513e-01
7.87015736e-01 -8.53871226e-01 -6.79286659e-01 9.80173200e-02
-4.20002669e-01 -5.08907914e-01 9.83673453e-01 -4.58359629e-01
1.94787532e-01 -2.15076685e+00 -1.48267701e-01 4.02718276e-01
3.68135184e-01 2.03929111e-01 7.25734308e-02 -7.11624250e-02
-6.27506226e-02 1.69922367e-01 -3.76826346e-01 -7.10243464e-01
-1.38665428e-02 1.76092923e-01 -2.94286489e-01 4.78279799e-01
1.19224131e-01 1.09190214e+00 -4.32840198e-01 -4.23482239e-01
1.00730026e+00 5.34898341e-01 -4.27920938e-01 1.91627201e-02
-3.62509519e-01 4.87789541e-01 -2.77969331e-01 8.32514644e-01
1.28042340e+00 -1.60530627e-01 -4.96355176e-01 -2.44470835e-01
-2.32950360e-01 1.82470188e-01 -1.07612896e+00 2.02229238e+00
-4.21870112e-01 5.05636871e-01 1.09211057e-01 -8.02145898e-01
1.10247481e+00 -3.20412926e-02 6.54299974e-01 -5.76303840e-01
3.05660933e-01 2.50714689e-01 -3.32813561e-01 -1.54414609e-01
6.64108574e-01 -7.98342526e-02 1.04944944e-01 2.04822533e-02
-1.33256868e-01 -7.59797156e-01 -2.59669095e-01 -3.30346912e-01
7.77055740e-01 2.92111099e-01 5.47773056e-02 -2.31678300e-02
4.29472476e-01 3.95827264e-01 4.32441622e-01 7.22968280e-01
-1.27575472e-01 9.42616761e-01 5.78789562e-02 -6.38446748e-01
-1.16928351e+00 -1.01278305e+00 -3.14525783e-01 4.77860451e-01
5.62584877e-01 2.33829483e-01 -5.87178707e-01 -3.02962065e-01
3.78410518e-01 6.82744086e-01 -2.49414936e-01 2.27271155e-01
-6.69235647e-01 -4.30175185e-01 2.11628482e-01 7.92177022e-01
7.94490159e-01 -1.09060454e+00 -7.19367027e-01 4.49831644e-03
3.38597298e-01 -1.18028581e+00 1.88902199e-01 3.37954581e-01
-1.55192256e+00 -8.04602444e-01 -6.54913425e-01 -1.06782544e+00
8.01583707e-01 6.86881959e-01 1.17834246e+00 1.46405101e-01
7.33073652e-02 8.60960260e-02 -1.18611239e-01 -7.62943566e-01
-1.00912087e-01 1.53658241e-01 -1.97781593e-01 -5.05308628e-01
7.16405571e-01 -6.90064430e-01 -5.72222710e-01 3.55642512e-02
-7.11227179e-01 2.26622358e-01 7.68207848e-01 5.47800720e-01
1.05795014e+00 -1.12708382e-01 -5.20215556e-03 -7.68572032e-01
5.35955429e-01 -2.46591456e-02 -8.73972118e-01 -2.98981279e-01
-2.83717930e-01 -3.51853937e-01 3.47802550e-01 1.83694392e-01
-6.72449946e-01 4.02130246e-01 -6.70584500e-01 -9.22859251e-01
-8.89578998e-01 2.90408760e-01 -1.03669688e-01 -4.69233125e-01
4.49384153e-01 2.24292912e-02 2.79369131e-02 -8.00918043e-01
4.05599415e-01 7.73364365e-01 5.18835664e-01 -2.46471256e-01
7.44086981e-01 7.34399796e-01 2.09515944e-01 -1.08062744e+00
-4.62431103e-01 -7.22775817e-01 -1.21962667e+00 -7.46516138e-02
1.02955246e+00 -9.22345936e-01 -8.57535541e-01 6.63797081e-01
-1.67311299e+00 -2.24575981e-01 -4.98008311e-01 3.82789910e-01
-6.59379065e-01 3.35354567e-01 -8.48719060e-01 -6.88964665e-01
-5.55362642e-01 -1.43652320e+00 1.47467744e+00 2.78025091e-01
2.62682348e-01 -8.01357746e-01 3.13467234e-02 3.22144121e-01
3.57991681e-02 4.54388440e-01 9.73672330e-01 -5.68923414e-01
-9.67402875e-01 -2.87828237e-01 -5.15660763e-01 4.89489019e-01
-8.99344981e-02 -2.75946409e-03 -1.10961223e+00 -6.81506693e-02
1.13223679e-01 -5.51545806e-02 8.33665371e-01 7.67169118e-01
1.64542651e+00 1.49989724e-01 -4.58246768e-01 9.01041806e-01
1.54029405e+00 1.70821518e-01 6.54325664e-01 3.61541748e-01
1.05373609e+00 3.74118596e-01 3.89900714e-01 8.06188136e-02
3.86767954e-01 3.11491668e-01 7.89726317e-01 -6.02328360e-01
-4.03902382e-02 7.49489143e-02 -2.33837649e-01 8.18859220e-01
-2.76761442e-01 2.97552824e-01 -1.30203569e+00 4.85875577e-01
-1.85563517e+00 -7.44780898e-01 -5.20376265e-01 1.86077094e+00
3.02977055e-01 3.17704618e-01 -1.74262464e-01 2.86772192e-01
5.92193186e-01 -1.27106160e-01 -5.71288884e-01 -5.79515874e-01
1.56282634e-01 5.57401538e-01 7.19202876e-01 4.07122105e-01
-1.37525511e+00 1.06856096e+00 6.19912958e+00 6.16790175e-01
-1.29637158e+00 -8.26292783e-02 4.92850572e-01 1.83960304e-01
2.54190624e-01 -4.61427569e-01 -6.37075484e-01 1.94408074e-01
4.51728016e-01 3.72163832e-01 -4.00205106e-02 1.16511250e+00
2.46099815e-01 1.20101753e-03 -1.26712227e+00 1.40813923e+00
-9.84017253e-02 -1.58554864e+00 2.64725685e-01 1.44158721e-01
7.75815666e-01 7.31820285e-01 -6.46648407e-02 2.35829726e-01
3.26024264e-01 -1.04799545e+00 7.72890687e-01 5.01501322e-01
4.28130060e-01 -9.01965201e-01 8.82015526e-01 4.65442568e-01
-9.68577683e-01 2.56222844e-01 -6.80598378e-01 -1.79495543e-01
6.90079033e-02 8.18935096e-01 -9.80367661e-01 5.03261268e-01
9.65817750e-01 7.96566665e-01 -3.77980769e-01 1.42782283e+00
-1.16906680e-01 2.09965050e-01 -6.43773735e-01 8.00002590e-02
4.83728528e-01 -5.88960469e-01 4.85955656e-01 1.05602312e+00
2.13578045e-01 6.58835620e-02 4.34645534e-01 1.01076031e+00
-2.17159227e-01 -7.72990286e-02 -8.68697882e-01 5.58674932e-01
2.83727676e-01 1.13789880e+00 -9.86405134e-01 -3.75379592e-01
-5.32958448e-01 7.27540016e-01 1.75588131e-01 2.87469059e-01
-3.85047644e-01 -4.83303934e-01 7.70427048e-01 1.45631935e-02
4.30636883e-01 -6.98099613e-01 -1.20988631e+00 -7.31103301e-01
6.23976551e-02 -3.83532405e-01 -1.64159760e-01 -9.32672024e-01
-1.39328516e+00 6.02315247e-01 -5.19292429e-02 -1.35703444e+00
-2.19441690e-02 -9.97928262e-01 -5.69955349e-01 8.93046498e-01
-1.92332995e+00 -1.31383824e+00 -6.32938504e-01 3.76509368e-01
7.98160791e-01 8.82783383e-02 4.78078008e-01 2.94973910e-01
-2.65326351e-01 -5.36886714e-02 2.10459158e-01 2.86143750e-01
-3.15542240e-03 -1.51983356e+00 9.98033524e-01 7.56959617e-01
1.82518005e-01 4.24373060e-01 2.25919098e-01 -6.04968309e-01
-1.31070316e+00 -1.24515510e+00 3.57267678e-01 -2.87047595e-01
1.54578060e-01 -4.66734797e-01 -1.10009992e+00 7.40427077e-01
6.56777946e-03 -1.15425009e-02 4.77699310e-01 -2.32543144e-02
2.47846246e-01 1.29155561e-01 -1.29771101e+00 4.98119384e-01
1.00117970e+00 -2.75147289e-01 -9.31143939e-01 3.04764658e-01
7.84542739e-01 -8.45474958e-01 -7.00915039e-01 6.60323083e-01
1.95190966e-01 -1.10973740e+00 1.27541137e+00 -2.53286779e-01
3.76894146e-01 -4.06691432e-01 -1.90352112e-01 -1.27459645e+00
-5.00851929e-01 2.22177599e-02 -8.71685520e-03 6.15642428e-01
2.50392228e-01 -5.05478621e-01 1.31193960e+00 4.99499083e-01
-7.29436457e-01 -7.30604172e-01 -1.03714967e+00 -4.80336964e-01
3.43680412e-01 -7.99767494e-01 8.76931787e-01 7.80839741e-01
-6.96545660e-01 2.30968818e-01 3.30296338e-01 5.19633889e-01
8.24687481e-01 4.73983079e-01 1.00287414e+00 -1.74720716e+00
3.58291417e-01 -6.99564576e-01 -7.05962002e-01 -1.56289876e+00
1.21572964e-01 -1.02759778e+00 2.04590887e-01 -2.07229900e+00
-3.77010226e-01 -8.63258600e-01 -1.51954554e-02 3.50332767e-01
2.66961485e-01 4.42776382e-01 3.39855790e-01 4.69114810e-01
-2.23838374e-01 5.46644747e-01 1.22097504e+00 -3.78068805e-01
-5.83891392e-01 3.31818134e-01 -2.49664858e-01 1.12214005e+00
8.30684185e-01 -1.34841040e-01 -1.49798365e-02 -9.10521746e-01
-1.94981784e-01 -2.54979759e-01 5.90494633e-01 -1.25404346e+00
4.68430430e-01 -8.75999592e-03 7.85582483e-01 -1.67018116e+00
6.29709542e-01 -1.16936874e+00 -2.38586187e-01 4.42298204e-01
1.34409398e-01 4.28244919e-02 4.97595370e-01 2.96260923e-01
-2.78452694e-01 -2.38932937e-01 8.00652921e-01 -5.05489111e-01
-9.03580844e-01 5.97153485e-01 -8.92914087e-02 -6.51851654e-01
9.94833171e-01 -7.69735813e-01 1.55162722e-01 2.43823573e-01
-6.80843651e-01 3.36934388e-01 6.01346195e-01 2.85972387e-01
1.12828708e+00 -1.30332601e+00 -5.81261992e-01 4.05129969e-01
-1.85515493e-01 1.05772400e+00 -4.76552024e-02 4.35325354e-01
-1.22154212e+00 4.55083728e-01 -4.05068427e-01 -1.40908456e+00
-1.01121438e+00 4.58590567e-01 4.41686809e-01 9.42897797e-02
-1.00998926e+00 7.50530481e-01 1.27445027e-01 -8.06304336e-01
4.00804847e-01 -9.31827009e-01 -2.01460660e-01 -3.23919266e-01
1.74291551e-01 4.13778335e-01 5.54100871e-01 -5.28497040e-01
-2.66205728e-01 9.55448091e-01 -1.58595026e-01 1.33957550e-01
1.54827738e+00 1.03413850e-01 -2.80299932e-01 4.79995936e-01
1.10250115e+00 -5.68921268e-01 -1.09230328e+00 -1.65231079e-01
-6.58732951e-02 -6.03936076e-01 4.45451379e-01 -3.99652839e-01
-1.11555290e+00 1.28411031e+00 7.74192393e-01 3.27666759e-01
9.26662147e-01 -8.22353289e-02 8.67722154e-01 4.58070397e-01
5.93050182e-01 -8.32985938e-01 -4.90545720e-01 7.07226276e-01
8.90240431e-01 -1.30390596e+00 -5.30074164e-03 -5.51007390e-01
-2.47254372e-02 1.08225822e+00 6.37124062e-01 -6.62339389e-01
8.25314224e-01 1.73455700e-01 2.32488424e-01 -4.28451031e-01
-2.54922420e-01 -2.33989179e-01 2.47875571e-01 7.76190341e-01
5.82824945e-02 -6.03806749e-02 1.19191997e-01 1.84082821e-01
-4.44689542e-01 -8.26142058e-02 2.48421356e-01 9.58917260e-01
-4.90312010e-01 -7.09071994e-01 -7.87811100e-01 6.47571146e-01
-9.77360681e-02 1.57507583e-01 -3.50459278e-01 6.76811755e-01
1.68354467e-01 5.78759074e-01 6.71134889e-01 -3.18006009e-01
5.79972148e-01 -1.38267115e-01 3.91417891e-01 -7.53804564e-01
-7.45412409e-01 4.71925735e-02 -5.73561251e-01 -5.55886626e-01
-4.96039242e-01 -4.50805753e-01 -1.66728044e+00 -4.35223103e-01
-3.69140744e-01 -2.02029750e-01 1.16156328e+00 8.16129863e-01
5.70595026e-01 5.60466826e-01 6.51375592e-01 -1.78880191e+00
-2.58110315e-01 -8.32447350e-01 -4.70262587e-01 2.83088684e-01
4.56630707e-01 -6.36381805e-01 -4.02240902e-02 3.48213576e-02] | [7.951545715332031, -3.4316680431365967] |
6b414684-4784-4a30-8ff6-be55793d9dac | pfns-are-flexible-models-for-real-world | 2305.17535 | null | https://arxiv.org/abs/2305.17535v4 | https://arxiv.org/pdf/2305.17535v4.pdf | PFNs4BO: In-Context Learning for Bayesian Optimization | In this paper, we use Prior-data Fitted Networks (PFNs) as a flexible surrogate for Bayesian Optimization (BO). PFNs are neural processes that are trained to approximate the posterior predictive distribution (PPD) through in-context learning on any prior distribution that can be efficiently sampled from. We describe how this flexibility can be exploited for surrogate modeling in BO. We use PFNs to mimic a naive Gaussian process (GP), an advanced GP, and a Bayesian Neural Network (BNN). In addition, we show how to incorporate further information into the prior, such as allowing hints about the position of optima (user priors), ignoring irrelevant dimensions, and performing non-myopic BO by learning the acquisition function. The flexibility underlying these extensions opens up vast possibilities for using PFNs for BO. We demonstrate the usefulness of PFNs for BO in a large-scale evaluation on artificial GP samples and three different hyperparameter optimization testbeds: HPO-B, Bayesmark, and PD1. We publish code alongside trained models at https://github.com/automl/PFNs4BO. | ['Frank Hutter', 'Noah Hollmann', 'Matthias Feurer', 'Samuel Müller'] | 2023-05-27 | null | null | null | null | ['automl', 'hyperparameter-optimization', 'bayesian-optimization'] | ['methodology', 'methodology', 'methodology'] | [-1.81168213e-01 3.86259347e-01 -9.05699804e-02 -2.95749843e-01
-6.36823356e-01 -4.57695276e-01 7.34847188e-01 -2.51603544e-01
-2.83718169e-01 8.65997076e-01 2.98299551e-01 -2.79949397e-01
-4.18075502e-01 -5.67786098e-01 -1.02298188e+00 -8.28648031e-01
-2.03738526e-01 9.03620481e-01 5.32371178e-02 2.04916209e-01
1.99595079e-01 6.92630112e-01 -1.30167198e+00 -1.42525032e-01
6.99112713e-01 7.63923883e-01 1.78067103e-01 8.76335621e-01
3.77653927e-01 3.19780678e-01 -6.78693116e-01 -3.87470394e-01
3.13281238e-01 8.38119611e-02 -3.95066619e-01 -3.29668105e-01
-1.06110610e-02 -3.05156678e-01 -1.91616312e-01 1.00485694e+00
6.73817277e-01 4.79801923e-01 9.21483815e-01 -1.38224101e+00
-4.60421354e-01 4.41917986e-01 -2.08412230e-01 6.58233166e-02
8.35113227e-02 8.97140861e-01 8.58828306e-01 -6.94719672e-01
2.20803201e-01 1.73188591e+00 9.08760965e-01 4.82208312e-01
-1.86223781e+00 -4.19417977e-01 -8.16079136e-03 -2.52733946e-01
-1.33916521e+00 -6.90943956e-01 3.91462505e-01 -5.01037180e-01
1.01896024e+00 2.12900564e-02 6.49510086e-01 1.64040673e+00
3.34443986e-01 9.21673596e-01 7.53608584e-01 -2.72342116e-01
6.37460291e-01 -3.64339016e-02 1.33413360e-01 4.16955620e-01
1.75965250e-01 6.76535487e-01 -6.38833463e-01 -6.50259912e-01
7.95703173e-01 -3.47847611e-01 -4.56091344e-01 -5.34660399e-01
-9.12045538e-01 7.57369757e-01 1.45697653e-01 -4.79541928e-01
-4.17391151e-01 6.65667892e-01 -1.59017704e-02 -2.20914498e-01
2.94310361e-01 6.98804975e-01 -6.20426714e-01 -4.68635619e-01
-8.26020241e-01 5.96442401e-01 1.40862191e+00 1.12764192e+00
6.33377016e-01 1.06197290e-01 -6.19100451e-01 8.24802518e-01
7.96959341e-01 5.83685398e-01 2.48786956e-01 -1.56247067e+00
7.52394199e-02 -3.04596335e-01 6.31556153e-01 -6.15393817e-01
-4.15167958e-01 -6.02793634e-01 -4.38370436e-01 1.76894978e-01
5.48910141e-01 -4.73557115e-01 -1.11118913e+00 1.80657756e+00
1.93386927e-01 1.58085197e-01 1.36053756e-01 6.90822899e-01
3.94793630e-01 8.35313559e-01 8.02630484e-02 2.31198370e-01
1.12122846e+00 -1.04593170e+00 -3.81609261e-01 -4.25284028e-01
1.47620499e-01 -5.09221256e-01 1.21939623e+00 7.69053578e-01
-1.07508755e+00 -3.49905163e-01 -1.05887580e+00 2.51424283e-01
-2.44548302e-02 2.87696309e-02 6.06654167e-01 7.82636225e-01
-1.28744709e+00 1.30341411e+00 -1.50211549e+00 -3.17314148e-01
4.94614035e-01 5.20624936e-01 2.80882686e-01 9.68840048e-02
-9.61281955e-01 7.46382654e-01 8.39318037e-01 1.06761493e-01
-1.33773363e+00 -8.59449923e-01 -6.13161564e-01 3.60726297e-01
3.40622604e-01 -1.00272536e+00 1.55568945e+00 -4.05660838e-01
-2.13341403e+00 8.86927098e-02 -1.37545288e-01 -6.14021063e-01
4.14382726e-01 -5.52675605e-01 -1.46133184e-01 -6.78307787e-02
-3.32562000e-01 1.04073620e+00 1.05081320e+00 -1.09382594e+00
-2.33303040e-01 2.70622224e-02 -8.83883908e-02 2.34828204e-01
3.46537642e-02 -5.63263260e-02 -6.25637054e-01 -4.27401304e-01
-1.66430086e-01 -1.04367149e+00 -3.42745155e-01 -9.11066756e-02
-5.84107518e-01 -2.26282939e-01 3.05602103e-01 -7.00209439e-01
9.82410491e-01 -2.18504858e+00 1.07040167e-01 3.90549988e-01
3.06236278e-02 9.58482847e-02 -1.24003515e-01 3.14875096e-01
2.56859332e-01 3.50359410e-01 -3.27460289e-01 -6.09423220e-01
6.90695167e-01 6.16667092e-01 -3.66862506e-01 3.50172013e-01
3.41945320e-01 8.07278991e-01 -9.02985632e-01 -1.33847222e-01
1.52978231e-03 6.29535615e-01 -8.76492441e-01 1.93811521e-01
-7.84068108e-01 5.38954914e-01 -2.69822747e-01 5.75678587e-01
5.59026957e-01 -4.11747515e-01 -2.97478022e-04 -7.25591257e-02
1.99537307e-01 3.58269960e-01 -1.29941237e+00 1.52520430e+00
-2.50054270e-01 5.14757633e-01 2.59946615e-01 -4.97753114e-01
7.42912829e-01 1.60752371e-01 -2.79930178e-02 4.90786992e-02
1.05073128e-03 1.92832518e-02 -1.09902378e-02 2.45012203e-03
4.66261327e-01 1.15127191e-01 3.37485671e-01 3.75030935e-01
4.63996857e-01 -4.77079391e-01 3.85191023e-01 -2.73899108e-01
9.14414287e-01 7.68366635e-01 1.30012095e-01 -4.97106373e-01
-8.58574137e-02 -1.85385898e-01 7.05870509e-01 1.25048685e+00
-9.63077992e-02 5.82454205e-01 7.37914324e-01 -1.42860606e-01
-9.95502234e-01 -1.34995949e+00 -4.95113581e-01 1.13469398e+00
-4.16888505e-01 -5.82607567e-01 -6.14283562e-01 -1.49305895e-01
1.68689251e-01 1.24047554e+00 -2.90440619e-01 -1.50592610e-01
-2.88664848e-01 -1.31767106e+00 5.76213241e-01 4.63714719e-01
2.28047132e-01 -9.30742025e-01 -3.77976656e-01 3.67088944e-01
1.52440056e-01 -8.00322890e-01 -2.73908615e-01 6.19923055e-01
-9.35830176e-01 -5.33104241e-01 -6.35848641e-01 -6.37144521e-02
2.00369626e-01 -7.33164668e-01 1.26216578e+00 -6.72020912e-01
1.67085916e-01 5.35135508e-01 2.70354331e-01 -6.29769325e-01
-6.46413445e-01 5.50403148e-02 2.27306321e-01 -3.91497135e-01
1.80606186e-01 -9.66636360e-01 -5.28201222e-01 4.63023007e-01
-5.45179367e-01 -5.73082455e-02 4.38863575e-01 8.86032879e-01
5.15038669e-01 -1.68147802e-01 5.36964834e-02 -4.45804179e-01
1.05522704e+00 -6.48391664e-01 -1.18541527e+00 1.20463245e-01
-5.72745025e-01 5.36596656e-01 3.46837550e-01 -5.56266904e-01
-1.04377067e+00 -2.36347109e-01 -2.31216341e-01 -6.40688121e-01
-3.44534606e-01 5.69000900e-01 -1.26804918e-01 1.56782523e-01
8.38541150e-01 -1.58471256e-01 1.18680097e-01 -5.72648764e-01
4.63985264e-01 3.68333966e-01 6.22992337e-01 -1.32975054e+00
3.66778493e-01 1.26034573e-01 2.97042787e-01 -4.86182421e-01
-8.89324605e-01 -1.17519684e-01 -2.86857784e-01 2.31040537e-01
7.43967712e-01 -6.98015094e-01 -9.55395997e-01 3.40321004e-01
-1.14772880e+00 -9.53474820e-01 -4.27764356e-01 5.83064437e-01
-8.69302094e-01 5.97284101e-02 -6.20203435e-01 -8.52374673e-01
-2.66419619e-01 -1.28547251e+00 7.89656103e-01 5.04853487e-01
-4.20037061e-01 -1.25321519e+00 3.08755666e-01 -7.84122571e-02
6.46657944e-01 1.13493972e-01 6.97198093e-01 -9.41270590e-01
-5.06785333e-01 8.78399014e-02 3.93682383e-02 5.43780506e-01
-3.45656306e-01 4.24046636e-01 -1.23782921e+00 -2.84191966e-01
1.12523317e-01 -2.92381376e-01 6.81235969e-01 8.02647293e-01
1.03150642e+00 -4.11321431e-01 -4.86402303e-01 1.18051374e+00
1.03566265e+00 1.36478677e-01 4.34967428e-01 3.49506795e-01
4.93657947e-01 2.59942502e-01 1.50579721e-01 5.92254102e-01
9.50665027e-02 4.80152786e-01 4.54042107e-01 4.84853715e-01
2.80276805e-01 -3.11604291e-01 5.06891727e-01 3.15492928e-01
-2.08141655e-02 -5.42279422e-01 -1.36964524e+00 2.34796017e-01
-1.95422518e+00 -5.63286483e-01 2.71087110e-01 2.19838977e+00
9.64517772e-01 3.35600942e-01 -4.66938168e-02 -5.90922415e-01
7.31721938e-01 -6.58500642e-02 -8.12007844e-01 -3.85310918e-01
1.25289217e-01 3.62939864e-01 6.57335281e-01 6.14016831e-01
-1.19108891e+00 6.28660858e-01 6.82976294e+00 1.10335350e+00
-6.69837832e-01 9.69297066e-03 7.69273758e-01 -3.08003008e-01
-2.38129385e-02 1.71096385e-01 -1.23828888e+00 5.26266932e-01
1.46087122e+00 -4.13141176e-02 7.80434132e-01 9.71257448e-01
2.99559891e-01 -2.16961816e-01 -1.36652398e+00 7.21332669e-01
-4.66854870e-01 -1.13232744e+00 -2.24635109e-01 1.93961993e-01
6.85429811e-01 5.44184327e-01 2.25185499e-01 5.24570346e-01
1.24234414e+00 -1.29273331e+00 6.76634312e-01 9.19728279e-01
2.77855784e-01 -6.28128469e-01 5.53820074e-01 4.67827976e-01
-4.20439839e-01 -1.30483344e-01 -6.11436605e-01 1.61441714e-01
2.73218185e-01 6.23343706e-01 -1.02987814e+00 3.45712036e-01
8.29718411e-01 5.92664599e-01 -2.82037973e-01 1.36431468e+00
-5.40750623e-01 9.89875734e-01 -1.08945346e+00 -3.56223844e-02
2.39221901e-01 -3.92047673e-01 9.93202269e-01 1.02519500e+00
4.70878035e-01 -4.75492895e-01 2.25469284e-02 1.55518937e+00
1.26662031e-01 -4.68834877e-01 -1.45004258e-01 -1.97115138e-01
6.17183387e-01 9.53905582e-01 -4.03070450e-01 -1.13416463e-01
9.19106379e-02 4.47374076e-01 1.87327385e-01 7.52433002e-01
-7.63564706e-01 -1.29025295e-01 7.25493968e-01 -2.26141527e-01
5.13371408e-01 -2.64603257e-01 -2.27637067e-02 -8.95196676e-01
-3.57417375e-01 -8.14692736e-01 2.50029087e-01 -1.15440774e+00
-1.53882444e+00 1.52946800e-01 5.28241038e-01 -6.10955536e-01
-4.27141726e-01 -8.83247256e-01 -5.43022215e-01 1.36479509e+00
-1.16738248e+00 -6.39733136e-01 3.49918194e-02 1.57689601e-01
1.90167930e-02 6.38181763e-03 6.85696840e-01 -3.10875744e-01
-7.06977606e-01 2.42321894e-01 4.77037132e-01 -2.16109604e-01
7.16820717e-01 -1.49928951e+00 6.46756589e-01 6.48473561e-01
-9.33571905e-02 9.13035870e-01 1.09559977e+00 -5.91947556e-01
-1.12198758e+00 -8.29230130e-01 -7.50696808e-02 -6.84624016e-01
8.56124282e-01 -2.46060744e-01 -8.19256067e-01 8.17105234e-01
1.50846735e-01 -4.74590212e-02 5.41879892e-01 2.96793193e-01
1.01033248e-01 1.01331256e-01 -1.10714877e+00 8.27149928e-01
7.46074378e-01 -1.00283630e-01 -4.92616832e-01 3.43875706e-01
8.36160481e-01 -7.01898456e-01 -1.11197150e+00 4.40043837e-01
5.06624341e-01 -8.59601259e-01 1.09805143e+00 -5.36613107e-01
1.08300939e-01 -3.31295043e-01 -2.09401220e-01 -1.61530983e+00
-2.22057387e-01 -1.26978302e+00 -6.23634756e-01 1.18121493e+00
4.29913998e-01 -8.55478942e-01 7.14426398e-01 1.15169060e+00
-3.94115627e-01 -7.68060267e-01 -8.23990464e-01 -9.63489473e-01
2.64269948e-01 -6.84004068e-01 8.49455953e-01 3.26788455e-01
-4.76747006e-01 1.76210068e-02 -2.84539819e-01 5.46663880e-01
7.11556971e-01 -4.46371585e-01 8.67203116e-01 -1.20571291e+00
-1.09219253e+00 -6.40796423e-01 5.85314557e-02 -1.42136431e+00
-2.71225143e-02 -6.30634785e-01 3.26386094e-01 -1.20828986e+00
-8.33847895e-02 -3.75417590e-01 -1.61803767e-01 3.94665152e-01
-1.01668015e-01 -3.49046230e-01 7.72540942e-02 3.45792025e-01
-3.02801341e-01 8.26772332e-01 1.08130634e+00 2.52868026e-01
-3.37108999e-01 2.82313555e-01 -4.61520612e-01 1.03259408e+00
1.06873798e+00 -6.88238621e-01 -4.26975459e-01 -2.24257544e-01
2.46971264e-01 -2.40133293e-02 5.87419808e-01 -1.12519920e+00
3.37424278e-01 -1.93805605e-01 4.80967849e-01 -5.28967321e-01
6.50847554e-01 -4.27224010e-01 3.01533014e-01 1.42492279e-01
-9.02086720e-02 -8.37109610e-02 6.37864649e-01 7.33526051e-01
2.01059014e-01 -6.26382172e-01 5.97623765e-01 -1.55574322e-01
-2.73601502e-01 1.71667933e-01 -4.33597118e-01 2.64434129e-01
4.17393923e-01 2.84861755e-02 -4.79146183e-01 -4.20233488e-01
-1.18882668e+00 4.87932652e-01 5.83898425e-01 -1.78532735e-01
2.12329596e-01 -8.58895898e-01 -3.30640137e-01 1.53341785e-01
-2.83784539e-01 3.68467480e-01 -2.27785766e-01 8.09433818e-01
-6.13943934e-01 3.79507810e-01 6.25605136e-02 -9.75639820e-01
-4.44458127e-01 2.81144768e-01 6.84182107e-01 -4.56563681e-01
-3.23322654e-01 9.19454753e-01 -3.48089710e-02 -7.13131249e-01
4.32736486e-01 -4.52384293e-01 2.66915619e-01 -4.55444127e-01
1.09423004e-01 3.80899876e-01 -2.87054867e-01 1.46007910e-01
1.22490274e-02 -7.15756742e-03 2.55309522e-01 -4.90825176e-01
1.42814124e+00 5.86631410e-02 -1.13940232e-01 4.47041512e-01
5.68845570e-01 -2.08559290e-01 -2.16910839e+00 -2.74261869e-02
1.82905018e-01 -1.30034208e-01 1.21760257e-01 -9.29603994e-01
-4.83137071e-01 8.51102650e-01 3.60938668e-01 -1.42538801e-01
6.86834157e-01 -1.41263261e-01 5.12212999e-02 7.45588541e-01
2.03259781e-01 -9.32607830e-01 -1.08426556e-01 7.27761805e-01
9.59563017e-01 -8.22641075e-01 1.63181815e-02 7.02617764e-02
-4.60667431e-01 1.23194075e+00 4.02512461e-01 -1.93713367e-01
8.97435427e-01 3.79314929e-01 -4.34374630e-01 -1.36524320e-01
-9.48858380e-01 1.64685249e-01 3.66742253e-01 8.62242520e-01
-3.55131589e-02 -1.69403940e-01 5.06761670e-01 7.90894508e-01
-4.65518624e-01 -3.19075547e-02 3.72169435e-01 9.10963297e-01
-2.26116315e-01 -9.54223216e-01 -6.45669103e-01 4.99368936e-01
-4.92449731e-01 -4.99665946e-01 2.75847495e-01 6.95560277e-01
-2.65867651e-01 6.97109342e-01 -2.70926859e-02 7.58529156e-02
-1.63975023e-02 4.06753480e-01 4.32251245e-01 -6.64358318e-01
-2.42344782e-01 1.76598489e-01 3.71970445e-01 -7.17449188e-01
-3.19178812e-02 -7.66860902e-01 -6.20255351e-01 -2.70241350e-01
-9.14753526e-02 1.33843511e-01 7.25768805e-01 8.24359179e-01
5.59740722e-01 4.19556051e-01 -2.94018298e-01 -1.34217072e+00
-1.02021146e+00 -1.22934711e+00 -3.48355085e-01 -2.81332284e-01
1.58526793e-01 -8.16134214e-01 -7.73211122e-01 -4.76246141e-02] | [6.959787845611572, 3.8718695640563965] |
bd24c53a-0fda-4679-a85d-2e05936e1902 | advanced-feature-learning-on-point-clouds | 2205.09962 | null | https://arxiv.org/abs/2205.09962v1 | https://arxiv.org/pdf/2205.09962v1.pdf | Advanced Feature Learning on Point Clouds using Multi-resolution Features and Learnable Pooling | Existing point cloud feature learning networks often incorporate sequences of sampling, neighborhood grouping, neighborhood-wise feature learning, and feature aggregation to learn high-semantic point features that represent the global context of a point cloud. Unfortunately, the compounded loss of information concerning granularity and non-maximum point features due to sampling and max pooling could adversely affect the high-semantic point features from existing networks such that they are insufficient to represent the local context of a point cloud, which in turn may hinder the network in distinguishing fine shapes. To cope with this problem, we propose a novel point cloud feature learning network, PointStack, using multi-resolution feature learning and learnable pooling (LP). The multi-resolution feature learning is realized by aggregating point features of various resolutions in the multiple layers, so that the final point features contain both high-semantic and high-resolution information. On the other hand, the LP is used as a generalized pooling function that calculates the weighted sum of multi-resolution point features through the attention mechanism with learnable queries, in order to extract all possible information from all available point features. Consequently, PointStack is capable of extracting high-semantic point features with minimal loss of information concerning granularity and non-maximum point features. Therefore, the final aggregated point features can effectively represent both global and local contexts of a point cloud. In addition, both the global structure and the local shape details of a point cloud can be well comprehended by the network head, which enables PointStack to advance the state-of-the-art of feature learning on point clouds. The codes are available at https://github.com/kaist-avelab/PointStack. | ['Seung-Hyun Kong', 'Dong-Hee Paek', 'Kevin Tirta Wijaya'] | 2022-05-20 | null | null | null | null | ['3d-point-cloud-classification'] | ['computer-vision'] | [-5.11544406e-01 -3.38187903e-01 5.48171513e-02 -6.44384444e-01
-8.36846471e-01 -4.65444535e-01 4.22349185e-01 4.46239829e-01
-2.10702047e-01 2.69551337e-01 -1.65148273e-01 3.18351299e-01
-5.77875137e-01 -1.30538535e+00 -8.45366001e-01 -6.94504559e-01
-1.66422606e-01 2.22945586e-01 3.77856374e-01 -1.79372892e-01
2.49965265e-01 1.26963890e+00 -1.88736880e+00 4.60239291e-01
7.45708227e-01 1.43106413e+00 4.42975074e-01 1.50841355e-01
-6.69411600e-01 3.59984324e-03 -3.52040589e-01 7.70408586e-02
2.77768761e-01 4.02762741e-01 -4.87291813e-01 -9.91232321e-02
6.65172577e-01 -2.12100402e-01 -3.22889298e-01 1.05858874e+00
3.60788733e-01 2.30288088e-01 5.06282747e-01 -1.35537910e+00
-5.61088860e-01 7.52237365e-02 -4.51612771e-01 4.50574793e-02
-7.49070719e-02 2.05785349e-01 1.21427441e+00 -1.14819849e+00
2.68987268e-01 1.37381124e+00 5.52574873e-01 1.39558554e-01
-7.32980132e-01 -8.26429665e-01 4.19416219e-01 1.61079913e-01
-1.64340782e+00 -9.30834264e-02 1.09276509e+00 -3.40836048e-01
9.90458190e-01 3.53800625e-01 8.96852255e-01 5.20043194e-01
9.61365029e-02 5.76543987e-01 5.14906526e-01 2.52476096e-01
5.51245175e-02 -6.60027191e-02 8.86369944e-02 6.65160477e-01
7.36507401e-02 1.97005942e-02 -5.13582468e-01 -2.27889419e-01
1.14070308e+00 8.40140343e-01 -1.91452116e-01 -3.86128992e-01
-1.16945064e+00 6.59403861e-01 1.30757785e+00 4.22997534e-01
-5.42856574e-01 1.78477257e-01 7.05087855e-02 1.84368923e-01
4.21456486e-01 3.87867332e-01 -6.72867239e-01 2.38193527e-01
-7.92841733e-01 2.44775146e-01 2.43666112e-01 1.15876400e+00
1.60743427e+00 -2.45130002e-01 -6.59502819e-02 6.61410689e-01
4.33080375e-01 6.92142010e-01 1.98114395e-01 -7.28440046e-01
6.67923927e-01 1.19717765e+00 -4.00635265e-02 -1.23931015e+00
-5.01590312e-01 -3.51123363e-01 -7.73979723e-01 4.20644641e-01
1.26531953e-02 1.98647767e-01 -8.95712674e-01 1.41941071e+00
3.40164363e-01 2.47212514e-01 -1.74710378e-01 1.07144451e+00
1.15969908e+00 8.59001696e-01 -8.68585184e-02 3.41432601e-01
1.31641793e+00 -4.59876925e-01 -8.67422521e-02 -1.77227080e-01
3.19767684e-01 -4.89100605e-01 9.72985744e-01 -2.17514127e-01
-7.73292184e-01 -7.07964122e-01 -1.01939273e+00 -2.84347653e-01
-7.86455035e-01 2.52820756e-02 7.60209084e-01 -6.77017719e-02
-6.76486373e-01 7.72778273e-01 -8.24008346e-01 2.12485455e-02
8.68191302e-01 5.60906649e-01 -5.21430910e-01 -2.28611574e-01
-9.69710588e-01 5.37042737e-01 2.82381803e-01 2.88553953e-01
-3.67502868e-01 -9.49272692e-01 -8.96465898e-01 4.14797723e-01
1.37692124e-01 -6.94819450e-01 7.16048658e-01 -6.61512375e-01
-8.67767274e-01 4.88621414e-01 -2.10550711e-01 2.60769606e-01
2.16492891e-01 -1.50613964e-01 -2.58899480e-01 1.37673661e-01
2.49258935e-01 7.93301344e-01 7.83395171e-01 -1.23202944e+00
-9.08545554e-01 -7.69898891e-01 3.86123918e-02 3.10271591e-01
-1.24429114e-01 -3.79939854e-01 -3.75896990e-01 -3.45645994e-01
5.65134287e-01 -5.38918316e-01 -1.37256026e-01 2.66318262e-01
-2.41145670e-01 -5.33922195e-01 1.00857532e+00 -1.62906840e-01
6.92760766e-01 -2.53605628e+00 -1.22593857e-01 3.84733945e-01
2.67983675e-01 -1.18240103e-01 -2.59503067e-01 2.54343569e-01
2.79247761e-04 2.54870564e-01 -1.08215064e-01 -4.32936065e-02
-9.85081345e-02 2.61555552e-01 -2.43895039e-01 3.17416936e-01
6.20791137e-01 1.00370073e+00 -8.31579685e-01 -3.30145955e-01
6.36205196e-01 6.85480356e-01 -4.21645641e-01 6.41427487e-02
-2.68574178e-01 3.36270750e-01 -1.12214077e+00 7.67921388e-01
9.03301060e-01 -1.38769880e-01 -6.83545828e-01 -4.72356766e-01
-2.93462753e-01 8.47385544e-03 -1.14825737e+00 1.82087445e+00
-5.79760611e-01 3.19700897e-01 -1.64306015e-01 -5.26991606e-01
1.21102893e+00 5.12790009e-02 6.90073252e-01 -5.87347269e-01
-3.86350267e-02 3.21903497e-01 -3.47702265e-01 -3.64662945e-01
4.40896153e-01 -3.42591293e-02 -1.64431840e-01 -7.05584586e-02
8.32110420e-02 -2.09303617e-01 -4.38879281e-01 -1.75627366e-01
8.12456548e-01 -1.74905241e-01 -2.79943757e-02 -1.22048825e-01
5.08729875e-01 -6.78014830e-02 4.97743815e-01 5.29934108e-01
1.80749357e-01 8.43493283e-01 2.28625610e-01 -7.19054163e-01
-9.26677883e-01 -1.20565403e+00 -2.69040823e-01 7.81497717e-01
5.84399402e-01 -2.68145055e-01 -1.96568996e-01 -6.14271939e-01
5.29969037e-01 3.25513363e-01 -3.86363804e-01 -2.45563865e-01
-4.99054402e-01 -1.77972168e-01 4.46068831e-02 6.21275246e-01
5.25152266e-01 -1.28272474e+00 -5.77223420e-01 2.01462805e-01
7.16407597e-02 -9.63421643e-01 -3.68934423e-01 1.59589186e-01
-1.00373721e+00 -1.12763739e+00 -3.59683961e-01 -6.77734196e-01
7.94700801e-01 4.56187934e-01 8.49414766e-01 2.89861888e-01
-1.91935480e-01 1.95895880e-01 -2.87638605e-01 -3.97303104e-01
3.50055128e-01 2.25335017e-01 -1.70003131e-01 8.32956582e-02
7.04695940e-01 -7.93742597e-01 -6.69898748e-01 2.22201124e-01
-8.18288684e-01 -1.31793126e-01 7.46612370e-01 6.20493889e-01
9.56696272e-01 2.89438069e-01 4.96983230e-01 -2.53558040e-01
2.07442388e-01 -5.06661296e-01 -5.52751303e-01 2.36958545e-02
9.47341695e-03 3.14367488e-02 7.31434643e-01 -1.47864342e-01
-7.06600606e-01 8.98407102e-02 -1.94219053e-01 -8.35205436e-01
-3.70580286e-01 3.66657853e-01 -6.47437155e-01 -3.68798733e-01
3.83917391e-01 2.64074594e-01 -1.67038783e-01 -6.03807569e-01
3.27904314e-01 6.15729570e-01 1.66255489e-01 -5.72581291e-01
1.00686634e+00 6.66198611e-01 1.09509781e-01 -8.26199651e-01
-7.33353436e-01 -6.10589921e-01 -8.02360356e-01 -6.85748309e-02
7.74384141e-01 -9.83679473e-01 -7.28407681e-01 4.71383154e-01
-1.11180723e+00 1.69381678e-01 -4.89204675e-01 4.08902496e-01
-4.65952009e-01 2.81085018e-02 -3.58386785e-01 -5.34893870e-01
-3.38781118e-01 -1.15115082e+00 1.37063479e+00 5.02707481e-01
2.82152534e-01 -6.30558372e-01 -4.00702715e-01 -1.29460633e-01
2.52616495e-01 3.79952461e-01 9.70104992e-01 -5.23815095e-01
-1.11692941e+00 -3.55673224e-01 -5.68132281e-01 2.51065940e-01
2.99534827e-01 3.68002132e-02 -1.00312448e+00 -2.15147600e-01
-5.56226224e-02 -6.74787238e-02 8.60568821e-01 3.48524839e-01
1.50248110e+00 -1.03667393e-01 -4.23285812e-01 8.71362567e-01
1.63083422e+00 -1.10909514e-01 3.95394623e-01 2.95032769e-01
1.03201592e+00 4.61965591e-01 6.64525092e-01 5.08001447e-01
4.62631017e-01 4.99290824e-01 7.85483718e-01 -1.52671896e-02
1.81007653e-01 -4.28877532e-01 -1.74074396e-01 5.33454955e-01
-1.11122444e-01 2.18503460e-01 -9.26138699e-01 6.62402391e-01
-1.90242124e+00 -8.35679233e-01 8.06030259e-02 2.10182667e+00
2.54135847e-01 -7.97114745e-02 -2.77864158e-01 -2.17754409e-01
8.20182085e-01 3.61282229e-01 -7.33565211e-01 4.60104719e-02
-1.46888450e-01 5.23363873e-02 5.10414183e-01 1.67575285e-01
-1.08782113e+00 8.91643584e-01 4.29815197e+00 9.88970041e-01
-1.28825080e+00 -1.57367721e-01 1.77420899e-01 -2.62432605e-01
-5.04561663e-01 -6.62586465e-02 -7.87283778e-01 4.91708547e-01
3.61111760e-01 3.12308539e-02 3.02603960e-01 1.04047966e+00
2.79003046e-02 3.70067626e-01 -1.01254880e+00 1.20091105e+00
-3.27386767e-01 -1.30814326e+00 2.72204131e-01 1.53823003e-01
4.18498456e-01 4.71537918e-01 -1.93041023e-02 3.13316077e-01
-1.40639439e-01 -9.86653149e-01 6.16293192e-01 7.57390738e-01
7.89061606e-01 -1.01271939e+00 7.12185502e-01 4.24204677e-01
-1.69397664e+00 -3.31863463e-01 -8.45907688e-01 4.67308834e-02
-1.24526443e-02 6.40117586e-01 -3.23190391e-01 8.04964542e-01
9.66860235e-01 9.74931657e-01 -6.07681692e-01 1.30685055e+00
-3.40368710e-02 -8.12611580e-02 -7.91180968e-01 1.05004586e-01
5.11828959e-01 -8.86894539e-02 5.39265752e-01 9.35001433e-01
6.42962456e-01 2.58720040e-01 3.22355688e-01 9.82630074e-01
-1.32550329e-01 1.37743801e-01 -6.58360422e-01 2.75460362e-01
7.06077814e-01 1.39457107e+00 -5.82089961e-01 -1.38092712e-01
-4.52050149e-01 4.03302163e-01 6.55919433e-01 3.24264407e-01
-4.50291663e-01 -5.94252229e-01 1.20184910e+00 1.43433660e-01
5.21780789e-01 -2.91679859e-01 -4.44226146e-01 -1.14391840e+00
3.77329797e-01 -2.04383850e-01 2.43831500e-01 -7.90994167e-01
-1.53957593e+00 6.12327576e-01 1.02071213e-02 -1.58308768e+00
3.02294761e-01 -5.30940473e-01 -8.00023735e-01 1.17784405e+00
-1.80887318e+00 -1.30320370e+00 -7.55242467e-01 6.90872908e-01
3.90980840e-01 -9.36690792e-02 5.80587208e-01 1.77336246e-01
-1.77203089e-01 3.04538608e-01 -1.17427126e-01 1.30173683e-01
2.96528071e-01 -1.06068957e+00 4.49976057e-01 3.01407397e-01
8.50069597e-02 5.38763106e-01 3.56868119e-03 -6.45357192e-01
-1.25457358e+00 -1.37801325e+00 6.10218942e-01 -3.90761703e-01
4.22900081e-01 -3.74021232e-01 -1.37028205e+00 3.74612927e-01
-5.90973616e-01 5.46827734e-01 3.74440134e-01 1.49166524e-01
-4.69823718e-01 -4.25556630e-01 -1.22446322e+00 4.32174891e-01
1.08233178e+00 -6.91756070e-01 -6.04111314e-01 2.35802069e-01
8.36274624e-01 -4.01190192e-01 -1.22004676e+00 6.06992066e-01
3.45569909e-01 -7.84809470e-01 1.10160577e+00 -3.58443469e-01
3.19812179e-01 -5.29761076e-01 -3.06961864e-01 -1.23650038e+00
-7.30483115e-01 1.62843809e-01 1.36485115e-01 1.08362567e+00
2.94947296e-01 -7.74418652e-01 7.00591326e-01 5.22107542e-01
-3.15454334e-01 -9.33989406e-01 -1.25981390e+00 -6.21539652e-01
1.37120485e-01 -6.08867526e-01 1.33283341e+00 8.52888167e-01
-3.30086648e-01 3.22809117e-03 1.26957625e-01 6.74362838e-01
5.57273865e-01 6.38071895e-01 5.23178458e-01 -1.56309259e+00
3.90940249e-01 -6.32831156e-01 -9.58800912e-01 -9.49357867e-01
2.00025335e-01 -1.04893124e+00 -2.15786129e-01 -1.61069441e+00
-1.24435261e-01 -8.62303793e-01 -5.40492058e-01 6.80560112e-01
-5.14497384e-02 4.16553840e-02 4.26228374e-01 4.68389690e-01
-4.52892482e-01 8.10535669e-01 1.45280933e+00 -1.83518723e-01
-2.22202152e-01 1.11506008e-01 -5.61363935e-01 5.99669099e-01
9.00129616e-01 -4.74973142e-01 -1.16556644e-01 -6.52508199e-01
1.80766419e-01 -5.54791614e-02 6.90545201e-01 -1.12039721e+00
5.23060083e-01 -2.42326111e-01 8.46739650e-01 -9.41155732e-01
6.04637384e-01 -1.27826619e+00 3.88191044e-02 -7.38283619e-02
1.12838656e-01 -6.71665519e-02 2.80403644e-01 4.80249614e-01
-4.27978426e-01 -2.79637594e-02 5.33124328e-01 -3.65225405e-01
-8.98247480e-01 1.04989767e+00 3.62279058e-01 -3.44116598e-01
8.29632342e-01 -3.80806625e-01 -3.38886589e-01 4.93864343e-02
-6.39600813e-01 5.48885882e-01 7.15089440e-01 8.34514558e-01
9.89294291e-01 -1.47513151e+00 -5.50403178e-01 6.18768752e-01
4.71049547e-01 8.45908403e-01 8.03598881e-01 3.31152260e-01
-4.05934453e-01 1.58859208e-01 -4.15724725e-01 -8.41826618e-01
-8.42047215e-01 5.23125470e-01 3.97406369e-01 2.65766203e-01
-9.59502995e-01 6.19530857e-01 3.11377645e-01 -6.54744148e-01
-1.68460399e-01 -5.06946385e-01 -3.00717771e-01 7.02756271e-02
4.01859015e-01 5.83384037e-02 1.36580199e-01 -7.77262747e-01
-5.63216388e-01 1.18602550e+00 9.34515614e-03 4.82943505e-01
1.46297598e+00 3.22647132e-02 -2.30226159e-01 5.43814182e-01
1.41841781e+00 -1.05331175e-01 -1.37353742e+00 -4.44104254e-01
-2.82093734e-01 -7.09800243e-01 2.22152337e-01 -3.90605718e-01
-1.29987490e+00 9.22388852e-01 5.45158923e-01 1.52970314e-01
9.42248225e-01 3.08869332e-01 6.39048278e-01 4.83824611e-01
6.06855273e-01 -7.17402101e-01 -3.24545890e-01 5.14661670e-01
1.01210582e+00 -1.37036121e+00 -8.42489079e-02 -4.05294120e-01
-3.89474779e-01 1.28544188e+00 8.15204978e-01 -4.96595204e-01
9.39643860e-01 2.46698931e-02 -3.34000662e-02 -6.36316061e-01
-4.45504665e-01 -2.34989166e-01 5.32365024e-01 5.38850605e-01
-1.01618148e-01 2.33063638e-01 4.80749786e-01 7.04142451e-01
-4.37488019e-01 -3.15744698e-01 -7.08041415e-02 7.78293669e-01
-8.11965764e-01 -6.11909986e-01 -2.55651712e-01 7.12893188e-01
5.02806641e-02 2.87370458e-02 -1.19812302e-01 7.43756950e-01
3.83499116e-01 5.14833927e-01 4.85713124e-01 -4.81178254e-01
6.39357328e-01 -1.42459452e-01 8.65391716e-02 -6.67936802e-01
-5.91467559e-01 -2.39537433e-01 -6.34362876e-01 -7.44243622e-01
-8.88810679e-02 -3.51888776e-01 -1.47533143e+00 -1.65779978e-01
-2.31635645e-01 1.52679637e-01 7.29177356e-01 8.81246328e-01
6.19416773e-01 5.02793312e-01 7.58749843e-01 -1.33760726e+00
-2.14879185e-01 -7.86848187e-01 -7.02059031e-01 4.42556262e-01
6.32891119e-01 -7.59478092e-01 -3.78340065e-01 -7.25067675e-01] | [7.910632133483887, -3.4492123126983643] |
5d010847-5e44-4894-b749-b16c4f650122 | degree-aware-based-adversarial-graph | null | null | https://www.sciencedirect.com/science/article/pii/S0925231222001448 | https://www.sciencedirect.com/science/article/pii/S0925231222001448/pdfft?md5=1791f3ca52941e00736316a5605ed3ff&pid=1-s2.0-S0925231222001448-main.pdf | Degree aware based adversarial graph convolutional networks for entity alignment in heterogeneous knowledge graph | Entity alignment, as the vital technique for knowledge graph construction and integration, aims to match entities that refer to the same real-world identity in different knowledge graphs (KGs). Recently, much effort has been devoted to embedding-based methods for entity alignment. For most of such methods, the entity with a high degree is hard to be aligned with its equivalent counterpart with a low degree. This degree difference between equivalent entities poses a great challenge for entity alignment. To solve this problem, a novel entity alignment framework that integrates a graph convolutional network (GCN) based embedding initializer and a degree aware generative adversarial network is proposed. In particular, the embedding initializer utilizes a GCN with highway gates to generate the preliminary embedding of entities based on their topological characteristics in the KGs. By alleviating the relevance between embeddings and degree features, the degree aware GAN mitigates the impact of degree difference and generates the final degree level irrelevant alignment result. To quantify the heterogeneity between the KGs, an evaluation metric called heterogeneity entropy of degree (HED) that represents the degree difference is defined in this paper. Based on HED, the impact of KG heterogeneity on the performance of the proposed DAGCN model is investigated based on WK3l-15k and DBP15k datasets. The experimental results show that the proposed degree aware adversarial graph convolutional network (DAGCN) outperforms other state-of-the-art methods over all metrics, especially when the HED is large. | ['Tao Luo', 'Jianfeng Li', 'Yining Wang', 'Hanchen Wang'] | 2022-04-28 | null | null | null | neurocomputing-2022-4 | ['entity-alignment', 'entity-alignment'] | ['knowledge-base', 'natural-language-processing'] | [-2.83408463e-01 3.53150189e-01 -7.45905191e-02 -6.26194775e-02
-8.40754882e-02 -6.76715851e-01 4.95574951e-01 5.13148844e-01
-1.92402765e-01 6.17975414e-01 9.29025039e-02 -1.39933825e-01
-3.10455471e-01 -1.42325282e+00 -6.43738210e-01 -5.87258637e-01
-7.28067830e-02 3.10772240e-01 1.60644606e-01 -4.37950432e-01
4.94607314e-02 3.48676324e-01 -9.64173198e-01 -2.92923987e-01
1.29607165e+00 6.79225624e-01 -1.06167048e-01 -1.74720269e-02
-2.90920377e-01 4.94404405e-01 -6.94402874e-01 -9.23420548e-01
3.95441115e-01 -3.23161721e-01 -6.75215721e-01 -4.82595176e-01
3.74302536e-01 1.04445331e-01 -6.77979231e-01 1.35367703e+00
7.54690051e-01 -2.50065979e-02 6.06495798e-01 -1.57306838e+00
-8.92491698e-01 8.46061051e-01 -2.93451458e-01 1.50831893e-01
1.72210127e-01 3.24005708e-02 1.07867777e+00 -6.03803158e-01
7.16451466e-01 9.83378410e-01 7.55429447e-01 2.45101918e-02
-1.13558221e+00 -6.12900734e-01 1.14599302e-01 4.08925802e-01
-1.83788157e+00 1.72863081e-01 1.10850501e+00 -4.05560315e-01
6.79501593e-01 -3.07657588e-02 8.26387644e-01 6.97947383e-01
1.49825692e-01 2.70802468e-01 7.67236114e-01 -1.99113220e-01
1.01473905e-01 5.20074479e-02 -4.87046316e-03 8.36174428e-01
8.95962358e-01 -1.45051852e-01 -1.63670614e-01 1.15818731e-01
6.07926726e-01 -1.50150254e-01 -3.57580632e-01 -5.47017872e-01
-1.04958510e+00 6.54653072e-01 1.16388261e+00 2.92564631e-01
-2.08560169e-01 9.06519890e-02 4.06888187e-01 1.82792485e-01
9.74046364e-02 6.17587864e-01 -1.49062589e-01 2.05966741e-01
-5.79292297e-01 1.31918475e-01 7.50487328e-01 9.43374038e-01
9.77956414e-01 9.19365808e-02 -3.41624141e-01 3.50601137e-01
2.21621022e-01 3.33120853e-01 3.11659783e-01 -1.54819891e-01
6.98225915e-01 1.42141294e+00 -2.04665378e-01 -1.82199466e+00
-3.67316902e-01 -8.28692079e-01 -1.19428980e+00 -1.70745283e-01
9.02596563e-02 -4.76693138e-02 -7.58608520e-01 1.95336688e+00
5.29870391e-01 4.54849094e-01 3.65349680e-01 7.70981431e-01
1.19966161e+00 5.05522847e-01 8.97298902e-02 2.15991199e-01
1.27833080e+00 -7.37822294e-01 -5.61114311e-01 -1.29371911e-01
7.23531544e-01 -5.70617855e-01 7.14828610e-01 -4.56169605e-01
-5.72558463e-01 -5.30505180e-01 -1.36126041e+00 5.15978336e-02
-8.15589309e-01 -1.51487915e-02 5.96855521e-01 5.72577477e-01
-9.02450979e-01 5.76548219e-01 -4.37905788e-01 -2.05648750e-01
2.32842684e-01 2.56665736e-01 -5.40927708e-01 2.23424053e-03
-1.79664040e+00 8.18251550e-01 8.09223175e-01 3.09660584e-01
-3.52849454e-01 -8.61643791e-01 -9.76909637e-01 3.59693646e-01
2.11171955e-01 -8.20921719e-01 3.00973415e-01 -6.14571214e-01
-9.72199023e-01 7.38013864e-01 4.62003648e-01 -3.20872337e-01
4.36359197e-01 2.63286382e-01 -7.97457039e-01 -3.10422629e-02
2.75887027e-02 2.57395625e-01 4.56302136e-01 -1.06762779e+00
-2.67050207e-01 -4.17500407e-01 3.55485916e-01 3.52394491e-01
-3.91375989e-01 -5.84009349e-01 -5.08941829e-01 -7.89769948e-01
1.88383777e-02 -8.74954641e-01 4.39493768e-02 -4.09967899e-01
-7.54456401e-01 -5.98907024e-02 6.47644997e-01 -6.36340559e-01
1.50698423e+00 -1.90125668e+00 4.22938436e-01 5.44232249e-01
4.92907703e-01 5.09077251e-01 -4.78847809e-02 5.57468235e-01
-1.08039729e-01 1.53011724e-01 -8.64219666e-02 1.39976278e-01
6.18623123e-02 -2.83891503e-02 8.73981118e-02 2.88320601e-01
2.82886654e-01 1.20443642e+00 -1.10427713e+00 -5.48983097e-01
7.63925985e-02 4.42473114e-01 -5.14027178e-01 2.03114063e-01
-4.42579128e-02 2.30828211e-01 -4.49015170e-01 5.34681916e-01
9.03827786e-01 -4.16726530e-01 5.71636200e-01 -7.96774685e-01
1.66016355e-01 -1.00568987e-01 -1.30640447e+00 1.45654953e+00
-2.37699151e-01 3.48180979e-01 -3.59564751e-01 -8.76199901e-01
1.17768824e+00 2.99234334e-02 2.73203164e-01 -6.04467332e-01
5.23711070e-02 2.27670357e-01 1.34704098e-01 -1.11835584e-01
6.56973839e-01 3.73959213e-01 -3.10810745e-01 -8.87278169e-02
8.91464874e-02 2.20437407e-01 6.83296472e-02 5.20919621e-01
1.29483902e+00 -3.11167866e-01 1.62574112e-01 -2.22318605e-01
6.83068156e-01 -1.81398824e-01 7.97391117e-01 2.11423248e-01
-1.29165053e-02 1.91796958e-01 7.58612454e-01 -4.23877060e-01
-1.02854896e+00 -1.03723061e+00 1.51053682e-01 9.79669616e-02
7.72988141e-01 -5.67394555e-01 -6.34766579e-01 -9.11899149e-01
2.68997222e-01 2.97314405e-01 -6.27296388e-01 -6.79033756e-01
-4.94667321e-01 -6.00550354e-01 7.19626427e-01 4.39289510e-01
8.97816956e-01 -8.84020507e-01 1.47681296e-01 1.07675083e-01
-2.11790964e-01 -1.26988053e+00 -4.06274110e-01 -2.36942962e-01
-4.34925079e-01 -1.22585082e+00 -5.52885532e-01 -8.83285522e-01
1.05105543e+00 -9.31507498e-02 1.08317375e+00 2.24163130e-01
-1.04065850e-01 7.26130605e-02 -4.71773505e-01 1.67635027e-02
-4.01037723e-01 5.76013803e-01 -5.40743135e-02 7.09173456e-02
2.22859487e-01 -7.86112726e-01 -8.64390373e-01 3.48197937e-01
-9.41531181e-01 1.36139169e-01 7.62439132e-01 7.16642439e-01
6.06956065e-01 3.17051411e-01 6.81110263e-01 -9.81233120e-01
7.03287363e-01 -6.20843530e-01 -7.29331136e-01 4.95025337e-01
-8.05087507e-01 3.58921379e-01 8.85022938e-01 -3.22730094e-01
-5.89558244e-01 -2.53179073e-01 5.61943315e-02 -4.17710781e-01
4.80587363e-01 9.23068404e-01 -6.50340915e-01 -2.86585689e-01
4.28263992e-01 1.37629092e-01 -3.27425838e-01 -1.91678271e-01
5.33423603e-01 2.42877141e-01 5.51609874e-01 -3.78383696e-01
1.15497625e+00 2.40663905e-02 3.26328546e-01 -3.86927933e-01
-3.48464102e-01 -1.55623302e-01 -4.66124505e-01 -1.35123685e-01
7.47643828e-01 -8.57403874e-01 -5.67607284e-01 6.25820994e-01
-1.10871971e+00 1.00979386e-02 -1.55962691e-01 3.71506929e-01
-7.45481113e-03 4.80583191e-01 -2.89229095e-01 -1.86170712e-01
-5.87180138e-01 -1.04545009e+00 6.62507057e-01 4.87542629e-01
8.08269978e-02 -9.63340044e-01 1.72954693e-01 4.39998694e-02
3.83409202e-01 7.94843316e-01 1.28792465e+00 -6.56258821e-01
-8.88563812e-01 -4.02014613e-01 -4.35382336e-01 1.24951489e-01
3.20654035e-01 -5.95808551e-02 -3.22087288e-01 -3.94684970e-01
-7.67515004e-01 1.63321540e-01 6.03890717e-01 -1.46884680e-01
6.73204660e-01 -4.72895741e-01 -4.59786087e-01 8.09797525e-01
1.86021960e+00 -1.29899025e-01 9.35040116e-01 3.61445308e-01
1.23158300e+00 1.76063254e-01 4.39383626e-01 1.68384090e-01
7.34142780e-01 8.35825801e-01 5.42730689e-01 -1.75014034e-01
-1.80704743e-01 -6.52591467e-01 8.59724656e-02 1.06179392e+00
-7.48922601e-02 -4.90965456e-01 -8.94090593e-01 7.18228817e-01
-1.78585649e+00 -7.22627521e-01 -5.71937300e-02 2.14716172e+00
6.93926096e-01 1.66371912e-01 -2.49905616e-01 1.14994399e-01
1.08355999e+00 2.92000741e-01 -5.05768239e-01 -1.31250098e-01
-2.05532581e-01 1.69129685e-01 6.50019288e-01 2.98245966e-01
-8.03067565e-01 9.80009973e-01 4.20152092e+00 8.81570160e-01
-1.04978216e+00 -9.63606015e-02 1.89140454e-01 4.63968068e-01
-5.72491288e-01 1.67037576e-01 -8.01619530e-01 9.47918057e-01
4.13571030e-01 -4.94678915e-01 2.91407883e-01 7.23648369e-01
-3.01958382e-01 3.55925590e-01 -8.00653040e-01 7.84298539e-01
-1.60612166e-01 -1.34510064e+00 2.16965392e-01 4.86378931e-02
8.23171496e-01 -2.41399616e-01 -4.35310900e-02 3.56674552e-01
2.74227738e-01 -9.46326792e-01 3.05484533e-01 6.98079467e-01
9.24728811e-01 -1.01373005e+00 1.07855225e+00 -9.28862542e-02
-1.68702245e+00 1.71098009e-01 -3.10888767e-01 4.52170700e-01
1.11341402e-01 7.98886776e-01 -7.64125347e-01 1.29612541e+00
3.77633005e-01 5.09144127e-01 -7.36019850e-01 9.16725993e-01
-5.32231510e-01 1.36590600e-01 -1.67136192e-01 9.68986750e-02
-2.14342820e-03 -3.42978030e-01 6.53603196e-01 9.55197334e-01
4.11006957e-01 -1.00295395e-01 -1.89429685e-03 9.74839687e-01
-6.33072615e-01 2.96145469e-01 -6.40868008e-01 -1.72627047e-01
1.05782390e+00 1.46673799e+00 -6.32075131e-01 -1.04736030e-01
-1.93124384e-01 1.01854420e+00 6.13598228e-01 2.28422418e-01
-1.02128065e+00 -7.33262420e-01 7.73945689e-01 1.31056145e-01
3.44591469e-01 -6.46167248e-02 1.96251467e-01 -1.10756433e+00
3.20243150e-01 -6.69700027e-01 3.23637873e-01 -5.40541291e-01
-1.25523889e+00 7.33841956e-01 -2.73601621e-01 -1.27910519e+00
2.06652820e-01 -1.28902406e-01 -8.03287566e-01 1.01431978e+00
-1.62723315e+00 -1.34143901e+00 -8.67679596e-01 4.90136027e-01
-4.38183665e-01 -2.70515501e-01 6.97332144e-01 6.90913260e-01
-8.35546434e-01 1.08803701e+00 -6.81980252e-02 5.32721400e-01
5.38036466e-01 -1.23831737e+00 5.34900486e-01 9.11145985e-01
-6.17656447e-02 6.37779832e-01 4.48863924e-01 -8.63042593e-01
-1.41889155e+00 -1.53219402e+00 7.22730160e-01 -1.35220423e-01
5.17412782e-01 -1.55037284e-01 -9.56130147e-01 5.37130177e-01
-4.93048579e-02 3.05341750e-01 3.40665102e-01 -1.69891208e-01
-4.00476396e-01 -3.89108211e-01 -1.18601012e+00 5.30501306e-01
1.09293735e+00 -5.34481704e-01 -3.33486080e-01 -1.98739246e-02
9.81184900e-01 -4.76465076e-01 -1.40810955e+00 7.69017220e-01
4.79996651e-01 -6.54735982e-01 1.06406486e+00 -3.92318547e-01
4.41989988e-01 -7.14801371e-01 5.79899438e-02 -1.54124439e+00
-4.30351138e-01 -2.70812809e-01 -3.33874673e-01 1.62866580e+00
2.40245923e-01 -7.46246934e-01 9.13205326e-01 2.55004227e-01
4.67998497e-02 -7.72940993e-01 -6.72115207e-01 -8.45187902e-01
-1.66671813e-01 4.40858573e-01 1.11262095e+00 1.40910506e+00
-2.69510269e-01 4.87046719e-01 -1.23572752e-01 5.95435202e-01
5.25314689e-01 2.01330602e-01 9.01184380e-01 -1.29237342e+00
-1.27844149e-02 -4.23026383e-01 -1.21787393e+00 -4.98754144e-01
1.67828828e-01 -1.24918759e+00 -5.63817918e-01 -1.70789897e+00
-6.11632653e-02 -7.21930623e-01 -3.94732863e-01 4.16849762e-01
-4.55159783e-01 -5.60388193e-02 1.62021518e-01 3.03626005e-02
-3.98440689e-01 7.76275992e-01 1.10064840e+00 -4.31476086e-01
-1.33152753e-01 -4.22072083e-01 -7.00349331e-01 3.32470149e-01
7.92976916e-01 -3.98809880e-01 -4.91413027e-01 -2.09466040e-01
6.78503633e-01 -2.16126874e-01 4.35055494e-01 -1.18993878e+00
3.95774752e-01 1.91365689e-01 1.64644733e-01 -5.14723599e-01
-2.13407185e-02 -8.47551167e-01 6.69398963e-01 4.78981495e-01
6.65712655e-02 2.60597348e-01 1.46934435e-01 7.52054036e-01
-4.26251203e-01 9.60533135e-03 6.48173034e-01 2.28709608e-01
-8.20004940e-01 4.88368213e-01 4.60640460e-01 3.20589632e-01
1.08060336e+00 -2.32758343e-01 -5.91347218e-01 -1.56141788e-01
-4.66648936e-01 3.96418661e-01 6.63204074e-01 4.04650241e-01
5.02868950e-01 -1.72264278e+00 -7.22195268e-01 4.47594114e-02
4.16084707e-01 1.58304676e-01 4.68486786e-01 6.63384855e-01
-6.44549012e-01 1.76948756e-01 -2.86729336e-01 -1.73113152e-01
-1.05707335e+00 4.88262832e-01 4.11387473e-01 -7.83375621e-01
-3.50053430e-01 6.90493166e-01 6.39292505e-03 -5.50754726e-01
-8.64790678e-02 -5.28066345e-02 -3.16499323e-01 6.95017502e-02
-1.19537950e-01 3.85260820e-01 2.15941504e-01 -7.35367000e-01
-4.65584308e-01 5.52919865e-01 -1.76896807e-02 5.54356575e-01
1.09413528e+00 1.37705043e-01 -1.84511259e-01 -2.01990709e-01
1.22369385e+00 2.81222939e-01 -6.68774307e-01 -3.54459703e-01
-2.27514938e-01 -3.11525166e-01 -1.29082441e-01 -5.97913802e-01
-1.48167765e+00 5.02632320e-01 5.45496702e-01 1.49673313e-01
9.63175595e-01 -2.09374979e-01 8.78320873e-01 1.62591621e-01
3.27448845e-01 -8.82883966e-01 4.44965139e-02 2.97236472e-01
6.34413004e-01 -1.04587698e+00 9.95731577e-02 -6.14510894e-01
-4.24501866e-01 7.64162660e-01 9.22283709e-01 -9.87391993e-02
5.67988694e-01 -2.14622959e-01 -3.37743610e-01 -5.56165934e-01
6.23739287e-02 -1.80925190e-01 4.55505371e-01 4.95175183e-01
-5.81385382e-02 1.87270194e-01 -4.40459162e-01 7.45031953e-01
-3.27924728e-01 -2.48112962e-01 2.09460244e-01 5.47934592e-01
4.06581536e-02 -1.18676341e+00 1.71798736e-01 2.83131719e-01
-2.07540050e-01 -3.20211202e-01 -4.85526621e-01 9.07231927e-01
3.47329497e-01 3.82376969e-01 -3.58465202e-02 -7.73996413e-01
4.34808433e-01 -2.43052095e-01 3.20732117e-01 -3.20155114e-01
-4.94883984e-01 -8.17750692e-01 -2.81273965e-02 -2.38951862e-01
-2.97061473e-01 1.81480870e-02 -1.30378103e+00 -6.16265595e-01
-6.22879863e-01 1.96411937e-01 3.57902437e-01 5.62352955e-01
8.03107440e-01 7.71185100e-01 7.90647686e-01 -1.25590965e-01
-7.48704299e-02 -7.04336762e-01 -6.80630267e-01 6.99795842e-01
-2.07127124e-01 -7.55253851e-01 -2.10297450e-01 -5.10816693e-01] | [8.729898452758789, 7.9181742668151855] |
faf4c410-eb63-4d88-9297-b85069bca5f0 | secseq-semantic-coding-for-sequence-to | null | null | https://openreview.net/forum?id=B1x5GPUOTX | https://openreview.net/pdf?id=B1x5GPUOTX | SeCSeq: Semantic Coding for Sequence-to-Sequence based Extreme Multi-label Classification | Extreme multi-label classification (XMC) aims at assigning to an instance the most relevant subset of labels from a colossal label set. There has been some success in formulating the multi-label problem as sequence-to-sequence (Seq2Seq) learning, where the positive class labels of each input instance are used as the corresponding output sequence. Seq2Seq methods, nonetheless, have not yet been scalable to the XMC setting due to the softmax bottleneck. In this paper, we propose a semantic coding framework, namely SeCSeq, for a Seq2Seq approach to the XMC problem. To circumvent the softmax bottleneck, SeCSeq compresses labels into sequences of semantic-aware compact codes, on which Seq2Seq models are trained. For inference, the generated semantic codes are then decompressed into sequences of positive labels using ensemble techniques. Preliminary experiments on XMC benchmark datasets show that SeCSeq is competitive with the state-of-the-art while requiring significantly fewer model parameters. | ['Yiming Yang', 'Inderjit S. Dhillon', 'Hsiang-Fu Yu', 'Wei-Cheng Chang'] | 2018-11-13 | null | null | null | nips-workshop-cdnnria-2018 | ['extreme-multi-label-classification'] | ['methodology'] | [ 1.13004339e+00 9.22711343e-02 -4.07611161e-01 -8.77859890e-01
-1.30043209e+00 -7.69588530e-01 4.01341528e-01 -3.48197185e-02
-5.20225763e-01 7.53385603e-01 1.28899664e-01 -2.71730155e-01
1.05828427e-01 -3.36120516e-01 -5.78811288e-01 -7.31704116e-01
2.94023722e-01 6.35538578e-01 -1.99833184e-01 3.00636262e-01
4.97348934e-01 -2.28628684e-02 -1.95902622e+00 1.01102936e+00
3.62553924e-01 1.07427740e+00 3.20710748e-01 8.63481879e-01
-6.18452311e-01 1.10026586e+00 -5.61800420e-01 -3.03645641e-01
-3.78738716e-02 -7.36277401e-01 -1.13878238e+00 -2.28114739e-01
5.89431167e-01 2.15513542e-01 3.64730358e-01 1.00095332e+00
4.58085895e-01 1.29669577e-01 8.09712589e-01 -1.38588405e+00
-4.21119958e-01 7.38051772e-01 -5.58340967e-01 -3.00366998e-01
2.59378582e-01 -4.23253834e-01 1.35347331e+00 -7.63408124e-01
7.40274251e-01 1.48942721e+00 7.50938058e-01 1.09499478e+00
-1.25864363e+00 -6.71308458e-01 1.69261862e-02 2.11394668e-01
-1.11224401e+00 -2.67003626e-01 3.18009168e-01 -2.10669398e-01
1.30559480e+00 4.45969462e-01 -2.21498478e-02 1.17716587e+00
-1.12907458e-02 9.92533088e-01 1.21013165e+00 -6.66895866e-01
4.30783957e-01 -8.41652155e-02 4.97475564e-02 7.53617823e-01
-5.65451026e-01 3.87209305e-03 -5.20295978e-01 -2.33666033e-01
3.89646553e-02 -2.27475818e-02 1.54198930e-01 5.11526577e-02
-1.26407671e+00 1.01629460e+00 2.11359471e-01 7.42202029e-02
-1.63422078e-02 7.69557595e-01 8.98650348e-01 2.94567078e-01
7.42856860e-01 3.55968237e-01 -8.51560414e-01 -3.77999574e-01
-9.68020439e-01 4.95080799e-02 7.35372603e-01 1.26387036e+00
6.37147546e-01 -3.12340498e-01 -1.96415588e-01 1.20341480e+00
1.46288261e-01 4.15229112e-01 6.29484594e-01 -1.35957849e+00
5.65260291e-01 2.88580269e-01 -3.79997969e-01 -3.71225387e-01
-4.76873547e-01 -2.34407350e-01 -5.98264456e-01 3.28790545e-02
4.53586243e-02 -1.93575174e-01 -8.38519990e-01 1.85786557e+00
1.53961718e-01 5.90393126e-01 2.91108131e-01 5.52448511e-01
6.27049744e-01 9.19715106e-01 6.02046967e-01 -6.20679744e-02
1.32840431e+00 -1.21863139e+00 -4.41015065e-01 -4.15448427e-01
1.35389435e+00 -6.83896959e-01 6.99335277e-01 2.68680453e-01
-8.16959321e-01 -5.98892391e-01 -6.88457191e-01 -1.11289345e-01
-4.39242780e-01 1.34059593e-01 4.23746675e-01 5.51400721e-01
-1.06689727e+00 6.43289924e-01 -5.11886299e-01 -1.42371356e-01
4.44540441e-01 3.10531735e-01 -2.16826215e-01 -2.39063337e-01
-1.08561003e+00 8.31472039e-01 9.38944101e-01 -3.53256762e-01
-6.50430501e-01 -7.38205194e-01 -1.05143571e+00 1.49549216e-01
3.34841371e-01 -3.28111738e-01 1.45654547e+00 -1.31742823e+00
-1.30895412e+00 1.07842898e+00 -3.23831737e-01 -2.13779375e-01
1.04282901e-01 1.61629543e-01 -4.27880108e-01 2.11333156e-01
3.24890226e-01 1.51631129e+00 7.04332590e-01 -1.28126097e+00
-1.08244574e+00 -6.72575831e-02 -2.03483805e-01 1.15312636e-01
-1.43150121e-01 3.77637833e-01 -1.41234130e-01 -7.31114686e-01
-6.60737902e-02 -1.30566180e+00 -4.46805544e-02 -1.50097176e-01
-3.05533588e-01 -6.84410691e-01 6.68499827e-01 -3.78075242e-01
1.07419908e+00 -2.36214972e+00 3.40533316e-01 -7.80624747e-02
-1.72550157e-01 1.14075951e-01 -5.63038826e-01 3.33109915e-01
-4.10215646e-01 6.45555407e-02 -4.86292362e-01 -7.86251307e-01
8.36808085e-02 3.86098385e-01 -2.69606054e-01 2.69505024e-01
3.45315188e-01 9.87247288e-01 -1.24499106e+00 -8.64967704e-01
-1.05779052e-01 2.79779732e-01 -5.56726217e-01 7.84418285e-02
-6.63193047e-01 2.12054774e-01 -1.57204181e-01 6.45735025e-01
4.92270350e-01 -5.58792174e-01 3.66075546e-01 3.18599284e-01
1.56020805e-01 8.73841569e-02 -8.86039793e-01 1.94110382e+00
-8.13756526e-01 4.09511745e-01 -3.52845311e-01 -1.19378448e+00
7.69226313e-01 4.47048932e-01 5.13682544e-01 -6.71064794e-01
-1.14774361e-01 5.12778103e-01 -5.04838645e-01 -5.74439228e-01
4.23380017e-01 -6.06688976e-01 -5.60774505e-01 6.88720763e-01
3.62735212e-01 8.40213075e-02 2.04746664e-01 6.68372288e-02
9.27722096e-01 5.27055621e-01 2.79172331e-01 -8.57963338e-02
4.36781198e-01 -4.30834517e-02 6.55201018e-01 5.94655812e-01
-2.27342770e-02 6.52686834e-01 4.96463358e-01 -2.69308388e-01
-1.14523888e+00 -6.75869286e-01 -8.01322684e-02 1.58857393e+00
-3.50968987e-01 -3.61535609e-01 -8.53485882e-01 -1.14570379e+00
9.85119566e-02 1.13926291e+00 -4.76570219e-01 -1.84957951e-01
-6.55575097e-01 -5.03959835e-01 7.41975844e-01 6.88648462e-01
1.40660733e-01 -1.30724585e+00 -6.57726467e-01 4.09069508e-01
-2.91199148e-01 -1.04156387e+00 -6.80087686e-01 6.48767591e-01
-7.73743570e-01 -7.44344056e-01 -7.50246465e-01 -1.13787425e+00
5.20845473e-01 -3.28254001e-03 1.12718010e+00 -1.59095392e-01
-2.06626609e-01 -4.57117781e-02 -5.79036236e-01 -1.90478086e-01
-8.24264348e-01 2.30162710e-01 -4.04693842e-01 2.55938945e-03
4.25832778e-01 -8.04371834e-02 -1.13773800e-01 5.67951202e-02
-9.48025644e-01 3.79287958e-01 4.36332554e-01 9.51756775e-01
6.98308706e-01 -9.36567113e-02 1.11147785e+00 -1.46688783e+00
1.38814330e-01 -7.44126737e-01 -2.92643487e-01 5.04549563e-01
-5.66243887e-01 3.11426491e-01 1.02043927e+00 -3.52824867e-01
-1.00562358e+00 4.73001152e-01 -2.52694756e-01 -2.97836542e-01
-4.27341849e-01 5.26464224e-01 1.27610713e-01 2.06396863e-01
3.37763458e-01 2.10617483e-01 8.07487126e-03 -4.51745719e-01
6.03027403e-01 1.09859288e+00 4.69973177e-01 -6.11485362e-01
-1.10024117e-01 1.05771624e-01 2.47446701e-01 -2.82717466e-01
-1.51292145e+00 -8.58444214e-01 -5.93092203e-01 -2.81062365e-01
1.18717313e+00 -8.48438501e-01 -6.45133078e-01 4.06028777e-01
-1.18688166e+00 -5.78085423e-01 -1.21652715e-01 1.19978145e-01
-1.06329548e+00 9.67774764e-02 -7.64941514e-01 -7.20767140e-01
-1.39697969e-01 -1.16841829e+00 1.56090581e+00 1.58195980e-02
-3.63909990e-01 -1.12753284e+00 -7.17671737e-02 5.11269331e-01
1.97030723e-01 1.24298230e-01 1.39567506e+00 -7.69524515e-01
-1.74017668e-01 -8.66607428e-02 -2.96393752e-01 4.25992817e-01
-2.95354247e-01 -4.73375648e-01 -1.14670575e+00 -4.45447505e-01
-3.48412842e-01 -8.96451831e-01 1.05108654e+00 4.57409658e-02
1.76385462e+00 -1.14539281e-01 -4.96194422e-01 5.70395589e-01
1.78973043e+00 2.01793179e-01 3.04030359e-01 1.39492750e-01
5.92066050e-01 5.67415178e-01 6.57613993e-01 3.52785975e-01
2.96172380e-01 6.97371066e-01 4.28817868e-01 1.60987124e-01
-3.06964695e-01 -1.76519394e-01 3.04171652e-01 7.98545301e-01
5.51082671e-01 -5.47730803e-01 -8.74310195e-01 2.78684914e-01
-1.64262962e+00 -9.61222470e-01 -1.27537981e-01 1.80352676e+00
9.52708364e-01 -7.66432062e-02 -2.07582086e-01 -5.13570085e-02
9.06174362e-01 9.19773802e-02 -6.31832480e-01 -6.41213059e-01
-4.43079807e-02 3.60260904e-01 6.17362022e-01 4.34459984e-01
-1.34439623e+00 9.52889442e-01 6.19312954e+00 1.18244779e+00
-9.61393237e-01 3.70190203e-01 7.30959415e-01 -2.62867361e-01
-2.54186004e-01 4.02973630e-02 -1.13804138e+00 7.44461536e-01
1.62620449e+00 4.14433062e-01 4.84350294e-01 7.37848341e-01
-4.20809060e-01 -8.71972069e-02 -1.24842894e+00 1.02302957e+00
4.59137797e-01 -1.18504786e+00 -1.65822566e-01 -1.10138312e-01
1.10445678e+00 -3.11471988e-02 1.31777033e-01 6.63397491e-01
4.16162193e-01 -9.94783461e-01 8.52677286e-01 4.97196048e-01
1.44712794e+00 -9.00879979e-01 7.00983524e-01 4.12450880e-01
-9.71031129e-01 -4.55812693e-01 -3.12043190e-01 3.51616144e-01
3.00524682e-01 4.01990294e-01 -7.61400104e-01 3.95674825e-01
1.97319388e-01 9.66349602e-01 -5.31012654e-01 5.81098974e-01
-5.02392836e-02 6.89268351e-01 1.64072942e-02 -1.82619587e-01
6.30484343e-01 2.06201315e-01 -1.44085154e-01 1.53102708e+00
4.20933902e-01 2.44644172e-02 3.66589874e-01 5.86053729e-01
-2.66919971e-01 1.21272039e-02 -3.29384089e-01 -5.34116328e-02
5.01763940e-01 1.15339375e+00 -8.80251288e-01 -6.30446434e-01
-7.36579716e-01 1.40036166e+00 6.60403728e-01 1.38787448e-01
-8.17724347e-01 -4.86291587e-01 3.16892028e-01 -7.05059707e-01
3.35963011e-01 3.24628860e-01 -4.41043556e-01 -8.27007651e-01
-3.73695076e-01 -8.77043784e-01 6.80540204e-01 -8.22852552e-01
-1.18376040e+00 3.23495388e-01 -1.82263762e-01 -1.06517196e+00
-4.03118879e-01 -6.53187573e-01 -4.39708941e-02 9.41308141e-01
-1.50617325e+00 -1.15831244e+00 1.74620315e-01 -5.07742651e-02
9.66756165e-01 -2.89075434e-01 1.32589173e+00 3.26820672e-01
-3.24935317e-01 6.94762409e-01 6.26593351e-01 -2.27520481e-01
8.71076941e-01 -1.53781462e+00 3.99159640e-01 1.55632004e-01
7.52022192e-02 1.67249236e-02 2.86732346e-01 -6.10553384e-01
-8.14052820e-01 -1.59780550e+00 1.48642886e+00 -3.60754311e-01
3.49082649e-01 -4.07914191e-01 -6.78510904e-01 5.99045336e-01
-1.07542602e-02 8.65871683e-02 1.14356542e+00 2.25624703e-02
-6.05371952e-01 2.69867867e-01 -1.02464187e+00 1.82944909e-01
1.02968705e+00 -9.01892662e-01 -2.24464610e-01 4.58584011e-01
9.74488676e-01 -1.37377232e-01 -9.31454360e-01 1.71368107e-01
4.15495187e-01 -3.93769205e-01 8.85634780e-01 -7.09541023e-01
9.72562909e-01 6.22712970e-02 -4.94715512e-01 -1.48628891e+00
-4.12103564e-01 1.29669592e-01 2.33271513e-02 1.27823150e+00
6.78773165e-01 -3.13395739e-01 7.55864561e-01 4.03502844e-02
-3.88514727e-01 -9.37430263e-01 -9.57532704e-01 -7.96669304e-01
2.36858860e-01 -4.51360673e-01 5.14191508e-01 1.13221812e+00
1.50553644e-01 4.15813267e-01 -3.44849527e-01 -2.27351025e-01
6.78414285e-01 4.19122815e-01 -1.22917697e-01 -1.12653399e+00
-4.92172122e-01 -3.54372352e-01 -1.00383699e-01 -8.05319965e-01
9.97995734e-01 -1.75445986e+00 4.50322956e-01 -1.19634855e+00
5.51316559e-01 -6.83703423e-01 -4.97656643e-01 7.56030679e-01
-4.66488183e-01 4.10790563e-01 3.44133079e-01 1.07133009e-01
-1.03988421e+00 2.30890691e-01 9.34032381e-01 -1.74007565e-01
4.12562191e-01 -3.01467508e-01 -5.12188435e-01 2.70784855e-01
7.52820492e-01 -9.07934964e-01 -4.05530363e-01 -4.35804993e-01
3.28619182e-01 4.49245691e-01 7.50093237e-02 -9.04734135e-01
1.39518127e-01 -1.23019762e-01 9.83987451e-02 -5.23281574e-01
3.15494001e-01 -6.87020183e-01 1.23055823e-01 4.62555468e-01
-1.36856496e+00 -9.81306210e-02 -1.68949272e-02 5.51992774e-01
-1.32826582e-01 -8.57719183e-01 9.26917195e-01 -2.50672013e-01
-9.34324086e-01 -4.23244871e-02 -3.70488256e-01 3.23052198e-01
1.11877990e+00 1.47636324e-01 -1.18442856e-01 2.61856057e-03
-7.60429561e-01 2.94978708e-01 2.31407315e-01 4.25442874e-01
4.38188076e-01 -1.43757176e+00 -8.07876885e-01 2.26117939e-01
3.67512137e-01 -1.06124945e-01 2.83123106e-01 2.59851187e-01
-2.39244506e-01 8.07736337e-01 6.83775917e-02 -4.89574343e-01
-1.43086004e+00 8.40303898e-01 1.13257632e-01 -1.69723168e-01
-3.55025113e-01 1.10732746e+00 4.88229170e-02 -8.83921146e-01
2.78315157e-01 2.30614617e-02 -9.95234102e-02 2.54581630e-01
5.73989391e-01 4.17052150e-01 -2.22743838e-03 -7.19797552e-01
-1.87008336e-01 4.76065964e-01 -7.01856315e-02 -1.95113257e-01
1.27560878e+00 5.30976951e-02 -1.39297649e-01 6.63725734e-01
1.95424545e+00 -7.96155512e-01 -1.28556538e+00 -1.84599340e-01
4.67410415e-01 -1.65974125e-01 -6.22413568e-02 -1.09182847e+00
-9.01313424e-01 8.97582114e-01 4.82839406e-01 -2.06137419e-01
9.67696488e-01 1.54637784e-01 9.27680969e-01 2.75716692e-01
4.88909960e-01 -1.30226564e+00 3.45094614e-02 6.63366377e-01
3.89079303e-01 -1.12038386e+00 -6.29562676e-01 -3.97210002e-01
-7.04979777e-01 1.06965911e+00 4.55916733e-01 2.16846347e-01
4.26437378e-01 3.19219708e-01 6.91546202e-02 2.43040379e-02
-1.13770533e+00 -2.10919529e-02 -2.59787619e-01 3.82447124e-01
6.17101431e-01 2.71298438e-01 -1.18096486e-01 2.81119049e-02
1.35448322e-01 3.21758464e-02 2.68235117e-01 9.57702458e-01
-5.30507505e-01 -1.29854119e+00 -2.35567033e-01 5.83527505e-01
-5.68844974e-01 -3.19645792e-01 -1.37379076e-02 5.14639392e-02
4.58388448e-01 9.04157102e-01 1.99344501e-01 -2.02713162e-01
5.61826564e-02 8.43183696e-01 3.94528508e-01 -8.86117578e-01
-5.86547136e-01 -1.05259858e-01 3.23576748e-01 -5.03125608e-01
-3.81328344e-01 -1.02376211e+00 -1.57391298e+00 -1.28610050e-02
-2.27966502e-01 1.28054827e-01 8.78071666e-01 9.93979812e-01
3.63241076e-01 6.93495870e-01 6.72962785e-01 -5.69453239e-01
-8.84154439e-01 -9.49879646e-01 -5.21419346e-01 6.43961072e-01
2.57695168e-01 -3.75662386e-01 -1.78899646e-01 2.60134608e-01] | [9.565849304199219, 4.412591457366943] |
e9dab7af-f06b-4bec-811c-f5ac0458fa63 | robustfill-neural-program-learning-under | 1703.07469 | null | http://arxiv.org/abs/1703.07469v1 | http://arxiv.org/pdf/1703.07469v1.pdf | RobustFill: Neural Program Learning under Noisy I/O | The problem of automatically generating a computer program from some
specification has been studied since the early days of AI. Recently, two
competing approaches for automatic program learning have received significant
attention: (1) neural program synthesis, where a neural network is conditioned
on input/output (I/O) examples and learns to generate a program, and (2) neural
program induction, where a neural network generates new outputs directly using
a latent program representation.
Here, for the first time, we directly compare both approaches on a
large-scale, real-world learning task. We additionally contrast to rule-based
program synthesis, which uses hand-crafted semantics to guide the program
generation. Our neural models use a modified attention RNN to allow encoding of
variable-sized sets of I/O pairs. Our best synthesis model achieves 92%
accuracy on a real-world test set, compared to the 34% accuracy of the previous
best neural synthesis approach. The synthesis model also outperforms a
comparable induction model on this task, but we more importantly demonstrate
that the strength of each approach is highly dependent on the evaluation metric
and end-user application. Finally, we show that we can train our neural models
to remain very robust to the type of noise expected in real-world data (e.g.,
typos), while a highly-engineered rule-based system fails entirely. | ['Abdel-rahman Mohamed', 'Surya Bhupatiraju', 'Rishabh Singh', 'Jacob Devlin', 'Pushmeet Kohli', 'Jonathan Uesato'] | 2017-03-21 | robustfill-neural-program-learning-under-1 | https://icml.cc/Conferences/2017/Schedule?showEvent=661 | http://proceedings.mlr.press/v70/devlin17a/devlin17a.pdf | icml-2017-8 | ['program-induction'] | ['computer-code'] | [ 6.08476996e-01 3.16674203e-01 -4.85396832e-01 -4.51792657e-01
-8.66168559e-01 -5.62066734e-01 5.68185031e-01 2.61495948e-01
-8.00251365e-02 4.53043342e-01 -3.23411711e-02 -9.19865191e-01
3.47956032e-01 -1.06178796e+00 -1.34557366e+00 -1.24898545e-01
-2.34531667e-02 3.62157136e-01 9.68994871e-02 -1.14530995e-01
3.74962777e-01 5.72543107e-02 -1.88527071e+00 4.85580504e-01
8.27544510e-01 5.70444107e-01 1.30811706e-01 1.04484820e+00
-1.31907508e-01 1.08103514e+00 -7.87281632e-01 -2.16360837e-01
-1.02111939e-02 -4.03688967e-01 -8.01849365e-01 -2.87346333e-01
3.36807519e-01 -2.61493355e-01 -2.09656000e-01 1.04366755e+00
2.14136526e-01 -2.64279008e-01 4.67684656e-01 -1.27936995e+00
-7.63963819e-01 1.07528114e+00 -1.27285331e-01 -3.08666021e-01
4.96116430e-01 5.43657422e-01 1.09888613e+00 -6.02921009e-01
6.49162650e-01 1.07109392e+00 6.08706117e-01 5.96966267e-01
-1.71906602e+00 -5.59575856e-01 -1.22054398e-01 -3.44241798e-01
-1.01972711e+00 -2.93139100e-01 5.93407154e-01 -7.44492769e-01
1.48043656e+00 2.61108740e-03 2.40312576e-01 1.13120008e+00
1.86514094e-01 1.00134635e+00 9.39083278e-01 -6.72919810e-01
3.39675456e-01 1.95931658e-01 3.71400803e-01 9.90478754e-01
5.30649312e-02 5.16477823e-01 4.70664725e-03 -4.49815273e-01
6.10137582e-01 -2.06697553e-01 -2.30838791e-01 -4.05811399e-01
-1.30077696e+00 9.16691422e-01 2.86342174e-01 2.92865783e-01
-7.61903077e-02 6.40543640e-01 7.12537527e-01 5.69068730e-01
-3.62196146e-03 9.37067330e-01 -6.06337309e-01 -2.65498251e-01
-1.21192002e+00 5.10430038e-01 1.13183284e+00 1.31235576e+00
7.51117587e-01 5.37050366e-01 -2.77817667e-01 4.68138307e-01
1.20486557e-01 2.97805578e-01 6.87440455e-01 -7.24916577e-01
5.37426472e-01 6.68574631e-01 -4.02238183e-02 -5.30089021e-01
-8.62188712e-02 -1.55404851e-01 -5.54794490e-01 5.35306752e-01
4.08038884e-01 -4.17180240e-01 -8.35879266e-01 1.77528608e+00
-5.36759853e-01 -1.32759124e-01 3.07267215e-02 3.48186582e-01
6.09688818e-01 8.42150986e-01 -2.44184434e-01 7.21527934e-02
8.51021945e-01 -1.14395285e+00 -2.45051846e-01 -3.63067210e-01
8.61105621e-01 -4.07925576e-01 1.41009438e+00 5.23767769e-01
-1.01844311e+00 -7.62010813e-01 -1.26396537e+00 2.21682847e-01
-4.14144844e-01 3.11362684e-01 9.53365386e-01 7.07233548e-01
-1.31413543e+00 5.84499955e-01 -6.28424823e-01 -1.10803388e-01
1.22614525e-01 5.94952285e-01 -1.46477833e-01 1.04254752e-01
-8.93628776e-01 5.38842976e-01 6.52052224e-01 -2.99958557e-01
-1.08038771e+00 -5.60783863e-01 -1.02767503e+00 2.12503001e-01
4.28372025e-01 -6.61170900e-01 1.71149588e+00 -1.45615149e+00
-1.65655136e+00 8.09195638e-01 -1.38091728e-01 -6.95212841e-01
8.67557600e-02 4.15975526e-02 -3.09487671e-01 -5.60863376e-01
-7.78114721e-02 6.33739769e-01 6.67003453e-01 -1.26474094e+00
-5.50022542e-01 4.21885960e-02 2.09209710e-01 -6.18653476e-01
-1.97760314e-02 2.12788597e-01 -1.56108171e-01 -6.94618940e-01
-6.82807326e-01 -1.02803469e+00 -3.00316602e-01 -5.07836342e-01
-4.12869424e-01 -4.41084653e-01 5.55997670e-01 -4.98871684e-01
1.45149970e+00 -2.07523942e+00 2.28468940e-01 1.91092640e-01
6.08131252e-02 2.56045431e-01 -2.91148335e-01 3.27587724e-01
-4.54992503e-01 3.61835659e-01 -4.57490444e-01 1.61338016e-01
4.53898430e-01 8.21972787e-02 -8.03076267e-01 1.00240763e-02
5.05415857e-01 1.17558503e+00 -8.66649866e-01 -1.36430576e-01
-2.38291964e-01 -5.63812703e-02 -1.10350192e+00 5.20243168e-01
-1.03431451e+00 -2.14911953e-01 -2.13182956e-01 5.49929976e-01
-1.09939963e-01 -3.41062129e-01 2.03083485e-01 2.24025771e-01
-1.33944556e-01 4.10197914e-01 -9.04130459e-01 1.63886321e+00
-9.34589088e-01 9.70676720e-01 -1.76546052e-01 -8.96990955e-01
9.61879313e-01 3.68350744e-01 -1.89398527e-01 -4.92352843e-01
-6.20738566e-02 3.79275680e-01 2.20450968e-01 -4.56270546e-01
5.97609758e-01 1.74207374e-01 -6.48463011e-01 8.97683263e-01
1.03051789e-01 -4.66211259e-01 2.76651680e-01 5.76662943e-02
1.43510902e+00 4.18288410e-01 3.34349394e-01 -1.86937422e-01
3.09669644e-01 1.74327061e-01 3.34474087e-01 1.16851985e+00
2.93438375e-01 5.61479926e-01 1.07404792e+00 -5.38859546e-01
-1.25813735e+00 -6.47994161e-01 3.16967666e-01 1.33260036e+00
-5.36110401e-01 -6.00060165e-01 -9.30454791e-01 -8.55411828e-01
-2.36520190e-02 1.07322907e+00 -4.93481219e-01 -2.42580652e-01
-9.30735767e-01 -4.62198108e-01 8.72074723e-01 7.28384316e-01
3.63481902e-02 -1.49252999e+00 -6.05935156e-01 2.83249825e-01
2.80373633e-01 -7.26796389e-01 -5.19954860e-01 5.64957082e-01
-7.59267092e-01 -9.46777582e-01 -3.05772424e-01 -1.06102228e+00
7.55096495e-01 -5.99780560e-01 1.71368742e+00 3.29991519e-01
-5.64831458e-02 1.64871320e-01 -2.42148601e-02 -3.28471005e-01
-1.15809846e+00 3.41485918e-01 -1.99750036e-01 -4.28596407e-01
2.90191352e-01 -5.20421028e-01 1.26571327e-01 -1.42055050e-01
-1.10195184e+00 8.66455510e-02 8.46702456e-01 1.13806295e+00
2.00302020e-01 2.19643056e-01 5.00143826e-01 -1.37606001e+00
8.76047075e-01 -3.40387255e-01 -1.11618066e+00 2.49760851e-01
-7.44106352e-01 6.59556329e-01 1.03909147e+00 -5.16560614e-01
-7.87445843e-01 3.05804729e-01 -2.47788697e-01 -1.04496956e-01
-4.42641199e-01 7.43631840e-01 -1.81644097e-01 1.72043949e-01
9.68499064e-01 3.39534730e-01 -1.75228417e-01 -6.26398064e-03
3.95982951e-01 7.29914248e-01 6.62554622e-01 -1.01436067e+00
7.96248794e-01 -3.22446197e-01 -3.79626960e-01 -2.23933578e-01
-1.53536439e-01 3.08344275e-01 -4.27944750e-01 3.55362624e-01
5.21859646e-01 -6.98151946e-01 -6.86170757e-01 3.72852117e-01
-1.47332692e+00 -1.12062490e+00 -2.02081874e-01 -5.00770845e-03
-7.79732704e-01 2.43036486e-02 -6.39551997e-01 -5.74994147e-01
-3.25538695e-01 -1.72668171e+00 1.03510261e+00 -1.22259438e-01
-6.91175759e-01 -8.85747850e-01 -3.24291699e-02 -1.99729756e-01
4.73538607e-01 3.54622066e-01 1.60018098e+00 -7.43116915e-01
-8.96832049e-01 -3.63315463e-01 -1.29644170e-01 5.50107539e-01
-5.36507294e-02 3.89625072e-01 -7.32848406e-01 -3.84142175e-02
-2.23879322e-01 -4.32977855e-01 5.87004721e-01 2.00087279e-01
1.43130064e+00 -4.05780762e-01 -2.66164333e-01 4.15520310e-01
1.38205481e+00 3.24870944e-01 5.28987706e-01 4.11495529e-02
5.35578966e-01 2.82994062e-01 1.58890665e-01 1.16068937e-01
1.91385165e-01 5.61293602e-01 2.65848786e-01 3.43268663e-02
6.44555241e-02 -4.81867224e-01 7.06852138e-01 5.10552406e-01
2.70626098e-01 -1.31269947e-01 -1.28401232e+00 6.17943764e-01
-1.74255216e+00 -7.79311717e-01 1.18816137e-01 2.37040210e+00
1.15485501e+00 6.21183515e-01 3.52576107e-01 5.83790578e-02
5.50505519e-01 -9.64638442e-02 -4.41050470e-01 -8.05257142e-01
2.64037818e-01 7.92168617e-01 4.80411500e-01 5.55647135e-01
-8.89686048e-01 9.31917369e-01 7.12729311e+00 5.51980078e-01
-1.34340751e+00 -2.06472918e-01 5.16085505e-01 1.41787589e-01
-6.33453906e-01 8.40079263e-02 -5.50051510e-01 5.12526631e-01
1.35121381e+00 -4.51290518e-01 7.55314052e-01 1.23981643e+00
-1.49745733e-01 1.62479937e-01 -1.87118435e+00 6.06014311e-01
1.23717710e-02 -1.39497638e+00 -3.73813994e-02 -1.52028695e-01
8.20344090e-01 1.21464934e-02 -6.43053204e-02 1.12667918e+00
8.64908040e-01 -1.38026273e+00 6.89439297e-01 3.37203979e-01
8.78470480e-01 -8.15345645e-01 5.99502623e-01 5.12941420e-01
-8.76234472e-01 -2.23587289e-01 2.30569020e-01 -2.04268217e-01
-2.96639264e-01 6.35900557e-01 -9.17908430e-01 4.53140177e-02
1.15421139e-01 4.42651868e-01 -8.21839094e-01 6.83356583e-01
-4.43760544e-01 7.73041427e-01 -6.83985353e-02 -3.24475974e-01
1.67630985e-01 4.45847958e-01 2.23511189e-01 1.47023833e+00
3.24054301e-01 -3.74452502e-01 2.19849385e-02 1.63349366e+00
-2.70642132e-01 -2.58463293e-01 -9.39546108e-01 -4.63555783e-01
2.12288797e-01 8.42901707e-01 -3.60784352e-01 -5.69674015e-01
-4.89237517e-01 6.42660856e-01 3.11512083e-01 5.04433095e-01
-8.72241855e-01 -8.30923378e-01 1.97110012e-01 -1.59929767e-02
3.11799288e-01 -1.67310536e-01 -3.58279139e-01 -1.12261248e+00
7.72924274e-02 -1.48451710e+00 -1.93250664e-02 -7.12539911e-01
-7.49114215e-01 8.62806857e-01 1.17596230e-02 -6.84841633e-01
-1.04249287e+00 -7.17242420e-01 -7.66368389e-01 9.78795707e-01
-1.05246854e+00 -8.33530843e-01 -3.09458817e-03 -5.93802705e-02
4.84376788e-01 -5.00819266e-01 1.02813506e+00 2.58676022e-01
-5.36893547e-01 8.75024319e-01 -2.89254695e-01 3.83723736e-01
4.22359854e-01 -1.49738395e+00 9.96077299e-01 9.01894569e-01
-3.70854102e-02 1.01128232e+00 7.49728560e-01 -5.34138322e-01
-1.84703147e+00 -1.32789278e+00 8.56205523e-01 -5.97296476e-01
6.58544064e-01 -4.68560576e-01 -8.70326042e-01 1.09631133e+00
2.40789294e-01 -7.42882416e-02 4.77467656e-01 1.78705111e-01
-6.91106379e-01 1.25004247e-01 -7.32015908e-01 7.68821418e-01
6.73908472e-01 -7.56245673e-01 -6.33358002e-01 2.79266417e-01
9.58985806e-01 -5.34848809e-01 -8.06085587e-01 4.03829396e-01
5.84274530e-01 -9.20972168e-01 5.97724140e-01 -6.88184798e-01
1.04090083e+00 -2.98390001e-01 -1.43216819e-01 -1.29092777e+00
-7.14718923e-02 -7.42527544e-01 -3.21656823e-01 1.19959998e+00
7.03154266e-01 -4.23439473e-01 7.83571601e-01 7.21670032e-01
-2.16403693e-01 -8.10826898e-01 -7.52552673e-02 -6.06264889e-01
2.87062883e-01 -6.45107508e-01 8.71527910e-01 7.26770878e-01
1.72849029e-01 5.41882992e-01 -1.62469730e-01 -5.27604632e-02
1.64006591e-01 5.02065182e-01 1.05507338e+00 -8.73560250e-01
-1.07497311e+00 -9.71753597e-01 -1.09777302e-01 -1.04089928e+00
6.92888081e-01 -1.15425956e+00 2.71558046e-01 -9.91646588e-01
1.75306555e-02 -3.20512116e-01 2.86710877e-02 6.99313879e-01
-6.22920655e-02 -2.56274998e-01 -3.75915468e-02 -1.83354631e-01
-2.96169698e-01 7.47253373e-02 5.36256075e-01 -4.48791444e-01
-3.84197652e-01 2.26207078e-01 -8.67788911e-01 5.44563890e-01
6.63626909e-01 -5.27453601e-01 -3.07411969e-01 -5.05977154e-01
5.69758356e-01 4.81439143e-01 3.87446135e-01 -1.01034689e+00
2.06837580e-01 -1.78302601e-01 -2.37115815e-01 -1.15779057e-01
-4.67829853e-01 -5.57352722e-01 1.38388827e-01 5.73904574e-01
-8.02039087e-01 2.35385045e-01 3.18576038e-01 2.54455358e-01
-2.97826588e-01 -5.35966277e-01 5.13684332e-01 -2.08625153e-01
-7.65220165e-01 2.38664206e-02 -5.64081311e-01 -7.96607956e-02
6.16825402e-01 7.54280239e-02 -2.20060378e-01 -2.59911925e-01
-2.60594100e-01 -1.62562989e-02 5.82388937e-01 6.17904305e-01
3.73725146e-01 -1.25580096e+00 -5.84937274e-01 4.90668893e-01
3.07377249e-01 -4.59401309e-03 -4.48687017e-01 3.22783828e-01
-5.12839258e-01 5.84466577e-01 2.54203193e-03 -5.05169809e-01
-1.02205622e+00 8.05421710e-01 3.34472865e-01 -4.84292537e-01
-2.76274949e-01 4.50722903e-01 1.17939077e-01 -8.23320866e-01
3.57336760e-01 -8.83175433e-01 3.79781485e-01 -5.45448661e-01
4.37691748e-01 -1.32581517e-01 6.99180216e-02 -2.46489532e-02
-7.84280375e-02 7.34467432e-02 -1.15803508e-02 -9.45924688e-03
1.27225769e+00 9.01195705e-01 -3.59665304e-01 6.14937723e-01
1.19589937e+00 -1.73684031e-01 -8.58343422e-01 -3.15223932e-01
2.93139309e-01 2.78655607e-02 -3.85049805e-02 -8.52502465e-01
-8.32194090e-01 9.19548571e-01 1.56528652e-01 2.70987272e-01
9.84214008e-01 -3.47926974e-01 7.35597968e-01 7.87984610e-01
4.75256473e-01 -8.17746937e-01 1.62033275e-01 7.38198519e-01
7.62758493e-01 -1.20941556e+00 -3.81585926e-01 3.53019536e-02
-3.62669498e-01 1.29496956e+00 7.58518040e-01 -2.50674278e-01
2.15557441e-01 9.09306467e-01 -3.73445600e-01 2.28053927e-02
-1.12967801e+00 1.80736810e-01 1.15787312e-01 5.51760674e-01
9.73822892e-01 8.93512815e-02 2.45783418e-01 7.78253853e-01
-4.39725965e-01 4.97847289e-01 7.29309082e-01 1.01437259e+00
-1.01742372e-01 -1.37866473e+00 -2.09686369e-01 7.20077991e-01
-3.48109633e-01 -2.54699916e-01 -2.81603277e-01 8.70601833e-01
-9.93992761e-02 4.71362293e-01 -5.88206053e-02 -4.94981676e-01
3.51655185e-01 4.19233441e-01 5.37896991e-01 -1.09085619e+00
-8.83900940e-01 -3.07316571e-01 -2.56835949e-03 -4.24057722e-01
1.11659907e-01 -4.07214493e-01 -1.17526829e+00 -1.43333241e-01
-9.63701978e-02 4.07778211e-02 4.42755520e-01 6.68029606e-01
3.24442506e-01 8.15675378e-01 3.70227277e-01 -7.72704959e-01
-7.07120478e-01 -6.78314328e-01 -2.02084601e-01 3.27096909e-01
5.18707573e-01 -7.00700060e-02 -1.84041187e-01 4.05517489e-01] | [8.101204872131348, 7.51939582824707] |
4a0a549f-dc50-45be-9f77-90d774c5d1d3 | geometric-change-detection-in-digital-twins | 2103.08201 | null | https://arxiv.org/abs/2103.08201v1 | https://arxiv.org/pdf/2103.08201v1.pdf | Geometric Change Detection in Digital Twins using 3D Machine Learning | Digital twins are meant to bridge the gap between real-world physical systems and virtual representations. Both stand-alone and descriptive digital twins incorporate 3D geometric models, which are the physical representations of objects in the digital replica. Digital twin applications are required to rapidly update internal parameters with the evolution of their physical counterpart. Due to an essential need for having high-quality geometric models for accurate physical representations, the storage and bandwidth requirements for storing 3D model information can quickly exceed the available storage and bandwidth capacity. In this work, we demonstrate a novel approach to geometric change detection in the context of a digital twin. We address the issue through a combined solution of Dynamic Mode Decomposition (DMD) for motion detection, YOLOv5 for object detection, and 3D machine learning for pose estimation. DMD is applied for background subtraction, enabling detection of moving foreground objects in real-time. The video frames containing detected motion are extracted and used as input to the change detection network. The object detection algorithm YOLOv5 is applied to extract the bounding boxes of detected objects in the video frames. Furthermore, the rotational pose of each object is estimated in a 3D pose estimation network. A series of convolutional neural networks conducts feature extraction from images and 3D model shapes. Then, the network outputs the estimated Euler angles of the camera orientation with respect to the object in the input image. By only storing data associated with a detected change in pose, we minimize necessary storage and bandwidth requirements while still being able to recreate the 3D scene on demand. | ['Omer San', 'Mandar Tabib', 'Adil Rasheed', 'Julia Maria Graham', 'Tiril Sundby'] | 2021-03-15 | null | null | null | null | ['motion-detection'] | ['computer-vision'] | [ 3.59352201e-01 -4.21397328e-01 1.19259253e-01 3.09289575e-01
-1.82918996e-01 -3.83382946e-01 3.26720834e-01 -1.85772240e-01
-2.50237763e-01 9.55264345e-02 -6.72380745e-01 -1.98791400e-01
1.56034887e-01 -8.88603210e-01 -7.34292388e-01 -7.06301212e-01
-1.32699430e-01 2.03439489e-01 7.87283778e-01 7.49135688e-02
3.50695908e-01 1.16152442e+00 -1.85891247e+00 6.30878806e-02
3.92476320e-01 1.46166205e+00 4.73689497e-01 1.05917668e+00
-7.85694867e-02 4.46199834e-01 -7.13211358e-01 7.72056952e-02
5.08702695e-01 2.41182931e-02 -1.25168294e-01 5.18586874e-01
6.56488478e-01 -7.51076758e-01 -8.01777184e-01 7.33000100e-01
3.00566226e-01 5.84212579e-02 4.68255371e-01 -1.23496747e+00
-1.17104471e-01 -3.13116729e-01 -7.21620739e-01 4.19498652e-01
3.46077204e-01 2.01376095e-01 3.29858840e-01 -1.05945027e+00
6.60630167e-01 1.26559412e+00 6.61917567e-01 2.16736764e-01
-1.03539383e+00 -4.24164921e-01 2.18360499e-01 3.69418651e-01
-1.40910757e+00 -4.74405497e-01 9.81132925e-01 -3.49214524e-01
1.00869334e+00 4.52380151e-01 9.08748269e-01 4.52417344e-01
3.39708507e-01 5.19276381e-01 2.49440327e-01 -5.63716233e-01
1.63661912e-01 -2.70332783e-01 -2.55867809e-01 7.63422966e-01
2.73293287e-01 9.41018201e-03 -5.86874962e-01 -1.04187831e-01
1.32986891e+00 1.05009280e-01 -3.40785831e-01 -6.39064372e-01
-1.08591449e+00 1.13086425e-01 1.16465703e-01 -7.16618598e-02
-3.98509234e-01 5.39137185e-01 1.86215535e-01 8.25680345e-02
3.72870654e-01 1.85742453e-02 -4.70030248e-01 -1.65617645e-01
-8.01039994e-01 6.30165115e-02 4.05859560e-01 1.01199901e+00
5.36625922e-01 3.04144621e-01 4.27377284e-01 3.07079375e-01
3.66264164e-01 7.98126578e-01 3.45937371e-01 -1.06312835e+00
3.10063660e-01 8.31229448e-01 1.64350286e-01 -1.48115075e+00
-2.19379544e-01 2.04774626e-02 -6.88121676e-01 3.98140997e-01
3.36276054e-01 2.49930441e-01 -8.37473869e-01 1.24406695e+00
9.03547227e-01 5.00672936e-01 -2.60045171e-01 7.78730989e-01
5.47252297e-01 8.79520833e-01 -5.85819006e-01 -2.32997641e-01
1.36163402e+00 -3.24319571e-01 -5.23024261e-01 -1.92636311e-01
4.33587819e-01 -8.92157197e-01 4.08374757e-01 3.77343357e-01
-1.21860242e+00 -8.45988274e-01 -1.33365309e+00 3.69436592e-02
-1.94991723e-01 2.87823558e-01 1.78556681e-01 4.24432695e-01
-7.94472814e-01 6.76120341e-01 -1.33763838e+00 -6.04801551e-02
1.76014274e-01 5.69421470e-01 -5.97093552e-02 2.15767309e-01
-8.92576456e-01 9.23085392e-01 2.29293719e-01 3.24223518e-01
-6.75694823e-01 -6.36240125e-01 -8.38580251e-01 -5.16936332e-02
5.79651952e-01 -5.01753747e-01 1.26618087e+00 -1.04100943e+00
-1.40522146e+00 7.11912334e-01 5.47680557e-02 -2.46874377e-01
6.13402188e-01 -1.82225212e-01 -4.76060688e-01 5.87757945e-01
-2.38992363e-01 2.74418235e-01 1.42047155e+00 -1.18112576e+00
-6.77611530e-01 -3.71840030e-01 -1.55423611e-01 1.75982878e-01
-1.82599008e-01 -1.57836646e-01 -9.47402537e-01 -5.13412774e-01
6.71371460e-01 -8.64923775e-01 -1.18866703e-02 6.92863286e-01
-2.90432628e-02 2.07193732e-01 1.69796574e+00 -7.54259050e-01
1.01310015e+00 -2.23953152e+00 3.90096456e-02 4.05055620e-02
2.46288270e-01 5.20560205e-01 -2.79218554e-02 3.73862982e-02
-3.89669985e-02 -3.52451056e-01 1.42403260e-01 -2.59938478e-01
-3.42198908e-01 -4.11823057e-02 -3.22653234e-01 6.56600535e-01
3.83757442e-01 3.56096148e-01 -6.30277276e-01 -4.37202245e-01
6.37765467e-01 6.08507335e-01 -2.79477119e-01 1.37131289e-01
-3.43386024e-01 4.35865931e-02 -5.15760124e-01 5.25263429e-01
1.05760574e+00 -9.46473032e-02 3.70286167e-01 -6.00404799e-01
-2.94756263e-01 1.81785509e-01 -1.66043413e+00 1.16759253e+00
-3.69340569e-01 7.49048710e-01 1.64873570e-01 -7.36328185e-01
9.56538796e-01 2.18003020e-01 6.97079599e-01 -6.03864372e-01
2.11343244e-01 5.87972021e-03 -2.94823676e-01 -5.51475525e-01
8.07815552e-01 4.32536036e-01 7.45227933e-02 4.81250226e-01
-4.76196498e-01 -3.51880431e-01 -6.47091027e-03 1.58330590e-01
9.14219856e-01 4.40055341e-01 2.23392829e-01 2.32055619e-01
4.93679047e-01 1.08272303e-02 6.23999834e-01 3.47355843e-01
5.58702648e-03 6.14756525e-01 2.28147462e-01 -6.55504823e-01
-1.15465593e+00 -1.13713741e+00 7.30534792e-02 4.49066073e-01
6.05013549e-01 -3.33990455e-01 -5.18604696e-01 -2.07567245e-01
1.84437022e-01 3.60190988e-01 -3.18959326e-01 -2.52342165e-01
-1.11190367e+00 -4.28502917e-01 1.51866138e-01 5.17341793e-01
4.89505827e-01 -5.49648702e-01 -1.57067370e+00 2.49017224e-01
8.01179782e-02 -1.24256682e+00 -4.53041434e-01 -5.58537804e-02
-1.12573683e+00 -1.26597047e+00 -1.91453069e-01 -7.10260987e-01
8.47223341e-01 6.95071280e-01 6.86595500e-01 4.26901191e-01
-7.56116390e-01 5.00035346e-01 -7.23452270e-02 -9.84136313e-02
-4.98727381e-01 -5.52447617e-01 3.28732729e-01 7.53011554e-02
-2.27791467e-03 -4.59806383e-01 -7.08658457e-01 4.79622096e-01
-9.42667365e-01 3.98866832e-01 2.21638426e-01 3.40410948e-01
4.90106255e-01 9.75072160e-02 -3.04755550e-02 7.93503225e-02
-1.99107721e-01 1.51579231e-01 -1.16183090e+00 4.05219868e-02
-5.86842373e-02 -2.61100322e-01 4.07457441e-01 -8.69850934e-01
-1.03392935e+00 2.56467670e-01 5.00192344e-01 -9.72939610e-01
2.26025611e-01 1.42614692e-01 -2.61245221e-01 -3.71706337e-02
3.91740829e-01 2.56334841e-01 5.05093895e-02 -4.09600258e-01
1.88949108e-01 6.16143107e-01 7.10276604e-01 -1.86531618e-01
9.01792228e-01 7.01764941e-01 7.50143081e-02 -1.12114429e+00
-2.31857523e-01 -1.95572168e-01 -1.18257761e+00 -6.59249127e-01
6.40733242e-01 -6.86388612e-01 -7.18531191e-01 8.90409112e-01
-1.42330503e+00 -5.52126067e-03 -1.10955000e-01 4.46640402e-01
-5.10205507e-01 6.93886697e-01 -5.34920752e-01 -8.59326363e-01
-2.55006820e-01 -1.05445147e+00 1.09849715e+00 1.33770034e-01
-1.15302064e-01 -5.81771851e-01 -2.56108612e-01 7.88892657e-02
4.26240638e-02 2.86759287e-01 9.80158210e-01 2.40899712e-01
-1.18621373e+00 -5.24906039e-01 -5.69716245e-02 5.66320904e-02
2.73568124e-01 6.63716376e-01 -7.16006219e-01 -2.74562925e-01
2.74543107e-01 4.67832208e-01 3.08468997e-01 4.81668532e-01
1.12137222e+00 -3.49871591e-02 -6.54594123e-01 6.61438882e-01
1.31408966e+00 6.40202284e-01 4.65627879e-01 2.76735842e-01
7.02961743e-01 2.47781932e-01 8.02392542e-01 7.49622583e-01
2.80225556e-03 1.00801539e+00 5.69160759e-01 1.25501260e-01
-2.81893790e-01 6.52107000e-02 4.81187761e-01 8.78566384e-01
-7.06324428e-02 -1.21264443e-01 -8.62735033e-01 2.05259874e-01
-1.68729091e+00 -9.11133409e-01 -2.13215515e-01 2.50313663e+00
4.38647777e-01 2.26738125e-01 -1.53458759e-01 7.08901659e-02
1.05101061e+00 3.25847529e-02 -8.90974164e-01 -9.88741815e-02
3.88561711e-02 -2.04290286e-01 5.12026489e-01 1.21646315e-01
-9.95825231e-01 7.18783021e-01 5.69608307e+00 5.27653396e-01
-1.33782113e+00 -2.64454544e-01 2.95552850e-01 -4.43975717e-01
3.53731036e-01 -2.23502554e-02 -6.36335015e-01 3.49726886e-01
6.84391141e-01 -5.70418127e-02 2.13448018e-01 6.61965013e-01
4.02725458e-01 -4.46260333e-01 -1.06607580e+00 1.25282371e+00
1.24157228e-01 -1.42648649e+00 6.15845472e-02 2.18060240e-02
4.40431774e-01 -1.99049845e-01 1.73992515e-02 -3.07325155e-01
-3.39020461e-01 -3.70135009e-01 1.10020149e+00 6.61021531e-01
9.36592996e-01 -6.73400342e-01 2.17464298e-01 4.21970814e-01
-1.48603678e+00 -2.42839959e-02 -4.06909883e-01 -1.20645300e-01
2.94858903e-01 4.04149354e-01 -7.82902002e-01 3.34521949e-01
7.48871863e-01 4.15278882e-01 -3.71946067e-01 8.77672851e-01
2.18201250e-01 5.73436208e-02 -5.29287457e-01 1.69030368e-01
-2.35583559e-01 -2.12920994e-01 7.86317289e-01 7.97291160e-01
6.53192818e-01 2.23965824e-01 6.09228350e-02 7.17254043e-01
5.59645332e-02 -4.30774838e-01 -4.53271061e-01 3.09252739e-02
6.66749239e-01 1.08411098e+00 -1.04231727e+00 -5.79750001e-01
-3.79914463e-01 1.09019196e+00 -4.71569896e-02 9.29411128e-02
-9.63877082e-01 -3.07931095e-01 7.53941834e-01 3.14007372e-01
7.69789577e-01 -7.29948580e-01 -1.54866241e-02 -1.05202794e+00
1.98451504e-01 -7.77761459e-01 2.56703403e-02 -1.02399182e+00
-6.72987819e-01 2.64169753e-01 5.48822917e-02 -1.66801751e+00
-2.95661986e-01 -7.81231344e-01 -4.83772963e-01 5.71709156e-01
-9.02745008e-01 -7.53154635e-01 -4.14755374e-01 3.76555234e-01
5.47254503e-01 1.69061944e-01 4.71290827e-01 2.99049079e-01
-6.81253910e-01 1.34631738e-01 3.02456260e-01 5.49850650e-02
3.95321846e-01 -5.84586143e-01 7.09485114e-01 1.02256608e+00
-1.69600278e-01 2.02207863e-01 5.25696278e-01 -9.57473338e-01
-2.13805628e+00 -8.85841370e-01 3.70685369e-01 -3.42862874e-01
5.19742489e-01 -4.36147898e-01 -9.51419711e-01 4.94445503e-01
-4.84295249e-01 1.58318743e-01 4.65094671e-02 -7.44544148e-01
-6.98999912e-02 -1.70463711e-01 -9.94491279e-01 6.93987310e-01
1.08493018e+00 -4.26782787e-01 -3.54825020e-01 1.01169825e-01
7.71109939e-01 -8.37758064e-01 -6.62381470e-01 1.66574448e-01
6.58443570e-01 -7.49732971e-01 1.32447720e+00 -1.63491324e-01
1.12834983e-01 -6.76708937e-01 -1.88731365e-02 -7.73227155e-01
-5.97488359e-02 -6.91955447e-01 -5.72818100e-01 8.02702129e-01
-2.76355714e-01 -2.88754761e-01 8.78778338e-01 6.01867795e-01
-6.99718371e-02 -5.32010317e-01 -1.25318456e+00 -6.88716829e-01
-5.72943151e-01 -7.05822527e-01 6.20962918e-01 7.23806083e-01
-2.99589932e-01 1.50433853e-02 -1.53177232e-01 6.84317052e-01
5.31318724e-01 3.64071488e-01 9.63059247e-01 -1.11159790e+00
-1.39275223e-01 -2.19796538e-01 -8.81382346e-01 -1.23775589e+00
-3.19656819e-01 -4.28614527e-01 -1.41794309e-01 -1.11865950e+00
-1.14924744e-01 -3.12549174e-01 1.61405951e-01 -8.74994919e-02
1.87202305e-01 1.74834713e-01 4.15431350e-01 2.97659725e-01
-2.64630705e-01 4.39782143e-01 1.21864402e+00 -1.18376449e-01
-4.81444120e-01 5.35101593e-02 1.51547611e-01 9.91741598e-01
5.52314162e-01 -2.07666069e-01 -1.67801321e-01 -6.53739214e-01
8.24325159e-02 3.22736531e-01 6.04325354e-01 -1.08297408e+00
2.64043689e-01 -2.10477561e-01 9.10162807e-01 -1.33417213e+00
8.12344074e-01 -1.02646255e+00 4.87565190e-01 6.73436344e-01
2.79354513e-01 3.06523412e-01 4.27244902e-01 4.89798307e-01
2.66476482e-01 -2.39971831e-01 7.38935649e-01 1.04128700e-02
-8.20905685e-01 4.51800168e-01 -6.60250366e-01 -5.89329660e-01
1.18502605e+00 -8.50811243e-01 -9.46182609e-02 -1.64683595e-01
-5.96488059e-01 -1.04189120e-01 6.64797187e-01 3.58083993e-01
9.49196041e-01 -1.28973186e+00 -1.79745108e-01 5.23710132e-01
-2.42276222e-01 3.33664894e-01 4.83812362e-01 6.58679008e-01
-9.84081924e-01 7.88578689e-02 -1.18237406e-01 -8.63704324e-01
-1.60545182e+00 7.68962204e-01 5.72720587e-01 3.57705712e-01
-9.16707218e-01 4.83955890e-01 7.44096115e-02 3.99286970e-02
1.90464139e-01 -6.30631089e-01 2.40805641e-01 1.48969954e-02
7.80077934e-01 7.33677983e-01 1.16615742e-01 -7.56332994e-01
-3.16748172e-01 8.55738938e-01 6.56783208e-02 -4.43398282e-02
1.28570223e+00 -3.09459537e-01 -1.21290110e-01 2.91742206e-01
1.13891780e+00 -2.27157190e-01 -1.74496078e+00 -3.90925229e-01
-1.91106245e-01 -7.79983640e-01 2.53928155e-01 -1.37208670e-01
-1.08726442e+00 7.93317795e-01 9.37705815e-01 8.23724866e-02
1.08351517e+00 -1.57207429e-01 8.19573760e-01 4.20704722e-01
2.09023044e-01 -1.30398643e+00 2.33349115e-01 4.07842040e-01
6.44568443e-01 -5.99292338e-01 3.85238647e-01 -6.02650404e-01
7.45245069e-02 1.56977510e+00 7.64204621e-01 -8.20374414e-02
4.72216904e-01 5.70266962e-01 -4.17721495e-02 -1.83613241e-01
-6.73107207e-01 2.65882134e-01 2.02608123e-01 6.44951403e-01
-2.08084255e-01 -1.30738482e-01 4.36924338e-01 -9.12209675e-02
1.15531527e-01 -2.62396634e-01 5.23823142e-01 1.21968758e+00
-6.41795278e-01 -6.98541105e-01 -1.04547691e+00 1.83229417e-01
4.33338284e-02 2.69340426e-01 -1.69989035e-01 7.69521654e-01
1.00618854e-01 5.25142074e-01 8.00721586e-01 -2.45197192e-01
3.50844085e-01 -1.75425559e-01 8.33633542e-01 -3.02342623e-01
-9.50384885e-02 2.68406183e-01 -1.59584001e-01 -5.57364941e-01
-2.12247729e-01 -8.57182264e-01 -1.51818132e+00 -2.65346527e-01
-4.76017594e-01 -4.99069899e-01 8.07718158e-01 7.33864307e-01
3.98900568e-01 4.93366838e-01 6.79606140e-01 -1.40425229e+00
-3.90206367e-01 -5.62523961e-01 -3.86397064e-01 1.00818433e-01
4.58917111e-01 -7.49761879e-01 -2.53502786e-01 3.49614143e-01] | [6.947773456573486, -2.2430646419525146] |
c0834ad4-1e6a-46ba-8a9f-58f637584116 | boosting-knowledge-graph-generation-from | null | null | https://link.springer.com/chapter/10.1007/978-3-031-33455-9_29 | https://oa.upm.es/73463/1/_2023___ESWC__RML_Tabular_Views.pdf | Boosting Knowledge Graph Generation from Tabular Data with RML Views | A large amount of data is available in tabular form. RML is commonly used to declare how such data can be transformed into RDF. However, RML presents limitations that lead, in many cases, to the need for additional preprocessing using scripting. Although some proposed extensions (e.g., FnO or RML fields) address some of these limitations, they are verbose, unfamiliar to most data engineers, and implemented in systems that do not scale up when large volumes of data need to be processed. In this work, we expand RML views to tabular sources so as to address the limitations of this mapping language. In this way, transformation functions, complex joins, or mixed syntax can be defined directly in SQL queries. We present our extension of Morph-KGC to efficiently support RML views for tabular sources. We validate our implementation adapting R2RML test cases with views and compare it against state-of-the-art RML+FnO systems showing that our system is significantly more scalable. Moreover, we present specific examples of a real use case in the public procurement domain where basic RML mappings could not be used without additional preprocessing. | ['Oscar Corcho', 'María S. Pérez', 'María Navas-Loro', 'Ahmad Alobaid', 'Julián Arenas-Guerrero'] | 2023-05-22 | null | null | null | extended-semantic-web-conference-2023-5 | ['knowledge-graphs-data-curation', 'data-integration'] | ['knowledge-base', 'knowledge-base'] | [-1.81760326e-01 4.54931796e-01 -1.63204357e-01 -7.88532913e-01
-2.99226969e-01 -8.58286738e-01 7.06152022e-01 4.83000308e-01
-5.51893339e-02 7.14219868e-01 -2.55928282e-03 -5.98236978e-01
-4.26248997e-01 -1.42249274e+00 -5.80846548e-01 2.47361228e-01
5.11578023e-02 8.22568774e-01 6.32757425e-01 -5.96597612e-01
1.07729495e-01 7.58715391e-01 -1.87197018e+00 7.21663296e-01
8.84326518e-01 9.77783978e-01 -2.38112733e-03 -9.85756069e-02
-8.88432980e-01 8.55668604e-01 -6.17284954e-01 -5.34602463e-01
3.52925271e-01 -1.36470780e-01 -8.84560883e-01 -3.33565712e-01
3.00499558e-01 -1.58919111e-01 5.35919189e-01 6.29809260e-01
2.73736089e-01 -5.06004572e-01 7.58536607e-02 -1.48866308e+00
-2.03754246e-01 8.71333182e-01 -1.20283112e-01 -6.66698575e-01
6.61702037e-01 -2.56060481e-01 5.41195750e-01 -8.99317503e-01
1.36573148e+00 9.87055838e-01 4.35953915e-01 3.08432221e-01
-1.32890105e+00 -3.31571430e-01 -3.20090324e-01 5.60714453e-02
-1.11165452e+00 -5.50029099e-01 4.61462945e-01 -3.33713233e-01
1.32954967e+00 1.04360080e+00 5.84937871e-01 2.88298488e-01
9.13662910e-02 9.85802189e-02 1.28547788e+00 -4.14286792e-01
2.86030769e-01 8.01448524e-01 2.34147161e-02 2.54577994e-01
7.61860847e-01 -6.16899431e-01 -3.91826421e-01 -3.83665472e-01
3.84690613e-01 -2.89679974e-01 -1.16760470e-01 -1.07040703e+00
-1.03489125e+00 5.36262274e-01 1.12966290e-02 4.84700918e-01
-8.43339041e-02 -7.33293518e-02 6.00047886e-01 6.37056351e-01
8.67917761e-02 2.34889627e-01 -6.25340819e-01 -3.00955176e-01
-7.46416509e-01 1.50989011e-01 1.42188466e+00 1.54249120e+00
1.05942953e+00 -8.88347775e-02 4.76417333e-01 7.47021914e-01
4.12813336e-01 4.23944622e-01 -6.64223433e-02 -9.37856734e-01
6.83793426e-01 1.23427260e+00 1.95422724e-01 -7.56139338e-01
-3.63054574e-01 -8.10235962e-02 -5.76137640e-02 6.70920789e-01
2.23594695e-01 5.30822158e-01 -5.02245367e-01 9.90788996e-01
4.19047236e-01 -1.29896355e+00 1.91871345e-01 5.16523778e-01
7.71736801e-01 3.19547445e-01 -1.03045948e-01 -3.24363738e-01
1.21314085e+00 -2.74796098e-01 -8.75502706e-01 1.31565988e-01
7.52200305e-01 -8.73952806e-01 1.04972553e+00 3.58020484e-01
-1.43173838e+00 -1.43551111e-01 -8.68867874e-01 -7.80243650e-02
-1.00532436e+00 -3.14903647e-01 8.30910802e-01 1.06727707e+00
-9.65733588e-01 3.30460310e-01 -6.47867739e-01 -9.53027427e-01
-3.54284346e-02 4.58663493e-01 -5.95633447e-01 1.10962205e-01
-1.07010770e+00 9.91305947e-01 7.57876575e-01 9.82946604e-02
-1.42832965e-01 -7.73675263e-01 -6.85712516e-01 5.25871515e-02
6.74757063e-01 -4.87322003e-01 9.42778885e-01 -5.46236813e-01
-1.09263098e+00 8.71950865e-01 2.60803968e-01 -2.55924255e-01
8.44112635e-01 1.15951665e-01 -8.84465814e-01 -3.19924712e-01
2.57992446e-01 2.16796488e-01 -2.02927858e-01 -1.54289544e+00
-5.48759520e-01 -3.81678075e-01 5.71487606e-01 -2.35017747e-01
-5.00289761e-02 2.62386054e-01 -5.20077407e-01 -9.43730250e-02
8.80493596e-02 -4.34620827e-01 -4.17040428e-03 -1.17974482e-01
-2.41014026e-02 2.27771443e-03 6.98087990e-01 -4.11076725e-01
1.32310450e+00 -1.78837395e+00 -2.05811411e-01 6.20499611e-01
-2.82454044e-01 7.25673065e-02 2.05230936e-01 1.41986036e+00
-1.11335576e-01 6.35561109e-01 -1.47152320e-01 4.15759981e-01
4.24048781e-01 5.07707119e-01 -2.04955563e-01 3.39084268e-02
-1.69548213e-01 4.71344322e-01 -2.98189998e-01 -5.19284308e-01
1.58240736e-01 3.11238259e-01 -5.51632285e-01 7.46472925e-02
-7.41686046e-01 -8.16982687e-02 -1.24716692e-01 8.59155476e-01
1.04673040e+00 -2.34549074e-03 9.97513533e-01 -4.08676893e-01
-6.81257844e-01 5.59118569e-01 -1.67799890e+00 1.74986470e+00
-6.34422004e-01 -1.47627071e-01 1.03281356e-01 -6.23724520e-01
1.21252918e+00 3.22099477e-01 6.41763926e-01 -9.54118013e-01
-5.70873976e-01 6.57695174e-01 -4.10045415e-01 -6.25152707e-01
5.44485450e-01 1.67573839e-01 -2.26465493e-01 6.02827489e-01
-3.58588457e-01 -2.37516239e-01 6.96601033e-01 3.54636461e-02
7.70665288e-01 8.98636699e-01 6.44894719e-01 -2.11043388e-01
5.96665502e-01 4.54672337e-01 5.98004818e-01 3.84257168e-01
8.31551015e-01 2.53111005e-01 9.57585812e-01 -5.22805214e-01
-1.09056342e+00 -1.06703246e+00 -3.65842432e-01 7.55860150e-01
-1.66094795e-01 -1.10354459e+00 -4.88628060e-01 -6.64337039e-01
-4.03136872e-02 1.11597645e+00 -2.84950465e-01 6.13967955e-01
-7.78421104e-01 -3.69809389e-01 3.44420761e-01 3.37480992e-01
2.97142297e-01 -1.10252786e+00 -1.07157242e+00 4.12612736e-01
1.77900463e-01 -9.85165000e-01 3.16868216e-01 -1.06094070e-02
-9.33083653e-01 -1.23609185e+00 1.46937251e-01 -8.74005854e-02
5.66798329e-01 -1.72267377e-01 1.46436214e+00 1.85338393e-01
-2.64593750e-01 4.64900792e-01 -4.75423843e-01 -4.29417938e-01
-7.32602358e-01 8.44473094e-02 -7.93541431e-01 -3.66915852e-01
2.02498883e-01 -4.65327233e-01 -2.15376079e-01 6.85182691e-01
-1.63955772e+00 2.20780939e-01 1.19973712e-01 2.07736239e-01
5.01842618e-01 -2.08947673e-01 5.43534636e-01 -1.60449922e+00
3.81754160e-01 -4.58476961e-01 -1.12958074e+00 7.44852483e-01
-1.02810466e+00 1.71309024e-01 6.33993745e-01 3.95516157e-01
-1.12620950e+00 -7.35252276e-02 -1.72181234e-01 2.74040818e-01
-1.05481103e-01 1.13413739e+00 -7.57721484e-01 3.61957759e-01
3.22243005e-01 -1.70879126e-01 2.16122493e-01 -7.29031384e-01
3.91878128e-01 5.38734317e-01 -8.92835762e-03 -7.77407110e-01
8.37914348e-01 5.72309494e-01 8.98535848e-02 -2.21561924e-01
2.25224495e-01 -1.28126115e-01 -6.32580757e-01 -2.16105953e-01
3.78094077e-01 -4.51884836e-01 -6.60091102e-01 -1.10797144e-01
-1.08960319e+00 -1.67928278e-01 -7.08288729e-01 5.39539121e-02
-8.06937933e-01 3.26173045e-02 1.74754839e-02 -6.52760804e-01
-7.91487917e-02 -1.02103353e+00 6.79355800e-01 -3.34392160e-01
-2.32882231e-01 -7.12191045e-01 6.53633401e-02 3.72351497e-01
9.80361938e-01 5.19738793e-01 1.40271330e+00 -6.41522229e-01
-9.72767413e-01 -6.41800612e-02 -1.25517964e-01 -1.03481039e-01
2.48313636e-01 3.73810709e-01 -4.31851000e-01 -1.51144758e-01
-4.19869334e-01 5.08048721e-02 -2.49128878e-01 -7.05013394e-01
7.70648122e-01 -1.69670314e-01 -2.27376163e-01 4.14096594e-01
1.91816437e+00 3.82043868e-01 1.23036516e+00 9.44063783e-01
2.82065570e-01 1.17799604e+00 9.54626799e-01 5.49250245e-01
6.49005413e-01 1.10111022e+00 6.26138091e-01 -3.45015228e-02
-9.12602842e-02 -1.46999463e-01 2.31209904e-01 8.38597298e-01
-3.47544968e-01 -3.90798971e-03 -1.02572811e+00 1.50396824e-01
-1.89673936e+00 -1.09017742e+00 -8.49153519e-01 2.60431623e+00
7.71224082e-01 -1.36376202e-01 1.46125183e-01 9.50789973e-02
2.23112091e-01 -4.12177630e-02 -9.98581797e-02 -7.76404917e-01
-1.24368116e-01 3.01794857e-01 5.54946303e-01 3.93266916e-01
-2.96279341e-01 5.27914882e-01 5.49526501e+00 1.20139264e-01
-1.04565048e+00 2.19664037e-01 -2.19133034e-01 4.94518988e-02
-1.19444966e+00 6.74153268e-01 -8.46965849e-01 1.64938778e-01
1.09216869e+00 -4.85240698e-01 4.96181309e-01 8.43114555e-01
2.96043038e-01 -1.18987717e-01 -1.10257149e+00 3.87218416e-01
4.66277935e-02 -1.65478671e+00 2.55881585e-02 3.10809523e-01
1.82144895e-01 -1.40641063e-01 -7.04637825e-01 1.58212781e-01
6.68658242e-02 -8.80163610e-01 1.02484810e+00 7.36417592e-01
8.90692830e-01 -5.84688842e-01 5.80287516e-01 -4.06673886e-02
-1.05194223e+00 2.81393260e-01 -4.61602867e-01 5.64372182e-01
1.69021547e-01 6.56266868e-01 -8.89176488e-01 1.39444768e+00
9.55630898e-01 1.88253224e-01 -8.22899640e-01 9.20774758e-01
2.94057369e-01 -2.39857621e-02 -5.36009073e-01 1.22486241e-01
-3.62439334e-01 -3.62802148e-01 -2.81427018e-02 1.24191856e+00
6.10814631e-01 -5.99132955e-01 -2.23125666e-01 9.23237503e-01
2.78009832e-01 7.17985928e-01 -6.74950361e-01 1.59875274e-01
3.41491312e-01 1.13010216e+00 -5.99357784e-01 -2.33519927e-01
-8.21484089e-01 2.74516732e-01 6.09610751e-02 3.68573248e-01
-7.87580550e-01 -6.58539653e-01 2.37425163e-01 1.06336653e+00
3.37160617e-01 -1.56357065e-01 4.86865640e-02 -1.14324617e+00
5.82539439e-01 -1.23902965e+00 3.86515319e-01 -9.29023266e-01
-1.02217174e+00 7.00942993e-01 5.55182815e-01 -1.30868387e+00
-6.07774317e-01 -2.79347122e-01 6.55915737e-02 8.36122513e-01
-1.33808732e+00 -1.35592270e+00 -3.51002991e-01 4.70679015e-01
-1.56834155e-01 2.56300624e-02 1.00493228e+00 7.30090797e-01
3.59853730e-02 3.18807401e-02 -5.36050908e-02 -2.38037795e-01
7.98675179e-01 -1.25945258e+00 2.08876088e-01 5.16186833e-01
-2.36538470e-01 1.12421131e+00 5.32701135e-01 -7.24061310e-01
-1.82326162e+00 -7.79995859e-01 1.03418660e+00 -1.67429894e-01
6.48372948e-01 -8.45697880e-01 -1.02146757e+00 9.58621562e-01
5.77852905e-01 1.55532107e-01 7.14622319e-01 -1.34315506e-01
-5.23559511e-01 -9.00272012e-01 -1.39065564e+00 2.75705546e-01
9.04302776e-01 -4.46553767e-01 -4.06821728e-01 2.25196600e-01
2.32426047e-01 -2.24101201e-01 -1.74197388e+00 5.02428770e-01
6.84912562e-01 -1.57133317e+00 6.96689963e-01 -5.10255873e-01
-1.15867123e-01 -8.70420337e-01 -4.35463160e-01 -8.09246540e-01
4.34042335e-01 -5.35268426e-01 2.96798766e-01 1.83950400e+00
6.68636560e-01 -1.16019452e+00 4.98591423e-01 1.03914893e+00
1.98520534e-02 -3.64747256e-01 -7.51172423e-01 -1.00825357e+00
-2.58033097e-01 -4.66440082e-01 1.39549339e+00 1.13881767e+00
2.44883105e-01 -2.42383897e-01 1.25988061e-03 -1.34421945e-01
3.55468631e-01 5.46894550e-01 1.26271522e+00 -1.40928769e+00
-2.81207889e-01 4.85686958e-03 -2.65435785e-01 -2.00158060e-01
-3.78282160e-01 -1.06903648e+00 -6.22930825e-01 -2.25970578e+00
-2.06903070e-01 -1.02614141e+00 2.25399837e-01 6.66139662e-01
8.44866574e-01 -1.46129757e-01 4.82635617e-01 4.18652862e-01
-1.64094776e-01 -2.42981851e-01 6.09065354e-01 2.66968329e-02
-1.17936105e-01 -3.01301360e-01 -3.77788782e-01 3.41477245e-01
7.74723172e-01 -7.72952199e-01 -4.82527971e-01 -3.46909761e-01
1.17971873e+00 3.81134808e-01 1.02807194e-01 -9.86127019e-01
1.28226906e-01 -5.47377825e-01 2.94027645e-02 -8.88695776e-01
2.26326689e-01 -1.51593518e+00 1.56723773e+00 2.87926883e-01
-4.10310291e-02 3.20022315e-01 1.98037028e-02 -2.61480182e-01
-3.28789771e-01 -8.49175632e-01 1.88892066e-01 -3.54635775e-01
-5.02492964e-01 -3.17015573e-02 -1.81120023e-01 -1.64625674e-01
1.33011615e+00 -4.28520322e-01 -1.08213186e+00 -3.76208872e-02
-5.41921318e-01 -2.27566287e-02 1.16959262e+00 4.15942013e-01
3.03744018e-01 -1.23948264e+00 -2.90566742e-01 3.08437139e-01
4.90361631e-01 -2.25147471e-01 -3.44082296e-01 7.36952543e-01
-1.09377110e+00 5.89928925e-01 -5.16161561e-01 -2.33017057e-01
-1.15715766e+00 6.96209252e-01 2.02849314e-01 -1.84415236e-01
-3.69715810e-01 -5.38838148e-01 -4.71573383e-01 -9.18858886e-01
-2.96052039e-01 -3.64522606e-01 6.79684505e-02 3.86018455e-01
1.30581632e-01 3.93940985e-01 6.38962209e-01 -2.61327118e-01
-6.99863195e-01 6.11625910e-01 3.68774325e-01 -2.56445527e-01
1.70392692e+00 7.30390027e-02 -8.24073017e-01 6.07852697e-01
7.66657054e-01 9.04449821e-01 -3.75342786e-01 3.62910539e-01
4.09203500e-01 -5.80078959e-01 -7.60465682e-01 -1.03643608e+00
-9.44767773e-01 5.25571048e-01 2.74051696e-01 7.81713367e-01
1.06683266e+00 2.17608526e-01 1.77407175e-01 4.62114394e-01
1.11915612e+00 -9.92536604e-01 -8.90689552e-01 -1.17556810e-01
1.28435016e+00 -6.28142536e-01 3.67582262e-01 -8.57132912e-01
-4.20902938e-01 1.69762182e+00 3.89930159e-01 6.16511106e-01
4.80704397e-01 7.62114346e-01 3.89859319e-01 -5.05391121e-01
-8.24118137e-01 -1.34699687e-01 -4.02717859e-01 7.24033356e-01
7.67842889e-01 -1.71220198e-01 -1.09105849e+00 5.61016910e-02
-6.92829043e-02 6.51744902e-01 9.64151084e-01 1.46435475e+00
-1.28732398e-01 -2.07843685e+00 -5.05712867e-01 4.09315258e-01
-4.60055500e-01 8.21117759e-02 -9.77418348e-02 1.59764421e+00
1.81494996e-01 6.85480416e-01 1.64037794e-01 1.78050339e-01
1.01353550e+00 6.38993382e-02 4.32528079e-01 -6.25378072e-01
-1.19625330e+00 -8.98658335e-02 7.95473337e-01 -7.85930157e-01
-6.40833616e-01 -7.71737516e-01 -1.37999618e+00 -2.94630855e-01
5.64693585e-02 2.95102447e-01 1.07113993e+00 2.85289019e-01
5.32342792e-01 1.58053830e-01 9.98065397e-02 1.84744596e-01
-3.51182938e-01 -4.43019241e-01 -4.72149014e-01 5.36022902e-01
-6.30688787e-01 -3.78123462e-01 1.59941509e-01 3.18054378e-01] | [9.141754150390625, 7.745604991912842] |
a9faf60f-8194-40d0-bd38-9cfd1c668061 | lift-yourself-up-retrieval-augmented-text | 2305.02437 | null | https://arxiv.org/abs/2305.02437v2 | https://arxiv.org/pdf/2305.02437v2.pdf | Lift Yourself Up: Retrieval-augmented Text Generation with Self Memory | With direct access to human-written reference as memory, retrieval-augmented generation has achieved much progress in a wide range of text generation tasks. Since better memory would typically prompt better generation~(we define this as primal problem). The traditional approach for memory retrieval involves selecting memory that exhibits the highest similarity to the input. However, this method is constrained by the quality of the fixed corpus from which memory is retrieved. In this paper, by exploring the duality of the primal problem: better generation also prompts better memory, we propose a novel framework, selfmem, which addresses this limitation by iteratively employing a retrieval-augmented generator to create an unbounded memory pool and using a memory selector to choose one output as memory for the subsequent generation round. This enables the model to leverage its own output, referred to as self-memory, for improved generation. We evaluate the effectiveness of selfmem on three distinct text generation tasks: neural machine translation, abstractive text summarization, and dialogue generation, under two generation paradigms: fine-tuned small model and few-shot LLM. Our approach achieves state-of-the-art results in four directions in JRC-Acquis, XSum (50.3 ROUGE-1), and BigPatent (62.9 ROUGE-1), demonstrating the potential of self-memory in enhancing retrieval-augmented generation models. Furthermore, we conduct thorough analyses of each component in the selfmem framework to identify bottlenecks and provide insights for future research. | ['Rui Yan', 'Dongyan Zhao', 'Lemao Liu', 'Xiuying Chen', 'Di Luo', 'Xin Cheng'] | 2023-05-03 | null | null | null | null | ['dialogue-generation', 'abstractive-text-summarization', 'text-summarization', 'dialogue-generation'] | ['natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'speech'] | [ 5.42838037e-01 2.66361088e-01 -3.84447485e-01 4.12051566e-02
-1.20647752e+00 -5.10497272e-01 9.57330525e-01 1.24188885e-01
-3.13182592e-01 1.06636608e+00 6.40827179e-01 -3.06979388e-01
2.14503005e-01 -9.35948730e-01 -5.82824171e-01 -4.22308266e-01
3.98414552e-01 7.33502567e-01 -9.45494976e-03 -4.81357336e-01
7.22205341e-01 5.72414603e-03 -1.38207388e+00 4.60087538e-01
1.29478252e+00 4.34366673e-01 4.86471564e-01 9.34885502e-01
-3.02317411e-01 6.95648193e-01 -1.03076017e+00 -1.68137610e-01
-1.76222399e-01 -1.02907360e+00 -1.13700771e+00 -2.09445670e-01
2.72886038e-01 -3.17909330e-01 -2.47841552e-01 5.42569518e-01
9.23182666e-01 6.28141403e-01 6.69683874e-01 -7.33069837e-01
-8.96124005e-01 9.64135945e-01 -5.34726977e-01 4.48881060e-01
6.52559400e-01 2.65142083e-01 9.66970146e-01 -7.76692986e-01
8.29881787e-01 1.38157773e+00 1.78856373e-01 8.88206780e-01
-1.29070520e+00 -4.07874048e-01 -1.62631363e-01 -3.59505743e-01
-1.00109327e+00 -9.27118003e-01 3.70777190e-01 -8.61653034e-03
1.41246414e+00 4.71997350e-01 3.93434435e-01 1.07706749e+00
3.15374285e-01 1.08744550e+00 9.12416935e-01 -7.02750623e-01
1.76248580e-01 -6.72231689e-02 2.41164491e-01 4.86183077e-01
1.57930046e-01 3.80121283e-02 -9.25899923e-01 -2.59877324e-01
5.09700000e-01 -3.63774121e-01 -4.67981339e-01 4.18247133e-01
-1.27066946e+00 9.74739492e-01 1.23406634e-01 1.73385769e-01
-3.53780001e-01 1.32528082e-01 3.06443274e-01 4.17701632e-01
7.94815361e-01 1.19387579e+00 -1.18573248e-01 -4.50885028e-01
-1.23842943e+00 5.00516891e-01 9.79044557e-01 1.02114224e+00
5.82886636e-01 1.13158882e-01 -8.21829319e-01 1.00063992e+00
1.14321873e-01 6.42776668e-01 1.00282443e+00 -7.79024839e-01
7.69569874e-01 6.25964701e-01 3.88943292e-02 -7.14216411e-01
-1.80019185e-01 -3.55006605e-01 -7.10769296e-01 -4.69212532e-01
-3.55967842e-02 -3.00510764e-01 -1.03547621e+00 1.84942806e+00
-5.93890510e-02 -1.26638293e-01 3.53735000e-01 7.42414057e-01
9.11582768e-01 1.12196660e+00 -7.64391292e-03 -5.08560181e-01
9.44085777e-01 -1.16437960e+00 -5.18538654e-01 -3.51804167e-01
8.20496559e-01 -9.17046607e-01 1.08063674e+00 -1.61859050e-01
-1.48281610e+00 -3.68619591e-01 -8.79786968e-01 -8.92521143e-02
2.74852314e-03 2.64007598e-02 3.73294473e-01 2.58762538e-01
-1.33032680e+00 7.35123277e-01 -7.33541310e-01 -5.24569035e-01
8.36084504e-03 1.71769276e-01 5.54164164e-02 3.62370834e-02
-1.30703914e+00 9.88059402e-01 4.91282254e-01 -2.01176718e-01
-6.35938048e-01 -5.17244816e-01 -6.25266731e-01 5.46551868e-02
2.60609746e-01 -1.28920543e+00 1.54147053e+00 -7.53570139e-01
-1.58012009e+00 5.21876812e-01 -3.70764345e-01 -5.60947895e-01
2.98463881e-01 -4.37028795e-01 -1.59896567e-01 5.10643274e-02
1.67088330e-01 8.30745816e-01 8.18956494e-01 -1.02484953e+00
-4.18400526e-01 -1.11879624e-01 -1.71339363e-01 5.59456050e-01
-2.96891838e-01 -8.74657407e-02 -3.43619287e-01 -8.21748734e-01
-9.32547674e-02 -1.00064039e+00 -1.20956898e-01 -8.40680718e-01
-6.07360303e-01 -2.39498511e-01 4.41575617e-01 -5.44588923e-01
1.84467602e+00 -1.68755984e+00 4.54633355e-01 -2.54313201e-02
7.69878775e-02 4.09811676e-01 -5.96951365e-01 8.48126173e-01
4.23661172e-01 3.70424777e-01 -1.40002295e-01 -3.77493560e-01
-5.36591373e-02 -6.66232854e-02 -6.74364805e-01 -2.71036893e-01
2.72045940e-01 1.21303785e+00 -1.05924082e+00 -5.18948078e-01
-3.57412547e-01 1.32385448e-01 -5.04425526e-01 4.32049006e-01
-4.67329323e-01 1.18709363e-01 -4.84516084e-01 3.77007782e-01
6.76039755e-02 -3.67500514e-01 -1.16180279e-03 2.99184382e-01
-8.58169794e-02 5.41772664e-01 -7.82287776e-01 1.71300626e+00
-5.19370615e-01 5.98121166e-01 -3.35209191e-01 -2.00143903e-01
1.00367332e+00 3.07204336e-01 -5.93103282e-02 -8.84455025e-01
-1.43512264e-01 3.12751532e-01 -1.64192878e-02 -1.70921922e-01
1.45737934e+00 3.21552865e-02 -8.36877674e-02 1.18038845e+00
1.25980843e-03 -2.26653650e-01 5.30148029e-01 6.79008663e-01
1.15744555e+00 -7.53992004e-03 2.40214050e-01 -9.68833193e-02
1.07259698e-01 2.65155196e-01 2.10178763e-01 1.23819399e+00
2.02249542e-01 6.68396175e-01 1.10069059e-01 8.95177275e-02
-9.10246015e-01 -9.61503208e-01 3.66271764e-01 1.45910299e+00
2.43453886e-02 -4.84916240e-01 -9.23282802e-01 -3.99590015e-01
-1.41109467e-01 1.11073399e+00 -3.31633002e-01 -6.01180077e-01
-8.90056729e-01 -8.89743626e-01 8.25502396e-01 5.34174383e-01
4.01778877e-01 -1.55205774e+00 -7.57691920e-01 3.34628582e-01
-5.71479976e-01 -3.99706006e-01 -8.83084118e-01 -7.49343485e-02
-1.16387439e+00 -5.44512391e-01 -7.65362978e-01 -4.89985526e-01
5.57945073e-01 4.38663393e-01 1.59693384e+00 1.88847482e-01
1.84161123e-02 3.67749125e-01 -4.97541070e-01 -7.48429298e-02
-9.09376025e-01 8.86857390e-01 -1.13191128e-01 -5.10544181e-01
-4.36240584e-02 -2.49398381e-01 -6.30964875e-01 -6.70174137e-02
-9.84451890e-01 4.03195649e-01 9.75944698e-01 1.11319506e+00
3.24653715e-01 -5.00609636e-01 1.09591949e+00 -9.49992716e-01
1.46240532e+00 -6.45159781e-01 -1.11495897e-01 4.47345167e-01
-9.58204091e-01 4.95418340e-01 5.27826488e-01 -4.47438955e-01
-1.20527542e+00 -6.18476629e-01 1.25902891e-01 4.31657098e-02
2.53476351e-01 8.01678181e-01 2.32603326e-01 4.55405593e-01
8.06387842e-01 6.40518367e-01 1.27478302e-01 -3.52979243e-01
6.61986887e-01 6.35486424e-01 3.87649775e-01 -7.14195132e-01
4.25330251e-01 -3.61617804e-01 -4.57237452e-01 -6.15529120e-01
-6.95977390e-01 -3.33594024e-01 -1.60058677e-01 5.01798466e-02
4.55709398e-01 -7.55656183e-01 -1.08422741e-01 2.69558400e-01
-1.31047368e+00 -4.96259481e-01 -3.58118147e-01 5.30196354e-02
-5.74933231e-01 1.54021397e-01 -8.96084309e-01 -8.40752423e-01
-1.25839031e+00 -9.20563340e-01 1.21633685e+00 4.15742636e-01
-7.48672366e-01 -8.80447686e-01 4.15313780e-01 3.20288956e-01
7.01120377e-01 -1.47668734e-01 1.10018957e+00 -7.73854554e-01
-6.14815593e-01 -4.26624622e-03 -6.46446645e-02 -5.45101799e-02
-1.17323458e-01 -1.82849228e-01 -6.78143322e-01 -4.35494721e-01
-3.36230636e-01 -4.80766535e-01 1.20978928e+00 8.85133967e-02
5.20874798e-01 -4.01487887e-01 -3.15520138e-01 2.17574269e-01
1.06201851e+00 2.08901420e-01 6.46805942e-01 2.64730006e-01
4.77617770e-01 3.48787606e-01 6.54200733e-01 4.48492259e-01
3.54039699e-01 5.92757881e-01 -1.81892931e-01 1.20694064e-01
-2.46219993e-01 -4.44001406e-01 5.89736521e-01 1.24098814e+00
-2.49508452e-02 -8.13236296e-01 -8.09310317e-01 4.83613729e-01
-2.00801611e+00 -1.07031429e+00 1.70070454e-01 2.21573210e+00
1.32391095e+00 -3.22996639e-02 5.81435952e-03 -3.29996407e-01
6.18400276e-01 3.35068792e-01 -5.31691015e-01 -6.66009307e-01
-2.00591058e-01 3.55534136e-01 9.21693295e-02 5.41374803e-01
-5.77616751e-01 1.17148209e+00 6.73313522e+00 8.83263409e-01
-1.05212522e+00 -2.49489048e-03 7.53386736e-01 -4.54713613e-01
-4.76965547e-01 -2.57302169e-02 -1.02982092e+00 4.95127618e-01
1.27588618e+00 -7.34045088e-01 6.10287309e-01 4.23798203e-01
2.29667544e-01 -3.48060399e-01 -1.12316275e+00 5.38555562e-01
3.91299278e-01 -1.51578617e+00 6.13655567e-01 -2.75806244e-02
9.28522408e-01 -7.05543458e-02 2.92466190e-02 5.62931657e-01
3.18946272e-01 -9.51786041e-01 4.87074286e-01 6.76659107e-01
7.75142372e-01 -7.49003053e-01 4.32925075e-01 5.41893125e-01
-8.46213996e-01 1.55896306e-01 -3.59398395e-01 1.91011298e-02
3.13882440e-01 3.40487242e-01 -1.09010696e+00 4.95327502e-01
-1.22521736e-01 2.34449342e-01 -6.17743671e-01 7.21091509e-01
7.76805952e-02 6.62937224e-01 5.27528971e-02 -2.31313303e-01
1.38437122e-01 1.22893810e-01 7.10623801e-01 1.35183489e+00
5.37377417e-01 1.20364137e-01 2.57691085e-01 1.04820275e+00
-3.65699708e-01 1.32689252e-01 -5.18025935e-01 -3.85672659e-01
8.23949814e-01 1.05190647e+00 -4.35457736e-01 -5.12585104e-01
1.41242832e-01 1.12960076e+00 3.61132324e-01 3.69977832e-01
-4.72899884e-01 -5.00010312e-01 1.72981337e-01 1.00691855e-01
-1.17762141e-01 -9.10957456e-02 -2.63126135e-01 -1.06420910e+00
-2.27747768e-01 -1.06276298e+00 2.36117467e-01 -7.25071132e-01
-9.53892529e-01 7.81925321e-01 2.32171267e-02 -8.81628752e-01
-9.99624372e-01 1.00911856e-01 -8.31042945e-01 1.17167854e+00
-1.25114202e+00 -9.39934254e-01 -1.66801419e-02 1.60056219e-01
9.71684933e-01 -3.05950552e-01 9.50711489e-01 -6.92819152e-03
-6.54607713e-01 8.01426947e-01 1.39241740e-01 -2.57250637e-01
9.59002316e-01 -1.27402675e+00 6.76681340e-01 8.16241682e-01
1.43343606e-03 1.03330493e+00 5.95799685e-01 -9.00604665e-01
-1.73579669e+00 -1.19858229e+00 1.24224544e+00 -4.74093795e-01
3.54443043e-01 5.59156686e-02 -7.99967408e-01 3.95721823e-01
5.03958285e-01 -7.49330163e-01 6.90737247e-01 8.09279233e-02
2.17462033e-02 3.16218317e-01 -7.40982234e-01 8.11215222e-01
1.00083041e+00 -4.06527162e-01 -6.95454776e-01 1.59088209e-01
1.04693377e+00 -4.79105294e-01 -7.49403834e-01 6.02111742e-02
3.68733406e-01 -5.79298913e-01 6.48635149e-01 -4.78391349e-01
8.29099774e-01 3.29890251e-02 4.41171117e-02 -1.77274024e+00
-1.67009175e-01 -9.96140957e-01 -5.98396480e-01 1.39376163e+00
7.49206364e-01 -6.07378185e-01 5.33387721e-01 6.07613742e-01
-4.16612923e-01 -8.65957260e-01 -5.51223934e-01 -6.12753928e-01
2.48185918e-01 2.53603935e-01 6.92477763e-01 6.96257293e-01
2.55686760e-01 1.01128113e+00 -4.05042946e-01 -5.43063521e-01
1.86082110e-01 4.42779630e-01 9.60007489e-01 -8.58324766e-01
-4.75301802e-01 -6.22847497e-01 3.84785861e-01 -1.42791176e+00
1.74687698e-01 -1.01452208e+00 3.72592032e-01 -1.62563705e+00
5.86684585e-01 -3.59763443e-01 -3.77094261e-02 2.61026382e-01
-7.63308048e-01 1.14273587e-02 4.20925111e-01 5.57943404e-01
-8.30642939e-01 6.40577614e-01 1.33595061e+00 -8.02125037e-02
-6.79525793e-01 -7.78529719e-02 -1.13758194e+00 1.25663755e-02
8.28716159e-01 -2.26863816e-01 -5.94064713e-01 -6.25113308e-01
3.40820439e-02 5.88188350e-01 -1.25846326e-01 -7.62271762e-01
3.56929004e-01 -1.20068036e-01 1.72256976e-01 -5.73072731e-01
1.63910180e-01 2.07200631e-01 -1.04766395e-02 2.51558751e-01
-9.41299438e-01 6.13802671e-01 1.84000567e-01 4.37237173e-01
-1.54303029e-01 -3.34742188e-01 3.74703616e-01 -3.02713633e-01
-5.33232212e-01 3.71484049e-02 -3.92551720e-01 5.37180603e-01
4.26823467e-01 -1.39901996e-01 -7.98827410e-01 -5.43495953e-01
-3.11641335e-01 2.85719484e-01 5.23567855e-01 5.05562603e-01
7.25169957e-01 -1.16298354e+00 -9.78671134e-01 5.80432117e-02
6.54086769e-02 -1.33517995e-01 -4.60702814e-02 5.99664688e-01
-1.82338089e-01 5.29880762e-01 9.24983174e-02 -3.58384669e-01
-1.08083665e+00 1.66926131e-01 2.96608564e-02 -7.43956327e-01
-3.34630042e-01 6.67015553e-01 -1.52442783e-01 -2.78823048e-01
-2.21292391e-01 2.10262295e-02 -2.73706876e-02 9.36871171e-02
6.78940415e-01 6.61095798e-01 -1.74706262e-02 -5.21184146e-01
1.31218314e-01 -1.18144020e-01 -4.69078779e-01 -5.14171600e-01
8.93068016e-01 -1.51011318e-01 -3.95067126e-01 4.22192186e-01
8.94599199e-01 -1.19719483e-01 -7.37434626e-01 -2.64245749e-01
1.02764785e-01 -3.04383308e-01 -4.92785238e-02 -1.10363889e+00
-7.02799618e-01 5.70056558e-01 1.45048901e-01 1.29730599e-02
1.04066873e+00 -1.90826520e-01 1.20265222e+00 5.32105446e-01
2.62522906e-01 -1.18867314e+00 3.82117063e-01 9.27604556e-01
9.09404755e-01 -1.01654911e+00 -5.97124407e-03 1.28717348e-01
-7.29423523e-01 9.22624171e-01 8.54339182e-01 2.26979509e-01
-1.08290173e-01 -8.28442723e-03 -4.05824855e-02 -2.20797472e-02
-1.30301309e+00 5.09053543e-02 3.65998715e-01 1.25492677e-01
9.29179549e-01 1.73903078e-01 -5.45385063e-01 4.06133443e-01
-3.76322657e-01 -7.58198276e-02 5.19343853e-01 1.16927946e+00
-7.77189851e-01 -1.19698298e+00 -2.76535779e-01 8.66130412e-01
-2.67784983e-01 -5.81828654e-01 -6.44286335e-01 3.47124249e-01
-6.04666293e-01 1.24522126e+00 8.86679664e-02 -3.59902173e-01
1.83708772e-01 3.23813081e-01 3.73105407e-01 -9.73308742e-01
-9.79253888e-01 1.33005023e-01 4.00585920e-01 -3.21694314e-01
3.03121377e-03 -4.61583138e-01 -1.19947147e+00 -6.11850262e-01
-7.23733127e-01 3.82759690e-01 2.83500582e-01 7.17717409e-01
9.04980719e-01 5.54490209e-01 5.00214994e-01 -7.82199681e-01
-9.52188611e-01 -1.22741163e+00 -1.39990151e-01 2.53488600e-01
2.22958922e-02 -2.72741973e-01 -5.67700490e-02 -9.29590762e-02] | [12.067597389221191, 9.02849006652832] |
b5699ebb-2afd-4555-8bed-053041580c23 | low-complexity-three-dimensional-discrete | 2206.00124 | null | https://arxiv.org/abs/2206.00124v1 | https://arxiv.org/pdf/2206.00124v1.pdf | Low-complexity Three-dimensional Discrete Hartley Transform Approximations for Medical Image Compression | The discrete Hartley transform (DHT) is a useful tool for medical image coding. The three-dimensional DHT (3D DHT) can be employed to compress medical image data, such as magnetic resonance and X-ray angiography. However, the computation of the 3D DHT involves several multiplications by irrational quantities, which require floating-point arithmetic and inherent truncation errors. In recent years, a significant progress in wireless and implantable biomedical devices has been achieved. Such devices present critical power and hardware limitations. The multiplication operation demands higher hardware, power, and time consumption than other arithmetic operations, such as addition and bit-shifts. In this work, we present a set of multiplierless DHT approximations, which can be implemented with fixed-point arithmetic. We derive 3D DHT approximations by employing tensor formalism. Such proposed methods present prominent computational savings compared to the usual 3D DHT approach, being appropriate for devices with limited resources. The proposed transforms are applied in a lossy 3D DHT-based medical image compression algorithm, presenting practically the same level of visual quality ($>98\%$ in terms of SSIM) at a considerable reduction in computational effort ($100 \%$ multiplicative complexity reduction). Furthermore, we implemented the proposed 3D transforms in an ARM Cortex-M0+ processor employing the low-cost Raspberry Pi Pico board. The execution time was reduced by $\sim$70% compared to the usual 3D DHT and $\sim$90% compared to 3D DCT. | ['R. J. Cintra', 'F. M. Bayer', 'V. A. Coutinho'] | 2022-05-31 | null | null | null | null | ['pico'] | ['natural-language-processing'] | [ 6.01841807e-01 5.46995476e-02 1.03359818e-01 -3.84607241e-02
-4.27837878e-01 1.08355545e-01 3.04491132e-01 6.40511513e-01
-7.70481229e-01 5.49511373e-01 -7.09432811e-02 -5.04742742e-01
-2.45567992e-01 -8.46977234e-01 -4.34328884e-01 -5.63879132e-01
-4.24472868e-01 2.70361930e-01 2.46222332e-01 -1.13265701e-01
2.86434889e-01 5.38262904e-01 -1.31560791e+00 2.59413093e-01
5.88091135e-01 1.60723627e+00 1.07774571e-01 6.85267091e-01
1.43377364e-01 7.76848197e-01 -3.83106142e-01 -3.92394125e-01
2.76419997e-01 -3.52389544e-01 -5.42708755e-01 -1.28961563e-01
-1.01295590e-01 -6.65759325e-01 -3.73067498e-01 1.01337051e+00
7.75140941e-01 -2.98052430e-01 7.36906648e-01 -8.66507053e-01
-1.56976342e-01 2.13963687e-01 -8.75480831e-01 3.15655112e-01
3.57967496e-01 -2.20650155e-02 2.05756903e-01 -6.35316908e-01
4.48836476e-01 1.05622864e+00 7.77565300e-01 7.27097467e-02
-1.01152384e+00 -5.32620013e-01 -8.41184020e-01 2.99035698e-01
-1.58343697e+00 -2.38265470e-01 5.75077593e-01 -1.50913939e-01
1.27301443e+00 5.69936037e-01 8.95268738e-01 3.23776543e-01
9.61711705e-01 2.73583591e-01 1.33961642e+00 -6.06159627e-01
4.35597926e-01 -1.19510807e-01 6.21108897e-02 8.28533411e-01
5.57672381e-01 -1.17323540e-01 -3.70105863e-01 -4.72966790e-01
9.53116000e-01 1.88483104e-01 -2.42656320e-01 9.80957672e-02
-1.26633465e+00 6.36683702e-01 3.02931577e-01 5.34741700e-01
-6.50317132e-01 4.09311265e-01 7.23091304e-01 1.87358990e-01
2.23975390e-01 -2.27542937e-01 -9.18798521e-02 -3.31621021e-01
-9.73836124e-01 8.52251127e-02 6.31608784e-01 9.56530929e-01
1.96725503e-01 1.89160392e-01 -2.93854922e-02 5.71231663e-01
2.40306556e-01 6.95293605e-01 8.54465425e-01 -7.27212965e-01
3.24098557e-01 3.32526296e-01 -2.96460122e-01 -1.26081479e+00
-4.57009643e-01 -3.29841167e-01 -1.44910324e+00 4.58745688e-01
-1.67840153e-01 3.87843728e-01 -7.51035452e-01 1.10041761e+00
2.55286038e-01 -3.29780787e-01 2.12660924e-01 5.76979876e-01
6.60775065e-01 7.96843886e-01 -1.55743295e-02 -6.03496790e-01
1.62411797e+00 -3.24463606e-01 -1.00680149e+00 2.64707088e-01
6.63965404e-01 -1.17057431e+00 6.59162402e-01 5.89232862e-01
-1.59361708e+00 -4.39544201e-01 -1.35492933e+00 -2.11727664e-01
3.26965339e-02 -8.62947553e-02 5.15661240e-01 9.63334620e-01
-9.98650908e-01 6.96280062e-01 -1.04863369e+00 -1.04366027e-01
3.90221119e-01 6.34272158e-01 -2.26466835e-01 -1.70644075e-02
-8.76393378e-01 8.45266044e-01 1.63859978e-01 6.67849462e-03
-3.45502794e-01 -6.66863441e-01 -5.30984342e-01 1.36555448e-01
-2.74329960e-01 -7.06569076e-01 9.11444128e-01 -1.90602347e-01
-1.44323993e+00 8.67959917e-01 -1.10028719e-03 -8.29834461e-01
4.03627217e-01 1.61921203e-01 -3.72127622e-01 7.30194688e-01
-2.61490792e-01 4.34060931e-01 7.96206474e-01 -2.96742857e-01
-2.24904791e-01 -5.40650070e-01 -1.37450099e-01 2.32445076e-01
-3.80093724e-01 -1.47644117e-01 8.96600038e-02 -8.44376624e-01
5.11908054e-01 -8.29841375e-01 -2.31964231e-01 6.79919600e-01
3.16712074e-02 3.70288283e-01 7.34421134e-01 -6.83617592e-01
1.18150818e+00 -2.16314673e+00 -1.36476114e-01 1.70939982e-01
3.42447847e-01 4.31733906e-01 5.73353469e-01 3.29521537e-01
1.37987316e-01 -1.26593739e-01 -3.41259807e-01 5.07921278e-02
-3.21589410e-01 2.00704277e-01 1.39125392e-01 7.51363635e-01
-3.76061380e-01 5.48194647e-01 -2.30292916e-01 -6.76467240e-01
3.76937062e-01 1.07024324e+00 -7.49139667e-01 -3.22596669e-01
4.77938890e-01 1.13669917e-01 -4.28206921e-01 5.70917189e-01
9.80644464e-01 -1.39994308e-01 6.33904636e-02 -5.58538675e-01
-5.70829250e-02 4.47059460e-02 -1.05518520e+00 1.70163512e+00
-5.04369676e-01 3.94677013e-01 3.05586047e-02 -1.25595069e+00
1.06737697e+00 5.64708829e-01 7.50422776e-01 -1.02933741e+00
5.13679326e-01 6.89523816e-01 -1.30407944e-01 -5.44860959e-01
2.57629812e-01 -5.70172906e-01 6.29995316e-02 3.37454170e-01
-8.08567405e-02 -4.13788885e-01 -6.90021813e-02 -2.91039571e-02
1.17904449e+00 -3.06392431e-01 8.35509002e-01 -6.65319502e-01
6.66743457e-01 -1.31525099e-01 2.09781617e-01 1.25781119e-01
-1.00993790e-01 2.10935205e-01 9.60796177e-02 -6.06679201e-01
-1.06656814e+00 -8.05685282e-01 -4.67789233e-01 -1.89474728e-02
2.30603784e-01 -2.36551896e-01 -6.46992803e-01 2.51826257e-01
-2.68455356e-01 3.32561553e-01 4.83167619e-02 -1.42815933e-01
-8.20235789e-01 -8.74761879e-01 6.38818800e-01 1.26657426e-01
9.62782919e-01 -6.87761724e-01 -1.61554110e+00 4.62897986e-01
9.22458097e-02 -1.13955605e+00 -1.70239881e-01 1.58969358e-01
-1.75846303e+00 -6.84323370e-01 -8.83039832e-01 -8.01285565e-01
6.04373634e-01 1.20418198e-01 7.30180979e-01 1.85406312e-01
-7.30174243e-01 2.48852074e-01 -3.93354803e-01 -3.42127979e-01
-3.66050065e-01 -4.61722195e-01 4.76353504e-02 -4.17005539e-01
2.74573922e-01 -9.49893951e-01 -1.13027000e+00 -7.55522177e-02
-1.05855513e+00 1.42220497e-01 7.74975002e-01 7.64650881e-01
8.47741365e-01 2.82933444e-01 2.21561536e-01 -6.53502584e-01
4.95915800e-01 -1.80187318e-02 -5.31776428e-01 -2.30490655e-01
-7.86275268e-01 3.51011492e-02 8.46501529e-01 -1.70553982e-01
-6.88540399e-01 -1.14779681e-01 -5.74669123e-01 -1.52913854e-01
3.95161569e-01 2.73090899e-01 4.69267994e-01 -5.76280773e-01
6.43578887e-01 4.73968416e-01 3.98714334e-01 -4.11549062e-01
-1.91196874e-01 7.86827326e-01 5.75357854e-01 -1.60232767e-01
4.17595178e-01 6.85715258e-01 5.70963144e-01 -1.06128633e+00
2.32908726e-01 -9.18385535e-02 -1.18189655e-01 -4.25105840e-02
6.33345008e-01 -8.61687541e-01 -1.03620553e+00 1.98413208e-01
-1.04766846e+00 3.51945221e-01 -4.38095331e-01 9.32829857e-01
-7.97535360e-01 6.24830306e-01 -9.22470689e-01 -6.26076460e-01
-1.12320149e+00 -1.53456199e+00 8.66996169e-01 -5.22034317e-02
-2.62256473e-01 -6.53398395e-01 -4.66681033e-01 7.36370608e-02
7.08428741e-01 6.92569196e-01 1.32065618e+00 1.44113019e-01
-4.60236162e-01 -4.14398998e-01 -5.53576909e-02 1.49446473e-01
-1.28046621e-03 -6.56456947e-01 -6.84449673e-01 -4.76051152e-01
9.07770157e-01 -2.74201911e-02 2.61698365e-01 6.53207123e-01
1.19203866e+00 -4.13321465e-01 -2.70221412e-01 7.52099574e-01
1.77298689e+00 6.64662778e-01 8.81892562e-01 2.02349067e-01
4.15611379e-02 4.91915569e-02 4.15828496e-01 8.27496767e-01
-3.23422588e-02 5.87568521e-01 3.24800551e-01 -1.61918804e-01
-4.07646954e-01 1.82990551e-01 -6.73444988e-03 1.48865879e+00
-2.50304937e-01 1.42766073e-01 -5.45133531e-01 2.16348678e-01
-1.08352840e+00 -5.76481700e-01 -2.55007237e-01 2.38619494e+00
8.47944081e-01 3.60810101e-01 -2.44824395e-01 1.10739839e+00
3.75585735e-01 -1.80023268e-01 -3.57708722e-01 -8.33628237e-01
1.50344610e-01 9.62324202e-01 6.91831648e-01 4.07470524e-01
-4.78811890e-01 -7.77205676e-02 5.33678627e+00 1.02568758e+00
-1.33157170e+00 3.08419347e-01 5.79783618e-01 -1.26395598e-01
-9.48609188e-02 -2.59998798e-01 -3.09755653e-01 4.35698748e-01
1.01088250e+00 -2.97973096e-01 6.68436289e-02 5.95455050e-01
1.29342303e-01 -4.19757128e-01 -7.81747103e-01 1.56412983e+00
-1.04152344e-01 -1.28914094e+00 1.75131410e-02 2.82500893e-01
9.92812738e-02 -4.45314080e-01 1.57287285e-01 -2.76969314e-01
-7.79488385e-01 -8.01300108e-01 5.39209425e-01 -7.37585202e-02
1.34156775e+00 -8.34345639e-01 7.78154433e-01 2.89865524e-01
-1.14420116e+00 1.67462453e-01 -5.49791515e-01 -1.38808474e-01
2.72964388e-01 9.07445908e-01 -6.62578881e-01 4.22895193e-01
8.11308265e-01 1.42733321e-01 7.66731426e-02 8.99584949e-01
5.90399206e-01 2.11907461e-01 -5.75066984e-01 -1.31333455e-01
7.51598179e-02 -1.83587730e-01 4.90659237e-01 1.01265287e+00
8.95025074e-01 6.15768313e-01 -4.56575304e-01 2.08518192e-01
1.80232435e-01 2.98754990e-01 -5.44495702e-01 2.91568995e-01
1.44478023e-01 7.37255394e-01 -9.97783244e-01 -5.01225412e-01
-3.07949454e-01 1.11503029e+00 -4.99051243e-01 -2.45195016e-01
-8.82335007e-01 -7.77434647e-01 2.26177201e-01 5.41166425e-01
1.64302364e-01 -2.52524137e-01 -4.11847919e-01 -7.28279054e-01
1.33739531e-01 -7.47917235e-01 1.99977264e-01 -5.54427087e-01
-5.38450539e-01 8.87195766e-01 1.84735864e-01 -1.72230208e+00
-2.04111084e-01 -4.80727136e-01 1.73689887e-01 7.58240461e-01
-1.42172182e+00 -4.77500886e-01 -3.20883989e-01 7.26258457e-01
3.28316152e-01 4.71441932e-02 1.17709613e+00 7.50372589e-01
3.00145686e-01 5.75851977e-01 1.95629030e-01 -5.02481341e-01
1.77418008e-01 -4.77989316e-01 4.29180935e-02 3.57361376e-01
-3.99969846e-01 5.59352696e-01 6.88330352e-01 -5.09500444e-01
-1.85696423e+00 -6.02568150e-01 8.63563180e-01 7.17618287e-01
3.49814966e-02 7.25906640e-02 -5.95562458e-01 2.63207167e-01
2.54229277e-01 3.85087013e-01 7.11693287e-01 -9.44754779e-01
7.73705170e-02 -3.93388122e-01 -1.92798698e+00 3.66148144e-01
7.23353624e-01 -3.18469644e-01 -3.17506760e-01 4.77870464e-01
5.14640868e-01 -6.68126106e-01 -1.38125098e+00 3.91503334e-01
7.08857238e-01 -1.16350663e+00 1.18938386e+00 4.57830071e-01
3.22467804e-01 -3.06407660e-01 -4.10613775e-01 -6.77364409e-01
-1.47678405e-01 -7.42377281e-01 -2.64585571e-04 4.01195109e-01
-7.14117214e-02 -6.27329171e-01 4.93034810e-01 4.15182024e-01
-1.39019266e-01 -1.02147460e+00 -1.43945491e+00 -7.01706111e-01
-1.92283243e-01 -2.64941216e-01 3.13197494e-01 4.84925807e-01
4.43077773e-01 -2.93873567e-02 -2.50882059e-01 -3.00122559e-01
8.09240639e-01 -2.82365143e-01 1.51157109e-02 -9.15075779e-01
-4.42498296e-01 -5.75403012e-02 -1.22033882e+00 -9.96434033e-01
-8.94552708e-01 -9.91301060e-01 -5.37452221e-01 -1.21954656e+00
8.28984659e-03 -6.68273449e-01 -4.51748334e-02 8.31984878e-02
5.47244489e-01 7.46629596e-01 1.32367760e-01 2.75245368e-01
1.66446388e-01 4.58333880e-01 1.34975195e+00 -2.46217117e-01
8.96622315e-02 -1.94331482e-01 -2.39853770e-01 6.74262464e-01
6.39545023e-01 -5.56590617e-01 -5.73357821e-01 -6.08132303e-01
1.63725778e-01 7.02485204e-01 1.09685354e-01 -1.50816727e+00
3.01125258e-01 5.72320461e-01 4.15739805e-01 -7.23565161e-01
6.90283716e-01 -1.17417955e+00 6.45950317e-01 1.41058218e+00
-1.34768439e-02 4.78730649e-01 1.88795790e-01 2.05852121e-01
-3.66026461e-01 -3.42599571e-01 1.08427715e+00 -2.42080048e-01
-2.02972382e-01 3.77321703e-04 -5.18001080e-01 -4.98738736e-01
1.30597794e+00 -6.66822612e-01 3.70738650e-04 -1.63659230e-01
-6.30764663e-01 -6.72237992e-01 1.83114916e-01 -3.86243910e-01
1.04156768e+00 -1.41699815e+00 -4.64898288e-01 2.83648849e-01
-3.73454690e-01 -9.66041386e-02 5.03576219e-01 1.12703598e+00
-1.49970973e+00 5.89618742e-01 -6.59437299e-01 -6.85114920e-01
-1.35776913e+00 4.58343267e-01 2.15576179e-02 -2.95690835e-01
-1.23912525e+00 5.03473341e-01 -1.98900431e-01 4.29077566e-01
1.96678620e-02 -6.51929617e-01 -2.13784445e-02 -2.87969887e-01
7.40918577e-01 7.28212357e-01 5.81113696e-01 -4.69268501e-01
-3.62303764e-01 9.34199870e-01 2.21813232e-01 -1.75378278e-01
1.35796404e+00 -9.02915820e-02 -3.29370648e-01 -2.51859501e-02
1.63064027e+00 -3.16331685e-01 -4.04253542e-01 -7.49581233e-02
-4.69436646e-01 -5.14983833e-01 2.79894024e-01 -3.48346859e-01
-1.06736851e+00 9.46006298e-01 1.15583479e+00 5.64713180e-02
1.73872590e+00 -5.07523000e-01 1.32090986e+00 8.78251195e-02
9.21633720e-01 -7.67645121e-01 -2.63688743e-01 -2.02826470e-01
7.26543069e-01 -7.13868439e-01 7.07828820e-01 -6.38584256e-01
-1.04880773e-01 1.36313224e+00 -1.96468562e-01 -2.46556655e-01
1.05969989e+00 6.20283484e-01 -3.72701615e-01 -1.69062361e-01
-2.94626385e-01 4.52253908e-01 -1.97710648e-01 4.93595034e-01
5.96249282e-01 1.78332314e-01 -1.32540727e+00 1.94914192e-01
-3.30255479e-01 5.51600516e-01 5.93700826e-01 1.36981082e+00
-3.27871770e-01 -9.05185819e-01 -5.33054709e-01 5.96957028e-01
-9.40231085e-01 -1.61416233e-01 6.59287512e-01 5.93423069e-01
1.14417396e-01 7.50804722e-01 4.14302945e-02 -2.78289616e-01
2.79179633e-01 -2.62261242e-01 8.50416780e-01 1.37386575e-01
-5.99517703e-01 2.58133084e-01 -2.70922273e-01 -5.67811131e-01
-5.30718684e-01 -2.75167495e-01 -1.46927238e+00 -6.25666142e-01
1.86687380e-01 -1.15425520e-01 1.09177506e+00 4.64238137e-01
2.94220060e-01 6.43623054e-01 3.51116300e-01 -6.63732052e-01
-4.83754784e-01 -7.46416330e-01 -6.24944985e-01 1.39804229e-01
2.72174209e-01 -5.41480541e-01 -5.00078201e-02 2.07083002e-01] | [11.483774185180664, -2.2972264289855957] |
a73ddbfe-42a5-44f1-9dcf-4e9a3536fc7c | pose-guided-human-animation-from-a-single | 2012.03796 | null | https://arxiv.org/abs/2012.03796v2 | https://arxiv.org/pdf/2012.03796v2.pdf | Pose-Guided Human Animation from a Single Image in the Wild | We present a new pose transfer method for synthesizing a human animation from a single image of a person controlled by a sequence of body poses. Existing pose transfer methods exhibit significant visual artifacts when applying to a novel scene, resulting in temporal inconsistency and failures in preserving the identity and textures of the person. To address these limitations, we design a compositional neural network that predicts the silhouette, garment labels, and textures. Each modular network is explicitly dedicated to a subtask that can be learned from the synthetic data. At the inference time, we utilize the trained network to produce a unified representation of appearance and its labels in UV coordinates, which remains constant across poses. The unified representation provides an incomplete yet strong guidance to generating the appearance in response to the pose change. We use the trained network to complete the appearance and render it with the background. With these strategies, we are able to synthesize human animations that can preserve the identity and appearance of the person in a temporally coherent way without any fine-tuning of the network on the testing scene. Experiments show that our method outperforms the state-of-the-arts in terms of synthesis quality, temporal coherence, and generalization ability. | ['Christian Theobalt', 'Hyun Soo Park', 'Kripasindhu Sarkar', 'Vladislav Golyanik', 'Lingjie Liu', 'Jae Shin Yoon'] | 2020-12-07 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Yoon_Pose-Guided_Human_Animation_From_a_Single_Image_in_the_Wild_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Yoon_Pose-Guided_Human_Animation_From_a_Single_Image_in_the_Wild_CVPR_2021_paper.pdf | cvpr-2021-1 | ['pose-transfer'] | ['computer-vision'] | [ 4.30147201e-01 2.86276966e-01 3.41149956e-01 -2.73781717e-01
-3.82845968e-01 -6.72866821e-01 6.76664829e-01 -4.86792624e-01
-1.10496152e-02 5.69478631e-01 -2.64734887e-02 3.55094731e-01
4.78204995e-01 -5.85175216e-01 -1.08182204e+00 -6.56430542e-01
1.03306711e-01 5.88105142e-01 2.61196405e-01 -2.28954121e-01
-2.26649627e-01 6.01627767e-01 -1.76559317e+00 2.43921101e-01
4.58096564e-01 7.98850536e-01 -5.65537252e-02 8.67271483e-01
4.61872101e-01 5.89144289e-01 -7.77872860e-01 -3.45373839e-01
5.62344313e-01 -8.84395957e-01 -6.10955358e-01 4.37036633e-01
1.15031958e+00 -4.72413003e-01 -4.16702211e-01 7.95332432e-01
3.52333158e-01 3.34206909e-01 6.89393282e-01 -1.22178721e+00
-3.40800673e-01 6.97645620e-02 -3.83564234e-01 -4.84867573e-01
6.73486352e-01 4.53394473e-01 5.64296365e-01 -6.10532939e-01
1.20834219e+00 1.50086749e+00 5.62398434e-01 9.69132364e-01
-1.57273829e+00 -4.57304686e-01 2.09564850e-01 -1.22930437e-01
-1.24815154e+00 -5.49490392e-01 8.28702569e-01 -5.04274905e-01
3.16117167e-01 2.64523119e-01 9.94428992e-01 1.44454217e+00
1.89893574e-01 3.61861467e-01 9.04021978e-01 -4.97367412e-01
2.09845692e-01 -1.30612254e-01 -5.45135319e-01 9.49038506e-01
-4.01706286e-02 2.85581380e-01 -6.82246089e-01 7.65009150e-02
1.21490681e+00 -3.86170775e-01 -2.57238209e-01 -6.77317679e-01
-1.25271428e+00 3.42175007e-01 3.22359055e-01 -3.89600210e-02
-3.39089155e-01 5.42939782e-01 1.77707389e-01 2.66027421e-01
3.89055580e-01 5.07044017e-01 -2.80731261e-01 2.00277597e-01
-8.20363164e-01 5.03164232e-01 7.33606815e-01 7.28906572e-01
6.31358027e-01 4.00313824e-01 -3.32724601e-01 3.29762489e-01
-5.49047105e-02 6.09049737e-01 1.28539473e-01 -1.41156828e+00
2.24666037e-02 3.56512368e-01 1.94238067e-01 -1.09664142e+00
-3.62746090e-01 -3.35062504e-01 -6.80679798e-01 8.16954613e-01
6.54252529e-01 -3.67886871e-01 -1.19189072e+00 2.17020869e+00
6.91916287e-01 3.00122172e-01 -2.12043732e-01 1.00097132e+00
5.10752201e-01 5.27603626e-01 -2.03706278e-03 9.30162668e-02
1.17395759e+00 -9.31417286e-01 -6.53403223e-01 -5.09775639e-01
-1.73715632e-02 -7.51714885e-01 1.02103293e+00 2.50707895e-01
-1.48506427e+00 -9.46180880e-01 -1.11725628e+00 -1.22810043e-01
-2.70171575e-02 1.64106131e-01 2.09053844e-01 2.69713104e-01
-1.21123827e+00 9.25763965e-01 -9.40545499e-01 -4.24280047e-01
-6.77219853e-02 3.18383813e-01 -5.02860785e-01 2.53513783e-01
-1.01578450e+00 9.57649291e-01 9.34608504e-02 7.76882395e-02
-9.81702268e-01 -5.87477326e-01 -9.82302606e-01 -1.71524975e-02
2.33920813e-01 -1.25887835e+00 1.21352375e+00 -1.55085278e+00
-2.02825260e+00 9.08283770e-01 -1.19711151e-02 -1.09394059e-01
9.30138111e-01 4.07161713e-02 -8.05277750e-02 3.06558818e-01
-3.14686983e-03 1.14286232e+00 1.23407805e+00 -1.63281512e+00
-4.46053475e-01 -1.93709314e-01 -1.01346903e-01 2.78342932e-01
4.73156385e-02 -2.48904541e-01 -7.16767251e-01 -8.58967304e-01
3.58821563e-02 -1.24489403e+00 -6.79013804e-02 6.01959109e-01
-3.72463375e-01 3.78595084e-01 7.06737578e-01 -8.28462541e-01
6.11455858e-01 -2.06994200e+00 7.11230099e-01 1.69636682e-01
-1.20827794e-01 -1.73655637e-02 -3.34315300e-01 1.73767477e-01
-2.24216014e-01 -3.02265078e-01 -8.32149088e-02 -8.48945022e-01
-1.02698013e-01 1.25182584e-01 -2.00332642e-01 6.73252404e-01
2.06054449e-01 7.82256782e-01 -8.91095102e-01 -3.91955793e-01
2.83448368e-01 7.83243954e-01 -6.24742031e-01 6.28212929e-01
-6.10794604e-01 1.11293137e+00 -1.51685953e-01 2.42690489e-01
3.77873570e-01 -1.84986055e-01 3.86175096e-01 -4.65088397e-01
1.52247518e-01 -1.01104081e-02 -1.19975960e+00 2.07201433e+00
-3.11254859e-01 5.61073184e-01 9.92709622e-02 -5.83638191e-01
7.47863710e-01 3.97312522e-01 4.28056031e-01 -5.09754956e-01
2.52964169e-01 3.62239778e-02 -9.76576135e-02 -3.54343385e-01
2.10052118e-01 -1.85008332e-01 -5.76577149e-02 4.54515845e-01
5.01638092e-02 -3.02637309e-01 1.13095306e-01 -1.35270253e-01
7.80242324e-01 8.38776231e-01 -1.82477042e-01 5.88390343e-02
3.53875041e-01 -9.29606110e-02 4.12850201e-01 3.07929486e-01
2.14003503e-01 1.02611411e+00 3.73588026e-01 -7.10853517e-01
-1.43132854e+00 -1.29930210e+00 4.98076469e-01 9.82438445e-01
1.62910968e-01 -1.52436513e-02 -9.15519357e-01 -2.75611997e-01
-2.33367249e-01 4.89106566e-01 -8.71113300e-01 -3.44229192e-01
-9.54541624e-01 -1.87336616e-02 4.10440475e-01 2.82562405e-01
3.84658158e-01 -1.18538988e+00 -8.91546190e-01 7.63593391e-02
-2.91180104e-01 -1.27315795e+00 -7.42837012e-01 -3.54372829e-01
-5.33898056e-01 -1.06888354e+00 -7.32632041e-01 -9.13047135e-01
9.59955275e-01 -2.97034502e-01 1.07375300e+00 -1.62269324e-02
-4.68917757e-01 2.52233297e-01 2.80894432e-02 6.51047900e-02
-7.99275219e-01 -1.62584275e-01 1.07775919e-01 4.06948388e-01
-7.27136552e-01 -8.29311311e-01 -7.13540912e-01 1.70043319e-01
-8.30519617e-01 6.18427932e-01 1.22799642e-01 6.51415706e-01
4.22844440e-01 -3.11615109e-01 1.47988945e-01 -5.41953683e-01
3.16846460e-01 1.86747655e-01 -6.07337236e-01 2.67565072e-01
-5.55644780e-02 1.95404500e-01 7.55583525e-01 -8.64889324e-01
-1.18020678e+00 6.13061368e-01 1.73870787e-01 -6.54096842e-01
-1.12887196e-01 -1.22063346e-01 -1.48030696e-02 -1.24379709e-01
8.38623524e-01 6.94229156e-02 3.08768690e-01 -4.16231424e-01
6.78529739e-01 -1.65921181e-01 1.01919198e+00 -7.44871378e-01
9.69817460e-01 6.52586579e-01 3.25449467e-01 -5.35566211e-01
-6.14339292e-01 2.26551563e-01 -8.97547901e-01 -5.53511739e-01
1.05109465e+00 -8.96512926e-01 -8.44550788e-01 6.66458368e-01
-1.33339930e+00 -5.79301596e-01 -4.04327065e-01 1.76911131e-01
-8.20566118e-01 1.60984382e-01 -6.04903221e-01 -3.85813683e-01
-1.96468636e-01 -1.13090014e+00 1.31070101e+00 8.90789032e-02
-5.14238000e-01 -7.48685002e-01 2.42266908e-01 4.34178673e-02
1.77228764e-01 8.92099559e-01 8.33197057e-01 1.01463526e-01
-5.94546199e-01 -1.28461555e-01 1.75624147e-01 2.27746382e-01
1.69014648e-01 2.98943222e-01 -8.69094968e-01 -4.94084388e-01
-3.65276426e-01 -2.82322168e-01 4.07134622e-01 3.51012617e-01
8.20062339e-01 -4.40339595e-01 -1.64496988e-01 7.48364747e-01
1.12495279e+00 1.31529314e-03 5.02972841e-01 2.36071032e-02
8.10688794e-01 7.56310344e-01 1.22737795e-01 3.03234249e-01
9.80239809e-02 1.07398260e+00 3.58006716e-01 -3.77658486e-01
-6.31845415e-01 -5.14522076e-01 5.19950032e-01 2.44276673e-01
-2.66217738e-01 -8.02174285e-02 -4.93798018e-01 2.91153550e-01
-1.75856674e+00 -9.69158530e-01 3.91772091e-01 2.35177517e+00
8.97638142e-01 -8.53435919e-02 2.18094587e-01 -3.14883888e-01
8.22098017e-01 8.36798623e-02 -5.68257570e-01 -3.10104638e-01
8.55844934e-03 2.01623216e-01 1.52521715e-01 6.38127089e-01
-9.58486497e-01 1.02246547e+00 6.77275896e+00 3.38865191e-01
-1.44560754e+00 -2.06849530e-01 5.45839310e-01 -2.17333212e-01
-1.93009213e-01 -7.12026060e-02 -2.20263824e-01 1.78048968e-01
6.59278154e-01 -6.61413968e-02 6.97356343e-01 5.64814448e-01
2.44057372e-01 1.94657385e-01 -1.38834620e+00 7.71649659e-01
3.40299517e-01 -1.35163474e+00 2.52832919e-01 -2.29841381e-01
8.88466060e-01 -4.99575555e-01 2.81585723e-01 -1.02284193e-01
2.80383378e-01 -9.22174096e-01 1.20228052e+00 7.36922264e-01
1.10769880e+00 -5.52443445e-01 8.85353759e-02 1.58849955e-01
-9.61622059e-01 2.93938726e-01 1.47019610e-01 -9.58875790e-02
4.12553221e-01 -9.63875279e-02 -7.29959905e-01 3.35477680e-01
4.03522879e-01 3.61982018e-01 -5.42459428e-01 6.84132159e-01
-4.41765934e-01 -4.50185351e-02 -1.83386177e-01 4.70676750e-01
-1.57330289e-01 -2.83201277e-01 5.37287116e-01 7.53182292e-01
4.22482938e-01 -1.20481424e-01 4.09783125e-01 1.13691521e+00
-7.90276602e-02 -1.46839142e-01 -5.36478758e-01 2.30742604e-01
1.33142039e-01 1.09480274e+00 -4.89463687e-01 -4.03925955e-01
6.39338568e-02 1.44488525e+00 3.27301711e-01 5.48307955e-01
-8.83222103e-01 1.21502712e-01 6.43346667e-01 3.36072862e-01
1.96551606e-01 -1.57358468e-01 -1.54805422e-01 -1.11817241e+00
8.92417952e-02 -1.07860863e+00 -5.41056432e-02 -1.09537995e+00
-1.03713989e+00 9.17650104e-01 1.16982155e-01 -1.26633871e+00
-7.48326302e-01 -3.86033833e-01 -5.24304032e-01 8.17127168e-01
-8.29563260e-01 -1.46915519e+00 -4.63698655e-01 5.06307185e-01
4.14920449e-01 2.99160760e-02 8.54934633e-01 1.83981955e-01
-4.28226650e-01 4.71236795e-01 -4.61418718e-01 4.75237966e-02
1.08120370e+00 -1.02603173e+00 6.22295082e-01 9.18782294e-01
-6.31043836e-02 3.54101360e-01 1.17948890e+00 -5.90528786e-01
-1.19789004e+00 -1.02713895e+00 6.16774142e-01 -4.35136974e-01
3.32166642e-01 -4.23708528e-01 -6.33696735e-01 8.01611125e-01
2.25030020e-01 7.82273412e-02 5.72514534e-02 -3.35442752e-01
-3.37508500e-01 -1.76091000e-01 -1.05366063e+00 1.09132528e+00
9.05771315e-01 -4.72617447e-01 -2.76279628e-01 2.22296551e-01
6.88623011e-01 -8.31574380e-01 -5.88353157e-01 2.65254676e-01
9.30708051e-01 -8.81617069e-01 1.03679371e+00 -6.09941125e-01
5.92307806e-01 -4.78311241e-01 2.23734811e-01 -1.35126472e+00
-2.80805796e-01 -7.86156416e-01 1.04183376e-01 8.12865674e-01
2.30458021e-01 -2.59329140e-01 8.77201378e-01 7.68465638e-01
5.15320860e-02 -3.85149717e-01 -9.22697306e-01 -5.48144937e-01
-1.72771811e-02 6.10977449e-02 5.67792118e-01 6.73017859e-01
-2.31037080e-01 3.90689015e-01 -8.74057412e-01 2.82478809e-01
6.45918071e-01 2.78372794e-01 1.25672615e+00 -9.95535314e-01
-4.97038543e-01 -1.45147532e-01 -2.61148512e-01 -9.00394857e-01
2.89334625e-01 -4.49911743e-01 2.60694653e-01 -1.19835424e+00
5.33116683e-02 -5.49599193e-02 3.32417905e-01 5.28947055e-01
2.38803259e-04 4.14746135e-01 4.81614500e-01 1.19345747e-01
-1.89021260e-01 5.29863179e-01 1.69762313e+00 -1.12258092e-01
-2.11558104e-01 -1.37783751e-01 -7.65948519e-02 8.62911999e-01
6.29845262e-01 -2.55752385e-01 -4.54454273e-01 -5.38481474e-01
1.24439441e-01 3.59925270e-01 6.69812977e-01 -1.17032146e+00
1.32595465e-01 -9.36634615e-02 7.55562425e-01 -1.81768745e-01
8.61615658e-01 -7.77511716e-01 8.60079348e-01 6.08984172e-01
-4.68996257e-01 1.71885967e-01 1.75838307e-01 3.70006710e-01
3.31435092e-02 2.89883465e-01 9.67103004e-01 -1.84663847e-01
-4.24301594e-01 4.46418494e-01 -9.09177139e-02 -1.67511776e-01
8.53527606e-01 -2.02981323e-01 -8.80505219e-02 -6.21153414e-01
-1.06520367e+00 -9.19073075e-02 1.02527642e+00 4.95867103e-01
4.45908368e-01 -1.46451342e+00 -7.72313714e-01 5.41890502e-01
-1.03455365e-01 -5.73178306e-02 3.22025001e-01 3.34446371e-01
-7.77130663e-01 -1.36948660e-01 -6.73865795e-01 -5.34574389e-01
-1.30600536e+00 5.72096467e-01 7.67540872e-01 -1.19962551e-01
-7.81152308e-01 4.72675353e-01 5.09750903e-01 -2.66854346e-01
2.13906989e-01 -1.28337815e-01 1.79039270e-01 -2.61701822e-01
1.66323721e-01 1.04861714e-01 -3.48972261e-01 -8.17848504e-01
-1.09699808e-01 8.69489551e-01 4.08304155e-01 -5.81415772e-01
1.09143364e+00 2.52162311e-02 1.47198280e-03 4.18769896e-01
1.04290664e+00 -3.03282961e-02 -1.89908147e+00 9.76847261e-02
-8.15985739e-01 -5.03424823e-01 -4.67842788e-01 -8.08107078e-01
-1.19368386e+00 6.21855676e-01 5.28912127e-01 -3.01060319e-01
9.94785845e-01 -2.35953107e-01 6.93746448e-01 4.24828939e-02
3.87914062e-01 -1.05984235e+00 3.80533874e-01 2.60088384e-01
1.20623553e+00 -8.12365770e-01 -2.22389842e-03 -4.97583628e-01
-5.53605020e-01 1.06875980e+00 7.65609622e-01 -3.87366861e-01
2.97999740e-01 1.25844285e-01 2.04918355e-01 -1.09847881e-01
-6.97035074e-01 1.37284786e-01 6.28419816e-01 6.79335952e-01
2.15646058e-01 2.17911936e-02 1.18307143e-01 -1.94049567e-01
-2.83510596e-01 -1.79413617e-01 4.21864301e-01 5.75997829e-01
-1.85458004e-01 -1.23344898e+00 -4.25067097e-01 -2.62734741e-01
-1.44939065e-01 3.22638899e-01 -4.99168813e-01 8.25562298e-01
3.79028827e-01 4.57867473e-01 1.32694960e-01 -2.74980277e-01
4.80188906e-01 5.30173443e-02 9.27419007e-01 -5.57113707e-01
-5.02010584e-01 2.34752879e-01 1.47755146e-01 -7.62277186e-01
-4.44166839e-01 -5.76281965e-01 -1.09062719e+00 -2.37306610e-01
2.51063824e-01 -3.65483254e-01 3.78141761e-01 6.89380705e-01
2.91851014e-01 7.24306881e-01 5.47695756e-01 -1.43164837e+00
-2.81044275e-01 -6.57088041e-01 -4.21870381e-01 9.89195108e-01
4.41456527e-01 -5.97808957e-01 -2.43170410e-01 6.58402622e-01] | [11.578238487243652, -0.7588024139404297] |
33f9acfe-fb9e-4dcf-b31d-bb0cb723675f | swde-a-sub-word-and-document-embedding-based | 1808.00957 | null | http://arxiv.org/abs/1808.00957v1 | http://arxiv.org/pdf/1808.00957v1.pdf | SWDE : A Sub-Word And Document Embedding Based Engine for Clickbait Detection | In order to expand their reach and increase website ad revenue, media outlets
have started using clickbait techniques to lure readers to click on articles on
their digital platform. Having successfully enticed the user to open the
article, the article fails to satiate his curiosity serving only to boost
click-through rates. Initial methods for this task were dependent on feature
engineering, which varies with each dataset. Industry systems have relied on an
exhaustive set of rules to get the job done. Neural networks have barely been
explored to perform this task. We propose a novel approach considering
different textual embeddings of a news headline and the related article. We
generate sub-word level embeddings of the title using Convolutional Neural
Networks and use them to train a bidirectional LSTM architecture. An attention
layer allows for calculation of significance of each term towards the nature of
the post. We also generate Doc2Vec embeddings of the title and article text and
model how they interact, following which it is concatenated with the output of
the previous component. Finally, this representation is passed through a neural
network to obtain a score for the headline. We test our model over 2538 posts
(having trained it on 17000 records) and achieve an accuracy of 83.49%
outscoring previous state-of-the-art approaches. | ['Manish Shrivastava', 'Yash Kumar Lal', 'Vasudeva Varma', 'Vaibhav Kumar', 'Mrinal Dhar', 'Dhruv Khattar', 'Abhimanshu Mishra'] | 2018-08-02 | null | null | null | null | ['document-embedding', 'clickbait-detection'] | ['methodology', 'natural-language-processing'] | [-5.12239672e-02 1.52471542e-01 -1.74688041e-01 -3.08590710e-01
-7.15864778e-01 -6.90108955e-01 9.27842081e-01 4.98472601e-01
-7.40588605e-01 3.50199372e-01 4.35370743e-01 -6.89031422e-01
1.66398525e-01 -8.59008908e-01 -9.48190272e-01 -1.49178386e-01
1.23078480e-01 3.15063596e-01 2.60751098e-01 -3.55434746e-01
5.32107055e-01 8.05206597e-02 -1.41218400e+00 4.62255448e-01
5.85504830e-01 1.06710470e+00 1.38684779e-01 6.82065547e-01
-4.47617143e-01 6.94404960e-01 -5.45870721e-01 -5.78639150e-01
1.07476674e-01 -1.31213889e-01 -6.77775204e-01 -1.14035107e-01
4.58592147e-01 -3.26253742e-01 -6.63837254e-01 8.74702036e-01
2.73071766e-01 -8.20662528e-02 5.08480966e-01 -7.77103841e-01
-1.05416942e+00 9.96596217e-01 -4.30810332e-01 2.85003304e-01
2.08363384e-01 -1.92669839e-01 1.54306591e+00 -9.32482779e-01
7.87877142e-01 7.25855350e-01 4.43850428e-01 9.65162218e-02
-1.33885252e+00 -3.59761685e-01 1.34238154e-01 2.47525170e-01
-8.37311745e-01 -1.64349318e-01 5.57873249e-01 -5.27811885e-01
1.06984949e+00 1.87592387e-01 7.31500924e-01 1.34560442e+00
3.32305491e-01 6.60927534e-01 8.87232184e-01 -3.33492965e-01
-7.65430108e-02 3.90760809e-01 4.26260591e-01 3.26961100e-01
1.71235278e-01 -3.56619537e-01 -2.50346839e-01 7.09937811e-02
2.13365138e-01 6.87809065e-02 -2.83914432e-02 1.09376684e-01
-9.81951952e-01 1.11850822e+00 7.82486916e-01 5.04557073e-01
-6.61892593e-01 8.03595111e-02 3.87539536e-01 3.03078294e-01
4.18800503e-01 7.27313280e-01 -3.75164539e-01 -8.49419981e-02
-9.70407188e-01 5.19596517e-01 1.07198393e+00 4.00246799e-01
5.81266820e-01 -4.76754993e-01 -2.42890969e-01 9.23106253e-01
2.58743912e-01 1.50823593e-01 7.28496611e-01 -2.62993306e-01
4.62412328e-01 8.66657317e-01 1.01109091e-02 -1.14571619e+00
-2.24801123e-01 -7.66139925e-01 -2.58227855e-01 -1.76010597e-02
4.20280844e-01 -1.75393030e-01 -8.91555369e-01 1.18619800e+00
-8.21913257e-02 -2.51953423e-01 -3.89011532e-01 6.45747304e-01
5.99417508e-01 8.51117253e-01 5.67616299e-02 3.54578286e-01
1.32348263e+00 -1.02349997e+00 -6.36022866e-01 -3.45159590e-01
7.96309888e-01 -1.01795733e+00 1.06081367e+00 4.12740141e-01
-1.05571687e+00 -3.78662974e-01 -1.20640910e+00 -2.89152324e-01
-1.02368426e+00 1.22252792e-01 3.03461522e-01 1.40352041e-01
-7.21867323e-01 8.33775997e-01 -5.39672136e-01 -3.70546490e-01
4.70910609e-01 1.22745879e-01 -1.62038237e-01 9.71149728e-02
-1.34576738e+00 1.03641403e+00 2.09865496e-01 -2.26963833e-02
-5.22332370e-01 -7.77413368e-01 -5.53758264e-01 2.78247327e-01
2.21334338e-01 -3.96892875e-01 1.17420518e+00 -8.15631628e-01
-1.09607911e+00 6.30440116e-01 1.14870012e-01 -7.05010116e-01
3.82050902e-01 -3.64494413e-01 -6.10305488e-01 -2.64485091e-01
1.97691396e-01 4.59491313e-01 7.74354517e-01 -7.34671235e-01
-8.42224360e-01 -3.24147791e-01 2.22437531e-01 -8.69916603e-02
-7.13850617e-01 2.82750614e-02 -6.34553611e-01 -4.87071931e-01
-1.02396689e-01 -8.89217138e-01 -1.44373670e-01 -3.13806385e-01
-5.92810214e-01 -2.71174610e-01 7.10220039e-01 -9.37797546e-01
1.67624617e+00 -2.11827993e+00 -6.37786649e-03 2.12905198e-01
2.87095487e-01 3.38574946e-01 -8.32511559e-02 7.90607512e-01
5.61261512e-02 5.24151444e-01 3.01951468e-01 -1.57052442e-01
2.46159285e-01 -1.15543500e-01 -4.50179040e-01 1.52435154e-01
2.76928544e-01 8.92380834e-01 -6.20976567e-01 -4.79699560e-02
-1.49361998e-01 5.80135822e-01 -7.99914002e-01 7.41397515e-02
-4.28082079e-01 -3.30689043e-01 -4.50404018e-01 2.76580453e-01
4.42936659e-01 -5.07980466e-01 1.49857283e-01 -2.54273295e-01
-4.00981903e-01 8.73067737e-01 -7.18227088e-01 1.03298378e+00
-6.32338047e-01 9.74771917e-01 -3.31889361e-01 -6.65141523e-01
8.15712154e-01 -8.76859426e-02 2.38685697e-01 -9.24203753e-01
3.00790429e-01 3.07038486e-01 1.82621270e-01 -5.01545191e-01
7.56725669e-01 1.64450154e-01 -1.63968191e-01 4.76802409e-01
-7.37482756e-02 4.95834976e-01 2.30573177e-01 5.51048338e-01
1.28188539e+00 9.58364680e-02 -1.75624013e-01 -6.68620178e-03
4.53515977e-01 1.30388001e-02 -2.69242227e-01 6.73928380e-01
1.74009070e-01 5.17534852e-01 7.66864359e-01 -4.88770068e-01
-1.40271652e+00 -7.25533485e-01 -1.54778078e-01 1.32159400e+00
-3.10912073e-01 -5.20696223e-01 -7.08300233e-01 -7.31292725e-01
2.17044324e-01 1.08330560e+00 -9.23507035e-01 -4.80049588e-02
-4.25423175e-01 -3.42018276e-01 1.68949753e-01 3.68278533e-01
3.42769951e-01 -7.75801003e-01 -3.21507305e-01 5.37437797e-01
-2.05686670e-02 -9.88829076e-01 -3.32614571e-01 3.63471925e-01
-4.25298780e-01 -9.03525352e-01 -6.48318350e-01 -5.82509458e-01
5.20497501e-01 -9.30024683e-03 7.61219025e-01 -1.80028770e-02
-8.56861100e-02 1.71061251e-02 -3.74965608e-01 -3.64674240e-01
-3.74338180e-01 5.41265428e-01 -3.07553530e-01 1.91711724e-01
6.70896828e-01 -4.26521748e-01 -5.81493556e-01 8.60460401e-02
-1.04650581e+00 -2.84317553e-01 6.55592859e-01 7.72305906e-01
2.03409880e-01 -1.91878393e-01 4.52859223e-01 -9.98610139e-01
1.01935065e+00 -8.33411694e-01 -4.64858711e-01 -8.34580213e-02
-6.72581196e-01 1.87025651e-01 6.31503582e-01 -4.17415679e-01
-5.64841032e-01 -2.85217047e-01 -4.13007081e-01 -6.28328323e-02
2.85933524e-01 1.04498613e+00 4.03291523e-01 3.79735261e-01
6.52851880e-01 8.68007839e-02 1.38216689e-02 -6.33425236e-01
5.00721395e-01 8.77324879e-01 2.53478646e-01 4.99969646e-02
8.68071556e-01 1.48404524e-01 -5.15624046e-01 -7.53650069e-01
-9.18116570e-01 -4.18552339e-01 -3.62392008e-01 -1.64273515e-01
9.00181711e-01 -5.87965250e-01 -6.96692526e-01 8.61089155e-02
-1.17243505e+00 -2.22775154e-02 5.07165166e-03 4.09575284e-01
9.09178779e-02 -7.13721337e-03 -4.48361069e-01 -3.89975369e-01
-2.37600803e-02 -8.78752172e-01 5.68874896e-01 1.32880762e-01
-2.12415114e-01 -1.05738139e+00 6.11208118e-02 3.00351292e-01
8.27040613e-01 1.33465677e-02 9.49051976e-01 -1.44590449e+00
-3.17443967e-01 -7.60217726e-01 -4.29331452e-01 5.00526786e-01
-1.39176920e-01 1.72629580e-01 -9.13352847e-01 6.74260482e-02
-3.72322172e-01 -1.20334029e-01 1.00875461e+00 2.49496251e-01
1.23180985e+00 -6.34132862e-01 -4.07741815e-01 1.36418328e-01
1.32881403e+00 -1.95708379e-01 8.20182741e-01 9.99312341e-01
5.83757937e-01 4.77262944e-01 1.96899310e-01 4.24252629e-01
4.07230794e-01 6.56409681e-01 5.98295450e-01 7.97073916e-02
3.80288959e-02 -5.37021279e-01 4.99281377e-01 6.72763348e-01
1.87605456e-01 -3.74370784e-01 -6.78890884e-01 4.85567510e-01
-1.47494614e+00 -9.22534883e-01 -3.32503140e-01 2.07139730e+00
6.37596607e-01 7.11115718e-01 6.83952048e-02 1.54141439e-02
4.79119390e-01 2.67999232e-01 -3.30981106e-01 -8.73013794e-01
1.62101656e-01 1.76115185e-02 8.45579445e-01 4.85511243e-01
-9.83124137e-01 8.75930667e-01 5.83677864e+00 5.18860400e-01
-1.40949321e+00 -2.73535922e-02 5.84120035e-01 -6.03600405e-02
-5.33774912e-01 -8.03896412e-02 -1.10601950e+00 6.36682034e-01
1.35985219e+00 -1.65696934e-01 4.85967249e-01 9.70204055e-01
2.50726402e-01 4.11780775e-02 -1.00060689e+00 5.29190958e-01
9.27292854e-02 -1.59376800e+00 3.71002629e-02 3.98990661e-01
5.05040646e-01 3.60431850e-01 3.28588307e-01 6.45095646e-01
1.78177252e-01 -9.39107001e-01 6.01266682e-01 4.90109921e-01
3.24402750e-01 -5.18226981e-01 7.54631281e-01 1.30739421e-01
-5.56294978e-01 -2.67385006e-01 -2.39883706e-01 -2.03877375e-01
2.34095901e-01 5.22354722e-01 -1.17190754e+00 1.90778807e-01
4.56675887e-01 6.51088238e-01 -8.62945735e-01 8.46763372e-01
-1.86776668e-01 6.60131514e-01 -2.10059911e-01 -5.02172530e-01
8.26464295e-01 1.21090701e-02 5.27912080e-01 1.19474661e+00
2.44566977e-01 -3.97745609e-01 -2.09560201e-01 9.33603585e-01
-5.45365274e-01 3.24054718e-01 -4.56779242e-01 -8.01719666e-01
2.60978907e-01 1.54277778e+00 -4.48964924e-01 -4.18358386e-01
-6.67684793e-01 9.30928826e-01 4.80625927e-01 2.91676730e-01
-9.95636344e-01 -7.07137167e-01 4.37068641e-01 7.20649302e-01
7.06323087e-01 -2.38836050e-01 -1.81029961e-01 -9.92070973e-01
9.06909779e-02 -7.28617013e-01 1.49439633e-01 -8.46148014e-01
-1.20617259e+00 7.69981384e-01 -4.62336272e-01 -8.62798214e-01
-9.01371092e-02 -7.32572019e-01 -7.32587337e-01 9.68086600e-01
-1.52739108e+00 -9.97449756e-01 -4.25211601e-02 4.71629873e-02
5.24873435e-01 -4.70949672e-02 6.31623507e-01 6.85835242e-01
-4.84899670e-01 4.56087172e-01 1.94115058e-01 2.45874435e-01
7.32071340e-01 -1.20185244e+00 5.12016714e-01 5.68968475e-01
1.18353330e-01 9.08136308e-01 8.47179353e-01 -6.31886482e-01
-1.46229470e+00 -1.00048292e+00 1.52512515e+00 -5.29075265e-01
1.40027595e+00 -4.36024994e-01 -8.94181490e-01 7.95032322e-01
4.67561543e-01 -2.84371674e-01 6.65978253e-01 3.72364223e-01
-4.52485144e-01 1.42379515e-02 -5.73212624e-01 7.57150054e-01
4.19697970e-01 -4.55705673e-01 -7.37692118e-01 3.08603913e-01
7.34167278e-01 -3.71582173e-02 -6.70047104e-01 -2.78980374e-01
7.61254609e-01 -6.72177553e-01 6.73603833e-01 -9.64623153e-01
1.06670630e+00 1.06646240e-01 -1.78982556e-01 -1.27048135e+00
-4.35391456e-01 -3.30134332e-01 6.94503635e-02 1.12998080e+00
9.10449028e-01 -6.78512156e-01 5.65891147e-01 6.64363742e-01
-2.44059548e-01 -9.18267012e-01 -4.31617081e-01 -3.75642270e-01
6.32086918e-02 -3.83290619e-01 3.04319501e-01 7.20130146e-01
7.80624598e-02 5.71810603e-01 -1.83728099e-01 -9.96922106e-02
2.35653222e-01 -1.97332397e-01 6.99249446e-01 -1.12284660e+00
-2.39910334e-01 -7.50461936e-01 -1.40790939e-01 -1.17244112e+00
-1.85322002e-01 -1.23178470e+00 -3.87528092e-01 -1.73289526e+00
6.70287013e-02 -1.36630237e-01 -3.74925792e-01 3.68198752e-01
1.20812707e-01 4.46040004e-01 9.63490009e-02 1.79596528e-01
-4.03715968e-01 2.17627123e-01 8.81993532e-01 -2.50499487e-01
-2.27587640e-01 -1.44968331e-01 -1.08044970e+00 4.38862413e-01
7.26168156e-01 -4.01753813e-01 1.41189367e-01 -3.38878423e-01
7.20189869e-01 -4.65340734e-01 4.88585502e-01 -7.69363344e-01
2.76123494e-01 3.61529797e-01 5.48928857e-01 -6.11597538e-01
1.58378258e-01 -8.53612959e-01 -2.17483863e-01 3.80016506e-01
-9.51501787e-01 4.65759784e-02 9.54912156e-02 6.08271718e-01
-1.98494554e-01 -4.68957901e-01 4.33759093e-01 1.07962400e-01
-3.24904025e-01 9.71222892e-02 -6.25814319e-01 -1.66111648e-01
6.56343400e-01 -2.32797302e-02 -2.84106225e-01 -4.51693177e-01
-8.38830054e-01 1.78459257e-01 1.16533346e-01 6.92158520e-01
3.24106723e-01 -1.23466003e+00 -6.58808708e-01 2.62334734e-01
-5.72820306e-02 -6.23412728e-01 -1.57114081e-02 7.51632750e-01
-4.66060698e-01 5.86111903e-01 -1.31749809e-01 -2.36410022e-01
-9.62208152e-01 5.66652954e-01 4.30961214e-02 -3.52991551e-01
-4.20228988e-01 6.56346381e-01 -1.99423656e-01 -1.75742581e-01
1.12790786e-01 -3.52831066e-01 -5.55800080e-01 5.57840586e-01
6.95131660e-01 2.05298424e-01 2.69167066e-01 -4.18848485e-01
-9.79771018e-02 1.24287754e-01 -6.54344678e-01 -1.98862717e-01
1.74424553e+00 -4.28078556e-03 2.13679839e-02 4.12052691e-01
1.80373049e+00 1.27169237e-01 -7.74753988e-01 -1.00630887e-01
4.39627990e-02 -4.48584855e-01 3.45028162e-01 -8.74861062e-01
-8.06094885e-01 7.90293038e-01 2.23211080e-01 8.33486855e-01
3.99547398e-01 4.46233116e-02 9.74760711e-01 4.03833598e-01
-3.55541676e-01 -1.31720626e+00 1.52657837e-01 6.05026126e-01
8.97250891e-01 -1.11156285e+00 8.82761851e-02 2.01570049e-01
-6.10287845e-01 1.39800537e+00 1.70775503e-01 -2.88376391e-01
7.05860376e-01 -6.22835197e-02 -1.68244764e-01 -2.83585697e-01
-7.82347679e-01 3.66902947e-02 3.41938794e-01 -1.79059114e-02
6.96147025e-01 -2.33357713e-01 -6.10774159e-01 6.65511012e-01
-2.12815464e-01 -3.67144309e-02 6.70767307e-01 6.60458565e-01
-7.25405514e-01 -1.11328340e+00 1.00250550e-01 7.97138929e-01
-9.36999977e-01 -2.57283717e-01 -3.89157832e-01 7.25154400e-01
-2.52619624e-01 6.44514978e-01 2.22855389e-01 -5.75639486e-01
4.23730880e-01 2.84735024e-01 -2.18234852e-01 -4.56624776e-01
-7.81284809e-01 1.67380884e-01 4.21329170e-01 -3.27129006e-01
1.04672119e-01 -6.67788029e-01 -1.02425551e+00 -4.06336337e-01
-2.53899753e-01 1.91605777e-01 1.32707489e+00 7.24611104e-01
4.94766414e-01 7.47834861e-01 7.14300990e-01 -7.19211757e-01
-8.29584181e-01 -9.82181907e-01 -2.20850542e-01 4.75077242e-01
8.37151855e-02 -3.59738499e-01 -5.34155667e-01 -1.45203099e-01] | [7.805327415466309, 9.753251075744629] |
b0bceaa4-dfbc-4f57-aa77-ee7addb5cbca | born-for-auto-tagging-faster-and-better-with | 2206.07264 | null | https://arxiv.org/abs/2206.07264v1 | https://arxiv.org/pdf/2206.07264v1.pdf | Born for Auto-Tagging: Faster and better with new objective functions | Keyword extraction is a task of text mining. It is applied to increase search volume in SEO and ads. Implemented in auto-tagging, it makes tagging on a mass scale of online articles and photos efficiently and accurately. BAT is invented for auto-tagging which served as awoo's AI marketing platform (AMP). awoo AMP not only provides service as a customized recommender system but also increases the converting rate in E-commerce. The strength of BAT converges faster and better than other SOTA models, as its 4-layer structure achieves the best F scores at 50 epochs. In other words, it performs better than other models which require deeper layers at 100 epochs. To generate rich and clean tags, awoo creates new objective functions to maintain similar ${\rm F_1}$ scores with cross-entropy while enhancing ${\rm F_2}$ scores simultaneously. To assure the even better performance of F scores awoo revamps the learning rate strategy proposed by Transformer \cite{Transformer} to increase ${\rm F_1}$ and ${\rm F_2}$ scores at the same time. | ['Huang-Ting Shieh', 'Chiung-ju Liu'] | 2022-06-15 | null | null | null | null | ['keyword-extraction'] | ['natural-language-processing'] | [-9.38351974e-02 1.06001452e-01 -3.82821441e-01 -2.20961899e-01
-7.42445469e-01 -3.30529690e-01 2.16792434e-01 5.60295917e-02
-4.53407139e-01 4.02985901e-01 -1.13918617e-01 -2.18047157e-01
-3.59851986e-01 -7.48199046e-01 -3.71251404e-01 -6.10466897e-01
-3.73303086e-01 2.68910855e-01 2.11313263e-01 -2.92528719e-01
1.37611389e-01 1.15585387e-01 -1.64844561e+00 9.97836739e-02
1.00912917e+00 1.57661974e+00 4.00761545e-01 1.61110297e-01
-4.59913909e-01 6.88288450e-01 -3.64437073e-01 -1.08338523e+00
3.83791417e-01 -1.15602121e-01 -5.07028937e-01 -2.32642934e-01
2.48774931e-01 -3.56915481e-02 -2.87874401e-01 1.36933029e+00
4.04330462e-01 2.70008296e-01 5.44613600e-01 -1.03781796e+00
-8.24451804e-01 1.17701840e+00 -6.35734737e-01 3.81579548e-01
-3.55853736e-02 -2.96661764e-01 1.51526129e+00 -8.89911592e-01
3.42876047e-01 8.28095734e-01 6.83328152e-01 3.18978310e-01
-1.01070321e+00 -9.66604471e-01 3.27972531e-01 2.19695389e-01
-1.48139691e+00 -1.15015388e-01 7.23242283e-01 -1.09541364e-01
8.54062021e-01 5.31145573e-01 6.72437489e-01 8.57002854e-01
2.14229554e-01 1.18340206e+00 7.52904236e-01 -2.83390462e-01
7.12042749e-02 2.85176009e-01 1.13711983e-01 6.97489798e-01
3.29475254e-02 -3.47051173e-01 -5.25167584e-01 9.36564878e-02
5.02932489e-01 1.83085531e-01 3.47175524e-02 3.11343949e-02
-9.49407339e-01 9.46251392e-01 4.65828836e-01 4.53362316e-01
-5.07605076e-01 2.19864752e-02 7.89428279e-02 1.51313484e-01
3.71816188e-01 8.45284462e-01 -5.07553816e-01 -1.96483418e-01
-9.87553239e-01 1.77246198e-01 4.47393268e-01 1.22313488e+00
6.16498709e-01 3.02741081e-01 -2.40963444e-01 1.18061364e+00
1.49682283e-01 7.25836396e-01 5.72349370e-01 -7.23555207e-01
4.88130301e-01 8.65879297e-01 -1.04929343e-01 -1.15540409e+00
-5.43291867e-01 -1.21112084e+00 -8.99685204e-01 -4.95587468e-01
1.18678669e-02 -6.84180781e-02 -9.81035590e-01 1.61051476e+00
8.09729472e-02 -2.02654973e-01 -3.11780959e-01 6.80928171e-01
7.49591649e-01 8.77501786e-01 3.35860491e-01 -3.73221755e-01
1.65720391e+00 -9.86480296e-01 -9.43664610e-01 -5.62394619e-01
3.70385528e-01 -9.67953503e-01 1.02721298e+00 6.23252571e-01
-1.02910852e+00 -4.82549548e-01 -1.00866938e+00 9.82016400e-02
-4.61444050e-01 3.48631203e-01 9.83363986e-01 6.20263934e-01
-6.15813017e-01 6.30509794e-01 -2.70974696e-01 -4.63528633e-02
6.07625544e-01 5.84598064e-01 1.33638501e-01 1.86634943e-01
-1.43325341e+00 6.66552186e-01 2.23111585e-01 9.53055471e-02
-4.91411269e-01 -6.07929409e-01 -5.34328282e-01 2.99783200e-01
7.33054459e-01 -2.19696179e-01 1.14977849e+00 -9.03425932e-01
-1.31436241e+00 6.13719404e-01 2.19812766e-01 -5.24710596e-01
8.75155106e-02 -3.99004281e-01 -9.47371542e-01 -6.98562339e-02
1.07241169e-01 7.24158943e-01 9.64784086e-01 -8.75526130e-01
-8.95749331e-01 -3.55372280e-01 -1.68147653e-01 1.21034071e-01
-8.01990688e-01 -3.82192060e-02 -6.61638200e-01 -9.18396533e-01
1.44027382e-01 -8.84275913e-01 -2.39343792e-01 -6.32011533e-01
-2.51400143e-01 -3.30595493e-01 4.66668427e-01 -6.76850021e-01
1.84034288e+00 -2.09689069e+00 -1.56374797e-01 5.07609069e-01
5.49940050e-01 6.68128014e-01 -3.97101678e-02 2.70611107e-01
1.63324565e-01 1.39173493e-01 3.33338022e-01 1.53866798e-01
2.36716852e-01 -9.89824608e-02 -1.53001055e-01 -2.12128665e-02
3.87066081e-02 9.80901599e-01 -7.46924520e-01 -5.25150239e-01
-1.29436180e-01 2.90378094e-01 -4.97606367e-01 -1.05102971e-01
-3.20638359e-01 -4.62009192e-01 -7.69737065e-01 6.91768408e-01
4.18364793e-01 -6.52979732e-01 2.14020312e-01 -3.78285527e-01
-1.40042067e-01 1.41107887e-01 -1.10485435e+00 1.40092111e+00
-2.53574491e-01 2.21811056e-01 9.87661481e-02 -9.38752532e-01
1.28480816e+00 1.00083323e-02 8.15076709e-01 -1.07547855e+00
4.76234704e-01 6.86135739e-02 -8.62416103e-02 -2.83183247e-01
6.87995315e-01 1.04261085e-01 -2.42037877e-01 1.74542978e-01
2.36052901e-01 1.64840341e-01 2.82930970e-01 3.12966317e-01
1.02471590e+00 3.97041142e-02 1.05597556e-01 -4.08447146e-01
3.95130783e-01 4.25346270e-02 5.98710179e-01 5.98906696e-01
1.43214360e-01 8.81239250e-02 1.31330565e-01 -4.24127162e-01
-1.12816131e+00 -8.88017237e-01 -9.81882289e-02 1.31920469e+00
1.09094687e-01 -6.91547155e-01 -5.98976731e-01 -7.00848043e-01
6.86181709e-02 7.41780162e-01 -5.73307514e-01 -3.14315706e-01
-3.51868659e-01 -9.65031326e-01 2.34820440e-01 2.95862138e-01
6.43576980e-01 -1.11529136e+00 -1.03279471e-01 2.90179640e-01
-3.53646040e-01 -8.64911675e-01 -6.38437867e-01 5.77298403e-01
-6.51115716e-01 -4.91811842e-01 -6.63173735e-01 -7.60359943e-01
5.03261089e-01 2.53243983e-01 1.06251788e+00 -1.06452249e-01
-7.73723125e-02 -1.07643537e-01 -8.58522594e-01 -7.62289822e-01
-7.17625245e-02 6.01642787e-01 1.27554521e-01 8.75182673e-02
7.82748818e-01 -4.98085707e-01 -5.41550994e-01 5.03420770e-01
-8.29867244e-01 -6.63873255e-02 8.32714736e-01 7.09831178e-01
7.35187650e-01 2.39423633e-01 8.18750560e-01 -7.73455083e-01
6.56892896e-01 -3.66557926e-01 -7.52205491e-01 2.06444889e-01
-1.31534469e+00 -2.40803286e-02 7.45744348e-01 -5.93018293e-01
-5.95226169e-01 -2.73124248e-01 -1.65176392e-01 -3.52814436e-01
4.69088048e-01 5.59419632e-01 3.71705964e-02 3.40933084e-01
5.41753054e-01 2.17835858e-01 -3.07865560e-01 -7.54776895e-01
1.55847281e-01 7.64849544e-01 2.97699720e-01 5.39135712e-04
7.76111960e-01 -7.46648386e-02 -3.05298954e-01 -4.55528617e-01
-1.34813702e+00 -5.60109496e-01 -2.45959669e-01 -2.43607059e-01
6.80872738e-01 -8.18892479e-01 -1.07134235e+00 1.10471152e-01
-3.37541312e-01 1.08232871e-01 -3.29143107e-01 5.29952049e-01
-4.16168422e-02 1.43352181e-01 -1.80011198e-01 -9.44912016e-01
-8.35215330e-01 -7.01569796e-01 7.59584069e-01 5.50811648e-01
-1.13411486e-01 -4.78100777e-01 -4.94185209e-01 5.15153289e-01
5.15523374e-01 -4.44050759e-01 7.83842027e-01 -8.80095661e-01
-4.02694643e-01 -4.90453839e-01 -2.54272223e-01 5.08311152e-01
-8.98309052e-02 -1.91190958e-01 -8.27426970e-01 -1.77919835e-01
-1.62904620e-01 1.38286695e-01 7.61129618e-01 2.81455994e-01
1.32022727e+00 -5.54106593e-01 -2.68546671e-01 3.96329582e-01
1.11237359e+00 6.69385612e-01 6.68170273e-01 5.11538327e-01
5.14609098e-01 3.32443923e-01 6.60063148e-01 3.52479517e-01
3.90212625e-01 8.14316571e-01 3.98965180e-01 -4.05633971e-02
4.33385894e-02 -4.35721517e-01 3.92864525e-01 1.17834198e+00
6.48644790e-02 -3.80509108e-01 -7.13645458e-01 4.81264949e-01
-1.65926743e+00 -1.00698149e+00 9.01135653e-02 2.02681828e+00
9.29195285e-01 3.50724876e-01 2.58440554e-01 6.32002577e-02
7.17796385e-01 -2.19394937e-02 -4.46690083e-01 -2.56072462e-01
-1.63132042e-01 4.35300648e-01 5.91683984e-01 -3.72702032e-02
-1.12126434e+00 1.08283186e+00 5.35452890e+00 1.43778956e+00
-8.64075959e-01 -9.00417473e-03 5.87322295e-01 -4.52554107e-01
-4.03229594e-01 -2.00049803e-01 -1.05697811e+00 7.86240876e-01
8.29555571e-01 -9.33191702e-02 6.51830554e-01 1.07622027e+00
-1.84623852e-01 2.37985272e-02 -5.90269208e-01 1.00516009e+00
5.81924990e-03 -1.47115099e+00 2.49360446e-02 2.49190882e-01
6.96495473e-01 5.80557063e-02 2.43600920e-01 6.68930292e-01
4.38191473e-01 -8.88263762e-01 7.31505692e-01 3.07593375e-01
6.75736010e-01 -9.89423037e-01 9.08556104e-01 3.20111603e-01
-1.17703199e+00 -4.87828702e-01 -4.20929343e-01 2.68084675e-01
1.49206534e-01 8.80157888e-01 -7.21701741e-01 5.61598182e-01
9.90803361e-01 3.63174379e-01 -5.45770228e-01 8.67990136e-01
4.06957418e-02 6.68583691e-01 -3.79864752e-01 -5.66228867e-01
3.83138776e-01 -3.83155018e-01 5.39639473e-01 1.05096602e+00
4.84174997e-01 1.54965177e-01 2.27145270e-01 4.12971348e-01
-4.02066737e-01 6.36604190e-01 -9.12149400e-02 -3.58973175e-01
6.10709846e-01 1.51941895e+00 -7.46245205e-01 -2.51056701e-01
-1.43874303e-01 4.45985585e-01 3.70068587e-02 -9.24919359e-03
-9.39700425e-01 -6.31654143e-01 3.73942763e-01 3.48589718e-01
6.69737995e-01 -2.44065672e-02 -2.31078155e-02 -1.01580107e+00
-1.55531600e-01 -8.41891289e-01 4.62405264e-01 -6.93868220e-01
-1.29549849e+00 8.88808489e-01 -1.88305408e-01 -1.32786775e+00
-6.97179064e-02 -5.01503527e-01 1.16279326e-01 5.65237463e-01
-1.33959627e+00 -1.05853868e+00 -1.57041758e-01 5.06758451e-01
4.19844478e-01 -4.63146746e-01 7.60181546e-01 7.52767861e-01
-7.80333936e-01 9.35760200e-01 2.20988005e-01 1.69208601e-01
5.07104576e-01 -9.58378434e-01 2.40053579e-01 6.85451210e-01
4.93743688e-01 6.16344094e-01 5.87206244e-01 -7.09716678e-01
-1.36987019e+00 -9.21854794e-01 8.49945188e-01 -2.13840619e-01
8.31597328e-01 -3.63130867e-01 -5.82764566e-01 4.28172618e-01
1.15681188e-02 -4.32023346e-01 6.41633928e-01 2.84069479e-01
-2.25118577e-01 -7.49900758e-01 -1.01164484e+00 6.03155792e-01
8.88755739e-01 -2.55073130e-01 -2.69255221e-01 2.93853849e-01
6.82738841e-01 -6.74133971e-02 -1.19533157e+00 3.79024029e-01
6.41450644e-01 -6.26646817e-01 9.62366462e-01 -4.43305820e-01
2.36379474e-01 -4.89350036e-02 -1.45812884e-01 -8.85213196e-01
-6.82983935e-01 -9.04692709e-01 -2.34123453e-01 1.37892056e+00
9.57914293e-01 -5.41072726e-01 8.24763358e-01 2.47657448e-01
-6.52495492e-03 -9.29418564e-01 -7.98431933e-01 -7.28886127e-01
-2.88202792e-01 -5.17032146e-01 7.14556336e-01 9.72304523e-01
1.88516062e-02 5.97721577e-01 -5.75846195e-01 -3.50361675e-01
5.69102228e-01 1.49392441e-01 3.26733083e-01 -1.48975122e+00
-1.96897741e-02 -5.63366473e-01 -4.30889279e-02 -1.06363010e+00
-4.63359416e-01 -9.37708616e-01 -2.34865144e-01 -1.09791207e+00
2.03025684e-01 -8.07021320e-01 -6.55618787e-01 6.80958331e-01
-1.62478946e-02 5.05016446e-01 2.35467672e-01 3.05846363e-01
-8.18983078e-01 4.31086302e-01 1.29381859e+00 -1.08675592e-01
-7.23037943e-02 2.36212522e-01 -1.02364409e+00 5.84112585e-01
7.62309968e-01 -5.86329162e-01 -3.62489879e-01 -2.66052604e-01
8.62179458e-01 -2.39095286e-01 -1.27853990e-01 -8.08741629e-01
1.29929632e-01 -5.51679656e-02 4.64424849e-01 -6.58157468e-01
3.90122086e-01 -9.08782959e-01 3.28006893e-01 9.26835164e-02
-3.64161700e-01 2.21794084e-01 4.07819860e-02 3.71143341e-01
-9.42116752e-02 -4.42541748e-01 5.69784820e-01 -1.60935462e-01
-5.01108587e-01 3.59873444e-01 -3.16773951e-01 -1.19244017e-01
8.86093676e-01 -2.51129746e-01 -2.16331109e-01 -4.31342036e-01
-5.73589087e-01 3.71397763e-01 -2.74002925e-02 7.41129220e-01
2.93006957e-01 -1.38551712e+00 -3.88046801e-01 9.89254564e-02
2.78885593e-03 -2.13015750e-01 3.43157321e-01 8.64048898e-01
-4.63298745e-02 4.96185362e-01 -1.29338149e-02 -2.73456335e-01
-1.02882135e+00 4.99420166e-01 -1.17010199e-01 -6.78455174e-01
-4.78984475e-01 1.12751877e+00 -1.12360261e-01 -1.08018599e-01
3.72399449e-01 -1.08750006e-02 -5.39316416e-01 4.50557083e-01
6.04508042e-01 4.65229988e-01 2.20707431e-01 -3.51396680e-01
-2.98631907e-01 2.83318549e-01 -4.14204568e-01 1.38202965e-01
1.69423103e+00 -2.12118521e-01 1.72307529e-02 2.04585027e-02
9.30223525e-01 1.89565629e-01 -9.05026436e-01 -4.05269742e-01
7.11704642e-02 -2.98628926e-01 3.70952517e-01 -1.24140489e+00
-1.35011029e+00 4.93759900e-01 6.27868235e-01 6.31084740e-01
1.18598199e+00 3.01474184e-02 1.09034395e+00 3.12143445e-01
2.65085131e-01 -1.68807483e+00 2.18624145e-01 3.47072959e-01
4.77905482e-01 -8.86088431e-01 3.36589843e-01 -1.61097094e-01
-9.77641940e-01 7.43899345e-01 4.54531550e-01 1.70746431e-01
5.95190823e-01 3.59478325e-01 -4.37385887e-02 -4.36747491e-01
-4.65681016e-01 -2.69670755e-01 6.78520679e-01 2.84124851e-01
2.12838098e-01 -1.81743205e-02 -5.61126292e-01 1.10589874e+00
-4.38178033e-01 -5.47466278e-02 -3.19832154e-02 5.74410856e-01
-5.71208477e-01 -1.18744445e+00 3.09901237e-02 9.75062072e-01
-8.66420269e-01 -3.54701906e-01 -1.42761871e-01 7.46434331e-01
2.24554658e-01 8.81090462e-01 -2.03669399e-01 -9.04156029e-01
2.14261085e-01 1.49778172e-01 6.67771548e-02 -2.63228238e-01
-9.38787162e-01 4.30080146e-01 2.21356869e-01 -4.05022383e-01
-2.80056149e-01 -6.22026384e-01 -1.07075286e+00 -3.76059890e-01
-6.96932018e-01 7.25902736e-01 9.05325353e-01 7.19615817e-01
3.99008006e-01 5.10625124e-01 8.69052052e-01 -2.52713799e-01
-6.13753617e-01 -1.05609679e+00 -7.96381176e-01 6.58616945e-02
-2.99104154e-01 -7.57910490e-01 -1.15567697e-02 -1.18071578e-01] | [10.054065704345703, 5.684659957885742] |
0627e47e-09f5-42be-a672-d372b7056d66 | multiscale-fields-of-patterns | 1406.0924 | null | http://arxiv.org/abs/1406.0924v3 | http://arxiv.org/pdf/1406.0924v3.pdf | Multiscale Fields of Patterns | We describe a framework for defining high-order image models that can be used
in a variety of applications. The approach involves modeling local patterns in
a multiscale representation of an image. Local properties of a coarsened image
reflect non-local properties of the original image. In the case of binary
images local properties are defined by the binary patterns observed over small
neighborhoods around each pixel. With the multiscale representation we capture
the frequency of patterns observed at different scales of resolution. This
framework leads to expressive priors that depend on a relatively small number
of parameters. For inference and learning we use an MCMC method for block
sampling with very large blocks. We evaluate the approach with two example
applications. One involves contour detection. The other involves binary
segmentation. | ['Pedro F. Felzenszwalb', 'John G. Oberlin'] | 2014-06-04 | multiscale-fields-of-patterns-1 | http://papers.nips.cc/paper/5283-multiscale-fields-of-patterns | http://papers.nips.cc/paper/5283-multiscale-fields-of-patterns.pdf | neurips-2014-12 | ['contour-detection'] | ['computer-vision'] | [ 6.15804434e-01 -9.39819813e-02 -3.59475702e-01 -3.38955790e-01
-5.48648715e-01 -2.84017712e-01 8.42201054e-01 3.58835489e-01
-4.82099146e-01 6.51470065e-01 -5.77124916e-02 1.45203099e-01
-1.75004795e-01 -1.08213937e+00 -5.57141483e-01 -9.34615791e-01
-2.04365849e-02 4.28367794e-01 1.00479877e+00 3.04323249e-02
3.28679651e-01 7.61071324e-01 -1.83750784e+00 9.50945258e-01
2.90576279e-01 8.02349091e-01 4.50625777e-01 1.10198450e+00
-8.70720891e-04 8.12769592e-01 -4.14702892e-01 1.79744333e-01
-2.02903003e-02 -4.20675218e-01 -8.32678974e-01 7.05567837e-01
4.89689946e-01 1.95097744e-01 2.39367813e-01 1.38156652e+00
-5.02620451e-02 9.44252908e-02 9.85380292e-01 -5.86277843e-01
-4.64501753e-02 3.41415465e-01 -7.08396673e-01 4.08299029e-01
2.99942642e-01 -1.53000832e-01 9.47053671e-01 -7.49980927e-01
8.20431352e-01 1.52610672e+00 7.57884562e-01 -1.11898687e-02
-2.07359433e+00 2.43155345e-01 -2.12904871e-01 8.69096369e-02
-1.38253808e+00 -2.04468802e-01 6.67662561e-01 -6.53409064e-01
5.71781814e-01 4.48500663e-01 5.65963626e-01 5.93644202e-01
5.16305447e-01 4.89142537e-01 1.61921740e+00 -6.86968446e-01
4.50497180e-01 1.67850375e-01 2.87047356e-01 6.64790809e-01
4.13248092e-02 5.55640161e-02 -1.69363618e-01 -4.40709025e-01
1.39207375e+00 -2.62940191e-02 8.86704326e-02 -3.71220380e-01
-1.16086709e+00 9.19433773e-01 1.98218852e-01 8.00663352e-01
-5.02603769e-01 1.36500522e-01 2.47269589e-02 1.84331909e-01
5.68727136e-01 -1.35044366e-01 -1.89492628e-01 1.46615535e-01
-1.30985355e+00 1.69559404e-01 6.83981001e-01 4.82920319e-01
1.31630790e+00 -3.70656371e-01 1.05365232e-01 9.90043581e-01
1.10755689e-01 2.60753751e-01 2.56649226e-01 -1.18122685e+00
-3.29577565e-01 2.83349067e-01 1.91782892e-01 -1.17690969e+00
-2.75646716e-01 -6.11106344e-02 -9.87170756e-01 7.92312801e-01
5.66470504e-01 4.02732134e-01 -1.14648175e+00 1.43331015e+00
3.69718522e-01 3.75102311e-02 -3.78139317e-01 1.90717146e-01
2.11178258e-01 8.24329495e-01 6.37933761e-02 -4.07810450e-01
1.49325645e+00 -4.32441801e-01 -6.68010175e-01 1.03332289e-01
-8.08334798e-02 -9.67236280e-01 1.00169051e+00 4.23724622e-01
-1.25976169e+00 -8.67106020e-01 -8.58652711e-01 1.85145751e-01
-5.09950459e-01 8.30815136e-02 6.37319759e-02 4.81959701e-01
-1.24102390e+00 8.98044288e-01 -9.49951053e-01 -4.36698854e-01
1.86653629e-01 2.23178059e-01 -1.12603858e-01 2.61441648e-01
-7.32447207e-01 7.69398153e-01 5.19928157e-01 -1.80389404e-01
-6.11468077e-01 -4.33871418e-01 -7.87404299e-01 9.10282135e-04
-1.49942726e-01 -4.58683044e-01 1.04935420e+00 -1.06503010e+00
-1.15048754e+00 1.28805339e+00 -5.28741360e-01 -2.95825690e-01
6.00048304e-01 3.39096904e-01 -2.14993373e-01 5.07013917e-01
2.40501717e-01 4.90992397e-01 1.31838071e+00 -1.56442201e+00
-8.16070855e-01 -1.38327777e-01 -1.61563948e-01 -1.52059674e-01
2.41227239e-01 -3.41378674e-02 -2.01093704e-01 -9.18143570e-01
4.14504290e-01 -6.40897870e-01 -3.89139473e-01 -9.04890522e-02
-5.26043102e-02 -3.90585847e-02 7.43132710e-01 -3.04065734e-01
1.10433316e+00 -2.24213028e+00 6.65126601e-03 5.47786713e-01
6.57921238e-03 -3.14741433e-01 1.16990149e-01 2.93252170e-01
-9.12138298e-02 2.31447414e-01 -5.40035248e-01 -1.18271731e-01
-1.54177129e-01 5.11254132e-01 -2.23512813e-01 5.61866164e-01
1.50484592e-01 5.53267300e-01 -6.10356212e-01 -9.72492158e-01
4.83628660e-01 4.31587994e-01 -3.28635752e-01 -1.16971515e-01
-2.61243880e-01 3.55978191e-01 -4.32080895e-01 2.79544860e-01
7.05510139e-01 -5.92683494e-01 8.20114762e-02 -2.12074101e-01
-3.77754360e-01 -1.19314581e-01 -1.65719962e+00 1.18839872e+00
-2.68168479e-01 5.13269246e-01 3.06994408e-01 -9.06931341e-01
7.08530247e-01 3.02925199e-01 3.18217069e-01 -1.02773540e-01
-7.35273957e-02 3.80838439e-02 -1.88100025e-01 -3.55776519e-01
3.08781415e-01 -6.20973587e-01 1.89731747e-01 4.14613396e-01
-1.56158745e-01 -2.95051873e-01 4.03860569e-01 -1.55950755e-01
9.20625865e-01 -1.66218221e-01 7.47702539e-01 -7.58236051e-01
7.04475999e-01 1.32623747e-01 3.93083155e-01 1.00251710e+00
6.04608096e-02 6.92210734e-01 4.54380900e-01 -5.96312344e-01
-1.19710767e+00 -1.17281318e+00 -8.01948488e-01 1.05517399e+00
7.78814331e-02 -6.96658567e-02 -9.04118776e-01 -1.13138340e-01
-2.52866328e-01 1.36175692e-01 -1.01235664e+00 4.79508132e-01
-6.91620171e-01 -9.18669403e-01 1.58761740e-02 2.88278639e-01
5.26277065e-01 -1.28483927e+00 -8.84436488e-01 2.46696293e-01
-9.89620686e-02 -1.08703578e+00 -2.16228768e-01 8.25411230e-02
-1.28452063e+00 -7.67463863e-01 -6.10999584e-01 -9.24906313e-01
7.51432717e-01 -6.30596578e-02 1.28093553e+00 7.81436861e-02
-5.24763048e-01 4.54796672e-01 -8.46502110e-02 2.42469221e-01
-8.68204057e-01 -2.54092216e-01 -2.68807739e-01 4.42157984e-01
-3.39111909e-02 -6.58147395e-01 -3.34747761e-01 3.30638379e-01
-1.24434590e+00 -6.85554072e-02 7.76454151e-01 8.10360432e-01
1.13727486e+00 4.79683191e-01 -2.49207482e-01 -1.10161245e+00
4.04487759e-01 -2.14840308e-01 -6.23840034e-01 3.07041764e-01
-6.25295565e-02 1.47951588e-01 3.98360521e-01 -7.23691344e-01
-1.08064961e+00 1.60482243e-01 1.38882667e-01 -1.02598988e-01
-8.61846387e-01 3.99059474e-01 -2.49536410e-02 -1.03413954e-01
6.59259737e-01 3.09690773e-01 -2.34954000e-01 -5.79242706e-01
2.25273699e-01 2.27986321e-01 6.19540036e-01 -6.97355211e-01
3.76538724e-01 9.28318381e-01 4.23020899e-01 -1.45220983e+00
-4.41476017e-01 -5.12682259e-01 -9.48535144e-01 -6.80577457e-02
1.03593719e+00 -5.31466782e-01 -4.03626263e-01 3.31976801e-01
-1.07623434e+00 -3.94311309e-01 -7.09842205e-01 2.32894018e-01
-9.92209733e-01 4.11493868e-01 -8.25279176e-01 -8.03113580e-01
5.11312604e-01 -1.06569004e+00 1.12644970e+00 6.38537407e-02
-4.30266351e-01 -1.50984788e+00 3.67651850e-01 -1.76457852e-01
2.75711268e-01 4.88546163e-01 1.18671596e+00 -1.30036011e-01
-6.25457764e-01 -9.40078571e-02 -1.18493475e-01 1.06666073e-01
4.06864375e-01 1.50110260e-01 -9.96231079e-01 -2.39132009e-02
4.49370772e-01 -8.32716003e-03 8.74371767e-01 9.57746565e-01
9.20518517e-01 -2.58905798e-01 -3.76617730e-01 2.38887295e-01
1.78469372e+00 4.75388952e-03 6.04389906e-01 1.92395955e-01
1.94447532e-01 6.45956457e-01 5.36928773e-01 3.45675439e-01
-3.66005152e-01 6.51575744e-01 2.35768408e-02 -2.19696566e-01
-1.52673498e-01 -5.41246310e-02 1.47815607e-02 4.57873404e-01
-2.33566076e-01 3.84357452e-01 -8.89971793e-01 7.60374129e-01
-1.79296851e+00 -1.23230207e+00 -2.35500097e-01 2.10854125e+00
9.69101548e-01 3.13892871e-01 1.74697861e-01 1.70294747e-01
1.08137596e+00 3.56376559e-01 -1.56298339e-01 -1.88369066e-01
-2.42358983e-01 3.06579888e-01 2.07405567e-01 1.23148429e+00
-1.28989410e+00 6.25457168e-01 7.94654226e+00 1.23314190e+00
-7.91679859e-01 -3.56451981e-02 1.00875795e+00 5.31238258e-01
-1.55186728e-01 1.04328394e-01 -8.02750289e-01 4.19801772e-01
8.62077475e-01 2.60960281e-01 1.41049281e-01 5.54441810e-01
2.47742742e-01 -6.51029646e-01 -9.67737913e-01 7.05494463e-01
-4.08244193e-01 -1.37463295e+00 2.35979334e-01 2.16180608e-01
8.78728330e-01 -1.87111914e-01 1.12457350e-01 -5.64879358e-01
3.94835085e-01 -1.13420165e+00 6.36991203e-01 8.34225059e-01
6.24299645e-01 -5.54601014e-01 4.28451180e-01 5.29605150e-01
-1.32469332e+00 8.58293995e-02 -5.99861383e-01 -4.30042110e-02
9.47960168e-02 8.70121777e-01 -4.31603402e-01 -3.09156179e-02
7.02380538e-01 6.24805033e-01 -3.87577921e-01 8.58626604e-01
-6.89789886e-03 4.77171808e-01 -5.96351027e-01 1.53327391e-01
2.11680740e-01 -5.69578826e-01 5.68079710e-01 1.61764383e+00
4.76983897e-02 8.23016167e-02 2.64880031e-01 1.09192741e+00
3.55071336e-01 -3.35883461e-02 -5.90395033e-01 6.24308586e-01
1.90244570e-01 1.29116476e+00 -1.40006423e+00 -6.10300124e-01
-3.39523405e-01 7.68311679e-01 1.31885156e-01 4.32710499e-01
-3.98361564e-01 -5.06597124e-02 3.08439761e-01 3.20146918e-01
6.88584805e-01 -2.41413891e-01 -4.30002391e-01 -9.15214479e-01
-2.28968218e-01 -8.95861149e-01 4.20239300e-01 -5.33298016e-01
-1.20349145e+00 5.50916791e-01 5.10571241e-01 -1.07130480e+00
-2.96629637e-01 -7.09537387e-01 -7.05965281e-01 8.01310003e-01
-1.20828462e+00 -1.03493226e+00 9.47695896e-02 5.89287281e-01
6.51839554e-01 3.06930721e-01 8.35326314e-01 -3.04025561e-01
3.88328801e-03 -3.50133687e-01 1.74984634e-01 -2.01312397e-02
1.01057030e-01 -1.59829330e+00 7.79018775e-02 8.05561781e-01
3.92916203e-01 6.89349830e-01 1.09261680e+00 -5.61430335e-01
-2.82705605e-01 -6.94677293e-01 9.98265386e-01 -3.75772148e-01
6.63959920e-01 -2.43391722e-01 -9.73346114e-01 5.64146578e-01
2.85748631e-01 2.91443855e-01 5.09041011e-01 -2.62098670e-01
4.42783833e-02 7.63531104e-02 -1.27068627e+00 4.94936675e-01
4.38536823e-01 -6.74165606e-01 -5.92872858e-01 9.01291221e-02
-1.79993808e-01 1.63715497e-01 -1.15225351e+00 5.21589577e-01
5.16257048e-01 -1.34545386e+00 1.06477141e+00 -1.57324389e-01
1.56402349e-01 -4.34875786e-01 -2.80034810e-01 -9.19545710e-01
-7.29782283e-01 -4.92966741e-01 2.06958354e-01 9.71944571e-01
1.55438930e-01 -3.57946783e-01 7.72726297e-01 3.55505884e-01
7.11631298e-01 -5.47290504e-01 -1.04789376e+00 -5.74009478e-01
8.94480199e-02 -1.37428701e-01 1.66712284e-01 5.18766284e-01
-8.91555175e-02 1.51606038e-01 -6.86532333e-02 3.55785310e-01
1.03097105e+00 4.80795503e-01 2.56996214e-01 -1.39413893e+00
-5.06932378e-01 -6.13027573e-01 -5.60390472e-01 -1.14340734e+00
-2.08433583e-01 -3.49198520e-01 2.07214698e-01 -1.03911054e+00
5.32797039e-01 -2.11792991e-01 -1.13358878e-01 2.31891591e-02
-7.82070905e-02 5.87783039e-01 -1.84862182e-01 4.14063215e-01
-2.72696763e-01 -7.10271969e-02 1.11214995e+00 -2.96023171e-02
1.51728079e-01 1.82840407e-01 1.43221915e-01 1.27025783e+00
7.62272537e-01 -6.51999354e-01 -1.25036463e-01 1.57293662e-01
1.69468746e-02 4.55168597e-02 5.13049066e-01 -8.95386279e-01
2.02605739e-01 -3.47936571e-01 6.34338260e-01 -5.78391254e-01
4.16392863e-01 -8.82338464e-01 3.58762473e-01 5.11440694e-01
-4.00222093e-01 4.67351750e-02 7.49165267e-02 5.82756877e-01
-5.36093593e-01 -6.95640624e-01 1.53438723e+00 -5.80833793e-01
-6.29037559e-01 -7.18208216e-03 -8.59402478e-01 -2.39173219e-01
8.28064263e-01 -4.31041509e-01 1.80047140e-01 -4.03624445e-01
-1.51036096e+00 -4.94374067e-01 8.44288647e-01 -2.69685805e-01
6.40997231e-01 -1.25346410e+00 -5.51104188e-01 5.22245526e-01
-3.09477866e-01 -2.59754568e-01 2.06676475e-03 7.11327255e-01
-7.71412313e-01 1.36276394e-01 -2.50057757e-01 -1.02423120e+00
-1.66486514e+00 3.75734597e-01 3.66966993e-01 -5.44727266e-01
-5.55205464e-01 3.74635696e-01 5.30823112e-01 7.39979222e-02
-1.49454162e-01 -6.21678174e-01 -3.47338736e-01 6.70622811e-02
5.97802639e-01 6.03011310e-01 -4.27970171e-01 -9.11299527e-01
-8.86556134e-02 1.17848730e+00 1.43205549e-03 -5.83498478e-01
1.17711079e+00 -3.10308129e-01 -6.67815030e-01 1.08720720e+00
1.10145772e+00 -1.63033057e-03 -1.47584975e+00 -2.60424674e-01
3.77411962e-01 -5.13849676e-01 1.18919881e-02 -2.72358239e-01
-6.44365191e-01 8.96730363e-01 7.46185899e-01 8.28538358e-01
1.09538472e+00 2.34596640e-01 -4.94604446e-02 9.40095633e-03
3.68228495e-01 -1.09048879e+00 2.30124630e-02 3.81045580e-01
7.23795056e-01 -1.00643110e+00 2.49853715e-01 -6.44646049e-01
-2.84641594e-01 1.32462978e+00 -2.69689232e-01 -5.50378680e-01
1.08808994e+00 4.41613376e-01 1.63690634e-02 -2.37606511e-01
-6.21745944e-01 -3.21895182e-01 3.90306622e-01 4.45494771e-01
3.28573823e-01 2.50009820e-02 -2.51268655e-01 -1.27823636e-01
1.86920002e-01 -2.04385072e-01 4.53459173e-01 1.00685692e+00
-8.69122207e-01 -1.15543664e+00 -7.50332057e-01 1.66875809e-01
-7.88436353e-01 -4.87800390e-02 -2.15359136e-01 8.21331859e-01
3.27224642e-01 8.35285604e-01 3.08344066e-01 2.14116693e-01
8.69271532e-02 6.21359833e-02 8.46258581e-01 -6.48634851e-01
-1.00142226e-01 7.85506070e-01 -2.59550571e-01 -5.47264457e-01
-9.72937942e-01 -1.03943396e+00 -9.11576033e-01 -3.60869281e-02
5.35152368e-02 7.00864568e-02 1.83561549e-01 6.58550978e-01
-2.13870794e-01 2.37303495e-01 4.31820989e-01 -1.11897039e+00
-2.55825341e-01 -1.11737287e+00 -1.18707335e+00 5.46332955e-01
6.87081039e-01 -5.14167130e-01 -3.67959380e-01 7.66588449e-01] | [11.280797004699707, -2.41402006149292] |
ea5de8df-cabb-4c11-ae06-2ae23614769f | ncsu_sas_sam-deep-encoding-and-reconstruction | null | null | https://aclanthology.org/W15-4323 | https://aclanthology.org/W15-4323.pdf | NCSU\_SAS\_SAM: Deep Encoding and Reconstruction for Normalization of Noisy Text | null | ['Samuel Leeman-Munk', 'James Lester', 'James Cox'] | 2015-07-01 | null | null | null | ws-2015-7 | ['lexical-normalization'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.312983512878418, 3.8664746284484863] |
b2bfb07b-9ae0-458f-8129-6e921af4f067 | multilingual-epidemiological-text | null | null | https://aclanthology.org/2020.coling-main.543 | https://aclanthology.org/2020.coling-main.543.pdf | Multilingual Epidemiological Text Classification: A Comparative Study | In this paper, we approach the multilingual text classification task in the context of the epidemiological field. Multilingual text classification models tend to perform differently across different languages (low- or high-resourced), more particularly when the dataset is highly imbalanced, which is the case for epidemiological datasets. We conduct a comparative study of different machine and deep learning text classification models using a dataset comprising news articles related to epidemic outbreaks from six languages, four low-resourced and two high-resourced, in order to analyze the influence of the nature of the language, the structure of the document, and the size of the data. Our findings indicate that the performance of the models based on fine-tuned language models exceeds by more than 50{\%} the chosen baseline models that include a specialized epidemiological news surveillance system and several machine learning models. Also, low-resource languages are highly influenced not only by the typology of the languages on which the models have been pre-trained or/and fine-tuned but also by their size. Furthermore, we discover that the beginning and the end of documents provide the most salient features for this task and, as expected, the performance of the models was proportionate to the training data size. | ['Moses Odeo', 'Ga{\\"e}l Lejeune', 'Adam Jatowt', 'Antoine Doucet', 'Emanuela Boros', 'Stephen Mutuvi'] | 2020-12-01 | null | null | null | coling-2020-8 | ['multilingual-text-classification'] | ['miscellaneous'] | [-4.05193388e-01 -1.35151535e-01 -5.03256142e-01 -3.73882912e-02
-3.18512648e-01 -5.18844426e-01 1.04191494e+00 8.08708429e-01
-8.85194182e-01 5.27605772e-01 8.51802289e-01 -5.16252100e-01
-1.37471735e-01 -6.86167777e-01 -5.47074258e-01 -4.12798107e-01
7.58513734e-02 8.16792309e-01 -1.01954989e-01 -4.64909196e-01
2.79191099e-02 3.60833555e-02 -1.12962806e+00 3.86616707e-01
7.20177650e-01 5.24286926e-01 -3.87909636e-02 4.01363254e-01
-4.39098537e-01 7.66211331e-01 -7.87156701e-01 -3.57293367e-01
-1.24288291e-01 -2.04753011e-01 -6.64239645e-01 -1.23315446e-01
1.71012998e-01 1.43683448e-01 -2.23013580e-01 6.25118911e-01
5.50739825e-01 -6.14577532e-01 9.94996011e-01 -7.55352974e-01
-4.49293762e-01 8.38836551e-01 -6.74335361e-01 5.35938919e-01
3.36195767e-01 -9.73230898e-02 9.38578248e-01 -4.78989989e-01
1.14335608e+00 1.25931704e+00 8.81772459e-01 -1.60364255e-01
-8.65832806e-01 -5.54868519e-01 3.48601073e-01 -2.64761448e-01
-1.04403424e+00 -4.70347703e-01 5.14105260e-01 -1.15319395e+00
8.45673561e-01 -1.62805617e-01 4.08614993e-01 1.43622303e+00
7.26997614e-01 1.08563140e-01 1.15626252e+00 -5.04503846e-01
-7.99885541e-02 6.11879408e-01 2.49283448e-01 5.64547718e-01
5.09342670e-01 -2.45397121e-01 -3.80001217e-01 -4.35387731e-01
5.48568033e-02 -1.83480784e-01 -2.21771568e-01 -2.31635012e-02
-1.42241454e+00 1.19879758e+00 8.31533298e-02 9.36517179e-01
-4.41890568e-01 -6.00033402e-01 9.63391960e-01 5.49076736e-01
1.14700723e+00 5.23149848e-01 -7.73482084e-01 1.46250680e-01
-9.46535826e-01 1.36389032e-01 9.60067332e-01 1.57255441e-01
2.04857826e-01 -1.28752097e-01 -2.99020976e-01 9.62345779e-01
1.67741761e-01 4.71053898e-01 7.44017720e-01 -5.23119420e-02
8.77131462e-01 6.57067239e-01 -2.15290859e-01 -1.16289127e+00
-9.45280612e-01 -6.90308928e-01 -9.46712971e-01 -4.27553952e-01
4.55290496e-01 -4.72827852e-01 -5.52290320e-01 1.61274660e+00
1.29610091e-01 -8.38951051e-01 1.26035780e-01 4.28678155e-01
9.84639883e-01 7.12753057e-01 2.52342850e-01 -2.96517909e-01
1.48506057e+00 -6.64453566e-01 -5.89935303e-01 -2.93406755e-01
9.56381857e-01 -8.23267519e-01 9.71443832e-01 1.23343699e-01
-3.74429673e-01 -3.34395796e-01 -7.51018524e-01 1.28995955e-01
-8.52861345e-01 2.16277897e-01 2.22179383e-01 5.13514340e-01
-7.92480052e-01 5.23584709e-02 -2.88962901e-01 -6.93871558e-01
-9.00293048e-03 -1.97987199e-01 -5.36171556e-01 1.22529499e-01
-1.33178484e+00 1.25386620e+00 5.46788573e-01 -4.16060746e-01
-2.76547253e-01 -5.88178277e-01 -5.96178174e-01 -1.50497675e-01
-4.89866249e-02 -4.56421107e-01 4.75116640e-01 -9.21191275e-01
-7.57893860e-01 1.17590845e+00 2.36410946e-01 -2.98316032e-01
7.93376982e-01 1.07324548e-01 -5.54017603e-01 -1.24454834e-01
4.24569517e-01 1.10108927e-02 4.43462163e-01 -8.32526863e-01
-5.65121055e-01 -6.24693453e-01 -1.22019432e-01 1.98989548e-02
-4.55692947e-01 5.27725160e-01 -2.21444219e-01 -1.00393152e+00
-5.19980431e-01 -7.45296299e-01 4.17084768e-02 -4.88423169e-01
-2.26376876e-01 -2.42360309e-01 5.18256187e-01 -9.54686105e-01
1.34794259e+00 -2.06121850e+00 2.40773574e-01 1.11777522e-01
1.98563606e-01 -8.59694332e-02 -1.11122847e-01 7.08480775e-01
-2.78362613e-02 3.07161331e-01 1.37282267e-01 3.16026025e-02
-1.47238821e-01 -1.33176344e-02 -1.75490335e-01 6.40274584e-01
2.14925818e-02 3.54430586e-01 -5.69307268e-01 -4.59099501e-01
-2.09215641e-01 5.96401095e-01 -4.96425807e-01 -2.29543298e-01
-1.37132466e-01 4.64864641e-01 -3.48315686e-01 7.67983645e-02
1.33402139e-01 -3.02374274e-01 4.67357934e-01 -2.01807283e-02
-3.83553118e-01 6.13812923e-01 -6.61492646e-01 9.70679879e-01
-7.55850673e-01 8.97736847e-01 -3.38320658e-02 -7.40338683e-01
6.54634297e-01 4.54819709e-01 5.13582110e-01 -8.83231699e-01
9.04998928e-02 2.31953725e-01 2.74480492e-01 -6.22048080e-01
3.75937611e-01 1.09909259e-01 -2.31322780e-01 7.38345325e-01
-7.77105838e-02 2.98521429e-01 4.77139920e-01 1.86377317e-01
7.39671826e-01 -3.46648067e-01 4.83110100e-01 -5.98233998e-01
3.25166225e-01 9.33233500e-02 3.82647187e-01 5.71674287e-01
-8.19191861e-04 2.09747940e-01 8.66577446e-01 -5.48584878e-01
-1.17527509e+00 -3.16492170e-01 -7.44558454e-01 1.43297732e+00
-5.49791336e-01 -3.92036498e-01 -5.45405626e-01 -5.84056437e-01
3.73686850e-02 4.93781209e-01 -7.80316293e-01 -1.17669731e-01
-2.99336374e-01 -1.39406610e+00 5.88605404e-01 -6.25253096e-02
1.33927837e-01 -9.90714133e-01 -5.38537800e-01 1.00130990e-01
-3.71125430e-01 -1.23075378e+00 -2.04862222e-01 2.32429713e-01
-3.65057707e-01 -1.03866053e+00 -6.38630927e-01 -5.68954468e-01
3.78290564e-01 -3.67706358e-01 1.30568290e+00 -1.72221825e-01
2.10426435e-01 1.17486060e-01 -6.00420475e-01 -6.88665211e-01
-8.48011255e-01 7.82957494e-01 1.01396376e-02 -8.77510011e-02
3.26662451e-01 8.60586017e-02 -1.02124430e-01 -7.67232180e-02
-8.69977236e-01 -7.52606392e-02 4.61313695e-01 5.81723213e-01
-2.29730532e-02 2.37688646e-01 6.59439027e-01 -1.11405563e+00
9.22724128e-01 -9.47182596e-01 -2.13732228e-01 3.98066074e-01
-4.94863182e-01 1.95449367e-01 6.12296164e-01 -3.48834187e-01
-5.43606043e-01 -7.30199039e-01 -7.17035010e-02 4.25218523e-01
8.66345689e-03 1.02337146e+00 3.30699146e-01 4.11666930e-01
7.59190798e-01 -5.35502061e-02 -1.47606209e-01 -5.14370322e-01
-1.58686996e-01 9.82654333e-01 -3.37854773e-01 -3.10268790e-01
4.43205088e-01 2.27489248e-01 -5.81843555e-01 -9.71421599e-01
-6.64680839e-01 -2.55092144e-01 -5.67765832e-01 -4.19237912e-02
8.93647492e-01 -9.69192743e-01 -1.12683937e-01 5.47053456e-01
-1.02575111e+00 -3.53531986e-01 1.60650507e-01 6.58351779e-01
1.39742330e-01 -7.88469836e-02 -6.55564249e-01 -4.31137830e-01
-5.15960872e-01 -1.15660465e+00 9.09719348e-01 -4.98485327e-01
-3.83296490e-01 -1.31313062e+00 3.58097047e-01 3.71627986e-01
5.74353814e-01 4.28468883e-01 1.56369495e+00 -1.00385308e+00
4.36916798e-01 -1.68309405e-01 -2.19037861e-01 -1.63160682e-01
1.12444237e-01 2.59048879e-01 -6.57120585e-01 -4.03181076e-01
-1.80218630e-02 -1.99577674e-01 9.98598993e-01 3.54815125e-01
5.66155732e-01 -5.78913331e-01 -2.49484301e-01 1.62686542e-01
1.17821085e+00 -2.50862479e-01 -9.66259316e-02 6.32821500e-01
6.26017153e-01 9.18183088e-01 1.21166721e-01 3.62383604e-01
7.36824989e-01 7.91458130e-01 -1.84413835e-01 -1.69644326e-01
1.38787597e-01 -7.49183968e-02 5.04732132e-01 1.09958243e+00
2.03883782e-01 -2.11775720e-01 -1.39720166e+00 4.76383924e-01
-1.45714867e+00 -6.77709401e-01 -1.43369228e-01 1.95170081e+00
1.03327346e+00 2.48138860e-01 3.93314958e-01 8.53011459e-02
3.88530284e-01 5.03581941e-01 -1.14548700e-02 -6.32242858e-01
-4.17774588e-01 -2.08164543e-01 5.44160426e-01 3.46597522e-01
-1.06538665e+00 5.95815659e-01 6.53537512e+00 5.55067778e-01
-1.46169019e+00 2.84709215e-01 6.50753021e-01 -1.23949870e-01
-1.87241092e-01 -5.36222637e-01 -6.37162507e-01 7.64774680e-01
1.28652644e+00 -7.77240726e-04 1.00424573e-01 4.59668964e-01
3.46107453e-01 -6.83126897e-02 -8.11918020e-01 3.45329046e-01
1.82999060e-01 -1.10069799e+00 2.17948869e-01 2.02430546e-01
5.68763614e-01 6.69290602e-01 5.68901263e-02 2.78632283e-01
2.78248459e-01 -8.02027106e-01 9.80092287e-01 1.87657833e-01
7.36776829e-01 -4.16343838e-01 9.13354337e-01 4.69592363e-01
-6.36655867e-01 -1.12380236e-01 -9.53605250e-02 1.54330999e-01
-2.53738999e-01 9.35821056e-01 -6.06001139e-01 4.33343858e-01
9.03569698e-01 5.79766214e-01 -8.17725122e-01 4.01011437e-01
2.50409782e-01 6.38130009e-01 -1.78615242e-01 2.69395486e-02
2.83533782e-01 9.37219262e-02 3.74706894e-01 1.57947409e+00
8.15658718e-02 -5.16964972e-01 2.49138772e-01 2.50908941e-01
-1.24357693e-01 7.19646275e-01 -7.44441390e-01 -3.73536140e-01
8.81924778e-02 8.62326264e-01 -6.34602964e-01 -1.38905257e-01
-6.65409088e-01 2.93600231e-01 3.69654387e-01 3.24270755e-01
-4.54533607e-01 -1.44367725e-01 1.86392665e-01 4.06430274e-01
1.10402852e-01 -3.68868560e-02 -1.67288423e-01 -1.28945851e+00
-1.22488469e-01 -1.16208363e+00 6.41263962e-01 -4.02087510e-01
-1.36062443e+00 9.06986058e-01 -1.43175134e-02 -6.42400205e-01
-3.15307558e-01 -6.35438681e-01 -4.35892418e-02 7.30238438e-01
-1.42822731e+00 -1.17132401e+00 1.70540661e-01 4.03200269e-01
3.10450047e-01 -5.66434503e-01 6.77914679e-01 5.90298653e-01
-6.77573800e-01 3.18747818e-01 5.08761346e-01 6.35708630e-01
8.56827915e-01 -7.65799165e-01 5.53028136e-02 4.25998896e-01
7.60741020e-03 4.32864517e-01 6.82690561e-01 -7.75857687e-01
-7.48215497e-01 -9.60705400e-01 1.54719388e+00 -5.94962478e-01
9.45798516e-01 -4.81413931e-01 -5.77330589e-01 6.20711088e-01
3.90948981e-01 -5.78541934e-01 9.83057141e-01 5.83415031e-01
-7.16258883e-01 -5.10433391e-02 -9.99038875e-01 3.21083963e-01
4.69719291e-01 -6.38403594e-01 -4.94000137e-01 6.99049354e-01
5.15829682e-01 -5.96505776e-02 -9.80124116e-01 3.19007069e-01
5.07232904e-01 -7.54109561e-01 5.31586289e-01 -8.19577396e-01
7.15950787e-01 2.57670343e-01 1.79145746e-02 -1.42449808e+00
-3.58265072e-01 4.28391844e-01 3.55192631e-01 1.18591380e+00
7.45751023e-01 -9.26132441e-01 -9.79808047e-02 -1.59271434e-01
4.11496937e-01 -5.90838432e-01 -7.69829571e-01 -3.09855759e-01
5.92378080e-01 -2.07528263e-01 3.57875198e-01 1.38840890e+00
-7.38566890e-02 6.05911314e-01 -9.49425325e-02 -2.50301152e-01
1.17244937e-01 1.24288470e-01 4.38140124e-01 -1.48739660e+00
-1.99620388e-02 -6.58066809e-01 -5.98278753e-02 4.90504503e-02
3.65536988e-01 -1.02296126e+00 -4.68389362e-01 -1.39225459e+00
3.80826414e-01 -5.83116114e-01 -1.20654047e-01 3.75023752e-01
-1.20039426e-01 3.16963624e-03 -1.33291995e-02 3.66074651e-01
-2.50264347e-01 -2.66808700e-02 6.75292015e-01 -2.47466192e-01
-1.97572052e-01 -1.08493771e-02 -7.56414354e-01 6.60505593e-01
6.48234785e-01 -5.67998648e-01 -4.68249200e-03 -7.16956317e-01
8.22435021e-01 -2.44960681e-01 5.56747429e-02 -5.31961918e-01
-1.55852631e-01 -3.99714597e-02 3.71923953e-01 -2.31995851e-01
-3.16191196e-01 -7.87767470e-01 6.51041940e-02 6.56796515e-01
-6.45424247e-01 4.61007625e-01 1.50212407e-01 2.35748500e-01
-1.66629687e-01 -2.77440902e-03 6.08532906e-01 -1.98043391e-01
-4.44698036e-02 4.93861064e-02 -6.61376655e-01 4.80132252e-01
5.77642918e-01 2.64170527e-01 -6.58116460e-01 -2.77576983e-01
-3.79826605e-01 5.37891388e-02 3.47344965e-01 9.09437776e-01
-3.34498942e-01 -7.77159452e-01 -1.15221059e+00 8.22713450e-02
3.29201996e-01 -5.42814076e-01 -1.49484202e-01 9.86834645e-01
-6.10872149e-01 9.41526234e-01 -2.67442375e-01 -3.96780193e-01
-8.84719670e-01 5.15394330e-01 3.00703555e-01 -8.22079182e-01
-3.50962937e-01 4.72543836e-02 1.32626802e-01 -7.67946661e-01
2.44409904e-01 -2.81762660e-01 -6.45716429e-01 9.06984150e-01
1.76186830e-01 1.94991693e-01 2.77942121e-01 -1.08050394e+00
-3.84287477e-01 4.85043079e-01 -1.83147684e-01 2.71418015e-04
1.49478626e+00 -1.46473777e-02 -3.93800795e-01 8.23374689e-01
1.11851382e+00 2.80157685e-01 -2.84221977e-01 -2.99076378e-01
6.17881827e-02 2.43043929e-01 2.24182501e-01 -8.64380121e-01
-9.87280607e-01 5.19971728e-01 3.80462587e-01 3.05993080e-01
6.52166665e-01 1.99248586e-02 1.42372414e-01 -4.80861291e-02
2.13751271e-01 -1.14576674e+00 -2.62094885e-01 7.21459031e-01
9.34228480e-01 -1.15257835e+00 1.53079838e-01 1.67999700e-01
-5.37574589e-01 8.33538353e-01 1.79235712e-01 3.55431378e-01
9.13303375e-01 2.12724954e-01 2.24632546e-01 -2.94364303e-01
-7.50903368e-01 -1.67263336e-02 4.43045467e-01 3.71370614e-01
8.14290524e-01 7.45613351e-02 -7.18849301e-01 4.33201730e-01
-4.79971796e-01 -2.34123260e-01 2.83027202e-01 5.21881998e-01
-1.46706522e-01 -7.51804590e-01 -4.04410541e-01 6.35625482e-01
-9.89627361e-01 -2.04016328e-01 -6.34955585e-01 7.96514511e-01
3.52971643e-01 7.57468343e-01 3.57010335e-01 -1.12228014e-01
1.36913314e-01 2.08916545e-01 -4.22486365e-02 -5.62752187e-01
-1.11917818e+00 -1.65946990e-01 2.30832577e-01 1.62841335e-01
-5.13623118e-01 -5.98013401e-01 -6.82585835e-01 -4.22059208e-01
-8.49400461e-02 1.89191535e-01 7.57912755e-01 1.09583354e+00
4.10719693e-01 3.91099989e-01 4.29356128e-01 -3.12882692e-01
-2.33179450e-01 -1.24796009e+00 -3.04451644e-01 3.95951092e-01
3.97571683e-01 -4.55676109e-01 -3.84790003e-01 -1.74832150e-01] | [10.093439102172852, 9.920912742614746] |
d07e78a0-5298-4001-b795-42d10782012d | audio-captioning-using-pre-trained-large | 2012.07331 | null | https://arxiv.org/abs/2012.07331v1 | https://arxiv.org/pdf/2012.07331v1.pdf | Audio Captioning using Pre-Trained Large-Scale Language Model Guided by Audio-based Similar Caption Retrieval | The goal of audio captioning is to translate input audio into its description using natural language. One of the problems in audio captioning is the lack of training data due to the difficulty in collecting audio-caption pairs by crawling the web. In this study, to overcome this problem, we propose to use a pre-trained large-scale language model. Since an audio input cannot be directly inputted into such a language model, we utilize guidance captions retrieved from a training dataset based on similarities that may exist in different audio. Then, the caption of the audio input is generated by using a pre-trained language model while referring to the guidance captions. Experimental results show that (i) the proposed method has succeeded to use a pre-trained language model for audio captioning, and (ii) the oracle performance of the pre-trained model-based caption generator was clearly better than that of the conventional method trained from scratch. | ['Masahiro Yasuda', 'Daiki Takeuchi', 'Daisuke Niizumi', 'Yasunori Ohishi', 'Yuma Koizumi'] | 2020-12-14 | null | null | null | null | ['audio-captioning'] | ['audio'] | [ 6.06914163e-01 4.18316156e-01 4.21116799e-01 -1.96641058e-01
-1.66231167e+00 -6.03030264e-01 4.75145519e-01 1.22853518e-01
-7.79848844e-02 8.88341129e-01 5.06128788e-01 -1.07596204e-01
2.22575322e-01 -5.63172638e-01 -1.11023498e+00 -3.07418108e-01
1.69292703e-01 7.85506666e-01 1.39418349e-01 -1.95728093e-01
5.20117693e-02 -1.13758378e-01 -1.67850816e+00 7.08678246e-01
6.99197412e-01 9.15639997e-01 5.52559972e-01 8.02619040e-01
-4.61739808e-01 4.97845858e-01 -8.01065743e-01 -2.02352822e-01
1.29758045e-01 -1.12936330e+00 -9.70350504e-01 4.87610735e-02
1.45017698e-01 -1.39276549e-01 -1.07333045e-02 8.98553848e-01
4.97681439e-01 -4.89074625e-02 4.39198434e-01 -1.30213714e+00
-4.15755630e-01 9.89221513e-01 -7.38095045e-02 -1.89912885e-01
9.03898895e-01 -2.03365460e-01 1.22891533e+00 -6.58270001e-01
5.64820766e-01 9.84235883e-01 3.15395504e-01 7.97364950e-01
-1.19304705e+00 -4.94324714e-01 -6.17568269e-02 1.53900936e-01
-1.66493356e+00 -3.82642150e-01 9.20022488e-01 -4.37756360e-01
3.83476228e-01 2.92822331e-01 4.48747844e-01 1.21053243e+00
-3.52140367e-01 5.72148383e-01 7.22537160e-01 -9.12584186e-01
2.50892848e-01 5.11440516e-01 -2.71260113e-01 2.23919556e-01
-2.29283214e-01 -1.78740531e-01 -5.24932683e-01 -4.26306218e-01
5.88704944e-01 -6.89970672e-01 -3.06125492e-01 -5.36386259e-02
-1.34247112e+00 6.68231189e-01 7.40866214e-02 5.37141442e-01
-5.23602545e-01 -1.95246190e-02 4.51583326e-01 2.39987254e-01
2.21180156e-01 8.03433359e-01 -1.35649204e-01 -2.63705730e-01
-1.10804451e+00 1.37985900e-01 7.28853643e-01 9.96696770e-01
6.74825728e-01 -2.97560282e-02 -1.86396748e-01 9.01050091e-01
3.90110940e-01 7.01002061e-01 7.33524561e-01 -6.57994807e-01
9.67675924e-01 1.51301846e-01 3.05150598e-01 -8.75747919e-01
1.22057967e-01 -7.66472444e-02 -4.26320702e-01 -3.06102961e-01
3.34055185e-01 -3.32389474e-01 -7.23468304e-01 1.86129808e+00
-1.07061388e-02 3.50909144e-01 5.35788834e-01 8.66895139e-01
8.19755137e-01 1.32685709e+00 -3.56794298e-02 -3.41104835e-01
1.14581859e+00 -9.64191616e-01 -6.47026479e-01 -3.30761164e-01
4.34084535e-01 -1.07124186e+00 1.23080063e+00 1.91819966e-01
-9.39121962e-01 -8.13788056e-01 -9.11924183e-01 4.58132476e-01
1.39439121e-01 2.09076926e-01 1.81636989e-01 4.30968136e-01
-9.80548739e-01 -9.22046378e-02 -4.69436973e-01 -4.72840935e-01
-3.38229775e-01 -1.54141663e-02 -3.25711161e-01 -6.54133558e-02
-1.44204926e+00 2.59777129e-01 6.17003620e-01 -2.24201614e-03
-1.13138819e+00 -2.50459969e-01 -7.42459178e-01 1.54776737e-01
3.05299073e-01 -5.54019034e-01 1.65345407e+00 -1.37871003e+00
-1.57549453e+00 4.76426780e-01 -1.96788475e-01 -4.77413863e-01
3.00447255e-01 -4.84719053e-02 -5.46549499e-01 5.21391034e-01
3.11938167e-01 8.51208508e-01 8.59723270e-01 -1.53547645e+00
-5.55512965e-01 2.41645500e-01 4.51074392e-02 2.31837466e-01
-4.02752846e-01 1.09737240e-01 -5.25853574e-01 -4.47638929e-01
-1.13843828e-01 -1.00233853e+00 -2.71431245e-02 -4.94977742e-01
-4.16480601e-01 -2.57918146e-02 5.54120779e-01 -7.16882110e-01
1.20204222e+00 -2.16062188e+00 -2.23165527e-01 3.37117851e-01
-4.72467184e-01 4.04359251e-01 -5.81492484e-01 8.26246262e-01
-1.83491439e-01 2.72417933e-01 -2.17553824e-01 -2.08690226e-01
-8.93410742e-02 8.64008069e-02 -8.25497746e-01 -3.20424080e-01
1.86798647e-01 4.40010726e-01 -1.19748938e+00 -7.05955863e-01
-2.82027572e-01 4.60363805e-01 -5.88005543e-01 8.92184734e-01
-6.07982397e-01 6.51421428e-01 -5.57787478e-01 2.50178635e-01
2.00364724e-01 1.66281068e-03 7.56010935e-02 4.13514711e-02
1.97362885e-01 6.94858432e-01 -1.20377135e+00 1.90362298e+00
-8.24129164e-01 6.49982452e-01 -1.74327746e-01 -8.91907215e-01
1.20889008e+00 1.09440339e+00 4.57505673e-01 -3.58843446e-01
2.93559823e-02 4.69143718e-01 -4.62794341e-02 -7.62742102e-01
3.17681193e-01 -1.93434671e-01 -2.29560748e-01 8.28833640e-01
1.58960730e-01 -4.17020202e-01 2.96793878e-01 1.34407133e-01
9.89991486e-01 -8.61170515e-02 -1.21701442e-01 4.24395978e-01
7.86943197e-01 9.62438434e-02 1.72411695e-01 6.55090332e-01
2.57399350e-01 1.30155313e+00 3.47222328e-01 4.44739461e-02
-1.11569154e+00 -8.90460432e-01 2.57093996e-01 1.02187073e+00
-2.81449139e-01 -6.55341029e-01 -1.01234758e+00 -3.92638057e-01
-6.78723991e-01 7.28009403e-01 -2.80682921e-01 -2.75291987e-02
-4.53715086e-01 -8.24549347e-02 7.09525228e-01 3.18536311e-01
3.48108232e-01 -1.32247686e+00 -1.42311350e-01 5.48738897e-01
-8.62531662e-01 -1.20371306e+00 -5.61249614e-01 -1.89343959e-01
-6.63816452e-01 -6.97480738e-01 -8.49743366e-01 -1.12502694e+00
6.22765005e-01 1.22523285e-01 9.83157039e-01 -1.19278088e-01
2.59917498e-01 4.01239008e-01 -7.55633414e-01 -5.13521135e-01
-1.08975732e+00 4.00703073e-01 8.91843140e-02 3.20845515e-01
1.46940723e-01 -6.16367459e-01 2.11858191e-02 4.38751370e-01
-1.13737226e+00 1.43860608e-01 7.16011882e-01 7.59437859e-01
5.97293735e-01 -5.58551103e-02 1.12492383e+00 -5.35625756e-01
8.74783874e-01 -3.55929881e-01 -4.26350981e-01 2.27939501e-01
-1.71329752e-01 1.81996197e-01 8.39435816e-01 -5.91168761e-01
-9.55615520e-01 3.43862176e-01 -3.66876096e-01 -3.01570237e-01
-2.57516772e-01 7.81685710e-01 -3.47787529e-01 4.38202441e-01
7.03756392e-01 3.41600090e-01 -1.51071036e-02 -5.46643555e-01
2.80378431e-01 1.24487424e+00 6.87712431e-01 -7.63822734e-01
9.58309650e-01 -1.12949647e-01 -4.65678215e-01 -7.37193704e-01
-6.91380918e-01 -4.50937718e-01 -3.34005117e-01 -2.09174946e-01
7.68136144e-01 -1.10372484e+00 6.86729141e-03 -3.87978368e-02
-1.55992556e+00 -1.26974937e-02 -3.39212358e-01 7.66382217e-01
-8.31633508e-01 2.31999502e-01 -2.75031060e-01 -8.86632979e-01
-3.83596122e-01 -9.39197719e-01 1.04855311e+00 1.47586852e-01
-3.79475355e-01 -6.61280334e-01 5.03117502e-01 5.25697231e-01
2.64840752e-01 -4.61031348e-02 7.34496415e-01 -1.00759244e+00
-4.40928757e-01 -5.42227864e-01 7.44016245e-02 4.95886296e-01
2.29153126e-01 -1.57348305e-01 -1.08600914e+00 -5.38672917e-02
-8.19537044e-03 -5.94356835e-01 8.62777159e-02 -1.27192825e-01
8.85027289e-01 -3.83596122e-01 -3.09892856e-02 3.76139134e-02
1.12654448e+00 3.66021246e-01 6.77468657e-01 2.11988479e-01
3.86088222e-01 5.67297220e-01 7.70288646e-01 1.25636727e-01
4.27015163e-02 8.57073426e-01 1.44067958e-01 -2.38938257e-02
2.92487629e-02 -1.03984892e+00 6.10127449e-01 1.16947913e+00
2.29148671e-01 -3.06739926e-01 -8.87434900e-01 9.39552963e-01
-1.74742806e+00 -8.92128646e-01 1.18587673e-01 2.27192187e+00
1.06658900e+00 1.75146624e-01 1.54197752e-01 2.23475918e-01
1.05105281e+00 -2.88995475e-01 -1.98988229e-01 -4.76299822e-01
2.08092421e-01 -4.28952761e-02 -1.17935099e-01 3.97859305e-01
-7.21621573e-01 6.82898104e-01 6.09702635e+00 7.32714355e-01
-1.18884385e+00 8.26098919e-02 2.78791428e-01 9.15748700e-02
-4.34280396e-01 3.32473703e-02 -6.10336006e-01 4.27145034e-01
1.48275375e+00 -5.62624991e-01 3.95841926e-01 6.60358787e-01
4.67056930e-01 2.73588985e-01 -1.34396470e+00 9.26808536e-01
3.18118334e-01 -8.98030460e-01 4.47589725e-01 -2.38916501e-01
6.89281940e-01 -2.44282335e-01 -2.75127679e-01 3.62833470e-01
-2.69016713e-01 -7.69698679e-01 8.21643293e-01 2.56096214e-01
8.74885142e-01 -4.61768270e-01 8.86345029e-01 4.46478516e-01
-1.18323767e+00 1.76549390e-01 -3.17123741e-01 -3.54216397e-02
4.92505729e-01 3.41028780e-01 -1.54759240e+00 6.61989629e-01
2.51486778e-01 7.37218373e-03 -4.02321994e-01 1.46182048e+00
-5.05890310e-01 1.06208050e+00 -4.74186271e-01 -4.84561510e-02
2.59577125e-01 -4.52292114e-02 5.29535294e-01 1.13522577e+00
8.02871883e-01 -2.27723002e-01 2.04193279e-01 7.17080474e-01
-1.62976861e-01 5.02405882e-01 -7.48194396e-01 -6.66204810e-01
4.16390240e-01 1.06091666e+00 -4.09780502e-01 -3.46312255e-01
-3.83053273e-01 6.88196421e-01 -3.95884514e-01 2.88812131e-01
-9.24208641e-01 -6.59216762e-01 7.63525888e-02 3.28014970e-01
1.09218359e-01 1.29543990e-01 2.05980152e-01 -1.00859964e+00
2.45924354e-01 -1.06961858e+00 2.07586899e-01 -1.33293211e+00
-1.09552646e+00 1.40417445e+00 1.12436138e-01 -1.78648150e+00
-1.05285180e+00 -5.77633604e-02 -5.65766394e-01 1.03528845e+00
-1.19840252e+00 -1.01853240e+00 -7.36539513e-02 4.20220524e-01
6.38625681e-01 -3.14563781e-01 1.07082093e+00 4.29140300e-01
9.33558941e-02 5.47567487e-01 -2.14366972e-01 1.99758947e-01
9.62244749e-01 -9.66942489e-01 1.71519101e-01 8.39412749e-01
5.91834188e-01 3.54578823e-01 1.02834892e+00 -4.33001876e-01
-8.53094697e-01 -1.10808933e+00 1.28420115e+00 -1.56825006e-01
7.66438723e-01 -5.26561081e-01 -1.17098439e+00 4.22291607e-01
3.41027319e-01 -3.34799677e-01 9.71701920e-01 -2.19943941e-01
-2.33248472e-01 -2.30768308e-01 -8.32056105e-01 2.78100580e-01
6.30796850e-01 -9.59999740e-01 -1.06199491e+00 4.25622076e-01
9.69126523e-01 -2.58603632e-01 -4.40339088e-01 -1.00519083e-01
3.00001055e-01 -2.94675767e-01 7.04718649e-01 -6.17222965e-01
5.66410542e-01 -5.94778299e-01 -1.94535822e-01 -1.33900809e+00
1.86925560e-01 -9.21192646e-01 4.27633584e-01 1.85887229e+00
9.12063658e-01 -2.24779025e-01 6.87122047e-01 3.82095486e-01
-1.41481787e-01 -2.88153347e-02 -8.80332291e-01 -9.46812391e-01
-3.06454748e-01 -4.21000153e-01 7.37610519e-01 5.14509559e-01
3.40168327e-01 8.61575246e-01 -4.70755398e-01 3.04249883e-01
3.04827858e-02 2.10297331e-01 8.95281494e-01 -1.16443312e+00
-5.64005911e-01 3.81803036e-01 -2.46066034e-01 -1.02147770e+00
1.52706817e-01 -8.99494529e-01 7.43903697e-01 -1.64468086e+00
-2.38869172e-02 -4.11292434e-01 -1.94789499e-01 5.41159511e-01
-1.20882630e-01 3.29721540e-01 3.17663699e-01 2.22234949e-01
-4.84985948e-01 5.28377831e-01 1.09469318e+00 -2.76558459e-01
-5.52582979e-01 3.48791897e-01 -8.88033509e-01 3.15638602e-01
7.69250154e-01 -8.67377579e-01 -7.60414660e-01 -3.82827252e-01
2.16841340e-01 4.89044517e-01 7.69264400e-02 -1.29880798e+00
1.71935678e-01 3.23937349e-02 -3.79298359e-01 -3.38482708e-01
3.77695054e-01 -8.60820174e-01 2.26715565e-01 6.91069961e-02
-7.00932026e-01 6.39892444e-02 2.45110303e-01 4.53971595e-01
-7.79605985e-01 -7.10300088e-01 3.53876919e-01 -4.31479141e-02
-2.61035234e-01 -1.43776268e-01 -7.16370583e-01 1.12096600e-01
6.28058136e-01 1.74320236e-01 -2.55293008e-02 -9.90132809e-01
-6.18955314e-01 -2.15702597e-02 2.34049514e-01 5.53345501e-01
4.38132197e-01 -1.57861114e+00 -7.76234090e-01 1.94989607e-01
3.90115738e-01 3.22436318e-02 -1.70766860e-01 3.48878711e-01
-1.94946155e-01 6.43801749e-01 -2.28321180e-01 -5.99846363e-01
-1.16643739e+00 4.70335931e-01 -1.09907566e-02 -2.55117029e-01
-2.76675761e-01 4.81009841e-01 -2.09020469e-02 -1.91341966e-01
3.54603469e-01 -2.53003687e-01 -1.90993473e-01 -1.23534217e-01
7.26013482e-01 -4.49507922e-01 9.64860767e-02 -6.91906273e-01
-1.33170158e-01 4.66977090e-01 2.19121456e-01 -8.54898512e-01
1.23142743e+00 -7.90777355e-02 7.48448372e-02 6.02337122e-01
1.24701560e+00 2.21402541e-01 -7.87426054e-01 -2.09577069e-01
-1.74941257e-01 -2.27268219e-01 -2.07392141e-01 -6.71049476e-01
-5.19679248e-01 1.00832796e+00 3.80727053e-01 5.42743206e-01
1.13585389e+00 -1.74235310e-02 1.00311613e+00 6.88653648e-01
5.31098723e-01 -9.64120746e-01 1.79738790e-01 4.16571915e-01
1.07994711e+00 -9.02981460e-01 -6.92935407e-01 -3.90315294e-01
-7.75329113e-01 1.09840858e+00 5.09676933e-01 2.34333858e-01
1.69383004e-01 -1.87436596e-01 5.41423380e-01 2.93807209e-01
-6.79582000e-01 -1.92539170e-01 3.03563714e-01 7.19079137e-01
3.77772391e-01 -2.51726329e-01 -3.37341309e-01 7.78572619e-01
-5.24590194e-01 2.98597336e-01 8.30671132e-01 8.11142385e-01
-4.42145705e-01 -1.55237520e+00 -7.14115560e-01 6.76835924e-02
-3.01017493e-01 -9.85216256e-03 -5.62046170e-01 1.99219927e-01
-1.17981747e-01 1.30373490e+00 -7.04245865e-02 -5.09910941e-01
3.15244585e-01 4.24043089e-01 -1.15035474e-02 -1.08481622e+00
-3.35688144e-01 2.58223653e-01 2.77855009e-01 -1.17869452e-01
-3.56228471e-01 -3.88896108e-01 -1.11709201e+00 6.07595026e-01
-3.08744639e-01 1.21653116e+00 8.70093405e-01 8.91524434e-01
2.85743922e-01 4.02947456e-01 8.61397088e-01 -7.56425440e-01
-4.51368272e-01 -1.15527236e+00 -3.68693978e-01 5.72265863e-01
1.45087466e-01 -1.81146458e-01 -4.51873273e-01 6.78233743e-01] | [15.281113624572754, 4.882950305938721] |
93fc7bf6-cc9c-4455-aed1-782335c1c8f5 | super-resolution-method-for-coherent-doa | 2103.03271 | null | https://arxiv.org/abs/2103.03271v1 | https://arxiv.org/pdf/2103.03271v1.pdf | Super-resolution Method for Coherent DOA Estimation of Multiple Wideband Sources | We focus on coherent direction of arrival estimation of wideband sources based on spatial sparsity. This area of research is encountered in many applications such as passive radar, sonar, mining, and communication problems, in which an increasing attention has been devoted to improving the estimation accuracy and robustness to noise. By the development of super-resolution algorithms, narrowband direction of arrival estimation based on gridless sparse algorithms and atomic norm minimization has already been addressed. In this paper, a superresolution based method is proposed for coherent direction of arrival estimation of multiple wideband sources. First, unlike the conventional coherent methods, we develop a new focusing method to map the subband with the largest center frequency to the other ones, which leads to an accurate method with no requirement for initial estimates for DOAs. Then, we introduce an atomic norm problem by defining a new set of atoms and exploiting the signal joint sparsity of different frequency subbands in a continuous spatial domain. This problem is then cast as a semidefinite program, which leads to implementing a new coherent direction of arrival estimation method with higher resolution and more robustness to noise. Our method needs only one single snapshot for each frequency subband, leading to a small number of snapshots for the received wideband signal compared to the other coherent DOA techniques. Numerical simulations show the outperformance of the proposed method compared to the conventional ones. | ['Mohammad Hossein Kahaei', 'Milad Javadzadeh Jirhandeh'] | 2021-03-04 | null | null | null | null | ['direction-of-arrival-estimation'] | ['audio'] | [ 1.24586493e-01 -3.63128394e-01 1.60241202e-01 6.69872835e-02
-7.02246726e-01 -2.36297995e-01 2.68606663e-01 -5.26341647e-02
-2.73021907e-01 9.74499345e-01 3.22741508e-01 2.93855011e-01
-5.72443128e-01 -9.34170187e-01 -2.82909155e-01 -1.02907491e+00
-2.76193321e-01 -1.49416059e-01 8.17149356e-02 -1.79500014e-01
3.84250134e-02 2.99894035e-01 -1.17054284e+00 -4.39385146e-01
9.50741589e-01 9.38642085e-01 2.32242033e-01 3.07776630e-01
6.22485168e-02 3.65423739e-01 -4.50485587e-01 1.19761534e-01
3.08300167e-01 -7.31150508e-01 1.80496544e-01 1.28415704e-01
2.43178189e-01 -1.50860742e-01 -1.02953613e-01 1.33326411e+00
6.75142944e-01 3.43583494e-01 3.16762626e-01 -7.72180080e-01
-2.27318630e-01 4.95130479e-01 -9.60233688e-01 2.31359482e-01
2.65869975e-01 -5.87548375e-01 5.92025220e-01 -1.20064867e+00
2.74096161e-01 8.93433750e-01 7.98977077e-01 -1.14062347e-01
-9.25963879e-01 -6.17648542e-01 -9.56394672e-02 1.48533002e-01
-1.71821558e+00 -5.52925944e-01 1.10378182e+00 -2.19094455e-01
7.29992986e-02 1.50117576e-01 8.41231704e-01 5.51819563e-01
-8.84973258e-02 2.91189820e-01 1.09012997e+00 -5.87394059e-01
4.08951104e-01 -2.42464378e-01 2.80662656e-01 5.88483989e-01
8.60004544e-01 -4.92248051e-02 -6.52398646e-01 -2.61256993e-01
7.03805268e-01 1.83849931e-02 -7.52172172e-01 -4.46757764e-01
-1.28436971e+00 8.24918151e-01 2.53495812e-01 9.04248059e-01
-7.23914087e-01 5.74850217e-02 -1.58473954e-01 -4.52657379e-02
6.25875711e-01 3.14306855e-01 1.41544163e-01 1.17886521e-01
-1.26644182e+00 1.90779239e-01 8.64975154e-01 5.34148097e-01
8.46113563e-01 6.58166707e-01 1.04604855e-01 8.80612314e-01
3.15090686e-01 1.14781857e+00 2.09651232e-01 -7.59219587e-01
3.23966146e-01 2.54654009e-02 3.99234265e-01 -1.69342148e+00
-5.16914010e-01 -1.34822476e+00 -1.34057033e+00 -1.10359341e-01
3.53412867e-01 -6.02527559e-01 -4.27112132e-01 1.69292021e+00
5.74837506e-01 7.55603433e-01 2.32610777e-01 1.06665957e+00
2.54127443e-01 1.01767731e+00 -6.63438737e-01 -9.28761959e-01
1.12639701e+00 -5.01516163e-01 -1.08597481e+00 -1.58738613e-01
9.89271030e-02 -9.50984955e-01 2.27544025e-01 6.69546068e-01
-9.60431218e-01 -4.86997813e-01 -1.14404094e+00 6.99849606e-01
6.34321719e-02 1.05406381e-01 4.76571113e-01 7.61870801e-01
-7.48390913e-01 -1.39800280e-01 -6.30202711e-01 3.50637324e-02
-3.02992528e-03 -2.18015492e-01 -2.05467984e-01 -1.58517540e-01
-1.04997540e+00 5.50223470e-01 1.98316962e-01 3.75380427e-01
-4.26680744e-01 -8.43353927e-01 -8.64222348e-01 7.98783228e-02
3.97623956e-01 -5.61166465e-01 5.80891728e-01 -9.14686441e-01
-1.32098210e+00 -4.56175068e-03 -5.66827655e-01 -5.66927850e-01
1.18922360e-01 -2.86367595e-01 -6.64562702e-01 2.16898263e-01
3.24608654e-01 -3.59157324e-01 1.14906001e+00 -1.20880640e+00
-6.01267576e-01 -3.45250696e-01 -2.43630856e-01 1.33600131e-01
-4.24783379e-01 -5.26947916e-01 -1.59471512e-01 -8.41777623e-01
7.29998171e-01 -7.23212421e-01 -4.68561381e-01 -4.14620727e-01
-2.96663791e-02 2.40897879e-01 4.59165633e-01 -4.97478634e-01
1.29700482e+00 -2.22858596e+00 3.42916459e-01 5.12750268e-01
1.87279493e-01 1.11625232e-02 9.19308234e-03 5.80536902e-01
7.53696933e-02 -4.52506393e-01 -3.78880769e-01 -1.95818454e-01
-4.85794574e-01 -1.90990582e-01 -3.76408130e-01 8.76927197e-01
-3.53937387e-01 -6.08520303e-03 -9.50152516e-01 -2.52058744e-01
1.68282479e-01 6.37120664e-01 -6.13283813e-01 4.56917658e-02
3.26486230e-01 7.76001513e-01 -7.47954011e-01 3.97505909e-01
1.19100559e+00 -1.23469019e-02 -9.33625326e-02 -4.27593529e-01
-7.49696732e-01 -3.66159618e-01 -1.80508590e+00 1.79595053e+00
-8.05865228e-01 3.41049373e-01 6.25640035e-01 -1.38156128e+00
1.14451802e+00 3.90788555e-01 7.23409176e-01 -5.32712579e-01
-1.98553786e-01 5.69699228e-01 -2.45498031e-01 -3.60840529e-01
1.78938434e-01 -3.48295778e-01 -2.68577430e-02 1.27781406e-01
-1.92860961e-01 -7.94668198e-02 2.82813549e-01 -7.30929002e-02
7.84222901e-01 -3.35759908e-01 6.45476282e-01 -3.69142503e-01
9.48079586e-01 -1.64208815e-01 7.56288707e-01 6.99355364e-01
3.56231421e-01 3.43437344e-01 -2.86468178e-01 -2.91309923e-01
-5.85925639e-01 -6.78348362e-01 -1.97951034e-01 4.18785572e-01
3.88141274e-01 -4.41225357e-02 -4.06325996e-01 9.79410135e-04
-2.62198627e-01 4.23242211e-01 -3.00583363e-01 1.51395202e-01
-9.25460458e-01 -9.80350733e-01 1.44651949e-01 -1.02476634e-01
9.96736288e-01 -1.29021481e-01 -6.56299114e-01 4.17429179e-01
-3.78622293e-01 -1.06586468e+00 -1.54240668e-01 -7.80217946e-02
-9.01186943e-01 -7.49521196e-01 -1.25102735e+00 -6.35596454e-01
5.71321487e-01 8.87669206e-01 5.90994954e-01 -1.46815404e-01
-5.94211258e-02 5.79711914e-01 -5.63079655e-01 -3.16305190e-01
2.47024149e-01 -2.00543553e-01 2.75502384e-01 7.04516292e-01
-2.77761310e-01 -9.55833435e-01 -7.35741973e-01 2.02005759e-01
-7.83681691e-01 -6.84794635e-02 7.84632385e-01 8.23398054e-01
5.81026673e-01 3.06476772e-01 8.46687555e-01 -5.57117641e-01
5.66816151e-01 -7.64921665e-01 -1.00227380e+00 -1.10852480e-01
-3.17643225e-01 -1.67364731e-01 8.42188358e-01 -1.32599205e-01
-1.30380952e+00 6.64370358e-02 -3.12873684e-02 -8.69031027e-02
1.36092067e-01 9.85534787e-01 1.60265863e-02 -3.52367342e-01
5.39466679e-01 6.91913128e-01 -2.01597422e-01 -4.84723359e-01
3.68696630e-01 3.50792795e-01 4.22654986e-01 -2.55730957e-01
1.23297215e+00 7.90229976e-01 5.20640492e-01 -1.50409114e+00
-9.72326100e-01 -7.63125360e-01 -3.54355454e-01 -1.53648764e-01
4.38837469e-01 -9.26107347e-01 -3.17556888e-01 2.70047754e-01
-1.09431303e+00 3.84499371e-01 -1.51443824e-01 1.23206341e+00
-1.96247816e-01 7.07035720e-01 9.99906566e-03 -1.15014672e+00
-1.66076019e-01 -7.19605744e-01 6.10440552e-01 1.95108026e-01
1.75859153e-01 -9.23880637e-01 5.37084043e-01 3.13702114e-02
6.32569075e-01 3.65557700e-01 3.42277795e-01 -1.97067186e-01
-6.23377979e-01 -2.49899998e-01 -2.85563767e-02 1.12315767e-01
4.10678029e-01 -7.87783802e-01 -3.78810585e-01 -4.14659232e-01
7.11148322e-01 2.35861182e-01 5.62851369e-01 8.03186297e-01
3.77849400e-01 -3.69319350e-01 -4.30594623e-01 8.54987383e-01
1.74570560e+00 2.84073174e-01 2.55625725e-01 1.52051777e-01
3.42066944e-01 2.23792911e-01 8.17272425e-01 8.50452960e-01
-7.13214949e-02 7.95193076e-01 3.31142664e-01 -1.30863935e-01
-3.07337698e-02 2.05105320e-01 -8.96498282e-03 9.33147669e-01
-1.80476919e-01 -2.06471622e-01 -4.85391498e-01 8.41199875e-01
-1.93521130e+00 -1.16852582e+00 -3.85576367e-01 2.50270796e+00
3.69522780e-01 -3.80240291e-01 -1.01851709e-02 2.84929395e-01
7.31985986e-01 4.96824712e-01 -2.63649493e-01 2.79177338e-01
-9.83455107e-02 2.67167538e-01 5.45340359e-01 7.35182762e-01
-8.69003475e-01 3.41478080e-01 5.36925507e+00 9.26699042e-01
-1.12715137e+00 1.34783790e-01 6.25337334e-03 1.62201166e-01
-3.17420661e-01 6.21754825e-02 -8.20212185e-01 5.03288686e-01
4.79166180e-01 -3.88742626e-01 3.47794265e-01 6.16746485e-01
6.05489910e-01 -3.60038787e-01 -2.91372478e-01 1.21568990e+00
2.62180507e-01 -1.25316346e+00 -1.28680795e-01 -9.77154449e-02
8.88363183e-01 -4.65334862e-01 -7.13856891e-02 -2.06559971e-01
-2.36352757e-01 -6.08243346e-01 4.56455529e-01 7.49372959e-01
3.62179965e-01 -8.52783859e-01 7.78814912e-01 5.26606500e-01
-1.36657143e+00 -8.66444483e-02 -4.83670831e-01 -3.04727197e-01
5.52444637e-01 1.65887249e+00 -5.82787812e-01 9.90756214e-01
3.57909471e-01 7.15719640e-01 2.57714868e-01 1.41333699e+00
-1.38291478e-01 8.92585337e-01 -6.16930425e-01 -4.79523838e-02
2.20263049e-01 -5.29153287e-01 1.08952093e+00 1.10715103e+00
1.08711135e+00 5.75340688e-01 4.09876853e-01 5.16393006e-01
2.90858418e-01 3.21183026e-01 -6.97985411e-01 4.74854290e-01
6.69388413e-01 1.24419653e+00 -5.83517969e-01 -1.62973523e-01
-5.19671321e-01 5.96401930e-01 -3.03201020e-01 5.93500376e-01
-5.62344790e-01 -5.28729260e-01 1.77430436e-01 1.33259386e-01
4.53412771e-01 -7.59199560e-01 -8.65026861e-02 -1.26225114e+00
2.31188126e-02 -7.85623550e-01 2.29288101e-01 -1.65610000e-01
-8.51068974e-01 4.95907277e-01 7.85637200e-02 -1.69365740e+00
-2.50802457e-01 3.48461904e-02 -6.17655396e-01 8.95114481e-01
-1.70629144e+00 -9.25783813e-01 -4.63143200e-01 7.91533947e-01
5.04294395e-01 -1.55158266e-01 6.68649197e-01 4.31117833e-01
-3.19136411e-01 1.00751013e-01 4.22685802e-01 -1.37398645e-01
4.99713808e-01 -7.79694140e-01 -5.19496918e-01 1.27525437e+00
1.14051081e-01 7.08262086e-01 1.04647255e+00 -5.67183375e-01
-1.43245971e+00 -7.28997827e-01 6.54083967e-01 4.15905029e-01
6.33722425e-01 8.47972855e-02 -5.73611319e-01 2.10316807e-01
2.09906667e-01 1.43924579e-01 8.39520872e-01 -9.67409462e-02
1.71588972e-01 -5.32879472e-01 -7.61866629e-01 2.83559561e-01
7.67687559e-01 1.10805370e-01 -6.47720873e-01 3.94239694e-01
2.28149548e-01 -2.20316947e-01 -6.14578545e-01 4.38209444e-01
4.53729182e-01 -1.07155120e+00 1.13634241e+00 2.35043541e-01
-1.23555273e-01 -7.60466278e-01 -2.85946876e-01 -1.53286028e+00
-5.80062509e-01 -7.51631498e-01 -2.81434394e-02 1.08133078e+00
7.80163854e-02 -8.50081265e-01 5.65286338e-01 -3.64533246e-01
-1.00046024e-01 -3.83695364e-01 -1.17892122e+00 -7.70056903e-01
-6.12849474e-01 -3.46158624e-01 2.44730905e-01 9.19189095e-01
-1.63317576e-01 3.22417617e-01 -8.34610522e-01 8.49169791e-01
1.26989305e+00 7.30740130e-01 7.87485182e-01 -1.24055672e+00
-5.78029752e-01 5.88084513e-05 -1.83941334e-01 -1.43286347e+00
-1.71154276e-01 -6.10468566e-01 9.67648849e-02 -1.52459216e+00
-1.90189108e-01 -5.55693686e-01 -1.66616008e-01 -1.58127055e-01
-5.43337204e-02 4.46443230e-01 1.40487840e-02 1.86401308e-01
-3.89160365e-01 6.45893276e-01 1.05325758e+00 -9.57532134e-03
-3.46305639e-01 4.90069240e-01 -4.58960682e-01 9.66089666e-01
4.25163358e-01 -4.53845203e-01 -4.13471162e-01 -5.80401957e-01
3.77309591e-01 5.47240496e-01 1.58861905e-01 -1.37357128e+00
3.95422071e-01 -5.43116517e-02 2.52980411e-01 -4.95007396e-01
4.94465858e-01 -9.99843717e-01 2.66434431e-01 5.96243680e-01
1.46003082e-01 -4.28160042e-01 -1.17088452e-01 9.63428259e-01
-6.48196936e-01 -5.39409995e-01 8.33001971e-01 -1.12090923e-01
-6.47465765e-01 1.02518685e-01 -5.25398850e-01 -1.57231793e-01
8.77213657e-01 -1.99820518e-01 2.04333574e-01 -8.68832290e-01
-6.60884142e-01 1.92480478e-02 -1.47090763e-01 -2.00257808e-01
6.82140768e-01 -1.27058363e+00 -9.91787493e-01 1.46479502e-01
-2.97291785e-01 -1.87890798e-01 6.15863740e-01 1.02597916e+00
-2.94623226e-01 5.56349516e-01 -3.00515592e-02 -5.51123142e-01
-9.03995991e-01 1.83179721e-01 1.37404040e-01 -3.51971507e-01
-4.13298219e-01 6.21279836e-01 2.15480939e-01 1.63322330e-01
-1.31216705e-01 -1.06189989e-01 -4.75179464e-01 2.35609114e-01
5.90551853e-01 7.14661658e-01 -7.11425021e-02 -7.41545379e-01
-3.88115793e-01 1.24701512e+00 5.08496821e-01 -2.96015471e-01
1.42679954e+00 -3.66861761e-01 -2.94674993e-01 3.29224586e-01
1.15342772e+00 1.13590300e+00 -8.02707255e-01 -2.49282748e-01
-2.74800569e-01 -6.05996311e-01 3.42891842e-01 -2.51764536e-01
-9.51696098e-01 7.02265739e-01 6.56604886e-01 3.31772387e-01
1.39211059e+00 -4.41972673e-01 8.07956457e-01 4.61822748e-01
8.28343749e-01 -5.55722713e-01 -1.15620652e-02 4.39546466e-01
8.54163051e-01 -8.67306650e-01 2.12322637e-01 -6.00971460e-01
8.51369202e-02 1.35952413e+00 4.74076308e-02 -4.24734980e-01
8.16736996e-01 2.04335093e-01 -1.06770605e-01 3.34982872e-02
4.64811213e-02 -3.48943740e-01 1.89545080e-01 3.95528704e-01
3.09830993e-01 -6.29510805e-02 -9.06609476e-01 5.61935961e-01
-9.51260552e-02 1.27832800e-01 5.11677384e-01 6.17292523e-01
-8.47382307e-01 -9.96285975e-01 -9.42355871e-01 1.78918898e-01
-4.71707404e-01 -1.14048958e-01 4.38450426e-01 4.75877047e-01
2.62285322e-02 1.03262711e+00 -3.14013153e-01 1.22149140e-01
3.75790536e-01 -2.86449194e-01 1.86892346e-01 -5.43648779e-01
2.72536039e-01 5.47637820e-01 -3.34473439e-02 -2.41432130e-01
-6.31864130e-01 -5.03297985e-01 -1.01620996e+00 4.07710373e-01
-5.54936349e-01 8.82845819e-01 5.56704581e-01 8.13956499e-01
1.79467484e-01 4.00576741e-01 1.10273600e+00 -8.14280689e-01
-3.27885717e-01 -6.93899930e-01 -8.41340125e-01 -5.52765504e-02
5.68513989e-01 -6.92695737e-01 -5.92908442e-01 -2.11391285e-01] | [6.476180553436279, 1.3352352380752563] |
84b5ac8d-6880-41d7-be97-724ceec09647 | the-re-label-method-for-data-centric-machine | 2302.04391 | null | https://arxiv.org/abs/2302.04391v3 | https://arxiv.org/pdf/2302.04391v3.pdf | The Re-Label Method For Data-Centric Machine Learning | In industry deep learning application, our manually labeled data has a certain number of noisy data. To solve this problem and achieve more than 90 score in dev dataset, we present a simple method to find the noisy data and re-label the noisy data by human, given the model predictions as references in human labeling. In this paper, we illustrate our idea for a broad set of deep learning tasks, includes classification, sequence tagging, object detection, sequence generation, click-through rate prediction. The experimental results and human evaluation results verify our idea. | ['Tong Guo'] | 2023-02-09 | null | null | null | null | ['click-through-rate-prediction'] | ['miscellaneous'] | [-1.66458368e-01 -3.01409692e-01 3.96552794e-02 -7.34396935e-01
-7.14547515e-01 -5.89405775e-01 2.58550644e-01 2.22246513e-01
-6.44613206e-01 1.14748037e+00 9.95189473e-02 -1.10243171e-01
3.23842019e-01 -6.07110322e-01 -5.59874892e-01 -4.21309710e-01
1.09391645e-01 5.21024466e-01 4.82582927e-01 -7.80265555e-02
2.21416235e-01 8.81236866e-02 -1.40557587e+00 5.65069020e-01
4.27067906e-01 1.21213102e+00 3.35793614e-01 5.84324598e-01
-4.61927801e-01 9.03493583e-01 -1.30154848e+00 -7.23458529e-01
1.12294666e-01 -2.36540228e-01 -1.01250553e+00 1.09038457e-01
-6.60046116e-02 -3.52034897e-01 -1.31168410e-01 1.45642102e+00
9.03104305e-01 4.12947163e-02 5.65256119e-01 -1.38450348e+00
-6.76354349e-01 8.64482284e-01 -4.18759376e-01 2.85890847e-01
2.22942770e-01 1.42884225e-01 7.72371948e-01 -9.37830448e-01
8.28734517e-01 1.21877062e+00 7.84607649e-01 8.62066209e-01
-7.75400877e-01 -8.87609720e-01 4.45682667e-02 3.40638876e-01
-1.51451540e+00 1.76227853e-01 4.37323749e-01 -6.59902811e-01
8.21175516e-01 3.16241272e-02 3.48654419e-01 1.34068465e+00
1.48668051e-01 9.55047488e-01 9.15148914e-01 -1.50419593e-01
3.51812452e-01 1.74450964e-01 6.07336998e-01 4.53962326e-01
2.36624911e-01 1.16673715e-01 -4.34469521e-01 -1.66709170e-01
3.69828910e-01 1.51136741e-01 -1.18690826e-01 2.45603666e-01
-1.12782252e+00 8.45268667e-01 -4.66109477e-02 2.61955947e-01
-1.09559111e-01 1.85650706e-01 9.68081057e-01 3.89422059e-01
4.42724615e-01 2.91859448e-01 -1.09567523e+00 -3.75900835e-01
-6.71648502e-01 3.01088572e-01 9.74880695e-01 1.45039523e+00
6.05870903e-01 8.18277523e-02 -3.86445701e-01 8.48483443e-01
2.19313323e-01 1.20048627e-01 1.13587868e+00 -6.00995541e-01
4.07817274e-01 3.29731286e-01 2.58668751e-01 -7.51464725e-01
-4.62397963e-01 -3.91561955e-01 -1.15356374e+00 -1.18061835e-02
5.81602864e-02 -4.96537983e-01 -1.33806169e+00 1.41950893e+00
2.79570490e-01 1.11445040e-01 -9.54272784e-03 1.06724131e+00
1.01470864e+00 7.79609382e-01 4.71729904e-01 -2.99351513e-01
1.10687220e+00 -9.26182151e-01 -9.75568414e-01 9.53309909e-02
8.83484244e-01 -1.14633405e+00 9.16992605e-01 9.56255555e-01
-3.49585712e-01 -8.84080589e-01 -8.12101305e-01 -9.36508030e-02
-5.89434445e-01 1.69210121e-01 5.53796828e-01 5.44067681e-01
-4.77848977e-01 7.81866252e-01 -4.59588200e-01 -1.90964147e-01
5.87914348e-01 2.11234838e-01 -1.04471490e-01 4.74952273e-02
-1.61578035e+00 6.52987540e-01 9.46029961e-01 -5.96890524e-02
-1.29630017e+00 -2.15647370e-01 -5.49530506e-01 -1.39100522e-01
1.56782329e-01 -2.52165765e-01 1.80875969e+00 -7.25142896e-01
-1.05869019e+00 8.18001211e-01 1.71662107e-01 -6.36750638e-01
9.39764619e-01 -4.43486691e-01 -7.52728522e-01 -5.74969888e-01
5.43642826e-02 5.55340946e-01 3.84920955e-01 -1.05352414e+00
-9.34249043e-01 -8.90932232e-02 -4.79847044e-01 -3.57517332e-01
1.36325480e-02 1.71647400e-01 -2.30605617e-01 -9.43409979e-01
-1.09649353e-01 -6.58105612e-01 -5.60338497e-01 -4.01600927e-01
-6.89428389e-01 -8.09253871e-01 8.22984815e-01 -4.61838454e-01
1.37055147e+00 -1.98107684e+00 -5.64906299e-01 5.50284609e-02
3.92894089e-01 6.36118233e-01 -1.21392295e-01 3.44712019e-01
-1.59860849e-01 5.81315041e-01 1.13953747e-01 1.64237805e-02
-1.68387555e-02 2.85340279e-01 -2.20643073e-01 1.75367385e-01
7.74233565e-02 7.06272066e-01 -1.09419656e+00 -6.74181581e-01
-9.52869654e-03 9.71984193e-02 -9.80131030e-02 4.39689934e-01
-3.81360829e-01 1.20698802e-01 -4.88492370e-01 6.41139984e-01
7.29081035e-01 -4.38921422e-01 2.00094134e-02 -1.59698144e-01
3.28099281e-01 8.76338407e-02 -1.48041129e+00 1.55444860e+00
-1.29423961e-01 5.61848402e-01 -4.78849173e-01 -8.54331851e-01
1.33798659e+00 2.77388364e-01 2.40163878e-01 -5.31541705e-01
3.49473029e-01 1.55311823e-01 -2.19407782e-01 -1.13165128e+00
4.31750059e-01 3.40528786e-02 -1.60806865e-01 4.35014635e-01
4.31738496e-01 2.76649296e-01 1.32371053e-01 1.07086197e-01
1.19044018e+00 -1.59225184e-02 3.94133747e-01 -1.70592461e-02
2.17926964e-01 1.56128526e-01 7.47173607e-01 1.02284181e+00
-5.22809625e-01 7.21445739e-01 3.72681439e-01 -9.12005186e-01
-1.39393020e+00 -5.55013359e-01 3.15141156e-02 1.18715405e+00
2.23230109e-01 -3.53963464e-01 -8.79663348e-01 -1.22512853e+00
-2.39368439e-01 4.40152138e-01 -3.63889515e-01 -2.25557759e-01
-4.85702217e-01 -7.75095046e-01 6.53355122e-01 6.25584424e-01
4.85313416e-01 -1.26025200e+00 -1.40043184e-01 5.12681663e-01
-1.96294487e-01 -9.08093333e-01 -4.63400066e-01 6.49062812e-01
-4.94596690e-01 -1.00822651e+00 -6.00356281e-01 -1.23405397e+00
3.14882785e-01 -2.27945819e-02 1.26953471e+00 4.28129673e-01
-5.45387208e-01 -5.28034985e-01 -6.10044718e-01 -5.04235029e-01
-4.68696386e-01 2.68776566e-02 1.47213519e-01 -6.27466917e-01
1.12419701e+00 -1.14532836e-01 -4.38347638e-01 5.39322615e-01
-1.08779752e+00 -3.46677482e-01 4.12656277e-01 1.18898618e+00
6.10998034e-01 3.29745561e-01 7.23902285e-01 -1.14491272e+00
9.69121337e-01 -5.98205566e-01 -6.29503548e-01 2.72075504e-01
-6.90538585e-01 1.87620059e-01 6.73559010e-01 -7.75387287e-01
-8.14989567e-01 6.11953557e-01 -6.05526567e-01 -2.55434602e-01
-6.21869147e-01 4.08363968e-01 -2.84660995e-01 3.08579385e-01
1.01658487e+00 2.62181193e-01 -5.09810925e-01 -9.74303067e-01
3.65731716e-01 1.16683209e+00 3.27549160e-01 -3.78548801e-01
3.39434475e-01 -2.91819334e-01 -5.03529012e-01 -1.01413406e-01
-1.07693791e+00 -5.38213968e-01 -6.36947751e-01 9.46581457e-03
7.16492951e-01 -8.94641817e-01 -9.98027384e-01 4.31653947e-01
-1.66778958e+00 7.93133155e-02 -1.80448756e-01 5.10799766e-01
-3.13897848e-01 3.97608697e-01 -8.86298597e-01 -7.90878117e-01
-4.85735923e-01 -1.27181530e+00 1.02423537e+00 7.10529741e-03
-3.25730562e-01 -4.50693637e-01 4.58781607e-02 -2.41187774e-02
-6.04281388e-02 1.68457702e-01 6.84287906e-01 -1.53544974e+00
-4.85109240e-01 -4.31622803e-01 -2.06462607e-01 7.15519130e-01
-2.95124412e-01 -1.24656409e-01 -8.99629772e-01 1.63682520e-01
-1.50421321e-01 -6.00137949e-01 8.97598386e-01 -3.45805436e-02
1.79842722e+00 -2.72285551e-01 -3.96569818e-01 2.97994852e-01
1.49638033e+00 6.17155433e-01 4.06572610e-01 1.22974694e-01
6.21815622e-01 3.20560306e-01 9.43918467e-01 6.80144727e-01
-1.63245663e-01 2.86837965e-01 2.46682703e-01 1.50612280e-01
-2.25861296e-02 -4.30591226e-01 -2.33093351e-01 8.44015777e-01
2.94996679e-01 -7.80844510e-01 -9.90471244e-01 4.12455291e-01
-2.08285737e+00 -8.18198383e-01 -3.92635822e-01 1.66816735e+00
1.17916548e+00 3.94807845e-01 -5.00149392e-02 2.85481393e-01
1.17105126e+00 -4.12309259e-01 -4.25137192e-01 -2.44910389e-01
1.27511382e-01 4.05226648e-02 6.43299818e-01 2.14829534e-01
-1.38635564e+00 1.14422464e+00 7.71714592e+00 1.48726308e+00
-6.46677732e-01 2.92121887e-01 1.11844695e+00 2.07670093e-01
5.22480793e-02 -6.45837963e-01 -1.38506162e+00 9.39794362e-01
9.09537435e-01 -1.57618627e-01 1.47987055e-02 1.46971655e+00
1.15267962e-01 1.58876762e-01 -1.26069033e+00 1.24174321e+00
-2.28664652e-01 -1.30747926e+00 1.37714043e-01 -2.58337855e-01
8.00427675e-01 -1.46170005e-01 -1.77560061e-01 6.15905702e-01
1.01402688e+00 -1.01399624e+00 5.46419024e-01 4.85345393e-01
4.36084032e-01 -9.86866355e-01 1.42992222e+00 5.68183959e-01
-5.18690944e-01 5.72177470e-02 -7.23335683e-01 2.39567528e-03
1.72748342e-01 1.17825544e+00 -1.03182030e+00 -1.09780766e-02
8.84991586e-01 3.93313050e-01 -3.56536239e-01 1.35776556e+00
-1.54902413e-01 7.07757413e-01 -4.29046266e-02 -6.26877666e-01
1.87373489e-01 4.21618998e-01 6.44451901e-02 1.78326416e+00
2.65312850e-01 2.69832850e-01 4.99357462e-01 7.17293501e-01
-4.38738465e-01 1.76909432e-01 -4.61084008e-01 2.72459015e-02
4.95117247e-01 1.14493537e+00 -7.82649398e-01 -7.06356585e-01
-1.79688632e-01 9.88317132e-01 9.37075838e-02 2.93326259e-01
-1.01018643e+00 -8.17045569e-01 4.78089094e-01 -1.93219334e-01
7.16732144e-02 -1.06701709e-01 -4.75451589e-01 -9.03730214e-01
-8.50690007e-02 -8.31053734e-01 3.86067003e-01 -7.51162052e-01
-1.86076522e+00 8.74883950e-01 -4.24590796e-01 -1.41839957e+00
-8.25130790e-02 -6.51292682e-01 -9.61440876e-02 7.16048241e-01
-1.02096593e+00 -7.01486051e-01 -3.59029055e-01 4.79261160e-01
9.51153994e-01 -4.15001929e-01 5.38950026e-01 7.29066193e-01
-4.42847848e-01 6.87921405e-01 3.05630773e-01 6.62952840e-01
6.19278252e-01 -1.21396875e+00 8.19929302e-01 5.85769236e-01
1.14386246e-01 5.87668121e-01 8.11537802e-01 -9.46407914e-01
-4.46704179e-01 -1.40147197e+00 9.95899379e-01 -3.62928331e-01
6.83662832e-01 -4.04097885e-01 -8.80514085e-01 5.30198038e-01
1.47444203e-01 1.40259087e-01 8.83126974e-01 -2.34763533e-01
-2.27204219e-01 1.01139143e-01 -1.36303520e+00 6.72212094e-02
1.27348769e+00 1.02313433e-03 -5.15199959e-01 8.95549178e-01
1.30294728e+00 -5.64730227e-01 -6.30196571e-01 3.86460960e-01
1.73231214e-01 -5.56055665e-01 7.26014197e-01 -1.23164546e+00
5.33726215e-01 -3.18435103e-01 -1.20836101e-01 -1.33863688e+00
-4.03347611e-01 -4.52264279e-01 -2.44075595e-03 1.24271798e+00
4.84982222e-01 7.10744709e-02 1.02509546e+00 4.79095757e-01
-1.81826636e-01 -6.28159285e-01 -7.58398712e-01 -9.23402548e-01
-4.00338292e-01 -3.97104949e-01 7.89595783e-01 1.07173729e+00
-2.40271240e-01 4.03945535e-01 -9.88954186e-01 -1.69065759e-01
2.58318424e-01 -2.91820139e-01 4.06569839e-01 -1.28552258e+00
-8.27395990e-02 -2.01835539e-02 -7.14517653e-01 -1.27962673e+00
-2.73937266e-02 -5.14787138e-01 5.25309920e-01 -1.16171682e+00
1.53704569e-01 -2.89100111e-01 -5.31825364e-01 4.85819489e-01
-1.01417854e-01 2.83358186e-01 -1.58673644e-01 1.26734257e-01
-1.20251775e+00 1.92272708e-01 1.03851306e+00 -1.52093604e-01
1.48253351e-01 1.25142306e-01 -3.56581837e-01 7.24877119e-01
7.63917625e-01 -9.40370739e-01 -5.84240481e-02 -5.78367531e-01
1.94116265e-01 -2.74607956e-01 2.20789667e-02 -9.41100299e-01
2.09532231e-01 -3.02692503e-01 8.92813206e-01 -9.12427127e-01
-9.40910429e-02 -1.02679646e+00 3.90127786e-02 5.82198739e-01
-8.60925615e-01 3.41926664e-02 -1.36681363e-01 7.38821208e-01
-3.63709211e-01 -8.15969169e-01 7.98223019e-01 -4.03840303e-01
-1.13084769e+00 4.08244371e-01 -5.39994657e-01 2.38284767e-01
9.90960538e-01 1.59100398e-01 -3.45996261e-01 -2.55207002e-01
-1.26225758e+00 2.61553079e-01 -1.80858225e-01 5.69542706e-01
6.36563241e-01 -1.57048059e+00 -4.68479365e-01 -5.65438867e-02
2.43005991e-01 -9.95071754e-02 -3.81956659e-02 -1.49658024e-01
-7.92713344e-01 2.92918593e-01 2.60803476e-02 -5.09434104e-01
-1.25534391e+00 9.85379577e-01 2.45456293e-01 -1.86752871e-01
-3.57527703e-01 1.25641394e+00 -3.27245206e-01 -3.93349171e-01
8.73528540e-01 -3.95055473e-01 -1.39878586e-01 -1.67549893e-01
8.18952501e-01 4.59862649e-01 2.74529696e-01 -3.70630562e-01
-1.67988107e-01 1.14505559e-01 -2.56893218e-01 2.44446605e-01
1.12400579e+00 -6.51784837e-02 2.73431271e-01 3.11012000e-01
1.50290668e+00 -5.80797493e-01 -9.27331269e-01 -4.00673956e-01
3.55457515e-01 -5.58929503e-01 -2.89724588e-01 -1.14646673e+00
-1.17788386e+00 9.71910119e-01 1.14015222e+00 4.61475790e-01
7.37588406e-01 -5.51666506e-02 1.26368546e+00 8.21852386e-01
4.20514137e-01 -1.43518484e+00 1.59998566e-01 6.07222974e-01
3.59162182e-01 -1.59032428e+00 -2.92727739e-01 -2.69077241e-01
-7.91013837e-01 1.02438724e+00 1.09328663e+00 2.30645039e-03
8.47564697e-01 4.99740362e-01 2.34397918e-01 -1.22841835e-01
-7.73950875e-01 -2.03505367e-01 -2.66076565e-01 8.71456742e-01
5.79079449e-01 -3.16242746e-04 -8.30887377e-01 1.13817990e+00
-1.73389927e-01 3.64077270e-01 4.56771851e-01 8.45344841e-01
-1.13846064e+00 -1.03347027e+00 -5.68878092e-02 4.18986917e-01
-8.40522468e-01 -2.62854487e-01 -3.55275482e-01 5.01025677e-01
6.91517472e-01 8.73827517e-01 -6.98122010e-02 -8.64557803e-01
4.01787698e-01 2.31355682e-01 -2.03196377e-01 -7.05824554e-01
-4.95289415e-01 1.40182167e-01 1.70691963e-02 -3.71450722e-01
-3.84336635e-02 8.63407776e-02 -1.24191201e+00 -1.61824882e-01
-5.14463603e-01 3.75391155e-01 6.67303205e-01 8.17783713e-01
1.83092132e-01 3.64799708e-01 5.31427026e-01 -1.79051548e-01
-8.37174892e-01 -1.26810420e+00 -7.41404712e-01 8.31081986e-01
2.13376537e-01 -4.68201935e-01 -1.52603880e-01 6.06742740e-01] | [9.38431167602539, 3.9376440048217773] |
8aa162c9-6a58-42c3-b2de-31adca0ce4c1 | advances-in-black-box-vi-normalizing-flows | 2006.10343 | null | https://arxiv.org/abs/2006.10343v2 | https://arxiv.org/pdf/2006.10343v2.pdf | Advances in Black-Box VI: Normalizing Flows, Importance Weighting, and Optimization | Recent research has seen several advances relevant to black-box VI, but the current state of automatic posterior inference is unclear. One such advance is the use of normalizing flows to define flexible posterior densities for deep latent variable models. Another direction is the integration of Monte-Carlo methods to serve two purposes; first, to obtain tighter variational objectives for optimization, and second, to define enriched variational families through sampling. However, both flows and variational Monte-Carlo methods remain relatively unexplored for black-box VI. Moreover, on a pragmatic front, there are several optimization considerations like step-size scheme, parameter initialization, and choice of gradient estimators, for which there are no clear guidance in the existing literature. In this paper, we postulate that black-box VI is best addressed through a careful combination of numerous algorithmic components. We evaluate components relating to optimization, flows, and Monte-Carlo methods on a benchmark of 30 models from the Stan model library. The combination of these algorithmic components significantly advances the state-of-the-art "out of the box" variational inference. | ['Abhinav Agrawal', 'Daniel Sheldon', 'Justin Domke'] | 2020-06-18 | null | http://proceedings.neurips.cc/paper/2020/hash/c91e3483cf4f90057d02aa492d2b25b1-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/c91e3483cf4f90057d02aa492d2b25b1-Paper.pdf | neurips-2020-12 | ['variational-monte-carlo'] | ['miscellaneous'] | [ 7.04474328e-03 -1.81907967e-01 -1.94339648e-01 -2.30938405e-01
-9.60279882e-01 -5.11923850e-01 9.51790333e-01 -2.01347575e-01
-3.10792297e-01 1.08439493e+00 1.14869200e-01 -2.95092195e-01
-2.79339522e-01 -5.88596284e-01 -5.14475346e-01 -9.60037589e-01
2.61437356e-01 9.20482397e-01 3.38688702e-03 1.26951605e-01
2.44355306e-01 5.98009109e-01 -1.33720648e+00 -3.53673398e-01
9.95646894e-01 6.10026896e-01 -1.67749539e-01 5.61727464e-01
-2.82816321e-01 2.74448127e-01 -4.32600945e-01 -5.20087123e-01
-1.63143471e-01 -5.10248065e-01 -4.64978367e-01 -9.71177593e-02
2.38461494e-01 -3.60638827e-01 1.84230879e-01 1.20708477e+00
4.27112639e-01 4.28510249e-01 1.12639356e+00 -1.19813228e+00
-3.59681010e-01 6.42445266e-01 -5.44223666e-01 2.40880370e-01
-1.56756893e-01 2.46578008e-01 1.05716991e+00 -8.38622510e-01
6.83628738e-01 1.26620233e+00 5.29083073e-01 5.32360077e-01
-1.73095870e+00 -5.07304668e-01 7.62761086e-02 1.20684402e-02
-1.37460351e+00 -5.62163174e-01 7.28176236e-01 -8.06636393e-01
8.98060143e-01 1.80903405e-01 8.25956106e-01 1.38958335e+00
8.52566808e-02 7.89686441e-01 9.58929300e-01 -2.77300477e-01
6.77228451e-01 3.35631818e-01 1.99652776e-01 4.66048062e-01
2.92091161e-01 2.69011825e-01 -5.38694263e-01 -4.63537961e-01
8.33649158e-01 -4.45347339e-01 -1.53620183e-01 -6.72283173e-01
-7.33758688e-01 1.18360066e+00 -1.79215595e-01 1.93202153e-01
-2.14734286e-01 3.77776891e-01 3.46816748e-01 -1.88077584e-01
8.42670321e-01 3.66220236e-01 -2.77826279e-01 -5.38657904e-01
-1.59151173e+00 7.17925310e-01 1.00974762e+00 5.90081275e-01
9.32477415e-01 2.62397796e-01 -3.74071449e-01 6.93813562e-01
9.44243252e-01 5.04461050e-01 -1.57415688e-01 -1.36207247e+00
2.62541831e-01 9.04285610e-02 2.91421771e-01 -4.96745348e-01
-7.15704411e-02 -4.91593748e-01 -6.37779713e-01 4.02471483e-01
7.56782651e-01 -3.71554613e-01 -9.64011610e-01 1.72165310e+00
4.66709763e-01 9.15202945e-02 -2.29372904e-01 7.23872900e-01
4.08765793e-01 7.19064653e-01 1.13548227e-01 -3.72401059e-01
1.15203905e+00 -7.83234119e-01 -7.54397929e-01 -1.46987736e-01
2.47500628e-01 -6.28399312e-01 7.05952466e-01 3.41094196e-01
-1.32384861e+00 -2.86160886e-01 -9.79049742e-01 -9.22412723e-02
-2.56551325e-01 1.69832096e-01 7.78807402e-01 1.02983022e+00
-1.06108725e+00 7.80456185e-01 -1.40892589e+00 -1.55395001e-01
4.87162143e-01 1.38942108e-01 2.24354744e-01 3.29995155e-01
-1.07433844e+00 9.45089459e-01 1.39054388e-01 1.13724798e-01
-1.07642066e+00 -1.05106711e+00 -8.54172289e-01 2.27212310e-01
3.34464043e-01 -1.02234399e+00 1.18788040e+00 -5.67725420e-01
-1.93472910e+00 5.43305635e-01 -5.28142989e-01 -4.06184465e-01
9.77997124e-01 -1.58040524e-01 -1.79236066e-02 2.51236316e-02
-1.71730183e-02 6.83455050e-01 1.04399538e+00 -1.30853713e+00
-1.76488340e-01 -3.31027836e-01 -2.45267972e-01 7.35851228e-02
-8.56733136e-03 -5.14927357e-02 -7.80905128e-01 -4.47263956e-01
-2.11484596e-01 -7.35728502e-01 -3.58275384e-01 2.81327106e-02
-2.34366208e-01 -2.47581482e-01 3.81293625e-01 -5.36629975e-01
1.22177303e+00 -1.97727537e+00 3.40138167e-01 2.23802775e-01
2.02044308e-01 -1.97454076e-02 3.19177777e-01 3.72737348e-01
2.77680039e-01 2.18372479e-01 -5.62816560e-01 -7.94394612e-01
4.26243097e-01 1.05601981e-01 -4.21194494e-01 6.40434325e-01
1.63557798e-01 8.89249742e-01 -8.84144723e-01 -7.13113010e-01
5.32874525e-01 8.03061903e-01 -9.27183390e-01 -6.50425777e-02
-5.84580958e-01 6.73004866e-01 -4.96184170e-01 4.34140593e-01
5.98505437e-01 -2.91958481e-01 2.06773475e-01 -3.10556889e-02
-3.40934008e-01 2.91435480e-01 -1.39313841e+00 1.66898048e+00
-8.24077521e-03 6.56374872e-01 1.87220693e-01 -7.07542539e-01
6.25628948e-01 1.68231428e-01 5.21334052e-01 3.74889039e-02
9.77175608e-02 1.87766418e-01 -1.68143973e-01 -2.05812842e-01
7.93046713e-01 -1.42544016e-01 3.67504746e-01 3.26365709e-01
2.09591419e-01 -3.23030531e-01 6.01692200e-01 1.86854064e-01
6.59966648e-01 7.04342186e-01 1.41444460e-01 -4.87203091e-01
2.93930709e-01 -8.29714630e-03 4.18859154e-01 9.77005303e-01
-1.95976332e-01 6.15906596e-01 7.56947994e-01 2.48532072e-01
-1.03118443e+00 -1.38699007e+00 -6.77136183e-01 8.70942295e-01
-2.16562778e-01 -5.29590905e-01 -8.28517377e-01 -2.08160326e-01
5.71637833e-03 9.41938519e-01 -6.52141154e-01 3.30397457e-01
-3.19038033e-01 -1.00466740e+00 5.41981280e-01 3.43131453e-01
4.19474691e-02 -8.79884303e-01 -6.29341006e-01 3.43736142e-01
4.95943502e-02 -7.72897363e-01 -1.38976127e-01 1.06309325e-01
-1.15425587e+00 -7.09685981e-01 -8.95146847e-01 1.04238458e-01
4.09147620e-01 -3.85997623e-01 1.27946782e+00 -3.77119839e-01
-3.27573493e-02 3.17314774e-01 6.73991665e-02 -2.10016221e-01
-4.84589905e-01 2.54140794e-01 -1.45956978e-01 -2.14085296e-01
2.06834048e-01 -6.13696814e-01 -4.42025989e-01 1.00911848e-01
-7.51388848e-01 -6.74784109e-02 2.02365533e-01 8.01417410e-01
5.98149598e-01 -2.60976076e-01 5.25454730e-02 -7.34386027e-01
6.09394491e-01 -5.16103089e-01 -9.60486531e-01 3.02103788e-01
-6.09722972e-01 4.06636059e-01 2.73761123e-01 -2.45516181e-01
-1.42213893e+00 -3.67880434e-01 -4.91473436e-01 -5.12248576e-01
-7.41801113e-02 6.32127404e-01 -9.51950550e-02 3.53402078e-01
3.95126581e-01 -5.44008687e-02 2.42237523e-02 -6.17618024e-01
6.97468460e-01 2.26939693e-01 3.36442381e-01 -1.01826048e+00
5.42133987e-01 6.23926938e-01 5.22738509e-02 -1.01434445e+00
-6.64715052e-01 -2.18846425e-01 -3.36563498e-01 -9.13000032e-02
1.10546815e+00 -6.06007636e-01 -6.91386938e-01 4.03085023e-01
-1.07944739e+00 -5.02376497e-01 -3.85773540e-01 4.79028195e-01
-6.11856103e-01 5.14868021e-01 -4.89203990e-01 -1.13623059e+00
-2.56853044e-01 -1.64900041e+00 1.14252031e+00 3.48073155e-01
-5.27598798e-01 -1.33148551e+00 3.84463131e-01 3.03090423e-01
4.00913507e-01 -5.29587194e-02 8.11442137e-01 -2.23402843e-01
-6.51831925e-01 2.03203589e-01 3.16796312e-03 1.97898701e-01
-3.11531007e-01 6.73392117e-01 -1.03172910e+00 -2.45220795e-01
1.97638161e-02 -8.41242447e-02 1.00115824e+00 1.06083155e+00
9.50935245e-01 1.57931104e-01 -3.56064290e-01 9.81876373e-01
1.32798839e+00 -1.64656863e-02 7.01618493e-01 -3.92755792e-02
4.73761618e-01 3.58486384e-01 2.88055539e-01 5.85465193e-01
2.04725549e-01 6.56283617e-01 2.22113505e-01 2.81519294e-01
1.14094496e-01 -2.69110084e-01 2.11243987e-01 7.08050668e-01
-2.17867717e-01 -1.82735518e-01 -9.23397720e-01 5.32119453e-01
-1.82901692e+00 -1.13390076e+00 -1.91500768e-01 2.01916337e+00
7.80491710e-01 1.45052835e-01 -3.96315977e-02 -1.58812314e-01
5.01168668e-01 4.09181535e-01 -6.59382105e-01 -6.99667558e-02
2.37887695e-01 3.04436862e-01 2.48097435e-01 8.35457146e-01
-9.47925866e-01 8.77911747e-01 6.77951956e+00 1.11301339e+00
-8.39748323e-01 3.29939723e-01 5.57282209e-01 -2.56375164e-01
-6.41749203e-01 3.07306558e-01 -1.33859539e+00 5.38862526e-01
9.17086899e-01 5.52443489e-02 5.83404243e-01 7.76100457e-01
3.74071062e-01 -3.77472997e-01 -1.07869291e+00 7.93774664e-01
-1.49208039e-01 -1.59406257e+00 -1.51916012e-01 2.36604288e-01
8.42345357e-01 1.54115394e-01 -2.11948901e-02 4.14845496e-01
3.56743395e-01 -1.02005148e+00 8.32653105e-01 6.93739474e-01
5.84004104e-01 -6.56962335e-01 3.54664803e-01 1.44484460e-01
-8.96569967e-01 5.14508545e-01 -2.24591181e-01 1.09718271e-01
7.10936368e-01 8.30061018e-01 -3.73001605e-01 4.42457020e-01
4.97054905e-01 6.33100808e-01 -1.26720324e-01 9.47537661e-01
-4.34856087e-01 9.61273491e-01 -5.55949688e-01 -1.30999401e-01
2.70055890e-01 -7.50688851e-01 8.49974453e-01 1.07679224e+00
1.14183776e-01 -3.45928252e-01 -1.92919537e-01 1.66788542e+00
1.93117678e-01 -3.19613427e-01 -1.88423648e-01 -1.78832367e-01
5.59583604e-01 1.16943216e+00 -8.17541361e-01 -2.93969035e-01
-1.14815168e-01 4.85515654e-01 4.10208255e-01 7.70725787e-01
-1.01945484e+00 1.21560939e-01 9.47389841e-01 -1.55007616e-01
5.13873696e-01 -4.83194858e-01 -7.39007294e-01 -1.31780159e+00
-3.78619641e-01 -7.36863196e-01 2.27585420e-01 -6.10412955e-01
-1.23811293e+00 1.20609894e-01 5.78492999e-01 -3.97729039e-01
-6.35653436e-01 -6.02160871e-01 -5.05028009e-01 1.13075554e+00
-1.21415508e+00 -7.43614733e-01 7.90508911e-02 2.26498485e-01
3.96137297e-01 6.04025871e-02 6.15324318e-01 3.73093218e-01
-7.00743377e-01 3.85181993e-01 5.67170680e-01 -3.10827523e-01
2.03747109e-01 -1.20027900e+00 5.49437463e-01 1.02353299e+00
3.01922351e-01 9.49880362e-01 1.02255070e+00 -7.18548238e-01
-1.35382843e+00 -4.89761740e-01 6.27042890e-01 -5.99601626e-01
8.11670423e-01 -3.18917513e-01 -8.75739753e-01 5.80439091e-01
1.49454335e-02 -2.88531035e-01 5.10299921e-01 4.92565572e-01
-2.08100826e-02 2.33896181e-01 -9.34085786e-01 7.36646891e-01
6.22243524e-01 -4.87193733e-01 -3.43495905e-01 -8.46272241e-03
2.49011248e-01 -4.46380258e-01 -9.62907732e-01 2.20603272e-01
6.59992278e-01 -1.08895898e+00 1.08577251e+00 -3.37797999e-01
4.78188932e-01 -3.06573033e-01 -6.75740764e-02 -1.18015385e+00
8.82879086e-03 -7.83142924e-01 -4.74070221e-01 1.42089999e+00
4.55000132e-01 -7.08764315e-01 9.19772625e-01 7.44226933e-01
-7.28307143e-02 -7.51172125e-01 -1.02413237e+00 -4.61431772e-01
3.22715193e-01 -8.55670929e-01 4.13176149e-01 5.55933654e-01
-4.60611075e-01 2.67791599e-01 -3.55608165e-01 -2.02831492e-01
9.07218575e-01 2.87594441e-02 8.53307486e-01 -1.09051883e+00
-6.25653028e-01 -7.51825571e-01 1.80497453e-01 -1.29173267e+00
1.85754955e-01 -7.18126237e-01 -8.22222885e-03 -1.63217413e+00
3.00616503e-01 -2.35633343e-01 9.66838300e-02 -2.69810250e-03
-2.56047875e-01 8.18597302e-02 4.16834503e-02 9.61340964e-02
-2.62481809e-01 8.87530148e-01 9.96022046e-01 9.24391448e-02
-2.70806879e-01 2.99097039e-02 -4.64228421e-01 8.26336026e-01
6.43729270e-01 -5.11348963e-01 -4.43868279e-01 -3.58493239e-01
2.83225387e-01 2.13750079e-01 5.63206434e-01 -6.46582544e-01
3.16406153e-02 -2.77393401e-01 2.16149494e-01 -9.09416080e-01
6.62501752e-01 -3.44910622e-01 4.88333136e-01 8.29386190e-02
-8.22625160e-02 -2.31578186e-01 1.82785004e-01 4.25761402e-01
1.14983968e-01 -6.64642096e-01 9.54402924e-01 -6.79010153e-02
-4.53452140e-01 1.10067807e-01 -3.96937132e-01 4.48509067e-01
5.13588250e-01 -1.72786504e-01 -1.50235757e-01 -2.04614535e-01
-6.91245973e-01 6.23802505e-02 4.34061468e-01 -5.51309623e-02
1.32873267e-01 -8.84844899e-01 -6.57643378e-01 1.23624898e-01
-3.61296684e-01 -2.87709609e-02 2.88106769e-01 9.32022154e-01
-4.76969451e-01 3.27845454e-01 1.27299324e-01 -7.95210421e-01
-9.15510058e-01 8.65421593e-02 3.64213526e-01 -5.73150694e-01
-4.53617603e-01 8.13798368e-01 -2.68285349e-02 -3.26630831e-01
3.06334436e-01 -2.16227248e-01 3.65660265e-02 3.41169983e-01
2.21158132e-01 7.28984594e-01 -2.60955423e-01 -2.94512719e-01
-2.70674497e-01 3.57390344e-01 1.63000464e-01 -7.32566535e-01
1.16529477e+00 -9.77830216e-02 -1.26476973e-01 7.12169647e-01
9.64045048e-01 -1.81990236e-01 -1.61634171e+00 1.93261340e-01
-1.28496885e-01 -4.15446699e-01 2.21917242e-01 -6.35950863e-01
-8.51774275e-01 1.07169509e+00 3.20680350e-01 8.32805485e-02
5.15799761e-01 -1.40536621e-01 2.17526823e-01 -3.75637375e-02
1.41450226e-01 -1.11169577e+00 -3.31589878e-01 6.19252145e-01
6.16379321e-01 -1.07789946e+00 3.69607151e-01 -1.43835813e-01
-4.51112956e-01 8.22507501e-01 2.64832824e-01 6.86492622e-02
8.11233282e-01 3.35344195e-01 -2.06691206e-01 -2.95915723e-01
-5.55924833e-01 -1.92593098e-01 4.92564052e-01 3.52067530e-01
5.10710418e-01 2.51302756e-02 -4.44948584e-01 2.94199705e-01
-1.89668268e-01 2.98937336e-02 8.81275758e-02 4.98783886e-01
-1.82909295e-01 -1.06009281e+00 -3.25801998e-01 5.07743478e-01
-5.76973319e-01 -2.36338019e-01 1.68110356e-01 8.58143628e-01
-2.21959993e-01 8.60917270e-01 1.38219520e-01 3.81672859e-01
-2.23947585e-01 2.77725071e-01 7.66792774e-01 -3.55520099e-01
-4.02347833e-01 3.49199682e-01 1.39805451e-01 -2.07889467e-01
-4.63751048e-01 -1.07178771e+00 -8.66581380e-01 -4.09970164e-01
-2.66579062e-01 1.55708700e-01 9.70584989e-01 1.27624762e+00
-1.60040576e-02 5.40350318e-01 -2.26681337e-01 -1.27502286e+00
-9.80466962e-01 -9.60549653e-01 -6.15497172e-01 9.76735577e-02
1.32725085e-03 -8.57412398e-01 -5.95265985e-01 -2.15308785e-01] | [6.956808567047119, 3.867215633392334] |
2417aa81-de52-49ef-90b1-0be6b7cb86b6 | uncertainty-aware-contour-proposal-networks | null | null | https://openreview.net/forum?id=YtgRjBw-7GJ | https://openreview.net/pdf?id=YtgRjBw-7GJ | Uncertainty-Aware Contour Proposal Networks for Cell Segmentation in Multi-Modality High-Resolution Microscopy Images | We present a simple framework for cell segmentation, based on uncertainty-aware Contour Proposal Networks (CPNs).
It is designed to provide high segmentation accuracy while remaining computationally efficient, which makes it an ideal solution for high throughput microscopy applications. Each predicted cell is provided with four uncertainty estimations that give information about the localization accuracy of the detected cell boundaries. Such additional insights are valuable for downstream single-cell analysis in microscopy image-based biology and biomedical research. In the context of the NeurIPS 22 Cell Segmentation Challenge, the proposed solution is shown to generalize well in a multi-modality setting, while respecting domain-specific requirements such as focusing on specific cell types. Without an ensemble or test-time augmentation the method achieves an F1 score of 0.8986 on the challenge's validation set. Code and trained models are available at celldetection.org | ['Timo Dickscheid', 'Katrin Amunts', 'Stefan Harmeling', 'Eric Upschulte'] | 2022-11-30 | null | null | null | neurips-cellseg-2022-2022-11 | ['cell-segmentation'] | ['medical'] | [ 4.14674342e-01 2.07128093e-01 -4.47567226e-03 -2.45242700e-01
-1.09117115e+00 -6.56072557e-01 3.90737623e-01 4.91292208e-01
-8.11939478e-01 1.19536245e+00 -5.05114436e-01 -3.34813476e-01
1.85342893e-01 -2.57012516e-01 -7.77250051e-01 -1.09982145e+00
1.08983874e-01 8.83118570e-01 4.41045672e-01 3.35775733e-01
4.63162452e-01 8.12887728e-01 -1.11421335e+00 1.37513265e-01
6.41334116e-01 1.09759092e+00 2.60353744e-01 9.55798268e-01
1.68718323e-01 1.09435879e-01 -3.88160914e-01 -8.54959562e-02
-2.14303788e-02 -2.65099071e-02 -8.35660458e-01 -1.51303679e-01
3.30107182e-01 -4.50065322e-02 2.60646611e-01 7.91860759e-01
5.51274896e-01 -9.38302875e-02 9.24469352e-01 -9.31620002e-01
4.08368297e-02 1.75784677e-01 -3.88373435e-01 3.52104813e-01
-2.30928272e-01 1.66654930e-01 6.99005067e-01 -7.76473820e-01
1.19106925e+00 6.31405294e-01 7.23209381e-01 8.43515098e-01
-1.68811250e+00 -1.29045889e-01 1.93784968e-03 -3.19426507e-01
-1.32479239e+00 -6.38371527e-01 5.33285402e-02 -4.98071402e-01
7.07367897e-01 -3.97103233e-03 4.59773600e-01 8.24711800e-01
5.28788030e-01 7.37387776e-01 1.15760040e+00 -1.19057596e-01
6.83660746e-01 -4.58065383e-02 1.80167139e-01 5.43217540e-01
2.76412219e-01 -3.20222139e-01 -2.63047338e-01 -1.06430024e-01
1.09034646e+00 -1.62583157e-01 -2.97101498e-01 -2.88124800e-01
-1.38764322e+00 3.82168353e-01 1.65985644e-01 4.49262232e-01
5.96820079e-02 2.55305737e-01 3.71494710e-01 -2.88256437e-01
6.09321356e-01 2.95990020e-01 -8.12167108e-01 9.04164314e-02
-1.18708909e+00 4.70572144e-01 6.13497138e-01 8.25719655e-01
5.78829050e-01 -2.98674673e-01 -3.39291096e-01 4.63153958e-01
2.57428318e-01 2.81760275e-01 2.23887280e-01 -1.38845062e+00
-1.85056195e-01 4.71491814e-01 2.52412617e-01 -4.01224047e-01
-1.00275874e+00 -3.34777087e-01 -7.79515147e-01 3.09887081e-01
1.02645445e+00 -3.08942527e-01 -1.26852489e+00 1.68737829e+00
3.68259639e-01 3.08628500e-01 -1.19388282e-01 8.00983489e-01
7.78130949e-01 3.83068681e-01 3.81946424e-03 -4.18743193e-01
1.28890145e+00 -4.97089237e-01 -5.55353403e-01 5.28646521e-02
8.87338459e-01 -3.44296187e-01 6.43207371e-01 3.68414015e-01
-1.00307667e+00 -1.01960883e-01 -7.05102921e-01 -1.48219362e-01
-5.26151419e-01 1.69540718e-01 5.22329867e-01 4.73921239e-01
-1.26957142e+00 6.37185574e-01 -1.05555737e+00 -5.88096142e-01
1.03421950e+00 5.90043128e-01 -3.80805582e-01 1.80018812e-01
-3.10189009e-01 7.36143291e-01 3.16210777e-01 1.51927238e-02
-9.07293081e-01 -1.00432217e+00 -5.44435024e-01 -1.13399230e-01
-5.33425324e-02 -7.37242103e-01 1.11715710e+00 -1.26320139e-01
-1.37254345e+00 1.31552231e+00 -6.11212552e-01 -7.43690968e-01
6.81867659e-01 3.13979238e-01 2.64909416e-01 4.39666748e-01
9.69890878e-02 1.29614615e+00 9.92724225e-02 -1.30565381e+00
-6.88067138e-01 -5.76467514e-01 -4.77778703e-01 -1.77911654e-01
2.10717618e-01 -2.18832165e-01 -6.24494672e-01 -1.95724681e-01
1.08822562e-01 -8.22888494e-01 -5.92916727e-01 2.38280460e-01
-5.31344652e-01 1.92222267e-01 8.92749071e-01 -5.46197891e-01
3.92378390e-01 -1.69913447e+00 5.58661185e-02 2.22522050e-01
2.27447793e-01 1.59967229e-01 4.40849513e-02 -4.52407263e-03
4.24504846e-01 2.41304934e-01 -5.44689834e-01 -6.57220900e-01
-4.22022194e-01 -5.28052002e-02 1.48616865e-01 6.92755282e-01
3.04977477e-01 1.07218778e+00 -6.18498683e-01 -5.62357426e-01
1.48573488e-01 6.38170660e-01 -3.03954363e-01 1.31902145e-02
-5.18308759e-01 1.13548064e+00 -2.49274015e-01 9.06101406e-01
6.36943936e-01 -4.69544768e-01 1.84862286e-01 -1.39600947e-01
-2.26698406e-02 -4.41397130e-01 -6.02006435e-01 1.66468859e+00
-8.80699754e-02 6.08498454e-01 3.75223458e-01 -9.55997348e-01
7.07834601e-01 3.16001892e-01 6.14892006e-01 -2.07147095e-02
3.17662179e-01 3.42714727e-01 -8.73107091e-02 -6.90260604e-02
-6.71799853e-02 -2.03659937e-01 2.92019218e-01 1.05300635e-01
3.34303528e-01 -3.08366865e-01 4.61916715e-01 2.30977356e-01
1.11625993e+00 4.56844479e-01 -1.08822994e-03 -6.69550121e-01
4.17167276e-01 8.83526132e-02 7.28996217e-01 4.72347587e-01
-5.75007558e-01 9.45817530e-01 6.35152280e-01 -2.84208596e-01
-1.18915832e+00 -9.59454656e-01 -5.73468804e-01 5.35318911e-01
2.10028499e-01 2.08964527e-01 -9.95412767e-01 -4.45557266e-01
-1.48061011e-02 2.16901481e-01 -6.12680793e-01 4.18478549e-01
-1.95865899e-01 -9.56963062e-01 6.04030073e-01 3.83204937e-01
1.15433753e-01 -8.25714171e-01 -7.48227715e-01 2.08485603e-01
-1.50443017e-01 -1.46588027e+00 -1.36830762e-01 3.78315598e-01
-1.06448662e+00 -1.18127072e+00 -1.18885005e+00 -8.65467072e-01
9.18685317e-01 -1.70695767e-01 7.10161448e-01 -1.79761683e-03
-5.98393261e-01 2.52068549e-01 1.06341183e-01 -3.76704037e-01
-9.14289206e-02 1.07554697e-01 -7.60868937e-02 -3.69829148e-01
1.89861238e-01 -4.97486174e-01 -6.86842978e-01 1.88765049e-01
-7.61073530e-01 5.20311408e-02 2.74953634e-01 1.09049666e+00
1.20567751e+00 -3.11484247e-01 6.74335063e-01 -1.13684309e+00
2.61576295e-01 -1.38661668e-01 -8.37102830e-01 1.52698353e-01
-5.13225198e-01 -3.11530411e-01 5.33895791e-01 -1.81514174e-01
-9.96087134e-01 2.64275998e-01 -3.24739754e-01 1.73667946e-03
-5.96541584e-01 3.22297990e-01 1.41501606e-01 -4.33615446e-01
4.70362633e-01 4.88838479e-02 2.95899272e-01 -8.69066790e-02
-1.32934943e-01 1.92960173e-01 6.24675155e-01 -5.17410398e-01
6.41176850e-02 9.09639895e-01 6.04447305e-01 -7.47928321e-01
-6.34388745e-01 -7.03296661e-01 -6.75604820e-01 -2.41372794e-01
8.97097886e-01 -6.22371137e-01 -1.10305989e+00 5.66482604e-01
-1.10702634e+00 -6.61806285e-01 6.64633363e-02 2.16970190e-01
-9.62239683e-01 2.38340482e-01 -9.50417221e-01 -9.05376792e-01
-3.08314651e-01 -1.25953877e+00 1.36687064e+00 5.87782145e-01
-7.32546449e-02 -1.21114302e+00 -1.01383962e-02 3.71394902e-01
2.61381119e-01 4.42844540e-01 9.66320038e-01 -8.77926528e-01
-6.30632520e-01 -1.29535273e-01 -2.68311173e-01 -1.05711967e-01
-1.32893264e-01 4.05857146e-01 -1.25411904e+00 -2.06048518e-01
-5.13568878e-01 -4.22907829e-01 1.29270720e+00 1.18096852e+00
1.26893187e+00 1.82487249e-01 -9.52602327e-01 5.57787836e-01
1.43407428e+00 1.62659083e-02 4.53484595e-01 1.15007147e-01
2.94142842e-01 6.79596126e-01 3.87576342e-01 2.39915222e-01
1.73079431e-01 3.09456080e-01 5.42579532e-01 -1.62928596e-01
1.02315813e-01 2.38911211e-01 -2.94483572e-01 -2.97004580e-02
-1.00470327e-01 -5.29759049e-01 -1.10001719e+00 6.86333179e-01
-2.03776550e+00 -5.89323699e-01 -2.81643957e-01 2.06090498e+00
7.43795514e-01 1.46882489e-01 2.37938970e-01 -1.16835073e-01
7.01007664e-01 -3.97035539e-01 -7.86866128e-01 -6.91038221e-02
-3.34592909e-01 1.45671830e-01 5.41089296e-01 6.09265745e-01
-1.07494998e+00 9.89750862e-01 7.05432606e+00 9.27847803e-01
-8.86528254e-01 -1.43516988e-01 1.51177990e+00 1.12305515e-01
-1.32475600e-01 4.28333767e-02 -1.12263680e+00 3.73641074e-01
9.22757030e-01 1.23396844e-01 3.03690135e-02 1.04751043e-01
4.25267577e-01 -5.90505779e-01 -1.09423470e+00 6.49657130e-01
-4.40278858e-01 -2.02524042e+00 -8.17614645e-02 3.00314844e-01
7.11475551e-01 2.62002826e-01 9.77774262e-02 -1.99025720e-01
7.54951090e-02 -1.17797256e+00 2.08931059e-01 6.90738022e-01
8.24169159e-01 -6.69528902e-01 1.18732762e+00 6.08851373e-01
-5.60882151e-01 3.09969038e-01 -3.64521116e-01 1.81152686e-01
3.90021354e-01 8.55063975e-01 -1.07198560e+00 1.94691345e-01
5.86509943e-01 4.48678166e-01 -2.38418519e-01 1.23728704e+00
3.99258077e-01 5.99231124e-01 -6.17223501e-01 7.13235736e-02
2.02575270e-02 -2.20021456e-01 4.20690864e-01 1.36155164e+00
3.60566020e-01 1.84758101e-02 -7.83298314e-02 1.13813877e+00
-2.24379942e-01 -7.61120915e-02 -4.44552392e-01 -1.18267179e-01
3.93345028e-01 1.79109466e+00 -1.46254039e+00 -1.70675293e-01
1.99103028e-01 6.29343569e-01 3.90567750e-01 5.03562987e-01
-4.53847796e-01 -9.07894447e-02 3.70796204e-01 2.94038597e-02
2.65432358e-01 -1.33958235e-01 -6.83365107e-01 -7.63764381e-01
-3.63013178e-01 -8.39486197e-02 5.94129935e-02 -5.82131028e-01
-1.08562064e+00 8.49730670e-02 -3.79035681e-01 -6.56591773e-01
7.58250877e-02 -1.19151628e+00 -5.39066732e-01 7.35003173e-01
-1.49878848e+00 -1.32051802e+00 3.82463299e-02 -1.22631148e-01
1.78849697e-01 5.78560084e-02 9.59564388e-01 -1.30990371e-01
-5.93537152e-01 3.51042330e-01 2.83362240e-01 -1.52640408e-02
4.82033968e-01 -1.32664263e+00 2.20012709e-01 6.61066175e-01
-5.97242638e-02 6.16493344e-01 8.57139647e-01 -5.80008686e-01
-8.22607160e-01 -1.00627792e+00 7.19541311e-01 -5.25540769e-01
4.13163990e-01 -2.71451265e-01 -8.56429160e-01 4.90713298e-01
-1.45439938e-01 3.81745964e-01 8.82143319e-01 -1.14154495e-01
2.09906191e-01 1.50804460e-01 -1.66552687e+00 6.78571463e-01
7.58606136e-01 -1.99519210e-02 2.04032987e-01 4.98190373e-01
3.62394750e-01 -6.93923593e-01 -1.15110207e+00 5.93413949e-01
6.11367643e-01 -1.01036227e+00 6.88926637e-01 -4.15201694e-01
3.74682486e-01 -3.19313705e-01 2.77623665e-02 -9.16352987e-01
-9.82534140e-02 -5.15068054e-01 -9.65427905e-02 1.11338615e+00
7.58412421e-01 -5.83655894e-01 1.31112862e+00 5.59575319e-01
-2.11231083e-01 -1.14340043e+00 -1.38475788e+00 -6.65065765e-01
3.88904184e-01 -1.07812807e-01 8.27422272e-03 4.21175748e-01
1.11474648e-01 1.59102753e-01 1.89701989e-01 5.78128770e-02
8.44252050e-01 -9.17483345e-02 5.47991037e-01 -1.42278874e+00
1.50827274e-01 -6.69799209e-01 -4.49398756e-01 -9.76570427e-01
2.79859960e-01 -6.41468704e-01 3.04091930e-01 -1.46039963e+00
3.36817205e-01 -3.72494221e-01 -3.42036396e-01 2.48444840e-01
1.55320112e-02 5.19918144e-01 -2.30949312e-01 8.62782225e-02
-6.42282605e-01 8.37603882e-02 1.18852258e+00 -3.62747423e-02
6.69603720e-02 4.63094860e-02 -4.86096144e-01 6.59636199e-01
8.95940363e-01 -1.56731993e-01 -1.49025455e-01 -1.00863703e-01
-1.62887722e-01 1.04521461e-01 4.68066394e-01 -1.14017403e+00
4.04052645e-01 -1.03769697e-01 7.45145261e-01 -6.28803790e-01
5.58803022e-01 -6.33870423e-01 -4.54343483e-03 4.49038744e-01
-4.63372916e-01 -6.11424625e-01 4.20494378e-01 7.63242483e-01
7.16965050e-02 -1.55106544e-01 1.06647384e+00 -2.41040990e-01
-3.46160978e-01 4.35950786e-01 -6.49705768e-01 -2.44299974e-02
1.23057079e+00 -5.17701626e-01 -4.36566561e-01 -1.97385088e-01
-9.49086785e-01 3.76412898e-01 8.46951544e-01 -5.00285923e-01
5.06026447e-01 -7.44099259e-01 -4.48535442e-01 6.11034781e-02
2.54383355e-01 3.51565927e-01 3.21331710e-01 1.05284393e+00
-6.80725574e-01 7.25577235e-01 -2.23761544e-01 -1.07906556e+00
-1.07785332e+00 6.40531480e-02 5.58566213e-01 -4.15484339e-01
-2.88975626e-01 1.16480994e+00 1.32458270e-01 -6.23709559e-01
1.12070842e-02 -4.79074508e-01 -2.68882036e-01 -4.25510019e-01
3.42716396e-01 2.28101432e-01 3.08977421e-02 -5.44279397e-01
-4.22206581e-01 6.01372004e-01 -2.06316784e-01 -1.28657758e-01
1.19768691e+00 -2.52007931e-01 -1.84701607e-01 5.89732170e-01
8.77494931e-01 -4.60075378e-01 -1.66832662e+00 1.26894012e-01
2.09629521e-01 1.95955217e-01 1.05821639e-01 -9.81424510e-01
-8.06704640e-01 8.67373765e-01 4.72963721e-01 -1.71977490e-01
7.33274400e-01 6.75320700e-02 6.87934816e-01 4.09999341e-01
4.70274568e-01 -1.09119463e+00 -4.24840510e-01 5.35993695e-01
2.37230837e-01 -1.42078233e+00 -4.94013168e-02 -4.63532120e-01
-3.94771159e-01 1.05576992e+00 6.14519000e-01 1.53679043e-01
5.66181540e-01 5.53128600e-01 1.62867889e-01 -1.25927150e-01
-9.03885305e-01 -1.11501351e-01 -7.03147277e-02 1.07095456e+00
4.98168856e-01 -1.81694105e-01 -2.07289457e-01 5.16506016e-01
5.02260685e-01 1.95487648e-01 5.33352613e-01 8.49623322e-01
-7.78090835e-01 -8.13305914e-01 9.14373547e-02 6.41941607e-01
-7.20215082e-01 1.33377880e-01 -3.99908900e-01 5.43799639e-01
-1.34197026e-02 6.79094493e-01 2.36477166e-01 1.71180561e-01
-3.36510420e-01 6.84747472e-02 6.33254468e-01 -6.09002650e-01
-1.83410957e-01 3.54712129e-01 -2.35437244e-01 -3.73045564e-01
-5.82947373e-01 -5.90774953e-01 -2.00352955e+00 -1.03435539e-01
-2.21763879e-01 1.09942004e-01 4.97215062e-01 1.10239613e+00
5.80087900e-01 3.13112110e-01 -1.12425976e-01 -1.15795374e+00
1.85645238e-01 -7.73067832e-01 -7.43093848e-01 3.11723445e-02
3.59437168e-01 -3.93911541e-01 -2.27929309e-01 4.88781512e-01] | [14.516863822937012, -3.114067554473877] |
f66a726a-801c-414f-9f41-11b95ba23773 | deepwriting-making-digital-ink-editable-via | 1801.08379 | null | http://arxiv.org/abs/1801.08379v1 | http://arxiv.org/pdf/1801.08379v1.pdf | DeepWriting: Making Digital Ink Editable via Deep Generative Modeling | Digital ink promises to combine the flexibility and aesthetics of handwriting
and the ability to process, search and edit digital text. Character recognition
converts handwritten text into a digital representation, albeit at the cost of
losing personalized appearance due to the technical difficulties of separating
the interwoven components of content and style. In this paper, we propose a
novel generative neural network architecture that is capable of disentangling
style from content and thus making digital ink editable. Our model can
synthesize arbitrary text, while giving users control over the visual
appearance (style). For example, allowing for style transfer without changing
the content, editing of digital ink at the word level and other application
scenarios such as spell-checking and correction of handwritten text. We
furthermore contribute a new dataset of handwritten text with fine-grained
annotations at the character level and report results from an initial user
evaluation. | ['Fabrizio Pece', 'Emre Aksan', 'Otmar Hilliges'] | 2018-01-25 | null | null | null | null | ['handwritten-word-generation', 'handwriting-generation'] | ['computer-vision', 'computer-vision'] | [ 5.92351675e-01 -5.80315292e-02 2.92147905e-01 -2.12452292e-01
-1.17897697e-01 -1.23525691e+00 6.37495816e-01 3.93971987e-02
-1.16522104e-01 4.43983525e-01 9.41662937e-02 -3.58667940e-01
4.95595932e-02 -8.40191722e-01 -5.41003108e-01 -4.46409255e-01
4.55008775e-01 4.44242418e-01 1.30063698e-01 -1.13919392e-01
3.75701517e-01 1.06283975e+00 -1.33311689e+00 5.35781205e-01
9.03798521e-01 8.57146382e-01 3.69446009e-01 1.11997378e+00
-4.31227237e-01 5.29169500e-01 -8.37402105e-01 -6.13069832e-01
2.65596509e-01 -5.12743115e-01 -2.81741560e-01 4.81580198e-01
1.05182052e+00 -5.33295929e-01 -4.91150618e-01 8.30605090e-01
3.94535542e-01 -2.00562552e-01 6.78862095e-01 -8.30512524e-01
-1.49757671e+00 2.98749924e-01 -2.76245147e-01 -6.01434946e-01
2.60781735e-01 1.88217849e-01 7.49399304e-01 -6.44713223e-01
8.99341941e-01 1.02233875e+00 6.44845486e-01 8.27964544e-01
-1.38616335e+00 -3.30119878e-01 2.25048941e-02 -3.53258997e-01
-8.95683229e-01 -3.51642132e-01 6.05548561e-01 -5.53057969e-01
6.63884342e-01 6.70354664e-01 7.39274025e-01 1.03929007e+00
2.25571290e-01 8.98819149e-01 1.08821046e+00 -5.41007936e-01
1.41662121e-01 1.97384834e-01 -1.90037012e-01 6.45497322e-01
8.17926899e-02 -5.91749586e-02 -4.57521230e-01 2.10392877e-01
1.26359510e+00 -6.65735453e-02 -3.03141296e-01 -4.14274603e-01
-1.00357163e+00 2.76611090e-01 1.88831091e-01 1.79501455e-02
3.26590687e-02 2.99335450e-01 1.87585548e-01 3.30347717e-01
2.06191931e-02 8.59014392e-01 -6.16716035e-02 -2.84046739e-01
-1.14244080e+00 2.40847608e-03 8.10053706e-01 1.17868662e+00
3.66440296e-01 3.18591654e-01 -5.72826207e-01 8.65639091e-01
8.75618979e-02 6.89363062e-01 4.13421720e-01 -7.26950526e-01
3.16903144e-01 6.29747033e-01 2.53678292e-01 -8.78585756e-01
-4.92936559e-02 9.95788053e-02 -7.55791843e-01 1.00557244e+00
5.35234869e-01 -1.46872178e-01 -1.20825255e+00 1.14233625e+00
-2.92564631e-01 -7.89907098e-01 -3.64194036e-01 8.25427175e-01
5.40253818e-01 4.60108191e-01 -4.74198088e-02 4.06915814e-01
1.23305857e+00 -8.90108943e-01 -7.57875264e-01 -2.86431581e-01
-1.53126135e-01 -9.87331629e-01 1.32658696e+00 5.74911475e-01
-1.27969015e+00 -5.57094455e-01 -1.32669508e+00 -4.52648222e-01
-3.56819719e-01 6.68697953e-01 5.10837913e-01 6.88184202e-01
-1.11417544e+00 8.37362647e-01 -6.91285431e-01 -1.99691474e-01
4.34467077e-01 2.93762833e-01 -3.80056828e-01 3.26998740e-01
-6.64176166e-01 6.53465033e-01 1.22399911e-01 -5.10608107e-02
-3.32184821e-01 -7.26033151e-01 -4.37053502e-01 1.10193104e-01
4.08160617e-04 -4.57741886e-01 1.12142861e+00 -1.37118042e+00
-2.03259516e+00 8.01407099e-01 2.91120917e-01 8.24829787e-02
1.28336334e+00 -2.77103573e-01 -4.89801496e-01 -5.62236868e-02
-4.32706147e-01 6.89743817e-01 1.27431846e+00 -1.18379951e+00
-4.89010274e-01 -8.83840919e-02 -9.80521142e-02 -1.93385016e-02
-5.18024147e-01 -3.63510907e-01 -6.98386729e-01 -1.20730758e+00
-2.38310739e-01 -8.40607226e-01 2.49209210e-01 6.45754158e-01
-5.17824471e-01 4.26556557e-01 1.24123347e+00 -9.18526411e-01
1.00117993e+00 -2.27674556e+00 3.53108615e-01 1.93113461e-01
2.77357221e-01 4.16856647e-01 -2.53125697e-01 5.24151444e-01
1.78769872e-01 1.70223296e-01 -8.92886966e-02 -3.79988909e-01
1.17099471e-01 -1.33497501e-02 -3.59066427e-01 -2.52859183e-02
3.15907270e-01 9.77907121e-01 -6.85646117e-01 -1.51062608e-01
1.12371989e-01 4.92857367e-01 -3.38698119e-01 1.20721444e-01
-5.11412740e-01 8.55476782e-02 -1.26503155e-01 6.54631913e-01
6.39132857e-01 5.21873217e-03 4.20020878e-01 -3.26510757e-01
-2.24081457e-01 -2.00865373e-01 -1.16648412e+00 1.44815612e+00
-6.73743188e-01 1.32347572e+00 2.54510254e-01 2.90648520e-01
1.03283048e+00 -2.68527940e-02 -1.92745492e-01 -7.54855692e-01
6.27831444e-02 1.07301362e-01 -2.14693546e-01 -2.39287525e-01
1.23211002e+00 4.86213667e-03 9.78594199e-02 8.98183644e-01
-2.52865106e-01 -3.79695266e-01 5.07416874e-02 -1.57180987e-02
6.62168086e-01 3.14257056e-01 -2.29222253e-01 -1.62826627e-01
-1.89145014e-01 -2.70922750e-01 -3.10008824e-01 7.37109303e-01
3.00375313e-01 1.03867877e+00 4.88925695e-01 -3.46709549e-01
-1.49560821e+00 -1.09927261e+00 -8.60326272e-03 1.13786578e+00
-5.20650446e-02 -6.83596283e-02 -7.24434137e-01 -5.23729384e-01
3.35027099e-01 7.59267092e-01 -6.86195910e-01 3.83970328e-02
-6.40231192e-01 -1.66904837e-01 7.65525162e-01 7.49589145e-01
3.39262307e-01 -9.63464379e-01 -7.04594135e-01 1.51724622e-01
3.56757522e-01 -7.89254665e-01 -8.53344977e-01 1.42512321e-01
-7.21245468e-01 -6.59718871e-01 -1.01818752e+00 -6.71127081e-01
9.60885644e-01 -2.01497450e-01 8.50805700e-01 4.43999693e-02
-6.32795274e-01 2.96281129e-01 -1.76173970e-01 -1.29849598e-01
-9.02582407e-01 1.04540437e-01 -2.67266244e-01 7.00008795e-02
-5.20719647e-01 -4.19569314e-01 -4.57030445e-01 1.79553702e-01
-1.20708454e+00 3.95276636e-01 6.63581789e-01 7.23018646e-01
3.57201487e-01 -2.98866838e-01 1.17880486e-01 -9.80427980e-01
1.02531481e+00 3.16407859e-01 -5.43129563e-01 5.58192134e-01
-4.83357698e-01 2.57198781e-01 7.57315040e-01 -7.14172125e-01
-1.18770921e+00 3.93742174e-02 1.16409577e-01 -2.12362498e-01
-8.43920708e-02 -3.53092328e-02 -2.05989048e-01 -2.83987552e-01
6.31037831e-01 2.12312788e-01 3.58562879e-02 -6.75998509e-01
6.99724138e-01 8.62077951e-01 8.26917529e-01 -4.38310534e-01
6.63944840e-01 2.69255489e-01 -1.61134332e-01 -7.60985792e-01
-6.59677833e-02 4.75311548e-01 -8.15642297e-01 -1.49619102e-01
5.77413440e-01 -2.87765145e-01 -6.75520062e-01 9.37805831e-01
-1.03053093e+00 -7.34189749e-01 -5.53489506e-01 -2.15207636e-01
-4.69427109e-01 4.83199894e-01 -8.49137127e-01 -5.78637898e-01
-2.67219812e-01 -8.87297451e-01 1.02014840e+00 3.75951409e-01
-5.21738410e-01 -8.75270724e-01 -2.35710829e-01 -7.87931457e-02
6.51919901e-01 2.60401368e-01 1.23895431e+00 -1.12651087e-01
-6.50094986e-01 -6.60560071e-01 -5.49353004e-01 2.77529806e-01
4.97063577e-01 4.87483740e-01 -9.95583713e-01 -2.19392061e-01
-7.17972815e-01 -1.22458413e-01 7.68076658e-01 -1.09398626e-01
1.17319036e+00 -4.14945543e-01 -4.49299850e-02 6.33260667e-01
1.25770903e+00 3.77495706e-01 8.71164143e-01 3.87610495e-01
1.04262817e+00 1.46155581e-01 -6.44100308e-02 4.78547245e-01
-1.27106354e-01 7.78143942e-01 8.72597471e-02 -1.77543253e-01
-7.05270886e-01 -5.14417768e-01 1.05777636e-01 2.52492487e-01
-2.22832039e-01 -6.35965347e-01 -6.38522029e-01 1.96992159e-01
-1.36732876e+00 -7.60502219e-01 7.02298060e-02 2.08822989e+00
1.08036101e+00 7.14133531e-02 -1.65557023e-02 1.90924779e-02
6.65650368e-01 -1.01064458e-01 -6.50650382e-01 -8.85038257e-01
-1.76749006e-01 1.31109968e-01 7.02874899e-01 6.21477365e-01
-8.19330931e-01 9.93918836e-01 6.89389181e+00 6.69876873e-01
-1.60413313e+00 -4.54048634e-01 3.82269740e-01 -1.98302314e-01
-5.11966527e-01 -4.25613403e-01 -7.00525045e-01 4.89254147e-01
1.61352813e-01 1.03980400e-01 8.50714386e-01 4.18144256e-01
-6.83729723e-02 2.72864336e-03 -1.34405482e+00 8.42494965e-01
1.39082983e-01 -1.55044794e+00 4.87765223e-01 -1.18808642e-01
7.64322460e-01 -5.46161830e-01 4.67553705e-01 -3.69526267e-01
2.48899028e-01 -1.23578358e+00 1.29301286e+00 9.49159741e-01
1.54477119e+00 -4.30281013e-01 1.32988021e-01 -4.11738269e-02
-5.69289327e-01 -4.01294343e-02 1.03297263e-01 1.18814170e-01
-1.16461180e-01 2.32025281e-01 -7.97499180e-01 1.05766639e-01
2.10572526e-01 3.42084438e-01 -8.36400747e-01 1.01575220e+00
-2.35007092e-01 5.82262501e-02 -2.72677951e-02 -2.16408923e-01
7.62978336e-03 -8.42473432e-02 3.22537482e-01 1.52223265e+00
4.64582890e-01 -1.72410101e-01 -2.36746982e-01 1.16474056e+00
-2.94358790e-01 -2.65879214e-01 -1.79012462e-01 -7.25692153e-01
4.53858018e-01 1.13805032e+00 -8.24905634e-01 -2.23626360e-01
1.51418731e-01 1.66521335e+00 2.32207030e-01 4.42533016e-01
-4.64730203e-01 -8.90777588e-01 4.35515344e-01 2.19826251e-01
4.03287917e-01 -6.10038102e-01 -7.14194775e-01 -9.04454410e-01
5.85195161e-02 -8.84467781e-01 -4.41969246e-01 -9.26485181e-01
-1.06098926e+00 5.66203952e-01 -6.76872969e-01 -9.54866886e-01
5.35944812e-02 -9.20601428e-01 -5.92114329e-01 1.13926470e+00
-8.74387026e-01 -1.40171266e+00 -5.50101876e-01 4.87183295e-02
5.10641158e-01 -1.07430272e-01 8.28791857e-01 1.85084835e-01
-2.43810430e-01 9.16841388e-01 5.13803840e-01 3.24369669e-01
9.38421071e-01 -1.49585903e+00 7.84995317e-01 4.83377635e-01
3.76660265e-02 4.80441809e-01 6.46741152e-01 -7.66044140e-01
-1.54777360e+00 -8.82057428e-01 5.57858288e-01 -4.99363810e-01
4.31771100e-01 -5.80670774e-01 -7.13704228e-01 4.49877560e-01
1.41573444e-01 -4.92377758e-01 5.66457391e-01 -1.65632948e-01
-4.67007846e-01 -2.33295728e-02 -1.18643534e+00 9.20511663e-01
7.43556499e-01 -8.10097456e-01 -2.09161893e-01 -7.99292848e-02
2.16630265e-01 -7.09796309e-01 -6.17112815e-01 -2.60795802e-01
1.23604858e+00 -8.71170104e-01 7.35321224e-01 -4.45863605e-01
7.74871826e-01 -2.80965596e-01 4.73888777e-02 -1.42029643e+00
-5.96988201e-01 -7.00644910e-01 -3.77192348e-02 1.44805562e+00
4.91032869e-01 -1.25929892e-01 7.58676887e-01 9.69506264e-01
-4.84870896e-02 -3.58670712e-01 -3.48487586e-01 -6.48188353e-01
-7.06462413e-02 -1.84441254e-01 6.92523122e-01 7.11704373e-01
7.39223137e-02 -1.01187862e-01 -6.25520766e-01 -1.16721876e-01
2.76789844e-01 2.17930689e-01 4.68753070e-01 -1.03092039e+00
-6.70942366e-01 -8.60621810e-01 -3.45377743e-01 -9.41728055e-01
-3.77624005e-01 -7.84302294e-01 -5.07946014e-02 -1.59755158e+00
-9.94943604e-02 -5.99951863e-01 3.38610172e-01 5.00066578e-01
9.38120186e-02 5.18537998e-01 6.49561048e-01 1.62188977e-01
-9.81907248e-02 1.47502258e-01 1.65205765e+00 -3.17368925e-01
-2.86956966e-01 -1.63663313e-01 -6.44692421e-01 4.72708672e-01
6.44529223e-01 2.07018420e-01 -2.11348251e-01 -9.80458498e-01
2.66076297e-01 -1.80734061e-02 3.00410777e-01 -7.98972726e-01
1.88765034e-01 -2.10084736e-01 1.01957321e+00 -2.12889522e-01
1.63606063e-01 -7.55160272e-01 1.68230236e-01 2.65888572e-01
-6.90353751e-01 2.05409974e-02 4.19783860e-01 5.43848217e-01
9.29519162e-02 -3.09643179e-01 9.07723725e-01 -4.01168726e-02
-6.66591823e-01 -6.15295582e-02 -5.01244366e-01 -3.39205086e-01
7.77881145e-01 -5.57002544e-01 -4.34410214e-01 -3.32268208e-01
-8.76570702e-01 -2.73361325e-01 1.08190358e+00 6.47437930e-01
5.49413085e-01 -1.28234196e+00 -2.99282104e-01 6.32052541e-01
-2.30942462e-02 -4.65781838e-01 1.10093161e-01 3.49344313e-02
-1.16694009e+00 6.84059188e-02 -6.96284771e-01 -1.00422360e-01
-1.33060706e+00 3.75096142e-01 3.99054617e-01 2.03980625e-01
-6.37737632e-01 7.39967108e-01 -1.19320385e-01 -5.42079285e-03
4.53033119e-01 -3.26657534e-01 2.45621130e-01 1.41331896e-01
6.72823012e-01 4.75514978e-01 1.93083614e-01 2.61394009e-02
1.37056857e-01 6.93006456e-01 -1.54968128e-01 -9.26356837e-02
1.08162737e+00 -7.93269742e-03 -1.16838835e-01 3.40436429e-01
8.03555608e-01 4.26335692e-01 -1.73862517e+00 2.63035476e-01
-2.54731447e-01 -7.03593850e-01 -1.61695540e-01 -1.52858698e+00
-8.85192692e-01 9.38607693e-01 7.55790770e-01 2.71060646e-01
8.83807302e-01 -4.45723474e-01 7.01166928e-01 4.61284965e-01
9.05375034e-02 -1.44447899e+00 2.00457782e-01 5.12344897e-01
1.05451298e+00 -7.27703929e-01 -1.41354680e-01 -1.73831791e-01
-7.72043943e-01 1.67357516e+00 4.76700842e-01 -1.78723633e-02
6.65225983e-02 8.53888929e-01 2.24048406e-01 1.29472464e-01
-2.72651851e-01 4.36884910e-01 5.62682748e-01 6.57503545e-01
5.92397094e-01 3.42029661e-01 1.25810698e-01 3.89455438e-01
-2.70462185e-01 1.89932302e-01 6.68503046e-01 8.38762224e-01
-3.89359921e-01 -1.34365761e+00 -3.03153515e-01 4.31036711e-01
-1.24296509e-01 -9.76395234e-02 -9.69326615e-01 4.36053485e-01
1.13896288e-01 4.14254218e-01 2.87582606e-01 -3.04654658e-01
3.56881708e-01 1.95278913e-01 7.81748831e-01 -2.71445304e-01
-6.15057230e-01 -8.74071568e-02 1.52445529e-02 -9.22125354e-02
3.73616487e-01 -4.43670690e-01 -9.07947361e-01 -3.65892708e-01
-3.16423401e-02 -3.88480633e-01 8.07861388e-01 5.45647979e-01
4.59139019e-01 6.41687870e-01 2.81604409e-01 -9.56300676e-01
-3.03436905e-01 -6.51846886e-01 -8.21312845e-01 4.42923814e-01
5.04629970e-01 -1.07769771e-02 1.55239567e-01 4.20189291e-01] | [11.642932891845703, -0.054519400000572205] |
95135a80-62e7-41c0-a2a4-597ec02c55be | low-complexity-deep-video-compression-with-a | 2303.11599 | null | https://arxiv.org/abs/2303.11599v2 | https://arxiv.org/pdf/2303.11599v2.pdf | Low-complexity Deep Video Compression with A Distributed Coding Architecture | Prevalent predictive coding-based video compression methods rely on a heavy encoder to reduce temporal redundancy, which makes it challenging to deploy them on resource-constrained devices. Since the 1970s, distributed source coding theory has indicated that independent encoding and joint decoding with side information (SI) can achieve high-efficient compression of correlated sources. This has inspired a distributed coding architecture aiming at reducing the encoding complexity. However, traditional distributed coding methods suffer from a substantial performance gap to predictive coding ones. Inspired by the great success of learning-based compression, we propose the first end-to-end distributed deep video compression framework to improve the rate-distortion performance. A key ingredient is an effective SI generation module at the decoder, which helps to effectively exploit inter-frame correlations without computation-intensive encoder-side motion estimation and compensation. Experiments show that our method significantly outperforms conventional distributed video coding and H.264. Meanwhile, it enjoys 6-7x encoding speedup against DVC [1] with comparable compression performance. Code is released at https://github.com/Xinjie-Q/Distributed-DVC. | ['Jun Zhang', 'Jiawei Shao', 'Xinjie Zhang'] | 2023-03-21 | null | null | null | null | ['motion-estimation'] | ['computer-vision'] | [ 1.38191476e-01 -2.78976738e-01 -4.56905425e-01 -2.10544407e-01
-9.24749255e-01 7.26753324e-02 1.49902269e-01 -1.64949875e-02
8.29546228e-02 4.56249088e-01 6.23492956e-01 -1.81967914e-01
-1.55853242e-01 -7.60753989e-01 -7.31386542e-01 -7.93136537e-01
-4.75429952e-01 4.22759838e-02 2.74096221e-01 3.41021866e-02
2.53002644e-01 -1.82353824e-01 -1.46147418e+00 5.10880709e-01
6.25995159e-01 1.37360466e+00 8.10729325e-01 7.94314265e-01
2.63788085e-02 1.48250771e+00 -3.35869819e-01 -2.75440276e-01
2.73458838e-01 -8.96525860e-01 -4.17413712e-01 1.13010593e-02
-6.73773140e-02 -9.39177930e-01 -1.05993152e+00 1.11515963e+00
5.64031363e-01 -2.14188024e-01 3.30463827e-01 -1.11015356e+00
-6.91596746e-01 8.18313718e-01 -5.88478506e-01 1.95112050e-01
1.56056404e-01 4.47690003e-02 1.15181243e+00 -5.65006256e-01
4.12592232e-01 1.01774538e+00 3.63349706e-01 2.73689061e-01
-8.25224400e-01 -6.20966554e-01 -4.34615374e-01 7.56348014e-01
-1.59754741e+00 -7.74837852e-01 9.70456421e-01 7.38603175e-02
9.51177657e-01 1.52479974e-03 6.03707075e-01 8.25900733e-01
4.35261428e-01 9.35359657e-01 3.71339053e-01 -2.87717521e-01
6.07154906e-01 -7.54346430e-01 -7.67824709e-01 4.48002964e-01
2.27703050e-01 1.47477105e-01 -8.70131016e-01 3.07104271e-02
8.39516401e-01 4.12738353e-01 -4.82272506e-01 -2.43618190e-01
-1.10671067e+00 8.41680765e-01 3.98406565e-01 2.86434233e-01
-5.81389010e-01 6.33923829e-01 8.06007445e-01 3.85940820e-01
4.37381238e-01 -4.26198602e-01 -3.13112974e-01 -6.18869424e-01
-1.39151061e+00 1.36889890e-02 6.78174078e-01 1.36511207e+00
5.23323119e-01 2.20450744e-01 4.17567007e-02 1.01929986e+00
4.48308051e-01 6.79808319e-01 5.97583175e-01 -1.50842953e+00
7.58627832e-01 1.54469445e-01 -3.78336906e-01 -1.33965015e+00
2.61709094e-01 -3.81697237e-01 -1.50193024e+00 -2.41419181e-01
-4.07860339e-01 4.17501591e-02 -3.50693196e-01 1.26892543e+00
-6.86683506e-02 3.60514313e-01 -7.42927343e-02 9.51062262e-01
3.64770651e-01 1.10829926e+00 -2.71965265e-01 -5.30671954e-01
8.07967782e-01 -1.19090140e+00 -8.69744837e-01 7.59951863e-03
6.85969353e-01 -7.36987710e-01 3.48484665e-01 5.71833014e-01
-1.50853646e+00 -5.96207559e-01 -1.16321361e+00 -2.40913942e-01
5.49244225e-01 -1.37527213e-01 4.43177015e-01 3.40614110e-01
-1.15766644e+00 6.62909806e-01 -1.05408990e+00 3.63088787e-01
8.38121951e-01 2.32260153e-01 9.68075451e-03 -6.37564301e-01
-8.85988057e-01 1.26435012e-01 3.29262018e-01 -4.53676879e-01
-1.24271476e+00 -5.41349649e-01 -6.15890563e-01 4.21217710e-01
4.05560046e-01 -6.24038339e-01 1.23230076e+00 -1.03894889e+00
-1.50613976e+00 1.25543088e-01 -3.90109360e-01 -8.28691721e-01
1.67001098e-01 -3.06386769e-01 -2.72720248e-01 6.92657113e-01
-1.22643217e-01 4.05963570e-01 9.58704352e-01 -8.33826602e-01
-6.48462713e-01 -8.80428106e-02 -3.52355778e-01 -3.25225927e-02
-6.22795761e-01 2.28327699e-02 -7.16513276e-01 -7.75888443e-01
1.08822241e-01 -5.40629327e-01 -2.18238831e-01 2.07105517e-01
-1.09563515e-01 -5.21029644e-02 9.62457061e-01 -7.26118922e-01
1.56625915e+00 -2.33589172e+00 2.46933252e-01 -1.24164619e-01
7.19343007e-01 3.42242032e-01 -1.89200521e-01 6.36390984e-01
3.06766540e-01 -4.17863131e-02 -1.44806460e-01 -6.05378807e-01
-2.08015978e-01 1.88716576e-01 -2.63848424e-01 5.77488720e-01
-3.36616009e-01 7.33772218e-01 -8.22713554e-01 -6.79067850e-01
1.56177759e-01 7.48428285e-01 -1.03651106e+00 4.20213401e-01
-2.18469188e-01 2.61395156e-01 -4.05896991e-01 5.96510053e-01
7.78732657e-01 -6.15168393e-01 4.38722521e-01 -1.11353956e-02
1.00831352e-01 3.66613984e-01 -7.47128546e-01 2.07205486e+00
-5.72857022e-01 7.86788642e-01 1.94736436e-01 -1.32258666e+00
9.08853948e-01 5.65911591e-01 7.00767875e-01 -9.61416543e-01
3.08543742e-01 5.61191082e-01 -1.20794557e-01 -2.28999913e-01
5.22469521e-01 2.43587077e-01 2.91688263e-01 1.56038970e-01
4.36174907e-02 7.58976266e-02 3.42709050e-02 5.54802895e-01
1.47839165e+00 -1.71649054e-01 1.99532241e-01 6.09090701e-02
3.63095701e-01 -4.84074920e-01 7.70815551e-01 3.38306397e-01
-1.81925982e-01 7.28790462e-01 3.22838724e-01 -3.07323903e-01
-1.50293696e+00 -8.31166148e-01 2.39381716e-01 8.16924632e-01
2.44127288e-01 -1.11525202e+00 -4.17573124e-01 8.66742581e-02
-3.61039072e-01 3.94870520e-01 3.11858431e-02 -3.50064516e-01
-6.14414930e-01 -2.58388855e-02 2.42336825e-01 4.34940547e-01
8.33012223e-01 -5.41283906e-01 -8.17072570e-01 3.92206490e-01
-5.83726227e-01 -1.00409567e+00 -4.46076244e-01 2.52071545e-02
-1.15582311e+00 -5.70771873e-01 -1.04081082e+00 -7.66523838e-01
9.13876444e-02 7.21504152e-01 1.18733644e+00 3.75296712e-01
-9.59450990e-05 1.61402300e-01 -7.79729903e-01 5.57812490e-03
-5.79844892e-01 -4.89739068e-02 -3.62328172e-01 -2.01302856e-01
2.19764367e-01 -1.03475249e+00 -1.15050995e+00 1.93369552e-01
-9.02225018e-01 4.26896363e-01 6.87690318e-01 6.76230669e-01
7.15178609e-01 2.32736751e-01 3.85802060e-01 -3.21743637e-01
1.27042294e-01 -7.82564998e-01 -4.79482561e-01 -2.43698456e-03
-4.76074457e-01 -1.14117272e-01 1.02353990e+00 -3.36329937e-02
-7.16127515e-01 -1.95051089e-01 -2.99307108e-01 -9.23626244e-01
3.24149877e-01 5.86832345e-01 -3.68736312e-02 1.51788279e-01
2.89171994e-01 6.27825201e-01 1.05509236e-02 -3.95284206e-01
1.30721703e-01 8.63663971e-01 4.17116880e-01 -2.61332840e-01
4.25911099e-01 4.65350360e-01 2.13966936e-01 -6.28242016e-01
-2.51696020e-01 -4.88955259e-01 -2.67894715e-01 -1.43135622e-01
4.68205571e-01 -1.58095479e+00 -4.59072411e-01 3.30197632e-01
-1.23574162e+00 -4.49079335e-01 -7.09458441e-02 4.77216601e-01
-9.52733219e-01 6.47276461e-01 -8.67372692e-01 -5.37666619e-01
-4.70734239e-01 -1.20833838e+00 9.68312919e-01 -1.45627826e-01
2.94976443e-01 -8.19307983e-01 3.65543552e-02 2.15337709e-01
9.78931308e-01 -1.48857147e-01 5.65896571e-01 -9.61553454e-02
-1.16164184e+00 -6.94958195e-02 -1.66238472e-01 4.84107673e-01
6.54248148e-02 -3.13298166e-01 -5.75108051e-01 -5.09126127e-01
1.58427820e-01 -4.05988067e-01 1.02843440e+00 4.62726563e-01
1.52099991e+00 -4.11323637e-01 -1.58171162e-01 1.04556859e+00
1.80523252e+00 2.92527765e-01 1.05837357e+00 -8.18192959e-02
6.03640497e-01 -1.92247987e-01 2.58414298e-01 1.15704942e+00
3.53457808e-01 5.61789155e-01 5.70357203e-01 4.70337063e-01
-5.00694931e-01 -4.05716360e-01 4.42883670e-01 1.50978768e+00
-2.69455224e-01 -6.48389578e-01 -6.59547091e-01 5.11020660e-01
-1.79119790e+00 -1.14560032e+00 8.21768790e-02 2.16999936e+00
8.94655585e-01 -1.10402547e-01 -1.40489802e-01 4.27443445e-01
5.66416383e-01 5.66381574e-01 -4.55735266e-01 -1.25198746e-02
-9.51652154e-02 1.72699511e-01 6.98990047e-01 1.53696164e-01
-8.17604125e-01 5.22892952e-01 5.49170923e+00 1.32755852e+00
-8.51887047e-01 4.82585818e-01 9.00647938e-01 -4.56384689e-01
-3.51151407e-01 -7.57194776e-03 -5.40546477e-01 8.86293948e-01
1.63535583e+00 -4.11505342e-01 5.90381503e-01 1.11720669e+00
3.22566390e-01 -8.65407512e-02 -8.52669895e-01 1.50613916e+00
1.39539272e-01 -1.75585401e+00 -4.34402265e-02 2.70329624e-01
9.61873531e-01 3.10939759e-01 -6.36073202e-02 4.83246855e-02
8.09700266e-02 -7.69127071e-01 7.05286980e-01 2.75117397e-01
1.03095186e+00 -8.82672429e-01 7.29027212e-01 4.05942529e-01
-1.45225084e+00 -3.38788986e-01 -9.03260767e-01 -1.23544425e-01
5.68833947e-01 9.41759229e-01 -2.66150255e-02 3.56372386e-01
7.22683311e-01 1.18775547e+00 -7.05395117e-02 1.18665695e+00
-1.08589441e-01 9.19104218e-01 -1.84008390e-01 2.01692179e-01
3.32417935e-02 -1.11481519e-02 3.77849549e-01 9.36473906e-01
8.95620227e-01 1.02885649e-01 3.34208049e-02 4.78978992e-01
-3.93605232e-01 -9.51882899e-02 -6.16290927e-01 -1.86944604e-02
6.90657437e-01 6.92776084e-01 -3.62869591e-01 -5.27149618e-01
-7.19803393e-01 1.34891915e+00 7.38702491e-02 5.87316938e-02
-9.33818340e-01 -2.76928246e-01 4.79447424e-01 1.06904201e-01
8.28004003e-01 -5.41855037e-01 -7.19759166e-02 -1.13152349e+00
1.47979287e-02 -1.01185203e+00 -1.20614335e-01 -5.98696411e-01
-7.99642861e-01 4.12266374e-01 -2.10544601e-01 -1.56328273e+00
-2.99999893e-01 -3.71731333e-02 -2.17709780e-01 5.62053800e-01
-1.57746327e+00 -8.71902108e-01 -3.01525533e-01 8.16132724e-01
8.55483294e-01 -4.10617769e-01 6.05512559e-01 6.65120244e-01
-1.56741709e-01 5.13199806e-01 7.03791678e-01 4.28633615e-02
4.35015112e-01 -4.62050438e-01 1.25568613e-01 9.70448077e-01
-5.77459252e-03 2.51781970e-01 4.13525760e-01 -5.71268201e-01
-2.00416017e+00 -1.03697693e+00 8.68484676e-01 5.21682918e-01
6.10066712e-01 -1.81975096e-01 -8.31892669e-01 3.39303434e-01
5.56672812e-01 3.38320285e-01 7.26369560e-01 -4.15223598e-01
-4.08859253e-01 -3.16522866e-01 -7.31094837e-01 3.20771754e-01
1.22591078e+00 -5.44970989e-01 5.91570660e-02 5.42397916e-01
1.02792764e+00 -1.27015784e-01 -7.46022701e-01 9.07187932e-05
1.83232367e-01 -1.49669814e+00 9.47196186e-01 1.63958997e-01
1.34731960e+00 1.21008940e-02 -8.25107992e-01 -9.21401083e-01
-6.18313074e-01 -8.60343337e-01 -8.45960557e-01 9.01131630e-01
-1.52815998e-01 -1.50211319e-01 7.22638905e-01 2.54562914e-01
-3.13285321e-01 -8.06777477e-01 -1.11775589e+00 -1.02822542e+00
-8.12413245e-02 -5.91903985e-01 3.76537710e-01 2.32113436e-01
1.03784665e-01 -8.60111136e-03 -9.17590737e-01 -6.07615523e-03
7.54906833e-01 1.02941491e-01 5.99429965e-01 -7.07333446e-01
-8.71580482e-01 -3.16367388e-01 -6.83271527e-01 -1.90420711e+00
-1.47760734e-01 -7.35565543e-01 1.17144749e-01 -1.40984976e+00
3.79258692e-01 -3.61064166e-01 -2.09698215e-01 2.02724978e-01
2.12755531e-01 2.36863792e-01 4.60331827e-01 8.26931000e-01
-1.10108924e+00 1.04586864e+00 1.08472145e+00 -1.15368538e-01
1.74206525e-01 -2.93211669e-01 -4.97256786e-01 3.64396751e-01
9.03175414e-01 -5.16677558e-01 -7.10177898e-01 -9.92461562e-01
3.53219867e-01 5.81791937e-01 2.75253505e-01 -1.44745326e+00
6.40880883e-01 1.04017062e-02 3.43127362e-02 -5.09312272e-01
3.17429006e-01 -9.10926938e-01 1.07193716e-01 6.14665687e-01
-2.71640688e-01 9.28783640e-02 -3.29399824e-01 8.32782507e-01
-6.83133006e-01 -6.06390238e-02 7.39474237e-01 -1.15198754e-01
-7.29074359e-01 6.06066644e-01 -4.36370164e-01 -2.53877360e-02
9.40059602e-01 9.32005793e-02 -1.26739755e-01 -8.38857651e-01
-7.09921215e-03 -9.04846042e-02 4.63452786e-01 2.37524640e-02
1.13265979e+00 -1.28073859e+00 -8.71304810e-01 2.88334459e-01
-3.03994596e-01 -1.36652533e-02 7.01291203e-01 7.92134702e-01
-8.79847229e-01 4.92996812e-01 3.04760318e-02 -5.91281056e-01
-1.19821024e+00 7.49511003e-01 -2.81994283e-01 -8.61110613e-02
-8.90300035e-01 8.53052378e-01 -1.03731923e-01 8.18446219e-01
1.58624932e-01 6.83017075e-02 4.38915640e-01 -5.80572009e-01
9.86963153e-01 4.76795614e-01 -1.88741043e-01 -4.77657080e-01
-5.96911125e-02 2.74472803e-01 5.27414195e-02 1.16784588e-01
1.52934182e+00 -5.39983869e-01 -5.38603365e-02 1.20512083e-01
1.80252278e+00 -1.52389660e-01 -1.41498852e+00 -3.40316951e-01
-4.65713710e-01 -9.67620552e-01 4.98036534e-01 -1.96878687e-01
-1.62889385e+00 9.02699769e-01 4.84784633e-01 1.33228684e-02
1.59380436e+00 -1.94283184e-02 1.46263003e+00 1.39840081e-01
9.22545850e-01 -8.59717011e-01 1.93121895e-01 4.62475270e-01
7.75842607e-01 -1.12033284e+00 3.18054050e-01 -2.20935136e-01
-3.68479192e-01 1.15315628e+00 -1.82091758e-01 -2.77935475e-01
8.82847071e-01 5.23571968e-01 -5.05200982e-01 1.36297852e-01
-1.18523073e+00 2.81371206e-01 -1.57511562e-01 4.79559273e-01
5.52117050e-01 -6.85898140e-02 -2.98450410e-01 1.69730335e-01
7.71675482e-02 5.05451977e-01 3.60715717e-01 1.02747703e+00
-5.75876534e-01 -1.24026525e+00 2.93104979e-03 4.52704161e-01
-4.53276008e-01 -3.83395910e-01 5.12135327e-01 6.41258433e-02
1.64511502e-02 1.17568636e+00 2.16579422e-01 -6.56947851e-01
-5.59892297e-01 -4.46407825e-01 1.96485817e-01 -2.14641616e-01
-1.52016673e-02 5.21491647e-01 -2.84347296e-01 -9.82319355e-01
-6.34367645e-01 -4.96755064e-01 -9.87112701e-01 -8.38350654e-01
-3.46448980e-02 -3.11868079e-02 5.69003105e-01 5.24345100e-01
8.65772963e-01 4.56585735e-01 1.01323593e+00 -9.10834908e-01
-4.90017176e-01 -7.19156027e-01 -6.69350386e-01 2.38125741e-01
3.90265852e-01 -1.93101987e-01 -4.20788616e-01 4.19095337e-01] | [11.334227561950684, -1.6301425695419312] |
62b81e6b-0458-4429-b7f1-6fded56b49b9 | auxiliary-signal-guided-knowledge-encoder | 2006.03744 | null | https://arxiv.org/abs/2006.03744v1 | https://arxiv.org/pdf/2006.03744v1.pdf | Auxiliary Signal-Guided Knowledge Encoder-Decoder for Medical Report Generation | Beyond the common difficulties faced in the natural image captioning, medical report generation specifically requires the model to describe a medical image with a fine-grained and semantic-coherence paragraph that should satisfy both medical commonsense and logic. Previous works generally extract the global image features and attempt to generate a paragraph that is similar to referenced reports; however, this approach has two limitations. Firstly, the regions of primary interest to radiologists are usually located in a small area of the global image, meaning that the remainder parts of the image could be considered as irrelevant noise in the training procedure. Secondly, there are many similar sentences used in each medical report to describe the normal regions of the image, which causes serious data bias. This deviation is likely to teach models to generate these inessential sentences on a regular basis. To address these problems, we propose an Auxiliary Signal-Guided Knowledge Encoder-Decoder (ASGK) to mimic radiologists' working patterns. In more detail, ASGK integrates internal visual feature fusion and external medical linguistic information to guide medical knowledge transfer and learning. The core structure of ASGK consists of a medical graph encoder and a natural language decoder, inspired by advanced Generative Pre-Training (GPT). Experiments on the CX-CHR dataset and our COVID-19 CT Report dataset demonstrate that our proposed ASGK is able to generate a robust and accurate report, and moreover outperforms state-of-the-art methods on both medical terminology classification and paragraph generation metrics. | ['Xiaodan Liang', 'Xiaojun Chang', 'Mingjie Li', 'Fuyu Wang'] | 2020-06-06 | null | null | null | null | ['medical-report-generation'] | ['medical'] | [ 5.12169302e-01 7.65116453e-01 -2.08795428e-01 -3.71353567e-01
-9.31599200e-01 -1.66573420e-01 5.45928001e-01 1.38162911e-01
-2.08915733e-02 7.81960130e-01 5.10973513e-01 -2.14863971e-01
3.09387427e-02 -9.43466663e-01 -8.91156852e-01 -7.71379173e-01
3.62553209e-01 3.63656193e-01 1.51458994e-01 -9.28350091e-02
-2.78264433e-02 1.90721810e-01 -1.08503795e+00 6.09882593e-01
8.68340194e-01 8.01665664e-01 5.72055817e-01 4.82860893e-01
-2.97040880e-01 1.13156533e+00 -7.56015658e-01 -6.53322875e-01
-1.85860783e-01 -9.97733772e-01 -8.80427182e-01 4.51496869e-01
1.40290663e-01 -6.30205199e-02 -4.26427931e-01 1.19740903e+00
6.87494874e-01 -7.31485710e-02 9.62781072e-01 -7.40522623e-01
-1.16132355e+00 7.70080447e-01 -6.22784674e-01 2.73652285e-01
4.48796332e-01 2.95020461e-01 6.81022286e-01 -5.21908641e-01
8.48285735e-01 1.11814415e+00 3.84752184e-01 8.29039574e-01
-1.03688705e+00 -5.38646400e-01 1.21782027e-01 3.74717601e-02
-1.52561235e+00 -1.91152453e-01 9.08007920e-01 -4.00438070e-01
5.22409379e-01 4.36611652e-01 7.93587983e-01 1.40005028e+00
6.52651727e-01 7.07367957e-01 9.43445802e-01 -2.66911119e-01
1.45832554e-01 4.31177735e-01 -2.55519748e-01 8.98630857e-01
3.32273364e-01 -7.74969086e-02 -2.52984941e-01 5.30989468e-03
9.29310203e-01 -1.29323795e-01 -7.10850477e-01 -2.05052942e-01
-1.46516562e+00 9.91542935e-01 8.02621126e-01 4.86893177e-01
-6.00273132e-01 -1.01220936e-01 3.55614781e-01 -9.79823545e-02
3.19840491e-01 2.60223866e-01 2.15387806e-01 3.88287276e-01
-9.40830231e-01 3.40540186e-02 6.60134614e-01 1.04473448e+00
4.31052148e-01 -4.92056832e-02 -7.27243364e-01 6.05914295e-01
3.63653183e-01 5.43697238e-01 8.51773858e-01 -4.42330092e-01
5.71928144e-01 7.00902164e-01 -4.42009211e-01 -1.29379630e+00
-1.24775909e-01 -8.21356416e-01 -1.31524086e+00 -2.89487481e-01
-1.08700491e-01 2.07115725e-01 -1.24004269e+00 1.64552510e+00
1.66410401e-01 9.06973109e-02 3.11224937e-01 1.11767733e+00
1.47400331e+00 5.40187418e-01 2.47009486e-01 -4.61778849e-01
1.66975808e+00 -8.68792474e-01 -1.00354338e+00 -3.39763194e-01
4.05305147e-01 -6.18406236e-01 1.01987731e+00 -8.79603550e-02
-1.04924774e+00 -6.41013086e-01 -9.23438072e-01 4.11124676e-02
-5.16911373e-02 1.45701006e-01 4.69653457e-01 2.91941583e-01
-9.47667003e-01 -7.74028152e-02 -5.79741538e-01 -2.47507706e-01
5.78656316e-01 -1.02361880e-01 -3.20813835e-01 -2.49024227e-01
-1.23668063e+00 8.80103350e-01 7.58548617e-01 2.05147773e-01
-9.08715665e-01 -6.45264506e-01 -1.11717081e+00 -6.78724945e-02
4.02033716e-01 -1.23461604e+00 1.11042225e+00 -8.35119724e-01
-1.15292752e+00 1.21253407e+00 -3.67113352e-02 -3.73257726e-01
6.55215383e-01 3.47826749e-01 -5.02503812e-01 5.84440291e-01
3.52504730e-01 9.71036971e-01 8.24324250e-01 -1.54972899e+00
-2.49166936e-01 -7.28664473e-02 -9.28517580e-02 2.96803623e-01
2.79660374e-02 -4.05742466e-01 -7.34930992e-01 -1.13686097e+00
-2.17205863e-02 -7.97891557e-01 -4.43649143e-01 -2.28951141e-01
-8.14829409e-01 -6.12612348e-03 3.33712846e-01 -6.44679844e-01
1.28118372e+00 -2.04847169e+00 1.98790804e-02 1.98399886e-01
3.02195489e-01 -3.36233899e-02 2.86155846e-03 8.17165226e-02
-2.24644214e-01 1.27383932e-01 -4.97057557e-01 -6.86121508e-02
-2.82184184e-01 3.24605107e-01 -2.81032950e-01 2.29140803e-01
2.61262029e-01 1.30528557e+00 -1.02738917e+00 -1.17892218e+00
2.22331032e-01 4.46075946e-01 -4.24443930e-01 4.26636159e-01
-1.10716179e-01 6.75471127e-01 -7.60529101e-01 4.83667284e-01
4.93008345e-01 -7.20174849e-01 -3.69850770e-02 -5.71672440e-01
4.06558245e-01 7.91517422e-02 -6.61368012e-01 1.92429221e+00
-4.52021271e-01 2.26588234e-01 -2.93567896e-01 -1.08242238e+00
8.32810760e-01 5.04126668e-01 3.38151217e-01 -8.21414590e-01
1.57820493e-01 5.79749346e-02 -4.14894558e-02 -8.63565028e-01
1.09214462e-01 -5.24640083e-01 -9.26888585e-02 2.41845489e-01
4.99593802e-02 -4.24956769e-01 1.75100770e-02 5.02971649e-01
9.59892690e-01 -1.33419424e-01 6.38586044e-01 1.01573626e-02
7.75295973e-01 7.44847506e-02 4.43922967e-01 8.85153413e-01
6.65055290e-02 9.78072882e-01 4.27963227e-01 -1.81046396e-01
-7.27875233e-01 -1.24108267e+00 -8.38112906e-02 2.88431466e-01
3.20114166e-01 -2.08809316e-01 -7.03480005e-01 -1.04250693e+00
-5.05788267e-01 1.01767135e+00 -8.52274597e-01 -4.57906663e-01
-3.65344912e-01 -6.97216809e-01 3.49170983e-01 4.73752320e-01
5.67334831e-01 -1.53399801e+00 -5.94648421e-01 2.48358190e-01
-5.59538245e-01 -1.29538262e+00 -7.46308744e-01 -1.34363681e-01
-6.04300141e-01 -1.02457345e+00 -1.05243897e+00 -8.88111651e-01
1.21056044e+00 6.42208904e-02 1.33709919e+00 1.44273728e-01
-5.56401134e-01 3.66320521e-01 -4.74620998e-01 -3.69248658e-01
-9.37157393e-01 -1.69648379e-01 -5.56312919e-01 9.62945595e-02
9.09334570e-02 -2.34241828e-01 -8.41108441e-01 -7.04672039e-02
-1.13208485e+00 6.72165453e-01 1.12383258e+00 1.08316875e+00
9.07078922e-01 3.65220532e-02 4.35334742e-01 -9.45187747e-01
7.84226537e-01 -5.47450483e-01 6.50913715e-02 3.84814739e-01
-4.63451594e-01 1.37687728e-01 6.02648675e-01 -3.42083275e-01
-1.12349820e+00 2.58070305e-02 -2.29355395e-01 -5.15169442e-01
-3.51802737e-01 4.56286848e-01 -3.87615599e-02 2.29580149e-01
6.57984436e-01 8.89773309e-01 8.06350932e-02 3.73369008e-02
3.72910529e-01 5.30335486e-01 6.94523096e-01 -2.59941638e-01
7.57206976e-01 4.41129625e-01 -3.33112516e-02 -5.46656370e-01
-1.15737021e+00 -2.07371488e-01 -2.38313496e-01 -2.41872609e-01
1.28196132e+00 -8.69993269e-01 -3.99575949e-01 -1.88508704e-01
-1.28255069e+00 2.29421571e-01 -5.21795511e-01 5.08546412e-01
-6.75750792e-01 3.70089382e-01 -5.27207315e-01 -4.68813539e-01
-6.43041253e-01 -1.49488342e+00 1.31980717e+00 2.22349510e-01
-1.92575976e-01 -1.01919687e+00 -1.50563955e-01 4.02152002e-01
1.22485384e-01 4.57704425e-01 1.01867592e+00 -4.01442647e-01
-5.43123603e-01 7.02632442e-02 -3.42113316e-01 3.86574596e-01
3.77586305e-01 -5.31105220e-01 -7.75566816e-01 -1.43243372e-01
1.89745054e-01 -2.57435501e-01 8.95751536e-01 4.88531768e-01
1.28868604e+00 -4.84791398e-01 -5.61127484e-01 5.75901151e-01
1.30847907e+00 2.69602895e-01 7.52810419e-01 4.73900512e-02
6.78173482e-01 5.51990151e-01 5.00131905e-01 5.69562316e-02
5.57009816e-01 4.64763045e-01 3.82715106e-01 -6.00326002e-01
-4.90607888e-01 -4.18609351e-01 8.62127841e-02 1.00760436e+00
5.05153909e-02 -3.53351414e-01 -7.51698494e-01 7.38040626e-01
-1.72269034e+00 -9.30866182e-01 -3.25004160e-02 1.70287514e+00
1.11019659e+00 5.15299886e-02 -3.57451916e-01 -8.77509862e-02
6.66351795e-01 1.08020730e-01 -3.23917627e-01 -1.74232304e-01
-1.19526580e-01 6.88864738e-02 3.07909995e-01 1.33099303e-01
-9.38416958e-01 6.20881855e-01 5.86207342e+00 1.10669506e+00
-1.10713875e+00 1.59836322e-01 8.86135638e-01 2.61534423e-01
-5.67318499e-01 -3.86664599e-01 -4.94384825e-01 5.53298473e-01
6.05592549e-01 -1.82477444e-01 -2.29902893e-01 6.19620621e-01
1.93858787e-01 9.80102569e-02 -9.72272754e-01 1.14486361e+00
4.39233482e-01 -1.56335402e+00 7.04132617e-01 2.90306620e-02
6.87340617e-01 -4.36758280e-01 4.16418118e-03 2.13306680e-01
4.32389565e-02 -1.17083991e+00 6.56997859e-01 8.59726429e-01
1.02756095e+00 -5.93541026e-01 9.99242187e-01 3.47842634e-01
-1.05881369e+00 2.74103463e-01 -3.89963120e-01 7.35047400e-01
2.18081594e-01 6.03488624e-01 -1.27271998e+00 1.17233288e+00
4.99369502e-01 5.69622636e-01 -6.36649370e-01 8.20946336e-01
-4.61673528e-01 5.38861692e-01 5.55034205e-02 1.60373263e-02
3.33247274e-01 5.50615601e-02 4.96558338e-01 1.47478163e+00
4.42906350e-01 3.09189111e-01 2.39744812e-01 1.22506297e+00
-8.50372910e-02 2.99747109e-01 -7.37431169e-01 7.70767778e-02
-5.74168563e-02 1.13922679e+00 -8.02830100e-01 -5.36955535e-01
-4.56736654e-01 1.07587039e+00 2.25360729e-02 3.61221284e-01
-9.95074630e-01 -1.73240706e-01 -3.99024524e-02 2.57648051e-01
2.07915559e-01 3.88807714e-01 -1.57755479e-01 -1.15258074e+00
1.18512213e-01 -8.23387623e-01 4.93295789e-01 -9.45672214e-01
-1.37862241e+00 8.99284065e-01 -2.87222844e-02 -1.58645499e+00
-4.67357457e-01 -1.71374768e-01 -5.12341857e-01 6.59102499e-01
-1.54022110e+00 -1.45735252e+00 -6.28643394e-01 7.02695489e-01
6.57773018e-01 -2.12709785e-01 8.70704532e-01 1.16789937e-01
-1.71885788e-01 6.09690726e-01 -5.74459314e-01 2.55517811e-01
4.86700386e-01 -1.16159010e+00 2.43124515e-02 7.46885478e-01
7.37918392e-02 5.45493305e-01 6.14057779e-01 -7.36406446e-01
-9.15735960e-01 -1.40165150e+00 6.51717007e-01 -1.36014640e-01
3.35053891e-01 -1.13438696e-01 -1.04646444e+00 5.36544204e-01
3.49806160e-01 4.70675379e-02 6.24676049e-01 -7.36214280e-01
9.46839433e-03 7.26748034e-02 -1.14292014e+00 7.17500031e-01
1.07523441e+00 -3.35652620e-01 -9.80495691e-01 4.70654428e-01
8.77780676e-01 -5.98116875e-01 -8.45176995e-01 4.61690277e-01
9.88371596e-02 -6.93423986e-01 9.55767810e-01 -3.36081773e-01
7.57007837e-01 -3.70992661e-01 1.10129528e-01 -1.35424328e+00
-3.45464319e-01 -3.26921910e-01 3.00680082e-02 1.11201406e+00
3.84735823e-01 -3.61925334e-01 5.06311476e-01 -3.32255065e-02
-3.09717238e-01 -8.68061304e-01 -8.12133253e-01 -3.45897317e-01
-1.84417307e-01 -3.82444739e-01 4.75627303e-01 1.06336915e+00
-2.89894551e-01 4.85933363e-01 -1.82499066e-01 2.52705097e-01
6.25886977e-01 1.44696295e-01 3.86180490e-01 -7.62446225e-01
-2.90346384e-01 -4.26571876e-01 -5.96969008e-01 -9.26676095e-01
1.76516324e-01 -1.31496584e+00 9.89521593e-02 -2.11467671e+00
6.41616762e-01 -1.81111649e-01 -2.66157508e-01 5.25316000e-01
-4.04554546e-01 3.77558082e-01 -6.51369914e-02 1.38198152e-01
-5.03278434e-01 6.35181010e-01 1.99735844e+00 -5.49611926e-01
9.37963277e-02 -7.71215707e-02 -1.01008046e+00 5.76332510e-01
3.76181096e-01 -4.85308290e-01 -5.95393300e-01 -5.13474420e-02
5.58400154e-02 4.15505886e-01 5.75424373e-01 -8.10534120e-01
1.51377663e-01 -1.72155425e-01 5.98643303e-01 -6.82642817e-01
7.07101226e-02 -8.01833570e-01 2.25861102e-01 5.85231781e-01
-3.90488982e-01 -2.44465172e-01 8.28236043e-02 5.80588400e-01
-5.02740681e-01 -1.57689705e-01 7.60690391e-01 -4.38896865e-01
-5.42625368e-01 2.61243969e-01 -2.04606920e-01 6.12562560e-02
1.18818295e+00 -2.66059339e-01 -2.67358065e-01 -4.35540408e-01
-7.87487328e-01 1.71580076e-01 1.80363044e-01 5.33327341e-01
1.08974862e+00 -1.32689691e+00 -1.10121799e+00 1.43049389e-01
4.58343089e-01 2.98391998e-01 6.17920518e-01 1.00968564e+00
-5.96654117e-01 5.08661747e-01 1.75312534e-01 -7.54042923e-01
-9.25832093e-01 7.52264380e-01 4.74854708e-01 -4.80895132e-01
-1.16525161e+00 6.43892527e-01 8.51153433e-01 -8.40472654e-02
-3.30184028e-02 -4.57159579e-01 -3.06564569e-01 -1.79364011e-01
5.70265532e-01 -4.76289123e-01 -4.90152352e-02 -6.82103932e-01
-3.57539892e-01 6.34875357e-01 -2.20032126e-01 1.10742308e-01
1.03202391e+00 -1.30958229e-01 -2.19443608e-02 2.10385457e-01
1.01127243e+00 -9.23418924e-02 -6.44595444e-01 -2.32139915e-01
-3.50334764e-01 -1.04965672e-01 9.15717147e-03 -8.85273039e-01
-1.22720015e+00 6.76966786e-01 5.27370751e-01 1.04896434e-01
1.35739613e+00 5.32790542e-01 6.42259717e-01 5.57295233e-02
1.59706503e-01 -7.01911688e-01 2.58638531e-01 -1.07134633e-01
1.32373011e+00 -1.26523948e+00 1.33681456e-02 -5.99224985e-01
-1.12705970e+00 9.62996364e-01 5.21983087e-01 1.46624878e-01
2.85212100e-01 1.83342397e-01 2.19525352e-01 -5.22902787e-01
-6.36797130e-01 -1.32250234e-01 6.64433956e-01 5.96789479e-01
4.56461936e-01 5.98223656e-02 -4.67515439e-01 6.42454326e-01
-3.42839330e-01 -4.61836867e-02 4.63567466e-01 6.90792441e-01
-1.38406202e-01 -5.17348111e-01 -4.71659541e-01 4.40016419e-01
-5.72252393e-01 -1.64972126e-01 -8.24781358e-02 8.24015379e-01
2.26011023e-01 7.73083091e-01 -3.18719149e-02 -1.06456682e-01
3.78175139e-01 -1.38354674e-01 4.11716431e-01 -8.54599118e-01
-3.87850195e-01 1.82851344e-01 -1.14625901e-01 -4.12050307e-01
-4.90729034e-01 -2.01455534e-01 -1.58820033e+00 2.56432444e-01
-1.43222198e-01 1.45018086e-01 4.38691378e-01 9.24147487e-01
2.02849403e-01 1.29545128e+00 4.42563355e-01 -2.84566909e-01
-2.83190548e-01 -1.08098507e+00 -4.66328949e-01 6.28980875e-01
2.89723039e-01 -4.86633003e-01 -1.64209545e-01 3.84750038e-01] | [15.055386543273926, -1.3882713317871094] |
227bda08-dac6-47ee-87a5-cecb71779f27 | a-new-approach-for-trading-based-on-long | 2001.03333 | null | https://arxiv.org/abs/2001.03333v1 | https://arxiv.org/pdf/2001.03333v1.pdf | A new approach for trading based on Long Short Term Memory technique | The stock market prediction has always been crucial for stakeholders, traders and investors. We developed an ensemble Long Short Term Memory (LSTM) model that includes two-time frequencies (annual and daily parameters) in order to predict the next-day Closing price (one step ahead). Based on a four-step approach, this methodology is a serial combination of two LSTM algorithms. The empirical experiment is applied to 417 NY stock exchange companies. Based on Open High Low Close metrics and other financial ratios, this approach proves that the stock market prediction can be improved. | ['Saaid Achchab', 'Zineb Lanbouri'] | 2020-01-10 | null | null | null | null | ['stock-market-prediction'] | ['time-series'] | [-6.50712371e-01 -3.25656414e-01 -7.31664374e-02 -1.25129402e-01
-1.92229077e-01 -6.33044243e-01 7.97678173e-01 -1.53809354e-01
-5.24803102e-01 1.14507210e+00 2.78198630e-01 -6.31452799e-01
-1.70593694e-01 -1.19340241e+00 -1.62986130e-01 -4.43306565e-01
-4.91267949e-01 2.49504447e-01 7.86638185e-02 -4.80404973e-01
8.43920827e-01 3.79188746e-01 -1.35796583e+00 5.09002745e-01
5.00919461e-01 1.46718657e+00 -6.96425661e-02 4.88712251e-01
-7.55730748e-01 1.28605080e+00 -4.21597749e-01 -4.03754860e-01
6.32477880e-01 -1.66506633e-01 -5.68493366e-01 -6.21969819e-01
-4.35216278e-01 -5.56096792e-01 1.68110952e-01 1.02085352e+00
2.90338665e-01 2.99036577e-02 2.08565116e-01 -7.25076556e-01
-7.00311720e-01 1.02682424e+00 -2.81077236e-01 6.52397752e-01
-1.76272959e-01 -4.47315648e-02 9.74459589e-01 -9.34180856e-01
2.92643696e-01 7.65207171e-01 9.04399455e-01 -2.26647913e-01
-7.62215436e-01 -8.25265527e-01 1.36794552e-01 1.54770032e-01
-7.73385525e-01 2.32652366e-01 6.81951344e-01 -6.00477934e-01
1.35679400e+00 3.89882885e-02 1.16342354e+00 8.33203256e-01
9.28184032e-01 2.59459198e-01 1.27611864e+00 -3.93206745e-01
1.97914377e-01 2.78226614e-01 3.52224469e-01 -1.88229144e-01
4.18580204e-01 6.72837615e-01 -2.92217433e-01 -1.67666227e-01
8.20739329e-01 2.45138928e-01 1.32276639e-01 5.70835590e-01
-1.19226229e+00 1.04516351e+00 2.97323793e-01 1.27397287e+00
-1.10129178e+00 4.10220437e-02 4.04197633e-01 9.04678106e-01
8.26923609e-01 4.39718962e-01 -8.43139887e-01 -5.14212549e-01
-1.24873638e+00 3.02349508e-01 1.25131810e+00 1.69713229e-01
3.50022495e-01 7.29696274e-01 6.36570677e-02 2.87931561e-01
3.21524382e-01 5.17751932e-01 1.15252805e+00 -3.97090882e-01
1.33912206e-01 5.43770075e-01 3.90641063e-01 -1.12330782e+00
-5.47174633e-01 -8.88308048e-01 -8.74697864e-01 5.85728765e-01
2.15626545e-02 -6.67890847e-01 -5.87060571e-01 8.80458415e-01
-3.34581316e-01 5.61092556e-01 2.85888135e-01 3.38307679e-01
5.16147792e-01 1.15380466e+00 1.19693242e-02 -4.28081483e-01
9.82400775e-01 -5.73905528e-01 -1.07134628e+00 1.53947741e-01
5.13026476e-01 -8.19253266e-01 1.92349210e-01 5.71349442e-01
-9.34680879e-01 -6.05648518e-01 -1.07071626e+00 6.85250521e-01
-1.09284103e+00 -2.09593832e-01 3.38274658e-01 5.92275679e-01
-1.02685165e+00 1.23612356e+00 -3.68990690e-01 2.14372531e-01
-2.84089386e-01 3.30435723e-01 2.31406167e-01 1.19929278e+00
-1.81059754e+00 1.39860225e+00 9.38493490e-01 5.30274987e-01
1.55833676e-01 -5.81751823e-01 -2.39467949e-01 4.46054377e-02
-3.88184637e-01 -4.57672030e-01 1.38028562e+00 -1.15970039e+00
-1.80067301e+00 3.73223543e-01 3.99176598e-01 -1.29694438e+00
7.61426985e-01 -1.94715559e-01 -9.44563508e-01 -2.86818594e-01
-4.30722564e-01 5.57552418e-03 6.30359292e-01 -5.94523311e-01
-1.01902008e+00 -1.47034988e-01 -3.55447620e-01 -1.57763466e-01
-3.22776318e-01 2.05496103e-01 9.64304626e-01 -1.24639523e+00
-2.77820174e-02 -5.05065441e-01 -2.04077780e-01 -1.04512513e+00
2.06853285e-01 -6.85568452e-02 6.38362825e-01 -1.28559124e+00
1.52282012e+00 -1.58867025e+00 -6.03408813e-01 5.65914690e-01
-2.93837517e-01 3.50188881e-01 3.03859770e-01 7.34826744e-01
-5.53807855e-01 2.83550262e-01 -4.58591711e-03 5.51707819e-02
2.80091882e-01 -1.41800591e-03 -1.01824880e+00 -2.54003983e-02
2.29180157e-02 1.17209923e+00 -3.08229983e-01 5.08043803e-02
3.91411543e-01 3.61607462e-01 2.37101212e-01 -8.60649422e-02
-2.72764653e-01 8.45472440e-02 -3.24310571e-01 4.33864921e-01
7.28917539e-01 -8.79169330e-02 -1.32955030e-01 1.70799971e-01
-8.45002651e-01 7.80053213e-02 -1.20282173e+00 9.82150257e-01
-4.64028627e-01 5.82666039e-01 -6.71552181e-01 -6.40500486e-01
1.50401580e+00 6.85759068e-01 4.94683504e-01 -8.95789921e-01
4.38753307e-01 1.03118277e+00 -3.58434841e-02 -3.33870262e-01
4.84788448e-01 -6.08784974e-01 2.42395729e-01 6.81969523e-01
-2.47832015e-01 4.74925816e-01 1.69181332e-01 -6.35082126e-01
3.06737244e-01 2.30726257e-01 3.95039648e-01 -4.43520069e-01
5.72614849e-01 -2.44416177e-01 4.84061241e-01 4.27660018e-01
7.87030458e-02 -8.27312469e-03 4.16551262e-01 -1.06725359e+00
-1.03169620e+00 -3.94385010e-01 -2.25084990e-01 5.19855917e-01
-7.82504022e-01 2.90586859e-01 -4.36068386e-01 -7.77643621e-02
2.36609548e-01 1.16202605e+00 -6.05721354e-01 3.32780987e-01
-4.43490714e-01 -9.70836759e-01 3.44403177e-01 4.87600476e-01
6.73874021e-01 -1.38843584e+00 -1.02447784e+00 8.87530208e-01
3.99895102e-01 -6.33265615e-01 1.61527812e-01 2.80589640e-01
-1.14052284e+00 -6.03014231e-01 -1.15501571e+00 -5.40317655e-01
-1.65127650e-01 -5.96138418e-01 1.14236128e+00 -2.13290274e-01
4.11311597e-01 -2.05798611e-01 -3.40544850e-01 -8.72095764e-01
-4.36604321e-01 3.37362252e-02 -1.67672217e-01 3.77918333e-01
6.47551656e-01 -7.73685873e-01 -4.27485913e-01 -1.79649889e-01
-6.81777477e-01 -2.01263562e-01 6.85478568e-01 5.90963542e-01
2.80492455e-01 1.30984083e-01 1.15835321e+00 -6.30142629e-01
9.10817921e-01 -6.41173244e-01 -1.25767660e+00 3.03114563e-01
-1.13940477e+00 7.20047504e-02 4.11709607e-01 -1.82901070e-01
-1.16448283e+00 -4.14104074e-01 -2.20342830e-01 -2.24560678e-01
3.65568072e-01 1.02970839e+00 6.16995513e-01 -1.17660150e-01
1.43986166e-04 5.95639229e-01 7.02156723e-02 -8.27656746e-01
-2.02242181e-01 5.59999168e-01 1.55012846e-01 2.18880251e-02
5.41301370e-01 -5.82317635e-02 -2.58419812e-01 -5.81090927e-01
-3.96173954e-01 9.77649465e-02 -8.62824023e-01 -5.84511161e-01
5.83973944e-01 -8.70922446e-01 -8.20087850e-01 9.38669741e-01
-1.25012076e+00 -1.21064214e-02 -4.00966346e-01 7.98643410e-01
-2.16386467e-01 -1.32673100e-01 -8.56047869e-01 -1.29655039e+00
-7.99339473e-01 -5.68090677e-01 1.87081978e-01 1.98970914e-01
-1.80569917e-01 -1.53321016e+00 4.18999255e-01 -5.40524364e-01
9.73706365e-01 6.64250493e-01 5.31594753e-01 -1.26654136e+00
-3.96119386e-01 -5.88527739e-01 2.35357992e-02 9.11275446e-01
-1.58910275e-01 2.33196661e-01 -8.32386374e-01 3.16316374e-02
6.75913751e-01 2.79733539e-01 8.81166816e-01 7.03092098e-01
1.82958633e-01 -4.03358459e-01 3.45594943e-01 3.91052037e-01
1.65871048e+00 8.75896037e-01 6.17518127e-01 1.11019802e+00
1.00694381e-01 4.66460258e-01 4.87690538e-01 6.98617458e-01
1.34884506e-01 -1.84176713e-01 1.02239668e-01 1.89975306e-01
5.87154150e-01 -1.40517443e-01 6.80302978e-01 1.19103086e+00
-3.56081694e-01 2.31022596e-01 -9.19270098e-01 2.55415440e-01
-1.81309128e+00 -1.41114473e+00 -1.76285088e-01 2.04543447e+00
3.15020055e-01 7.94219971e-01 2.76120901e-01 4.35271114e-01
6.64119303e-01 4.04044122e-01 -4.27035123e-01 -7.07042575e-01
-4.44666535e-01 4.09768999e-01 8.62498879e-01 2.84776539e-01
-1.01940644e+00 4.07547563e-01 7.24041367e+00 5.11414647e-01
-1.49188709e+00 -1.23130456e-01 7.99289584e-01 1.62383839e-01
-5.34400940e-01 -4.09606807e-02 -9.03164387e-01 9.08402443e-01
1.56375539e+00 -8.64549577e-01 2.49773543e-02 6.46084607e-01
3.92642587e-01 1.08987726e-01 -5.54692268e-01 6.38543785e-01
-3.73199672e-01 -1.79501557e+00 2.11652920e-01 2.94775873e-01
7.37648845e-01 1.63226143e-01 2.11109012e-01 3.36501896e-01
-6.63627237e-02 -7.83099353e-01 9.42319274e-01 1.36325824e+00
-1.05674885e-01 -1.13801062e+00 1.52757883e+00 4.61901993e-01
-1.23796833e+00 -4.80332881e-01 -1.84851229e-01 -5.99766672e-01
2.44120702e-01 6.78002238e-01 -3.51092726e-01 7.93804586e-01
7.48151124e-01 8.98538113e-01 -1.96310654e-01 8.96576345e-01
1.96482182e-01 3.29563677e-01 -4.15537506e-01 -5.12285292e-01
7.96909094e-01 -6.26531661e-01 3.19563121e-01 9.33895588e-01
9.18257833e-01 7.61244521e-02 -3.78302187e-01 9.61171687e-01
3.24843764e-01 3.03663522e-01 -6.67423010e-01 -1.78494886e-01
2.44803175e-01 7.80264676e-01 -5.59045851e-01 -4.11122471e-01
-5.88780165e-01 3.96968871e-01 -5.36051273e-01 3.63868505e-01
-6.49087012e-01 -4.59043622e-01 1.54637083e-01 5.09176180e-02
4.27094698e-01 4.93534990e-02 -7.44813859e-01 -9.48975265e-01
1.16357468e-01 -4.91869837e-01 4.95918989e-01 -5.51437199e-01
-1.39848948e+00 7.26126373e-01 -1.94847465e-01 -1.60517251e+00
-7.68666625e-01 -8.33626926e-01 -1.00176299e+00 1.23372197e+00
-1.80342710e+00 -8.86822283e-01 5.13691545e-01 2.72688091e-01
2.73040444e-01 -7.66433060e-01 8.81585956e-01 2.60623753e-01
-4.75014269e-01 -1.65939942e-01 5.73068857e-01 1.24861129e-01
-1.39208734e-01 -1.24774885e+00 7.13200450e-01 8.28396261e-01
-1.13621376e-01 2.77205259e-01 6.72216475e-01 -7.98749864e-01
-5.38416207e-01 -8.20704758e-01 1.39527535e+00 4.56788614e-02
1.28768659e+00 3.64207238e-01 -9.72521424e-01 1.03032255e+00
8.48412871e-01 -5.43505013e-01 6.88228846e-01 -3.53404373e-01
2.56652832e-01 -3.59529644e-01 -1.02476442e+00 5.87378927e-02
-3.00967209e-02 -1.21896073e-01 -1.07558548e+00 -2.38491327e-01
6.00308478e-01 3.37874033e-02 -1.36239755e+00 4.74836409e-01
7.67651320e-01 -1.41183734e+00 5.03875971e-01 -2.86758095e-01
-8.04771483e-02 7.65222684e-02 3.61837931e-02 -1.21845841e+00
-1.42970249e-01 -6.92267656e-01 -2.43981764e-01 1.06111598e+00
6.95169210e-01 -1.51319027e+00 4.89670157e-01 6.90514445e-01
2.56658375e-01 -6.35692954e-01 -1.03002870e+00 -1.01995623e+00
3.73725891e-01 -3.34139138e-01 1.14764321e+00 8.71221244e-01
-1.25725612e-01 -1.04307204e-01 -4.44827944e-01 -2.89655447e-01
3.51335317e-01 5.05215108e-01 1.18723527e-01 -1.63389349e+00
1.02068760e-01 -9.97967422e-01 -1.98492363e-01 -3.77431333e-01
1.11569896e-01 -5.07041991e-01 -8.05641353e-01 -1.19555044e+00
-4.80753630e-01 8.17472637e-02 -1.14077675e+00 1.55630499e-01
3.88317198e-01 -2.72017747e-01 2.24686578e-01 2.02707574e-01
3.73013645e-01 4.73149896e-01 8.58196437e-01 1.43191293e-01
-1.78430766e-01 4.60710138e-01 -2.12450266e-01 8.37957203e-01
9.83938575e-01 -2.11034536e-01 9.92091671e-02 6.91884831e-02
7.09377229e-01 1.06963649e-01 2.67660260e-01 -1.18342793e+00
2.74620265e-01 -3.21974903e-02 5.63349605e-01 -1.40433669e+00
1.92082580e-02 -8.30872953e-01 7.75147080e-01 1.15836763e+00
-1.71022236e-01 7.49303043e-01 3.74260157e-01 1.59954011e-01
-9.04230058e-01 -3.94616753e-01 5.35316467e-01 -3.20479095e-01
-8.79002988e-01 5.31196706e-02 -4.31572765e-01 -5.31401634e-01
1.30480933e+00 -3.98699135e-01 -4.87203486e-02 -2.62725443e-01
-7.99762070e-01 2.20783994e-01 -1.50139287e-01 4.07565385e-01
4.15130705e-01 -1.49144971e+00 -9.72682774e-01 3.00867468e-01
-5.22052348e-01 -7.19241321e-01 2.77941674e-01 6.58809125e-01
-9.62405801e-01 7.64677227e-01 -5.87981939e-01 8.61237720e-02
-5.09919465e-01 2.40900725e-01 8.25921774e-01 -5.76742291e-01
-5.58066249e-01 6.88994527e-01 -7.78091073e-01 -5.58791459e-02
8.60594362e-02 -6.39453232e-01 -8.45376670e-01 9.77340043e-01
6.70798242e-01 5.59060752e-01 1.35690406e-01 -4.97613072e-01
-4.48665917e-02 7.37775028e-01 2.19361842e-01 -3.60601813e-01
1.99332035e+00 1.04867376e-01 -4.58683640e-01 1.43963706e+00
9.73190427e-01 -3.44006360e-01 -7.76135147e-01 -5.90277743e-03
1.01074266e+00 -1.98587239e-01 1.00919329e-01 -7.41216004e-01
-1.13008130e+00 6.41336620e-01 7.04100609e-01 9.41191673e-01
1.03346896e+00 -1.09142673e+00 1.37685144e+00 4.27773863e-01
3.36792588e-01 -1.52840602e+00 -6.79247141e-01 6.47829533e-01
9.41034794e-01 -9.12986338e-01 -3.22176635e-01 5.17138720e-01
-3.36307168e-01 1.77786493e+00 -1.57665849e-01 -4.38787252e-01
1.45960188e+00 3.97845656e-01 2.97275424e-01 1.35658607e-01
-1.17610776e+00 3.30269597e-02 7.84946755e-02 -1.07090391e-01
5.03967404e-01 1.01944990e-01 -4.73646432e-01 8.93774986e-01
-4.88350838e-01 4.47768480e-01 4.88867283e-01 8.37020993e-01
-6.53084934e-01 -1.09294510e+00 -4.95913178e-01 5.76550066e-01
-8.11064541e-01 -1.22257628e-01 -2.15943053e-01 9.24983084e-01
-1.21111348e-01 5.96882164e-01 3.89742643e-01 -4.61284280e-01
4.23630714e-01 4.21176851e-01 -2.29135230e-01 -5.42258769e-02
-1.22963369e+00 1.96361333e-01 5.81204367e-04 -2.25598872e-01
-5.53210020e-01 -7.60147750e-01 -1.23438299e+00 -7.18908966e-01
-5.72412461e-02 2.03336403e-01 7.54776418e-01 8.82678449e-01
1.91645056e-01 5.48192918e-01 1.14798760e+00 -7.42981493e-01
-1.02969503e+00 -1.23063195e+00 -1.17333901e+00 -2.34425217e-01
4.40333039e-01 -2.64913768e-01 -5.05813003e-01 -1.65506855e-01] | [4.52601957321167, 4.198631763458252] |
d7e7b5fd-7d9e-4fd1-a081-b37917f965b5 | shadowformer-global-context-helps-image | 2302.01650 | null | https://arxiv.org/abs/2302.01650v1 | https://arxiv.org/pdf/2302.01650v1.pdf | ShadowFormer: Global Context Helps Image Shadow Removal | Recent deep learning methods have achieved promising results in image shadow removal. However, most of the existing approaches focus on working locally within shadow and non-shadow regions, resulting in severe artifacts around the shadow boundaries as well as inconsistent illumination between shadow and non-shadow regions. It is still challenging for the deep shadow removal model to exploit the global contextual correlation between shadow and non-shadow regions. In this work, we first propose a Retinex-based shadow model, from which we derive a novel transformer-based network, dubbed ShandowFormer, to exploit non-shadow regions to help shadow region restoration. A multi-scale channel attention framework is employed to hierarchically capture the global information. Based on that, we propose a Shadow-Interaction Module (SIM) with Shadow-Interaction Attention (SIA) in the bottleneck stage to effectively model the context correlation between shadow and non-shadow regions. We conduct extensive experiments on three popular public datasets, including ISTD, ISTD+, and SRD, to evaluate the proposed method. Our method achieves state-of-the-art performance by using up to 150X fewer model parameters. | ['Bihan Wen', 'Hao Cheng', 'Ding Liu', 'Siyu Huang', 'Lanqing Guo'] | 2023-02-03 | null | null | null | null | ['shadow-removal', 'image-shadow-removal'] | ['computer-vision', 'computer-vision'] | [ 3.86915088e-01 -2.64754564e-01 3.30439746e-01 -4.84198004e-01
-4.78516966e-01 -1.18079573e-01 4.64349121e-01 -5.47118306e-01
2.80003510e-02 7.08472431e-01 5.91718197e-01 -4.70066547e-01
3.68006974e-01 -5.05080104e-01 -5.65877020e-01 -1.00520301e+00
3.28130990e-01 -7.27893189e-02 6.49309397e-01 -1.36607900e-01
-1.31607857e-02 4.60997820e-01 -1.11627913e+00 3.48127931e-01
1.14623439e+00 9.18146610e-01 7.29146004e-01 5.22005379e-01
-2.10820865e-02 7.91481793e-01 -6.90127075e-01 1.27626836e-01
2.31691509e-01 -4.05444056e-01 -1.65741101e-01 4.39381413e-03
3.23791891e-01 -9.92052317e-01 -6.05026364e-01 5.73874712e-01
7.58276820e-01 1.53453993e-02 3.09542120e-01 -1.16777587e+00
-6.01405561e-01 5.71453087e-02 -9.12373841e-01 1.06619097e-01
8.90785381e-02 1.93729460e-01 6.49169266e-01 -1.03154564e+00
2.51528323e-01 1.46705139e+00 6.89759433e-01 1.25234917e-01
-8.93361688e-01 -8.90364945e-01 5.07938802e-01 3.47052991e-01
-1.20108366e+00 -4.95505482e-01 8.56114388e-01 2.37955958e-01
6.24071300e-01 3.32362503e-01 6.31279469e-01 1.15890419e+00
5.25424600e-01 1.19001019e+00 1.58063376e+00 -2.37344801e-01
1.28877405e-02 -1.19683504e-01 -2.03170460e-02 7.03149438e-01
-3.74523886e-02 2.20456362e-01 -6.37488067e-01 -5.79792522e-02
7.94377863e-01 2.43856192e-01 -7.37902939e-01 -3.26162279e-01
-8.44487011e-01 5.49626648e-01 6.83946013e-01 -7.09922761e-02
-4.01965499e-01 2.42205024e-01 6.87895641e-02 -2.00904265e-01
7.24892914e-01 -9.75994542e-02 -4.08701569e-01 3.85286301e-01
-8.38526666e-01 1.21711463e-01 5.39839149e-01 7.88106918e-01
7.39421368e-01 8.78104270e-02 -5.68489611e-01 7.92819619e-01
5.87781847e-01 8.60418379e-01 -8.09346810e-02 -8.10362220e-01
3.10870886e-01 2.60817081e-01 2.95079976e-01 -1.04653621e+00
-3.52835983e-01 -7.40862668e-01 -9.53218400e-01 2.45517984e-01
-1.32972330e-01 -2.35629501e-03 -1.38671339e+00 1.56447446e+00
5.15435696e-01 8.19033623e-01 8.21364149e-02 1.19400406e+00
1.09679210e+00 7.62620568e-01 -1.70301154e-01 -2.71760464e-01
1.12952483e+00 -1.34920561e+00 -8.65823388e-01 -3.93614203e-01
6.24588504e-02 -1.00863588e+00 1.28528607e+00 9.53940153e-02
-6.52548790e-01 -3.43738586e-01 -1.10609031e+00 -2.33047426e-01
-1.70018598e-01 1.00580700e-01 7.62737572e-01 3.40129018e-01
-8.62872183e-01 5.53201623e-02 -6.76828086e-01 -1.90609112e-01
6.56823575e-01 5.95558994e-02 1.81264237e-01 -5.05248666e-01
-1.08029711e+00 6.87383890e-01 -2.07680538e-01 6.54895127e-01
-1.28078592e+00 -8.67996871e-01 -6.27057433e-01 6.27875030e-02
6.10853851e-01 -4.98473197e-01 9.62819159e-01 -9.78535116e-01
-1.32103086e+00 2.05363795e-01 -5.60550213e-01 -1.39889374e-01
4.88468081e-01 -5.69642663e-01 -3.13316733e-01 -1.08483344e-01
7.05067674e-03 2.06483051e-01 8.44336689e-01 -2.22285795e+00
-4.19158280e-01 -1.90076202e-01 1.61489442e-01 6.26874268e-01
-9.84807685e-02 1.12545183e-02 -9.58397031e-01 -8.40881586e-01
8.63144770e-02 -8.47016454e-01 -4.47048321e-02 1.54821008e-01
-7.91434824e-01 2.35714227e-01 1.44724584e+00 -9.84286189e-01
1.16353059e+00 -2.09890056e+00 -2.71038681e-01 1.71839982e-01
4.83681560e-01 4.46066856e-01 -2.81873494e-01 1.54104874e-01
2.57510483e-01 -3.94082695e-01 -5.22580504e-01 -6.38359666e-01
1.43824972e-03 5.09828210e-01 -7.08436012e-01 4.75653350e-01
-2.20931277e-01 9.05723095e-01 -7.73438215e-01 -5.00261009e-01
4.07458365e-01 1.01396966e+00 -1.17911853e-01 4.66824621e-01
-1.37893260e-01 4.71127659e-01 -6.03294730e-01 9.07440186e-01
1.40461707e+00 -9.06220451e-02 4.66165617e-02 -4.09378111e-01
-8.20702612e-02 1.61286145e-01 -9.22770083e-01 1.41665447e+00
-8.30126286e-01 8.74576747e-01 4.27354157e-01 -2.39713788e-01
4.94603902e-01 -4.20159847e-02 2.82253981e-01 -1.11336648e+00
-1.28718512e-02 5.55677339e-02 -2.09186718e-01 -2.08456218e-01
4.46305782e-01 4.12144559e-03 3.42220515e-01 4.96723086e-01
-5.57687640e-01 2.59076118e-01 -6.29267812e-01 4.34399754e-01
1.22344637e+00 3.91631603e-01 -5.42773306e-02 -2.41001755e-01
3.78010333e-01 -6.11967862e-01 7.81406581e-01 7.50450015e-01
-1.88294083e-01 9.40480888e-01 1.86421141e-01 -2.26612091e-01
-4.24869955e-01 -1.25383890e+00 1.03260212e-01 1.28620541e+00
7.43445992e-01 -2.37331584e-01 -6.01707697e-01 -6.61850214e-01
-2.38394435e-03 6.93693876e-01 -7.95041502e-01 -9.71170813e-02
-6.14362657e-01 -1.05092716e+00 2.68527418e-01 3.33897412e-01
1.03474581e+00 -1.00547957e+00 -6.40029252e-01 1.07560262e-01
-3.50819856e-01 -1.05578351e+00 -5.50209522e-01 1.76048845e-01
-1.79901719e-01 -8.85619462e-01 -9.07459795e-01 -3.67513269e-01
5.15238285e-01 1.13628733e+00 1.09683657e+00 4.85066295e-01
-1.97054267e-01 6.54698536e-02 -4.66618627e-01 -6.76725149e-01
1.08275071e-01 -3.26827794e-01 -4.05212939e-01 3.21098894e-01
-2.53847390e-01 -6.25775993e-01 -1.04962015e+00 4.90755141e-01
-9.14625525e-01 4.95965540e-01 1.01252317e+00 9.22467887e-01
5.03108203e-01 1.18705472e-02 1.82781652e-01 -1.00190163e+00
3.41415435e-01 -5.10736167e-01 -3.73718530e-01 3.64603072e-01
-6.31184459e-01 -2.67621487e-01 5.59643805e-01 -4.06590626e-02
-1.85971916e+00 -6.44556582e-02 1.64030075e-01 -5.37418008e-01
1.40213683e-01 5.21805026e-02 -8.13038409e-01 -2.58820534e-01
7.46289790e-02 3.88488531e-01 -6.48112833e-01 -5.36591530e-01
1.95314363e-01 5.49691081e-01 4.13337290e-01 -2.91756064e-01
1.16513455e+00 1.00183356e+00 -6.73554614e-02 -8.08739424e-01
-1.21606684e+00 -3.23777288e-01 -3.27110916e-01 -2.21884519e-01
6.57354057e-01 -1.07656789e+00 -3.01672399e-01 8.97726715e-01
-1.14639258e+00 -9.97887433e-01 2.21740291e-01 -4.28878814e-02
1.51634766e-02 4.75938231e-01 -2.60206699e-01 -9.20590341e-01
-4.24480200e-01 -1.11378312e+00 1.48945999e+00 4.38520342e-01
5.27216136e-01 -8.76551509e-01 -1.29313126e-01 3.05111825e-01
6.09526932e-01 1.51792809e-01 7.10634649e-01 1.05130993e-01
-1.14066291e+00 3.14098179e-01 -9.72852468e-01 3.25209022e-01
3.10602605e-01 -5.02282858e-01 -1.33665371e+00 -2.74629653e-01
-1.80781230e-01 -8.33854005e-02 1.41926408e+00 4.14081931e-01
1.33138335e+00 -1.22707248e-01 -5.42276204e-01 9.99660194e-01
1.39367414e+00 6.22092746e-03 8.78056467e-01 1.44677639e-01
1.22729230e+00 1.82615921e-01 8.49939585e-01 4.84575480e-01
6.50644183e-01 6.88342810e-01 5.48035920e-01 -8.91943634e-01
-7.26831973e-01 -8.78181979e-02 2.74320692e-01 3.55406523e-01
8.95292386e-02 -7.23548710e-01 -6.58358157e-01 4.14815485e-01
-1.86788094e+00 -5.02008736e-01 -1.71202227e-01 1.92904449e+00
5.66021442e-01 1.05973810e-01 -6.29808903e-01 -3.13600779e-01
2.70919591e-01 7.57705808e-01 -7.64806926e-01 7.16227815e-02
-4.82963592e-01 1.47178024e-01 6.53315485e-01 8.23218286e-01
-1.04457521e+00 1.22905326e+00 5.37155771e+00 9.80824590e-01
-1.03200722e+00 1.89608306e-01 5.16178846e-01 4.44677919e-02
-6.47072196e-01 2.50981301e-01 -5.46569228e-01 6.29444540e-01
2.10106909e-01 5.54541409e-01 6.30248368e-01 3.81897688e-01
6.51417196e-01 -6.56072080e-01 -4.01995778e-01 7.59987831e-01
2.65496224e-01 -8.85436475e-01 -1.70266658e-01 4.68066335e-02
9.09824550e-01 8.79450142e-02 1.24235237e-02 2.38788486e-01
3.79438609e-01 -9.70839202e-01 5.72295010e-01 5.80041111e-01
6.43347025e-01 -6.41554475e-01 7.79396534e-01 -4.37171571e-02
-1.52276742e+00 -9.24629644e-02 -2.75088668e-01 1.79441348e-01
9.74170044e-02 9.79115486e-01 -7.33241737e-01 7.82896101e-01
8.79241765e-01 6.57150924e-01 -5.83760142e-01 9.27220762e-01
-4.10315633e-01 6.50173664e-01 -2.66937017e-01 4.28586483e-01
1.18851312e-01 -1.62774220e-01 3.97850513e-01 1.29285121e+00
2.79561151e-02 2.98458219e-01 2.36439079e-01 6.65437400e-01
-1.59572169e-01 -3.43126833e-01 -1.90343082e-01 6.01236045e-01
5.48725247e-01 1.31764877e+00 -7.94723988e-01 -3.29821765e-01
-5.06805599e-01 1.41444254e+00 -1.72261506e-01 8.96923244e-01
-1.35749400e+00 -2.65078962e-01 7.22607613e-01 -1.92697883e-01
2.42099524e-01 -1.83906108e-01 -3.05490106e-01 -1.00445485e+00
6.88709691e-02 -8.62604678e-01 -4.56213728e-02 -1.24178755e+00
-9.31220531e-01 2.79911011e-01 -2.01945364e-01 -8.97391140e-01
7.62550712e-01 -9.83612537e-02 -9.58521545e-01 1.08538556e+00
-2.21776128e+00 -1.66482055e+00 -1.06174541e+00 7.45947778e-01
5.04872143e-01 2.00286329e-01 4.61895972e-01 4.25079137e-01
-6.35770559e-01 4.67712939e-01 3.28902990e-01 -1.17277980e-01
1.01485240e+00 -1.20295107e+00 3.27543646e-01 1.03126419e+00
-3.66531104e-01 2.70240217e-01 7.28020310e-01 -8.28050435e-01
-1.49157965e+00 -1.34043419e+00 4.95226562e-01 -1.34593293e-01
2.25475714e-01 -5.57602465e-01 -9.99551773e-01 4.31143641e-01
2.38363266e-01 1.49417773e-01 2.93273747e-01 -5.44183329e-02
-4.01340663e-01 -4.20011699e-01 -8.21069598e-01 8.16450179e-01
1.32494617e+00 -5.48722625e-01 -1.17421687e-01 5.63840270e-01
8.93909812e-01 -5.76740563e-01 -8.06048140e-02 5.12718678e-01
5.74580193e-01 -1.32717478e+00 1.20260870e+00 2.67740369e-01
3.19399625e-01 -5.10549605e-01 -3.88202310e-01 -1.22746241e+00
-1.38508767e-01 -6.28618896e-01 -3.59860241e-01 1.32896173e+00
6.14860700e-03 -6.44856632e-01 6.93285227e-01 1.29439369e-01
-6.97068870e-01 -1.04368639e+00 -7.51979470e-01 -4.79925722e-01
-4.82063621e-01 -3.23257267e-01 6.82231128e-01 6.30786359e-01
-1.08681285e+00 3.10875475e-01 -7.05679059e-01 6.19522214e-01
8.27659011e-01 6.22692645e-01 9.94856477e-01 -8.08526993e-01
-2.80606985e-01 -2.79028807e-02 4.29493994e-01 -1.29299879e+00
2.46520564e-01 -3.45943838e-01 5.97688198e-01 -1.92712843e+00
5.68145752e-01 -5.71286738e-01 -5.48347890e-01 6.78855538e-01
-5.32438934e-01 4.00567174e-01 1.11480504e-01 1.69218272e-01
-8.49841177e-01 1.04344058e+00 1.54530025e+00 -2.58887233e-03
-2.89777249e-01 -8.08378607e-02 -7.13728249e-01 7.87783265e-01
5.24806142e-01 -2.57038772e-01 -5.07448256e-01 -7.51580954e-01
-3.34464401e-01 -1.52666911e-01 8.20138276e-01 -8.54142010e-01
-1.19349249e-01 -3.43480319e-01 6.78013384e-01 -9.42545772e-01
7.98274636e-01 -8.17411423e-01 -1.79611340e-01 2.32713714e-01
7.76641816e-02 -5.19129157e-01 1.00296356e-01 8.89520943e-01
1.51733235e-01 6.56267047e-01 8.10621023e-01 2.38337386e-02
-7.97210693e-01 3.15715164e-01 -2.06637666e-01 -2.16915905e-01
7.74976373e-01 -1.01526072e-02 -5.07067621e-01 -5.85523009e-01
-6.25235215e-02 4.74661887e-01 5.11419833e-01 3.09667110e-01
8.88822496e-01 -1.05368674e+00 -7.24080205e-01 2.23105341e-01
-5.25829867e-02 3.17215323e-02 6.35841489e-01 9.17086124e-01
-3.79858255e-01 2.20057160e-01 9.64395776e-02 -3.43115538e-01
-1.52876508e+00 1.38174146e-01 4.41345543e-01 -2.65843064e-01
-8.78852606e-01 9.06632543e-01 1.07570481e+00 -2.44063899e-01
4.10127193e-01 -2.87514776e-01 1.63693935e-01 -3.08734417e-01
3.74688983e-01 3.03272486e-01 -1.77633226e-01 -5.34471333e-01
-4.56903011e-01 3.74098092e-01 7.68863410e-02 1.15444839e-01
1.22529805e+00 -5.59921741e-01 -1.31750524e-01 1.00416131e-01
1.05605900e+00 2.55782038e-01 -1.88884437e+00 -4.43152845e-01
-5.50473213e-01 -8.75734568e-01 5.47256410e-01 -1.31197524e+00
-1.22728109e+00 8.91877472e-01 9.27383184e-01 -2.77666628e-01
1.65825427e+00 -7.57829919e-02 1.36550081e+00 1.65384188e-01
1.97411418e-01 -8.80531669e-01 2.19760373e-01 6.92924321e-01
1.10197175e+00 -1.32271242e+00 3.87295663e-01 -5.73882103e-01
-4.77106780e-01 5.92272043e-01 5.98590016e-01 2.80745756e-02
6.90495491e-01 4.22238827e-01 1.82861999e-01 -4.92394902e-03
-4.42356139e-01 -3.38436961e-01 2.39476383e-01 6.96919441e-01
5.08068912e-02 8.01402777e-02 1.46983236e-01 5.72569847e-01
3.76711518e-01 -2.40352005e-01 4.07021821e-01 1.01888788e+00
-5.05831242e-01 -8.69875789e-01 -5.85692823e-01 4.30658668e-01
1.47327362e-03 -5.45429111e-01 -5.34791470e-01 7.18767226e-01
3.25998396e-01 1.08545434e+00 -2.99388915e-01 -4.34027970e-01
8.17827359e-02 -5.67651808e-01 2.42400259e-01 -3.35909307e-01
-2.28717819e-01 5.16978860e-01 1.33194895e-02 -9.14610863e-01
-3.44520569e-01 -4.47314143e-01 -1.33188546e+00 -3.51404577e-01
-2.88228869e-01 -1.99523449e-01 5.12214780e-01 1.11025155e+00
4.04146552e-01 9.61594760e-01 7.24710882e-01 -1.10873687e+00
1.15587249e-01 -8.04213822e-01 -6.57100141e-01 -6.41947240e-02
7.59275019e-01 -9.06178951e-01 -2.65648782e-01 -3.05085599e-01] | [10.830449104309082, -4.076090335845947] |
3235960f-bd36-4a7f-9cb3-931369761a7f | towards-fully-automated-segmentation-of-rat | 2109.04188 | null | https://arxiv.org/abs/2109.04188v1 | https://arxiv.org/pdf/2109.04188v1.pdf | Towards Fully Automated Segmentation of Rat Cardiac MRI by Leveraging Deep Learning Frameworks | Automated segmentation of human cardiac magnetic resonance datasets has been steadily improving during recent years. However, these methods are not directly applicable in preclinical context due to limited datasets and lower image resolution. Successful application of deep architectures for rat cardiac segmentation, although of critical importance for preclinical evaluation of cardiac function, has to our knowledge not yet been reported. We developed segmentation models that expand on the standard U-Net architecture and evaluated separate models for systole and diastole phases, 2MSA, and one model for all timepoints, 1MSA. Furthermore, we calibrated model outputs using a Gaussian Process (GP)-based prior to improve phase selection. Resulting models approach human performance in terms of left ventricular segmentation quality and ejection fraction (EF) estimation in both 1MSA and 2MSA settings (S{\o}rensen-Dice score 0.91 +/- 0.072 and 0.93 +/- 0.032, respectively). 2MSA achieved a mean absolute difference between estimated and reference EF of 3.5 +/- 2.5 %, while 1MSA resulted in 4.1 +/- 3.0 %. Applying Gaussian Processes to 1MSA allows to automate the selection of systole and diastole phases. Combined with a novel cardiac phase selection strategy, our work presents an important first step towards a fully automated segmentation pipeline in the context of rat cardiac analysis. | ['Leif Hultin', 'Patrik Kagelid', 'Peter Konings', 'Magdalena Zurek', 'Arijit Patra', 'Harris Vince', 'Andrea Gondova', 'Daniel Fernandez-Llaneza'] | 2021-09-09 | null | null | null | null | ['cardiac-segmentation'] | ['medical'] | [ 2.60784775e-01 8.43526050e-02 1.84497818e-01 -3.88651311e-01
-6.33800626e-01 -7.09822059e-01 3.21655065e-01 6.29895270e-01
-7.13759303e-01 7.41240799e-01 -3.30884844e-01 -4.89292741e-01
-1.44933537e-01 -5.62552154e-01 -4.72345740e-01 -6.65044188e-01
-3.21217895e-01 1.02591777e+00 4.74116206e-01 2.79686093e-01
1.97567031e-01 6.98434353e-01 -5.06782055e-01 1.17960140e-01
8.42048287e-01 5.92870474e-01 2.26369217e-01 9.53370631e-01
1.33381218e-01 2.87707776e-01 -4.56937909e-01 -5.15213497e-02
1.65774912e-01 -6.34438932e-01 -8.16753924e-01 -2.98093587e-01
2.51592338e-01 -3.19661230e-01 1.93976723e-02 7.16315150e-01
9.30835128e-01 -6.64917706e-03 6.03338599e-01 -7.03589678e-01
4.84548174e-02 5.96486628e-01 -4.74484950e-01 6.99556291e-01
-2.55222380e-01 4.62603599e-01 4.06195819e-01 -3.21598649e-01
6.70184016e-01 6.42176151e-01 8.50641549e-01 5.02451539e-01
-1.84930944e+00 -5.52589595e-01 -3.09170336e-01 -3.74625713e-01
-1.05612493e+00 -2.14548171e-01 3.56041253e-01 -9.78690863e-01
9.12217915e-01 5.89872003e-02 8.60792637e-01 5.27136624e-01
6.34299338e-01 4.59648103e-01 1.22989929e+00 -1.95023596e-01
8.08299854e-02 -1.07979164e-01 2.68124670e-01 3.56094450e-01
2.70753771e-01 1.50252897e-02 2.72795171e-01 -8.22862461e-02
1.19089806e+00 -2.83254623e-01 -2.17210159e-01 -4.24194962e-01
-1.50495899e+00 7.40734220e-01 1.75023139e-01 5.08385718e-01
-4.74865735e-01 3.25438797e-01 5.33275366e-01 -2.95784753e-02
2.74386913e-01 4.90526140e-01 -4.12646592e-01 -1.32241830e-01
-1.51357329e+00 4.00300980e-01 4.15066332e-01 5.47712266e-01
1.64695665e-01 1.17968045e-01 -4.57780987e-01 8.05790126e-01
4.00902718e-01 5.05974174e-01 4.70437706e-01 -1.18467474e+00
5.98089099e-02 3.04698020e-01 -1.32678319e-02 -7.19353974e-01
-9.49898541e-01 -6.98971748e-01 -8.31022978e-01 3.30333382e-01
8.14738572e-01 -3.57089221e-01 -1.12717330e+00 1.57641649e+00
2.31501371e-01 -3.06372829e-02 -2.27617607e-01 9.44352746e-01
5.83685935e-01 1.67362258e-01 4.85144854e-01 -3.04315776e-01
1.30514479e+00 -4.79137421e-01 -4.37461346e-01 -9.35229957e-02
7.80528903e-01 -6.42069995e-01 7.32615232e-01 2.28542864e-01
-1.33644295e+00 -4.12659585e-01 -9.59866226e-01 4.02442276e-01
2.07096115e-01 1.57649115e-01 3.76321584e-01 8.87257040e-01
-1.16277838e+00 1.10964251e+00 -1.40227699e+00 -8.17611217e-02
7.99881220e-01 6.10377789e-01 -2.12865502e-01 2.63558507e-01
-1.07156479e+00 1.02953053e+00 3.79585594e-01 9.07381326e-02
-7.69001782e-01 -1.15964568e+00 -5.11483192e-01 -1.08426176e-01
-3.31101157e-02 -8.51972640e-01 1.14142871e+00 -2.51708299e-01
-1.36843657e+00 1.09916604e+00 7.74589926e-02 -9.21531856e-01
8.83454621e-01 4.73284088e-02 -8.28143209e-02 4.62044477e-01
1.95987433e-01 7.40178823e-01 2.93770850e-01 -9.70279217e-01
-9.07517821e-02 -5.17184675e-01 -4.97743160e-01 -1.62945688e-02
5.13247907e-01 1.71862647e-01 -2.08591402e-01 -6.72827244e-01
2.09480077e-01 -9.62859392e-01 -6.30746424e-01 -7.22640902e-02
4.32355776e-02 5.80779970e-01 2.95876533e-01 -1.05954981e+00
1.23720348e+00 -1.61595631e+00 -1.84350848e-01 3.36989820e-01
5.53213596e-01 5.20449936e-01 2.82789975e-01 -1.86514407e-01
-3.08500171e-01 3.76299351e-01 -6.59804642e-01 -1.14907593e-01
-4.63580310e-01 -1.17061153e-01 2.58192301e-01 5.45948327e-01
3.70704010e-02 9.70072865e-01 -1.04134595e+00 -5.68565845e-01
5.06991923e-01 5.70199013e-01 -5.02888799e-01 -3.09149981e-01
1.83701769e-01 9.31565762e-01 -2.78014362e-01 1.78187996e-01
6.33413434e-01 -9.33372229e-02 4.07596231e-01 -1.18932702e-01
-3.58378649e-01 -1.20995171e-01 -8.11048388e-01 1.65195513e+00
-2.20222205e-01 5.70106447e-01 1.42266765e-01 -8.47363114e-01
8.74343097e-01 6.20452344e-01 1.09821427e+00 -6.56174898e-01
4.72485900e-01 2.83073515e-01 6.79148793e-01 -1.70212314e-01
-1.42344674e-02 -7.33650327e-01 3.04570317e-01 2.55838811e-01
1.71468467e-01 -3.30416739e-01 4.24144864e-01 -6.96327239e-02
9.28968906e-01 2.52221674e-01 3.96536216e-02 -6.43180251e-01
4.81481344e-01 8.82077292e-02 6.22251511e-01 7.17323184e-01
-8.17847610e-01 1.16434014e+00 7.64587104e-01 -4.21998590e-01
-1.10462308e+00 -1.09042096e+00 -4.49232757e-01 2.74414390e-01
-2.18482047e-01 -1.70652807e-01 -9.29083228e-01 -6.03727579e-01
-1.53482676e-01 7.28518307e-01 -3.91545087e-01 1.55058235e-01
-7.98969865e-01 -1.18808389e+00 6.96902454e-01 5.60580671e-01
1.60883144e-01 -1.08033824e+00 -1.07608056e+00 6.81139886e-01
4.59120143e-03 -9.50981855e-01 -1.47522256e-01 1.72294974e-01
-1.34418917e+00 -1.06151450e+00 -1.25758171e+00 -5.07930994e-01
4.07884300e-01 -4.09131646e-01 1.27979338e+00 -2.99381204e-02
-3.91888380e-01 -3.28673683e-02 1.30956069e-01 -3.47241879e-01
-5.79582930e-01 -1.79603919e-01 -2.46353567e-01 -5.76104522e-01
-1.07060038e-01 -5.86851954e-01 -1.00107682e+00 2.79273361e-01
-5.94855905e-01 1.07451409e-01 3.72698188e-01 6.42419219e-01
9.21666920e-01 -4.25954342e-01 6.09250426e-01 -1.07784069e+00
4.25785124e-01 -2.57131934e-01 -7.33129740e-01 -2.48698890e-02
-8.88545275e-01 -2.23668098e-01 4.24908817e-01 -2.19569698e-01
-7.12973654e-01 1.40168995e-01 -1.94990829e-01 -4.49085623e-01
-3.75586957e-01 3.50822687e-01 5.28424144e-01 3.03818155e-02
7.16954052e-01 -1.13797411e-01 4.18997854e-01 -1.54661536e-01
7.97676221e-02 9.57625657e-02 4.93157864e-01 -4.71593916e-01
2.30933696e-01 5.37286699e-01 3.95879805e-01 -6.40360296e-01
-7.95902759e-02 -3.55197102e-01 -9.74964678e-01 -2.47270882e-01
1.26361918e+00 -5.41531324e-01 -5.42982459e-01 4.62828219e-01
-8.19647491e-01 -7.65546739e-01 -3.48973334e-01 8.07678521e-01
-8.20428252e-01 5.94379008e-01 -7.47179806e-01 -3.47506076e-01
-9.36245322e-01 -1.67282641e+00 5.67367315e-01 7.65652508e-02
-6.81604505e-01 -1.16234899e+00 8.51080716e-02 4.82758105e-01
6.25957251e-01 6.99728549e-01 8.72883260e-01 -7.86887646e-01
-2.65154600e-01 -1.97351769e-01 -3.44414404e-03 4.19589579e-01
-1.47429213e-01 -4.77193948e-03 -5.48033416e-01 -4.60759401e-02
-5.30830119e-03 3.42349350e-01 5.83640158e-01 1.25017130e+00
8.33429098e-01 4.29421395e-01 -2.19818830e-01 4.08885419e-01
1.12495279e+00 6.50666356e-01 6.69886947e-01 2.21072093e-01
4.96078551e-01 2.38061413e-01 3.38697135e-01 2.24893600e-01
2.08961964e-02 4.89398748e-01 3.98041978e-02 -1.53072461e-01
-2.94428408e-01 2.09003404e-01 -1.92271501e-01 3.15097779e-01
-4.92668338e-02 2.28896230e-01 -1.41361880e+00 7.10661173e-01
-1.47214448e+00 -7.26650476e-01 -7.20408738e-01 2.43685484e+00
7.41521478e-01 5.33113837e-01 2.34493524e-01 -1.39777929e-01
7.47048020e-01 -2.23359674e-01 -4.00992930e-01 -1.98770091e-01
2.04293713e-01 5.90118110e-01 5.98799348e-01 6.18263721e-01
-1.04014063e+00 6.18016422e-01 6.17269278e+00 3.27594459e-01
-1.20502305e+00 1.31988041e-02 9.96313751e-01 2.23286692e-02
-5.28551936e-02 3.09608132e-01 -3.97716939e-01 6.29490197e-01
1.07508063e+00 -8.39485750e-02 8.49907100e-02 3.70170563e-01
5.07634759e-01 -3.05130988e-01 -9.45894539e-01 6.96632326e-01
-3.73638868e-01 -1.41859937e+00 -3.97459179e-01 -4.50915620e-02
6.42460644e-01 1.78636506e-01 -1.55055612e-01 1.94255143e-01
-2.54589528e-01 -1.01295543e+00 5.05248487e-01 6.15784645e-01
8.68263304e-01 -4.84451652e-01 8.47753048e-01 2.23267704e-01
-7.67650306e-01 4.73938406e-01 9.21145976e-02 3.12556416e-01
4.69080240e-01 7.20525265e-01 -1.25804532e+00 3.26393992e-01
5.01139402e-01 3.64675730e-01 -2.98017144e-01 1.29583418e+00
6.84231073e-02 7.16671765e-01 -4.09886003e-01 5.08540750e-01
1.16991967e-01 -4.59976137e-01 6.24280691e-01 1.24428844e+00
2.33242497e-01 -5.11503071e-02 -7.52581432e-02 1.11069179e+00
1.55496627e-01 2.64968038e-01 -4.00860934e-03 7.44374003e-03
2.30609089e-01 1.28754711e+00 -1.44142234e+00 -4.48734254e-01
-1.46929055e-01 5.32906950e-01 -3.41996670e-01 1.70832083e-01
-7.92860389e-01 -3.38771641e-01 1.14726141e-01 7.48327076e-01
-5.66033311e-02 -4.46403213e-03 -8.01242352e-01 -6.55007064e-01
-3.01363945e-01 -5.57654202e-01 4.48737800e-01 -5.52302420e-01
-6.87648714e-01 4.19884384e-01 3.14161271e-01 -8.14376175e-01
-2.73168772e-01 -4.49657738e-01 -7.37686515e-01 1.41151369e+00
-1.02424741e+00 -7.24465311e-01 -4.34818119e-02 -8.03886726e-02
2.08125904e-01 1.42606154e-01 8.13946724e-01 4.09959525e-01
-4.00893062e-01 1.32938221e-01 -8.28120187e-02 1.69689789e-01
7.29524255e-01 -1.53820479e+00 4.88874078e-01 7.77908385e-01
-4.09827113e-01 8.23258877e-01 7.94585824e-01 -7.79898703e-01
-5.47217429e-01 -8.45341444e-01 8.15872967e-01 -4.19187665e-01
5.31159580e-01 3.18628043e-01 -8.11685801e-01 6.04532599e-01
-6.40466064e-02 1.57161251e-01 6.93761528e-01 -3.25735718e-01
4.81693804e-01 6.03956766e-02 -1.42066181e+00 4.82616723e-01
2.42969155e-01 7.70409927e-02 -3.21740568e-01 1.73900545e-01
2.75500864e-01 -7.81008422e-01 -1.50503993e+00 7.01155305e-01
5.33327579e-01 -9.47530627e-01 1.07122803e+00 -3.06976765e-01
1.72536075e-01 -3.58488977e-01 4.36216265e-01 -9.39850569e-01
-2.49172673e-01 -6.30209923e-01 9.82027426e-02 7.49171495e-01
3.45701188e-01 -5.68952799e-01 8.77759695e-01 8.56316924e-01
-5.04725397e-01 -9.16087866e-01 -9.34620023e-01 -2.41717726e-01
6.00809515e-01 -4.68428254e-01 -8.22677836e-03 7.21972883e-01
-2.75023460e-01 5.52417859e-02 1.23509869e-01 -1.07001118e-01
7.52884328e-01 3.06807458e-02 2.62251168e-01 -1.28851402e+00
-1.08966045e-01 -8.18373799e-01 -4.56423730e-01 -4.45677668e-01
-1.51910424e-01 -1.07109630e+00 -1.88707069e-01 -1.88732612e+00
8.15006047e-02 -4.89651591e-01 -4.25971419e-01 2.51513600e-01
-1.47215143e-01 1.28487870e-01 2.02470258e-01 2.29145512e-02
3.05489413e-02 -6.32890221e-03 1.36001897e+00 2.00795338e-01
-4.62275386e-01 1.97501287e-01 -4.56546783e-01 6.83643460e-01
1.17306304e+00 -4.37708467e-01 -4.52546209e-01 -1.03856131e-01
-1.63410798e-01 4.56675619e-01 4.77681071e-01 -1.21175694e+00
-1.85505509e-01 2.53164351e-01 7.48047113e-01 -6.32567167e-01
-1.30669817e-01 -5.50447464e-01 4.32409495e-01 8.39161873e-01
-2.56307125e-01 -5.37968911e-02 3.81234705e-01 4.67416570e-02
1.17945068e-01 -3.66807967e-01 1.26758611e+00 -3.98057669e-01
-1.88039199e-01 3.86236191e-01 -6.14825904e-01 2.29360476e-01
1.06267726e+00 -4.22829419e-01 1.88183099e-01 -4.78740446e-02
-1.52527916e+00 3.88319016e-01 2.82841623e-01 -8.91250148e-02
2.96626717e-01 -8.61834407e-01 -7.74918199e-01 -8.66524354e-02
-3.92621219e-01 1.48554832e-01 8.24859142e-01 1.49957371e+00
-1.14106607e+00 3.95284444e-01 -3.45113933e-01 -9.29523289e-01
-1.22796822e+00 2.38081500e-01 7.48281181e-01 -6.11254215e-01
-7.07747877e-01 4.93872404e-01 7.83375353e-02 -1.96812943e-01
-2.71545887e-01 -5.68273604e-01 -2.30424523e-01 1.54339867e-02
9.51662958e-02 4.68381643e-01 1.49597213e-01 -7.53589332e-01
-4.34020907e-01 4.84185815e-01 -2.95569338e-02 -8.69061798e-02
1.07917440e+00 -1.46391675e-01 4.88050766e-02 2.76193261e-01
8.61080945e-01 -3.00952315e-01 -1.32091188e+00 2.96174675e-01
-5.37407759e-04 7.38343671e-02 2.68310994e-01 -1.02065611e+00
-1.13143933e+00 1.11427522e+00 8.79295826e-01 -1.56774744e-01
8.68113697e-01 -3.87654036e-01 6.69215679e-01 -4.22655553e-01
-1.29985996e-02 -9.50204551e-01 -4.05449122e-01 2.68010616e-01
6.01497948e-01 -9.37803626e-01 1.63346484e-01 -1.97476789e-01
-7.99447417e-01 1.15742242e+00 2.29768738e-01 -1.09758332e-01
4.91882533e-01 3.31318974e-01 2.26378575e-01 -2.23791897e-01
-2.05594137e-01 1.40362278e-01 2.45617926e-01 5.46814919e-01
9.12982225e-01 1.17512196e-01 -6.72826707e-01 5.45155346e-01
7.93925673e-02 2.18873575e-01 3.95651042e-01 8.93063903e-01
-2.59488791e-01 -1.04155374e+00 -1.86213329e-01 5.34599006e-01
-9.71175075e-01 -1.55324981e-01 3.24493051e-01 7.78379798e-01
-1.06960004e-02 4.80583638e-01 -6.32957220e-02 2.55026549e-01
3.22035611e-01 3.86861265e-01 5.55222869e-01 -5.85364938e-01
-9.01453733e-01 5.77931404e-01 -1.99648380e-01 -3.56925160e-01
-3.68986845e-01 -8.60946417e-01 -1.76043344e+00 -3.86462472e-02
-6.73271790e-02 -7.07868189e-02 6.15010560e-01 6.87072337e-01
2.58393168e-01 8.06593776e-01 3.51644233e-02 -7.07939804e-01
-2.29832292e-01 -8.80960345e-01 -4.78170693e-01 2.22956181e-01
-1.91535056e-02 -4.59391952e-01 -6.34841099e-02 2.56822109e-01] | [14.134992599487305, -2.508458137512207] |
057cc273-bc0d-4125-ae35-6bbade13de37 | factorization-of-multi-agent-sampling-based | 2304.00342 | null | https://arxiv.org/abs/2304.00342v1 | https://arxiv.org/pdf/2304.00342v1.pdf | Factorization of Multi-Agent Sampling-Based Motion Planning | Modern robotics often involves multiple embodied agents operating within a shared environment. Path planning in these cases is considerably more challenging than in single-agent scenarios. Although standard Sampling-based Algorithms (SBAs) can be used to search for solutions in the robots' joint space, this approach quickly becomes computationally intractable as the number of agents increases. To address this issue, we integrate the concept of factorization into sampling-based algorithms, which requires only minimal modifications to existing methods. During the search for a solution we can decouple (i.e., factorize) different subsets of agents into independent lower-dimensional search spaces once we certify that their future solutions will be independent of each other using a factorization heuristic. Consequently, we progressively construct a lean hypergraph where certain (hyper-)edges split the agents to independent subgraphs. In the best case, this approach can reduce the growth in dimensionality of the search space from exponential to linear in the number of agents. On average, fewer samples are needed to find high-quality solutions while preserving the optimality, completeness, and anytime properties of SBAs. We present a general implementation of a factorized SBA, derive an analytical gain in terms of sample complexity for PRM*, and showcase empirical results for RRG. | ['Emilio Frazzoli', 'Andrea Censi', 'Pietro Zullo', 'Alessandro Zanardi'] | 2023-04-01 | null | null | null | null | ['motion-planning'] | ['robots'] | [ 7.28921220e-02 4.03291374e-01 4.19409983e-02 1.57565251e-01
-6.54514849e-01 -9.17468786e-01 2.27057815e-01 9.35968459e-02
-4.70281094e-01 1.04109037e+00 -1.29388899e-01 -3.37442100e-01
-7.50939727e-01 -9.44604158e-01 -8.57903004e-01 -8.26676488e-01
-4.58689392e-01 1.13336349e+00 2.29485899e-01 -9.53752100e-02
1.89040601e-01 6.20137393e-01 -1.44418168e+00 -4.91287887e-01
9.03734744e-01 4.47514176e-01 5.82872152e-01 7.49268949e-01
1.75566331e-01 1.99238896e-01 -4.43574250e-01 4.89758607e-03
6.90468788e-01 -2.83590257e-01 -1.01525915e+00 5.80814481e-01
-2.12270796e-01 -2.56144434e-01 -3.03226888e-01 1.08239555e+00
2.08584405e-03 3.27110291e-01 4.48860615e-01 -1.88011265e+00
-6.64178282e-02 7.51643479e-01 -6.78828478e-01 -3.12840998e-01
3.10467631e-01 1.50302211e-02 7.58774996e-01 -3.97138178e-01
1.00606751e+00 1.33605051e+00 4.50296104e-01 2.62640387e-01
-1.44712615e+00 -3.35590750e-01 5.30383825e-01 5.44582754e-02
-1.18374729e+00 -3.42308342e-01 2.05827013e-01 -1.37810349e-01
9.27644789e-01 2.51010567e-01 9.76176500e-01 5.02177715e-01
3.35178405e-01 8.08344960e-01 7.36147404e-01 -3.48204017e-01
5.16022980e-01 -2.69703209e-01 -5.70625886e-02 8.07489038e-01
9.13048625e-01 1.31273242e-02 -4.12347615e-01 -3.99536103e-01
8.26471508e-01 -9.34473053e-02 -2.67364740e-01 -1.04762435e+00
-1.39589155e+00 8.35459709e-01 3.31636965e-01 -7.61422738e-02
-4.54091966e-01 4.08146501e-01 1.04335934e-01 6.28323853e-01
-5.51749319e-02 8.37798238e-01 -4.04714108e-01 1.41060166e-03
-3.53290796e-01 7.73820162e-01 1.13031685e+00 1.20087242e+00
1.03450596e+00 -3.04334760e-01 6.80774212e-01 3.33741039e-01
1.90829068e-01 3.94956052e-01 -1.99936599e-01 -1.46296144e+00
4.78027463e-01 5.22197247e-01 7.19418645e-01 -9.95763242e-01
-7.24571347e-01 -3.41252625e-01 -3.69465321e-01 6.09303117e-01
4.68680739e-01 -4.97656047e-01 -7.55908072e-01 1.73970139e+00
7.89409935e-01 -2.80941755e-01 2.77765006e-01 8.66740584e-01
-2.99900442e-01 5.91500401e-01 -6.42435730e-01 -5.94295859e-01
1.05551553e+00 -1.20795822e+00 -4.97163415e-01 -5.54999471e-01
6.80109262e-01 -4.67882991e-01 5.24943888e-01 5.40533185e-01
-1.22578835e+00 1.63665682e-01 -1.15677524e+00 4.23937291e-01
9.73913968e-02 -2.82545745e-01 8.16671133e-01 4.31068003e-01
-1.16744280e+00 4.59236711e-01 -1.11755896e+00 -5.05272150e-01
5.71020655e-02 7.04997182e-01 -4.79913980e-01 -4.42020923e-01
-5.21531761e-01 9.66623366e-01 3.64247143e-01 1.81564271e-01
-1.04197598e+00 -3.32677186e-01 -6.65959597e-01 -1.04242653e-01
1.03407288e+00 -7.70141602e-01 1.44408250e+00 -5.41206241e-01
-1.36786950e+00 -4.31548059e-02 -1.89772904e-01 -3.65143955e-01
4.47234869e-01 -3.67432013e-02 3.56749296e-01 1.68167487e-01
3.69711459e-01 3.95639300e-01 5.84280491e-01 -1.46837497e+00
-9.26074505e-01 -4.07615393e-01 5.80463290e-01 6.22800410e-01
-2.24420670e-02 -3.02749693e-01 -3.67001235e-01 -4.40275595e-02
4.79180992e-01 -1.43220770e+00 -1.06108487e+00 -3.09615768e-02
-1.18510574e-01 -1.87104970e-01 3.84217143e-01 -8.35359320e-02
6.17649198e-01 -1.91104341e+00 8.94817472e-01 4.76962686e-01
3.91808301e-01 -4.18265641e-01 -3.41518044e-01 8.09855580e-01
4.02621448e-01 -2.24144265e-01 -6.70183748e-02 -2.89254248e-01
8.85378197e-02 6.10660017e-01 6.69821650e-02 9.45016861e-01
-3.31224144e-01 5.56554914e-01 -1.18722832e+00 -1.66694820e-01
-1.03501985e-02 -3.12952340e-01 -9.17442560e-01 -2.73752749e-01
-3.74037206e-01 6.70180321e-02 -6.30554855e-01 2.84420371e-01
6.69217110e-01 -1.03023931e-01 7.12442160e-01 2.98645139e-01
-2.28992298e-01 6.31278530e-02 -1.63707662e+00 1.78782439e+00
-4.11468565e-01 3.95699948e-01 7.88502038e-01 -8.74535203e-01
4.62456167e-01 6.42960593e-02 7.20315933e-01 -1.36526525e-01
8.77337754e-02 4.19962555e-01 2.57278010e-02 -1.25398651e-01
7.01006472e-01 -6.25810176e-02 -3.11875999e-01 7.45109200e-01
-2.40161926e-01 -5.68400323e-02 6.52499676e-01 2.07482889e-01
1.44401741e+00 -1.14773028e-02 1.20848589e-01 -3.58086795e-01
3.46198156e-02 6.40059352e-01 6.27339005e-01 8.11896384e-01
-1.31397899e-02 -1.88063048e-02 5.69342375e-01 -2.73291796e-01
-1.03727293e+00 -7.57708788e-01 4.71595228e-01 6.80851042e-01
7.09825754e-01 -5.85862875e-01 -8.32606018e-01 -3.81852388e-01
1.58479720e-01 5.45242310e-01 -6.01818502e-01 1.17786914e-01
-5.28240561e-01 -6.54080033e-01 -6.45453334e-02 1.33948073e-01
1.27500802e-01 -5.61736166e-01 -1.21096766e+00 5.28258562e-01
-9.45577100e-02 -9.29517746e-01 -2.42220789e-01 3.79840612e-01
-6.84841633e-01 -1.24724233e+00 -3.94936860e-01 -7.89412022e-01
1.10194433e+00 8.13480616e-01 5.91313660e-01 -6.20992184e-02
-1.43594682e-01 5.67274213e-01 -4.44122821e-01 -1.99667200e-01
-2.38698602e-01 1.28067762e-01 2.96954453e-01 -4.94063556e-01
-2.40970775e-01 -5.23860216e-01 -3.82228911e-01 3.46273154e-01
-5.58615267e-01 9.31612700e-02 6.57789767e-01 6.54210687e-01
5.05499303e-01 7.32113838e-01 2.68914819e-01 -2.47923926e-01
8.14437389e-01 -5.01909733e-01 -1.02730882e+00 1.66362271e-01
-3.83570731e-01 2.91119307e-01 3.79469335e-01 -4.61291015e-01
-7.02297091e-01 4.85029668e-01 7.33889103e-01 -4.03478742e-01
2.08229989e-01 8.24136734e-01 -8.45146365e-04 -2.27263242e-01
4.95032281e-01 -9.41659212e-02 4.52836037e-01 -1.73123389e-01
5.25174618e-01 2.60528922e-01 3.09390843e-01 -6.39380574e-01
8.12825739e-01 4.59976465e-01 4.66616303e-01 -6.29134178e-01
-1.41972631e-01 -2.93117881e-01 -2.24393338e-01 -3.98649603e-01
4.59422886e-01 -5.58654249e-01 -1.02953172e+00 1.83831096e-01
-1.07533002e+00 -3.82459760e-01 -2.34321401e-01 4.88332003e-01
-9.08316612e-01 2.61928588e-01 -1.28656834e-01 -1.11444163e+00
3.72636944e-01 -1.37501073e+00 7.38763213e-01 8.68594944e-02
-2.47393072e-01 -4.00686711e-01 2.39532530e-01 1.55967083e-02
-8.60336237e-03 2.34870702e-01 7.88178563e-01 -3.56675595e-01
-9.04029131e-01 -3.28450650e-01 1.54183149e-01 -6.29689813e-01
1.04401059e-01 -2.40413100e-01 4.14161533e-02 -6.58027470e-01
-1.46801144e-01 -1.09677494e-01 2.07659960e-01 3.76386374e-01
1.86854064e-01 -4.85425234e-01 -9.16680098e-01 1.10171894e-02
1.38992834e+00 3.16287369e-01 2.37428352e-01 7.20366120e-01
2.98989028e-01 7.24733233e-01 9.57850814e-01 6.52390480e-01
4.56255436e-01 6.20696068e-01 5.87589443e-01 5.20234823e-01
3.55891615e-01 -2.83908024e-02 4.23182666e-01 3.23915243e-01
-2.66649891e-02 -4.16176915e-01 -8.58860731e-01 8.13673437e-01
-2.33576560e+00 -7.17847943e-01 -2.61624344e-03 2.07867575e+00
3.21027160e-01 -1.32941261e-01 4.57609057e-01 1.86138898e-01
6.28153145e-01 -3.72087151e-01 -7.48783290e-01 -2.27035552e-01
3.51784825e-01 -4.63581204e-01 1.00850403e+00 8.10055196e-01
-7.23860323e-01 7.14124680e-01 6.58098841e+00 3.46455723e-01
-4.10083532e-01 -8.69995207e-02 2.74484325e-02 -4.33768541e-01
-3.56628090e-01 2.25296333e-01 -4.00012136e-01 -4.52606156e-02
7.18792975e-01 -3.70015204e-01 1.10645127e+00 1.01699615e+00
6.52711540e-02 -4.29233849e-01 -1.09538293e+00 6.65178597e-01
-3.01410675e-01 -1.19467139e+00 -3.13525259e-01 6.36764646e-01
7.66481280e-01 7.51432925e-02 -3.10470551e-01 -1.52387500e-01
1.00243235e+00 -6.96228743e-01 1.07132566e+00 -5.36333174e-02
1.38475522e-01 -1.18933380e+00 3.87643933e-01 6.44399107e-01
-1.22427571e+00 -4.92014587e-01 -4.66495037e-01 -3.10647249e-01
5.08931816e-01 2.18919605e-01 -9.36767459e-01 9.12709951e-01
6.18554354e-01 1.20787017e-01 2.45608822e-01 1.30757105e+00
6.43045008e-02 -2.70515442e-01 -8.70056987e-01 -3.88337493e-01
5.28321147e-01 -5.49910426e-01 8.47277462e-01 3.27833921e-01
4.61267054e-01 2.74668038e-01 5.53677201e-01 5.04943609e-01
4.70358551e-01 -2.60781884e-01 -5.71173489e-01 -2.35975012e-01
6.91702962e-01 1.10816765e+00 -1.31115317e+00 -2.55981158e-03
-2.41280720e-01 7.53395259e-01 4.67593163e-01 3.85107219e-01
-6.68857634e-01 -4.44616377e-01 8.47330213e-01 -1.63687527e-01
5.25787354e-01 -8.97191107e-01 -2.05565959e-01 -8.27949226e-01
8.88449848e-02 -1.03856254e+00 1.10770613e-01 -4.03238297e-01
-7.55780935e-01 5.31591773e-01 1.09894603e-01 -1.05907953e+00
-5.96384287e-01 -4.36032206e-01 -5.54309934e-02 4.13275182e-01
-9.56012726e-01 -6.85081720e-01 -5.02530634e-02 3.79483432e-01
6.39371753e-01 4.42163050e-02 6.43316329e-01 -1.99395657e-01
-4.83927190e-01 2.77870242e-03 2.97649324e-01 -6.49310052e-01
6.49886280e-02 -9.01619434e-01 2.80684292e-01 1.08520365e+00
-1.50531650e-01 6.07028902e-01 1.10680699e+00 -8.18371832e-01
-2.16288757e+00 -6.34011567e-01 4.15199399e-01 -9.44387987e-02
8.42847109e-01 -2.79774487e-01 -1.05996780e-01 7.86727190e-01
1.07354566e-01 -4.61810410e-01 2.16367900e-01 3.72589946e-01
1.78833529e-01 1.73377082e-01 -1.06749356e+00 1.02285540e+00
1.23956370e+00 1.48068339e-01 -2.00575307e-01 4.57363307e-01
7.02345788e-01 -3.28783661e-01 -6.02120996e-01 2.02844888e-01
4.90610629e-01 -5.98237634e-01 6.84400618e-01 -4.07092512e-01
-6.45828098e-02 -7.99559951e-01 -2.45197982e-01 -1.56245971e+00
-5.21947563e-01 -1.11202741e+00 3.77026312e-02 5.22426307e-01
4.90633100e-01 -8.47099483e-01 9.05259788e-01 8.25071096e-01
-2.32907906e-01 -7.58282900e-01 -9.93870080e-01 -1.15568686e+00
-9.20831412e-02 -1.81791529e-01 7.06451178e-01 4.93250221e-01
5.60552537e-01 4.21875387e-01 -1.96418941e-01 5.48157036e-01
7.76986480e-01 3.85639489e-01 1.03342021e+00 -1.12978518e+00
-4.74046350e-01 -3.65466386e-01 -1.42150491e-01 -1.20575666e+00
6.78341985e-02 -5.74354529e-01 4.65286374e-01 -1.97090626e+00
5.02496213e-02 -9.26842213e-01 4.02542830e-01 3.96876991e-01
2.99414456e-01 -3.27329189e-01 4.27884161e-01 2.16040224e-01
-7.15382040e-01 4.49603081e-01 1.31016982e+00 5.37689626e-02
-3.43749940e-01 -5.64377606e-02 -6.15067720e-01 6.68627977e-01
7.17460871e-01 -4.11897868e-01 -7.56502032e-01 -6.66574717e-01
5.29428959e-01 3.96822363e-01 2.68322546e-02 -8.53809476e-01
4.27985668e-01 -7.01605976e-01 -1.84037924e-01 -4.90120858e-01
4.14411306e-01 -1.08722723e+00 7.38024056e-01 8.25803697e-01
-1.10849902e-01 2.70685494e-01 3.95089313e-02 1.01827073e+00
3.03734660e-01 -6.34459198e-01 3.18108439e-01 -3.21997285e-01
-4.04767811e-01 1.01234838e-01 -9.02602673e-01 -6.35101736e-01
1.49298048e+00 -4.05244529e-01 -2.24618167e-01 -3.38978946e-01
-5.88855267e-01 7.30824351e-01 9.12577748e-01 4.92575169e-02
4.49152619e-01 -1.01737130e+00 -3.14561009e-01 -2.18606144e-01
-2.69731581e-01 4.96692300e-01 8.36155415e-02 8.63147914e-01
-5.88909149e-01 3.99882346e-01 -2.21223205e-01 -2.20547467e-01
-1.16474807e+00 8.86514068e-01 4.92631756e-02 -1.76176026e-01
-7.31958449e-01 7.06856847e-01 7.38135204e-02 -3.03620219e-01
1.64229363e-01 -2.48719126e-01 1.55306667e-01 3.07740644e-04
3.69494289e-01 7.96602309e-01 -3.40049833e-01 -2.02533990e-01
-4.74485636e-01 3.59396875e-01 -1.93510294e-01 -5.46733797e-01
1.61235905e+00 -4.07999158e-01 -3.48957628e-01 -3.19018364e-02
6.84495747e-01 1.25218043e-03 -1.31035924e+00 9.32422727e-02
1.47838406e-02 -5.72938204e-01 -1.49463341e-01 -3.68154407e-01
-9.93005455e-01 -1.38647169e-01 -1.40488178e-01 5.99954247e-01
9.59287703e-01 2.07842797e-01 4.71917450e-01 8.79162550e-01
1.27729535e+00 -1.05798614e+00 -6.88215494e-02 4.07878876e-01
8.42801332e-01 -4.47180092e-01 2.29745299e-01 -7.01437771e-01
-4.46909457e-01 1.07153130e+00 4.24102306e-01 -4.42558110e-01
1.56437039e-01 3.57176304e-01 -4.06369686e-01 -1.62653029e-01
-9.11000729e-01 -1.58373252e-01 -5.61814427e-01 7.30484009e-01
-5.86901844e-01 9.48145390e-02 -4.09485459e-01 3.14874947e-01
-4.17393863e-01 -2.33757660e-01 9.52114940e-01 1.37871599e+00
-7.65135169e-01 -1.18455553e+00 -5.20966530e-01 2.34746724e-01
3.12033713e-01 5.53278208e-01 -2.68651307e-01 7.74195910e-01
-2.63453692e-01 1.24381447e+00 -6.35263175e-02 -3.30483109e-01
2.86866903e-01 -3.84779721e-01 9.25152779e-01 -3.78439218e-01
4.95545268e-02 2.07736835e-01 5.26864529e-01 -7.57923782e-01
-5.19205987e-01 -1.09696114e+00 -1.38579988e+00 -5.33182442e-01
-5.03914475e-01 3.66500854e-01 6.05829120e-01 9.81485426e-01
4.64861631e-01 4.38611537e-01 4.09781575e-01 -1.16453683e+00
-5.53638160e-01 -3.34384382e-01 -3.92762333e-01 -4.40874130e-01
3.11248004e-01 -1.12365937e+00 -2.55699068e-01 -3.52585375e-01] | [4.865657806396484, 1.7048355340957642] |
78b010e6-2b2d-4455-881d-6b9fe6e571da | a-clarification-of-misconceptions-myths-and | 2008.05607 | null | https://arxiv.org/abs/2008.05607v1 | https://arxiv.org/pdf/2008.05607v1.pdf | A clarification of misconceptions, myths and desired status of artificial intelligence | The field artificial intelligence (AI) has been founded over 65 years ago. Starting with great hopes and ambitious goals the field progressed though various stages of popularity and received recently a revival in the form of deep neural networks. Some problems of AI are that so far neither 'intelligence' nor the goals of AI are formally defined causing confusion when comparing AI to other fields. In this paper, we present a perspective on the desired and current status of AI in relation to machine learning and statistics and clarify common misconceptions and myths. Our discussion is intended to uncurtain the veil of vagueness surrounding AI to see its true countenance. | ['Olli Yli-Harja', 'Frank Emmert-Streib', 'Matthias Dehmer'] | 2020-08-03 | null | null | null | null | ['misconceptions'] | ['miscellaneous'] | [ 2.11437613e-01 4.04400140e-01 -1.97185636e-01 -5.47846854e-01
-1.21852346e-02 -3.00200194e-01 1.03894532e+00 4.30751182e-02
-4.48409766e-01 8.93734217e-01 2.42210925e-01 -4.64326531e-01
-4.25663054e-01 -7.44639575e-01 -1.67005658e-01 -6.19667530e-01
-2.78011709e-01 6.18360162e-01 -2.65771121e-01 -4.91167396e-01
5.51182091e-01 5.22094786e-01 -1.53865921e+00 1.92808956e-01
7.79942811e-01 1.07209885e+00 -4.03297961e-01 5.63584924e-01
-2.95411050e-01 1.48178923e+00 -1.00175977e+00 -5.52994788e-01
1.71912849e-01 -3.75598758e-01 -1.01341236e+00 -3.99811156e-02
2.37658441e-01 6.08037636e-02 -4.11218077e-01 1.04835916e+00
5.55886589e-02 -1.94098845e-01 6.11497223e-01 -1.42435002e+00
-8.74434292e-01 5.86550593e-01 -2.16648549e-01 4.49378729e-01
1.00531392e-01 2.48265103e-01 1.00226414e+00 -4.50786829e-01
1.36079475e-01 1.24370551e+00 7.84368217e-01 5.53642094e-01
-8.35687578e-01 -4.81833518e-01 1.81094140e-01 3.29929084e-01
-1.32817888e+00 -3.45203668e-01 6.47421122e-01 -3.94879222e-01
7.71306098e-01 4.91434544e-01 8.66521478e-01 8.16283107e-01
4.58860636e-01 1.00441658e+00 1.14396143e+00 -5.24840713e-01
3.93887967e-01 1.59148633e-01 2.28705153e-01 5.34894049e-01
5.05271375e-01 1.99539825e-01 -2.13637218e-01 -1.37665868e-03
8.46197009e-01 -1.11915909e-01 9.17909071e-02 -7.12301433e-02
-1.33442819e+00 1.12916291e+00 4.29026514e-01 1.18413126e+00
-5.26379108e-01 6.98903278e-02 3.72358501e-01 3.46795827e-01
2.11559609e-01 8.69576335e-01 -5.82583368e-01 -1.78853974e-01
-6.90731704e-01 1.36084691e-01 1.12337971e+00 4.99874383e-01
5.88926375e-01 3.83903056e-01 4.90894049e-01 6.85998261e-01
3.61672603e-02 1.48128524e-01 5.74044406e-01 -1.03181005e+00
-2.25539804e-01 5.34249783e-01 5.59556708e-02 -1.28577626e+00
-5.71951807e-01 -5.82767487e-01 -1.13180137e+00 3.52894098e-01
5.61399877e-01 -4.11210269e-01 -1.01824093e+00 1.35272038e+00
-2.72647083e-01 -1.81839213e-01 2.34954417e-01 8.69307995e-01
1.01495957e+00 5.91700137e-01 2.83912033e-01 -2.69545428e-02
9.69585419e-01 -4.45254773e-01 -9.12699461e-01 -5.55273056e-01
3.02697629e-01 -5.71162939e-01 5.77706695e-01 7.91329801e-01
-9.47608292e-01 -3.41185242e-01 -1.02675545e+00 -8.34517032e-02
-8.63292158e-01 -3.32920790e-01 1.32115543e+00 8.07443559e-01
-9.63157535e-01 5.38727999e-01 -6.78216398e-01 -3.85908633e-01
5.01189828e-01 5.16462445e-01 -4.59517799e-02 1.36223257e-01
-1.22364020e+00 1.49366808e+00 8.39252472e-01 2.34479994e-01
-2.76700705e-01 -4.17084187e-01 -7.80764461e-01 -3.31137717e-01
8.00181687e-01 -6.86640859e-01 1.22326374e+00 -1.29014754e+00
-1.33119130e+00 1.11434484e+00 -5.38134351e-02 -8.50977302e-01
2.84869909e-01 -1.23950623e-01 -8.52640927e-01 -1.93908438e-01
-1.47369325e-01 4.89905804e-01 1.25943348e-01 -1.22327983e+00
-9.64637995e-01 -4.57944810e-01 -1.45153720e-02 4.15220037e-02
1.86853796e-01 2.33523265e-01 5.87635115e-02 -6.62443697e-01
-8.82610306e-03 -5.57260692e-01 -4.39543605e-01 -1.60420701e-01
9.20590088e-02 -5.70623696e-01 4.29992974e-01 -1.17374197e-01
1.31115222e+00 -1.99464285e+00 -1.90560609e-01 9.89495292e-02
5.93839109e-01 3.00078541e-01 2.16122180e-01 5.06637335e-01
-1.38777122e-01 9.03139636e-02 -3.08193505e-01 6.43673599e-01
2.45180696e-01 5.69513857e-01 -3.80696841e-02 5.68489671e-01
3.14747632e-01 1.27022159e+00 -1.07266521e+00 -4.59158063e-01
5.33852994e-01 4.46695507e-01 -9.09985676e-02 -1.88362926e-01
-1.05754979e-01 -3.43526304e-02 -5.89767039e-01 8.17421854e-01
3.42261046e-01 -4.45043445e-01 6.00312240e-02 5.63802570e-02
-3.79777730e-01 6.47747517e-01 -8.41354012e-01 1.33533549e+00
-4.40414548e-02 1.04467344e+00 -4.59822651e-04 -1.60161555e+00
1.10579205e+00 3.69060248e-01 7.74940729e-01 -6.49336398e-01
5.90193868e-01 4.43832636e-01 7.09692955e-01 -5.03737867e-01
3.40130419e-01 -3.81060213e-01 -8.50116983e-02 2.27695554e-01
-1.40625596e-01 -5.46529531e-01 3.26134935e-02 -1.43826038e-01
8.49288583e-01 -1.17963314e-01 8.34584177e-01 -6.00378156e-01
4.88985002e-01 4.24909562e-01 3.30762893e-01 7.66166329e-01
-3.48066628e-01 3.12187076e-01 2.85164237e-01 -1.41650558e+00
-1.05745459e+00 -1.07217634e+00 -3.57663214e-01 9.47156549e-01
-1.80491656e-01 -2.35559583e-01 -5.62704980e-01 -4.36446190e-01
-1.56465888e-01 1.02612031e+00 -8.17464352e-01 1.57545856e-03
-4.50921834e-01 -9.98497009e-01 2.49615058e-01 4.12929535e-01
6.14212930e-01 -1.44687688e+00 -1.00936842e+00 1.73369840e-01
2.28885099e-01 -8.58400881e-01 5.04938841e-01 3.46748382e-01
-7.11198330e-01 -8.97483170e-01 -7.05106080e-01 -7.41050839e-01
2.54207999e-01 -3.26808542e-02 1.45298433e+00 4.94530611e-02
-2.03684479e-01 1.97600484e-01 -1.05665877e-01 -1.10991788e+00
-6.39278233e-01 -4.80962433e-02 -1.30487084e-01 -5.24577916e-01
1.38821018e+00 -5.47851682e-01 -3.86207014e-01 -1.71040431e-01
-1.04204416e+00 -2.45168746e-01 9.68370974e-01 5.20858943e-01
-2.96140909e-01 6.86560720e-02 6.31791174e-01 -9.86364484e-01
8.14890027e-01 -5.18903732e-01 -2.60861337e-01 1.66995943e-01
-7.61061251e-01 -1.56596884e-01 3.27497214e-01 -2.72995263e-01
-7.19780505e-01 -4.88887012e-01 -2.27650896e-01 8.30705017e-02
-7.05758750e-01 6.33463025e-01 2.01457113e-01 3.20760310e-02
8.04273307e-01 1.04169741e-01 6.73820674e-02 -2.53321618e-01
2.77497500e-01 9.52532589e-01 7.27958560e-01 -4.99281615e-01
4.65206832e-01 4.46843445e-01 -1.20044500e-01 -1.22618115e+00
-9.78679180e-01 -8.12049657e-02 -5.95258951e-01 -1.87328935e-01
5.74073255e-01 -4.26557094e-01 -9.15488660e-01 2.96675324e-01
-1.17325616e+00 1.88029885e-01 -6.42594278e-01 5.23335934e-01
-4.24837947e-01 2.17046812e-01 -4.00672048e-01 -8.94860685e-01
-2.34409332e-01 -8.83013666e-01 3.54359418e-01 3.76299858e-01
-7.47607827e-01 -1.25334549e+00 -8.41400325e-02 1.65723354e-01
6.08437777e-01 4.58424687e-01 6.92910671e-01 -9.77897525e-01
6.64947033e-02 -4.67692465e-01 -2.65724987e-01 3.79532367e-01
3.60175192e-01 -1.19943492e-01 -9.98089731e-01 2.79245555e-01
3.11755240e-01 -3.82827193e-01 5.77073276e-01 4.84591484e-01
9.64216352e-01 -6.25896573e-01 -3.42986733e-02 6.83379620e-02
1.46960354e+00 9.52459455e-01 7.58648396e-01 6.22828960e-01
2.11194575e-01 7.24252343e-01 1.90672219e-01 4.58050281e-01
1.81482852e-01 2.70648729e-02 3.00419092e-01 -1.14956550e-01
1.73925817e-01 1.74987867e-01 -1.37639299e-01 6.85966969e-01
-4.27571386e-01 -2.24507928e-01 -1.30427837e+00 6.39740765e-01
-1.44443738e+00 -9.07333851e-01 -4.50289957e-02 1.94577122e+00
7.14167416e-01 6.45652413e-01 2.17434570e-01 2.57164717e-01
5.63562095e-01 2.36747831e-01 -4.08669978e-01 -9.42553043e-01
-1.78201824e-01 3.03245306e-01 1.75437808e-01 5.47700644e-01
-1.08389401e+00 1.00153208e+00 7.95560598e+00 7.28230774e-01
-1.20294344e+00 -3.45940262e-01 9.33446586e-01 4.76859361e-01
-2.70387948e-01 -2.89071053e-02 -3.37872028e-01 2.70665705e-01
8.09676945e-01 -4.27824140e-01 3.31652641e-01 7.31779993e-01
-1.26442865e-01 -1.83750182e-01 -9.96472538e-01 7.45302975e-01
1.82344884e-01 -1.41398954e+00 5.81057519e-02 1.76326167e-02
4.62442100e-01 2.75298029e-01 2.56895632e-01 1.82415739e-01
8.97343695e-01 -1.64443767e+00 3.67653310e-01 4.14263070e-01
3.89624596e-01 -7.10343897e-01 7.75852621e-01 2.86377072e-01
-4.32337165e-01 -1.62044212e-01 -5.53648114e-01 -8.88348758e-01
-2.43985564e-01 6.64710104e-01 -8.96400273e-01 3.51037085e-01
5.04495144e-01 5.60755253e-01 -3.02612275e-01 9.73696232e-01
9.23130363e-02 4.13953602e-01 -5.17017901e-01 -6.13371253e-01
1.00369310e+00 -2.41706371e-01 4.81441379e-01 1.42233670e+00
-2.63503850e-01 3.98466676e-01 -4.30959612e-02 1.05908465e+00
3.94844592e-01 -9.09843445e-02 -8.16590726e-01 -5.33440173e-01
2.26180851e-01 8.73611689e-01 -7.73662031e-01 -4.24287677e-01
-7.48230755e-01 4.26335335e-01 -1.50389552e-01 3.57295811e-01
-3.72116536e-01 -3.03889543e-01 5.64906180e-01 -1.81198284e-01
-1.47425339e-01 -1.82300076e-01 -9.35820878e-01 -1.04938257e+00
-3.81803930e-01 -9.90903080e-01 3.91340047e-01 -4.97306377e-01
-1.39145994e+00 7.67769814e-01 7.61466175e-02 -7.30036855e-01
-4.93620932e-01 -9.24362183e-01 -5.11432707e-01 6.47851467e-01
-1.02289605e+00 -9.12032127e-01 -1.01218253e-01 4.03574295e-02
5.07665217e-01 -3.71853322e-01 1.09607434e+00 -1.58178821e-01
-2.34170854e-01 1.64016068e-01 1.99124530e-01 4.45040911e-01
1.51214257e-01 -1.20300519e+00 4.68661547e-01 2.36359000e-01
2.34415933e-01 4.38105106e-01 1.26331484e+00 -2.72939682e-01
-1.05810475e+00 -3.21941495e-01 1.09934568e+00 -4.07515287e-01
8.96892667e-01 7.30443895e-02 -7.47940421e-01 8.95463586e-01
6.33315682e-01 -3.82913202e-01 5.19428134e-01 2.56425947e-01
-6.42019883e-02 8.25412199e-03 -1.04308605e+00 7.23739088e-01
4.17700320e-01 -1.87782552e-02 -1.19981718e+00 2.07893014e-01
3.46195728e-01 8.67110863e-03 -8.85067105e-01 7.83039093e-01
8.60005319e-01 -1.13946557e+00 8.37606311e-01 -7.42542267e-01
4.66697872e-01 2.77873635e-01 2.90132482e-02 -9.63084042e-01
-4.16130960e-01 -7.82883286e-01 3.11206490e-01 7.42710650e-01
2.44750530e-01 -7.78252482e-01 1.06089735e+00 8.03050458e-01
-8.07058662e-02 -9.55314398e-01 -8.39921951e-01 -5.81637204e-01
5.71817577e-01 -5.78834653e-01 4.33968872e-01 1.31401718e+00
2.72623360e-01 6.22000694e-01 -8.28935355e-02 -5.31566501e-01
2.84215838e-01 8.17028284e-02 4.43359941e-01 -1.56205380e+00
2.85923034e-01 -1.05737376e+00 -8.39920580e-01 -8.12954426e-01
-1.66052520e-01 -4.54397023e-01 -2.23641455e-01 -1.48689163e+00
-1.02744833e-01 -1.69223949e-01 -4.66043979e-01 4.59620476e-01
2.74118751e-01 2.35852852e-01 1.30915239e-01 -1.01109043e-01
-5.19391119e-01 -2.15599164e-02 1.10045981e+00 -1.39463142e-01
-1.06223285e-01 -1.42819015e-02 -1.16134751e+00 1.20353460e+00
1.23179233e+00 3.70337479e-02 -2.63960451e-01 -3.87482017e-01
4.99298424e-01 -2.49074101e-01 2.24335790e-01 -1.16118217e+00
1.58408135e-01 -4.72398490e-01 5.69061041e-01 -4.73389000e-01
2.85249233e-01 -8.96782696e-01 6.10321760e-02 5.07382572e-01
-5.14019132e-01 2.71001421e-02 3.81347984e-01 -9.34851393e-02
-3.95073831e-01 -3.89466852e-01 8.81189585e-01 -6.26860917e-01
-9.07654583e-01 -2.04221904e-02 -6.45860553e-01 2.14114204e-01
1.03256989e+00 -3.93269926e-01 -1.61429927e-01 -5.64690769e-01
-6.76803350e-01 9.77332145e-02 3.75580460e-01 3.84818852e-01
6.42773569e-01 -8.07655394e-01 -7.73276925e-01 -8.06870908e-02
-9.23337564e-02 -1.23253755e-01 -2.60932028e-01 5.70013702e-01
-7.53508270e-01 9.46614683e-01 -5.06393313e-01 -2.19332695e-01
-6.30684376e-01 8.05051982e-01 4.11593676e-01 6.85540512e-02
-9.08056378e-01 6.56142056e-01 2.51507133e-01 -6.36666268e-02
4.98428524e-01 -2.84209192e-01 -5.55904388e-01 -2.63893485e-01
6.23286605e-01 3.88077766e-01 -1.96603134e-01 -4.11921889e-01
-3.40045482e-01 1.39366537e-01 -2.11188674e-01 -1.13407128e-01
1.57966483e+00 1.58576280e-01 -3.73863548e-01 1.08337784e+00
9.19167638e-01 -4.06368256e-01 -5.11127353e-01 -8.02116543e-02
3.17493111e-01 -2.77172089e-01 1.59509167e-01 -1.29948461e+00
-8.22741449e-01 9.40242648e-01 3.48643035e-01 6.91300750e-01
1.07964945e+00 1.55042872e-01 5.81898928e-01 6.43333495e-01
2.42831707e-01 -1.05363107e+00 -3.41074377e-01 6.55043781e-01
8.94466639e-01 -1.28160429e+00 1.40003309e-01 -1.61107868e-01
-7.77063608e-01 1.28860950e+00 3.88722807e-01 -2.88521409e-01
6.72415555e-01 4.17361706e-01 1.72601387e-01 -5.05138576e-01
-8.07015359e-01 -2.63553619e-01 3.23456585e-01 7.94851780e-01
1.00650454e+00 7.52289146e-02 -6.82622850e-01 4.15548027e-01
-6.07304692e-01 3.85174513e-01 3.33743811e-01 8.76941204e-01
-8.66889536e-01 -7.02148557e-01 -5.45995712e-01 5.73550761e-01
-9.50356007e-01 -1.25071123e-01 -9.45112586e-01 1.25314879e+00
3.87019098e-01 8.21250379e-01 2.16567934e-01 -3.22244972e-01
-1.72526330e-01 6.25667125e-02 5.16718566e-01 -3.47799718e-01
-5.12661815e-01 4.31155302e-02 1.66639924e-01 -1.83095261e-01
-5.40128827e-01 -4.91277516e-01 -1.06403017e+00 -6.99238300e-01
-4.22998816e-02 5.67866743e-01 6.76250219e-01 1.25669968e+00
-2.69790262e-01 2.91211843e-01 4.93843555e-02 -3.50489080e-01
-3.95245552e-01 -1.05478787e+00 -6.67208731e-01 8.80573876e-03
4.12647843e-01 -3.62830400e-01 -2.57991970e-01 -1.92172185e-01] | [9.039913177490234, 6.399143218994141] |
c3413361-2e2b-4087-a5f2-2e46467caf0e | affective-decoding-for-empathetic-response | 2108.08102 | null | https://arxiv.org/abs/2108.08102v3 | https://arxiv.org/pdf/2108.08102v3.pdf | Affective Decoding for Empathetic Response Generation | Understanding speaker's feelings and producing appropriate responses with emotion connection is a key communicative skill for empathetic dialogue systems. In this paper, we propose a simple technique called Affective Decoding for empathetic response generation. Our method can effectively incorporate emotion signals during each decoding step, and can additionally be augmented with an auxiliary dual emotion encoder, which learns separate embeddings for the speaker and listener given the emotion base of the dialogue. Extensive empirical studies show that our models are perceived to be more empathetic by human evaluations, in comparison to several strong mainstream methods for empathetic responding. | ['Chengkun Zeng', 'Zhigang Chen', 'Ruizhe Li', 'Chenghua Lin', 'Guanyi Chen'] | 2021-08-18 | null | https://aclanthology.org/2021.inlg-1.37 | https://aclanthology.org/2021.inlg-1.37.pdf | inlg-acl-2021-8 | ['empathetic-response-generation'] | ['natural-language-processing'] | [-3.25076729e-01 7.11108506e-01 9.07209665e-02 -7.97889948e-01
-6.17200255e-01 -2.87036330e-01 4.95511472e-01 -3.06011885e-02
-3.68390232e-01 7.62215614e-01 1.06760490e+00 4.20525521e-01
4.57017928e-01 -5.95915079e-01 1.56978548e-01 -4.61079329e-01
3.68441314e-01 4.82844502e-01 -9.61978257e-01 -1.08594978e+00
2.31189281e-02 9.54934955e-02 -9.91426528e-01 6.85307086e-01
5.07040858e-01 9.24071729e-01 -4.92921770e-01 8.93571258e-01
-1.76401392e-01 1.75243819e+00 -8.62512052e-01 -1.01663280e+00
-3.42650980e-01 -8.02171528e-01 -1.40775990e+00 -3.45056117e-01
-4.86207306e-01 -5.64303458e-01 -2.01692045e-01 5.52933514e-01
1.01423430e+00 6.56627595e-01 7.44239450e-01 -1.27868533e+00
-9.91974950e-01 1.02387488e+00 -3.55514288e-02 -3.07372808e-01
1.05365622e+00 -4.45214920e-02 1.28381026e+00 -1.00040984e+00
4.30442542e-01 1.46597028e+00 8.38774025e-01 1.26983631e+00
-7.58322001e-01 -7.24874496e-01 -1.51915073e-01 2.54892468e-01
-6.69313669e-01 -7.50550091e-01 1.09450603e+00 -1.06361493e-01
9.93124545e-01 1.72549903e-01 7.06219971e-01 1.75422692e+00
-1.28648847e-01 9.45390761e-01 7.97728717e-01 -3.21916312e-01
1.27550319e-01 5.68765342e-01 2.49826908e-01 6.68774322e-02
-9.38090444e-01 6.23083785e-02 -8.66267979e-01 -4.64115769e-01
2.78390080e-01 -4.54375893e-01 -2.99434304e-01 3.69647175e-01
-9.30067241e-01 1.67709482e+00 4.61649358e-01 -1.67908147e-03
-9.06651855e-01 2.11868778e-01 8.28096747e-01 7.63475657e-01
6.37623549e-01 9.14062083e-01 3.13878693e-02 -7.47276843e-01
-6.88897371e-02 2.19851956e-01 1.34094226e+00 6.45111263e-01
2.66685098e-01 1.92685470e-01 -2.52900004e-01 1.39512122e+00
2.00461432e-01 3.10834080e-01 5.53254426e-01 -1.34899855e+00
-1.12572961e-01 1.81061879e-01 3.10670525e-01 -8.69780242e-01
-6.35039806e-01 2.01733634e-01 -6.10478818e-01 -2.41716625e-03
-1.26892358e-01 -8.39939535e-01 1.90707162e-01 2.11424232e+00
3.29847455e-01 -3.83377433e-01 9.28752065e-01 1.01539838e+00
1.35971689e+00 1.06506550e+00 5.22151530e-01 -3.39922577e-01
1.17415750e+00 -1.34190166e+00 -1.09731770e+00 -4.52659070e-01
6.78796947e-01 -8.34899366e-01 1.13604116e+00 2.56995797e-01
-1.29021382e+00 -2.99872637e-01 -5.98719597e-01 -2.58349806e-01
9.06487405e-02 1.04263246e-01 1.02261257e+00 2.73627311e-01
-1.00792241e+00 1.92108288e-01 9.37392339e-02 -3.14585477e-01
-1.20865842e-02 2.53476769e-01 -5.66899598e-01 3.45018774e-01
-1.89046741e+00 1.55845869e+00 -2.93482453e-01 1.69313148e-01
-2.89101988e-01 -3.57665151e-01 -1.11612892e+00 -9.16134287e-03
-6.84371889e-01 -7.35427737e-01 1.89370346e+00 -1.37316608e+00
-2.48208404e+00 9.42362130e-01 3.06937043e-02 -2.15021953e-01
1.02184474e-01 -3.76855910e-01 -3.92544925e-01 3.65292758e-01
-2.67525911e-01 1.04792356e+00 6.79560304e-01 -1.05643356e+00
-1.72098801e-01 3.88465659e-03 2.62944847e-01 7.60103762e-01
-5.74375212e-01 5.69066167e-01 4.41673338e-01 -4.47651595e-01
-4.98290122e-01 -7.53011703e-01 -4.04310048e-01 -1.81636989e-01
5.02659343e-02 -6.04299903e-01 3.26589167e-01 -5.67889273e-01
9.43592906e-01 -2.02556539e+00 3.48479837e-01 -1.12552807e-01
2.36026332e-01 -1.15460418e-01 -4.07133937e-01 9.66633081e-01
-2.58065224e-01 -4.73594397e-01 1.51626855e-01 -5.53761184e-01
3.81901354e-01 9.07771103e-03 -6.74418688e-01 3.44535768e-01
2.40021050e-01 9.62369084e-01 -8.78282607e-01 -2.58894920e-01
2.83640381e-02 6.79796994e-01 -8.13679516e-01 9.33092415e-01
1.87175959e-01 4.95016754e-01 -4.88874525e-01 7.42882723e-04
2.07825273e-01 7.94329122e-02 -1.76168755e-02 6.23236001e-02
2.01070592e-01 6.06711924e-01 -4.18574244e-01 1.56856716e+00
-1.09097373e+00 7.40816355e-01 2.93339700e-01 -6.72828734e-01
1.41740060e+00 9.90844488e-01 3.72375280e-01 -7.08507359e-01
6.29085064e-01 -2.13013247e-01 -1.04155712e-01 -6.01090789e-01
8.70824635e-01 -9.12199199e-01 -8.17017376e-01 1.32746422e+00
-7.31385825e-03 -4.09181744e-01 -5.58336139e-01 3.24267924e-01
7.64285266e-01 -2.05115736e-01 4.42040831e-01 1.36002481e-01
5.85189819e-01 -1.05148531e-01 2.17469916e-01 2.96336263e-01
-6.80987537e-01 1.03688262e-01 5.43958306e-01 -5.56279778e-01
-5.85655332e-01 -6.75808609e-01 1.85583472e-01 1.73325598e+00
5.02286330e-02 -1.53338924e-01 -7.06634939e-01 -3.13550204e-01
-4.28689420e-01 1.08054459e+00 -8.09189558e-01 -7.31098711e-01
-4.63798046e-01 -3.62765372e-01 8.92080843e-01 7.29295671e-01
-5.30643836e-02 -1.75101328e+00 -6.14063621e-01 4.36567754e-01
-7.51375437e-01 -9.48247075e-01 -3.80394340e-01 2.47855052e-01
-1.43715993e-01 -4.61578965e-01 -4.96916085e-01 -8.91776323e-01
2.80565500e-01 -6.78520203e-02 1.16289675e+00 -2.09254757e-01
1.87031507e-01 6.45219147e-01 -7.97788203e-01 -3.92691880e-01
-1.00126612e+00 -3.21019500e-01 1.75886169e-01 -1.59061611e-01
5.88569224e-01 -5.39442122e-01 -5.90964139e-01 2.43059155e-02
-3.49897772e-01 1.14766896e-01 1.23555064e-01 9.53127265e-01
-3.33420783e-01 -9.40000653e-01 1.48068130e+00 -7.24412739e-01
1.71904385e+00 -8.62945974e-01 8.07327271e-01 -2.53333244e-02
-8.54792669e-02 -2.05659747e-01 9.71290112e-01 -6.21073663e-01
-1.43393743e+00 -8.62138942e-02 -8.82108212e-01 6.52385429e-02
-9.05173272e-02 3.03683937e-01 1.73137322e-01 5.68644367e-02
9.47499454e-01 -2.91079134e-01 3.53446394e-01 -4.39851265e-03
9.09197807e-01 1.39255893e+00 6.68560803e-01 -8.70954275e-01
-1.66126620e-02 1.19461894e-01 -8.23548198e-01 -4.69455987e-01
-9.93077040e-01 -4.06058073e-01 -6.57998770e-02 -6.67185605e-01
7.89925992e-01 -1.07618916e+00 -1.17560232e+00 2.84544647e-01
-1.72958910e+00 -4.28762674e-01 -3.71944487e-01 5.29931724e-01
-1.03523290e+00 2.26883173e-01 -1.33378911e+00 -8.76884520e-01
-9.74860191e-01 -7.52423465e-01 7.75330544e-01 3.39293808e-01
-1.38394296e+00 -1.30885661e+00 6.27536118e-01 4.82167661e-01
6.68168247e-01 2.22113691e-02 7.44506598e-01 -9.69975889e-01
1.06658494e+00 -3.66454482e-01 1.07037872e-02 3.78462821e-01
-6.60823062e-02 -2.27291644e-01 -1.30334902e+00 3.08666527e-01
4.49817955e-01 -1.36815631e+00 1.97260290e-01 -5.02322577e-02
4.32664365e-01 -7.08295345e-01 4.01224792e-01 3.25337857e-01
4.93077576e-01 1.03896689e-02 5.64454257e-01 -1.87645536e-02
1.91291839e-01 1.24352884e+00 7.84864724e-01 1.10334909e+00
7.77264953e-01 5.51609874e-01 1.70696124e-01 -2.09398985e-01
3.04503202e-01 -1.90027475e-01 9.21600223e-01 1.06180835e+00
2.42790103e-01 -2.30914503e-01 -4.57531750e-01 2.66469508e-01
-1.85855889e+00 -1.30182445e+00 -1.31918505e-01 1.28206027e+00
1.35389960e+00 -7.42528915e-01 1.22135215e-01 -2.38771871e-01
5.16338229e-01 3.37530911e-01 -2.40133226e-01 -1.60749745e+00
-1.37438819e-01 6.11676812e-01 -4.58360553e-01 8.97057533e-01
-6.96875453e-01 1.18768418e+00 7.21446466e+00 1.27531826e-01
-9.45935905e-01 8.76877382e-02 3.88280183e-01 -1.43702239e-01
-5.19422829e-01 -2.85635859e-01 -1.36573508e-01 5.43618985e-02
1.20477664e+00 -5.11085808e-01 2.70800322e-01 1.10173321e+00
2.86122173e-01 4.13216978e-01 -1.32527733e+00 1.04702640e+00
4.03584480e-01 -8.50919664e-01 -2.26461887e-01 -5.80382645e-01
4.29002881e-01 -5.63677013e-01 5.49864061e-02 8.01572025e-01
7.11387873e-01 -1.23988760e+00 5.02362370e-01 3.99707973e-01
5.29073119e-01 -9.90529478e-01 9.63072121e-01 2.09360480e-01
-4.82306868e-01 -1.54324308e-01 -2.98165232e-01 -6.23502433e-01
5.68267822e-01 -2.43643299e-01 -6.98641062e-01 -1.73937827e-01
4.03690457e-01 6.87617421e-01 1.99411541e-01 6.21631108e-02
-5.84433794e-01 4.83742863e-01 6.41307458e-02 -3.19142491e-01
3.06172103e-01 -1.61457494e-01 1.72321796e-01 1.41372478e+00
6.10559471e-02 5.19511044e-01 -2.00173885e-01 7.31357813e-01
-1.51737168e-01 5.82530260e-01 -6.16767168e-01 -6.85214996e-02
5.37648499e-01 1.42774475e+00 2.92740077e-01 -2.24133626e-01
-1.17333807e-01 1.35526431e+00 5.82320333e-01 1.80651769e-01
-1.01742911e+00 -3.42996687e-01 8.59407246e-01 -7.95671701e-01
-3.80735725e-01 4.89928216e-01 -2.19917923e-01 -7.44508147e-01
-5.57736278e-01 -1.06109357e+00 4.68843877e-01 -1.20415926e+00
-1.69903278e+00 8.27504814e-01 -5.72406650e-01 -9.18016374e-01
-6.90961361e-01 -4.02320802e-01 -1.15733242e+00 8.75003695e-01
-1.14851797e+00 -1.18506670e+00 -2.78257996e-01 6.97843075e-01
5.09205639e-01 -4.38967347e-02 1.67642188e+00 -3.65174538e-03
-4.59726840e-01 7.09106028e-01 -5.49952149e-01 8.26671049e-02
1.31275499e+00 -9.12903428e-01 -3.88382189e-02 -4.07675691e-02
-3.80193442e-01 5.41891575e-01 9.65379357e-01 1.81701899e-01
-9.92282391e-01 -5.06892204e-01 1.20515430e+00 -3.28234404e-01
7.68867850e-01 6.42477721e-03 -7.81689465e-01 6.81049526e-01
9.81621981e-01 -6.77248359e-01 1.72728598e+00 3.89332205e-01
-4.21107411e-01 4.09445822e-01 -1.27880716e+00 7.26097703e-01
4.79536325e-01 -6.38344228e-01 -1.05507326e+00 3.87662023e-01
8.14875722e-01 -2.18917415e-01 -1.07763481e+00 1.47964023e-02
6.03869259e-01 -9.76510763e-01 8.52850437e-01 -1.10999823e+00
1.13135731e+00 5.40989995e-01 -1.01916611e-01 -1.69358909e+00
-2.16334894e-01 -1.09673131e+00 -1.00659005e-01 1.14246678e+00
2.05383137e-01 -5.20470977e-01 3.92086923e-01 9.85668540e-01
-3.80743355e-01 -6.02942884e-01 -7.73405135e-01 1.59817904e-01
5.43541789e-01 -3.42524201e-01 5.23128033e-01 1.30069780e+00
1.26540506e+00 1.07563436e+00 -1.01128006e+00 -3.53588611e-01
-4.38772850e-02 7.94344544e-02 8.67944598e-01 -1.01201558e+00
-2.70790070e-01 -8.11204970e-01 7.18564242e-02 -1.00286555e+00
1.05064440e+00 -7.46839821e-01 4.08613056e-01 -1.60136580e+00
-2.82787029e-02 -1.17307857e-01 2.38384251e-02 4.30955768e-01
-2.78052300e-01 5.27221113e-02 -4.85768355e-02 -8.36990029e-02
-4.76069123e-01 1.16368520e+00 1.04616642e+00 1.58580750e-01
-2.39134908e-01 -8.61851033e-03 -1.30027354e+00 7.66297638e-01
9.32813406e-01 -3.57154429e-01 -3.14886063e-01 -1.32713750e-01
6.62266910e-01 6.85890794e-01 9.41511169e-02 -3.22665215e-01
2.34919012e-01 -2.94945747e-01 -2.64333170e-02 7.26758763e-02
9.55071747e-01 -4.62966442e-01 -3.57888728e-01 9.20947790e-02
-1.15669084e+00 1.50323719e-01 4.19154689e-02 2.17011809e-01
-5.39247334e-01 -3.70730788e-01 9.21475291e-01 8.89363512e-02
-4.48548883e-01 -9.23957005e-02 -8.99558425e-01 5.57633415e-02
9.00822163e-01 1.49442963e-02 -3.77804965e-01 -1.48824966e+00
-7.12052763e-01 4.37936544e-01 1.66062459e-01 5.32725334e-01
8.79662931e-01 -1.63942158e+00 -1.04487169e+00 -3.37260216e-01
3.59639764e-01 -4.99563336e-01 5.33623099e-01 8.45053375e-01
-2.28072092e-01 -1.80147752e-01 -4.60251391e-01 2.45421246e-01
-1.13918710e+00 1.70083314e-01 4.72233802e-01 -2.52664853e-02
-6.98730350e-01 1.27212977e+00 9.25220624e-02 -8.26781750e-01
2.97396541e-01 5.31062543e-01 -6.62234008e-01 4.67714161e-01
6.48357630e-01 3.96396481e-02 -5.08679986e-01 -1.03519213e+00
-8.37013572e-02 6.27069101e-02 1.42730558e-02 -4.51019824e-01
1.36918890e+00 -1.77197933e-01 -2.43578404e-01 4.96358842e-01
1.19919062e+00 1.91847190e-01 -6.47308230e-01 -8.91407654e-02
-4.10250664e-01 1.61355885e-03 -2.85529017e-01 -8.16522479e-01
-6.04322374e-01 1.02218080e+00 -1.16795674e-01 -3.21215838e-02
8.77427936e-01 1.64045196e-04 1.30715036e+00 6.63369775e-01
-9.92378369e-02 -1.48955870e+00 6.31835401e-01 7.42109239e-01
1.26697004e+00 -1.22629869e+00 -3.33457768e-01 6.89940900e-02
-1.75482595e+00 1.38178158e+00 8.47456872e-01 -5.27609885e-02
2.93617666e-01 2.75081247e-01 1.03203177e+00 -2.58542538e-01
-1.35380590e+00 2.81844616e-01 -3.71545613e-01 6.63614571e-01
1.00596428e+00 1.25796288e-01 -3.26256096e-01 1.36612225e+00
-8.68149161e-01 -4.53294069e-01 8.28646064e-01 5.17548084e-01
-4.61863667e-01 -1.06667340e+00 -7.52761513e-02 -2.61811078e-01
-3.14591557e-01 -8.56040418e-03 -1.13687849e+00 3.85446012e-01
-6.46721542e-01 1.42201483e+00 1.89104732e-02 -5.39944708e-01
4.16263551e-01 3.93753678e-01 1.29876018e-01 -6.58428371e-01
-1.36045289e+00 -5.12287915e-01 7.04105020e-01 -5.07633626e-01
-4.56453741e-01 -3.55304778e-01 -1.60576022e+00 -5.96805215e-01
-1.12712398e-01 6.19669437e-01 4.09806073e-01 8.22684407e-01
1.79132804e-01 1.56824112e-01 1.33943009e+00 -7.58157134e-01
-1.01120436e+00 -1.26593816e+00 -2.51340479e-01 8.25150132e-01
-2.77732350e-02 -2.78675586e-01 -7.35194027e-01 -1.69981673e-01] | [13.142956733703613, 7.6380696296691895] |
a3e054b8-0c82-4da9-be6c-4d61a2649278 | the-effect-of-information-type-on-human | 2302.09069 | null | https://arxiv.org/abs/2302.09069v1 | https://arxiv.org/pdf/2302.09069v1.pdf | The Effect of Information Type on Human Cognitive Augmentation | When performing a task alone, humans achieve a certain level of performance. When humans are assisted by a tool or automation to perform the same task, performance is enhanced (augmented). Recently developed cognitive systems are able to perform cognitive processing at or above the level of a human in some domains. When humans work collaboratively with such cogs in a human/cog ensemble, we expect augmentation of cognitive processing to be evident and measurable. This paper shows the degree of cognitive augmentation depends on the nature of the information the cog contributes to the ensemble. Results of an experiment are reported showing conceptual information is the most effective type of information resulting in increases in cognitive accuracy, cognitive precision, and cognitive power. | ['Samuel McGaha', 'Ron Fulbright'] | 2023-02-15 | null | null | null | null | ['type'] | ['speech'] | [ 2.99143314e-01 4.50432420e-01 5.23112059e-01 -1.56700000e-01
1.02576762e-01 -6.98794365e-01 6.14266634e-01 6.68652594e-01
-5.04178643e-01 5.18597424e-01 2.86910057e-01 -2.53555447e-01
-6.86792731e-01 -8.35531414e-01 -1.18958749e-01 -1.81689054e-01
2.99386263e-01 5.35702050e-01 2.94575065e-01 -4.44245726e-01
6.77283645e-01 3.66930485e-01 -1.95439601e+00 3.77525926e-01
1.11360288e+00 5.32895207e-01 7.39763916e-01 6.85117364e-01
-1.02636695e-01 9.55951393e-01 -9.24346507e-01 2.15149056e-02
1.88399985e-01 -3.26763868e-01 -6.98777914e-01 2.55662560e-01
-1.22409463e-01 1.25506029e-01 3.95306200e-01 9.17634904e-01
4.81473535e-01 2.98732460e-01 4.72673357e-01 -9.78460789e-01
-5.60670614e-01 3.68598789e-01 2.18212679e-01 2.55379558e-01
8.19419682e-01 1.35866031e-01 9.48279127e-02 -5.13323545e-01
3.39964390e-01 1.28793859e+00 5.75287521e-01 -1.52706698e-01
-1.11567020e+00 -4.95607167e-01 -1.84010431e-01 -6.76036850e-02
-1.04436350e+00 -2.02545330e-01 5.27760863e-01 -6.85693145e-01
1.17606235e+00 2.20378935e-02 8.88815284e-01 2.79307187e-01
5.55714011e-01 -2.49430686e-01 1.57092881e+00 -9.06391978e-01
4.53322321e-01 4.55171883e-01 7.30013907e-01 4.45684314e-01
9.40737247e-01 2.73989320e-01 -7.18399465e-01 7.94657022e-02
4.54260200e-01 3.13716568e-02 -1.22108711e-02 2.03250796e-01
-1.13569295e+00 3.39406848e-01 2.04220176e-01 8.72964025e-01
-1.21354270e+00 -2.25961685e-01 2.08455980e-01 4.90120262e-01
1.85336739e-01 1.06819534e+00 -1.07785046e-01 -2.96733230e-01
-5.08531034e-01 4.06722397e-01 7.88934290e-01 8.33343565e-01
5.29456079e-01 -4.01282720e-02 -5.57052195e-01 3.22951406e-01
5.44503033e-02 3.89003754e-01 5.07347226e-01 -1.11503875e+00
-4.66392636e-02 1.41011858e+00 5.08802891e-01 -9.14039195e-01
-7.53551245e-01 -6.94648623e-01 -2.16916159e-01 9.00556862e-01
2.23341122e-01 -3.24895203e-01 -6.14288867e-01 1.37286484e+00
3.96363810e-02 -1.00836527e+00 -1.32338814e-02 6.91525578e-01
9.87783670e-02 1.17183372e-01 6.25086963e-01 -1.40211567e-01
1.80015254e+00 -6.08517230e-01 -1.25189602e+00 -5.27975023e-01
5.61584115e-01 -6.70676768e-01 1.06365144e+00 4.78901446e-01
-1.25843501e+00 -1.12666690e+00 -1.41240847e+00 2.13085786e-02
-8.08572173e-01 -1.76997498e-01 6.27966762e-01 9.62357104e-01
-1.41462314e+00 5.63258886e-01 -2.12468177e-01 -4.51177418e-01
-1.76367685e-02 2.07220137e-01 -7.29811490e-02 -1.48153320e-01
-1.12746382e+00 1.48868442e+00 1.02518523e+00 -3.48963916e-01
-2.22499862e-01 -6.38181388e-01 -2.95868248e-01 3.29981834e-01
3.24596077e-01 -9.39139247e-01 1.01278555e+00 -7.75649488e-01
-8.85163665e-01 6.61026001e-01 4.07020718e-01 -2.83008575e-01
5.55040777e-01 -3.66264701e-01 -6.07799828e-01 2.47673225e-02
3.47889453e-01 5.23479164e-01 2.53396213e-01 -9.83259141e-01
-7.55258858e-01 -6.79434776e-01 4.37383562e-01 3.96310538e-01
-2.83283561e-01 1.94641635e-01 2.14426473e-01 -2.70374745e-01
1.40159905e-01 -9.14717674e-01 1.27715215e-01 -4.52028483e-01
3.20208699e-01 -3.79217297e-01 7.51378179e-01 -8.57187212e-01
1.18029594e+00 -1.92666781e+00 -2.25380674e-01 1.63217559e-01
5.41010261e-01 2.56145626e-01 5.09504676e-01 8.06384087e-01
2.15512991e-01 1.58981442e-01 2.58896917e-01 1.78543761e-01
3.81395608e-01 9.05650407e-02 4.73787189e-01 -1.81784630e-01
-3.05745274e-01 6.41053736e-01 -8.88317227e-01 -3.33261371e-01
-3.06038726e-02 1.84205219e-01 -1.26438126e-01 2.04567507e-01
6.22860380e-02 1.47545695e-01 -2.88122267e-01 1.82791814e-01
3.16020876e-01 -2.73739219e-01 3.99595112e-01 3.04814458e-01
-4.81272548e-01 2.44316757e-01 -8.27253759e-01 1.30367732e+00
-3.23442757e-01 4.09957469e-01 -8.26385319e-02 -4.32249010e-01
1.18072855e+00 5.28767109e-01 -2.47479036e-01 -8.66745532e-01
7.53795132e-02 -1.51137024e-01 6.71152413e-01 -4.09835875e-01
7.28998244e-01 -2.90721562e-02 8.17074627e-02 1.02415645e+00
1.18912859e-02 -2.67452866e-01 5.12034833e-01 1.99554279e-01
1.22975326e+00 -1.16610475e-01 7.39823759e-01 -8.89109850e-01
1.18802629e-01 3.13284516e-01 1.04591101e-01 1.01386309e+00
-3.13308865e-01 -4.68155771e-01 6.12223744e-02 -2.41779715e-01
-1.08822715e+00 -7.37008333e-01 2.91990131e-01 1.59111130e+00
-7.61301965e-02 -3.28957647e-01 -8.13916206e-01 -8.29120427e-02
-3.07808697e-01 1.25840378e+00 -3.78944784e-01 -3.77478033e-01
5.96033828e-03 -3.33108902e-01 2.43647337e-01 7.17616558e-01
1.03813970e+00 -1.20421290e+00 -1.34412920e+00 2.31877357e-01
-9.71107781e-02 -9.37599182e-01 2.79016402e-02 1.21241011e-01
-1.02895641e+00 -7.94755876e-01 -6.27036914e-02 -4.43515420e-01
5.69322884e-01 5.54413378e-01 1.07283890e+00 5.25607765e-01
-1.45892501e-02 5.65216601e-01 -4.99851227e-01 -1.26470292e+00
-4.17546898e-01 -3.50096464e-01 2.98549025e-03 -7.94280112e-01
4.57789570e-01 -4.04331088e-01 -3.63577843e-01 1.44572750e-01
-5.98781586e-01 4.75948125e-01 6.36113346e-01 1.73415273e-01
-6.05492182e-02 6.73916757e-01 7.24285185e-01 -5.62972724e-01
1.57219410e+00 -1.85990632e-01 7.27224350e-03 2.69700170e-01
-1.04262567e+00 -7.09365010e-02 4.58629802e-02 -2.38478884e-01
-1.46860409e+00 -2.58886278e-01 7.18000650e-01 2.80014515e-01
-2.89211571e-01 7.09686220e-01 7.72862583e-02 5.60874410e-04
9.89502490e-01 6.35652244e-02 1.35522798e-01 -1.81682244e-01
-3.83030213e-02 4.50053424e-01 4.03630078e-01 -6.23520970e-01
3.23728830e-01 -1.01659134e-01 2.63357814e-02 -4.74380702e-01
-5.06430924e-01 -9.20376629e-02 -7.96548903e-01 -9.45042610e-01
9.73367095e-01 -7.00363219e-01 -9.56353128e-01 -6.84344247e-02
-8.51152599e-01 -2.69277453e-01 -1.79089636e-01 6.96431637e-01
-5.83479881e-01 -1.22621208e-01 -7.56334290e-02 -1.28972304e+00
-3.06307554e-01 -5.22189140e-01 4.07389551e-01 4.36655492e-01
-7.83692956e-01 -1.05145788e+00 -2.95331746e-01 5.40341675e-01
6.19424343e-01 2.47552872e-01 9.84570742e-01 -7.88390875e-01
-3.40819389e-01 -6.29427671e-01 -8.26004520e-02 8.31394345e-02
7.07375035e-02 -7.31937826e-01 -6.87334895e-01 3.15324287e-03
3.90231729e-01 -9.72863361e-02 9.12631750e-02 7.70149007e-02
4.28950280e-01 4.68494035e-02 -2.79181957e-01 -5.00459850e-01
1.10731900e+00 7.59321511e-01 6.28754973e-01 4.00995255e-01
-1.15208082e-01 1.07734418e+00 8.51595283e-01 1.91983774e-01
2.42635787e-01 5.56517363e-01 -9.38451886e-02 2.94674903e-01
-1.15502536e-01 -5.23361713e-02 7.81117454e-02 4.79480535e-01
-6.88260198e-01 6.29013926e-02 -1.21210790e+00 5.63148372e-02
-1.83676565e+00 -9.02323306e-01 -1.85594857e-01 2.05332136e+00
3.38719398e-01 3.94821107e-01 1.10128425e-01 6.27827406e-01
9.51978981e-01 -6.11668587e-01 -1.69716477e-01 -4.91942585e-01
2.65469611e-01 -1.56790093e-01 2.29924768e-01 2.82563239e-01
-3.26840311e-01 7.05115199e-01 7.43655491e+00 1.68377802e-01
-2.09983051e-01 2.08670408e-01 9.13454518e-02 1.67594533e-02
1.97430506e-01 -2.60072321e-01 -2.89950341e-01 4.20492440e-01
1.19475293e+00 -8.26364815e-01 4.36351240e-01 5.91648400e-01
3.82053763e-01 -6.78709328e-01 -9.18352842e-01 4.72595334e-01
-1.75088063e-01 -8.87993157e-01 -1.37990996e-01 3.63378048e-01
2.87989974e-01 -8.92886221e-01 -2.53702879e-01 5.53751111e-01
6.01864457e-01 -8.15301538e-01 8.51078033e-01 1.14199388e+00
3.19513381e-01 -8.09737086e-01 8.63719463e-01 8.08609962e-01
-7.70536602e-01 -2.05550104e-01 -9.93150026e-02 -1.11991823e+00
2.68546134e-01 4.46373373e-01 -8.62714171e-01 4.63769138e-01
5.43986142e-01 -2.69237727e-01 -8.78204644e-01 5.98694861e-01
-2.75160044e-01 7.78245404e-02 -8.98876786e-02 -2.80171096e-01
-2.62710333e-01 -2.26232670e-02 3.44520181e-01 8.51063788e-01
4.85504270e-01 7.23235130e-01 2.32520655e-01 8.07904243e-01
4.72032428e-01 -1.52678058e-01 -6.62085533e-01 -3.16450506e-01
8.84854496e-01 1.07323980e+00 -1.02055573e+00 -9.42032397e-01
6.54483167e-03 6.59498692e-01 7.84220546e-02 3.56626175e-02
-2.12016925e-01 -3.32987309e-01 1.38012826e-01 2.45831028e-01
-5.30590296e-01 -4.13468719e-01 -8.93448532e-01 5.39676063e-02
-3.32448334e-02 -5.04940271e-01 1.76475555e-01 -1.42758405e+00
-8.34632337e-01 3.68123293e-01 1.35092616e-01 -7.52560079e-01
-3.65502149e-01 -6.38718545e-01 -4.54514593e-01 1.26480210e+00
-7.45503157e-02 -7.56289482e-01 -9.53960955e-01 3.38909119e-01
1.53977856e-01 -1.95996642e-01 9.51231956e-01 -1.52976587e-01
3.37684870e-01 -1.74657971e-01 -7.01219261e-01 -4.76515412e-01
4.53465372e-01 -1.30920303e+00 -1.84214935e-01 6.48164153e-01
-6.72512770e-01 7.61300981e-01 9.39393520e-01 -1.26224792e+00
-1.17286050e+00 -5.94294310e-01 9.03438985e-01 -5.82279384e-01
1.60828024e-01 -4.89683747e-02 -7.63366997e-01 5.18674910e-01
5.40959179e-01 -1.03877509e+00 9.04157043e-01 2.43904442e-01
1.60083145e-01 2.47729823e-01 -1.46414220e+00 6.05193138e-01
1.49123824e+00 -5.77904642e-01 -1.43171871e+00 3.40748250e-01
8.60402763e-01 1.60376996e-01 -7.96753705e-01 4.19946946e-02
6.69461131e-01 -1.20765638e+00 8.85918081e-01 -2.71801263e-01
1.66641295e-01 -2.94226050e-01 -3.72868939e-03 -1.51516569e+00
-8.18631351e-01 -1.76261634e-01 9.13642868e-02 8.53575885e-01
-3.22241895e-02 -9.68987405e-01 -9.42272618e-02 1.06132293e+00
-5.47656238e-01 2.21069068e-01 -1.43237665e-01 -8.89104247e-01
-4.61884201e-01 -3.78811598e-01 7.90047884e-01 8.13576281e-01
8.12260330e-01 4.47313458e-01 3.30787003e-01 -7.81826600e-02
6.31265044e-01 -3.72167647e-01 5.70143223e-01 -2.00610232e+00
1.10948436e-01 -3.64724904e-01 -6.13075197e-01 -2.74999253e-02
-3.33757281e-01 -7.52328694e-01 -2.72218853e-01 -2.00691104e+00
5.50841205e-02 2.44899675e-01 -1.64484024e-01 4.07207280e-01
-2.04909325e-01 -1.90166473e-01 5.91426432e-01 1.84800416e-01
-3.51143807e-01 -3.17624569e-01 1.18889511e+00 4.89155471e-01
-5.58520854e-01 -2.57358700e-01 -1.33503115e+00 8.46219301e-01
1.11460924e+00 8.56792461e-03 -5.21778107e-01 1.65065080e-01
4.06674355e-01 3.70887667e-01 3.27317685e-01 -1.64472306e+00
6.10704362e-01 1.51955366e-01 8.10973585e-01 -4.73773688e-01
3.00872207e-01 -8.68293881e-01 6.54399455e-01 9.13504243e-01
-3.55533302e-01 5.39679706e-01 3.19269985e-01 3.01379472e-01
7.15901107e-02 -2.72233427e-01 5.33517778e-01 -2.03367010e-01
-6.34388864e-01 -7.73878276e-01 -9.86194611e-01 -5.74697554e-01
1.04131603e+00 -4.93756324e-01 -6.33458614e-01 -3.52722794e-01
-1.02391839e+00 -4.46977913e-02 3.55790615e-01 3.31977725e-01
5.73935872e-03 -9.86591458e-01 -4.23079967e-01 -1.09301917e-01
2.86299363e-02 -7.23681748e-01 2.07420707e-01 8.36811602e-01
-3.43744129e-01 7.66839743e-01 -9.96356666e-01 -6.99064359e-02
-1.16373932e+00 6.51710987e-01 2.52492772e-03 -2.18623713e-01
-5.50591171e-01 1.06548175e-01 1.77462488e-01 1.11644529e-01
-2.56091535e-01 1.65510178e-01 -5.04318774e-01 2.70636857e-01
9.06661034e-01 9.91089940e-01 1.75336018e-01 -2.39899769e-01
-1.14786148e-01 -2.64995778e-03 1.84010044e-01 -5.34068048e-01
7.66553521e-01 -2.40923643e-01 -1.03745513e-01 5.30758202e-01
1.26766665e-02 -1.85934544e-01 -8.71097207e-01 3.42003889e-02
2.72178173e-01 -4.51499403e-01 1.07483484e-01 -1.56415009e+00
-1.02026410e-01 5.13963223e-01 7.03105748e-01 6.67851746e-01
1.15146875e+00 -1.88573137e-01 -2.58006245e-01 6.57251835e-01
8.84621978e-01 -1.71403182e+00 1.77550361e-01 3.57411861e-01
1.15843844e+00 -7.46009707e-01 1.18983679e-01 -4.62320268e-01
-7.02974439e-01 9.35458243e-01 6.29023790e-01 -2.92784069e-03
4.77772206e-01 2.89457589e-01 -3.18469286e-01 -5.19365609e-01
-7.78881252e-01 -5.43176711e-01 9.94814485e-02 9.67761457e-01
6.33860886e-01 4.18969303e-01 -1.05850768e+00 8.41992438e-01
-6.62348092e-01 5.43228447e-01 5.50472021e-01 1.31143379e+00
-1.23311353e+00 -4.16818887e-01 -8.90434504e-01 5.07340133e-01
-1.32797971e-01 2.54845977e-01 -7.86884725e-01 7.75589406e-01
2.49101207e-01 1.43683338e+00 2.35514447e-01 -2.10440844e-01
5.55959105e-01 8.88429821e-01 5.32441378e-01 -6.32809937e-01
-9.01685596e-01 -1.38610497e-01 4.39536095e-01 -4.01815146e-01
-3.89707267e-01 -5.37237227e-01 -1.41323125e+00 -4.98812050e-01
1.73976824e-01 2.22716987e-01 7.82934666e-01 9.99360442e-01
4.06978279e-01 1.28006816e+00 -3.50042880e-01 -8.75422597e-01
-1.79949943e-02 -1.46443355e+00 -7.67707646e-01 3.22347015e-01
-6.02835655e-01 -9.56290960e-01 -1.71282470e-01 1.67228132e-01] | [9.028462409973145, 6.351901054382324] |
9220fd04-369f-473e-a7f4-effd53e1015a | low-rank-representation-of-head-impact | 2004.12979 | null | https://arxiv.org/abs/2004.12979v1 | https://arxiv.org/pdf/2004.12979v1.pdf | Low-rank representation of head impact kinematics: A data-driven emulator | Head motion induced by impacts has been deemed as one of the most important measures in brain injury prediction, given that the majority of brain injury metrics use head kinematics as input. Recently, researchers have focused on using fast approaches, such as machine learning, to approximate brain deformation in real-time for early brain injury diagnosis. However, those requires large number of kinematic measurements, and therefore data augmentation is required given the limited on-field measured data available. In this study we present a principal component analysis-based method that emulates an empirical low-rank substitution for head impact kinematics, while requiring low computational cost. In characterizing our existing data set of 537 head impacts, consisting of 6 degrees of freedom measurements, we found that only a few modes, e.g. 15 in the case of angular velocity, is sufficient for accurate reconstruction of the entire data set. Furthermore, these modes are predominantly low frequency since over 70% to 90% of the angular velocity response can be captured by modes that have frequencies under 40Hz. We compared our proposed method against existing impact parametrization methods and showed significantly better performance in injury prediction using a range of kinematic-based metrics -- such as head injury criterion and rotational injury criterion (RIC) -- and brain tissue deformation-metrics -- such as brain angle metric, maximum principal strain (MPS) and axonal fiber strains (FS). In all cases, our approach reproduced injury metrics similar to the ground truth measurements with no significant difference, whereas the existing methods obtained significantly different (p<0.01) values as well as poor injury classification sensitivity and specificity. This emulator will enable us to provide the necessary data augmentation to build a head impact kinematic data set of any size. | ['Kaveh Laksari', 'Hessam Babaee', 'Nima Toosizadeh', 'Patricio Arrue'] | 2020-04-27 | null | null | null | null | ['injury-prediction'] | ['playing-games'] | [-2.58318126e-01 -1.53774306e-01 -1.62455216e-02 -2.65703145e-02
-6.45498633e-01 -2.34286532e-01 1.59842908e-01 5.87454718e-03
-5.45855105e-01 7.14280248e-01 5.42096615e-01 -2.23696470e-01
-4.69730049e-01 -5.30726790e-01 -5.89344800e-01 -8.65776300e-01
-4.27870125e-01 4.72977668e-01 5.08732259e-01 -2.49064758e-01
2.93148637e-01 6.22597694e-01 -1.46271408e+00 -2.47217417e-02
7.18871832e-01 7.18374789e-01 2.34879956e-01 3.07946533e-01
2.81227171e-01 2.05903098e-01 -3.58767390e-01 -5.22578247e-02
9.89087392e-03 -1.56164795e-01 -9.16217268e-01 -4.93654817e-01
5.77674434e-02 -5.10538816e-01 -5.30780435e-01 5.20178318e-01
9.28367138e-01 2.60183394e-01 8.97997618e-01 -1.02601254e+00
-1.88040450e-01 4.95904118e-01 -3.62004042e-01 4.37919497e-01
7.33177602e-01 1.04294263e-01 4.57429647e-01 -8.34834278e-01
6.70705438e-01 7.17849791e-01 8.21639955e-01 7.04735756e-01
-1.12394881e+00 -6.49963975e-01 -4.19205427e-01 8.75771224e-01
-1.20782959e+00 -6.02149703e-02 9.23807144e-01 -9.13596153e-01
1.04256642e+00 4.75514442e-01 9.25061703e-01 1.12490141e+00
6.71374202e-01 2.66426653e-01 1.20202160e+00 -1.72320530e-01
1.34547025e-01 -4.80874777e-01 4.20855045e-01 2.33800262e-01
4.36039388e-01 2.26679280e-01 -4.23423767e-01 -3.05725932e-01
6.11632645e-01 -1.58743754e-01 -8.73556256e-01 1.50623649e-01
-1.11620986e+00 5.89233339e-01 3.50356519e-01 4.28537369e-01
-5.42781234e-01 -1.80321448e-02 4.83854562e-01 -4.76376675e-02
2.84298778e-01 7.12473989e-02 -1.79042459e-01 -3.88825029e-01
-9.23352718e-01 4.19315726e-01 5.50631940e-01 4.23577905e-01
2.75950670e-01 1.71236753e-01 -1.17071774e-02 7.68743277e-01
1.20532066e-01 5.31858742e-01 7.12512612e-01 -7.22431958e-01
3.00845951e-01 3.96824658e-01 -2.56202042e-01 -8.60164464e-01
-1.13125157e+00 -3.68199170e-01 -9.13526773e-01 1.88764736e-01
5.27439415e-01 -2.45970219e-01 -7.97225714e-01 1.67971325e+00
2.77797580e-01 1.58586994e-01 -4.65353638e-01 1.38312268e+00
6.48542345e-01 1.36041611e-01 3.56660485e-02 -4.87273335e-01
1.38672137e+00 -9.86220762e-02 -4.93603259e-01 1.16253756e-01
6.51282609e-01 -8.34794521e-01 1.04400635e+00 2.86053240e-01
-9.89699602e-01 -1.36170581e-01 -8.14133406e-01 4.75410074e-01
1.45378754e-01 -1.96184397e-01 3.67550284e-01 6.13798261e-01
-9.87792909e-01 7.19757318e-01 -1.09513092e+00 -3.63162100e-01
7.09498078e-02 3.06867570e-01 -5.90524375e-01 3.68722603e-02
-1.06666958e+00 1.31667423e+00 -2.10481081e-02 9.83293429e-02
-5.76343894e-01 -1.08909249e+00 -2.48401418e-01 -2.07720891e-01
-2.51719207e-01 -6.97285891e-01 8.72133315e-01 1.26319870e-01
-1.30202484e+00 5.24757981e-01 -1.44084305e-01 -2.31773004e-01
6.12884939e-01 -3.16978544e-01 -4.05734509e-01 4.78471160e-01
-2.15003267e-01 5.26367761e-02 4.27763820e-01 -1.17630005e+00
2.08884589e-02 -5.12817681e-01 -2.06380770e-01 -1.56776771e-01
-3.39311689e-01 3.01732630e-01 8.11220929e-02 -6.92542374e-01
4.99505103e-01 -1.19565475e+00 -5.00960760e-02 -3.75024170e-01
-2.44416475e-01 1.80067182e-01 6.21123075e-01 -1.15207171e+00
1.27695286e+00 -1.69060504e+00 1.35274351e-01 3.00566047e-01
2.34364986e-01 6.58325329e-02 2.65515268e-01 4.70581532e-01
-4.71239120e-01 -9.48498920e-02 -3.82208228e-01 1.16517648e-01
-4.19387728e-01 7.03071952e-02 -1.26587525e-01 9.32865024e-01
-7.58432522e-02 4.89824802e-01 -5.46374619e-01 -3.38000447e-01
1.09409340e-01 5.96806049e-01 -8.32979083e-01 1.43935546e-01
6.06274009e-01 7.00534523e-01 -1.54160514e-01 5.08619905e-01
7.74028242e-01 5.29348016e-01 -4.15474206e-01 -5.82482815e-01
-6.70071170e-02 -3.02329343e-02 -1.12598443e+00 1.18591559e+00
-3.11841190e-01 6.35732234e-01 -3.36370915e-02 -1.13432765e+00
6.29383743e-01 6.95399165e-01 1.15283358e+00 -3.81524235e-01
3.63862604e-01 3.04826319e-01 3.62126797e-01 -9.32456851e-01
-7.19398931e-02 -2.46416897e-01 3.06800276e-01 4.41470295e-01
1.98965780e-02 6.20222986e-02 6.25406802e-02 -2.13036969e-01
1.18299007e+00 -1.15563937e-01 -8.52046162e-02 -4.70722526e-01
4.47118938e-01 1.58399660e-02 4.23502535e-01 1.73460066e-01
-2.14258552e-01 1.00081778e+00 3.13471943e-01 -1.43759355e-01
-9.92725015e-01 -9.61663544e-01 -6.97004676e-01 6.42655909e-01
-1.02823719e-01 -5.60741648e-02 -1.14654732e+00 1.18768208e-01
4.20725867e-02 5.38349688e-01 -2.57779628e-01 -3.62317920e-01
-1.02948356e+00 -1.33824277e+00 6.90887809e-01 7.25819468e-01
-2.27642618e-02 -9.16667938e-01 -8.49494815e-01 1.77287623e-01
-3.70753199e-01 -9.80054021e-01 -2.57365793e-01 5.04423911e-03
-1.03971517e+00 -1.22123516e+00 -8.56167138e-01 -5.58732212e-01
6.57978833e-01 -1.04139773e-02 4.54469800e-01 1.81873560e-01
-5.06714821e-01 2.38714531e-01 -5.34380317e-01 -1.80147648e-01
-2.73851603e-01 -2.27937773e-01 6.45303905e-01 -2.75311649e-01
-2.56811548e-02 -1.03473592e+00 -8.91930997e-01 3.65113825e-01
-6.76622510e-01 -2.45319605e-01 7.06333399e-01 7.03856528e-01
2.50081629e-01 -4.59939361e-01 7.79541373e-01 -3.65455687e-01
7.71704197e-01 -5.34637034e-01 1.23795830e-02 -2.56768256e-01
-6.54349327e-01 -1.25135571e-01 6.96747124e-01 -7.75344849e-01
-7.76527882e-01 -2.60307610e-01 -4.52788949e-01 -4.09463167e-01
-3.56399506e-01 8.38981152e-01 1.16448730e-01 -3.44848663e-01
9.88393188e-01 2.46099114e-01 1.07823655e-01 -6.70383990e-01
-1.87333807e-01 6.56351686e-01 8.05753827e-01 -7.13588357e-01
5.06727934e-01 3.40222180e-01 3.98091853e-01 -9.55848098e-01
1.39375031e-01 -7.24785507e-01 -8.40654314e-01 -6.83124661e-01
7.60889113e-01 -4.40596700e-01 -8.66538584e-01 6.90314710e-01
-1.10193598e+00 -5.32295294e-02 1.68042406e-01 1.30675268e+00
-5.84566534e-01 5.77115715e-01 -7.98942089e-01 -7.33915448e-01
-6.31345868e-01 -1.35624087e+00 6.36336684e-01 -1.39639854e-01
-2.98362136e-01 -6.49905443e-01 1.49911270e-01 4.37357068e-01
5.79445601e-01 7.69227862e-01 1.15643001e+00 -3.78990382e-01
-3.21226418e-02 -5.70358276e-01 5.77206537e-02 4.37665246e-02
-2.26015851e-01 -5.91635294e-02 -8.07789087e-01 -5.23934424e-01
3.45848173e-01 1.57223254e-01 4.75675464e-01 5.04570782e-01
9.17973220e-01 -1.20182544e-01 -2.24341989e-01 5.89691877e-01
1.33914435e+00 1.34400800e-01 5.45788705e-01 3.31039846e-01
6.86546922e-01 6.49039149e-01 1.78339109e-01 4.50843692e-01
1.43081352e-01 7.88907290e-01 4.14094687e-01 2.27973148e-01
-3.96408170e-01 3.12468380e-01 3.69551539e-01 1.36030376e+00
-1.01080132e+00 2.55912513e-01 -1.32417393e+00 5.79638183e-01
-1.51776814e+00 -1.03157902e+00 -6.70952439e-01 2.47082758e+00
6.89175189e-01 8.44466835e-02 4.42074418e-01 6.39596760e-01
6.25829816e-01 -4.32281047e-01 -3.49412143e-01 -1.05483249e-01
1.03059381e-01 2.86753774e-01 5.98279357e-01 3.32551271e-01
-5.24769902e-01 1.02222234e-01 6.48855066e+00 3.02425683e-01
-1.44965887e+00 3.45696956e-01 -1.75605580e-01 -2.07903862e-01
-6.73561618e-02 -9.34864581e-02 -3.77174497e-01 6.97797835e-01
1.23871911e+00 -3.33528906e-01 3.35491419e-01 5.95508099e-01
5.20779431e-01 -1.81149274e-01 -7.19310701e-01 8.84046793e-01
9.64455865e-03 -9.24077213e-01 -1.62495047e-01 1.25612959e-01
2.67842054e-01 1.98874831e-01 -4.96545970e-01 -1.37937278e-01
-7.13139474e-01 -8.42171729e-01 9.15446579e-01 8.10821354e-01
8.14278841e-01 -6.78463161e-01 9.42490041e-01 5.54755330e-01
-1.17349064e+00 -9.04423594e-02 -2.54169762e-01 -6.16932698e-02
6.35509074e-01 6.38880491e-01 -7.34540880e-01 7.01962769e-01
7.88724065e-01 9.41214990e-03 -1.97392046e-01 1.57937884e+00
5.63376360e-02 1.00283766e+00 -5.89921832e-01 1.77266508e-01
-3.56474906e-01 -1.48585722e-01 9.64886427e-01 1.28195512e+00
5.65526783e-01 3.28033775e-01 -4.01865691e-02 4.85108614e-01
6.28601849e-01 2.02151790e-01 -2.59949595e-01 6.12055719e-01
3.53668511e-01 9.63819802e-01 -6.25067413e-01 2.11066186e-01
-5.24721265e-01 3.31020772e-01 -1.51893916e-02 3.70703548e-01
-9.25898731e-01 -2.95888543e-01 6.79916203e-01 7.62658000e-01
-4.87736285e-01 -5.12211442e-01 -5.59065819e-01 -9.90045249e-01
1.71177402e-01 -3.87774706e-01 2.31237903e-01 -5.66518903e-01
-1.03862417e+00 5.35137773e-01 6.78175449e-01 -1.30403495e+00
-5.40127993e-01 -7.26280808e-01 -8.19895148e-01 1.07916915e+00
-8.92098904e-01 -9.85026658e-01 -5.93877792e-01 6.09162629e-01
8.46858397e-02 1.31886393e-01 6.89595640e-01 6.23825848e-01
-6.27818525e-01 5.21167159e-01 -2.31478915e-01 1.52737470e-02
6.40394866e-01 -9.57623005e-01 -9.23053548e-02 8.86043191e-01
-3.99531603e-01 6.22111261e-01 1.10477114e+00 -6.36845469e-01
-1.54562593e+00 -6.58295631e-01 5.29572725e-01 -1.82417497e-01
6.65147185e-01 8.11338574e-02 -1.29260445e+00 2.64977664e-01
-2.48484045e-01 -2.25661993e-02 6.15076840e-01 -4.22845662e-01
2.50313044e-01 2.16112748e-01 -1.06724501e+00 4.35713410e-01
1.25150824e+00 -4.51606125e-01 -8.73713136e-01 2.65334219e-01
1.70391858e-01 -5.01500666e-01 -1.23073268e+00 8.59891236e-01
7.78855979e-01 -7.12430418e-01 1.09710717e+00 -5.43180704e-01
1.50146767e-01 -2.21058711e-01 5.28380647e-02 -1.41370332e+00
-5.70372760e-01 -2.67027706e-01 -2.54912943e-01 9.50857639e-01
1.02277912e-01 -7.92514682e-01 7.70198405e-01 6.95412278e-01
-5.65883040e-01 -1.13834131e+00 -1.34981096e+00 -8.51180613e-01
2.93739706e-01 -7.72565186e-01 3.02951604e-01 7.13765204e-01
4.14532036e-01 -2.75665075e-01 3.37029458e-04 2.33226001e-01
6.61185920e-01 -2.22956866e-01 2.63042986e-01 -1.21983016e+00
1.05665237e-01 -5.78872263e-01 -9.07691002e-01 -2.34483138e-01
1.15123242e-01 -1.00302744e+00 -1.15447931e-01 -1.51076567e+00
1.97378740e-01 -2.95544297e-01 -2.02014104e-01 4.11809236e-01
-5.10909744e-02 1.97702140e-01 -5.29715419e-02 5.54172158e-01
7.97955573e-01 2.21965104e-01 1.19490039e+00 2.88200378e-01
-2.08649695e-01 -1.67787839e-02 -1.29938319e-01 7.94043899e-01
6.69352114e-01 -4.29037452e-01 -2.20387474e-01 -2.68070519e-01
-2.43932098e-01 2.96753615e-01 3.83099765e-01 -1.41754591e+00
4.04525846e-01 -2.64323950e-01 3.80423874e-01 -4.12878007e-01
1.52762458e-01 -5.64226508e-01 6.10514939e-01 6.53432429e-01
2.76162177e-01 1.80632666e-01 1.12674229e-01 8.15719068e-02
-1.06205888e-01 -3.39222163e-01 6.86469257e-01 4.18215007e-01
-4.53241348e-01 1.98199674e-01 -5.82418025e-01 -3.99426706e-02
9.67236102e-01 -3.72364342e-01 -3.92618209e-01 -1.88539490e-01
-7.61353076e-01 -4.00586516e-01 3.13980401e-01 2.36527592e-01
6.91746712e-01 -1.20917118e+00 -8.67079258e-01 1.61225364e-01
-7.55004361e-02 -1.51135638e-01 4.80384290e-01 1.66202176e+00
-8.41230452e-01 5.53871691e-01 -5.67897618e-01 -6.40236676e-01
-1.12531209e+00 2.70403206e-01 1.13872349e-01 1.77870482e-01
-8.69524837e-01 4.69722152e-01 -1.96309596e-01 9.99701470e-02
-1.10591583e-01 -5.40323794e-01 -4.28009838e-01 2.16731578e-01
2.81605721e-01 8.34755421e-01 6.47043943e-01 -1.18467748e+00
-7.04144061e-01 9.39975977e-01 4.93086755e-01 -5.88636138e-02
1.46302009e+00 1.32346734e-01 -4.99110632e-02 1.21988915e-01
1.25623798e+00 -5.50295785e-02 -9.13879991e-01 4.27246630e-01
-1.22424416e-01 -2.35831678e-01 2.62642562e-01 -5.12365341e-01
-1.05216289e+00 7.51995087e-01 8.29618156e-01 -4.63195108e-02
1.19650769e+00 7.91417137e-02 1.01573789e+00 1.00258291e-01
7.83495247e-01 -8.10907066e-01 -3.21233094e-01 2.66065836e-01
1.22439694e+00 -6.02793574e-01 1.48109151e-02 -4.90508944e-01
-3.16580713e-01 1.24783683e+00 5.91581821e-01 -3.75616401e-01
8.24211359e-01 4.71427441e-01 -1.06110521e-01 -7.06473142e-02
-3.46953303e-01 7.36083686e-02 5.87624729e-01 6.04268670e-01
5.67131162e-01 2.60213435e-01 -9.54091430e-01 9.32045817e-01
-7.59142399e-01 7.70563353e-03 5.68033993e-01 1.06247878e+00
-3.97135794e-01 -8.61764550e-01 -7.58711874e-01 7.81218946e-01
-5.98263621e-01 3.33694965e-01 1.37079820e-01 6.90374136e-01
6.94648502e-03 6.24367237e-01 -1.98596224e-01 -7.72983730e-01
7.02062070e-01 2.17563778e-01 6.13242686e-01 -2.68035680e-01
-4.05923486e-01 -3.44814174e-02 -8.31746385e-02 -6.08130932e-01
-3.08031857e-01 -8.14210415e-01 -1.56179070e+00 -4.98925328e-01
-5.34706235e-01 -1.27014294e-01 7.94700325e-01 1.13713610e+00
8.82030204e-02 4.38550442e-01 3.60271931e-01 -1.27483702e+00
-4.96005237e-01 -1.30110526e+00 -5.31941533e-01 6.98225975e-01
1.72425941e-01 -1.17416775e+00 -5.47392368e-01 1.50189742e-01] | [14.068229675292969, -1.885719895362854] |
dc37164c-314d-41a4-bfb9-d0ee9a4818b3 | adaptive-template-enhancement-for-improved | 2201.01218 | null | https://arxiv.org/abs/2201.01218v1 | https://arxiv.org/pdf/2201.01218v1.pdf | Adaptive Template Enhancement for Improved Person Recognition using Small Datasets | A novel instance-based method for the classification of electroencephalography (EEG) signals is presented and evaluated in this paper. The non-stationary nature of the EEG signals, coupled with the demanding task of pattern recognition with limited training data as well as the potentially noisy signal acquisition conditions, have motivated the work reported in this study. The proposed adaptive template enhancement mechanism transforms the feature-level instances by treating each feature dimension separately, hence resulting in improved class separation and better query-class matching. The proposed new instance-based learning algorithm is compared with a few related algorithms in a number of scenarios. A clinical grade 64-electrode EEG database, as well as a low-quality (high-noise level) EEG database obtained with a low-cost system using a single dry sensor have been used for evaluations in biometric person recognition. The proposed approach demonstrates significantly improved classification accuracy in both identification and verification scenarios. In particular, this new method is seen to provide a good classification performance for noisy EEG data, indicating its potential suitability for a wide range of applications. | ['Farzin Deravi', 'Sanaul Hoque', 'Su Yang'] | 2022-01-03 | null | null | null | null | ['person-recognition'] | ['computer-vision'] | [ 7.64083624e-01 -4.25458670e-01 4.41688299e-01 -4.54689741e-01
-8.49655628e-01 -3.34835768e-01 3.49543184e-01 3.64323944e-01
-6.90299094e-01 1.09654260e+00 -3.38460058e-01 3.06255311e-01
-8.15219879e-01 -3.71032417e-01 -2.71935552e-01 -1.11532581e+00
-2.73710102e-01 1.68591157e-01 -3.39980394e-01 1.09069027e-01
4.34411734e-01 8.37772727e-01 -1.96665418e+00 1.91716775e-01
8.28689158e-01 1.39561224e+00 2.46919677e-01 2.36897692e-01
3.55452269e-01 -9.86897796e-02 -1.10447919e+00 -1.49104208e-01
1.51544571e-01 -3.21387678e-01 -3.22833121e-01 1.64935619e-01
2.38351360e-01 2.12411344e-01 9.55168679e-02 8.33940268e-01
1.06376016e+00 1.01539142e-01 7.32689202e-01 -1.13951576e+00
1.93019703e-01 1.77324012e-01 -2.88136750e-01 5.68183005e-01
5.95830739e-01 -4.37455662e-02 4.57392335e-01 -8.24960589e-01
1.10951066e-01 2.79928595e-01 4.96985197e-01 3.32392395e-01
-1.55266905e+00 -1.05224061e+00 -2.96153575e-01 4.73984480e-01
-1.75171936e+00 -4.28918839e-01 1.05486178e+00 -4.68305349e-01
1.10923564e+00 3.70992720e-01 9.64748561e-01 1.12169755e+00
4.10195887e-01 8.25318918e-02 1.78683293e+00 -3.96635979e-01
4.66695189e-01 5.04517257e-01 3.61078620e-01 1.62066706e-02
4.34896857e-01 9.41093788e-02 -8.01900148e-01 -4.01411444e-01
3.87798965e-01 -5.30640297e-02 -6.02112830e-01 -3.50290388e-01
-1.06424630e+00 1.17985442e-01 8.21790025e-02 8.68896127e-01
-8.74362946e-01 -3.12452048e-01 3.93484384e-01 3.93086404e-01
3.95092130e-01 6.70732379e-01 -4.20211107e-01 -4.73767906e-01
-1.20248091e+00 7.07493201e-02 7.52213299e-01 6.43224776e-01
2.47236058e-01 2.06780463e-01 -1.02042928e-01 9.39921200e-01
-1.13975108e-01 4.86322552e-01 5.02939045e-01 -8.55391771e-02
4.71305609e-01 4.44967479e-01 9.19786394e-02 -1.01928961e+00
-6.86206520e-01 -7.36651361e-01 -8.94823968e-01 2.69999504e-01
2.32030213e-01 -1.04882821e-01 -5.88616967e-01 1.54671764e+00
-1.35165215e-01 3.20533901e-01 9.40273181e-02 7.57646084e-01
5.74567735e-01 4.44933295e-01 -2.25131260e-03 -5.69705129e-01
1.31136549e+00 1.20854832e-01 -7.66790450e-01 8.90967175e-02
-7.81587064e-02 -6.08904600e-01 6.30133331e-01 1.05395973e+00
-9.24313724e-01 -5.52468300e-01 -1.22893333e+00 8.36913168e-01
-4.56497133e-01 2.43521899e-01 1.64081290e-01 9.52757537e-01
-8.27714264e-01 4.82122689e-01 -7.99029887e-01 -4.79024738e-01
3.34494919e-01 1.01016188e+00 -6.46294355e-01 2.59951383e-01
-9.66420233e-01 8.86012077e-01 3.48851413e-01 2.63833374e-01
-1.46442279e-01 -4.46870714e-01 -5.23177207e-01 1.49868473e-01
-1.64523244e-01 -9.15419161e-02 5.60561359e-01 -6.93878889e-01
-1.30210555e+00 5.32631338e-01 -1.65119052e-01 -3.42179567e-01
3.51097286e-01 1.69655383e-01 -7.39438236e-01 2.23565742e-01
-4.15960521e-01 -1.38435289e-02 1.01460660e+00 -9.27299678e-01
-2.63820946e-01 -8.35260987e-01 -5.23930788e-01 2.44051680e-01
-5.39112270e-01 -3.46113890e-02 4.73469585e-01 -5.49550474e-01
1.92949310e-01 -3.63007337e-01 3.81971598e-01 -7.29619622e-01
1.67704925e-01 9.41404179e-02 5.09145677e-01 -6.90978706e-01
1.10741806e+00 -2.11382818e+00 8.90399516e-02 8.39477479e-01
-2.31996089e-01 1.34950608e-01 8.79085138e-02 4.91953462e-01
-4.71306890e-01 -4.62983042e-01 -4.67656225e-01 -1.45921037e-01
-2.75156528e-01 -8.56057927e-02 4.89634573e-02 7.38068998e-01
2.91810542e-01 3.43859047e-01 -5.05479097e-01 -5.83233871e-02
4.07038182e-01 8.01253140e-01 1.08449213e-01 2.86959201e-01
7.18546808e-01 6.19949281e-01 -3.50422673e-02 6.20184660e-01
5.82978666e-01 3.62810552e-01 -7.81400576e-02 -4.01711851e-01
2.51486171e-02 -5.56437951e-03 -1.53174484e+00 1.47200930e+00
-3.88212979e-01 6.56261146e-01 -1.24973236e-02 -1.21270823e+00
1.34299123e+00 8.31305206e-01 7.34629869e-01 -9.37319994e-01
2.75435865e-01 4.54295069e-01 3.73154551e-01 -6.06753945e-01
2.33004779e-01 -1.96936488e-01 2.65308678e-01 2.82991767e-01
2.62434691e-01 -5.84283657e-02 -4.19576690e-02 -5.11470616e-01
8.31324279e-01 -5.12567237e-02 5.72256625e-01 -6.66205823e-01
8.12187076e-01 -4.52177137e-01 2.12096810e-01 6.06197238e-01
-4.62322719e-02 3.45723450e-01 3.18198577e-02 -1.20824896e-01
-6.01597667e-01 -6.65172935e-01 -8.02040279e-01 1.82869315e-01
1.29513696e-01 -5.55222780e-02 -8.35878611e-01 4.40185107e-02
-1.10062607e-01 5.17199397e-01 -3.73604119e-01 -1.73903584e-01
-3.30453157e-01 -1.12743986e+00 4.45805341e-01 3.25642765e-01
2.51217425e-01 -1.12815905e+00 -1.06620729e+00 5.14815986e-01
1.34015456e-01 -8.02887857e-01 2.16348186e-01 6.11750901e-01
-1.07592964e+00 -1.08002579e+00 -6.65294766e-01 -6.42169893e-01
6.12007678e-01 -2.56254733e-01 7.07560480e-01 -2.25580841e-01
-5.78293264e-01 7.05157459e-01 -2.83928990e-01 -6.37574732e-01
5.10772057e-02 -1.07232384e-01 5.58144212e-01 4.80610788e-01
6.69138968e-01 -8.59196424e-01 -5.37784934e-01 3.44040632e-01
-7.34098196e-01 -6.52013540e-01 5.65274179e-01 1.24980962e+00
5.04750192e-01 4.23048556e-01 1.09381747e+00 -2.38863513e-01
1.19488800e+00 -1.32805824e-01 -5.88787913e-01 1.76702306e-01
-7.93617308e-01 -3.55317742e-01 5.42787075e-01 -4.65445340e-01
-9.42111194e-01 -3.68757285e-02 -1.80210710e-01 1.49391159e-01
-5.79402268e-01 2.98969805e-01 -3.42242271e-01 -4.80583072e-01
7.64043987e-01 4.59674567e-01 8.49306863e-03 -3.97022367e-01
-6.43468082e-01 1.19237280e+00 5.87889731e-01 -4.92357075e-01
4.01576459e-01 -7.16940835e-02 1.52908072e-01 -1.17111051e+00
3.32607239e-01 -6.22254729e-01 -7.31928647e-01 -2.60771871e-01
4.52424824e-01 -8.46294403e-01 -8.60594273e-01 8.05073500e-01
-9.28744495e-01 1.28600493e-01 -8.00862163e-02 9.53546584e-01
-5.03097653e-01 5.76524027e-02 -1.43315405e-01 -1.37090266e+00
-7.36298919e-01 -1.22594523e+00 7.76831031e-01 1.05437867e-01
-3.63352269e-01 -6.67071998e-01 -2.89084256e-01 5.96054755e-02
3.68499547e-01 2.55683213e-01 6.81938648e-01 -1.02650142e+00
-1.11059368e-01 -7.40378082e-01 1.65816814e-01 4.51613069e-01
4.10687476e-01 -5.48644125e-01 -1.40971541e+00 -6.42858624e-01
5.46715498e-01 -8.94383416e-02 2.79473037e-01 1.86014250e-01
9.50663209e-01 3.33714068e-01 -1.35622516e-01 3.11807990e-01
1.51799226e+00 8.76695812e-01 5.78185320e-01 2.76716143e-01
8.12063962e-02 4.00956750e-01 5.16640127e-01 6.39370501e-01
-4.34683204e-01 7.53714681e-01 5.79986647e-02 4.09763120e-02
3.75283331e-01 5.34761012e-01 -6.04320131e-02 6.38934553e-01
-2.97353506e-01 -2.40186099e-02 -8.01231623e-01 3.53670448e-01
-1.35191154e+00 -8.98290575e-01 1.57502323e-01 2.40953875e+00
2.84583896e-01 1.53146312e-02 1.69046417e-01 1.12727976e+00
6.36687934e-01 -3.97904962e-01 -4.42559838e-01 -3.21682781e-01
-2.15897352e-01 8.85912120e-01 2.42059246e-01 4.54226956e-02
-7.14203656e-01 -1.43099368e-01 5.80344009e+00 5.41321397e-01
-1.33498335e+00 -9.64709744e-02 2.72065789e-01 -2.21939087e-01
4.58559722e-01 -6.79788291e-01 -3.58836234e-01 7.32645273e-01
1.03572106e+00 -2.73079395e-01 4.07160729e-01 3.29144686e-01
2.91830719e-01 -3.53922218e-01 -1.27485693e+00 1.81905782e+00
3.90966833e-01 -8.46343398e-01 -2.33249068e-01 1.98241249e-01
2.42684796e-01 -3.04977626e-01 -5.76525629e-02 -2.05668375e-01
-1.29697955e+00 -1.06540668e+00 5.94540775e-01 8.56961966e-01
7.89156437e-01 -1.05735767e+00 1.16758502e+00 2.75892347e-01
-9.62099135e-01 -2.95906276e-01 -9.60157532e-03 -1.39392376e-01
-7.62686133e-02 4.89000142e-01 -6.15918219e-01 7.79940546e-01
6.76289141e-01 5.30823529e-01 -4.65192556e-01 1.53619790e+00
2.58931547e-01 6.12892926e-01 -4.26767915e-01 -1.80061668e-01
-3.11388850e-01 -3.11442554e-01 6.72173262e-01 1.24399006e+00
6.56675816e-01 3.98514599e-01 -4.82253700e-01 7.26615191e-01
2.48214483e-01 2.28719488e-01 -6.05679452e-01 1.09515026e-01
4.04085904e-01 1.25698376e+00 -7.78684378e-01 -2.96275038e-02
-2.37772286e-01 7.07409918e-01 -2.33307630e-01 2.73460269e-01
-3.82127315e-01 -8.26321125e-01 3.68882895e-01 1.44200847e-02
-8.42953175e-02 -2.12646071e-02 -5.35205185e-01 -1.00142694e+00
3.56898516e-01 -9.06279147e-01 1.87585294e-01 -5.83783090e-01
-1.19410527e+00 1.06262791e+00 2.62106329e-01 -1.46208358e+00
-3.57643336e-01 -6.34832323e-01 -4.37727243e-01 1.28806484e+00
-1.08642220e+00 -5.43686986e-01 -4.01419014e-01 7.68196166e-01
2.24734321e-01 -6.00093722e-01 1.14399159e+00 5.61578155e-01
-3.00757676e-01 7.22455204e-01 1.75035596e-01 -3.09514701e-01
5.01129925e-01 -1.01117289e+00 -1.86407521e-01 6.43341064e-01
5.25379181e-02 7.13673294e-01 7.14908421e-01 -2.91826785e-01
-1.32489705e+00 -4.35019642e-01 5.78596115e-01 -7.61613175e-02
3.03905040e-01 -5.20385027e-01 -1.07244253e+00 -1.63517162e-01
2.24291474e-01 -3.09719712e-01 8.37249458e-01 -1.91383719e-01
3.53510708e-01 -7.17264414e-01 -1.57558870e+00 1.22988522e-02
6.97502375e-01 -6.20122373e-01 -9.72384930e-01 -4.84346785e-02
-5.05510509e-01 -2.40546316e-01 -1.17939818e+00 4.23779309e-01
7.68221736e-01 -8.65441501e-01 5.47553658e-01 -1.10814549e-01
-3.40367466e-01 -1.07702084e-01 -5.50822504e-02 -1.51799667e+00
-2.23631889e-01 -5.46777427e-01 1.79111451e-01 1.22541249e+00
3.14257294e-01 -1.03038859e+00 5.68434775e-01 6.89714670e-01
2.54851162e-01 -8.83605957e-01 -1.22374415e+00 -8.91512454e-01
-4.30605114e-01 -4.24492925e-01 7.00441599e-01 3.39136571e-01
5.03408551e-01 7.55191669e-02 -2.11220801e-01 1.99198928e-02
7.24245131e-01 -8.56025219e-02 2.39788830e-01 -1.44404113e+00
-1.26982346e-01 -2.03987092e-01 -1.37022364e+00 -8.70534182e-02
5.95765002e-02 -8.31684053e-01 -1.12123057e-01 -1.03723991e+00
8.25038850e-02 -2.99057990e-01 -8.59299302e-01 2.56632287e-02
1.23715043e-01 5.22360742e-01 -1.33300304e-01 -3.32757682e-02
2.90802628e-01 2.24422395e-01 3.80738467e-01 -1.45661771e-01
-3.63416344e-01 3.30410779e-01 -3.27812076e-01 2.08482474e-01
8.54481459e-01 -3.27423424e-01 -5.79511940e-01 6.70861006e-02
-2.84227163e-01 1.01502061e-01 2.20811784e-01 -1.56576586e+00
2.74077475e-01 5.36522746e-01 8.53164911e-01 -3.31396282e-01
7.41891026e-01 -1.39967000e+00 5.69315970e-01 4.27229106e-01
-2.71312088e-01 2.95549631e-01 4.05924112e-01 4.29815292e-01
-4.35812265e-01 -2.36031741e-01 8.23208034e-01 2.45629683e-01
-4.69603330e-01 -3.68982814e-02 -3.52259010e-01 -5.42365730e-01
1.21673286e+00 -9.14322495e-01 1.49713993e-01 -1.71864942e-01
-1.01133132e+00 -3.29218715e-01 1.08806737e-01 1.77761391e-01
7.75825560e-01 -1.09394681e+00 -4.56768185e-01 8.27979803e-01
4.88284498e-01 -6.18378937e-01 2.43097380e-01 1.20333087e+00
8.37395266e-02 4.10020083e-01 -7.85920322e-01 -7.94582129e-01
-1.64004970e+00 1.44872189e-01 4.24416125e-01 3.61987442e-01
-6.66729093e-01 5.48067868e-01 -4.67733949e-01 2.06075817e-01
4.83856261e-01 -2.78919429e-01 -5.00195026e-01 3.08297753e-01
7.48878658e-01 4.25468951e-01 9.14593458e-01 -5.94153762e-01
-5.95002413e-01 5.51134884e-01 2.21111998e-01 -1.78852573e-01
1.62359822e+00 1.07044898e-01 -1.66527405e-01 6.88952863e-01
9.86322939e-01 -2.86758304e-01 -7.03150213e-01 -7.36005083e-02
2.31892079e-01 -4.92929310e-01 -1.08952880e-01 -1.13192809e+00
-8.06255341e-01 9.12195206e-01 1.27072799e+00 2.56332278e-01
1.63185942e+00 -4.60638195e-01 1.70358956e-01 4.69643980e-01
9.54681277e-01 -1.10197866e+00 -3.91023755e-01 -1.41498238e-01
9.40405250e-01 -8.76037836e-01 7.92170465e-02 -2.94220690e-02
-3.48211437e-01 1.21678245e+00 2.60290146e-01 -3.99456657e-02
5.67279160e-01 4.83029485e-01 -1.59274608e-01 -4.00833078e-02
-4.10152286e-01 1.49381056e-01 4.39748824e-01 9.74769533e-01
4.18842047e-01 2.23017689e-02 -6.93543613e-01 8.06115389e-01
-7.58406296e-02 2.62203276e-01 4.62691873e-01 9.66073573e-01
-1.21999651e-01 -1.02778971e+00 -6.99189186e-01 9.86515045e-01
-5.38950026e-01 4.15788591e-02 -3.11594516e-01 8.56499910e-01
1.66615322e-01 1.19221926e+00 -9.13651939e-03 -4.63185847e-01
7.37616897e-01 3.86539042e-01 8.77875328e-01 -1.86957210e-01
-9.52604175e-01 1.20702319e-01 -1.49164662e-01 -4.31271464e-01
-5.95245957e-01 -8.39553475e-01 -8.79577219e-01 3.46522033e-01
-4.12413359e-01 4.28510159e-01 1.07332087e+00 9.58335817e-01
3.14145029e-01 5.13302684e-01 5.66707492e-01 -8.98351133e-01
-5.60316563e-01 -1.09887373e+00 -1.16467297e+00 4.11496282e-01
3.51698011e-01 -9.30752695e-01 -3.70786190e-01 -5.56930788e-02] | [13.250313758850098, 3.3363819122314453] |
e010470c-98fa-4d9c-9efe-f9137d94983a | monocular-bev-perception-of-road-scenes-via | 2211.08144 | null | https://arxiv.org/abs/2211.08144v1 | https://arxiv.org/pdf/2211.08144v1.pdf | Monocular BEV Perception of Road Scenes via Front-to-Top View Projection | HD map reconstruction is crucial for autonomous driving. LiDAR-based methods are limited due to expensive sensors and time-consuming computation. Camera-based methods usually need to perform road segmentation and view transformation separately, which often causes distortion and missing content. To push the limits of the technology, we present a novel framework that reconstructs a local map formed by road layout and vehicle occupancy in the bird's-eye view given a front-view monocular image only. We propose a front-to-top view projection (FTVP) module, which takes the constraint of cycle consistency between views into account and makes full use of their correlation to strengthen the view transformation and scene understanding. In addition, we also apply multi-scale FTVP modules to propagate the rich spatial information of low-level features to mitigate spatial deviation of the predicted object location. Experiments on public benchmarks show that our method achieves the state-of-the-art performance in the tasks of road layout estimation, vehicle occupancy estimation, and multi-class semantic estimation. For multi-class semantic estimation, in particular, our model outperforms all competitors by a large margin. Furthermore, our model runs at 25 FPS on a single GPU, which is efficient and applicable for real-time panorama HD map reconstruction. | ['Jia Pan', 'Shengfeng He', 'Yuexin Ma', 'Yuanlong Yu', 'Jiaxin Cai', 'Weixiang Yang', 'Qi Li', 'Wenxi Liu'] | 2022-11-15 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 1.07082218e-01 -1.27810091e-01 -1.57775268e-01 -7.78881013e-01
-5.84416330e-01 -5.00627041e-01 4.93873417e-01 -3.73396456e-01
-2.60172188e-01 3.60519469e-01 -6.60730526e-02 -3.57228398e-01
2.90923446e-01 -1.37724125e+00 -1.17912030e+00 -4.72321332e-01
6.28511667e-01 4.76185590e-01 9.51882541e-01 -1.89395934e-01
2.31479257e-01 5.08862555e-01 -2.12417626e+00 1.01306334e-01
9.85146523e-01 1.07152486e+00 7.08718657e-01 5.50393045e-01
-1.45055288e-02 7.01596439e-01 1.10740885e-01 -3.75970721e-01
3.50450158e-01 1.29682049e-01 -3.03228617e-01 2.26683900e-01
8.63566279e-01 -8.10914516e-01 -6.34116411e-01 9.49423790e-01
3.60178024e-01 -2.89518628e-02 2.53720492e-01 -1.48851359e+00
9.49104875e-02 -8.22115242e-02 -7.43325293e-01 2.15574298e-02
2.04665810e-01 2.69250214e-01 6.02956593e-01 -1.11353207e+00
6.88502431e-01 1.19861197e+00 4.82223988e-01 5.40638305e-02
-8.06192636e-01 -8.68936419e-01 4.67183530e-01 6.19991541e-01
-1.45444632e+00 -4.56782609e-01 6.29310846e-01 -2.71629155e-01
8.44751537e-01 2.78402269e-01 7.31194615e-01 6.09291553e-01
3.18896621e-01 7.98003554e-01 1.26348901e+00 1.92851335e-01
6.70584757e-03 1.61845088e-01 -4.06531617e-02 7.50649929e-01
2.31099635e-01 3.30995679e-01 -6.73553228e-01 3.54468405e-01
4.60144162e-01 2.57557034e-01 -1.50320008e-01 -6.54517651e-01
-1.12309468e+00 6.99039638e-01 4.35951084e-01 -4.44061607e-01
-1.42353415e-01 6.89892918e-02 1.41644940e-01 -1.50056019e-01
5.11673689e-01 -3.23344767e-01 -3.81755829e-01 -3.04169338e-02
-7.73134351e-01 2.40986198e-01 4.04162228e-01 1.19950819e+00
1.22402096e+00 -2.79431909e-01 2.00750977e-01 6.23747826e-01
2.56144494e-01 1.10681367e+00 -3.07338178e-01 -1.26334405e+00
8.36339474e-01 5.78872085e-01 3.51038128e-02 -9.28716123e-01
-4.42569166e-01 -2.18413860e-01 -6.37469947e-01 1.87605396e-01
8.24692920e-02 2.06097335e-01 -1.02684331e+00 1.27332354e+00
6.82459176e-01 4.28947806e-01 -1.28192276e-01 9.63876247e-01
8.75213385e-01 8.11987221e-01 -3.67901951e-01 -1.59546621e-02
1.26612139e+00 -1.21484673e+00 -5.61246991e-01 -7.23609507e-01
4.34116095e-01 -5.81437409e-01 8.93560946e-01 2.35845640e-01
-7.99741447e-01 -6.36844873e-01 -1.17048240e+00 -4.83931929e-01
-4.92560178e-01 1.15205068e-02 5.55380344e-01 5.93110025e-01
-7.90132165e-01 5.33811897e-02 -7.31304705e-01 -1.46188393e-01
3.25677693e-01 6.24578632e-02 -3.46741498e-01 -7.07736969e-01
-1.10551643e+00 1.05001044e+00 3.20485413e-01 1.86779588e-01
-7.26530194e-01 -9.89582598e-01 -1.09005690e+00 -9.17044133e-02
5.69463372e-01 -7.25410581e-01 8.65211010e-01 -2.00373650e-01
-1.41002369e+00 7.52606273e-01 -5.59609294e-01 -3.76848876e-01
7.67027497e-01 -2.45820135e-01 -4.06666785e-01 3.55449482e-03
3.15322727e-01 9.93101716e-01 5.85424840e-01 -1.38399768e+00
-1.32379699e+00 -6.08820915e-01 3.00016627e-02 5.18281341e-01
9.15385559e-02 -5.55591702e-01 -1.01543009e+00 2.53372967e-01
4.87929463e-01 -1.04307151e+00 -1.87182456e-01 2.68613230e-02
-3.09293330e-01 1.39045686e-01 1.11664534e+00 -5.74573696e-01
9.49374795e-01 -2.13582039e+00 -2.06374958e-01 1.39650047e-01
1.37240961e-01 -2.26145349e-02 3.07506561e-01 1.05317205e-01
4.90691364e-01 -4.69327956e-01 -2.65543699e-01 -3.28258991e-01
-5.85872084e-02 3.95598352e-01 -4.88473564e-01 6.25600278e-01
-1.94651797e-01 9.08981323e-01 -8.10999453e-01 -5.45617342e-01
9.54655230e-01 5.18061697e-01 -5.71478248e-01 1.99794769e-01
-2.06212401e-02 4.10741270e-01 -2.99259365e-01 4.96165603e-01
1.50772965e+00 -1.34713622e-02 5.22361435e-02 -3.14023346e-01
-4.85697865e-01 3.60882610e-01 -1.28843021e+00 1.66317475e+00
-7.41506577e-01 8.26710939e-01 -1.38680056e-01 -5.49725890e-01
8.66430461e-01 -4.14414525e-01 2.91498989e-01 -1.17204916e+00
-2.26523012e-01 2.09014699e-01 -5.48570633e-01 -1.81842268e-01
8.94235253e-01 3.26509416e-01 -1.06952816e-01 5.34293428e-02
-5.67953348e-01 -3.78238767e-01 -5.28436899e-02 1.33273676e-01
6.86246395e-01 2.52092481e-01 1.99284151e-01 -1.07803196e-03
5.23402214e-01 5.18358424e-02 7.03024745e-01 3.94279689e-01
3.39509100e-02 6.76650405e-01 2.22754061e-01 -5.19008994e-01
-1.20602274e+00 -1.16464281e+00 -3.26630890e-01 8.84285092e-01
9.22165751e-01 -2.42843062e-01 -6.13724232e-01 -5.65946043e-01
1.68148696e-01 7.98159122e-01 -4.50226724e-01 7.06939027e-02
-7.51139581e-01 -6.54660046e-01 8.81564394e-02 6.25321150e-01
9.12641227e-01 -4.76564676e-01 -1.01802492e+00 5.12111150e-02
-5.35227656e-01 -1.72668195e+00 -4.59749073e-01 -1.25795394e-01
-7.10882723e-01 -9.89176214e-01 -3.00612807e-01 -4.10587788e-01
5.82991898e-01 1.03359652e+00 1.07649100e+00 -2.74793148e-01
1.07666431e-02 -5.97610250e-02 6.11283481e-02 -3.27224612e-01
-4.73818630e-02 6.77604750e-02 -3.36898863e-01 -9.23182294e-02
3.76852602e-01 -4.52437609e-01 -8.50549102e-01 8.73968959e-01
-5.63551068e-01 8.10096264e-01 6.41260624e-01 4.68427896e-01
9.01869059e-01 3.45175415e-02 -7.21964538e-02 -8.59589577e-01
-5.27390122e-01 -4.31088984e-01 -1.10136890e+00 -6.35615140e-02
-8.23227167e-01 -2.17880756e-01 3.18239599e-01 1.69470161e-01
-1.33173561e+00 5.72337270e-01 -2.23722786e-01 -5.36602974e-01
-1.26097560e-01 -5.56432493e-02 -6.12052619e-01 5.43566514e-03
2.36650258e-01 5.55398643e-01 -1.95555955e-01 -1.16531178e-01
5.12277842e-01 6.45672560e-01 7.60854006e-01 2.76549496e-02
7.88669825e-01 1.10661757e+00 1.02180488e-01 -8.20281386e-01
-9.52857256e-01 -7.51891673e-01 -7.06723094e-01 -4.79649007e-01
8.17610264e-01 -1.38394451e+00 -5.66655517e-01 5.64788580e-01
-1.14560509e+00 -2.34533772e-01 1.60255313e-01 3.27199072e-01
-7.08111346e-01 3.09370190e-01 -1.81292266e-01 -6.22324586e-01
-9.53559279e-02 -1.29582810e+00 1.51057863e+00 1.10452868e-01
6.44532979e-01 -4.87708569e-01 -1.61655694e-01 7.10129559e-01
1.00141361e-01 9.79667529e-02 4.49250042e-01 3.42781186e-01
-1.42660987e+00 1.42986774e-01 -6.95434332e-01 1.11156680e-01
-3.49593669e-01 -1.55956253e-01 -1.26355076e+00 2.10903492e-02
-2.86714405e-01 2.26692706e-01 1.18956542e+00 5.02394617e-01
1.26565576e+00 1.81586564e-01 -6.03451014e-01 1.08585763e+00
1.49525821e+00 -1.15296751e-01 1.05323982e+00 3.78044724e-01
1.19735718e+00 7.61940002e-01 1.25548673e+00 3.64639938e-01
1.21560526e+00 9.93106246e-01 8.45263004e-01 -2.08397284e-01
-1.62370294e-01 -5.88292003e-01 3.32568288e-01 6.42887831e-01
1.27426356e-01 -1.98261350e-01 -8.40712726e-01 5.29064238e-01
-1.95840859e+00 -9.36889768e-01 -6.19887650e-01 2.27595687e+00
2.37214342e-01 3.09106052e-01 2.30463352e-02 -6.61820322e-02
6.01873100e-01 4.22346860e-01 -7.42901683e-01 -1.20032951e-02
-1.43714517e-01 -4.10468400e-01 1.39619541e+00 6.48127437e-01
-1.08453679e+00 1.10521317e+00 5.39096451e+00 9.36833322e-01
-9.72633719e-01 2.81538337e-01 5.32356024e-01 6.89322948e-02
-5.98233819e-01 1.42249372e-03 -1.33106804e+00 5.70451617e-01
8.32959354e-01 2.11758822e-01 3.19621533e-01 9.98139322e-01
2.23478258e-01 -6.18144393e-01 -8.94645989e-01 1.20366919e+00
6.61590695e-02 -1.44939673e+00 -2.58657515e-01 4.76792097e-01
9.03192043e-01 5.91612875e-01 -1.65557601e-02 4.69587483e-02
1.21875502e-01 -6.85154974e-01 8.83903623e-01 3.93334150e-01
9.30449188e-01 -9.90360022e-01 7.40321100e-01 6.43511236e-01
-1.68376446e+00 1.71990823e-02 -5.26552796e-01 1.08988121e-01
5.22884667e-01 7.84029007e-01 -7.33575225e-01 6.47996008e-01
8.27261090e-01 9.26515579e-01 -5.11321902e-01 7.43729115e-01
-1.46137372e-01 1.51915289e-03 -4.85094756e-01 4.09184575e-01
9.56624076e-02 -2.70427555e-01 2.44541407e-01 1.01273561e+00
4.70691919e-01 3.12659554e-02 1.60289586e-01 7.25347281e-01
1.52265817e-01 -3.13122869e-01 -8.61562550e-01 6.93205118e-01
7.36793935e-01 1.22034097e+00 -7.73181736e-01 -3.72479320e-01
-5.60113907e-01 7.97014713e-01 1.34644166e-01 8.03234577e-02
-1.27819288e+00 -1.27524853e-01 6.84596300e-01 5.91565609e-01
5.89824736e-01 -2.64400899e-01 -5.55851877e-01 -1.00974369e+00
3.05683434e-01 -2.84815252e-01 -6.84651434e-02 -8.67408156e-01
-5.70851266e-01 3.70372355e-01 5.33727482e-02 -1.47757888e+00
-1.69407606e-01 -3.22906613e-01 -7.33620822e-02 6.63464427e-01
-2.20241570e+00 -1.21668363e+00 -1.00105929e+00 5.96990705e-01
7.85769582e-01 3.69822592e-01 2.50424981e-01 6.44164741e-01
-3.46227705e-01 2.31567442e-01 9.04580355e-02 -3.57656062e-01
6.08138561e-01 -9.35337186e-01 8.51841152e-01 9.71483588e-01
-1.42353877e-01 -1.63074180e-01 5.51301897e-01 -5.53683162e-01
-1.50247335e+00 -1.52324200e+00 7.82833993e-01 -6.54278338e-01
3.22602466e-02 -5.08851469e-01 -6.81416631e-01 5.38874030e-01
-2.55742610e-01 3.09017509e-01 -4.53292094e-02 -3.27335268e-01
-2.65895605e-01 -5.42478561e-01 -8.82350802e-01 3.54979932e-01
1.40182042e+00 -5.99293709e-01 -2.57384637e-03 2.57076830e-01
7.72601247e-01 -9.99840379e-01 -4.95884538e-01 6.65753126e-01
7.85519481e-01 -1.43329227e+00 1.19485855e+00 4.46693420e-01
4.02881742e-01 -7.05416024e-01 -4.48177993e-01 -8.84111524e-01
1.18871834e-02 -7.61518925e-02 -4.82672602e-02 8.53763700e-01
2.21720338e-01 -7.50818849e-01 8.59827399e-01 3.85130465e-01
-5.34385145e-01 -6.53032660e-01 -8.51176441e-01 -6.04135931e-01
-4.76223648e-01 -7.81975627e-01 8.83263886e-01 5.11473775e-01
-5.61877072e-01 2.89443225e-01 -5.36502898e-01 5.75527072e-01
7.05640912e-01 4.48695183e-01 1.35374105e+00 -1.15648913e+00
1.14103451e-01 -6.49933591e-02 -4.76104438e-01 -1.56501746e+00
1.58782884e-01 -4.53274488e-01 2.81709880e-01 -1.63787842e+00
3.10095370e-01 -4.86895502e-01 1.32217333e-01 -9.95742530e-02
-1.17675856e-01 3.85284692e-01 2.08472982e-02 1.43066719e-01
-7.15496063e-01 6.02036238e-01 1.21959972e+00 5.39613813e-02
-5.63188642e-03 2.88437065e-02 -2.52229035e-01 8.65368307e-01
5.98299861e-01 -3.90719950e-01 -6.27863348e-01 -5.48239589e-01
3.40278387e-01 6.59459084e-03 4.75523353e-01 -1.04182529e+00
3.05476815e-01 -3.18448544e-01 2.95323014e-01 -1.50495386e+00
6.80272043e-01 -9.33125794e-01 3.40214521e-01 2.96100259e-01
3.36369872e-01 1.90658662e-02 8.20403621e-02 7.73370802e-01
-1.60480261e-01 3.58832717e-01 8.14758837e-01 1.10877618e-01
-1.26502132e+00 6.06194139e-01 4.35091667e-02 -4.67457399e-02
1.19149852e+00 -5.61187744e-01 -6.11343145e-01 -3.11403990e-01
1.05513133e-01 5.55160940e-01 7.68258095e-01 5.91477692e-01
8.16229582e-01 -1.27637923e+00 -5.78156054e-01 4.95754778e-01
4.42138791e-01 4.00884688e-01 7.32133150e-01 9.54618573e-01
-6.57351255e-01 7.24327922e-01 -1.65991440e-01 -1.13322997e+00
-1.28352404e+00 4.77555633e-01 1.93896830e-01 3.05729695e-02
-1.01621294e+00 4.47756827e-01 7.78530836e-01 -6.47253990e-01
-1.29086822e-01 -3.52151752e-01 3.65935005e-02 -1.84892625e-01
5.50698042e-01 6.18934274e-01 2.91797519e-01 -9.71772671e-01
-4.38876718e-01 9.18984294e-01 1.40498221e-01 -1.15655512e-01
1.08910298e+00 -7.65569687e-01 1.79747954e-01 2.53195077e-01
1.12241673e+00 -2.99274419e-02 -1.81322277e+00 -2.29907036e-01
-5.16551852e-01 -8.95495892e-01 4.62249577e-01 -3.87704581e-01
-1.14865160e+00 1.06353974e+00 7.23286748e-01 -3.37999910e-01
9.58379626e-01 -1.77846327e-02 1.14894462e+00 2.44895071e-01
7.39055157e-01 -1.27996492e+00 -4.64748442e-01 5.24089813e-01
4.67656076e-01 -1.53583097e+00 1.81105942e-01 -1.00082302e+00
-6.39196992e-01 8.76959443e-01 8.60778809e-01 1.41994476e-01
5.39925754e-01 3.13751340e-01 -2.02103704e-01 -1.47809595e-01
-5.75418353e-01 -3.12672704e-01 2.57937372e-01 5.62736988e-01
-3.14522415e-01 3.07089388e-01 2.18512982e-01 -1.16655380e-02
-2.97255158e-01 -1.52894616e-01 4.50818717e-01 5.94728410e-01
-7.74962544e-01 -5.75821161e-01 -2.91655242e-01 4.04558659e-01
1.50340587e-01 3.79859023e-02 2.42167771e-01 8.05307031e-01
4.24790323e-01 8.58985603e-01 3.11729580e-01 -5.35988927e-01
5.65096617e-01 -4.37288463e-01 3.86319935e-01 -2.71872282e-01
2.51679957e-01 -4.51488160e-02 1.65105700e-01 -1.14266181e+00
-3.42319340e-01 -7.27408409e-01 -1.16855872e+00 -8.13698471e-01
-2.19430864e-01 -5.02779186e-01 9.59850132e-01 9.11423206e-01
4.44547325e-01 4.02053684e-01 8.76042724e-01 -9.67057526e-01
-1.78612247e-02 -5.59782684e-01 -5.65644622e-01 5.80208749e-03
3.06889951e-01 -8.31924319e-01 -1.54928356e-01 -1.14693977e-01] | [8.197169303894043, -2.1952428817749023] |
165b7fdb-bc2c-4410-b81e-3dc93256e959 | crisp-curriculum-inducing-primitive-informed | 2304.03535 | null | https://arxiv.org/abs/2304.03535v1 | https://arxiv.org/pdf/2304.03535v1.pdf | CRISP: Curriculum inducing Primitive Informed Subgoal Prediction for Hierarchical Reinforcement Learning | Hierarchical reinforcement learning is a promising approach that uses temporal abstraction to solve complex long horizon problems. However, simultaneously learning a hierarchy of policies is unstable as it is challenging to train higher-level policy when the lower-level primitive is non-stationary. In this paper, we propose a novel hierarchical algorithm by generating a curriculum of achievable subgoals for evolving lower-level primitives using reinforcement learning and imitation learning. The lower level primitive periodically performs data relabeling on a handful of expert demonstrations using our primitive informed parsing approach. We provide expressions to bound the sub-optimality of our method and develop a practical algorithm for hierarchical reinforcement learning. Since our approach uses a handful of expert demonstrations, it is suitable for most robotic control tasks. Experimental evaluation on complex maze navigation and robotic manipulation environments show that inducing hierarchical curriculum learning significantly improves sample efficiency, and results in efficient goal conditioned policies for solving temporally extended tasks. | ['Vinay P Namboodiri', 'Utsav Singh'] | 2023-04-07 | null | null | null | null | ['hierarchical-reinforcement-learning'] | ['methodology'] | [ 2.16564819e-01 3.84242296e-01 -3.09111118e-01 -2.19742022e-02
-1.00746906e+00 -8.29392731e-01 5.29998302e-01 9.44214091e-02
-6.64660871e-01 1.16689026e+00 4.40995432e-02 -4.57679331e-01
-4.06286359e-01 -6.47663057e-01 -8.84170711e-01 -7.71801949e-01
-7.10658371e-01 7.19275892e-01 6.19456053e-01 -2.91720212e-01
3.52551430e-01 6.35677099e-01 -1.61565912e+00 1.83349565e-01
9.18378413e-01 4.62484300e-01 7.05192089e-01 1.11568987e+00
8.29212219e-02 9.36953008e-01 -5.30187666e-01 4.07956183e-01
4.17996138e-01 -5.66701889e-01 -1.07087588e+00 1.08364329e-01
5.31210452e-02 -6.20792806e-01 -9.37284455e-02 9.33024287e-01
1.77646384e-01 6.36793196e-01 4.73637223e-01 -1.25535810e+00
1.54517889e-01 8.40858936e-01 -2.20541552e-01 1.03493601e-01
3.18692476e-01 4.07246351e-01 1.00812078e+00 1.92440413e-02
7.11769044e-01 1.46935081e+00 2.99477905e-01 7.76422381e-01
-1.14283001e+00 -4.69954371e-01 6.59189165e-01 -8.03425501e-04
-6.79415762e-01 1.11273319e-01 3.79325777e-01 -5.58643162e-01
1.02143526e+00 -1.96876645e-01 7.65245438e-01 1.08496964e+00
4.98986363e-01 8.90189826e-01 1.48733163e+00 -3.26783627e-01
6.08558357e-01 -4.85499442e-01 4.88275699e-02 1.06556892e+00
-1.32148147e-01 8.04143369e-01 -2.38941789e-01 -2.80576557e-01
1.19038236e+00 -3.21026623e-01 9.90584791e-02 -8.78848374e-01
-1.16128874e+00 8.64121079e-01 1.57718733e-01 9.77625474e-02
-4.66383487e-01 5.70839643e-01 2.71696746e-01 6.24429643e-01
-4.10357505e-01 8.45440269e-01 -6.91561162e-01 -4.01758879e-01
-5.44815004e-01 7.53722131e-01 9.72679496e-01 1.13136947e+00
6.42012954e-01 1.85238242e-01 -1.44655928e-01 5.13405204e-01
-5.32331094e-02 9.09039751e-02 3.67329359e-01 -1.84168124e+00
4.08631623e-01 8.84578601e-02 5.76898873e-01 -5.60301542e-02
-6.22532368e-01 -6.12262003e-02 -7.77262673e-02 9.86935735e-01
6.09356999e-01 -3.77939105e-01 -1.02456677e+00 2.00228453e+00
5.34539163e-01 -1.04224682e-01 3.19981992e-01 6.22158527e-01
1.95388985e-03 8.50617468e-01 3.34532261e-01 -5.18397450e-01
9.96195853e-01 -1.19374526e+00 -4.08751547e-01 -8.79454762e-02
6.94886386e-01 -1.09716661e-01 1.21712172e+00 5.77954412e-01
-1.22759080e+00 -6.43293798e-01 -9.68194246e-01 2.08506525e-01
-5.61563559e-02 -4.02542681e-01 8.35514069e-01 3.50692779e-01
-1.07893634e+00 1.23849380e+00 -1.26141667e+00 -2.81014949e-01
5.61802648e-02 6.48992181e-01 5.37460633e-02 1.16764463e-01
-8.26828718e-01 9.22336221e-01 8.35818172e-01 -4.94094282e-01
-1.84872985e+00 -2.95955420e-01 -9.66103315e-01 9.42715406e-02
6.60733342e-01 -5.29875040e-01 1.96933711e+00 -4.26152915e-01
-2.07416654e+00 2.42801756e-01 3.20645422e-01 -4.94667590e-01
3.92163843e-01 -2.04371855e-01 3.30744326e-01 2.91376173e-01
2.08546489e-01 1.00338435e+00 8.54199529e-01 -1.24498308e+00
-1.23762262e+00 -1.61634728e-01 7.12484062e-01 5.81531942e-01
5.36808930e-02 -2.01910108e-01 -3.28291468e-02 -2.44784251e-01
1.21830292e-02 -1.19496608e+00 -9.37881529e-01 -3.88145804e-01
2.07092777e-01 -6.08086646e-01 4.46180612e-01 -3.36879671e-01
7.47471154e-01 -1.83487105e+00 7.69166052e-01 -8.35225284e-02
-1.94758445e-01 -3.20933282e-01 -3.59256059e-01 5.11450112e-01
1.92254975e-01 -2.30223864e-01 -4.59106356e-01 1.08656108e-01
2.14479148e-01 7.12887645e-01 -3.83747816e-01 1.33150697e-01
-4.05955277e-02 6.58121943e-01 -1.41158116e+00 -5.30410767e-01
1.04233705e-01 -1.80193186e-01 -1.02886355e+00 4.09882993e-01
-7.86492527e-01 9.39160645e-01 -7.98939824e-01 4.81067300e-01
-2.39576116e-01 1.16360411e-01 3.62902761e-01 6.26283765e-01
-3.68747532e-01 4.55508053e-01 -1.00777364e+00 2.10768723e+00
-4.67574507e-01 1.31295070e-01 1.99005619e-01 -9.42647755e-01
5.02780080e-01 1.93970099e-01 6.28217638e-01 -5.11483967e-01
-9.80812088e-02 2.84538679e-02 1.99521869e-01 -6.26210749e-01
6.06655240e-01 -2.11617902e-01 -6.85031652e-01 1.64819494e-01
1.42545611e-01 -8.35135162e-01 4.94976819e-01 -2.95932051e-02
1.36695576e+00 1.11220098e+00 4.89661634e-01 -3.02075356e-01
2.47225210e-01 5.07668138e-01 7.50783801e-01 1.24863529e+00
-4.64057148e-01 -1.93479881e-01 6.75260901e-01 -1.40448764e-01
-1.15097451e+00 -1.15482652e+00 2.47998416e-01 1.36157095e+00
1.74175486e-01 -2.94511676e-01 -9.07347143e-01 -7.35553324e-01
-1.85563087e-01 8.96563053e-01 -4.05758530e-01 1.02731571e-01
-1.10644281e+00 -2.82091528e-01 2.13422224e-01 5.07534564e-01
3.74940872e-01 -1.55194259e+00 -1.45215070e+00 4.98132378e-01
1.04001604e-01 -1.06041849e+00 -2.03013942e-01 5.90693474e-01
-1.22930276e+00 -1.15627110e+00 -5.24439573e-01 -1.07136607e+00
5.00059068e-01 -1.52844161e-01 8.95686805e-01 -1.71268940e-01
-1.34888932e-01 8.53117406e-01 -3.11664224e-01 -9.25854445e-02
-7.46400893e-01 3.86790000e-02 1.04401469e-01 -1.03340232e+00
-3.77981156e-01 -7.42422879e-01 -2.46813014e-01 -5.47851548e-02
-4.99039948e-01 9.88741294e-02 5.62775016e-01 1.09686220e+00
5.20534098e-01 3.71671289e-01 2.47166485e-01 -5.68998754e-01
7.72384584e-01 -1.31931588e-01 -1.26942539e+00 5.64335175e-02
-3.65290791e-01 5.20397484e-01 7.32782960e-01 -5.97681463e-01
-1.22539282e+00 3.85887504e-01 7.33087361e-02 -1.02908418e-01
-4.06149715e-01 3.15466762e-01 1.59031302e-01 2.07112744e-01
6.09237790e-01 8.25850219e-02 -3.29497576e-01 -1.97290942e-01
3.80250335e-01 5.38654439e-02 5.77293515e-01 -1.53821373e+00
6.00904584e-01 -2.31808590e-04 2.15067357e-01 -7.46030092e-01
-5.46841919e-01 -1.86308950e-01 -5.75195372e-01 -2.56640792e-01
9.19207215e-01 -5.91716886e-01 -1.13556397e+00 4.02495712e-02
-9.42773461e-01 -1.32755101e+00 -5.92269003e-01 4.40062374e-01
-1.57964694e+00 3.30420166e-01 -8.53355348e-01 -9.95454013e-01
1.88724130e-01 -1.56251156e+00 9.99984324e-01 2.57754833e-01
-2.24754766e-01 -5.77363312e-01 2.45642334e-01 -1.12595759e-01
-2.60147810e-01 4.15033787e-01 1.10922325e+00 -1.33105144e-01
-9.15087700e-01 5.25539398e-01 3.75291735e-01 -1.65128231e-01
-1.67464107e-01 -3.47858578e-01 -3.35699618e-01 -6.65215790e-01
-2.61172019e-02 -9.28127527e-01 4.93512839e-01 3.85936111e-01
1.29277337e+00 -2.39546910e-01 -2.61958301e-01 2.38445535e-01
1.20561552e+00 7.55766332e-01 3.47211361e-01 6.23903692e-01
1.71468303e-01 9.13121402e-01 1.27828026e+00 4.57260460e-01
1.45862222e-01 4.97369349e-01 5.39149761e-01 6.37967110e-01
1.88049108e-01 -3.93424749e-01 7.03249395e-01 4.31177258e-01
-1.35082334e-01 2.55664647e-01 -7.17418849e-01 5.13266265e-01
-2.04167700e+00 -9.56517577e-01 4.01951075e-01 1.89316869e+00
1.00894690e+00 4.06840384e-01 4.30754036e-01 -2.35230148e-01
2.81158358e-01 -1.27513841e-01 -8.62340748e-01 -4.07880634e-01
5.74100733e-01 3.88744533e-01 4.01080251e-01 8.55892062e-01
-9.46534872e-01 1.39077020e+00 7.05505085e+00 4.72434461e-01
-4.33145016e-01 -4.91036288e-02 -6.32848665e-02 4.75022718e-02
5.53799197e-02 2.17096210e-01 -9.79936123e-01 5.87313883e-02
9.64714110e-01 -1.28705248e-01 7.85503983e-01 1.04734540e+00
1.59753025e-01 -3.51728767e-01 -1.44603860e+00 4.77225810e-01
-7.60424972e-01 -1.06583881e+00 -1.95296586e-01 3.25483233e-02
7.80501842e-01 -6.50381595e-02 -9.05995667e-02 9.06816006e-01
1.39826763e+00 -7.71921515e-01 6.78618789e-01 -5.78310564e-02
4.17615891e-01 -9.00785029e-01 -4.31852490e-02 8.93964767e-01
-1.06348252e+00 -6.15025103e-01 -3.19129199e-01 -2.69159645e-01
1.57324225e-01 -5.30715406e-01 -7.67700016e-01 3.19728822e-01
7.32356906e-01 2.77511626e-01 9.71846431e-02 8.53290319e-01
-6.01597071e-01 2.59477496e-01 -2.37142280e-01 -2.63503522e-01
8.95094991e-01 -3.38909775e-01 5.00967085e-01 9.82482970e-01
2.38770783e-01 5.74267268e-01 9.02985692e-01 5.43680370e-01
4.91482377e-01 -3.18071961e-01 -7.20833778e-01 -1.33444846e-01
2.82908410e-01 9.67276275e-01 -9.06577945e-01 -4.19242114e-01
-4.41375226e-02 8.68630826e-01 7.39042222e-01 5.00305831e-01
-9.18267310e-01 -2.03288309e-02 6.14794612e-01 -5.78913927e-01
4.85757023e-01 -7.59028018e-01 1.62788615e-01 -8.81456792e-01
-4.19735402e-01 -1.09669387e+00 6.25061154e-01 -6.65230274e-01
-5.98104596e-01 3.22959661e-01 4.88512516e-01 -1.07378590e+00
-8.73326838e-01 -6.67737007e-01 -1.31023616e-01 3.77497494e-01
-1.38657212e+00 -5.29453814e-01 -1.29666016e-01 6.46426916e-01
1.14803433e+00 -7.26148561e-02 1.01529574e+00 -4.07237291e-01
-2.29671001e-01 -4.47593909e-03 -1.05548911e-01 -3.07479322e-01
7.54373372e-02 -1.64711761e+00 1.35566667e-01 6.54420495e-01
-2.30072722e-01 4.90585536e-01 1.06968284e+00 -7.71713734e-01
-1.18317807e+00 -6.14360750e-01 -8.82684216e-02 -1.89948454e-01
8.65379393e-01 -2.84788031e-02 -5.97884238e-01 9.59390163e-01
3.17763954e-01 -6.92866087e-01 2.78842211e-01 1.85044445e-02
-1.77217647e-02 3.48101258e-01 -9.63648856e-01 1.09051049e+00
1.25506842e+00 -2.49156252e-01 -1.08927870e+00 3.97146195e-01
9.25647616e-01 -8.09137464e-01 -9.81548905e-01 6.06899619e-01
4.51795340e-01 -7.50024438e-01 7.81029642e-01 -9.09292281e-01
4.49318260e-01 -2.64819503e-01 -4.76544648e-02 -1.48461521e+00
-4.27236348e-01 -9.54209268e-01 -2.80028522e-01 5.38802564e-01
3.28142717e-02 -3.55912209e-01 9.69621301e-01 2.14284584e-01
-4.63650912e-01 -5.62468886e-01 -6.80323660e-01 -1.23123121e+00
4.87097710e-01 -3.50115076e-02 1.58367261e-01 5.76194465e-01
3.68826032e-01 2.36522928e-01 -2.75376260e-01 4.42157060e-01
9.43755507e-01 3.88032585e-01 7.68578589e-01 -9.28020716e-01
-7.95037031e-01 -5.01037121e-01 2.64897704e-01 -1.50571978e+00
5.27308106e-01 -6.69055760e-01 5.96447587e-01 -1.44886839e+00
-1.00079715e-01 -5.78189671e-01 -2.01742128e-01 4.30725843e-01
1.57115608e-01 -7.77456403e-01 2.21033856e-01 -3.99501361e-02
-6.73255563e-01 6.34047210e-01 1.70829058e+00 -4.95069772e-02
-5.73677719e-01 -2.10649800e-02 -1.91993490e-01 8.46457005e-01
1.11977196e+00 -4.86846596e-01 -7.25216568e-01 -1.21931642e-01
-2.38371074e-01 8.02610815e-01 7.45157227e-02 -1.10617328e+00
1.87591508e-01 -6.54661655e-01 9.79842544e-02 -7.07221389e-01
4.50730681e-01 -7.72692382e-01 -1.66606411e-01 1.07987261e+00
-7.08132863e-01 1.94918171e-01 3.17219019e-01 6.99697137e-01
1.26711518e-01 -5.92205465e-01 8.40805769e-01 -6.54034138e-01
-1.10479283e+00 1.74588472e-01 -8.77467036e-01 2.76970938e-02
1.30733466e+00 2.22200211e-02 9.74078178e-02 -2.32018843e-01
-1.11188579e+00 7.90579796e-01 3.19694459e-01 3.24430078e-01
4.89543527e-01 -8.69921565e-01 -2.02358931e-01 -1.56790271e-01
-1.46473512e-01 1.30201459e-01 -1.97997093e-01 2.93834716e-01
-3.81886899e-01 4.00600702e-01 -7.59900570e-01 -6.28303885e-01
-1.20327723e+00 7.77533293e-01 2.91602433e-01 -6.78226233e-01
-8.32272410e-01 5.61206222e-01 2.77732790e-01 -5.58067381e-01
5.77618480e-01 -6.83675468e-01 -2.64973819e-01 -3.22917640e-01
2.77016580e-01 2.85113573e-01 -6.98351979e-01 1.34540394e-01
1.22644879e-01 4.21569943e-01 -7.74022862e-02 -6.45613313e-01
1.30293679e+00 5.53040691e-02 4.19493839e-02 5.65822363e-01
5.68600774e-01 -5.19369900e-01 -2.07911086e+00 8.21693912e-02
4.69418257e-01 -4.27741073e-02 -4.19073224e-01 -4.73116577e-01
-1.55083179e-01 6.65462852e-01 1.89073876e-01 1.43752351e-01
8.33189428e-01 -3.54523100e-02 3.92397672e-01 1.16582620e+00
1.21776581e+00 -1.26539910e+00 6.11566484e-01 9.55966771e-01
9.22039568e-01 -8.19313943e-01 -8.27018544e-02 -2.56285757e-01
-4.05468076e-01 1.13174582e+00 1.03683817e+00 -2.59297848e-01
4.51814197e-02 3.09506595e-01 -4.24965858e-01 -7.37683177e-02
-8.42882156e-01 -3.58278304e-01 -4.78476584e-01 7.24084735e-01
-1.74020961e-01 -1.04908329e-02 -3.71021092e-01 8.04297477e-02
-3.77567351e-01 -1.42760485e-01 7.11473286e-01 1.57023549e+00
-1.03471398e+00 -1.31071329e+00 -3.89103085e-01 2.46294327e-02
-1.92728981e-01 2.25428209e-01 3.07683628e-02 1.19099915e+00
-1.38951838e-01 6.72872126e-01 -1.47569463e-01 1.00681521e-01
2.98177868e-01 1.11927405e-01 1.36397254e+00 -9.73635972e-01
-3.85188460e-01 2.30958477e-01 2.66216189e-01 -9.65999603e-01
-4.30783838e-01 -9.22609329e-01 -1.90869141e+00 2.87938803e-01
3.43370378e-01 4.27684873e-01 3.46176147e-01 9.04200017e-01
-2.39867374e-01 7.86679983e-01 4.19711977e-01 -1.07791531e+00
-1.08039713e+00 -5.33220351e-01 -4.63249177e-01 2.24865437e-01
4.28686768e-01 -9.45756733e-01 -1.79886278e-02 5.04412949e-02] | [4.209881782531738, 1.4591076374053955] |
69aafbfd-3e35-4d5d-bf04-4bb8bd6be7c7 | multi-attribute-open-set-recognition | 2208.06809 | null | https://arxiv.org/abs/2208.06809v1 | https://arxiv.org/pdf/2208.06809v1.pdf | Multi-Attribute Open Set Recognition | Open Set Recognition (OSR) extends image classification to an open-world setting, by simultaneously classifying known classes and identifying unknown ones. While conventional OSR approaches can detect Out-of-Distribution (OOD) samples, they cannot provide explanations indicating which underlying visual attribute(s) (e.g., shape, color or background) cause a specific sample to be unknown. In this work, we introduce a novel problem setup that generalizes conventional OSR to a multi-attribute setting, where multiple visual attributes are simultaneously recognized. Here, OOD samples can be not only identified but also categorized by their unknown attribute(s). We propose simple extensions of common OSR baselines to handle this novel scenario. We show that these baselines are vulnerable to shortcuts when spurious correlations exist in the training dataset. This leads to poor OOD performance which, according to our experiments, is mainly due to unintended cross-attribute correlations of the predicted confidence scores. We provide an empirical evidence showing that this behavior is consistent across different baselines on both synthetic and real world datasets. | ['Volker Fischer', 'Mauricio Munoz', 'Claudia Blaiotta', 'Chaithanya Kumar Mummadi', 'Piyapat Saranrittichai'] | 2022-08-14 | null | null | null | null | ['open-set-learning'] | ['miscellaneous'] | [ 4.67949957e-01 6.73680156e-02 -3.96619171e-01 -6.02592409e-01
-9.96759474e-01 -9.48995709e-01 7.61647403e-01 3.39003980e-01
5.17921820e-02 6.49278641e-01 -1.36168674e-01 -1.94468215e-01
-1.99900325e-02 -7.19561756e-01 -1.00393236e+00 -6.12103164e-01
-8.69621187e-02 7.19846427e-01 1.01378791e-01 1.87844455e-01
2.43697003e-01 3.81012112e-01 -2.05372167e+00 4.34202313e-01
6.91925466e-01 1.01880789e+00 -2.86526233e-01 7.00934112e-01
-1.82745233e-01 4.12738591e-01 -7.88388073e-01 -4.05349106e-01
4.15819526e-01 -9.32030454e-02 -6.20077074e-01 3.02664459e-01
1.08870840e+00 -1.70072913e-01 -1.37068376e-01 1.13675654e+00
1.85176149e-01 -7.63991773e-02 1.31364512e+00 -1.78677785e+00
-1.00722373e+00 4.27040040e-01 -6.20785534e-01 5.89182228e-02
4.23779041e-01 2.16566458e-01 1.31314170e+00 -9.29959476e-01
5.43829024e-01 1.48060679e+00 5.15639842e-01 5.75613081e-01
-1.80389309e+00 -6.89442813e-01 3.75579327e-01 1.77646324e-01
-1.50689089e+00 -2.43457451e-01 5.45394599e-01 -3.89733106e-01
3.36013317e-01 4.74564701e-01 1.27258837e-01 1.60144150e+00
-1.96517389e-02 1.03034496e+00 1.53622186e+00 -4.59546268e-01
3.55577737e-01 4.11390632e-01 6.36856616e-01 2.73221701e-01
6.59140170e-01 2.82846361e-01 -4.35619652e-01 -2.85863489e-01
3.21917057e-01 -4.51243296e-02 -1.92655534e-01 -7.69101799e-01
-1.24775887e+00 7.59601414e-01 4.51334149e-01 -2.28490591e-01
4.15047929e-02 -8.75420272e-02 1.49960279e-01 5.41479468e-01
4.67341393e-01 5.33709764e-01 -7.11015701e-01 2.89851904e-01
-3.92721832e-01 -5.13151772e-02 8.32561553e-01 1.00560141e+00
9.82501209e-01 -2.68448502e-01 -2.98122615e-01 9.21741307e-01
2.11075515e-01 5.71367204e-01 4.76671100e-01 -7.46352196e-01
2.48075411e-01 4.92694944e-01 1.47689000e-01 -8.61804843e-01
-1.17681116e-01 -4.77035731e-01 -5.09374797e-01 3.87139410e-01
7.93738127e-01 1.91126183e-01 -1.29805636e+00 1.79458022e+00
2.84989715e-01 1.69615090e-01 2.92357504e-01 7.98530936e-01
8.94249439e-01 3.64861965e-01 1.26646921e-01 7.31587261e-02
1.42487681e+00 -6.69553459e-01 -4.32574302e-01 -5.73764741e-01
4.48235720e-01 -6.31258667e-01 1.28949249e+00 3.81210744e-01
-1.70127466e-01 -4.70494270e-01 -1.13726413e+00 2.09579200e-01
-6.71251476e-01 9.24559012e-02 6.29837751e-01 7.88783848e-01
-6.50944769e-01 5.08176148e-01 -3.19437653e-01 -4.82616246e-01
5.39656997e-01 3.17936569e-01 -4.97627646e-01 -2.94815361e-01
-9.23166454e-01 5.85805118e-01 2.86350965e-01 -2.46977627e-01
-1.05311489e+00 -6.16777718e-01 -7.86333442e-01 2.07522116e-03
7.45618045e-01 -4.94610637e-01 1.06673086e+00 -1.35486460e+00
-6.46616876e-01 1.22967613e+00 -2.32716173e-01 -2.25085422e-01
7.28719652e-01 -1.69546932e-01 -5.07655382e-01 -1.39784038e-01
4.11003202e-01 6.19015932e-01 1.12320256e+00 -1.81492102e+00
-4.46579397e-01 -6.88489079e-01 2.67209690e-02 -3.42500098e-02
-2.55926400e-01 -1.83062613e-01 -4.00858261e-02 -8.64320457e-01
4.06535149e-01 -9.95817363e-01 -7.79752061e-02 1.70467004e-01
-8.97381485e-01 -2.48723045e-01 6.52842462e-01 1.10966533e-01
6.52188659e-01 -2.29211569e+00 -3.38834643e-01 2.75048465e-01
2.61722386e-01 -1.03593789e-01 -2.78425962e-01 1.18886344e-01
-3.80581290e-01 3.47598225e-01 -2.32260421e-01 -7.57992044e-02
1.04646042e-01 3.93005013e-01 -6.10510588e-01 5.71686983e-01
4.22482133e-01 6.30325735e-01 -6.26333177e-01 -3.84031773e-01
-4.52432707e-02 2.03023955e-01 9.02100187e-03 2.75235027e-01
-4.03385609e-01 1.52065948e-01 -3.67189914e-01 1.04131687e+00
7.29124367e-01 -3.88009310e-01 -1.12204589e-01 -1.96129829e-02
3.50605458e-01 5.07184453e-02 -1.43468785e+00 8.33565533e-01
-1.81082174e-01 6.46805167e-01 -4.56468165e-01 -8.85093689e-01
1.19626820e+00 1.82916112e-02 5.16541414e-02 -2.98369825e-01
-1.26431733e-01 3.24047476e-01 -1.98141914e-02 -2.28163451e-01
4.27128077e-01 -9.49569270e-02 -1.48267206e-02 6.00413024e-01
3.31585780e-02 2.98257172e-01 -2.95647755e-02 1.54328018e-01
9.71775472e-01 3.24656442e-02 5.05911112e-01 -3.21926624e-01
2.61526346e-01 -7.96290413e-02 6.79639220e-01 1.34261274e+00
-4.50314850e-01 9.76501107e-01 7.50169337e-01 -3.94488424e-01
-1.06598878e+00 -1.32745826e+00 -5.11235714e-01 1.03578305e+00
4.82282132e-01 -1.79077417e-01 -3.50998402e-01 -1.07602847e+00
4.17603940e-01 6.93572521e-01 -1.06900060e+00 -2.76821464e-01
1.70650050e-01 -8.75036955e-01 5.47039866e-01 6.04040086e-01
-1.37101024e-01 -7.94439793e-01 -2.31391907e-01 -1.05933115e-01
-1.28753155e-01 -1.28773820e+00 -2.32162789e-01 5.36662459e-01
-7.68431425e-01 -1.29834950e+00 -5.86607397e-01 -4.49738026e-01
8.04327488e-01 3.99134904e-01 1.36048722e+00 1.73584893e-01
-3.31727177e-01 4.60945845e-01 -4.28838551e-01 -5.51741540e-01
-7.00797737e-01 -1.50120214e-01 2.16002285e-01 5.75977564e-01
6.95661128e-01 -2.96627581e-01 -4.02967930e-01 6.66407526e-01
-7.45078862e-01 -3.21571261e-01 7.24149287e-01 7.57616341e-01
6.17938280e-01 -1.74851015e-01 5.36128461e-01 -1.30928004e+00
1.85362563e-01 -6.93998992e-01 -4.29452032e-01 5.74478030e-01
-8.53950143e-01 2.98042685e-01 5.55718958e-01 -9.49376225e-01
-7.65443146e-01 2.08880007e-01 3.41710269e-01 -5.48249960e-01
-8.41309547e-01 -2.76966002e-02 -3.47160816e-01 1.12807818e-01
8.54576051e-01 1.59692138e-01 -1.24258623e-01 -3.99214745e-01
9.96691212e-02 8.11862290e-01 5.05532146e-01 -7.71159112e-01
9.75485802e-01 6.15494728e-01 -4.25752699e-02 -6.78790271e-01
-1.23697746e+00 -5.14173388e-01 -6.51601434e-01 -3.64521854e-02
5.66711485e-01 -1.06458020e+00 -4.11405683e-01 4.61851031e-01
-8.82872820e-01 -9.13411528e-02 -1.39804378e-01 3.77868146e-01
-5.19508958e-01 2.93602437e-01 -2.31475234e-01 -8.09692442e-01
2.70018786e-01 -1.09457326e+00 1.14454281e+00 1.54923409e-01
-3.02470982e-01 -7.33420908e-01 -3.11020583e-01 3.08243155e-01
8.43621511e-03 2.54199475e-01 9.43101645e-01 -1.34706235e+00
-5.56346893e-01 -2.91781455e-01 -3.70446146e-01 2.38178387e-01
1.53973714e-01 3.15575957e-01 -1.41701066e+00 -2.01369047e-01
-3.32858503e-01 -7.22617626e-01 1.03804529e+00 2.00508654e-01
1.19430614e+00 -1.34889752e-01 -4.30515438e-01 4.75958049e-01
1.37586319e+00 -8.04459304e-02 5.22576869e-01 4.75316793e-01
6.02001309e-01 6.21528506e-01 8.43967557e-01 2.87632942e-01
2.35162422e-01 7.17230737e-01 5.38001537e-01 -2.92993665e-01
-1.36816651e-01 -2.38542557e-01 5.06938100e-01 -3.18076879e-01
2.38313928e-01 -4.13847506e-01 -8.18700373e-01 4.22519445e-01
-1.60323131e+00 -7.45769083e-01 -4.40569013e-01 2.53183389e+00
8.02343667e-01 2.78843790e-01 1.31033748e-01 1.67217359e-01
1.05461109e+00 -5.58665842e-02 -8.24894726e-01 -8.98152441e-02
-4.45069730e-01 -1.36188880e-01 6.23003423e-01 2.56751120e-01
-1.30140138e+00 6.07881546e-01 6.44000244e+00 6.41682327e-01
-9.30228829e-01 -3.05458069e-01 9.63077426e-01 3.40087205e-01
-4.38281834e-01 9.81145538e-03 -9.60726500e-01 3.30200046e-01
4.97382015e-01 6.45965487e-02 1.59568444e-01 1.03733706e+00
-5.08308291e-01 -1.54137552e-01 -1.73022807e+00 9.31319118e-01
3.36693496e-01 -8.89923513e-01 2.53359646e-01 2.12101132e-01
5.95615804e-01 -1.41580239e-01 1.05537370e-01 3.38004798e-01
5.68307698e-01 -1.08741450e+00 7.31839180e-01 6.08488880e-02
8.17115963e-01 -5.29113948e-01 6.76970482e-01 1.87208325e-01
-8.44327748e-01 -2.15544283e-01 -2.92075127e-01 -2.89289840e-02
-4.63557333e-01 4.94126409e-01 -8.96230638e-01 3.39062363e-01
7.49952078e-01 6.52990997e-01 -1.18060040e+00 9.23436344e-01
-3.20135027e-01 6.62045062e-01 -3.42076331e-01 2.26129934e-01
-5.78838475e-02 1.42383069e-01 6.38159990e-01 8.18808317e-01
9.43546295e-02 -2.73422867e-01 3.38690072e-01 1.04996800e+00
-4.75834943e-02 -1.05955333e-01 -7.87274837e-01 7.81907141e-02
5.52018583e-01 9.22560275e-01 -1.02983177e+00 -2.90121883e-01
-5.26113212e-01 9.27154243e-01 3.33002031e-01 3.53902429e-01
-6.16057277e-01 -2.07102835e-01 8.22223961e-01 3.71209695e-03
4.37637538e-01 2.46671721e-01 -3.55625302e-01 -1.44722891e+00
2.30533406e-02 -9.13999498e-01 8.38720262e-01 -6.77502215e-01
-1.97943592e+00 3.34504604e-01 -1.54273778e-01 -1.41049588e+00
7.48164160e-03 -8.73210311e-01 -2.82854438e-01 4.91322070e-01
-1.73076689e+00 -1.13484406e+00 -3.10142875e-01 4.42932576e-01
6.11086726e-01 -5.11063188e-02 8.86371255e-01 2.12259591e-02
-5.06919205e-01 8.06459606e-01 1.03596337e-01 4.03385013e-01
1.09841526e+00 -1.66626883e+00 4.13552910e-01 8.53106380e-01
6.40648484e-01 4.39025789e-01 9.88400817e-01 -4.36209202e-01
-1.13320601e+00 -1.03232181e+00 5.49252868e-01 -9.27956104e-01
6.66961789e-01 -6.28250241e-01 -1.16271734e+00 8.67123365e-01
-4.97459143e-01 7.03292489e-01 8.11745286e-01 5.30040026e-01
-1.04141963e+00 -1.13385625e-01 -1.05296993e+00 3.73748988e-01
8.85824919e-01 -4.61675495e-01 -5.73059261e-01 1.92196131e-01
5.46932518e-01 -2.15604320e-01 -7.11464465e-01 5.23116589e-01
5.28770804e-01 -1.03890657e+00 1.15335202e+00 -9.94970918e-01
5.44327497e-01 -3.87773007e-01 -4.72846389e-01 -1.28761268e+00
-7.43557885e-02 -2.29132950e-01 -5.06631173e-02 1.46150935e+00
5.61496854e-01 -7.69940197e-01 5.91788709e-01 6.38135552e-01
8.65690410e-02 -3.79465371e-01 -7.58971870e-01 -1.22263706e+00
1.08865425e-02 -5.98740339e-01 6.70189917e-01 1.08261919e+00
-5.13695180e-01 4.98602241e-01 -2.29948163e-01 6.14482284e-01
8.24827850e-01 4.39810872e-01 9.29180920e-01 -1.66199028e+00
-2.42930010e-01 -1.84099466e-01 -7.42028713e-01 -7.32394457e-01
3.02132815e-01 -7.50218272e-01 1.54359385e-01 -9.28661585e-01
5.42064190e-01 -9.42079544e-01 -4.62047756e-01 5.25702894e-01
-3.96776497e-01 4.75899190e-01 1.44006580e-01 2.75714040e-01
-6.63212419e-01 3.07645917e-01 9.21774030e-01 -2.60482877e-01
5.14503941e-02 2.16138795e-01 -9.87035871e-01 6.79812610e-01
5.49362898e-01 -8.10126066e-01 -1.99431404e-01 -1.19474150e-01
1.29643530e-01 -1.22966155e-01 8.37752759e-01 -8.48020852e-01
-2.67297924e-01 -3.06640267e-01 6.22266829e-01 -3.73433352e-01
1.67487308e-01 -9.48737323e-01 -1.94196284e-01 2.85275906e-01
-6.24640703e-01 -3.07560354e-01 -5.51235070e-03 9.93469238e-01
-5.78783303e-02 -2.21265435e-01 7.97678351e-01 -1.44787766e-02
-8.37426662e-01 2.81942993e-01 -1.19883336e-01 3.83933634e-01
1.24737155e+00 -5.46727896e-01 -6.13327503e-01 -3.40612561e-01
-6.51159465e-01 1.37652770e-01 7.45131671e-01 6.37095571e-01
3.47148597e-01 -1.07834780e+00 -7.39847660e-01 3.61248314e-01
8.58086705e-01 -1.26928642e-01 -1.26776591e-01 4.44335431e-01
1.35335729e-01 5.19144870e-02 -8.12927783e-02 -9.72024739e-01
-1.45117736e+00 7.31970549e-01 2.76674449e-01 1.53728709e-01
-5.38249433e-01 6.52514338e-01 5.16716182e-01 -5.78952730e-01
3.73337895e-01 -8.14449564e-02 -9.30225477e-02 1.24588631e-01
4.74520385e-01 9.59504619e-02 5.74876517e-02 -6.69369936e-01
-4.14102912e-01 4.05683309e-01 -1.96208879e-01 3.62626016e-01
1.12511480e+00 -8.52483660e-02 2.32563570e-01 8.00413251e-01
1.03395450e+00 5.66034578e-02 -1.34772444e+00 -3.43054563e-01
2.11834058e-01 -8.10113788e-01 -3.05573612e-01 -9.33492899e-01
-7.62700975e-01 7.65011370e-01 7.42840886e-01 2.95693189e-01
7.32231200e-01 5.47214806e-01 2.22726017e-01 3.59921992e-01
2.12669641e-01 -7.59750664e-01 1.51139542e-01 2.15536863e-01
5.98546922e-01 -1.82961714e+00 -3.10223680e-02 -7.81498373e-01
-7.04405069e-01 1.20999432e+00 8.76640379e-01 1.05871372e-01
3.15354228e-01 2.03207389e-01 3.10892463e-01 4.94500855e-03
-8.56659889e-01 -2.91434973e-01 3.60207140e-01 9.23940718e-01
2.45778903e-01 2.07723618e-01 2.80801028e-01 6.03637695e-01
-2.13082954e-02 -4.39253122e-01 8.58377397e-01 5.62451959e-01
-2.06959963e-01 -9.89501417e-01 -8.05159569e-01 6.61528826e-01
-3.98428172e-01 1.63222104e-01 -8.40060174e-01 8.00840974e-01
7.40477210e-03 7.39153862e-01 4.07044083e-01 -2.07904115e-01
1.67281866e-01 5.54773584e-02 3.95554453e-01 -7.82937109e-01
-8.09269696e-02 -3.95086348e-01 5.84911294e-02 -3.10657680e-01
-2.85993427e-01 -8.39940965e-01 -9.47442710e-01 1.55099541e-01
-4.49737519e-01 -2.35920087e-01 3.35892916e-01 8.63764882e-01
2.42138430e-01 2.27741867e-01 7.40781903e-01 -3.35265726e-01
-8.74591589e-01 -6.68634772e-01 -6.38574958e-01 8.56616139e-01
6.91929698e-01 -1.00672889e+00 -7.90099025e-01 -5.29490504e-03] | [9.71666431427002, 2.8500876426696777] |
d4490687-93e2-4f24-a492-b704adf0ec5f | zero-shot-personalized-speech-enhancement | 2105.03542 | null | https://arxiv.org/abs/2105.03542v1 | https://arxiv.org/pdf/2105.03542v1.pdf | Zero-Shot Personalized Speech Enhancement through Speaker-Informed Model Selection | This paper presents a novel zero-shot learning approach towards personalized speech enhancement through the use of a sparsely active ensemble model. Optimizing speech denoising systems towards a particular test-time speaker can improve performance and reduce run-time complexity. However, test-time model adaptation may be challenging if collecting data from the test-time speaker is not possible. To this end, we propose using an ensemble model wherein each specialist module denoises noisy utterances from a distinct partition of training set speakers. The gating module inexpensively estimates test-time speaker characteristics in the form of an embedding vector and selects the most appropriate specialist module for denoising the test signal. Grouping the training set speakers into non-overlapping semantically similar groups is non-trivial and ill-defined. To do this, we first train a Siamese network using noisy speech pairs to maximize or minimize the similarity of its output vectors depending on whether the utterances derive from the same speaker or not. Next, we perform k-means clustering on the latent space formed by the averaged embedding vectors per training set speaker. In this way, we designate speaker groups and train specialist modules optimized around partitions of the complete training set. Our experiments show that ensemble models made up of low-capacity specialists can outperform high-capacity generalist models with greater efficiency and improved adaptation towards unseen test-time speakers. | ['Minje Kim', 'Aswin Sivaraman'] | 2021-05-08 | null | null | null | null | ['speech-denoising'] | ['speech'] | [ 1.54123649e-01 1.82097375e-01 1.74952269e-01 -4.89607960e-01
-1.25218678e+00 -4.14303571e-01 1.41605362e-01 -1.47722647e-01
-3.84402305e-01 2.74208695e-01 6.18445694e-01 4.13370207e-02
-2.15971276e-01 -4.76792693e-01 -3.17824453e-01 -1.00542092e+00
2.46346984e-02 7.26162255e-01 -1.10288702e-01 -2.57803142e-01
-1.77393869e-01 2.68751919e-01 -1.76922977e+00 3.40761095e-01
1.02178943e+00 6.06622159e-01 5.00885606e-01 9.10351932e-01
-5.47174476e-02 2.03744456e-01 -9.35105085e-01 -1.48104623e-01
6.54117614e-02 -8.26786697e-01 -4.60499346e-01 5.58849096e-01
9.69898999e-02 5.24017066e-02 -1.68877840e-01 9.08070683e-01
1.13054168e+00 7.46730149e-01 4.39373612e-01 -8.09060037e-01
-2.54359096e-01 7.79964983e-01 8.05412829e-02 2.42190659e-01
4.68387920e-03 -2.00697202e-02 8.37485969e-01 -9.89740968e-01
2.72144258e-01 1.15160310e+00 4.80970919e-01 9.69590724e-01
-1.35831261e+00 -5.45506477e-01 2.38159776e-01 2.13651597e-01
-1.30578315e+00 -1.21488333e+00 8.52266669e-01 -5.51624335e-02
9.41597879e-01 5.95691383e-01 5.44929028e-01 1.05819702e+00
-4.23368007e-01 8.30895245e-01 4.83848125e-01 -3.46631110e-01
6.71795428e-01 4.84026670e-01 5.65836988e-02 5.44633925e-01
-4.83254671e-01 -2.20986426e-01 -8.99143100e-01 -3.19955558e-01
1.03731751e-01 -2.92360663e-01 -5.10847330e-01 -3.31754327e-01
-8.33967924e-01 9.04011428e-01 -8.76632612e-03 4.83764648e-01
-4.61701989e-01 -3.02920073e-01 3.12242538e-01 4.86429691e-01
7.57698536e-01 4.03044224e-01 -3.95753205e-01 -1.54431298e-01
-1.43848836e+00 -3.08304757e-01 1.00128365e+00 5.35954356e-01
8.02369475e-01 4.94902432e-01 -1.12662472e-01 1.22475517e+00
4.20749098e-01 1.44370228e-01 8.92236531e-01 -9.92841601e-01
2.91985512e-01 2.20337361e-01 -2.35957086e-01 -3.60180974e-01
5.79196122e-03 -7.78824329e-01 -7.52391160e-01 3.95217463e-02
-3.60178873e-02 -3.82005960e-01 -1.10076320e+00 1.85124707e+00
4.55995470e-01 5.50946951e-01 4.36454594e-01 7.06315219e-01
6.88803017e-01 8.57905090e-01 5.92641421e-02 -5.18559337e-01
1.06862485e+00 -9.73839402e-01 -5.63154638e-01 -4.59832430e-01
5.63662827e-01 -7.15433776e-01 7.42997706e-01 3.20510209e-01
-1.26219630e+00 -6.31724596e-01 -9.99149382e-01 4.00218368e-01
-1.95693791e-01 2.16375645e-02 6.18895628e-02 9.67738688e-01
-1.21805584e+00 4.05282080e-01 -8.39019060e-01 -2.28218794e-01
2.40933895e-01 4.77379650e-01 -2.67544419e-01 5.73161021e-02
-1.11140120e+00 6.84843242e-01 8.65062773e-02 -1.46290008e-02
-1.12219918e+00 -7.37642109e-01 -1.01594865e+00 5.02418876e-01
1.92013815e-01 -8.32587242e-01 1.12925875e+00 -1.18746686e+00
-1.77943110e+00 5.57614505e-01 -5.34603953e-01 -5.18652618e-01
1.07413687e-01 1.64459884e-01 -5.53360343e-01 7.73072019e-02
-3.69012654e-02 3.83244187e-01 1.25398386e+00 -1.13985848e+00
-3.35710526e-01 -4.33563560e-01 -5.27048349e-01 5.15350580e-01
-6.94527268e-01 5.00197597e-02 -4.77960050e-01 -5.65474093e-01
3.98025066e-01 -5.84786594e-01 -2.80725926e-01 -4.14802045e-01
-1.98165745e-01 -2.65998244e-01 9.24784660e-01 -7.38298774e-01
1.30607426e+00 -2.72656322e+00 4.47916180e-01 4.49365467e-01
4.63077351e-02 2.49741226e-01 -2.37585574e-01 2.20770538e-01
-2.17771053e-01 -1.69569820e-01 -3.90966117e-01 -8.21469605e-01
4.78567369e-02 3.58928293e-01 -1.70573503e-01 3.10749084e-01
1.53328717e-01 2.75081068e-01 -7.22305655e-01 -4.11996931e-01
3.18626523e-01 7.52034843e-01 -7.30086625e-01 4.02875066e-01
1.24231139e-02 4.36054856e-01 -2.06649601e-01 4.09831136e-01
2.95741349e-01 1.29597529e-03 1.89851865e-01 1.01060411e-02
1.88061535e-01 4.15014595e-01 -1.44172442e+00 1.69459033e+00
-6.16072893e-01 4.99712467e-01 6.06682360e-01 -1.18497062e+00
9.69953418e-01 6.88769639e-01 5.65264225e-01 -2.16873616e-01
-1.24229401e-01 -1.32966535e-02 3.83262075e-02 -4.82998580e-01
2.27832645e-01 -3.70813996e-01 -1.17316991e-02 5.80526292e-01
4.61021900e-01 -1.03324719e-01 1.08484983e-01 8.21077451e-02
1.11528611e+00 -4.67945188e-01 1.37665495e-01 -5.67250280e-03
5.35223186e-01 -4.94239360e-01 5.89202046e-01 6.10551298e-01
-4.76755977e-01 5.87470293e-01 -9.82155465e-03 2.16213852e-01
-8.37471426e-01 -1.24225676e+00 -9.86579508e-02 1.65147054e+00
-3.62518370e-01 -3.03156227e-01 -9.64210987e-01 -4.86328989e-01
-2.78555036e-01 9.15546000e-01 -2.68744588e-01 -4.52776313e-01
-5.17064989e-01 -6.94362819e-01 3.68522584e-01 2.71757215e-01
2.24045202e-01 -9.45720851e-01 -1.63436949e-01 4.78206009e-01
-3.01298380e-01 -5.26887774e-01 -7.13707805e-01 6.23876929e-01
-6.13164723e-01 -3.56542498e-01 -7.83078432e-01 -1.08392322e+00
7.00952590e-01 2.31042668e-01 9.55704212e-01 -2.37105325e-01
1.39318675e-01 5.84572196e-01 -1.91300988e-01 -2.19638929e-01
-5.95500767e-01 4.26067039e-02 2.88026482e-01 4.75826949e-01
4.74903673e-01 -5.30399382e-01 -3.52601767e-01 2.42475078e-01
-6.81080222e-01 -4.45208162e-01 2.30646849e-01 9.68873203e-01
5.28559566e-01 2.49789596e-01 7.75997877e-01 -4.04866308e-01
6.89941168e-01 -6.14699662e-01 -4.78961430e-02 2.91474819e-01
-2.75031716e-01 1.78146899e-01 4.27652329e-01 -6.92102611e-01
-1.14003432e+00 3.37016732e-01 -2.95754284e-01 -4.97326285e-01
-1.34756491e-01 3.89157414e-01 -5.04360616e-01 2.89688170e-01
7.62696207e-01 5.72705388e-01 1.00888811e-01 -4.17646617e-01
2.88696766e-01 1.04017186e+00 5.18231690e-01 -2.09800005e-01
5.05411983e-01 2.07126871e-01 -7.87009537e-01 -1.25178647e+00
-5.64144194e-01 -8.09843779e-01 -3.08865756e-01 -3.98376793e-01
8.03072095e-01 -1.04183173e+00 -1.71781033e-01 6.41412064e-02
-8.10354769e-01 -1.99231341e-01 -6.17342949e-01 6.72220707e-01
-5.34968615e-01 3.66902471e-01 -2.97222316e-01 -1.03508985e+00
-4.56432462e-01 -1.40596306e+00 1.06422138e+00 7.16760457e-02
-5.61383963e-01 -9.28153396e-01 1.78947955e-01 5.03446639e-01
4.41065818e-01 -6.05458856e-01 5.42069614e-01 -1.05878282e+00
-3.15295994e-01 -1.68166950e-01 5.92840374e-01 7.06913471e-01
3.34694907e-02 -2.52056569e-01 -1.40006423e+00 -3.49038869e-01
5.49597859e-01 -4.06552553e-02 1.03260744e+00 6.31694734e-01
1.08126032e+00 -1.16261423e-01 -1.62784874e-01 5.52873075e-01
8.29875648e-01 4.47462201e-02 3.90280545e-01 1.53976521e-02
3.80147278e-01 7.56345153e-01 9.49800164e-02 4.04348135e-01
1.15496092e-01 3.81643057e-01 -3.98791172e-02 1.29289165e-01
-9.77208167e-02 6.96241707e-02 8.22519362e-01 1.37399495e+00
3.98277402e-01 -4.33185637e-01 -6.29311264e-01 7.49780893e-01
-1.43210745e+00 -1.23539114e+00 4.69357610e-01 2.25192451e+00
8.41417670e-01 -2.97801066e-02 1.19006641e-01 3.63735080e-01
8.64983141e-01 1.76897898e-01 -4.88294572e-01 -3.80698204e-01
-1.88317031e-01 3.93354297e-01 -5.47926426e-02 6.88679814e-01
-1.01824117e+00 6.45187616e-01 5.71692562e+00 8.08449328e-01
-1.06317341e+00 3.63771915e-01 5.59727252e-01 -6.22360349e-01
-3.47284853e-01 -2.32124478e-01 -7.78164983e-01 5.61289608e-01
1.51432383e+00 -2.51675069e-01 5.03394186e-01 7.37882972e-01
4.23677593e-01 3.26453030e-01 -9.93572295e-01 9.34500456e-01
4.78256971e-01 -9.14214253e-01 -3.54267806e-01 4.59624873e-03
6.52624309e-01 1.71350390e-01 2.56381989e-01 4.91124511e-01
2.04160109e-01 -7.01758265e-01 4.92003471e-01 5.29903829e-01
2.75517344e-01 -8.53702545e-01 3.27426732e-01 6.19834006e-01
-9.58426058e-01 -2.36170381e-01 -2.99102455e-01 4.83148724e-01
1.93416595e-01 5.38677454e-01 -9.82266784e-01 7.55822212e-02
6.57986224e-01 2.86852241e-01 -1.65672421e-01 1.02587318e+00
6.42567268e-03 9.12034094e-01 -5.14142275e-01 1.02233119e-01
7.72631243e-02 -3.84141728e-02 8.27866435e-01 1.27018106e+00
6.18257344e-01 7.30953291e-02 8.16333443e-02 4.45762902e-01
-1.30276605e-01 1.02651201e-01 -4.31138963e-01 1.12937413e-01
5.31789720e-01 1.09152937e+00 -3.26082140e-01 -5.57177305e-01
-2.08307326e-01 1.16154623e+00 1.70641035e-01 5.60553670e-01
-4.89139497e-01 -1.78673565e-01 9.19589221e-01 -1.49285838e-01
6.31668627e-01 8.67836550e-03 -2.00620279e-01 -1.14769161e+00
-1.00938700e-01 -9.34315026e-01 4.61107224e-01 -6.42155349e-01
-1.23269022e+00 7.93926656e-01 -4.42838252e-01 -1.17703152e+00
-6.58548355e-01 -4.51716408e-02 -7.42695034e-01 1.05387151e+00
-9.93687093e-01 -5.67289770e-01 6.85329288e-02 5.59599936e-01
9.67526793e-01 -5.56367815e-01 1.17634034e+00 1.81109384e-01
-7.89091825e-01 7.80617952e-01 4.67944860e-01 -1.12993546e-01
7.05247700e-01 -1.09774554e+00 1.31670147e-01 8.75720084e-01
3.54636282e-01 5.64004004e-01 7.69380867e-01 -4.21795458e-01
-1.04539645e+00 -1.13865340e+00 1.18349361e+00 -3.09455186e-01
4.32178974e-01 -3.71205479e-01 -1.15530074e+00 3.94802660e-01
3.22903931e-01 -3.63824427e-01 1.23835254e+00 2.34031454e-01
-5.96073829e-02 -1.84538215e-01 -1.09370267e+00 4.74510401e-01
6.51916862e-01 -8.46714437e-01 -7.17788696e-01 3.20384681e-01
5.27612925e-01 5.20102903e-02 -7.95925498e-01 -2.46982072e-02
6.39937446e-02 -7.55466163e-01 9.41884041e-01 -4.61910576e-01
-1.67294979e-01 -1.48846462e-01 -9.98398960e-02 -1.68539178e+00
-3.90659660e-01 -6.90913081e-01 -1.70183167e-01 1.34850788e+00
5.32916307e-01 -3.23078901e-01 9.62106287e-01 5.03149986e-01
-5.64382553e-01 -5.90621591e-01 -1.26128328e+00 -4.88173127e-01
-3.79136115e-01 -6.12808466e-01 5.18162310e-01 8.33316863e-01
9.51007456e-02 5.72980165e-01 -1.82961479e-01 3.99752617e-01
6.79774702e-01 -2.35628843e-01 4.90803719e-01 -1.04883516e+00
-6.01344764e-01 -4.21020389e-01 -3.53990465e-01 -9.16006982e-01
4.33481187e-01 -1.01926923e+00 3.49965543e-01 -1.27779531e+00
-1.23842783e-01 -1.42913759e-01 -3.87948662e-01 2.36480534e-01
-2.39825979e-01 -5.14125451e-02 -1.39165312e-01 -2.77566370e-02
-5.54040611e-01 7.33987510e-01 7.38659501e-01 -2.66249478e-01
-5.76794147e-01 4.15025920e-01 -7.00311482e-01 5.23448110e-01
8.23908687e-01 -4.42115337e-01 -6.14758313e-01 -4.02981758e-01
-3.37004423e-01 1.39589995e-01 4.96168770e-02 -9.68569398e-01
5.12279332e-01 2.67906904e-01 3.42288464e-01 -4.38743532e-01
8.14090848e-01 -6.80654466e-01 1.15612425e-01 2.38187358e-01
-3.55116308e-01 -4.64799821e-01 7.58251920e-02 6.36770904e-01
-3.26811343e-01 -5.00311673e-01 9.28643048e-01 -1.05445392e-01
-5.49200296e-01 8.03542957e-02 -7.14400113e-01 -2.49268636e-01
8.95674706e-01 -3.38411689e-01 2.81720787e-01 -5.92186332e-01
-1.31811547e+00 1.18950441e-01 2.78905760e-02 1.89450845e-01
7.53042459e-01 -1.07329702e+00 -9.59083080e-01 7.20491827e-01
-2.74586142e-04 -2.36820877e-01 5.60764849e-01 5.60192049e-01
2.30742380e-01 6.17905743e-02 3.30964446e-01 -6.37597024e-01
-1.47941363e+00 2.64053732e-01 4.69591111e-01 1.38063338e-02
-1.19575500e-01 1.43224800e+00 7.80459195e-02 -5.97287297e-01
4.30539995e-01 2.44378075e-02 -2.69195735e-01 4.13989216e-01
5.54261982e-01 3.41301829e-01 3.10146749e-01 -8.50024283e-01
-2.22242236e-01 8.85396227e-02 -8.15932304e-02 -3.89233023e-01
1.52480447e+00 -3.30436587e-01 1.48688301e-01 4.00757223e-01
1.35718536e+00 -1.50751574e-02 -1.17017794e+00 -3.84672791e-01
-2.79981285e-01 -7.04347566e-02 3.18919569e-01 -5.43627620e-01
-9.92906570e-01 6.77471995e-01 8.03915203e-01 2.70430833e-01
1.33610320e+00 3.27896416e-01 6.22338474e-01 1.93846256e-01
-1.59320146e-01 -1.34030950e+00 1.67571614e-03 4.18091148e-01
5.47051787e-01 -1.13137221e+00 -4.07037169e-01 6.22494917e-05
-6.89363480e-01 7.67980456e-01 3.92558277e-01 1.48372531e-01
8.35141599e-01 1.30819187e-01 2.09235296e-01 -6.89476281e-02
-9.74035144e-01 -3.41264367e-01 4.62101519e-01 6.19699359e-01
4.13742185e-01 1.97939172e-01 2.07062617e-01 6.08911872e-01
-5.41744471e-01 -6.09896302e-01 2.84027513e-02 6.44691408e-01
-8.84889066e-01 -1.04013824e+00 -5.98461509e-01 4.63913202e-01
-1.29002005e-01 -1.60106793e-01 -2.21348286e-01 -1.30359065e-02
8.81496966e-02 1.30234241e+00 1.96456596e-01 -4.20993805e-01
3.56706411e-01 6.77866876e-01 2.04272732e-01 -9.65459824e-01
-8.32536638e-01 5.97189426e-01 1.02476954e-01 -7.25063905e-02
-1.81211397e-01 -8.81618798e-01 -9.90350723e-01 -9.12746564e-02
-3.85114461e-01 4.71287519e-01 8.15311015e-01 9.70843732e-01
3.90651882e-01 6.09600902e-01 8.51747513e-01 -7.78409064e-01
-8.94122422e-01 -1.05428183e+00 -5.15326977e-01 2.54644871e-01
4.71812546e-01 -2.04631805e-01 -8.10423553e-01 2.31258869e-01] | [14.608619689941406, 6.166618824005127] |
3ebd2b95-fcaa-4d83-832c-dfee0e6b023c | explainable-machine-learning-for-hydrocarbon | 2212.07563 | null | https://arxiv.org/abs/2212.07563v1 | https://arxiv.org/pdf/2212.07563v1.pdf | Explainable Machine Learning for Hydrocarbon Prospect Risking | Hydrocarbon prospect risking is a critical application in geophysics predicting well outcomes from a variety of data including geological, geophysical, and other information modalities. Traditional routines require interpreters to go through a long process to arrive at the probability of success of specific outcomes. AI has the capability to automate the process but its adoption has been limited thus far owing to a lack of transparency in the way complicated, black box models generate decisions. We demonstrate how LIME -- a model-agnostic explanation technique -- can be used to inject trust in model decisions by uncovering the model's reasoning process for individual predictions. It generates these explanations by fitting interpretable models in the local neighborhood of specific datapoints being queried. On a dataset of well outcomes and corresponding geophysical attribute data, we show how LIME can induce trust in model's decisions by revealing the decision-making process to be aligned to domain knowledge. Further, it has the potential to debug mispredictions made due to anomalous patterns in the data or faulty training datasets. | ['Ghassan AlRegib', 'Ahmad Mustafa'] | 2022-12-15 | null | null | null | null | ['geophysics'] | ['miscellaneous'] | [ 1.45175263e-01 8.88688564e-01 -1.50148928e-01 -8.28921318e-01
-6.27466679e-01 -4.72870171e-01 8.07119548e-01 6.26996100e-01
1.97620362e-01 6.13272071e-01 4.37074900e-01 -1.18279719e+00
-2.58017004e-01 -9.99639273e-01 -8.95607829e-01 -3.26324761e-01
-3.33207101e-01 7.88703024e-01 -5.96384471e-03 -1.96138978e-01
8.06699336e-01 5.28637171e-01 -1.16270363e+00 3.79061073e-01
7.66082764e-01 7.05614090e-01 -2.28708565e-01 5.93351483e-01
6.79159388e-02 1.16275001e+00 -2.29386985e-01 -1.84864238e-01
1.44064710e-01 -6.35102540e-02 -8.73735905e-01 -2.51189083e-01
-2.87793785e-01 -3.21260720e-01 -2.90393806e-03 9.76281583e-01
-2.13775888e-01 -2.37449467e-01 7.94184923e-01 -1.28658259e+00
-4.29368705e-01 7.26353109e-01 -1.24510676e-01 2.49483868e-01
3.85787457e-01 7.83744872e-01 9.88222897e-01 -7.56419361e-01
3.59618664e-01 1.27923763e+00 7.58943677e-01 3.33624810e-01
-1.28483033e+00 -4.34842378e-01 1.44864365e-01 7.63703436e-02
-8.99156868e-01 -5.16952574e-01 3.74600291e-01 -9.09711897e-01
1.20298469e+00 3.91187757e-01 6.74176931e-01 9.52251434e-01
5.53588748e-01 2.94487387e-01 1.21969616e+00 -2.67545462e-01
7.38905013e-01 1.68484211e-01 2.05403522e-01 7.34360933e-01
5.60879767e-01 6.86634123e-01 -8.29638839e-01 -6.23071253e-01
5.35384417e-01 2.51325574e-02 -1.57202765e-01 2.19271481e-01
-1.05446064e+00 8.09660792e-01 4.44295466e-01 -2.63512671e-01
-7.07050681e-01 3.61169159e-01 4.39220779e-02 3.50387633e-01
4.74079341e-01 1.22187269e+00 -6.88102543e-01 1.83493321e-04
-6.20544791e-01 5.52223444e-01 1.05813265e+00 5.54244697e-01
8.08454573e-01 2.96865463e-01 3.14129412e-01 -1.33633986e-01
8.64207745e-01 2.79312074e-01 2.46316060e-01 -8.66798520e-01
2.72630811e-01 7.15076208e-01 4.80743349e-01 -1.00277734e+00
-2.61210561e-01 -2.27527227e-02 -1.61453396e-01 7.87942648e-01
3.28530490e-01 -1.75989047e-01 -1.02570736e+00 1.18854821e+00
-6.83674812e-02 1.19094267e-01 6.49281681e-01 9.55832958e-01
3.49204332e-01 6.13860428e-01 3.09216440e-01 4.04321343e-01
9.19187129e-01 -3.17439675e-01 -1.77098453e-01 -8.76675963e-01
6.00340903e-01 -1.33202031e-01 1.05677462e+00 3.10123354e-01
-9.17328596e-01 3.52430493e-02 -1.16327381e+00 3.74372214e-01
-2.61235803e-01 -6.62151515e-01 9.50520694e-01 1.89616337e-01
-8.43931913e-01 9.70127523e-01 -1.02739596e+00 -1.52251944e-01
4.07757759e-01 1.69748217e-01 -2.56820887e-01 1.82655305e-01
-1.11628711e+00 1.16785491e+00 3.96558136e-01 2.35664263e-01
-1.24631631e+00 -8.29291165e-01 -8.49685967e-01 2.77105421e-01
3.02032143e-01 -5.99418581e-01 1.38825190e+00 -6.82907760e-01
-9.37431097e-01 5.44155478e-01 -2.98653990e-01 -6.26153409e-01
6.19034827e-01 -9.83247459e-02 -5.67863464e-01 -1.81605369e-01
2.68730998e-01 3.26323509e-01 6.31719410e-01 -1.26203215e+00
-4.82431471e-01 -4.36499834e-01 -1.01025723e-01 -1.11874126e-01
3.06251049e-01 -1.71985537e-01 3.73046547e-01 -2.15382993e-01
6.13586307e-01 -6.29057527e-01 -7.65813887e-01 -1.23743057e-01
-5.26151836e-01 1.11421920e-01 3.95377845e-01 -7.63414383e-01
9.68565643e-01 -1.97115374e+00 -3.68420482e-01 7.80455649e-01
2.98864990e-01 -2.82278895e-01 4.85146850e-01 5.69931388e-01
-1.41543314e-01 7.62482285e-01 -5.96981823e-01 1.01437226e-01
1.21897884e-01 5.09958208e-01 -8.96647632e-01 3.03552538e-01
6.02558196e-01 7.03312993e-01 -1.03600395e+00 1.36593925e-02
9.44905914e-03 -2.30098486e-01 -4.48988914e-01 3.43143880e-01
-8.33401680e-01 6.05564952e-01 -7.81582117e-01 8.83051217e-01
2.23900259e-01 -3.68691325e-01 2.63131946e-01 5.80809116e-01
-1.98191151e-01 7.57900238e-01 -7.16823399e-01 1.05018818e+00
-1.10669039e-01 5.14621854e-01 -2.73973137e-01 -8.18222821e-01
1.04058564e+00 4.79149014e-01 -2.12519690e-01 -3.68890613e-01
-3.97091776e-01 5.19980788e-01 1.43146291e-01 -9.04904604e-01
5.17148197e-01 -5.51513314e-01 -1.10454485e-01 8.39395344e-01
-7.38472104e-01 -4.71567690e-01 -5.95197558e-01 8.99036005e-02
1.30392587e+00 -4.83528525e-03 3.55354190e-01 -3.71086329e-01
-1.85180962e-01 9.19781744e-01 6.52007639e-01 9.16561246e-01
3.01270843e-01 4.24975812e-01 7.47337699e-01 -9.30012763e-01
-1.22920859e+00 -7.39087462e-01 -1.68961331e-01 4.52653617e-01
-9.73924547e-02 -1.27413109e-01 -4.22081649e-02 -5.61538160e-01
3.99328411e-01 1.51156437e+00 -7.45791912e-01 -3.34096521e-01
-8.82291123e-02 -6.96554661e-01 5.40118396e-01 7.51326859e-01
1.52648836e-01 -1.02114785e+00 -9.95975018e-01 4.38353360e-01
3.02737027e-01 -5.79299510e-01 5.30104816e-01 3.87514502e-01
-1.20839465e+00 -1.02626443e+00 3.53794456e-01 1.44765660e-01
9.86655354e-01 -2.54608929e-01 1.23122382e+00 4.55522567e-01
1.13071032e-01 1.55010402e-01 -1.07863367e-01 -7.66160131e-01
-1.04419637e+00 -4.26340371e-01 -8.15264955e-02 -3.82921040e-01
4.69623268e-01 -5.61964989e-01 -3.02681714e-01 1.72090113e-01
-7.53929496e-01 2.23153323e-01 3.86687785e-01 6.44313276e-01
6.38427213e-02 -2.31292583e-02 4.80911046e-01 -1.24386191e+00
6.21374369e-01 -1.16190362e+00 -6.37879133e-01 3.80707502e-01
-1.28732169e+00 4.34421897e-01 3.65760267e-01 4.79706638e-02
-1.10407686e+00 -3.23460549e-01 3.19441676e-01 -6.99496418e-02
-3.27397019e-01 1.27968264e+00 1.13545761e-01 3.92111421e-01
8.83099556e-01 -1.07724823e-01 -2.60648411e-02 -3.73274803e-01
-1.01224124e-01 3.76130432e-01 7.47087181e-01 -9.41941321e-01
7.41391242e-01 3.27347487e-01 5.39338170e-03 -1.08030252e-01
-6.97857559e-01 1.41813323e-01 -2.70390481e-01 -2.58430928e-01
5.07810652e-01 -1.02036119e+00 -4.53339159e-01 6.04623221e-02
-1.00068057e+00 -6.44949853e-01 -3.37066501e-01 4.36869353e-01
-2.17520550e-01 -1.20836288e-01 -1.69698566e-01 -1.02688742e+00
2.22220290e-02 -1.01808286e+00 5.75060964e-01 2.81178504e-01
-9.70157325e-01 -1.15807438e+00 -4.31995764e-02 4.42100829e-03
4.10715967e-01 4.82159883e-01 1.38458920e+00 -1.13848090e+00
-8.66822064e-01 -3.66102368e-01 -6.44272268e-02 -2.16089755e-01
-1.98779583e-01 2.45569587e-01 -1.18734622e+00 2.36155674e-01
5.24681173e-02 -1.79406300e-01 5.22430539e-01 5.11722919e-03
1.06386936e+00 -7.59591281e-01 -3.46210539e-01 3.46855879e-01
1.12435901e+00 3.21823478e-01 5.64908922e-01 7.32966006e-01
2.43195742e-01 9.48281467e-01 6.20837331e-01 5.70434332e-01
4.11809295e-01 3.97986136e-02 6.44852340e-01 -1.80510446e-01
6.40619218e-01 -4.97335047e-01 3.33190322e-01 -1.14545792e-01
-9.19347703e-02 -3.36149298e-02 -1.75072312e+00 5.82699001e-01
-2.17633796e+00 -9.59130108e-01 -3.29906821e-01 2.16764021e+00
4.90234375e-01 4.64364022e-01 -5.75955212e-01 -4.74221595e-02
1.41293645e-01 4.08538841e-02 -9.99463379e-01 -6.09181702e-01
1.85632277e-02 -7.31237009e-02 6.79580569e-01 8.23385894e-01
-4.54133153e-01 7.83303201e-01 7.01410389e+00 -1.54205963e-01
-1.00485802e+00 -1.71230912e-01 8.58157933e-01 3.44179094e-01
-1.19802535e+00 7.71886408e-01 -2.90060163e-01 2.31695071e-01
1.21905279e+00 -3.59343231e-01 2.57646978e-01 8.73125315e-01
5.23590744e-01 -5.31192780e-01 -1.54338348e+00 2.80399360e-02
-3.26918453e-01 -1.70603156e+00 -9.88279432e-02 2.03902900e-01
5.35045505e-01 -7.02494681e-02 -5.15962020e-02 -1.17845424e-01
1.17158258e+00 -1.51369846e+00 1.11602020e+00 9.98645127e-01
4.67970341e-01 -4.31630462e-01 6.23041749e-01 4.18546736e-01
-4.17718083e-01 -4.04192358e-01 -1.82573944e-01 -7.68961787e-01
-6.23302385e-02 5.79876900e-01 -1.70875239e+00 2.76761621e-01
5.25755763e-01 6.79652750e-01 -4.75654423e-01 5.96949935e-01
-5.58089793e-01 1.09045947e+00 -3.07242483e-01 2.56046474e-01
2.86177486e-01 -1.11822873e-01 5.13119698e-01 7.65771210e-01
4.72107708e-01 3.73362064e-01 -3.76743800e-03 1.26809108e+00
4.39781219e-01 -7.13060379e-01 -8.94320667e-01 -1.60200968e-01
6.55103266e-01 5.34416795e-01 -3.43113869e-01 -5.86459875e-01
-4.05031554e-02 3.47498745e-01 9.48362947e-02 4.63429838e-01
-2.96087325e-01 1.96605414e-01 8.01438034e-01 3.26513857e-01
-3.05194348e-01 -2.10938051e-01 -1.10588861e+00 -8.85394692e-01
-4.23108004e-02 -9.94739711e-01 3.25209320e-01 -1.30901897e+00
-1.13845766e+00 2.57655650e-01 6.42712936e-02 -8.04305255e-01
-5.92089951e-01 -4.34173614e-01 -1.36465025e+00 1.12355483e+00
-1.50148106e+00 -7.88737416e-01 -1.14008307e-01 -8.93441122e-03
3.12077284e-01 -2.01847717e-01 9.62253273e-01 -7.19875216e-01
-3.87846380e-01 -2.77341515e-01 -2.75312096e-01 -1.54221669e-01
3.71816248e-01 -1.34744334e+00 9.88193393e-01 1.04812419e+00
-3.20055574e-01 7.20673800e-01 1.19634140e+00 -1.10484004e+00
-1.26250255e+00 -9.04617608e-01 1.18223703e+00 -7.65056193e-01
9.63802934e-01 1.70074895e-01 -1.38181984e+00 1.12649930e+00
-1.47812292e-01 -3.15565497e-01 6.95589721e-01 1.53872013e-01
-3.00103307e-01 1.45251811e-01 -1.22204101e+00 6.11070037e-01
6.32676601e-01 -6.03379071e-01 -9.53879178e-01 3.03124011e-01
4.35956299e-01 -3.26370567e-01 -8.91368508e-01 3.29955757e-01
3.79548430e-01 -1.01646674e+00 5.40731430e-01 -1.23948741e+00
8.79606962e-01 -5.09026408e-01 -2.62657553e-01 -1.27900422e+00
-1.80957451e-01 -6.17675841e-01 2.55425125e-01 9.74769711e-01
9.68973696e-01 -1.08698952e+00 5.67932785e-01 1.74351883e+00
-3.66970003e-01 -4.82660472e-01 -5.78976274e-01 -1.05865769e-01
-6.03495203e-02 -7.99051225e-01 1.30726874e+00 1.06813633e+00
2.97625989e-01 -2.13360906e-01 -5.89352287e-02 1.10909796e+00
5.81667900e-01 2.53290087e-01 7.41223216e-01 -1.27490377e+00
-3.46977770e-01 -1.76582336e-01 -1.65085092e-01 -4.21715856e-01
-4.73294742e-02 -7.42374420e-01 2.15975791e-01 -1.42410231e+00
-1.15523942e-01 -1.03284204e+00 -3.72986756e-02 9.88142371e-01
-3.02979201e-01 -5.28707027e-01 -5.46006821e-02 6.46826804e-01
1.53204173e-01 1.47834927e-01 7.18604028e-01 2.94992868e-02
8.20959061e-02 1.38415564e-02 -1.02318025e+00 1.06210959e+00
8.48544300e-01 -7.98206210e-01 -2.22551122e-01 -5.72531760e-01
7.08046138e-01 5.21138489e-01 9.83912587e-01 -8.60572755e-01
3.33479404e-01 -7.16666758e-01 4.92597103e-01 -1.59333408e-01
4.46936414e-02 -7.20544398e-01 7.20114887e-01 7.27239847e-01
-7.38379359e-01 4.58523035e-01 1.51317134e-01 7.20619321e-01
-1.11438178e-01 -2.94524193e-01 1.94560498e-01 -5.16746879e-01
-9.73166287e-01 -4.51421402e-02 -7.07541585e-01 -1.98709399e-01
8.54468226e-01 -1.43077374e-01 -4.81711566e-01 -2.87925512e-01
-7.35164225e-01 6.45280004e-01 8.47794950e-01 1.09259402e-02
6.48794532e-01 -6.15875125e-01 -7.81369150e-01 6.06135964e-01
2.50693113e-01 3.00566047e-01 -1.94394633e-01 4.98236358e-01
-5.64322829e-01 1.04085438e-01 -3.21423262e-03 -5.20103633e-01
-5.74941218e-01 2.95511447e-03 5.00565469e-01 -1.40819415e-01
-8.59298587e-01 6.74599528e-01 -3.62871699e-02 -4.94045287e-01
-2.37333387e-01 -5.58673561e-01 2.03869820e-01 -3.94854426e-01
4.10865754e-01 6.97165877e-02 -8.80777538e-02 -2.07115207e-02
-2.60203063e-01 -1.00691013e-01 1.64881587e-01 -4.87105817e-01
1.65687835e+00 1.26368478e-01 -5.21618798e-02 7.48261273e-01
7.56791979e-02 -2.56173342e-01 -1.58762002e+00 -4.88499925e-02
5.92717290e-01 -7.94776559e-01 -4.51517366e-02 -1.33291757e+00
-4.26156938e-01 9.46779251e-01 -3.60645391e-02 2.16197312e-01
3.85261804e-01 1.21727586e-02 2.32489899e-01 5.65414310e-01
2.85922080e-01 -8.09084833e-01 -3.30394477e-01 2.80417711e-01
1.07733953e+00 -1.42591310e+00 1.14090607e-01 2.00474784e-01
-8.93794954e-01 1.24097824e+00 5.71321011e-01 1.16342641e-01
4.96593565e-01 3.17960978e-01 3.28660458e-01 -6.94449663e-01
-1.16191411e+00 4.43051219e-01 -1.28368452e-01 4.16677952e-01
-2.67727626e-03 4.25840884e-01 6.10289514e-01 5.27756155e-01
-2.57743895e-01 5.57654612e-02 8.63808811e-01 1.05468440e+00
-6.91195965e-01 -7.67652452e-01 -7.12985694e-01 6.66027665e-01
-2.58418739e-01 -1.88787505e-01 -3.58307242e-01 6.19563937e-01
-2.20512480e-01 1.06016052e+00 1.85914978e-01 -2.85867631e-01
1.20649420e-01 3.55710953e-01 -4.19632554e-01 -8.88434887e-01
-6.95181072e-01 -4.22996998e-01 6.11637414e-01 -6.78657115e-01
7.27177132e-03 -8.75838816e-01 -1.57977092e+00 -6.04981542e-01
1.84657007e-01 4.49371994e-01 7.48055518e-01 1.26360154e+00
5.02979159e-01 3.93542498e-01 4.46326375e-01 -7.37891972e-01
-8.55141699e-01 -6.57906473e-01 -5.18943071e-01 3.25361550e-01
3.85683417e-01 -3.63517135e-01 -4.87155855e-01 -3.37408334e-02] | [8.750967025756836, 5.839333534240723] |
68dc7561-3950-42d2-a39b-884a9bbe4707 | self-distillation-for-unsupervised-3d-domain | 2210.08226 | null | https://arxiv.org/abs/2210.08226v1 | https://arxiv.org/pdf/2210.08226v1.pdf | Self-Distillation for Unsupervised 3D Domain Adaptation | Point cloud classification is a popular task in 3D vision. However, previous works, usually assume that point clouds at test time are obtained with the same procedure or sensor as those at training time. Unsupervised Domain Adaptation (UDA) instead, breaks this assumption and tries to solve the task on an unlabeled target domain, leveraging only on a supervised source domain. For point cloud classification, recent UDA methods try to align features across domains via auxiliary tasks such as point cloud reconstruction, which however do not optimize the discriminative power in the target domain in feature space. In contrast, in this work, we focus on obtaining a discriminative feature space for the target domain enforcing consistency between a point cloud and its augmented version. We then propose a novel iterative self-training methodology that exploits Graph Neural Networks in the UDA context to refine pseudo-labels. We perform extensive experiments and set the new state-of-the-art in standard UDA benchmarks for point cloud classification. Finally, we show how our approach can be extended to more complex tasks such as part segmentation. | ['Luigi Di Stefano', 'Samuele Salti', 'Pierluigi Zama Ramirez', 'Riccardo Spezialetti', 'Adriano Cardace'] | 2022-10-15 | null | null | null | null | ['point-cloud-reconstruction', 'point-cloud-classification'] | ['computer-vision', 'computer-vision'] | [ 3.47300023e-01 3.87931354e-02 -2.56201386e-01 -4.94080752e-01
-7.67257035e-01 -7.75832236e-01 6.08510315e-01 2.18703866e-01
-1.02912314e-01 1.89304188e-01 -5.45513690e-01 -2.91713327e-01
4.42916062e-03 -7.35703886e-01 -1.02822340e+00 -6.10079110e-01
2.58098125e-01 1.12678850e+00 4.80645031e-01 1.53724089e-01
2.04218388e-01 9.14447844e-01 -1.47806764e+00 -8.27679485e-02
8.73850822e-01 8.84116113e-01 2.44274348e-01 3.62803608e-01
-3.94921511e-01 1.95560545e-01 -4.17891294e-01 -2.07177065e-02
6.80560172e-01 -2.04463795e-01 -9.45366979e-01 7.76250482e-01
7.65928090e-01 -3.12201474e-02 -1.86538264e-01 1.16426241e+00
7.11148977e-02 1.25019670e-01 9.61797118e-01 -1.42360854e+00
-4.18152392e-01 8.79691690e-02 -6.03730619e-01 -1.85177624e-01
3.61269191e-02 -2.27451604e-02 9.66057658e-01 -8.62067997e-01
8.64153862e-01 8.43947291e-01 7.00946212e-01 5.36750376e-01
-1.35235393e+00 -5.42899311e-01 5.16869426e-01 1.04008690e-01
-1.24943757e+00 -9.29145366e-02 1.31522727e+00 -6.67974710e-01
8.28165591e-01 -3.76314344e-03 7.77617395e-01 9.03309584e-01
-4.19629931e-01 8.60624790e-01 1.13681614e+00 -4.22202826e-01
5.17460585e-01 -3.09319459e-02 3.83547843e-01 5.01228571e-01
2.05475256e-01 -9.87522751e-02 -6.09226152e-02 -1.22197971e-01
8.42566609e-01 2.15861738e-01 -1.62784189e-01 -1.33855355e+00
-1.14943910e+00 7.87749231e-01 5.52198470e-01 2.85691828e-01
-3.80330533e-01 -1.74418747e-01 1.59192219e-01 3.97447109e-01
6.86322570e-01 3.59687746e-01 -5.61980426e-01 2.94517040e-01
-9.29653406e-01 2.03052610e-01 8.66309702e-01 1.30107474e+00
1.14686871e+00 -2.33016789e-01 1.95297733e-01 7.13461280e-01
4.50664252e-01 4.37031090e-01 2.00422183e-01 -9.14690256e-01
3.82219315e-01 8.41430902e-01 -1.79681517e-02 -5.23553491e-01
-2.77484208e-01 -4.72285897e-01 -5.17756164e-01 4.87388492e-01
5.16680539e-01 1.65046483e-01 -1.47093880e+00 1.40485787e+00
7.46014416e-01 7.34540343e-01 1.22228801e-01 9.36449647e-01
6.70178175e-01 3.88140231e-01 -2.58987248e-01 5.96810132e-02
8.44867587e-01 -1.08807743e+00 -2.74821389e-02 -4.08918381e-01
5.30577242e-01 -5.58565021e-01 9.04220343e-01 3.47905010e-01
-7.28672445e-01 -6.41355038e-01 -1.06569970e+00 -1.93538200e-02
-3.58801842e-01 -3.57376896e-02 4.53123659e-01 3.98472637e-01
-9.37067628e-01 7.28440225e-01 -9.72295284e-01 -7.05847740e-01
6.97384119e-01 4.96055096e-01 -4.97590810e-01 -2.86593258e-01
-3.84899914e-01 7.28882432e-01 3.52251410e-01 -1.86640829e-01
-8.79766107e-01 -5.81161618e-01 -9.11357939e-01 -3.82496297e-01
4.01727587e-01 -8.97339582e-01 1.28613365e+00 -1.02060926e+00
-1.50615549e+00 1.35264850e+00 -6.77678958e-02 -3.81675601e-01
2.95918554e-01 -1.54850200e-01 8.34096670e-02 1.13872096e-01
2.51707256e-01 6.87035859e-01 1.08114827e+00 -1.81024539e+00
-6.38359666e-01 -7.88370669e-01 1.00406103e-01 2.48861521e-01
6.77471310e-02 -4.82831657e-01 -7.37922728e-01 -2.88423866e-01
6.28774226e-01 -1.20612025e+00 -2.79452324e-01 2.36606956e-01
-3.96807373e-01 -4.28332388e-01 1.11638725e+00 -1.38164237e-01
2.61474192e-01 -2.09472489e+00 4.37318683e-01 5.06689608e-01
2.25278407e-01 2.00332627e-01 -1.87871486e-01 6.91030249e-02
-1.90886065e-01 -2.13699996e-01 -7.80069888e-01 -7.37851858e-01
6.91818222e-02 6.95495009e-01 -2.36158893e-01 8.50031376e-01
4.41909403e-01 7.24720061e-01 -9.84704852e-01 -6.10036731e-01
4.73886967e-01 1.55966163e-01 -4.54396307e-01 2.43040472e-01
-5.24785638e-01 7.80401886e-01 -6.15013659e-01 8.09671104e-01
1.01623452e+00 -4.67643142e-01 -7.32366443e-02 1.10964097e-01
-1.71618685e-02 1.11313768e-01 -1.11847162e+00 2.30859256e+00
-4.68149096e-01 2.99880415e-01 9.41345394e-02 -1.43115473e+00
1.07585645e+00 -8.18725973e-02 7.95571446e-01 -2.26329088e-01
1.03877574e-01 3.49350274e-01 -2.35787183e-01 -3.11128974e-01
2.90321022e-01 -1.65252045e-01 1.13782622e-01 1.87083364e-01
2.72321582e-01 -7.32184589e-01 -2.46301949e-01 -2.32855454e-01
1.00309956e+00 6.04959011e-01 1.68277249e-01 1.01709381e-01
4.56644714e-01 4.85696763e-01 3.90624434e-01 6.47479415e-01
-1.80199668e-01 9.87843454e-01 1.84004098e-01 -2.03746811e-01
-1.20749474e+00 -1.10013175e+00 -2.40298390e-01 6.38649464e-01
5.41668534e-01 -3.38666812e-02 -5.53540051e-01 -1.36291957e+00
3.02960426e-01 6.37668371e-01 -5.49236774e-01 7.95736611e-02
-4.97862309e-01 -1.61628753e-01 9.24327224e-02 4.75757539e-01
3.39702010e-01 -7.57249773e-01 -3.57386142e-01 -4.12489772e-02
1.78774521e-01 -1.27656519e+00 -2.09321633e-01 4.08289135e-01
-1.29774630e+00 -1.02608263e+00 -7.54823387e-01 -1.09141529e+00
9.50352430e-01 5.20533085e-01 1.36851895e+00 -2.93490496e-02
2.28385195e-01 7.62645900e-01 -4.88314629e-01 -5.01369178e-01
-2.48241678e-01 2.32588634e-01 -1.40959457e-01 1.99136194e-02
5.63188195e-01 -7.90345073e-01 -3.11396539e-01 3.15874398e-01
-8.22557747e-01 -7.59342313e-02 6.44830048e-01 7.81538785e-01
1.13018382e+00 -2.44531557e-01 1.30609035e-01 -1.11373103e+00
-1.41593948e-01 -5.01179576e-01 -7.15649545e-01 1.15667656e-01
-4.76633430e-01 7.74975792e-02 3.83130014e-01 -3.53479415e-01
-5.88304400e-01 6.31563008e-01 -1.58898681e-01 -1.35777712e+00
-6.19985580e-01 2.89857894e-01 -3.50912541e-01 -3.92353505e-01
5.41070759e-01 2.51037240e-01 1.37850903e-02 -6.53329670e-01
3.88482273e-01 5.26893497e-01 5.41121244e-01 -6.32274806e-01
1.28762996e+00 8.34521532e-01 2.27417842e-01 -7.46165454e-01
-7.54803598e-01 -1.17454362e+00 -1.11983132e+00 -7.30385855e-02
8.89809430e-01 -8.32741082e-01 -1.84985444e-01 3.46077770e-01
-1.28273356e+00 -2.83216387e-01 -6.95429862e-01 4.03998226e-01
-8.23012650e-01 5.85577250e-01 1.73315927e-01 -6.14513934e-01
2.95449849e-02 -1.08582294e+00 1.54085875e+00 -5.29331826e-02
1.78696916e-01 -9.68832493e-01 2.65478969e-01 1.58554003e-01
-3.39768618e-01 3.25090408e-01 6.76029265e-01 -9.78708684e-01
-7.03829885e-01 -1.81659400e-01 -1.96538016e-01 5.42169631e-01
2.44108841e-01 -2.42734194e-01 -9.69503343e-01 -2.45693117e-01
2.01168001e-01 -2.02257797e-01 8.25751662e-01 2.47948915e-01
1.18470204e+00 1.56988412e-01 -5.57086706e-01 8.18072379e-01
1.49317169e+00 -5.71091622e-02 3.13197941e-01 3.95418465e-01
1.01959217e+00 6.07284248e-01 8.75648975e-01 1.09337151e-01
3.52583170e-01 7.04335630e-01 7.88239956e-01 -1.54723555e-01
-5.97518086e-02 -3.13521445e-01 -4.07389831e-03 6.46796763e-01
-7.98957348e-02 -3.80263701e-02 -1.18407154e+00 8.05036545e-01
-1.95227253e+00 -4.64296341e-01 -1.38512224e-01 2.23003197e+00
4.64503556e-01 2.01441988e-01 2.18387350e-01 1.03762679e-01
6.53734386e-01 -8.00348148e-02 -8.59820187e-01 3.72545570e-02
4.40191589e-02 4.37352747e-01 6.83692396e-01 3.02449852e-01
-1.49068630e+00 1.01450527e+00 5.51641226e+00 5.35500407e-01
-1.15388548e+00 1.02444991e-01 1.89578488e-01 3.23914647e-01
-1.96541011e-01 1.88189626e-01 -5.37801564e-01 1.77853957e-01
4.53779578e-01 1.25777707e-01 1.35493025e-01 1.30698216e+00
-3.46666902e-01 2.92435974e-01 -1.54251635e+00 1.13934875e+00
1.12239636e-01 -1.21107066e+00 -6.79146796e-02 1.29041523e-01
8.71672690e-01 4.32283610e-01 -1.14957727e-01 2.98563272e-01
2.13913634e-01 -6.27965510e-01 6.85059309e-01 4.11232769e-01
5.82065344e-01 -4.31726038e-01 5.41675329e-01 4.53278720e-01
-1.05906093e+00 3.18045676e-01 -4.19442534e-01 1.46361724e-01
-1.62458658e-01 4.32626545e-01 -1.11369121e+00 9.28635299e-01
5.85420310e-01 1.28230405e+00 -5.47026932e-01 1.25694299e+00
-1.72167793e-01 4.58866924e-01 -5.20457923e-01 2.88637459e-01
3.20059806e-01 -5.17730832e-01 8.77654135e-01 9.66101706e-01
2.95293599e-01 -1.45322993e-01 5.42345226e-01 8.48855078e-01
-4.89090160e-02 -6.34816587e-02 -1.10375667e+00 1.33171678e-01
2.43611649e-01 1.10276210e+00 -8.59845340e-01 -2.91145027e-01
-6.79150224e-01 1.05289519e+00 3.85260075e-01 2.01482996e-01
-5.70060432e-01 -2.44602724e-03 7.47300982e-01 5.41712567e-02
6.89673960e-01 -5.60077608e-01 -5.18132508e-01 -1.23013997e+00
2.07577750e-01 -4.80325341e-01 2.01239258e-01 -6.43814564e-01
-1.66901326e+00 4.23768431e-01 1.28152505e-01 -2.01209092e+00
-3.19866955e-01 -7.97883451e-01 -4.86508101e-01 7.89731205e-01
-1.95707893e+00 -1.47456121e+00 -4.73088324e-01 7.90389776e-01
7.38913178e-01 -4.09344304e-03 6.37299180e-01 8.20582733e-02
-1.62766039e-01 1.48702204e-01 2.02886075e-01 5.89160658e-02
5.17018020e-01 -1.57558024e+00 5.27255058e-01 7.47253001e-01
5.27058005e-01 2.81751126e-01 5.16535223e-01 -5.96920192e-01
-1.37185252e+00 -1.40487027e+00 4.53799009e-01 -6.03597641e-01
5.89710474e-01 -5.29033244e-01 -1.17221451e+00 7.70006835e-01
-1.04093969e-01 4.21527654e-01 4.00000632e-01 7.57507831e-02
-3.48984361e-01 2.67950539e-02 -1.12131524e+00 2.36153394e-01
1.30356073e+00 -5.77227235e-01 -7.35764205e-01 5.91342688e-01
7.30351150e-01 -7.50567436e-01 -8.68902743e-01 5.65503716e-01
8.05472583e-02 -6.72030449e-01 1.02955306e+00 -7.09775686e-01
2.45068505e-01 -5.92029810e-01 -1.82773814e-01 -1.32001078e+00
-5.01880273e-02 -2.33959422e-01 -1.37335792e-01 1.08293724e+00
2.43796736e-01 -5.86264133e-01 1.29369354e+00 2.54485518e-01
-5.20272911e-01 -4.68746275e-01 -1.07357812e+00 -1.13439035e+00
2.05216765e-01 -6.18866265e-01 6.25645101e-01 1.12508428e+00
-6.27267540e-01 2.82274663e-01 2.54547358e-01 6.16512299e-01
8.28542709e-01 7.03870773e-01 1.10881364e+00 -1.72028196e+00
-2.46983215e-01 -2.34630927e-01 -7.65345633e-01 -1.26472414e+00
6.13932192e-01 -1.24372983e+00 2.57050157e-01 -1.62670875e+00
-1.16234183e-01 -7.74615705e-01 -1.84601769e-01 4.61681485e-01
1.26202017e-01 2.07504168e-01 2.19150826e-01 5.78339458e-01
-5.65563619e-01 4.82860118e-01 1.29598105e+00 -5.01996756e-01
-3.58947635e-01 2.91546404e-01 -2.96279758e-01 7.95352459e-01
6.81789458e-01 -5.40425420e-01 -4.72374201e-01 -5.93613923e-01
-3.59869808e-01 -1.60483390e-01 6.69234633e-01 -1.14419985e+00
2.43478060e-01 -2.91660011e-01 3.01427484e-01 -8.14286172e-01
5.35135329e-01 -1.31465125e+00 2.84547228e-02 -1.03297746e-02
7.73039609e-02 -2.22246930e-01 1.70867354e-01 8.07488084e-01
-2.23218724e-01 -4.52010751e-01 7.20585465e-01 -9.51030478e-02
-8.65884244e-01 6.02686882e-01 2.31284648e-01 -1.93257138e-01
1.22698748e+00 -6.13778472e-01 4.34776917e-02 9.51132849e-02
-9.32746053e-01 2.79167473e-01 1.09658849e+00 3.18349451e-01
8.39278758e-01 -1.21540070e+00 -5.70231676e-01 2.86072254e-01
5.88704884e-01 8.11508060e-01 -2.80446350e-03 6.69394851e-01
-3.90017331e-01 2.18640566e-01 -1.63645506e-01 -1.39457047e+00
-1.14558733e+00 8.08374941e-01 3.62786293e-01 -6.63028434e-02
-8.63553524e-01 6.62114680e-01 3.00843179e-01 -7.08433628e-01
1.04691401e-01 -5.02786458e-01 -1.88266814e-01 -2.05991998e-01
-3.06446314e-01 -8.20765793e-02 3.31088692e-01 -7.64299989e-01
-4.68144417e-01 9.91484344e-01 1.82670057e-02 4.76489849e-02
1.40409613e+00 1.20610707e-01 5.64458370e-02 6.86774611e-01
1.25292230e+00 -1.88695744e-01 -1.36303246e+00 -6.01637244e-01
2.09694192e-01 -6.23518765e-01 3.93084921e-02 -3.62040877e-01
-1.08852088e+00 7.54923284e-01 6.05582774e-01 1.91067755e-01
1.13764799e+00 5.14829993e-01 4.43675160e-01 4.51873481e-01
5.24696827e-01 -8.67833495e-01 -1.13942929e-01 5.16860843e-01
7.61004925e-01 -1.38709116e+00 -4.72033210e-02 -6.99527562e-01
-4.47008610e-01 1.00353026e+00 4.81610537e-01 -5.82552433e-01
7.61233449e-01 -1.35043815e-01 -6.83664232e-02 -3.39458346e-01
-2.70914733e-01 -6.60551548e-01 2.98782349e-01 1.03963971e+00
-1.26780525e-01 -5.63725494e-02 1.87658802e-01 1.84752733e-01
-1.55216139e-02 -8.24686736e-02 1.92591652e-01 1.20622563e+00
-1.37563124e-01 -1.36925173e+00 -4.64579850e-01 4.83964860e-01
1.32335767e-01 3.50419044e-01 -7.23190784e-01 1.01513100e+00
2.73024887e-01 6.71569884e-01 2.85085142e-01 -4.05848444e-01
5.87673545e-01 4.15022932e-02 7.48852909e-01 -9.95841444e-01
-2.76482671e-01 1.06209040e-01 -2.21908092e-01 -3.89812320e-01
-8.66823137e-01 -1.02315271e+00 -1.22124958e+00 2.33925909e-01
-3.75370264e-01 1.85533762e-01 9.40215468e-01 8.89748931e-01
2.95169383e-01 3.26398194e-01 7.71908700e-01 -1.11338496e+00
-3.70902598e-01 -7.79462516e-01 -6.25671506e-01 5.71979165e-01
5.26380122e-01 -9.53046799e-01 -2.42190883e-01 1.06435925e-01] | [8.012557029724121, -3.1193437576293945] |
cf03782b-05bc-4227-94eb-71cffd0ad687 | convergence-of-uncertainty-estimates-in | 2301.12649 | null | https://arxiv.org/abs/2301.12649v2 | https://arxiv.org/pdf/2301.12649v2.pdf | Convergence of uncertainty estimates in Ensemble and Bayesian sparse model discovery | Sparse model identification enables nonlinear dynamical system discovery from data. However, the control of false discoveries for sparse model identification is challenging, especially in the low-data and high-noise limit. In this paper, we perform a theoretical study on ensemble sparse model discovery, which shows empirical success in terms of accuracy and robustness to noise. In particular, we analyse the bootstrapping-based sequential thresholding least-squares estimator. We show that this bootstrapping-based ensembling technique can perform a provably correct variable selection procedure with an exponential convergence rate of the error rate. In addition, we show that the ensemble sparse model discovery method can perform computationally efficient uncertainty estimation, compared to expensive Bayesian uncertainty quantification methods via MCMC. We demonstrate the convergence properties and connection to uncertainty quantification in various numerical studies on synthetic sparse linear regression and sparse model discovery. The experiments on sparse linear regression support that the bootstrapping-based sequential thresholding least-squares method has better performance for sparse variable selection compared to LASSO, thresholding least-squares, and bootstrapping-based LASSO. In the sparse model discovery experiment, we show that the bootstrapping-based sequential thresholding least-squares method can provide valid uncertainty quantification, converging to a delta measure centered around the true value with increased sample sizes. Finally, we highlight the improved robustness to hyperparameter selection under shifting noise and sparsity levels of the bootstrapping-based sequential thresholding least-squares method compared to other sparse regression methods. | ['J. Nathan Kutz', 'Steven L. Brunton', 'Urban Fasel', 'L. Mars Gao'] | 2023-01-30 | null | null | null | null | ['variable-selection', 'model-discovery'] | ['methodology', 'miscellaneous'] | [ 1.58021688e-01 -1.35589868e-01 -3.85752171e-02 -3.74702960e-02
-9.94889975e-01 -3.64624619e-01 2.85062790e-01 -2.57039696e-01
1.13754958e-01 1.38127816e+00 -2.06416368e-01 -1.36703223e-01
-4.84181464e-01 -5.54494262e-01 -8.76835704e-01 -1.03998458e+00
-4.12551254e-01 7.65315175e-01 -3.41219634e-01 2.19165370e-01
-4.37661037e-02 1.99888691e-01 -1.35418558e+00 -4.11700130e-01
9.34658527e-01 1.07265949e+00 -1.53024122e-01 4.03976947e-01
3.95645887e-01 4.34476018e-01 -2.56358534e-01 1.87564582e-01
3.10962915e-01 -5.78426898e-01 -6.77739158e-02 -3.27012658e-01
2.65672207e-01 -1.57520659e-02 3.54331657e-02 1.16726637e+00
5.96810579e-01 1.15196593e-01 8.77960622e-01 -1.39943969e+00
-3.26464593e-01 8.30898523e-01 -8.03836286e-01 1.64893955e-01
2.35894412e-01 -5.63345291e-02 5.47566056e-01 -1.32926393e+00
4.84762043e-01 1.13697457e+00 1.17441463e+00 1.88676029e-01
-1.72973990e+00 -9.31783736e-01 -2.64254391e-01 -5.00965655e-01
-1.92140245e+00 -7.64096200e-01 5.10361433e-01 -7.87636399e-01
5.20279348e-01 3.18205893e-01 4.04770285e-01 1.14335787e+00
2.63139278e-01 1.22634351e-01 1.24965847e+00 -3.20842326e-01
8.04497719e-01 3.83052602e-02 2.79479623e-01 8.47172022e-01
9.27412570e-01 6.41582072e-01 -5.84371448e-01 -8.16627979e-01
1.13176358e+00 -1.82016134e-01 -3.36975843e-01 -4.72188264e-01
-1.30721235e+00 9.45277750e-01 -2.30748534e-01 2.38423452e-01
-5.40975094e-01 6.02060378e-01 2.14400619e-01 2.46409193e-01
7.62564600e-01 5.29946625e-01 -3.17057848e-01 -4.46093716e-02
-1.60970962e+00 1.09904028e-01 1.07187796e+00 9.00113344e-01
4.03240293e-01 8.01259637e-01 -1.50405109e-01 6.16771281e-01
5.30258834e-01 1.23513758e+00 1.59964412e-01 -1.28603292e+00
3.91850211e-02 -2.14682162e-01 3.47073823e-01 -1.26482606e+00
-4.43141162e-01 -7.11419642e-01 -1.48896742e+00 7.17535987e-02
4.32119548e-01 -4.29550081e-01 -6.67486370e-01 1.81350458e+00
2.93014467e-01 7.62920439e-01 -7.99696073e-02 7.18105376e-01
4.14348274e-01 6.00011051e-01 -2.65884876e-01 -1.16484427e+00
8.69511664e-01 -2.18482599e-01 -8.44888031e-01 2.31871083e-01
4.39644068e-01 -2.93306291e-01 4.93156999e-01 4.20868486e-01
-9.90992069e-01 -1.48531094e-01 -6.61033392e-01 7.69836664e-01
4.18404520e-01 4.51970659e-02 4.72564369e-01 6.29756451e-01
-6.29875720e-01 4.73882109e-01 -1.14417827e+00 -1.30750299e-01
2.08774731e-01 2.83264518e-01 -3.72916132e-01 1.80094808e-01
-9.95062888e-01 5.32851636e-01 7.15270713e-02 2.37906829e-01
-1.02981746e+00 -1.02199209e+00 -9.56117511e-01 1.82180688e-01
3.56284738e-01 -9.55967546e-01 7.81906426e-01 -7.08582461e-01
-1.45233417e+00 3.91056031e-01 -6.35912657e-01 -7.64618754e-01
4.58146155e-01 1.07247740e-01 -1.35531062e-02 6.79993350e-03
2.94547826e-01 -1.58050403e-01 1.35703731e+00 -1.44203913e+00
1.26028940e-01 -2.44926065e-01 -9.23825383e-01 -3.20549726e-01
2.57239282e-01 -8.38078484e-02 2.58649051e-01 -8.41109276e-01
5.96631885e-01 -8.97107542e-01 -4.65696245e-01 -2.52176285e-01
-3.31227005e-01 3.26561391e-01 4.99959648e-01 -5.00202477e-01
1.45415854e+00 -2.11611962e+00 3.52433205e-01 8.16599548e-01
2.49764547e-01 -3.51344079e-01 2.02094108e-01 1.97751507e-01
-1.59774125e-01 2.80102789e-01 -8.55084956e-01 -2.98202783e-01
-1.84952065e-01 6.75356612e-02 -4.83290672e-01 9.03597593e-01
6.27149716e-02 5.25821865e-01 -8.12865078e-01 -5.62247276e-01
1.66083038e-01 2.62647182e-01 -4.92319047e-01 -1.00313067e-01
1.06568649e-01 6.76733792e-01 -3.19567025e-01 1.02205288e+00
5.13209164e-01 -5.78907788e-01 1.44664705e-01 -3.98993902e-02
4.73831035e-02 -6.49275959e-01 -1.67780840e+00 1.22172534e+00
-3.25039238e-01 4.52403307e-01 6.28510177e-01 -1.21355832e+00
1.00434041e+00 5.44012189e-01 5.76586127e-01 7.65827820e-02
1.85708761e-01 5.73608577e-01 -2.61323243e-01 -1.57346487e-01
1.31188348e-01 -6.32895172e-01 -2.22296253e-01 2.04887345e-01
-8.55704211e-03 -4.40102816e-01 -2.08254289e-02 2.50996292e-01
9.27713752e-01 -1.94029823e-01 7.49901295e-01 -7.33960092e-01
3.47944707e-01 -6.52587786e-02 7.69199491e-01 1.10413170e+00
-1.26445312e-02 5.89467585e-01 4.74674404e-01 -3.38747576e-02
-1.03199637e+00 -9.83817399e-01 -7.51652241e-01 4.91968364e-01
-9.60882008e-02 -1.21856771e-01 -4.18456018e-01 -1.25312939e-01
4.45209891e-01 6.91489279e-01 -6.64287806e-01 -4.70030159e-02
-2.77260691e-01 -1.23092818e+00 4.93383259e-01 3.39630365e-01
5.92276417e-02 -3.92869264e-01 -8.33812431e-02 1.45570710e-01
-2.30097082e-02 -8.54636967e-01 -2.52175421e-01 3.55155855e-01
-9.07571495e-01 -9.03959930e-01 -7.49600410e-01 -3.83967131e-01
7.95012712e-01 -2.34069616e-01 8.75849664e-01 -2.48295411e-01
-5.36770187e-02 5.85329294e-01 -6.76686913e-02 -1.31562456e-01
-3.96718144e-01 -4.09933656e-01 7.58525252e-01 1.09225623e-02
-4.55641933e-02 -6.83518827e-01 4.24380153e-02 4.21120405e-01
-3.91703755e-01 -4.78668004e-01 1.45113006e-01 1.47395861e+00
9.67245579e-01 1.86935455e-01 7.92920470e-01 -5.66573024e-01
5.46389759e-01 -9.74587917e-01 -1.04176462e+00 2.38779616e-02
-6.63971603e-01 1.42514825e-01 5.33743024e-01 -5.00272453e-01
-7.84351349e-01 2.19647095e-01 4.00655329e-01 -8.25199127e-01
3.95080835e-01 9.34480250e-01 2.71614969e-01 -3.58649671e-01
7.95253456e-01 2.73745865e-01 3.01183254e-01 -3.25969905e-01
3.05793267e-02 4.62553501e-01 4.17404592e-01 -9.01101351e-01
8.51023793e-01 4.93366271e-01 5.76150537e-01 -1.13723397e+00
-5.89888871e-01 -2.04215750e-01 -2.47680292e-01 -8.68818313e-02
3.76412511e-01 -1.12294877e+00 -6.34277999e-01 2.82065094e-01
-6.49544597e-01 -1.91551536e-01 -6.14578187e-01 7.49345958e-01
-8.48775208e-01 3.69132936e-01 -3.27912569e-01 -1.45577991e+00
-2.46397138e-01 -9.64766920e-01 9.73097622e-01 -8.17749426e-02
-3.18661451e-01 -1.00594401e+00 2.44906619e-01 -6.98571801e-02
5.20370662e-01 6.42318308e-01 4.91722316e-01 -7.18015909e-01
-2.45996088e-01 -1.82471648e-01 2.21846774e-02 1.11814931e-01
-3.15787464e-01 3.92527610e-01 -5.23931265e-01 -5.02354503e-01
3.67859602e-01 -7.31943324e-02 8.80601585e-01 1.28197825e+00
8.56014371e-01 -3.48682761e-01 -5.86771667e-01 8.56116951e-01
1.36791027e+00 -3.03610265e-01 1.69325069e-01 -1.22543968e-01
3.75858009e-01 1.89086407e-01 5.32687068e-01 8.93230021e-01
-3.03140163e-01 3.60495955e-01 1.68293402e-01 2.14817837e-01
3.77188742e-01 -9.43079665e-02 3.15563172e-01 7.95544446e-01
-6.38797383e-06 7.83077404e-02 -9.58477378e-01 3.38731617e-01
-1.95966423e+00 -1.37936091e+00 -4.65169907e-01 2.64988422e+00
9.24664319e-01 -1.91927284e-01 2.60182209e-02 -7.42803663e-02
8.94795835e-01 -1.42133251e-01 -6.04415834e-01 1.59849338e-02
-3.88365597e-01 6.75198585e-02 6.91050649e-01 7.98813283e-01
-9.05223131e-01 4.95559961e-01 7.57410908e+00 8.26408565e-01
-6.70299351e-01 2.32642859e-01 5.78908622e-01 -3.06731194e-01
-2.81737268e-01 1.18538663e-01 -7.93923795e-01 8.00682247e-01
1.14601171e+00 -4.93551791e-01 4.28677797e-01 8.56396914e-01
7.23600745e-01 -3.67097199e-01 -9.17593062e-01 1.15798318e+00
-1.38554320e-01 -1.30193818e+00 -3.56984049e-01 1.51917994e-01
1.14257395e+00 -7.37376362e-02 -9.88800675e-02 1.03939272e-01
4.33361828e-01 -1.31984818e+00 4.92319822e-01 9.56885636e-01
9.85522866e-01 -4.82700318e-01 8.37197542e-01 4.71028149e-01
-9.34466958e-01 -2.34302014e-01 -3.85777652e-01 -1.02394119e-01
4.32197571e-01 1.59152329e+00 -4.76502836e-01 2.68837690e-01
5.81211686e-01 7.65887082e-01 1.66531801e-01 1.03646505e+00
1.05521224e-01 1.15185630e+00 -9.96750593e-01 1.30225360e-01
-3.39919478e-01 -5.67678213e-01 1.21036398e+00 9.79789376e-01
8.33447397e-01 2.62032866e-01 1.18420042e-01 1.32890463e+00
1.68899223e-01 -2.01752320e-01 -6.94150388e-01 -9.35904160e-02
9.55747962e-01 7.59276092e-01 -5.22095740e-01 -4.26651150e-01
4.33502272e-02 2.81860828e-01 -1.62545651e-01 6.03202164e-01
-7.40447879e-01 5.20351380e-02 3.65162164e-01 1.33716762e-01
3.34591895e-01 -3.62589628e-01 -7.08550632e-01 -1.36147022e+00
-2.72234589e-01 -1.03486073e+00 2.14549810e-01 -7.10229099e-01
-1.47922206e+00 1.61379591e-01 3.74674678e-01 -9.78341341e-01
-4.13492411e-01 -2.23456860e-01 -4.22320455e-01 8.74503076e-01
-6.70698225e-01 -6.63168132e-01 5.64554408e-02 3.36387396e-01
1.60997540e-01 -3.19300413e-01 6.64358675e-01 -1.28574014e-01
-8.94177675e-01 3.40747088e-01 8.83478820e-01 -3.16401184e-01
5.35625160e-01 -1.00374758e+00 -3.38269740e-01 9.54390109e-01
-2.12921903e-01 9.10676301e-01 1.26049757e+00 -1.01117337e+00
-1.39871979e+00 -7.48939097e-01 4.59293336e-01 -4.20904547e-01
9.19471085e-01 -4.01388928e-02 -9.61124897e-01 7.11285055e-01
-4.99561220e-01 3.77958655e-01 7.10367560e-01 2.83596486e-01
-9.18081254e-02 4.53323908e-02 -1.56708872e+00 3.91424783e-02
6.37284458e-01 -2.38369703e-01 -6.26157463e-01 4.14548934e-01
4.19496626e-01 -2.71402180e-01 -1.24779582e+00 6.30220950e-01
6.95331216e-01 -5.76814473e-01 9.53943074e-01 -5.05282342e-01
1.43852323e-01 -5.12310803e-01 -4.45692509e-01 -1.13587546e+00
-4.30123955e-01 -9.59554851e-01 -6.09924495e-01 1.08603811e+00
3.65419209e-01 -8.59336376e-01 5.25573969e-01 4.99501616e-01
2.56506860e-01 -5.57777047e-01 -1.48253167e+00 -1.09852457e+00
1.23779535e-01 -3.80147517e-01 1.18620381e-01 1.04316640e+00
3.92375402e-02 1.47807486e-02 -6.44295454e-01 4.27482367e-01
1.27283204e+00 1.24138907e-01 5.84064007e-01 -1.45848346e+00
-4.46436077e-01 -1.29587576e-01 -3.26397717e-01 -5.04404843e-01
2.98418880e-01 -6.83571160e-01 1.55143023e-01 -9.55522954e-01
3.21446121e-01 -6.21456504e-01 -8.78480896e-02 2.12261483e-01
-1.46589473e-01 1.80761352e-01 -1.47941232e-01 4.66233045e-01
-2.73019612e-01 5.53306937e-01 7.86549211e-01 2.30291262e-02
-4.16233361e-01 -1.29752476e-02 -4.14193153e-01 7.37844646e-01
5.23964405e-01 -9.19618487e-01 -8.73195007e-02 3.04066211e-01
5.26586890e-01 6.00211561e-01 6.63161576e-01 -9.04525042e-01
2.25850180e-01 -2.69897938e-01 2.35224962e-01 -3.72770905e-01
2.31950149e-01 -7.98155189e-01 6.84875309e-01 5.25627077e-01
-1.79040745e-01 -3.38995427e-01 1.72415286e-01 1.00666821e+00
-7.97687769e-02 -4.10902977e-01 8.96914542e-01 -1.43413335e-01
3.56453774e-03 3.04010622e-02 -5.13820946e-01 2.78435916e-01
1.02764010e+00 -5.78792766e-02 -2.18478069e-02 -5.11471272e-01
-1.07597911e+00 2.46925414e-01 4.44061428e-01 -5.64204931e-01
5.46555281e-01 -1.20157635e+00 -1.14645767e+00 4.03947145e-01
-3.56060326e-01 -7.77784288e-02 4.18195724e-02 1.45724905e+00
-1.11442894e-01 2.25990027e-01 3.70442927e-01 -1.02163613e+00
-9.47073877e-01 4.81409520e-01 5.23375630e-01 -2.52804509e-03
-1.56519473e-01 8.22384417e-01 -2.56342530e-01 -2.46201664e-01
8.10733810e-02 -2.85380721e-01 4.11895066e-01 -2.27564406e-02
3.17403167e-01 9.08710539e-01 -3.88404280e-01 -5.22487640e-01
-4.52173591e-01 7.81602383e-01 5.89523852e-01 -1.43720374e-01
1.17996264e+00 -2.26995692e-01 -4.74491924e-01 7.74628818e-01
9.15242612e-01 1.78873628e-01 -1.06078339e+00 1.89219974e-02
-1.75623119e-01 -2.69662946e-01 2.59427547e-01 -5.90903223e-01
-7.96367586e-01 3.63585502e-01 3.37690890e-01 3.19349796e-01
7.36439049e-01 -1.11154467e-01 9.39201340e-02 3.19265425e-01
6.60242021e-01 -8.15742552e-01 -5.91666043e-01 4.37084347e-01
1.02444685e+00 -1.43948138e+00 3.92956883e-01 -3.97442579e-01
-2.46342078e-01 8.23046446e-01 9.06399488e-02 -3.74650210e-01
9.30597663e-01 6.06548667e-01 -6.21823967e-01 -5.67747839e-02
-7.97004044e-01 2.87226707e-01 1.94180861e-01 6.17798924e-01
1.23253159e-01 1.20259978e-01 -3.63028675e-01 8.96867394e-01
-1.17023058e-01 2.60656804e-01 4.78447974e-01 5.30908287e-01
-6.01619720e-01 -3.46242607e-01 -9.62809920e-01 9.46878135e-01
-1.51323289e-01 -3.36339384e-01 -4.13687043e-02 6.57901764e-01
-4.58611101e-01 9.83700037e-01 -1.08004687e-03 8.03857744e-02
-1.90123558e-01 2.44736552e-01 2.51007110e-01 -4.47456688e-01
-1.80142671e-01 2.48665228e-01 6.42901734e-02 -5.77226400e-01
-3.41541201e-01 -1.03388667e+00 -1.07484603e+00 -4.63011116e-01
-8.20715547e-01 4.22544718e-01 4.30606514e-01 8.51466715e-01
3.61475617e-01 1.96336597e-01 5.43357790e-01 -7.13587284e-01
-1.11133146e+00 -9.51049149e-01 -1.00413895e+00 -1.75123602e-01
4.84977037e-01 -8.80074680e-01 -1.18773150e+00 -2.54874062e-02] | [7.094308853149414, 4.435424327850342] |
69dbcee6-e3e8-4005-8e95-10d6f8b17216 | htmot-hierarchical-topic-modelling-over-time | 2112.03104 | null | https://arxiv.org/abs/2112.03104v2 | https://arxiv.org/pdf/2112.03104v2.pdf | HTMOT : Hierarchical Topic Modelling Over Time | Over the years, topic models have provided an efficient way of extracting insights from text. However, while many models have been proposed, none are able to model topic temporality and hierarchy jointly. Modelling time provide more precise topics by separating lexically close but temporally distinct topics while modelling hierarchy provides a more detailed view of the content of a document corpus. In this study, we therefore propose a novel method, HTMOT, to perform Hierarchical Topic Modelling Over Time. We train HTMOT using a new implementation of Gibbs sampling, which is more efficient. Specifically, we show that only applying time modelling to deep sub-topics provides a way to extract specific stories or events while high level topics extract larger themes in the corpus. Our results show that our training procedure is fast and can extract accurate high-level topics and temporally precise sub-topics. We measured our model's performance using the Word Intrusion task and outlined some limitations of this evaluation method, especially for hierarchical models. As a case study, we focused on the various developments in the space industry in 2020. | ['Ashwin Ittoo', 'Judicael Poumay'] | 2021-11-22 | null | null | null | null | ['topic-models'] | ['natural-language-processing'] | [-2.05056369e-01 2.74127632e-01 -3.23185116e-01 -3.32355380e-01
-9.94083703e-01 -3.68912935e-01 1.21869576e+00 2.53465772e-01
-1.52014911e-01 6.54765725e-01 5.75486779e-01 -1.72928542e-01
-5.79822622e-02 -9.52718735e-01 -2.25324512e-01 -5.83233297e-01
-4.87299412e-01 6.94165170e-01 5.29392123e-01 6.78127781e-02
4.83880043e-01 2.49264836e-02 -1.58835900e+00 4.17969555e-01
5.58751523e-01 6.19361818e-01 5.07643759e-01 5.23591697e-01
-5.00528097e-01 5.53941369e-01 -8.80108058e-01 1.29710332e-01
-2.08928987e-01 -4.86546010e-01 -1.03688884e+00 2.86949694e-01
-9.65340957e-02 -1.55692771e-01 -9.14744735e-02 2.59658754e-01
6.65713400e-02 2.94375092e-01 6.57701433e-01 -1.21190727e+00
2.84432560e-01 9.95747268e-01 -5.86363852e-01 2.40856051e-01
1.49758190e-01 -3.49556923e-01 1.22622502e+00 -6.64798498e-01
9.61620569e-01 1.19026124e+00 5.83153367e-01 2.56105214e-01
-1.30514908e+00 -5.14001489e-01 6.58232808e-01 2.38551527e-01
-1.22309864e+00 -4.89277989e-02 9.08185005e-01 -5.74923038e-01
1.28866446e+00 2.75482774e-01 8.67448092e-01 1.24110591e+00
5.23736358e-01 1.04832363e+00 1.16851974e+00 -3.71087402e-01
4.05739129e-01 1.90917328e-01 4.29630011e-01 -7.61988387e-02
-1.85180455e-01 -1.60869122e-01 -9.58788574e-01 -4.16068137e-01
3.94366354e-01 1.32376030e-01 7.16950744e-02 -9.59623307e-02
-1.35546482e+00 1.03092170e+00 -1.02437302e-01 7.33946383e-01
-5.87997377e-01 1.85452804e-01 3.73710304e-01 -1.22146800e-01
1.11205220e+00 3.11050236e-01 -4.55116242e-01 -5.46229005e-01
-1.70902300e+00 4.22878951e-01 8.60545576e-01 7.88732886e-01
7.29857683e-01 -2.84436971e-01 -2.97459275e-01 7.08311260e-01
3.27466100e-01 -2.14859217e-01 6.41169906e-01 -6.43132567e-01
3.23033720e-01 1.59468427e-01 2.44177818e-01 -8.71057272e-01
-5.46822250e-01 -3.28624129e-01 -3.34478557e-01 -2.74406493e-01
-3.88532765e-02 -9.68029648e-02 -1.15283084e+00 1.58786786e+00
2.96204686e-01 1.50381207e-01 -2.84741074e-01 2.62920886e-01
4.38946307e-01 1.32739425e+00 3.34320784e-01 -5.69252670e-01
1.59070647e+00 -7.56858766e-01 -1.17731094e+00 -2.30659828e-01
4.29619372e-01 -6.33582413e-01 8.01172435e-01 6.45627022e-01
-1.17227042e+00 -2.69046277e-01 -9.27806318e-01 -2.22160318e-03
-6.67898476e-01 -2.29566053e-01 8.69417548e-01 4.37883675e-01
-9.81697738e-01 6.36387944e-01 -1.33255303e+00 -6.90620124e-01
1.97341323e-01 4.18200456e-02 2.00164154e-01 3.87497574e-01
-1.46784890e+00 8.51806760e-01 5.41701615e-01 -3.09121430e-01
-1.15654898e+00 -6.36639893e-01 -6.20568991e-01 1.13759942e-01
3.21719855e-01 -3.88509572e-01 1.56158698e+00 -2.56091088e-01
-1.36794722e+00 6.32250130e-01 -5.73845565e-01 -7.86687672e-01
1.88474759e-01 -3.93490672e-01 -4.50172961e-01 1.67549849e-01
2.26019174e-01 7.94672847e-01 6.24015629e-01 -1.12946010e+00
-1.09916663e+00 -2.77941972e-01 -2.29626358e-01 8.93669128e-02
-4.35265154e-01 3.30615163e-01 -5.10427296e-01 -6.74407661e-01
8.88297409e-02 -7.45749235e-01 -2.77527392e-01 -7.81841218e-01
-4.89777535e-01 -6.58842027e-01 1.01228881e+00 -7.53709316e-01
1.74162352e+00 -2.01115203e+00 9.50938184e-03 -1.48033217e-01
1.20660186e-01 -6.70521557e-01 5.26666999e-01 9.18949723e-01
2.19903216e-02 3.21554393e-01 -3.17326248e-01 -7.23848999e-01
2.32460797e-01 1.75798655e-01 -9.09146488e-01 1.70041144e-01
-5.02039827e-02 5.45980096e-01 -7.68015325e-01 -6.64436162e-01
2.33879581e-01 5.04642010e-01 -3.29858184e-01 -1.27117932e-01
-5.84864199e-01 -1.26048997e-02 -4.79132652e-01 1.50088295e-01
2.17273921e-01 -2.04348147e-01 2.44754270e-01 3.24747562e-01
-4.41886723e-01 1.07326591e+00 -8.48897457e-01 1.92899001e+00
-5.63947439e-01 1.20006156e+00 -3.26568365e-01 -7.13340342e-01
8.25047553e-01 8.03797543e-01 1.00034881e+00 -3.49894345e-01
-1.51921913e-01 -2.26637006e-01 -5.23201227e-01 -2.74318792e-02
1.08672369e+00 -4.89914566e-01 -3.35422754e-01 7.47818470e-01
1.27942890e-01 -3.79119426e-01 2.88562238e-01 3.08055490e-01
9.10373628e-01 2.49421492e-01 3.01545054e-01 -4.87775087e-01
-3.16965997e-01 3.06589156e-01 4.26945567e-01 5.07410824e-01
-6.58105463e-02 5.93915880e-01 5.06632626e-01 -3.30232978e-01
-1.06799638e+00 -9.75107551e-01 -2.93170124e-01 1.20507872e+00
-1.29706562e-01 -1.13514054e+00 -5.27795434e-01 -5.80678821e-01
-4.39809978e-01 1.38281453e+00 -7.46451139e-01 2.72842646e-01
-5.32527745e-01 -9.63223338e-01 4.18165401e-02 3.27456385e-01
2.19235420e-01 -1.03440940e+00 -9.75288153e-01 5.07775128e-01
-7.52789140e-01 -8.05370152e-01 -1.56555086e-01 4.37741339e-01
-1.21208298e+00 -4.98791426e-01 -8.09704065e-01 -4.84170735e-01
-1.81640163e-02 1.27356425e-01 1.29039562e+00 -8.30869794e-01
-2.44567618e-01 2.84389406e-01 -2.03204796e-01 -8.77793491e-01
-4.16562632e-02 4.86029685e-01 -3.73160869e-01 -3.39426011e-01
6.56258047e-01 -7.22778738e-01 -5.41517138e-01 2.49750257e-01
-8.99305820e-01 3.50729853e-01 1.94022939e-01 3.57256383e-01
3.59650612e-01 6.61891341e-01 4.36386049e-01 -8.72146666e-01
6.74534976e-01 -7.62063980e-01 -3.48880231e-01 -2.33456656e-01
-7.81533599e-01 -2.91090216e-02 -4.23448682e-02 -3.39751959e-01
-1.42260945e+00 -4.31114256e-01 5.54021038e-02 1.67608216e-01
-3.51418644e-01 7.71302342e-01 2.24186867e-01 1.17295778e+00
4.53092933e-01 2.81827807e-01 -4.86175835e-01 -6.44507527e-01
4.35426980e-01 5.89436471e-01 1.32928222e-01 -7.37510920e-01
3.78352970e-01 9.28393960e-01 -5.11983752e-01 -1.00756991e+00
-8.10610533e-01 -8.79347205e-01 -4.64229494e-01 -2.78862953e-01
8.63092184e-01 -9.49016929e-01 -7.41793662e-02 1.81199461e-01
-1.21969163e+00 -4.55667377e-01 -5.44716716e-01 5.49913049e-01
-7.45039821e-01 2.07702667e-01 -7.03228474e-01 -9.99145210e-01
-2.06942469e-01 -5.98580241e-01 1.33272004e+00 -6.06605299e-02
-8.21300983e-01 -1.27978098e+00 4.66949850e-01 -1.26857370e-01
4.00095671e-01 4.55603957e-01 8.11192572e-01 -6.66600347e-01
-4.91117954e-01 1.96046047e-02 2.68145770e-01 -2.05552787e-01
2.42606997e-01 6.97375834e-02 -1.17534137e+00 -1.90924600e-01
4.16516423e-01 2.18970209e-01 1.11746144e+00 6.85811937e-01
9.37998950e-01 -4.30053025e-01 -8.67614865e-01 -5.16582765e-02
1.22991335e+00 4.16695684e-01 8.59177828e-01 7.86671996e-01
1.72729000e-01 8.13489199e-01 9.37294424e-01 7.00423539e-01
6.06955588e-01 8.01837325e-01 1.23600416e-01 -6.24378771e-02
1.40470490e-01 -1.44504696e-01 5.20278871e-01 9.22058105e-01
-3.25860344e-02 -2.23799542e-01 -1.06964576e+00 1.31362462e+00
-1.82816732e+00 -1.05500507e+00 -1.18734032e-01 1.97134006e+00
1.13643456e+00 5.15590906e-01 3.07639271e-01 1.38398483e-02
5.43475926e-01 5.33111989e-01 -4.05219495e-02 -1.96550593e-01
1.74334660e-01 5.02993278e-02 1.58260897e-01 4.63683844e-01
-1.22833598e+00 9.25999403e-01 7.24691343e+00 9.01882172e-01
-8.49521220e-01 2.93085277e-01 4.61693257e-01 -4.08769816e-01
-4.73084658e-01 4.12014723e-01 -1.00715017e+00 5.05774021e-01
1.39703667e+00 -4.59156990e-01 -4.42162305e-01 1.02379799e+00
5.28996229e-01 -3.23926061e-01 -1.11412454e+00 4.41588163e-01
-1.59945861e-01 -1.16003096e+00 -9.97302458e-02 4.82530475e-01
8.17762911e-01 2.18088143e-02 -1.91841766e-01 4.81517583e-01
4.52095628e-01 -7.26539016e-01 8.54719579e-01 4.39316213e-01
1.63638487e-01 -7.88212001e-01 4.44781959e-01 3.84830713e-01
-1.21531844e+00 2.71853328e-01 -2.26228654e-01 -1.34437874e-01
6.66131437e-01 1.07063508e+00 -1.21446025e+00 7.06940830e-01
8.20679605e-01 8.31461787e-01 -1.57457039e-01 9.63241518e-01
-1.05826522e-03 9.50744331e-01 -5.49057066e-01 7.80345127e-02
4.57352757e-01 1.22597806e-01 6.74858689e-01 1.58965898e+00
5.09729981e-01 -1.50205955e-01 1.97972119e-01 7.88278043e-01
4.46680278e-01 -7.35627189e-02 -6.24403119e-01 -1.71060354e-01
2.86981404e-01 1.12853646e+00 -1.14987350e+00 -5.41922569e-01
-2.48918653e-01 5.84490180e-01 -3.44742954e-01 4.22881693e-01
-9.21278536e-01 -3.56011182e-01 6.57662868e-01 2.78156251e-01
4.73397076e-01 -6.48238361e-01 -3.94313842e-01 -1.03876388e+00
-2.74449617e-01 -4.53625202e-01 6.20392263e-01 -7.71370411e-01
-8.56155872e-01 5.91748714e-01 6.60400689e-01 -9.76209939e-01
-6.06442451e-01 -7.57740885e-02 -6.31507814e-01 7.36566961e-01
-1.15244782e+00 -8.87457967e-01 3.02423779e-02 2.09717974e-01
1.22453952e+00 2.79965580e-01 7.08193779e-01 -1.28465310e-01
-1.62018567e-01 -3.89182299e-01 9.11732856e-03 -3.90608579e-01
6.57974005e-01 -1.54326594e+00 7.13551879e-01 7.71274507e-01
3.79471600e-01 8.56598198e-01 1.13847935e+00 -8.18900466e-01
-6.16283476e-01 -8.26438427e-01 1.34350336e+00 -6.07171655e-01
8.00536513e-01 -6.03973091e-01 -8.26248229e-01 1.00158572e+00
7.78040826e-01 -1.11671174e+00 8.21924686e-01 7.77185559e-01
-1.14176581e-02 1.43861085e-01 -4.92729008e-01 5.05241990e-01
4.10225511e-01 -6.51217818e-01 -1.02562785e+00 4.19983476e-01
1.01369596e+00 -3.44371311e-02 -9.78740335e-01 2.41534069e-01
3.59297365e-01 -8.80254507e-01 7.69436359e-01 -3.31341416e-01
4.71886665e-01 -9.76728126e-02 5.68052344e-02 -1.18936551e+00
-2.25796983e-01 -8.09284270e-01 -4.68865961e-01 1.49591720e+00
5.37339449e-01 -3.71952295e-01 8.43130589e-01 4.86899942e-01
-7.85704181e-02 -6.95905507e-01 -7.78325438e-01 -7.43496478e-01
4.90860157e-02 -9.03810740e-01 4.53297436e-01 1.03555393e+00
4.17775631e-01 4.08101916e-01 -3.27358156e-01 -1.62660286e-01
4.18671995e-01 3.89639854e-01 4.87220794e-01 -1.36747086e+00
-7.78014958e-02 -5.80688179e-01 -4.12328131e-02 -1.03897607e+00
-1.12790696e-01 -3.03546637e-01 3.47684294e-01 -2.03489995e+00
3.54851216e-01 -1.35204777e-01 -1.89299226e-01 3.25278729e-01
8.99899676e-02 -2.84962863e-01 -5.81314489e-02 3.94143313e-01
-6.06403172e-01 5.90401173e-01 7.45925069e-01 -1.03905357e-01
-3.66144210e-01 2.57404774e-01 -5.01613855e-01 6.24201775e-01
8.51392925e-01 -6.43929958e-01 -5.81592381e-01 -6.09003529e-02
1.85034931e-01 -4.40919064e-02 9.48849693e-02 -8.45299780e-01
2.57057130e-01 -1.31008163e-01 1.23780541e-01 -1.52619898e+00
5.26066422e-01 -5.34522831e-01 2.70507216e-01 1.27546161e-01
-3.56920242e-01 -2.80687720e-01 4.90681380e-01 6.46242917e-01
-5.24001479e-01 -6.88521340e-02 2.05740392e-01 -2.97760010e-01
-7.30680227e-01 1.10554099e-01 -8.60247433e-01 -2.60432184e-01
9.25895572e-01 -8.41385573e-02 9.21509415e-02 -6.57853246e-01
-1.03376305e+00 2.00745448e-01 2.47441158e-01 5.59017360e-01
2.74931610e-01 -1.13247108e+00 -4.90026206e-01 -4.84774917e-01
-6.11985661e-02 4.34127264e-02 1.96304336e-01 6.12041652e-01
3.53782997e-02 1.11846411e+00 2.93407232e-01 -8.45604420e-01
-1.07108498e+00 6.14590526e-01 -1.77847847e-01 -7.16853380e-01
-8.72431159e-01 5.58152437e-01 6.04072809e-01 1.30449489e-01
3.83533746e-01 -6.80411279e-01 -4.05352503e-01 6.39010966e-01
4.45756853e-01 3.26765537e-01 -8.29977021e-02 -1.34569854e-01
-2.31866404e-01 2.73286730e-01 -3.48189324e-01 -8.74643862e-01
1.49107945e+00 -4.52918380e-01 -1.34376630e-01 1.39271438e+00
8.42862248e-01 -2.27993354e-01 -1.25480676e+00 -1.47326797e-01
6.24085486e-01 -8.16650316e-02 3.15291584e-01 -5.40214956e-01
-3.05036753e-01 9.57995594e-01 1.55847564e-01 1.07523775e+00
1.12347162e+00 2.79907346e-01 7.55352855e-01 -1.62497088e-02
3.52828413e-01 -1.38072002e+00 -1.25619784e-01 4.53091145e-01
7.68849909e-01 -7.72123635e-01 1.87657207e-01 -3.35984766e-01
-5.21059990e-01 1.06226707e+00 2.22041637e-01 2.48336226e-01
9.04150963e-01 5.49476482e-02 -2.89245784e-01 -5.66192508e-01
-1.21701097e+00 -1.75172552e-01 2.33607814e-01 1.70487866e-01
6.48280382e-01 6.21725917e-02 -4.59379435e-01 5.61862528e-01
-4.71092582e-01 -1.32386237e-01 4.82425183e-01 1.06675029e+00
-7.64816225e-01 -1.02155685e+00 -3.13334525e-01 1.97834671e-01
-6.70877814e-01 -4.37725745e-02 -3.31561565e-01 1.11219025e+00
-2.39571154e-01 9.48229849e-01 3.10944855e-01 -1.71464533e-01
-1.16542071e-01 4.87823337e-01 1.25935674e-01 -9.31125522e-01
-3.18884909e-01 7.02225924e-01 3.73553634e-01 -5.56937277e-01
-5.22207499e-01 -9.87272620e-01 -1.08101475e+00 1.29426152e-01
-3.34042192e-01 7.13947654e-01 1.12241173e+00 1.07605648e+00
2.45266721e-01 9.16923046e-01 5.30523658e-01 -9.26037669e-01
5.09251654e-02 -1.33932149e+00 -7.15610147e-01 -1.08122490e-02
1.79769441e-01 -5.96965909e-01 -2.86824495e-01 4.02532965e-01] | [10.371562957763672, 7.026930332183838] |
3517314e-3ccd-4d72-94b3-f1cf050cf4ea | chbias-bias-evaluation-and-mitigation-of | 2305.11262 | null | https://arxiv.org/abs/2305.11262v1 | https://arxiv.org/pdf/2305.11262v1.pdf | CHBias: Bias Evaluation and Mitigation of Chinese Conversational Language Models | \textit{\textbf{\textcolor{red}{Warning}:} This paper contains content that may be offensive or upsetting.} Pretrained conversational agents have been exposed to safety issues, exhibiting a range of stereotypical human biases such as gender bias. However, there are still limited bias categories in current research, and most of them only focus on English. In this paper, we introduce a new Chinese dataset, CHBias, for bias evaluation and mitigation of Chinese conversational language models. Apart from those previous well-explored bias categories, CHBias includes under-explored bias categories, such as ageism and appearance biases, which received less attention. We evaluate two popular pretrained Chinese conversational models, CDial-GPT and EVA2.0, using CHBias. Furthermore, to mitigate different biases, we apply several debiasing methods to the Chinese pretrained models. Experimental results show that these Chinese pretrained models are potentially risky for generating texts that contain social biases, and debiasing methods using the proposed dataset can make response generation less biased while preserving the models' conversational capabilities. | ['Mykola Pechenizkiy', 'Ling Chen', 'Yitong Li', 'Zijing Shi', 'Meng Fang', 'Jiaxu Zhao'] | 2023-05-18 | null | null | null | null | ['response-generation'] | ['natural-language-processing'] | [-1.88409500e-02 3.13696802e-01 -1.76599368e-01 -6.11830175e-01
2.70567685e-02 -4.44866776e-01 9.35925961e-01 -1.68880582e-01
-4.47417766e-01 9.97085273e-01 6.83023453e-01 -3.55724424e-01
4.21576649e-01 -7.80228913e-01 -1.98436767e-01 -7.38354802e-01
5.33625841e-01 3.61603558e-01 -2.53235877e-01 -9.30443108e-01
8.34218144e-01 -5.63777573e-02 -1.08969891e+00 5.52741408e-01
1.31125891e+00 5.22762954e-01 -2.78357685e-01 5.12971878e-01
-4.23506126e-02 1.09908271e+00 -1.23764932e+00 -1.06360614e+00
-3.10759574e-01 -4.97983664e-01 -1.02953815e+00 -6.40489161e-01
4.00222510e-01 -6.90469563e-01 -1.96715757e-01 1.37853038e+00
1.04085922e+00 4.33183163e-02 7.07863748e-01 -1.56851494e+00
-1.29556930e+00 1.06878972e+00 -3.24945986e-01 2.33193651e-01
3.51058424e-01 3.73644739e-01 5.57255030e-01 -6.08737826e-01
3.62077773e-01 1.94113040e+00 6.07492208e-01 1.53342116e+00
-1.03789186e+00 -1.41501129e+00 5.99590242e-01 5.18143289e-02
-6.16446853e-01 -2.51323909e-01 9.47994232e-01 -5.30365527e-01
6.82751417e-01 5.16280949e-01 5.40642083e-01 2.40567136e+00
4.84751552e-01 8.11006427e-01 1.75413620e+00 7.47560486e-02
1.50748804e-01 5.37354052e-01 6.32080615e-01 1.95857927e-01
3.72438282e-01 4.51187670e-01 -7.27703035e-01 -6.44097567e-01
9.91589427e-02 -8.97999480e-02 -2.44960085e-01 4.02933329e-01
-1.22794116e+00 1.12989068e+00 5.11761308e-01 1.96738150e-02
-1.57265142e-01 -1.15878731e-01 6.01147413e-01 3.22315782e-01
9.01549041e-01 5.70539176e-01 -1.25937134e-01 -2.21106216e-01
-2.30979826e-02 6.14888072e-01 8.61235380e-01 1.10933924e+00
6.25544846e-01 2.34298378e-01 -5.39891601e-01 1.09383595e+00
2.61011392e-01 9.54717517e-01 6.85273707e-01 -7.30846882e-01
5.48198164e-01 6.58897460e-01 1.99966401e-01 -1.59244585e+00
-4.49983060e-01 -2.44492784e-01 -1.13521850e+00 -1.22013874e-01
2.50393778e-01 -7.25126207e-01 -4.40765709e-01 1.99675333e+00
1.25618786e-01 -6.23921275e-01 2.73715854e-01 8.96478176e-01
1.15125763e+00 6.75491214e-01 4.51900721e-01 -4.06967849e-02
1.18839741e+00 -8.74226034e-01 -9.35842752e-01 -6.34313345e-01
6.31635785e-01 -8.01063240e-01 1.70487297e+00 3.94684821e-01
-8.25276792e-01 -2.84889817e-01 -8.82275999e-01 4.00089705e-03
-4.93905395e-01 -3.20572346e-01 5.74842215e-01 1.27152288e+00
-7.93131948e-01 1.37775719e-01 1.58491239e-01 -3.54283273e-01
2.58433193e-01 -3.73086333e-02 2.33640909e-01 8.24504718e-02
-1.94786263e+00 1.29331064e+00 4.51506637e-02 1.34355903e-01
-1.11354363e+00 -6.03172839e-01 -5.95570683e-01 -2.80644536e-01
4.10951257e-01 -5.19464672e-01 1.37581658e+00 -1.25700390e+00
-1.61530173e+00 9.49301422e-01 4.92788963e-02 -2.56540060e-01
8.07972908e-01 -4.83108908e-01 -5.21530628e-01 -3.70607108e-01
-4.67911959e-02 8.26857924e-01 8.40042353e-01 -1.43780112e+00
-2.57630765e-01 -3.02982301e-01 3.15700233e-01 1.86133251e-01
-7.48565912e-01 4.49128449e-01 4.80488867e-01 -1.07607067e+00
-6.12390161e-01 -1.13910913e+00 -2.02828515e-02 -4.29175705e-01
-7.26704180e-01 -4.19128537e-01 6.57274306e-01 -7.48295426e-01
1.44470894e+00 -1.63250339e+00 -8.67034271e-02 7.26535171e-02
1.74032167e-01 3.10503602e-01 3.04262973e-02 4.97643858e-01
1.60741180e-01 6.44126832e-01 -3.86482738e-02 -6.64296821e-02
1.98435262e-01 3.20426524e-02 -5.79328179e-01 2.50371873e-01
-1.18294924e-01 7.14007139e-01 -1.01886725e+00 -5.50780416e-01
-1.61807299e-01 2.70172119e-01 -9.24141884e-01 3.95761758e-01
-1.54761955e-01 3.19499016e-01 -4.25210476e-01 5.03316998e-01
8.95839095e-01 3.62669647e-01 1.92315858e-02 1.37163550e-01
2.50995513e-02 5.31074941e-01 -3.29855680e-01 9.96365905e-01
-2.00090885e-01 5.56810021e-01 6.68237638e-03 -3.24766040e-01
1.20150173e+00 -6.49938434e-02 -1.99661911e-01 -6.01162195e-01
6.28858864e-01 -7.72709493e-04 2.11496547e-01 -6.00671291e-01
9.03589249e-01 -2.42963076e-01 -4.07979935e-01 6.43731356e-01
-5.73764920e-01 -3.81133586e-01 -1.13592647e-01 4.86987650e-01
3.51005256e-01 -3.72601867e-01 -1.11493923e-01 -7.82354116e-01
5.36633134e-01 -1.51630351e-02 6.25291526e-01 6.74193799e-01
-6.96093500e-01 1.88241780e-01 9.19113934e-01 -4.11227405e-01
-7.34806597e-01 -4.92678314e-01 2.15996221e-01 1.71562493e+00
2.54012048e-01 -2.16015667e-01 -1.19179285e+00 -9.93915796e-01
-1.80464357e-01 1.26580369e+00 -9.79116261e-01 -7.63715744e-01
-4.61620361e-01 -1.14263499e+00 9.72696185e-01 2.38914609e-01
8.92367005e-01 -1.30819595e+00 -3.51314694e-01 -1.84078619e-01
-6.69030190e-01 -5.29521406e-01 -7.49084771e-01 -4.54510003e-01
-4.07278836e-01 -9.19508100e-01 -7.09757507e-01 -5.74449062e-01
6.97453856e-01 3.34217578e-01 1.13480306e+00 3.83252501e-01
4.11283761e-01 5.65354899e-02 -3.88082862e-01 -1.21574175e+00
-8.69820535e-01 4.21153046e-02 1.86589777e-01 -1.74114466e-01
7.44162917e-01 6.78427517e-02 -6.89433634e-01 7.95239091e-01
-5.88887215e-01 2.00080305e-01 1.71614319e-01 9.41300035e-01
-6.85037374e-01 -6.84257329e-01 8.80952358e-01 -1.19528890e+00
1.44000328e+00 -4.99927580e-01 3.49159241e-02 9.18930769e-03
-7.50398159e-01 -3.47697109e-01 4.39816266e-01 -8.82431209e-01
-1.74639344e+00 -1.16716897e+00 -1.82159558e-01 3.15613836e-01
-1.20142512e-01 1.12842888e-01 -3.25489670e-01 1.92616835e-01
1.07942533e+00 -3.40782013e-03 3.05621296e-01 -1.31651968e-01
7.27167726e-02 1.09823561e+00 9.27790441e-03 -8.86068225e-01
3.37788701e-01 4.04631495e-01 -8.20374370e-01 -6.33537054e-01
-6.93178892e-01 3.47120345e-01 1.04137488e-01 -6.68445110e-01
8.57313812e-01 -7.60956109e-01 -9.16124225e-01 1.06113517e+00
-1.41293955e+00 -5.69025040e-01 5.34649849e-01 1.78124189e-01
-8.62099081e-02 3.60296607e-01 -8.76531303e-01 -9.44366992e-01
-6.03477538e-01 -9.19601262e-01 4.28308547e-01 3.00701529e-01
-9.05170083e-01 -8.44414353e-01 1.65807039e-01 6.47774756e-01
7.39768863e-01 9.13738385e-02 1.17584884e+00 -1.00081909e+00
2.94805765e-01 5.23922034e-02 -1.23991556e-01 6.45552516e-01
2.67528743e-02 5.68125360e-02 -1.18976724e+00 -3.54554772e-01
1.09910905e-01 -6.40755594e-01 6.30086184e-01 -1.90663058e-02
1.11291802e+00 -9.58930254e-01 -2.80803978e-01 8.73629898e-02
4.64167833e-01 5.01719117e-01 6.87905312e-01 3.63490820e-01
6.60453320e-01 1.25974524e+00 7.21346855e-01 3.48666012e-01
7.51220882e-01 8.87072831e-02 5.43092132e-01 -7.24637657e-02
3.80744278e-01 -3.42251569e-01 8.68429542e-01 7.18835294e-01
-1.76115576e-02 -4.45066571e-01 -8.64878297e-01 4.26096827e-01
-1.69357979e+00 -1.11319327e+00 -1.94643989e-01 1.65955532e+00
1.06841695e+00 1.99291959e-01 7.87426755e-02 -2.82704294e-01
1.07084250e+00 5.48882961e-01 -5.56873441e-01 -9.89892244e-01
-3.58534127e-01 -4.35938567e-01 5.77374063e-02 4.73110080e-01
-7.75028110e-01 1.07383275e+00 6.51545095e+00 4.54391330e-01
-1.28646886e+00 7.43857352e-03 1.17665291e+00 -1.35310426e-01
-7.10029364e-01 -1.93483770e-01 -7.05612242e-01 7.80612409e-01
5.08032501e-01 -3.84737998e-01 2.13956848e-01 1.19694412e+00
2.25643218e-01 8.49339068e-02 -9.89261389e-01 8.25952888e-01
5.04507244e-01 -6.97521985e-01 3.67394447e-01 -1.14332125e-01
8.22372675e-01 -4.09980446e-01 3.85501325e-01 8.78445625e-01
8.28317165e-01 -1.11616719e+00 7.53437638e-01 2.09847748e-01
2.28095978e-01 -8.33963692e-01 7.91686237e-01 4.54891592e-01
1.46939650e-01 -2.04665318e-01 -7.26481795e-01 -5.14205873e-01
-2.12728679e-01 3.14409435e-01 -7.06230581e-01 -9.20013897e-03
9.98691082e-01 4.44471210e-01 -6.47947013e-01 1.82782248e-01
-2.97427207e-01 6.78638339e-01 3.11905682e-01 -9.64154661e-01
1.99683100e-01 -1.25919566e-01 5.47093332e-01 1.40122795e+00
8.71280655e-02 2.46221736e-01 -1.56147078e-01 9.04002726e-01
-7.58071467e-02 1.98644996e-01 -7.95857191e-01 1.50326744e-01
6.29439652e-01 9.97911274e-01 4.34391238e-02 -5.97024560e-01
-7.27287903e-02 4.27530795e-01 2.83200443e-01 3.77186537e-01
-9.53206062e-01 -2.18960375e-01 9.83030021e-01 -2.94768810e-01
-7.23595500e-01 2.39677042e-01 -6.76689625e-01 -1.16705036e+00
-4.60356563e-01 -1.80490291e+00 4.38737601e-01 -8.35258126e-01
-1.68120944e+00 6.52700245e-01 2.86886245e-01 -7.75436819e-01
-1.08354151e-01 -5.53859115e-01 -9.03604150e-01 9.20042753e-01
-1.17559946e+00 -1.01050496e+00 -5.30564487e-01 5.43391168e-01
7.52050638e-01 -4.26248699e-01 4.95604217e-01 2.67381016e-02
-8.57192576e-01 7.26589501e-01 -5.10171652e-01 1.14938475e-01
1.39369261e+00 -1.05373383e+00 3.01263422e-01 4.86400068e-01
-1.05888295e+00 1.16561460e+00 9.94998515e-01 -9.28543031e-01
-8.31856370e-01 -8.45003307e-01 8.91999483e-01 -6.56834424e-01
5.36419570e-01 -5.18029511e-01 -9.72419441e-01 8.04353297e-01
9.36953902e-01 -1.10147548e+00 8.55441630e-01 2.27884024e-01
-4.96765465e-01 1.19982399e-02 -1.19503820e+00 1.26357281e+00
1.14254379e+00 -3.54305476e-01 -9.05376911e-01 2.24118382e-01
7.67094076e-01 -2.61240512e-01 -2.55684197e-01 2.00137615e-01
4.58082587e-01 -1.18732178e+00 7.29550004e-01 -1.00224924e+00
7.20683336e-01 3.90442014e-01 1.72248721e-01 -1.97498035e+00
-3.94390166e-01 -8.82237673e-01 4.40547556e-01 1.52369499e+00
4.43540961e-01 -1.12423480e+00 5.94007432e-01 9.76535082e-01
-8.06102827e-02 -3.19926739e-01 -4.74858761e-01 -2.94034183e-01
7.49108374e-01 -1.68230280e-01 1.00076067e+00 1.50010824e+00
3.18849325e-01 4.87924725e-01 -9.24489558e-01 -1.26755431e-01
4.60106969e-01 -2.16124281e-01 1.29051614e+00 -9.02113914e-01
3.95550579e-01 -7.58955598e-01 5.76152921e-01 -9.09204483e-01
4.06816155e-01 -5.60155749e-01 2.35398374e-02 -9.61421072e-01
3.23157132e-01 -3.22846502e-01 1.07874222e-01 2.34955117e-01
-7.75991261e-01 -1.90809712e-01 8.01566914e-02 -6.49836287e-02
4.69084606e-02 7.48181820e-01 1.37838304e+00 -3.98957878e-01
1.14807136e-01 -1.27431780e-01 -1.41911972e+00 9.92492020e-01
1.06238174e+00 -4.34356332e-01 -4.25466210e-01 -5.18528700e-01
5.16915262e-01 -5.01675904e-01 4.23965394e-01 -3.29035103e-01
-8.93147476e-03 -6.97427511e-01 1.07102633e-01 -4.71259594e-01
1.40342399e-01 -3.24623913e-01 -4.17437762e-01 7.57782102e-01
-7.77554750e-01 4.47041392e-01 1.06708601e-01 3.20836753e-01
6.02099784e-02 -8.41572285e-02 8.20336461e-01 -2.61457413e-01
-2.60576308e-01 -4.16157506e-02 -8.29606712e-01 3.29596281e-01
7.62110651e-01 1.21460646e-01 -1.21414185e+00 -8.18531573e-01
-3.05451546e-02 6.38113022e-01 2.67352521e-01 9.52470958e-01
4.73746061e-01 -1.30907571e+00 -1.03142631e+00 -2.19521418e-01
2.16976628e-01 -2.60000020e-01 6.22082293e-01 6.08272314e-01
-3.23413700e-01 1.87648043e-01 -5.67511797e-01 -2.98958570e-01
-1.44905877e+00 7.11180389e-01 4.44903404e-01 6.40087798e-02
1.51015773e-01 9.24958050e-01 7.62996912e-01 -8.83902073e-01
3.16597462e-01 -4.39432770e-01 -4.68323827e-01 2.43339702e-01
6.84209108e-01 8.06126893e-01 -2.97995239e-01 -5.29291153e-01
-3.58432353e-01 1.69810485e-02 -3.69820923e-01 -9.57525522e-03
6.80761516e-01 -1.86812043e-01 -4.01763856e-01 2.41816178e-01
7.17489541e-01 2.05848798e-01 -9.62921858e-01 -7.58928359e-02
-1.27865165e-01 -6.49962068e-01 -4.51692194e-01 -1.19512689e+00
-6.71708286e-01 9.94463861e-01 2.41654485e-01 3.25093120e-01
5.99957645e-01 -5.39392769e-01 5.79781830e-01 4.14996654e-01
2.97146380e-01 -1.52159178e+00 5.44492841e-01 8.46310914e-01
1.50907969e+00 -1.31558764e+00 -1.54494643e-01 -3.56597126e-01
-1.41310561e+00 5.31307638e-01 1.48426628e+00 8.48738104e-03
3.09582800e-01 -1.39095098e-01 6.75005496e-01 -5.59745505e-02
-1.11211038e+00 4.69898105e-01 -3.48984846e-03 8.68730247e-01
4.88714457e-01 1.03319086e-01 -7.60305583e-01 9.35353100e-01
-7.71536648e-01 -5.17573059e-01 8.94880295e-01 7.29114771e-01
-2.59355426e-01 -7.46035159e-01 -7.10439265e-01 9.13290605e-02
-4.83306170e-01 -2.62718439e-01 -1.36243236e+00 7.48062849e-01
-3.80385816e-02 1.38322592e+00 -6.75085112e-02 -7.25816309e-01
4.63056207e-01 1.05651349e-01 -1.57795355e-01 -2.52407491e-01
-1.28661454e+00 -3.70294124e-01 7.91968465e-01 -1.97047070e-01
-3.08283806e-01 -4.07919317e-01 -6.45826161e-01 -1.09465766e+00
-2.22318456e-01 1.57889009e-01 2.22344592e-01 6.14885986e-01
1.67103618e-01 2.73253381e-01 7.03143775e-01 -6.96255684e-01
-8.00579131e-01 -1.54532158e+00 2.49809604e-02 7.49140143e-01
1.32649824e-01 -6.01069629e-01 -4.94312137e-01 -4.52125579e-01] | [9.1924467086792, 10.245473861694336] |
688d25e9-62be-4240-a911-b1620d01d98c | ba-net-dense-bundle-adjustment-network | 1806.04807 | null | https://arxiv.org/abs/1806.04807v3 | https://arxiv.org/pdf/1806.04807v3.pdf | BA-Net: Dense Bundle Adjustment Network | This paper introduces a network architecture to solve the structure-from-motion (SfM) problem via feature-metric bundle adjustment (BA), which explicitly enforces multi-view geometry constraints in the form of feature-metric error. The whole pipeline is differentiable so that the network can learn suitable features that make the BA problem more tractable. Furthermore, this work introduces a novel depth parameterization to recover dense per-pixel depth. The network first generates several basis depth maps according to the input image and optimizes the final depth as a linear combination of these basis depth maps via feature-metric BA. The basis depth maps generator is also learned via end-to-end training. The whole system nicely combines domain knowledge (i.e. hard-coded multi-view geometry constraints) and deep learning (i.e. feature learning and basis depth maps learning) to address the challenging dense SfM problem. Experiments on large scale real data prove the success of the proposed method. | ['Chengzhou Tang', 'Ping Tan'] | 2018-06-13 | null | null | null | null | ['depth-and-camera-motion'] | ['computer-vision'] | [ 0.00583111 0.33326465 0.10432972 -0.70674634 -0.845899 -0.3932679
0.5960693 -0.39422062 -0.38360098 0.3168311 0.23172316 0.14866479
0.08201252 -1.0263184 -0.9454983 -0.6004654 0.21335925 0.5290956
0.2515009 -0.18724403 0.50399023 0.5983545 -1.5412257 0.14118655
0.71549904 1.2946031 0.5288773 0.68518114 0.07046542 0.4883203
-0.13658923 -0.04524947 0.7633822 -0.04500242 -0.841463 0.40444586
0.9084408 -0.78637075 -0.45011395 0.7847311 0.36501116 0.1400822
0.4009458 -0.9776975 -0.3660486 -0.20888634 -0.6213354 0.02380518
0.36948413 0.18247879 1.1052973 -1.13952 0.72792834 1.2375515
0.4508184 0.33755752 -0.8542609 -0.15626614 0.2715154 0.07247399
-1.1169897 -0.1481105 0.9466708 -0.5489045 1.1188699 -0.15095165
0.86033684 0.6914872 0.47038016 0.67161345 0.8117777 -0.31604603
0.07369952 -0.23836035 -0.36289364 1.1012834 -0.21983507 0.37680498
-0.54809284 0.13033931 1.3628287 0.20308015 -0.23523979 -0.87640333
-1.4118959 1.0047002 0.79177135 -0.2415657 -0.20781499 0.14842413
0.02165825 0.05488922 0.34638304 0.43423447 -0.66916645 0.10553133
-0.85276985 0.30149436 0.35803995 1.1283756 1.5266714 0.04419208
0.21829322 0.5664466 0.6905331 0.55441475 0.30941314 -1.4199198
0.6858304 0.7336597 0.01109203 -0.925321 -0.58744055 -0.24224438
-0.771867 0.56430125 0.41986406 -0.08683103 -1.0517424 1.5206354
0.6967827 0.10678848 -0.11879598 1.1907831 0.6338887 0.528538
-0.6421765 0.21827807 0.98533225 -1.2657249 -0.21217883 -0.45433664
0.53545135 -0.55204725 1.0031703 0.41057336 -1.311088 -0.49384204
-1.2037419 -0.58850896 -0.05825371 0.08391939 0.7374342 0.1350646
-1.1066625 0.60463506 -1.0245786 -0.1371082 0.3565924 0.50918627
-0.59962606 -0.19506937 -0.73773384 0.57839924 0.16854383 0.38656265
-1.1213163 -0.70114934 -1.3208606 -0.23548494 0.2231673 -1.3790073
1.0827667 -0.8906311 -1.782866 1.1140634 -0.14311709 0.0261852
0.6852518 -0.38293067 0.33186874 0.52565354 0.26628047 0.8202323
0.8917514 -1.1105719 -0.87967384 -0.66808915 0.38683838 0.43772036
-0.04927899 -0.5215059 -0.43584132 -0.3246377 0.64909816 -0.7044784
-0.42399877 0.29358876 -0.3518749 0.06694192 0.47849846 -0.6141746
0.6252976 -1.9074341 0.58295804 0.26791707 0.39815378 -0.4857697
-0.1603289 0.19193129 0.15338045 -0.31491962 -0.3564362 -0.62920696
-0.20453349 0.31834617 0.06233702 0.8096637 0.31263983 0.86305594
-0.9459724 -0.26258036 0.63524085 0.508961 -0.90958685 0.5509625
-0.24941777 0.68582296 -0.40576792 0.60722685 0.79913837 -0.21962632
-0.33262163 -0.3305968 -0.25193742 0.11225141 -1.1577444 2.5345793
-0.9145688 0.32915133 0.14454167 -0.7091237 1.0448016 -0.14399697
0.6114662 -0.4567506 0.14471497 0.18161988 -0.3802836 -0.41890168
0.40339777 -0.11733837 0.1068856 0.13106272 0.22328827 -0.553414
-0.34234062 0.08575927 0.8763986 0.5545995 0.19376385 -0.13411812
0.7450687 -0.15021423 0.6799601 0.21701138 0.13904154 1.1194229
0.297234 -0.7806172 -1.2602148 -1.15104 -0.00983464 0.50255406
0.35391957 -0.27681342 -0.66961044 -0.647181 -0.09859367 0.03914318
-0.71650183 -0.03239829 -0.5500578 -0.3309061 -0.15048748 0.6566748
0.7034152 -0.83019114 -0.7719899 0.10782597 -0.02536377 -1.1306744
-0.6581388 0.11240117 -1.0326449 -1.0702403 -0.55345774 -0.81360525
0.8138556 0.44834965 1.0077858 -0.00825027 -0.2884107 0.37756094
-0.1657599 0.20917821 -0.07320219 0.14340757 -0.16425271 0.21270102
-0.01039097 -0.9427593 -1.0186932 0.32053244 -0.8492381 0.1998805
0.57698804 0.7680497 1.080055 -0.22655989 0.03599717 -0.5952501
-0.20967667 -0.304685 -0.91385084 -0.05123995 -0.53790206 0.09662816
0.44308457 0.06526259 -0.80412245 0.6041447 -0.2936857 -0.63881063
-0.06820978 0.22120367 -0.58196086 -0.22781107 0.41188347 0.28788614
0.04769737 -0.44906282 0.5451633 0.3281109 0.64325136 -0.36277726
0.7848843 0.9888732 0.22953154 -0.4709095 -1.1273289 -0.57378954
-1.2757677 -0.11054437 1.0634427 -1.3234544 -0.5143386 0.5933456
-1.2388963 -0.4715604 -0.19684955 0.5156395 -1.1313167 0.41070047
-0.45607847 -0.25277466 -0.23145014 -1.2788403 1.6525468 0.09014504
0.16327605 -1.1186441 -0.01437082 0.41061056 0.01171244 0.61131895
0.54382217 0.29030868 -1.2341874 0.04815369 -0.10032737 0.4146272
-0.09695549 0.02913163 -1.0184224 -0.38334733 0.2701003 -0.22027218
0.9054323 0.65445393 1.0076426 0.01058126 -0.03183639 1.604702
1.7005063 -0.32063523 0.79127145 0.77067894 1.2361468 0.70088327
0.78876555 0.618947 0.80796593 0.71836835 0.91096073 -0.16334794
-0.05065443 -0.32688168 0.33908018 0.612156 0.0872202 0.2686173
-0.734056 0.3827046 -1.7909657 -0.730892 -0.20724997 2.008004
0.47460386 0.0566726 -0.0062412 -0.08994213 0.27796733 0.19329901
-0.6936785 -0.26813903 -0.13846941 0.02030707 0.4824485 0.9901128
-1.0893997 0.9937919 5.4134946 0.35829845 -1.208988 0.04907325
0.5296247 -0.0769325 -0.60201097 0.00883973 -0.9706154 0.07589379
0.42640394 0.1851039 0.47588116 0.99388266 0.2223791 -0.26657432
-1.2047112 1.2964637 0.1554857 -1.4939934 -0.10545238 0.2965797
0.9129821 0.54217523 -0.08480833 -0.15097001 0.07564537 -0.76428044
0.96991664 0.49396893 0.82906294 -0.8371814 0.49880314 0.47656548
-1.1702317 -0.1632776 -0.6417356 -0.2028122 0.33068904 0.87125415
-0.56625855 0.6446215 0.62761885 1.2580489 -0.5089343 0.88483256
-0.3606724 -0.2968767 -0.41691658 0.70831466 0.374434 -0.4443603
0.37165874 0.836617 0.3743506 -0.28450933 0.19173068 0.98818135
0.19836272 -0.06479714 -0.7469019 0.62494147 0.1581734 1.6189865
-0.6631917 -0.04612767 -0.5115867 1.2903299 0.6104776 0.22954935
-0.4717137 -0.09130111 0.72978294 0.4594245 0.48452228 -0.47164494
-0.4418951 -1.3759314 0.29419616 -0.40217385 0.1319596 -0.72692627
-1.2675561 0.69245917 -0.21703348 -1.50795 -0.4994506 -0.6196345
-0.5427122 1.0101566 -1.878655 -1.3309903 -0.7562896 0.7271037
0.5438513 -0.03821956 0.5511792 0.11415268 -0.34811816 0.3628742
-0.10363762 -0.08347806 0.48138544 -1.4113399 0.6883036 0.7116403
-0.04350987 0.26168215 0.07427038 -0.40047795 -1.5482681 -1.2442615
0.6014832 -0.7535926 0.44813472 -0.5336217 -0.5566427 0.63575184
-0.14407514 0.3877952 0.25046894 -0.29373664 -0.3817926 -0.11069879
-1.1669717 0.11990479 1.2021893 -0.62537766 -0.2187783 0.3052273
0.7835826 -0.8870596 -0.9914104 0.33099675 0.44615138 -1.3740575
1.085637 -0.28512645 0.7932803 -0.516365 -0.39892855 -1.2573161
-0.3160889 -0.6158045 -0.24555059 0.8709171 0.20191659 -0.40076125
0.9436658 0.41190097 -0.5737522 -1.2120981 -1.0179304 -0.49838296
0.08026435 -0.26770517 0.57135916 0.74595267 -0.4419007 0.38607726
-0.1731258 0.50340074 0.6899032 0.24074857 1.0671589 -1.0510951
-0.58888185 -0.21211816 -0.70511055 -1.5614096 0.12939388 -0.98178875
0.07454657 -1.7182024 -0.07655518 -0.26980406 0.20012489 0.08035179
-0.07408863 0.07999588 -0.03194118 0.03673063 -0.30163538 0.7985206
1.7252421 0.22782965 -0.208144 -0.07165721 -0.4912959 0.8393207
0.47715282 -0.21043791 -0.48803982 -0.84808624 0.3006376 0.2547563
0.5287947 -0.9030877 0.09596799 -0.27500278 0.46211576 -0.8172238
0.5494251 -0.8336088 -0.2229224 0.1068415 0.07834227 0.10594516
-0.20648125 0.673891 -0.26357046 0.03433133 0.8393636 -0.28865242
-0.839612 0.7764601 0.37188676 0.12857208 1.0099591 -0.5661261
0.06771351 -0.35220268 -0.8094376 0.36038703 0.81562185 0.27162027
0.9991845 -1.4489377 -0.63526475 0.68761975 0.21661295 1.0875709
0.2807505 0.7174046 -0.89637274 0.11126702 -0.24949102 -1.0100746
-0.645423 0.30589157 0.68645287 -0.1865461 -0.9573217 1.1008879
0.62762934 -0.66223466 0.19740269 -0.41279718 0.20538397 -0.29736772
0.57948124 0.20959818 0.05069179 -0.6872572 -0.13280155 1.1108779
-0.03620341 -0.22996283 1.7338237 -0.42211056 -0.1907417 0.23206964
1.6728193 -0.319124 -2.0346808 -0.16713077 -0.30051455 -0.82738215
0.31685868 -0.28096613 -1.3250972 1.1672162 0.44346607 -0.5683995
1.0444375 -0.09850227 0.9622993 0.31579673 0.53346664 -0.96045226
0.3683537 0.6964274 1.028041 -1.3411794 -0.00549275 -0.39773235
-0.26996055 1.3235765 0.9150543 -0.6367645 0.8432938 0.04945795
0.14084044 -0.43026227 -0.4035824 -0.0139512 0.20251273 0.6796979
0.06012674 -0.23002948 0.18923421 0.17423275 -0.45840698 -0.15703137
0.5635008 0.5948407 -0.4395429 -0.9848676 -0.23724177 0.2104344
0.06640911 0.03573812 -0.03190216 0.612996 0.29668188 0.5766775
0.27491575 -0.5150436 0.50586826 -0.37154615 0.78207964 -0.87157995
-0.39219037 0.16194318 -0.4208508 -1.1798964 -0.4189136 -0.62063456
-1.4739821 -0.17101717 -0.06661922 -0.47527525 0.78993094 0.8979037
0.35716745 0.24287975 1.1404431 -1.3400595 -0.44928524 -0.7390491
-0.7027709 0.47446346 0.625734 -0.748219 -0.44036168 -0.04527497] | [8.601398468017578, -2.6735174655914307] |
c473d621-639e-4ccd-aaef-8c5788dfaf13 | multivariate-time-series-classification-with-1 | 2010.05649 | null | https://arxiv.org/abs/2010.05649v2 | https://arxiv.org/pdf/2010.05649v2.pdf | Multivariate Time Series Classification with Hierarchical Variational Graph Pooling | With the advancement of sensing technology, multivariate time series classification (MTSC) has recently received considerable attention. Existing deep learning-based MTSC techniques, which mostly rely on convolutional or recurrent neural networks, are primarily concerned with the temporal dependency of single time series. As a result, they struggle to express pairwise dependencies among multivariate variables directly. Furthermore, current spatial-temporal modeling (e.g., graph classification) methodologies based on Graph Neural Networks (GNNs) are inherently flat and cannot aggregate hub data in a hierarchical manner. To address these limitations, we propose a novel graph pooling-based framework MTPool to obtain the expressive global representation of MTS. We first convert MTS slices to graphs by utilizing interactions of variables via graph structure learning module and attain the spatial-temporal graph node features via temporal convolutional module. To get global graph-level representation, we design an "encoder-decoder" based variational graph pooling module for creating adaptive centroids for cluster assignments. Then we combine GNNs and our proposed variational graph pooling layers for joint graph representation learning and graph coarsening, after which the graph is progressively coarsened to one node. At last, a differentiable classifier takes this coarsened representation to get the final predicted class. Experiments on ten benchmark datasets exhibit MTPool outperforms state-of-the-art strategies in the MTSC task. | ['Zhongbin Xu', 'Ziheng Duan', 'Wei Wang', 'Yizhou Sun', 'Yueyang Wang', 'Anni Ren', 'Yida Huang', 'Haoyan Xu'] | 2020-10-12 | null | null | null | null | ['graph-structure-learning'] | ['graphs'] | [ 5.65470904e-02 -5.09914458e-02 -1.15010403e-01 -2.61005223e-01
-5.21376014e-01 -2.71292090e-01 5.11033058e-01 4.12678719e-01
5.65879159e-02 4.59424317e-01 9.33029503e-02 -2.92471170e-01
-1.82168931e-01 -1.00729787e+00 -7.82479525e-01 -8.07728052e-01
-4.94319409e-01 1.33999540e-02 1.86121151e-01 -1.93644818e-02
-4.90181856e-02 4.04749066e-01 -9.86940980e-01 7.34593645e-02
8.99739802e-01 1.03095639e+00 1.75260723e-01 6.48866594e-01
-2.97090083e-01 8.96612704e-01 -4.47496146e-01 -8.04591700e-02
-8.90742913e-02 -6.05376124e-01 -4.71707255e-01 1.49955720e-01
-4.58216667e-02 2.88546771e-01 -6.86956465e-01 1.02822804e+00
4.05964166e-01 3.89017880e-01 5.35127640e-01 -1.17860258e+00
-1.26002741e+00 9.69129384e-01 -7.23039687e-01 1.68018728e-01
1.02490723e-01 2.24403385e-02 1.15321398e+00 -6.30912304e-01
3.84385407e-01 1.29001224e+00 7.63287723e-01 3.17594290e-01
-1.18833339e+00 -4.69481081e-01 7.62237191e-01 3.42131793e-01
-1.53436887e+00 -5.05178869e-02 1.34772575e+00 -3.89452845e-01
1.19223249e+00 2.11406022e-01 8.19498479e-01 9.52515423e-01
4.75848675e-01 5.82462251e-01 6.98991239e-01 2.49175392e-02
9.66026708e-02 -6.77263021e-01 -8.61515775e-02 9.63566780e-01
-9.43155140e-02 -4.18023348e-01 -4.45415407e-01 1.32900625e-01
8.08392763e-01 4.32424664e-01 -1.67154506e-01 -4.72536199e-02
-1.38789499e+00 7.59569407e-01 1.19188356e+00 4.13241088e-01
-6.56063378e-01 5.94677389e-01 3.62758428e-01 1.55071110e-01
7.38879859e-01 1.89175218e-01 -2.55629301e-01 2.52072990e-01
-7.07149863e-01 -3.83449122e-02 4.19745982e-01 9.31481063e-01
1.00148785e+00 3.08607161e-01 -3.73370320e-01 7.03970551e-01
4.13388878e-01 3.33767354e-01 5.65710187e-01 -3.31002146e-01
7.03554094e-01 1.15184486e+00 -6.05662763e-01 -1.47848523e+00
-6.70875907e-01 -8.37010801e-01 -1.40107667e+00 -3.73178005e-01
-2.39946470e-01 -2.89693654e-01 -1.14684081e+00 1.62884569e+00
1.77061126e-01 6.86877251e-01 -8.46939981e-02 7.93129385e-01
8.66677999e-01 8.56997371e-01 1.25299618e-01 -3.03286165e-01
8.72434199e-01 -8.76600385e-01 -7.40879118e-01 -9.94391292e-02
5.93331099e-01 -2.22626209e-01 8.03510606e-01 -7.03629851e-02
-6.02091134e-01 -5.08135021e-01 -1.06952691e+00 -9.81322899e-02
-6.50231481e-01 -1.42552778e-01 7.22079456e-01 1.22085132e-01
-1.23214936e+00 6.92649066e-01 -1.25003362e+00 -2.77607232e-01
5.42313755e-01 4.52110767e-01 -1.90243557e-01 8.78226310e-02
-1.19127238e+00 4.08547163e-01 3.17446351e-01 5.69107652e-01
-7.82133996e-01 -5.01652241e-01 -1.11602247e+00 1.44843295e-01
2.57239014e-01 -5.72270930e-01 5.71215689e-01 -6.83172286e-01
-1.41350842e+00 2.16643482e-01 -2.08411515e-01 -5.71926296e-01
8.63608252e-03 3.04319203e-01 -6.20655119e-01 -9.24154222e-02
1.82568654e-01 4.11635488e-01 9.25102592e-01 -9.15571094e-01
-4.02731627e-01 -5.90537608e-01 -6.68090582e-02 1.51691824e-01
-2.84980595e-01 -3.76754135e-01 -5.19383729e-01 -8.93578768e-01
4.19454664e-01 -7.79802620e-01 -6.10577524e-01 -5.26868045e-01
-6.05211139e-01 -4.39161748e-01 1.12966716e+00 -6.01918399e-01
1.70786238e+00 -2.02576637e+00 5.53884506e-01 2.31436029e-01
6.05613530e-01 -1.03538826e-01 -1.11575864e-01 4.31055456e-01
-1.52383283e-01 3.19634527e-01 -3.55864733e-01 -6.54291809e-01
-3.95361036e-02 2.86282271e-01 -2.57772636e-02 6.56359613e-01
4.33862388e-01 1.25509477e+00 -1.00107765e+00 -5.69785357e-01
4.03865784e-01 5.45956910e-01 -5.18934965e-01 8.75595510e-02
-4.24612731e-01 5.55359662e-01 -7.79792070e-01 5.69194555e-01
5.31662703e-01 -6.58851624e-01 2.56332815e-01 -3.05904180e-01
-1.46390274e-01 -1.26092106e-01 -7.28001535e-01 1.94060326e+00
-3.36127013e-01 4.66922909e-01 -2.84225076e-01 -1.52276957e+00
1.21002197e+00 1.87225059e-01 8.20855796e-01 -5.48169076e-01
4.13424894e-02 -5.26535325e-03 -1.71204209e-01 -3.76524746e-01
3.37281197e-01 2.51293276e-02 -2.47763857e-01 1.16589898e-02
2.39863083e-01 1.77559063e-01 -1.53334945e-01 1.69869781e-01
1.18991983e+00 5.78610897e-02 2.74309330e-02 -1.13344744e-01
6.39535606e-01 -2.12813422e-01 7.08208203e-01 2.73292512e-01
-1.28551647e-01 6.41176403e-01 6.44915521e-01 -4.10956562e-01
-6.03168368e-01 -9.91694927e-01 4.82942820e-01 9.67538357e-01
1.78350002e-01 -6.55369580e-01 -4.92987663e-01 -5.58698118e-01
-9.21793133e-02 2.18833745e-01 -8.71126056e-01 -2.35562935e-01
-5.86844325e-01 -7.51691341e-01 3.30540866e-01 5.39929450e-01
5.86869001e-01 -1.01682603e+00 -9.11122710e-02 5.52149594e-01
-7.73581192e-02 -1.05085337e+00 -5.39310277e-01 1.15075320e-01
-7.67504573e-01 -9.03555214e-01 -6.51264846e-01 -9.42946851e-01
5.85135698e-01 1.62493140e-01 7.56931245e-01 2.16182634e-01
-1.51085760e-02 1.73673064e-01 -5.31955302e-01 -8.38466063e-02
2.06282705e-01 5.16106486e-01 -3.27219844e-01 5.09981394e-01
1.03464290e-01 -1.01626825e+00 -7.93598950e-01 -8.70726854e-02
-7.57790267e-01 1.47828355e-01 4.93081629e-01 6.50101960e-01
8.73848796e-01 2.16926381e-01 7.93078840e-01 -7.86561847e-01
7.71980584e-01 -6.86069489e-01 -5.65514863e-01 4.24270689e-01
-3.96701008e-01 1.56529486e-01 1.05658603e+00 -3.79643619e-01
-6.63878083e-01 9.53864083e-02 1.75051317e-01 -9.84247506e-01
2.65035659e-01 1.37521172e+00 -1.00308530e-01 -8.96306057e-03
4.73803073e-01 3.44434768e-01 -8.45696107e-02 -3.43459189e-01
5.11484921e-01 2.84345448e-01 5.19280851e-01 -3.38065505e-01
6.58927679e-01 1.95247576e-01 2.71264672e-01 -7.52841651e-01
-6.08772457e-01 -2.32899353e-01 -6.19425833e-01 -4.60281074e-01
1.32165396e+00 -8.53348553e-01 -7.43764341e-01 5.43792069e-01
-1.06056118e+00 -5.95221221e-01 5.19117638e-02 4.81080920e-01
-2.74194211e-01 2.05242783e-01 -5.19935369e-01 -7.23368824e-01
-2.75982440e-01 -9.87242043e-01 1.36956739e+00 2.95266479e-01
2.39129707e-01 -1.36720610e+00 -1.16478866e-02 -1.22543573e-01
4.37903464e-01 9.59128678e-01 8.76027465e-01 -2.81689197e-01
-7.85887957e-01 -2.60554105e-01 -1.83939144e-01 -2.79219337e-02
3.95683646e-01 3.18802409e-02 -5.69099844e-01 -3.13295603e-01
-3.18197370e-01 2.19247192e-01 9.25614178e-01 5.59958398e-01
1.60173035e+00 -3.67534548e-01 -4.79587555e-01 8.48962545e-01
1.55492818e+00 9.83243585e-02 3.41211110e-01 -1.01541452e-01
1.58486128e+00 2.56067395e-01 6.06291667e-02 4.84815121e-01
7.66413748e-01 2.72802532e-01 6.22944534e-01 1.88332777e-02
-8.91861022e-02 -3.77818346e-01 2.56842017e-01 1.40485847e+00
-2.70028621e-01 -3.92395526e-01 -1.05738628e+00 6.16231740e-01
-2.26458979e+00 -7.57118404e-01 -1.75349236e-01 1.76988721e+00
2.25200236e-01 8.05306584e-02 -8.38532820e-02 -2.71962211e-02
9.89083350e-01 7.97755241e-01 -6.41437471e-01 -1.01232424e-01
-7.79082477e-02 1.62212461e-01 6.94956243e-01 4.11432683e-01
-1.19735301e+00 1.07470548e+00 5.02993870e+00 7.03609645e-01
-1.47090852e+00 6.99010342e-02 5.49500525e-01 2.00306013e-01
-6.19972885e-01 -6.27093688e-02 -1.12916701e-01 4.17366773e-01
8.31524253e-01 -1.91924468e-01 7.37428606e-01 4.33315754e-01
2.95845598e-01 6.25065923e-01 -7.84279704e-01 1.14145005e+00
-1.74535930e-01 -1.54637170e+00 2.45371968e-01 -4.92952429e-02
8.89259636e-01 3.04484665e-01 1.28639750e-02 5.63745737e-01
2.83071816e-01 -1.32278562e+00 5.20895720e-01 8.40521574e-01
6.01137638e-01 -7.24899828e-01 4.70771313e-01 1.34136051e-01
-2.06395030e+00 -8.37515965e-02 -3.97454232e-01 -1.43646330e-01
3.76715064e-02 4.73264068e-01 -4.83453810e-01 1.17760050e+00
5.11002064e-01 1.46355629e+00 -7.81492889e-01 7.07407534e-01
-8.79443958e-02 6.09018803e-01 -1.76518515e-01 -3.06875348e-01
5.25880158e-01 -3.14126372e-01 3.99581671e-01 1.03185570e+00
3.47178847e-01 1.26660496e-01 4.76347625e-01 9.88692701e-01
-2.20179170e-01 1.70752481e-02 -5.29630303e-01 -4.04149354e-01
5.07564485e-01 1.28945363e+00 -9.00581658e-01 -1.27168357e-01
-4.78196502e-01 9.86297607e-01 7.25581944e-01 7.26023555e-01
-8.81092846e-01 -5.13993502e-01 3.65547478e-01 -2.55685449e-01
4.45550352e-01 -5.57653248e-01 -2.96907037e-01 -1.31440961e+00
7.49015659e-02 -4.23554748e-01 4.67502117e-01 -6.49136364e-01
-1.33666039e+00 7.14646101e-01 -2.81802267e-01 -1.11835039e+00
-3.14231589e-02 -3.41508836e-01 -1.04146147e+00 1.02474844e+00
-1.23992014e+00 -1.62673771e+00 -5.64803123e-01 9.56922889e-01
3.70748401e-01 -4.05988023e-02 5.47303796e-01 2.38357604e-01
-8.90825570e-01 4.86128002e-01 -1.37234762e-01 2.82198370e-01
4.19012271e-02 -1.35872746e+00 5.86223304e-01 1.04328811e+00
-4.42140326e-02 5.38712978e-01 3.23532403e-01 -8.63335490e-01
-1.71178639e+00 -1.81262779e+00 5.67523122e-01 -6.05083182e-02
1.00850403e+00 -5.39620996e-01 -1.10936785e+00 8.55273604e-01
2.33639367e-02 4.56500262e-01 2.59954304e-01 2.73559332e-01
-1.51720181e-01 -3.15180361e-01 -6.29813433e-01 5.75659513e-01
1.08726025e+00 -9.66484070e-01 -1.36199608e-01 3.37062120e-01
1.22589207e+00 -3.39578658e-01 -1.23338819e+00 5.07701576e-01
2.16342092e-01 -5.23195505e-01 6.37230396e-01 -5.11070549e-01
2.70560563e-01 -4.80924368e-01 -1.65336564e-01 -1.42866039e+00
-5.92884839e-01 -7.88623095e-01 -1.33117154e-01 1.15072799e+00
5.42490959e-01 -8.34178627e-01 7.42046118e-01 1.15360662e-01
-5.80768228e-01 -1.03853011e+00 -9.21846271e-01 -4.97839391e-01
-6.32589087e-02 -4.66574162e-01 1.06154847e+00 1.20286274e+00
4.01638635e-02 5.91836452e-01 -1.97196230e-01 4.64496493e-01
6.07368290e-01 2.13688985e-01 5.31747937e-01 -1.07286334e+00
-9.70562398e-02 -7.04419732e-01 -7.49104202e-01 -8.07441890e-01
2.58958608e-01 -1.13760495e+00 7.42895808e-03 -1.98947930e+00
-2.56486446e-01 -3.13119888e-01 -7.26875603e-01 3.63778681e-01
-4.00795877e-01 -1.11940451e-01 -4.87087332e-02 1.23460226e-01
-7.25498080e-01 9.05338824e-01 1.49918747e+00 -3.71367037e-01
-3.09897542e-01 -2.82557428e-01 -4.26519901e-01 1.52888894e-01
9.13512766e-01 -1.73387796e-01 -6.41551852e-01 -5.58222950e-01
1.85510769e-01 3.45641702e-01 4.88455743e-01 -1.01718199e+00
5.42727470e-01 -1.36381149e-01 2.86016315e-01 -6.98467374e-01
1.50864378e-01 -5.56652188e-01 3.29794616e-01 1.80931389e-01
-2.88987100e-01 2.57642061e-01 -3.70102152e-02 9.69356835e-01
-3.79328907e-01 5.63989997e-01 3.66924077e-01 -1.49897942e-02
-7.02754200e-01 1.07525158e+00 -2.05921903e-01 -1.59392759e-01
8.16362560e-01 2.15279385e-02 -2.74767458e-01 -2.80682772e-01
-9.53684390e-01 6.01394832e-01 8.57957676e-02 4.15042818e-01
6.59159660e-01 -1.68996501e+00 -5.81180930e-01 1.33779481e-01
1.13794647e-01 3.91636908e-01 4.65154022e-01 1.00058913e+00
-4.23562884e-01 2.14120001e-01 -1.74932666e-02 -7.11526871e-01
-6.43834233e-01 7.32986510e-01 5.00490010e-01 -3.08118701e-01
-7.01126695e-01 8.09209406e-01 1.69808030e-01 -3.31532687e-01
-6.22538989e-03 -7.67766893e-01 -2.77614832e-01 -6.68662786e-02
-1.54942706e-01 1.29059553e-01 -4.15610336e-02 -6.40661418e-01
-5.39359987e-01 5.23730338e-01 4.11140174e-01 2.17344373e-01
1.39195335e+00 -2.27745697e-01 -3.28901142e-01 6.45238698e-01
1.52838576e+00 -4.06239599e-01 -1.21053314e+00 -2.67711699e-01
-4.07520458e-02 6.11304045e-02 1.04500078e-01 -7.11214021e-02
-1.47026801e+00 9.40747261e-01 2.98164606e-01 7.06540942e-01
1.28090668e+00 8.94440636e-02 7.04960525e-01 1.61640868e-01
7.30970576e-02 -7.92079449e-01 -3.86953750e-03 5.32303751e-01
8.39834809e-01 -9.78674352e-01 -1.32164046e-01 -2.43654296e-01
-3.78450423e-01 1.14358366e+00 5.35383165e-01 -5.05526483e-01
1.05029202e+00 -2.55412996e-01 -3.41581225e-01 -7.14165747e-01
-6.66544259e-01 -3.52744281e-01 5.53067565e-01 5.70889592e-01
5.06780922e-01 6.11541569e-01 -7.08221048e-02 6.97686672e-01
-1.74122676e-01 -2.60111064e-01 2.05027908e-01 5.69929361e-01
-2.03784227e-01 -6.47299111e-01 6.17112368e-02 6.68802619e-01
-2.57143259e-01 -3.23755927e-02 -3.08306962e-01 6.33509159e-01
9.03725997e-02 8.77678871e-01 1.53061643e-01 -9.46247339e-01
2.11838797e-01 -3.27949435e-01 2.01784492e-01 -6.83820486e-01
-4.45672542e-01 2.46934861e-01 -2.21920282e-01 -6.54595613e-01
-6.19366705e-01 -5.97015023e-01 -1.38990819e+00 -1.93316936e-01
-2.43290961e-01 1.54605031e-01 2.83100694e-01 9.19455171e-01
4.34972614e-01 1.13410676e+00 7.47506022e-01 -9.85495389e-01
1.46702588e-01 -9.89491165e-01 -6.82096004e-01 2.68317699e-01
5.34959674e-01 -5.69274306e-01 -1.14923194e-01 -1.39243737e-01] | [6.818709373474121, 2.8210859298706055] |
4ebc83cd-ef1b-4a09-9f27-5ac100500d37 | improving-policy-learning-via-language | 2210.00066 | null | https://arxiv.org/abs/2210.00066v1 | https://arxiv.org/pdf/2210.00066v1.pdf | Improving Policy Learning via Language Dynamics Distillation | Recent work has shown that augmenting environments with language descriptions improves policy learning. However, for environments with complex language abstractions, learning how to ground language to observations is difficult due to sparse, delayed rewards. We propose Language Dynamics Distillation (LDD), which pretrains a model to predict environment dynamics given demonstrations with language descriptions, and then fine-tunes these language-aware pretrained representations via reinforcement learning (RL). In this way, the model is trained to both maximize expected reward and retain knowledge about how language relates to environment dynamics. On SILG, a benchmark of five tasks with language descriptions that evaluate distinct generalization challenges on unseen environments (NetHack, ALFWorld, RTFM, Messenger, and Touchdown), LDD outperforms tabula-rasa RL, VAE pretraining, and methods that learn from unlabeled demonstrations in inverse RL and reward shaping with pretrained experts. In our analyses, we show that language descriptions in demonstrations improve sample-efficiency and generalization across environments, and that dynamics modelling with expert demonstrations is more effective than with non-experts. | ['Tim Rocktäschel', 'Edward Grefenstette', 'Luke Zettlemoyer', 'Jesse Mu', 'Victor Zhong'] | 2022-09-30 | null | null | null | null | ['nethack'] | ['playing-games'] | [-2.33526364e-01 -9.56076533e-02 -2.04324380e-01 -1.90795869e-01
-6.26462817e-01 -8.58879983e-01 8.14257979e-01 7.53380731e-02
-7.66608119e-01 1.01313436e+00 5.17851233e-01 -5.06171048e-01
1.08785003e-01 -4.27986503e-01 -1.15130222e+00 -4.29492354e-01
-6.38722956e-01 7.74896622e-01 -3.81666310e-02 -4.27069455e-01
-8.58615991e-03 5.14460742e-01 -1.45037913e+00 8.86840671e-02
8.79674971e-01 4.11952794e-01 6.33481860e-01 9.40858006e-01
-1.04335740e-01 1.16884637e+00 -4.41464514e-01 5.03862143e-01
3.26610535e-01 -3.21959466e-01 -5.77379942e-01 -3.18443894e-01
2.75657009e-02 -6.96322978e-01 -7.00086594e-01 7.09252119e-01
4.66878742e-01 8.62144947e-01 6.98632479e-01 -1.19423127e+00
-9.30689096e-01 8.15727592e-01 -1.84257939e-01 2.32612520e-01
4.29344594e-01 1.04411852e+00 6.99196160e-01 -3.83411974e-01
7.68993616e-01 1.96931159e+00 2.74587333e-01 1.10590804e+00
-1.62135208e+00 -7.10033178e-01 7.93654859e-01 -5.83410226e-02
-8.51729274e-01 -3.13852519e-01 1.97005332e-01 -6.64316475e-01
1.56056523e+00 -3.36685032e-01 6.13668263e-01 1.84256005e+00
7.15631619e-02 1.05681086e+00 1.23611045e+00 -2.25573346e-01
4.58049476e-01 5.61353043e-02 -9.11704600e-02 6.69702172e-01
-1.23713955e-01 1.07699943e+00 -4.93108898e-01 -1.09765776e-01
7.46701002e-01 1.64537691e-02 -2.03474194e-01 -8.00167263e-01
-1.36964142e+00 6.88381910e-01 5.22875309e-01 -1.02182746e-01
-4.31630671e-01 6.89239085e-01 6.05295360e-01 6.84677720e-01
-1.05689215e-02 1.12612152e+00 -7.08483160e-01 -4.94396389e-01
-1.58727795e-01 6.69220507e-01 8.51901054e-01 1.05687082e+00
7.59076118e-01 5.11397481e-01 -2.64328569e-01 4.34947252e-01
1.38711676e-01 9.87464547e-01 9.17999685e-01 -1.19696796e+00
5.16365290e-01 2.63559729e-01 4.36827272e-01 -5.12351468e-02
-5.03404081e-01 -3.14224958e-01 -2.64461964e-01 6.91202760e-01
2.98838496e-01 -5.64717889e-01 -1.19082403e+00 2.23559999e+00
-1.43225836e-02 3.76744568e-01 7.46947050e-01 8.83048058e-01
5.31764567e-01 7.70746946e-01 4.96051759e-01 1.63906515e-01
6.12977684e-01 -1.24736416e+00 -4.50248361e-01 -5.65204144e-01
7.87388921e-01 -8.67775977e-02 1.51568604e+00 3.22483838e-01
-7.75944948e-01 -8.76093626e-01 -7.43748903e-01 -5.44624925e-02
-4.49933887e-01 -1.22500747e-01 4.65884387e-01 -2.45377440e-02
-1.09608042e+00 7.79301763e-01 -1.24355543e+00 -5.34516275e-01
9.22771990e-02 5.14523506e-01 -1.33035392e-01 4.99191284e-02
-1.14436162e+00 1.28177452e+00 6.63256764e-01 -3.31935704e-01
-2.19666886e+00 -5.47259867e-01 -1.18008626e+00 8.11147764e-02
5.42730272e-01 -6.36856914e-01 1.72632027e+00 -7.07058251e-01
-1.98909307e+00 2.87653685e-01 1.10472888e-01 -1.08050299e+00
4.12477016e-01 -5.81237495e-01 -4.61129956e-02 -2.69681126e-01
-1.10015366e-02 1.18216562e+00 7.26666868e-01 -1.36131978e+00
-5.31732678e-01 8.52275789e-02 3.77019078e-01 5.38913548e-01
1.39498487e-01 -4.70420361e-01 -3.66582945e-02 -1.44160852e-01
-5.14079690e-01 -1.06859446e+00 -5.92910647e-01 -1.69050381e-01
-7.19289333e-02 -3.41840059e-01 7.24505246e-01 -5.36601543e-01
7.07520187e-01 -2.25120521e+00 5.65969944e-01 4.07530069e-02
9.89339352e-02 1.12289570e-01 -7.29280293e-01 4.58072096e-01
2.27346838e-01 -5.96890114e-02 5.49018346e-02 -3.58970881e-01
5.46207249e-01 6.26062572e-01 -7.17238307e-01 1.66342527e-01
8.68105236e-03 1.09646976e+00 -1.45528030e+00 2.81211585e-02
3.85459632e-01 2.08445847e-01 -7.37099707e-01 6.03503287e-01
-1.00318623e+00 8.63489866e-01 -5.06960869e-01 2.55426735e-01
-5.54216430e-02 -1.52810723e-01 2.21267402e-01 5.46259522e-01
-4.33107987e-02 4.35925275e-01 -8.52462709e-01 2.07057858e+00
-9.09151852e-01 6.02170289e-01 2.11762916e-02 -7.73601294e-01
9.12987709e-01 4.61052991e-02 6.89020008e-02 -8.58739555e-01
-2.25195780e-01 1.14710055e-01 1.79669276e-01 -7.62932301e-01
4.98770505e-01 -5.59339859e-02 -2.84778923e-01 5.02957344e-01
1.96639910e-01 -5.97750127e-01 5.81536926e-02 2.82068402e-01
1.19833076e+00 8.78440440e-01 2.41750643e-01 2.52441913e-01
1.09882660e-01 1.74169794e-01 2.56127447e-01 1.22300196e+00
-3.53534997e-01 -1.88377097e-01 3.62129956e-01 -3.19012046e-01
-1.02954674e+00 -1.36296213e+00 6.45723164e-01 1.66587734e+00
8.72722194e-02 -3.30219001e-01 -3.75459731e-01 -6.46793485e-01
4.57787246e-01 1.40042686e+00 -7.48676717e-01 -4.55506921e-01
-6.34828389e-01 -3.01246196e-02 3.47060144e-01 7.25955129e-01
1.56840727e-01 -1.50699461e+00 -9.40240622e-01 1.74420580e-01
3.21062893e-01 -8.69774878e-01 -4.82461840e-01 7.01313853e-01
-7.35259295e-01 -7.43629396e-01 -6.02454007e-01 -6.55743778e-01
4.79977220e-01 -2.88784690e-02 1.15378404e+00 -2.20372528e-01
-4.95003499e-02 9.15938318e-01 -2.18952432e-01 -2.45570704e-01
-7.56802022e-01 9.91006866e-02 5.20162940e-01 -8.91499043e-01
1.90399855e-01 -6.80213392e-01 -1.61334887e-01 -1.28682315e-01
-4.34840173e-01 -1.01501010e-01 7.24510491e-01 1.14935338e+00
4.72226858e-01 -2.92187214e-01 4.96488154e-01 -4.41547692e-01
1.03257501e+00 -5.69312632e-01 -9.61990237e-01 1.53741226e-01
-5.22338629e-01 8.03831935e-01 8.25546741e-01 -8.87633979e-01
-1.07863772e+00 1.88384727e-01 3.38628441e-01 -8.91233206e-01
-3.79795074e-01 3.84874195e-01 3.02537471e-01 2.15406597e-01
1.07655990e+00 4.45019633e-01 1.88215375e-01 -4.19862866e-01
8.57797146e-01 1.22558393e-01 5.81287444e-01 -1.32805669e+00
9.30912495e-01 1.67949386e-02 -3.50978345e-01 -5.30630767e-01
-5.38670778e-01 -1.96606398e-01 -3.13663810e-01 2.38938764e-01
8.62646043e-01 -1.08687508e+00 -9.71146047e-01 -9.50920284e-02
-8.85544717e-01 -1.53330553e+00 -7.50778496e-01 8.07364404e-01
-1.07245839e+00 -9.25625637e-02 -4.97523278e-01 -1.12751198e+00
1.83208063e-01 -1.39954448e+00 8.59108448e-01 1.92379430e-01
-2.71160394e-01 -1.04786825e+00 2.64909416e-01 -3.76366884e-01
5.57035029e-01 1.85835168e-01 9.66623843e-01 -8.14662576e-01
-7.20917940e-01 5.13701916e-01 1.96693420e-01 2.69711584e-01
-6.20729849e-02 -4.46951061e-01 -7.05709219e-01 -6.17646217e-01
-4.84106928e-01 -1.19824624e+00 8.15516353e-01 2.56205320e-01
1.15343750e+00 -3.62236977e-01 -3.82091731e-01 5.60292423e-01
1.00657094e+00 4.28333879e-01 -1.19524319e-02 6.26176119e-01
3.51012945e-01 4.01483655e-01 5.52493572e-01 3.83622378e-01
3.57696027e-01 4.10281479e-01 5.79117298e-01 1.90872833e-01
-2.12503597e-02 -8.30891132e-01 9.66671705e-01 2.21243829e-01
2.25424692e-01 -1.34881973e-01 -7.67370999e-01 4.66128856e-01
-1.95725071e+00 -1.01458704e+00 7.73786485e-01 1.99853337e+00
8.45582247e-01 1.19819306e-01 3.73475403e-01 -8.49504054e-01
1.74556524e-01 3.66144292e-02 -1.43422961e+00 -3.83923143e-01
5.02956733e-02 1.39033943e-01 3.34713131e-01 9.42783833e-01
-8.26227367e-01 1.31222510e+00 6.55647087e+00 3.88393551e-01
-1.03271019e+00 -1.88917175e-01 2.37540394e-01 -3.01451355e-01
-2.18678594e-01 -1.40629366e-01 -7.38386214e-01 1.57474145e-01
1.12109029e+00 -3.98218870e-01 1.21940625e+00 1.23996568e+00
4.10488576e-01 6.43198714e-02 -1.58802736e+00 6.08243108e-01
-3.91316921e-01 -9.43536401e-01 -7.01502562e-02 -3.26267593e-02
7.95323133e-01 6.05020463e-01 4.25707519e-01 1.44256163e+00
1.51773238e+00 -1.29352963e+00 6.85365140e-01 4.65877771e-01
5.33120751e-01 -4.51237559e-01 2.89729416e-01 8.69069874e-01
-7.78084815e-01 -4.94184792e-01 -3.63496959e-01 -3.21894616e-01
1.04480386e-01 -6.41500235e-01 -1.39602959e+00 8.86772275e-02
4.39633727e-01 7.95772612e-01 -2.52417833e-01 6.36378467e-01
-5.65817595e-01 4.29968178e-01 -2.87718415e-01 -4.31449950e-01
6.90587461e-01 -4.75372374e-02 5.33879340e-01 1.08364296e+00
2.74842232e-01 2.33589280e-02 9.63616252e-01 1.18457174e+00
1.77346528e-01 -4.10842031e-01 -8.89341354e-01 -4.45404589e-01
4.67610061e-01 6.47919655e-01 -7.19949305e-02 -5.95420480e-01
-9.40546468e-02 7.77564526e-01 6.46772146e-01 9.15560305e-01
-6.33529663e-01 1.64473087e-01 9.88616824e-01 -2.67004102e-01
2.24116221e-01 -5.60267091e-01 3.67155313e-01 -1.05463374e+00
-5.63451767e-01 -1.19461143e+00 1.39867455e-01 -9.14039731e-01
-1.19215262e+00 4.68141735e-01 3.53155553e-01 -1.05163014e+00
-9.47635829e-01 -8.08271706e-01 -2.69409716e-01 1.07797587e+00
-1.58111978e+00 -7.60978103e-01 -1.39685661e-01 5.16495764e-01
8.71339917e-01 -4.89889860e-01 9.68357980e-01 -4.58069116e-01
-3.44088942e-01 3.16780746e-01 2.81070799e-01 -9.40928310e-02
6.26376152e-01 -1.61146009e+00 7.02743649e-01 3.82974207e-01
-3.13580334e-02 9.47296083e-01 1.04875875e+00 -6.92117453e-01
-1.51303041e+00 -1.07657254e+00 -3.18743102e-02 -6.17337763e-01
9.94823635e-01 -3.74548435e-01 -9.53291893e-01 1.03315008e+00
2.82750100e-01 3.24649066e-02 1.42739013e-01 9.18565542e-02
-4.78110462e-01 2.56873757e-01 -8.65849733e-01 7.61842012e-01
1.17828453e+00 -7.19557106e-01 -9.63577151e-01 5.32973111e-01
1.15692174e+00 -8.37591648e-01 -5.27614355e-01 -8.74451995e-02
5.18408239e-01 -4.46967214e-01 1.00437081e+00 -1.33164239e+00
2.21255094e-01 -3.15817326e-01 -2.51740038e-01 -1.97200942e+00
-1.60402715e-01 -8.36943507e-01 -4.06647414e-01 4.64715600e-01
4.51400161e-01 -6.79759860e-01 4.84492660e-01 2.85155952e-01
-2.44335279e-01 -4.96976644e-01 -4.76279885e-01 -1.28120160e+00
4.27951992e-01 -1.06045775e-01 6.66481197e-01 4.57397997e-01
-2.26572212e-02 3.76708508e-01 -3.20097387e-01 1.53841406e-01
2.72833824e-01 8.28398764e-03 1.20561910e+00 -7.57569253e-01
-8.78564835e-01 -3.65406871e-01 1.59004644e-01 -1.61819434e+00
9.23146009e-01 -8.78030419e-01 5.12372136e-01 -1.47508633e+00
-1.23985473e-03 -5.04494727e-01 -3.31288189e-01 5.59543014e-01
-9.32073072e-02 -8.90004516e-01 4.08720374e-01 -6.14922121e-03
-9.29134250e-01 9.40733552e-01 1.52944243e+00 -3.21578771e-01
-6.18065834e-01 -1.43429682e-01 -3.68749231e-01 5.43238521e-01
7.07231224e-01 -1.82469696e-01 -1.00026298e+00 -6.28588438e-01
-1.28526300e-01 1.08561553e-01 3.84993732e-01 -1.02669072e+00
8.65463689e-02 -5.22885740e-01 3.65784109e-01 -3.42644721e-01
5.60562491e-01 -5.33334136e-01 -3.72351259e-01 7.33225048e-01
-1.07445133e+00 2.98323125e-01 6.65713608e-01 9.56884086e-01
2.36569971e-01 -2.67547332e-02 4.12677675e-01 -5.82104146e-01
-1.25118208e+00 1.81374907e-01 -5.54713130e-01 3.16640884e-01
8.10265303e-01 2.22050607e-01 -4.25842702e-01 -7.68784523e-01
-1.02053642e+00 9.46095347e-01 4.80782330e-01 5.56580722e-01
6.24819338e-01 -1.10633409e+00 -4.87365127e-01 7.61051178e-02
1.39759645e-01 1.46346778e-01 3.11789531e-02 3.30948740e-01
-7.18771517e-02 4.27866787e-01 -2.22150579e-01 -5.17499804e-01
-7.40694284e-01 9.32380974e-01 5.55846095e-01 -4.32934165e-01
-6.83443189e-01 9.11500216e-01 4.77931738e-01 -1.03546584e+00
6.68610930e-01 -8.82411480e-01 -7.66299292e-02 -6.52577519e-01
3.73516649e-01 9.76761132e-02 -7.08582222e-01 -1.34531185e-01
-1.48301860e-02 1.64475635e-01 -1.43858239e-01 -5.03436983e-01
1.21294892e+00 6.99330121e-02 5.48478544e-01 6.93284452e-01
7.97235787e-01 -3.84286076e-01 -2.08646226e+00 -3.02663624e-01
9.99409929e-02 -6.23344220e-02 -1.98138013e-01 -1.18441832e+00
-1.94272235e-01 1.06088197e+00 6.30036950e-01 -2.65266359e-01
6.39582932e-01 -8.00345019e-02 3.92677575e-01 1.29537630e+00
5.43758333e-01 -1.01690841e+00 6.58345222e-01 9.56752002e-01
1.15673864e+00 -1.35634875e+00 -3.16913009e-01 6.01873875e-01
-8.65375102e-01 9.40535128e-01 1.23647606e+00 -3.37572753e-01
2.14017913e-01 2.64421403e-01 -1.42199665e-01 1.36641681e-01
-1.21100771e+00 -3.54895055e-01 -2.34398339e-02 1.07426357e+00
1.31417653e-02 2.31074363e-01 6.52463317e-01 2.53553182e-01
-2.15858608e-01 -9.02979448e-03 2.40376189e-01 9.14820313e-01
-6.67011678e-01 -9.14628088e-01 -3.03800404e-01 1.45334020e-01
3.47829700e-01 7.47288465e-02 -2.69659251e-01 1.19657278e+00
-8.90567601e-02 5.77645063e-01 -3.53006236e-02 -4.18605834e-01
3.23371470e-01 9.44077000e-02 5.24883091e-01 -1.05223882e+00
-5.60735106e-01 -6.64326251e-02 4.96293120e-02 -8.27255249e-01
1.09210968e-01 -6.58079982e-01 -1.68228960e+00 1.49695016e-02
2.00347021e-01 3.54595423e-01 4.79470909e-01 9.15755212e-01
3.00567329e-01 9.42116380e-01 1.93518907e-01 -1.00463235e+00
-1.52329600e+00 -1.02916694e+00 -3.30869824e-01 3.71095300e-01
9.37184513e-01 -8.84293139e-01 -4.86542612e-01 -1.72976196e-01] | [4.186910629272461, 1.3837281465530396] |
f90f2046-fe80-476b-9d9b-1e8145020ef6 | attack-agnostic-adversarial-detection | 2206.00489 | null | https://arxiv.org/abs/2206.00489v1 | https://arxiv.org/pdf/2206.00489v1.pdf | Attack-Agnostic Adversarial Detection | The growing number of adversarial attacks in recent years gives attackers an advantage over defenders, as defenders must train detectors after knowing the types of attacks, and many models need to be maintained to ensure good performance in detecting any upcoming attacks. We propose a way to end the tug-of-war between attackers and defenders by treating adversarial attack detection as an anomaly detection problem so that the detector is agnostic to the attack. We quantify the statistical deviation caused by adversarial perturbations in two aspects. The Least Significant Component Feature (LSCF) quantifies the deviation of adversarial examples from the statistics of benign samples and Hessian Feature (HF) reflects how adversarial examples distort the landscape of the model's optima by measuring the local loss curvature. Empirical results show that our method can achieve an overall ROC AUC of 94.9%, 89.7%, and 94.6% on CIFAR10, CIFAR100, and SVHN, respectively, and has comparable performance to adversarial detectors trained with adversarial examples on most of the attacks. | ['Wael AbdAlmageed', 'Jay Billa', 'Mohamed Hussein', 'Jiaxin Cheng'] | 2022-06-01 | null | null | null | null | ['adversarial-attack-detection', 'adversarial-attack-detection'] | ['computer-vision', 'knowledge-base'] | [-1.41884252e-01 -1.22964554e-01 1.88552514e-01 -4.42261398e-02
-7.61324704e-01 -1.37424457e+00 6.85961843e-01 4.03428040e-02
-3.73117268e-01 3.24341267e-01 -6.36993200e-02 -5.24059117e-01
1.23319395e-01 -7.28683829e-01 -6.94402635e-01 -7.12768197e-01
-5.75191677e-01 1.80126145e-01 4.06285286e-01 -4.78973776e-01
3.43828201e-01 7.88395286e-01 -5.14618695e-01 2.53442019e-01
4.74887162e-01 9.46916759e-01 -8.77651811e-01 8.73059034e-01
4.02019441e-01 5.79647958e-01 -1.15997040e+00 -7.54415274e-01
9.18818593e-01 -3.46678168e-01 -4.00738657e-01 -5.63834429e-01
4.91614997e-01 -3.49776179e-01 -7.24239826e-01 1.61324811e+00
4.28564638e-01 -5.59091987e-03 7.18587458e-01 -1.13421953e+00
-4.99653339e-01 5.35257518e-01 -4.85522866e-01 6.30005836e-01
3.12062353e-01 5.80221891e-01 9.00387406e-01 -5.07104397e-01
2.56269962e-01 1.37965584e+00 5.72374284e-01 7.44947612e-01
-9.63890374e-01 -9.58374321e-01 4.25541073e-01 -3.65201496e-02
-1.30198026e+00 -3.22887480e-01 6.83912396e-01 -5.07491946e-01
3.30090284e-01 5.48804402e-01 2.49715105e-01 1.34109354e+00
5.37132859e-01 3.57793093e-01 7.02714324e-01 -1.18553661e-01
3.48421067e-01 2.76548833e-01 -1.48344606e-01 4.71808761e-01
3.89916509e-01 7.56598949e-01 6.04269244e-02 -6.59120560e-01
4.49136227e-01 8.57649669e-02 -3.50709260e-01 -2.06526369e-01
-5.78489721e-01 1.15119386e+00 6.57120466e-01 -1.70565695e-02
-1.98460251e-01 -7.12779686e-02 4.68193918e-01 6.05877340e-01
2.90373594e-01 8.98604751e-01 -2.38102660e-01 1.87005132e-01
-4.15534973e-01 2.81156898e-01 8.48634005e-01 2.33275682e-01
3.40741947e-02 5.78982770e-01 5.31593058e-03 3.02528828e-01
1.45674884e-01 8.47627819e-01 3.70587736e-01 -4.55454558e-01
5.79225421e-01 3.73400837e-01 -5.62306568e-02 -1.19416177e+00
-1.57816652e-02 -8.84372294e-01 -6.97671950e-01 5.43332875e-01
6.40850544e-01 -4.87026155e-01 -7.86509931e-01 1.73286903e+00
2.09702715e-01 5.81461452e-02 2.04318702e-01 7.39302814e-01
6.27730042e-02 4.38324451e-01 9.16563943e-02 6.72788247e-02
6.92908764e-01 -4.23113495e-01 -2.18561962e-01 -6.11998081e-01
3.84605318e-01 -5.52721560e-01 8.31423938e-01 3.22600424e-01
-8.19002986e-01 -2.01722920e-01 -1.24500179e+00 1.07990932e+00
-4.80579853e-01 -7.22239614e-01 2.76063770e-01 1.18510580e+00
-3.21206748e-01 7.65474975e-01 -6.36664987e-01 2.84207284e-01
4.25969005e-01 2.41677314e-01 -3.74120235e-01 2.35339567e-01
-1.62065780e+00 8.58580768e-01 3.43992412e-01 -1.89450532e-01
-1.43407536e+00 -7.38156676e-01 -5.37506402e-01 2.39189401e-01
1.67836323e-01 6.35771751e-02 9.78958905e-01 -7.66174614e-01
-9.52922404e-01 6.30795360e-01 6.01085186e-01 -9.09047008e-01
8.63579512e-01 -1.90643221e-01 -7.95542836e-01 6.32188320e-02
-2.76750028e-01 -2.44169518e-01 1.20769727e+00 -9.63033140e-01
-4.20246601e-01 -6.09666765e-01 2.18061730e-01 2.56060231e-02
-5.65530300e-01 9.47947800e-02 3.13317269e-01 -9.03739810e-01
-1.14426747e-01 -9.69468951e-01 -3.64755630e-01 -1.46226615e-01
-5.44424951e-01 5.09658754e-01 1.16130316e+00 -4.19960767e-01
1.31454396e+00 -2.41707945e+00 -2.22393990e-01 5.99914253e-01
2.72627234e-01 8.41880977e-01 7.88364559e-03 4.17443365e-01
-3.52782100e-01 2.15898991e-01 -2.24582955e-01 4.32432950e-01
-6.15063757e-02 -1.96090981e-01 -1.10658646e+00 8.62244010e-01
-7.95270205e-02 6.15499794e-01 -8.72201622e-01 1.89134106e-01
-5.36901802e-02 2.12256268e-01 -7.16785371e-01 2.29160503e-01
1.50228158e-01 3.22892189e-01 -7.67830968e-01 5.65026283e-01
6.81703329e-01 2.87147433e-01 -1.63542584e-01 3.59406382e-01
4.46752220e-01 -1.16946779e-01 -1.15486169e+00 6.32428348e-01
-7.08145574e-02 5.14725745e-01 1.80103391e-01 -8.81995857e-01
9.42968011e-01 2.06516027e-01 1.20799288e-01 -4.48401064e-01
2.99320638e-01 8.39320943e-02 4.86323506e-01 1.68479338e-01
-2.49338657e-01 -1.23889714e-01 -3.78856957e-01 4.74027187e-01
-2.73809582e-01 9.67800897e-03 -2.53863692e-01 4.07319188e-01
1.32396972e+00 -8.42738092e-01 2.77076900e-01 -1.28269285e-01
6.92963481e-01 -2.13284388e-01 5.56589365e-01 1.04119873e+00
-5.87147474e-01 7.48045594e-02 5.50701141e-01 -7.33861327e-01
-7.48842001e-01 -1.71187830e+00 -1.60947803e-03 9.09402966e-01
5.68723539e-03 -2.56978363e-01 -8.85592103e-01 -1.20683742e+00
2.74642438e-01 8.64812732e-01 -7.15852618e-01 -9.20658588e-01
-5.31955302e-01 -8.70868385e-01 8.73898447e-01 3.59790444e-01
5.45992494e-01 -6.78098261e-01 -3.18040520e-01 -2.07089353e-03
3.75077814e-01 -9.94290769e-01 -5.62138975e-01 1.06632886e-02
-4.66442525e-01 -1.29799294e+00 -1.69562623e-01 -2.56699085e-01
5.98233163e-01 -1.58548132e-02 7.20684469e-01 6.50275573e-02
-5.02811253e-01 2.04240680e-01 -3.54182810e-01 -7.82461166e-01
-7.16334879e-01 -3.36750358e-01 6.55438721e-01 1.75922409e-01
7.89731517e-02 -6.07059658e-01 -4.73521054e-01 4.57454711e-01
-7.65464485e-01 -1.06243265e+00 4.94982541e-01 7.57974565e-01
3.06020617e-01 4.58936393e-01 5.05395174e-01 -1.01603055e+00
8.46096277e-01 -5.66846430e-01 -7.44240165e-01 9.84759182e-02
-4.36067790e-01 -9.99414083e-03 1.35277271e+00 -8.25341046e-01
-6.63920701e-01 -1.05308862e-02 -2.90240318e-01 -6.67201877e-01
-3.13064791e-02 -1.77810818e-01 -2.55313039e-01 -5.22404134e-01
1.43957019e+00 2.40702212e-01 -1.98758602e-01 -2.48810679e-01
2.45502278e-01 2.67330110e-01 5.91573060e-01 -3.43728513e-01
1.88648951e+00 3.60306710e-01 6.83277100e-02 -6.87283039e-01
-1.00135255e+00 -2.45650038e-01 -3.56388450e-01 -1.78448692e-01
4.11543548e-01 -4.24797773e-01 -6.55460000e-01 4.48075294e-01
-7.38584697e-01 3.65679562e-02 -3.64328861e-01 2.37843201e-01
-7.50166774e-02 4.61807966e-01 -4.60970581e-01 -9.46842194e-01
-4.67525631e-01 -9.91117239e-01 2.49167293e-01 1.30957603e-01
4.39490825e-02 -1.06588686e+00 2.31009096e-01 -1.54722080e-01
6.15585506e-01 8.59390259e-01 7.08184302e-01 -1.50209260e+00
-1.10560119e-01 -1.26431501e+00 3.29564601e-01 7.10447729e-01
-9.55026895e-02 -4.74168956e-02 -1.17511535e+00 -7.44549811e-01
3.44422519e-01 -9.37164351e-02 6.18444085e-01 1.62683934e-01
1.16933966e+00 -9.16330040e-01 -2.47639865e-01 7.50552893e-01
1.13760519e+00 6.62954867e-01 5.95752001e-01 3.35722536e-01
5.05251229e-01 3.04145873e-01 3.24448258e-01 3.19422185e-01
-5.77796459e-01 4.72175121e-01 8.84864211e-01 5.71284331e-02
4.71679121e-01 -3.93087953e-01 6.37228131e-01 1.02026970e-04
3.91194284e-01 -1.97046950e-01 -9.19896483e-01 2.00790190e-03
-1.22631824e+00 -1.40899801e+00 2.51642346e-01 2.50511861e+00
3.51811141e-01 9.61914182e-01 2.86362052e-01 3.98801237e-01
6.38828754e-01 4.05117095e-01 -8.02496612e-01 -7.36950338e-01
2.11548060e-03 1.29518300e-01 7.55702138e-01 5.70042789e-01
-1.48107159e+00 1.01753247e+00 6.62336159e+00 8.61405373e-01
-1.01304972e+00 -3.38764548e-01 7.72353530e-01 -1.79748356e-01
2.48375595e-01 -8.71944577e-02 -5.99783361e-01 5.51799417e-01
1.16582370e+00 -4.23986316e-01 4.22520518e-01 1.26745427e+00
-1.95396125e-01 5.91434121e-01 -8.46122384e-01 4.75505888e-01
-1.62792340e-01 -1.11815286e+00 1.06788561e-01 2.54065543e-01
4.02396560e-01 -6.93578050e-02 8.69045913e-01 4.94023442e-01
8.07037055e-01 -1.09866619e+00 4.37707543e-01 1.87867835e-01
4.74615633e-01 -1.02874422e+00 7.75549293e-01 5.96308410e-01
-1.04368472e+00 -4.28004146e-01 -3.25575441e-01 6.93387762e-02
-1.64237693e-01 2.18957588e-01 -9.23112154e-01 9.13674310e-02
4.97468442e-01 2.02472702e-01 -6.42496884e-01 7.79590845e-01
-3.16118807e-01 8.43967080e-01 -7.78914467e-02 1.21263072e-01
5.23270488e-01 9.82821956e-02 1.06585693e+00 1.00105357e+00
-3.48182410e-01 -1.20015554e-02 3.43487084e-01 5.35628259e-01
-4.87474399e-03 -5.34653440e-02 -7.80657589e-01 -2.01987267e-01
7.06945300e-01 9.73617733e-01 -3.75982463e-01 4.93287928e-02
8.57585222e-02 7.66527295e-01 -2.97503527e-02 2.29135692e-01
-9.57717896e-01 -6.91033542e-01 1.08684087e+00 1.35041878e-01
-4.61723171e-02 1.38845230e-02 6.25569671e-02 -1.08861017e+00
1.29744178e-02 -1.26231003e+00 7.79571176e-01 1.57988399e-01
-1.55803394e+00 9.35335994e-01 -3.40771884e-01 -1.41924059e+00
-5.05765378e-01 -7.06945062e-01 -1.24957824e+00 8.49779248e-01
-8.11794579e-01 -6.76181972e-01 2.04831332e-01 6.81794584e-01
2.88236439e-01 -6.96977139e-01 8.85327160e-01 -8.31624269e-02
-7.16999173e-01 1.21129787e+00 1.69686586e-01 8.21144342e-01
4.86439764e-01 -1.12027383e+00 1.01595700e+00 1.36733222e+00
2.40809768e-01 5.11929035e-01 8.18808138e-01 -5.87585092e-01
-1.11131620e+00 -1.11068857e+00 2.45113280e-02 -9.07196045e-01
1.14565551e+00 -3.91434222e-01 -1.05335927e+00 5.92342615e-01
-5.94036341e-01 2.58098364e-01 6.53414607e-01 -7.57303685e-02
-9.95807350e-01 7.01079816e-02 -1.54203558e+00 8.22723687e-01
6.75626576e-01 -5.21746278e-01 -5.33630490e-01 5.03831625e-01
5.80200076e-01 -4.42658156e-01 -6.01836443e-01 3.21066797e-01
3.82492542e-01 -9.02786314e-01 1.45659947e+00 -1.22257805e+00
-1.45452917e-01 7.64485076e-03 -2.46579438e-01 -1.37141120e+00
-4.74127054e-01 -8.49277198e-01 -3.37917387e-01 6.72191501e-01
5.08418024e-01 -9.60969746e-01 7.89039969e-01 3.31341833e-01
1.80325508e-01 -8.18612754e-01 -9.62434769e-01 -1.10878611e+00
5.08191407e-01 -2.63421535e-01 4.18099940e-01 9.27925348e-01
-1.90900207e-01 -1.55298382e-01 -3.54722589e-01 6.76635444e-01
9.17656541e-01 -4.13806289e-01 7.37319887e-01 -1.19933534e+00
-4.94776756e-01 -6.46749496e-01 -8.66567492e-01 -2.93225288e-01
4.74139601e-02 -7.33702719e-01 -4.26014990e-01 -5.91602445e-01
-2.39973456e-01 -2.35909089e-01 -5.21792114e-01 3.00947100e-01
-3.35859120e-01 1.88441798e-01 1.88191310e-01 3.05331469e-01
-2.80395865e-01 2.60974467e-01 8.86299372e-01 -3.72018814e-01
1.92739926e-02 5.94125748e-01 -9.03241396e-01 1.15380239e+00
8.82560551e-01 -6.87104106e-01 -3.19523424e-01 8.67945030e-02
-1.69822410e-01 -2.13421956e-01 3.41965884e-01 -1.25783932e+00
-6.18442968e-02 -2.33386084e-01 7.69450665e-01 -6.64909091e-03
1.85093045e-01 -6.61036313e-01 -3.06474656e-01 9.88736689e-01
-4.80432570e-01 2.35858381e-01 2.41969004e-01 9.23026085e-01
-6.43457547e-02 -5.65284006e-02 1.47977245e+00 -4.29301299e-02
-2.72322685e-01 6.19473994e-01 -2.97518313e-01 5.97740352e-01
1.32382548e+00 4.43657078e-02 -4.85724002e-01 -5.11420369e-01
-6.72665119e-01 1.11055978e-01 2.15360671e-01 3.74897599e-01
7.68297374e-01 -1.06967056e+00 -7.90639222e-01 5.33427835e-01
-4.14633229e-02 -7.57714570e-01 2.44616717e-01 2.11041987e-01
-4.39956874e-01 1.99943826e-01 -2.39074305e-01 -7.50743374e-02
-1.28952622e+00 9.21024859e-01 7.14548707e-01 -4.74265665e-01
-5.69310069e-01 1.10575461e+00 2.93894976e-01 -1.47280633e-01
4.17805046e-01 4.70780581e-01 2.78291572e-02 -4.34472501e-01
8.51639569e-01 2.88623720e-01 -1.40784770e-01 -6.61521077e-01
-5.38041532e-01 2.03079358e-01 -5.38776577e-01 6.01364225e-02
9.41337705e-01 6.25741959e-01 4.18191612e-01 -2.01895699e-01
1.12589514e+00 3.22135299e-01 -1.19883847e+00 -9.31126699e-02
-4.64692190e-02 -7.43497252e-01 -1.18323028e-01 -9.66238439e-01
-1.09091675e+00 8.72258484e-01 7.41096854e-01 6.00837827e-01
1.03514349e+00 -2.42845729e-01 6.69799924e-01 4.33707356e-01
1.38321534e-01 -6.42459214e-01 2.45668337e-01 5.20263553e-01
8.26359034e-01 -9.43885565e-01 -1.23716742e-01 -8.29283595e-02
-8.43584597e-01 9.12852824e-01 5.85210383e-01 -6.40308380e-01
7.40555406e-01 2.06150889e-01 1.48392886e-01 -1.16406284e-01
-5.05509973e-01 5.27177632e-01 6.10181987e-01 5.64102888e-01
-2.36627162e-01 2.23050803e-01 3.28568786e-01 4.21731144e-01
-4.60476905e-01 -1.29136825e+00 3.62556368e-01 7.57825613e-01
-7.12150037e-01 -8.95676613e-01 -6.14362478e-01 4.12249744e-01
-8.52590382e-01 -2.68486943e-02 -9.71359432e-01 9.08052683e-01
-3.29422742e-01 9.24982846e-01 -2.86197096e-01 -9.18736815e-01
6.05195999e-01 -1.42839272e-02 -8.98041651e-02 -4.71972376e-01
-8.32287669e-01 -4.23934937e-01 -3.02517414e-01 -7.05898643e-01
7.34680593e-01 -4.24034208e-01 -7.61217058e-01 -5.98024309e-01
-2.69602358e-01 5.36422193e-01 5.10414362e-01 7.18175411e-01
1.21368781e-01 4.00196135e-01 1.38541186e+00 -3.88879120e-01
-1.39279854e+00 -6.48776710e-01 -5.72283566e-01 5.86758435e-01
5.71605265e-01 -6.22499228e-01 -1.19801116e+00 -5.88213086e-01] | [5.749630928039551, 7.822292804718018] |
2c9b34f2-ec18-4292-bf5f-3ead57d5e6e4 | towards-semi-supervised-learning-of-automatic | 2204.03896 | null | https://arxiv.org/abs/2204.03896v1 | https://arxiv.org/pdf/2204.03896v1.pdf | Towards Semi-Supervised Learning of Automatic Post-Editing: Data-Synthesis by Infilling Mask with Erroneous Tokens | Semi-supervised learning that leverages synthetic training data has been widely adopted in the field of Automatic post-editing (APE) to overcome the lack of human-annotated training data. In that context, data-synthesis methods to create high-quality synthetic data have also received much attention. Considering that APE takes machine-translation outputs containing translation errors as input, we propose a noising-based data-synthesis method that uses a mask language model to create noisy texts through substituting masked tokens with erroneous tokens, yet following the error-quantity statistics appearing in genuine APE data. In addition, we propose corpus interleaving, which is to combine two separate synthetic data by taking only advantageous samples, to further enhance the quality of the synthetic data created with our noising method. Experimental results reveal that using the synthetic data created with our approach results in significant improvements in APE performance upon using other synthetic data created with different existing data-synthesis methods. | ['Jong-Hyeok Lee', 'Baikjin Jung', 'Seong-Hwan Heo', 'WonKee Lee'] | 2022-04-08 | null | null | null | null | ['automatic-post-editing', 'automatic-post-editing'] | ['computer-vision', 'natural-language-processing'] | [ 9.10637558e-01 3.98328841e-01 9.88323167e-02 -3.37976187e-01
-1.34293735e+00 -4.84293818e-01 9.19894576e-01 2.73125201e-01
-5.97660065e-01 9.46819365e-01 3.99127334e-01 -3.02786976e-01
2.38545522e-01 -5.85387945e-01 -8.85476291e-01 -3.34123909e-01
5.57201445e-01 3.61442804e-01 -1.32028773e-01 -2.85259813e-01
3.09643865e-01 -2.35559717e-01 -1.29793906e+00 6.32329166e-01
1.44836426e+00 3.35779786e-01 1.01754501e-01 5.63686311e-01
-2.02637196e-01 4.97551471e-01 -1.00588226e+00 -7.77535975e-01
5.18879294e-01 -8.64500523e-01 -5.21487772e-01 1.20133325e-01
1.78234920e-01 3.48805338e-02 1.30551398e-01 9.61275339e-01
6.10024989e-01 -2.00343892e-01 4.90238637e-01 -9.51665282e-01
-7.04127312e-01 1.16557741e+00 -9.60891470e-02 -1.55340657e-01
3.36011708e-01 2.22434253e-01 8.48100781e-01 -1.38502192e+00
7.54880011e-01 1.07077062e+00 5.90578437e-01 5.86219430e-01
-1.59687030e+00 -5.46869516e-01 -4.58033532e-01 -1.91844814e-02
-1.14866960e+00 -7.14134991e-01 6.61179960e-01 -1.60560891e-01
8.70177567e-01 4.31578338e-01 3.10943514e-01 1.58499253e+00
-1.02525622e-01 9.45129573e-01 1.38855207e+00 -1.18795598e+00
1.68122038e-01 3.66911978e-01 -3.82889748e-01 2.24443093e-01
3.02973509e-01 6.10448569e-02 -6.97487533e-01 3.19010653e-02
1.86766505e-01 -3.71405602e-01 -1.35286942e-01 1.33870944e-01
-1.58111811e+00 6.03590548e-01 -2.89482206e-01 3.45060229e-01
-4.90785837e-01 -2.85363108e-01 6.18280053e-01 6.73251987e-01
8.24037552e-01 9.81033325e-01 -6.28243148e-01 -5.53682327e-01
-1.33290386e+00 4.72265273e-01 6.06946766e-01 1.05199575e+00
6.70196056e-01 6.02523237e-02 -5.28767169e-01 1.04780543e+00
-1.12347409e-01 6.76764071e-01 7.22501934e-01 -6.42197728e-01
1.04830348e+00 7.98785448e-01 1.92693263e-01 -5.19328952e-01
1.63336664e-01 -1.88619524e-01 -6.93298817e-01 1.73561443e-02
3.74051005e-01 -3.37036818e-01 -7.84151256e-01 1.56276309e+00
1.74894065e-01 -1.66571125e-01 4.05103594e-01 5.03237188e-01
3.28613669e-01 5.58244586e-01 -3.41204733e-01 -3.04100543e-01
9.12937105e-01 -1.04768896e+00 -9.47443306e-01 -2.09195889e-03
1.17275262e+00 -1.07770181e+00 1.64428294e+00 6.28808320e-01
-8.79544973e-01 -4.75771487e-01 -1.07705963e+00 -6.71207681e-02
-3.52849901e-01 4.03541297e-01 -8.36027116e-02 8.16180229e-01
-6.46061599e-01 7.68422008e-01 -5.32233477e-01 -1.34828404e-01
3.54982972e-01 -1.31552815e-01 -3.28509092e-01 -1.61610723e-01
-1.34064627e+00 9.53942895e-01 6.17266059e-01 -5.34280166e-02
-5.20729125e-01 -8.09812963e-01 -7.42534518e-01 -2.41719857e-01
6.91293657e-01 -1.99268475e-01 1.38746750e+00 -1.28257740e+00
-1.77626240e+00 5.72714746e-01 -9.48246270e-02 -4.43266749e-01
9.61360693e-01 -6.45272076e-01 -5.25382936e-01 -4.25541580e-01
8.06505159e-02 4.11092430e-01 7.27269173e-01 -1.28977156e+00
-4.51842457e-01 -1.80092938e-02 -4.78564531e-01 1.45354509e-01
-3.69502515e-01 1.92881376e-01 -1.86643064e-01 -1.19556367e+00
-1.02803536e-01 -9.22928810e-01 -2.35184029e-01 -3.62647980e-01
-7.60171235e-01 -1.10013425e-01 5.93744576e-01 -9.62925911e-01
1.52905190e+00 -1.86692441e+00 1.32298410e-01 3.79878551e-01
-1.60574242e-01 7.87143707e-01 -3.64230812e-01 6.81243658e-01
1.24696575e-01 3.87826294e-01 -7.09222913e-01 -7.41883099e-01
6.75477535e-02 3.29815984e-01 -3.85846078e-01 -1.13696456e-01
7.96237886e-01 8.35324168e-01 -1.09199131e+00 -4.51873451e-01
-3.46309058e-02 7.83485770e-02 -5.78751385e-01 3.28102589e-01
-5.36463618e-01 4.23510432e-01 -9.88211483e-02 4.08882111e-01
4.95097488e-01 2.68199950e-01 2.15970084e-01 1.79023087e-01
-2.21717238e-01 5.63429832e-01 -1.15924621e+00 1.77175057e+00
-5.21228671e-01 6.32058024e-01 -4.32666153e-01 -5.52104115e-01
1.03935194e+00 4.71436262e-01 -8.00723117e-03 -9.02973890e-01
3.53882574e-02 6.34905934e-01 4.27897573e-02 -5.15059531e-01
1.08369136e+00 -1.46458119e-01 -7.09765851e-02 6.56154335e-01
3.64688262e-02 -3.14293027e-01 4.87731695e-01 2.24145219e-01
1.15980899e+00 4.58059430e-01 5.92121370e-02 2.62967169e-01
5.01001656e-01 1.73865512e-01 6.58522189e-01 7.92946041e-01
1.19900249e-01 8.92570198e-01 3.76453251e-01 -5.52157464e-04
-1.67068815e+00 -5.79631209e-01 1.33251250e-01 6.95210338e-01
-4.44181025e-01 -6.34109855e-01 -1.14050639e+00 -8.13017368e-01
-3.12089920e-01 1.11455715e+00 -4.98636842e-01 -1.77162707e-01
-7.79997170e-01 -6.68736339e-01 1.07489681e+00 1.34522185e-01
3.55961263e-01 -1.29447377e+00 -3.79888326e-01 4.68884379e-01
-6.29638851e-01 -9.70019817e-01 -4.33791399e-01 2.45995373e-01
-7.19788849e-01 -8.03698063e-01 -7.18073010e-01 -5.20949662e-01
7.23689556e-01 -1.37289599e-01 1.00681090e+00 -3.76274884e-02
1.77496821e-01 -3.60691339e-01 -1.04794252e+00 -5.53054810e-01
-1.39053774e+00 2.86256254e-01 1.35199204e-01 8.54528025e-02
4.21462059e-01 -3.10549229e-01 -1.96957844e-03 1.70189098e-01
-1.29036689e+00 4.22441959e-01 9.35717762e-01 9.93264377e-01
5.17825246e-01 -2.86192983e-01 7.53293633e-01 -1.31472993e+00
9.55895364e-01 -3.40149790e-01 -1.97884232e-01 4.56297457e-01
-8.86693180e-01 4.07910168e-01 1.05157876e+00 -6.65263057e-01
-1.16674840e+00 -3.55361730e-01 6.65641064e-03 -2.24891737e-01
-2.32629273e-02 6.31374538e-01 -2.65895218e-01 5.98278403e-01
1.00543392e+00 4.10733342e-01 5.54962456e-02 -7.10260212e-01
6.76155746e-01 1.21762311e+00 5.83562613e-01 -6.14923716e-01
7.98631191e-01 -1.41082510e-01 -4.35062319e-01 -4.09812331e-01
-5.81801534e-01 -4.07689176e-02 -6.63179994e-01 6.13141023e-02
3.51820588e-01 -7.17814982e-01 3.01684767e-01 5.88062942e-01
-1.21820176e+00 -3.99025172e-01 -4.70574588e-01 4.75731730e-01
-4.47767377e-01 3.85461092e-01 -3.77904773e-01 -7.47428536e-01
-4.76358950e-01 -1.13743365e+00 1.02527082e+00 -3.49660993e-01
-5.38982570e-01 -5.25042474e-01 4.24562067e-01 4.00248826e-01
3.62017900e-01 3.46603304e-01 8.21593344e-01 -1.06696630e+00
-3.11219573e-01 -2.36716941e-01 -5.31631298e-02 8.14494669e-01
3.77775967e-01 1.32129818e-01 -8.17570150e-01 -5.90141453e-02
-1.69778094e-01 -4.53696340e-01 3.53046805e-01 -3.20766121e-01
9.12798107e-01 -7.83792973e-01 7.13287741e-02 1.79355919e-01
1.07856762e+00 -2.30550721e-01 6.68282390e-01 3.90196711e-01
6.75850451e-01 7.43267059e-01 7.72955954e-01 4.24257398e-01
1.31768035e-02 7.29080200e-01 -1.33192495e-01 -6.37887120e-02
-2.32566521e-01 -5.49170315e-01 6.41900957e-01 1.26141083e+00
1.58402085e-01 -4.12581593e-01 -7.52102375e-01 6.67834878e-01
-1.76011670e+00 -8.05651784e-01 -2.77090490e-01 2.24295497e+00
1.57789719e+00 3.13582271e-01 -6.02989458e-02 5.39705396e-01
5.66407859e-01 -1.38845757e-01 -2.67003328e-01 -5.67676365e-01
-4.33649480e-01 4.51354980e-01 5.64929605e-01 3.00954074e-01
-7.01453090e-01 9.65791702e-01 5.39971828e+00 1.16143692e+00
-1.03763103e+00 1.59483850e-01 2.76077360e-01 -8.76474231e-02
-7.63769329e-01 5.81947640e-02 -5.22490621e-01 8.05196226e-01
1.26837778e+00 -8.83119181e-02 3.62880290e-01 4.84082758e-01
7.00735211e-01 -4.71015554e-03 -1.17773557e+00 7.38489449e-01
1.60906300e-01 -1.19505346e+00 2.61706561e-01 3.34989140e-03
9.31203008e-01 -2.32391521e-01 -8.94172937e-02 3.90584260e-01
4.19795543e-01 -7.52921224e-01 1.17133403e+00 4.91289079e-01
9.57873106e-01 -5.47406077e-01 8.35676253e-01 6.33988142e-01
-3.37694585e-01 2.42441148e-01 -1.30543903e-01 -1.10781707e-01
1.44630790e-01 1.08241916e+00 -1.35688603e+00 7.85473168e-01
1.61046430e-01 3.75766546e-01 -7.93937743e-01 6.36983812e-01
-3.34647447e-01 1.04605877e+00 -1.96136400e-01 -3.76832932e-02
-1.58785272e-03 -2.83405751e-01 5.64368784e-01 1.50033045e+00
4.72999781e-01 -4.62500215e-01 -1.24431022e-01 9.67622519e-01
-4.33851779e-01 5.56593239e-01 -6.04638696e-01 -3.11534077e-01
6.63383067e-01 9.56630826e-01 -2.42260441e-01 -6.13326073e-01
-3.13574284e-01 1.27619457e+00 2.67394036e-01 1.46057993e-01
-5.29240310e-01 -6.34217680e-01 2.78953552e-01 -4.12750244e-03
1.61870703e-01 -1.01514608e-01 -6.09367132e-01 -1.18697786e+00
5.07199645e-01 -1.57539487e+00 -1.76115185e-01 -5.54222047e-01
-1.19258833e+00 5.85161805e-01 -1.68017805e-01 -1.47718418e+00
-3.74804199e-01 -2.45403156e-01 -3.25835168e-01 9.13327098e-01
-1.19726455e+00 -1.05052662e+00 1.19203657e-01 -3.50437462e-02
8.95299792e-01 -2.08010808e-01 7.97595739e-01 3.81052226e-01
-4.84561473e-01 9.70882237e-01 4.72291231e-01 1.60261229e-01
1.05205107e+00 -1.24430740e+00 8.34457457e-01 1.17379284e+00
4.05599535e-01 6.77430332e-01 7.59386778e-01 -9.37883556e-01
-1.05532277e+00 -1.38986290e+00 1.47977614e+00 -7.38082170e-01
4.37219828e-01 -6.52773738e-01 -1.09516001e+00 4.58212316e-01
3.13550621e-01 -2.91714638e-01 6.41108274e-01 -2.54080325e-01
-4.05876845e-01 1.68818086e-01 -1.12065423e+00 8.10765445e-01
9.38305140e-01 -5.57284176e-01 -8.66244972e-01 1.85274318e-01
9.54645753e-01 -4.69832003e-01 -6.01724923e-01 1.98858127e-01
2.81520933e-01 -5.72905779e-01 3.20027381e-01 -5.80237925e-01
7.76725352e-01 -3.60906005e-01 -1.76396415e-01 -1.88488638e+00
3.83434325e-01 -9.47945774e-01 9.23643485e-02 1.57203269e+00
8.29124808e-01 -2.78426409e-01 4.96296078e-01 4.49855715e-01
-3.16041321e-01 -4.88062233e-01 -7.93098629e-01 -9.65232432e-01
1.16385192e-01 -5.56028426e-01 8.09474349e-01 1.02258730e+00
4.66969162e-02 1.67072639e-01 -6.78379416e-01 -2.98910826e-01
3.46485317e-01 -4.84739929e-01 1.24603021e+00 -7.42876709e-01
-2.62476414e-01 -7.53059685e-02 1.77311167e-01 -7.85260677e-01
-7.93484822e-02 -1.01772368e+00 4.19496477e-01 -1.11649644e+00
2.29946300e-02 -5.76027989e-01 3.97827104e-02 5.67471862e-01
-5.90823054e-01 2.89921433e-01 2.32299402e-01 3.07065189e-01
-2.84801185e-01 6.37339413e-01 1.10150063e+00 1.06981561e-01
-4.16018873e-01 -2.50495940e-01 -6.08201921e-01 2.29135603e-01
7.73987949e-01 -9.03212726e-01 -1.67299360e-01 -4.64837998e-01
4.04628128e-01 -3.87596965e-01 -8.65451247e-02 -8.95150363e-01
-1.92103565e-01 -7.95601383e-02 -4.30914089e-02 -3.44541818e-01
-2.43898243e-01 -5.84895670e-01 1.69636399e-01 2.08394513e-01
-7.37762690e-01 3.47250760e-01 -5.06243631e-02 3.47204596e-01
-2.60321110e-01 -3.57910484e-01 4.09567118e-01 -2.78119184e-02
-1.71291098e-01 -3.73119265e-01 -4.01212007e-01 3.19767028e-01
5.41606903e-01 -1.89515576e-01 -1.27086893e-01 -9.58255380e-02
-2.09974155e-01 -6.64504915e-02 6.02599621e-01 5.51030338e-01
3.39785904e-01 -1.36532831e+00 -1.18100274e+00 3.60863090e-01
4.72965181e-01 1.02465205e-01 -3.73088032e-01 7.21458614e-01
-3.19530129e-01 9.60109979e-02 2.22300384e-02 -3.19226414e-01
-1.15253639e+00 2.07121119e-01 -1.69684529e-01 -5.10993361e-01
-4.66086984e-01 3.68266553e-01 -7.69525170e-01 -1.08197892e+00
3.75804976e-02 -5.24311960e-01 2.37968147e-01 -7.88630545e-02
3.97188157e-01 4.56742972e-01 5.82496762e-01 -4.33359027e-01
1.60242632e-01 -2.51049697e-01 -1.93794489e-01 -6.63528860e-01
1.29831135e+00 -1.80439502e-01 9.93093997e-02 5.23342013e-01
9.56002057e-01 4.79241908e-01 -1.02928829e+00 -6.31894588e-01
6.19387746e-01 -5.80686152e-01 -2.98285842e-01 -1.09954774e+00
-3.73386562e-01 7.30610609e-01 3.16398770e-01 1.94599722e-02
8.66532266e-01 -5.09516835e-01 9.52963471e-01 4.63083744e-01
2.88962036e-01 -1.46634793e+00 2.09711164e-01 6.98887706e-01
8.69626522e-01 -1.41974068e+00 -2.26580068e-01 -2.53865272e-01
-7.38519311e-01 8.90801311e-01 4.59889829e-01 2.49119878e-01
-1.36966929e-01 1.44803628e-01 4.48601782e-01 3.54333460e-01
-6.62765324e-01 2.10829824e-02 4.10262309e-02 5.54868221e-01
4.57142562e-01 1.40902862e-01 -7.36771882e-01 6.44802451e-01
-3.48377258e-01 3.13100219e-01 7.97963083e-01 1.03896368e+00
-1.17451534e-01 -1.87720132e+00 -4.48652267e-01 4.57758754e-01
-4.93103236e-01 -5.61951101e-01 -6.92600965e-01 4.88281012e-01
2.66619295e-01 9.94949102e-01 -2.16359079e-01 -4.77061003e-01
4.93597180e-01 5.09265780e-01 2.97196865e-01 -8.57223213e-01
-8.80785406e-01 -6.04237318e-02 3.90999198e-01 -1.91724166e-01
-3.38858545e-01 -6.36175871e-01 -1.01683426e+00 -8.66803750e-02
-5.57547688e-01 2.19822437e-01 8.87048841e-01 1.07781792e+00
5.48600852e-01 3.72657329e-01 8.47541928e-01 -6.03037596e-01
-9.22109067e-01 -1.48278713e+00 -6.17200881e-02 8.21324587e-01
2.82269269e-01 -2.15481207e-01 -2.78528601e-01 3.55214149e-01] | [11.584450721740723, 9.74827766418457] |
51ffcdad-b72c-4085-815e-0712ac3a11e9 | geometry-aware-reference-synthesis-for-multi | 2207.08601 | null | https://arxiv.org/abs/2207.08601v2 | https://arxiv.org/pdf/2207.08601v2.pdf | Geometry-Aware Reference Synthesis for Multi-View Image Super-Resolution | Recent multi-view multimedia applications struggle between high-resolution (HR) visual experience and storage or bandwidth constraints. Therefore, this paper proposes a Multi-View Image Super-Resolution (MVISR) task. It aims to increase the resolution of multi-view images captured from the same scene. One solution is to apply image or video super-resolution (SR) methods to reconstruct HR results from the low-resolution (LR) input view. However, these methods cannot handle large-angle transformations between views and leverage information in all multi-view images. To address these problems, we propose the MVSRnet, which uses geometry information to extract sharp details from all LR multi-view to support the SR of the LR input view. Specifically, the proposed Geometry-Aware Reference Synthesis module in MVSRnet uses geometry information and all multi-view LR images to synthesize pixel-aligned HR reference images. Then, the proposed Dynamic High-Frequency Search network fully exploits the high-frequency textural details in reference images for SR. Extensive experiments on several benchmarks show that our method significantly improves over the state-of-the-art approaches. | ['Chenxi Ma', 'Weimin Tan', 'Bo Yan', 'Yuqi Sun', 'Ri Cheng'] | 2022-07-18 | null | null | null | null | ['video-super-resolution'] | ['computer-vision'] | [ 5.50360084e-01 -3.95777225e-01 -2.15208396e-01 -1.40567228e-01
-1.06015933e+00 -2.89813310e-01 2.16568395e-01 -8.88479531e-01
2.48971004e-02 7.56202877e-01 4.37063366e-01 2.40390331e-01
-3.46432701e-02 -8.84733558e-01 -6.25795305e-01 -6.78058863e-01
4.84249920e-01 -3.19487780e-01 5.45052230e-01 -3.62185776e-01
4.18271720e-01 8.14878345e-01 -1.77596164e+00 7.69734323e-01
5.78695476e-01 8.60791028e-01 8.05735290e-01 7.54489541e-01
2.35242665e-01 7.38065362e-01 -1.08411506e-01 1.38157859e-01
4.95571375e-01 -3.22089195e-01 -6.17367327e-01 9.87042040e-02
1.11338961e+00 -9.61074769e-01 -5.86856782e-01 1.03634119e+00
6.43940628e-01 3.38832140e-01 -1.80889130e-01 -4.19135809e-01
-7.81449378e-01 2.56171495e-01 -1.15053427e+00 7.34041810e-01
7.21063852e-01 -6.97259754e-02 7.75875866e-01 -1.49317467e+00
1.02128160e+00 1.41753161e+00 2.22894266e-01 3.79044175e-01
-1.10893714e+00 -6.03217661e-01 7.55384862e-02 2.50018865e-01
-1.42925549e+00 -7.45666742e-01 9.36698556e-01 2.27617055e-01
7.31215417e-01 3.54413331e-01 3.29145521e-01 8.41423869e-01
1.50270298e-01 3.05508882e-01 1.24884272e+00 -2.15990707e-01
-1.77910864e-01 -1.38264582e-01 -3.04458886e-01 3.92672002e-01
2.10695669e-01 5.11935413e-01 -1.05098462e+00 1.20232813e-01
1.55446708e+00 2.96053529e-01 -7.98551261e-01 -2.04175159e-01
-1.35447407e+00 3.54987204e-01 3.09007376e-01 4.29539472e-01
-4.15105194e-01 -1.93378001e-01 -2.61007168e-04 1.44810349e-01
4.57516879e-01 2.24492475e-01 -3.18315595e-01 3.62041801e-01
-9.24935341e-01 -2.38836575e-02 7.47059099e-03 9.88237798e-01
7.44992197e-01 2.57587224e-01 1.25638796e-02 1.10537362e+00
1.79947857e-02 5.91922581e-01 1.65476874e-01 -1.31511509e+00
8.19929779e-01 1.90890670e-01 1.66670188e-01 -1.20142913e+00
-1.47714466e-01 -4.93282825e-01 -1.25210023e+00 6.35627061e-02
-8.58200118e-02 3.00706327e-01 -7.54045665e-01 1.31821895e+00
4.61668462e-01 1.20241322e-01 2.66673595e-01 1.40512109e+00
1.18063498e+00 8.01369905e-01 -6.52321756e-01 -7.71992505e-01
1.04004514e+00 -9.83029008e-01 -6.71981812e-01 -2.94776887e-01
-3.92418534e-01 -1.07398069e+00 7.60143459e-01 6.07012689e-01
-1.75531709e+00 -1.13855112e+00 -1.13766694e+00 -4.11751956e-01
2.56084651e-01 1.09782055e-01 8.49898160e-02 1.45721555e-01
-1.15736651e+00 4.06713158e-01 -4.22128141e-01 5.96356578e-02
1.84948817e-01 -1.52261723e-02 -4.15386289e-01 -8.74307096e-01
-1.09236133e+00 6.01258636e-01 2.35618398e-01 1.46818087e-01
-6.96794689e-01 -8.27483237e-01 -9.48839545e-01 -5.57504669e-02
6.02682054e-01 -9.17163670e-01 7.41423488e-01 -7.08992302e-01
-1.40034521e+00 7.50751317e-01 -4.27889526e-01 7.29481801e-02
1.18315883e-01 -2.39887550e-01 -7.54504859e-01 6.89228952e-01
3.99121121e-02 3.29736620e-01 1.06138492e+00 -1.81322062e+00
-9.06528115e-01 -6.03791952e-01 -8.89927428e-03 5.99612713e-01
2.46339872e-01 -2.83792871e-03 -6.78259432e-01 -6.93500102e-01
7.05282688e-01 -3.36598366e-01 -1.30536422e-01 -2.27850184e-01
-2.62684137e-01 5.33778965e-01 9.64508832e-01 -9.41192865e-01
1.09464419e+00 -2.16081166e+00 4.40647483e-01 -1.52940169e-01
3.95748526e-01 8.87307376e-02 -2.01067090e-01 -2.55521107e-03
-1.70106485e-01 -2.30294839e-01 2.51291305e-01 -1.40330583e-01
-8.66545737e-01 3.45941298e-02 -4.50512826e-01 5.21911442e-01
-1.37472898e-01 6.11551464e-01 -6.76053464e-01 -4.99187589e-01
5.19654334e-01 9.32710350e-01 -3.86063814e-01 3.92233849e-01
2.54448175e-01 8.75183165e-01 -3.90130311e-01 6.12130165e-01
1.25949323e+00 -4.72975940e-01 2.24273428e-01 -9.78207767e-01
-3.88051599e-01 -9.47713330e-02 -1.38950264e+00 2.01736379e+00
-7.45879591e-01 3.25104952e-01 2.37961188e-01 -3.61462593e-01
9.57033515e-01 -2.71453969e-02 5.68486035e-01 -1.02820814e+00
-3.32482606e-01 2.49173164e-01 -5.11594892e-01 -2.76344270e-01
9.21960950e-01 3.13454345e-02 3.48995358e-01 1.50562689e-01
-2.87831724e-01 -3.18597034e-02 -1.63740307e-01 4.75709774e-02
5.20209074e-01 1.55482784e-01 5.46744525e-01 3.64421487e-01
1.11672604e+00 -6.10079110e-01 5.76819718e-01 3.32454860e-01
2.86214650e-01 1.15801859e+00 -3.23396176e-01 -6.39228582e-01
-1.30770910e+00 -1.45410359e+00 -2.70510893e-02 8.65750372e-01
5.92742503e-01 -2.38082767e-01 -2.47955695e-01 -1.12083286e-01
-6.02208138e-01 3.45992684e-01 -2.87003696e-01 2.47433603e-01
-1.08792198e+00 -2.97884166e-01 -3.55943918e-01 3.41522813e-01
9.43505585e-01 -6.94084227e-01 -7.64623523e-01 9.77965072e-03
-7.08126009e-01 -1.67479157e+00 -7.71549881e-01 -4.14894760e-01
-9.34672236e-01 -9.53333914e-01 -9.53473330e-01 -5.74091256e-01
4.98714387e-01 1.18256545e+00 1.22477305e+00 -3.88332307e-02
-2.45647579e-01 2.04283416e-01 -2.61170030e-01 6.27526104e-01
-1.58542290e-01 -2.81018794e-01 7.72490259e-03 2.17639774e-01
-3.21679652e-01 -8.92660022e-01 -9.77928936e-01 6.13487661e-01
-9.13709939e-01 5.96723795e-01 8.06476772e-01 8.57143879e-01
1.32695436e+00 2.28466883e-01 2.98051178e-01 -5.30065358e-01
-8.22159052e-02 -7.56527483e-02 -6.77971065e-01 3.25180441e-01
-4.39419210e-01 -1.64830774e-01 7.79633880e-01 -1.59863278e-01
-1.48387468e+00 1.28290439e-02 1.10987298e-01 -9.48692918e-01
2.54227575e-02 3.04241478e-03 -4.01194602e-01 -2.37490669e-01
2.98967540e-01 7.98219442e-01 -3.33575785e-01 -4.78812724e-01
4.34132278e-01 4.81884927e-01 8.53923619e-01 -1.40018582e-01
9.08401966e-01 8.86917591e-01 2.13540390e-01 -9.18793261e-01
-1.24159110e+00 -4.55337077e-01 -6.41884089e-01 -2.91584522e-01
1.06621313e+00 -1.34138727e+00 -4.16289419e-01 1.15907229e-01
-1.13437462e+00 1.17620461e-01 -7.78474957e-02 5.02555132e-01
-7.73586094e-01 6.61757708e-01 -5.51720977e-01 -5.70706844e-01
-5.68990111e-01 -1.25292087e+00 1.44097579e+00 5.25593936e-01
5.01767993e-01 -4.12383527e-01 -1.83649242e-01 6.79994285e-01
6.03333473e-01 5.32757379e-02 4.04327720e-01 3.27549368e-01
-1.22692287e+00 5.36154807e-01 -6.64137125e-01 2.45880708e-01
2.92903394e-01 -3.91260058e-01 -8.38434815e-01 -5.40200889e-01
2.39716277e-01 -1.64442211e-02 6.54726624e-01 6.22368813e-01
1.35465467e+00 -2.73380905e-01 -6.35801926e-02 1.22306156e+00
1.96146464e+00 -5.01302653e-04 9.22813356e-01 3.58679503e-01
1.01396549e+00 4.02778804e-01 1.00034797e+00 5.13881028e-01
3.91104490e-01 1.25281978e+00 3.97109181e-01 -2.19232380e-01
-4.81344759e-01 -2.05163404e-01 2.96604037e-01 6.77696943e-01
-4.07339633e-01 -1.27795693e-02 -2.78340459e-01 1.49639875e-01
-1.47430611e+00 -1.36886775e+00 -8.60018134e-02 2.19963479e+00
6.44140899e-01 -2.66631126e-01 -1.69187501e-01 -1.08618662e-01
8.15984547e-01 6.63639307e-01 -7.62367725e-01 3.86305563e-02
-5.36808193e-01 1.19689278e-01 5.61842680e-01 4.88891393e-01
-6.75549686e-01 6.51132762e-01 5.68996811e+00 9.16283727e-01
-1.09785199e+00 1.41165301e-01 7.73684442e-01 -3.66534829e-01
-3.39145780e-01 -3.82919133e-01 -9.90782022e-01 1.27350226e-01
5.07510364e-01 -3.42468098e-02 8.55004251e-01 5.57133377e-01
3.24136645e-01 -2.59368747e-01 -8.94985199e-01 1.69237423e+00
4.61444557e-01 -1.54415965e+00 1.75637618e-01 6.17747009e-02
1.05584836e+00 -9.02155861e-02 3.29961121e-01 -3.43379855e-01
-9.46420729e-02 -9.87837732e-01 3.63975763e-01 7.14530051e-01
1.58240843e+00 -1.13166177e+00 3.72227699e-01 8.13692287e-02
-1.68690085e+00 -1.92344159e-01 -4.92482543e-01 6.17226720e-01
5.34965336e-01 5.32110929e-01 2.23147823e-03 1.13916647e+00
1.11422157e+00 9.61860478e-01 -4.24838990e-01 3.11785340e-01
1.68658361e-01 -3.21646124e-01 1.63234264e-01 1.03400004e+00
-2.68140644e-01 -1.59435287e-01 7.08776355e-01 5.91203451e-01
5.49146295e-01 6.27597451e-01 1.10837497e-01 8.61673951e-01
1.49325386e-01 -1.90022036e-01 -6.71054065e-01 5.53448796e-01
4.59383696e-01 1.33034253e+00 -3.62440675e-01 -2.42962867e-01
-5.86857319e-01 1.26939690e+00 -5.98892868e-02 4.99868810e-01
-6.29068017e-01 5.14587760e-02 5.38915753e-01 2.92995155e-01
4.84172672e-01 -2.31190428e-01 -1.97726622e-01 -1.46707654e+00
2.08357692e-01 -1.17722428e+00 3.66574347e-01 -1.48738706e+00
-1.08876407e+00 8.33846152e-01 -7.39661604e-02 -1.63155901e+00
-3.56318891e-01 -2.62831189e-02 -2.32570502e-03 1.19443727e+00
-2.01458240e+00 -1.09278166e+00 -7.52435029e-01 1.00328362e+00
1.10504103e+00 -7.88005367e-02 2.38009691e-01 2.89562434e-01
-1.55966520e-01 3.34911376e-01 -5.91615699e-02 -2.78050512e-01
8.66677940e-01 -6.07538223e-01 1.87950104e-01 1.05398178e+00
-2.03450561e-01 4.90162343e-01 5.54095387e-01 -6.28017366e-01
-1.70436072e+00 -1.06856322e+00 3.64414692e-01 -1.28843307e-01
-9.34083089e-02 1.62006035e-01 -9.56623733e-01 3.92821133e-01
4.09498513e-02 5.22027850e-01 2.58755416e-01 -5.44444859e-01
-4.51740772e-01 -3.28815043e-01 -1.18782139e+00 4.93221700e-01
1.32354414e+00 -5.37182093e-01 -3.15616250e-01 -1.86948255e-01
9.41269517e-01 -8.65113556e-01 -1.09749126e+00 5.24341643e-01
6.10313833e-01 -1.61351609e+00 1.88865113e+00 6.95240498e-02
9.25033808e-01 -5.36752284e-01 -7.04127073e-01 -1.07785046e+00
-4.12000179e-01 -4.05092925e-01 -3.52618963e-01 8.84213686e-01
-1.97413743e-01 -4.64617580e-01 3.29711586e-01 8.63123313e-02
-3.13931592e-02 -6.38695359e-01 -7.97532320e-01 -5.02166271e-01
-5.69549680e-01 1.73672631e-01 5.29956341e-01 9.17497277e-01
-6.51582837e-01 4.19719815e-01 -7.47604907e-01 6.52008533e-01
1.15909719e+00 7.35900760e-01 7.60040522e-01 -9.02943611e-01
-4.74351376e-01 6.42737746e-03 -1.53733734e-02 -1.21981370e+00
-2.68258214e-01 -3.73321414e-01 -2.42973134e-01 -1.55050540e+00
4.00647163e-01 1.20644588e-02 1.40654715e-02 -3.56548816e-01
-1.51219800e-01 4.65474337e-01 2.37397999e-01 3.29468161e-01
-5.73510051e-01 3.74670833e-01 1.87099135e+00 3.02795410e-01
-3.07564706e-01 -2.36668885e-01 -6.99749172e-01 5.38322806e-01
4.53949571e-01 1.14261329e-01 -4.76630837e-01 -5.02719223e-01
3.03855985e-01 9.81656611e-01 4.81061876e-01 -8.37423503e-01
2.31815428e-01 -4.91830915e-01 8.84266675e-01 -1.27463663e+00
6.69097424e-01 -8.16212296e-01 4.94712532e-01 5.68640791e-03
-2.86542982e-01 2.16478109e-01 -8.41180906e-02 6.90764785e-01
-3.64447325e-01 4.18444246e-01 1.20359969e+00 -4.40135896e-01
-7.96982706e-01 5.90044439e-01 2.96578556e-01 -8.65789056e-02
6.95654511e-01 -5.07895231e-01 -5.64753771e-01 -2.61109054e-01
-6.89979851e-01 -1.00748472e-01 8.02815676e-01 4.47233617e-01
1.42082465e+00 -1.32278240e+00 -7.54932404e-01 4.51498866e-01
-8.89965966e-02 3.29336196e-01 1.13035500e+00 6.77150667e-01
-4.26665992e-01 3.44257891e-01 -5.95799983e-01 -6.33150697e-01
-1.65598989e+00 7.95717120e-01 3.53173435e-01 -3.45744073e-01
-1.06586802e+00 4.28072989e-01 6.00229204e-01 1.22860506e-01
-1.28052697e-01 9.60331410e-02 -5.04158199e-01 -3.10050517e-01
1.15392983e+00 5.88905573e-01 -1.01406619e-01 -8.78675818e-01
-7.54975751e-02 1.28482807e+00 -2.54365802e-01 -2.82411128e-01
1.69778299e+00 -9.79436755e-01 -4.91517410e-02 2.70283788e-01
1.24488211e+00 2.76038319e-01 -1.23942351e+00 -5.67434430e-01
-7.45105326e-01 -1.22809720e+00 3.28095138e-01 -4.00615394e-01
-1.54567802e+00 6.63967907e-01 7.28145540e-01 -2.98794597e-01
1.75141525e+00 -1.32711470e-01 9.79878008e-01 -3.38231921e-02
7.52476335e-01 -9.59614158e-01 3.43681693e-01 2.23961815e-01
1.12810075e+00 -1.20721233e+00 4.08617854e-01 -7.91606307e-01
-5.02697945e-01 1.43169498e+00 8.97140265e-01 8.47353879e-03
2.69268245e-01 1.05594568e-01 -2.70015776e-01 -1.96912676e-01
-7.84763932e-01 -7.69323669e-04 3.09841394e-01 5.79119802e-01
2.54941583e-01 -4.41783518e-01 -2.85091400e-02 1.22633986e-01
1.61056742e-02 2.39163749e-02 8.35541964e-01 4.16545272e-01
-4.30994421e-01 -6.55729055e-01 -6.33256793e-01 1.36976629e-01
-5.08574903e-01 -2.62486845e-01 1.15976587e-01 4.89531994e-01
-1.25885680e-01 9.94559944e-01 -7.07815960e-02 -4.43371773e-01
3.68769258e-01 -6.61479712e-01 6.67295158e-01 -4.53170717e-01
-8.82787332e-02 3.09943318e-01 -1.84170052e-01 -1.26670814e+00
-7.65331149e-01 -4.67391491e-01 -8.82915974e-01 -2.86056459e-01
5.33427782e-02 -4.53966111e-01 3.16414773e-01 3.83134574e-01
4.19585615e-01 5.09725034e-01 1.11622727e+00 -1.16986156e+00
-2.64771163e-01 -5.08830845e-01 -7.44293630e-01 1.64763123e-01
6.97876036e-01 -4.48170394e-01 -4.92559582e-01 3.12386472e-02] | [10.808774948120117, -2.139523983001709] |
ec21df2c-03e5-4dac-80f9-123576a7aaed | automated-segmentation-of-hip-and-thigh | 1906.11484 | null | https://arxiv.org/abs/1906.11484v1 | https://arxiv.org/pdf/1906.11484v1.pdf | Automated Segmentation of Hip and Thigh Muscles in Metal Artifact-Contaminated CT using Convolutional Neural Network-Enhanced Normalized Metal Artifact Reduction | In total hip arthroplasty, analysis of postoperative medical images is important to evaluate surgical outcome. Since Computed Tomography (CT) is most prevalent modality in orthopedic surgery, we aimed at the analysis of CT image. In this work, we focus on the metal artifact in postoperative CT caused by the metallic implant, which reduces the accuracy of segmentation especially in the vicinity of the implant. Our goal was to develop an automated segmentation method of the bones and muscles in the postoperative CT images. We propose a method that combines Normalized Metal Artifact Reduction (NMAR), which is one of the state-of-the-art metal artifact reduction methods, and a Convolutional Neural Network-based segmentation using two U-net architectures. The first U-net refines the result of NMAR and the muscle segmentation is performed by the second U-net. We conducted experiments using simulated images of 20 patients and real images of three patients to evaluate the segmentation accuracy of 19 muscles. In simulation study, the proposed method showed statistically significant improvement (p<0.05) in the average symmetric surface distance (ASD) metric for 14 muscles out of 19 muscles and the average ASD of all muscles from 1.17 +/- 0.543 mm (mean +/- std over all patients) to 1.10 +/- 0.509 mm over our previous method. The real image study using the manual trace of gluteus maximus and medius muscles showed ASD of 1.32 +/- 0.25 mm. Our future work includes training of a network in an end-to-end manner for both the metal artifact reduction and muscle segmentation. | ['Masaki Takao', 'Yoshito Otake', 'Yoshinobu Sato', 'Nobuhiko Sugano', 'Yuta Hiasa', 'Mitsuki Sakamoto', 'Yuki Suzuki'] | 2019-06-27 | null | null | null | null | ['metal-artifact-reduction'] | ['medical'] | [ 3.09661180e-02 2.61408985e-01 3.58198524e-01 2.30296031e-01
-7.29775071e-01 8.35739002e-02 -2.34592050e-01 -1.85973570e-01
-6.78752363e-01 6.95211351e-01 7.28463158e-02 -6.56129047e-02
-8.32819864e-02 -6.39678657e-01 -8.38438570e-01 -6.59272909e-01
-3.22831810e-01 5.09538770e-01 5.92346013e-01 -9.10119563e-02
2.50132382e-01 5.72584748e-01 -8.97952735e-01 3.10594231e-01
6.93411112e-01 1.01550722e+00 3.78127545e-01 3.61494064e-01
1.39535412e-01 4.85889465e-01 -5.03728986e-01 9.91344601e-02
5.52034616e-01 -3.01395386e-01 -6.46932602e-01 1.55460775e-01
2.30956942e-01 -1.59152776e-01 -2.40921184e-01 1.12002707e+00
7.55697012e-01 -1.28188625e-01 5.51877499e-01 -6.21904612e-01
-3.07107985e-01 7.08485007e-01 -9.31447625e-01 2.10953087e-01
-2.89498866e-01 1.26526922e-01 6.38947189e-02 -6.54811919e-01
7.18129456e-01 7.03836620e-01 9.28678215e-01 7.12721348e-01
-9.81853902e-01 -8.91008973e-01 -4.63241845e-01 -1.46625906e-01
-1.29938877e+00 2.16252044e-01 7.87444711e-01 -7.35602200e-01
7.76724875e-01 1.68224424e-01 9.10693705e-01 2.18382239e-01
1.11818111e+00 6.28220737e-01 1.03082740e+00 -4.29112375e-01
1.31772757e-01 -3.36053222e-01 -3.95244062e-02 6.54165924e-01
5.57733119e-01 9.43270624e-02 1.33204237e-01 8.53339136e-02
1.35485482e+00 4.30235863e-02 -3.55897278e-01 -3.56774777e-01
-1.07595360e+00 5.24312079e-01 6.34963214e-01 4.18079257e-01
-6.21041477e-01 5.29911458e-01 4.07472819e-01 -1.14499629e-01
3.87960225e-01 4.18657720e-01 -3.76511157e-01 5.02984598e-02
-8.85082424e-01 1.69478238e-01 3.53320032e-01 5.03164113e-01
1.52183354e-01 9.33246538e-02 -6.73891604e-02 1.00326133e+00
3.71056378e-01 2.73512095e-01 1.13351333e+00 -9.15020168e-01
1.41599014e-01 5.89060009e-01 -1.54262617e-01 -6.15403950e-01
-7.65975296e-01 -6.51823401e-01 -8.35258186e-01 6.19777560e-01
4.34479505e-01 -4.46447104e-01 -1.41318691e+00 1.37649512e+00
1.90038323e-01 9.47234966e-03 -2.99070746e-01 1.12245119e+00
5.47115505e-01 3.98552045e-02 -1.82186365e-01 -3.81577700e-01
1.29067540e+00 -9.79146123e-01 -6.52608275e-01 -1.50142655e-01
8.72600019e-01 -9.87382114e-01 9.27728474e-01 4.93848860e-01
-1.39908481e+00 -3.42588723e-01 -1.10473776e+00 3.23145419e-01
1.97824240e-01 2.89730966e-01 2.40465105e-01 3.80411536e-01
-8.84733438e-01 9.24338520e-01 -1.23891795e+00 -1.98873907e-01
3.87473613e-01 8.98296535e-01 -3.37805420e-01 3.19746643e-01
-8.65659595e-01 1.01644921e+00 2.19769161e-02 9.07648429e-02
-4.29847658e-01 -8.89737606e-01 -4.27288920e-01 -2.99622267e-01
2.14497656e-01 -5.66162407e-01 1.28924978e+00 -9.34912503e-01
-1.53244054e+00 8.73385549e-01 3.62591445e-01 -4.51800346e-01
7.04361916e-01 -4.72530931e-01 -6.92614093e-02 9.05731469e-02
1.77138686e-01 3.42411250e-01 1.62497982e-01 -1.20388806e+00
-3.44590098e-01 -6.40207946e-01 -6.09226465e-01 -2.17079222e-02
9.71532390e-02 -9.10496935e-02 -5.49685001e-01 -1.10157919e+00
6.16180837e-01 -1.28888726e+00 -4.95740801e-01 2.41114900e-01
-2.63049603e-01 2.26444378e-01 6.23900235e-01 -1.09255874e+00
1.25243545e+00 -1.95419073e+00 -6.65347353e-02 3.65132660e-01
1.21925607e-01 1.08735561e-01 3.68409723e-01 -1.95615441e-01
-3.46587062e-01 -9.84071419e-02 -6.00717247e-01 -1.35614127e-01
-8.39864850e-01 8.42755660e-02 7.08751261e-01 6.52183950e-01
-2.90524483e-01 6.17644668e-01 -3.23984861e-01 -6.83621466e-01
1.53244242e-01 2.28738472e-01 -5.74901164e-01 -3.60495690e-03
1.93475157e-01 5.23129642e-01 -2.79882938e-01 6.38916314e-01
8.07531834e-01 -7.54677085e-03 7.20089898e-02 -4.52836663e-01
-4.11270596e-02 -2.75771946e-01 -1.18126297e+00 1.92161274e+00
-4.07512933e-01 3.29229325e-01 1.47956893e-01 -5.60950935e-01
7.46658981e-01 4.90660429e-01 1.27310491e+00 -7.32253015e-01
6.06359899e-01 5.42248547e-01 5.27929902e-01 -6.70501828e-01
-6.02618232e-02 -5.69973111e-01 9.21788663e-02 4.42131817e-01
-2.60426283e-01 -1.11570232e-01 1.40574379e-02 -1.44206747e-01
9.37559724e-01 1.03991210e-01 1.24166593e-01 -5.47307193e-01
3.34874690e-01 1.74464002e-01 6.25674784e-01 2.06711322e-01
-2.77490973e-01 1.05739141e+00 2.00020537e-01 -4.52326030e-01
-9.57384884e-01 -1.12492633e+00 -2.50739455e-01 6.65788800e-02
2.17879876e-01 1.75739422e-01 -1.12795520e+00 -4.34214503e-01
7.49511868e-02 3.07521462e-01 -6.56475663e-01 -2.80113637e-01
-9.66355681e-01 -7.46732950e-01 2.50217170e-01 8.24970245e-01
4.24182266e-01 -1.12465322e+00 -7.98098028e-01 4.38001633e-01
-1.14750810e-01 -7.59950101e-01 -5.50883114e-01 2.92495310e-01
-1.54661107e+00 -1.10769916e+00 -9.26149964e-01 -8.44101071e-01
8.07491064e-01 -1.16284333e-01 6.75285041e-01 2.23902076e-01
-6.21086359e-01 -1.55177698e-01 4.53230627e-02 -3.90247494e-01
-3.03968936e-01 -2.15578303e-01 -5.08301482e-02 -4.14333284e-01
8.08006823e-02 -3.83224726e-01 -8.85235965e-01 4.10896480e-01
-8.93411458e-01 1.12117015e-01 9.29406345e-01 5.95939994e-01
8.51955175e-01 2.27600243e-02 -3.36079560e-02 -8.73252451e-01
6.32022083e-01 -1.98518395e-01 -4.12568897e-01 -3.91485244e-02
-6.15787745e-01 1.34937495e-01 2.25714713e-01 -5.62651455e-01
-8.96848023e-01 3.74957502e-01 -1.27015680e-01 -4.90599185e-01
1.48284599e-01 4.64053571e-01 3.03114086e-01 -2.53142983e-01
7.00644016e-01 -2.46424377e-01 4.95794952e-01 -4.54426199e-01
-2.43843585e-01 5.66015184e-01 6.99977875e-01 -3.81545454e-01
2.17704684e-01 4.95561361e-01 1.33740485e-01 -6.34159863e-01
-2.60892838e-01 -4.15546566e-01 -6.40344977e-01 -4.83962357e-01
1.00475204e+00 -6.06468379e-01 -4.39155728e-01 5.26846945e-01
-1.00335968e+00 -4.80800271e-01 -2.36362711e-01 9.85036194e-01
-6.48048520e-01 5.11744976e-01 -1.02574205e+00 -2.97767639e-01
-8.78666520e-01 -1.63388193e+00 7.37981677e-01 1.22083582e-01
-4.81645793e-01 -6.19244814e-01 -2.32195891e-02 4.15663600e-01
3.32879096e-01 4.58673000e-01 8.07150900e-01 -2.27948368e-01
-2.53484040e-01 -3.98166627e-01 9.40873325e-02 4.98479575e-01
3.00754696e-01 -3.17246884e-01 -4.40854758e-01 -6.37801141e-02
4.18093681e-01 2.72624791e-01 5.10195017e-01 1.09018755e+00
1.31040478e+00 2.14980662e-01 -5.12107909e-01 4.29519981e-01
1.76207280e+00 5.54600716e-01 9.54581499e-01 4.92692709e-01
6.26487553e-01 2.69438893e-01 5.99061251e-01 2.44646549e-01
-2.40901560e-01 7.49482691e-01 5.97084165e-01 -1.56419918e-01
-5.97785175e-01 3.18195820e-01 7.30665177e-02 8.68445218e-01
-4.43883359e-01 3.13849896e-01 -9.09346879e-01 6.84400916e-01
-1.68448603e+00 -2.60994911e-01 -6.64558530e-01 2.33236480e+00
9.46724653e-01 3.22633684e-01 -4.20050621e-02 7.65728801e-02
7.01672554e-01 -6.97510481e-01 -3.62076074e-01 -1.49819165e-01
5.17301321e-01 6.66684508e-01 8.96141291e-01 5.62024832e-01
-8.19742382e-01 5.57024062e-01 5.72690630e+00 6.02331698e-01
-1.42553139e+00 2.48780310e-01 2.40187496e-01 -2.55046248e-01
3.32500935e-01 -3.67200881e-01 -2.14416072e-01 6.58025920e-01
6.39468193e-01 1.44567251e-01 -1.36489943e-01 7.99955428e-01
4.01499420e-01 -6.12384140e-01 -6.70214117e-01 7.20005095e-01
-1.46069061e-02 -1.31217265e+00 -2.77350813e-01 1.62715763e-01
6.85359120e-01 1.08521603e-01 -3.70653063e-01 -1.33501336e-01
-4.04904559e-02 -8.08335185e-01 5.59270620e-01 5.89255333e-01
5.32970428e-01 -5.75502217e-01 1.19596255e+00 1.35375271e-02
-1.00436842e+00 1.19049199e-01 -2.04740912e-01 -3.74643616e-02
1.90714478e-01 6.25079811e-01 -7.86027312e-01 2.73622304e-01
8.20629716e-01 1.39956310e-01 -1.50463432e-01 1.33820724e+00
-5.40036820e-02 3.37422460e-01 -4.03954715e-01 1.51934370e-01
2.43972726e-02 -1.49516925e-01 3.60951364e-01 8.52675736e-01
4.08549815e-01 1.46433726e-01 -6.75832201e-03 7.16658235e-01
-1.48354145e-02 2.74143040e-01 -1.50248632e-01 6.26694679e-01
3.06393742e-03 1.04138041e+00 -1.04491162e+00 -2.19558179e-01
-3.23408127e-01 8.34022641e-01 -1.72519907e-01 3.53241190e-02
-7.58323312e-01 -1.81570992e-01 3.24301749e-01 7.28998303e-01
-1.06467023e-01 -1.01571888e-01 -9.97702479e-01 -5.70608497e-01
2.30705351e-01 -5.30750275e-01 3.51118296e-01 -7.93187022e-01
-8.44493508e-01 5.78934491e-01 -1.37075290e-01 -1.35581756e+00
1.92949548e-02 -8.04902792e-01 -5.70020854e-01 8.30874801e-01
-6.55370474e-01 -8.00340474e-01 -5.57871878e-01 4.10289615e-01
6.91014349e-01 1.32246152e-01 4.34867173e-01 4.56753492e-01
-3.76568735e-01 3.49795967e-01 -2.06673648e-02 3.22286546e-01
7.08443999e-01 -9.49161112e-01 -1.18660994e-01 8.21768105e-01
-5.72304964e-01 5.04829526e-01 7.21906245e-01 -1.18848491e+00
-9.99497473e-01 -8.91197920e-01 3.77961040e-01 8.49976391e-02
5.50850809e-01 2.89742947e-01 -7.15064347e-01 6.96982443e-01
7.63222799e-02 8.58579874e-02 4.85050350e-01 -5.48601747e-01
6.05366766e-01 -1.48934135e-02 -1.37477183e+00 6.33262336e-01
6.10380530e-01 3.05644393e-01 -5.39829910e-01 1.13416210e-01
3.97637516e-01 -7.81037033e-01 -1.24306226e+00 8.50871980e-01
8.20362926e-01 -6.78243399e-01 8.91664982e-01 -1.42119348e-01
6.84359848e-01 -3.77387047e-01 1.44730667e-02 -1.14327538e+00
-2.96433210e-01 4.94095758e-02 2.67456979e-01 4.65843618e-01
4.14638460e-01 -3.42058599e-01 1.25573766e+00 6.54244959e-01
-7.16469467e-01 -1.11018145e+00 -9.40064013e-01 -6.25066817e-01
3.79542708e-01 -4.89462286e-01 -1.56618863e-01 6.49959326e-01
-4.56198603e-02 -2.40356356e-01 -5.27942069e-02 7.25271627e-02
6.46800876e-01 -2.10653737e-01 3.91670883e-01 -1.09693396e+00
-3.02889198e-01 -4.00615573e-01 -4.92438495e-01 -5.38602769e-01
-5.20532787e-01 -9.26220119e-01 1.94354683e-01 -2.03000927e+00
1.28562152e-01 -2.38440841e-01 -3.04306746e-01 3.39807004e-01
3.00113540e-02 3.99027973e-01 -4.81579863e-02 3.19443852e-01
4.18211222e-01 1.27517179e-01 1.81442130e+00 -3.23499627e-02
-3.85257423e-01 1.86344430e-01 -3.08437616e-01 8.93532693e-01
9.26336884e-01 -6.46135032e-01 -1.13934815e-01 -5.25074840e-01
-3.53760123e-01 1.15213752e-01 1.45925388e-01 -1.39434779e+00
2.44218096e-01 1.52631685e-01 5.34491420e-01 -7.13989079e-01
8.32269415e-02 -1.19897878e+00 3.86783689e-01 1.32078683e+00
1.09160356e-01 6.79011717e-02 4.40923333e-01 -7.16418773e-02
-9.78885442e-02 -3.43992621e-01 1.19003999e+00 -5.53346515e-01
-3.75909120e-01 1.18018940e-01 -5.01710832e-01 -1.98965266e-01
1.23131156e+00 -5.63041985e-01 2.70944953e-01 1.03752054e-01
-1.17936718e+00 -6.62324876e-02 1.67372480e-01 7.82275349e-02
7.35222399e-01 -1.33870244e+00 -5.31844914e-01 1.13314405e-01
-2.44812682e-01 2.24755540e-01 3.48079324e-01 1.42962503e+00
-1.26373363e+00 7.14634806e-02 -4.33237612e-01 -7.49151826e-01
-1.49630618e+00 2.71861762e-01 8.82417977e-01 -4.46474016e-01
-7.26589084e-01 8.11374068e-01 1.34080023e-01 -1.76519573e-01
1.69025809e-01 -7.46177137e-01 2.51286149e-01 -5.27789056e-01
2.17769053e-02 6.09352827e-01 3.46609682e-01 -3.55417043e-01
-3.84847850e-01 9.67602551e-01 -1.64101183e-01 -1.66279972e-02
1.29137123e+00 1.15680419e-01 -2.85258353e-01 9.38784331e-02
1.08498418e+00 -4.49491329e-02 -9.82425392e-01 1.20458223e-01
-7.70530850e-02 -2.72243083e-01 2.97673255e-01 -9.65821862e-01
-1.68695939e+00 7.23600626e-01 1.36362302e+00 -5.05774498e-01
1.11678779e+00 -6.04180358e-02 1.18846047e+00 -2.77922034e-01
5.13046145e-01 -1.31503785e+00 9.13248137e-02 -4.05030213e-02
9.49683547e-01 -9.75385368e-01 2.10550174e-01 -7.49180734e-01
-4.86625731e-01 1.21929967e+00 8.66167665e-01 -5.75984418e-01
9.06413138e-01 7.11184919e-01 3.56276989e-01 -4.20841366e-01
7.89277107e-02 1.66397005e-01 2.75621533e-01 2.07497239e-01
8.12047124e-01 1.48498863e-01 -9.75331604e-01 6.88660026e-01
-1.84639156e-01 5.34660697e-01 5.19902825e-01 1.38513589e+00
-4.93357837e-01 -9.35872972e-01 -5.88735819e-01 7.02076972e-01
-9.11561906e-01 7.00763017e-02 -3.18347923e-02 1.09149110e+00
2.95835763e-01 6.29247665e-01 5.19624315e-02 -3.35255563e-01
5.74100554e-01 -1.95460424e-01 6.17098272e-01 -6.06083333e-01
-9.02798533e-01 6.74575746e-01 -1.71808541e-01 -5.68561971e-01
-3.10236067e-01 -4.99085724e-01 -1.98842120e+00 5.33589721e-03
-3.87272418e-01 -9.55351517e-02 8.98361564e-01 8.19651365e-01
-1.47605250e-02 1.06415224e+00 1.93239570e-01 -6.68320358e-01
-1.54982105e-01 -1.23632276e+00 -7.89005876e-01 4.72040951e-01
-1.85075611e-01 -9.21111047e-01 -2.35623196e-01 6.99286610e-02] | [13.614556312561035, -2.5695197582244873] |
9a77047b-9547-4cec-8086-7527f0ef0c0f | bidirectional-inference-networks-a-class-of | 1902.02037 | null | http://arxiv.org/abs/1902.02037v1 | http://arxiv.org/pdf/1902.02037v1.pdf | Bidirectional Inference Networks: A Class of Deep Bayesian Networks for Health Profiling | We consider the problem of inferring the values of an arbitrary set of
variables (e.g., risk of diseases) given other observed variables (e.g.,
symptoms and diagnosed diseases) and high-dimensional signals (e.g., MRI images
or EEG). This is a common problem in healthcare since variables of interest
often differ for different patients. Existing methods including Bayesian
networks and structured prediction either do not incorporate high-dimensional
signals or fail to model conditional dependencies among variables. To address
these issues, we propose bidirectional inference networks (BIN), which stich
together multiple probabilistic neural networks, each modeling a conditional
dependency. Predictions are then made via iteratively updating variables using
backpropagation (BP) to maximize corresponding posterior probability.
Furthermore, we extend BIN to composite BIN (CBIN), which involves the
iterative prediction process in the training stage and improves both accuracy
and computational efficiency by adaptively smoothing the optimization
landscape. Experiments on synthetic and real-world datasets (a sleep study and
a dermatology dataset) show that CBIN is a single model that can achieve
state-of-the-art performance and obtain better accuracy in most inference tasks
than multiple models each specifically trained for a different task. | ['Ming-Min Zhao', 'Tommi S. Jaakkola', 'Chengzhi Mao', 'Hao Wang', 'Dina Katabi', 'Hao He'] | 2019-02-06 | null | null | null | null | ['sleep-quality-prediction'] | ['medical'] | [ 4.16202813e-01 2.54252583e-01 -3.72614831e-01 -7.43897736e-01
-6.18213713e-01 6.00092001e-02 3.50599289e-01 1.33263394e-01
-4.56807613e-01 1.35429156e+00 2.09021702e-01 -1.83386266e-01
-6.86819613e-01 -7.61847973e-01 -7.57075012e-01 -8.76729608e-01
-1.26921922e-01 9.11010742e-01 2.61961762e-02 4.83550012e-01
-6.11295775e-02 2.63615876e-01 -1.05054474e+00 -2.76552811e-02
8.48409176e-01 9.24803555e-01 3.80526751e-01 4.37524855e-01
2.51087338e-01 8.42202485e-01 -4.51782107e-01 -4.73771989e-01
-3.43352318e-01 -2.72921622e-01 -6.41810179e-01 -1.72571495e-01
-3.37599456e-01 -1.41427636e-01 -2.03259677e-01 9.82802749e-01
3.66204232e-01 -2.38227081e-02 8.18633974e-01 -1.18713582e+00
-4.51287955e-01 9.20011282e-01 -4.09466058e-01 8.57224539e-02
-1.26772135e-01 -9.42356810e-02 1.02136421e+00 -4.01254237e-01
3.02666426e-01 1.20977759e+00 6.16012096e-01 4.11163211e-01
-1.74258566e+00 -6.75799847e-01 3.15017700e-01 4.25662100e-01
-1.34291542e+00 -1.14772834e-01 6.79961383e-01 -4.76527810e-01
7.61397898e-01 2.44915605e-01 5.33084571e-01 1.39191735e+00
5.95863163e-01 5.44825137e-01 1.09324837e+00 -5.74760437e-02
3.60409021e-01 2.28730232e-01 2.68699169e-01 4.56482798e-01
2.28278160e-01 1.26567647e-01 -6.36466146e-01 -5.30779779e-01
5.68137050e-01 3.04736644e-01 -3.90864402e-01 4.22367500e-03
-1.37096334e+00 8.01331580e-01 3.55043262e-01 -3.40391062e-02
-7.37994015e-01 1.55247286e-01 -4.81345132e-02 -2.52535045e-01
3.23458403e-01 3.53294849e-01 -7.25616097e-01 4.02031653e-02
-8.95662308e-01 1.40867755e-01 8.13917339e-01 5.01778364e-01
5.69017649e-01 -3.50191861e-01 -4.15322393e-01 9.76868331e-01
5.49602985e-01 3.66799057e-01 3.16218346e-01 -8.45513105e-01
2.11959258e-01 2.94440627e-01 1.51163518e-01 -9.07871842e-01
-7.51430809e-01 -6.24409378e-01 -1.18427730e+00 -2.22476885e-01
3.22518021e-01 -4.76072967e-01 -1.07209301e+00 2.06832504e+00
2.20567226e-01 4.56601351e-01 -1.07480466e-01 6.28060222e-01
6.15005732e-01 5.63953757e-01 2.25732669e-01 -3.36985856e-01
1.23046052e+00 -6.36830330e-01 -9.30997491e-01 -5.04120588e-01
6.19515702e-02 -1.02572292e-01 3.78420442e-01 6.65251195e-01
-1.07171249e+00 -1.62453815e-01 -7.37175047e-01 2.39833459e-01
-2.41949946e-01 1.07223183e-01 5.91527939e-01 3.60401571e-01
-7.98785329e-01 5.66783667e-01 -1.32457852e+00 5.21866307e-02
4.91413921e-01 6.43531859e-01 -7.05206543e-02 3.45288739e-02
-1.38639998e+00 9.53927398e-01 4.16508913e-01 2.73034990e-01
-8.51375937e-01 -7.68143296e-01 -6.19147480e-01 2.34674275e-01
4.39774811e-01 -1.01396024e+00 8.59195054e-01 -7.21182823e-01
-1.43151581e+00 2.42177114e-01 -4.36475277e-01 -5.76328516e-01
2.84772694e-01 -2.35264763e-01 -4.17645663e-01 -7.22641721e-02
-9.77102220e-02 7.62878537e-01 8.00848722e-01 -9.37349737e-01
-3.69281977e-01 -5.45940936e-01 -1.86242446e-01 4.65783142e-02
-2.44763285e-01 -2.27499250e-02 -5.67615151e-01 -4.49744463e-01
2.25072473e-01 -9.19646740e-01 -4.61641550e-01 -2.48859357e-02
-9.05534923e-01 -1.40873432e-01 1.42312199e-01 -9.61980462e-01
1.34186268e+00 -1.94074178e+00 6.16445839e-01 4.29687083e-01
3.03020507e-01 -4.15900379e-01 4.25427318e-01 -1.13827005e-01
4.83132014e-03 1.10009499e-03 -6.87458396e-01 -4.14626688e-01
-1.88295886e-01 7.68826962e-01 5.99927530e-02 4.62775111e-01
2.65078396e-01 6.93158805e-01 -7.52056718e-01 -7.16716826e-01
-4.23159115e-02 7.46005595e-01 -6.80772960e-01 1.86619744e-01
-1.91489652e-01 6.38489485e-01 -5.11862934e-01 4.46462780e-01
2.92525113e-01 -8.10460508e-01 6.06074393e-01 -1.97481871e-01
3.91399413e-01 2.99740195e-01 -1.15816486e+00 1.26555538e+00
-4.22065526e-01 3.09649020e-01 -1.29097059e-01 -1.32850266e+00
6.56461358e-01 4.13181245e-01 5.66891372e-01 -2.58368880e-01
1.18874282e-01 -1.39325663e-01 2.17777967e-01 -5.48220277e-01
-1.36678312e-02 -4.56037015e-01 5.70130870e-02 1.14157334e-01
5.29625602e-02 2.51561403e-01 -1.30525544e-01 -1.37248158e-01
1.05334091e+00 -1.56458497e-01 3.56962979e-01 2.34394595e-02
2.56917387e-01 -3.69394481e-01 1.19716835e+00 8.32796693e-01
-6.23635342e-03 4.76109296e-01 8.30269456e-01 -8.24498683e-02
-5.97643077e-01 -1.28062749e+00 -6.83176279e-01 6.00510061e-01
2.21338565e-03 -1.67829394e-01 -4.43646282e-01 -4.99358028e-01
1.12169482e-01 9.55511570e-01 -8.69310439e-01 -1.78455070e-01
-2.39340201e-01 -1.52705979e+00 1.73001111e-01 7.57090390e-01
3.68467495e-02 -9.68044937e-01 -5.61342239e-01 4.35680032e-01
-3.93899262e-01 -9.50327396e-01 1.15347095e-01 5.00086308e-01
-9.41044867e-01 -7.87600875e-01 -3.88024479e-01 -3.40343148e-01
7.68185079e-01 -6.49638772e-01 1.27720857e+00 -2.83134013e-01
-1.38264105e-01 5.56382351e-02 1.83747306e-01 -4.22754496e-01
-1.50843531e-01 -1.52872920e-01 8.03270638e-02 1.26533747e-01
2.86068618e-01 -6.37682259e-01 -6.44138932e-01 3.01440716e-01
-7.27950454e-01 2.86187351e-01 7.20919907e-01 1.28518236e+00
8.92006457e-01 2.58363426e-01 6.11540139e-01 -1.04264665e+00
4.56526458e-01 -9.20025229e-01 -5.95197976e-01 4.22699034e-01
-7.93348610e-01 2.33946368e-01 3.94327104e-01 -5.36222875e-01
-1.16610563e+00 2.54831742e-02 -2.00701386e-01 -2.10073829e-01
-2.86539555e-01 9.28546011e-01 -9.18383300e-02 5.87792218e-01
2.72871763e-01 1.97166175e-01 -7.48553425e-02 -5.21836519e-01
1.64149608e-03 6.23948157e-01 5.32781422e-01 -4.38835800e-01
1.40355334e-01 3.78400028e-01 1.39541164e-01 -4.03386414e-01
-1.03313804e+00 5.93877733e-02 -4.55541909e-01 -4.11029123e-02
1.14703071e+00 -8.49259377e-01 -7.93642581e-01 3.20401788e-01
-9.85068202e-01 -3.33284348e-01 2.23294720e-01 8.88077974e-01
-3.55300426e-01 -1.18802883e-01 -4.94997650e-01 -8.76723647e-01
-1.95485398e-01 -1.30150485e+00 8.97822440e-01 2.68721133e-01
-4.40982431e-01 -1.26064169e+00 -1.49951437e-02 3.77713084e-01
2.22975925e-01 2.35827431e-01 1.23448062e+00 -5.12676895e-01
-3.74328494e-01 -4.31044661e-02 -1.15749493e-01 2.64104098e-01
8.15159529e-02 -1.14350110e-01 -6.89530611e-01 3.95449735e-02
2.30003912e-02 -9.84518602e-02 6.85018063e-01 9.75459039e-01
1.80855751e+00 -3.92107338e-01 -7.04669952e-01 6.06500685e-01
1.23870432e+00 2.95711547e-01 4.38201398e-01 2.44466346e-02
5.00356615e-01 5.08753836e-01 2.65805662e-01 5.95943749e-01
7.26937473e-01 7.35662937e-01 5.25678158e-01 1.88138243e-02
3.46343935e-01 3.09366565e-02 4.73400913e-02 5.88040531e-01
2.77075637e-02 -2.29572296e-01 -9.82448161e-01 5.77634573e-01
-1.97493303e+00 -6.03456259e-01 -4.57492918e-02 2.13297248e+00
1.26898551e+00 2.91343838e-01 -1.10188581e-01 -1.27837107e-01
6.61778033e-01 -2.14324504e-01 -9.50105667e-01 -1.06742680e-01
1.33113667e-01 1.57223821e-01 5.24960637e-01 4.48100120e-01
-7.83445299e-01 3.58041734e-01 6.60865688e+00 4.62521642e-01
-1.01746809e+00 2.08808631e-01 1.08470285e+00 -3.34144503e-01
-3.04917723e-01 -2.63509005e-01 -7.62805343e-01 7.99560726e-01
1.23419964e+00 1.50215253e-01 5.46816349e-01 4.31757003e-01
2.39525378e-01 -2.16342777e-01 -1.19973946e+00 7.81914949e-01
-9.85264257e-02 -1.18644893e+00 -2.93774307e-01 -4.26218733e-02
6.16170466e-01 3.10896188e-01 -5.29139750e-02 1.12001196e-01
7.01473057e-01 -1.17633593e+00 4.48923796e-01 1.12127495e+00
3.45142841e-01 -5.43646574e-01 7.18966126e-01 5.00759780e-01
-3.83818358e-01 -2.09937066e-01 -9.66208577e-02 1.23955846e-01
4.45911318e-01 1.06000364e+00 -7.66331196e-01 3.62704694e-01
9.81404364e-01 8.65777612e-01 -2.42902443e-01 1.10820985e+00
-5.60208917e-01 1.04227197e+00 -4.84890759e-01 -8.35425779e-02
-1.08547322e-02 -2.11585835e-01 3.90639573e-01 9.75110054e-01
2.25363418e-01 3.16337161e-02 -6.80710375e-02 1.21360278e+00
4.72363792e-02 -4.20589209e-01 -2.61565559e-02 2.16168821e-01
4.58310246e-01 1.05974436e+00 -4.79574054e-01 -3.60464871e-01
-3.71374518e-01 4.88163620e-01 3.26914608e-01 5.51586092e-01
-1.05514538e+00 -1.16926074e-01 6.93554401e-01 -2.69700050e-01
3.51872534e-01 8.02644994e-03 -6.12041354e-01 -9.32278037e-01
-6.17346391e-02 -5.50678194e-01 4.69098210e-01 -8.32307875e-01
-1.53500545e+00 5.24154067e-01 7.46851861e-02 -7.22809434e-01
-3.67464781e-01 -5.20608425e-01 -3.50347489e-01 1.02306437e+00
-1.66400361e+00 -6.59139156e-01 -4.74599394e-04 5.63148022e-01
1.62982404e-01 2.36625418e-01 8.51958334e-01 2.29560301e-01
-9.92130220e-01 2.90686458e-01 3.31022143e-01 -4.64122035e-02
5.85443676e-01 -1.35163367e+00 -1.75148770e-01 4.88879770e-01
-1.59490213e-01 5.52830935e-01 5.99928379e-01 -7.36279964e-01
-1.04554939e+00 -1.06472707e+00 9.00831282e-01 -4.18929309e-01
7.71133482e-01 -3.23716730e-01 -9.73250389e-01 8.53863537e-01
-1.50979668e-01 -1.46405920e-02 9.54938173e-01 5.29584527e-01
1.04864389e-01 -2.09083021e-01 -1.14583993e+00 4.22893286e-01
6.89446688e-01 -3.26109171e-01 -5.80012679e-01 5.04548371e-01
5.26477814e-01 -4.31745768e-01 -1.14344597e+00 3.96658182e-01
6.59449637e-01 -7.24128664e-01 1.15056121e+00 -7.78341830e-01
6.40575349e-01 -8.34896266e-02 -1.12539999e-01 -1.39452183e+00
-4.07854319e-01 -1.71287298e-01 -4.78452653e-01 9.22818005e-01
7.45584190e-01 -7.54100502e-01 5.27231932e-01 1.19729197e+00
2.20975354e-01 -1.15028715e+00 -1.06212926e+00 -4.39033657e-01
-1.85278207e-01 -7.01792300e-01 5.90788126e-01 8.96098912e-01
5.18632159e-02 3.56632888e-01 -6.11483157e-01 5.76596975e-01
7.77359605e-01 1.01996221e-01 5.51400781e-02 -1.30873132e+00
-8.00115943e-01 -3.25301737e-01 -2.38915339e-01 -8.29439163e-01
2.03795180e-01 -6.75571859e-01 2.76329011e-01 -1.62636733e+00
5.79629123e-01 -5.59627831e-01 -7.37865269e-01 6.68686509e-01
-6.28662527e-01 -4.29503545e-02 -3.21717262e-01 -4.53852452e-02
-3.30618918e-01 4.95377004e-01 9.74845171e-01 -1.60128757e-01
-1.70231134e-01 3.02954704e-01 -7.70216167e-01 7.81552732e-01
6.64599657e-01 -8.65618587e-01 -4.52134788e-01 -4.80796397e-01
2.88829684e-01 6.01291358e-01 5.11160612e-01 -7.08966732e-01
3.24464232e-01 -3.57492983e-01 5.98188043e-01 -5.15500665e-01
5.82205415e-01 -8.46569121e-01 3.69729221e-01 4.51891303e-01
-5.04842877e-01 -1.52879700e-01 -5.72808348e-02 8.31705451e-01
-1.63712680e-01 -5.13222180e-02 5.72801411e-01 1.42834157e-01
-8.91941786e-02 3.02879363e-01 -2.81518757e-01 -2.06660822e-01
7.17105091e-01 1.66442811e-01 -5.13111800e-02 -3.37350070e-01
-1.27955544e+00 6.39251471e-01 -2.59002239e-01 1.50479987e-01
6.58105612e-01 -9.53456342e-01 -7.56549358e-01 1.05418146e-01
-2.34713644e-01 2.27838367e-01 3.16164523e-01 1.22822511e+00
1.27841994e-01 2.94322848e-01 5.78809269e-02 -8.37007642e-01
-1.08341324e+00 2.77878195e-01 4.33900118e-01 -5.41887462e-01
-5.11263728e-01 8.82914782e-01 1.96667522e-01 -4.78755832e-01
4.31008250e-01 -3.90974075e-01 -3.78105134e-01 -4.50546071e-02
3.72723669e-01 2.42194086e-01 -2.00205445e-01 -2.62086362e-01
-5.18366694e-01 7.80105293e-02 -9.93318632e-02 -2.24992260e-03
1.79669690e+00 -7.07825273e-02 -2.72136241e-01 6.09271169e-01
8.06533515e-01 -6.04574442e-01 -1.38079858e+00 -4.41039354e-01
-1.13872670e-01 -2.00331911e-01 5.27824104e-01 -1.30768931e+00
-1.20597410e+00 7.17455328e-01 5.60228884e-01 1.38465822e-01
1.22688615e+00 2.13094026e-01 4.46608365e-01 2.87745565e-01
1.18166983e-01 -9.73101079e-01 -2.85918117e-01 1.66717380e-01
6.72535658e-01 -1.12931132e+00 4.41012233e-02 -1.56390876e-01
-7.08815217e-01 8.75087142e-01 4.62805808e-01 2.24372581e-01
1.10290825e+00 3.73262256e-01 -2.47249529e-01 -1.60975456e-01
-1.11376524e+00 2.17534959e-01 3.58058095e-01 3.19364041e-01
2.38875106e-01 4.33916777e-01 -2.10879967e-01 1.06408989e+00
-5.97265363e-02 2.53391743e-01 2.59592563e-01 6.30971193e-01
8.59285891e-02 -9.75381792e-01 -5.44633687e-01 1.09275758e+00
-6.33752525e-01 -1.81891993e-01 7.65798613e-02 3.53834540e-01
1.49625972e-01 8.19417596e-01 1.78481370e-01 -1.91747785e-01
-6.36431351e-02 1.14040487e-01 3.18020731e-01 -4.47325975e-01
-2.40671977e-01 4.19651680e-02 8.64108056e-02 -4.73069131e-01
-4.62271333e-01 -1.00855029e+00 -1.13118279e+00 1.10142335e-01
-3.01905513e-01 -6.64130552e-03 7.78764129e-01 1.28054428e+00
3.35178167e-01 1.16927028e+00 2.84416676e-01 -5.44128418e-01
-4.24395829e-01 -9.43567574e-01 -6.71978951e-01 1.76430523e-01
4.01563019e-01 -9.06750023e-01 -2.39171073e-01 1.25398546e-01] | [7.813964366912842, 5.239936828613281] |
f2f5af39-c536-499f-9d1c-6adb3f6d5d8f | epipolar-guided-deep-object-matching-for | 2007.15540 | null | https://arxiv.org/abs/2007.15540v1 | https://arxiv.org/pdf/2007.15540v1.pdf | Epipolar-Guided Deep Object Matching for Scene Change Detection | This paper describes a viewpoint-robust object-based change detection network (OBJ-CDNet). Mobile cameras such as drive recorders capture images from different viewpoints each time due to differences in camera trajectory and shutter timing. However, previous methods for pixel-wise change detection are vulnerable to the viewpoint differences because they assume aligned image pairs as inputs. To cope with the difficulty, we introduce a deep graph matching network that establishes object correspondence between an image pair. The introduction enables us to detect object-wise scene changes without precise image alignment. For more accurate object matching, we propose an epipolar-guided deep graph matching network (EGMNet), which incorporates the epipolar constraint into the deep graph matching layer used in OBJCDNet. To evaluate our network's robustness against viewpoint differences, we created synthetic and real datasets for scene change detection from an image pair. The experimental results verified the effectiveness of our network. | ['Ken Sakurada', 'Shun Iwase', 'Ryuhei Hamaguchi', 'Yutaka Matsuo', 'Rio Yokota', 'Kento Doi'] | 2020-07-30 | null | null | null | null | ['scene-change-detection'] | ['computer-vision'] | [ 1.20376691e-01 -5.22213995e-01 1.58406779e-01 -4.14573133e-01
1.59690320e-01 -5.59544981e-01 4.13219631e-01 -2.70497531e-01
-3.38562638e-01 2.37202004e-01 -2.42848918e-01 -1.25485405e-01
1.05876334e-01 -9.66384947e-01 -8.57693434e-01 -3.50661993e-01
1.88445091e-01 -1.29557252e-01 7.58988321e-01 -2.22979099e-01
2.74453431e-01 7.93839455e-01 -1.27187645e+00 -1.29241750e-01
5.55105329e-01 8.30619752e-01 4.85699415e-01 7.05569744e-01
2.61242986e-01 7.65515327e-01 -3.28383088e-01 -1.90450296e-01
8.41562986e-01 -3.59729528e-01 -4.69653785e-01 4.74340677e-01
1.05493689e+00 -7.05528975e-01 -1.07033360e+00 1.38939071e+00
5.26428223e-01 2.02319875e-01 2.73217391e-02 -1.71333992e+00
-4.29416239e-01 7.62105063e-02 -6.48715615e-01 5.01814842e-01
5.05325139e-01 3.98400217e-01 5.29551089e-01 -7.06688046e-01
8.68065357e-01 1.19907868e+00 7.57246375e-01 1.87080488e-01
-8.13022614e-01 -7.22318769e-01 2.96886057e-01 5.58476031e-01
-1.47226274e+00 -4.94916797e-01 1.14034474e+00 -3.37716341e-01
9.24843788e-01 1.31731287e-01 9.23576176e-01 5.91491580e-01
3.80599976e-01 3.52739722e-01 5.67476630e-01 -2.03300193e-01
-8.12782720e-02 -1.97232053e-01 -1.25598758e-01 7.84940660e-01
5.46934247e-01 2.54816532e-01 -2.77836859e-01 3.15903872e-01
1.32087636e+00 3.26546341e-01 -6.49850845e-01 -7.05038965e-01
-1.52798545e+00 3.87817234e-01 9.39439833e-01 7.22304434e-02
-2.03938596e-02 4.02143031e-01 2.27393389e-01 4.59378660e-01
3.55680585e-02 2.75042742e-01 -8.93716216e-02 1.76527619e-01
-6.39650285e-01 8.75097662e-02 4.00651336e-01 1.49476063e+00
8.15204322e-01 7.25149810e-02 8.85757804e-02 4.41870242e-01
3.39056522e-01 4.78623331e-01 2.37278491e-01 -9.58339572e-01
5.98155200e-01 7.99187779e-01 4.55439165e-02 -1.89289665e+00
-5.55588245e-01 -2.62754768e-01 -1.05547225e+00 2.45327935e-01
3.69715810e-01 9.03519019e-02 -7.31840909e-01 1.52545524e+00
4.90521312e-01 4.85662788e-01 -1.96442097e-01 1.27328944e+00
9.28853512e-01 3.67350131e-01 -5.69498003e-01 -1.61349714e-01
1.01812768e+00 -1.06103122e+00 -7.32926011e-01 -2.25286886e-01
4.17805016e-01 -7.99470305e-01 7.37866402e-01 5.84712019e-04
-9.65944171e-01 -8.58637273e-01 -1.23128474e+00 -2.16968104e-01
-6.79147914e-02 -2.00469732e-01 3.37954789e-01 3.22833389e-01
-1.16910863e+00 4.04065669e-01 -8.24288905e-01 -5.31944215e-01
1.76361591e-01 3.30220997e-01 -4.83023494e-01 -2.97457129e-01
-1.09423089e+00 6.60451889e-01 3.46161544e-01 4.22095686e-01
-8.58101964e-01 -2.97563404e-01 -1.06469071e+00 -4.65089642e-02
3.35275739e-01 -9.66528773e-01 9.80160534e-01 -1.05765355e+00
-1.41135585e+00 9.10607994e-01 5.56917104e-04 -2.72211373e-01
7.34718680e-01 3.27429473e-02 -5.31398773e-01 2.21197829e-01
-1.23977780e-01 6.10728383e-01 7.10681796e-01 -1.20202398e+00
-6.37454450e-01 -2.54992634e-01 3.98004025e-01 4.15946484e-01
1.92345008e-01 3.81571576e-02 -9.00987566e-01 -4.91662920e-01
6.17750823e-01 -8.61481369e-01 -2.89302081e-01 3.61655265e-01
-4.36600924e-01 2.53196210e-01 1.17588234e+00 -3.76261801e-01
1.10196197e+00 -2.07824087e+00 -2.16568932e-01 1.10911138e-01
3.68943065e-01 2.06475154e-01 -2.47829095e-01 3.56692880e-01
-1.11361988e-01 -3.61144572e-01 -9.36082229e-02 -2.01705113e-01
-4.26157534e-01 1.67621642e-01 3.71106807e-03 8.46995950e-01
-1.81523323e-01 8.58445108e-01 -9.91095960e-01 -3.68195385e-01
5.59198558e-01 1.86295480e-01 -5.33771932e-01 2.21600786e-01
-3.72127816e-02 3.51299763e-01 -7.41925538e-02 4.78843719e-01
1.14769661e+00 -2.26967394e-01 7.96066597e-02 -6.94805801e-01
-1.48778275e-01 7.36470446e-02 -1.57949340e+00 1.82623124e+00
-2.33242605e-02 1.12740469e+00 -1.17702134e-01 -7.75622606e-01
9.02357280e-01 -7.05777407e-02 5.85566938e-01 -8.06733787e-01
2.81426787e-01 -1.07994393e-01 7.06726089e-02 -5.36882639e-01
7.75545359e-01 3.38568389e-01 3.82474929e-01 1.80000305e-01
-3.89412463e-01 -3.24290037e-01 1.96558964e-02 3.12057376e-01
9.71880496e-01 2.68892758e-02 4.21000928e-01 -6.04839735e-02
5.35230219e-01 -3.32231410e-02 9.37712789e-01 6.31408870e-01
-4.77437973e-01 9.90671098e-01 -8.61798506e-03 -8.79969001e-01
-8.67227316e-01 -9.09454346e-01 7.76717961e-02 1.68623507e-01
1.20189428e+00 -2.61344433e-01 -4.52328533e-01 -4.92954403e-01
-9.29711908e-02 3.84446830e-02 -1.22719832e-01 -2.28102684e-01
-7.40265191e-01 -3.45401496e-01 2.56947190e-01 4.23014998e-01
1.18712282e+00 -5.96625566e-01 -6.33877993e-01 1.67586520e-01
-2.87265807e-01 -1.52206945e+00 -9.07427549e-01 -4.65307474e-01
-7.70731866e-01 -1.55315542e+00 -1.44685537e-01 -9.50025201e-01
9.20240343e-01 1.13537896e+00 9.47571099e-01 2.18704939e-01
-3.06284398e-01 4.83213037e-01 -3.69977355e-02 -1.59804653e-02
-2.31627405e-01 -3.47422957e-01 2.38089740e-01 -3.53372283e-03
2.59191990e-01 -6.11826777e-01 -9.96590018e-01 7.70403028e-01
-1.01377892e+00 3.28009695e-01 6.66115507e-02 3.95538002e-01
4.09239739e-01 2.77256995e-01 -1.27750263e-02 -4.02117878e-01
1.77558050e-01 -2.66277399e-02 -1.12333465e+00 1.38407648e-01
-5.31481028e-01 -4.13156271e-01 3.70911390e-01 -2.80817032e-01
-8.36291134e-01 1.92888632e-01 2.44440287e-01 -5.37366390e-01
5.54647259e-02 1.84257507e-01 -3.70259583e-01 -5.45777738e-01
5.10279119e-01 8.49025846e-02 -7.37052262e-02 -4.98856343e-02
2.46803492e-01 4.68360782e-01 9.85222816e-01 9.56394672e-02
1.11533391e+00 8.70672703e-01 3.49801518e-02 -5.96509695e-01
-3.43196779e-01 -6.75920784e-01 -9.18114424e-01 -5.31949639e-01
8.07923853e-01 -1.31589282e+00 -7.46120930e-01 1.10193300e+00
-1.43450248e+00 -2.30350718e-01 3.53629887e-02 7.39601851e-01
-5.40532768e-01 8.00948799e-01 -4.75981474e-01 -5.08044511e-02
-1.71756178e-01 -1.02005196e+00 8.83993030e-01 3.55760366e-01
2.22819224e-01 -1.08497953e+00 7.81756788e-02 6.54736441e-03
2.02713400e-01 4.96040463e-01 3.75532508e-01 8.39262232e-02
-1.27441227e+00 -1.13179527e-01 -5.20553112e-01 2.71782041e-01
6.16264164e-01 2.15573624e-01 -7.43308008e-01 -5.05237579e-01
6.62691072e-02 4.78364736e-01 4.62214291e-01 4.89637375e-01
8.70347619e-01 -2.50917435e-01 -4.85060334e-01 1.17804790e+00
1.85038853e+00 4.17885929e-01 8.90458226e-01 7.28158951e-01
1.25308168e+00 6.54154196e-02 5.90416670e-01 2.21670896e-01
6.17754102e-01 9.89442289e-01 7.85994232e-01 -4.32336479e-01
-4.56720203e-01 -1.97013006e-01 2.78099954e-01 8.10974300e-01
3.82384881e-02 -4.74234909e-01 -6.98316693e-01 5.22052765e-01
-1.85449898e+00 -9.93592620e-01 -5.18783510e-01 2.23050666e+00
4.62313652e-01 1.63270220e-01 -1.46276727e-01 -1.21591352e-01
1.01554108e+00 3.06376666e-01 -7.28444576e-01 8.26505125e-02
-2.58844733e-01 -5.86319447e-01 7.62814760e-01 5.31837761e-01
-9.41969097e-01 9.16478097e-01 6.10606861e+00 2.03853756e-01
-1.27577794e+00 -7.17532709e-02 7.75475428e-02 -2.74145533e-03
-1.14165731e-01 4.62894738e-02 -7.15440214e-01 4.61663634e-01
7.09257275e-02 -2.50719160e-01 4.23084438e-01 6.98858976e-01
4.34565097e-01 -2.33760223e-01 -1.06610608e+00 1.43931925e+00
2.89702296e-01 -1.39434671e+00 3.87271009e-02 -2.19953567e-01
1.08182108e+00 2.94960886e-01 -4.06372726e-01 -2.86607116e-01
2.81190246e-01 -3.74733061e-01 5.43980241e-01 4.49958682e-01
6.27866447e-01 -4.90484029e-01 5.86263061e-01 1.18256472e-01
-1.34183550e+00 1.56207249e-01 -6.27551913e-01 -3.70614618e-01
2.93691039e-01 5.36933780e-01 -8.72789204e-01 6.88397825e-01
7.29433835e-01 1.26115489e+00 -6.80670977e-01 1.31275499e+00
-7.29033053e-02 4.33093384e-02 -3.68920952e-01 4.91379321e-01
5.47748618e-02 -4.36781704e-01 7.99875677e-01 8.61179292e-01
2.92682737e-01 2.15295106e-02 1.76006302e-01 8.46219122e-01
-1.80048838e-01 -3.73014659e-01 -9.45217550e-01 5.35674870e-01
5.44952214e-01 1.19782519e+00 -7.91171312e-01 -2.40095735e-01
-5.55936992e-01 1.32044792e+00 2.31888518e-02 4.82639909e-01
-9.00218844e-01 -4.92052078e-01 9.62834001e-01 2.18171954e-01
2.16702059e-01 -6.23436749e-01 1.37422280e-02 -1.43789268e+00
2.17429116e-01 -7.41523206e-01 1.71009496e-01 -1.16930640e+00
-9.40325558e-01 4.59147602e-01 -4.66323122e-02 -1.66423380e+00
2.25361988e-01 -4.26114321e-01 -6.95309401e-01 5.11785269e-01
-1.72105527e+00 -8.86611998e-01 -1.05569053e+00 9.81970966e-01
4.56274509e-01 1.34554893e-01 7.72526562e-02 4.88904983e-01
-6.32096052e-01 4.73501146e-01 5.26886955e-02 4.50609088e-01
7.52884150e-01 -8.46295059e-01 9.14605796e-01 1.44952917e+00
1.88498106e-02 5.01831651e-01 4.36629236e-01 -5.84697545e-01
-1.55423057e+00 -1.34102416e+00 5.32061279e-01 -4.39209998e-01
3.90526444e-01 -3.10537010e-01 -7.88194358e-01 8.84622991e-01
2.78116345e-01 3.13911915e-01 -1.15773410e-01 -6.14425838e-01
-2.05468625e-01 -4.73678172e-01 -1.00648773e+00 6.70141876e-01
1.60325778e+00 -5.37342250e-01 -3.39270622e-01 3.22759211e-01
1.03588235e+00 -9.29795206e-01 -6.48629606e-01 4.71686214e-01
3.39541703e-01 -1.08724689e+00 9.74817693e-01 -2.73743331e-01
1.22057505e-01 -1.04840004e+00 -2.71626431e-02 -1.21495116e+00
-4.72677886e-01 -7.47862875e-01 2.67423242e-01 1.07170439e+00
-3.29445362e-01 -8.30087900e-01 4.81933624e-01 4.77772027e-01
-2.31275827e-01 -1.89184621e-01 -7.35735416e-01 -8.93976867e-01
-6.91146791e-01 -3.99433315e-01 9.28051710e-01 1.21295106e+00
-3.70136231e-01 2.43664622e-01 -3.89820039e-01 7.01707900e-01
5.09720981e-01 1.89441279e-01 1.28560388e+00 -8.93923700e-01
-1.19655728e-02 -3.41275930e-01 -1.08823597e+00 -1.42400694e+00
-3.34653586e-01 -5.63798010e-01 7.41384998e-02 -1.66646528e+00
9.94589701e-02 -4.13349688e-01 -4.00175862e-02 2.53933556e-02
-1.91677883e-01 4.16478366e-01 2.31954649e-01 2.70146579e-01
-6.70606911e-01 4.42713857e-01 1.50638866e+00 -1.10661142e-01
-2.93896914e-01 -1.25505179e-01 -2.11551443e-01 8.02659452e-01
6.32554233e-01 -3.19847465e-01 -5.24179697e-01 -8.43671501e-01
4.74945873e-01 -2.10218877e-02 5.97123444e-01 -1.25426304e+00
6.21620655e-01 -2.52518773e-01 2.16275215e-01 -8.18418920e-01
2.25593336e-02 -1.20870912e+00 5.35064220e-01 5.36073923e-01
1.20307572e-01 5.34950614e-01 2.01164246e-01 5.90696514e-01
-3.11630875e-01 8.96210670e-02 8.48168015e-01 -4.16763052e-02
-1.11283910e+00 7.38841355e-01 -1.69869006e-01 -1.99515566e-01
1.13430953e+00 -6.63302481e-01 -6.27245545e-01 -4.80192691e-01
-1.29709184e-01 2.49397740e-01 9.20957029e-01 6.48625851e-01
8.26885641e-01 -1.67696571e+00 -3.06984752e-01 4.79473948e-01
2.82116234e-01 4.15901929e-01 2.75279641e-01 8.39982808e-01
-1.12940264e+00 1.10668018e-01 -3.16089213e-01 -9.22658980e-01
-1.48945951e+00 5.76916277e-01 8.95800233e-01 2.71756470e-01
-7.53850043e-01 6.16056859e-01 3.75607669e-01 -5.09496272e-01
1.36999398e-01 -8.61311078e-01 1.28602535e-01 -4.44573075e-01
4.18513000e-01 4.25499171e-01 3.84754539e-02 -5.85902452e-01
-4.06255215e-01 9.39712107e-01 4.54201214e-02 2.26060003e-01
9.58046257e-01 -7.54661858e-01 -1.99377820e-01 1.13972738e-01
1.25475073e+00 -1.60623178e-01 -1.53073072e+00 -5.83928406e-01
-4.88005787e-01 -1.00493145e+00 1.03920430e-01 -2.45917439e-01
-1.41254723e+00 5.58427930e-01 7.36623526e-01 1.84418838e-02
1.28981924e+00 -6.04132831e-01 8.86697352e-01 5.36379874e-01
4.72756296e-01 -1.01028085e+00 1.46893695e-01 4.83629555e-01
6.96606874e-01 -1.45194769e+00 2.69259751e-01 -5.74733794e-01
-1.55217901e-01 1.33532727e+00 1.04008603e+00 -1.68276861e-01
5.81505418e-01 -1.43443216e-02 2.77411997e-01 -3.77990305e-01
-3.43457997e-01 -1.28411055e-01 2.28961781e-01 5.79689622e-01
-1.65039852e-01 -3.21011066e-01 8.30413923e-02 -5.14576614e-01
8.51039601e-07 -5.77501394e-02 7.81958699e-01 8.08357537e-01
-3.89572829e-01 -6.39344811e-01 -3.80154282e-01 -1.74456894e-01
1.64012522e-01 3.71910743e-02 -2.82464683e-01 1.01342988e+00
8.05548280e-02 9.76871252e-01 3.93741369e-01 -5.13505578e-01
5.54441512e-01 -7.83356667e-01 6.07169628e-01 -3.79838943e-01
-2.45874599e-01 -3.11215222e-01 -2.26230308e-01 -9.46425140e-01
-8.36129725e-01 -5.10976374e-01 -1.16729259e+00 -4.86463279e-01
-3.80186200e-01 -4.85709041e-01 5.01729131e-01 6.41467690e-01
4.78488624e-01 4.66729730e-01 1.03188968e+00 -7.05164611e-01
3.98141108e-02 -6.24259651e-01 -5.12149572e-01 5.67567706e-01
5.94238698e-01 -4.23653215e-01 -3.97635639e-01 3.55299830e-01] | [8.715157508850098, -1.8501040935516357] |
f4c0cf70-be89-468c-a27c-b49e4726d6b6 | towards-adaptable-and-interactive-image | 2306.03500 | null | https://arxiv.org/abs/2306.03500v1 | https://arxiv.org/pdf/2306.03500v1.pdf | Towards Adaptable and Interactive Image Captioning with Data Augmentation and Episodic Memory | Interactive machine learning (IML) is a beneficial learning paradigm in cases of limited data availability, as human feedback is incrementally integrated into the training process. In this paper, we present an IML pipeline for image captioning which allows us to incrementally adapt a pre-trained image captioning model to a new data distribution based on user input. In order to incorporate user input into the model, we explore the use of a combination of simple data augmentation methods to obtain larger data batches for each newly annotated data instance and implement continual learning methods to prevent catastrophic forgetting from repeated updates. For our experiments, we split a domain-specific image captioning dataset, namely VizWiz, into non-overlapping parts to simulate an incremental input flow for continually adapting the model to new data. We find that, while data augmentation worsens results, even when relatively small amounts of data are available, episodic memory is an effective strategy to retain knowledge from previously seen clusters. | ['Daniel Sonntag', 'Mareike Hartmann', 'Aliki Anagnostopoulou'] | 2023-06-06 | null | null | null | null | ['image-captioning'] | ['computer-vision'] | [ 5.48183799e-01 5.30275047e-01 -1.22707233e-01 -4.44048554e-01
-4.55099702e-01 -5.98885596e-01 5.91834843e-01 3.94677132e-01
-5.99460006e-01 7.29301810e-01 2.58884251e-01 -3.94523889e-01
4.37359273e-01 -5.40479183e-01 -9.97684538e-01 -2.93415546e-01
-4.94184606e-02 8.47803771e-01 4.31735188e-01 -1.52378445e-02
1.73102140e-01 4.18643266e-01 -1.86441112e+00 6.09933674e-01
8.75667453e-01 3.60580206e-01 7.32451260e-01 8.66170168e-01
-4.16986644e-01 8.88428330e-01 -5.00896752e-01 -1.08477794e-01
3.98483910e-02 -6.46727383e-01 -1.11759746e+00 4.81135547e-01
3.60205531e-01 -3.54526013e-01 -2.75439441e-01 4.20717806e-01
3.24841499e-01 2.88655162e-01 3.24142069e-01 -1.28101861e+00
-8.61172557e-01 4.09039646e-01 -2.97001153e-01 2.53181279e-01
1.82338834e-01 4.95134205e-01 3.88806045e-01 -8.31093371e-01
9.40564573e-01 8.75655651e-01 5.61173856e-01 8.58547330e-01
-1.42685521e+00 -3.25107068e-01 5.42300344e-01 3.61250907e-01
-9.89093959e-01 -4.70990330e-01 6.10471964e-01 -3.93009961e-01
1.00080836e+00 1.95015892e-01 6.58484876e-01 1.04558241e+00
-2.36962020e-01 9.39917207e-01 8.58426392e-01 -7.21464038e-01
3.74640763e-01 6.32689178e-01 -3.03807016e-02 3.82087320e-01
-1.05624191e-01 -1.18002854e-01 -5.07762194e-01 -1.13078244e-01
7.08960235e-01 -6.78951889e-02 -1.08214140e-01 -4.00486887e-01
-1.14099002e+00 5.98272145e-01 4.96427983e-01 2.08377570e-01
-5.35756528e-01 -2.30024680e-02 3.65350634e-01 4.45851862e-01
6.03027582e-01 7.08371103e-01 -6.26610398e-01 -8.27126354e-02
-9.55016732e-01 5.83647415e-02 6.51422441e-01 7.62504518e-01
9.57590759e-01 -1.83243334e-01 -4.58313376e-01 8.39720368e-01
-1.59927309e-02 2.38117069e-01 7.43735552e-01 -9.50747728e-01
2.03804329e-01 7.53574252e-01 2.98369735e-01 -5.00425637e-01
-1.49425328e-01 -1.96781114e-01 -5.37196279e-01 1.12502106e-01
3.19250345e-01 6.67173192e-02 -1.56458688e+00 1.70376539e+00
2.67779469e-01 3.90915990e-01 5.71286157e-02 5.52759051e-01
6.10777676e-01 8.42270613e-01 4.94388849e-01 -3.36232901e-01
9.89625394e-01 -1.02452099e+00 -7.39156485e-01 -5.15649259e-01
6.67162955e-01 -4.65159714e-01 1.49819696e+00 1.44789919e-01
-9.47949171e-01 -6.38370931e-01 -8.22455525e-01 8.86371210e-02
-4.75227326e-01 -2.00477213e-01 5.06529927e-01 2.81561196e-01
-1.21246195e+00 4.44934040e-01 -1.02549779e+00 -6.20471895e-01
5.18890679e-01 2.36622408e-01 -3.58011484e-01 -3.75932097e-01
-1.01886523e+00 9.32350099e-01 6.23950481e-01 -1.23341069e-01
-8.85443628e-01 -6.88553870e-01 -8.32946002e-01 -1.27365053e-01
4.40496594e-01 -5.99700689e-01 1.48002791e+00 -1.45804083e+00
-1.18490541e+00 8.61249626e-01 -4.18910801e-01 -5.37130892e-01
3.36761236e-01 -1.50388598e-01 -1.47122994e-01 5.96338511e-02
-2.30148241e-01 1.36219740e+00 8.96562219e-01 -1.54668725e+00
-3.52946997e-01 -1.16021708e-01 -4.40672450e-02 3.82929802e-01
-5.71062684e-01 -2.14289844e-01 -6.94753885e-01 -4.59752470e-01
-1.26060396e-01 -1.05379903e+00 -4.48270828e-01 -8.64462256e-02
-6.73492253e-02 7.71056265e-02 9.92896557e-01 -6.76107943e-01
1.23774993e+00 -2.04341698e+00 1.13475047e-01 7.24564120e-02
-1.11689925e-01 5.00228107e-01 -4.06028897e-01 4.16536033e-01
-9.27838981e-02 1.87620685e-01 -4.71237093e-01 -6.06004775e-01
-4.69616860e-01 6.33680582e-01 -3.00860971e-01 -1.29350483e-01
5.47455788e-01 1.00259244e+00 -9.95793641e-01 -2.86363304e-01
1.25093341e-01 4.22949195e-01 -6.52264893e-01 5.82176805e-01
-9.17368233e-01 6.79382324e-01 1.73733681e-01 2.76362985e-01
5.74699283e-01 -6.35679126e-01 5.48231341e-02 1.69552773e-01
5.95321506e-02 5.65072484e-02 -9.47375536e-01 1.80866063e+00
-4.65417236e-01 7.60891616e-01 -2.48732328e-01 -6.69069767e-01
6.51874602e-01 3.51362824e-01 1.70122519e-01 -8.65827978e-01
-3.08861524e-01 -1.51787013e-01 -1.16552457e-01 -6.91407800e-01
6.49925709e-01 1.12613484e-01 2.19169930e-01 8.53314340e-01
1.65031910e-01 1.98528871e-01 3.12390596e-01 5.08727431e-01
1.08078897e+00 1.31750584e-01 8.36948976e-02 2.56682426e-01
1.66235030e-01 5.02620339e-01 3.29408348e-01 9.92542446e-01
-8.00894946e-02 8.83402348e-01 1.56859547e-01 -6.35346949e-01
-1.43502963e+00 -8.83965611e-01 1.07486434e-01 1.41722369e+00
-1.22898325e-01 -2.23366722e-01 -7.07721651e-01 -8.83152843e-01
-2.30646074e-01 9.97845590e-01 -7.51228690e-01 -4.11232799e-01
-6.24157250e-01 -8.43821526e-01 1.59412593e-01 5.52548289e-01
4.80966926e-01 -1.69062912e+00 -7.60002136e-01 2.79819310e-01
-2.59512275e-01 -7.17688382e-01 -5.03026664e-01 2.37329334e-01
-8.93012166e-01 -8.35592151e-01 -6.62776291e-01 -8.47648919e-01
1.13763416e+00 1.30235374e-01 1.41115141e+00 3.88236314e-01
-3.25139940e-01 6.26305878e-01 -5.61067402e-01 -4.53663915e-01
-8.97882938e-01 2.59220749e-01 -1.78429887e-01 -1.74061403e-01
2.74214000e-01 -2.89958179e-01 -6.29295290e-01 2.45556802e-01
-1.27894866e+00 5.59803188e-01 5.55507839e-01 8.44583213e-01
6.22990072e-01 -2.71688372e-01 7.79034853e-01 -1.30533218e+00
6.34440899e-01 -6.60034776e-01 -2.79786050e-01 4.02785331e-01
-6.83421969e-01 3.05621803e-01 4.27768201e-01 -8.96242559e-01
-1.27335191e+00 4.69329357e-01 -5.72884306e-02 -4.51878697e-01
-2.77813464e-01 5.69767892e-01 6.03369288e-02 2.60510862e-01
8.67431700e-01 3.84994596e-01 1.92680016e-01 -4.44846690e-01
8.11412632e-01 7.11996734e-01 6.85354531e-01 -2.94313937e-01
5.45518994e-01 3.05525243e-01 -8.42426479e-01 -4.28312927e-01
-7.15273440e-01 -1.87780201e-01 -9.30740058e-01 -3.76644909e-01
6.48791850e-01 -8.06948781e-01 -2.40486532e-01 3.93725961e-01
-1.09499717e+00 -8.59864354e-01 -6.48797870e-01 1.12923346e-01
-4.15800542e-01 -4.67721969e-02 -4.10909414e-01 -6.64200068e-01
-2.15428442e-01 -9.65395212e-01 7.22606003e-01 3.49944055e-01
-4.26589578e-01 -1.02061212e+00 2.93897659e-01 1.65824816e-01
5.73632598e-01 8.49974155e-02 7.97549248e-01 -9.36248183e-01
-5.77550113e-01 -1.82129711e-01 -1.01121254e-01 3.88833821e-01
1.58521727e-01 -1.96776137e-01 -9.94296014e-01 -3.95101219e-01
-1.84714392e-01 -6.32905662e-01 7.94822216e-01 -1.21921793e-01
1.26767039e+00 -5.26669145e-01 -3.74859422e-01 2.00130939e-01
1.17051888e+00 3.13517570e-01 8.27219963e-01 5.56002080e-01
5.49669385e-01 5.34202933e-01 5.77282906e-01 2.87413090e-01
6.05633616e-01 2.92881548e-01 4.07124639e-01 -2.51375824e-01
-2.34995693e-01 -4.53515947e-01 1.43514439e-01 4.52018887e-01
3.54088843e-01 -3.33110988e-01 -1.05498993e+00 7.76226640e-01
-2.00516081e+00 -8.39249015e-01 2.31871068e-01 2.42813921e+00
1.14140260e+00 1.20987356e-01 1.02600642e-02 -3.10882628e-01
6.37551367e-01 -2.50492334e-01 -1.01531446e+00 -2.25377306e-01
1.01760097e-01 -9.84567590e-03 4.30175662e-01 4.49126184e-01
-8.78624678e-01 1.10833108e+00 7.09784365e+00 2.69404143e-01
-1.27276003e+00 1.70186996e-01 8.54481339e-01 -2.54921407e-01
-3.86271268e-01 5.25542460e-02 -4.57699329e-01 6.74590349e-01
1.31782722e+00 -7.41953477e-02 4.27959710e-01 6.39889479e-01
1.53551534e-01 -4.24967229e-01 -1.17495024e+00 6.86449289e-01
2.20833465e-01 -1.48553038e+00 2.11608768e-01 -1.56310782e-01
6.03388250e-01 1.76098779e-01 4.20216732e-02 4.12883729e-01
4.53920543e-01 -9.07988429e-01 3.57847363e-01 6.76978052e-01
5.56936681e-01 -6.05661690e-01 5.96692443e-01 5.95432580e-01
-4.85329360e-01 -3.65832411e-02 -1.49064183e-01 -1.00433424e-01
2.62856424e-01 3.82378548e-01 -1.46930397e+00 1.57802831e-02
7.46297121e-01 3.57573003e-01 -1.10530388e+00 1.18843853e+00
-2.45300308e-02 7.38752604e-01 -3.50585103e-01 4.21936512e-01
-5.03023565e-02 2.51509130e-01 2.73663104e-01 1.17834949e+00
1.48533940e-01 1.14540935e-01 1.61267191e-01 6.82000279e-01
-1.64582416e-01 -5.01840040e-02 -6.89105809e-01 -1.62537582e-02
5.78086555e-01 1.03685093e+00 -8.38945508e-01 -6.58596337e-01
-3.27102959e-01 1.44328535e+00 6.24607742e-01 4.84057039e-01
-5.33360004e-01 -5.31756096e-02 1.93670794e-01 3.84460479e-01
3.45972568e-01 -1.99907124e-01 -2.11187214e-01 -1.02930999e+00
-5.92357554e-02 -7.28558600e-01 5.80749571e-01 -1.15214050e+00
-9.59826827e-01 7.22573638e-01 -2.48987577e-03 -9.13039327e-01
-5.18597305e-01 -1.18739798e-03 -6.67955697e-01 6.98637247e-01
-1.33832836e+00 -1.04745722e+00 -4.25867289e-01 4.05250847e-01
7.97926068e-01 5.61018847e-02 8.86901081e-01 1.93432242e-01
-5.72241783e-01 5.26789963e-01 1.93218756e-02 -1.69300094e-01
9.43371654e-01 -1.25570750e+00 4.82591063e-01 8.52856874e-01
2.33071908e-01 5.63356698e-01 9.77153301e-01 -9.83855486e-01
-8.73483062e-01 -1.33843803e+00 7.35459864e-01 -7.14489281e-01
2.77081102e-01 -6.34708762e-01 -1.66353166e+00 1.02728069e+00
5.80686629e-01 2.98952367e-02 8.60639036e-01 3.53792161e-02
-1.97347403e-01 1.79417387e-01 -1.18898129e+00 6.16999388e-01
8.68432701e-01 -3.94605935e-01 -5.88330030e-01 4.54525232e-01
9.93571818e-01 -3.54399949e-01 -5.03805041e-01 3.25813591e-01
8.06848556e-02 -5.10070205e-01 6.92794204e-01 -7.27268875e-01
1.30740464e-01 -4.63529766e-01 3.22370827e-01 -1.47190261e+00
-3.11114997e-01 -5.07002115e-01 -3.50383997e-01 1.26823890e+00
6.24276280e-01 -3.36317569e-01 9.48011756e-01 1.10861313e+00
-6.63525462e-02 -4.51261282e-01 -6.06856048e-01 -4.85164106e-01
-1.51469305e-01 -3.98871630e-01 4.02769923e-01 9.60580528e-01
1.41516611e-01 3.57642740e-01 -5.16585886e-01 6.17543459e-02
2.41333678e-01 -1.06279008e-01 7.21792996e-01 -1.01604140e+00
-2.53465176e-01 1.01933353e-01 -9.10519622e-03 -9.55817044e-01
-9.93463919e-02 -8.25843751e-01 2.13420406e-01 -1.66007578e+00
4.24542218e-01 -5.63029528e-01 -3.01731020e-01 8.57051194e-01
-4.61535722e-01 2.79397219e-01 2.68871397e-01 4.81544405e-01
-9.36483920e-01 4.30608958e-01 1.11315048e+00 -9.63471271e-03
-7.60551691e-01 -7.76363611e-02 -7.06708789e-01 3.82695526e-01
8.13350558e-01 -4.61514950e-01 -7.96704531e-01 -5.44154763e-01
1.72486395e-01 -3.13725293e-01 1.76212430e-01 -9.35282826e-01
5.14375150e-01 -9.92976278e-02 5.01202345e-01 -5.00537932e-01
5.84038869e-02 -7.39092171e-01 3.05927426e-01 2.47573659e-01
-6.48929775e-01 8.19244385e-02 6.78522229e-01 5.94347894e-01
-1.11223258e-01 -2.38976881e-01 7.27449656e-01 -2.83322245e-01
-1.01770461e+00 2.02838987e-01 -6.63226426e-01 -1.29417181e-01
1.16128635e+00 -2.24233955e-01 -2.41614908e-01 -4.75993186e-01
-1.21756887e+00 4.69908088e-01 7.40259409e-01 6.73476577e-01
5.98471284e-01 -1.27606559e+00 -5.68657637e-01 3.65643471e-01
3.76370132e-01 1.52748123e-01 3.18978935e-01 3.27586353e-01
-3.93627703e-01 7.14024231e-02 -2.14051232e-01 -5.70731103e-01
-1.19665861e+00 9.01055217e-01 1.26531839e-01 -1.67192668e-01
-7.16882646e-01 5.96756279e-01 -7.51481950e-02 -5.06640673e-01
4.68465179e-01 -8.44051540e-02 -2.07602262e-01 4.16374169e-02
8.24678361e-01 -1.51364222e-01 1.61394060e-01 -2.52164900e-01
-2.82203127e-03 -2.59051532e-01 -6.82720482e-01 -1.96330175e-01
1.41784883e+00 -4.18268234e-01 8.59741345e-02 6.60420954e-01
9.00618136e-01 -6.55165315e-01 -1.73763943e+00 -4.69618022e-01
1.30325615e-01 -3.27036738e-01 -2.26255134e-01 -1.21381593e+00
-7.46673107e-01 5.26855767e-01 8.94528687e-01 1.26575798e-01
1.24995804e+00 3.45676959e-01 6.08330071e-01 5.00643432e-01
7.03339726e-02 -1.17223918e+00 3.86670738e-01 4.82982367e-01
8.23715925e-01 -1.52780938e+00 -2.86032200e-01 -6.10194653e-02
-9.01624441e-01 8.40407193e-01 1.02734208e+00 2.77992278e-01
4.64769542e-01 9.86361429e-02 1.98659346e-01 -4.99575660e-02
-1.09402478e+00 -1.85273096e-01 -4.21322472e-02 7.14538097e-01
2.52747267e-01 -4.49291050e-01 1.33835450e-01 1.21055432e-01
1.61866352e-01 2.80498773e-01 7.11464465e-01 1.20529091e+00
-5.11036515e-01 -1.21391594e+00 -2.97894537e-01 5.38301408e-01
7.76639581e-02 -1.65461987e-01 -3.85493040e-01 6.07013524e-01
7.05617368e-02 6.40265405e-01 6.17902100e-01 -1.32803470e-01
3.47630978e-01 4.01123136e-01 3.07368875e-01 -8.83534372e-01
-4.24927294e-01 -3.18979561e-01 -1.85020119e-01 -3.16510588e-01
-3.64509940e-01 -7.31417418e-01 -1.33313346e+00 5.12510464e-02
1.41450502e-02 7.06681311e-02 6.64848268e-01 8.56536090e-01
8.17469895e-01 5.71334481e-01 3.32712501e-01 -8.88823211e-01
1.22322641e-01 -1.03397465e+00 5.96784949e-02 7.59250462e-01
3.78331512e-01 -2.30625570e-01 -2.41862252e-01 7.00629115e-01] | [10.28110408782959, 1.8547966480255127] |
f184ae9f-9990-4e8c-a2ed-a1ac7d1e762f | an-aiot-enabled-autonomous-dementia | 2207.00804 | null | https://arxiv.org/abs/2207.00804v1 | https://arxiv.org/pdf/2207.00804v1.pdf | An AIoT-enabled Autonomous Dementia Monitoring System | An autonomous Artificial Internet of Things (AIoT) system for elderly dementia patients monitoring in a smart home is presented. The system mainly implements two functions based on the activity inference of the sensor data, which are real time abnormal activity monitoring and trend prediction of disease related activities. Specifically, CASAS dataset is employed to train a Random Forest (RF) model for activity inference. Then, another RF model trained by the output data of activity inference is used for abnormal activity monitoring. Particularly, RF is chosen for these tasks because of its balanced trade offs between accuracy, time efficiency, flexibility, and interpretability. Moreover, Long Short Term Memory (LSTM) is utilised to forecast the disease related activity trend of a patient. Consequently, the accuracy of two RF classifiers designed for activity inference and abnormal activity detection is greater than 99 percent and 94 percent, respectively. Furthermore, using the duration of sleep as an example, the LSTM model achieves accurate and evident future trends prediction. | ['Jinyang Li', 'Xingyu Wu'] | 2022-07-02 | null | null | null | null | ['activity-detection'] | ['computer-vision'] | [ 3.00040513e-01 2.56178260e-01 -4.65212852e-01 -5.83249867e-01
-5.75619899e-02 3.74035299e-01 5.74107766e-01 1.52930841e-01
-4.60933834e-01 1.16204417e+00 6.09049380e-01 -3.72986972e-01
-3.37130010e-01 -1.07927465e+00 2.11432166e-02 -8.84166718e-01
-3.04612011e-01 2.48345181e-01 2.03705132e-02 3.50762337e-01
-1.88004181e-01 3.03819627e-01 -1.58315825e+00 3.89921039e-01
1.00463021e+00 1.46665084e+00 2.37748206e-01 3.83506536e-01
5.44148730e-03 1.29837811e+00 -6.89794421e-01 9.41650718e-02
-4.00255799e-01 -1.87971696e-01 -6.65069878e-01 -7.64335096e-02
-5.14302373e-01 -6.18378341e-01 -2.23493904e-01 4.87067282e-01
3.85818183e-01 1.22990958e-01 5.14226973e-01 -1.15064967e+00
-1.28371507e-01 4.49669510e-01 1.02556460e-01 4.90977138e-01
3.11768353e-01 -2.18957718e-02 5.47089219e-01 -1.00746594e-01
-1.56273898e-02 5.89650393e-01 5.10285616e-01 4.67888236e-01
-6.36854291e-01 -4.99249607e-01 2.56884009e-01 5.52288771e-01
-1.04486752e+00 -4.95584995e-01 3.50983143e-01 -3.83616865e-01
1.35826123e+00 2.12076783e-01 1.44015658e+00 1.51726973e+00
8.73719692e-01 7.26992071e-01 7.98372388e-01 -2.90780246e-01
7.46163189e-01 -7.22529069e-02 2.74390817e-01 4.64743555e-01
4.33364123e-01 -4.88408543e-02 -4.54610616e-01 -2.80153990e-01
4.25290644e-01 7.70752728e-01 -1.09169729e-01 1.09244414e-01
-1.33842683e+00 2.96924025e-01 4.19835627e-01 5.73732734e-01
-8.22290719e-01 -4.22901325e-02 2.91032702e-01 9.96892676e-02
4.41485375e-01 -2.97077000e-02 -7.16767788e-01 -3.49844068e-01
-8.28388453e-01 -1.26143202e-01 7.31530011e-01 5.52016616e-01
2.35209614e-02 8.49177316e-02 -5.77493370e-01 6.15036190e-01
6.94237173e-01 4.49431211e-01 1.04195869e+00 -7.20891654e-01
2.47382760e-01 1.16454422e+00 2.25782335e-01 -4.58689094e-01
-9.06636894e-01 -8.39861035e-01 -1.13728189e+00 -1.37510002e-01
1.77013967e-02 -3.43141556e-01 -7.64378369e-01 1.41061604e+00
8.07578489e-02 1.35119528e-01 2.47220293e-01 1.35543361e-01
4.61742640e-01 3.48829538e-01 6.39299095e-01 -6.65991306e-01
1.64571428e+00 -5.77389598e-01 -1.05545259e+00 -6.86234593e-01
8.50102782e-01 3.75000596e-01 6.45118296e-01 1.70560583e-01
-8.29010248e-01 -2.65939474e-01 -1.12142324e+00 3.77700508e-01
-4.13363695e-01 2.61463583e-01 6.62514210e-01 5.98502934e-01
-8.29382360e-01 4.77949291e-01 -1.44081020e+00 -5.07614911e-01
7.65475929e-01 5.64103484e-01 -1.42848715e-02 2.28282034e-01
-1.21761191e+00 1.05168951e+00 4.44606006e-01 3.37482728e-02
-2.96180874e-01 -3.99274856e-01 -4.62500215e-01 2.06827521e-01
-4.47279632e-01 -1.11600840e+00 1.17978501e+00 -6.13162637e-01
-1.34605253e+00 5.78413725e-01 -2.45392874e-01 -8.49844396e-01
7.06793308e-01 -2.75933057e-01 -8.24845314e-01 -1.62710935e-01
5.76453507e-02 9.92023721e-02 5.78391314e-01 -2.08321840e-01
-9.23735559e-01 -9.41031694e-01 -3.80078167e-01 1.26036391e-01
-5.63622236e-01 -3.02700579e-01 2.73802966e-01 -4.86859918e-01
1.87813528e-02 -5.86158454e-01 -1.78029835e-01 -1.93632260e-01
-1.61847606e-01 -4.27458316e-01 7.31933117e-01 -9.81349766e-01
1.73489475e+00 -1.88913488e+00 -6.07193172e-01 2.14819893e-01
3.60182792e-01 -5.50938249e-02 7.20107079e-01 -6.15584739e-02
1.42036006e-02 -3.83467913e-01 -7.27965012e-02 1.92293376e-02
-3.78850967e-01 2.73243248e-01 1.22601762e-01 1.74615338e-01
-1.26728892e-01 1.11736143e+00 -7.25203454e-01 -3.53104889e-01
3.36881340e-01 4.62612569e-01 -2.04404682e-01 2.13067040e-01
-9.33041647e-02 5.64572990e-01 -6.81005657e-01 7.04326868e-01
-7.86917806e-02 -4.18154180e-01 5.21105766e-01 3.32109109e-02
-5.65337129e-02 5.07498682e-01 -5.22107124e-01 1.20935357e+00
-5.23577452e-01 1.43604159e-01 -7.10851789e-01 -1.04326010e+00
1.02607405e+00 7.12765932e-01 9.95489001e-01 -9.03049648e-01
1.97795719e-01 9.57612619e-02 -5.12119941e-03 -8.37843776e-01
-3.66269857e-01 2.04596281e-01 1.02687322e-01 4.46570635e-01
-2.95220226e-01 1.02441621e+00 -6.82234243e-02 -4.18013036e-01
1.80326498e+00 2.26233542e-01 9.50666130e-01 -5.65816909e-02
4.26177353e-01 -1.36337742e-01 5.77845573e-01 4.83652532e-01
-4.34835523e-01 -3.81439477e-01 2.13481355e-02 -8.09021592e-01
-5.33317268e-01 -9.82635021e-01 -1.08723506e-01 1.05459404e+00
-3.89185220e-01 -2.38192633e-01 -5.27354658e-01 -7.07127094e-01
-3.53562713e-01 8.75189960e-01 -5.09342134e-01 -4.68651921e-01
-5.36821127e-01 -1.09501684e+00 1.86346486e-01 9.57830608e-01
1.17429769e+00 -1.55561888e+00 -1.38163948e+00 4.76650238e-01
-4.63582784e-01 -8.15343797e-01 2.01563627e-01 5.85851490e-01
-1.45494437e+00 -1.01422489e+00 -4.32734042e-01 -7.38236547e-01
5.56830883e-01 -4.95500751e-02 7.89114296e-01 -1.68386772e-01
1.49544343e-01 1.07542098e-01 8.07604194e-03 -5.17969549e-01
-5.23106121e-02 7.69686103e-02 2.76024342e-01 4.46840981e-03
1.00775945e+00 -1.13316739e+00 -9.80078220e-01 1.83585718e-01
-3.14884603e-01 -5.28281070e-02 9.12846565e-01 5.21073818e-01
4.27799881e-01 2.09381342e-01 8.64387512e-01 -3.88997048e-01
5.54608107e-01 -7.35650361e-01 -9.87176597e-02 3.76572430e-01
-1.09751427e+00 1.73037857e-01 3.73982877e-01 -4.92100537e-01
-1.22721291e+00 3.11853886e-01 -7.50046372e-02 2.95112818e-01
-3.30494791e-01 2.63589293e-01 -5.35719872e-01 8.47867250e-01
5.00500441e-01 4.69885856e-01 -1.00484915e-01 -2.95768142e-01
-4.35488641e-01 1.20955706e+00 4.12847102e-01 1.32878125e-01
-1.02244146e-01 3.57326508e-01 -2.24722564e-01 -7.30125725e-01
-7.28236377e-01 -3.24895084e-01 -5.56767285e-01 -3.35147321e-01
9.57144380e-01 -9.57625806e-01 -9.78468835e-01 7.12325335e-01
-7.32275724e-01 -3.15045148e-01 -2.35353827e-01 6.67388737e-01
-7.13683784e-01 -3.57403934e-01 -2.38573343e-01 -1.08685338e+00
-9.64330018e-01 -6.75427675e-01 7.30395019e-01 1.12620100e-01
-8.06140721e-01 -1.02250576e+00 -7.21801072e-02 4.44462717e-01
4.73042160e-01 3.80973667e-01 1.07562149e+00 -8.08038652e-01
-2.76414514e-01 -6.53003097e-01 -6.25887653e-03 2.56759822e-02
5.33511698e-01 -6.84237897e-01 -1.01425302e+00 -6.59397244e-02
3.14130574e-01 3.58919501e-01 4.14346486e-01 7.67824590e-01
1.10472405e+00 -9.02001023e-01 -8.62877488e-01 2.60957301e-01
8.95200014e-01 7.85872817e-01 1.02782607e+00 8.02163124e-01
2.75285035e-01 2.89568424e-01 3.72300297e-01 5.94795406e-01
6.30401909e-01 6.08277380e-01 3.93859088e-01 3.62629354e-01
2.61969328e-01 -9.77552012e-02 6.77244544e-01 3.76764685e-01
-4.23176527e-01 -7.79986382e-02 -6.99239552e-01 2.49346659e-01
-2.15088224e+00 -1.06094599e+00 7.43961260e-02 2.20578527e+00
4.24801111e-01 2.70880222e-01 2.81879306e-01 8.19191754e-01
5.04345596e-01 5.59927784e-02 -8.51701021e-01 -9.47501883e-02
3.47157925e-01 -7.80922621e-02 4.56365854e-01 -3.24974775e-01
-1.11492515e+00 2.54343182e-01 5.86184883e+00 1.00686207e-01
-6.50090218e-01 3.75977874e-01 6.52970076e-01 -2.29983211e-01
3.32026303e-01 -5.75623333e-01 -5.75173259e-01 9.45295751e-01
1.38560200e+00 2.38411501e-01 2.63305336e-01 1.01757610e+00
8.40272844e-01 -1.82869643e-01 -8.14666688e-01 8.68367791e-01
-5.12020946e-01 -8.66711199e-01 -8.17716867e-02 3.16646278e-01
1.49782375e-01 -2.06646305e-02 -5.36791980e-01 2.68489838e-01
2.58017391e-01 -8.24484766e-01 4.12445992e-01 9.49828148e-01
5.19897163e-01 -6.50400400e-01 8.53335559e-01 6.03606582e-01
-1.14471471e+00 -9.19628859e-01 2.62143105e-01 -2.49726817e-01
2.70391613e-01 7.78046370e-01 -7.91875243e-01 9.44788530e-02
1.08317053e+00 8.66669178e-01 -4.38443720e-01 8.87127995e-01
-3.45219314e-01 5.77694714e-01 -2.58579224e-01 -2.10890576e-01
-1.93680406e-01 -1.04354098e-02 1.61775216e-01 7.95947313e-01
4.94780302e-01 3.66912670e-02 -5.51186576e-02 2.92639643e-01
1.43332168e-01 -1.90694764e-01 -6.46284759e-01 6.48496002e-02
4.12182987e-01 9.20592487e-01 -3.97906393e-01 -3.96247864e-01
-3.39573026e-01 7.26443231e-01 9.35086533e-02 1.64655730e-01
-6.68846071e-01 -2.12426726e-02 5.90253234e-01 5.70786119e-01
-1.70288831e-02 1.75514266e-01 -6.83541358e-01 -9.72867608e-01
1.44480377e-01 -3.73062730e-01 6.33387148e-01 -7.11077690e-01
-8.75873983e-01 4.54845130e-01 -1.94335163e-01 -1.46006060e+00
-5.81317484e-01 -3.75951350e-01 -8.61437023e-01 3.27856869e-01
-1.05602992e+00 -9.27471220e-01 -6.79504097e-01 7.42166460e-01
4.82586771e-01 -1.81673557e-01 1.12761140e+00 2.66783625e-01
-7.90432572e-01 1.57921076e-01 -3.34124677e-02 5.98965213e-02
-1.24130594e-02 -1.01322234e+00 2.11435050e-01 6.11730993e-01
-4.48807240e-01 4.18099463e-01 2.19238982e-01 -9.00056124e-01
-8.18728328e-01 -1.44040442e+00 1.18546999e+00 -4.42499906e-01
4.58609134e-01 1.77142635e-01 -5.12574852e-01 9.16005671e-01
-4.31161255e-01 -1.05554320e-01 8.63591194e-01 -2.06144378e-01
3.48359406e-01 -4.12838906e-01 -1.41289127e+00 5.25355995e-01
1.25705040e+00 -4.26401244e-03 -6.30417645e-01 1.79559603e-01
3.32938045e-01 1.50402158e-01 -1.19816220e+00 6.47042692e-01
1.09855640e+00 -8.56613457e-01 9.34308112e-01 -5.53745329e-01
7.30701685e-02 6.00892194e-02 -1.23814583e-01 -9.89749670e-01
-3.84014785e-01 -1.90370202e-01 -1.06616211e+00 7.44348407e-01
3.18539709e-01 -1.01075733e+00 7.91440070e-01 7.04635143e-01
-1.43093839e-02 -9.54325140e-01 -8.58799219e-01 -5.29276609e-01
-9.38079238e-01 -3.63506705e-01 7.69166589e-01 5.98058820e-01
3.93031299e-01 5.57821751e-01 -2.92931855e-01 -3.66964415e-02
4.66648966e-01 -3.84695023e-01 1.75416812e-01 -1.77088773e+00
-6.24301732e-02 -1.24849342e-01 -6.06459856e-01 -7.21457958e-01
-2.83368468e-01 -5.35480559e-01 -2.49653831e-01 -1.98155975e+00
3.02920435e-02 -1.41108274e-01 -6.89864635e-01 8.03467572e-01
4.96962517e-02 -5.75045049e-02 -5.50710618e-01 3.45638275e-01
-4.11164969e-01 4.11255002e-01 8.39623749e-01 -2.80758291e-01
-7.63830245e-01 1.21399009e+00 -5.13166547e-01 8.63947868e-01
1.27646422e+00 -2.28349179e-01 -4.58460271e-01 -7.13367248e-03
1.55803049e-02 2.65889347e-01 3.39010268e-01 -1.41808975e+00
3.62580046e-02 -8.80725831e-02 1.07864618e+00 -6.14943266e-01
4.12264764e-01 -9.28527057e-01 2.83851534e-01 1.11401868e+00
-1.89554766e-01 -1.27252052e-02 -3.53451818e-01 7.40576088e-01
3.26803565e-01 2.07354799e-01 5.95705748e-01 -2.32602239e-01
-4.35168445e-01 2.85002559e-01 -9.62942004e-01 -5.38627625e-01
1.12670457e+00 -6.05463743e-01 -3.29563677e-01 -2.40000546e-01
-1.03765750e+00 6.10152185e-02 -2.17784762e-01 4.39317286e-01
5.00796497e-01 -1.24955356e+00 -1.38347134e-01 2.91381329e-01
3.53094071e-01 -7.80713037e-02 1.14707246e-01 1.19416237e+00
-1.57546446e-01 6.20636940e-01 -3.35130155e-01 -4.11189944e-01
-1.08076000e+00 3.43956292e-01 5.75494170e-01 -6.61650062e-01
-9.70172167e-01 -5.96361756e-02 -4.45482016e-01 1.39884248e-01
5.45414627e-01 -5.74591160e-01 -7.45978355e-01 8.81314799e-02
9.22759533e-01 8.54769826e-01 4.23093081e-01 -5.90306595e-02
-6.37329698e-01 -4.48223054e-02 1.26878977e-01 2.45305896e-01
1.58354688e+00 -1.45440370e-01 9.62833613e-02 4.30467337e-01
5.02062500e-01 -7.11867750e-01 -1.08128905e+00 -3.99892032e-02
5.57743371e-01 3.23960066e-01 1.72103807e-01 -1.07997370e+00
-7.36390948e-01 4.55441445e-01 1.16450787e+00 2.97366709e-01
1.31661582e+00 -1.35809168e-01 8.57246220e-01 6.88969076e-01
3.75228167e-01 -9.38517332e-01 -1.12658054e-01 3.96650136e-02
3.78980607e-01 -8.24648321e-01 -1.48413062e-01 3.48619938e-01
-4.48094696e-01 9.11931098e-01 5.42085588e-01 2.40738660e-01
8.16553712e-01 1.47317767e-01 -1.80545345e-01 -7.73918480e-02
-7.52466679e-01 -1.03384748e-01 -8.39227289e-02 8.99424970e-01
9.42233503e-02 3.07773262e-01 -1.05393991e-01 8.22640657e-01
-3.01654816e-01 6.95468068e-01 9.93731525e-03 1.00886858e+00
-7.58927464e-01 -4.75157768e-01 -8.38630721e-02 1.24149764e+00
-5.73251665e-01 1.92189813e-01 1.59651134e-02 1.59462929e-01
1.13137297e-01 1.17865825e+00 1.96315467e-01 -4.34119910e-01
3.45004290e-01 5.28451622e-01 8.72556418e-02 -2.60713279e-01
-4.30584818e-01 -4.71655935e-01 1.40323371e-01 -6.29554927e-01
-5.66940665e-01 -4.33415711e-01 -1.30826247e+00 -4.45547402e-02
3.68600078e-02 -1.67803168e-01 5.11161029e-01 1.61696637e+00
4.53867078e-01 9.88093615e-01 5.81308961e-01 -3.56121302e-01
-1.97790265e-01 -1.47795856e+00 -5.30914307e-01 -2.30080202e-01
2.27065846e-01 -6.94062531e-01 -1.42053589e-01 7.04055279e-02] | [7.415951251983643, 0.8565438985824585] |
75549586-3a0e-4833-ad0a-b592b7388136 | accurate-spectral-super-resolution-from | 1806.03575 | null | http://arxiv.org/abs/1806.03575v3 | http://arxiv.org/pdf/1806.03575v3.pdf | Accurate Spectral Super-resolution from Single RGB Image Using Multi-scale CNN | Different from traditional hyperspectral super-resolution approaches that
focus on improving the spatial resolution, spectral super-resolution aims at
producing a high-resolution hyperspectral image from the RGB observation with
super-resolution in spectral domain. However, it is challenging to accurately
reconstruct a high-dimensional continuous spectrum from three discrete
intensity values at each pixel, since too much information is lost during the
procedure where the latent hyperspectral image is downsampled (e.g., with x10
scaling factor) in spectral domain to produce an RGB observation. To address
this problem, we present a multi-scale deep convolutional neural network (CNN)
to explicitly map the input RGB image into a hyperspectral image. Through
symmetrically downsampling and upsampling the intermediate feature maps in a
cascading paradigm, the local and non-local image information can be jointly
encoded for spectral representation, ultimately improving the spectral
reconstruction accuracy. Extensive experiments on a large hyperspectral dataset
demonstrate the effectiveness of the proposed method. | ['Yanning Zhang', 'Jun Li', 'Wei Wei', 'Lei Zhang', 'Yiqi Yan'] | 2018-06-10 | null | null | null | null | ['spectral-reconstruction', 'spectral-super-resolution'] | ['computer-vision', 'computer-vision'] | [ 1.21513426e+00 -3.55399609e-01 1.07947126e-01 -2.53649235e-01
-8.51076841e-01 -5.69225371e-01 1.51784703e-01 -3.67883772e-01
-8.76481924e-03 9.41755354e-01 1.24910586e-01 -1.18531249e-01
-4.74546909e-01 -1.24117994e+00 -6.69334412e-01 -1.09454060e+00
3.43264222e-01 -1.07577205e-01 -4.24942553e-01 -1.67359054e-01
-1.06330164e-01 6.90860510e-01 -1.69815743e+00 3.61542106e-01
1.26710975e+00 9.53176081e-01 5.31785011e-01 5.31881154e-01
-7.99256116e-02 3.64275336e-01 -2.48048916e-01 3.84669155e-01
6.01145685e-01 -5.71337819e-01 -7.46390045e-01 6.21330678e-01
5.93185306e-01 -6.47211909e-01 -2.99294412e-01 1.72813535e+00
2.29908377e-01 1.48880333e-01 2.01115906e-01 -5.12714446e-01
-1.01687074e+00 5.84621310e-01 -1.03143024e+00 -1.79109558e-01
-8.80616754e-02 -1.42781660e-01 8.29445481e-01 -8.27942431e-01
2.95189142e-01 9.29926753e-01 5.68284035e-01 -4.79806885e-02
-1.44727743e+00 -6.89643562e-01 -1.33960277e-01 -7.63945878e-02
-1.48518276e+00 -5.35098985e-02 9.10970330e-01 -1.41959384e-01
3.54153216e-01 3.22702616e-01 7.57014990e-01 6.26542091e-01
-3.68393719e-01 1.89763591e-01 1.47778368e+00 -1.44900009e-01
-2.39333627e-03 -4.38076556e-01 -1.51797950e-01 2.93770850e-01
3.38939905e-01 2.52092898e-01 -5.66635668e-01 4.09897380e-02
1.14208281e+00 4.76938665e-01 -6.99824333e-01 -4.80559142e-03
-1.11659575e+00 6.22425675e-01 1.00473487e+00 2.46733457e-01
-8.21883976e-01 -2.67863184e-01 -1.84433699e-01 5.25678806e-02
7.04645634e-01 4.02850002e-01 -4.42553729e-01 4.29985762e-01
-1.16328490e+00 -1.64978549e-01 2.31587626e-02 5.46622813e-01
1.33100331e+00 3.09253335e-01 3.06224406e-01 9.27204549e-01
-3.71862538e-02 7.54342377e-01 3.18128496e-01 -1.09819710e+00
1.51399866e-01 5.51329732e-01 3.26642901e-01 -3.98457885e-01
-4.19004351e-01 -7.89456129e-01 -1.29403698e+00 3.78384531e-01
3.31839696e-02 -1.21001519e-01 -1.03598595e+00 1.50051916e+00
2.31486812e-01 2.84152389e-01 3.39049935e-01 1.38602459e+00
6.54357433e-01 9.73031044e-01 -2.44108215e-02 -5.06011665e-01
1.14286995e+00 -7.27211773e-01 -6.59203053e-01 -2.96827674e-01
-7.80940130e-02 -5.30240178e-01 9.94309545e-01 2.95272589e-01
-8.06906104e-01 -6.25394106e-01 -1.32389629e+00 -1.68146566e-01
-4.10471380e-01 2.14420751e-01 7.15327978e-01 2.68066823e-01
-8.14472616e-01 7.39733636e-01 -5.77396095e-01 5.54334484e-02
5.43578923e-01 1.16781265e-01 -4.53155428e-01 -1.92541957e-01
-1.21462011e+00 3.20924759e-01 7.77225792e-01 3.47379714e-01
-6.42701507e-01 -9.38303292e-01 -5.57666957e-01 2.58881807e-01
1.65698498e-01 -3.81665528e-01 7.68070936e-01 -1.23601794e+00
-1.43029904e+00 5.57066083e-01 -2.73305684e-01 -6.32393220e-03
-5.27073033e-02 4.13095839e-02 -6.44508660e-01 4.21319336e-01
1.28253579e-01 2.85613626e-01 1.01269591e+00 -1.42894506e+00
-8.18429768e-01 -7.87926674e-01 -2.35738177e-02 5.28913915e-01
-4.34540421e-01 -3.29847634e-01 1.84381872e-01 -3.34226251e-01
8.23461711e-01 -6.58545196e-01 -1.71342000e-01 -6.08301014e-02
-4.34733182e-01 6.64032161e-01 8.59399617e-01 -7.41448283e-01
4.89857197e-01 -2.35392690e+00 3.33801508e-01 -7.60220885e-02
3.54001224e-01 2.83117533e-01 -1.56194761e-01 1.13883689e-01
-6.74610078e-01 2.08912537e-01 -7.21870601e-01 1.40122205e-01
-4.49458838e-01 -2.25404091e-02 -4.26560104e-01 5.98563313e-01
1.87230736e-01 7.83797622e-01 -9.18541431e-01 9.29560885e-02
2.91826457e-01 8.37818384e-01 2.60617007e-02 4.12413329e-02
-9.88814905e-02 7.02670872e-01 -3.95653695e-01 7.37216890e-01
1.41642416e+00 -5.44917464e-01 1.34033188e-01 -5.84959447e-01
-5.00874758e-01 -7.36406446e-02 -1.03347206e+00 1.90270889e+00
-4.36848611e-01 3.79228175e-01 4.65637118e-01 -8.29849243e-01
8.12054694e-01 2.09127322e-01 6.95235252e-01 -7.88675606e-01
-3.70309949e-01 2.30693161e-01 -5.23036897e-01 -2.17819273e-01
7.04114735e-01 -7.87653923e-01 4.07645404e-01 3.90518725e-01
-2.97692746e-01 -1.69079334e-01 -3.43764633e-01 -4.49646980e-01
2.65223444e-01 1.90657809e-01 1.47407323e-01 -1.08011067e-01
4.95078772e-01 2.74534732e-01 3.95299375e-01 4.86641228e-01
7.47449547e-02 7.18319893e-01 -4.76519373e-04 -3.18376869e-01
-1.30614352e+00 -1.12047970e+00 -3.93040955e-01 9.79522705e-01
2.56911248e-01 4.70343620e-01 -4.94908184e-01 9.37993005e-02
-4.77843657e-02 4.73120600e-01 -4.97436911e-01 -8.05267617e-02
-1.45079732e-01 -1.24495602e+00 3.04165632e-01 3.44407916e-01
1.14854991e+00 -7.66679585e-01 -3.60482663e-01 2.08754718e-01
-3.21102440e-01 -1.02805007e+00 8.65257159e-02 2.80204862e-01
-1.01130390e+00 -8.78706157e-01 -4.95484263e-01 -3.93101245e-01
4.74866480e-01 1.11110067e+00 5.37588418e-01 -3.31017464e-01
-3.78438950e-01 -3.37246239e-01 -4.24202204e-01 -1.88007906e-01
-1.57480821e-01 -9.13043320e-02 -2.21960783e-01 3.08566839e-01
3.03127527e-01 -8.20839465e-01 -6.70119226e-01 -2.79653966e-01
-1.37217093e+00 5.28918028e-01 7.29740202e-01 8.86386335e-01
8.91997337e-01 8.36777925e-01 4.20897812e-01 -7.49825716e-01
3.58166516e-01 -4.15104955e-01 -7.25209594e-01 1.30410969e-01
-4.55905944e-01 -1.18747167e-01 8.22546661e-01 -1.56965971e-01
-1.57466686e+00 3.35058361e-01 1.73144728e-01 -2.82194763e-01
-3.77557397e-01 9.40880895e-01 -1.39042273e-01 -3.05729330e-01
7.89900780e-01 7.68531978e-01 -1.71977524e-02 -4.69667852e-01
5.34582794e-01 8.64103019e-01 8.43374074e-01 -3.33663262e-02
1.07296610e+00 1.02220345e+00 2.54115820e-01 -9.67235029e-01
-1.31427550e+00 -5.06338298e-01 -8.36098194e-01 3.88167016e-02
9.04375196e-01 -1.33792233e+00 -4.29632306e-01 5.02333283e-01
-5.80503941e-01 -1.26360938e-01 -1.82569697e-01 5.51119566e-01
-2.31229201e-01 3.45059216e-01 -4.94780302e-01 -6.60626292e-01
-3.13407004e-01 -9.71668482e-01 1.27607846e+00 4.81209099e-01
7.21809566e-01 -7.26393342e-01 -2.28237957e-01 4.50449646e-01
5.16863585e-01 4.27652955e-01 8.66540551e-01 1.69941336e-01
-7.81897068e-01 -2.45359000e-02 -9.88843799e-01 3.03542078e-01
7.61265099e-01 -1.24817379e-01 -1.23705149e+00 -2.43392095e-01
2.40335733e-01 -4.45129663e-01 1.01914835e+00 3.95989835e-01
1.38571572e+00 -6.81179613e-02 2.16736093e-01 1.26431024e+00
2.11954927e+00 -3.69268395e-02 5.05686998e-01 4.23947394e-01
8.67064357e-01 4.11484152e-01 5.67098022e-01 5.48420906e-01
-8.99518728e-02 1.22456744e-01 8.87595177e-01 -5.95416188e-01
-1.34494931e-01 -9.40081328e-02 -1.51756167e-01 1.25468448e-01
-2.66798913e-01 1.49379998e-01 -5.78080356e-01 2.95425326e-01
-1.44034779e+00 -1.14639068e+00 -2.86409199e-01 2.10055661e+00
9.15082753e-01 -6.16412401e-01 -3.98119330e-01 1.60001621e-01
9.50183570e-01 6.34036660e-01 -9.98794794e-01 4.51844960e-01
-6.30266130e-01 3.02530885e-01 8.36869538e-01 6.46881282e-01
-1.11208534e+00 9.95157838e-01 6.11490631e+00 4.10763681e-01
-1.64434659e+00 3.47460397e-02 3.48611474e-01 -1.15303546e-01
-5.20312548e-01 -1.50097966e-01 -3.27098459e-01 3.21830958e-02
7.55137861e-01 -2.26889312e-01 1.14094341e+00 3.77517343e-01
3.66580069e-01 7.52595719e-03 -3.56139392e-01 1.17446792e+00
-2.79257745e-01 -1.34814286e+00 3.21198463e-01 1.69730827e-01
9.43603992e-01 1.73611462e-01 3.70842785e-01 -2.71730214e-01
3.28819782e-01 -1.14010417e+00 4.23212677e-01 5.59057951e-01
1.39590204e+00 -9.17589605e-01 3.32863092e-01 3.16505790e-01
-1.32461512e+00 -3.57844830e-01 -8.20188224e-01 -1.32388920e-02
-6.63829446e-02 8.33875716e-01 -5.70470095e-01 1.06198931e+00
6.91296339e-01 8.03317010e-01 -2.65328974e-01 5.98287106e-01
-2.61751920e-01 4.55282092e-01 -1.38832390e-01 7.35020399e-01
3.48513901e-01 -9.01682317e-01 1.96191087e-01 7.26835430e-01
6.17298424e-01 6.67536736e-01 -8.52767974e-02 1.43118501e+00
-1.25054657e-01 -2.65533179e-01 -4.38450575e-01 -3.88116091e-01
3.89075607e-01 1.53454602e+00 -3.65679771e-01 -2.40864858e-01
-4.34829324e-01 1.09310210e+00 3.59645151e-02 9.28960979e-01
-6.50631070e-01 -1.83505908e-01 7.93130696e-01 -2.33656392e-01
2.55133212e-01 -2.16690868e-01 -5.85701942e-01 -1.18238997e+00
-1.47103742e-01 -8.81463289e-01 1.03566304e-01 -1.47394478e+00
-9.29188490e-01 5.91178060e-01 -3.77425939e-01 -1.39151525e+00
1.35266542e-01 -5.61359763e-01 -5.61177470e-02 1.57029533e+00
-2.15450311e+00 -1.43685627e+00 -1.07414663e+00 6.94268346e-01
1.86311796e-01 3.79069567e-01 9.04327393e-01 5.02083786e-02
-5.07613659e-01 -3.16292793e-01 5.02137423e-01 -2.65323639e-01
4.63042110e-01 -1.25121009e+00 6.01581708e-02 1.01643217e+00
-4.07100976e-01 3.95745099e-01 5.37435472e-01 -5.39724827e-01
-1.37919927e+00 -1.53994358e+00 5.85585423e-02 3.88550222e-01
7.61205077e-01 1.81045383e-01 -1.19121516e+00 4.95830923e-01
-1.98793784e-02 1.97204277e-01 7.74610400e-01 -3.03376347e-01
-5.52483797e-01 -4.14884448e-01 -1.17772579e+00 3.30820203e-01
1.02919626e+00 -1.15621221e+00 -1.66260853e-01 5.79412937e-01
8.99551690e-01 -2.76097327e-01 -1.24685967e+00 5.91308653e-01
3.50077868e-01 -1.08234191e+00 1.20400691e+00 -3.54750037e-01
7.49951839e-01 -6.02631807e-01 -3.83444548e-01 -1.47991586e+00
-7.92298615e-01 -1.62699878e-01 4.08384740e-01 5.89463055e-01
2.65503734e-01 -7.41020143e-01 6.56414568e-01 3.20045829e-01
-1.61564276e-01 -5.64353280e-02 -6.25817597e-01 -4.41790611e-01
-2.20381487e-02 4.59240004e-02 1.16563451e+00 1.17928600e+00
-4.27485257e-01 -5.43495342e-02 -3.24756503e-01 1.18567336e+00
1.03337359e+00 8.44245851e-01 2.31824026e-01 -1.42230380e+00
3.95036191e-02 -4.08169419e-01 1.31061554e-01 -7.53819346e-01
1.36416495e-01 -7.41097569e-01 7.14004412e-02 -1.58603060e+00
4.75866437e-01 -8.70006997e-03 -4.37164098e-01 4.10756826e-01
-4.33537722e-01 5.03182948e-01 -1.72632605e-01 4.35948312e-01
6.78405985e-02 5.69650173e-01 1.61164141e+00 -2.96291739e-01
-1.62527859e-01 -3.15834105e-01 -8.58791053e-01 3.60794932e-01
8.71372283e-01 -1.54660821e-01 -4.12648797e-01 -6.03899002e-01
3.45606834e-01 5.74291229e-01 5.49169481e-01 -8.64355743e-01
8.43644813e-02 -5.71212590e-01 6.32456779e-01 -7.18381286e-01
5.07575810e-01 -8.88444602e-01 5.19284606e-01 8.38758126e-02
-2.01006338e-01 -8.26493561e-01 1.14371404e-01 4.15499926e-01
-4.32344317e-01 -8.65938738e-02 9.38125551e-01 -3.04733187e-01
-8.62348676e-01 7.05301464e-01 1.30623057e-01 -7.48057842e-01
5.93491554e-01 -3.19952041e-01 -5.70631564e-01 -1.74516425e-01
-7.10655808e-01 -2.68490255e-01 8.25940669e-01 -4.39494774e-02
6.26696825e-01 -1.21073544e+00 -8.31676245e-01 4.51131165e-01
2.16674447e-01 3.96788150e-01 7.32195318e-01 4.86362070e-01
-7.66159534e-01 2.02382103e-01 -4.79068607e-01 -5.16789138e-01
-9.65090394e-01 5.52064300e-01 8.26379657e-01 1.36844382e-01
-5.90624273e-01 7.19467103e-01 2.78938204e-01 -4.93697882e-01
-4.57877010e-01 -1.58364758e-01 -1.72585562e-01 -2.28835661e-02
8.36809814e-01 1.92270011e-01 -1.34215225e-02 -7.11293280e-01
1.13772668e-01 7.18586504e-01 4.66159970e-01 -1.19845711e-01
1.83572888e+00 -4.11085427e-01 -7.62774706e-01 3.50764513e-01
1.21733332e+00 -1.72361895e-01 -1.71673739e+00 -5.09969652e-01
-7.08146214e-01 -7.17257261e-01 6.83790147e-01 -6.61964595e-01
-1.22537899e+00 8.49137425e-01 5.65358162e-01 3.98190647e-01
1.77165031e+00 -3.16420555e-01 4.96693194e-01 4.42400694e-01
2.65102029e-01 -9.12339509e-01 -2.78717905e-01 2.76582569e-01
8.57153118e-01 -1.34471238e+00 2.67735720e-01 -5.71221590e-01
-2.40301266e-01 1.43872166e+00 2.49046385e-01 2.41159812e-01
3.40677053e-01 -4.90557514e-02 5.60008269e-03 -3.31499070e-01
-8.73570740e-02 -3.37984085e-01 4.48050685e-02 5.10351062e-01
2.56649017e-01 3.84503782e-01 3.94511938e-01 1.37783334e-01
-2.03709081e-01 1.48640826e-01 6.75717413e-01 4.89127249e-01
-7.26573467e-01 -5.12077332e-01 -6.45393848e-01 2.91754931e-01
-3.17082293e-02 -4.01941270e-01 -2.29601264e-01 3.17492396e-01
1.43396944e-01 8.89522314e-01 -9.22761038e-02 -3.11036766e-01
4.13860679e-02 3.75419036e-02 1.74067482e-01 -6.64394796e-01
2.23944291e-01 3.00704509e-01 -3.38997930e-01 -5.10773182e-01
-8.11488330e-01 -4.68159586e-01 -1.25121510e+00 -2.26289287e-01
-2.59166688e-01 2.02764175e-03 8.09968472e-01 8.14197123e-01
1.24933116e-01 7.20530152e-01 8.89830589e-01 -1.07829487e+00
-5.83763421e-01 -1.06033182e+00 -1.42036092e+00 3.55087906e-01
8.06448042e-01 -2.77956396e-01 -6.24524891e-01 9.00625288e-02] | [10.210756301879883, -2.0200018882751465] |
6a0e9ba3-daab-4895-9a7e-2d15c4922f2d | morel-multi-omics-relational-learning-1 | 2203.08149 | null | https://arxiv.org/abs/2203.08149v1 | https://arxiv.org/pdf/2203.08149v1.pdf | MoReL: Multi-omics Relational Learning | Multi-omics data analysis has the potential to discover hidden molecular interactions, revealing potential regulatory and/or signal transduction pathways for cellular processes of interest when studying life and disease systems. One of critical challenges when dealing with real-world multi-omics data is that they may manifest heterogeneous structures and data quality as often existing data may be collected from different subjects under different conditions for each type of omics data. We propose a novel deep Bayesian generative model to efficiently infer a multi-partite graph encoding molecular interactions across such heterogeneous views, using a fused Gromov-Wasserstein (FGW) regularization between latent representations of corresponding views for integrative analysis. With such an optimal transport regularization in the deep Bayesian generative model, it not only allows incorporating view-specific side information, either with graph-structured or unstructured data in different views, but also increases the model flexibility with the distribution-based regularization. This allows efficient alignment of heterogeneous latent variable distributions to derive reliable interaction predictions compared to the existing point-based graph embedding methods. Our experiments on several real-world datasets demonstrate enhanced performance of MoReL in inferring meaningful interactions compared to existing baselines. | ['Xiaoning Qian', 'Nick Duffield', 'Ehsan Hajiramezanali', 'Arman Hasanzadeh'] | 2022-03-15 | morel-multi-omics-relational-learning | https://openreview.net/forum?id=DnG75_KyHjX | https://openreview.net/pdf?id=DnG75_KyHjX | iclr-2022-4 | ['relational-reasoning'] | ['natural-language-processing'] | [ 1.62896365e-01 -3.79467495e-02 -1.97397158e-01 -2.68612355e-01
-4.27084953e-01 -7.19253957e-01 3.73160958e-01 3.16607475e-01
3.52921993e-01 8.22875738e-01 6.24182940e-01 -1.27613187e-01
-4.69310433e-01 -7.74573267e-01 -7.50529051e-01 -1.19992900e+00
5.85962981e-02 9.43738878e-01 -1.60692230e-01 3.09627980e-01
-1.82720765e-01 3.10210049e-01 -8.79135311e-01 4.43263352e-01
8.28489482e-01 5.32187879e-01 1.19921811e-01 5.12819707e-01
-6.78036287e-02 -3.71945382e-04 -1.40213013e-01 -4.55529749e-01
1.39132932e-01 -2.45967612e-01 -2.48584598e-01 1.15517095e-01
3.75887990e-01 5.73486388e-02 -3.84651273e-01 1.06431592e+00
5.87523937e-01 -1.99686691e-01 9.36863601e-01 -1.23683763e+00
-9.61189449e-01 1.84185624e-01 -7.84922659e-01 -8.75780452e-03
1.16597258e-01 3.78549069e-01 1.33816636e+00 -6.07916117e-01
1.08311605e+00 1.68336773e+00 6.47870749e-02 2.24894211e-01
-1.94967055e+00 -4.14426059e-01 2.91805118e-01 9.89693031e-02
-1.14546609e+00 -5.68853207e-02 6.63115442e-01 -7.33905256e-01
8.08052778e-01 3.49061340e-01 8.64875972e-01 1.33616483e+00
7.55513668e-01 3.58853489e-01 8.04670393e-01 2.76645750e-01
1.74796790e-01 -3.20362568e-01 1.54526621e-01 7.27880299e-01
5.34126043e-01 -2.96527684e-01 -5.27944088e-01 -8.71203244e-01
2.97166169e-01 8.16957057e-01 -2.45363638e-01 -8.06844056e-01
-1.51738715e+00 7.98641503e-01 2.55695581e-01 1.44579902e-01
-3.80549341e-01 2.12892875e-01 2.90689707e-01 2.50274986e-02
6.97949171e-01 3.31378758e-01 -6.15773916e-01 4.21468288e-01
-3.58358443e-01 1.86783120e-01 4.44878191e-01 8.34089220e-01
8.34131539e-01 -2.71489024e-01 -1.37753710e-01 7.02839434e-01
8.23484063e-01 4.02152926e-01 1.49983928e-01 -6.43222868e-01
4.38637197e-01 9.89273727e-01 -9.76019874e-02 -1.29198897e+00
-4.21190530e-01 -3.41792375e-01 -1.15521026e+00 -3.60902399e-01
1.53065756e-01 -3.71429436e-02 -8.65275085e-01 1.78908539e+00
8.05334985e-01 3.07142973e-01 -1.50131986e-01 6.29575133e-01
7.86521375e-01 8.26243460e-01 1.66557774e-01 -3.13882530e-01
1.59189332e+00 -4.75652397e-01 -8.18050146e-01 1.43433362e-01
5.19774914e-01 -4.33459610e-01 6.88026369e-01 1.57805145e-01
-6.21350706e-01 -7.40807652e-02 -8.05325329e-01 -1.95613489e-01
-4.19722348e-01 -1.59414172e-01 7.06093907e-01 4.36350219e-02
-6.12507105e-01 3.00514132e-01 -8.50742698e-01 -2.52926230e-01
5.29121339e-01 3.57595265e-01 -6.85845733e-01 -3.46992999e-01
-1.00302339e+00 3.43485445e-01 2.29334906e-01 9.72180218e-02
-9.53314960e-01 -1.04690564e+00 -7.55137265e-01 3.02271247e-01
3.53535086e-01 -1.16646087e+00 2.33795661e-02 -6.11140840e-02
-1.16885602e+00 4.97638047e-01 -4.02603269e-01 2.64769495e-01
-4.85614799e-02 5.45631163e-02 -3.43786329e-01 1.38893425e-01
4.08175141e-02 4.11633968e-01 5.47643900e-01 -1.12620294e+00
2.21576810e-01 -1.02500713e+00 -1.90673113e-01 2.28689015e-01
-9.09581035e-02 -4.24925357e-01 -1.62777603e-01 -4.36835140e-01
5.12338638e-01 -1.05211294e+00 -3.15824002e-01 2.68367082e-01
-7.40184367e-01 -9.30526406e-02 8.70071352e-01 -5.56916475e-01
7.47148514e-01 -1.85916770e+00 1.07184446e+00 1.33973524e-01
9.45042849e-01 -5.34621440e-02 -3.04723084e-01 9.24795806e-01
-2.76738077e-01 4.32890534e-01 -2.44626522e-01 -5.65621853e-02
-3.08272671e-02 6.52894601e-02 -2.11958960e-01 7.62739420e-01
5.76874614e-02 1.10777116e+00 -8.85000408e-01 -1.04061797e-01
1.04868576e-01 7.67355502e-01 -5.83241105e-01 2.33953759e-01
-5.55030942e-01 7.83714771e-01 -7.77315497e-01 7.36696064e-01
7.15212286e-01 -7.72910416e-01 7.77086020e-01 -5.64265132e-01
3.59413654e-01 -1.39370605e-01 -9.23614383e-01 1.70608890e+00
4.24145302e-03 2.58561432e-01 -6.24438040e-02 -1.09143400e+00
6.09166026e-01 3.96668136e-01 6.60235941e-01 -9.08251405e-02
-1.14045233e-01 -2.41598710e-01 2.15396643e-01 -4.68966126e-01
-1.43760875e-01 -8.30873623e-02 1.22100167e-01 3.41927201e-01
1.57100514e-01 3.46058577e-01 -2.36311048e-01 4.79245275e-01
1.19361746e+00 -8.65005031e-02 1.14997581e-01 -4.50374961e-01
2.57152528e-01 -4.41514909e-01 8.53567004e-01 6.19844198e-02
-1.82506815e-02 3.70187134e-01 1.03181231e+00 -4.26092476e-01
-1.08546889e+00 -1.16345572e+00 -1.28146708e-01 6.90153301e-01
1.31487921e-01 -5.70969403e-01 -1.32575050e-01 -6.49843276e-01
2.98537225e-01 1.87217191e-01 -6.95232809e-01 -2.85589069e-01
1.33769035e-01 -1.40219760e+00 2.00681180e-01 7.21250623e-02
-1.91550046e-01 -1.98050395e-01 4.34307009e-01 1.43461332e-01
-7.97482729e-02 -8.70732427e-01 -4.96138006e-01 1.46769270e-01
-9.19366419e-01 -1.21486521e+00 -5.02844512e-01 -1.22036256e-01
9.58734691e-01 4.66973782e-01 6.86706424e-01 -2.99700528e-01
-6.16026103e-01 1.03642471e-01 -3.16984132e-02 -9.35885906e-02
-2.15281278e-01 -2.99336106e-01 1.89695776e-01 2.66397238e-01
4.57061172e-01 -7.72803724e-01 -9.25734222e-01 2.17561141e-01
-1.14622796e+00 2.38694429e-01 3.92949939e-01 1.08328962e+00
9.09497797e-01 -3.74632329e-01 5.72661161e-01 -1.06629896e+00
4.85830992e-01 -9.29952264e-01 -7.16100276e-01 4.91343826e-01
-6.26435637e-01 3.14023048e-01 5.65921009e-01 -2.52791882e-01
-7.73288190e-01 -1.47072241e-01 2.74427891e-01 -5.28843462e-01
-4.51024026e-02 7.16754615e-01 -6.93429291e-01 1.69622451e-01
1.69316277e-01 1.77527815e-01 1.44427806e-01 -4.21789199e-01
5.58917165e-01 3.60428989e-01 -4.51546013e-01 -5.06245613e-01
3.28948855e-01 7.36067116e-01 5.94941437e-01 -8.00805151e-01
-5.25683343e-01 -5.00821888e-01 -8.05569232e-01 2.12208197e-01
1.17767191e+00 -1.05213141e+00 -1.02513206e+00 1.89246342e-01
-9.46342051e-01 2.13151246e-01 3.53368729e-01 5.20484149e-01
-2.11086825e-01 6.44263983e-01 -7.79267371e-01 -4.19927925e-01
-2.28985578e-01 -1.39968991e+00 1.42387247e+00 -1.42627805e-01
-3.52917537e-02 -1.22788811e+00 6.71884179e-01 6.19453013e-01
-1.02681525e-01 5.14278352e-01 1.55841804e+00 -4.81531501e-01
-7.72018790e-01 -4.54800092e-02 -2.99810350e-01 -1.58333063e-01
4.23136383e-01 1.74670532e-01 -7.28755295e-01 -3.46008599e-01
-3.58484477e-01 -1.31552309e-01 8.68915856e-01 5.54872274e-01
1.10976577e+00 -3.44105124e-01 -7.35131085e-01 8.06987345e-01
1.31494391e+00 -7.06284568e-02 3.67216796e-01 -4.67899948e-01
1.33214855e+00 6.77254260e-01 2.66361594e-01 4.27783340e-01
3.22519273e-01 8.10194254e-01 4.82111216e-01 -8.34434573e-03
4.50217843e-01 -3.54349673e-01 2.79174447e-01 9.19003963e-01
9.56315100e-02 -5.56163490e-01 -6.86604023e-01 1.97666913e-01
-2.15597725e+00 -9.38584626e-01 -4.63763475e-01 2.10005713e+00
6.82472527e-01 -6.26507640e-01 -5.52967303e-02 -7.10778415e-01
7.93797612e-01 4.32882011e-01 -1.04470587e+00 -3.01060695e-02
-2.93097794e-01 -2.99239546e-01 1.87484771e-01 4.15782154e-01
-6.88559115e-01 5.93726337e-01 6.08138132e+00 5.25374949e-01
-9.50140536e-01 1.39283419e-01 7.90615976e-01 -3.79648179e-01
-8.95648539e-01 1.94601089e-01 -8.34275484e-01 5.66230714e-01
8.95248771e-01 -7.21473545e-02 2.48404190e-01 4.21742439e-01
4.52319145e-01 1.00961201e-01 -1.17026734e+00 9.35404301e-01
-1.10613987e-01 -1.65273213e+00 4.01924223e-01 6.44278467e-01
6.76962674e-01 1.62936524e-01 -1.32340685e-01 -3.08116168e-01
4.84509259e-01 -9.69069600e-01 -5.57322167e-02 6.00398898e-01
5.33490181e-01 -6.26607180e-01 3.31786543e-01 3.17287713e-01
-9.95788991e-01 4.30338234e-01 -8.33861113e-01 3.61307621e-01
1.39421955e-01 1.12628019e+00 -8.12396049e-01 7.28114665e-01
2.62301087e-01 1.16819477e+00 -1.70446292e-01 4.47771817e-01
1.68321617e-02 3.13653350e-01 -1.21603243e-01 2.15730861e-01
-1.77186221e-01 -9.89310920e-01 7.58688152e-01 6.62003696e-01
3.19927454e-01 2.76163537e-02 2.71692544e-01 1.25068438e+00
-1.43267348e-01 4.49401513e-02 -8.68891418e-01 -6.80053949e-01
2.37743393e-01 1.53394186e+00 -5.52826464e-01 -1.35462373e-01
-4.24074769e-01 8.55886817e-01 5.56704640e-01 6.52380228e-01
-7.54658401e-01 2.30750218e-01 1.16018605e+00 2.08001345e-01
4.75653745e-02 -4.69122559e-01 1.77262768e-01 -1.67168605e+00
-1.68974921e-01 -7.47364759e-01 4.61936772e-01 -6.13162637e-01
-1.78590608e+00 1.26261652e-01 4.42367643e-02 -1.09591556e+00
3.05339277e-01 -6.59375370e-01 -3.33278775e-01 1.01568389e+00
-1.20656168e+00 -1.45032763e+00 9.09487009e-02 1.98740363e-01
1.22030444e-01 -1.37490734e-01 9.82941329e-01 2.16166943e-01
-9.90010023e-01 7.58281723e-02 5.47285080e-01 -4.03784126e-01
8.60317111e-01 -1.12974870e+00 1.87552154e-01 4.37434763e-01
1.21210106e-02 1.04435337e+00 5.79650223e-01 -1.00843656e+00
-2.06004834e+00 -1.25566936e+00 2.38538995e-01 -5.65048873e-01
7.94367135e-01 -8.16374540e-01 -1.02825081e+00 7.66657829e-01
9.61819366e-02 1.44605428e-01 1.56495333e+00 4.71281141e-01
-6.63388431e-01 8.40002298e-02 -9.86843109e-01 6.28819704e-01
1.02605009e+00 -5.13928056e-01 -1.33884385e-01 6.91311002e-01
9.10342813e-01 -3.37130316e-02 -1.36481583e+00 2.99649686e-01
5.69292486e-01 -5.94011188e-01 1.05852520e+00 -1.22431672e+00
4.14044261e-01 -5.52197218e-01 -3.01343650e-01 -1.54790795e+00
-6.15774691e-01 -4.02499497e-01 -2.84800798e-01 9.51279879e-01
4.24197584e-01 -8.12504292e-01 5.46687067e-01 6.18987262e-01
4.56281938e-02 -8.15483451e-01 -9.09186542e-01 -2.72753119e-01
-1.38487354e-01 5.85746393e-02 5.02679467e-01 1.20813227e+00
2.42075119e-02 6.28162146e-01 -5.53184152e-01 3.53320688e-01
7.74616063e-01 2.99634814e-01 9.17740345e-01 -1.48897803e+00
-5.82088351e-01 3.34911793e-02 -6.99123263e-01 -6.73871040e-01
1.61215648e-01 -1.06922996e+00 -4.14894015e-01 -1.66545987e+00
7.98394680e-01 -1.56937856e-02 -2.79201925e-01 2.99887210e-01
-3.74886721e-01 -1.74507946e-01 -1.04402058e-01 2.61251628e-01
-5.16823530e-01 9.49711859e-01 1.49441040e+00 -3.79523128e-01
7.01718479e-02 -4.63810235e-01 -8.69512975e-01 4.11882132e-01
2.48637870e-01 -5.73472083e-01 -7.47246027e-01 -2.01083168e-01
6.18406236e-01 1.67657733e-01 3.93428653e-01 -1.33256972e-01
-7.31277466e-02 -4.57381219e-01 5.99544406e-01 -4.61934596e-01
4.93594676e-01 -8.38330567e-01 1.00571501e+00 3.82063240e-01
-3.25292468e-01 -5.61058596e-02 6.59061745e-02 1.45416009e+00
-5.82291326e-03 3.66984218e-01 5.44130981e-01 -1.07705206e-01
-2.02117804e-02 8.70924652e-01 -1.63085386e-01 -1.75582916e-01
1.17020595e+00 1.50419980e-01 -7.20172703e-01 -2.00419441e-01
-9.41058397e-01 2.40845978e-01 5.65361559e-01 3.28145266e-01
5.83015621e-01 -1.38282478e+00 -6.71749532e-01 2.40757838e-01
4.20953989e-01 4.64887638e-03 7.58064449e-01 9.68372285e-01
-2.54318774e-01 3.34341466e-01 -4.65145446e-02 -9.31579351e-01
-1.40118027e+00 5.48694611e-01 5.70989475e-02 -4.87818033e-01
-5.66928267e-01 6.12985194e-01 1.04118216e+00 -6.10831678e-01
-4.95653361e-01 -1.09146424e-01 -1.94293931e-02 2.55641252e-01
3.71634156e-01 1.79403067e-01 -1.63126126e-01 -5.16740382e-01
-3.71355712e-01 5.34530818e-01 -3.01260799e-01 2.29312018e-01
1.58715832e+00 -1.45705417e-01 -6.11992002e-01 7.34611988e-01
1.45329809e+00 -7.62400310e-03 -1.10638869e+00 -6.08844906e-02
-4.52854842e-01 -5.65330148e-01 7.16848820e-02 -5.28793573e-01
-9.49420989e-01 1.01093709e+00 2.76620239e-01 -4.41630073e-02
5.69542706e-01 7.80785456e-02 5.10905325e-01 3.50684941e-01
2.87952751e-01 -5.61671317e-01 -7.54368529e-02 6.00173995e-02
9.25852716e-01 -1.18347061e+00 3.60754937e-01 -6.97041571e-01
-5.87071180e-01 9.86437798e-01 4.11243290e-01 1.36689534e-02
7.87918806e-01 -1.11990355e-01 -2.56889254e-01 -7.52863765e-01
-1.22591293e+00 3.04385155e-01 5.37515163e-01 4.73064303e-01
6.02139473e-01 2.33204797e-01 -1.82786316e-01 4.33064967e-01
5.92357814e-01 -3.90810728e-01 2.63879955e-01 4.11230057e-01
-1.28182203e-01 -1.39643443e+00 4.26238403e-02 7.86810696e-01
-4.17892247e-01 -1.46139339e-01 -5.34676611e-01 3.03920507e-01
-2.68687814e-01 5.84550798e-01 -1.09149329e-01 -2.80464709e-01
-1.18200421e-01 1.72038510e-01 2.81037986e-01 -9.30163383e-01
-7.28000402e-02 6.21924579e-01 -2.15360820e-01 -5.99837124e-01
-3.45586747e-01 -7.23236859e-01 -1.11106074e+00 -2.65652031e-01
-3.81641805e-01 1.23348050e-01 5.32544494e-01 8.07753325e-01
1.35336220e+00 7.10022390e-01 5.45998931e-01 -6.03832364e-01
-2.45468259e-01 -7.58589625e-01 -8.32148254e-01 4.27365303e-01
1.10536672e-01 -8.23867202e-01 -3.29935849e-01 2.54196860e-02] | [6.0380635261535645, 5.705058574676514] |
733eef2f-165b-46b0-9a9f-44a7dd58a996 | symbiotic-message-passing-model-for-transfer | 2304.07017 | null | https://arxiv.org/abs/2304.07017v1 | https://arxiv.org/pdf/2304.07017v1.pdf | Symbiotic Message Passing Model for Transfer Learning between Anti-Fungal and Anti-Bacterial Domains | Machine learning, and representation learning in particular, has the potential to facilitate drug discovery by screening billions of compounds. For example, a successful approach is representing the molecules as a graph and utilizing graph neural networks (GNN). Yet, these approaches still require experimental measurements of thousands of compounds to construct a proper training set. While in some domains it is easier to acquire experimental data, in others it might be more limited. For example, it is easier to test the compounds on bacteria than perform in-vivo experiments. Thus, a key question is how to utilize information from a large available dataset together with a small subset of compounds where both domains are measured to predict compounds' effect on the second, experimentally less available domain. Current transfer learning approaches for drug discovery, including training of pre-trained modules or meta-learning, have limited success. In this work, we develop a novel method, named Symbiotic Message Passing Neural Network (SMPNN), for merging graph-neural-network models from different domains. Using routing new message passing lanes between them, our approach resolves some of the potential conflicts between the different domains, and implicit constraints induced by the larger datasets. By collecting public data and performing additional high-throughput experiments, we demonstrate the advantage of our approach by predicting anti-fungal activity from anti-bacterial activity. We compare our method to the standard transfer learning approach and show that SMPNN provided better and less variable performances. Our approach is general and can be used to facilitate information transfer between any two domains such as different organisms, different organelles, or different environments. | ['Yonatan Savir', 'Tanya Wasserman', 'Ronen Taub'] | 2023-04-14 | null | null | null | null | ['drug-discovery'] | ['medical'] | [ 5.7374442e-01 -1.9788805e-02 -3.8155919e-01 9.9390402e-02
-4.0749002e-01 -7.5315917e-01 5.1812607e-01 6.5088117e-01
-2.0480128e-01 1.2177567e+00 -3.8263279e-01 -8.5250986e-01
-1.3073683e-01 -1.0947503e+00 -1.1074167e+00 -6.7091995e-01
-2.6967984e-02 7.6712430e-01 3.6775929e-01 -1.8389726e-01
2.8616217e-01 6.7189574e-01 -9.3519020e-01 3.6988911e-01
7.5886649e-01 5.6613660e-01 4.6049753e-01 4.0131101e-01
-3.7219456e-01 6.5662199e-01 -7.3096585e-01 -2.1632731e-01
1.2147407e-01 -5.6698054e-01 -9.0018082e-01 -2.6377615e-01
-4.5778034e-03 1.1023126e-01 1.3626583e-01 9.1547394e-01
4.3136129e-01 -9.3601376e-02 5.7419014e-01 -1.1216959e+00
-6.7157698e-01 2.6618758e-01 -2.5132948e-01 -2.2097790e-01
3.6796910e-01 1.7067480e-01 5.4407156e-01 -4.7325438e-01
8.9172202e-01 9.9755841e-01 6.2032968e-01 6.0535526e-01
-1.4883220e+00 -7.6614994e-01 -8.9344993e-02 1.5850623e-01
-1.0750009e+00 -1.9035656e-02 5.0870675e-01 -5.2267849e-01
1.0378616e+00 5.2984364e-02 6.7838895e-01 1.2679170e+00
3.2988274e-01 3.7394524e-01 1.1083710e+00 -2.8960568e-01
4.7668210e-01 2.6266709e-01 -2.6635695e-01 5.6745112e-01
6.5081018e-01 2.6305582e-02 -4.0342730e-01 -5.0932908e-01
7.2204798e-01 2.0426759e-01 -4.4489735e-01 -7.2274697e-01
-1.0413107e+00 9.9431837e-01 4.7214070e-01 4.7828212e-01
-2.2913738e-01 -8.4233865e-02 2.8900459e-01 4.6335942e-01
4.9506846e-01 9.7712576e-01 -8.5886908e-01 2.4689980e-01
-4.7530890e-01 -2.2387872e-02 1.1633795e+00 4.8035961e-01
9.6570647e-01 -2.3791294e-01 5.2316874e-01 5.8493584e-01
2.8148752e-01 -5.7299100e-02 3.9739907e-01 -4.4555148e-01
2.4575451e-01 7.0624566e-01 4.7340155e-02 -9.7543794e-01
-4.2843637e-01 -2.0535688e-01 -7.2614336e-01 1.5219350e-01
7.9959774e-01 -1.8498269e-01 -8.7881643e-01 1.8285503e+00
4.0340725e-01 5.6973886e-01 2.9034856e-01 5.2171963e-01
6.8165362e-01 6.9866478e-01 2.9084951e-01 -3.6331132e-02
9.9836588e-01 -6.4062250e-01 -1.5594727e-01 -1.4346954e-02
1.0901102e+00 -6.2088096e-01 5.5051106e-01 5.1808709e-01
-6.0136515e-01 -2.1243472e-01 -1.1617335e+00 2.9664040e-01
-1.1498353e+00 -4.1359144e-01 8.3874291e-01 4.5814866e-01
-8.2899344e-01 1.2559021e+00 -6.7033112e-01 -6.7438477e-01
4.7025126e-01 8.2789773e-01 -6.9739282e-01 -2.7263200e-01
-1.2578906e+00 8.6703843e-01 6.1782432e-01 -3.7438419e-01
-1.0173204e+00 -8.8175583e-01 -6.9629306e-01 -1.4660726e-02
2.5938788e-01 -8.4019601e-01 7.0083559e-01 -9.3768752e-01
-1.5108266e+00 6.2968719e-01 2.6714736e-01 -4.4648591e-01
1.4401214e-01 1.3798960e-01 -1.5202141e-01 -9.6651660e-03
-1.4122805e-01 7.1457183e-01 2.8156987e-01 -1.1069682e+00
-3.0522308e-01 -5.3445214e-01 2.2800107e-01 -1.0033105e-01
-3.9389902e-01 -1.5884195e-01 -9.0023004e-02 -3.3085603e-01
-3.4068510e-01 -9.5864618e-01 -4.5324296e-01 -8.4301397e-02
-3.2113791e-01 -2.1760973e-01 8.9136124e-01 -4.9907035e-01
5.7646275e-01 -1.7320268e+00 4.3319431e-01 2.6266766e-01
2.9426330e-01 6.7953312e-01 -5.6894505e-01 7.5091666e-01
-1.6567068e-01 3.6497873e-01 -2.6352328e-01 3.8867548e-01
-6.2114698e-01 2.4476700e-01 6.9635592e-02 2.5408581e-01
5.3629303e-01 7.1309650e-01 -1.1203283e+00 -1.6804163e-01
2.1583363e-02 7.6637322e-01 -4.9489623e-01 1.8397647e-01
-5.9093535e-01 1.0466973e+00 -5.2539027e-01 4.8029119e-01
5.7871550e-01 -7.5949371e-01 7.7031213e-01 9.4665959e-02
9.7713932e-02 3.1729311e-01 -9.6879643e-01 1.6363363e+00
-2.5657815e-01 3.3123845e-01 -1.7119129e-01 -1.4966394e+00
9.6200335e-01 2.8881788e-01 6.3391221e-01 -6.5170115e-01
-8.4139984e-03 3.7474248e-01 2.6284799e-01 -2.4183460e-01
-1.3955268e-01 -1.0237280e-01 4.6056065e-01 2.3327072e-01
2.8993025e-01 -1.3549156e-01 2.7873296e-01 -3.7023298e-02
1.4656079e+00 2.8717256e-01 4.1858944e-01 -4.5611776e-02
5.2122724e-01 2.6827678e-01 4.9155742e-01 4.3971270e-01
4.5944728e-02 1.9549176e-01 7.5580698e-01 -4.7686094e-01
-1.0796311e+00 -7.1363187e-01 -9.5307365e-02 9.5053399e-01
8.9948356e-02 -3.7934071e-01 -6.3094628e-01 -8.0357838e-01
1.7878911e-01 3.9630914e-01 -5.9565181e-01 -1.9828847e-01
-4.1295877e-01 -9.2121202e-01 5.5278432e-01 1.3430960e-01
1.1813979e-01 -8.2897693e-01 -1.0457909e-01 5.0806510e-01
2.5159174e-01 -8.3623743e-01 3.2828575e-01 5.9964710e-01
-9.0365905e-01 -1.4238138e+00 -8.6567467e-01 -8.7799644e-01
6.7477018e-01 2.8858012e-01 1.0574907e+00 2.3158512e-01
-2.7765983e-01 -5.8973230e-02 -2.5085098e-01 -7.2386789e-01
-6.3567817e-01 1.7906092e-01 -1.4929757e-01 -2.8860417e-01
3.7278384e-01 -6.6818714e-01 -5.2137005e-01 4.1967696e-01
-9.3369591e-01 -1.6196376e-02 6.0198939e-01 1.1187476e+00
5.1963866e-01 -1.2673852e-01 9.4414681e-01 -1.2800668e+00
6.6465169e-01 -8.7479955e-01 -6.3918954e-01 3.2498559e-01
-6.7024690e-01 2.3563271e-02 8.6453092e-01 -7.9785359e-01
-5.3508246e-01 8.1847996e-02 -5.1560984e-03 -8.5035108e-02
-2.2413008e-01 8.4972972e-01 -2.1352951e-01 -4.5965663e-01
7.9537439e-01 5.5496269e-03 2.8483111e-01 -4.2003673e-01
5.7492912e-02 5.7076728e-01 -1.3448682e-01 -6.9408244e-01
5.9012192e-01 1.9244242e-01 3.0227244e-01 -7.5679612e-01
-3.9713883e-01 -4.3755275e-01 -4.8605379e-01 2.5953245e-01
7.8945655e-01 -7.3567694e-01 -6.6065758e-01 3.2420239e-01
-1.2339106e+00 -6.0939205e-01 1.1629493e-03 5.2229059e-01
-3.0210418e-01 3.4748015e-01 -6.4366925e-01 -2.7960664e-01
-1.2171850e-01 -1.3080704e+00 8.0219334e-01 2.3512378e-01
-1.6903517e-01 -1.3158753e+00 5.9425908e-01 2.0666261e-01
4.8190898e-01 4.3805856e-01 1.2360014e+00 -1.1636409e+00
-6.8340993e-01 -1.0546853e-01 -2.9386139e-01 1.4535123e-01
6.5268058e-01 -1.4957878e-01 -7.5738895e-01 -4.3021211e-01
-3.5574356e-01 -6.5622598e-01 7.2230226e-01 1.3762340e-01
1.0979451e+00 -1.3650122e-01 -7.1464157e-01 3.7187910e-01
1.4908928e+00 6.0257900e-01 8.8368332e-01 4.8844367e-01
7.1883219e-01 7.5664586e-01 2.3628685e-01 1.2725294e-02
1.0933294e-01 5.4117203e-01 4.1797677e-01 -4.7341242e-01
4.9551036e-02 -1.2630421e-01 2.9309851e-01 6.2440395e-01
-2.4987940e-01 -5.5790520e-01 -9.2440873e-01 2.2445376e-01
-1.7337862e+00 -7.1515828e-01 1.2408817e-01 2.3802006e+00
1.0128260e+00 -5.1322199e-02 8.0016710e-02 -1.4510989e-01
6.4246911e-01 -4.9518737e-01 -7.9237032e-01 -3.9671317e-01
-1.9815254e-01 6.1758029e-01 3.7773299e-01 3.4430963e-01
-1.0410639e+00 8.4763384e-01 6.0399361e+00 7.1709597e-01
-1.4919907e+00 -1.9288892e-01 7.3585832e-01 3.5513651e-01
-1.6024575e-01 2.6003328e-01 -5.8566684e-01 2.4268445e-01
1.3564740e+00 -1.4543751e-01 4.3177241e-01 6.1834455e-01
-8.1357650e-02 5.3852797e-02 -1.4190109e+00 5.8937633e-01
-6.8001442e-02 -1.8656375e+00 2.6126283e-01 3.6544681e-01
6.4962256e-01 2.1969214e-01 -2.3533936e-01 1.8793176e-01
7.1389496e-01 -1.2523350e+00 -2.4689536e-01 2.8626072e-01
4.6912852e-01 -5.7144386e-01 6.7256552e-01 3.7581706e-01
-8.4238321e-01 1.9998936e-01 -5.2950650e-01 -5.7772975e-02
-4.0162975e-01 4.8906058e-01 -1.3451931e+00 8.9401042e-01
3.2327414e-01 8.3223361e-01 -7.5610650e-01 1.0936964e+00
1.0068087e-01 4.7309208e-01 -1.6249883e-01 -3.7009770e-01
1.0155827e-01 -4.5775399e-01 6.0294341e-02 1.0650527e+00
3.0508932e-01 -2.7256688e-01 3.6831659e-01 7.4845427e-01
-2.4364042e-01 3.6231571e-01 -1.1967089e+00 -5.5072719e-01
1.6313975e-01 1.2301958e+00 -7.0525682e-01 -3.6330187e-01
-5.3777629e-01 7.0464313e-01 5.2105016e-01 4.9538422e-01
-6.6073853e-01 -4.5149809e-01 4.8793787e-01 7.3476084e-02
2.4884129e-01 -8.2822479e-02 8.2331963e-02 -1.0309188e+00
-5.9003603e-01 -1.1714633e+00 3.7803656e-01 -5.4535252e-01
-1.4030796e+00 3.6678624e-01 -2.6137495e-01 -1.1326834e+00
-6.4293616e-02 -1.0676526e+00 -3.6433959e-01 7.7506632e-01
-1.6287072e+00 -1.0227749e+00 6.9739409e-02 3.6969340e-01
1.2962304e-01 -2.6510444e-01 1.1139520e+00 5.4090226e-01
-5.9769350e-01 1.5196529e-01 2.6972875e-01 -1.6330127e-02
9.1243804e-01 -1.0405800e+00 2.8057879e-02 1.8372647e-01
2.1398096e-01 7.2183454e-01 5.2995634e-01 -7.9206723e-01
-1.3492658e+00 -1.2176154e+00 5.8475000e-01 -4.3430719e-01
8.3251381e-01 -3.5816225e-01 -1.2785742e+00 5.5695659e-01
1.8950948e-01 -1.9931050e-01 1.1910865e+00 1.2008593e-01
-5.1245528e-01 2.3199239e-01 -1.1127561e+00 5.1537657e-01
8.4626162e-01 -3.9035690e-01 -1.5209644e-01 8.1264496e-01
7.0020771e-01 -1.7796023e-01 -1.0898000e+00 5.1327240e-01
2.7872607e-01 -4.8997831e-01 9.9816906e-01 -1.2607999e+00
5.2469474e-01 -3.9776558e-01 -9.1829754e-02 -1.5498173e+00
-1.8537821e-01 -3.4476009e-01 1.1069886e-01 8.9590394e-01
7.9733390e-01 -9.3118370e-01 9.3320101e-01 2.0200403e-01
1.6574387e-01 -8.3052909e-01 -6.9394284e-01 -9.3925840e-01
4.6607247e-01 3.3310167e-02 5.4881406e-01 1.3414499e+00
1.1467910e-01 6.2937617e-01 -2.4212487e-01 1.0586368e-02
2.7576423e-01 -1.4582798e-02 1.0151522e+00 -1.6366352e+00
-6.4673001e-01 -1.5151148e-01 -5.5443209e-01 -6.4422387e-01
2.6613313e-01 -1.2268909e+00 -3.5493606e-01 -1.5729110e+00
2.6912445e-01 -4.8969620e-01 -5.1370496e-01 6.8200713e-01
9.8455392e-02 3.6468986e-02 -1.0530329e-01 6.9823757e-02
-4.6712789e-01 1.4385572e-01 1.4289482e+00 -4.8724896e-01
-3.3594841e-01 -3.5524639e-01 -7.2414321e-01 6.7259961e-01
9.9205130e-01 -5.6925505e-01 -3.9710465e-01 -1.7323267e-01
2.4760433e-01 -4.1877031e-02 2.7248389e-01 -9.2256796e-01
2.0693414e-01 -3.5674313e-01 6.1603016e-01 -9.5359847e-02
1.4844529e-01 -6.8364859e-01 5.6704903e-01 8.8265687e-01
-7.7120796e-02 -1.8034181e-01 4.5852333e-01 8.0901229e-01
-1.2310238e-01 -8.9310549e-02 6.4799052e-01 -3.8448286e-01
-3.8655236e-01 2.3671360e-01 -4.1858727e-01 -2.3135963e-01
1.0488695e+00 -1.7068064e-01 -6.3404191e-01 -1.4268672e-01
-6.3089263e-01 -4.7198117e-02 6.7451346e-01 2.0480736e-01
5.1256841e-01 -8.5741234e-01 -5.3570563e-01 8.8081047e-02
2.2822642e-01 -2.2292075e-01 -2.1468043e-01 6.9626105e-01
-6.2563658e-01 4.5700350e-01 -4.0456229e-01 -6.4141327e-01
-1.1737595e+00 7.4872530e-01 3.2968101e-01 -6.1933237e-01
-1.3481648e-01 4.7783184e-01 3.6040431e-01 -7.1574986e-01
-2.5348628e-02 4.9725855e-03 -3.0903104e-01 -5.6791056e-02
2.9192328e-01 8.8141542e-03 2.2709544e-01 -1.5205133e-01
-2.5641474e-01 4.3190727e-01 -3.4270969e-01 3.4290802e-01
1.6435550e+00 5.6679606e-01 -2.3298457e-01 2.0293382e-01
1.1300735e+00 -2.3121743e-01 -8.8871402e-01 -9.2679091e-02
1.9097805e-01 -1.2021442e-01 -4.0213078e-01 -7.5804627e-01
-8.0940151e-01 8.3195913e-01 5.9728277e-01 2.1662664e-01
8.5974061e-01 7.4573927e-02 5.5209601e-01 8.8960135e-01
5.2174819e-01 -8.3936483e-01 3.4180427e-01 3.5615832e-01
6.0992450e-01 -1.2982891e+00 3.5592694e-02 -5.4173267e-01
4.3334413e-02 1.2204342e+00 5.9733742e-01 6.1107557e-02
4.8861742e-01 7.5771928e-02 -6.6788957e-02 -3.1966755e-01
-7.7894264e-01 -2.8927600e-02 -6.5284170e-02 9.5746326e-01
6.2276244e-01 -1.0285337e-01 -2.5833142e-01 1.0758457e-02
3.5131019e-01 3.4423491e-01 5.7431573e-01 1.0690010e+00
-2.7963153e-01 -1.9194698e+00 -2.0808446e-01 4.8338562e-01
-5.2486986e-01 -1.5137883e-01 -8.9229453e-01 8.6759692e-01
1.5565769e-01 9.4643801e-01 -2.2154097e-01 -3.7895203e-01
1.5418920e-01 1.3753536e-01 6.6609919e-01 -8.4721100e-01
-4.3610391e-01 -5.4308265e-03 2.3084220e-01 -2.8599024e-01
-6.3375193e-01 -2.6300448e-01 -1.1330082e+00 -2.5012568e-01
-5.4042375e-01 3.3513656e-01 7.4193752e-01 7.9078615e-01
8.0111998e-01 5.0050008e-01 4.5299798e-01 -7.9090440e-01
-1.0289788e-01 -7.0745438e-01 -4.7209662e-01 4.7607192e-01
3.6326014e-02 -6.2330782e-01 -1.5864113e-01 1.1687912e-01] | [5.348282337188721, 5.80991792678833] |
2a3cfc46-61da-426e-bd1e-cd369cf7d870 | interactive-segmentation-as-gaussion-process | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Zhou_Interactive_Segmentation_As_Gaussion_Process_Classification_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Zhou_Interactive_Segmentation_As_Gaussion_Process_Classification_CVPR_2023_paper.pdf | Interactive Segmentation As Gaussion Process Classification | Click-based interactive segmentation (IS) aims to extract the target objects under user interaction. For this task, most of the current deep learning (DL)-based methods mainly follow the general pipelines of semantic segmentation. Albeit achieving promising performance, they do not fully and explicitly utilize and propagate the click information, inevitably leading to unsatisfactory segmentation results, even at clicked points. Against this issue, in this paper, we propose to formulate the IS task as a Gaussian process (GP)-based pixel-wise binary classification model on each image. To solve this model, we utilize amortized variational inference to approximate the intractable GP posterior in a data-driven manner and then decouple the approximated GP posterior into double space forms for efficient sampling with linear complexity. Then, we correspondingly construct a GP classification framework, named GPCIS, which is integrated with the deep kernel learning mechanism for more flexibility. The main specificities of the proposed GPCIS lie in: 1) Under the explicit guidance of the derived GP posterior, the information contained in clicks can be finely propagated to the entire image and then boost the segmentation; 2) The accuracy of predictions at clicks has good theoretical support. These merits of GPCIS as well as its good generality and high efficiency are substantiated by comprehensive experiments on several benchmarks, as compared with representative methods both quantitatively and qualitatively. Codes will be released at https://github.com/zmhhmz/GPCIS_CVPR2023. | ['Yefeng Zheng', 'Deyu Meng', 'Yawen Huang', 'Yuexiang Li', 'Qian Zhao', 'Hong Wang', 'Minghao Zhou'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['interactive-segmentation'] | ['computer-vision'] | [ 7.33113810e-02 1.00951627e-01 -3.01960558e-01 -2.79496223e-01
-1.06409252e+00 -3.89625400e-01 4.19620216e-01 -2.13702649e-01
-2.80053914e-01 6.44287705e-01 -3.12733471e-01 -2.44548872e-01
-5.22034802e-02 -6.87865078e-01 -8.29299867e-01 -9.82396603e-01
5.51138341e-01 3.15835893e-01 5.33411264e-01 3.95489872e-01
2.43384838e-01 2.13761494e-01 -1.40627062e+00 -1.16076037e-01
1.34488225e+00 1.14330614e+00 3.06009799e-01 4.55340087e-01
-2.06007719e-01 2.59114534e-01 -1.55824304e-01 -7.28286028e-01
1.31732114e-02 -1.09858118e-01 -6.35394394e-01 3.23434323e-01
1.36903882e-01 -3.35802823e-01 -2.30205715e-01 1.34390903e+00
3.20078611e-01 8.82449225e-02 7.91414380e-01 -1.19383180e+00
-5.69020629e-01 2.67723382e-01 -9.75844145e-01 -1.84446126e-01
-3.51351947e-02 2.94129282e-01 1.14670122e+00 -1.04729497e+00
3.85183513e-01 1.06290448e+00 4.99799669e-01 3.20840329e-01
-1.40133274e+00 -4.53214377e-01 3.24085921e-01 9.54692811e-02
-1.54118204e+00 -4.38985936e-02 8.07391584e-01 -5.24493814e-01
2.70957083e-01 2.38901809e-01 8.06689918e-01 1.18098664e+00
-4.88811452e-03 1.44290924e+00 1.04880452e+00 -6.53350279e-02
3.92567933e-01 2.49953091e-01 3.41690272e-01 6.96968734e-01
1.24403395e-01 -2.15553492e-01 -3.72954220e-01 -8.22141841e-02
7.52958834e-01 -1.48657234e-02 -2.78249681e-01 -4.17059362e-01
-8.64667356e-01 8.48018646e-01 5.20386577e-01 2.42494065e-02
-4.55276996e-01 1.61192283e-01 1.71885744e-01 -5.56659997e-01
6.35131717e-01 -1.98445648e-01 -3.12911749e-01 -1.09221645e-01
-1.24643207e+00 5.43071151e-01 7.28556573e-01 1.08038259e+00
8.60271335e-01 -2.71151006e-01 -4.09942687e-01 9.17390049e-01
6.18728936e-01 4.50409144e-01 9.32205245e-02 -1.06153643e+00
3.01855505e-01 4.02824670e-01 1.92063004e-01 -9.68351126e-01
-6.98120072e-02 -6.68689728e-01 -6.80366039e-01 6.68627471e-02
3.79182875e-01 -2.01644465e-01 -8.94037306e-01 1.47678447e+00
5.88640392e-01 1.98855534e-01 -3.81938130e-01 8.79856050e-01
5.33363163e-01 7.30431020e-01 2.66060203e-01 -2.95657595e-03
1.47484851e+00 -1.15203714e+00 -5.93537867e-01 -1.40782595e-01
3.28728080e-01 -5.67889154e-01 1.31765187e+00 5.59390008e-01
-1.01081717e+00 -4.98071939e-01 -8.32316637e-01 -1.07908942e-01
-1.35042071e-01 4.34603602e-01 6.50252044e-01 5.43546379e-01
-9.29787219e-01 5.16353309e-01 -1.11177135e+00 -9.17978361e-02
8.58386815e-01 1.57599345e-01 2.66087055e-01 1.32215083e-01
-9.94592726e-01 2.90577859e-01 3.73295814e-01 3.11940432e-01
-8.29292953e-01 -7.10300982e-01 -5.35975397e-01 1.85278401e-01
5.51280677e-01 -7.36328959e-01 1.38656580e+00 -6.99171364e-01
-1.66061497e+00 6.77605450e-01 -3.58570307e-01 -4.18987304e-01
1.01788890e+00 -4.24457163e-01 1.44406199e-01 2.21643254e-01
2.20952615e-01 9.01417613e-01 8.91163290e-01 -1.38936019e+00
-8.19616437e-01 -3.34822327e-01 4.08825651e-02 2.18804777e-01
-2.29823083e-01 -1.94307014e-01 -1.17879951e+00 -5.32793999e-01
1.63995191e-01 -9.22807515e-01 -2.96749890e-01 1.91993400e-01
-8.09633195e-01 -3.52198929e-01 6.22241139e-01 -8.49694371e-01
1.25343466e+00 -2.12956786e+00 1.72177553e-02 -1.87413134e-02
3.29599053e-01 3.14936012e-01 3.26805532e-01 2.02070490e-01
3.79887551e-01 1.55107290e-01 -5.26570916e-01 -5.86629152e-01
2.88530201e-01 7.64607564e-02 -2.58745819e-01 4.09868062e-01
2.73736387e-01 1.15800261e+00 -7.59410858e-01 -8.01249504e-01
2.96126068e-01 5.62094688e-01 -5.74782252e-01 -1.20542459e-02
-5.55264533e-01 5.98199189e-01 -8.32256496e-01 7.16153204e-01
1.03634083e+00 -5.16011417e-01 -1.36200130e-01 -3.21030229e-01
-1.01144575e-01 -1.59115389e-01 -1.02188909e+00 1.62715650e+00
-9.15009901e-02 3.05810481e-01 2.61618614e-01 -1.02159810e+00
6.89455271e-01 4.39177714e-02 3.82876724e-01 -4.59369540e-01
6.26497045e-02 2.10445359e-01 -4.04691845e-01 -4.80871886e-01
2.69608587e-01 9.55493227e-02 1.62676975e-01 7.77619779e-02
6.49928525e-02 -1.39552772e-01 2.41152704e-01 9.70012173e-02
5.08839786e-01 5.97519100e-01 -5.52386530e-02 -1.18384629e-01
4.86707866e-01 2.56974753e-02 6.00687802e-01 8.22508395e-01
-4.06922907e-01 5.98431051e-01 7.11550891e-01 2.11810365e-01
-8.82482469e-01 -1.13204932e+00 -4.28963274e-01 8.87093067e-01
3.33056450e-01 -1.75900206e-01 -1.18717408e+00 -7.91016757e-01
-8.38714987e-02 8.55140150e-01 -4.01884168e-01 1.88572109e-01
-2.21785545e-01 -9.07732427e-01 3.09743673e-01 4.14296180e-01
8.42350841e-01 -7.96048820e-01 -2.48252034e-01 1.00889742e-01
-1.90621108e-01 -1.03734744e+00 -2.51660287e-01 -1.63314238e-01
-7.92104483e-01 -9.13262367e-01 -1.16799414e+00 -5.28834283e-01
5.28461576e-01 9.05660167e-02 5.85519314e-01 -2.60807782e-01
-2.14488655e-01 4.09836769e-01 -1.22184083e-01 -1.98484480e-01
5.92228211e-02 1.16003148e-01 -3.41569066e-01 3.07010233e-01
3.98135513e-01 -5.11642337e-01 -9.47893381e-01 2.45778412e-01
-7.60250092e-01 3.54123682e-01 6.53127909e-01 7.70446599e-01
9.86419082e-01 2.64605284e-01 2.93452322e-01 -8.97478342e-01
5.41613877e-01 -4.81583744e-01 -8.80565822e-01 1.37495324e-01
-5.74459374e-01 -1.97764158e-01 1.51522070e-01 -2.36503854e-01
-1.49398124e+00 5.11160009e-02 -4.61537659e-01 -4.77701247e-01
-2.61361927e-01 5.17310560e-01 -2.94247091e-01 -4.06569503e-02
1.88288420e-01 5.41986704e-01 -4.18654755e-02 -5.56844115e-01
5.48604786e-01 6.25750542e-01 4.12187308e-01 -7.85737634e-01
5.39980233e-01 5.82720697e-01 -2.09207699e-01 -7.22313404e-01
-1.08117461e+00 -3.64738464e-01 -5.35016477e-01 -2.65691638e-01
1.11626101e+00 -7.57074296e-01 -7.68462420e-01 6.72692478e-01
-9.97160733e-01 -4.25319046e-01 2.81913839e-02 3.88903946e-01
-7.48523653e-01 6.02803826e-01 -7.81662464e-01 -1.09997964e+00
-1.96505398e-01 -1.39336991e+00 1.17457986e+00 6.99607491e-01
1.68809295e-01 -1.04514515e+00 -2.52502203e-01 6.76812112e-01
1.97466567e-01 8.39090571e-02 7.13766932e-01 -5.65477788e-01
-9.74362731e-01 -3.67170423e-01 -6.52446210e-01 5.43621898e-01
-1.59318089e-01 2.41479829e-01 -1.06634891e+00 8.41354206e-02
7.22430795e-02 -9.03380737e-02 7.79983938e-01 8.36462796e-01
1.55481386e+00 -4.04450372e-02 -5.10047019e-01 6.34148955e-01
1.45767796e+00 -1.93232037e-02 4.41529244e-01 1.20380022e-01
7.61949778e-01 5.39021969e-01 8.28087628e-01 3.48848045e-01
3.60965997e-01 6.02115571e-01 4.59294766e-01 -1.40016628e-02
-4.99832779e-02 -4.79354769e-01 3.10939830e-02 6.13546729e-01
1.81725156e-02 -2.42711246e-01 -8.21871817e-01 3.65468383e-01
-2.05266404e+00 -4.44806218e-01 -4.08576548e-01 2.04548764e+00
8.65786016e-01 2.96528190e-01 -4.94201705e-02 -2.32990965e-01
8.18765223e-01 2.26088718e-01 -7.43755996e-01 1.09061465e-01
9.80027169e-02 9.59962234e-03 5.06105542e-01 4.45522696e-01
-1.31363988e+00 1.10875261e+00 5.44989967e+00 1.45946276e+00
-8.83084536e-01 2.60248482e-01 9.43182766e-01 1.13416635e-01
-1.66954085e-01 5.41162193e-02 -1.02166522e+00 8.22897017e-01
5.89199722e-01 -3.30756120e-02 2.67421216e-01 1.00263202e+00
4.74535108e-01 -3.13693076e-01 -7.06999004e-01 9.93822396e-01
-2.86062896e-01 -1.21369338e+00 -1.50620714e-01 1.37394637e-01
5.75387001e-01 -1.83824506e-02 2.59374201e-01 3.80498677e-01
1.71623051e-01 -7.28082895e-01 7.92773068e-01 6.72267079e-01
4.27940607e-01 -5.55047572e-01 4.31924939e-01 5.79795361e-01
-8.58912408e-01 9.02815387e-02 -3.60960335e-01 2.81644136e-01
4.29655254e-01 8.37626517e-01 -4.44568962e-01 5.82796276e-01
7.78661370e-01 6.95866942e-01 -4.20551777e-01 1.28990686e+00
-6.71378553e-01 9.48680162e-01 -3.69366705e-01 -4.39753197e-02
4.36377913e-01 -5.90445995e-01 6.22227252e-01 1.18000567e+00
1.93069041e-01 -2.44774953e-01 -4.13026102e-02 1.48738468e+00
1.55958280e-01 2.00580403e-01 -3.05532105e-02 -1.29729643e-01
3.44497979e-01 1.39372528e+00 -9.07273829e-01 -3.80934745e-01
-4.89625633e-01 1.10575736e+00 3.34581435e-01 8.48498702e-01
-1.12893236e+00 -2.67989367e-01 4.66118962e-01 -3.36528383e-02
6.12479746e-01 -2.02053130e-01 -5.52828133e-01 -1.16749728e+00
1.93334743e-01 -4.71656084e-01 6.18417636e-02 -7.94565260e-01
-1.26300335e+00 1.83508128e-01 6.59477338e-02 -9.31128502e-01
2.57047176e-01 -6.46117270e-01 -6.53591514e-01 1.03098941e+00
-1.44284749e+00 -1.25015473e+00 -3.23413372e-01 3.36895734e-01
6.55461609e-01 4.08743143e-01 3.56531918e-01 3.70376498e-01
-1.04933953e+00 4.61816996e-01 1.74126506e-01 -4.46786620e-02
3.69794637e-01 -1.28790390e+00 2.45059714e-01 7.60912180e-01
6.54415265e-02 5.59830189e-01 6.16945863e-01 -6.91828310e-01
-1.04499519e+00 -1.06971061e+00 3.92265409e-01 -2.82577276e-01
9.13221121e-01 -4.17770803e-01 -1.02738559e+00 5.19202113e-01
7.31598958e-03 -4.13480848e-02 5.09452343e-01 -1.27606884e-01
-2.25179456e-02 9.97790024e-02 -8.87011170e-01 7.59813011e-01
7.44665980e-01 -3.19031805e-01 -2.72983462e-01 4.53068256e-01
7.90719688e-01 -4.35493708e-01 -7.51276851e-01 3.21105868e-01
3.81785274e-01 -9.81097877e-01 8.99495482e-01 -2.45778412e-01
4.94260371e-01 -3.16641301e-01 -7.55576864e-02 -8.05635631e-01
-1.87414080e-01 -4.61347967e-01 -2.99954265e-01 1.47021449e+00
3.93139064e-01 -6.97091103e-01 1.12681854e+00 8.39829683e-01
-1.04720913e-01 -1.19537354e+00 -7.25510657e-01 -4.52664703e-01
9.47618932e-02 -6.39253795e-01 3.31231803e-01 3.81498903e-01
-4.49778438e-01 4.75071296e-02 -2.33058855e-01 3.29303235e-01
1.03848720e+00 1.21377036e-01 6.62003934e-01 -1.06620359e+00
-4.63834941e-01 -6.44552588e-01 -4.25434038e-02 -1.53572476e+00
4.82103825e-02 -6.03723705e-01 1.40779808e-01 -1.62580049e+00
3.40157062e-01 -5.31092644e-01 -7.77304247e-02 1.24038927e-01
-5.10605037e-01 4.74009626e-02 -5.58349453e-02 3.65375429e-01
-6.99709415e-01 7.84101903e-01 1.35110438e+00 2.71504894e-02
-6.26650676e-02 5.09836614e-01 -6.55919075e-01 8.93742263e-01
7.69120872e-01 -2.45073810e-01 -4.50052768e-01 -2.14782864e-01
3.57084386e-02 -9.05351788e-02 6.35369360e-01 -7.79155672e-01
2.51030654e-01 -1.81931704e-01 2.29871258e-01 -8.13175976e-01
4.60421443e-01 -5.93278885e-01 -1.12271179e-02 1.69676945e-01
-1.81741655e-01 -8.60814989e-01 -4.68084179e-02 9.51565862e-01
-2.78927922e-01 -4.12050277e-01 7.22063184e-01 4.44035186e-03
-6.07227981e-01 6.16881847e-01 -1.70535684e-01 -9.30504501e-02
1.12025142e+00 -2.15432525e-01 -2.31032781e-02 -2.15744421e-01
-9.76004899e-01 3.78603846e-01 3.60374421e-01 -2.05687666e-03
2.81669706e-01 -1.13495767e+00 -3.49550575e-01 -1.43224178e-02
-7.29119778e-02 3.80093247e-01 6.94171131e-01 1.34195995e+00
-4.13785219e-01 5.05749047e-01 3.50087851e-01 -9.16514039e-01
-7.52510309e-01 4.19999093e-01 3.04936141e-01 -1.51681483e-01
-6.98530376e-01 1.05148375e+00 6.12980127e-01 -2.31468558e-01
3.90826672e-01 -2.22651750e-01 -6.89993501e-02 -1.43319997e-03
3.13577920e-01 3.17677975e-01 -2.45937169e-01 -2.99605638e-01
-1.76227894e-02 4.98237461e-01 -1.56260058e-01 -1.10484496e-01
1.02137184e+00 -3.46188784e-01 1.27719864e-02 3.77739608e-01
9.92391646e-01 -2.27439180e-01 -1.95670056e+00 -2.37326562e-01
-3.64979059e-02 -4.59350169e-01 1.74956545e-01 -5.53566635e-01
-1.12339723e+00 1.18705153e+00 3.66950244e-01 2.42412567e-01
8.41848731e-01 1.64715603e-01 8.34526539e-01 -1.96131825e-01
2.16636330e-01 -1.15838850e+00 -3.94267924e-02 6.36286885e-02
5.37434280e-01 -1.16497087e+00 -2.55031168e-01 -8.31983387e-01
-8.54568481e-01 8.73662710e-01 5.29854655e-01 -1.55141458e-01
8.29283595e-01 -4.18156832e-02 -2.48783231e-01 -1.97230130e-01
-3.38314980e-01 -1.14120409e-01 2.00133413e-01 4.61821437e-01
1.31435290e-01 1.74310848e-01 -5.07787824e-01 1.02116954e+00
1.18684717e-01 2.51495808e-01 9.36910044e-04 7.23474801e-01
-4.34932679e-01 -8.44159245e-01 -5.65354861e-02 4.62338269e-01
-4.42100674e-01 -8.83780196e-02 5.39266951e-02 5.57348371e-01
1.47236073e-02 6.67973280e-01 -2.20845059e-01 -1.16666295e-02
3.60084102e-02 4.64495346e-02 1.62386611e-01 -4.51164961e-01
8.38501379e-02 5.21142840e-01 -2.48705938e-01 -7.47208655e-01
-1.51316434e-01 -7.07662463e-01 -1.19772780e+00 -3.26721184e-02
-4.43597257e-01 5.10613471e-02 7.68314481e-01 9.48678136e-01
3.90648574e-01 4.70535785e-01 1.96460724e-01 -1.03250396e+00
-6.91321731e-01 -7.73886085e-01 -7.30681479e-01 1.13207266e-01
-1.05364591e-01 -8.20026755e-01 -4.65815753e-01 -9.94734280e-03] | [9.482627868652344, -0.0005439297528937459] |
beafcf57-bda5-44cb-9a48-c4cdce746ba0 | personalized-image-aesthetics-assessment-via | null | null | https://ieeexplore.ieee.org/abstract/document/9115059 | https://ieeexplore.ieee.org/abstract/document/9115059 | Personalized Image Aesthetics Assessment via Meta-Learning With Bilevel Gradient Optimization | Typical image aesthetics assessment (IAA) is modeled for the generic aesthetics perceived by an ``average'' user. However, such generic aesthetics models neglect the fact that users' aesthetic preferences vary significantly depending on their unique preferences. Therefore, it is essential to tackle the issue for personalized IAA (PIAA). Since PIAA is a typical small sample learning (SSL) problem, existing PIAA models are usually built by fine-tuning the well-established generic IAA (GIAA) models, which are regarded as prior knowledge. Nevertheless, this kind of prior knowledge based on ``average aesthetics'' fails to incarnate the aesthetic diversity of different people. In order to learn the shared prior knowledge when different people judge aesthetics, that is, learn how people judge image aesthetics, we propose a PIAA method based on meta-learning with bilevel gradient optimization (BLG-PIAA), which is trained using individual aesthetic data directly and generalizes to unknown users quickly. The proposed approach consists of two phases: 1) meta-training and 2) meta-testing. In meta-training, the aesthetics assessment of each user is regarded as a task, and the training set of each task is divided into two sets: 1) support set and 2) query set. Unlike traditional methods that train a GIAA model based on average aesthetics, we train an aesthetic meta-learner model by bilevel gradient updating from the support set to the query set using many users' PIAA tasks. In meta-testing, the aesthetic meta-learner model is fine-tuned using a small amount of aesthetic data of a target user to obtain the PIAA model. The experimental results show that the proposed method outperforms the state-of-the-art PIAA metrics, and the learned prior model of BLG-PIAA can be quickly adapted to unseen PIAA tasks. | ['and Guangming Shi', 'Guiguang Ding', 'Sicheng Zhao', 'Jinjian Wu', 'Leida Li', 'Hancheng Zhu'] | 2020-06-11 | null | null | null | ieee-transactions-on-cybernetics-2020-6 | ['aesthetics-quality-assessment'] | ['computer-vision'] | [ 4.89903167e-02 -8.21486413e-02 1.86341032e-01 -4.43143278e-01
-9.14570332e-01 -1.51510477e-01 1.82508856e-01 -5.19511104e-02
-3.11192602e-01 1.41272262e-01 -3.57178599e-02 1.73108906e-01
-2.17674121e-01 -7.89896071e-01 -4.45186764e-01 -7.28937864e-01
2.42220059e-01 6.10424817e-01 -1.28040865e-01 -3.57827276e-01
2.56739736e-01 -1.55225903e-01 -1.77800310e+00 6.21801317e-01
1.54950404e+00 1.42041099e+00 2.01644585e-01 4.54172343e-01
-2.81441450e-01 -6.41270056e-02 -7.32545018e-01 -8.62269461e-01
3.12160343e-01 -7.44439244e-01 -7.06648767e-01 1.90866917e-01
3.08127135e-01 -8.21335390e-02 4.76985902e-01 1.04654944e+00
5.62553704e-01 3.06746066e-01 8.30316067e-01 -1.50208414e+00
-1.24082577e+00 1.87148735e-01 -5.86256623e-01 -4.09869790e-01
8.54919851e-02 4.33814019e-01 1.27640855e+00 -8.21502864e-01
1.50100544e-01 1.44699490e+00 6.20360136e-01 7.18567908e-01
-1.01055002e+00 -6.20956481e-01 4.50465232e-01 4.88663346e-01
-1.22521055e+00 5.02970479e-02 1.02102411e+00 -1.79881439e-01
3.97595525e-01 4.19160098e-01 1.10930097e+00 1.01022494e+00
-1.65454254e-01 1.18894482e+00 1.55143797e+00 -3.49878550e-01
6.56046569e-01 5.03882408e-01 -1.19690038e-01 5.05938709e-01
-5.67897297e-02 -1.25513479e-01 -5.76276362e-01 -7.75350034e-02
5.19801497e-01 -4.41682935e-01 1.55966273e-02 -4.27435696e-01
-8.15249264e-01 8.11973155e-01 5.70485473e-01 8.94122571e-03
-3.00635636e-01 -1.05748042e-01 5.75430989e-01 4.72901791e-01
4.54662919e-01 5.82559228e-01 -4.64377016e-01 -3.56731303e-02
-4.51918930e-01 -3.92219163e-02 5.89896262e-01 8.60207677e-01
9.46651042e-01 -2.26617947e-01 -3.18043351e-01 1.25111699e+00
5.76507390e-01 5.18719196e-01 8.65453959e-01 -8.19805741e-01
1.70199513e-01 8.39826405e-01 8.58186781e-02 -9.87567902e-01
-7.99912959e-02 -6.51671290e-01 -7.92819738e-01 5.14124870e-01
2.14750215e-01 5.88869490e-02 -8.13163877e-01 1.87213886e+00
4.32171464e-01 -2.48187751e-01 1.30647704e-01 9.35945153e-01
1.00828743e+00 8.15720022e-01 2.99001813e-01 -4.17302251e-01
1.30888534e+00 -1.32177424e+00 -3.99796784e-01 -3.41616243e-01
4.38427687e-01 -8.58171642e-01 1.96437454e+00 7.46699393e-01
-1.07405102e+00 -9.05043364e-01 -8.80281746e-01 1.46186262e-01
-5.04201293e-01 2.25381464e-01 7.04956710e-01 1.07487178e+00
-1.10428488e+00 4.47314620e-01 -7.09428862e-02 -3.91529024e-01
4.96990114e-01 2.38362253e-01 -4.96168956e-02 -7.01368451e-02
-1.09043217e+00 7.29554951e-01 4.31806743e-01 1.45108268e-01
-7.45066822e-01 -5.35237372e-01 -4.95230883e-01 1.24028943e-01
2.83472747e-01 -1.10742283e+00 1.02818131e+00 -1.71795011e+00
-1.69925439e+00 9.14356947e-01 8.31030160e-02 3.32039386e-01
4.83352900e-01 -8.30314830e-02 -4.78708595e-01 -1.84780300e-01
1.12023063e-01 5.90178907e-01 9.19578493e-01 -1.83822143e+00
-5.90330243e-01 -3.06526810e-01 2.53710806e-01 6.74667597e-01
-7.15900123e-01 -3.47091585e-01 -5.84268510e-01 -5.43824434e-01
-1.02240555e-01 -7.79728949e-01 -1.32518381e-01 1.29935965e-01
-1.77499130e-01 -4.31817740e-01 4.61524189e-01 -4.43371981e-01
1.30254221e+00 -2.08094978e+00 1.92056701e-01 5.64369738e-01
-1.19890518e-01 3.11158061e-01 -6.28421128e-01 1.83835343e-01
7.25293532e-02 1.79323286e-01 -3.77166271e-03 -5.50900340e-01
2.39268109e-01 1.66499019e-01 2.87494153e-01 -3.29105288e-01
-1.95362359e-01 8.53605211e-01 -1.18558347e+00 -7.08099067e-01
1.02754749e-01 2.87877709e-01 -6.14057660e-01 4.87244874e-01
-1.92098722e-01 4.40294623e-01 -5.67606151e-01 7.25639045e-01
6.80409253e-01 -3.91078800e-01 -5.45199811e-02 -5.95945835e-01
2.51722991e-01 -4.83737499e-01 -1.26493430e+00 1.77955699e+00
-1.01932001e+00 -9.09221619e-02 -3.01740468e-01 -6.22953296e-01
1.04028308e+00 6.06322736e-02 3.55751008e-01 -9.11900282e-01
2.94791818e-01 2.32725069e-01 -1.26433372e-01 -6.24714077e-01
2.36795649e-01 -2.30249569e-01 -1.73796430e-01 5.70723295e-01
8.72382745e-02 -8.45174864e-02 8.92541260e-02 -1.60762772e-01
5.86125910e-01 2.17687309e-01 2.78118581e-01 -7.47500360e-02
7.92836666e-01 -1.52845055e-01 4.90088701e-01 6.08580470e-01
-4.14398730e-01 6.36132002e-01 7.93053806e-02 -5.43459117e-01
-9.81158137e-01 -1.37061679e+00 1.39870420e-01 1.51576293e+00
4.23988968e-01 -3.96858335e-01 -9.07387435e-01 -7.94803858e-01
-4.86660570e-01 8.87891591e-01 -8.75339925e-01 -4.56469327e-01
-8.18913057e-02 -1.02257025e+00 -2.05488473e-01 2.20514312e-01
9.74212646e-01 -1.45678163e+00 -3.90803933e-01 -3.19972932e-02
-4.64151263e-01 -4.08062667e-01 -5.80162644e-01 -4.76831406e-01
-6.20517075e-01 -7.99969018e-01 -7.94855356e-01 -7.28409529e-01
7.70480573e-01 3.24608266e-01 1.27850842e+00 8.01229626e-02
-1.48402408e-01 7.75760889e-01 -6.64151251e-01 -4.13078964e-01
-2.94071615e-01 -1.84158683e-01 -1.14256747e-01 5.80861151e-01
3.73382151e-01 -7.64950037e-01 -1.05352175e+00 5.30842841e-01
-8.28784645e-01 2.18135819e-01 1.15694296e+00 8.29843163e-01
7.93819666e-01 1.72348723e-01 6.79337263e-01 -7.41847217e-01
1.12468231e+00 -3.02601904e-01 6.13537524e-03 8.39855194e-01
-1.01804996e+00 -2.12434426e-01 5.61571300e-01 -7.40774751e-01
-1.31971157e+00 -1.53304622e-01 -3.09498087e-02 -4.19833362e-01
-1.46120355e-01 4.92917269e-01 -6.74518049e-01 -2.49385327e-01
6.71181560e-01 2.47025043e-01 2.08805334e-02 -3.32286149e-01
5.08203387e-01 6.73126698e-01 3.16377461e-01 -9.12974417e-01
7.78089643e-01 1.96823984e-01 -2.45434135e-01 -3.95299107e-01
-1.25423706e+00 -4.78992611e-01 -4.22048718e-01 -6.94280207e-01
7.80654371e-01 -6.11157537e-01 -8.59473467e-01 5.39954960e-01
-8.46091390e-01 -2.66730279e-01 -5.13821125e-01 1.62383124e-01
-7.62398064e-01 4.33419615e-01 -1.62924364e-01 -9.13964391e-01
-6.59186482e-01 -1.18683743e+00 1.04700291e+00 5.70484757e-01
-1.71192303e-01 -1.10178280e+00 1.51489392e-01 7.54824758e-01
4.49985147e-01 1.81941345e-01 1.05972743e+00 -2.28337169e-01
-2.57495970e-01 -2.17347741e-01 -2.94005185e-01 5.62530994e-01
1.83350340e-01 -1.52770072e-01 -1.14578986e+00 -2.55102217e-01
5.64907230e-02 -8.38855445e-01 4.67225224e-01 2.30327234e-01
1.45887983e+00 -6.29092932e-01 2.33870968e-01 5.56066871e-01
1.50704789e+00 -9.39560402e-03 7.64854252e-01 7.16780961e-01
5.00315249e-01 5.57453632e-01 9.40709174e-01 4.58914816e-01
6.17972672e-01 4.21492785e-01 4.86320704e-01 -8.50039572e-02
-8.64440799e-02 -2.62482882e-01 3.46647203e-01 8.92177165e-01
-5.21207750e-01 -1.04098275e-01 -3.88006032e-01 3.91106606e-01
-1.99040461e+00 -8.18806708e-01 2.04897016e-01 2.44900560e+00
8.43485773e-01 1.31322471e-02 3.45295221e-01 -7.89608620e-03
6.25151813e-01 4.70446572e-02 -7.67718494e-01 -8.14374864e-01
-7.68873096e-02 -3.12189795e-02 -3.10950279e-01 2.90108085e-01
-9.11701381e-01 9.12396312e-01 5.20842648e+00 1.26539099e+00
-7.85589278e-01 2.53968388e-01 8.81218612e-01 5.39575629e-02
-6.62095308e-01 -1.44479707e-01 -1.62272975e-01 4.76997733e-01
6.13482177e-01 -2.09540814e-01 6.54834986e-01 1.09448647e+00
8.45505521e-02 2.05605328e-02 -6.90696061e-01 1.32796478e+00
2.11451873e-01 -4.44483757e-01 5.73787332e-01 -1.60849854e-01
8.92899096e-01 -7.82146811e-01 6.70587063e-01 7.03602254e-01
2.20785320e-01 -8.32139969e-01 4.73249346e-01 6.93436086e-01
8.06034803e-01 -8.65244687e-01 7.05650270e-01 2.32431740e-01
-1.09665334e+00 -1.97846398e-01 -6.16419494e-01 2.73995310e-01
3.30896378e-02 4.22516733e-01 -3.54308993e-01 4.68740523e-01
1.00055873e+00 2.76177496e-01 -1.01307237e+00 1.21623135e+00
-1.71282843e-01 2.71627843e-01 8.12197104e-02 -1.95435047e-01
2.90775478e-01 -4.89479035e-01 2.55176216e-01 8.37338269e-01
4.49933290e-01 8.10557306e-02 1.22276925e-01 7.63642073e-01
1.46581411e-01 7.92541564e-01 -2.46498846e-02 1.63631752e-01
1.08460085e-02 1.64937019e+00 -4.96721715e-01 -2.49879166e-01
-3.67255419e-01 1.41011226e+00 3.54270458e-01 3.11041355e-01
-5.99765241e-01 -1.27231911e-01 2.73516923e-01 -2.36224875e-01
-4.92922543e-03 3.67294908e-01 -5.31456172e-01 -1.07981122e+00
1.54504413e-03 -9.11643326e-01 6.34989560e-01 -1.15987265e+00
-1.84873426e+00 6.39866114e-01 -1.59642279e-01 -1.58947694e+00
1.47243887e-01 -5.40625811e-01 -9.92919743e-01 8.36177409e-01
-1.25502300e+00 -1.69558668e+00 -8.18994045e-01 5.64468622e-01
5.48891664e-01 -2.01696470e-01 8.83248806e-01 -1.16890021e-01
-2.60872304e-01 9.34263945e-01 -4.27755900e-02 -3.15054864e-01
7.82603204e-01 -1.40838063e+00 1.56727079e-02 2.68456310e-01
-1.92074403e-01 4.77085829e-01 7.20534563e-01 -4.73669738e-01
-8.49115312e-01 -8.93116474e-01 6.33689821e-01 -2.15250120e-01
3.27829868e-01 -4.26627025e-02 -9.35625374e-01 2.12305084e-01
2.88747430e-01 -2.99838573e-01 1.29300499e+00 6.54741466e-01
-4.53716695e-01 -5.64824879e-01 -1.31868744e+00 7.72070289e-01
1.15036142e+00 -8.59762728e-02 -4.22523588e-01 4.32471812e-01
3.85650277e-01 6.01737797e-02 -8.99190903e-01 3.44379067e-01
8.24250221e-01 -1.26972997e+00 1.03460145e+00 -5.40443242e-01
5.08390903e-01 -2.30498686e-01 -6.23364076e-02 -1.73581481e+00
-6.22121632e-01 -3.10631275e-01 1.94465853e-02 1.28632832e+00
2.66304016e-01 -3.89827430e-01 7.89573491e-01 7.64782786e-01
-1.90657243e-01 -1.05698609e+00 -7.36490548e-01 -7.57853270e-01
-1.37140667e-02 -3.75520706e-01 7.29911089e-01 7.89291859e-01
-9.40059870e-02 4.60601687e-01 -4.54504967e-01 9.94511992e-02
8.69223714e-01 3.05091321e-01 7.71390259e-01 -1.34074640e+00
-5.64943373e-01 -8.06834161e-01 -1.06225029e-01 -5.81241608e-01
-2.37177685e-01 -7.26836205e-01 -2.84885596e-02 -1.59290397e+00
4.02563632e-01 -6.77378178e-01 -6.49921656e-01 2.86569953e-01
-5.46220601e-01 3.18117142e-01 6.41839504e-02 2.37325430e-01
-1.00942957e+00 9.98111188e-01 1.73330426e+00 -2.71953613e-01
-3.67509037e-01 3.35864365e-01 -1.04990113e+00 7.50461757e-01
8.37734878e-01 3.28650065e-02 -9.18814182e-01 -2.61656374e-01
6.05476618e-01 -2.97443449e-01 2.68467993e-01 -1.07462907e+00
5.46755195e-02 -4.80824351e-01 6.22426927e-01 -2.64244199e-01
5.01950085e-01 -7.03385234e-01 -7.36624822e-02 1.42563939e-01
-2.34713674e-01 -9.18276049e-03 -4.97294702e-02 5.73376060e-01
3.39842252e-02 -3.56280684e-01 7.63785183e-01 -5.14064014e-01
-8.78429472e-01 5.35771370e-01 1.62844911e-01 1.22657090e-01
8.59213114e-01 -4.24956262e-01 -1.26627432e-02 -4.86931950e-01
-9.28851187e-01 2.63154089e-01 5.53776801e-01 4.92201507e-01
8.38116169e-01 -1.75171208e+00 -5.46066582e-01 -9.62373167e-02
5.87626696e-01 -1.91826537e-01 9.08889830e-01 4.68308002e-01
-7.96360150e-02 -3.94225180e-01 -4.44367975e-01 -5.60894668e-01
-1.13874865e+00 7.76942968e-01 3.71554345e-01 -2.60687679e-01
-3.97118703e-02 9.68956113e-01 5.60729682e-01 -6.12327993e-01
5.47803454e-02 3.35050970e-01 -4.13445890e-01 8.71282890e-02
5.11245489e-01 3.48208696e-01 -2.87306011e-01 -6.07320964e-01
9.76985246e-02 8.91691625e-01 -4.15379144e-02 -1.36196762e-01
1.20866001e+00 -4.02614921e-01 -5.34835085e-02 6.61645412e-01
1.08277607e+00 -2.04100579e-01 -1.05932772e+00 -2.37529635e-01
-4.48139459e-01 -7.51470923e-01 -2.06904471e-01 -1.26855326e+00
-1.07133675e+00 9.14114952e-01 1.01343131e+00 -7.89043307e-02
1.66397905e+00 -6.35612011e-02 7.15386569e-01 2.83493370e-01
3.93426508e-01 -1.60528505e+00 7.56780207e-01 -3.93689685e-02
1.16790688e+00 -1.45819557e+00 -2.07380757e-01 -1.12232938e-01
-1.14838099e+00 1.00894701e+00 1.11809003e+00 9.65941995e-02
5.30063808e-01 -7.81745970e-01 4.40547228e-01 -8.19059983e-02
-5.00230908e-01 -2.57060587e-01 7.29868650e-01 7.72294760e-01
1.18636675e-01 1.54085368e-01 -3.34986836e-01 1.02682757e+00
-3.83564800e-01 -1.39919907e-01 1.38041899e-01 4.35378641e-01
-5.54813266e-01 -1.15162992e+00 -1.98841453e-01 3.07438850e-01
2.54522890e-01 2.90806070e-02 -3.71780246e-01 5.37031353e-01
5.08569479e-01 8.52277696e-01 -3.22191536e-01 -7.08909035e-01
5.15517890e-01 -3.03165931e-02 3.65337819e-01 -4.03063267e-01
-6.65511370e-01 -1.29885515e-02 -1.36372507e-01 -5.71441829e-01
-3.78139198e-01 -5.66137671e-01 -7.13724911e-01 -3.04176301e-01
-1.19035058e-01 2.76822358e-01 4.94389355e-01 9.53466475e-01
1.06310882e-01 3.25244814e-01 9.16427135e-01 -7.59552300e-01
-3.86299103e-01 -8.64216983e-01 -6.28499210e-01 9.36040163e-01
-2.02902064e-01 -5.48865080e-01 -3.98956716e-01 -1.71428666e-01] | [11.52184772491455, -1.055650234222412] |
2b4c78a9-eb83-44f1-878c-52804edf5044 | hyperspectral-pansharpening-a-review | 1504.04531 | null | http://arxiv.org/abs/1504.04531v1 | http://arxiv.org/pdf/1504.04531v1.pdf | Hyperspectral pansharpening: a review | Pansharpening aims at fusing a panchromatic image with a multispectral one,
to generate an image with the high spatial resolution of the former and the
high spectral resolution of the latter. In the last decade, many algorithms
have been presented in the literature for pansharpening using multispectral
data. With the increasing availability of hyperspectral systems, these methods
are now being adapted to hyperspectral images. In this work, we compare new
pansharpening techniques designed for hyperspectral data with some of the state
of the art methods for multispectral pansharpening, which have been adapted for
hyperspectral data. Eleven methods from different classes (component
substitution, multiresolution analysis, hybrid, Bayesian and matrix
factorization) are analyzed. These methods are applied to three datasets and
their effectiveness and robustness are evaluated with widely used performance
indicators. In addition, all the pansharpening techniques considered in this
paper have been implemented in a MATLAB toolbox that is made available to the
community. | ['Jean-Yves Tourneret', 'Miguel Simões', 'José M. Bioucas-Dias', 'Xavier Briottet', 'Qi Wei', 'Naoto Yokoya', 'Giorgio A. Licciardi', 'Nicolas Dobigeon', 'Laetitia Loncan', 'Sophie Fabre', 'Luis B. Almeida', 'Jocelyn Chanussot', 'Gemine Vivone', 'Wenzhi Liao', 'Miguel A. Veganzones'] | 2015-04-17 | null | null | null | null | ['pansharpening'] | ['computer-vision'] | [ 9.75735247e-01 -6.13607109e-01 5.09081185e-02 3.02702207e-02
-5.97721159e-01 -6.32513404e-01 5.31536162e-01 -2.06005171e-01
-2.95664340e-01 8.42930615e-01 -2.72767730e-02 -5.46343885e-02
-9.87683773e-01 -1.04649723e+00 1.29793361e-02 -1.20330536e+00
2.77640760e-01 2.30905071e-01 4.28838916e-02 -4.02072757e-01
1.40032217e-01 8.10057402e-01 -1.99453366e+00 1.87245667e-01
1.20000219e+00 7.47599244e-01 3.01232666e-01 7.23775029e-01
3.63836437e-01 2.14081071e-02 -1.82271495e-01 -2.04426330e-02
6.38647556e-01 -3.45975578e-01 -4.13228005e-01 4.35225308e-01
4.96424705e-01 4.02478501e-02 -5.70050143e-02 1.47916102e+00
3.00037742e-01 1.93609893e-01 5.62791109e-01 -8.60121548e-01
-4.47760940e-01 5.38887084e-01 -1.10097790e+00 5.19500747e-02
5.24460077e-02 -1.13106921e-01 8.43259871e-01 -5.47668457e-01
1.74548760e-01 8.15909863e-01 4.88796234e-01 -1.80057868e-01
-1.23900163e+00 -5.14748216e-01 -4.71209794e-01 2.82262415e-01
-1.16205704e+00 5.45394607e-02 9.70871568e-01 -4.89889264e-01
5.68298161e-01 7.43630230e-01 8.81274283e-01 4.11394358e-01
4.22868431e-02 3.52718860e-01 1.73281598e+00 -6.49098217e-01
5.36459237e-02 -8.28141570e-02 3.54754031e-01 8.18025973e-03
5.10054648e-01 6.17597818e-01 -1.35355160e-01 -2.72385508e-01
2.93335557e-01 2.25992948e-02 -5.25102615e-01 -4.14036602e-01
-8.27626288e-01 8.44210684e-01 4.61522341e-01 7.11081862e-01
-8.13029885e-01 -6.38073444e-01 1.02793105e-01 1.19858123e-01
4.77040857e-01 4.83373493e-01 -3.53304178e-01 4.19879079e-01
-1.24527800e+00 2.60704815e-01 3.96237940e-01 3.78128290e-01
1.06247926e+00 2.01669685e-03 2.02619776e-01 1.09715438e+00
3.00218731e-01 9.80231524e-01 6.33171320e-01 -5.37796438e-01
2.73970049e-02 5.71552038e-01 2.41146591e-02 -9.83435273e-01
-4.20501053e-01 -2.51125723e-01 -9.59239483e-01 4.11224455e-01
-2.64740586e-01 -2.12019920e-01 -9.34978187e-01 9.56306458e-01
2.87046999e-01 3.30305919e-02 4.38798130e-01 7.89479077e-01
5.46902597e-01 1.22393179e+00 -1.25138402e-01 -5.63804686e-01
1.14129758e+00 -8.48610997e-01 -5.58223903e-01 -2.17387721e-01
-1.41465992e-01 -1.29632449e+00 3.77887815e-01 1.05518079e+00
-5.13849199e-01 -5.53994417e-01 -1.19196308e+00 5.37865460e-01
-6.52398050e-01 3.50066364e-01 6.06215239e-01 1.29284942e+00
-8.01239491e-01 6.32084370e-01 -5.17228007e-01 -4.47698683e-01
-6.19850010e-02 1.51424989e-01 -2.23676115e-01 -1.36724174e-01
-1.03371000e+00 9.45060492e-01 1.07324207e+00 1.41202092e-01
-2.81925738e-01 -5.06292045e-01 -4.53797579e-01 -1.96952939e-01
3.00456226e-01 -1.98232993e-01 8.94078076e-01 -1.39635956e+00
-1.64727998e+00 6.20034277e-01 3.32114875e-01 -1.92265019e-01
2.19035089e-01 -2.78802484e-01 -8.95929277e-01 4.04474288e-01
-1.05045833e-01 1.14562608e-01 1.02880096e+00 -1.37888277e+00
-6.38054729e-01 -7.03359127e-01 -2.92201906e-01 2.17314184e-01
-1.90912291e-01 4.96778786e-02 1.87376767e-01 -7.43067026e-01
4.16112930e-01 -9.89059448e-01 -1.52397156e-01 -5.08090794e-01
-1.60589576e-01 3.72557521e-01 1.14217520e+00 -9.75367129e-01
6.87868655e-01 -2.17480159e+00 5.47297239e-01 6.92257106e-01
-5.54586172e-01 8.08407307e-01 -1.65917665e-01 6.38372660e-01
-5.96275687e-01 -2.06743181e-01 -9.10077691e-01 1.82840884e-01
-4.25545514e-01 2.47358441e-01 -3.83041680e-01 5.80354750e-01
-3.12026531e-01 1.87767580e-01 -6.33663118e-01 2.17970069e-02
8.21491778e-01 6.90725565e-01 4.02565673e-02 -5.30724525e-02
-1.09810278e-01 5.25460243e-02 -2.39693448e-01 7.51593649e-01
1.38102412e+00 4.51509863e-01 1.47418827e-01 -5.11753500e-01
-6.35494828e-01 -7.86484540e-01 -1.39649951e+00 1.30767810e+00
-1.37687907e-01 1.86133921e-01 3.59422237e-01 -9.33812916e-01
9.75027919e-01 5.31750262e-01 7.15651095e-01 -3.41315091e-01
-4.77838963e-02 4.53497857e-01 -1.18443869e-01 -2.63838470e-01
8.79232347e-01 -4.69305277e-01 7.98980415e-01 4.07465965e-01
-6.33255839e-02 -1.00802898e+00 5.66495001e-01 -3.98603946e-01
-2.80334242e-02 2.62436837e-01 5.27186215e-01 -1.81329072e-01
8.77841651e-01 2.82921076e-01 -1.90214701e-02 3.13035011e-01
1.50132880e-01 3.82238090e-01 -1.09780371e-01 1.98350698e-02
-1.15446997e+00 -7.59742737e-01 -2.88077235e-01 7.90961921e-01
-1.41027510e-01 4.10704166e-02 -6.77473307e-01 2.35142820e-02
-1.60915598e-01 6.56548619e-01 -2.85773814e-01 1.24908559e-01
8.84816051e-02 -1.70045233e+00 1.70540467e-01 -1.49836838e-01
7.91989684e-01 -8.46032739e-01 -6.80949271e-01 2.96089232e-01
-3.96027155e-02 -8.25224936e-01 3.88172954e-01 -3.44899036e-02
-1.15560305e+00 -1.32782865e+00 -8.51954043e-01 -9.96436998e-02
1.64979532e-01 8.79929304e-01 5.44822991e-01 -4.26833302e-01
-4.22508925e-01 4.38379049e-01 -9.99553680e-01 -5.46670675e-01
-4.98712838e-01 -2.01265529e-01 -4.04519349e-01 4.63361651e-01
1.09430999e-01 -7.25091577e-01 -1.37250274e-01 2.19829470e-01
-1.67714417e+00 1.16990237e-02 1.02535963e+00 5.41285157e-01
7.71151602e-01 8.84709120e-01 5.14603518e-02 -7.95252979e-01
3.50897610e-01 -2.68536776e-01 -1.07060993e+00 2.62435049e-01
-8.04847121e-01 -4.56382304e-01 3.20577592e-01 -1.64703161e-01
-1.38248396e+00 2.98147380e-01 -2.03172993e-02 2.10762117e-02
-3.54984134e-01 9.47191715e-01 -9.42754373e-02 -5.29059350e-01
9.75259423e-01 4.05144602e-01 6.86978083e-03 -6.18571162e-01
5.08135855e-01 8.03394794e-01 6.05111539e-01 -3.79746705e-01
9.48108077e-01 6.77339315e-01 1.68695435e-01 -1.50517058e+00
-7.71353483e-01 -8.50207925e-01 -6.37476802e-01 -3.56447458e-01
7.04698324e-01 -6.55520499e-01 2.13876396e-01 1.08605003e+00
-6.39199495e-01 8.48046411e-03 -1.00388139e-01 5.98444700e-01
-4.52934027e-01 8.33928227e-01 1.05524585e-02 -8.32445979e-01
-4.65542912e-01 -1.03727508e+00 6.98894978e-01 4.86917466e-01
5.40466607e-01 -6.49784625e-01 5.09055197e-01 5.27158320e-01
4.45842296e-01 4.06217963e-01 6.77959144e-01 -1.24764033e-01
-1.31543279e-01 -2.24075437e-01 -4.30236429e-01 7.14565217e-01
5.12472391e-01 5.41075468e-01 -1.10906041e+00 -3.12438279e-01
1.73473224e-01 -3.01853418e-02 1.01653278e+00 6.84144616e-01
8.33865225e-01 1.19332073e-03 -1.41420677e-01 5.76424301e-01
1.84567761e+00 3.69275600e-01 8.65345776e-01 5.18582523e-01
3.34015578e-01 6.18657291e-01 7.82852530e-01 4.48288918e-01
-5.50087273e-01 3.88599753e-01 7.48928308e-01 -5.74378595e-02
1.91488832e-01 3.85084420e-01 8.30949247e-02 3.18599254e-01
-7.28765130e-01 -1.49370104e-01 -8.47069681e-01 3.06100070e-01
-1.73061633e+00 -1.18905389e+00 -5.04424214e-01 2.20389199e+00
5.44569254e-01 -3.65724534e-01 -4.05521579e-02 7.18617082e-01
1.02153790e+00 5.89759052e-01 -2.24905431e-01 -3.12206894e-02
-5.99945426e-01 6.20965242e-01 7.37907112e-01 5.87125063e-01
-1.46649539e+00 5.80181897e-01 5.74448347e+00 5.95207572e-01
-1.27068889e+00 1.31365821e-01 1.27595142e-01 5.05178988e-01
-1.19392298e-01 5.78191131e-02 -1.38113230e-01 -1.52317062e-01
7.95173109e-01 -1.80073604e-01 6.68177128e-01 5.70237696e-01
1.85181811e-01 -6.07851624e-01 -6.55258670e-02 1.13615513e+00
2.73260057e-01 -7.36094892e-01 1.78616986e-01 -8.44482034e-02
1.17454755e+00 2.65391320e-01 2.06993103e-01 -3.89545202e-01
4.94479993e-03 -7.31194615e-01 4.05397952e-01 6.12489462e-01
2.95914471e-01 -9.85986888e-01 8.66253674e-01 1.30679667e-01
-9.00638640e-01 -3.01630974e-01 -4.27393407e-01 1.07281424e-01
1.52581364e-01 1.16628349e+00 -2.02233568e-01 1.57736111e+00
5.74474692e-01 8.92104924e-01 -5.59224427e-01 1.56451726e+00
-4.24444020e-01 4.39394683e-01 -4.56365347e-01 4.00626421e-01
2.90051639e-01 -9.35566843e-01 7.47191012e-01 9.73454356e-01
7.34098136e-01 3.23705673e-01 -3.76670212e-02 5.73276281e-01
8.24945450e-01 5.29506445e-01 -4.15006220e-01 -5.60621440e-01
-6.68407083e-02 1.50100708e+00 -6.39050126e-01 -2.66124189e-01
-3.40546429e-01 6.59491181e-01 -7.76403487e-01 2.99367189e-01
-4.81116474e-01 -2.85469651e-01 3.78628463e-01 -3.53549004e-01
1.70051694e-01 -1.36422738e-01 -8.89655203e-02 -9.85813320e-01
-5.01020253e-01 -1.11119473e+00 6.70334637e-01 -1.30873144e+00
-1.08188033e+00 5.10777891e-01 4.53763664e-01 -1.00996768e+00
1.34301886e-01 -1.02076101e+00 -1.35525003e-01 1.13172376e+00
-1.71092892e+00 -1.23155832e+00 -5.75052440e-01 5.72385967e-01
2.04483807e-01 -1.45429671e-01 9.90428150e-01 1.85409591e-01
-4.00773466e-01 -6.78219140e-01 4.87438560e-01 -5.58910131e-01
5.77243209e-01 -9.27321315e-01 -3.69210601e-01 1.35299408e+00
4.37462181e-02 8.40185583e-02 9.41728830e-01 -5.83899856e-01
-1.08935833e+00 -9.49338973e-01 2.98578709e-01 4.29594904e-01
7.85279870e-01 7.42723286e-01 -8.12319517e-01 4.59416628e-01
3.08806777e-01 -4.30144876e-01 7.26620317e-01 -3.77055258e-01
-4.21506107e-01 -3.79681706e-01 -1.13795376e+00 1.05926111e-01
4.74048331e-02 -3.09735328e-01 -7.29589701e-01 3.62663984e-01
-1.40140146e-01 -1.37078896e-01 -8.20865810e-01 6.86786294e-01
4.91454363e-01 -1.42515242e+00 1.02270460e+00 1.57390773e-01
2.79003441e-01 -5.85542083e-01 -4.11405951e-01 -1.59486890e+00
-4.16711807e-01 -4.74098504e-01 5.78277886e-01 7.48505652e-01
2.47184277e-01 -7.58285046e-01 2.97091752e-01 3.55253518e-02
-1.46266108e-03 3.09680790e-01 -6.48587704e-01 -5.82051098e-01
-1.35232240e-01 -7.10917190e-02 6.88350081e-01 8.87917936e-01
-3.79783064e-01 -6.27544373e-02 -3.56943727e-01 7.59021163e-01
7.92530537e-01 5.39334774e-01 4.28764910e-01 -1.72696626e+00
-4.02924985e-01 -6.79017305e-01 -2.71015286e-01 -1.43219247e-01
-5.33734411e-02 -6.56801045e-01 -2.33569235e-01 -1.49290252e+00
2.85881549e-01 -2.84224935e-02 -2.69452333e-01 4.55399930e-01
-7.83534646e-02 4.21487838e-01 1.87060699e-01 1.01090319e-01
6.50494456e-01 2.24177107e-01 1.00425887e+00 -4.58248287e-01
-4.09645170e-01 1.55052409e-01 -5.05662978e-01 6.79299593e-01
1.12431610e+00 -2.06945881e-01 -3.92024219e-01 -1.04981460e-01
3.60836536e-01 -1.64679870e-01 4.38039631e-01 -1.38783383e+00
-1.99105144e-01 -3.21621358e-01 8.39513075e-03 -7.54936516e-01
5.64953208e-01 -1.19391394e+00 9.32718813e-01 6.12682283e-01
3.43464851e-01 -4.27896738e-01 3.53487462e-01 2.91212231e-01
-4.41665053e-01 -6.46543801e-01 1.38180983e+00 -1.78600460e-01
-9.77254868e-01 1.89252704e-01 -3.31067652e-01 -8.19141328e-01
1.14033628e+00 -2.04514995e-01 -3.22026163e-01 -1.97947696e-01
-7.43419468e-01 -5.52122712e-01 3.03683043e-01 -2.24576704e-02
4.08180594e-01 -8.08127463e-01 -8.37566197e-01 1.03358023e-01
8.13178420e-02 -4.97065961e-01 7.10143745e-01 7.69846559e-01
-8.19682181e-01 3.98181200e-01 -8.60836923e-01 -3.01450282e-01
-1.53261399e+00 7.21607506e-01 5.65450430e-01 -8.19452479e-03
-8.80521238e-02 3.36677641e-01 -3.19533497e-01 -4.03903514e-01
-6.53271019e-01 -1.65494367e-01 -6.64345682e-01 4.69768286e-01
6.12841547e-01 7.40235984e-01 3.24664623e-01 -1.13089442e+00
1.29812816e-02 1.00727296e+00 5.40176809e-01 -2.88710237e-01
1.65418732e+00 1.15294091e-01 -9.13134098e-01 2.47077256e-01
5.44982731e-01 2.32105374e-01 -7.93528914e-01 -2.88485810e-02
-3.11471641e-01 -6.12169445e-01 5.99714100e-01 -9.77699101e-01
-1.08234203e+00 7.61765420e-01 7.19502568e-01 7.11761653e-01
1.69212496e+00 -6.74273312e-01 7.62514323e-02 2.88585246e-01
2.48210147e-01 -1.07592499e+00 -5.79919279e-01 2.57205695e-01
9.43304598e-01 -9.47935939e-01 5.64917088e-01 -8.80287528e-01
-4.14940268e-01 1.54994249e+00 -1.63699463e-02 1.10875778e-01
7.31743157e-01 -1.15032464e-01 5.96516430e-02 -2.29940135e-02
1.44464821e-01 -4.12067384e-01 2.87795246e-01 7.27437735e-01
2.70067960e-01 3.89148563e-01 -6.63277447e-01 -8.59676301e-02
-1.08847320e-01 1.39852658e-01 6.72818422e-01 6.12287283e-01
-7.76968837e-01 -1.30122101e+00 -1.16384554e+00 3.48664641e-01
-2.30703980e-01 -6.83438182e-02 -4.49504763e-01 7.41824865e-01
1.64641127e-01 1.19484746e+00 -5.22728562e-01 -4.56158035e-02
3.73454660e-01 1.27756879e-01 4.94875193e-01 -3.80205691e-01
-2.32908070e-01 3.58432919e-01 2.56462283e-02 -1.05903514e-01
-1.43193924e+00 -9.50883269e-01 -4.65133250e-01 -3.99307795e-02
-5.63627481e-01 3.65604639e-01 8.87172163e-01 6.81392372e-01
-2.59043604e-01 2.82276511e-01 4.84650195e-01 -1.19374669e+00
-3.89021873e-01 -1.11131358e+00 -1.32686567e+00 -8.23686868e-02
1.38028055e-01 -6.19908452e-01 -2.87886709e-01 2.45167129e-02] | [10.08407974243164, -2.102548837661743] |
9190845b-7aaa-46b7-a129-9af5c997fceb | coarse-to-fine-recursive-speech-separation | 2203.16054 | null | https://arxiv.org/abs/2203.16054v1 | https://arxiv.org/pdf/2203.16054v1.pdf | Coarse-to-Fine Recursive Speech Separation for Unknown Number of Speakers | The vast majority of speech separation methods assume that the number of speakers is known in advance, hence they are specific to the number of speakers. By contrast, a more realistic and challenging task is to separate a mixture in which the number of speakers is unknown. This paper formulates the speech separation with the unknown number of speakers as a multi-pass source extraction problem and proposes a coarse-to-fine recursive speech separation method. This method comprises two stages, namely, recursive cue extraction and target speaker extraction. The recursive cue extraction stage determines how many computational iterations need to be performed and outputs a coarse cue speech by monitoring statistics in the mixture. As the number of recursive iterations increases, the accumulation of distortion eventually comes into the extracted speech and reminder. Therefore, in the second stage, we use a target speaker extraction network to extract a fine speech based on the coarse target cue and the original distortionless mixture. Experiments show that the proposed method archived state-of-the-art performance on the WSJ0 dataset with a different number of speakers. Furthermore, it generalizes well to an unseen large number of speakers. | ['Xiangdong Su', 'Xiang Hao', 'Zhenhao Jin'] | 2022-03-30 | null | null | null | null | ['target-speaker-extraction'] | ['audio'] | [ 2.40406901e-01 -1.46949738e-01 1.74417660e-01 -1.53180003e-01
-1.03845227e+00 -5.66849411e-01 6.43547833e-01 -1.15442865e-01
-2.72691786e-01 5.83523452e-01 2.86813289e-01 -2.73104191e-01
1.14761107e-02 -2.48721421e-01 -2.98596710e-01 -9.88563657e-01
1.59047306e-01 4.70706999e-01 4.01658714e-01 -1.59084901e-01
-9.72541142e-03 4.66272265e-01 -1.54796898e+00 -4.69738469e-02
8.37443531e-01 8.55434775e-01 3.10087383e-01 9.74989414e-01
-2.32135832e-01 4.62530345e-01 -1.13423681e+00 -1.17693484e-01
2.66423911e-01 -6.40075743e-01 -4.12292272e-01 5.26831508e-01
1.46934465e-01 -9.70623791e-02 -8.33265334e-02 1.30062199e+00
7.57549882e-01 3.16947877e-01 7.16667593e-01 -1.03448641e+00
-3.11593443e-01 9.12651300e-01 -5.00171840e-01 5.84268987e-01
4.48762476e-02 -1.48577601e-01 6.44811153e-01 -1.10788357e+00
-7.75973499e-02 1.30933881e+00 1.55379549e-01 5.54038048e-01
-9.92577076e-01 -9.63244915e-01 4.26792204e-01 -1.70540050e-01
-1.36323309e+00 -1.12055457e+00 9.63741064e-01 -4.66177344e-01
5.40639937e-01 3.85497749e-01 2.90740669e-01 8.24364901e-01
-4.34917837e-01 7.38007665e-01 8.47406805e-01 -5.15908241e-01
4.53044236e-01 2.24193960e-01 3.70715499e-01 2.48826846e-01
1.53506294e-01 2.33723521e-01 -5.64101815e-01 -1.06908884e-02
4.75685358e-01 -2.99030364e-01 -5.17564416e-01 2.39200160e-01
-9.79967237e-01 5.75285733e-01 -1.98460430e-01 5.46493173e-01
-3.45855594e-01 -5.04532933e-01 5.16669750e-02 1.42702967e-01
5.31976223e-01 7.33722076e-02 -2.56247908e-01 3.01112328e-02
-1.51211417e+00 4.34469096e-02 9.90961015e-01 9.79225576e-01
4.78732288e-01 3.96253109e-01 -2.70687342e-02 1.10833764e+00
3.50475580e-01 6.37210906e-01 6.58626437e-01 -5.49620926e-01
7.13456750e-01 1.05595872e-01 1.47170752e-01 -7.55123734e-01
-4.72036824e-02 -8.81903291e-01 -9.42925215e-01 1.86480761e-01
4.36892569e-01 -3.23907405e-01 -1.08109820e+00 1.70257223e+00
4.41073567e-01 4.38078284e-01 3.01518530e-01 8.84709001e-01
7.27990806e-01 9.25241530e-01 -4.03040022e-01 -6.78098440e-01
1.37608016e+00 -1.23170626e+00 -1.05761278e+00 -5.96341312e-01
-3.28151822e-01 -1.18948519e+00 5.98091662e-01 6.02338672e-01
-1.20244777e+00 -7.17824459e-01 -1.25822532e+00 4.16094601e-01
-2.68623680e-01 4.85875458e-01 7.87467659e-02 9.19855475e-01
-9.03441429e-01 9.55862552e-02 -6.68911040e-01 -5.67762554e-02
4.01281305e-02 3.00367951e-01 6.07916191e-02 1.53429434e-01
-1.04017627e+00 5.57128608e-01 4.17066336e-01 3.35979164e-01
-1.09633172e+00 -3.18363696e-01 -6.83364630e-01 2.89386153e-01
5.91962755e-01 -4.23043311e-01 1.38142085e+00 -9.85932469e-01
-1.73596406e+00 3.51512313e-01 -5.66461444e-01 -1.94739446e-01
5.20674586e-01 -2.34280780e-01 -9.09228802e-01 3.41123715e-02
9.69862193e-02 9.90289375e-02 1.48471475e+00 -1.41870892e+00
-9.77805197e-01 -2.81312644e-01 -3.39790523e-01 4.93183881e-01
-1.87957704e-01 1.38028786e-01 -7.66364276e-01 -8.14303756e-01
2.44040772e-01 -6.85331821e-01 2.03391865e-01 -6.19544566e-01
-6.57979608e-01 -6.74751401e-02 7.81440556e-01 -8.63686204e-01
1.31217611e+00 -2.52835345e+00 1.83490098e-01 1.36203036e-01
2.47396022e-01 3.52263808e-01 -1.21805176e-01 3.57649922e-02
-1.75425574e-01 -1.37960687e-01 -2.85153300e-01 -6.31566644e-01
5.21512106e-02 -2.14008093e-01 -3.79691750e-01 3.25782448e-01
7.46252313e-02 1.61894783e-01 -8.16891730e-01 -5.37523210e-01
6.91679791e-02 4.74112868e-01 -3.05215091e-01 5.01347244e-01
2.60999918e-01 3.56173396e-01 -1.92907363e-01 4.49470490e-01
9.67672646e-01 1.10042915e-01 -1.46360904e-01 -1.97099745e-02
-2.87294745e-01 3.25514615e-01 -1.79366660e+00 1.31126535e+00
-2.27127612e-01 5.99362135e-01 7.61724174e-01 -7.03312099e-01
8.70582819e-01 6.50001407e-01 2.24720225e-01 -4.71617952e-02
1.92323625e-01 4.39771622e-01 1.94693118e-01 -2.48429745e-01
3.96458417e-01 -2.97434866e-01 1.14828162e-01 1.82412371e-01
8.41703713e-02 -1.66824907e-01 3.49803656e-01 3.30113694e-02
7.98233211e-01 -4.67289776e-01 3.16468060e-01 -1.96479782e-01
8.20950687e-01 -5.47307312e-01 6.90650702e-01 5.62348545e-01
-3.06663901e-01 4.71622467e-01 7.24362209e-03 2.82240450e-01
-6.64634228e-01 -1.27959919e+00 1.42706186e-01 9.99998808e-01
8.66231993e-02 -1.14991859e-01 -1.10496545e+00 -1.94231451e-01
-4.01790142e-01 6.31702602e-01 -1.91350624e-01 -1.11852810e-01
-6.21124148e-01 -7.57313192e-01 4.95408237e-01 1.75497442e-01
5.88366389e-01 -8.49174917e-01 -1.79833531e-01 2.84955800e-01
-5.00306070e-01 -1.26237249e+00 -1.05682600e+00 2.17621654e-01
-5.18746614e-01 -7.14120030e-01 -1.00608349e+00 -1.07180488e+00
7.07302570e-01 4.69884515e-01 7.11875498e-01 -2.92305291e-01
2.53314376e-01 -5.51241226e-02 -1.60752445e-01 -4.71426457e-01
-6.72224641e-01 1.20643735e-01 3.44759166e-01 4.68620569e-01
2.69080907e-01 -5.10160446e-01 -2.32281551e-01 2.91323990e-01
-8.38856459e-01 -1.45779469e-03 6.50723219e-01 5.62859237e-01
3.23517352e-01 9.08557415e-01 6.37912214e-01 -3.70794207e-01
8.70842218e-01 -2.92208970e-01 -5.03626227e-01 3.92258093e-02
-2.21410260e-01 7.98133668e-03 8.73729646e-01 -8.18044841e-01
-1.22615850e+00 -1.79812755e-03 -2.61206776e-01 -2.97977000e-01
-4.15524960e-01 2.61882305e-01 -7.54576027e-01 5.48704207e-01
3.09486598e-01 4.92821753e-01 -4.18934137e-01 -6.82853878e-01
1.96883515e-01 1.03120089e+00 7.18750119e-01 -3.35117131e-01
1.04881740e+00 6.07207716e-02 -4.77333993e-01 -1.22275555e+00
-7.07801461e-01 -6.97907448e-01 -5.70510626e-01 -6.39258400e-02
7.04352140e-01 -9.83228445e-01 -5.28552830e-01 8.85072351e-01
-1.11439610e+00 -9.40803885e-02 -1.48804352e-01 6.73764348e-01
-1.61362767e-01 3.40645850e-01 -4.46810812e-01 -1.13807416e+00
-2.11167112e-01 -1.31355309e+00 9.12335694e-01 3.70542586e-01
2.40005534e-02 -6.13383293e-01 -7.21816495e-02 2.68505335e-01
2.71200895e-01 -2.88743764e-01 5.50448477e-01 -9.51561928e-01
-5.33282578e-01 -5.30027524e-02 2.39821553e-01 4.27728295e-01
6.91590667e-01 -5.01758493e-02 -1.28308570e+00 -3.21995974e-01
5.46523809e-01 4.15792733e-01 7.74971247e-01 4.45934951e-01
5.99131346e-01 -5.01484275e-01 -1.62702501e-01 4.21571940e-01
8.47195029e-01 6.48225367e-01 2.26699069e-01 -6.18433803e-02
5.98213434e-01 5.47710836e-01 3.62120777e-01 1.08795233e-01
1.98791131e-01 5.22145092e-01 5.33567257e-02 -1.80238858e-01
-4.88702118e-01 -8.23020097e-03 5.92232347e-01 1.51184535e+00
1.83990106e-01 -4.58798021e-01 -5.93720794e-01 7.36683667e-01
-1.42806160e+00 -1.01901650e+00 2.19069093e-01 2.38101125e+00
1.00918913e+00 5.14057338e-01 3.00068110e-01 6.50482595e-01
1.25084460e+00 2.58307159e-01 -5.74890375e-01 -3.19781601e-02
-2.50425279e-01 -5.81972711e-02 1.49294972e-01 9.44450915e-01
-9.80880857e-01 8.44911754e-01 5.64316559e+00 1.01577735e+00
-1.14536965e+00 4.02231840e-03 4.37518328e-01 -1.46392167e-01
-2.04685777e-02 -3.49996269e-01 -1.08502305e+00 6.35985315e-01
1.02728724e+00 -1.21190012e-01 7.49882102e-01 3.28936666e-01
2.83499807e-01 -4.10699695e-02 -1.10411608e+00 1.10475802e+00
2.40624741e-01 -6.21331275e-01 -4.48201187e-02 -5.91778345e-02
4.52595472e-01 -2.88664579e-01 2.93310266e-02 2.26711035e-01
3.03448379e-01 -6.86435759e-01 1.08475959e+00 3.54738981e-01
4.64079946e-01 -8.75264108e-01 3.71665299e-01 7.75655508e-01
-1.41618633e+00 -1.55936092e-01 -7.20661283e-02 2.43829057e-01
2.62899935e-01 9.69505072e-01 -8.16709578e-01 4.57063973e-01
4.37152475e-01 9.80945528e-02 -3.09199214e-01 1.04484856e+00
-4.25104916e-01 7.96839476e-01 -3.84163499e-01 2.50021994e-01
-7.97261298e-02 -1.61516160e-01 9.63765502e-01 1.46731591e+00
4.95029449e-01 1.35950804e-01 2.37991646e-01 6.53330088e-01
-4.86717708e-02 -5.76754212e-02 -3.61467227e-02 6.60571605e-02
8.24648380e-01 1.16788280e+00 -8.19895446e-01 -4.99675006e-01
-1.89076245e-01 9.73548472e-01 6.18157014e-02 7.40218222e-01
-7.35902190e-01 -4.84037727e-01 5.68000197e-01 -1.64110035e-01
5.18541396e-01 -1.76312938e-01 -1.07679181e-01 -1.22193217e+00
1.49718687e-01 -9.99432564e-01 1.09205700e-01 -5.02574742e-01
-1.15800047e+00 9.00856793e-01 -2.37660185e-01 -1.22680128e+00
-3.32158417e-01 -3.21526140e-01 -6.78317368e-01 1.25967360e+00
-1.70895422e+00 -6.36173308e-01 -1.48292571e-01 5.76413870e-01
1.07373142e+00 -3.04785967e-01 6.14975393e-01 4.15353775e-01
-8.69885802e-01 5.40329278e-01 1.34192780e-01 3.35882664e-01
6.96894586e-01 -1.25029206e+00 2.42090002e-01 1.36463022e+00
1.65372863e-02 4.71195281e-01 8.54478240e-01 -5.22157669e-01
-8.08066130e-01 -9.28978562e-01 1.03177106e+00 3.72293033e-02
3.69012922e-01 -6.07999086e-01 -7.27293730e-01 3.57024401e-01
1.47933930e-01 -2.47936010e-01 6.01586103e-01 -2.80896366e-01
-5.51766008e-02 -3.64836127e-01 -9.31424081e-01 6.07175171e-01
6.46779656e-01 -5.26717246e-01 -8.58629704e-01 -1.64897814e-01
7.35434413e-01 -4.80892837e-01 -4.49021220e-01 6.54774997e-03
2.00856090e-01 -7.26379931e-01 1.01120877e+00 -9.66277942e-02
-1.09659582e-01 -6.72032118e-01 -2.33274862e-01 -1.62794340e+00
-3.68951291e-01 -8.65416408e-01 -3.57631207e-01 1.89748633e+00
4.71941829e-01 -6.04355991e-01 2.25746483e-01 3.88187677e-01
-1.51121110e-01 -1.94943711e-01 -8.57303083e-01 -9.46282506e-01
-1.46590203e-01 -2.73196340e-01 8.26091170e-01 8.63915026e-01
-2.66799033e-01 6.16949320e-01 -3.06009233e-01 7.49449074e-01
8.25866520e-01 5.87774143e-02 6.02293670e-01 -1.33303356e+00
-4.02241021e-01 -5.52263618e-01 2.29872182e-01 -1.37767589e+00
5.17904311e-02 -6.03168011e-01 5.36426485e-01 -1.32862079e+00
-4.80245091e-02 -2.05788702e-01 -3.56339842e-01 2.66265329e-02
-6.06336236e-01 -3.65099937e-01 1.95591018e-01 3.32041234e-01
-1.89833269e-01 5.51784873e-01 1.07204342e+00 -4.09864515e-01
-6.42120719e-01 6.02471948e-01 -8.87182176e-01 8.15566599e-01
6.93654120e-01 -5.45155168e-01 -5.80835998e-01 -3.23726326e-01
-4.32170212e-01 1.73106194e-01 7.38810524e-02 -1.20711076e+00
4.30333436e-01 -5.94531372e-02 2.41348028e-01 -7.88180709e-01
6.06480777e-01 -9.12030280e-01 -3.25368196e-02 8.71677697e-02
-2.46558741e-01 -3.26259285e-01 2.11582378e-01 3.57487798e-01
-3.62254232e-01 -3.45503360e-01 8.04541767e-01 4.95255291e-02
-2.33289942e-01 2.64847040e-01 -4.69879538e-01 2.24944606e-01
7.31354177e-01 -3.50339502e-01 -2.13368446e-01 -4.00713950e-01
-1.00848722e+00 -2.90968213e-02 -1.25208810e-01 4.67025161e-01
5.12452126e-01 -1.29093480e+00 -8.60031247e-01 4.15657818e-01
-3.36657941e-01 1.88142672e-01 1.56267554e-01 6.46557927e-01
1.45417243e-01 2.73803711e-01 3.88252527e-01 -4.74804997e-01
-1.33079231e+00 7.68840551e-01 3.56518030e-01 -8.71087238e-02
-3.09567839e-01 9.67011452e-01 4.36883092e-01 -9.58939046e-02
5.80687821e-01 -3.33366275e-01 -4.46554154e-01 3.29904377e-01
7.91989505e-01 5.41450322e-01 2.58025583e-02 -9.22909141e-01
-3.32148045e-01 5.21438360e-01 -5.09483367e-02 -5.48300266e-01
1.20175493e+00 -6.05725348e-01 4.09846902e-02 7.09369540e-01
9.92490530e-01 6.05549753e-01 -1.35598040e+00 -4.36380863e-01
-1.02211967e-01 -3.46685886e-01 9.90481526e-02 -7.67226279e-01
-1.10948038e+00 9.41392660e-01 5.31220615e-01 4.96811181e-01
1.48635948e+00 1.26631008e-02 6.54836535e-01 4.05128710e-02
-5.04147187e-02 -1.14918053e+00 -1.70859229e-02 6.28061950e-01
7.72686183e-01 -9.72353101e-01 -3.81858230e-01 -6.46829367e-01
-3.04249376e-01 9.70752180e-01 4.22242075e-01 2.97460556e-01
7.96193421e-01 5.92099786e-01 2.56145716e-01 1.47714779e-01
-3.85591000e-01 -3.21445405e-01 5.01037836e-01 5.49662530e-01
1.57622501e-01 1.92234039e-01 2.07840294e-01 9.95135248e-01
-6.86495125e-01 -3.67247611e-01 5.17727017e-01 6.49675667e-01
-7.69863307e-01 -9.00132716e-01 -9.61495519e-01 4.19685729e-02
-4.63925958e-01 -3.50268245e-01 -3.41256171e-01 2.17489272e-01
2.34617293e-01 1.65586400e+00 -1.44737184e-01 -7.23589361e-02
4.14919138e-01 1.80345118e-01 1.31633401e-01 -6.67458355e-01
-3.68849844e-01 8.47400427e-01 -1.79659445e-02 7.85263479e-02
-4.23409641e-01 -6.92141593e-01 -1.16460621e+00 -6.82004914e-02
-6.60856307e-01 4.82033134e-01 6.67066336e-01 1.13179410e+00
-1.51599608e-02 9.85708833e-01 7.76421309e-01 -8.74527752e-01
-5.32432854e-01 -1.15734732e+00 -7.82129109e-01 9.35544223e-02
1.08077967e+00 -3.27483922e-01 -6.87717319e-01 4.10462767e-01] | [14.860955238342285, 5.907175064086914] |
faf41f38-b8bb-495e-8882-7f8c885d6065 | take-5-interpretable-image-classification | 2303.13166 | null | https://arxiv.org/abs/2303.13166v1 | https://arxiv.org/pdf/2303.13166v1.pdf | Take 5: Interpretable Image Classification with a Handful of Features | Deep Neural Networks use thousands of mostly incomprehensible features to identify a single class, a decision no human can follow. We propose an interpretable sparse and low dimensional final decision layer in a deep neural network with measurable aspects of interpretability and demonstrate it on fine-grained image classification. We argue that a human can only understand the decision of a machine learning model, if the features are interpretable and only very few of them are used for a single decision. For that matter, the final layer has to be sparse and, to make interpreting the features feasible, low dimensional. We call a model with a Sparse Low-Dimensional Decision SLDD-Model. We show that a SLDD-Model is easier to interpret locally and globally than a dense high-dimensional decision layer while being able to maintain competitive accuracy. Additionally, we propose a loss function that improves a model's feature diversity and accuracy. Our more interpretable SLDD-Model only uses 5 out of just 50 features per class, while maintaining 97% to 100% of the accuracy on four common benchmark datasets compared to the baseline model with 2048 features. | ['Bodo Rosenhahn', 'Marco Rudolph', 'Thomas Norrenbrock'] | 2023-03-23 | null | null | null | null | ['fine-grained-image-classification', 'interpretable-machine-learning'] | ['computer-vision', 'methodology'] | [ 3.36449057e-01 7.15436816e-01 -9.42948982e-02 -8.74541163e-01
-3.67861032e-01 -5.60703516e-01 5.47507107e-01 1.75761133e-01
-2.52541900e-01 5.87807953e-01 2.29509488e-01 -6.10240817e-01
-1.75534293e-01 -5.29774070e-01 -7.38171518e-01 -6.56714618e-01
5.56464531e-02 7.36449182e-01 -2.20963165e-01 3.14501852e-01
-3.48799676e-02 4.93292838e-01 -1.62409651e+00 8.59089077e-01
5.46456814e-01 1.48497081e+00 -1.85993910e-01 5.09897470e-01
2.48423852e-02 8.82318318e-01 -5.25860965e-01 -4.48822051e-01
3.50097716e-01 -2.44994447e-01 -1.23909450e+00 2.15212509e-01
8.51447582e-01 -5.74740946e-01 4.28440198e-02 6.90425932e-01
-1.06447034e-01 -2.74606735e-01 1.20388639e+00 -1.30900896e+00
-1.00252903e+00 6.54778123e-01 -2.83738256e-01 1.07618265e-01
-7.16922507e-02 9.69066769e-02 1.43673384e+00 -1.03005326e+00
2.84539014e-01 1.47949457e+00 6.73898578e-01 4.44326133e-01
-1.32838178e+00 -4.19544220e-01 5.57459056e-01 9.46745425e-02
-1.08359528e+00 -4.78368253e-01 2.55768180e-01 -5.98772466e-01
1.03433001e+00 5.03800035e-01 5.99450886e-01 1.07430625e+00
4.31213856e-01 5.86000681e-01 1.09852087e+00 -3.49550158e-01
5.53209901e-01 1.24075618e-02 7.05704391e-01 1.01747155e+00
6.13700926e-01 7.42906006e-03 -5.45363307e-01 -5.83834164e-02
7.18224525e-01 2.88112819e-01 -7.10982755e-02 -2.53031403e-01
-1.07420909e+00 1.10988152e+00 8.09285104e-01 1.52557075e-01
-3.91911209e-01 4.00010973e-01 1.32933676e-01 5.67104518e-01
5.70207000e-01 7.89784968e-01 -6.93931043e-01 2.02322304e-01
-4.84759688e-01 -4.96248901e-02 8.40493321e-01 6.10089779e-01
9.62671101e-01 -3.55325490e-02 -1.39434159e-01 3.92034292e-01
2.43277162e-01 2.33323485e-01 3.64849120e-01 -1.21846783e+00
1.31169900e-01 8.62100780e-01 -1.13293089e-01 -1.14032304e+00
-4.38799530e-01 -7.44715154e-01 -1.33232582e+00 4.88155603e-01
2.62156725e-01 -7.92204887e-02 -1.11911500e+00 1.55485582e+00
-1.78589135e-01 -9.12733376e-02 -4.72547160e-03 9.84686673e-01
6.27304852e-01 5.38168371e-01 -8.13977867e-02 2.38733202e-01
1.35977685e+00 -9.48477626e-01 -2.40738213e-01 -7.40451753e-01
7.42664993e-01 -1.97215572e-01 1.19723535e+00 6.58670068e-01
-7.52024174e-01 -5.04950225e-01 -1.29742992e+00 -4.18657660e-01
-2.86650360e-01 2.09960267e-01 1.11663246e+00 3.97611618e-01
-1.27253735e+00 5.21752954e-01 -8.13793004e-01 -6.08546287e-02
7.68640816e-01 6.38679445e-01 -6.93465233e-01 9.77808889e-03
-8.64444256e-01 1.03470671e+00 3.61325055e-01 1.21532820e-01
-7.72223353e-01 -4.46056902e-01 -8.44094276e-01 3.64662290e-01
-1.10149428e-01 -1.07615972e+00 1.24625909e+00 -1.53377616e+00
-8.97382259e-01 1.03744841e+00 -4.97368753e-01 -8.16806674e-01
4.45294768e-01 -2.52545416e-01 -3.07460967e-02 1.27870262e-01
1.72690243e-01 1.03060591e+00 1.06504428e+00 -1.28090274e+00
-7.37643301e-01 -5.38208604e-01 4.08108622e-01 1.50341734e-01
-4.83498394e-01 -3.75104427e-01 2.64537963e-03 -4.90745455e-01
3.33325207e-01 -9.56483245e-01 -2.65430659e-01 6.25465631e-01
-5.33941925e-01 -3.78760010e-01 8.53079677e-01 -3.23155791e-01
9.52509940e-01 -2.06540680e+00 8.57413858e-02 1.49071664e-01
1.10800207e+00 9.13006663e-02 -2.65343457e-01 -2.63786048e-01
-3.74515295e-01 6.36126459e-01 -3.03136259e-01 -6.29484653e-01
1.80470496e-01 5.48968136e-01 -4.35581684e-01 4.57218468e-01
7.22142875e-01 8.32080185e-01 -6.43171072e-01 1.08853817e-01
7.93959051e-02 3.27524155e-01 -5.75753927e-01 -1.43482625e-01
2.58116014e-02 1.03326172e-01 -5.04574001e-01 3.74344677e-01
5.85219681e-01 -7.56350398e-01 -5.38783483e-02 -2.40352094e-01
3.29895079e-01 3.00510764e-01 -9.79740739e-01 1.23663568e+00
-7.29000270e-01 1.15991390e+00 -3.06459218e-01 -1.13025081e+00
7.24516094e-01 -5.28542092e-03 -8.78646299e-02 -4.88942355e-01
-9.84614715e-02 3.08240652e-01 1.86333567e-01 -2.68080812e-02
1.43196791e-01 -9.01906267e-02 -1.59097865e-01 7.32758880e-01
3.19639295e-02 2.50923038e-01 -2.76011586e-01 1.00054219e-01
1.13600624e+00 -5.45239449e-01 5.54148257e-01 -6.19884253e-01
1.14151530e-01 -1.93828553e-01 4.73678857e-01 1.03810251e+00
7.20523819e-02 7.92105556e-01 8.47660959e-01 -1.28872788e+00
-9.79682982e-01 -1.09304464e+00 -2.69555837e-01 1.15722418e+00
3.88289569e-03 -2.70142227e-01 -6.63826764e-01 -9.06911194e-01
1.27967194e-01 7.14534819e-01 -1.07618749e+00 -2.62955219e-01
-2.07318127e-01 -4.79855955e-01 4.91641879e-01 7.01892912e-01
4.37386841e-01 -7.04869628e-01 -6.33121848e-01 -1.15141161e-01
5.35297655e-02 -8.69880915e-01 -1.49387047e-01 9.09840405e-01
-7.54858434e-01 -9.15544331e-01 -1.56429276e-01 -7.90065169e-01
1.10254002e+00 1.24699622e-01 1.49980915e+00 3.64524901e-01
2.00386178e-02 -2.75418252e-01 -2.35915810e-01 -5.86353898e-01
-4.00228590e-01 1.87300697e-01 2.69874722e-01 8.78444389e-02
5.50981045e-01 -4.17143643e-01 -5.96044600e-01 2.61510074e-01
-8.83556426e-01 3.55109811e-01 7.89010286e-01 1.02905655e+00
4.45474595e-01 -2.04123214e-01 5.10725498e-01 -9.45311368e-01
3.90716374e-01 -4.19030696e-01 -1.77887142e-01 7.90577680e-02
-7.00824559e-01 6.37476146e-01 9.65388358e-01 -1.98543876e-01
-5.55852652e-01 1.83971420e-01 -9.27494466e-03 -3.67756456e-01
-3.32902819e-01 2.95635253e-01 2.99859550e-02 -8.51096362e-02
1.01115191e+00 -9.54105780e-02 -1.04948640e-01 -4.54077929e-01
3.46725762e-01 7.88259149e-01 1.87647015e-01 -4.32291001e-01
6.40809774e-01 4.44020987e-01 1.28622418e-02 -4.76709932e-01
-1.37424433e+00 -9.96123627e-02 -9.15562689e-01 3.69510651e-01
6.29055738e-01 -1.00248981e+00 -4.78141308e-01 4.45666790e-01
-1.47042370e+00 -1.03145592e-01 -5.33218503e-01 4.01492901e-02
-3.58643770e-01 -8.99489075e-02 -5.25927842e-01 -3.69817436e-01
-1.90942124e-01 -1.25691915e+00 1.29253531e+00 -1.80632576e-01
-6.63556814e-01 -1.17089999e+00 -5.34802020e-01 1.20907195e-01
2.80964196e-01 3.32540423e-01 1.20705080e+00 -8.57299685e-01
-5.13990819e-01 -2.42633410e-02 -4.78782117e-01 5.87314785e-01
3.14652503e-01 -1.93671092e-01 -1.30036557e+00 -1.58521146e-01
6.54023066e-02 -4.86253113e-01 1.35816443e+00 3.57167304e-01
1.70268381e+00 -1.06588733e+00 -1.25840262e-01 7.00295806e-01
1.13233876e+00 -2.13209257e-01 3.48573864e-01 1.41520917e-01
6.84141815e-01 3.43252718e-01 2.63989806e-01 3.39251846e-01
3.42302710e-01 3.18095833e-01 6.53002381e-01 -4.72010016e-01
2.27605142e-02 -5.69598451e-02 1.66685134e-01 4.21207935e-01
-2.11435836e-02 -4.50676739e-01 -9.65453744e-01 5.08322477e-01
-1.81375229e+00 -5.45264065e-01 -6.95509762e-02 1.93781316e+00
7.66385436e-01 3.73689204e-01 -2.85945058e-01 2.73950905e-01
2.74159640e-01 1.28069773e-01 -8.96450758e-01 -9.56924915e-01
-3.06020021e-01 2.53501713e-01 2.80678898e-01 6.46807373e-01
-1.33110356e+00 5.42546809e-01 6.48843861e+00 6.09736264e-01
-1.13477039e+00 -6.70976639e-02 1.57924211e+00 8.88279267e-03
-6.88987494e-01 -1.77357227e-01 -7.92069793e-01 2.09261775e-01
1.01180565e+00 6.96911886e-02 6.91462830e-02 9.55370009e-01
1.36605397e-01 3.58412489e-02 -1.91694748e+00 9.64214563e-01
-9.03778747e-02 -1.67871535e+00 6.01628065e-01 3.01934421e-01
6.35158181e-01 2.53622923e-02 3.06253344e-01 4.15979922e-02
3.73270571e-01 -1.88018537e+00 9.48841155e-01 3.39917302e-01
9.63129938e-01 -6.67292893e-01 7.32927442e-01 4.69872355e-01
-8.40672314e-01 -2.24286124e-01 -4.21854466e-01 -4.67358828e-01
-5.30679345e-01 6.58653021e-01 -8.08569074e-01 1.55612268e-02
5.01239717e-01 8.93346727e-01 -8.70821476e-01 4.81947780e-01
-2.73904532e-01 6.12908363e-01 -6.15258694e-01 8.01876113e-02
6.53447926e-01 2.28989199e-01 1.60388157e-01 1.09325421e+00
2.47355461e-01 3.54612917e-02 3.45462300e-02 9.34349477e-01
-2.32384667e-01 -5.67377567e-01 -6.99098706e-01 8.74389187e-02
2.32978538e-01 1.06983972e+00 -3.89445126e-01 -3.26611340e-01
-5.51568083e-02 1.34262371e+00 7.02904344e-01 1.98246777e-01
-5.97409308e-01 -6.47231238e-03 9.99320805e-01 1.52925342e-01
2.90005863e-01 5.42738028e-02 -9.54501987e-01 -1.22574413e+00
2.43156120e-01 -9.59038973e-01 1.79061994e-01 -6.90454781e-01
-1.41875494e+00 9.05901432e-01 -3.06379020e-01 -1.12025523e+00
-5.98595321e-01 -1.08103001e+00 -3.88742745e-01 8.37177396e-01
-1.27528083e+00 -1.16764808e+00 -2.05245256e-01 2.01720744e-01
5.87683141e-01 -2.87659951e-02 1.21438348e+00 -2.34910935e-01
-2.94714004e-01 7.89954066e-01 1.19549677e-01 1.70965701e-01
2.74205565e-01 -1.56675935e+00 6.77525818e-01 5.23394108e-01
3.02136391e-01 8.08723390e-01 5.86108744e-01 1.07706934e-01
-1.04997337e+00 -1.35381365e+00 1.02394724e+00 -5.55592000e-01
3.65062177e-01 -4.01982069e-01 -9.15798128e-01 1.00853682e+00
8.19881558e-02 2.31974632e-01 8.25210750e-01 4.33318049e-01
-6.33879840e-01 -2.58182973e-01 -1.16536462e+00 5.33580899e-01
1.08256996e+00 -4.49819803e-01 -7.22390473e-01 4.07385647e-01
8.24006617e-01 -1.14304684e-01 -7.10472465e-01 2.42738068e-01
7.25173533e-01 -1.09741223e+00 8.36719573e-01 -1.00981700e+00
7.40961492e-01 -2.77302265e-01 -1.34222388e-01 -1.45286691e+00
-6.59055650e-01 -2.55346030e-01 -2.54041821e-01 5.44843674e-01
8.19858670e-01 -9.29790676e-01 7.25046813e-01 7.65688241e-01
-3.11227888e-02 -1.07650566e+00 -8.83876443e-01 -5.54833353e-01
2.58617669e-01 -4.01884168e-01 4.89273399e-01 8.28443766e-01
-3.73972774e-01 5.59476554e-01 -1.06385544e-01 1.39008150e-01
6.19095802e-01 2.80207723e-01 2.93059558e-01 -1.63271523e+00
-3.07115227e-01 -3.87290627e-01 -7.28018582e-01 -1.28816593e+00
3.23137462e-01 -8.08420420e-01 -1.39760613e-01 -1.49603224e+00
3.56078267e-01 -5.02552688e-01 -2.98418105e-01 9.93982255e-01
1.96522940e-02 4.20753360e-01 1.12969622e-01 1.20491393e-01
-5.79206049e-01 2.77008355e-01 1.07341421e+00 -3.04794252e-01
3.30724627e-01 -7.76975974e-02 -1.30453277e+00 1.21619129e+00
6.08015358e-01 -5.53307652e-01 -3.78084660e-01 -9.70706582e-01
1.33408636e-01 -4.40637350e-01 6.93034410e-01 -8.63634586e-01
8.75608623e-02 -1.54968947e-01 9.29637015e-01 -1.83558181e-01
4.30896789e-01 -6.94742143e-01 7.47675821e-02 5.40665030e-01
-7.60870576e-01 7.81106129e-02 2.14333516e-02 5.65294921e-01
-1.50047064e-01 4.49271910e-02 7.10266292e-01 -1.91049706e-02
-8.49648952e-01 3.74490440e-01 -3.85301292e-01 -4.78578880e-02
8.38492811e-01 -3.85239184e-01 -3.71989161e-01 -3.85713667e-01
-8.53825152e-01 5.41513078e-02 5.21438718e-01 3.51218104e-01
8.11040759e-01 -1.21158290e+00 -6.94346249e-01 3.97329360e-01
1.63572565e-01 4.28145707e-01 -1.74864888e-01 2.12734908e-01
-6.85901046e-01 6.66724443e-01 -1.08404040e-01 -7.81491816e-01
-1.06476963e+00 -1.10231884e-01 5.86192369e-01 -3.56879920e-01
-5.10285318e-01 1.03208816e+00 8.66275847e-01 -3.99159342e-01
2.16162190e-01 -9.49130416e-01 6.16270453e-02 -4.30937223e-02
6.50408447e-01 -2.21894696e-01 1.86202005e-01 -5.21743417e-01
-4.88998413e-01 4.39010233e-01 -2.60657996e-01 2.53470182e-01
1.40235269e+00 -6.79229200e-02 -8.08402225e-02 5.10672510e-01
1.37374854e+00 -5.13172328e-01 -1.42938960e+00 -2.34987408e-01
-1.91024289e-01 -2.39694163e-01 1.06397979e-01 -9.06736553e-01
-9.91110027e-01 1.17694557e+00 3.77628088e-01 3.47273916e-01
1.19067192e+00 1.15383312e-01 3.69629234e-01 1.08168566e+00
2.57344067e-01 -4.70028132e-01 -9.32329893e-02 7.09799647e-01
1.17742574e+00 -1.38269663e+00 -1.28561616e-01 -3.29970926e-01
-6.42549336e-01 1.38144565e+00 4.79785800e-01 -2.07959458e-01
6.36026740e-01 3.55536342e-01 -9.38828960e-02 -1.59590229e-01
-1.12590718e+00 1.74307138e-01 5.49049258e-01 5.57260334e-01
1.04989775e-01 2.93189198e-01 3.66570294e-01 8.02458227e-01
-4.65170443e-01 -2.82946881e-02 3.32624257e-01 4.09030586e-01
-6.29309535e-01 -6.29804552e-01 1.01523429e-01 8.98545086e-01
-3.46474439e-01 -3.53642911e-01 -7.24584997e-01 4.95966136e-01
2.11121306e-01 8.88836384e-01 4.82330203e-01 -5.60221374e-01
-8.12122971e-02 -1.60409018e-01 1.06272832e-01 -7.97804117e-01
-4.94025797e-01 -7.16241956e-01 3.46549273e-01 -5.70834875e-01
-2.21940368e-01 -4.55732584e-01 -1.22544885e+00 -6.10126972e-01
1.00368848e-02 -1.48296520e-01 3.95495564e-01 1.30727613e+00
6.92916095e-01 3.37171584e-01 6.99191153e-01 -5.60909748e-01
-6.85352385e-01 -6.68842971e-01 -4.73043263e-01 3.24956447e-01
9.11704063e-01 -4.69899029e-01 -7.12658048e-01 1.91628337e-01] | [8.923942565917969, 5.676281929016113] |
f1796795-c34e-42a1-8f88-957c89a333ad | a-primer-on-getting-neologisms-from-foreign | 2304.10495 | null | https://arxiv.org/abs/2304.10495v1 | https://arxiv.org/pdf/2304.10495v1.pdf | A primer on getting neologisms from foreign languages to under-resourced languages | Mainly due to lack of support, most under-resourced languages have a reduced lexicon in most realms and domains of increasing importance, then their speakers need to significantly augment it. Although neologisms should arise from the languages themselves, external sources are widely accepted. However, we dispute the "common sense" of using the imposed official languages, which are highly probably a legacy of colonialism, as the only source, and we propose to introduce neologisms from any language as long as these neologisms "sound like" native words of the target languages. | ['Luis Camacho'] | 2023-03-07 | null | null | null | null | ['common-sense-reasoning'] | ['reasoning'] | [-3.30139369e-01 2.58789897e-01 -3.41896743e-01 -2.47251481e-01
-1.27690896e-01 -8.60691190e-01 9.45291519e-01 -1.91914603e-01
-8.73644471e-01 1.02285290e+00 7.76613772e-01 -9.21083689e-01
-9.95117053e-02 -6.32939517e-01 -2.96392977e-01 -3.88045549e-01
3.21600050e-01 3.69819999e-01 2.18106255e-01 -1.01565206e+00
5.11035562e-01 3.63043159e-01 -1.10263741e+00 -1.76665694e-01
1.26096177e+00 -2.83550292e-01 3.07204872e-01 5.35641722e-02
-5.28393984e-01 9.43290651e-01 -7.46901810e-01 -4.85454738e-01
1.13384925e-01 -6.26944542e-01 -1.01783442e+00 -1.08659603e-01
4.82688583e-02 -2.95114130e-01 -2.53859669e-01 1.27389622e+00
8.37920085e-02 -2.10193306e-01 4.76611525e-01 -5.13779283e-01
-1.07821679e+00 9.21300352e-01 -6.50609508e-02 1.60467908e-01
3.38187128e-01 -1.13164470e-01 8.44418705e-01 -8.10750306e-01
9.36647654e-01 1.39229405e+00 4.72508192e-01 5.52763522e-01
-1.19822145e+00 -5.57210505e-01 5.43959737e-01 -2.59845912e-01
-1.45073378e+00 -6.21584713e-01 4.95579600e-01 -6.04563594e-01
1.06113720e+00 2.83246487e-01 9.02287900e-01 8.93719077e-01
4.37135309e-01 -2.87009150e-01 1.36216247e+00 -7.71602750e-01
-2.33938232e-01 7.76884258e-01 -3.44231635e-01 5.35607398e-01
8.14104676e-01 -7.63703138e-02 -5.95066726e-01 -2.26776794e-01
7.47788906e-01 -2.68441200e-01 -2.70623803e-01 2.88384199e-01
-1.09371519e+00 9.26336288e-01 8.45894888e-02 1.21855497e+00
-2.18960941e-01 -1.07032515e-01 2.72940636e-01 6.87889457e-01
4.88254637e-01 6.25588298e-01 -6.27040386e-01 -2.95340121e-01
-6.46003664e-01 1.06491521e-01 1.12516749e+00 5.75215757e-01
5.18234611e-01 2.30371729e-01 9.25502479e-01 1.13059914e+00
6.19789362e-01 7.45890737e-01 8.60810637e-01 -8.46445918e-01
-1.40952483e-01 4.92532283e-01 -2.93723810e-02 -9.59999740e-01
-7.20880777e-02 -4.85032022e-01 -3.25587600e-01 2.31624156e-01
4.54031378e-01 1.29959760e-02 -5.35922110e-01 1.89471114e+00
5.02863601e-02 -8.94812822e-01 4.06383336e-01 6.36640847e-01
4.68230069e-01 5.96771479e-01 1.89899147e-01 -5.31162620e-01
1.24436593e+00 -4.99524206e-01 -6.27107084e-01 -5.57248473e-01
7.07246184e-01 -1.06591845e+00 1.34705842e+00 3.01682979e-01
-9.85214651e-01 -1.95820928e-02 -9.37835574e-01 1.04888536e-01
-6.23034298e-01 -2.75183201e-01 7.11918831e-01 1.02879751e+00
-1.31153643e+00 4.74544883e-01 -5.15911996e-01 -6.54618382e-01
-2.45328456e-01 7.92860091e-02 -4.24453408e-01 5.76392531e-01
-1.20986032e+00 1.40541863e+00 3.85487944e-01 2.63034105e-01
-4.38668340e-01 -2.26598650e-01 -6.25361025e-01 -6.35733545e-01
3.39544147e-01 -1.61982909e-01 9.65199769e-01 -1.60340643e+00
-1.14193439e+00 1.47252941e+00 1.34933470e-02 6.15776330e-03
4.56397474e-01 -2.26632461e-01 -9.83704031e-01 3.38086560e-02
7.56401837e-01 1.46393016e-01 2.92940855e-01 -1.33736646e+00
-6.66102529e-01 -9.94438455e-02 2.84102470e-01 1.83278680e-01
-4.19974029e-01 5.72537720e-01 -3.53114419e-02 -7.39363551e-01
3.41854781e-01 -8.33782017e-01 -1.94662005e-01 -3.43066454e-01
-1.05376050e-01 5.36525398e-02 3.17597896e-01 -1.06732905e+00
1.32832384e+00 -2.27388597e+00 -2.31436551e-01 2.14318246e-01
9.91828367e-02 -6.52930513e-02 2.65669703e-01 7.24170029e-01
-5.42529896e-02 8.48101497e-01 -7.53905252e-02 4.77310479e-01
9.71655771e-02 7.05648482e-01 -4.85217988e-01 5.85842609e-01
-1.10638134e-01 4.27272469e-01 -9.89234507e-01 -4.60388988e-01
-1.26147434e-01 3.33652407e-01 -3.59523267e-01 -3.86571139e-01
7.72946104e-02 2.29299590e-01 -2.75844753e-01 6.43503189e-01
5.37884235e-01 -1.34098362e-02 7.88257957e-01 4.62017179e-01
-6.89002931e-01 7.59963036e-01 -9.02645290e-01 1.41719174e+00
-4.18671906e-01 6.18786216e-01 3.03384632e-01 -5.37419617e-01
6.69887900e-01 3.26742679e-01 -1.21855736e-03 -6.42458797e-01
-5.34635894e-02 1.10629082e+00 7.27358460e-01 -4.94017959e-01
8.32651377e-01 -6.49212778e-01 -2.38328978e-01 3.53164852e-01
-3.96313429e-01 -2.14167982e-01 1.92025840e-01 3.54767501e-01
6.09911740e-01 4.41305012e-01 8.19073021e-01 -1.17361486e+00
5.49726725e-01 3.40916902e-01 7.82974660e-01 2.98649669e-01
-1.51496977e-01 -2.47862693e-02 4.19607788e-01 -3.38987827e-01
-9.72603261e-01 -1.15210283e+00 -6.89759493e-01 1.22286749e+00
2.49692407e-02 -3.19481403e-01 -5.45631051e-01 -4.21790689e-01
-3.13880354e-01 8.28028679e-01 -3.93842310e-01 2.43303120e-01
-6.83515608e-01 -7.35556006e-01 6.78234279e-01 4.93239611e-02
1.19995642e-02 -9.78070617e-01 -7.81970501e-01 4.91793305e-01
1.60826873e-02 -8.11829805e-01 -1.30755901e-01 3.06022078e-01
-6.61758482e-01 -6.70683086e-01 -8.30989778e-01 -9.11449134e-01
5.88041961e-01 1.66836716e-02 8.53103220e-01 5.85147917e-01
2.48386487e-01 9.28104967e-02 -4.50777709e-01 -2.19638765e-01
-1.15227938e+00 1.25511914e-01 2.81195551e-01 -6.32818937e-01
4.19073910e-01 -7.07934499e-01 -5.48406178e-03 -1.79287940e-01
-8.07462931e-01 -2.85349309e-01 3.88265818e-01 5.17045081e-01
-1.05944954e-01 -1.89761952e-01 1.06528771e+00 -1.47000599e+00
5.30057251e-01 -4.08076406e-01 -2.78206408e-01 1.26684517e-01
-7.14519083e-01 -1.14901781e-01 6.36233389e-01 -3.06026787e-01
-1.13621294e+00 -9.13160682e-01 -4.59700942e-01 5.86094975e-01
-1.52370840e-01 6.46375418e-01 -2.38033205e-01 2.56132632e-02
9.28933203e-01 1.81633964e-01 -7.74103925e-02 -5.16672432e-01
3.27816635e-01 7.40118086e-01 5.18487096e-01 -9.00145054e-01
7.78310716e-01 4.42427427e-01 -4.97226298e-01 -1.33706427e+00
-3.33657026e-01 -1.02413416e-01 -4.76224273e-01 3.53983343e-02
5.56530893e-01 -1.07923663e+00 -7.13567510e-02 -4.33066115e-02
-1.38522100e+00 -2.70692140e-01 -4.26412046e-01 6.67602003e-01
-1.34999398e-02 4.97688770e-01 -6.31095409e-01 -9.12217796e-01
2.20416877e-02 -8.75452280e-01 2.84982145e-01 7.91870579e-02
-5.64056277e-01 -1.67048836e+00 1.74819499e-01 -1.26098141e-01
4.13928956e-01 3.19093376e-01 1.39876390e+00 -6.25473142e-01
3.26583944e-02 2.08651811e-01 5.41165173e-02 3.49573761e-01
7.07497239e-01 6.61135390e-02 -8.16264451e-01 -7.85902441e-02
4.52618748e-01 -1.24030173e-01 1.10292070e-01 -9.64101329e-02
-1.00686394e-01 -6.69275045e-01 -1.50161177e-01 3.49398047e-01
1.88300705e+00 5.28917551e-01 4.43825483e-01 5.77074826e-01
2.71234602e-01 1.01571226e+00 -6.12453446e-02 2.02149272e-01
3.22866678e-01 2.15675592e-01 -1.86604574e-01 -6.03139102e-02
2.40748059e-02 -3.80566269e-01 8.95118833e-01 1.80224288e+00
-5.43690324e-02 1.06144994e-01 -1.25286436e+00 8.12172294e-01
-1.38973093e+00 -8.13733399e-01 -5.90112686e-01 2.01422787e+00
1.08926594e+00 3.81233156e-01 1.55919045e-01 -2.46604562e-01
4.79839981e-01 2.77690619e-01 -1.80804774e-01 -6.76396787e-01
-7.40749598e-01 1.81166545e-01 4.73863691e-01 9.35885727e-01
-2.72566319e-01 1.42945969e+00 8.22408485e+00 5.52902997e-01
-1.30518425e+00 2.92843580e-01 1.93466380e-01 3.11389327e-01
-8.88299644e-01 5.95584035e-01 -4.71548319e-01 2.52805889e-01
7.42386639e-01 -6.25120461e-01 2.91487664e-01 5.70420444e-01
2.21466035e-01 -1.10145859e-01 -6.52006984e-01 1.99569508e-01
9.58879367e-02 -8.89014125e-01 1.38097659e-01 3.72210890e-01
5.38896263e-01 2.33609840e-01 -1.56068161e-01 -8.48096889e-03
5.75493991e-01 -1.14989948e+00 1.14203513e+00 5.78583367e-02
1.02771270e+00 -3.71913403e-01 4.76739109e-01 3.24566126e-01
-9.46604967e-01 4.47904617e-01 -7.35427201e-01 -6.88534021e-01
3.44984978e-01 4.37693864e-01 -1.77229643e-01 3.66830736e-01
3.14609498e-01 2.39762247e-01 -7.43551910e-01 3.57726067e-01
-4.09100413e-01 7.99200416e-01 -3.04520845e-01 -8.25430974e-02
4.85654294e-01 -4.27566588e-01 5.84785581e-01 1.26944602e+00
2.19179004e-01 -4.38850038e-02 1.31562278e-01 5.87481618e-01
4.04535741e-01 7.89639175e-01 -9.79271293e-01 -3.51003826e-01
4.60272193e-01 8.90695453e-01 -9.98856902e-01 -3.17954302e-01
-8.87341440e-01 8.85125458e-01 2.65687723e-02 4.31177169e-01
-4.82687443e-01 -3.72606218e-01 6.69536054e-01 4.59118664e-01
-2.26084724e-01 -2.68798411e-01 -2.42598355e-01 -1.32599270e+00
3.15717645e-02 -1.20161605e+00 4.34171967e-02 -1.91724524e-01
-1.08196938e+00 9.41119492e-01 -3.23370956e-02 -6.87105119e-01
-4.16268289e-01 -8.83626938e-01 -1.77442089e-01 9.84869897e-01
-9.90150630e-01 -9.56108451e-01 5.49735367e-01 3.62355053e-01
2.02089012e-01 -2.90379703e-01 9.74735081e-01 1.34729639e-01
1.26393422e-01 4.17509228e-01 3.26736122e-01 8.09998363e-02
6.05205834e-01 -9.53536212e-01 2.24061444e-01 1.00917375e+00
2.78577656e-01 1.10817397e+00 9.60082293e-01 -8.68929505e-01
-7.35702872e-01 -3.15416485e-01 1.59758198e+00 -4.57856864e-01
1.19570827e+00 -4.29603636e-01 -8.40298355e-01 9.22165751e-01
8.50949824e-01 -9.68860507e-01 7.90208638e-01 4.01724428e-01
-4.70885098e-01 1.00431010e-01 -9.24867630e-01 8.68424416e-01
1.18752897e+00 -7.78113842e-01 -9.99386966e-01 4.65249568e-01
8.95221233e-01 3.64350349e-01 -8.10883224e-01 -5.00519201e-02
7.45599806e-01 -7.70950735e-01 4.47980016e-01 -4.49923038e-01
8.46581236e-02 -2.71816552e-01 -3.91735405e-01 -1.23562789e+00
-3.66862684e-01 -6.48148775e-01 1.03224051e+00 1.48312223e+00
7.64982581e-01 -1.34152782e+00 4.09318149e-01 3.10357600e-01
-9.11284089e-02 -1.00281864e-01 -1.12351072e+00 -8.97829831e-01
7.24550247e-01 -4.29982513e-01 3.80612254e-01 1.59886253e+00
3.71994138e-01 6.18442237e-01 -3.53868961e-01 -1.59969464e-01
1.79549128e-01 -1.42202124e-01 4.88221675e-01 -1.19500959e+00
-3.20818007e-01 -7.05651402e-01 -1.11218989e-01 -6.16450489e-01
2.56568879e-01 -9.94568884e-01 4.42332216e-02 -1.39489198e+00
-2.30141431e-01 -5.01387000e-01 2.66672671e-01 1.78835392e-01
2.08580941e-01 8.98638740e-02 2.60468245e-01 4.96212959e-01
1.69729948e-01 1.97053596e-01 8.97029817e-01 3.67707103e-01
-4.38165575e-01 -5.54167926e-01 -1.51139820e+00 1.31487894e+00
7.04411864e-01 -6.25459075e-01 -2.88051784e-01 -6.61289394e-01
8.18320811e-01 -2.26684600e-01 5.39713986e-02 -7.57579267e-01
-6.54607937e-02 -8.12413812e-01 -3.79251577e-02 1.98467836e-01
-5.82713261e-03 -8.69353175e-01 5.59869528e-01 1.16918659e+00
2.85461023e-02 4.01963204e-01 2.00478569e-01 -2.19807640e-01
-1.65251359e-01 -6.13275766e-01 7.82085001e-01 -4.48042035e-01
-6.59282565e-01 -4.21717048e-01 -9.83649254e-01 3.47193718e-01
7.23919809e-01 -4.73985463e-01 -3.83261561e-01 -2.69988805e-01
-4.63098049e-01 -3.19099903e-01 1.16171861e+00 1.74858257e-01
2.30279863e-01 -1.09001875e+00 -8.93913090e-01 4.32018898e-02
1.92067456e-02 -5.97878754e-01 -1.46007344e-01 6.50011420e-01
-1.26781380e+00 4.02816296e-01 -3.87553394e-01 2.56888419e-01
-5.12205780e-01 3.84226292e-01 2.98022717e-01 1.57235429e-01
-7.97988534e-01 4.85666066e-01 7.09940612e-01 -5.51700056e-01
-3.35898995e-01 -4.29209089e-03 -2.66974360e-01 1.00775443e-01
3.43254209e-01 -4.36907215e-03 -4.36663359e-01 -1.25354898e+00
-8.34022462e-01 2.48563945e-01 -2.93180317e-01 -7.97809780e-01
1.02624881e+00 -4.20798182e-01 -6.60331368e-01 9.39299583e-01
6.66901410e-01 1.08158624e+00 -2.88577259e-01 1.15147354e-02
8.69878381e-02 -6.49707675e-01 -5.63947856e-01 -8.17119360e-01
-4.40595269e-01 5.42115510e-01 4.85014394e-02 1.66743249e-01
9.20359731e-01 1.79777607e-01 2.96793252e-01 1.76909775e-01
5.42820036e-01 -1.44707537e+00 -5.26167452e-01 6.17271543e-01
9.53437567e-01 -4.31785703e-01 1.68390289e-01 -3.08227211e-01
-4.62799847e-01 7.86955833e-01 4.50513750e-01 -1.70211792e-01
7.79871166e-01 2.11308420e-01 6.37908936e-01 -1.29125714e-01
-7.83815742e-01 -1.68597713e-01 -3.73942703e-01 7.36721277e-01
1.05962408e+00 2.47925296e-01 -1.59596241e+00 6.72131717e-01
-8.79294097e-01 -3.96373987e-01 9.71125007e-01 9.52710092e-01
-6.95432663e-01 -1.41094780e+00 -6.51345432e-01 1.28390834e-01
-9.77513373e-01 -5.47289670e-01 -7.06569254e-01 1.35946023e+00
6.82480454e-01 7.88463116e-01 3.56335603e-02 -2.73226917e-01
-5.72477989e-02 -2.12962311e-02 2.62248427e-01 -6.57354653e-01
-8.10809135e-01 4.21193212e-01 4.77282077e-01 1.85822949e-01
-5.78217328e-01 -6.24056756e-01 -1.22849417e+00 -8.88233244e-01
-1.53565750e-01 5.37422299e-01 4.44974214e-01 1.12179327e+00
-3.36449891e-01 -3.57203409e-02 1.96359485e-01 -9.32431873e-03
-5.39038360e-01 -8.33190143e-01 -1.07846940e+00 2.56688171e-03
-1.72194570e-01 -1.91404209e-01 -4.54363376e-01 7.48129934e-02] | [10.451817512512207, 9.90169906616211] |
bd47f959-7fa4-4c81-81c3-7d119d3f0dec | exploring-text-representations-for-generative | null | null | https://aclanthology.org/2022.clinicalnlp-1.12 | https://aclanthology.org/2022.clinicalnlp-1.12.pdf | Exploring Text Representations for Generative Temporal Relation Extraction | Sequence-to-sequence models are appealing because they allow both encoder and decoder to be shared across many tasks by formulating those tasks as text-to-text problems. Despite recently reported successes of such models, we find that engineering input/output representations for such text-to-text models is challenging. On the Clinical TempEval 2016 relation extraction task, the most natural choice of output representations, where relations are spelled out in simple predicate logic statements, did not lead to good performance. We explore a variety of input/output representations, with the most successful prompting one event at a time, and achieving results competitive with standard pairwise temporal relation extraction systems. | ['Guergana Savova', 'Timothy Miller', 'Steven Bethard', 'Dmitriy Dligach'] | null | null | null | null | naacl-clinicalnlp-2022-7 | ['temporal-relation-extraction'] | ['natural-language-processing'] | [ 8.95065248e-01 6.60179436e-01 -5.22679687e-01 -5.68163216e-01
-9.86285686e-01 -7.19641805e-01 9.41301227e-01 6.91634893e-01
-3.35551083e-01 1.08511400e+00 7.02629387e-01 -9.53811288e-01
-3.74015540e-01 -6.14665926e-01 -5.45159340e-01 -3.42257112e-01
-3.53345394e-01 7.47412264e-01 5.58060408e-02 -3.26618701e-01
-2.33270675e-01 1.07852168e-01 -8.81463706e-01 8.77753854e-01
3.39339048e-01 6.88382268e-01 -3.28340799e-01 7.99093902e-01
6.01362437e-02 1.38589299e+00 -4.64300156e-01 -6.94152474e-01
-4.44085002e-02 -6.48888171e-01 -1.35786116e+00 -5.02156615e-01
-3.00212577e-03 -1.12937517e-01 -4.78930742e-01 5.47218919e-01
5.27256131e-01 -1.07952364e-01 9.84874725e-01 -8.83700967e-01
-6.81354761e-01 1.08615541e+00 -3.09491128e-01 6.22286081e-01
8.08328092e-01 1.01522110e-01 1.59332871e+00 -4.59492981e-01
1.04878414e+00 1.17572284e+00 4.23913151e-01 5.06421745e-01
-1.81425738e+00 -5.78078449e-01 6.90210685e-02 1.63840711e-01
-1.15101361e+00 -5.20199120e-01 1.49523914e-01 -5.13872981e-01
1.87601352e+00 4.67697471e-01 2.92766541e-01 1.36588418e+00
6.37414098e-01 7.63471961e-01 9.40676630e-01 -2.99691528e-01
-2.09643617e-01 -3.77938271e-01 2.11275980e-01 4.62373734e-01
1.91834450e-01 1.88770682e-01 -7.20258117e-01 -4.39746559e-01
3.43898475e-01 -1.42652392e-01 -2.27121159e-01 2.06050649e-01
-1.31196880e+00 6.79293215e-01 3.39443296e-01 1.85971648e-01
-2.50841796e-01 3.18600833e-01 8.86082828e-01 4.42743570e-01
3.69400799e-01 5.96189201e-01 -7.79375494e-01 -2.89379880e-02
-9.62070227e-01 5.06494701e-01 8.90323877e-01 9.82890308e-01
2.24578641e-02 -8.14848959e-01 -5.95828652e-01 4.93303567e-01
4.06286716e-02 -6.04052730e-02 3.26042503e-01 -3.22122246e-01
8.79214227e-01 3.58022958e-01 -1.42275572e-01 -7.26412117e-01
-8.23357046e-01 5.98818362e-02 -7.14794993e-01 -2.13767871e-01
3.25921744e-01 -3.32755089e-01 -8.64472508e-01 1.79576707e+00
-1.54233530e-01 -1.70529604e-01 2.50155151e-01 4.08351153e-01
8.85722995e-01 3.27278286e-01 5.43740988e-01 -4.82758641e-01
1.62035489e+00 -5.09178758e-01 -1.08092153e+00 -5.30442297e-01
1.17460179e+00 -6.42652094e-01 2.15305358e-01 3.00930500e-01
-1.14254737e+00 1.86198637e-01 -9.23424184e-01 -4.34058428e-01
-3.77656102e-01 -3.69888097e-01 9.60263193e-01 -2.88081300e-02
-9.97188687e-01 6.34255767e-01 -8.67066562e-01 -4.97813106e-01
5.54518938e-01 5.15227437e-01 -6.52828455e-01 1.43978596e-01
-1.50155187e+00 1.42025125e+00 6.72552764e-01 2.20524892e-02
-6.91464484e-01 -7.29902923e-01 -9.73640621e-01 -8.61840099e-02
5.77083230e-01 -1.01711392e+00 1.50766921e+00 -3.77278417e-01
-1.03298521e+00 9.77854729e-01 -2.85056800e-01 -1.01735020e+00
4.50752616e-01 -9.16606858e-02 -6.19913042e-01 3.16809714e-02
1.53243706e-01 5.53805172e-01 2.36917034e-01 -2.79839754e-01
-3.02302718e-01 -8.18906873e-02 -1.65588349e-01 1.35517687e-01
2.69774884e-01 4.43649590e-01 -1.04383506e-01 -6.00801349e-01
-1.28076330e-01 -6.02949917e-01 -5.13143182e-01 -1.01063654e-01
-7.97796249e-01 -4.82949257e-01 3.19342524e-01 -6.53115869e-01
1.30346811e+00 -1.83053994e+00 7.16209412e-02 -6.92276005e-03
3.24254513e-01 -8.92664641e-02 2.80837491e-02 8.70666504e-01
-8.46463859e-01 4.04349208e-01 -2.43096545e-01 -2.05466986e-01
-1.20294407e-01 3.91801596e-01 -4.15038496e-01 4.11329776e-01
8.13534260e-01 1.39120567e+00 -1.01156545e+00 -8.78403425e-01
-1.07419118e-01 1.87350377e-01 -5.72767019e-01 6.29743487e-02
-4.55980986e-01 2.56953359e-01 -5.26955426e-01 3.99929136e-01
4.38755602e-02 -5.95540762e-01 6.76912129e-01 1.56935111e-01
2.77466297e-01 1.30180800e+00 -6.45801127e-01 1.58214569e+00
-3.27611625e-01 5.61020017e-01 -4.73356247e-01 -7.83602536e-01
4.60701704e-01 8.42106104e-01 6.97131276e-01 -6.22568727e-01
-2.92602591e-02 7.51652569e-02 4.49750423e-01 -6.68029428e-01
2.30718568e-01 -7.44778812e-01 -1.00325078e-01 5.05470574e-01
1.77830234e-01 -1.21557236e-01 3.01633418e-01 4.90044713e-01
1.56995606e+00 2.29565307e-01 9.49273646e-01 -1.62291944e-01
2.98368391e-02 1.16466902e-01 5.65491259e-01 4.73016560e-01
8.68915841e-02 8.02801490e-01 1.16163838e+00 -4.56288010e-01
-7.92026281e-01 -7.90490031e-01 -4.64948684e-01 8.18366170e-01
-4.93746012e-01 -1.05303526e+00 1.51014412e-02 -8.52224529e-01
-1.03426509e-01 6.97539032e-01 -9.75855947e-01 -4.01890874e-02
-3.96698773e-01 -8.35219622e-01 1.00806808e+00 7.11828887e-01
-4.19681638e-01 -9.80715990e-01 -4.45156306e-01 5.94201207e-01
-2.10077554e-01 -1.26646149e+00 -3.80981177e-01 8.28640640e-01
-8.55376959e-01 -1.18498015e+00 -5.72191179e-01 -4.60470110e-01
3.82256448e-01 -5.63058674e-01 1.37358618e+00 -1.14137076e-01
-3.67617428e-01 -2.83426285e-01 -2.23962814e-01 -5.19973218e-01
-5.28276742e-01 -1.71712153e-02 -1.81709900e-01 -4.39410031e-01
5.29424727e-01 -3.75170469e-01 -3.77963066e-01 2.37312559e-02
-1.02673316e+00 4.20469522e-01 5.59293330e-01 9.03457999e-01
5.03195941e-01 -3.83635998e-01 5.44583261e-01 -1.41392219e+00
7.43847132e-01 -7.34237850e-01 6.46144822e-02 1.13252126e-01
-7.02940464e-01 2.99544483e-01 5.06790042e-01 -2.90199071e-01
-4.86106694e-01 2.51545687e-03 -4.12913650e-01 9.06782150e-02
-1.02724910e-01 8.08838427e-01 3.82531166e-01 7.09354937e-01
1.08870590e+00 1.57133281e-01 2.21551657e-01 -2.25484252e-01
6.14676654e-01 2.54198074e-01 3.43218416e-01 -4.56531614e-01
3.60997647e-01 1.91167936e-01 5.79778776e-02 -2.08697885e-01
-1.00052655e+00 -2.81714410e-01 -4.19184715e-01 6.09628618e-01
1.17260134e+00 -9.01250005e-01 -9.26627040e-01 -3.02940816e-01
-1.19088483e+00 -4.20158893e-01 -4.70072716e-01 4.15855855e-01
-6.91246748e-01 5.28828464e-02 -9.77979422e-01 -4.50173914e-01
-3.14226091e-01 -1.05177140e+00 1.15840256e+00 -3.10180724e-01
-1.11934090e+00 -1.06809008e+00 6.57354966e-02 -1.60686731e-01
5.81459552e-02 4.85235959e-01 1.36539769e+00 -1.15882087e+00
-2.03909531e-01 -1.43868461e-01 -1.66147724e-01 -1.95465639e-01
3.20061237e-01 -9.98935010e-03 -7.47256875e-01 1.38326377e-01
-2.97387958e-01 -5.98792851e-01 8.56314600e-01 2.38041788e-01
1.20844162e+00 -5.81051528e-01 -8.73029470e-01 3.39757323e-01
1.00308943e+00 -3.44998017e-02 6.02871299e-01 -8.65406245e-02
3.32309186e-01 3.75276208e-01 2.80389667e-01 2.12224975e-01
5.04502833e-01 6.63819790e-01 1.84570059e-01 -6.99036196e-02
-1.49481162e-01 -4.00191873e-01 2.88797617e-01 2.61001796e-01
1.30797997e-01 -2.80057549e-01 -1.26475060e+00 6.48496330e-01
-1.86075079e+00 -9.41047490e-01 -2.75438905e-01 1.63442385e+00
1.43513906e+00 4.94291216e-01 4.45104726e-02 9.62062646e-03
2.60319281e-02 2.46402606e-01 -2.71867067e-01 -5.83631098e-01
-1.55393809e-01 6.38951302e-01 5.17634153e-01 5.35170734e-01
-9.46759760e-01 8.30174327e-01 7.56756639e+00 6.18564904e-01
-8.15490365e-01 3.74126323e-02 7.15416789e-01 -3.07659864e-01
-6.25213265e-01 1.54739872e-01 -7.64335096e-01 2.13691771e-01
1.29224551e+00 -1.41124696e-01 8.66760612e-02 1.69003755e-01
1.27826825e-01 -1.79313049e-02 -1.93161786e+00 8.09255421e-01
-1.74625918e-01 -1.77940524e+00 2.79131848e-02 4.33447510e-02
3.37352723e-01 1.69088349e-01 -3.67330045e-01 2.43900329e-01
7.55001664e-01 -1.76282060e+00 5.69002450e-01 4.50545430e-01
9.39042926e-01 -2.12943226e-01 6.44736409e-01 1.92651272e-01
-9.60239410e-01 1.38064533e-01 1.29232615e-01 -1.68093726e-01
4.79449630e-01 5.20758331e-01 -1.37476122e+00 9.42050934e-01
2.25237623e-01 1.08973098e+00 -3.75392884e-01 6.55525148e-01
-4.89872783e-01 5.68570733e-01 -3.57428849e-01 -1.54244661e-01
1.50652111e-01 3.33371252e-01 2.99009383e-01 1.70988917e+00
-1.55500904e-01 3.65281582e-01 2.88814213e-02 8.59208584e-01
-1.15299754e-01 5.55662066e-02 -1.10911226e+00 -3.74694109e-01
3.76299359e-02 8.49301875e-01 -4.86233801e-01 -3.41418207e-01
-4.14069921e-01 6.73716843e-01 4.57263529e-01 3.79441202e-01
-7.35283315e-01 -1.83710799e-01 7.02818394e-01 7.12461174e-02
-1.19768577e-02 -1.84717923e-01 -5.96358120e-01 -8.68698120e-01
-1.17089935e-01 -9.93283212e-01 1.06404972e+00 -7.20729411e-01
-1.41287792e+00 5.60083628e-01 9.00793374e-02 -9.60589111e-01
-6.84799194e-01 -5.99747241e-01 -3.42409283e-01 1.27200639e+00
-1.13243294e+00 -9.80205357e-01 6.60528600e-01 5.39549887e-01
3.27975959e-01 3.40816468e-01 1.22323859e+00 2.77658761e-01
-3.05887401e-01 4.80731606e-01 -6.61501348e-01 3.10006857e-01
8.41957092e-01 -1.13898754e+00 6.93822145e-01 6.44115686e-01
2.43118897e-01 9.80479658e-01 8.10625136e-01 -7.14438558e-01
-1.42153275e+00 -9.64783132e-01 1.49798024e+00 -9.65116620e-01
9.83171225e-01 -5.29284298e-01 -7.61983156e-01 1.33223987e+00
4.64278102e-01 5.19018807e-02 1.00335431e+00 5.83939731e-01
-5.04022181e-01 4.77121681e-01 -7.10873663e-01 7.62767434e-01
1.10814989e+00 -9.10416603e-01 -1.01321793e+00 7.28404999e-01
1.02434480e+00 -8.10062289e-01 -1.16064835e+00 3.97552490e-01
1.79085672e-01 -1.95852965e-01 8.83697212e-01 -1.43068099e+00
9.20377314e-01 -2.26485699e-01 -1.69956125e-02 -1.06228352e+00
-2.92946398e-01 -8.80057573e-01 -3.28085393e-01 8.68735969e-01
1.23833632e+00 -5.83989561e-01 2.96414733e-01 7.23851085e-01
-1.09496742e-01 -1.05152905e+00 -7.73895800e-01 -4.40625817e-01
2.21373320e-01 -7.20407665e-01 3.78995031e-01 1.01492441e+00
7.36547053e-01 8.37498605e-01 -1.64014623e-01 -3.31937559e-02
-2.98875924e-02 1.04353294e-01 1.15706034e-01 -8.95852983e-01
-5.66926837e-01 -4.93693888e-01 -3.06531668e-01 -8.27997386e-01
1.17679067e-01 -1.51953697e+00 -1.11887090e-01 -1.91246891e+00
3.18667591e-01 -3.16819102e-01 -3.58798176e-01 1.19809592e+00
-3.50111097e-01 -8.42613913e-03 -7.70444348e-02 -5.94880953e-02
-4.30892080e-01 2.11601809e-01 1.14614558e+00 -2.76479453e-01
1.29509315e-01 -1.21220790e-01 -1.11167645e+00 3.51407140e-01
4.29562241e-01 -7.66801298e-01 -4.64170009e-01 -4.13843572e-01
6.88382387e-01 6.14364207e-01 1.34210244e-01 -3.45035523e-01
2.26637587e-01 -3.32608134e-01 2.64650881e-01 -4.33685213e-01
3.36071759e-01 -3.98698956e-01 2.48198122e-01 5.12542605e-01
-9.43321705e-01 1.68986201e-01 3.71681780e-01 5.51269770e-01
-2.29936838e-01 2.33859811e-02 4.70223039e-01 -5.52732795e-02
-3.70360404e-01 3.63150537e-01 -3.16528052e-01 4.45321590e-01
9.76104319e-01 2.28386357e-01 -4.90886599e-01 -3.31750125e-01
-1.11447132e+00 3.92473310e-01 -8.98925737e-02 4.30115581e-01
4.43078130e-01 -1.02162313e+00 -8.61203909e-01 -1.65836439e-01
3.30735356e-01 1.39884884e-02 -1.31700397e-01 1.18749249e+00
-3.00820440e-01 7.28897512e-01 1.15350917e-01 -4.54377860e-01
-1.29776669e+00 8.64814699e-01 4.90777463e-01 -7.42759824e-01
-1.02684772e+00 9.25400138e-01 1.84217557e-01 -1.14258476e-01
1.41530067e-01 -5.77594697e-01 -9.43180770e-02 1.68573976e-01
6.20026648e-01 -4.78119969e-01 3.08038145e-01 -1.28698483e-01
-9.00113881e-01 -7.00764731e-02 -3.91865313e-01 -2.66614050e-01
1.41579294e+00 3.48140389e-01 -1.70198560e-01 4.47303355e-01
1.10235262e+00 -3.97058517e-01 -7.79214203e-01 -3.10831845e-01
3.69607687e-01 1.85985699e-01 -3.02174807e-01 -1.06676221e+00
-4.41580415e-01 8.27842236e-01 -2.67698858e-02 1.68018922e-01
9.06430960e-01 2.12420881e-01 5.62137485e-01 2.85180897e-01
3.06497723e-01 -5.96473873e-01 -2.79132545e-01 7.82041550e-01
7.40567565e-01 -1.05025971e+00 2.29130164e-01 -3.54239762e-01
-8.15174520e-01 1.03536928e+00 2.22975180e-01 1.87891603e-01
5.37308693e-01 7.10805833e-01 -1.70611531e-01 -4.69407022e-01
-1.48027313e+00 -2.41600171e-01 3.47821712e-01 2.95644104e-01
1.18524837e+00 2.97218829e-01 -4.31669354e-01 5.89082181e-01
-1.73878193e-01 2.53630251e-01 3.92560810e-01 9.52549160e-01
2.95617253e-01 -1.47018337e+00 1.52693123e-01 8.99353206e-01
-1.03238070e+00 -6.39011204e-01 -4.41923887e-01 6.89077079e-01
-2.84878790e-01 8.89109313e-01 -7.45667964e-02 -2.53198177e-01
2.36323446e-01 4.58291322e-01 4.64581400e-01 -1.10326004e+00
-9.81254876e-01 -1.01363994e-01 9.48809326e-01 -7.02919424e-01
-2.21957132e-01 -6.90211594e-01 -1.60143757e+00 -1.21412329e-01
2.28750855e-02 6.80927979e-03 1.03801340e-01 1.28117275e+00
2.25904152e-01 1.03396201e+00 -5.22873513e-02 -1.80492401e-01
-4.09039289e-01 -9.65800703e-01 -3.27342659e-01 5.32027185e-01
5.93114495e-01 -3.35566103e-01 3.01291078e-01 8.32930729e-02] | [8.505514144897461, 8.918716430664062] |
0ba28b81-cb2e-4e1c-b016-e1f0866c9031 | rep-predicting-the-time-course-of-drug | 1907.11911 | null | https://arxiv.org/abs/1907.11911v1 | https://arxiv.org/pdf/1907.11911v1.pdf | REP: Predicting the Time-Course of Drug Sensitivity | The biological processes involved in a drug's mechanisms of action are oftentimes dynamic, complex and difficult to discern. Time-course gene expression data is a rich source of information that can be used to unravel these complex processes, identify biomarkers of drug sensitivity and predict the response to a drug. However, the majority of previous work has not fully utilized this temporal dimension. In these studies, the gene expression data is either considered at one time-point (before the administration of the drug) or two timepoints (before and after the administration of the drug). This is clearly inadequate in modeling dynamic gene-drug interactions, especially for applications such as long-term drug therapy. In this work, we present a novel REcursive Prediction (REP) framework for drug response prediction by taking advantage of time-course gene expression data. Our goal is to predict drug response values at every stage of a long-term treatment, given the expression levels of genes collected in the previous time-points. To this end, REP employs a built-in recursive structure that exploits the intrinsic time-course nature of the data and integrates past values of drug responses for subsequent predictions. It also incorporates tensor completion that can not only alleviate the impact of noise and missing data, but also predict unseen gene expression levels (GELs). These advantages enable REP to estimate drug response at any stage of a given treatment from some GELs measured in the beginning of the treatment. Extensive experiments on a dataset corresponding to 53 multiple sclerosis patients treated with interferon are included to showcase the effectiveness of REP. | ['Amin Emad', 'Cheng Qian', 'Nicholas D. Sidiropoulos'] | 2019-07-27 | null | null | null | null | ['drug-response-prediction'] | ['medical'] | [ 3.52422267e-01 -7.08424628e-01 -4.38059896e-01 -2.43635967e-01
-3.12822193e-01 -5.98297298e-01 4.96136755e-01 5.47694504e-01
-1.82043210e-01 6.30600452e-01 1.46853551e-01 -3.81716430e-01
-6.65253937e-01 -7.08350480e-01 -2.46269912e-01 -1.19173455e+00
-5.78371584e-01 6.86344802e-01 -1.94571838e-01 -2.94391006e-01
1.85784966e-01 8.50323737e-01 -1.16732430e+00 5.40910661e-01
5.71525097e-01 6.47784650e-01 1.31402854e-02 6.67630017e-01
-1.24638915e-01 2.67933160e-01 -2.08984599e-01 2.52826333e-01
-1.09372325e-01 -3.39333922e-01 -6.58770025e-01 3.74403507e-01
-2.07172632e-01 2.92896796e-02 -2.54456580e-01 7.57310510e-01
5.04592836e-01 5.51652648e-02 4.95852679e-01 -9.30427969e-01
7.49660507e-02 7.96391293e-02 -5.94344079e-01 2.28648603e-01
3.01487714e-01 4.53511029e-01 8.45137179e-01 -8.06773067e-01
6.32327259e-01 1.10607445e+00 5.56154370e-01 3.84505212e-01
-1.76768136e+00 -5.62004782e-02 -1.14186414e-01 1.84269324e-01
-1.37444091e+00 -4.20800269e-01 6.23547375e-01 -1.00502563e+00
8.78513396e-01 4.30971563e-01 6.82828724e-01 9.60120618e-01
5.42791903e-01 4.51501071e-01 1.37640846e+00 -8.61132219e-02
4.31958199e-01 -5.69326222e-01 2.50798732e-01 4.45243299e-01
-3.71246904e-01 1.91323474e-01 -3.63656461e-01 -5.98120928e-01
3.80897462e-01 5.07481992e-01 -4.59409177e-01 -1.22091137e-01
-1.24295151e+00 6.21054471e-01 1.54820934e-01 8.20815384e-01
-7.38407373e-01 -1.85901329e-01 5.06168544e-01 3.56443793e-01
7.86243856e-01 2.48715565e-01 -7.04026759e-01 -2.49093607e-01
-9.07581806e-01 1.13703467e-01 5.10123730e-01 -6.64326474e-02
6.84486032e-01 -4.19397235e-01 -3.42568874e-01 9.25976038e-01
1.51156887e-01 1.11704186e-01 6.60017729e-01 -2.45927989e-01
3.15814242e-02 7.95624673e-01 6.58858567e-02 -1.12530208e+00
-8.61873150e-01 -4.62583452e-01 -1.10915220e+00 2.59811133e-02
6.84638560e-01 -8.70787650e-02 -9.68791783e-01 1.63645172e+00
5.81667066e-01 5.76684773e-01 -1.41577154e-01 8.25215220e-01
2.16561139e-01 6.49703622e-01 1.84473455e-01 -8.37489247e-01
1.37218404e+00 -1.84311390e-01 -7.36810267e-01 1.50184721e-01
9.45036650e-01 -5.29359818e-01 6.96519196e-01 4.30985123e-01
-6.41506970e-01 -1.29846439e-01 -7.41184294e-01 2.18149871e-01
-3.95402908e-01 4.82293591e-02 7.86311269e-01 3.16924512e-01
-6.21417284e-01 1.18102396e+00 -1.20501423e+00 -3.60264242e-01
2.95138180e-01 6.61388755e-01 -5.62457025e-01 -3.01441342e-01
-1.12994301e+00 5.61756253e-01 -7.29277283e-02 3.98859739e-01
-7.02613413e-01 -9.96031940e-01 -2.63366610e-01 -1.46472171e-01
2.39611417e-01 -5.97886443e-01 8.12136710e-01 -5.17059982e-01
-1.47573006e+00 5.53331971e-01 -3.51131469e-01 -7.09135011e-02
2.12306514e-01 3.41506958e-01 -1.71887726e-01 -1.38709128e-01
-3.61266136e-02 -3.12139858e-02 4.98263180e-01 -3.86003643e-01
-1.52792260e-01 -1.18603909e+00 -2.24361226e-01 2.50407066e-02
-9.13559198e-02 -3.74690220e-02 -3.77629846e-01 -5.29534876e-01
1.52966827e-01 -1.19128287e+00 -5.36854506e-01 -2.77455389e-01
-3.17553669e-01 -1.56718075e-01 5.67907274e-01 -5.10031819e-01
1.15727222e+00 -2.03748918e+00 5.59289277e-01 1.64050952e-01
2.17197835e-01 2.53913194e-01 -2.22698092e-01 8.93574297e-01
-5.86524665e-01 2.81015933e-01 -2.81714946e-01 9.08151269e-02
-4.95104849e-01 3.03269684e-01 -1.09218307e-01 6.71161890e-01
1.75826073e-01 8.96586418e-01 -8.46164703e-01 1.90683842e-01
7.48089980e-03 5.91647863e-01 -2.42715240e-01 9.10094678e-02
-5.01268923e-01 7.77934670e-01 -7.71933317e-01 5.85920691e-01
5.23346782e-01 -3.59540671e-01 5.70845842e-01 -3.30301285e-01
-9.25300121e-02 -4.92684059e-02 -8.22594702e-01 1.49021304e+00
-2.03178182e-01 1.06821120e-01 -2.15204805e-01 -1.12296963e+00
5.66941738e-01 5.73557377e-01 1.20136416e+00 -6.48973942e-01
2.26749063e-01 2.00165063e-01 2.09509715e-01 -7.15128422e-01
-3.09303701e-01 -4.08537656e-01 1.86313733e-01 3.72789353e-01
-3.03522646e-01 2.12319225e-01 3.83925557e-01 -2.72497773e-01
1.58562636e+00 -1.68918580e-01 1.94728941e-01 -5.15010357e-02
5.99838138e-01 1.36933252e-02 7.49569654e-01 2.28502646e-01
-6.10729642e-02 2.66703039e-01 8.71220589e-01 -5.11211693e-01
-1.07590282e+00 -4.08895701e-01 -3.18353206e-01 8.38367522e-01
-5.16856670e-01 -2.66428143e-01 -3.71329814e-01 -3.76240641e-01
1.40074000e-01 3.30775939e-02 -9.69158113e-01 4.85244058e-02
-2.78683782e-01 -1.58484793e+00 1.76325560e-01 8.44394267e-02
-1.12884946e-01 -4.38024819e-01 7.60267004e-02 6.47183776e-01
-7.41655231e-02 -7.22297430e-01 -2.40383953e-01 1.86362222e-01
-1.32094157e+00 -1.29170251e+00 -4.82706100e-01 -2.22000510e-01
6.88001752e-01 -2.84163170e-02 5.30937076e-01 -1.15532475e-03
-5.03877103e-01 1.00109324e-01 -1.08312547e-01 -4.54939017e-03
-3.01914036e-01 -2.71957874e-01 2.09019437e-01 4.66041416e-01
2.02115908e-01 -8.21925640e-01 -6.04642868e-01 3.48004103e-01
-1.10116243e+00 -1.14280798e-01 4.55191731e-01 1.16288805e+00
8.33413184e-01 2.51345217e-01 4.23038810e-01 -1.00260842e+00
7.93520808e-01 -8.35754275e-01 -5.03112972e-01 1.98572367e-01
-6.14464641e-01 1.60332432e-03 5.78691602e-01 -5.37806988e-01
-6.85778916e-01 2.59934455e-01 -2.08978161e-01 -3.11509192e-01
-8.67213458e-02 1.13978493e+00 1.63635761e-01 1.55265927e-01
5.00996113e-01 1.90109953e-01 4.17234868e-01 -7.25964427e-01
-3.47919464e-02 6.56210601e-01 1.83288872e-01 -3.71890247e-01
1.44263208e-01 6.16416812e-01 7.12365568e-01 -7.66082942e-01
-4.74561989e-01 -5.99763095e-01 -7.93155730e-01 1.36233708e-02
4.21142429e-01 -6.33386970e-01 -1.03918636e+00 7.56452203e-01
-8.49590719e-01 -3.26101780e-01 2.01796189e-01 4.82259333e-01
-4.32292670e-01 3.12579125e-01 -8.22212636e-01 -5.28932810e-01
-2.27540538e-01 -1.10358834e+00 7.65819013e-01 -1.88351497e-01
-5.80410920e-02 -1.19541562e+00 5.05277336e-01 1.66748703e-01
2.42199168e-01 5.53960025e-01 1.24177408e+00 -5.72323382e-01
-2.77947336e-01 -4.63124067e-01 2.50911117e-01 -8.48141499e-03
4.78574336e-01 2.20376760e-01 -5.63863873e-01 -4.50435519e-01
2.01872051e-01 1.91727132e-01 5.10079920e-01 5.34418762e-01
8.78457010e-01 -2.01492414e-01 -5.88097513e-01 3.35673481e-01
1.46036744e+00 4.58579779e-01 7.69797027e-01 -9.45452303e-02
4.10019904e-01 7.67442822e-01 6.98539078e-01 5.71903825e-01
-1.62478134e-01 1.16843569e+00 1.61783606e-01 -5.92713915e-02
1.58636168e-01 2.29729235e-01 2.34610975e-01 5.75234890e-01
1.55376755e-02 -1.54662043e-01 -1.03613091e+00 5.73538184e-01
-1.86909986e+00 -7.22809136e-01 -7.05709934e-01 2.19523478e+00
9.87582386e-01 -3.44044596e-01 7.43274465e-02 1.71796679e-01
5.79207718e-01 -2.18708038e-01 -5.43004870e-01 -3.37103486e-01
-9.83285829e-02 1.00634098e-01 2.00711340e-01 5.24739087e-01
-8.58325958e-01 4.92744744e-01 6.46743917e+00 7.79835045e-01
-1.43339252e+00 -7.83184692e-02 9.22650397e-01 -4.45065014e-02
1.36917844e-01 8.32898840e-02 -3.54519427e-01 4.91702229e-01
1.21980250e+00 -2.59809643e-01 4.81255740e-01 2.40934551e-01
9.60403919e-01 -2.87492216e-01 -1.21080709e+00 6.63103342e-01
-4.94670928e-01 -1.23559749e+00 -3.13097537e-01 2.48021483e-01
3.83218706e-01 1.72016770e-01 1.57672226e-01 -2.30403826e-01
2.79380288e-02 -9.09940302e-01 -3.69564652e-01 7.80597746e-01
5.50678015e-01 -5.91155589e-01 7.79610276e-01 4.24479574e-01
-7.32243180e-01 -1.38853848e-01 -4.69228253e-02 -1.33817703e-01
1.56748652e-01 1.23431730e+00 -1.41788399e+00 6.39064133e-01
1.09223969e-01 9.72867548e-01 -4.04233932e-01 9.99649346e-01
1.77709467e-03 5.83364904e-01 -2.05912471e-01 2.08165020e-01
1.10655472e-01 -3.72583300e-01 3.99671733e-01 8.82644355e-01
3.40164959e-01 4.94647086e-01 2.51308292e-01 4.45201159e-01
3.96414995e-01 3.65832597e-01 -5.44491529e-01 -5.62133133e-01
-9.73754972e-02 1.14150774e+00 -6.14580870e-01 -6.07881732e-02
-2.92750806e-01 8.77837360e-01 6.70576319e-02 4.35811371e-01
-2.16007933e-01 1.61878616e-02 1.00859022e+00 2.54151106e-01
-1.39612062e-02 -3.64920974e-01 -1.40211970e-01 -9.04467344e-01
-2.72865802e-01 -8.61480296e-01 4.81412649e-01 -4.99824733e-01
-1.37931395e+00 2.25749210e-01 -2.61819094e-01 -1.09462976e+00
-4.25431073e-01 -7.23296106e-01 -2.77966946e-01 1.12358677e+00
-1.08605957e+00 -8.08965862e-01 1.35985211e-01 4.12865400e-01
2.43263140e-01 1.37868509e-01 9.40259814e-01 4.66471910e-01
-9.41520751e-01 -1.23583533e-01 5.41983724e-01 -1.07459806e-01
4.24581409e-01 -8.14013243e-01 -7.04883039e-02 4.96160328e-01
-1.66493207e-01 7.98782349e-01 7.96973526e-01 -8.91219199e-01
-1.73264563e+00 -8.63638937e-01 9.70915079e-01 -3.07953894e-01
9.58817422e-01 -1.36801481e-01 -1.16398168e+00 2.94204742e-01
-4.38413084e-01 1.47991151e-01 1.05645883e+00 3.45765054e-01
-8.28991905e-02 -2.41809845e-01 -1.02016771e+00 6.93014205e-01
5.37021697e-01 -5.49721479e-01 -2.09741741e-02 6.91597044e-01
7.84316212e-02 -1.91537678e-01 -1.34863198e+00 5.52435696e-01
5.54403126e-01 -6.87036633e-01 8.06110919e-01 -1.06461251e+00
2.37273246e-01 -5.90035915e-01 1.34956567e-02 -1.47458518e+00
-5.20755529e-01 -5.04332423e-01 1.04947336e-01 8.30916107e-01
2.41656914e-01 -4.63158906e-01 8.01613688e-01 7.89676309e-01
1.79995909e-01 -1.23241222e+00 -1.01081872e+00 -5.54816782e-01
-2.43575364e-01 -3.73577684e-01 4.88829046e-01 1.08570254e+00
2.59227961e-01 3.61674130e-01 -4.38990176e-01 9.01346132e-02
1.29466817e-01 3.39158267e-01 5.26338398e-01 -1.21597207e+00
-3.80984128e-01 -2.27962911e-01 -6.77289605e-01 -4.41859066e-01
-1.18312486e-01 -9.47080970e-01 -2.30099916e-01 -1.24495745e+00
4.05702919e-01 -5.22478640e-01 -4.64650095e-01 7.53231525e-01
-1.83424428e-01 1.15726456e-01 -2.81874478e-01 2.95459241e-01
7.71907642e-02 3.18656057e-01 1.25871408e+00 -2.13836163e-01
-3.84917587e-01 6.95458576e-02 -4.01937723e-01 3.59147251e-01
5.83185494e-01 -5.42399108e-01 -3.63436311e-01 6.51280805e-02
1.75015956e-01 5.49166858e-01 1.50694907e-01 -4.73686516e-01
2.00949535e-01 -5.43390274e-01 3.61731619e-01 -3.44415843e-01
3.04044366e-01 -7.68838108e-01 8.88351917e-01 7.35887051e-01
-9.55339894e-02 2.54549878e-03 1.70466885e-01 7.44751036e-01
-3.11749935e-01 -8.34689140e-02 4.91618246e-01 2.96811517e-02
-4.18864876e-01 6.55436158e-01 -5.03166676e-01 -6.23302400e-01
9.49833155e-01 7.52364099e-02 -1.18148938e-01 4.73582223e-02
-1.35661614e+00 8.49598199e-02 3.18583369e-01 2.70406395e-01
4.26691830e-01 -1.38870025e+00 -7.19459653e-01 2.84779575e-02
9.46265161e-02 -5.29332042e-01 6.59507096e-01 1.43879902e+00
-1.24766015e-01 4.46636319e-01 -6.88338801e-02 -8.41663718e-01
-1.54956436e+00 8.53783309e-01 3.01533461e-01 -6.32717133e-01
-3.28043640e-01 4.29211706e-01 1.91761181e-01 -6.97405785e-02
-3.79973829e-01 5.08886985e-02 -4.94891167e-01 4.33180541e-01
5.54049253e-01 1.37177229e-01 5.56470752e-01 -5.36916673e-01
-3.29550505e-01 6.35782123e-01 -1.99148983e-01 3.89176980e-02
1.70466793e+00 -3.83628868e-02 -7.15077221e-01 7.61985064e-01
1.41738272e+00 -2.89071918e-01 -7.59350359e-01 -1.76763579e-01
1.72155023e-01 -6.57862246e-01 2.78717756e-01 -9.52190995e-01
-9.32931006e-01 7.14342713e-01 8.42096210e-01 1.94202028e-02
1.25682735e+00 -3.21336359e-01 4.63378131e-01 6.75250962e-02
2.27479488e-01 -8.03401172e-01 -7.29609728e-02 4.20946062e-01
8.53511333e-01 -6.99752092e-01 -7.24929869e-02 -6.43015981e-01
-2.69986778e-01 1.23301435e+00 -1.47858277e-01 3.05252671e-01
6.11168623e-01 2.52495450e-03 -1.48675159e-01 -4.09367979e-01
-1.13194919e+00 -1.04712635e-01 2.19116628e-01 2.48731524e-01
6.72695696e-01 1.59937635e-01 -7.29772687e-01 2.27351338e-01
5.58943629e-01 4.06662464e-01 3.99629444e-01 9.05755401e-01
1.31382095e-02 -1.71647501e+00 -3.27593684e-01 5.66877127e-01
-5.59855938e-01 -3.39160338e-02 -3.54326159e-01 2.14432225e-01
4.98171262e-02 8.59638214e-01 -1.95043817e-01 -3.50610286e-01
4.55332607e-01 4.17198330e-01 5.04977226e-01 -5.61023831e-01
-3.85679811e-01 3.72502267e-01 -3.82529548e-03 -6.35091245e-01
-3.52707356e-01 -1.01203680e+00 -1.12918353e+00 -3.34620208e-01
-1.41906172e-01 1.14806026e-01 8.25500309e-01 9.36286688e-01
8.01394343e-01 5.46040714e-01 1.05625844e+00 -5.39014041e-01
-2.14179218e-01 -8.35854471e-01 -9.31234658e-01 5.31556606e-01
4.00304884e-01 -6.56997800e-01 -2.19365507e-01 8.91681612e-02] | [6.033640384674072, 5.608170509338379] |
54eb687d-19b7-4870-80ba-162153e4b677 | don-t-stop-self-supervision-accent-adaptation | 2307.00453 | null | https://arxiv.org/abs/2307.00453v1 | https://arxiv.org/pdf/2307.00453v1.pdf | Don't Stop Self-Supervision: Accent Adaptation of Speech Representations via Residual Adapters | Speech representations learned in a self-supervised fashion from massive unlabeled speech corpora have been adapted successfully toward several downstream tasks. However, such representations may be skewed toward canonical data characteristics of such corpora and perform poorly on atypical, non-native accented speaker populations. With the state-of-the-art HuBERT model as a baseline, we propose and investigate self-supervised adaptation of speech representations to such populations in a parameter-efficient way via training accent-specific residual adapters. We experiment with 4 accents and choose automatic speech recognition (ASR) as the downstream task of interest. We obtain strong word error rate reductions (WERR) over HuBERT-large for all 4 accents, with a mean WERR of 22.7% with accent-specific adapters and a mean WERR of 25.1% if the entire encoder is accent-adapted. While our experiments utilize HuBERT and ASR as the downstream task, our proposed approach is both model and task-agnostic. | ['Katrin Kirchhoff', 'Sravan Bodapati', 'Karthik Gopalakrishnan', 'Saket Dingliwal', 'Sanchit Sinha', 'Anshu Bhatia'] | 2023-07-02 | null | null | null | null | ['speech-recognition', 'automatic-speech-recognition'] | ['speech', 'speech'] | [ 1.89628109e-01 6.03172719e-01 8.96708760e-03 -8.16041708e-01
-1.25276971e+00 -6.42841578e-01 6.07732594e-01 -1.98372900e-01
-5.94749033e-01 6.97935939e-01 7.03529298e-01 -3.87900203e-01
2.94949621e-01 -2.31195822e-01 -6.02845073e-01 -5.80359638e-01
1.69123739e-01 7.20917642e-01 -1.08777411e-01 -5.25583863e-01
-3.13351572e-01 2.41504550e-01 -1.05608606e+00 -4.06352207e-02
8.17450941e-01 5.58786035e-01 2.29361817e-01 7.72771835e-01
1.59160867e-02 5.13713002e-01 -9.56671357e-01 -6.11817956e-01
1.29584029e-01 -4.16562021e-01 -1.01834786e+00 1.46027207e-01
4.22163576e-01 1.01249322e-01 -4.95047241e-01 1.04290462e+00
7.58028984e-01 4.14017826e-01 5.49561381e-01 -4.86204863e-01
-1.11121976e+00 1.28609765e+00 -1.87891573e-01 3.97679508e-01
-1.47097915e-01 3.58326942e-01 9.64078903e-01 -1.00171936e+00
2.19039977e-01 1.26011336e+00 3.57745588e-01 1.02923679e+00
-1.31348479e+00 -6.84211195e-01 2.39120334e-01 -3.29084806e-02
-1.53760111e+00 -1.20821226e+00 7.54511774e-01 -5.42678079e-03
1.22896886e+00 3.05336982e-01 2.24447362e-02 1.38198352e+00
-3.98586035e-01 7.62081981e-01 9.73536730e-01 -4.64259207e-01
4.37910348e-01 6.66967109e-02 -3.36640701e-02 2.47499928e-01
-1.42529681e-01 1.67414621e-01 -7.59080470e-01 2.27152258e-01
5.50982296e-01 -5.63224077e-01 -1.86669677e-01 2.64788777e-01
-1.28038323e+00 8.67508531e-01 3.82674724e-01 4.24917579e-01
-4.57231939e-01 -7.71642243e-03 4.15332824e-01 4.76976901e-01
7.35020995e-01 4.03027326e-01 -8.25509787e-01 -3.47872585e-01
-8.40250731e-01 -9.82212275e-02 6.58059418e-01 1.29986477e+00
7.18630731e-01 9.42061126e-01 -1.79348201e-01 1.55298364e+00
5.13263308e-02 6.96931064e-01 1.07416081e+00 -6.68024361e-01
6.05331719e-01 1.16879724e-01 -3.61642927e-01 8.07152018e-02
-2.11587772e-01 -7.18512475e-01 -9.21910048e-01 -1.61222786e-01
4.06046689e-01 -5.30910909e-01 -1.35579920e+00 2.26701999e+00
7.70726800e-02 -1.75946966e-01 5.58524430e-01 7.54871786e-01
6.85648501e-01 8.04862797e-01 3.82106155e-01 -3.30019951e-01
1.28832018e+00 -1.12258267e+00 -7.55584657e-01 -7.98890650e-01
6.77888870e-01 -8.44766974e-01 1.36803126e+00 -8.84117410e-02
-1.27229106e+00 -6.96490943e-01 -8.31055582e-01 5.26819751e-02
-3.19549531e-01 5.87301403e-02 1.76037833e-01 9.89233971e-01
-1.27739954e+00 8.37515444e-02 -6.84816957e-01 -4.61092830e-01
1.49736539e-01 4.82467502e-01 -4.20378327e-01 3.41425866e-01
-1.25676262e+00 1.01562560e+00 3.90866220e-01 -1.67467654e-01
-1.01188588e+00 -8.10073614e-01 -8.50193262e-01 2.09304333e-01
1.21519327e-01 -3.40260148e-01 1.54822624e+00 -9.77647305e-01
-2.01768374e+00 8.49454701e-01 -1.89341322e-01 -7.01474190e-01
7.67281130e-02 -3.61153126e-01 -7.98147142e-01 -2.32949764e-01
-2.02761576e-01 6.08239472e-01 7.59192348e-01 -1.01903105e+00
-4.11619961e-01 -3.39119226e-01 -3.08556229e-01 3.76708865e-01
-4.77825850e-01 2.13185817e-01 4.35450636e-02 -9.43899810e-01
-1.21914513e-01 -9.12719727e-01 -2.34285101e-01 -1.00720572e+00
-3.58940899e-01 -2.71229982e-01 5.02370179e-01 -7.10098088e-01
1.14927804e+00 -2.21381497e+00 1.74676236e-02 -5.29906200e-03
-2.72896945e-01 6.27734721e-01 -4.16282237e-01 2.72995770e-01
-1.85832500e-01 1.26255050e-01 -4.61394072e-01 -4.48392451e-01
3.79064158e-02 1.86807722e-01 -2.22482815e-01 2.75605977e-01
5.86836874e-01 6.54938936e-01 -7.86197126e-01 -6.96433857e-02
1.83200255e-01 4.73562330e-01 -6.07867241e-01 8.49152982e-01
1.43920332e-02 4.27787662e-01 -4.20776242e-03 5.24563432e-01
3.17150295e-01 3.96796465e-01 1.80070579e-01 1.81792229e-01
8.02621692e-02 7.51772940e-01 -7.35747695e-01 1.58148634e+00
-8.83883178e-01 4.09088671e-01 2.52575010e-01 -1.08311594e+00
1.40748489e+00 5.75408816e-01 -1.76165938e-01 -4.95518982e-01
-7.60388374e-02 3.60671461e-01 5.00884116e-01 -1.78938247e-02
4.69718695e-01 -5.69542706e-01 -2.06987053e-01 1.95605755e-01
6.86757147e-01 -9.50839594e-02 -1.62937820e-01 -1.40524849e-01
1.23417652e+00 -2.85464108e-01 4.49956506e-01 -5.18070579e-01
6.00809276e-01 -2.62687683e-01 5.74810982e-01 6.62341774e-01
-3.88650596e-01 7.98396885e-01 -1.09385446e-01 -2.55767584e-01
-1.05141377e+00 -1.45622969e+00 -1.41524270e-01 1.58329642e+00
-5.99513829e-01 -1.84414908e-01 -9.83179331e-01 -6.26881957e-01
-3.85514140e-01 1.18386900e+00 -2.53208637e-01 -3.57301921e-01
-9.73603725e-01 -6.55040145e-01 8.14102411e-01 6.95351303e-01
3.36895198e-01 -1.31420910e+00 2.63638884e-01 3.91896844e-01
1.43563211e-01 -1.18647528e+00 -7.70019352e-01 6.04119301e-01
-4.88649100e-01 -3.36751848e-01 -1.00398707e+00 -1.05903542e+00
5.04703164e-01 -2.35537708e-01 1.15461540e+00 -4.86016482e-01
3.28978062e-01 1.35783136e-01 -3.93045932e-01 -2.73362756e-01
-9.72967863e-01 6.82416201e-01 5.22069991e-01 2.63497949e-01
4.77575958e-01 -4.89058167e-01 -3.23664397e-01 3.44756663e-01
-4.82471287e-01 -4.39802974e-01 6.70112371e-01 9.93493438e-01
4.10329521e-01 -5.94272494e-01 1.34660816e+00 -1.16556585e+00
4.52099741e-01 -5.26764095e-01 -3.91359597e-01 1.13797197e-02
-5.00285327e-01 1.73046917e-01 9.68982995e-01 -5.88292241e-01
-1.42148900e+00 2.09922910e-01 -7.02798903e-01 -3.07489067e-01
-6.84540391e-01 1.27404124e-01 -4.53062564e-01 4.59464550e-01
8.40187132e-01 3.05933148e-01 -1.61378175e-01 -6.01610243e-01
6.50859118e-01 1.15075481e+00 8.03693593e-01 -5.87792933e-01
8.61529350e-01 -3.64051372e-01 -7.83303916e-01 -1.29389167e+00
-7.12645113e-01 -4.16169554e-01 -3.72269630e-01 1.96811378e-01
7.51850605e-01 -1.13180697e+00 -2.58844793e-01 2.48693332e-01
-1.02032566e+00 -6.36794090e-01 -5.53647995e-01 5.76232374e-01
-7.92096674e-01 -9.78943184e-02 -5.99386573e-01 -1.01245689e+00
-5.75313389e-01 -1.19339693e+00 9.44278359e-01 3.44210081e-02
-6.20382130e-01 -9.97736037e-01 1.59691766e-01 4.81744915e-01
8.80948007e-01 -5.70194662e-01 8.01851153e-01 -1.42505538e+00
-2.97884028e-02 5.07258438e-02 2.86508918e-01 7.98185587e-01
3.53339821e-01 -5.20315588e-01 -1.55089462e+00 -3.48131567e-01
-1.02074139e-01 -3.97693992e-01 5.74837744e-01 2.71782905e-01
9.06204522e-01 -5.04905164e-01 2.51936108e-01 7.01083779e-01
8.43306065e-01 3.12565476e-01 3.30121696e-01 6.60374910e-02
6.38133168e-01 6.64640009e-01 1.21916682e-01 -1.05433181e-01
1.10579342e-01 6.28944814e-01 -7.00204745e-02 -1.44146591e-01
-5.07754147e-01 -3.76278877e-01 7.80666232e-01 1.67451036e+00
7.64245458e-04 -2.45920792e-01 -9.91611302e-01 9.58703935e-01
-1.37079990e+00 -7.05417752e-01 2.97474802e-01 2.29836655e+00
1.33856499e+00 5.36714569e-02 3.34463298e-01 -1.07818358e-01
9.78368819e-01 3.16463858e-01 -4.29793209e-01 -7.01854110e-01
-2.88175792e-01 6.06454015e-01 4.39852118e-01 6.58540189e-01
-1.00286579e+00 1.56931329e+00 6.69051170e+00 7.80008256e-01
-9.55653548e-01 3.55004817e-01 8.34085703e-01 -9.89120156e-02
-1.48929968e-01 -1.33338422e-01 -8.57963145e-01 3.15605491e-01
1.82810295e+00 -2.27843598e-01 6.21475279e-01 9.47419286e-01
2.16267537e-02 6.94027543e-01 -1.12450039e+00 8.35368395e-01
-1.03063665e-01 -9.82971370e-01 -3.97076719e-02 -9.80645716e-02
6.70243561e-01 3.68494570e-01 2.12106973e-01 7.09571064e-01
7.63169169e-01 -1.07287884e+00 8.11205506e-01 -2.15503335e-01
1.16032374e+00 -8.89601409e-01 7.83407032e-01 8.77406299e-02
-8.19319725e-01 1.53909639e-01 -4.19994742e-01 1.09898195e-01
1.98418647e-01 2.66659677e-01 -1.20102382e+00 4.97132316e-02
4.24960613e-01 3.36565562e-02 -3.74276459e-01 3.24680120e-01
-2.18200684e-01 1.40782058e+00 -2.91379422e-01 1.51210185e-02
1.87323138e-01 4.86017056e-02 4.35422629e-01 1.65098894e+00
2.96430886e-01 -7.77917430e-02 -1.59581050e-01 6.36223018e-01
-4.72464561e-01 3.84964257e-01 -6.30469143e-01 -2.43944615e-01
8.32109213e-01 1.01113713e+00 -8.77190232e-02 -3.10919017e-01
-4.41148430e-01 9.13305581e-01 7.79061019e-01 5.94357967e-01
-3.83367300e-01 -4.66899961e-01 1.13925421e+00 -5.37405051e-02
2.62118518e-01 -2.08860308e-01 -2.42990956e-01 -1.05064321e+00
-2.07495198e-01 -1.05866218e+00 7.31212273e-02 -4.14376497e-01
-1.56807160e+00 1.09887218e+00 -2.29890510e-01 -7.48833656e-01
-7.40385890e-01 -6.40277982e-01 -7.32240260e-01 1.07261479e+00
-1.46038401e+00 -1.18853080e+00 3.49038482e-01 6.17649913e-01
1.26562977e+00 -9.47466910e-01 1.18215001e+00 1.03055440e-01
-6.67697847e-01 1.06661367e+00 2.83716228e-02 3.88574839e-01
7.95515060e-01 -1.53580272e+00 9.39055622e-01 1.00224268e+00
3.70350510e-01 7.42675781e-01 5.10294318e-01 -3.33926797e-01
-1.19024813e+00 -1.26961541e+00 1.01413393e+00 -3.39852482e-01
7.85797417e-01 -6.22369409e-01 -1.03607976e+00 1.00853837e+00
5.81517458e-01 8.49310309e-02 8.48025739e-01 6.93086982e-01
-4.99043405e-01 -1.75175592e-01 -8.75175118e-01 7.69961715e-01
1.27307093e+00 -7.66852856e-01 -9.50257421e-01 3.05118471e-01
1.00149429e+00 -2.03217492e-01 -9.59989190e-01 1.23943463e-01
-5.88854440e-02 -6.69316769e-01 8.25650811e-01 -8.42650294e-01
-1.07300952e-01 -3.45391557e-02 -4.22796965e-01 -2.02496219e+00
-4.73178744e-01 -1.08192146e+00 1.66657880e-01 1.83197069e+00
8.60050321e-01 -6.50610566e-01 5.46248138e-01 2.96589673e-01
-7.59845555e-01 -1.60511032e-01 -1.24718940e+00 -1.04771876e+00
4.83171999e-01 -4.86928403e-01 7.39886642e-01 8.97597671e-01
7.33284205e-02 9.57740724e-01 -1.76908255e-01 2.49595344e-01
3.20662707e-01 -5.78863740e-01 6.54464900e-01 -7.77003527e-01
-2.83647746e-01 -4.36979175e-01 -3.39625239e-01 -8.75224471e-01
5.85986555e-01 -1.11870539e+00 4.11554396e-01 -7.48440146e-01
-2.65395999e-01 -4.20486629e-01 -3.48248303e-01 4.40398425e-01
-3.64121348e-01 -5.65887876e-02 1.02950901e-01 -9.13672969e-02
-1.73828691e-01 7.89919555e-01 6.54106319e-01 -2.03053758e-01
-1.53125018e-01 1.74161643e-01 -8.30997109e-01 6.28729999e-01
1.12191439e+00 -3.50040823e-01 -5.42528212e-01 -5.74912310e-01
-4.69727457e-01 -1.28050908e-01 -3.90885383e-01 -9.04629171e-01
3.89619134e-02 -1.26541167e-01 1.20078631e-01 1.31361671e-02
4.61913824e-01 -4.58902150e-01 -2.06620336e-01 -2.07982790e-02
-7.44699419e-01 1.44986212e-01 9.23440680e-02 2.47507825e-01
-4.17955399e-01 -1.89222574e-01 1.08132863e+00 -1.49775043e-01
-5.41201353e-01 1.24914825e-01 -4.96121109e-01 5.32755733e-01
5.11152089e-01 2.38595366e-01 -3.78599495e-01 -5.15258431e-01
-7.29237378e-01 -2.56194472e-01 1.59123197e-01 6.01085663e-01
4.39541489e-01 -1.28356433e+00 -1.25154662e+00 7.05753624e-01
2.80478120e-01 -2.24245749e-02 7.95724615e-02 2.37157136e-01
-1.18719108e-01 4.15545970e-01 5.61211221e-02 -3.05743068e-01
-8.19052696e-01 4.07842249e-01 2.80257255e-01 9.25686434e-02
-3.45186204e-01 1.08557665e+00 5.89546263e-01 -7.54306257e-01
1.69294953e-01 -6.48175701e-02 9.66438875e-02 -3.94350886e-01
4.86913890e-01 1.59188107e-01 3.90973330e-01 -9.98871803e-01
-4.29351270e-01 1.00682482e-01 -2.52779216e-01 -3.75103652e-01
1.26848292e+00 -2.20919430e-01 5.58815897e-01 4.48077232e-01
1.11809337e+00 2.62156844e-01 -1.08532441e+00 -2.06698596e-01
1.11168616e-01 5.76912798e-02 2.26419494e-01 -7.85664618e-01
-1.01716650e+00 9.08412397e-01 5.13378978e-01 2.41607666e-01
9.04192448e-01 3.01979899e-01 7.98604131e-01 3.32533062e-01
5.19248918e-02 -1.27495062e+00 -3.46359402e-01 8.70718718e-01
1.01613808e+00 -1.21884048e+00 -7.39794672e-01 -2.31813118e-01
-9.63999987e-01 7.36667812e-01 7.35108495e-01 2.02315003e-02
4.65009570e-01 3.82569313e-01 5.63591599e-01 2.05900252e-01
-9.10256624e-01 -5.41957974e-01 1.70175865e-01 1.05447042e+00
1.03221357e+00 3.64698201e-01 3.35596986e-02 6.10423505e-01
-8.05526137e-01 -6.63269937e-01 2.40814433e-01 4.48849916e-01
-3.77118111e-01 -1.05234528e+00 -3.10008973e-01 1.61863640e-01
-4.43611115e-01 -4.04167652e-01 -3.43682110e-01 5.65459251e-01
-4.38392878e-01 1.23534477e+00 2.99360782e-01 -4.42219317e-01
5.54376662e-01 5.43412149e-01 4.29179482e-02 -9.76110458e-01
-8.93645167e-01 2.04962656e-01 5.49458921e-01 -1.98691502e-01
-2.36513745e-02 -6.34188116e-01 -1.08464634e+00 -2.19737599e-03
-1.00531764e-01 1.70203596e-01 6.80561960e-01 9.31169331e-01
2.83719867e-01 5.34677386e-01 7.05366611e-01 -5.17337263e-01
-1.00416088e+00 -1.46169949e+00 -5.97180188e-01 6.30270660e-01
4.11783993e-01 -4.00108308e-01 -4.18004215e-01 1.97159111e-01] | [14.385852813720703, 6.723005294799805] |
91180242-2cf0-44e1-96fb-8a403945c447 | accurate-and-efficient-stereo-matching-via | 2209.12699 | null | https://arxiv.org/abs/2209.12699v2 | https://arxiv.org/pdf/2209.12699v2.pdf | Accurate and Efficient Stereo Matching via Attention Concatenation Volume | Stereo matching is a fundamental building block for many vision and robotics applications. An informative and concise cost volume representation is vital for stereo matching of high accuracy and efficiency. In this paper, we present a novel cost volume construction method, named attention concatenation volume (ACV), which generates attention weights from correlation clues to suppress redundant information and enhance matching-related information in the concatenation volume. The ACV can be seamlessly embedded into most stereo matching networks, the resulting networks can use a more lightweight aggregation network and meanwhile achieve higher accuracy. We further design a fast version of ACV to enable real-time performance, named Fast-ACV, which generates high likelihood disparity hypotheses and the corresponding attention weights from low-resolution correlation clues to significantly reduce computational and memory cost and meanwhile maintain a satisfactory accuracy. The core idea of our Fast-ACV is volume attention propagation (VAP) which can automatically select accurate correlation values from an upsampled correlation volume and propagate these accurate values to the surroundings pixels with ambiguous correlation clues. Furthermore, we design a highly accurate network ACVNet and a real-time network Fast-ACVNet based on our ACV and Fast-ACV respectively, which achieve the state-of-the-art performance on several benchmarks (i.e., our ACVNet ranks the 2nd on KITTI 2015 and Scene Flow, and the 3rd on KITTI 2012 and ETH3D among all the published methods; our Fast-ACVNet outperforms almost all state-of-the-art real-time methods on Scene Flow, KITTI 2012 and 2015 and meanwhile has better generalization ability) | ['Xin Yang', 'Jinhui Tang', 'Junda Cheng', 'Yun Wang', 'Gangwei Xu'] | 2022-09-23 | null | null | null | null | ['stereo-matching-1'] | ['computer-vision'] | [-1.38662560e-02 -4.83474374e-01 -5.11564966e-03 -1.03370331e-01
-3.61005694e-01 -6.73745526e-04 3.60837758e-01 -3.60466987e-01
-5.48604190e-01 7.44599342e-01 2.80298263e-01 -1.55107081e-01
-2.60736018e-01 -1.01701295e+00 -7.16504276e-01 -5.32548428e-01
1.46106735e-01 4.11328673e-01 6.69307470e-01 -3.54967862e-01
3.25277925e-01 4.04560417e-01 -1.84146261e+00 -3.96871418e-02
1.29310822e+00 1.20815766e+00 4.46202427e-01 3.64275575e-01
-2.26151481e-01 8.02906632e-01 -3.06020826e-01 -3.66933167e-01
5.57800114e-01 -1.28191262e-01 -6.89560473e-01 -2.64311314e-01
9.54194665e-01 -4.65881079e-01 -7.83834636e-01 1.08070588e+00
6.96230233e-01 3.96694899e-01 4.63564247e-01 -1.34482563e+00
-4.88497466e-01 4.24632907e-01 -9.22478318e-01 3.99757236e-01
1.20789222e-01 5.57691514e-01 8.42429578e-01 -1.02698815e+00
6.89515531e-01 1.52874720e+00 7.04219759e-01 3.24566007e-01
-1.06916475e+00 -1.12304723e+00 3.36321950e-01 5.78525245e-01
-1.35362971e+00 -1.55577108e-01 9.01860356e-01 -5.30113399e-01
8.42470467e-01 8.19564313e-02 1.01976013e+00 7.06273973e-01
-9.07619484e-03 8.20667565e-01 8.08169305e-01 1.73439190e-01
-1.28538713e-01 -3.80032331e-01 6.07330985e-02 9.03834760e-01
2.81153709e-01 4.72930074e-01 -6.18671238e-01 2.90889710e-01
1.31837201e+00 4.16771472e-02 -7.68119454e-01 -2.81760216e-01
-1.55153704e+00 7.81761289e-01 1.13086855e+00 4.13315855e-02
-4.03843880e-01 3.34288299e-01 6.36359811e-01 -1.02323422e-04
1.52374223e-01 2.42827073e-01 -2.39287212e-01 -2.00187396e-02
-6.06492698e-01 2.69069970e-01 2.30425894e-01 1.07725155e+00
1.02068329e+00 1.81932539e-01 -6.03821814e-01 8.29794466e-01
9.83627886e-02 5.96047640e-01 3.46807092e-01 -1.20891738e+00
8.73861849e-01 6.93925560e-01 -1.96857855e-01 -1.28198647e+00
-3.52632374e-01 -5.45032561e-01 -1.20216310e+00 2.11114570e-01
4.28334773e-02 1.22219861e-01 -8.58529806e-01 1.70140600e+00
3.15796793e-01 5.97172499e-01 -5.46030328e-02 1.34653437e+00
1.39187932e+00 8.09209287e-01 -8.87112543e-02 -5.92775084e-02
1.04742384e+00 -1.28285635e+00 -5.03104627e-01 -2.81054348e-01
2.28875712e-01 -6.85867488e-01 1.16781497e+00 2.35254347e-01
-9.36094701e-01 -1.07529902e+00 -1.20694637e+00 -4.59587216e-01
-2.30608340e-02 -7.22032487e-02 1.02148056e+00 3.19488160e-02
-9.13758278e-01 6.71968997e-01 -3.44208896e-01 8.85489210e-03
7.91019499e-01 3.41023743e-01 -3.45219284e-01 -4.04263765e-01
-1.15960062e+00 6.21599376e-01 3.72042984e-01 3.24929804e-01
-4.59774017e-01 -1.07111168e+00 -1.30671060e+00 -6.80354014e-02
2.35214368e-01 -9.95521963e-01 8.73868525e-01 -6.32484674e-01
-1.38660276e+00 6.61295235e-01 -2.33436882e-01 -3.53836656e-01
7.69433200e-01 -7.18206614e-02 -1.24408059e-01 8.04424509e-02
2.63510525e-01 1.24082637e+00 4.33971554e-01 -9.75797594e-01
-8.57601821e-01 -1.54166609e-01 -1.72084525e-01 2.81111836e-01
-1.15365185e-01 -5.47761202e-01 -6.82691991e-01 -6.02824152e-01
3.11227292e-01 -5.18703938e-01 -2.35750988e-01 4.72506821e-01
-2.87479550e-01 -3.00266147e-01 1.04586101e+00 -4.55883831e-01
1.13872349e+00 -1.95042741e+00 2.83190876e-01 7.21381307e-02
4.24841046e-01 5.09572208e-01 -2.17416242e-01 -2.73490578e-01
7.10388869e-02 -2.19956324e-01 -3.75379056e-01 -2.96293646e-01
-2.60447234e-01 2.32117742e-01 -2.55910337e-01 4.86872047e-01
1.39701635e-01 1.06063664e+00 -1.14223659e+00 -7.92967916e-01
1.00665843e+00 6.96006536e-01 -7.87288010e-01 1.81405395e-01
8.99040401e-02 4.16114271e-01 -4.30750132e-01 5.47649205e-01
9.80513930e-01 -1.44253224e-01 -5.48159063e-01 -6.17544174e-01
-3.43628615e-01 -4.80490662e-02 -1.25535548e+00 1.77055013e+00
-5.43046415e-01 1.04119229e+00 -5.08326031e-02 -7.53460348e-01
1.22490776e+00 -2.54643530e-01 4.53832924e-01 -1.08735156e+00
2.94151366e-01 3.02277774e-01 -7.09338114e-02 -3.93110275e-01
6.02415025e-01 3.67526412e-01 1.83486342e-01 -1.64867997e-01
-1.97159294e-02 -1.54347032e-01 2.87714452e-01 1.48561954e-01
6.02346778e-01 1.84788436e-01 4.73303683e-02 -4.33446348e-01
1.18996048e+00 -1.05821222e-01 1.13123405e+00 4.20706242e-01
-5.49531102e-01 7.62962103e-01 1.87954903e-01 -7.33389139e-01
-9.87831533e-01 -9.73265290e-01 -2.09971979e-01 5.10144293e-01
8.83327782e-01 -1.78815469e-01 -3.24933290e-01 -4.45082605e-01
1.49329767e-01 2.86276251e-01 -5.30198574e-01 -1.26181379e-01
-8.90427709e-01 -2.89568901e-01 1.24244571e-01 7.41096735e-01
1.40349448e+00 -1.33069801e+00 -4.86055046e-01 2.63793707e-01
-5.68267465e-01 -1.13737416e+00 -8.63038063e-01 -4.61617053e-01
-6.71542823e-01 -1.21787441e+00 -7.91274905e-01 -8.95873368e-01
5.72515488e-01 5.14290631e-01 1.21867669e+00 3.13761741e-01
-4.16197330e-01 -3.07521075e-01 3.23462207e-03 -2.13339299e-01
2.98326939e-01 -4.33894619e-02 -2.56015182e-01 6.68347031e-02
1.48101017e-01 -6.15075827e-01 -1.01227713e+00 3.96144420e-01
-6.35839760e-01 4.87779140e-01 4.35145408e-01 9.93916810e-01
7.49515474e-01 -3.31106871e-01 2.65750051e-01 -4.71648246e-01
2.42083430e-01 -5.33772036e-02 -1.04032254e+00 -1.33789614e-01
-3.77971679e-01 1.52847528e-01 7.43367910e-01 -2.86937892e-01
-8.80658090e-01 -4.89402823e-02 -3.75403136e-01 -9.08245027e-01
3.81414384e-01 -2.97021661e-02 -1.04876600e-01 -4.73654568e-01
3.93842846e-01 2.26958111e-01 -1.38861015e-01 -1.37689427e-01
3.05924475e-01 3.06073070e-01 9.97651935e-01 -3.53159338e-01
7.13325500e-01 5.35590351e-01 3.56961757e-01 -5.38531721e-01
-5.68474293e-01 -5.56085646e-01 -4.36350524e-01 -4.83827859e-01
8.93660963e-01 -8.32118809e-01 -1.11151683e+00 7.64133334e-01
-1.35065222e+00 -2.61730760e-01 -1.78503782e-01 6.07504904e-01
-4.70480055e-01 4.28716749e-01 -5.69306970e-01 -5.90229273e-01
-7.67847776e-01 -1.57523608e+00 1.09408629e+00 6.85013354e-01
1.34561718e-01 -8.52037787e-01 -1.97695151e-01 1.86016932e-01
4.48910594e-01 2.81818330e-01 5.24964392e-01 1.33230001e-01
-1.01564336e+00 3.75507593e-01 -9.59141850e-01 1.52521938e-01
-1.82485625e-01 1.45177171e-01 -9.69938338e-01 -1.67066991e-01
-5.45400620e-01 -1.17413484e-01 1.32517338e+00 7.28745461e-01
1.37162030e+00 6.96136951e-02 -3.30153733e-01 1.31806993e+00
1.68469334e+00 2.03732729e-01 9.18227255e-01 5.32886982e-01
1.01194096e+00 5.85031271e-01 6.47447646e-01 1.79104879e-01
5.09141386e-01 7.32478142e-01 5.65586925e-01 -3.16630006e-01
-4.76592451e-01 -4.33972657e-01 -7.68057853e-02 7.67850459e-01
-2.15297982e-01 -2.30828114e-02 -5.29942274e-01 6.02292418e-01
-1.97472107e+00 -1.05399013e+00 -4.47636932e-01 2.27722549e+00
4.71981883e-01 1.83784336e-01 -1.05528884e-01 1.79154888e-01
8.70984137e-01 4.89830703e-01 -5.75288236e-01 -1.15382649e-01
-2.78938711e-01 2.03336790e-01 6.49725795e-01 8.14378738e-01
-1.14567196e+00 1.04084885e+00 4.99238491e+00 9.86702323e-01
-1.09642649e+00 -1.18639588e-01 6.74887359e-01 1.10648461e-01
-3.54928225e-01 -1.08742580e-01 -7.34730065e-01 6.61160529e-01
-7.81583264e-02 -4.73746024e-02 3.48782003e-01 8.57896447e-01
1.81422114e-01 -2.19645072e-02 -8.32839906e-01 1.50349104e+00
-6.30041659e-02 -1.75599182e+00 2.32545659e-01 -1.36568919e-01
7.91962683e-01 1.73110217e-01 -1.23412170e-01 3.21629733e-01
1.52279675e-01 -8.52974474e-01 4.57369685e-01 4.07347620e-01
9.31220591e-01 -8.44086647e-01 1.04060566e+00 -7.65403360e-02
-1.71618581e+00 -1.21202357e-01 -6.85934901e-01 5.84551468e-02
3.80503327e-01 6.26986146e-01 1.03038261e-02 7.19298959e-01
8.22085500e-01 1.31974113e+00 -3.87146354e-01 1.48642838e+00
-1.42207831e-01 9.56130400e-03 -2.54672229e-01 -9.81765315e-02
4.68412817e-01 -4.30992663e-01 6.65303409e-01 1.04893661e+00
2.88193345e-01 -1.11609173e-03 2.65462101e-01 9.03256893e-01
-7.43715763e-02 1.09776862e-01 -4.27321047e-01 7.98720896e-01
7.00132966e-01 1.17583251e+00 -4.98979896e-01 -5.38838744e-01
-2.71844327e-01 8.01084042e-01 3.49020600e-01 2.48631418e-01
-9.03462410e-01 -7.03715384e-01 9.74618673e-01 -2.21423209e-01
2.55178392e-01 1.64643258e-01 -3.93924147e-01 -1.15007675e+00
1.36152998e-01 -3.91987622e-01 4.27216560e-01 -1.12402236e+00
-1.24234951e+00 8.85427058e-01 -2.21458226e-01 -1.54603314e+00
1.62435532e-01 -4.76870745e-01 -8.46077323e-01 1.01522350e+00
-1.88235497e+00 -8.00715208e-01 -1.16484690e+00 7.42747962e-01
7.42760777e-01 -6.92997351e-02 6.60095364e-02 5.63099146e-01
-6.14905536e-01 4.86579746e-01 -3.59323978e-01 1.54962763e-01
5.52648902e-01 -8.70992064e-01 5.40785134e-01 9.84479845e-01
-1.50897801e-01 1.74362302e-01 3.37412536e-01 -5.02644658e-01
-1.02328718e+00 -1.43728960e+00 6.56008959e-01 5.09368815e-02
2.40010738e-01 -3.31403464e-02 -9.86369848e-01 3.47800761e-01
1.94139108e-02 3.37409407e-01 -3.22674781e-01 -4.89626706e-01
-4.13084269e-01 -5.31241000e-01 -1.15135741e+00 7.20786810e-01
1.75804222e+00 -2.13347390e-01 -5.32638788e-01 1.00465842e-01
1.16341817e+00 -7.24809945e-01 -4.45213616e-01 6.92881644e-01
4.20922726e-01 -1.40987682e+00 1.24837291e+00 7.43359998e-02
6.74025118e-01 -6.11010075e-01 6.06885180e-02 -1.07815528e+00
-5.89772165e-01 -6.37035787e-01 3.32725160e-02 1.14270473e+00
7.52953738e-02 -9.39925015e-01 8.03098202e-01 2.55056888e-01
-5.04199684e-01 -9.54455078e-01 -9.74918902e-01 -5.54059029e-01
-3.15436929e-01 -4.84938502e-01 8.05095851e-01 7.74051428e-01
-5.40428400e-01 1.94918886e-01 -3.20362836e-01 -8.51148292e-02
9.68930483e-01 2.45088577e-01 7.91951537e-01 -1.23956096e+00
1.16116300e-01 -8.83904934e-01 -9.39123988e-01 -1.59455657e+00
-2.20301021e-02 -7.43418276e-01 1.52990809e-02 -1.65546465e+00
8.05633888e-02 -5.57153761e-01 -6.68027392e-03 7.28686228e-02
-4.11690384e-01 3.40969175e-01 2.85495371e-01 2.11870611e-01
-3.17421108e-01 9.10245836e-01 1.79626131e+00 -2.29718238e-01
-4.11973983e-01 -2.36457482e-01 -3.62669766e-01 6.20964706e-01
4.29103076e-01 6.04288988e-02 -5.31619728e-01 -5.32113016e-01
-2.74300098e-01 3.58232036e-02 5.24235487e-01 -1.31869888e+00
4.25184727e-01 -9.57583562e-02 4.67589021e-01 -1.06284046e+00
3.06047171e-01 -6.27255976e-01 -4.34518494e-02 6.66124642e-01
-9.24108103e-02 1.81167752e-01 1.59443915e-01 3.96592081e-01
-4.45790261e-01 2.36038625e-01 8.71083260e-01 2.77847759e-02
-1.32500565e+00 1.03872705e+00 3.32581758e-01 3.44671130e-01
9.43193138e-01 -4.76118654e-01 -5.92930317e-01 -1.92568988e-01
4.37013851e-03 7.02212334e-01 2.79196233e-01 5.96171260e-01
9.51928973e-01 -1.68421936e+00 -7.43672907e-01 5.04592955e-01
1.09963000e-01 6.49755418e-01 6.34090900e-01 6.94265485e-01
-1.00950491e+00 4.62878108e-01 -5.20381033e-01 -9.38969195e-01
-1.05674720e+00 5.33404708e-01 5.35935760e-01 -5.32729141e-02
-1.00843656e+00 1.15395534e+00 5.66615939e-01 -2.50196010e-01
3.02788585e-01 -3.65029722e-01 -4.14856941e-01 -2.93215930e-01
6.13593400e-01 5.53666413e-01 -2.11712644e-01 -7.65642107e-01
-3.58508170e-01 1.20756924e+00 2.62193948e-01 1.51576459e-01
1.00560176e+00 1.15894675e-02 -4.60738093e-02 -2.06666157e-01
1.48630345e+00 -3.07988733e-01 -1.65400529e+00 -3.54793340e-01
-6.81564927e-01 -1.05584717e+00 2.48540282e-01 -1.85657158e-01
-1.78081000e+00 1.03959990e+00 5.20825148e-01 -4.61247683e-01
1.21989632e+00 -2.93340534e-01 1.23819721e+00 2.47137416e-02
4.36579198e-01 -8.49634230e-01 1.20836377e-01 4.17384267e-01
1.02922690e+00 -1.27433777e+00 8.82355422e-02 -7.55313873e-01
-5.88872373e-01 1.03665841e+00 1.35148156e+00 -3.87858272e-01
3.86374176e-01 -4.42353599e-02 -3.96601669e-02 -8.64615664e-02
-3.81832242e-01 -3.76845837e-01 4.57516760e-01 9.01459932e-01
2.84443386e-02 -2.82610893e-01 -2.45514542e-01 1.96453691e-01
-2.67258853e-01 -6.77732304e-02 1.83246270e-01 4.08971220e-01
-3.00435871e-01 -5.86260319e-01 -1.94140062e-01 5.44435620e-01
9.87487808e-02 -2.19527781e-01 -6.97168857e-02 8.96874905e-01
3.21143001e-01 6.35426879e-01 5.19015610e-01 -7.54166901e-01
6.40273094e-01 -7.19178915e-01 2.32268840e-01 -1.88380703e-02
-4.89026189e-01 -2.26691619e-01 -3.82272117e-02 -1.01680160e+00
-4.50745285e-01 -2.88642317e-01 -1.24648333e+00 -6.81494415e-01
-2.05480650e-01 -1.59948722e-01 1.20455943e-01 7.74773777e-01
3.43081594e-01 6.44119918e-01 6.45131111e-01 -1.20630276e+00
1.47258893e-01 -7.01596975e-01 -5.76778986e-02 5.21849036e-01
3.81922483e-01 -9.68758881e-01 -3.23538721e-01 -3.89497519e-01] | [8.916604042053223, -2.2264482975006104] |
d9877afb-2320-407a-ba6c-ca4468b8afdc | spiking-sampling-network-for-image-sparse | 2211.04166 | null | https://arxiv.org/abs/2211.04166v1 | https://arxiv.org/pdf/2211.04166v1.pdf | Spiking sampling network for image sparse representation and dynamic vision sensor data compression | Sparse representation has attracted great attention because it can greatly save storage re- sources and find representative features of data in a low-dimensional space. As a result, it may be widely applied in engineering domains including feature extraction, compressed sensing, signal denoising, picture clustering, and dictionary learning, just to name a few. In this paper, we propose a spiking sampling network. This network is composed of spiking neurons, and it can dynamically decide which pixel points should be retained and which ones need to be masked according to the input. Our experiments demonstrate that this approach enables better sparse representation of the original image and facilitates image reconstruction compared to random sampling. We thus use this approach for compressing massive data from the dynamic vision sensor, which greatly reduces the storage requirements for event data. | ['Yilei Zhang', 'Chunming Jiang'] | 2022-11-08 | null | null | null | null | ['data-compression'] | ['time-series'] | [ 5.63312411e-01 -5.73401511e-01 1.36654660e-01 -5.73202111e-02
-1.84260145e-01 -2.31045306e-01 1.78498238e-01 2.48146690e-02
-4.96616960e-01 6.77680314e-01 7.54194781e-02 3.60362619e-01
-8.11435133e-02 -7.82569230e-01 -5.45243204e-01 -1.07281625e+00
1.71795383e-01 -6.74642669e-03 4.12954777e-01 1.58182129e-01
5.81561089e-01 6.79389000e-01 -1.91701901e+00 1.20306209e-01
4.53456402e-01 1.35670507e+00 9.37820613e-01 -2.33365353e-02
1.83397140e-02 5.48148394e-01 -4.07674551e-01 3.49871933e-01
1.25117913e-01 -4.15157855e-01 -2.13354118e-02 1.75351039e-01
-1.59366503e-02 -1.94703519e-01 -8.36236417e-01 1.34274864e+00
5.70313334e-01 5.90434559e-02 3.83158952e-01 -8.20520461e-01
-5.45647383e-01 5.30241609e-01 -5.73116779e-01 4.55221385e-01
8.83184001e-03 9.00793597e-02 3.74142766e-01 -7.97500134e-01
7.33021975e-01 9.84970212e-01 2.92887241e-01 4.29301679e-01
-1.17999828e+00 -6.35029316e-01 -1.24066144e-01 4.79875624e-01
-1.43886364e+00 -7.28842318e-01 1.04415834e+00 -5.70702627e-02
6.93184316e-01 5.02929762e-02 1.02348769e+00 7.04470575e-01
3.18179488e-01 5.62243223e-01 9.90742207e-01 -2.36828193e-01
7.09277749e-01 -4.88676757e-01 1.10667320e-02 5.52991807e-01
4.55420285e-01 -1.05709910e-01 -8.30587924e-01 -9.23915356e-02
1.02623928e+00 7.12227404e-01 -6.00118220e-01 -2.76152849e-01
-1.40147650e+00 5.44486523e-01 3.00185472e-01 5.23767948e-01
-6.51980817e-01 2.42337689e-01 2.32482418e-01 2.55859822e-01
-1.08475611e-01 1.33885279e-01 9.11205485e-02 -1.64274067e-01
-9.36663926e-01 2.12573275e-01 4.92254198e-01 7.12786555e-01
7.43982553e-01 4.19330835e-01 4.59632762e-02 8.76355886e-01
-3.09518576e-02 7.49900699e-01 1.01092017e+00 -1.34313905e+00
-1.10859172e-02 5.95033824e-01 -1.52464494e-01 -1.36987507e+00
-2.75714636e-01 -1.27241373e-01 -1.47128606e+00 4.79673184e-02
-1.10319801e-01 1.21094741e-01 -7.48269022e-01 1.49736023e+00
-3.73091809e-02 4.57736909e-01 -5.51238935e-03 9.91837144e-01
6.40176535e-01 7.00481236e-01 -4.32012320e-01 -3.60478759e-01
1.16109335e+00 -2.15565383e-01 -7.60786772e-01 -2.54998386e-01
-1.48193598e-01 -5.68181872e-01 6.55072987e-01 6.31853640e-01
-1.03074861e+00 -4.95800495e-01 -1.02896798e+00 1.02964245e-01
-1.18299484e-01 1.75006092e-01 6.52055264e-01 -5.44867404e-02
-9.50363338e-01 7.13382185e-01 -9.79384601e-01 -2.18335956e-01
7.34884501e-01 4.10996079e-01 -3.13280314e-01 -4.53770965e-01
-6.79320395e-01 3.47597510e-01 2.24140614e-01 1.61374379e-02
-8.45566213e-01 -1.24682732e-01 -7.84599125e-01 3.81646454e-01
-4.25394624e-02 -4.14358765e-01 6.37319207e-01 -3.16313922e-01
-1.18690062e+00 5.66386998e-01 -5.66845536e-01 -6.20351493e-01
-2.94487447e-01 4.33070540e-01 -3.31080645e-01 4.33483213e-01
6.65528923e-02 5.71729004e-01 1.13332093e+00 -7.69839168e-01
-3.93410653e-01 -7.44782567e-01 -5.00617325e-01 -1.20610399e-02
-5.12964725e-01 -2.52243906e-01 -2.88256407e-01 -7.71605313e-01
7.26968646e-01 -6.11902118e-01 -4.95359659e-01 2.75398016e-01
-7.52109960e-02 1.23228513e-01 1.23039269e+00 -3.07296515e-01
1.03772092e+00 -2.60563588e+00 2.59381980e-01 2.79194742e-01
4.10286874e-01 1.27011791e-01 -2.45525539e-02 2.68616468e-01
2.12884456e-01 -3.24363500e-01 -4.41334367e-01 -2.37260133e-01
-5.54025292e-01 3.23892891e-01 -3.79136503e-01 6.98390782e-01
1.32853221e-02 6.85574949e-01 -4.80343461e-01 -3.41550916e-01
2.96796352e-01 5.58187962e-01 -3.52330685e-01 5.08490391e-02
9.83039662e-02 5.08094847e-01 -5.68857729e-01 7.14811862e-01
6.21540785e-01 -5.13140082e-01 -6.16584420e-02 -3.63231510e-01
-2.66936660e-01 -1.50730386e-01 -1.33798254e+00 1.90151608e+00
-3.32376696e-02 6.60393715e-01 2.84941465e-01 -1.42926967e+00
1.37020969e+00 2.56275423e-02 7.50870526e-01 -9.86420095e-01
3.23364496e-01 1.79398805e-01 3.19654215e-03 -4.73061204e-01
2.90101916e-01 7.18016103e-02 1.21319279e-01 3.81741166e-01
-1.41865909e-01 -2.71007456e-02 1.17289178e-01 1.34945437e-01
1.27759731e+00 -4.86232132e-01 3.66660595e-01 -1.68975934e-01
3.41495901e-01 -3.33229899e-02 7.91722834e-01 4.35551375e-01
-7.58727267e-02 4.99767840e-01 1.67688072e-01 -3.86084616e-01
-9.34569120e-01 -6.86328769e-01 -9.31340083e-02 2.65276790e-01
5.73906243e-01 -9.69051719e-02 -4.97087419e-01 3.24035317e-01
5.57471439e-02 8.24766830e-02 -3.78244579e-01 -4.39146489e-01
-5.69172621e-01 -3.98233086e-01 9.11143795e-02 1.85326412e-01
7.05354035e-01 -1.23820293e+00 -1.10926092e+00 4.14887071e-01
-1.51654050e-01 -9.26125050e-01 -2.58596182e-01 4.68314052e-01
-1.16317284e+00 -7.92888165e-01 -7.09722638e-01 -1.11743855e+00
8.43290210e-01 7.34071255e-01 6.03647649e-01 -4.66222167e-02
-5.44679999e-01 2.27902997e-02 -3.15264553e-01 -2.05378264e-01
2.83376038e-01 -3.64259392e-01 1.99179754e-01 1.47032097e-01
5.38850129e-01 -1.02300501e+00 -7.43792892e-01 1.26234502e-01
-1.11880803e+00 -1.60736874e-01 6.29739702e-01 9.24289107e-01
1.25163102e+00 1.72688112e-01 6.68059289e-01 -6.58121109e-01
5.75151980e-01 -3.74151438e-01 -6.32138252e-01 -1.52318045e-01
-2.18478501e-01 5.89465313e-02 9.31076825e-01 -5.84688067e-01
-4.18793410e-01 4.39361274e-01 2.19954863e-01 -7.02274561e-01
-8.07695538e-02 5.07203639e-01 -2.34042555e-01 -4.77329582e-01
3.14919591e-01 8.84896219e-01 4.62351471e-01 -6.29286766e-01
-1.45456210e-01 5.66628575e-01 5.90310514e-01 -1.28645286e-01
4.05603558e-01 7.90695071e-01 2.56625891e-01 -9.39261019e-01
-8.49726498e-02 -4.21183765e-01 -2.83262372e-01 -2.21904907e-02
4.09541637e-01 -9.39827144e-01 -7.47577012e-01 5.64340115e-01
-9.58604693e-01 9.95732918e-02 -4.84604865e-01 4.05523181e-01
-5.32076478e-01 4.55239028e-01 -5.20572007e-01 -5.32685697e-01
-2.23211408e-01 -1.11805928e+00 7.70674050e-01 3.03745806e-01
1.06147520e-01 -3.37752849e-01 -2.30412036e-01 -2.18215942e-01
5.33369303e-01 2.66115397e-01 6.92011297e-01 -6.05014339e-02
-8.80209982e-01 -1.59568563e-01 -1.15701862e-01 1.13202490e-01
2.41705820e-01 -4.81888890e-01 -5.97571254e-01 -5.29453695e-01
5.45797110e-01 -2.72294760e-01 1.10899746e+00 6.96141779e-01
1.67985451e+00 -2.70408869e-01 -4.87373441e-01 8.84222984e-01
1.55568564e+00 2.53114194e-01 7.27712154e-01 8.51463899e-02
5.17435133e-01 2.35707939e-01 4.32113558e-01 8.13822448e-01
4.86995354e-02 4.41011101e-01 6.26263559e-01 1.37938187e-01
-7.11390153e-02 -2.17625238e-02 1.73350126e-01 1.09591913e+00
2.47376785e-01 6.92387437e-03 -4.33175504e-01 6.68079317e-01
-1.78814149e+00 -1.13028109e+00 2.70176619e-01 1.98523164e+00
6.40736878e-01 -2.60742426e-01 -1.63624793e-01 3.46341878e-01
8.49436462e-01 2.81335622e-01 -1.13835704e+00 1.19628794e-01
-5.36547244e-01 2.45497465e-01 3.93298209e-01 -7.33993948e-02
-6.10703468e-01 6.85380995e-01 6.17198086e+00 8.86028707e-01
-1.47714233e+00 -8.61993507e-02 4.64581430e-01 -2.10838765e-01
-2.62228101e-01 -2.26006478e-01 -6.01928890e-01 8.66789997e-01
6.13021970e-01 -2.90727556e-01 8.50732088e-01 7.04001129e-01
3.27183783e-01 -2.93172270e-01 -7.43236542e-01 1.77628684e+00
1.16597466e-01 -1.65240169e+00 2.92322993e-01 2.02561066e-01
6.24067068e-01 6.36208355e-02 9.82190669e-02 -5.13942778e-01
-1.23338379e-01 -9.58548903e-01 3.73350173e-01 6.35262370e-01
8.18031430e-01 -7.39627421e-01 4.93797392e-01 5.42988300e-01
-1.09230673e+00 -3.07716042e-01 -9.55412447e-01 -1.77309960e-01
2.33859286e-01 1.14621770e+00 -1.97005674e-01 -2.12685540e-02
7.46851087e-01 1.04071271e+00 -8.19154754e-02 1.26475739e+00
3.80198568e-01 3.96186739e-01 -4.89475876e-01 -2.10222200e-01
-6.38599843e-02 -3.65653694e-01 5.12271762e-01 5.05703866e-01
8.11706960e-01 4.20995265e-01 1.49122223e-01 8.15899193e-01
-1.09410003e-01 -1.69595122e-01 -9.58753824e-01 2.31249933e-03
8.82356703e-01 1.05330360e+00 -9.27619338e-01 -2.44248018e-01
-2.92869836e-01 9.51987624e-01 1.87648669e-01 2.67255187e-01
-2.31431648e-01 -5.75168788e-01 5.95556676e-01 1.40650824e-01
7.32446730e-01 -3.81117076e-01 -3.01050365e-01 -1.18357325e+00
5.71950637e-02 -7.63927877e-01 -8.30450207e-02 -6.61202073e-01
-9.67680931e-01 4.76079702e-01 -5.08034587e-01 -1.46976936e+00
-8.34528878e-02 -4.59581614e-01 -4.35810983e-01 7.40505636e-01
-1.42557406e+00 -3.70055646e-01 -5.59066176e-01 1.09995389e+00
4.04701382e-01 -4.60830003e-01 7.51124859e-01 2.75063753e-01
-3.63926142e-01 3.21863708e-03 4.63647097e-01 -1.40591308e-01
2.72277087e-01 -5.08932173e-01 3.84902884e-03 8.93396258e-01
1.83077767e-01 5.96634269e-01 4.73759621e-01 -5.72439432e-01
-1.83001924e+00 -9.28981364e-01 4.47410673e-01 6.02058411e-01
3.60823005e-01 -9.57964584e-02 -8.89433563e-01 1.73378870e-01
-1.67939067e-01 3.32599461e-01 5.72642386e-01 -7.75760531e-01
-2.67376658e-02 -4.84262228e-01 -1.38715088e+00 4.44522381e-01
9.08185899e-01 -2.61054516e-01 -2.15373829e-01 1.32623747e-01
4.09195721e-01 -1.41992308e-02 -7.13649631e-01 6.83684498e-02
3.68999630e-01 -9.32586968e-01 9.17663395e-01 7.18543828e-02
2.37338379e-01 -4.82146829e-01 -4.46890712e-01 -1.30993080e+00
-6.20936155e-01 -4.02801156e-01 -3.07924122e-01 8.86403024e-01
-2.10756838e-01 -6.33873522e-01 8.64825547e-01 1.10259324e-01
-5.28475568e-02 -7.56740391e-01 -1.18302345e+00 -5.45059323e-01
-6.14801109e-01 7.26203173e-02 6.17579341e-01 6.71862364e-01
1.15452958e-02 5.17112426e-02 -3.73489588e-01 -2.15179864e-02
9.65753973e-01 5.31691790e-01 2.60924816e-01 -1.43124521e+00
-1.43529803e-01 -3.17548990e-01 -7.77934611e-01 -1.39657259e+00
-4.52825651e-02 -7.64709711e-01 2.05076903e-01 -1.32025456e+00
1.60030872e-01 -4.73193645e-01 -3.66536558e-01 4.23193604e-01
1.72849342e-01 4.87602025e-01 6.99098110e-02 6.14654124e-01
-5.43809593e-01 6.48912907e-01 1.23601532e+00 -1.91240698e-01
-1.21440552e-02 -3.02146524e-01 -7.91591048e-01 4.78775024e-01
1.00776064e+00 -4.72891688e-01 -2.26338416e-01 -6.31970525e-01
-1.45442829e-01 1.34145364e-01 3.85092705e-01 -1.27030432e+00
5.96572816e-01 -1.54827178e-01 5.83169878e-01 -5.19393444e-01
5.23404658e-01 -1.14554632e+00 3.30857515e-01 6.55251503e-01
-4.73353118e-02 5.49314581e-02 8.68106857e-02 7.57602215e-01
-5.27710736e-01 -4.00331050e-01 7.56288588e-01 -4.58371520e-01
-8.77982676e-01 4.38079745e-01 -4.73518908e-01 -2.89298594e-01
9.13839221e-01 -4.31174755e-01 -1.68281123e-01 -2.21100092e-01
-5.31760216e-01 -4.47131991e-02 6.56253159e-01 -1.62391245e-01
1.27458858e+00 -1.42359197e+00 -4.20264989e-01 6.30314529e-01
-3.89856510e-02 1.45043224e-01 3.57419133e-01 6.40273035e-01
-3.59277785e-01 2.35731080e-01 -6.12538338e-01 -7.27859318e-01
-1.11960995e+00 6.61944151e-01 -2.92120457e-01 2.08689839e-01
-9.49492514e-01 6.36880994e-01 -1.76335469e-01 3.63001376e-01
2.95785785e-01 -2.18269989e-01 -2.44320214e-01 -2.23510206e-01
8.90276194e-01 3.63862097e-01 -7.57681280e-02 -5.28468013e-01
-2.98844516e-01 8.46955836e-01 1.17344536e-01 1.40603721e-01
1.64083993e+00 -2.82791346e-01 -5.72149992e-01 5.59715629e-01
1.13669991e+00 -2.34141856e-01 -1.30710137e+00 -4.51802313e-01
-3.48197728e-01 -6.07408404e-01 3.68439883e-01 -1.68097332e-01
-1.44624531e+00 8.89622390e-01 6.63567483e-01 2.68134743e-01
1.41866970e+00 -4.21682931e-02 1.04982948e+00 6.91458046e-01
8.01959693e-01 -9.01805997e-01 -3.66536602e-02 3.15789491e-01
7.65164256e-01 -8.88878644e-01 -1.30643817e-02 -1.60670385e-01
-3.64779681e-01 1.01201701e+00 3.32389295e-01 -6.85534954e-01
7.69609094e-01 5.32371521e-01 -3.85377616e-01 -2.11632654e-01
-5.80748200e-01 -6.44483343e-02 -3.36865097e-01 7.34989822e-01
-2.35325858e-01 -1.00791864e-01 -1.42615348e-01 4.07445133e-01
-6.18730448e-02 3.18766445e-01 3.85732740e-01 9.56865728e-01
-8.41886163e-01 -8.68640304e-01 -4.16674942e-01 9.47133005e-01
-1.48017630e-01 -1.58798367e-01 -2.64929719e-02 -1.26741067e-01
3.34193036e-02 6.37203395e-01 1.73902437e-01 -4.11683440e-01
6.51885197e-02 -2.71836966e-01 4.60065693e-01 -4.68354285e-01
-1.00385539e-01 1.57609478e-01 -6.10997856e-01 -7.37407684e-01
-6.24490798e-01 -8.07282984e-01 -1.30713952e+00 -3.03484440e-01
-8.42104331e-02 8.94631967e-02 6.83600843e-01 6.00347400e-01
7.72600472e-01 4.36327606e-01 7.51164138e-01 -1.07061565e+00
-4.39291596e-01 -7.42978692e-01 -9.18684125e-01 1.54913828e-01
4.67164069e-01 -6.95162714e-01 -2.80240387e-01 3.36108118e-01] | [11.105327606201172, -1.7430795431137085] |
977cd459-8338-4bd5-8d35-29a1e87b4b70 | efficient-micro-structured-weight-unification | 2106.08301 | null | https://arxiv.org/abs/2106.08301v2 | https://arxiv.org/pdf/2106.08301v2.pdf | Efficient Micro-Structured Weight Unification and Pruning for Neural Network Compression | Compressing Deep Neural Network (DNN) models to alleviate the storage and computation requirements is essential for practical applications, especially for resource limited devices. Although capable of reducing a reasonable amount of model parameters, previous unstructured or structured weight pruning methods can hardly truly accelerate inference, either due to the poor hardware compatibility of the unstructured sparsity or due to the low sparse rate of the structurally pruned network. Aiming at reducing both storage and computation, as well as preserving the original task performance, we propose a generalized weight unification framework at a hardware compatible micro-structured level to achieve high amount of compression and acceleration. Weight coefficients of a selected micro-structured block are unified to reduce the storage and computation of the block without changing the neuron connections, which turns to a micro-structured pruning special case when all unified coefficients are set to zero, where neuron connections (hence storage and computation) are completely removed. In addition, we developed an effective training framework based on the alternating direction method of multipliers (ADMM), which converts our complex constrained optimization into separately solvable subproblems. Through iteratively optimizing the subproblems, the desired micro-structure can be ensured with high compression ratio and low performance degradation. We extensively evaluated our method using a variety of benchmark models and datasets for different applications. Experimental results demonstrate state-of-the-art performance. | ['Songnan Li', 'Shan Liu', 'Yanzhi Wang', 'Kaidi Xu', 'Wei Wang', 'Wei Jiang', 'Sheng Lin'] | 2021-06-15 | null | null | null | null | ['neural-network-compression', 'neural-network-compression'] | ['methodology', 'miscellaneous'] | [ 4.04920310e-01 -9.22373980e-02 -4.78353240e-02 -4.28432643e-01
-1.97288617e-02 1.41968295e-01 1.24896348e-01 -9.54492614e-02
-8.43402565e-01 7.46086419e-01 -2.43026853e-01 -2.94400334e-01
-3.01478177e-01 -9.28421676e-01 -6.91178322e-01 -9.29677725e-01
1.69622242e-01 2.05133334e-01 1.99701980e-01 5.99495247e-02
1.93650320e-01 3.89923394e-01 -1.79911089e+00 1.50726467e-01
1.04078770e+00 1.26110840e+00 9.03327227e-01 1.23546861e-01
-1.23915225e-01 7.08341062e-01 -3.53658050e-01 -4.72233862e-01
3.78780514e-01 -3.85580510e-02 -4.30621296e-01 -1.71584042e-03
4.41924304e-01 -5.58470905e-01 -5.55016220e-01 1.47419119e+00
3.59623820e-01 1.21638305e-01 1.56242177e-01 -8.00979793e-01
-2.75379866e-01 6.78372622e-01 -4.79990810e-01 2.13964254e-01
-5.02622247e-01 -2.05834419e-01 8.92735183e-01 -8.99972677e-01
5.18902063e-01 9.34378207e-01 5.91606081e-01 5.50959527e-01
-1.16303694e+00 -7.46446073e-01 4.58351582e-01 4.90830600e-01
-1.42236352e+00 -6.15022838e-01 7.68035531e-01 5.62276281e-02
1.17114115e+00 3.27975422e-01 9.58288729e-01 6.45168245e-01
-9.61188227e-02 6.33053422e-01 4.50771540e-01 -3.07037532e-01
3.69231820e-01 -7.47535229e-02 3.67632866e-01 1.02601397e+00
7.06475317e-01 -3.08001071e-01 -4.17477071e-01 1.26840577e-01
6.76052392e-01 3.01052213e-01 -3.50069761e-01 -2.33203471e-01
-1.02929914e+00 6.88607872e-01 4.35593635e-01 2.40595043e-01
-3.48371565e-01 1.69435740e-01 4.05032933e-01 2.21913710e-01
1.54028729e-01 -1.31436497e-01 -5.74471593e-01 -3.57115321e-04
-1.25900733e+00 1.21523142e-01 6.33575261e-01 1.00248778e+00
6.91704810e-01 5.97519636e-01 6.19881004e-02 9.97323930e-01
1.20471425e-01 1.05577052e-01 7.90054739e-01 -7.66198575e-01
8.70574713e-01 8.11370969e-01 -3.80521506e-01 -1.14978874e+00
-4.17265922e-01 -8.44886720e-01 -1.51470876e+00 -9.28195864e-02
1.40674531e-01 -1.09159388e-01 -7.47826636e-01 1.77108943e+00
1.74305469e-01 1.53991193e-01 -6.87609762e-02 8.70753348e-01
7.07922220e-01 7.73805618e-01 1.55607322e-02 -5.46797037e-01
1.40840912e+00 -1.10197055e+00 -6.90971613e-01 -4.47284937e-01
8.94142151e-01 -4.03472811e-01 8.55944574e-01 4.83584106e-01
-1.43655109e+00 -5.45988560e-01 -1.38097405e+00 -3.66426528e-01
-1.40288845e-01 5.23757756e-01 7.53759503e-01 3.17413270e-01
-8.36186767e-01 8.98346424e-01 -1.10732722e+00 4.04491633e-01
6.79268122e-01 6.66770816e-01 -2.05350220e-01 -1.94216356e-01
-1.12021708e+00 8.22834730e-01 6.36729479e-01 5.47869086e-01
-4.20476288e-01 -9.30270314e-01 -6.73268080e-01 6.33404493e-01
2.57887989e-01 -6.21288717e-01 8.19409370e-01 -6.16339386e-01
-1.23642349e+00 5.17418802e-01 -3.22036028e-01 -7.36426055e-01
3.41074973e-01 -1.22837454e-01 -3.22684854e-01 8.26121643e-02
-4.45009351e-01 6.44269466e-01 9.14869010e-01 -5.90068102e-01
-6.79520011e-01 -4.31159705e-01 -1.63605452e-01 1.42150357e-01
-1.18615258e+00 -1.71873868e-01 -6.38009608e-01 -7.98757076e-01
5.44610083e-01 -6.61559582e-01 -3.71836662e-01 1.93242937e-01
-1.43104762e-01 -3.57710868e-02 9.45869625e-01 -7.13837683e-01
1.51890838e+00 -2.19695830e+00 4.36211348e-01 7.59263337e-02
4.17493373e-01 6.86344683e-01 -4.43129241e-02 -8.28087255e-02
1.31466016e-02 -2.31501922e-01 -3.61811250e-01 -6.05746627e-01
-1.57798141e-01 5.63595176e-01 -2.56747425e-01 3.01762313e-01
8.54091123e-02 5.44277668e-01 -6.08539820e-01 -5.43152750e-01
1.36847660e-01 7.01223969e-01 -8.69529963e-01 -1.63480356e-01
-2.51849219e-02 -1.95537910e-01 -3.72447610e-01 5.47732115e-01
8.83336782e-01 -1.03181019e-01 3.34222287e-01 -5.53962231e-01
-7.36356750e-02 4.91110235e-01 -1.39421606e+00 1.63845944e+00
-5.47214746e-01 3.47599149e-01 3.69987160e-01 -1.54187179e+00
9.98332381e-01 1.12790771e-01 3.85590166e-01 -7.23756969e-01
2.20892489e-01 2.94147730e-01 1.07764184e-01 -1.69049650e-01
3.76619041e-01 1.85694695e-01 3.28288168e-01 1.20589212e-01
2.33676985e-01 2.79857755e-01 5.01214504e-01 4.11773659e-02
7.68719256e-01 -2.54772574e-01 -3.17665488e-02 -4.90354091e-01
6.43910170e-01 -1.90031111e-01 9.40768719e-01 2.81963974e-01
4.29629646e-02 2.70839155e-01 4.53676909e-01 -6.69706225e-01
-9.82903659e-01 -5.96035540e-01 -3.74087334e-01 8.28226566e-01
-6.08211271e-02 -5.08254170e-01 -8.21220279e-01 -6.03887066e-02
-2.27932617e-01 4.13208634e-01 -2.54707038e-01 -3.82475823e-01
-8.66373241e-01 -1.09901702e+00 3.44824344e-01 3.82083565e-01
8.29391539e-01 -7.87714481e-01 -8.51884961e-01 3.66026342e-01
1.16180396e-03 -1.20821261e+00 -1.86092019e-01 5.39502978e-01
-1.47341692e+00 -7.05152750e-01 -5.12411237e-01 -1.11626923e+00
9.73030448e-01 4.67709273e-01 7.82698870e-01 4.74569410e-01
-2.73200154e-01 -7.59812534e-01 3.02657578e-02 -2.30237901e-01
2.38818154e-01 2.30861902e-01 2.09512085e-01 -4.50096279e-02
1.50011912e-01 -9.22908187e-01 -5.96730769e-01 3.13506089e-02
-9.96170342e-01 4.46216136e-01 6.21061325e-01 9.54475582e-01
7.70788193e-01 3.64574224e-01 3.69958013e-01 -7.68543541e-01
3.51511985e-01 -1.39325589e-01 -9.21011388e-01 1.94181174e-01
-6.84908390e-01 2.10810170e-01 1.00131440e+00 -6.59398675e-01
-8.73596907e-01 1.23691827e-01 -2.18782172e-01 -5.62554538e-01
3.93993199e-01 5.62911570e-01 -2.72915334e-01 -2.17548475e-01
1.62644655e-01 4.63191658e-01 -6.81047738e-02 -6.49296880e-01
1.76317785e-02 2.66805917e-01 5.51761627e-01 -3.61511201e-01
5.86318552e-01 3.73462290e-01 2.68595248e-01 -7.21835375e-01
-7.32290089e-01 8.24686661e-02 -5.50050616e-01 5.70329465e-02
4.28193867e-01 -1.08261919e+00 -6.33569717e-01 3.46968353e-01
-1.07352340e+00 -1.01318918e-01 -1.56913131e-01 6.68506086e-01
5.96772023e-02 4.82034683e-01 -7.91677654e-01 -4.80956107e-01
-7.14211881e-01 -1.28449452e+00 5.46148300e-01 1.35274544e-01
1.01995945e-01 -7.05432177e-01 -4.87972826e-01 1.78849295e-01
6.19881094e-01 -1.32390767e-01 1.22869849e+00 -2.57584304e-01
-5.05571783e-01 -2.23878741e-01 -4.78921741e-01 3.74192595e-01
-2.34294787e-01 -1.84399217e-01 -6.60930812e-01 -2.75309771e-01
5.10055184e-01 -2.46073604e-01 1.16100180e+00 2.49596074e-01
1.63799095e+00 -5.52488804e-01 -2.13100955e-01 1.03955853e+00
1.55210412e+00 3.41693535e-02 6.31727457e-01 2.71369904e-01
7.69686222e-01 3.23477656e-01 1.63177013e-01 5.36814809e-01
9.94264111e-02 5.15561581e-01 5.52012920e-01 8.95339325e-02
-7.45700672e-02 -5.32428995e-02 1.06595099e-01 1.51875854e+00
-3.06029290e-01 7.92134181e-02 -4.98628318e-01 3.88112098e-01
-1.86482608e+00 -7.92292714e-01 -2.01203063e-01 2.18002367e+00
1.07719803e+00 3.45235705e-01 -4.53884810e-01 5.36961138e-01
4.63232040e-01 2.43071824e-01 -6.02787256e-01 -2.84676164e-01
-1.75578743e-01 3.73909831e-01 5.58736145e-01 2.47556657e-01
-7.86470294e-01 7.25610197e-01 5.49801111e+00 1.22023106e+00
-1.03382099e+00 1.96070462e-01 6.03839934e-01 -6.37485266e-01
-8.57709441e-03 -1.54451504e-01 -1.26017118e+00 5.00583231e-01
8.64246070e-01 1.11376807e-01 5.96137345e-01 1.08376300e+00
2.86835879e-02 1.54786319e-01 -9.51555729e-01 1.27236807e+00
-2.45093331e-01 -1.54554391e+00 2.48066932e-01 8.58845338e-02
7.42677867e-01 1.19123850e-02 -8.58583972e-02 3.21543783e-01
-2.66712427e-01 -6.78497553e-01 7.45289564e-01 2.71039158e-01
6.27583683e-01 -9.19478834e-01 6.86968088e-01 5.28305769e-01
-1.06910956e+00 -2.18255043e-01 -1.14747071e+00 -3.53894413e-01
9.74843577e-02 1.11787081e+00 9.36754271e-02 2.28139028e-01
7.23888397e-01 5.59226751e-01 -2.25701213e-01 7.24905372e-01
4.71251644e-03 4.51875538e-01 -5.54543257e-01 -1.30512631e-02
2.43385226e-01 -5.97499013e-01 2.48553813e-01 1.05134201e+00
3.55514228e-01 2.34927937e-01 -2.39434332e-01 6.74675107e-01
-3.24489743e-01 -4.10284847e-02 -1.79306284e-01 1.98474750e-01
5.78076661e-01 1.33179760e+00 -6.67413473e-01 -4.46267158e-01
-5.09157181e-01 6.96286559e-01 7.55624712e-01 1.84620082e-01
-8.32077086e-01 -5.13860345e-01 6.29119098e-01 -7.26915672e-02
5.47671676e-01 -4.15813208e-01 -6.66672707e-01 -1.28902018e+00
4.35738087e-01 -6.53613091e-01 5.87560311e-02 -2.31640756e-01
-5.60292721e-01 6.71341121e-01 -2.00770527e-01 -1.15652049e+00
1.06259100e-01 -5.91616750e-01 -4.57111716e-01 6.13934457e-01
-1.64211249e+00 -6.25324309e-01 -3.03230554e-01 6.11057341e-01
5.43382525e-01 -2.39224210e-01 7.80311704e-01 7.84571052e-01
-1.15797961e+00 8.70863438e-01 1.26667321e-01 -1.03806771e-01
1.00027531e-01 -5.50849140e-01 4.82729375e-02 1.01288378e+00
8.12949538e-02 8.23338389e-01 3.13454092e-01 -3.15995365e-01
-1.70400894e+00 -1.23142385e+00 1.09339559e+00 5.70254385e-01
5.42975008e-01 -4.49440747e-01 -1.12708414e+00 2.54108042e-01
-2.24284485e-01 2.33488023e-01 4.11726087e-01 -1.74711153e-01
-1.02610432e-01 -5.43559909e-01 -1.04845023e+00 7.98481882e-01
1.28747785e+00 -9.04392749e-02 -4.46530104e-01 5.50395131e-01
7.43804693e-01 -3.49446625e-01 -8.35427701e-01 4.98419076e-01
4.68662620e-01 -8.72125804e-01 1.12410498e+00 -4.16808456e-01
7.01492727e-01 -1.63118690e-01 -3.29351723e-01 -7.80270159e-01
-4.73391324e-01 -2.98323572e-01 -8.54590118e-01 1.11993825e+00
3.26549768e-01 -5.29978752e-01 9.42094326e-01 8.56741190e-01
-3.22055221e-01 -1.34715271e+00 -1.02941382e+00 -6.76915765e-01
-3.78972262e-01 -4.94395167e-01 5.19757807e-01 7.84394026e-01
-7.08399117e-02 1.97489753e-01 -4.52246904e-01 7.27929994e-02
6.25005841e-01 1.42982438e-01 1.21872835e-01 -1.15051627e+00
-3.30796987e-01 -6.55657053e-01 -3.00787181e-01 -1.29523873e+00
6.41667917e-02 -9.14599597e-01 -2.83798397e-01 -1.42725587e+00
9.78519022e-02 -5.73650002e-01 -3.96322489e-01 6.47852421e-01
1.28111392e-02 3.90078664e-01 3.05189937e-01 4.23300296e-01
-3.15205663e-01 7.57682085e-01 1.14211214e+00 -1.16339967e-01
-1.56593949e-01 -1.12676203e-01 -5.56843281e-01 8.73099327e-01
7.35600173e-01 -5.98069966e-01 -5.72185159e-01 -1.09707308e+00
1.77821323e-01 -1.51864327e-02 1.56597242e-01 -1.24887061e+00
4.84928131e-01 4.79488894e-02 3.55281562e-01 -6.51544690e-01
5.21953046e-01 -1.00580597e+00 -1.83623731e-02 9.00671124e-01
-1.56767249e-01 -5.97860441e-02 1.86125010e-01 3.07884812e-01
-3.85038674e-01 -6.88624442e-01 8.90861928e-01 1.22184970e-03
-6.41555250e-01 4.44140762e-01 -1.11152805e-01 -3.83958787e-01
7.80840814e-01 -3.68954659e-01 -6.41793162e-02 2.85373300e-01
-5.70851088e-01 2.16781020e-01 3.93251190e-03 6.20260052e-02
7.83769608e-01 -1.30436289e+00 -4.16309148e-01 5.43124855e-01
-5.99986076e-01 4.46772039e-01 4.63896573e-01 8.61026824e-01
-6.48872852e-01 5.97528994e-01 -3.62014174e-01 -3.62828404e-01
-1.12877464e+00 5.17914891e-01 -5.41552678e-02 -5.54622233e-01
-8.52366924e-01 9.94481027e-01 3.03831808e-02 -5.89625053e-02
6.73654616e-01 -6.52628005e-01 -1.09851576e-01 5.23780510e-02
7.92540014e-01 4.85728532e-01 5.13315499e-01 -3.37299764e-01
-2.59045422e-01 4.94851887e-01 -2.38723412e-01 4.99737799e-01
1.66626537e+00 8.90410915e-02 -3.56822729e-01 -1.71307281e-01
1.29762423e+00 -4.07993406e-01 -1.27824903e+00 -3.21932286e-01
-3.41864914e-01 -1.90445125e-01 5.91075242e-01 -7.05858842e-02
-1.80352187e+00 9.81274664e-01 3.55748862e-01 -2.21983865e-01
1.43801463e+00 -7.93740571e-01 1.20323360e+00 8.53726149e-01
4.63858992e-01 -1.20346153e+00 -4.15263385e-01 5.45337439e-01
5.80297947e-01 -7.89954841e-01 4.48860407e-01 -4.40080196e-01
-2.76978850e-01 1.22476029e+00 6.16666198e-01 -3.71212274e-01
8.10551882e-01 4.06331390e-01 -6.59040153e-01 1.08604059e-01
-8.17306519e-01 2.82175541e-01 1.28184885e-01 2.16118738e-01
1.36458069e-01 -2.19601840e-01 -6.39158249e-01 9.39339161e-01
-3.46214503e-01 2.59389002e-02 1.19841687e-01 8.30111146e-01
-5.65457463e-01 -8.54578316e-01 2.33033802e-02 7.80906796e-01
-4.68711734e-01 -3.47614288e-01 3.42432141e-01 5.68010211e-01
3.48293185e-01 6.03598177e-01 2.41072580e-01 -4.33874547e-01
2.07539335e-01 -3.77997756e-02 3.93995762e-01 -3.70256960e-01
-4.17913944e-01 1.07968673e-01 -5.30083925e-02 -4.61708039e-01
-2.66204715e-01 -3.18052918e-01 -1.30847442e+00 -4.29614305e-01
-3.71068448e-01 -1.03703044e-01 7.24396050e-01 9.88993108e-01
3.57733697e-01 6.83073580e-01 2.58728892e-01 -1.02632165e+00
-7.39929140e-01 -9.51061666e-01 -4.92574751e-01 1.27193600e-01
-2.31021754e-02 -5.38375735e-01 -1.13613747e-01 1.04521319e-01] | [8.53030776977539, 3.071462631225586] |
d2ad1c1b-916c-4519-aaf5-1aa10177220a | differentiable-hierarchical-graph-grouping | 2007.11864 | null | https://arxiv.org/abs/2007.11864v1 | https://arxiv.org/pdf/2007.11864v1.pdf | Differentiable Hierarchical Graph Grouping for Multi-Person Pose Estimation | Multi-person pose estimation is challenging because it localizes body keypoints for multiple persons simultaneously. Previous methods can be divided into two streams, i.e. top-down and bottom-up methods. The top-down methods localize keypoints after human detection, while the bottom-up methods localize keypoints directly and then cluster/group them for different persons, which are generally more efficient than top-down methods. However, in existing bottom-up methods, the keypoint grouping is usually solved independently from keypoint detection, making them not end-to-end trainable and have sub-optimal performance. In this paper, we investigate a new perspective of human part grouping and reformulate it as a graph clustering task. Especially, we propose a novel differentiable Hierarchical Graph Grouping (HGG) method to learn the graph grouping in bottom-up multi-person pose estimation task. Moreover, HGG is easily embedded into main-stream bottom-up methods. It takes human keypoint candidates as graph nodes and clusters keypoints in a multi-layer graph neural network model. The modules of HGG can be trained end-to-end with the keypoint detection network and is able to supervise the grouping process in a hierarchical manner. To improve the discrimination of the clustering, we add a set of edge discriminators and macro-node discriminators. Extensive experiments on both COCO and OCHuman datasets demonstrate that the proposed method improves the performance of bottom-up pose estimation methods. | ['Sheng Jin', 'Wentao Liu', 'Ping Luo', 'Chen Qian', 'Wenhai Wang', 'Wanli Ouyang', 'Enze Xie'] | 2020-07-23 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/386_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123520698.pdf | eccv-2020-8 | ['2d-human-pose-estimation'] | ['computer-vision'] | [-0.18000409 0.02465307 -0.189461 -0.12481438 -0.40891722 -0.3211083
0.21108074 0.31311077 -0.44392437 0.15857895 -0.01135644 0.28218958
-0.01619366 -0.69465214 -0.6862085 -0.36471966 -0.31623834 0.71108264
0.50052226 -0.11927637 -0.15263726 0.41020784 -1.4487704 -0.10573889
0.77237314 0.72879535 0.15030225 0.6894783 0.3990454 0.24103665
-0.5883692 -0.3566487 0.37375677 -0.40004918 -0.3882197 0.1695343
0.7866569 -0.07633909 -0.4276783 0.9830425 0.8372617 0.34676576
0.47057557 -1.5285873 -0.17136383 0.34162828 -0.9025759 -0.09105869
0.64373726 -0.17211346 0.8609274 -0.8720103 0.42802423 1.5508074
0.8988836 0.58554167 -0.99068725 -0.60830283 0.6676828 0.20224757
-1.5786717 -0.13667454 0.846394 -0.49577245 0.5842681 0.13067777
1.03116 0.6675406 0.1433069 1.1074528 0.6293637 -0.32509223
-0.01488199 -0.31825104 0.16813931 1.1490796 0.46769807 -0.0745152
-0.6425433 -0.13162765 0.97583914 0.34392145 -0.28580227 -0.8921532
-1.3932651 0.63111717 1.0486329 -0.03692331 -0.36036602 0.3047074
0.37165272 -0.06239273 0.25122666 0.06625702 -0.19869585 0.56074643
-0.87279356 0.46190348 0.61628693 1.1493294 0.7560507 -0.40047213
-0.3988715 0.7236351 0.5187776 0.2819295 0.3173295 -0.35424194
0.6571575 0.8908131 0.0257161 -1.5118843 -0.8520346 -0.5090928
-0.93848187 -0.08436011 0.5103893 -0.26842606 -1.112902 1.6712205
0.74108696 0.07747665 -0.5281904 1.2035599 0.88851136 0.4600466
0.13314013 -0.01702894 1.6269695 -1.4957347 -0.5629391 -0.5456579
0.45645115 -0.22718981 0.72515756 0.3259032 -0.9164732 -1.080839
-1.0089887 -0.02476626 -0.3234939 0.4485362 0.57510513 0.50475526
-0.9284605 0.50655556 -0.97233075 -0.40375283 0.13961944 0.44722554
-0.48816738 0.04637791 -1.0847808 0.42730463 0.66866595 0.49666777
-0.5634833 -0.2971341 -1.1893146 -0.12025601 0.7844864 -1.1680462
0.8567702 -0.42054242 -1.2549946 0.7204149 -0.1816357 -0.15787265
0.74913704 -0.45421484 -0.142463 0.4089925 0.4078392 0.8016682
1.0121055 -1.238208 -0.96862143 -0.6646079 -0.11985902 0.49868459
-0.13333768 -0.12265153 -1.3117081 -0.70952594 0.44591737 -1.2213082
-0.32435405 -0.0550014 -0.88598204 -0.5366837 0.57198405 -0.62170035
1.2618283 -1.9517269 0.5310097 0.33239982 0.612683 -0.11934963
0.18358706 0.5125221 -0.12135549 -0.24905543 0.1654269 -0.5895821
0.09383948 -0.15277337 0.51977664 0.8143045 -0.20546313 1.0128938
-0.9384442 -0.7004301 0.3950423 0.35212317 -0.4909374 0.18199475
0.18346047 0.27879167 -0.3441921 0.76224375 0.5045031 -0.31137007
0.22466771 -0.5003498 0.21802737 -0.40450296 -1.622536 1.8547245
0.04701023 0.07741161 0.05737455 -0.6804522 0.6107827 0.0811852
0.4048415 -0.05554077 0.07204059 -0.23386197 -0.14668198 -0.1871968
0.5060146 0.02321179 -0.47047186 -0.10493523 0.17274453 0.58776355
0.3219613 0.252102 0.8144376 0.23899975 0.47030163 0.05491762
0.563045 -0.08396507 0.7216526 0.6964196 -0.39683253 0.7417313
0.21735741 -0.37949473 -0.46445912 -0.94856423 0.3830172 1.4535953
0.68484014 -1.095166 -0.81655747 -1.0126058 0.25324294 -0.16383772
-0.6322603 -0.20271297 -0.75047314 -0.60305107 0.33519417 0.73098034
0.5914967 -0.8647048 -0.6522738 0.10658884 -0.32088566 -1.2276242
-0.8293981 -0.06883172 -0.54057807 -1.3338068 -1.0833015 -1.0322236
1.0043135 0.4813051 0.8910003 0.21548396 -0.2808044 0.5637154
-0.40201512 -0.26341325 0.12825441 0.24419764 0.42680603 0.17741056
0.220018 -0.1671537 -0.9503118 0.548706 -0.15254892 0.1487829
0.80916476 0.5358951 0.6647022 0.31176558 0.27626583 -0.65242046
0.39356804 0.02220744 -0.23348902 0.24874136 -0.26785928 -0.31507918
0.60862 -0.4170935 -0.38736352 0.6370825 0.06872371 -0.6580686
-0.29271394 0.36305347 -0.49013773 -0.25264165 0.35971335 0.03134595
-0.14835374 -0.51954806 0.41195327 0.27674678 0.7794662 -0.4516028
1.0737164 0.33264187 -0.02874166 -0.86589867 -0.5766063 -0.96854717
-1.0371263 -0.74604005 1.1443423 -1.0762439 -0.92360973 0.60796267
-0.7916783 -0.16037591 0.11305638 0.4240182 -0.3275737 0.6681518
-0.5618207 -0.7084062 -0.31114233 -0.97604287 1.5848324 0.46152338
-0.16348991 -0.81919616 -0.1583373 0.1974468 -0.41465557 0.63783306
0.38280496 -0.5343822 -0.3295442 -0.67721885 -0.09144309 -0.18275574
-0.08863625 -0.49988353 -0.50384945 -0.87396526 -0.49581066 -0.18303487
0.86247885 0.47879747 0.7964416 -0.08676206 -0.9487321 0.76634794
1.2028499 -0.34102273 0.13360283 0.31760857 1.4243414 0.6867705
0.7035152 0.21798465 0.7319041 0.9640147 0.17957039 -0.41049144
-0.09784868 -0.6592748 0.29326215 0.48991567 -0.3014242 -0.08843613
-0.76897436 0.27126178 -2.2669997 -0.6859314 -0.14215599 2.1036305
0.31957373 0.2384116 0.87061363 0.07559007 1.3215942 0.22680409
-0.47667822 0.4975432 0.38083547 -0.33349532 0.4581277 0.18292408
-1.670522 1.0065103 5.278994 0.74306214 -0.5569055 -0.14819333
0.09519717 0.06498472 0.531988 -0.25100854 -1.10073 0.34589732
0.22478323 0.13057816 -0.03091852 1.0277339 -0.0404352 0.0164217
-1.3131647 1.4383243 0.24416341 -0.7065032 -0.01119137 -0.08210814
0.27334237 -0.38400474 -0.44166544 0.51644397 -0.05174683 -0.7620521
0.74648154 0.31088245 0.5158912 -0.96064025 0.5827353 0.6352007
-1.9505383 -0.08296332 -0.41725022 -0.0118298 0.30244935 0.20950708
-0.47144705 0.92201173 0.8360128 0.6777108 -0.9293826 1.2444487
-0.46999452 0.06637459 -0.3251014 -0.04069421 -0.00855468 -0.05903054
0.659752 1.2815632 -0.0066234 -0.05967584 0.95765567 0.3815815
0.11667108 0.05765413 -0.24527873 0.4049055 0.32279322 1.5658563
-1.1158035 -0.32078764 -0.21157822 1.2823642 0.43841493 0.3255674
-0.91450876 -0.6327668 0.23040995 0.41787246 0.21144164 -0.3820017
0.35520267 -1.2146496 0.15288076 -0.77823895 0.7858608 -0.5237087
-1.3092635 0.31722158 0.3123673 -1.2059957 -0.2562293 -0.6095032
-0.41643387 0.6024038 -0.97807485 -1.4976251 -0.6767328 0.8466488
0.40348247 0.12183543 0.48225114 0.31105587 -0.8037847 0.9146942
-0.60301095 0.73037744 0.71776557 -1.550413 0.50890505 1.1784238
0.03979335 0.7689183 0.43384206 -1.0707623 -1.4365411 -1.2719523
0.563769 -0.2427832 0.26524496 -0.8323407 -0.5727286 0.6763231
-0.22844373 0.09051567 0.5292696 0.41962174 -0.0886576 0.0383136
-0.84292245 0.6774124 1.3359965 -0.16754241 -0.6391091 0.49060693
0.5186854 -0.6989036 -0.73448527 0.44093037 0.50849324 -0.8880585
1.2808584 -0.28012794 -0.11772713 -0.6879028 0.3844976 -1.2408832
-0.75188637 -0.72318214 -0.42808706 1.0305827 0.02385182 -0.5239153
1.0925832 0.3179225 -0.07253521 -0.6780852 -0.69380987 -0.7560694
-0.47954515 -0.13062559 0.23521931 0.63746107 0.01965754 0.67100453
-0.5594189 0.6159847 1.047651 0.18772362 1.2425056 -1.3684042
-0.5993897 -0.26790488 -0.8846617 -1.5556687 -0.18949713 -0.8637094
0.2058591 -1.7578003 0.2222516 -0.1003544 -0.11088613 0.59435946
-0.7495435 0.27819687 0.47450924 0.31307998 -0.8650979 0.32371032
1.1552095 -0.07110471 -0.4992328 0.29746622 -0.5719952 0.8868202
0.5007482 -0.3440163 -0.3676069 0.01927529 0.11670505 0.08161681
0.76799595 -1.4575937 0.65875703 0.26027012 0.90262115 -0.99086374
0.2944466 -0.5639693 -0.08661077 0.605533 0.09202561 0.09941065
-0.07353207 0.96756196 -0.05049245 0.21426961 0.6834856 -0.29935467
-0.97304446 0.64411074 0.09186325 -0.05545693 1.1597615 -0.4505662
0.09447113 -0.19148852 -1.0528156 0.62761325 0.35784504 0.5225596
0.67632407 -1.4192557 -0.5331142 0.12412295 0.20042595 0.3365529
0.33547097 0.992854 -0.34304887 0.07755272 0.09833965 -0.95335174
-1.5347862 0.8183926 0.43880433 -0.2261205 -0.9673652 0.98305184
0.4319982 -0.39908442 0.5239197 -0.20886414 -0.39803502 0.30157578
0.46824977 0.378584 -0.28170085 -0.88661236 -0.6188019 1.0746762
0.03402347 0.2152519 0.9632067 -0.17838322 0.09362761 0.3894406
1.0024107 -0.00899791 -1.0650951 -0.01695781 -0.22014146 -0.39943248
-0.31641954 -0.3960527 -0.8971124 0.6375141 0.46339804 0.26928127
1.0610752 0.11715644 0.84628016 0.33115107 0.5381489 -1.1880137
0.12427465 0.06202577 0.7553977 -1.1084287 0.28674608 -0.82316226
-0.37909964 0.9804071 0.95219564 -0.39581278 0.5071142 -0.21522331
-0.17450984 -0.43809408 -0.08472371 -0.54072887 0.742956 0.6715779
0.1288548 0.13132271 -0.07588441 0.77122056 -0.20353544 -0.21304531
-0.11212234 0.88712496 -0.49480084 -0.78005034 -0.6821894 0.1903133
-0.1874867 0.3228811 -0.5239773 1.2494053 0.32918888 0.94962656
-0.2958153 -0.74490786 0.85679585 -0.07497003 0.52290475 -0.7820164
-0.5751434 0.44326255 0.01208588 -0.6710725 -0.4635962 -0.5372355
-1.3237431 -0.07914569 -0.49735942 0.2104464 0.08226427 0.7851203
0.27536774 0.7174479 0.30817804 -1.4830616 -0.16183387 -0.8677525
-0.51791465 0.48835403 0.15534098 -0.90590554 -0.1244818 -0.09922796] | [7.119457721710205, -0.7864536046981812] |
5e0f9a69-3c89-41b7-9fd0-745db0638291 | multi-view-keypoints-for-reliable-6d-object | 2303.16833 | null | https://arxiv.org/abs/2303.16833v1 | https://arxiv.org/pdf/2303.16833v1.pdf | Multi-View Keypoints for Reliable 6D Object Pose Estimation | 6D Object pose estimation is a fundamental component in robotics enabling efficient interaction with the environment. It is particularly challenging in bin-picking applications, where many objects are low-feature and reflective, and self-occlusion between objects of the same type is common. We propose a novel multi-view approach leveraging known camera transformations from an eye-in-hand setup to combine heatmap and keypoint estimates into a probability density map over 3D space. The result is a robust approach that is scalable in the number of views. It relies on a confidence score composed of keypoint probabilities and point-cloud alignment error, which allows reliable rejection of false positives. We demonstrate an average pose estimation error of approximately 0.5mm and 2 degrees across a variety of difficult low-feature and reflective objects in the ROBI dataset, while also surpassing the state-of-art correct detection rate, measured using the 10% object diameter threshold on ADD error. | ['Angela P. Schoellig', 'Alan Li'] | 2023-03-29 | null | null | null | null | ['6d-pose-estimation'] | ['computer-vision'] | [ 1.04198970e-01 -1.13909230e-01 2.86557049e-01 -7.80285522e-02
-7.76126683e-01 -8.89997363e-01 5.07463038e-01 3.41135204e-01
-6.13347411e-01 4.14790571e-01 -4.01602626e-01 1.81437612e-01
-3.45062256e-01 -4.17868972e-01 -9.61142242e-01 -4.67132568e-01
-1.52105121e-02 1.07534432e+00 6.70972705e-01 9.68992390e-06
5.00614524e-01 8.96595716e-01 -1.85577571e+00 -2.09966585e-01
5.07022858e-01 1.08946240e+00 5.08408546e-01 7.46104956e-01
3.73349905e-01 1.60167459e-02 -5.09720981e-01 -3.23943645e-01
5.54181814e-01 3.90169531e-01 -1.29666626e-01 -4.82850783e-02
8.19414377e-01 -5.53675830e-01 1.11104779e-01 9.01970327e-01
4.50438440e-01 -7.95943756e-03 7.92360306e-01 -1.37044108e+00
1.27109274e-01 -1.31848469e-01 -7.85746992e-01 -3.50192040e-02
7.26057589e-01 1.57755226e-01 7.02872455e-01 -1.03568006e+00
6.97413087e-01 1.15921891e+00 7.82263339e-01 1.92533061e-01
-1.26507461e+00 -6.08084381e-01 1.65744826e-01 -1.43123744e-02
-1.50746047e+00 -7.82179832e-02 3.46400529e-01 -7.27858484e-01
1.03249574e+00 2.61151463e-01 8.85782719e-01 7.27346480e-01
5.58033824e-01 3.00166845e-01 1.08219004e+00 -2.61363238e-01
2.83415139e-01 1.77159607e-01 -3.35776150e-01 3.90770286e-01
8.38883400e-01 6.77104592e-02 -7.29526818e-01 -1.46499231e-01
8.69361997e-01 2.67125815e-01 -1.18439114e-02 -1.21393228e+00
-1.41545630e+00 3.86015505e-01 4.25601572e-01 -5.87481320e-01
-2.67552465e-01 1.23985209e-01 -1.24737658e-01 -3.81981403e-01
2.00607345e-01 5.07777631e-01 -3.47618550e-01 -4.11986440e-01
-2.57404298e-01 6.39134943e-01 6.88649714e-01 1.56191587e+00
6.03147566e-01 -5.48325598e-01 3.42680424e-01 4.19427097e-01
3.56646210e-01 1.04410660e+00 -1.75944179e-01 -9.71589386e-01
5.14256001e-01 8.63829374e-01 5.43649256e-01 -8.59538317e-01
-5.20142317e-01 -1.99540451e-01 -1.52505338e-01 7.92763293e-01
4.89462882e-01 2.98536271e-01 -1.01533687e+00 1.06622612e+00
7.34345257e-01 -5.30926168e-01 -3.13710481e-01 9.89091456e-01
5.15414834e-01 -2.66250856e-02 -4.65484381e-01 4.54748869e-02
1.38549232e+00 -5.15112340e-01 -1.69896930e-01 -6.14334941e-01
-4.22236584e-02 -9.40618813e-01 6.76376760e-01 6.86064541e-01
-9.54454422e-01 -1.36975050e-01 -1.14300263e+00 -1.16221711e-01
-3.09333116e-01 5.61896302e-02 5.21792173e-01 3.41194630e-01
-5.24080992e-01 3.06311309e-01 -8.74213636e-01 -5.99593699e-01
2.34055400e-01 5.74944079e-01 -7.05539048e-01 -1.48454458e-01
-1.90838019e-03 1.35627043e+00 3.72242838e-01 -2.25336440e-02
-8.36270273e-01 -8.99185598e-01 -7.42624104e-01 -3.11425805e-01
6.28138900e-01 -5.48712611e-01 1.24940050e+00 2.27263331e-01
-1.34482980e+00 8.98188949e-01 -4.57881466e-02 -9.79941562e-02
1.05484819e+00 -8.50824535e-01 8.44522715e-02 1.66636631e-01
8.50505903e-02 4.48795468e-01 8.07737350e-01 -1.57868659e+00
-7.65232205e-01 -9.33680415e-01 4.60232943e-02 6.09666884e-01
3.93938601e-01 -3.17434490e-01 -3.50191116e-01 6.66636089e-03
7.65603483e-01 -1.13138533e+00 -1.24233998e-01 3.15094799e-01
-3.40982944e-01 7.26616830e-02 8.89636397e-01 -2.06036866e-01
2.78715342e-01 -1.91020763e+00 2.00727239e-01 1.48460642e-01
1.53387070e-01 -2.25463748e-01 4.44198877e-01 4.80829000e-01
3.61611009e-01 -3.08442652e-01 1.47214249e-01 -9.57842991e-02
-3.34700756e-02 -1.19282953e-01 -3.10087707e-02 9.29251373e-01
1.13527961e-01 4.84287351e-01 -9.69928145e-01 -1.02482922e-01
5.08168936e-01 3.84088576e-01 -3.63871992e-01 3.18869054e-01
-6.76665604e-02 2.61891663e-01 -1.17904231e-01 8.37995946e-01
9.66751575e-01 9.98806506e-02 -2.29929537e-02 -3.03834826e-01
-3.62078965e-01 1.72028020e-01 -1.60905397e+00 1.71193016e+00
-3.17223310e-01 3.27763557e-01 1.90314889e-01 -1.14939258e-01
9.72295642e-01 -2.13856518e-01 4.32059884e-01 -2.75934692e-02
1.01328082e-01 3.62028360e-01 -1.51571870e-01 -1.46432608e-01
8.05720091e-01 -2.84203351e-03 -4.29860502e-02 1.32269710e-01
1.35229975e-01 -8.10294807e-01 9.49361250e-02 -9.29853134e-03
1.04877329e+00 4.98193443e-01 4.52708244e-01 -1.92910507e-01
-1.29693761e-01 9.77956578e-02 1.50098532e-01 7.04417825e-01
-7.41614355e-03 9.27616656e-01 1.39954627e-01 -2.42849752e-01
-1.15222347e+00 -1.51845527e+00 -5.08687377e-01 4.14166451e-01
7.61404097e-01 -3.50470483e-01 -3.34375501e-01 -4.17688966e-01
6.30481958e-01 4.00057614e-01 -3.14380199e-01 -6.66082874e-02
-2.88283616e-01 -4.52126473e-01 -2.56974339e-01 6.01907551e-01
1.63933132e-02 -3.73426020e-01 -1.27662551e+00 1.10501446e-01
1.36771142e-01 -1.38041794e+00 2.94269752e-02 4.72051680e-01
-7.57436812e-01 -1.34068501e+00 -6.02937937e-01 -1.83964118e-01
8.73555541e-01 6.80725276e-01 1.10140073e+00 -5.08223236e-01
-5.89232504e-01 6.31582558e-01 -4.39309716e-01 -7.78104007e-01
5.08428691e-03 -2.10834160e-01 4.87204283e-01 -5.53929389e-01
5.51358461e-01 -4.77602184e-01 -6.26364648e-01 6.22631669e-01
-2.62445033e-01 -4.38747881e-03 6.31810009e-01 4.65688765e-01
7.00820863e-01 -3.45390528e-01 -2.55249947e-01 -2.07489520e-01
-1.45345002e-01 -2.12133780e-01 -1.04814065e+00 4.69344631e-02
-3.47688735e-01 -2.63650894e-01 -5.52202687e-02 -5.69134474e-01
-6.59030318e-01 4.26080495e-01 5.34119010e-01 -5.78042388e-01
-2.66270965e-01 -1.19640440e-01 2.50622556e-02 -4.78299946e-01
6.80501282e-01 -2.18682125e-01 6.46005720e-02 -3.31512958e-01
2.43923187e-01 5.05532742e-01 4.83178645e-01 -3.86863559e-01
8.35017323e-01 8.65252972e-01 4.00694460e-01 -6.90641642e-01
-6.57372415e-01 -8.90375614e-01 -1.05604351e+00 -6.31972134e-01
6.44928277e-01 -9.52525377e-01 -1.22623444e+00 3.59161139e-01
-1.10516143e+00 -1.43176713e-03 -1.38627261e-01 9.34727907e-01
-7.62052000e-01 9.68165845e-02 5.40158451e-02 -1.11398602e+00
1.33980634e-02 -1.25401914e+00 1.60985124e+00 3.22949439e-01
-6.80896267e-02 -1.02697760e-01 -2.58725137e-01 2.66630977e-01
-1.33098409e-01 3.02535266e-01 2.34403223e-01 -3.19401681e-01
-1.06879711e+00 -6.42522335e-01 -2.61693388e-01 -3.50183606e-01
3.97134162e-02 4.24263105e-02 -9.64502513e-01 -4.57226008e-01
-7.04531223e-02 -2.23970726e-01 5.03382921e-01 2.49021441e-01
5.85198581e-01 4.14447248e-01 -6.41341686e-01 3.82487744e-01
1.53767860e+00 -1.47783300e-02 2.35803083e-01 5.70787132e-01
6.06948674e-01 5.15778363e-01 1.03898561e+00 6.15606368e-01
4.54478830e-01 1.04485083e+00 1.04964685e+00 4.47575539e-01
1.32632926e-01 -1.96758255e-01 9.72447544e-02 2.13268027e-01
-2.95448631e-01 6.81871250e-02 -1.08341432e+00 3.76773328e-01
-1.79492378e+00 -6.10099196e-01 -2.73472160e-01 2.77981043e+00
2.09746152e-01 3.74708861e-01 1.97377726e-01 -8.52889940e-02
6.27431452e-01 -4.00932580e-01 -8.35450172e-01 1.28372818e-01
2.89975345e-01 -2.88377523e-01 1.07041454e+00 3.87428313e-01
-1.00935411e+00 6.69502556e-01 6.45461178e+00 1.60618708e-01
-7.81866133e-01 -1.82298034e-01 -2.67537326e-01 -3.55502427e-01
9.89920348e-02 -9.06205457e-03 -1.41558611e+00 1.59969687e-01
4.10603464e-01 5.22653013e-02 3.77995074e-01 1.19034386e+00
-4.88139957e-01 -8.54329467e-01 -1.15093493e+00 1.16888976e+00
4.05006498e-01 -7.18369305e-01 -3.98687333e-01 2.95902997e-01
3.57379645e-01 2.92759627e-01 -1.58548467e-02 -1.33075520e-01
3.68890077e-01 -6.32532001e-01 1.17958438e+00 3.54421824e-01
7.57542491e-01 -6.30682409e-01 6.49426818e-01 6.40740693e-01
-1.14562511e+00 -3.35008116e-03 -5.20211339e-01 -1.43291771e-01
2.15466753e-01 5.79977334e-01 -1.34381628e+00 4.81894851e-01
1.16041172e+00 3.82985950e-01 -5.68390667e-01 1.47747302e+00
-1.07649550e-01 -3.38848144e-01 -8.63773882e-01 -1.72348261e-01
-2.93317318e-01 -1.38819620e-01 9.68284190e-01 7.91263878e-01
5.50242722e-01 -7.95143098e-02 2.36876577e-01 7.51068771e-01
2.87800640e-01 -3.04550678e-01 -7.10603535e-01 4.78592634e-01
6.19330704e-01 1.55178583e+00 -9.99993145e-01 1.73989147e-01
-1.91169694e-01 7.45927513e-01 3.28724355e-01 -9.37168524e-02
-5.80517709e-01 -3.25601101e-01 6.32535398e-01 1.73786789e-01
6.28116250e-01 -6.23964369e-01 -2.47620046e-01 -1.07195079e+00
6.97210491e-01 -3.18400145e-01 -1.55534431e-01 -9.83884811e-01
-1.05844176e+00 3.25090379e-01 4.96913165e-01 -1.70552969e+00
-2.08273217e-01 -1.05970514e+00 8.44706371e-02 7.72684813e-01
-1.23858845e+00 -1.24216282e+00 -7.60752439e-01 5.77932224e-02
4.08691883e-01 1.77441210e-01 8.08405340e-01 -1.21748656e-01
2.50439886e-02 1.89256623e-01 8.39117989e-02 -5.65708876e-01
6.82638705e-01 -1.36811578e+00 2.18722999e-01 6.84243381e-01
5.05884551e-02 5.74683726e-01 9.31263208e-01 -8.64840031e-01
-1.87681019e+00 -6.71120346e-01 1.05615050e-01 -1.02961755e+00
4.17814195e-01 -9.20555651e-01 -5.45644104e-01 6.57233596e-01
-4.46092039e-01 2.75390863e-01 3.41428041e-01 2.44379625e-01
-3.27943683e-01 -9.53563601e-02 -1.31627262e+00 3.98310125e-01
1.20592451e+00 -2.26912856e-01 -5.98734021e-01 3.35588098e-01
2.86429286e-01 -1.28060925e+00 -8.77849340e-01 7.55839944e-01
1.10789692e+00 -9.81007099e-01 1.13050222e+00 -2.36858930e-02
3.89395617e-02 -6.43628597e-01 -2.92339802e-01 -1.19686902e+00
-3.29106629e-01 -4.47935522e-01 -2.21817255e-01 8.20272624e-01
1.14462100e-01 -3.92462075e-01 9.47559118e-01 5.88178515e-01
-8.61420706e-02 -3.63592952e-01 -1.28286660e+00 -1.03106868e+00
-5.58459461e-01 -3.90367508e-01 2.14135081e-01 1.60257772e-01
-4.67260107e-02 4.91478778e-02 6.66333884e-02 6.36485994e-01
9.65349734e-01 2.26073310e-01 1.34818232e+00 -1.65819788e+00
2.16174759e-02 -1.12070881e-01 -8.69472504e-01 -9.92305875e-01
-2.64601916e-01 -4.88133013e-01 5.19447803e-01 -1.41721535e+00
2.40708143e-01 -3.32741022e-01 2.37792835e-01 1.06297567e-01
-8.06705281e-02 3.79625469e-01 2.53197104e-01 2.34858915e-01
-5.54028094e-01 3.08532566e-01 9.68602955e-01 1.92408085e-01
4.57388274e-02 4.27982677e-03 -1.93602905e-01 9.70904231e-01
3.72551799e-01 -4.84552979e-01 -1.22695509e-02 -3.84368092e-01
2.41805792e-01 -2.67651290e-01 6.67309701e-01 -1.41136980e+00
2.88703650e-01 -2.88472295e-01 6.38789535e-01 -1.20389783e+00
1.03554916e+00 -1.22223508e+00 2.98627794e-01 3.41541439e-01
1.94831863e-01 1.06813639e-01 3.80060226e-01 7.66190708e-01
4.18196440e-01 -2.43431136e-01 5.67419589e-01 -4.08847004e-01
-6.50003850e-01 1.24869965e-01 9.29672197e-02 -2.23667189e-01
1.43018639e+00 -6.03936017e-01 -3.65710467e-01 -8.77904370e-02
-3.49488556e-01 1.18262336e-01 9.87237334e-01 4.40656722e-01
7.36008584e-01 -1.14976883e+00 -4.37116146e-01 3.70588154e-01
5.50445378e-01 6.59445703e-01 -6.60460740e-02 8.61442626e-01
-7.31119275e-01 2.52384037e-01 -1.71379328e-01 -1.33641505e+00
-1.39574182e+00 3.02372545e-01 4.24819626e-02 3.94717216e-01
-6.58865690e-01 1.04896963e+00 -7.67117068e-02 -5.64811110e-01
1.77433819e-01 -3.66908103e-01 2.12629855e-01 -2.50208139e-01
5.68336666e-01 6.26252651e-01 3.44243586e-01 -6.12693965e-01
-5.25674403e-01 9.50642169e-01 -1.94788635e-01 -1.41690895e-01
1.42858160e+00 -7.79783726e-02 7.98157975e-02 6.86690629e-01
6.38510168e-01 1.36021897e-01 -1.75889277e+00 4.90273945e-02
-1.38565198e-01 -1.02143145e+00 -2.09217995e-01 -6.88247919e-01
-2.68397540e-01 6.71779573e-01 8.73093367e-01 1.46462321e-01
3.24656963e-01 3.44702214e-01 1.55193552e-01 4.88273770e-01
1.00198519e+00 -9.77378070e-01 4.30230945e-02 5.13871849e-01
1.22319019e+00 -1.40315747e+00 6.90251946e-01 -7.22406864e-01
-3.49431038e-01 1.17257464e+00 8.42975199e-01 -2.96598703e-01
2.04592139e-01 5.40676057e-01 -1.04780354e-01 -3.05171013e-01
-2.75807381e-01 -8.76038074e-02 4.11607146e-01 8.86001527e-01
-9.76109281e-02 4.99428473e-02 2.27122217e-01 2.19861362e-02
-4.00341302e-01 -2.84607321e-01 4.11555231e-01 1.29869843e+00
-6.38289392e-01 -6.51612461e-01 -6.85982049e-01 4.53418702e-01
-8.54522213e-02 4.12061930e-01 -2.61506736e-01 8.95529449e-01
-1.59819275e-01 6.80748165e-01 3.53116721e-01 -4.82021838e-01
7.51863003e-01 -2.61288375e-01 1.01902294e+00 -7.74374545e-01
-1.82550102e-01 8.74530599e-02 -1.35682046e-01 -8.26202214e-01
-4.32207108e-01 -9.02855456e-01 -8.46224844e-01 6.18509315e-02
-8.43436420e-01 -4.65099484e-01 1.17861760e+00 6.81707442e-01
3.47360075e-01 1.32215783e-01 2.01814130e-01 -1.64480984e+00
-5.89632392e-01 -9.32673633e-01 -6.37457430e-01 2.35567600e-01
2.78921217e-01 -1.38418055e+00 -4.10921365e-01 -3.26605499e-01] | [7.255337715148926, -2.3009254932403564] |
f6d3dae9-5f5b-4ce0-b6cf-b3dd0855774e | a-dense-material-segmentation-dataset-for | 2207.10614 | null | https://arxiv.org/abs/2207.10614v1 | https://arxiv.org/pdf/2207.10614v1.pdf | A Dense Material Segmentation Dataset for Indoor and Outdoor Scene Parsing | A key algorithm for understanding the world is material segmentation, which assigns a label (metal, glass, etc.) to each pixel. We find that a model trained on existing data underperforms in some settings and propose to address this with a large-scale dataset of 3.2 million dense segments on 44,560 indoor and outdoor images, which is 23x more segments than existing data. Our data covers a more diverse set of scenes, objects, viewpoints and materials, and contains a more fair distribution of skin types. We show that a model trained on our data outperforms a state-of-the-art model across datasets and viewpoints. We propose a large-scale scene parsing benchmark and baseline of 0.729 per-pixel accuracy, 0.585 mean class accuracy and 0.420 mean IoU across 46 materials. | ['Ransen Niu', 'Paul Upchurch'] | 2022-07-21 | null | null | null | null | ['scene-parsing', 'material-classification', 'material-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 6.05892658e-01 2.62501091e-02 -2.97490120e-01 -5.34012318e-01
-1.09589553e+00 -1.03821361e+00 3.25326800e-01 -1.59123372e-02
-1.68037593e-01 4.31780457e-01 1.51980430e-01 -2.00789660e-01
4.91429418e-01 -9.62390304e-01 -1.26562572e+00 -3.29378933e-01
4.02862698e-01 5.02924681e-01 6.28771424e-01 1.38348639e-01
1.87215015e-01 2.87320793e-01 -1.32009387e+00 6.47504449e-01
6.81002676e-01 1.20546317e+00 2.16186479e-01 8.66469860e-01
-2.05381230e-01 6.18306220e-01 -4.68143225e-01 -7.81650841e-01
3.45544964e-01 -9.46833640e-02 -1.22815895e+00 4.52057332e-01
1.32758641e+00 -3.88082653e-01 -2.38583058e-01 9.90209162e-01
2.20123768e-01 -1.94691330e-01 5.48844337e-01 -7.45011389e-01
-6.47897720e-01 6.50628030e-01 -8.14173639e-01 5.07603586e-02
3.99880350e-01 2.34628811e-01 1.02846336e+00 -4.04828161e-01
8.16141665e-01 1.06478989e+00 8.28366637e-01 6.65490925e-01
-1.12512851e+00 -4.48650122e-01 4.99745995e-01 -2.64851302e-01
-9.93695974e-01 -2.19627380e-01 4.94284153e-01 -4.35087770e-01
9.80039358e-01 3.46942157e-01 6.67963445e-01 1.21724033e+00
1.07896030e-01 1.03539240e+00 1.21557415e+00 -1.79275021e-01
1.13215461e-01 -1.08179584e-01 3.06283712e-01 8.58628750e-01
3.90414149e-01 -4.32011724e-01 -2.66382962e-01 1.03634171e-01
6.97246492e-01 -2.15469480e-01 1.79984689e-01 -2.49266863e-01
-1.08977628e+00 5.18638790e-01 5.36616623e-01 -2.21298441e-01
-1.29072368e-01 3.42846811e-01 2.27601960e-01 -2.56944358e-01
4.34356362e-01 4.84250784e-01 -8.52134049e-01 -5.82872629e-02
-5.66935122e-01 2.12827563e-01 8.25357795e-01 1.27636731e+00
7.36752391e-01 -1.74142480e-01 2.56579310e-01 9.13047314e-01
9.23573300e-02 7.67927110e-01 -6.43568933e-02 -1.56751692e+00
6.96283340e-01 4.77974802e-01 -6.14005364e-02 -6.29457653e-01
-4.71659839e-01 -2.42327079e-01 -2.46318102e-01 -1.36790439e-01
7.45769739e-01 -1.54828668e-01 -1.50993443e+00 1.44397497e+00
5.15031338e-01 -1.13326736e-01 -2.83907324e-01 5.88023126e-01
9.46692884e-01 7.02549994e-01 2.55388051e-01 4.42374200e-01
1.41330922e+00 -1.32893932e+00 -1.09339610e-01 -8.31137419e-01
3.79508704e-01 -9.60953653e-01 1.25776875e+00 7.79239595e-01
-1.11569083e+00 -5.34735620e-01 -8.80730331e-01 -3.85074854e-01
-4.70707029e-01 -1.51923656e-01 1.02926648e+00 8.81366313e-01
-8.21468234e-01 3.56610268e-01 -7.15305507e-01 -5.89293957e-01
7.95977950e-01 1.97114691e-01 -1.35478258e-01 -5.01602709e-01
-3.48165125e-01 3.76592606e-01 1.88065320e-01 -2.97780633e-01
-1.04572058e+00 -7.59690225e-01 -8.13553393e-01 -4.35478836e-01
4.22648072e-01 -6.45976365e-01 1.33689034e+00 -9.54709828e-01
-1.21509969e+00 1.29007590e+00 -5.40516898e-02 -2.57522970e-01
4.81848508e-01 -5.81259429e-01 -3.99504483e-01 2.85189271e-01
3.40514690e-01 1.00446939e+00 5.37260592e-01 -1.63350868e+00
-8.00741196e-01 -3.06497723e-01 3.56935441e-01 -1.41011532e-02
1.01635426e-01 6.20908551e-02 -9.47622657e-01 -5.17277360e-01
2.76874900e-01 -1.02042639e+00 -2.99198776e-01 -9.55693871e-02
-8.78365755e-01 1.26087129e-01 4.61231649e-01 -8.02745521e-01
7.01863527e-01 -2.05073524e+00 -3.17548901e-01 -1.21149510e-01
-2.06892211e-02 -2.65055180e-01 -2.22089335e-01 -2.14026738e-02
2.70637035e-01 4.02076632e-01 -4.01033282e-01 -4.35509443e-01
-1.88039526e-01 1.73516154e-01 -1.62013054e-01 3.04490954e-01
6.56708404e-02 8.52076113e-01 -7.25559473e-01 -7.50447273e-01
2.76898086e-01 2.65821457e-01 -6.49404645e-01 -7.25231618e-02
-5.23720682e-01 3.76846492e-01 -5.01868010e-01 1.33998632e+00
8.68278801e-01 -4.96021211e-01 3.46889533e-02 -2.02416629e-01
-1.87032372e-02 3.38007748e-01 -7.78428912e-01 2.19274044e+00
-2.83391327e-01 6.90573812e-01 1.08905666e-01 -3.57885987e-01
4.90215987e-01 -3.11835378e-01 6.25976741e-01 -5.94068348e-01
2.39330426e-01 4.34077084e-02 -1.07696839e-01 -4.18462545e-01
6.38777256e-01 2.91001767e-01 -4.83915299e-01 1.41229361e-01
-6.31321454e-03 -4.76461619e-01 1.94886953e-01 2.29927421e-01
1.51647174e+00 3.21949244e-01 -1.04278520e-01 -4.07351315e-01
-2.50710756e-01 4.28700715e-01 5.34093916e-01 8.20955992e-01
-3.40993851e-01 9.78299677e-01 3.81271631e-01 -5.88414073e-01
-1.19636047e+00 -1.33863306e+00 -3.92446011e-01 1.26360154e+00
4.54703122e-01 -2.11615682e-01 -1.20188081e+00 -8.17791939e-01
-1.05357207e-01 4.19965684e-01 -6.66967630e-01 3.79196107e-01
-6.51803732e-01 -8.89984488e-01 3.81451905e-01 9.11863625e-01
7.70254314e-01 -1.07368684e+00 -3.83337855e-01 2.87709609e-02
-2.62939692e-01 -1.75779927e+00 -3.31121802e-01 1.95023820e-01
-8.28378320e-01 -1.14796805e+00 -3.12273175e-01 -9.46110070e-01
6.65418148e-01 1.69156536e-01 1.85403061e+00 -2.98253506e-01
-4.64796901e-01 3.39222938e-01 -3.88823301e-01 -2.95340687e-01
-2.82603085e-01 2.44357392e-01 -4.92585123e-01 -4.52676296e-01
1.92542002e-01 -1.83383018e-01 -6.34352982e-01 3.04607809e-01
-6.72520339e-01 2.92319536e-01 6.22456908e-01 3.08061004e-01
8.32253873e-01 -6.69253543e-02 2.97838151e-02 -1.38638282e+00
-3.54329407e-01 -4.47562754e-01 -5.33081889e-01 4.08657402e-01
-1.27447292e-01 -2.79221803e-01 3.83936673e-01 -2.66012669e-01
-1.22545111e+00 4.26144153e-01 -3.77752453e-01 1.82910264e-01
-7.43168533e-01 -2.09814921e-01 -5.64931810e-01 3.45775373e-02
4.94002312e-01 -1.84735730e-01 -9.00181055e-01 -5.40668428e-01
6.12496674e-01 5.01456559e-01 8.20958495e-01 -1.03292179e+00
4.76807624e-01 7.51262963e-01 -2.60375887e-01 -7.96925366e-01
-1.31310570e+00 -4.67755079e-01 -7.95273721e-01 1.21576255e-02
1.30160129e+00 -1.04697263e+00 -3.33470792e-01 7.18286932e-01
-1.12036920e+00 -9.44205403e-01 -2.83244491e-01 8.24933872e-02
-4.75912482e-01 2.35218406e-01 -1.07245684e+00 -4.21028227e-01
-1.89365774e-01 -1.00875151e+00 1.42052293e+00 3.53161275e-01
-7.98895657e-02 -6.83949351e-01 -2.10880592e-01 9.48031008e-01
-4.74807583e-02 5.67659736e-01 6.90736294e-01 -9.15745348e-02
-9.00346279e-01 -5.78053519e-02 -5.71289897e-01 3.33974987e-01
1.98069543e-01 3.31161827e-01 -1.21444559e+00 1.79242924e-01
-3.89474124e-01 -5.70956051e-01 1.15094340e+00 6.15888178e-01
1.71105862e+00 8.30805153e-02 -4.12255228e-01 7.99524486e-01
1.63542593e+00 3.10266525e-01 7.77126789e-01 4.70083803e-01
1.23738241e+00 5.98570704e-01 5.68181634e-01 1.00186788e-01
5.69550812e-01 2.46440172e-01 5.88921428e-01 -2.14080229e-01
-4.02712524e-01 -2.56936520e-01 1.77174196e-01 3.67664963e-01
-4.81673842e-03 -5.44930160e-01 -1.06977510e+00 6.89847469e-01
-1.30683410e+00 -6.20738149e-01 -3.32603782e-01 1.71389735e+00
8.18883300e-01 6.27960324e-01 3.31621796e-01 -2.84665018e-01
6.35614753e-01 3.26203257e-01 -8.77870262e-01 -4.29565668e-01
-3.53259802e-01 2.79393733e-01 1.03014100e+00 3.28600109e-01
-1.51495016e+00 1.27437639e+00 7.54766560e+00 7.02162862e-01
-7.89023697e-01 -8.51510018e-02 1.24614120e+00 9.73600075e-02
-3.05667192e-01 -2.16990218e-01 -9.43265319e-01 4.84529912e-01
8.20813000e-01 7.17614353e-01 4.15313900e-01 1.19849026e+00
-4.81252611e-01 -3.01701844e-01 -1.00026643e+00 8.35523188e-01
1.71007350e-01 -1.18936431e+00 -2.05381930e-01 1.90444142e-01
1.29680586e+00 4.31812465e-01 1.16701148e-01 1.67930219e-02
8.27178597e-01 -9.11808312e-01 1.04434609e+00 5.09456396e-02
7.92945981e-01 -3.66410971e-01 3.30442578e-01 -1.76139530e-02
-1.22528470e+00 1.02641143e-01 -2.34800681e-01 8.09065998e-02
1.62473425e-01 6.51337266e-01 -4.60812867e-01 9.42557827e-02
1.18106627e+00 6.91280127e-01 -8.79418612e-01 9.28197145e-01
-1.74143404e-01 7.63583004e-01 -5.11888802e-01 1.88749626e-01
1.36133716e-01 4.43679132e-02 -1.80619210e-01 1.46201205e+00
-3.14571410e-02 1.09940797e-01 3.19866955e-01 4.89978790e-01
-4.37257737e-01 -2.14817673e-01 -4.71674204e-01 2.97865588e-02
3.99211943e-01 1.28906417e+00 -1.38821208e+00 -6.34587348e-01
-5.22062004e-01 1.06745446e+00 7.92407095e-02 2.90949374e-01
-1.10749507e+00 -4.50948700e-02 7.38609612e-01 2.02340633e-01
4.15708929e-01 -2.42436260e-01 -8.19725871e-01 -1.08527267e+00
6.88453689e-02 -7.91875660e-01 2.93838799e-01 -8.84470165e-01
-1.52380371e+00 4.92720664e-01 -1.53008804e-01 -7.88558602e-01
2.94891685e-01 -1.04489350e+00 -2.05366388e-01 1.79208413e-01
-1.31099391e+00 -1.45414245e+00 -5.30763388e-01 3.62600088e-01
9.40865099e-01 5.18557727e-01 5.46684146e-01 3.86402577e-01
-6.80422187e-01 2.49781758e-01 -6.29113466e-02 6.01417243e-01
4.58939046e-01 -1.49057674e+00 1.10571742e+00 9.05805171e-01
1.57334298e-01 2.18562454e-01 5.56670129e-01 -6.34541750e-01
-1.34942234e+00 -1.18307185e+00 3.49981219e-01 -1.09249675e+00
6.48868978e-01 -6.35174751e-01 -4.12224442e-01 9.90644217e-01
2.66982436e-01 8.91057327e-02 6.85757458e-01 2.51777232e-01
-6.86635196e-01 4.87437611e-03 -1.33829796e+00 4.23164308e-01
1.62985659e+00 -4.10031497e-01 -1.43325597e-01 5.53163886e-01
1.07807457e+00 -7.96005785e-01 -1.02383840e+00 4.91776198e-01
7.61446416e-01 -1.04112816e+00 1.17133999e+00 -4.39526916e-01
8.36886048e-01 3.68879810e-02 -6.92711473e-01 -7.70683289e-01
-3.04694265e-01 -2.30769411e-01 2.22663730e-02 1.29892635e+00
5.51948667e-01 -3.06967109e-01 1.27845335e+00 8.95757496e-01
-5.06524503e-01 -7.40741432e-01 -4.73408073e-01 -7.37264633e-01
3.38247359e-01 -7.62093306e-01 6.06836081e-01 6.65874362e-01
-7.07857668e-01 2.03107297e-01 -4.82392125e-02 1.46676019e-01
8.46395135e-01 4.30578291e-01 8.09964538e-01 -1.06246698e+00
-3.10363173e-01 -3.51402789e-01 -1.12697601e-01 -1.44209743e+00
-6.45545870e-02 -4.03248072e-01 2.15635329e-01 -1.98985159e+00
4.58022177e-01 -8.13759685e-01 -1.38283834e-01 5.45683146e-01
-8.48577321e-02 6.61738515e-01 5.13446070e-02 -1.18074678e-01
-1.05513465e+00 -1.81466073e-01 1.41174817e+00 -3.83646429e-01
1.51698425e-01 -2.41219193e-01 -9.34662700e-01 1.08214521e+00
9.36734557e-01 -2.82084376e-01 -1.41827285e-01 -1.11059606e+00
1.80811971e-01 -4.59509641e-01 2.33080342e-01 -1.08739674e+00
-3.76899660e-01 -4.23200637e-01 7.75743186e-01 -7.45475054e-01
6.11547410e-01 -6.34391308e-01 -9.39192483e-04 2.28124738e-01
-1.03121161e-01 -8.33424479e-02 2.80507058e-01 3.55656713e-01
2.92929977e-01 4.47576120e-02 7.28196800e-01 -5.03149807e-01
-1.13539791e+00 3.15922379e-01 1.55093044e-01 4.35903966e-01
9.29531574e-01 -3.16013336e-01 -8.58289719e-01 2.10965991e-01
-6.28375173e-01 -9.11167040e-02 9.57436442e-01 3.99636805e-01
2.63600528e-01 -1.02931738e+00 -3.58136088e-01 -3.01459003e-02
1.27249748e-01 4.19178516e-01 3.22799593e-01 2.05280527e-01
-9.09326613e-01 1.80189654e-01 -2.10379679e-02 -7.39003420e-01
-1.01370621e+00 3.71630132e-01 1.33941650e-01 -4.65233624e-02
-5.09494722e-01 1.33148301e+00 4.67942715e-01 -4.49403167e-01
-3.58225815e-02 -8.53037238e-01 4.24056411e-01 -2.40094885e-01
1.77483752e-01 4.46454316e-01 -2.23302264e-02 -6.45708799e-01
-4.15229291e-01 1.13312852e+00 2.36520562e-02 1.20989472e-01
1.22489071e+00 2.89345961e-02 7.98573196e-02 3.96073014e-01
1.27372074e+00 2.57456839e-01 -1.71888530e+00 5.75655587e-02
-2.69545317e-01 -5.83288133e-01 -3.36002171e-01 -1.06568491e+00
-1.20075214e+00 6.49123013e-01 4.51039225e-01 2.89231062e-01
1.18868160e+00 5.71646631e-01 1.12744021e+00 2.12336242e-01
7.46866703e-01 -1.26608407e+00 1.42359942e-01 4.24375921e-01
2.70118028e-01 -1.66884375e+00 1.69708550e-01 -1.10129011e+00
-6.29351437e-01 6.98916316e-01 8.07078362e-01 -1.75814450e-01
5.80203235e-01 5.38501143e-01 8.48361775e-02 -2.61004299e-01
-3.46449524e-01 -2.53092498e-01 1.33566201e-01 6.40826762e-01
2.32247114e-01 4.65995014e-01 2.49705106e-01 3.46440792e-01
-5.31462610e-01 -4.14705694e-01 2.75947154e-01 1.09081984e+00
-6.00329697e-01 -9.49334025e-01 -3.13217252e-01 5.66844523e-01
-6.73236847e-01 -3.25371288e-02 -5.53568244e-01 6.74438119e-01
4.98929650e-01 1.19674802e+00 2.54355520e-01 -4.46387351e-01
2.55245537e-01 -4.75173086e-01 8.52730632e-01 -7.94516861e-01
-3.89959902e-01 1.80618539e-01 3.63882780e-01 -7.83510983e-01
-5.10889232e-01 -9.22152460e-01 -1.34208751e+00 -2.46782377e-01
-1.61232904e-01 -5.08103013e-01 8.38869810e-01 6.57630682e-01
2.09399417e-01 5.78842700e-01 3.93284351e-01 -7.62343109e-01
3.34659517e-01 -6.25066876e-01 -5.38946509e-01 5.85197091e-01
2.49872431e-02 -3.26101512e-01 -1.11895405e-01 4.00416374e-01] | [9.65163516998291, 0.3722502887248993] |
363745f2-d294-4826-89af-43fd7a0091d8 | antisymmetricrnn-a-dynamical-system-view-on | 1902.09689 | null | http://arxiv.org/abs/1902.09689v1 | http://arxiv.org/pdf/1902.09689v1.pdf | AntisymmetricRNN: A Dynamical System View on Recurrent Neural Networks | Recurrent neural networks have gained widespread use in modeling sequential
data. Learning long-term dependencies using these models remains difficult
though, due to exploding or vanishing gradients. In this paper, we draw
connections between recurrent networks and ordinary differential equations. A
special form of recurrent networks called the AntisymmetricRNN is proposed
under this theoretical framework, which is able to capture long-term
dependencies thanks to the stability property of its underlying differential
equation. Existing approaches to improving RNN trainability often incur
significant computation overhead. In comparison, AntisymmetricRNN achieves the
same goal by design. We showcase the advantage of this new architecture through
extensive simulations and experiments. AntisymmetricRNN exhibits much more
predictable dynamics. It outperforms regular LSTM models on tasks requiring
long-term memory and matches the performance on tasks where short-term
dependencies dominate despite being much simpler. | ['Ed H. Chi', 'Eldad Haber', 'Bo Chang', 'Minmin Chen'] | 2019-02-26 | antisymmetricrnn-a-dynamical-system-view-on-1 | https://openreview.net/forum?id=ryxepo0cFX | https://openreview.net/pdf?id=ryxepo0cFX | iclr-2019-5 | ['sequential-image-classification'] | ['computer-vision'] | [-1.20928951e-01 3.24076451e-02 -1.67565718e-01 -1.63458109e-01
-3.14167179e-02 -4.25473839e-01 6.90118551e-01 -4.73363340e-01
-4.98356521e-01 8.48876953e-01 1.94043607e-01 -5.60301304e-01
-3.66971530e-02 -3.09276342e-01 -6.59606218e-01 -7.38862336e-01
-3.04015011e-01 1.92008689e-01 1.50174834e-02 -6.31317914e-01
2.92842746e-01 6.82090819e-01 -9.97235954e-01 6.54160902e-02
6.08929574e-01 7.08278537e-01 1.98155671e-01 8.19043636e-01
-2.16383457e-01 1.38298464e+00 -2.64465421e-01 -4.81977239e-02
1.44474491e-01 -4.48025167e-01 -9.41489220e-01 -5.01036286e-01
1.15817465e-01 -1.01668954e-01 -7.12164164e-01 7.10250258e-01
4.25790846e-01 3.71790439e-01 4.66130942e-01 -7.97899187e-01
-7.47026384e-01 8.69019449e-01 -2.75141150e-01 7.33878314e-01
-7.35026821e-02 -1.37794092e-01 1.15946448e+00 -8.61525357e-01
4.54077333e-01 1.13547921e+00 1.17122650e+00 9.26554024e-01
-1.39529455e+00 -6.21678114e-01 6.20428503e-01 -1.84639573e-01
-8.99327219e-01 -4.40139323e-01 6.89566731e-01 -1.75804287e-01
1.59627640e+00 5.60172573e-02 4.81747359e-01 1.35940778e+00
4.41908360e-01 9.59226012e-01 8.07557404e-01 -2.21065789e-01
-2.52581447e-01 -6.96590841e-02 5.46234071e-01 5.21652937e-01
-1.15556009e-02 1.86994016e-01 -5.38461804e-01 1.00234360e-01
9.75188673e-01 2.49386683e-01 -1.85114652e-01 -1.23415649e-01
-8.74210119e-01 7.75433540e-01 6.04345918e-01 7.78934062e-01
-3.61167222e-01 5.52097857e-01 5.59274256e-01 7.80913234e-01
6.32453680e-01 4.72051114e-01 -6.21472836e-01 -2.72697747e-01
-7.57264853e-01 3.90706547e-02 9.51900184e-01 7.08827972e-01
3.58915716e-01 5.91097534e-01 6.83252513e-02 8.10915291e-01
3.45717184e-02 2.49443814e-01 9.26388204e-01 -6.20750725e-01
3.52449775e-01 1.68155938e-01 -2.80030072e-02 -9.19202387e-01
-8.53989959e-01 -8.16375792e-01 -1.24750340e+00 -8.02552998e-02
2.11385176e-01 -5.96983969e-01 -1.01083672e+00 1.93757463e+00
-3.43393177e-01 2.42562279e-01 3.56862575e-01 5.74299037e-01
2.86843896e-01 1.04291332e+00 -1.31488711e-01 -3.33889127e-01
4.30607259e-01 -1.04336691e+00 -9.35133636e-01 -3.40833604e-01
8.89447510e-01 -3.70836347e-01 6.22521996e-01 8.57076943e-02
-1.08250475e+00 -3.39864671e-01 -7.52795100e-01 -8.92069712e-02
-3.62270057e-01 -9.31394249e-02 7.96364486e-01 9.79224965e-02
-1.65453458e+00 1.15271318e+00 -1.23338366e+00 -2.28920087e-01
-1.39551669e-01 7.08139420e-01 6.41551986e-02 6.24122679e-01
-1.58793914e+00 1.26036203e+00 1.44849271e-01 7.35605061e-01
-6.13078535e-01 -9.63915884e-01 -9.29848015e-01 -9.55000706e-03
-3.13957892e-02 -4.02917027e-01 1.67876232e+00 -1.00670624e+00
-1.79883790e+00 3.75067115e-01 -4.34402555e-01 -9.89415944e-01
4.85358983e-01 -5.10430515e-01 -2.42725864e-01 -3.39437097e-01
-3.97547901e-01 3.06664556e-01 8.18735242e-01 -6.36778057e-01
-3.75362426e-01 -8.06931481e-02 -2.20062723e-03 1.77930608e-01
-4.27785337e-01 -1.16932094e-01 6.27611578e-02 -8.29021096e-01
-1.69823244e-02 -1.22073352e+00 -7.35655308e-01 -5.51876366e-01
-2.66171724e-01 -3.86159986e-01 1.01359999e+00 -5.08233786e-01
1.52380049e+00 -1.85488939e+00 2.97540039e-01 1.47307590e-01
2.05904156e-01 7.03564048e-01 -4.32237647e-02 5.22908628e-01
-3.77711773e-01 1.93040460e-01 -1.01342566e-01 -4.19950247e-01
-2.43720740e-01 4.11674052e-01 -7.17589319e-01 3.09619874e-01
3.42749625e-01 1.20497298e+00 -8.22180510e-01 3.25800069e-02
-7.02053076e-03 7.64360011e-01 -2.45004982e-01 -6.01402074e-02
-1.40053481e-01 2.23843485e-01 -3.38411063e-01 -6.74799085e-02
1.21432386e-01 -3.99003863e-01 3.33438396e-01 2.48524696e-01
-3.89649332e-01 6.28862798e-01 -7.54063904e-01 1.41046548e+00
-9.69206691e-01 1.22347188e+00 -1.15114912e-01 -1.32388926e+00
9.27813530e-01 4.53891426e-01 4.07092810e-01 -8.41329277e-01
-8.37306604e-02 3.41997236e-01 1.08956032e-01 -2.79643804e-01
3.41394216e-01 -4.37290519e-01 1.03113174e-01 6.54398620e-01
2.71594729e-02 2.63879061e-01 2.10686132e-01 3.11776847e-01
1.11051500e+00 3.75039615e-02 -6.27478883e-02 -4.77241904e-01
4.08046663e-01 -4.47917461e-01 5.43489814e-01 9.14796591e-01
8.32850337e-02 3.26402545e-01 5.13634145e-01 -7.24029958e-01
-1.12178516e+00 -6.54850483e-01 -5.45423180e-02 1.04429007e+00
-3.21452916e-01 -2.92327821e-01 -3.46359670e-01 -3.30553949e-01
-5.24481619e-03 6.14024818e-01 -8.62662494e-01 -2.64152884e-01
-1.10135067e+00 -8.42452765e-01 6.87650621e-01 8.78621578e-01
3.26403648e-01 -1.38625419e+00 -6.31729603e-01 4.85496223e-01
5.31555302e-02 -8.12284946e-01 -4.97706801e-01 5.89488447e-01
-1.28137398e+00 -3.88987482e-01 -1.17050648e+00 -8.88243854e-01
2.86947638e-01 5.95843233e-02 1.16002655e+00 6.60670847e-02
1.73374135e-02 1.66266233e-01 -4.26732842e-03 -2.21885949e-01
-3.44129205e-01 5.72569370e-01 2.99069047e-01 -1.99688196e-01
1.10435754e-01 -8.82098794e-01 -3.57177824e-01 4.61848117e-02
-8.04871738e-01 -3.63526084e-02 6.05777740e-01 9.75461304e-01
1.52476519e-01 -2.20805749e-01 7.33107805e-01 -9.92676497e-01
8.91922593e-01 -3.44619691e-01 -5.61309636e-01 1.86531439e-01
-7.94287622e-01 6.96188569e-01 1.00761306e+00 -6.02103114e-01
-1.19876218e+00 -1.30607516e-01 -1.74421817e-01 -3.47122848e-01
3.84890556e-01 7.85674155e-01 7.33709514e-01 -1.31371811e-01
3.30401868e-01 2.75287062e-01 5.87866157e-02 -5.78589678e-01
1.68477520e-01 2.06388190e-01 1.75939173e-01 -1.65326118e-01
5.69963694e-01 2.67989725e-01 -4.05846126e-02 -1.09882021e+00
-1.16049910e+00 -3.18095595e-01 -6.27859771e-01 4.09300663e-02
4.63064313e-01 -6.91509426e-01 -7.29588389e-01 7.83673108e-01
-1.41289377e+00 -6.62260473e-01 -3.36050630e-01 5.60292482e-01
-4.85773057e-01 -7.82688186e-02 -1.32227492e+00 -1.06930196e+00
-4.23786849e-01 -4.79053259e-01 4.14341539e-01 2.22423866e-01
-2.27044493e-01 -1.68876922e+00 4.26895976e-01 -6.67074025e-01
8.99683475e-01 -5.01695611e-02 7.15027869e-01 -4.71865386e-01
-4.23159115e-02 -9.79882181e-02 -1.47692477e-02 5.42875528e-01
-2.21656393e-02 1.18805200e-01 -7.29113817e-01 -2.21620664e-01
2.29896724e-01 -2.07123667e-01 1.28666770e+00 6.25073195e-01
7.73307681e-01 -1.78291678e-01 -2.91776568e-01 5.64094305e-01
1.28449142e+00 3.03694248e-01 4.46732670e-01 1.65735140e-01
9.44858551e-01 4.83572900e-01 -3.65768969e-02 2.46416777e-01
1.83498219e-01 3.38321567e-01 4.14053835e-02 -2.93518454e-01
1.37198418e-01 -1.43781319e-01 8.44242990e-01 1.37935078e+00
-2.68714339e-01 -5.09296991e-02 -1.05600703e+00 4.39056396e-01
-2.13703346e+00 -1.17775869e+00 -1.77876338e-01 1.74145973e+00
8.22551906e-01 4.33737606e-01 1.10969096e-01 -3.23958173e-02
5.62053025e-01 4.03203160e-01 -8.32959294e-01 -9.94957089e-01
-2.64650792e-01 2.67678559e-01 5.83409429e-01 7.32103288e-01
-8.22835028e-01 1.04086459e+00 7.69113493e+00 3.80434126e-01
-1.56915963e+00 -9.71844867e-02 4.37834173e-01 -2.62767762e-01
-1.38560846e-01 -2.04900816e-01 -1.17229927e+00 1.54962376e-01
1.58856595e+00 -2.38494024e-01 3.62127483e-01 5.53452671e-01
2.98316211e-01 3.14862520e-01 -1.16496754e+00 9.42893326e-01
-4.16446570e-03 -1.56450200e+00 -6.80550262e-02 -1.63587347e-01
8.65274191e-01 4.29586411e-01 3.58823150e-01 4.98958558e-01
6.29078746e-01 -1.20969772e+00 3.84458005e-01 7.32219577e-01
4.15637314e-01 -8.26053619e-01 5.60006320e-01 3.89981180e-01
-1.14077818e+00 -3.16366106e-01 -3.41261178e-01 -6.30000532e-01
1.95710003e-01 5.82527339e-01 -3.24872345e-01 1.69268206e-01
5.67381322e-01 1.33536029e+00 -2.30379656e-01 5.74919701e-01
-1.47747621e-01 6.15035772e-01 -4.62000698e-01 -3.22519451e-01
8.44871283e-01 -2.81786680e-01 3.63645345e-01 1.58413303e+00
7.92157874e-02 4.33284529e-02 -2.02667385e-01 6.41665161e-01
1.52105326e-02 -2.19848186e-01 -9.20032084e-01 -2.53731757e-01
-1.15576573e-01 7.96950042e-01 -5.26779711e-01 -1.90228403e-01
-4.21568125e-01 1.09935200e+00 7.31657624e-01 8.20847690e-01
-7.95256317e-01 -4.47890699e-01 7.08813787e-01 -1.76514655e-01
4.32112843e-01 -4.98229265e-01 -2.60856837e-01 -1.32808828e+00
6.08829595e-02 -5.60593605e-01 2.92059571e-01 -4.03365284e-01
-1.10281301e+00 8.61928403e-01 -3.10060710e-01 -6.57119870e-01
-7.64237583e-01 -7.61488378e-01 -6.21547520e-01 9.24707949e-01
-1.45577121e+00 -7.35917926e-01 3.52756590e-01 5.64243495e-01
6.30169511e-01 1.71408087e-01 8.31788242e-01 1.77904397e-01
-9.58170235e-01 4.75772351e-01 4.98756319e-01 1.14421867e-01
9.38956290e-02 -1.26046586e+00 8.10452223e-01 7.73199797e-01
1.96528926e-01 9.10162330e-01 9.12535548e-01 -3.86354506e-01
-1.27050972e+00 -9.42556381e-01 1.26626563e+00 -3.66522104e-01
1.11061990e+00 -4.81171101e-01 -1.03892219e+00 1.15148699e+00
2.42448300e-01 -6.88770115e-02 2.58399487e-01 4.13138509e-01
-3.00985873e-01 -2.14398056e-02 -3.71538967e-01 7.96798885e-01
1.09550977e+00 -8.87435138e-01 -5.05027413e-01 2.24656135e-01
7.69867122e-01 -2.24853516e-01 -5.42083979e-01 3.78734916e-01
6.69179022e-01 -9.31247890e-01 6.53797090e-01 -9.79858220e-01
3.09295923e-01 3.29114944e-01 2.95504332e-01 -1.55688798e+00
-3.18892270e-01 -1.16296959e+00 -3.80718440e-01 8.28208864e-01
8.47992182e-01 -8.42857122e-01 7.04909921e-01 7.13402271e-01
-2.84779165e-03 -9.70375478e-01 -5.54484785e-01 -1.08001125e+00
5.50911367e-01 -4.04170603e-01 -7.73494616e-02 9.62019384e-01
1.34253427e-01 6.97693110e-01 -7.03891397e-01 -2.26864770e-01
1.17400967e-01 6.94847405e-02 1.68290794e-01 -1.15641034e+00
-1.58021986e-01 -6.45476937e-01 -1.80792496e-01 -1.51292539e+00
6.43894672e-01 -7.30391324e-01 6.33623526e-02 -1.32698011e+00
-1.49949431e-01 -4.79842603e-01 -7.03426301e-01 3.40427428e-01
1.74234331e-01 1.28570929e-01 1.19089425e-01 2.34231830e-01
-4.53750521e-01 5.49557447e-01 8.95472288e-01 2.55597889e-01
-6.34479880e-01 3.44669610e-01 -3.54785442e-01 8.63817096e-01
1.13672483e+00 -4.05663550e-01 -5.04305124e-01 -6.32013679e-01
5.33823967e-01 1.42827526e-01 9.75894630e-02 -8.07534397e-01
4.16268051e-01 8.73464271e-02 1.67183787e-01 -6.38470471e-01
3.04027200e-01 -5.00922561e-01 -5.73873408e-02 7.97890484e-01
-8.18757534e-01 5.56630433e-01 2.49995768e-01 3.80451232e-01
-2.79576212e-01 1.38484780e-02 7.84535706e-01 -1.37413874e-01
-5.84181070e-01 2.74139255e-01 -8.27976048e-01 1.17588542e-01
5.79384565e-01 4.67731506e-02 3.64631452e-02 -7.68765628e-01
-9.28830862e-01 2.04725519e-01 -1.61108389e-01 4.32374239e-01
6.26495004e-01 -1.16977048e+00 -4.42030430e-01 9.95200053e-02
-7.40949154e-01 -2.74879098e-01 1.02340773e-01 1.12517810e+00
-3.54848504e-01 9.63723183e-01 4.52635698e-02 -3.15618366e-01
-9.28895116e-01 4.77300227e-01 7.45290279e-01 -8.04222047e-01
-9.14095700e-01 1.01138735e+00 2.19067171e-01 -4.38289881e-01
4.23173547e-01 -5.57451069e-01 -2.96482712e-01 1.96782008e-01
4.94861126e-01 2.90571630e-01 2.66689509e-02 -4.53788012e-01
-2.74465948e-01 4.42141652e-01 -5.28548181e-01 -1.37602270e-01
1.49886572e+00 -1.90318570e-01 -1.88563794e-01 1.11404824e+00
1.37896180e+00 -5.01403213e-01 -1.32506347e+00 -4.46257204e-01
4.97816652e-01 4.01437193e-01 1.05526894e-01 -3.04609418e-01
-1.16809559e+00 1.15387881e+00 2.88070887e-01 4.99135435e-01
8.16646755e-01 -3.61481279e-01 1.03463125e+00 8.92447054e-01
2.17503589e-02 -1.05564010e+00 1.15279280e-01 1.48683226e+00
7.50600696e-01 -9.60996568e-01 -4.24834937e-01 2.62951195e-01
-4.64945346e-01 1.38228071e+00 5.31742930e-01 -6.03794992e-01
9.53749657e-01 4.92442846e-01 1.69471636e-01 -7.28849769e-02
-1.25360119e+00 1.11941723e-02 -3.43909184e-03 3.03421021e-01
7.44350970e-01 -2.82998770e-01 -1.96963266e-01 9.48689580e-02
-1.22146294e-01 -1.80177152e-01 5.09879529e-01 8.46950233e-01
-1.61156595e-01 -9.18534577e-01 1.47413522e-01 3.79195452e-01
-7.09886074e-01 -4.40943837e-01 -2.82545477e-01 7.69785225e-01
-8.46509695e-01 6.22822165e-01 2.57633001e-01 -3.40071529e-01
2.00550631e-01 1.42681271e-01 2.49329925e-01 -3.93653125e-01
-8.40652049e-01 -1.32429838e-01 3.25163342e-02 -6.11925423e-01
-4.35957968e-01 -6.61787987e-01 -1.48329818e+00 -5.11421323e-01
-2.87147552e-01 2.51192808e-01 5.76194942e-01 1.15600049e+00
3.77289265e-01 5.84877491e-01 5.79534769e-01 -9.35483158e-01
-9.13134634e-01 -9.13322031e-01 -5.77621341e-01 1.85810775e-01
9.46659446e-01 -2.81879544e-01 -4.64680135e-01 -1.46232978e-01] | [7.6716718673706055, 3.4292569160461426] |
dd860527-b366-4f68-be6b-5f236ad3306b | takelab-at-semeval-2017-task-6 | null | null | https://aclanthology.org/S17-2066 | https://aclanthology.org/S17-2066.pdf | TakeLab at SemEval-2017 Task 6: \#RankingHumorIn4Pages | This paper describes our system for humor ranking in tweets within the SemEval 2017 Task 6: {\#}HashtagWars (6A and 6B). For both subtasks, we use an off-the-shelf gradient boosting model built on a rich set of features, handcrafted to provide the model with the external knowledge needed to better predict the humor in the text. The features capture various cultural references and specific humor patterns. Our system ranked 2nd (officially 7th) among 10 submissions on the Subtask A and 2nd among 9 submissions on the Subtask B. | ['Jan {\\v{S}}najder', "Domagoj Alagi{\\'c}", "Antonio {\\v{S}}ajatovi{\\'c}", "Ivan Mr{\\v{s}}i{\\'c}", 'Marin Kukova{\\v{c}}ec', 'Juraj Malenica'] | 2017-08-01 | null | null | null | semeval-2017-8 | ['humor-detection'] | ['natural-language-processing'] | [-7.40589261e-01 -9.41855684e-02 -1.14132658e-01 -2.01300308e-01
-5.86012602e-01 -5.23232758e-01 9.40133154e-01 4.30417806e-01
-4.13497627e-01 8.00673366e-01 8.06287050e-01 -3.53309810e-01
1.60032585e-01 -6.65739357e-01 -3.72114718e-01 -2.54666984e-01
-8.28874707e-02 4.77485955e-01 3.00435096e-01 -1.27188861e+00
6.88999832e-01 -5.60583081e-04 -1.25600517e+00 8.43708634e-01
7.78677464e-01 6.94395304e-01 -2.81950563e-01 7.78071821e-01
-2.37699766e-02 1.75493515e+00 -5.08832693e-01 -9.89263773e-01
4.11082134e-02 -6.15796924e-01 -1.14938533e+00 -6.48729563e-01
6.42966807e-01 8.14711973e-02 -6.58696353e-01 6.48934066e-01
4.68367100e-01 3.10311079e-01 5.42163014e-01 -1.09385085e+00
-4.17740047e-01 1.15845966e+00 -3.03337812e-01 4.91529018e-01
5.69849432e-01 -1.40535673e-02 1.37437344e+00 -1.36577797e+00
7.06913292e-01 9.99172568e-01 8.36649716e-01 6.09463215e-01
-8.22627544e-01 -8.46823573e-01 -5.56272984e-01 8.46963942e-01
-8.74926031e-01 -2.40653232e-01 1.05586874e+00 -1.03147209e+00
6.45034671e-01 4.69750047e-01 7.78403580e-01 1.07011700e+00
-2.21482441e-02 1.01356423e+00 1.47288716e+00 -2.59312857e-02
2.94511523e-02 3.88909549e-01 8.18146110e-01 6.94906890e-01
-9.51036140e-02 -2.73183316e-01 -1.15762854e+00 -3.94648254e-01
-2.42588431e-01 -3.05372924e-01 -2.63233334e-01 3.59643698e-01
-8.59897733e-01 1.40153193e+00 8.13999355e-01 2.75736153e-01
-2.29551211e-01 -3.29707742e-01 6.30195200e-01 3.05816710e-01
5.80977261e-01 7.76536345e-01 -2.42785029e-02 -2.35961989e-01
-1.18115330e+00 8.38471413e-01 1.07789540e+00 4.73302901e-01
8.42601895e-01 -2.97394782e-01 -5.29913723e-01 8.94661188e-01
-2.11276323e-01 4.47362542e-01 5.69779754e-01 -5.36067069e-01
4.56517965e-01 6.35077119e-01 2.97947437e-01 -1.33315241e+00
-7.94691741e-01 -7.55801082e-01 -5.27622581e-01 -1.44706577e-01
4.57344443e-01 -1.24788694e-02 -3.16726893e-01 1.27179527e+00
-4.69906889e-02 -7.40422010e-02 -4.24054414e-01 9.83634114e-01
1.61534023e+00 6.83512628e-01 -2.22756520e-01 -7.66509175e-02
1.17910242e+00 -1.45450962e+00 -4.14437443e-01 -1.74033433e-01
6.46820664e-01 -1.03048038e+00 1.21925700e+00 3.64986271e-01
-1.08302104e+00 -3.01032931e-01 -1.13275909e+00 -3.38745177e-01
-4.98467684e-01 -3.60752314e-01 2.30568409e-01 3.74853611e-01
-9.09619451e-01 6.97980046e-01 5.52311242e-02 -1.98493659e-01
5.87174334e-02 -5.48602119e-02 5.84317837e-03 8.46902728e-02
-1.77236032e+00 1.45227385e+00 1.25426084e-01 -3.90131384e-01
-9.49048519e-01 -7.73078144e-01 -4.83690083e-01 -1.70998439e-01
-3.07601750e-01 -3.87417704e-01 1.17997122e+00 -6.54730618e-01
-1.42415738e+00 1.35118961e+00 1.23910168e-02 -5.91812074e-01
5.26202142e-01 -3.28152508e-01 -3.84431481e-01 -1.96729764e-01
1.76075786e-01 -7.51752555e-02 6.34517014e-01 -1.09049129e+00
-5.93450725e-01 -5.00527062e-02 6.10538274e-02 -5.14079034e-02
-7.73824513e-01 4.85644698e-01 1.04337700e-01 -2.92531222e-01
-2.87665635e-01 -8.49854946e-01 2.64128745e-01 -1.40228617e+00
-2.54959583e-01 -3.47976327e-01 7.15212584e-01 -1.05026186e+00
1.79756749e+00 -1.67064595e+00 2.00031877e-01 1.89364538e-01
5.48944533e-01 7.20535517e-02 3.90310064e-02 8.00706863e-01
2.71798968e-01 -1.50191143e-01 4.16193008e-01 -2.55414188e-01
3.62778865e-02 -3.90763223e-01 -6.12726569e-01 3.98870468e-01
-5.35649478e-01 7.82442212e-01 -1.39491129e+00 -3.86897475e-01
-1.47295266e-01 -4.85286396e-03 -3.85918796e-01 4.00097907e-01
1.37002289e-01 4.12968487e-01 -2.28474692e-01 2.96183199e-01
4.88811404e-01 -3.46540391e-01 -1.12634823e-01 2.61940300e-01
-2.71949887e-01 9.15557861e-01 -1.35606155e-01 9.33113873e-01
-4.68295515e-01 9.35714722e-01 -3.22147995e-01 -1.81677237e-01
1.16337836e+00 -1.00708976e-02 3.13686341e-01 -7.71272480e-01
2.35551298e-01 3.77179146e-01 -1.10616744e-01 -3.98510903e-01
9.24429476e-01 -2.72222757e-01 -4.76670682e-01 4.74975616e-01
-4.57330160e-02 -1.65545464e-01 3.08992833e-01 6.91534817e-01
1.13455105e+00 -4.39150244e-01 3.78561765e-01 -7.38887548e-01
8.47889483e-01 7.14172125e-02 3.17963660e-01 1.00675106e+00
-2.27562830e-01 5.90703070e-01 4.41717803e-01 -1.17035687e+00
-1.19907665e+00 -5.23069024e-01 -1.99129041e-02 1.90244758e+00
-1.48116261e-01 -8.42563510e-01 -4.63443369e-01 -7.08763182e-01
6.38113767e-02 9.32406902e-01 -9.39955294e-01 9.62161422e-02
-6.64179027e-01 -9.12790477e-01 5.91281533e-01 1.16380103e-01
4.62472945e-01 -9.14426446e-01 -6.77041292e-01 4.32345532e-02
-8.09673786e-01 -7.99121857e-01 -4.12288100e-01 2.93917149e-01
-4.17141825e-01 -8.70498896e-01 -3.62906098e-01 -6.72017157e-01
-1.41466490e-03 4.02358830e-01 1.74760079e+00 6.12121165e-01
2.81221777e-01 -1.94486126e-01 -8.10991168e-01 -4.10825938e-01
-2.73360342e-01 3.10501575e-01 -1.07471496e-01 -1.92929342e-01
5.66029191e-01 -2.94324309e-01 -3.86820346e-01 2.19004691e-01
-1.93360880e-01 1.69134468e-01 -2.13867292e-01 1.20174253e+00
-3.92370969e-01 -4.17463452e-01 4.53053147e-01 -9.22282636e-01
7.98981071e-01 -9.06811118e-01 -8.85562450e-02 -1.95696995e-01
-4.65707242e-01 -4.67962354e-01 9.36989963e-01 -9.97784808e-02
-7.17123926e-01 -4.25622880e-01 9.20532085e-03 1.56659201e-01
5.41176140e-01 8.24766338e-01 4.30051595e-01 -4.61889394e-02
1.36955059e+00 3.09070721e-02 -6.05831563e-01 -5.45047820e-01
2.68078595e-01 8.97457957e-01 5.23568690e-01 -4.53822553e-01
9.98903990e-01 1.58917010e-01 -5.86893298e-02 -6.75766706e-01
-1.74217987e+00 -8.88423741e-01 -5.85149705e-01 -5.62579036e-01
5.35836637e-01 -1.03387201e+00 -7.84251928e-01 5.60433447e-01
-1.32229495e+00 -3.95677030e-01 2.68599838e-01 9.28441137e-02
-2.48265237e-01 5.61886616e-02 -1.08018935e+00 -8.16804290e-01
-8.58937919e-01 -3.88222992e-01 4.22035396e-01 6.94479570e-02
-6.88103676e-01 -8.39717865e-01 6.20167434e-01 1.05058920e+00
3.79864156e-01 2.34288767e-01 9.74971652e-01 -1.01771009e+00
1.07761987e-01 -3.20302635e-01 -6.45782277e-02 -7.94065967e-02
-5.88454783e-01 -2.28027374e-01 -1.20106554e+00 -2.17062339e-01
-1.86119989e-01 -8.65979075e-01 1.22911477e+00 -1.22862928e-01
8.45324516e-01 -7.22331405e-01 1.96059927e-01 1.68116808e-01
8.77789378e-01 -4.85780150e-01 7.60049224e-01 9.67457831e-01
5.65022051e-01 7.26445377e-01 4.30482566e-01 9.20563698e-01
7.31144428e-01 7.24136353e-01 4.52978760e-01 1.33259505e-01
1.49498703e-02 -6.36558473e-01 5.86698472e-01 1.36474013e+00
-2.25087330e-01 1.13269491e-02 -1.29309261e+00 6.96354687e-01
-2.01241517e+00 -1.42255366e+00 -8.97937775e-01 2.09882045e+00
1.08049726e+00 9.56719816e-02 8.29078496e-01 -8.29722509e-02
4.71664369e-01 4.95376498e-01 1.88715264e-01 -6.89767182e-01
-3.91522408e-01 -4.61182073e-02 1.64640769e-01 9.17677760e-01
-1.01617014e+00 1.18914485e+00 6.41352177e+00 7.33323693e-01
-8.55042815e-01 3.06003302e-01 4.83644485e-01 -3.36376309e-01
-4.26766396e-01 6.65697232e-02 -8.42774272e-01 5.56517601e-01
1.03082609e+00 -3.89889032e-01 5.54339707e-01 7.72540092e-01
1.96144313e-01 1.69032551e-02 -7.95690536e-01 7.73325741e-01
6.11234307e-01 -1.28469443e+00 -1.42500460e-01 -2.50557125e-01
1.17193985e+00 4.17503297e-01 -8.07221457e-02 8.49724531e-01
5.32656491e-01 -1.19620574e+00 1.01639760e+00 3.65441620e-01
-1.31459264e-02 -8.27034652e-01 1.21369183e+00 5.89565516e-01
-4.52031583e-01 -2.55696267e-01 -3.21575671e-01 -7.46614635e-01
-1.83534212e-02 8.39563549e-01 -1.10496247e+00 8.27293769e-02
8.59905362e-01 7.94258296e-01 -9.28222060e-01 9.76640344e-01
-5.09145021e-01 1.03042698e+00 2.41802156e-01 -6.46463096e-01
3.92771393e-01 2.41865695e-01 7.39798903e-01 1.77815545e+00
1.70059726e-02 2.84083989e-02 1.31110355e-01 4.43563759e-01
-5.54001927e-01 6.07197106e-01 -3.45148653e-01 2.50526726e-01
3.52497846e-01 1.56821418e+00 -2.11359754e-01 -3.69172931e-01
-2.20252555e-02 6.68429375e-01 7.45462120e-01 -8.50718021e-02
-7.91606784e-01 -3.84507000e-01 -1.48586193e-02 4.37088966e-01
-1.68332890e-01 2.09646467e-02 -7.36460507e-01 -1.22473550e+00
-5.82097292e-01 -9.24612701e-01 7.79692233e-01 -7.33162344e-01
-1.70212650e+00 7.49396324e-01 -2.82507658e-01 -7.03456581e-01
2.85035037e-02 -5.79527855e-01 -9.79171395e-01 7.60837615e-01
-1.19241405e+00 -1.31506491e+00 -6.30309343e-01 5.47364473e-01
3.27484667e-01 -4.49999303e-01 4.72732782e-01 1.59692883e-01
-1.40967384e-01 5.49451530e-01 1.62109226e-01 2.19974265e-01
8.42565715e-01 -1.34712887e+00 -5.47114713e-03 3.85984212e-01
-1.42071545e-01 5.79128146e-01 1.25272441e+00 -5.46920478e-01
-8.30140471e-01 -7.28061438e-01 2.04475451e+00 -9.90669072e-01
1.24806154e+00 -3.21819752e-01 -7.54310250e-01 3.47114146e-01
6.20367229e-01 -7.09733248e-01 8.91137898e-01 9.29900765e-01
-8.51160944e-01 -8.99454430e-02 -7.57234454e-01 4.38733429e-01
4.84507203e-01 -6.38915658e-01 -8.19931865e-01 4.92887735e-01
1.79891586e-01 -3.71254385e-01 -6.88823283e-01 2.52988666e-01
7.57852137e-01 -1.12090361e+00 6.10143960e-01 -1.10137379e+00
1.36414707e+00 1.44476555e-02 -2.40667075e-01 -1.26399601e+00
-9.54831004e-01 -8.61606598e-01 -1.53863475e-01 8.72206569e-01
5.88259339e-01 -6.87921792e-02 6.74173415e-01 2.96054274e-01
-1.03936426e-01 -4.85285848e-01 -6.73462510e-01 -4.25029814e-01
6.62976146e-01 -4.91433367e-02 3.54344755e-01 1.27302110e+00
9.26900506e-01 8.76649022e-01 -1.18179631e+00 -6.56700134e-01
4.93165016e-01 3.70728672e-01 1.00577092e+00 -1.22042799e+00
-9.48365107e-02 -8.05207014e-01 -3.41101408e-01 -7.27598608e-01
3.58107388e-01 -1.22136199e+00 7.13717118e-02 -1.02145672e+00
1.20447540e+00 1.66842937e-01 -5.56844473e-01 4.30538595e-01
-4.19498652e-01 6.80712461e-01 4.80635941e-01 3.64927083e-01
-9.67934906e-01 5.20299137e-01 8.79290819e-01 -2.17868000e-01
-1.58635639e-02 6.86329650e-03 -7.07871437e-01 7.10935473e-01
8.23683619e-01 -4.80256736e-01 3.43367696e-01 -1.72489546e-02
8.75345528e-01 -2.32485011e-01 5.22943199e-01 -7.49561965e-01
2.16366902e-01 -2.99040854e-01 6.68370277e-02 -8.36176813e-01
8.84156004e-02 -9.44174733e-03 -2.95434147e-01 3.54923964e-01
-5.43273628e-01 8.30605440e-03 -1.76764682e-01 -6.76701739e-02
-2.06078261e-01 -3.72913718e-01 1.02627277e+00 -1.49266377e-01
-3.74357462e-01 -1.24371640e-01 -4.42897469e-01 4.12179530e-01
4.01387423e-01 4.29448813e-01 -8.27808380e-01 -9.46800172e-01
-3.95651221e-01 2.25902155e-01 3.92391533e-01 4.28285003e-01
4.86179262e-01 -1.36291635e+00 -1.48772252e+00 -5.32462299e-01
2.90713340e-01 -9.69342709e-01 1.93863302e-01 1.15549266e+00
-5.00506163e-01 3.52327615e-01 -5.01394570e-01 1.95467565e-02
-1.34396160e+00 3.39823395e-01 2.64043659e-01 -8.13359559e-01
-1.88841686e-01 9.84844387e-01 -1.59796670e-01 -5.13058484e-01
-1.72702089e-01 9.13937807e-01 -5.73987782e-01 3.84909779e-01
8.25861096e-01 1.07594156e+00 1.72109976e-01 -9.91206467e-01
-3.26942682e-01 -1.07330784e-01 -1.54306397e-01 -1.45063564e-01
1.39041162e+00 -3.82062304e-03 -6.27334833e-01 7.04444468e-01
1.16043591e+00 3.13336998e-01 -3.96930128e-01 -2.20635056e-01
3.31457913e-01 -4.79554355e-01 2.19470575e-01 -1.14475012e+00
-4.23261017e-01 9.30172324e-01 -3.19002494e-02 2.97522038e-01
6.07449889e-01 -1.46229997e-01 1.17288470e+00 3.77901107e-01
5.00777364e-01 -1.51489127e+00 4.67421889e-01 1.48979127e+00
1.15852487e+00 -1.16749072e+00 2.24758029e-01 2.77729005e-01
-9.20441031e-01 1.27137864e+00 5.89599848e-01 -9.09805298e-02
4.07687277e-01 -2.78276443e-01 1.25972167e-01 -3.51673931e-01
-8.53895485e-01 -5.68191968e-02 7.38358080e-01 -8.43107104e-02
8.71723652e-01 2.35833555e-01 -1.03499329e+00 9.78730083e-01
-8.60115826e-01 -3.37649047e-01 7.41935670e-01 3.33733916e-01
-1.05275619e+00 -6.06111407e-01 -2.96170264e-01 2.93397486e-01
-6.01903737e-01 -3.18067968e-01 -1.19515228e+00 4.77294952e-01
1.71901137e-01 1.11413431e+00 -5.88412464e-01 -1.26214886e+00
-5.57897892e-03 -2.11793948e-02 1.25412807e-01 -2.24019542e-01
-1.45255220e+00 -3.40475082e-01 7.24959850e-01 -2.80685395e-01
6.84632966e-03 -6.85608864e-01 -8.29379082e-01 -1.10982406e+00
-7.81523213e-02 4.88266945e-01 3.52607191e-01 8.11123669e-01
-2.27691203e-01 -2.89566100e-01 1.11183536e+00 -6.63176060e-01
-5.79248548e-01 -1.33859324e+00 -5.76705933e-01 1.05101776e+00
2.14379162e-01 -3.62553328e-01 -5.26763916e-01 -1.41483635e-01] | [8.841609954833984, 11.05250072479248] |
7a1d2f19-d479-43a3-a81b-bd7cf967049a | a-novel-method-using-machine-learning-to | 2301.12340 | null | https://arxiv.org/abs/2301.12340v1 | https://arxiv.org/pdf/2301.12340v1.pdf | A novel method using machine learning to integrate features from lung and epicardial adipose tissue for detecting the severity of COVID-19 infection | Objectives: To investigate the value of radiomics features of epicardial adipose tissue (EAT) combined with lung for detecting the severity of Coronavirus Disease 2019 (COVID-19) infection. Methods: The retrospective study included data from 515 COVID-19 patients (Cohort1: 415, cohort2: 100) from the two centers between January 2020 and July 2020. A deep learning method was developed to extract the myocardium and visceral pericardium from chest CTs, and then a threshold was applied for automatic EAT extraction. Lung segmentation was achieved according to a published method. Radiomics features of both EAT and lung were extracted for the severity prediction. In a derivation cohort (290, cohort1), univariate analysis and Pearson correlation analysis were used to identify predictors of the severity of COVID-19. A generalized linear regression model for detecting the severity of COVID-19 was built in a derivation cohort and evaluated in internal (125, cohort1) and external (100, cohort2) validation cohorts. Results: For EAT extraction, the Dice similarity coefficients (DSC) of the two centers were 0.972 (0.011) and 0.968 (0.005), respectively. For severity detection, the AUC, net reclassification improvement (NRI), and integrated discrimination improvement (IDI) of the model with radiomics features of both lung and EAT increased by 0.09 (p<0.001), 22.4%, and 17.0%, respectively, compared with the model with lung radiomics features, in the internal validation cohort. The AUC, NRI, and IDI increased by 0.04 (p<0.001), 11.1%, and 8.0%, respectively, in the external validation cohort. Conclusion: Radiomics features of EAT combined with lung have incremental value in detecting the severity of COVID-19. | ['Weihua Zhou', 'Neng Dai', 'Fubao Zhu', 'Chuang Han', 'Yanting Li', 'Alair Augusto Sarmet Moreira Damas dos Santos', 'Wolney de Andrade Martins', 'Claudio Tinoco Mesquita', 'Chen Zhao', 'Daniel Gama das Neves', 'Yanhui Tian', 'Ni Yao'] | 2023-01-29 | null | null | null | null | ['severity-prediction'] | ['computer-vision'] | [-1.82498530e-01 -2.37040445e-01 -2.81625569e-01 1.24308892e-01
-7.38032997e-01 -8.01864922e-01 1.12388112e-01 3.14045936e-01
-3.91700029e-01 6.51941180e-01 3.86286154e-02 -4.83981729e-01
-1.98603630e-01 -6.50316834e-01 -1.96732730e-01 -7.20385671e-01
-5.06581724e-01 7.75181115e-01 1.97737113e-01 5.71960151e-01
-3.92688602e-01 7.18689442e-01 -5.12416005e-01 3.58870387e-01
8.49777281e-01 1.14141023e+00 3.02543938e-01 9.94116902e-01
1.97726458e-01 7.26694405e-01 -2.79659778e-01 -4.96278554e-02
3.55369836e-01 -7.41005421e-01 -3.70231509e-01 -4.36323315e-01
7.24096375e-04 -6.45078421e-01 -1.82635821e-02 3.31960797e-01
7.06002831e-01 -4.61114317e-01 1.13968623e+00 -9.38666403e-01
-3.49179050e-03 2.48147994e-01 -7.82274723e-01 4.25077826e-01
-4.08175558e-01 3.20204884e-01 5.82057476e-01 -7.59910107e-01
5.53201973e-01 4.24984813e-01 1.21502709e+00 3.01555514e-01
-1.03906310e+00 -7.29346216e-01 -4.63720500e-01 -2.80698091e-01
-1.26702499e+00 2.04467356e-01 8.52476507e-02 -9.45053935e-01
5.33064485e-01 4.56933558e-01 1.25541055e+00 4.25760657e-01
5.25668263e-01 3.65773410e-01 1.00788379e+00 -2.25859694e-02
-1.00093685e-01 2.24756628e-01 -7.46032074e-02 9.81804490e-01
6.91553771e-01 1.25010967e-01 4.78805572e-01 -7.05785096e-01
9.79128897e-01 2.09325239e-01 -8.46009180e-02 -2.89237767e-01
-1.27662063e+00 1.02014267e+00 1.61226749e-01 3.43529165e-01
-5.81933439e-01 -2.14131460e-01 8.17238331e-01 -1.62691791e-02
3.13097119e-01 1.08833984e-01 -6.71151161e-01 2.00652018e-01
-9.59563673e-01 -1.74710870e-01 4.16063905e-01 2.84289598e-01
-1.49391994e-01 -1.23600200e-01 -3.76460552e-01 7.79249668e-01
4.16603357e-01 1.17375684e+00 6.45334661e-01 -8.72023106e-01
2.62802213e-01 5.70599735e-01 -3.53080546e-03 -6.73102319e-01
-1.00194120e+00 -6.75776482e-01 -1.30930817e+00 -2.86659539e-01
4.10719454e-01 -6.64809465e-01 -7.37694263e-01 1.61127818e+00
2.20214918e-01 -4.33164574e-02 1.44322187e-01 9.23893452e-01
8.82929504e-01 4.60534871e-01 4.02972430e-01 -5.47500432e-01
1.59654784e+00 -5.08158982e-01 -1.30315587e-01 2.48640776e-01
1.16848612e+00 -5.47012925e-01 6.68327928e-01 1.84394956e-01
-9.42705929e-01 -3.06627005e-01 -5.96440971e-01 7.00983882e-01
8.65609646e-02 6.20654643e-01 2.40577087e-01 8.51310372e-01
-8.45673501e-01 4.41673934e-01 -7.73052871e-01 -5.59015036e-01
5.73007941e-01 2.50728786e-01 -1.96713120e-01 5.33007123e-02
-1.27090740e+00 9.10995007e-01 3.95935960e-02 -1.66549638e-01
-8.26893389e-01 -9.98288929e-01 -2.80617058e-01 -1.06816947e-01
-1.50221124e-01 -1.18020809e+00 6.67403281e-01 -6.68988407e-01
-8.03892374e-01 1.14388645e+00 2.51817077e-01 -3.87217760e-01
4.84919190e-01 1.07010119e-01 -3.09166580e-01 6.80954337e-01
3.10207903e-01 2.05725893e-01 3.62688601e-01 -1.05634582e+00
-6.80654228e-01 -5.62887728e-01 -4.77440506e-01 4.09548804e-02
9.77506191e-02 4.22993809e-01 -2.82301214e-02 -8.47964883e-01
-1.74686655e-01 -1.23097658e+00 -3.29704583e-01 9.62215438e-02
7.95986727e-02 2.95880307e-02 4.74458843e-01 -1.28300965e+00
1.04335749e+00 -2.17208600e+00 -4.55587864e-01 3.92222047e-01
9.13027942e-01 5.79062939e-01 2.22483929e-02 -1.60863996e-01
-4.42234874e-02 5.92803419e-01 -3.30148578e-01 2.16326579e-01
-4.25183237e-01 -4.24275219e-01 2.53226131e-01 7.11419880e-01
-2.74826363e-02 1.07857180e+00 -8.07160258e-01 -8.97439063e-01
2.56079823e-01 3.01413804e-01 -4.92895007e-01 2.81427145e-01
4.38949138e-01 4.50114489e-01 -5.31946480e-01 3.02263200e-01
6.87602818e-01 -4.40772593e-01 6.03196383e-01 -1.83770850e-01
2.45712455e-02 -3.05155754e-01 -4.85389054e-01 1.08421350e+00
-3.03999424e-01 6.00989938e-01 1.39496177e-01 -5.45161486e-01
1.03550303e+00 6.24671519e-01 1.22742403e+00 -4.99396712e-01
3.53509158e-01 3.48081499e-01 4.10094172e-01 -5.03402472e-01
-5.15049636e-01 -8.58020782e-01 -5.48850931e-02 4.80632275e-01
-4.58538890e-01 -7.02901855e-02 -1.06077328e-01 1.88177466e-01
9.18871880e-01 -3.56697738e-01 4.11384463e-01 -2.84763724e-01
6.07882917e-01 2.18106836e-01 6.20773375e-01 7.70215809e-01
-5.62856853e-01 7.12541580e-01 6.68324828e-01 -4.62855130e-01
-8.73638391e-01 -1.39100623e+00 -4.32572097e-01 3.62856865e-01
-2.08590880e-01 -1.76384166e-01 -5.80834150e-01 -1.01767993e+00
9.63039324e-02 3.04624349e-01 -4.17979419e-01 -2.02454492e-01
-6.60084784e-01 -1.03889883e+00 9.99660492e-01 8.57533276e-01
3.69194269e-01 -6.88943505e-01 -9.30350065e-01 1.55201703e-01
-4.52780336e-01 -8.75580013e-01 -2.42636025e-01 -1.52170524e-01
-1.24875128e+00 -1.31095648e+00 -1.17533314e+00 -5.73079407e-01
2.46344522e-01 -3.79072204e-02 9.55974996e-01 4.33464348e-01
-4.75168377e-01 3.30485970e-01 -1.35074019e-01 -3.98637712e-01
-4.80158657e-01 -1.47946626e-01 1.84943080e-01 -2.93704569e-01
-1.58092026e-02 -1.38582677e-01 -9.33869243e-01 4.42388266e-01
-4.41125721e-01 -1.73236117e-01 7.84483254e-01 7.00201035e-01
5.74196577e-01 -3.35062146e-01 7.10718691e-01 -6.32883668e-01
7.04913214e-02 -7.07253933e-01 -4.27951515e-01 1.08351916e-01
-8.52956772e-01 -7.42991865e-01 6.67044163e-01 -1.36560753e-01
-6.11265123e-01 -1.98065028e-01 1.85730070e-01 -6.58236921e-01
5.63755706e-02 4.13519323e-01 3.29149127e-01 4.66099948e-01
3.24553072e-01 -6.88681975e-02 4.03804928e-01 -3.47704530e-01
-1.55749023e-01 5.11393547e-01 1.15119666e-01 -2.81326652e-01
6.85579300e-01 3.32520843e-01 3.72895747e-01 -5.19566357e-01
-4.80746627e-01 -6.53245151e-01 -8.96213114e-01 -1.01261579e-01
1.38893318e+00 -9.08504844e-01 -5.20569444e-01 3.70291144e-01
-8.19526374e-01 -4.88188356e-01 -3.37286919e-01 1.10877061e+00
-5.30214071e-01 5.52782536e-01 -6.45113230e-01 -4.30273473e-01
-9.76292312e-01 -1.02556789e+00 5.33471942e-01 -1.68756694e-01
-5.33258617e-01 -1.08437192e+00 4.03165340e-01 3.77702713e-01
4.93684173e-01 7.19789207e-01 1.26418412e+00 -9.67022896e-01
-3.54656912e-02 -4.14580315e-01 -5.84498823e-01 3.86326551e-01
2.09245041e-01 2.44654454e-02 -3.98308635e-01 -2.29162425e-01
8.49006474e-02 3.74678187e-02 5.20641088e-01 8.49068344e-01
8.88193846e-01 -2.12833330e-01 -3.84995431e-01 6.37546837e-01
1.35451400e+00 5.85847199e-01 3.77579242e-01 8.75974633e-03
4.33023721e-01 3.24038893e-01 5.52340448e-01 5.51579475e-01
1.71077419e-02 5.08183897e-01 1.51431531e-01 -5.45610309e-01
-1.94771484e-01 9.72628370e-02 1.75074250e-01 6.87190354e-01
-4.04977143e-01 -3.54105979e-02 -1.20999110e+00 6.54846370e-01
-1.15759289e+00 -7.55151331e-01 -7.89131522e-01 2.12735844e+00
3.29105198e-01 -8.50124657e-02 5.72784543e-01 -2.81731158e-01
7.89105415e-01 -7.29061291e-02 -2.59763718e-01 -4.95932907e-01
1.02877855e-01 2.64981985e-02 4.67225552e-01 -4.17105816e-02
-1.00429845e+00 1.31422818e-01 6.14750338e+00 3.59504730e-01
-1.05579400e+00 3.34229469e-01 8.97181630e-01 -8.93776938e-02
2.23960698e-01 -1.65753528e-01 -1.68698892e-01 5.48062682e-01
1.17541599e+00 4.80175242e-02 -6.19627088e-02 4.84090477e-01
2.96224535e-01 -7.46233836e-02 -4.37529176e-01 6.43029988e-01
3.80712934e-02 -1.26686692e+00 -3.52958530e-01 1.47721827e-01
5.68415880e-01 3.53905648e-01 -2.35430039e-02 2.52243489e-01
-1.34543777e-01 -7.52063990e-01 1.16710961e-01 6.32600784e-01
1.59240294e+00 -3.99660587e-01 1.44407272e+00 1.80241570e-01
-1.29016149e+00 1.07199430e-01 1.28149763e-01 3.99537474e-01
2.45901957e-01 7.52582669e-01 -1.30984783e+00 3.40380043e-01
6.77810192e-01 3.13733786e-01 -3.20976347e-01 8.28654826e-01
1.97748959e-01 9.53798294e-01 -4.05227005e-01 3.14272791e-02
6.49061352e-02 -1.83998227e-01 5.66701353e-01 1.31838298e+00
1.38649046e-01 4.56109285e-01 -5.47891110e-02 6.99888527e-01
1.07020989e-01 5.12620628e-01 -4.08031911e-01 -1.09830923e-01
1.55148521e-01 1.40881062e+00 -1.05528712e+00 -7.29856730e-01
-4.96505052e-01 2.94257760e-01 -4.21315521e-01 1.49060308e-03
-1.27054465e+00 -2.63209075e-01 2.73251265e-01 7.46313632e-01
2.76280865e-02 2.17264250e-01 -6.42318547e-01 -8.04455936e-01
-4.66130942e-01 -6.30247593e-01 9.16998625e-01 -7.18639195e-01
-1.09127069e+00 3.68701786e-01 -2.55400203e-02 -1.34695280e+00
-5.84367895e-03 -4.54544216e-01 -6.27117395e-01 1.12284648e+00
-1.00360632e+00 -9.32417274e-01 -2.85597950e-01 4.73040231e-02
1.02852568e-01 -2.22749412e-01 9.56569195e-01 1.71996072e-01
-3.40830237e-01 5.67076802e-01 3.35404515e-01 4.97223884e-01
5.68066835e-01 -8.79617751e-01 -3.15691650e-01 2.25075245e-01
-6.37753427e-01 6.75520718e-01 -2.76492596e-01 -9.50022459e-01
-7.28899181e-01 -1.44339156e+00 9.68092740e-01 -5.55622578e-01
3.20092738e-01 2.38865301e-01 -5.33534825e-01 5.29143035e-01
-3.25193614e-01 -1.17529286e-02 1.14361179e+00 -4.68539953e-01
-1.21637277e-01 1.20313140e-02 -1.67178297e+00 2.43085697e-01
5.03928304e-01 -2.18261987e-01 -5.31965792e-01 1.85838059e-01
4.25340652e-01 -1.21638373e-01 -1.68987060e+00 9.80987906e-01
1.12969589e+00 -8.23747277e-01 1.20357132e+00 -6.99227631e-01
9.85183939e-02 -1.68533370e-01 -2.07178704e-02 -5.59166253e-01
-7.38554358e-01 -7.76917338e-02 2.52930045e-01 7.53197610e-01
4.47334170e-01 -7.62569189e-01 5.36024451e-01 1.73846468e-01
-1.37743251e-02 -1.09384620e+00 -7.51467168e-01 -5.77474177e-01
3.34295273e-01 -1.61676690e-01 3.57785940e-01 8.31638515e-01
-3.37565213e-01 5.37478104e-02 1.14639953e-01 9.76672694e-02
3.28216463e-01 2.18245342e-01 2.77942985e-01 -1.46332037e+00
-2.02484980e-01 -4.48670000e-01 -1.32690877e-01 -6.29896820e-02
-2.31412962e-01 -1.24936295e+00 -6.00936890e-01 -2.04511690e+00
9.11780953e-01 -8.50427628e-01 -7.05313444e-01 3.77411038e-01
-2.14242250e-01 4.12978739e-01 4.82166231e-01 5.79840779e-01
5.93677647e-02 5.58274277e-02 1.10702038e+00 1.22700676e-01
-3.04278672e-01 2.74004787e-01 -3.92119199e-01 8.58931780e-01
1.27131879e+00 -7.93006957e-01 -3.14394593e-01 2.11707547e-01
-1.46261930e-01 3.90102178e-01 7.05346584e-01 -7.61161149e-01
-6.46934211e-01 -2.66155720e-01 9.16602671e-01 -8.69152308e-01
9.45386849e-03 -6.92578673e-01 3.03834587e-01 1.28906536e+00
7.05613717e-02 2.30726093e-01 4.14491653e-01 2.16614351e-01
4.97027874e-01 -9.06183943e-02 1.02558756e+00 -1.02217555e-01
2.31954399e-02 2.61160374e-01 -9.41908181e-01 4.99832392e-01
1.04566789e+00 -1.32134497e-01 -1.01973414e-01 -8.80624950e-02
-9.18557167e-01 -1.70226797e-01 3.22410315e-01 -5.69376647e-02
4.49151248e-01 -1.06655455e+00 -9.12775218e-01 -3.99732962e-03
-1.86833635e-01 -2.15321705e-01 3.41497540e-01 1.92425525e+00
-8.04422915e-01 4.95781749e-01 -6.89192638e-02 -7.53498018e-01
-1.37204087e+00 5.53239346e-01 6.42919481e-01 -6.34936750e-01
-5.65474331e-01 4.86981809e-01 6.50023937e-01 -2.56020576e-01
-4.18398231e-01 -3.31755102e-01 -1.32454105e-03 1.13350049e-01
7.71027654e-02 6.77166224e-01 -1.30874321e-01 -6.51331007e-01
-7.31111765e-01 1.00273407e+00 3.44898671e-01 1.93893015e-01
1.04245317e+00 -1.04881018e-01 -2.79626042e-01 3.30298036e-01
1.36353350e+00 1.93869889e-01 -4.59035993e-01 1.58076704e-01
-3.72843921e-01 2.16740780e-02 -2.00075731e-01 -1.12245619e+00
-1.13361192e+00 7.68807352e-01 1.28631461e+00 -3.46468240e-02
1.14516151e+00 2.26363525e-01 8.66274714e-01 -4.16618228e-01
-2.17156690e-02 -6.32289350e-01 -2.05349728e-01 1.86646029e-01
6.57669842e-01 -7.59232581e-01 -3.19794454e-02 -2.02235818e-01
-9.18393552e-01 9.25448298e-01 1.97192743e-01 3.65031622e-02
7.60061264e-01 2.51835436e-01 2.42175728e-01 -2.72099644e-01
-3.88007075e-01 2.34488487e-01 2.25959122e-01 5.93315303e-01
5.30679762e-01 5.25293469e-01 -6.65215433e-01 1.13200343e+00
2.22295791e-01 3.58341224e-02 1.13116980e-01 5.16020358e-01
-3.83696020e-01 -4.47683752e-01 -4.45453942e-01 1.03837097e+00
-6.28205597e-01 -8.38078409e-02 -3.55518430e-01 1.04817224e+00
5.33245921e-01 6.79449618e-01 -3.98502871e-02 -8.69717225e-02
1.35433599e-01 3.22280794e-01 9.34742391e-02 -3.47755313e-01
-7.57126629e-01 2.68828660e-01 6.24600314e-02 -1.96941629e-01
-3.51759255e-01 -7.70918906e-01 -1.70403838e+00 7.82015473e-02
-3.32588166e-01 3.45048577e-01 4.10819113e-01 5.44615030e-01
3.97705704e-01 4.30938721e-01 7.93951750e-01 3.32054235e-02
-4.16391343e-01 -8.92823637e-01 -8.87271643e-01 1.39422510e-02
2.56437153e-01 -5.21246493e-01 -6.29977942e-01 -7.14458292e-03] | [15.457290649414062, -1.844844102859497] |
48df79b5-bbd6-454b-beb7-f4b2b04d13fc | if-net-an-illumination-invariant-feature | 2008.03897 | null | https://arxiv.org/abs/2008.03897v1 | https://arxiv.org/pdf/2008.03897v1.pdf | IF-Net: An Illumination-invariant Feature Network | Feature descriptor matching is a critical step is many computer vision applications such as image stitching, image retrieval and visual localization. However, it is often affected by many practical factors which will degrade its performance. Among these factors, illumination variations are the most influential one, and especially no previous descriptor learning works focus on dealing with this problem. In this paper, we propose IF-Net, aimed to generate a robust and generic descriptor under crucial illumination changes conditions. We find out not only the kind of training data important but also the order it is presented. To this end, we investigate several dataset scheduling methods and propose a separation training scheme to improve the matching accuracy. Further, we propose a ROI loss and hard-positive mining strategy along with the training scheme, which can strengthen the ability of generated descriptor dealing with large illumination change conditions. We evaluate our approach on public patch matching benchmark and achieve the best results compared with several state-of-the-arts methods. To show the practicality, we further evaluate IF-Net on the task of visual localization under large illumination changes scenes, and achieves the best localization accuracy. | ['Kuan-Wen Chen', 'Zu-Kuan Huang', 'Zhao-Xu Luo', 'Po-Heng Chen', 'Chun Yang'] | 2020-08-10 | null | null | null | null | ['image-stitching', 'patch-matching'] | ['computer-vision', 'computer-vision'] | [ 1.87551886e-01 -8.69587958e-01 -3.43969584e-01 -1.89535379e-01
-5.20288825e-01 -3.60214978e-01 5.25044143e-01 4.80374880e-02
-3.17725092e-01 3.78327698e-01 -7.80602843e-02 7.59134665e-02
-3.74160528e-01 -6.70103133e-01 -5.95007718e-01 -8.54780316e-01
2.15412840e-01 1.17826588e-01 4.16396677e-01 -1.91425487e-01
4.71275985e-01 6.83263958e-01 -1.82689154e+00 1.18604600e-02
8.56819093e-01 1.28905141e+00 2.32518733e-01 -9.63818282e-02
-2.99615599e-02 3.85479152e-01 -5.03394306e-01 -2.42369652e-01
4.34806168e-01 -1.73816711e-01 -2.98284769e-01 2.25055978e-01
4.41996187e-01 -1.87170655e-01 -2.29418814e-01 1.03497779e+00
7.99279213e-01 6.47575855e-02 7.13613808e-01 -1.40199006e+00
-3.48517627e-01 2.29243860e-01 -8.02200198e-01 1.79775327e-01
2.87394762e-01 2.36943409e-01 8.09123099e-01 -9.58418965e-01
5.98520577e-01 1.08996844e+00 6.51468515e-01 3.00878525e-01
-9.60654318e-01 -9.39384341e-01 3.90485115e-02 4.28283274e-01
-1.42604685e+00 -5.76403797e-01 1.16002214e+00 -2.61486024e-01
4.27078664e-01 3.93157527e-02 2.85792917e-01 9.71518576e-01
4.68306579e-02 8.66666198e-01 1.22903264e+00 -4.53463674e-01
3.30861211e-02 9.44671184e-02 5.02373204e-02 4.22296315e-01
1.03277251e-01 1.24932446e-01 -3.43078941e-01 -1.07004233e-01
5.68948150e-01 3.49376887e-01 -3.58822465e-01 -5.22250831e-01
-1.24439812e+00 5.43427765e-01 5.89552701e-01 4.18309391e-01
-2.26695687e-01 3.98917720e-02 4.51883286e-01 4.43509191e-01
2.09852636e-01 2.66961843e-01 -2.18268707e-01 6.77826926e-02
-8.55466008e-01 1.18225642e-01 3.86544108e-01 7.24779069e-01
7.75207222e-01 -3.58985841e-01 -3.24317008e-01 1.09052706e+00
1.58052832e-01 5.85595965e-01 4.32249039e-01 -3.77885610e-01
3.93397421e-01 7.13854432e-01 -3.72550189e-02 -1.51575005e+00
-1.90962866e-01 -6.27858102e-01 -1.01425529e+00 -1.06897205e-01
1.99458212e-01 2.42747933e-01 -7.01864362e-01 1.49323618e+00
3.87083679e-01 2.80695468e-01 -2.20636562e-01 1.07066441e+00
9.37878907e-01 5.12473285e-01 -1.70495436e-01 -2.29381144e-01
1.43046796e+00 -9.17983174e-01 -6.36511564e-01 -4.88088429e-02
4.39962327e-01 -1.19224977e+00 7.65731633e-01 4.24656689e-01
-5.74657261e-01 -8.24641645e-01 -9.41269100e-01 8.66618529e-02
-2.99347073e-01 5.02987742e-01 5.94232976e-01 4.48239326e-01
-6.32687151e-01 5.49914539e-01 -4.02789176e-01 -5.18645287e-01
3.61012846e-01 3.24687690e-01 -5.94367683e-01 -3.67378622e-01
-1.10641456e+00 7.76401043e-01 2.73062497e-01 1.34222612e-01
-5.26003718e-01 -4.78861511e-01 -5.89367807e-01 -1.03606239e-01
4.05627578e-01 -3.81719440e-01 6.75876200e-01 -8.48387539e-01
-1.04655802e+00 7.36018896e-01 -7.88481310e-02 -1.62239283e-01
6.17657900e-01 2.82259937e-02 -5.18410504e-01 -1.14207901e-03
1.77757561e-01 5.69045246e-01 1.06700397e+00 -1.14899182e+00
-6.35956645e-01 -4.20695722e-01 -1.11095630e-01 1.04717173e-01
-5.79580784e-01 1.39121011e-01 -8.32568467e-01 -6.93298399e-01
1.33674711e-01 -9.13930655e-01 -6.52943924e-02 1.74226239e-01
-6.14258870e-02 -4.63235080e-01 8.66935015e-01 -4.43615049e-01
1.26654422e+00 -2.54737663e+00 -1.07187070e-01 4.54457670e-01
-8.11154619e-02 3.77167493e-01 -1.16816685e-01 4.32380676e-01
1.17321732e-02 -3.01005185e-01 -4.24392382e-03 -3.47409189e-01
6.65544719e-02 -9.59955994e-03 -3.10786992e-01 5.84807038e-01
2.16940627e-01 5.59887886e-01 -6.01019144e-01 -7.11086452e-01
2.58302152e-01 4.60306048e-01 -3.68345499e-01 2.37759650e-01
2.53377706e-02 4.71318424e-01 -6.23174906e-01 9.53634560e-01
1.03792667e+00 -1.22548126e-01 -2.23477751e-01 -6.35005951e-01
-1.32965267e-01 -3.69175076e-01 -1.48884237e+00 1.77759457e+00
-4.20318753e-01 3.23844135e-01 -1.16660520e-01 -9.86959577e-01
1.29000068e+00 -2.30333637e-02 5.30869544e-01 -1.03932989e+00
3.29623252e-01 4.53021586e-01 6.88275229e-03 -5.80448449e-01
3.05902243e-01 1.49453461e-01 7.44364113e-02 1.81319013e-01
-2.28516713e-01 1.92251995e-01 1.73422098e-01 -1.27060831e-01
9.63285983e-01 4.51733805e-02 2.22803012e-01 -1.33307770e-01
8.34933400e-01 -1.95722044e-01 6.32696986e-01 4.99724865e-01
-2.65734136e-01 6.85020506e-01 3.37729007e-01 -5.61569154e-01
-8.92926037e-01 -4.78899300e-01 -4.95859921e-01 8.04149508e-01
6.62015378e-01 -2.21452296e-01 -4.61315304e-01 -6.22919202e-01
2.28742540e-01 1.12671033e-01 -4.98864591e-01 -2.22746953e-01
-4.30385888e-01 -7.08658576e-01 4.79566991e-01 2.76263893e-01
8.05340052e-01 -1.13173866e+00 -2.52618581e-01 5.95068038e-02
-2.10598692e-01 -1.14854896e+00 -6.27684057e-01 -1.06179260e-01
-4.86917824e-01 -1.15416694e+00 -8.05983424e-01 -1.05967295e+00
7.84525275e-01 6.15146101e-01 7.31441319e-01 3.31932098e-01
-5.04034638e-01 -7.27549046e-02 -4.39442277e-01 -1.77436322e-01
-7.42486417e-02 2.58067548e-01 8.81294161e-03 4.48556185e-01
3.40407640e-01 -5.96980274e-01 -8.56156826e-01 7.87423551e-01
-1.10654271e+00 -2.30340749e-01 8.93693805e-01 9.20923948e-01
5.42610288e-01 1.09513842e-01 3.64537448e-01 -5.63248515e-01
5.19241035e-01 -2.40659624e-01 -7.10404575e-01 3.79905432e-01
-5.77953398e-01 1.66264370e-01 6.55518949e-01 -6.10009611e-01
-6.72936320e-01 2.18816072e-01 -1.76134095e-01 -5.11579990e-01
-2.16594964e-01 2.27281913e-01 -2.94379681e-01 -6.02668166e-01
5.93543053e-01 3.84895831e-01 8.18082988e-02 -4.98650968e-01
-7.24333003e-02 7.78247893e-01 2.80429959e-01 -5.83916068e-01
1.03032911e+00 3.83414567e-01 2.98830658e-01 -5.97650826e-01
-5.47050118e-01 -7.98254371e-01 -4.17892456e-01 -1.22413278e-01
4.23675120e-01 -7.46022522e-01 -7.61852741e-01 5.72835267e-01
-1.16669512e+00 3.44788015e-01 2.46669263e-01 3.84127200e-01
-2.28859052e-01 5.36310792e-01 -2.06946582e-01 -5.59689879e-01
-3.96400273e-01 -1.40317333e+00 1.10747099e+00 3.22682023e-01
3.78200620e-01 -6.92996979e-01 1.64560378e-01 2.21810356e-01
6.63174629e-01 3.52193683e-01 5.75693905e-01 -5.40106893e-01
-6.28754973e-01 -2.36385062e-01 -5.52841127e-01 3.11013848e-01
1.27664551e-01 -1.11508876e-01 -9.09330010e-01 -4.12174493e-01
-2.46264547e-01 -2.10379973e-01 1.03008842e+00 3.91237997e-02
1.21712732e+00 -2.13603452e-02 -5.93728960e-01 8.57286215e-01
1.44602895e+00 9.47666615e-02 8.26776922e-01 5.46791315e-01
5.20309985e-01 6.11824870e-01 1.08717036e+00 4.58475679e-01
2.05133960e-01 1.06462216e+00 5.07569790e-01 -4.12826806e-01
-1.92957804e-01 -1.94137231e-01 1.84695244e-01 6.24283433e-01
1.19177721e-01 -3.33022662e-02 -6.18269861e-01 5.31803787e-01
-1.88807535e+00 -9.46572959e-01 8.55259690e-03 2.28792167e+00
6.99476004e-01 -3.73143852e-02 -4.19793501e-02 2.95495540e-01
8.17720652e-01 2.85339206e-01 -3.73619914e-01 1.78865045e-01
-2.43521184e-01 2.99659431e-01 4.90806073e-01 1.31188966e-02
-1.29132569e+00 7.84679353e-01 4.68484259e+00 1.28456640e+00
-1.43464768e+00 -6.84254942e-03 3.26989621e-01 4.35264885e-01
-1.07303560e-01 1.18566908e-01 -8.56706977e-01 6.75591886e-01
6.73771426e-02 1.45163208e-01 2.82714069e-01 8.39402199e-01
-5.42316064e-02 2.66970545e-02 -9.00629163e-01 1.36704469e+00
3.19076449e-01 -8.85258496e-01 3.19734477e-02 1.44282430e-02
5.89135408e-01 -2.88423747e-01 1.06162354e-01 1.46454364e-01
-4.66091245e-01 -7.94963300e-01 3.62545043e-01 4.40873981e-01
6.59132302e-01 -7.71861851e-01 1.06979537e+00 3.70792061e-01
-1.31916034e+00 -4.49956208e-02 -6.24605298e-01 3.10003191e-01
-6.96600899e-02 7.91434467e-01 -5.22556067e-01 7.98794627e-01
6.14120185e-01 9.12402630e-01 -8.32079530e-01 1.59808433e+00
-1.37901723e-01 2.96938896e-01 -3.51299822e-01 -8.34490359e-02
1.56321395e-02 -6.36808500e-02 4.19375360e-01 1.06687152e+00
4.77266163e-01 -3.16758938e-02 3.44945937e-01 5.92527449e-01
-1.73061728e-01 5.28843939e-01 -7.20656455e-01 3.73065352e-01
5.01014948e-01 1.42920053e+00 -6.58305228e-01 -5.97048365e-02
-1.54200777e-01 9.29308951e-01 1.07604407e-01 1.00741051e-01
-9.32255268e-01 -6.06173217e-01 4.67747808e-01 1.64488390e-01
4.19082552e-01 -8.16520602e-02 2.54952788e-01 -1.18376243e+00
3.81664842e-01 -9.56003666e-01 2.13001549e-01 -4.39318508e-01
-1.32387483e+00 5.27894914e-01 -2.40699887e-01 -1.67982066e+00
1.84279427e-01 -5.84281802e-01 -6.17396653e-01 5.04249692e-01
-2.03125238e+00 -1.37394750e+00 -7.18409300e-01 7.82930911e-01
3.17462981e-01 -2.34620467e-01 4.72163469e-01 9.36899066e-01
-6.11824572e-01 9.54189241e-01 1.41105175e-01 1.20249398e-01
1.26935768e+00 -5.86043715e-01 1.37070045e-01 8.03602040e-01
8.18647072e-02 5.93252242e-01 5.21613955e-01 -3.82182866e-01
-1.61039698e+00 -1.02924347e+00 6.49646938e-01 -4.08009849e-02
4.53364491e-01 -2.78149039e-01 -8.29980254e-01 3.51865031e-02
-9.49808285e-02 4.36341047e-01 3.16810161e-01 -6.84302524e-02
-4.72446978e-01 -6.74201131e-01 -1.15377116e+00 2.75003314e-01
1.13102198e+00 -3.19860429e-01 -3.07660192e-01 2.95052826e-01
3.53762329e-01 -3.93733054e-01 -7.98161805e-01 6.72031522e-01
6.42896771e-01 -1.03259456e+00 9.57024574e-01 -8.44789967e-02
2.23844260e-01 -6.77987278e-01 -4.35779355e-02 -1.09909999e+00
-2.82082409e-01 -4.57283527e-01 3.12849432e-01 1.63401675e+00
4.29137386e-02 -8.22429061e-01 5.80488920e-01 7.58557245e-02
1.16785936e-01 -6.41943634e-01 -7.90542483e-01 -8.38952243e-01
-4.65064585e-01 -1.23279072e-01 7.51215577e-01 1.03926110e+00
-4.35362875e-01 5.85630052e-02 -5.56548119e-01 1.29243344e-01
7.40298212e-01 3.49610955e-01 1.15905464e+00 -1.17871487e+00
-7.55690485e-02 -3.79521966e-01 -8.04547012e-01 -9.88226473e-01
6.14947714e-02 -6.37231052e-01 -1.88431132e-03 -1.23162317e+00
3.26106071e-01 -8.10711980e-01 -4.50763732e-01 4.34227526e-01
-1.70253664e-01 3.47577900e-01 1.51417062e-01 5.20265639e-01
-7.77377009e-01 6.38842762e-01 1.19564462e+00 -3.16964775e-01
-4.70024981e-02 8.81882384e-03 -4.27941531e-01 2.52942592e-01
5.79419494e-01 -4.23408598e-01 -2.25143820e-01 -2.93507010e-01
-3.04426532e-02 -1.89100206e-01 4.71497655e-01 -1.10029447e+00
5.59340656e-01 -1.52661070e-01 3.69787097e-01 -5.51037490e-01
1.45723045e-01 -1.09843349e+00 2.72555351e-01 4.60081339e-01
4.47151484e-03 1.37306541e-01 8.13627467e-02 3.35784972e-01
-6.07162356e-01 -1.54611945e-01 8.57923806e-01 1.61098376e-01
-9.46204960e-01 5.16179562e-01 5.32491922e-01 -1.37436807e-01
1.15659249e+00 -2.94022977e-01 -4.24047410e-01 -1.18286215e-01
-8.30831602e-02 2.41454303e-01 6.40080273e-01 5.94919682e-01
6.06694281e-01 -1.50662446e+00 -6.45974934e-01 3.68266046e-01
6.12034023e-01 -3.11343729e-01 2.11504072e-01 1.08915925e+00
-4.90928173e-01 1.06346294e-01 -3.28949928e-01 -7.30158985e-01
-1.33990204e+00 7.40684211e-01 1.73228800e-01 -1.68136626e-01
-3.02822739e-01 3.98586363e-01 5.53589053e-02 -2.32173905e-01
4.76977825e-01 -1.25505716e-01 -4.49245751e-01 1.56980529e-01
4.93938237e-01 1.92376852e-01 2.36791089e-01 -6.08586431e-01
-5.55257320e-01 1.24399626e+00 -1.47894353e-01 4.19380039e-01
1.15591407e+00 5.82729094e-02 -2.52379954e-01 -1.20834962e-01
1.38387370e+00 1.15029342e-01 -9.34443772e-01 -4.65145171e-01
-1.57769531e-01 -8.02732587e-01 -1.43844143e-01 -4.90236014e-01
-1.20489621e+00 9.65214074e-01 1.01825070e+00 1.49907982e-02
1.35164046e+00 -2.10104540e-01 9.38306093e-01 2.76363105e-01
5.52641869e-01 -1.00711727e+00 1.00095876e-01 1.96574345e-01
7.64892757e-01 -1.52402711e+00 1.23977646e-01 -4.29156959e-01
-3.30654353e-01 1.10458887e+00 5.57168663e-01 -1.69632569e-01
6.70318723e-01 6.36044592e-02 -7.11432695e-02 -1.16902791e-01
-3.51033628e-01 -4.16160762e-01 4.97979581e-01 4.63795096e-01
2.07324356e-01 -2.49743819e-01 -5.01165509e-01 2.34751105e-01
1.90597028e-01 -2.32531782e-02 -2.05778092e-01 6.73820674e-01
-4.07762796e-01 -1.35730171e+00 -5.71841478e-01 2.12959364e-01
-3.76897722e-01 5.25509529e-02 -2.85612643e-01 7.71096587e-01
4.59252179e-01 8.09445918e-01 -2.44928867e-01 -4.62657213e-01
5.32664180e-01 -4.64227974e-01 3.75501722e-01 -9.44556370e-02
-6.78269386e-01 -2.17577200e-02 -2.71582782e-01 -4.94360656e-01
-5.62591195e-01 -5.28531671e-01 -8.37474883e-01 -3.87580842e-01
-5.84506571e-01 1.41211689e-01 7.12740958e-01 7.65262663e-01
4.15555388e-01 1.37190089e-01 1.14657104e+00 -6.99118376e-01
-6.82900250e-01 -7.90505230e-01 -4.25169945e-01 9.02957499e-01
2.97412187e-01 -8.88208568e-01 -5.03981590e-01 -2.12454110e-01] | [10.821693420410156, 0.47993242740631104] |
4b4f1e70-2007-453d-87ec-5a3f05b6b195 | i-see-dead-people-gray-box-adversarial-attack | 2306.07591 | null | https://arxiv.org/abs/2306.07591v1 | https://arxiv.org/pdf/2306.07591v1.pdf | I See Dead People: Gray-Box Adversarial Attack on Image-To-Text Models | Modern image-to-text systems typically adopt the encoder-decoder framework, which comprises two main components: an image encoder, responsible for extracting image features, and a transformer-based decoder, used for generating captions. Taking inspiration from the analysis of neural networks' robustness against adversarial perturbations, we propose a novel gray-box algorithm for creating adversarial examples in image-to-text models. Unlike image classification tasks that have a finite set of class labels, finding visually similar adversarial examples in an image-to-text task poses greater challenges because the captioning system allows for a virtually infinite space of possible captions. In this paper, we present a gray-box adversarial attack on image-to-text, both untargeted and targeted. We formulate the process of discovering adversarial perturbations as an optimization problem that uses only the image-encoder component, meaning the proposed attack is language-model agnostic. Through experiments conducted on the ViT-GPT2 model, which is the most-used image-to-text model in Hugging Face, and the Flickr30k dataset, we demonstrate that our proposed attack successfully generates visually similar adversarial examples, both with untargeted and targeted captions. Notably, our attack operates in a gray-box manner, requiring no knowledge about the decoder module. We also show that our attacks fool the popular open-source platform Hugging Face. | ['Moshe Sipper', 'Raz Lapid'] | 2023-06-13 | null | null | null | null | ['adversarial-attack'] | ['adversarial'] | [ 7.08676875e-01 3.86157244e-01 3.92408103e-01 -1.64025620e-01
-1.00821185e+00 -1.08448267e+00 7.82912970e-01 -5.41200399e-01
-3.29478830e-02 4.42390591e-01 -9.57389101e-02 -5.74172616e-01
5.09295166e-01 -7.08118141e-01 -1.48823822e+00 -5.83312750e-01
2.32355997e-01 2.35885620e-01 -5.89500926e-02 -2.92927742e-01
-8.53253379e-02 4.03319299e-01 -1.16525841e+00 4.95952278e-01
4.36331302e-01 9.81053472e-01 -2.66944408e-01 1.10029709e+00
1.70136452e-01 1.11744416e+00 -8.49591136e-01 -1.12619162e+00
6.76401436e-01 -5.41778266e-01 -7.04464078e-01 2.72662938e-01
8.35762262e-01 -4.20819610e-01 -9.76594925e-01 1.21969581e+00
6.08608603e-01 -2.32731238e-01 7.55449414e-01 -1.78698969e+00
-1.33383787e+00 4.60081428e-01 -4.47804868e-01 -2.16889009e-01
4.50980037e-01 6.85728371e-01 6.26134396e-01 -7.32410431e-01
4.94107991e-01 1.51142931e+00 3.85618627e-01 1.05268097e+00
-1.15855157e+00 -8.92128646e-01 -5.63150980e-02 1.90208808e-01
-1.38535964e+00 -5.96672714e-01 7.08335876e-01 -4.93173659e-01
6.95508838e-01 4.39961642e-01 1.32165164e-01 1.54945135e+00
2.53264040e-01 6.58726454e-01 9.47326005e-01 -4.28230077e-01
6.10830523e-02 3.91132832e-01 -6.56346202e-01 5.46389699e-01
-8.43814760e-02 4.38087016e-01 -1.68102324e-01 -2.08181068e-01
5.95412910e-01 -3.23575169e-01 -3.43420863e-01 -2.48497546e-01
-8.53293657e-01 8.20988536e-01 4.07891542e-01 -1.27343968e-01
-3.18866931e-02 5.33588469e-01 3.58288944e-01 4.21878934e-01
1.19222894e-01 3.47088367e-01 -5.98326065e-02 2.07213223e-01
-6.86466932e-01 1.27276868e-01 6.45214260e-01 1.11149871e+00
5.68897009e-01 2.92147458e-01 -4.07178074e-01 4.73302513e-01
2.65915781e-01 9.54556167e-01 4.47024405e-01 -5.90756238e-01
6.59535646e-01 1.03616185e-01 -5.63222766e-02 -1.20392895e+00
3.30046266e-01 -1.13692638e-02 -9.24563706e-01 4.69051957e-01
3.36606979e-01 -3.47542286e-01 -1.11951399e+00 1.90168571e+00
2.02038556e-01 2.97946453e-01 4.56175476e-01 6.90970719e-01
7.08485484e-01 7.85011590e-01 4.22253273e-02 1.68778196e-01
1.37118196e+00 -1.00959241e+00 -5.44463158e-01 -5.79359710e-01
3.13955039e-01 -8.84408891e-01 1.12775552e+00 1.20311327e-01
-1.19348657e+00 -4.71631259e-01 -1.03324306e+00 -7.69437328e-02
-5.38688064e-01 -1.10639937e-01 1.74420867e-02 9.04021800e-01
-1.11926603e+00 2.87643820e-01 -1.65621623e-01 -1.40048072e-01
6.04514718e-01 2.47946799e-01 -5.63275337e-01 -2.73008853e-01
-1.25588822e+00 9.18610871e-01 3.59210610e-01 -2.99617052e-02
-1.45664489e+00 -5.79166114e-01 -1.03847337e+00 3.75315100e-02
3.37498307e-01 -7.08907783e-01 1.26198709e+00 -1.46082282e+00
-1.52076542e+00 9.67268705e-01 8.69297311e-02 -6.15778267e-01
9.41948593e-01 1.09912008e-01 -5.67057848e-01 3.69736493e-01
-1.77592039e-01 8.93842399e-01 1.63002908e+00 -1.52931070e+00
-2.32121646e-01 -5.42240553e-02 3.76277328e-01 8.88983831e-02
-2.29958400e-01 1.34103760e-01 -6.57205045e-01 -9.43849325e-01
-6.57038093e-01 -1.07287431e+00 -1.10065013e-01 2.60428488e-01
-8.17700684e-01 5.12665808e-01 1.15499890e+00 -7.95694649e-01
7.71598577e-01 -2.32083797e+00 -4.02783304e-02 1.62353799e-01
1.00084037e-01 3.96412969e-01 -4.91214126e-01 4.59657252e-01
-6.00852489e-01 4.10245121e-01 -4.29860950e-01 -3.02807689e-01
1.38681114e-01 1.57916814e-01 -8.26401472e-01 3.95329595e-01
5.10500968e-01 1.29945016e+00 -7.60499954e-01 -4.64336276e-01
2.91160971e-01 5.74121773e-01 -4.18668061e-01 4.34427857e-01
-3.46326709e-01 3.29720438e-01 -1.93743467e-01 5.22460222e-01
8.69881094e-01 -4.56702299e-02 -1.27714694e-01 -1.51081860e-01
4.92110789e-01 -5.53340375e-01 -7.38212824e-01 1.27476287e+00
-5.26412666e-01 9.37106669e-01 -1.28273562e-01 -6.96394205e-01
6.24567866e-01 4.64987963e-01 -8.16051513e-02 -5.80515862e-01
2.21636951e-01 -4.88970801e-02 -1.08727656e-01 -5.14158964e-01
2.55392313e-01 -9.47862044e-02 -2.92701364e-01 3.59700084e-01
-4.92933728e-02 -3.21360588e-01 1.05428940e-03 5.99094987e-01
1.28894305e+00 -1.06593713e-01 2.26298720e-02 3.34430486e-01
6.58636153e-01 -2.04347834e-01 -4.67284657e-02 9.17397916e-01
-1.10957056e-01 9.43041086e-01 6.99075460e-01 -1.66791245e-01
-1.25961196e+00 -1.13711190e+00 1.66077614e-01 6.87014282e-01
1.05854936e-01 -2.14601055e-01 -1.05818415e+00 -1.05099213e+00
-5.17502986e-02 7.52466321e-01 -9.02356863e-01 -4.57423091e-01
-2.83091426e-01 -2.46502608e-01 1.09288168e+00 9.16612968e-02
5.10133982e-01 -1.20788884e+00 -1.28324077e-01 -9.94893312e-02
-8.27459916e-02 -1.47363138e+00 -8.29570353e-01 -3.04425001e-01
-6.07410595e-02 -1.16497779e+00 -8.50962222e-01 -8.01812708e-01
9.65159118e-01 6.35214299e-02 1.06065261e+00 6.51877597e-02
-4.45508629e-01 8.14569354e-01 -3.25781733e-01 -5.51037014e-01
-1.15238154e+00 -5.15411615e-01 -2.27211609e-01 3.35722476e-01
-9.57683697e-02 -4.29933280e-01 -5.20715356e-01 3.43412995e-01
-1.41641772e+00 1.23601571e-01 7.21241653e-01 6.96047068e-01
3.57684910e-01 1.46915168e-01 4.34431374e-01 -7.94277847e-01
4.78661120e-01 -5.10593414e-01 -6.67755246e-01 3.74169648e-01
-4.28712964e-02 -4.49906923e-02 1.15350699e+00 -7.87282944e-01
-7.42938697e-01 3.18656504e-01 -1.47444025e-01 -1.04119551e+00
-1.42559186e-01 -2.73521505e-02 -6.85556948e-01 -5.61154127e-01
8.54984045e-01 5.98182619e-01 1.12254755e-03 5.31340204e-02
8.20997536e-01 6.52661204e-01 9.56420660e-01 -4.72230911e-01
1.71956849e+00 4.58585411e-01 -7.60484636e-02 -6.51828766e-01
-3.84864181e-01 1.54654160e-01 -2.12011963e-01 -3.45588148e-01
8.59317958e-01 -8.43349397e-01 -7.99661875e-01 6.88098073e-01
-1.34745836e+00 -2.43464813e-01 -2.99261898e-01 -2.32702315e-01
-8.01573873e-01 5.37204862e-01 -4.05899167e-01 -5.24416506e-01
-4.21011358e-01 -1.43168485e+00 1.17884684e+00 2.74261963e-02
2.06721917e-01 -8.37789595e-01 -2.16778815e-01 3.80976588e-01
3.88931900e-01 6.15164578e-01 8.06384087e-01 -7.24473238e-01
-7.05724001e-01 -4.72198606e-01 -3.10363084e-01 6.92844927e-01
3.20438296e-02 -3.98788927e-03 -1.14911020e+00 -4.18985397e-01
-1.03425801e-01 -6.24754369e-01 4.55222011e-01 -5.41089624e-02
1.49785411e+00 -8.83465827e-01 -1.39618546e-01 7.76945829e-01
1.50081170e+00 1.98860839e-01 1.20796621e+00 1.66165888e-01
9.86069798e-01 3.42835367e-01 2.71170944e-01 1.32570535e-01
4.07865271e-02 6.33326828e-01 9.27235067e-01 -2.96299130e-01
-9.65108573e-02 -5.39325655e-01 6.96472168e-01 4.21623886e-03
6.75482333e-01 -7.87473738e-01 -6.76747561e-01 4.70668197e-01
-1.55320990e+00 -1.00546241e+00 2.23554790e-01 2.03402805e+00
9.00125742e-01 5.79766184e-02 -1.53469265e-01 1.52884886e-01
8.38670552e-01 1.53126419e-01 -6.95214927e-01 -7.04355001e-01
-1.84558362e-01 2.29707614e-01 6.50639534e-01 4.42335457e-01
-1.16873980e+00 1.15603197e+00 5.94422245e+00 1.02196503e+00
-1.10798812e+00 -5.48638105e-02 7.79299617e-01 2.52456069e-02
-3.17111820e-01 -1.06186613e-01 -3.13144028e-01 5.25641322e-01
9.66964066e-01 -4.40576077e-01 5.93012035e-01 8.35931659e-01
8.43084455e-02 3.89361262e-01 -1.23334444e+00 9.46498632e-01
3.98610055e-01 -1.25086927e+00 6.79957271e-01 -2.28914525e-03
6.10742807e-01 -3.30532700e-01 5.61656475e-01 1.20172195e-01
4.39211577e-01 -1.37047577e+00 9.63766515e-01 2.16288000e-01
1.31254220e+00 -7.61214495e-01 3.00920427e-01 -4.42517176e-02
-9.12278891e-01 -5.53778149e-02 -2.78474212e-01 3.95473599e-01
1.04858428e-01 1.88192308e-01 -1.03072059e+00 5.06853521e-01
4.23742563e-01 4.51522827e-01 -6.37721062e-01 6.86366975e-01
-3.01087260e-01 4.52416778e-01 4.01226729e-02 4.92368639e-01
2.74226993e-01 2.02085882e-01 7.05706954e-01 1.04994440e+00
2.89234638e-01 -1.43328846e-01 -5.72033338e-02 7.63061404e-01
-4.97696757e-01 1.95995364e-02 -1.05072486e+00 -2.05417559e-01
4.05390650e-01 1.14340389e+00 -2.69792140e-01 -2.82696933e-01
-2.85856217e-01 1.58708298e+00 -4.47591357e-02 5.01278996e-01
-1.36779892e+00 -5.08431852e-01 7.42182314e-01 4.61558141e-02
2.86665022e-01 1.98698446e-01 6.09116554e-02 -9.75938141e-01
1.22114077e-01 -1.55602896e+00 1.12727746e-01 -1.17048705e+00
-1.32088661e+00 8.47827673e-01 -1.86383098e-01 -1.26808357e+00
-3.75437856e-01 -6.60241723e-01 -6.86135471e-01 9.88445818e-01
-1.45938516e+00 -1.50641835e+00 -2.75252253e-01 1.07757270e+00
4.91081566e-01 -2.63664603e-01 6.99620545e-01 1.72807932e-01
-5.51815152e-01 1.30222464e+00 9.13939327e-02 4.62903619e-01
7.18092859e-01 -1.01801300e+00 8.91696930e-01 1.23720741e+00
1.18257947e-01 2.76343115e-02 7.81744838e-01 -5.12412488e-01
-1.62047029e+00 -1.64340293e+00 2.92797089e-01 -5.14714539e-01
7.18606770e-01 -6.79792821e-01 -6.90207362e-01 9.59105909e-01
4.27944750e-01 1.06007524e-01 3.79638612e-01 -1.00917768e+00
-6.49718285e-01 1.48487821e-01 -1.43647051e+00 1.04194856e+00
8.97812605e-01 -7.28023946e-01 -3.78089696e-01 6.25035703e-01
1.05013669e+00 -6.59951508e-01 -6.47283792e-01 1.10270515e-01
4.26608235e-01 -6.99491799e-01 1.28157663e+00 -6.83485985e-01
8.36782277e-01 -3.44172984e-01 -2.58739263e-01 -1.22547007e+00
-3.59737873e-02 -1.09925497e+00 -3.47044063e-03 1.22580719e+00
4.02459741e-01 -5.72044671e-01 5.56241930e-01 5.59329033e-01
-2.32256036e-02 -3.30496460e-01 -8.18551779e-01 -8.18315208e-01
1.39072120e-01 -5.85164547e-01 6.69651866e-01 7.31203377e-01
-4.99760956e-01 3.92841958e-02 -8.09478998e-01 5.97241282e-01
6.19593799e-01 -5.04080594e-01 1.06639838e+00 -4.65630382e-01
-4.89276767e-01 -1.26163244e-01 -5.70140421e-01 -6.92606926e-01
3.97536904e-01 -8.34216654e-01 2.07046136e-01 -1.02474189e+00
9.48096961e-02 -4.69412804e-02 1.87117487e-01 7.00960815e-01
-1.78594708e-01 7.87313163e-01 5.16683459e-01 1.09000176e-01
-2.01610148e-01 3.90573651e-01 1.30374682e+00 -5.98291695e-01
3.42730671e-01 -3.08126509e-01 -8.89281571e-01 5.81936359e-01
5.99536657e-01 -5.96016645e-01 -6.08255088e-01 -4.97718602e-01
7.70060644e-02 -3.18783969e-02 7.48265147e-01 -1.00057793e+00
-1.36092806e-03 -7.10259974e-02 3.95662308e-01 -1.48992121e-01
5.40832818e-01 -9.86516833e-01 2.79223144e-01 3.78485471e-01
-3.91956389e-01 5.54995751e-03 5.51065683e-01 6.36465013e-01
-1.03410393e-01 -1.71633109e-01 1.05076516e+00 -9.56832469e-02
-5.55031776e-01 5.47731698e-01 -2.73714423e-01 1.60649568e-01
1.41732812e+00 -1.59243673e-01 -5.98246038e-01 -7.99486279e-01
-6.71166718e-01 6.23640344e-02 6.90452158e-01 6.77028060e-01
8.48253071e-01 -1.32386708e+00 -9.42785561e-01 2.56232858e-01
2.15698585e-01 -2.36948222e-01 5.12592316e-01 1.78537071e-02
-5.90962827e-01 1.07007861e-01 -1.38141453e-01 -4.76312846e-01
-1.29098845e+00 1.06468225e+00 6.03891611e-01 -1.48313239e-01
-4.02253062e-01 8.45878363e-01 7.62131751e-01 -9.77339372e-02
1.45407841e-01 2.67971873e-01 1.40504897e-01 -4.16353583e-01
6.32913291e-01 -1.71420395e-01 -1.32230699e-01 -9.52516854e-01
-1.24979548e-01 5.32667398e-01 -1.63904741e-01 -1.93345606e-01
6.98459685e-01 -1.17360696e-01 1.94752607e-02 -2.75160730e-01
1.35072076e+00 -9.46285799e-02 -1.15683258e+00 -7.80936107e-02
-6.28432930e-01 -6.46174192e-01 -2.89641410e-01 -8.91636014e-01
-1.15640855e+00 8.54808569e-01 6.07286334e-01 2.09237486e-01
1.36029756e+00 -1.75721645e-01 8.68780911e-01 2.09829554e-01
9.23900530e-02 -4.92789626e-01 2.82033086e-01 2.98033088e-01
1.35921574e+00 -1.17222297e+00 -4.51280266e-01 -4.05859649e-01
-8.60878289e-01 7.71978199e-01 6.53446496e-01 5.52811213e-02
3.47309411e-01 2.85439581e-01 2.26615489e-01 1.65312767e-01
-6.97813511e-01 1.43783152e-01 2.99971044e-01 9.71212268e-01
-2.82798707e-01 -2.22391814e-01 4.11409974e-01 3.29912692e-01
-3.98692489e-01 -1.53091401e-01 8.61026764e-01 7.34844506e-01
1.08574063e-01 -1.14874315e+00 -7.68169165e-01 1.45961449e-01
-6.19355440e-01 -5.06243706e-01 -6.59823954e-01 7.42827713e-01
3.38845514e-02 1.16375494e+00 -9.14896578e-02 -5.74879825e-01
2.59006411e-01 4.77701761e-02 4.30953890e-01 -3.96376431e-01
-6.37382388e-01 -4.62670535e-01 -7.62850940e-02 -6.00623310e-01
2.92318203e-02 -1.65466249e-01 -7.38698661e-01 -4.74388242e-01
-1.05586715e-01 1.35794915e-02 7.25950480e-01 8.64346623e-01
5.22470474e-01 4.26333159e-01 1.09489572e+00 -8.34164619e-01
-5.45768499e-01 -6.36385500e-01 -2.97148377e-01 6.67077899e-01
3.66485327e-01 -1.34111777e-01 -5.70000827e-01 4.37688321e-01] | [5.7564239501953125, 7.819818019866943] |
6d655468-451c-4aca-b114-86d9e3a26bd8 | point-bert-pre-training-3d-point-cloud | 2111.14819 | null | https://arxiv.org/abs/2111.14819v2 | https://arxiv.org/pdf/2111.14819v2.pdf | Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modeling | We present Point-BERT, a new paradigm for learning Transformers to generalize the concept of BERT to 3D point cloud. Inspired by BERT, we devise a Masked Point Modeling (MPM) task to pre-train point cloud Transformers. Specifically, we first divide a point cloud into several local point patches, and a point cloud Tokenizer with a discrete Variational AutoEncoder (dVAE) is designed to generate discrete point tokens containing meaningful local information. Then, we randomly mask out some patches of input point clouds and feed them into the backbone Transformers. The pre-training objective is to recover the original point tokens at the masked locations under the supervision of point tokens obtained by the Tokenizer. Extensive experiments demonstrate that the proposed BERT-style pre-training strategy significantly improves the performance of standard point cloud Transformers. Equipped with our pre-training strategy, we show that a pure Transformer architecture attains 93.8% accuracy on ModelNet40 and 83.1% accuracy on the hardest setting of ScanObjectNN, surpassing carefully designed point cloud models with much fewer hand-made designs. We also demonstrate that the representations learned by Point-BERT transfer well to new tasks and domains, where our models largely advance the state-of-the-art of few-shot point cloud classification task. The code and pre-trained models are available at https://github.com/lulutang0608/Point-BERT | ['Jiwen Lu', 'Jie zhou', 'Tiejun Huang', 'Yongming Rao', 'Lulu Tang', 'Xumin Yu'] | 2021-11-29 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Yu_Point-BERT_Pre-Training_3D_Point_Cloud_Transformers_With_Masked_Point_Modeling_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Yu_Point-BERT_Pre-Training_3D_Point_Cloud_Transformers_With_Masked_Point_Modeling_CVPR_2022_paper.pdf | cvpr-2022-1 | ['3d-point-cloud-linear-classification', 'few-shot-point-cloud-classification', 'few-shot-3d-point-cloud-classification', 'point-cloud-segmentation', 'point-cloud-classification'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [-1.95371449e-01 1.67273253e-01 -1.10408038e-01 -2.15314075e-01
-1.21213007e+00 -5.10130048e-01 6.84707999e-01 -3.59578013e-01
1.71747789e-01 1.37635693e-01 -3.14115435e-01 -2.79225767e-01
1.70747131e-01 -1.10409713e+00 -1.49375606e+00 -5.88312745e-01
-6.83076233e-02 9.01409209e-01 1.27266183e-01 -2.12515563e-01
7.09969625e-02 7.77154267e-01 -1.63058257e+00 2.20287353e-01
7.72971869e-01 1.04391229e+00 4.35875088e-01 3.48390341e-01
-4.03755367e-01 5.11934876e-01 -4.91694093e-01 -3.10386300e-01
4.89353865e-01 3.94765615e-01 -6.39150620e-01 1.76694959e-01
6.25887632e-01 -3.29819649e-01 -5.88369250e-01 8.16238165e-01
2.01131076e-01 1.44988209e-01 7.21710980e-01 -1.23935354e+00
-1.02731073e+00 4.81852800e-01 -5.56762218e-01 -1.25380039e-01
-2.61397719e-01 3.82877469e-01 1.11972344e+00 -1.52951515e+00
2.96598375e-01 1.03314674e+00 9.61177468e-01 5.05303144e-01
-1.12726140e+00 -9.05305624e-01 8.86967257e-02 -3.17875785e-03
-1.60520124e+00 -1.69601709e-01 9.83913064e-01 -5.38998723e-01
1.15946686e+00 1.46897301e-01 7.70443380e-01 1.05583310e+00
-8.45140293e-02 8.31767976e-01 3.09622675e-01 -3.58842686e-02
2.92003900e-01 -2.31620729e-01 -9.21561867e-02 6.32565618e-01
-1.66224569e-01 2.04940900e-01 -2.73036540e-01 -3.73131454e-01
1.16402507e+00 5.39700210e-01 -5.48198633e-02 -5.54156482e-01
-1.24947608e+00 9.31585431e-01 1.04473615e+00 1.77989274e-01
-4.95444417e-01 6.48207366e-01 1.24762706e-01 1.03359088e-01
7.76993513e-01 3.27979147e-01 -3.62362444e-01 1.28449902e-01
-1.05047393e+00 3.74354422e-01 3.20990771e-01 1.54890871e+00
1.25469148e+00 3.39029282e-01 1.44352419e-02 7.65275776e-01
2.89661169e-01 7.68045008e-01 7.64522851e-02 -1.08318543e+00
5.74594676e-01 5.12767851e-01 7.05115944e-02 -5.80309033e-01
1.84258521e-01 -5.48586190e-01 -7.74420738e-01 3.35370690e-01
-1.64939985e-01 2.96760380e-01 -1.24680877e+00 1.29919517e+00
2.61774361e-01 7.49491274e-01 -1.29544571e-01 7.59354055e-01
8.23582947e-01 1.09604013e+00 -3.57128656e-03 4.98286009e-01
1.12670946e+00 -8.85519147e-01 1.11936018e-01 -1.27840117e-01
3.26516777e-01 -4.33339536e-01 1.34640360e+00 2.07854286e-01
-1.07560444e+00 -6.96620703e-01 -1.01249313e+00 -3.69780749e-01
-1.36386141e-01 9.50290337e-02 5.58152795e-01 -2.94005256e-02
-9.47980285e-01 8.66992414e-01 -1.29379821e+00 1.17909081e-01
9.05487180e-01 2.95674950e-01 -9.65968221e-02 -1.60095528e-01
-6.60565734e-01 5.03586113e-01 -2.01994568e-01 4.22938913e-02
-1.35528433e+00 -1.38356626e+00 -8.46727848e-01 2.57389694e-01
5.16786650e-02 -9.74641442e-01 1.49586058e+00 -6.35003507e-01
-1.34306395e+00 9.78531539e-01 -2.98709989e-01 -4.61421400e-01
4.02104676e-01 -1.51798189e-01 -5.18267276e-04 1.96111262e-01
4.20599103e-01 9.43183899e-01 1.19857156e+00 -1.51885688e+00
-4.88437593e-01 -1.42878845e-01 -2.59207841e-02 -8.55384842e-02
1.06228776e-01 -4.34341460e-01 -6.46370232e-01 -4.89794493e-01
5.46563044e-02 -8.50440562e-01 -1.96533218e-01 2.69224137e-01
-3.96655053e-01 -5.63077509e-01 8.80754113e-01 -2.74230808e-01
2.50305265e-01 -2.41131639e+00 1.38884842e-01 6.49549067e-02
5.76869071e-01 4.92856354e-02 -3.05256218e-01 4.63693798e-01
-2.67868251e-01 5.36462441e-02 -5.28747499e-01 -1.08996534e+00
2.08266065e-01 5.53126216e-01 -1.00119960e+00 4.68263656e-01
4.56678629e-01 1.19399750e+00 -7.45251775e-01 -2.06117243e-01
7.27901161e-01 6.58041954e-01 -7.31065452e-01 1.72157452e-01
-7.53804862e-01 4.16576505e-01 -4.90374118e-01 9.58561540e-01
1.05846953e+00 -5.26614070e-01 -5.87163746e-01 -3.21550220e-02
-1.11087173e-01 3.43879819e-01 -5.39326370e-01 2.07762456e+00
-5.97998619e-01 4.91526097e-01 -5.35289710e-03 -1.01694918e+00
9.48785365e-01 2.63513833e-01 8.17450821e-01 -2.39079878e-01
-7.02116117e-02 1.36239618e-01 -6.34748578e-01 -1.96848720e-01
4.38986957e-01 -3.71835679e-01 -4.34814692e-02 -6.03044592e-02
3.07015002e-01 -6.44437730e-01 -6.01134539e-01 5.25328442e-02
1.10017014e+00 2.36535221e-01 -2.54658818e-01 -2.00683344e-02
3.39376442e-02 3.89245003e-01 4.72066134e-01 6.64602280e-01
2.23456606e-01 1.17196822e+00 1.52992055e-01 -5.45648217e-01
-1.33025408e+00 -1.62505162e+00 -1.55491456e-01 6.78041637e-01
1.77029133e-01 -4.91184503e-01 -3.90286654e-01 -5.60961366e-01
4.63661879e-01 8.28625321e-01 -5.07309616e-01 -1.46771520e-01
-5.44793248e-01 -1.54829368e-01 5.16961694e-01 6.92571104e-01
2.02055961e-01 -9.20953631e-01 -2.57805407e-01 1.36324510e-01
-5.42887207e-03 -1.00362909e+00 -2.22980663e-01 4.35289815e-02
-1.09168482e+00 -7.89935589e-01 -7.11409450e-01 -8.99250925e-01
5.97756624e-01 4.86003608e-01 1.47435439e+00 1.97798729e-01
7.01894313e-02 3.30159515e-01 -3.19536090e-01 -5.49323440e-01
-1.23628303e-01 2.54422843e-01 -1.12213217e-01 -1.92785084e-01
4.53781188e-01 -1.07603645e+00 -5.52195489e-01 1.41385555e-01
-7.43288636e-01 -1.53580233e-02 6.45204544e-01 7.55394816e-01
1.09771228e+00 -1.16916239e-01 -1.02372011e-02 -4.60633785e-01
-1.32768571e-01 -7.39801824e-01 -6.68093145e-01 -2.32831925e-01
-1.04646504e-01 -1.00409053e-01 6.14342153e-01 -4.10285503e-01
-5.25498986e-01 1.51826978e-01 -5.33847392e-01 -1.63655400e+00
-6.60137311e-02 1.99219659e-01 -1.25782102e-01 -2.25088388e-01
5.88545144e-01 3.54383916e-01 -9.94593501e-02 -7.09539413e-01
4.49648082e-01 4.17142481e-01 7.66743302e-01 -9.54465866e-01
1.37965786e+00 9.25650060e-01 -2.44871065e-01 -9.16232824e-01
-6.40076220e-01 -5.40665686e-01 -5.56222260e-01 1.32076740e-01
7.59764791e-01 -1.27522922e+00 -6.01274371e-01 1.89586610e-01
-1.36063147e+00 -5.36036968e-01 -8.29808235e-01 7.01941252e-02
-8.17661881e-01 1.65443420e-01 -5.09205520e-01 -4.72146332e-01
-4.33197677e-01 -1.17872369e+00 1.82048368e+00 -3.10261518e-01
2.54219413e-01 -5.81472754e-01 -1.57288499e-02 1.29670873e-01
1.04134388e-01 2.10890502e-01 8.63788664e-01 -2.25407124e-01
-1.28944588e+00 -1.64816394e-01 -4.45184894e-02 4.46970373e-01
-1.90670062e-02 9.90934577e-03 -1.16936541e+00 -1.29931748e-01
1.77409410e-01 -9.74125117e-02 9.10930276e-01 4.84599024e-01
1.65867448e+00 -1.17210090e-01 -5.21675229e-01 1.19855177e+00
1.51435304e+00 -3.30695003e-01 6.29308224e-01 9.89482030e-02
1.09344172e+00 2.43824106e-02 3.56283635e-01 3.98388177e-01
4.65999871e-01 4.66200203e-01 9.23774183e-01 -7.51900971e-02
3.59366206e-03 -7.62126982e-01 1.11295231e-01 7.70189703e-01
-2.18447819e-01 4.47935462e-02 -1.06268084e+00 7.19532609e-01
-1.79317856e+00 -8.71360123e-01 7.71183521e-02 1.89202249e+00
4.83500123e-01 4.28387709e-02 -1.07553285e-02 -1.22108251e-01
5.84261119e-01 3.57793510e-01 -7.26059377e-01 2.42358983e-01
2.15314761e-01 6.52903080e-01 6.37563944e-01 4.46838528e-01
-9.91363287e-01 1.15886402e+00 5.57810068e+00 9.60440099e-01
-1.24552369e+00 3.87877792e-01 1.77424163e-01 -1.78576991e-01
-6.78802252e-01 2.73209028e-02 -5.57929277e-01 5.06660640e-01
7.86225080e-01 -5.59820086e-02 5.20186424e-01 1.35126650e+00
3.33687663e-02 8.21378648e-01 -1.20931673e+00 1.24233961e+00
-2.38195077e-01 -1.78339958e+00 2.38486096e-01 2.20556855e-01
4.85835046e-01 8.83257747e-01 3.01982135e-01 6.70454741e-01
4.05099034e-01 -1.02100146e+00 1.08072841e+00 4.40074861e-01
1.01715755e+00 -5.58983386e-01 2.04246238e-01 5.54154694e-01
-1.31812346e+00 1.00593202e-01 -9.51288283e-01 -3.00222691e-02
1.19642399e-01 7.03879356e-01 -8.98664296e-01 6.64251149e-01
8.74263704e-01 1.13041520e+00 -1.90860495e-01 9.84401524e-01
-1.97780520e-01 6.95059836e-01 -7.49376714e-01 3.26441944e-01
3.73186320e-01 -1.54839605e-01 7.72799373e-01 7.46323049e-01
5.60399771e-01 5.92292622e-02 -1.80547368e-02 1.63919294e+00
-3.64720643e-01 -4.27589089e-01 -8.16037655e-01 2.38203600e-01
6.94402814e-01 1.06788898e+00 -2.58184373e-01 -3.82728785e-01
-4.83206660e-01 8.55241120e-01 5.10504901e-01 3.90246272e-01
-1.03043187e+00 -1.71257555e-01 1.05362833e+00 3.09991121e-01
8.05088341e-01 -4.60362524e-01 -4.74793583e-01 -1.31525242e+00
2.33990550e-01 -4.54443157e-01 -9.36360955e-02 -1.14570689e+00
-1.53444624e+00 6.48064375e-01 1.10787079e-01 -1.80376446e+00
-9.36614424e-02 -5.83476782e-01 -9.40655112e-01 8.77781093e-01
-1.49801421e+00 -1.59241712e+00 -4.13995922e-01 4.99971896e-01
7.72379756e-01 5.52198337e-03 6.17957890e-01 2.87609726e-01
-1.77639395e-01 4.13471967e-01 1.15040883e-01 1.10468626e-01
1.50059447e-01 -1.31188917e+00 1.16554308e+00 5.89824677e-01
4.21021819e-01 5.47585964e-01 3.92146647e-01 -5.20789623e-01
-1.53129578e+00 -1.56107295e+00 4.07979935e-01 -9.54197943e-01
5.00138938e-01 -7.50548422e-01 -1.05900192e+00 1.24615896e+00
-2.03057438e-01 3.04029554e-01 9.03236344e-02 -4.85214218e-02
-4.62717772e-01 -7.56171718e-02 -1.08843076e+00 4.60083485e-01
1.22223771e+00 -6.96342528e-01 -9.38887060e-01 6.15640640e-01
1.17261171e+00 -7.86556721e-01 -8.71064603e-01 4.91854936e-01
-2.71474272e-02 -6.30885363e-01 1.52693450e+00 -3.63125682e-01
6.74912989e-01 -2.71169275e-01 -3.61471623e-01 -1.32532656e+00
-5.56435823e-01 -4.12537724e-01 -2.33053476e-01 9.57444787e-01
3.28619540e-01 -6.01324916e-01 1.19318080e+00 3.38801712e-01
-8.81950915e-01 -8.31752717e-01 -1.11739564e+00 -8.59769642e-01
6.11158729e-01 -8.28888476e-01 1.23196328e+00 7.98767388e-01
-5.69336593e-01 1.86341368e-02 -1.79876946e-02 6.69313729e-01
8.83819818e-01 3.45615625e-01 1.00244749e+00 -1.20871437e+00
-3.32504809e-01 -2.58429229e-01 -4.22348857e-01 -1.48799074e+00
3.85374099e-01 -1.11195624e+00 1.11761592e-01 -1.49644601e+00
-2.96464026e-01 -8.64748001e-01 -4.97992374e-02 6.54730439e-01
4.34125476e-02 1.88035160e-01 2.75595516e-01 7.14800477e-01
-1.40971333e-01 1.11628175e+00 1.09024465e+00 -5.59766471e-01
-1.01001620e-01 2.59605944e-01 -5.97810090e-01 5.12233496e-01
5.42220175e-01 -6.62453413e-01 -3.60802591e-01 -1.03659546e+00
-1.15940772e-01 -9.19825062e-02 8.30917239e-01 -1.18533909e+00
1.38711751e-01 -8.23755711e-02 3.60150754e-01 -1.04821372e+00
8.40763748e-01 -9.68509197e-01 2.67721891e-01 8.01273882e-02
1.43788502e-01 6.06558174e-02 2.54981488e-01 5.75235724e-01
-7.53614455e-02 3.39890197e-02 6.02882206e-01 -3.24127614e-01
-6.27038836e-01 9.66615260e-01 1.53162211e-01 -2.66674817e-01
8.45481277e-01 -1.29896507e-01 -1.75284818e-01 -7.95429572e-02
-6.92297816e-01 3.18937361e-01 8.35266650e-01 4.40870106e-01
8.76057148e-01 -1.41776764e+00 -6.02023423e-01 4.18596804e-01
2.38701671e-01 1.06934738e+00 3.48602653e-01 4.33424652e-01
-6.94159508e-01 1.20279521e-01 1.35319039e-01 -1.27913988e+00
-5.18453181e-01 7.09603906e-01 5.29294252e-01 2.85006016e-01
-1.49207735e+00 1.02566326e+00 3.97526801e-01 -7.96872735e-01
2.75159739e-02 -8.60509753e-01 5.46193898e-01 -6.40159845e-01
1.14638783e-01 -1.66533254e-02 2.86268353e-01 -4.52560425e-01
-2.95470417e-01 7.77572095e-01 1.43355981e-01 -1.95711777e-02
1.66186714e+00 4.14287686e-01 1.08360760e-01 5.23006916e-01
1.33210194e+00 -9.77648124e-02 -1.49942744e+00 -3.37922394e-01
-5.12865543e-01 -4.52660799e-01 3.35007645e-02 -1.90455183e-01
-1.18693042e+00 1.06368995e+00 1.47525579e-01 -3.78213503e-04
6.28746033e-01 5.49929082e-01 9.80679572e-01 3.58533323e-01
6.47761643e-01 -3.31050098e-01 -1.89569578e-01 4.11320627e-01
1.06325996e+00 -1.02032161e+00 -3.62947136e-01 -4.76010144e-01
-4.25913811e-01 7.71944106e-01 4.41277295e-01 -9.02627707e-01
8.51768970e-01 2.41811261e-01 -9.70170125e-02 -6.19553924e-01
-7.76622117e-01 -3.16290855e-02 8.54358822e-02 7.53605962e-01
-3.19750160e-01 1.86278537e-01 8.14547598e-01 6.63334668e-01
-6.97303772e-01 1.45720214e-01 7.96578377e-02 8.42447102e-01
-3.47966343e-01 -7.08034515e-01 -4.59776878e-01 5.08464158e-01
8.29811916e-02 -1.58686265e-01 -1.49020001e-01 7.89005995e-01
1.23337574e-01 4.53472912e-01 5.98913908e-01 -6.13908529e-01
5.56048632e-01 -8.92241001e-02 3.63325983e-01 -8.85648727e-01
-3.51483881e-01 -1.38514042e-01 -4.98289019e-01 -5.91699183e-01
-1.89623982e-02 -5.64989209e-01 -1.29033256e+00 -5.68721652e-01
-7.58848339e-02 3.10733646e-01 5.57178497e-01 6.87298179e-01
7.25948870e-01 5.19784212e-01 7.17020512e-01 -1.65883124e+00
-7.51233220e-01 -9.14310694e-01 -4.38407063e-01 2.29773954e-01
7.42373705e-01 -8.54098260e-01 -5.26406884e-01 -9.73703191e-02] | [8.082572937011719, -3.447631359100342] |
77684837-08b0-48bc-a578-b81da20409ae | program-induction-by-rationale-generation | 1705.04146 | null | http://arxiv.org/abs/1705.04146v3 | http://arxiv.org/pdf/1705.04146v3.pdf | Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problems | Solving algebraic word problems requires executing a series of arithmetic
operations---a program---to obtain a final answer. However, since programs can
be arbitrarily complicated, inducing them directly from question-answer pairs
is a formidable challenge. To make this task more feasible, we solve these
problems by generating answer rationales, sequences of natural language and
human-readable mathematical expressions that derive the final answer through a
series of small steps. Although rationales do not explicitly specify programs,
they provide a scaffolding for their structure via intermediate milestones. To
evaluate our approach, we have created a new 100,000-sample dataset of
questions, answers and rationales. Experimental results show that indirect
supervision of program learning via answer rationales is a promising strategy
for inducing arithmetic programs. | ['Wang Ling', 'Chris Dyer', 'Phil Blunsom', 'Dani Yogatama'] | 2017-05-11 | null | null | null | null | ['program-induction'] | ['computer-code'] | [ 2.51582444e-01 2.37186745e-01 -2.29750276e-01 -8.59133363e-01
-1.03653216e+00 -9.88970637e-01 3.94309819e-01 4.06278312e-01
-6.84392676e-02 5.38011968e-01 2.40644207e-03 -1.06354320e+00
1.64694980e-01 -1.36400604e+00 -1.02482283e+00 1.33211643e-01
6.18736446e-02 4.37497199e-01 2.86398470e-01 -3.91071171e-01
6.13680720e-01 8.95009190e-02 -1.46276760e+00 6.55414283e-01
1.23910058e+00 5.19108117e-01 1.25053842e-02 9.88446593e-01
-7.68394351e-01 1.46965086e+00 -5.82542300e-01 -7.55375087e-01
6.37365654e-02 -5.81231236e-01 -1.24925745e+00 -4.10107851e-01
7.50270665e-01 -4.45532143e-01 2.31600598e-01 1.08553660e+00
-1.30759209e-01 6.49073124e-02 5.93436539e-01 -1.20048010e+00
-8.92990410e-01 9.10933197e-01 -1.26263097e-01 -9.83691588e-02
1.04989552e+00 4.19257969e-01 1.68797529e+00 -8.79059196e-01
4.70309198e-01 1.24060774e+00 7.03913152e-01 6.12270296e-01
-1.60597444e+00 -4.75641727e-01 -3.01805705e-01 2.40211189e-02
-9.82231140e-01 -2.85623699e-01 7.10210204e-01 -5.15092969e-01
9.88387704e-01 3.23564708e-01 6.33297086e-01 2.86295652e-01
5.16131055e-03 8.75851035e-01 7.98883259e-01 -5.72210431e-01
1.60414323e-01 2.45204926e-01 9.26184058e-01 1.40825784e+00
1.70636177e-01 -1.98831260e-01 -1.26930520e-01 -3.73179168e-01
6.05275929e-01 -3.45669627e-01 -2.35754296e-01 -3.10559601e-01
-9.14532661e-01 1.23559785e+00 2.73692697e-01 -3.88807766e-02
1.11469164e-01 3.97297561e-01 2.85821199e-01 9.26766276e-01
-2.40332335e-01 9.94706273e-01 -5.92874825e-01 8.74530990e-03
-7.54290700e-01 6.79532051e-01 1.32003891e+00 9.12678599e-01
1.01309335e+00 -3.86022478e-02 -1.53032184e-01 2.68684924e-01
2.07491308e-01 4.52194303e-01 1.83789916e-02 -1.06171501e+00
5.99767745e-01 9.31382477e-01 1.12855598e-01 -1.12569976e+00
-1.54038697e-01 1.35948613e-01 -3.09193969e-01 1.63154244e-01
9.26351666e-01 -1.68845862e-01 -4.46536213e-01 1.85474777e+00
2.08106831e-01 -8.93511325e-02 1.75745599e-02 5.71063817e-01
1.02583468e+00 9.34167027e-01 -4.11721356e-02 8.73491317e-02
1.25867105e+00 -1.04292274e+00 -4.35463279e-01 -1.00147530e-01
1.41351867e+00 -4.73877251e-01 1.85142899e+00 3.56087983e-01
-1.30002213e+00 -6.13112867e-01 -7.82678962e-01 -3.56788397e-01
-1.91877484e-01 1.54648898e-02 1.11570835e+00 6.62050724e-01
-1.10331929e+00 4.48219866e-01 -2.95028031e-01 2.46607542e-01
1.58389643e-01 3.10551882e-01 -2.08663985e-01 -1.42466784e-01
-1.21589768e+00 8.41683269e-01 3.29468131e-01 -2.86991715e-01
-5.54418206e-01 -1.10951674e+00 -1.30812907e+00 1.47774607e-01
5.17310798e-01 -8.44923496e-01 1.56686366e+00 -7.33351886e-01
-1.45474994e+00 1.10133886e+00 -2.49878928e-01 -5.16498208e-01
1.43206287e-02 -1.53490573e-01 3.51529121e-02 8.71166438e-02
-4.60320637e-02 6.06153309e-01 7.00786769e-01 -7.94320524e-01
-2.73854136e-01 -8.58945400e-02 8.98308575e-01 -4.20281500e-01
-1.09753154e-01 5.38499296e-01 -1.05487421e-01 -3.25554132e-01
-8.25158060e-02 -7.18049705e-01 -3.67427170e-01 9.95005667e-02
-2.63609260e-01 -7.38569260e-01 1.08120181e-01 -6.55257881e-01
1.42088068e+00 -1.82639050e+00 1.53047532e-01 2.35257983e-01
3.67779732e-01 1.82647198e-01 -3.95121813e-01 7.16346055e-02
-2.55667925e-01 4.45069790e-01 -3.33237022e-01 2.60828495e-01
4.74891841e-01 1.30794570e-01 -6.90260947e-01 -3.74706723e-02
6.53044581e-01 1.37806284e+00 -1.00433218e+00 -4.73957151e-01
2.86951866e-02 -3.26224029e-01 -1.33064055e+00 6.36830688e-01
-1.11777258e+00 9.08499882e-02 -4.27609384e-01 4.35700089e-01
5.18032074e-01 -4.53001320e-01 -1.56419635e-01 2.03479484e-01
1.88995495e-01 7.94644952e-01 -1.17795289e+00 1.63778496e+00
-7.82181621e-01 6.26922667e-01 -1.33981004e-01 -1.18400836e+00
9.98158813e-01 4.46616486e-03 -2.66869724e-01 -3.74053478e-01
1.13864159e-02 3.82216305e-01 4.84053008e-02 -1.00419104e+00
5.39295852e-01 -3.74308318e-01 -6.21781170e-01 6.60746574e-01
-2.46731360e-02 -1.07037282e+00 6.36869550e-01 3.92737210e-01
1.19127059e+00 1.04314484e-01 3.42004299e-01 -1.47337317e-01
9.88667488e-01 3.88421714e-01 2.96893775e-01 8.14512849e-01
1.96200103e-01 2.70567477e-01 9.66511726e-01 -7.02455819e-01
-9.73950088e-01 -1.26406920e+00 3.33838046e-01 1.05187058e+00
-4.82851565e-01 -7.15974689e-01 -7.16919541e-01 -8.11730504e-01
-1.18178748e-01 1.15607607e+00 -3.37931216e-01 -1.87221140e-01
-1.08889604e+00 -5.66518791e-02 6.51092470e-01 5.40934563e-01
2.85397142e-01 -1.06625819e+00 -4.53442484e-01 -1.42843497e-03
-2.79833317e-01 -9.03190970e-01 -5.00249028e-01 1.63163036e-01
-7.74803817e-01 -1.32093191e+00 -2.18365356e-01 -9.94674861e-01
1.03984487e+00 2.10729223e-02 1.89017987e+00 6.80538476e-01
-1.39254853e-01 3.24088991e-01 -3.03924326e-02 -2.06986383e-01
-7.67065644e-01 8.02748352e-02 -5.07634044e-01 -5.22180438e-01
5.95297337e-01 -4.92859453e-01 5.58849201e-02 -3.07076275e-01
-8.96389186e-01 4.38662991e-02 2.20589608e-01 6.83511198e-01
2.49191850e-01 -9.60368365e-02 3.29198450e-01 -1.23043394e+00
8.14425826e-01 -3.10740709e-01 -1.17722130e+00 5.30090749e-01
-3.38237174e-02 6.90175831e-01 1.14027095e+00 -3.47162694e-01
-8.09564412e-01 1.40748784e-01 -2.71773458e-01 2.05574945e-01
-1.76910982e-01 8.21485043e-01 -7.16735125e-02 7.29943961e-02
9.10606205e-01 2.17693880e-01 -2.49685850e-02 -3.35663334e-02
7.43083119e-01 1.50654346e-01 7.05132663e-01 -1.38003230e+00
1.30920386e+00 -1.81243882e-01 -8.25981647e-02 -5.04983068e-01
-1.16416264e+00 -2.02818438e-01 -4.74145323e-01 1.93833426e-01
7.38687694e-01 -6.37730122e-01 -7.33677268e-01 4.63168845e-02
-1.75507545e+00 -5.70560038e-01 -2.56732285e-01 6.37143701e-02
-5.31859994e-01 2.63868570e-01 -4.62779939e-01 -5.88819563e-01
-2.59671777e-01 -1.17350352e+00 7.36493945e-01 1.45155817e-01
-9.51019883e-01 -1.14879966e+00 1.18850589e-01 3.95893931e-01
2.76817948e-01 1.20127082e-01 1.78269958e+00 -5.87806463e-01
-1.00106215e+00 -1.82773784e-01 -1.74484611e-01 4.94895935e-01
-1.10464670e-01 2.78600276e-01 -4.45829570e-01 3.59431803e-01
2.07822129e-01 -8.35834801e-01 5.10741591e-01 -8.57274532e-02
1.38002706e+00 -5.71948409e-01 3.08232039e-01 4.64184135e-01
1.33021390e+00 -3.86660993e-01 6.77078724e-01 -3.30817029e-02
3.29951555e-01 6.33162379e-01 2.84726322e-01 1.51029453e-01
6.86323047e-01 1.69228107e-01 1.48269594e-01 2.56667942e-01
3.83481905e-02 -5.01499832e-01 5.72591960e-01 8.53495777e-01
3.00263494e-01 3.69541258e-01 -1.05992889e+00 5.55143952e-01
-1.36199272e+00 -1.01006246e+00 -7.94643462e-01 2.07855248e+00
1.48130643e+00 5.38118482e-02 2.28614472e-02 2.82217741e-01
2.94334441e-01 2.47702301e-02 -9.30428058e-02 -6.64127827e-01
2.76452124e-01 9.70923066e-01 -2.07036972e-01 8.65461946e-01
-6.29186273e-01 9.76970196e-01 6.60618114e+00 6.50073469e-01
-8.82595062e-01 -2.79782951e-01 2.77613491e-01 2.80571312e-01
-1.10681081e+00 1.89578041e-01 -6.79869115e-01 1.59201771e-01
9.73486781e-01 -3.08409423e-01 6.96901202e-01 9.17881966e-01
-1.61200151e-01 -1.64028138e-01 -1.69966245e+00 6.27514839e-01
-1.31873816e-01 -1.54625177e+00 2.13613078e-01 -7.75358558e-01
6.77161992e-01 -7.16861069e-01 -4.00233641e-02 7.67323852e-01
7.81255722e-01 -1.30872631e+00 6.92456722e-01 3.42438459e-01
5.04167557e-01 -5.73467612e-01 3.77607942e-01 5.45860887e-01
-1.12456954e+00 -1.03836931e-01 -3.39601457e-01 -6.56533062e-01
-9.22917277e-02 6.91094875e-01 -6.25943899e-01 2.30570406e-01
5.51620126e-02 2.93165684e-01 -8.31040263e-01 7.56619275e-01
-1.03801191e+00 6.01143479e-01 -2.23708391e-01 -5.79404056e-01
2.24095345e-01 -1.94734782e-01 -7.74797648e-02 1.07374191e+00
2.30413720e-01 4.35766697e-01 8.53362828e-02 1.51903296e+00
-8.83012265e-02 8.51063356e-02 -7.80240178e-01 -3.11811507e-01
3.51059109e-01 1.09721923e+00 -3.69996786e-01 -5.40252090e-01
-6.19324386e-01 5.44819176e-01 7.18760669e-01 1.89831778e-01
-1.07728899e+00 -8.46053123e-01 4.41493124e-01 -6.44551665e-02
-1.07151985e-01 -4.91266102e-01 -6.94830239e-01 -1.40743613e+00
3.43468964e-01 -1.50923634e+00 3.54861110e-01 -9.64382946e-01
-9.19465840e-01 2.04561397e-01 -6.62513375e-02 -5.65747738e-01
-5.53363264e-01 -8.60374153e-01 -9.20485973e-01 9.16241169e-01
-1.44829273e+00 -6.53179169e-01 -2.74362296e-01 5.22387147e-01
3.83680969e-01 2.29565710e-01 1.12136400e+00 8.29386041e-02
-2.15505287e-01 6.06691599e-01 -7.38824487e-01 3.41315359e-01
3.20504278e-01 -1.41118586e+00 4.01477039e-01 6.56929076e-01
2.74162024e-01 1.11240327e+00 8.82715106e-01 -1.81147620e-01
-1.84922695e+00 -8.52101147e-01 1.35352254e+00 -8.69094372e-01
1.17653775e+00 -3.83354336e-01 -1.09013581e+00 7.03042626e-01
1.26490787e-01 -1.18116237e-01 7.18320191e-01 1.54906780e-01
-8.10927510e-01 2.05672309e-01 -8.11423004e-01 8.47325146e-01
7.92699158e-01 -1.02572227e+00 -1.27950311e+00 3.81002605e-01
1.05615127e+00 -3.58225167e-01 -5.55347264e-01 2.48645261e-01
1.87665626e-01 -7.30138898e-01 9.38550770e-01 -1.03023601e+00
1.19170821e+00 -3.61250579e-01 -1.03471935e-01 -1.10522962e+00
2.49794558e-01 -8.32478702e-01 -2.16058016e-01 1.17793846e+00
7.35384822e-01 -4.21817869e-01 8.62834573e-01 1.12935007e+00
1.38848154e-02 -6.65855885e-01 -1.70047119e-01 -6.17643416e-01
6.14229977e-01 -7.91516483e-01 7.64095187e-01 7.82828510e-01
3.35417598e-01 6.27443492e-01 1.10044196e-01 -8.51418823e-02
4.12118614e-01 6.37317002e-01 1.25597239e+00 -1.18743742e+00
-6.08788073e-01 -5.69935024e-01 1.64733440e-01 -1.34144390e+00
7.40840733e-01 -1.41695917e+00 1.24657750e-01 -1.38536239e+00
3.39717558e-03 -5.25610685e-01 4.60540116e-01 4.82658207e-01
-3.67867172e-01 -2.74611175e-01 -4.68411110e-02 -4.80390847e-01
-6.24670208e-01 1.95065573e-01 1.24441564e+00 -3.33298624e-01
1.19876504e-01 1.00526355e-01 -8.50957096e-01 1.06222045e+00
8.07785511e-01 -5.22882283e-01 -6.45646214e-01 -7.26556540e-01
9.81174111e-01 5.35396695e-01 4.30594355e-01 -6.11525357e-01
4.64602232e-01 -4.15630430e-01 -4.15789992e-01 -5.27694881e-01
-1.22395001e-01 -4.63447183e-01 -4.12604630e-01 6.21351540e-01
-7.99736381e-01 2.12283686e-01 1.17198840e-01 5.42958863e-02
-3.80725622e-01 -1.13085771e+00 5.30125737e-01 -4.70713437e-01
-5.79114676e-01 -1.29980985e-02 -3.25674593e-01 7.24777162e-01
6.23491883e-01 5.02664670e-02 -2.15162560e-01 -3.45224291e-01
-3.34964633e-01 3.93207073e-01 3.62160981e-01 5.05766571e-02
6.38012946e-01 -1.11940789e+00 -6.35823429e-01 2.39319324e-01
8.58595595e-02 1.89728767e-01 -3.71647775e-01 4.56857502e-01
-8.14828515e-01 5.65296173e-01 1.96376100e-01 -2.57376194e-01
-1.31334424e+00 5.93090057e-01 1.56456262e-01 -6.31983578e-01
-3.24108414e-02 1.12291539e+00 1.42499119e-01 -1.29348385e+00
2.51166731e-01 -9.49959397e-01 -5.11782765e-02 -2.41737992e-01
7.04685986e-01 -1.77941889e-01 -2.54055917e-01 2.35336572e-01
4.07348350e-02 5.95646560e-01 1.55829668e-01 2.56409675e-01
1.41639781e+00 4.42914665e-01 -8.70242774e-01 1.17393874e-01
1.28048706e+00 1.42794594e-01 -5.66558957e-01 -3.80092800e-01
5.77272654e-01 -4.22700077e-01 -6.57732010e-01 -4.11897659e-01
-6.37223065e-01 1.18720508e+00 -4.63182837e-01 2.21215099e-01
8.42064500e-01 2.93167122e-02 8.34485650e-01 1.21150875e+00
5.11113822e-01 -6.36065364e-01 5.97948611e-01 1.03667021e+00
8.74408424e-01 -1.26803052e+00 -4.58431065e-01 -7.10058212e-01
2.75842436e-02 1.45462716e+00 7.60888040e-01 -1.83497876e-01
3.50946814e-01 4.66299921e-01 -2.67507404e-01 -2.21259981e-01
-1.00600958e+00 -8.15626793e-03 6.14603162e-02 3.99681658e-01
7.37745285e-01 -2.03365892e-01 -1.34814575e-01 9.77063358e-01
-7.55481601e-01 2.59237915e-01 7.25628257e-01 8.33994865e-01
-5.50945640e-01 -1.26056647e+00 -4.55974400e-01 3.99887383e-01
-3.38442683e-01 -5.55712223e-01 -3.03607136e-01 5.36491275e-01
-2.73190767e-01 9.06518579e-01 -2.46460766e-01 -2.29346082e-01
9.86699089e-02 3.82522166e-01 8.96983325e-01 -1.21014369e+00
-6.64679110e-01 -1.11538541e+00 4.61156696e-01 -2.87730455e-01
-9.88339037e-02 -2.69965351e-01 -1.80621755e+00 -6.59056067e-01
-2.84310043e-01 3.85821432e-01 2.64257282e-01 1.05888021e+00
-7.67612532e-02 4.57284123e-01 5.22007525e-01 -1.20114304e-01
-1.25315404e+00 -3.44472378e-01 6.91591436e-03 4.43214417e-01
1.87208965e-01 9.10231322e-02 -3.96098435e-01 3.71647298e-01] | [9.584199905395508, 7.43862771987915] |
3601d06d-e150-4633-bef9-9f5f35f90a74 | collaborative-neural-rendering-using-anime | 2207.05378 | null | https://arxiv.org/abs/2207.05378v5 | https://arxiv.org/pdf/2207.05378v5.pdf | Collaborative Neural Rendering using Anime Character Sheets | Drawing images of characters with desired poses is an essential but laborious task in anime production. Assisting artists to create is a research hotspot in recent years. In this paper, we present the Collaborative Neural Rendering (CoNR) method, which creates new images for specified poses from a few reference images (AKA Character Sheets). In general, the diverse hairstyles and garments of anime characters defies the employment of universal body models like SMPL, which fits in most nude human shapes. To overcome this, CoNR uses a compact and easy-to-obtain landmark encoding to avoid creating a unified UV mapping in the pipeline. In addition, the performance of CoNR can be significantly improved when referring to multiple reference images, thanks to feature space cross-view warping in a carefully designed neural network. Moreover, we have collected a character sheet dataset containing over 700,000 hand-drawn and synthesized images of diverse poses to facilitate research in this area. Our code and demo are available at https://github.com/megvii-research/IJCAI2023-CoNR. | ['Zhewei Huang', 'Ailin Huang', 'Zuzeng Lin'] | 2022-07-12 | null | null | null | null | ['image-to-video', 'image-to-3d'] | ['computer-vision', 'computer-vision'] | [ 8.77821967e-02 -2.05191284e-01 2.14128733e-01 -3.03409308e-01
-3.15304369e-01 -6.18433714e-01 5.83559930e-01 -5.39037883e-01
-3.39648165e-02 5.18737555e-01 2.63872415e-01 1.07777402e-01
2.20046118e-01 -9.41785932e-01 -6.10941172e-01 -3.71050656e-01
3.85553509e-01 4.62796420e-01 7.22190365e-02 -4.99943048e-01
1.45001218e-01 7.61780620e-01 -1.33748245e+00 1.33939400e-01
7.17952192e-01 6.14582777e-01 3.23374242e-01 5.78368068e-01
2.90939938e-02 3.47473353e-01 -6.80620790e-01 -9.10866082e-01
3.84119660e-01 -5.45662940e-01 -3.51815820e-01 1.56984121e-01
6.18859708e-01 -5.03336608e-01 -3.56858492e-01 8.74940693e-01
7.15523064e-01 1.12573564e-01 6.35454118e-01 -9.75419402e-01
-8.63750994e-01 4.87721533e-01 -8.05162847e-01 -3.95451456e-01
4.42318231e-01 2.17813060e-01 7.14107394e-01 -8.81116569e-01
8.92866552e-01 1.21589112e+00 6.59456670e-01 7.17511177e-01
-1.17345953e+00 -9.09787714e-01 -6.05294779e-02 -9.89466906e-02
-1.48716307e+00 -3.52113038e-01 1.19251716e+00 -4.02517408e-01
3.49293143e-01 4.91038024e-01 1.08096385e+00 1.33109081e+00
8.53198469e-02 7.56140113e-01 8.90341461e-01 -1.81335464e-01
4.44371179e-02 -1.79903015e-01 -5.70395470e-01 6.22950792e-01
8.65322426e-02 -1.40343934e-01 -5.51518977e-01 3.05470079e-02
1.54830694e+00 -2.30475795e-02 -2.32156724e-01 -5.41606009e-01
-1.18989336e+00 7.27373779e-01 5.72115958e-01 1.30023643e-01
-2.63606876e-01 3.27395558e-01 2.17950851e-01 -5.37846312e-02
3.93264055e-01 4.99357402e-01 -4.32594493e-03 -1.20044217e-01
-9.30982053e-01 3.01582336e-01 4.45679277e-01 1.02729368e+00
4.63328421e-01 2.76386082e-01 7.24427179e-02 1.01192987e+00
1.93241537e-01 4.28638220e-01 2.20398545e-01 -1.07184422e+00
4.59974349e-01 4.79099214e-01 -5.05667254e-02 -1.24555326e+00
-2.90378779e-01 -2.55199820e-01 -1.19444215e+00 4.02045637e-01
4.19804305e-01 -1.58120945e-01 -8.66169631e-01 1.43885040e+00
3.98907870e-01 -1.15936160e-01 -2.36650184e-01 1.09331167e+00
9.11887705e-01 6.39074683e-01 -1.27458304e-01 3.37080419e-01
1.24852753e+00 -8.67737770e-01 -6.13701940e-01 -1.50202304e-01
5.86388893e-02 -8.94615412e-01 1.08372581e+00 3.25467408e-01
-1.15677178e+00 -6.29411459e-01 -1.07756579e+00 -4.70238090e-01
-1.80908307e-01 3.86919022e-01 7.74411142e-01 5.01652718e-01
-9.40577865e-01 6.44542992e-01 -5.78553438e-01 -2.46202633e-01
6.02563858e-01 1.15581580e-01 -4.77490127e-01 5.10658324e-02
-9.36210454e-01 6.88168108e-01 1.24796741e-01 3.85672897e-01
-6.25414968e-01 -4.81085032e-01 -7.39929318e-01 -3.01173210e-01
3.05985004e-01 -7.06101716e-01 8.03522766e-01 -8.21034908e-01
-1.86484897e+00 7.58862376e-01 1.71839178e-01 3.65205714e-03
1.07085502e+00 -4.44224358e-01 -9.88348722e-02 3.19888219e-02
-1.99258447e-01 9.42889392e-01 7.77428389e-01 -1.41427982e+00
8.49652141e-02 -2.40055636e-01 3.92952785e-02 2.89228231e-01
-4.31368873e-02 4.96890582e-02 -8.65488350e-01 -1.06082225e+00
2.06117466e-01 -1.00795519e+00 -2.56977767e-01 4.51459944e-01
-5.66658318e-01 1.48563281e-01 6.66984975e-01 -9.96147513e-01
8.39612007e-01 -1.89356422e+00 4.27702397e-01 2.20257387e-01
2.30911717e-01 1.88895851e-01 -1.01956204e-01 5.08963704e-01
2.70884708e-02 8.78654420e-02 -1.91087186e-01 -4.48174089e-01
-5.59658110e-02 -2.32088361e-02 -1.54826775e-01 4.22608614e-01
1.09874301e-01 1.03735077e+00 -6.31965697e-01 -5.52405596e-01
3.78328264e-01 8.26008320e-01 -3.91267389e-01 1.69421181e-01
-1.84519216e-01 8.51126671e-01 -2.92249829e-01 7.59984016e-01
7.15838194e-01 -7.69112855e-02 2.45154873e-01 -3.50526541e-01
-4.11609709e-02 -2.94166595e-01 -1.09443212e+00 2.15293288e+00
-3.86425525e-01 7.83826053e-01 -1.73721969e-01 -2.29350463e-01
1.32451975e+00 1.47465274e-01 2.97364891e-01 -5.14242649e-01
3.38487893e-01 9.20223743e-02 -1.93425700e-01 -2.90411890e-01
6.85143530e-01 1.93221703e-01 1.12635000e-02 3.21424067e-01
-1.59845889e-01 -4.01943445e-01 1.77834421e-01 1.62953958e-01
5.85182011e-01 7.86738098e-01 1.52882069e-01 3.69412303e-02
2.69699812e-01 -2.16042727e-01 5.59396088e-01 2.79443026e-01
1.22195616e-01 1.22373247e+00 3.31705004e-01 -7.38761365e-01
-1.43489468e+00 -1.09600425e+00 -9.39608514e-02 8.19749177e-01
9.30740014e-02 -3.59914452e-01 -6.77589297e-01 -2.63262004e-01
-1.59838170e-01 5.38005888e-01 -7.00243771e-01 2.34415576e-01
-9.70015109e-01 -4.01785284e-01 5.35612524e-01 5.93092859e-01
6.68975174e-01 -1.20777643e+00 -7.58710861e-01 8.30757394e-02
-1.97880030e-01 -8.16056311e-01 -5.58422804e-01 -2.68963784e-01
-6.67338073e-01 -7.88681686e-01 -1.41556156e+00 -5.88888347e-01
8.36453974e-01 -1.34795094e-02 1.07572067e+00 6.08075075e-02
-4.00867939e-01 -1.39448613e-01 -4.13854480e-01 -2.10012957e-01
-3.70799810e-01 1.67641938e-02 -1.67891279e-01 -2.51314659e-02
-7.51864761e-02 -6.81184471e-01 -5.74490190e-01 4.16682452e-01
-7.51481116e-01 7.56112814e-01 4.08141702e-01 5.54597855e-01
5.49112022e-01 -4.47757065e-01 2.19595969e-01 -6.46185696e-01
5.72492540e-01 -1.09606951e-01 -6.76108420e-01 2.55110562e-01
7.77179375e-02 -2.24951014e-01 6.11894071e-01 -5.39146602e-01
-9.33641732e-01 1.92956746e-01 -2.45047942e-01 -5.25569320e-01
-9.22494233e-02 -1.33442814e-02 -4.24180388e-01 -6.64628446e-02
6.18827939e-01 2.18497574e-01 3.05183660e-02 -5.90871751e-01
5.50366938e-01 5.26738286e-01 6.87436879e-01 -7.72484303e-01
1.03325057e+00 3.75230461e-01 -7.88480602e-03 -7.75991261e-01
-2.48289615e-01 2.52265036e-01 -8.89489233e-01 -6.95787609e-01
8.26560795e-01 -8.18089068e-01 -5.23493707e-01 7.43339241e-01
-1.16243923e+00 -4.25170302e-01 -5.10709696e-02 2.83150613e-01
-5.76144159e-01 3.01425844e-01 -5.62374234e-01 -6.80475116e-01
-2.93859184e-01 -1.20731199e+00 1.06209886e+00 4.74501312e-01
-4.08090115e-01 -6.15000427e-01 -1.77094303e-02 4.85837460e-01
3.75360906e-01 7.32704222e-01 4.72731680e-01 2.47291345e-02
-6.61638618e-01 -6.94724917e-02 -2.13817388e-01 9.64217260e-02
1.88252971e-01 3.38307321e-01 -7.90140927e-01 -2.22153008e-01
-5.41673958e-01 -3.74003351e-01 5.56098282e-01 2.38794565e-01
1.19065452e+00 -7.20538348e-02 -1.93207606e-03 9.98919308e-01
1.23185384e+00 3.01223636e-01 8.39386523e-01 2.75962472e-01
1.13053298e+00 6.52706265e-01 4.68846321e-01 6.04648054e-01
2.64233857e-01 8.04863691e-01 2.06377760e-01 -3.36500525e-01
-3.67858559e-01 -5.10783672e-01 1.17112650e-02 6.48411095e-01
-6.40648365e-01 -2.25065053e-01 -9.37418759e-01 2.48257026e-01
-1.59941721e+00 -7.57646143e-01 -5.62843494e-02 2.08319020e+00
9.87903833e-01 -2.68990844e-01 5.95272817e-02 -1.28369972e-01
8.40109885e-01 3.02022696e-01 -6.09464705e-01 -2.47683764e-01
-2.50210345e-01 1.75436269e-02 2.97540307e-01 2.20497310e-01
-9.16334629e-01 1.11371922e+00 5.61820650e+00 8.24950516e-01
-1.19939947e+00 -1.69019789e-01 6.58637583e-01 -2.10125774e-01
-2.81628162e-01 -2.76001036e-01 -4.62421566e-01 5.45724511e-01
3.71744335e-01 -3.13750072e-03 5.63884795e-01 8.13307583e-01
6.12739176e-02 1.63202570e-03 -8.39616120e-01 1.28714335e+00
2.51070440e-01 -1.52197266e+00 1.06311783e-01 5.10575883e-02
7.46708989e-01 -2.50795394e-01 1.81944370e-01 -1.61373228e-01
2.96950579e-01 -1.10814500e+00 8.86750221e-01 6.76270008e-01
1.17545915e+00 -8.98265123e-01 4.31286842e-01 6.59139529e-02
-1.10306311e+00 3.44569415e-01 -4.82277602e-01 1.57826394e-01
4.46742862e-01 3.00550461e-01 -4.75858629e-01 5.93790293e-01
7.40234137e-01 6.18741930e-01 -6.90394044e-01 1.01327920e+00
-4.41687644e-01 5.69478571e-02 -2.92305648e-01 -6.02965020e-02
-1.72348186e-01 -6.18779838e-01 3.73812258e-01 8.68058980e-01
4.54470009e-01 -5.27138561e-02 -1.39515504e-01 9.78789151e-01
-1.82301685e-01 1.59141049e-01 -5.68193555e-01 -7.17766806e-02
3.31308722e-01 1.34887922e+00 -9.22388494e-01 -1.54671788e-01
1.89022869e-02 1.35662472e+00 2.00646088e-01 2.95972764e-01
-1.04850638e+00 -2.39824012e-01 4.37185913e-01 1.70901880e-01
1.12116113e-01 -5.54869294e-01 -3.40428740e-01 -1.18887842e+00
1.16266936e-01 -9.32893455e-01 -2.44599849e-01 -1.06060565e+00
-1.06155050e+00 8.41148019e-01 2.44776420e-02 -1.41064608e+00
-9.58660692e-02 -4.12860036e-01 -5.45599878e-01 9.62116063e-01
-8.98910880e-01 -1.60242319e+00 -5.03477156e-01 3.42785567e-01
5.91868162e-01 -1.49842530e-01 7.25061476e-01 3.15271646e-01
-6.61424935e-01 5.69752455e-01 -1.21668242e-01 5.00201941e-01
7.66027391e-01 -9.32076454e-01 9.40371096e-01 7.87994504e-01
1.87099814e-01 6.52540624e-01 5.46784520e-01 -9.30777073e-01
-1.32322371e+00 -9.96461332e-01 4.54799831e-01 -2.99269378e-01
3.17139804e-01 -6.39840126e-01 -7.81095147e-01 6.35424614e-01
3.17149758e-01 -1.19050145e-01 6.52818203e-01 -1.09331883e-01
-2.09255189e-01 1.69327229e-01 -1.00880885e+00 1.02698898e+00
1.30092430e+00 -2.24367723e-01 -3.20992202e-01 1.88577324e-01
2.65649110e-01 -8.09861720e-01 -8.87566090e-01 1.12645596e-01
9.64069724e-01 -9.80552018e-01 1.00984049e+00 -2.09299192e-01
7.50224590e-01 -4.44993526e-01 5.05828019e-03 -1.25924397e+00
-2.64673293e-01 -6.00423813e-01 8.43943655e-02 1.34567714e+00
6.33019283e-02 -4.43833649e-01 8.26923430e-01 6.69902325e-01
1.23484492e-01 -8.15569580e-01 -6.73882604e-01 -6.03648245e-01
7.42145032e-02 -2.05879509e-01 8.98804367e-01 8.28383982e-01
-5.16948104e-01 1.01536000e-02 -8.37054908e-01 -1.82904795e-01
6.59526765e-01 2.08058968e-01 1.19407547e+00 -1.11876416e+00
-2.28544265e-01 -3.07953358e-01 -4.32501107e-01 -1.09500086e+00
-1.41122729e-01 -6.83953226e-01 -1.83662653e-01 -1.64235115e+00
1.95561033e-02 -5.55404425e-01 4.44413215e-01 5.10949552e-01
4.40139025e-02 7.18730032e-01 6.35406077e-01 2.92176932e-01
-1.73240110e-01 7.38472998e-01 1.70222425e+00 -3.99430171e-02
-2.16009870e-01 -2.65439600e-01 -4.61466670e-01 8.48133147e-01
1.08284760e+00 -2.02739477e-01 -2.43511543e-01 -7.30554402e-01
2.84621030e-01 3.35269347e-02 4.07960206e-01 -1.12933743e+00
7.97681287e-02 -1.43349737e-01 9.62311208e-01 -8.83765399e-01
8.39234710e-01 -6.62203312e-01 8.26662004e-01 2.31984034e-01
-1.79748714e-01 2.26599261e-01 -6.38558641e-02 1.52817339e-01
1.32145986e-01 -1.13649376e-01 7.65098512e-01 -3.33588123e-01
-6.61679864e-01 2.98372895e-01 2.72785407e-02 -2.68615782e-01
1.04907548e+00 -5.30630708e-01 -1.70096546e-01 -3.80266190e-01
-4.52674747e-01 -4.48270924e-02 9.81324434e-01 6.06025755e-01
7.77902305e-01 -1.47765970e+00 -8.67822945e-01 1.55075103e-01
-4.70692441e-02 3.15266609e-01 5.44146299e-01 3.94666791e-01
-1.14526141e+00 -9.70236212e-02 -7.47160196e-01 -4.09070611e-01
-1.29466677e+00 2.35855728e-01 1.51240721e-01 1.99558631e-01
-9.03480947e-01 8.23311090e-01 1.13894209e-01 -4.24198568e-01
-3.97784039e-02 8.81222412e-02 -2.05222607e-01 4.37492095e-02
4.89237249e-01 3.81287336e-01 -3.24713707e-01 -9.36619878e-01
-8.86080861e-02 8.22604895e-01 5.76420650e-02 -1.07660972e-01
1.37824738e+00 1.00289449e-01 5.50515726e-02 2.82267779e-01
9.32089686e-01 2.75410503e-01 -1.59764719e+00 1.14288270e-01
-6.35014832e-01 -8.07788551e-01 -2.44921893e-01 -7.55900085e-01
-1.35248625e+00 7.86166728e-01 4.28204030e-01 -3.84037346e-01
9.26035285e-01 -1.93562493e-01 7.79697716e-01 5.46269119e-02
6.61209583e-01 -1.09328461e+00 6.00640252e-02 2.42236480e-01
1.53556204e+00 -9.27947462e-01 1.60146326e-01 -3.63503784e-01
-8.96255136e-01 1.32379305e+00 6.64829493e-01 -2.32207328e-01
1.15210578e-01 1.55275404e-01 3.21309090e-01 -5.17607965e-02
-1.35469973e-01 5.37346788e-02 2.27614373e-01 7.23701775e-01
4.25870270e-01 1.23965897e-01 -1.97543129e-01 3.03190410e-01
-6.49334788e-01 -1.50347754e-01 5.14722109e-01 8.49767625e-01
8.30662847e-02 -1.21058726e+00 -5.23835063e-01 2.09578216e-01
-2.51000702e-01 7.54947811e-02 -5.35992682e-01 7.55084276e-01
1.59574866e-01 5.46179891e-01 3.70735489e-02 -4.52141851e-01
1.80734619e-01 -2.30043590e-01 5.35684228e-01 -3.39751482e-01
-4.63444620e-01 1.99079931e-01 -7.18845949e-02 -5.02174795e-01
-2.49925569e-01 -5.71443558e-01 -9.37658846e-01 -4.09515917e-01
-7.02877641e-02 -3.29757452e-01 8.28961372e-01 3.07672143e-01
2.52449512e-01 5.93960583e-01 4.89973098e-01 -1.47571731e+00
-9.66458488e-03 -8.99102390e-01 -4.99128997e-01 3.77840579e-01
-2.98086673e-01 -7.62444437e-01 3.81970480e-02 1.35645092e-01] | [11.829695701599121, -0.48429644107818604] |
67bf78ab-6dfc-4b82-b609-4ac967e1b2f1 | incremental-natural-language-processing | null | null | https://aclanthology.org/C18-1253 | https://aclanthology.org/C18-1253.pdf | Incremental Natural Language Processing: Challenges, Strategies, and Evaluation | Incrementality is ubiquitous in human-human interaction and beneficial for human-computer interaction. It has been a topic of research in different parts of the NLP community, mostly with focus on the specific topic at hand even though incremental systems have to deal with similar challenges regardless of domain. In this survey, I consolidate and categorize the approaches, identifying similarities and differences in the computation and data, and show trade-offs that have to be considered. A focus lies on evaluating incremental systems because the standard metrics often fail to capture the incremental properties of a system and coming up with a suitable evaluation scheme is non-trivial. | ['Arne K{\\"o}hn'] | 2018-08-01 | incremental-natural-language-processing-2 | https://aclanthology.org/C18-1253 | https://aclanthology.org/C18-1253.pdf | coling-2018-8 | ['dialogue-understanding'] | ['natural-language-processing'] | [ 1.70889303e-01 1.29263118e-01 -2.55283922e-01 -4.53676552e-01
-3.21126610e-01 -9.33299780e-01 9.35109138e-01 5.86172462e-01
-3.94112915e-01 3.67987037e-01 2.23551840e-01 -2.86255687e-01
-5.96452296e-01 -3.30349445e-01 -4.97225635e-02 -2.87683427e-01
-9.04229060e-02 8.32260072e-01 3.17258507e-01 -4.08156157e-01
4.96138901e-01 4.92715597e-01 -1.73047137e+00 3.81445527e-01
6.53944731e-01 6.38327777e-01 -9.17255804e-02 8.25489461e-01
-3.75807285e-01 8.63562167e-01 -5.98746717e-01 -5.39394259e-01
3.38577092e-01 -6.13712072e-01 -1.34091330e+00 -1.19688660e-01
2.50423819e-01 1.73267424e-01 2.05996409e-01 1.01245642e+00
7.27098346e-01 -1.84457451e-02 6.89673007e-01 -1.62911975e+00
-1.71890244e-01 6.20469689e-01 6.77942857e-02 2.95903623e-01
1.01401162e+00 2.85047982e-02 1.27148044e+00 -7.90838897e-01
9.00882781e-01 1.10919869e+00 8.14495444e-01 3.08617920e-01
-1.28831208e+00 6.18674979e-02 6.36245459e-02 2.75411308e-01
-1.44060063e+00 -5.42389512e-01 5.71769893e-01 -6.47196651e-01
1.18849027e+00 5.57376742e-01 5.06064713e-01 6.87588632e-01
-2.34216861e-02 8.67642105e-01 1.19984543e+00 -7.01472938e-01
2.13243738e-01 3.77567261e-01 6.16352856e-01 2.59779215e-01
1.96802244e-01 -1.88177601e-01 -4.49097276e-01 -3.09959710e-01
2.52251059e-01 -2.60467350e-01 -1.51687786e-01 -6.74652338e-01
-1.22551525e+00 5.11238039e-01 -1.97990745e-01 1.10875845e+00
-3.15597475e-01 -2.84375727e-01 6.80059612e-01 7.58813560e-01
2.89271623e-01 8.69260669e-01 -6.46376729e-01 -7.71999240e-01
-9.52498078e-01 7.45545745e-01 1.50257123e+00 1.04358029e+00
5.38352489e-01 -5.56000710e-01 -2.65337855e-01 5.88852644e-01
-2.08138868e-01 -4.96498160e-02 4.68662173e-01 -1.04365599e+00
2.41580218e-01 1.02022719e+00 3.82462293e-01 -1.04754913e+00
-5.57352483e-01 -3.07198428e-02 -4.70740736e-01 1.21430479e-01
5.36375284e-01 -6.17475286e-02 -3.27726483e-01 1.44446862e+00
3.53281885e-01 -4.16976452e-01 -2.17057746e-02 5.05307198e-01
5.46196997e-01 3.86122584e-01 6.52927905e-02 -4.84149724e-01
1.29524362e+00 -9.56728101e-01 -7.53027022e-01 -7.96141773e-02
8.80942762e-01 -9.53663647e-01 1.06932199e+00 4.31900620e-01
-1.30381262e+00 -4.84786272e-01 -7.81964183e-01 -2.06121713e-01
-6.51413798e-01 -3.00444186e-01 5.28174102e-01 6.66980982e-01
-1.16137016e+00 7.98334301e-01 -6.49179101e-01 -7.72657335e-01
-3.71237934e-01 4.06684935e-01 -2.75182247e-01 9.39550698e-02
-1.10802472e+00 1.30002379e+00 4.68169391e-01 -8.00972506e-02
-1.43819898e-01 -6.32510364e-01 -5.27686775e-01 -2.17213738e-03
6.57352090e-01 -7.79839039e-01 1.74303198e+00 -8.85039628e-01
-1.58821774e+00 8.11251521e-01 -2.50516832e-01 -4.82620299e-01
9.23502922e-01 -2.83373088e-01 -2.75429547e-01 -2.17455804e-01
6.03429154e-02 3.09046537e-01 3.28925997e-01 -8.00080776e-01
-7.21832037e-01 7.04860268e-03 5.07762372e-01 6.39498830e-01
-2.49193326e-01 6.01522267e-01 -4.14189547e-01 -3.62594783e-01
-1.80452958e-01 -9.58009720e-01 -2.60508269e-01 -2.76708424e-01
-5.82963228e-02 -6.75126195e-01 7.90970147e-01 -2.96595842e-01
1.62830102e+00 -1.80081856e+00 3.75013411e-01 2.59328056e-02
1.20598689e-01 3.17410111e-01 1.61318749e-01 1.11291480e+00
-2.17410345e-02 2.26894736e-01 -2.79494584e-01 -1.62615120e-01
3.55458766e-01 -2.87917089e-02 -7.35192969e-02 2.43362755e-01
-1.34963006e-01 9.63987231e-01 -1.05377924e+00 -6.92186117e-01
2.39877135e-01 2.35292345e-01 -3.40100199e-01 3.33343625e-01
-3.15361708e-01 1.51415005e-01 -2.77438253e-01 4.93045360e-01
2.82602131e-01 -2.35398412e-01 4.01748329e-01 1.25237033e-01
-3.57521296e-01 4.17647034e-01 -1.47854376e+00 1.63207126e+00
-2.76959300e-01 5.69476604e-01 -1.18223175e-01 -8.98525178e-01
5.70971072e-01 7.21692264e-01 7.01080978e-01 -6.17926896e-01
7.09718019e-02 3.88761312e-01 2.07584456e-01 -7.61795819e-01
7.46970892e-01 -1.14393830e-01 -2.75520593e-01 8.16128671e-01
-1.06528297e-01 -2.23927528e-01 6.87168598e-01 6.18394166e-02
1.10966265e+00 9.55107063e-02 9.32026684e-01 -2.48419598e-01
7.91036963e-01 2.49599099e-01 2.37910748e-01 6.63296938e-01
-3.86081457e-01 3.03434908e-01 5.88065088e-01 -5.35785437e-01
-1.05166292e+00 -5.71212053e-01 -7.89342821e-03 1.20125675e+00
8.70334134e-02 -7.80119359e-01 -8.34163010e-01 -8.61030757e-01
-2.97310948e-01 4.72645640e-01 -4.60280895e-01 2.08536059e-01
-5.45828342e-01 -4.66944575e-01 4.99928117e-01 2.89272517e-01
1.50529280e-01 -1.20839393e+00 -1.08741987e+00 3.61756116e-01
-2.67742425e-01 -1.11037219e+00 -1.51920676e-01 5.49950488e-02
-8.54842663e-01 -1.11725211e+00 -6.30164027e-01 -6.05386853e-01
4.06016409e-01 9.15919319e-02 1.45324433e+00 -3.22841145e-02
-1.28794611e-01 7.49662578e-01 -5.42958856e-01 -4.01149541e-01
-4.39831287e-01 3.60650420e-01 6.21793866e-02 -4.74306434e-01
5.21169722e-01 -4.39097792e-01 -3.41433197e-01 4.61803406e-01
-7.34611809e-01 -2.40528509e-01 4.32394207e-01 6.91797554e-01
1.57328993e-01 2.06489146e-01 1.67521417e-01 -1.26170182e+00
1.20295131e+00 -2.41876885e-01 -1.81653202e-01 4.72771049e-01
-8.94232213e-01 3.80340554e-02 6.17575467e-01 -4.13880467e-01
-8.47453654e-01 1.48608014e-01 -9.49393436e-02 2.34181702e-01
-3.26434642e-01 6.66751802e-01 -2.95181572e-01 -3.70591804e-02
6.27896905e-01 -1.91483423e-01 -2.53518879e-01 -4.16126758e-01
7.13442683e-01 6.78306758e-01 2.03804344e-01 -5.20634472e-01
4.95558679e-01 1.47471264e-01 -1.35006249e-01 -9.44315553e-01
-5.63412488e-01 -8.99456739e-01 -9.39763486e-01 -1.03615470e-01
4.02203888e-01 -2.11620986e-01 -5.67917228e-01 3.57082129e-01
-1.29628575e+00 -1.77389219e-01 -6.85649395e-01 2.00195119e-01
-6.82094812e-01 5.66515803e-01 -3.06791484e-01 -9.73548591e-01
-4.41335112e-01 -9.76328075e-01 7.90159404e-01 -3.94901820e-03
-1.00740159e+00 -1.13583624e+00 5.27296960e-01 2.40110271e-02
5.41695118e-01 5.84905110e-02 7.90027738e-01 -1.03571045e+00
-1.52070880e-01 -5.16215146e-01 -3.92503664e-02 2.77818710e-01
2.18445182e-01 1.72223672e-02 -9.20662820e-01 -1.45687968e-01
5.86335547e-02 1.96892712e-02 1.14611365e-01 1.48634970e-01
5.74919224e-01 -4.09536839e-01 -4.49911535e-01 -1.14224426e-01
1.23989987e+00 1.46782383e-01 5.39804280e-01 2.91931510e-01
2.86403269e-01 1.07800770e+00 8.55632842e-01 2.77563512e-01
3.43623191e-01 9.35094535e-01 -1.18538462e-01 3.16088885e-01
9.38196853e-02 -1.29143685e-01 2.49937713e-01 8.93484056e-01
-1.39592066e-01 -2.46159226e-01 -1.08460760e+00 6.24885261e-01
-2.14349174e+00 -9.76730466e-01 -7.06861168e-02 2.05928636e+00
7.97832370e-01 1.56886071e-01 5.37717581e-01 5.25140226e-01
5.42958081e-01 6.54141754e-02 -8.05101916e-02 -7.13286459e-01
2.29507372e-01 -5.95473871e-02 4.17649299e-02 5.99475026e-01
-8.17714036e-01 8.46833110e-01 7.85977745e+00 6.46554828e-01
-1.02375925e+00 -3.04297395e-02 2.83416986e-01 7.22484440e-02
-6.24751747e-02 3.65911603e-01 -6.85020626e-01 3.47502381e-01
1.18797469e+00 -6.34219646e-01 3.31941515e-01 8.53105903e-01
7.66751468e-02 -4.50787209e-02 -1.62064004e+00 8.31457913e-01
1.03084475e-01 -9.39288855e-01 -2.09345847e-01 -1.27376094e-01
4.79381472e-01 -8.21511894e-02 -2.39603564e-01 3.84264529e-01
1.72037527e-01 -9.18799341e-01 7.08552837e-01 5.83557129e-01
2.72275597e-01 -3.62762034e-01 7.72667050e-01 6.17492616e-01
-1.25344217e+00 4.22731601e-02 3.68345603e-02 -5.05557001e-01
3.98143381e-01 6.20015681e-01 -8.97004247e-01 6.27621174e-01
6.98639989e-01 3.48566711e-01 -6.01075292e-01 1.19979310e+00
-1.31699145e-01 1.42667696e-01 -4.22806323e-01 -3.51023674e-01
2.81662755e-02 -1.79861650e-01 6.83461130e-01 1.58010507e+00
2.28036195e-01 5.39184455e-03 2.36467347e-01 5.78406274e-01
2.05040321e-01 5.98572850e-01 -8.25683475e-01 -3.84361416e-01
4.25430000e-01 1.19397402e+00 -8.26429725e-01 -4.00019616e-01
-3.20760518e-01 9.77889597e-01 3.69244665e-01 -6.76198527e-02
-5.55954993e-01 -5.76620281e-01 4.04555440e-01 4.02661443e-01
-8.73816535e-02 -4.92196202e-01 -3.53460848e-01 -8.15878749e-01
2.23738387e-01 -1.11309123e+00 4.86864835e-01 -4.21100885e-01
-1.13118529e+00 6.49771512e-01 3.64694089e-01 -1.22356796e+00
-7.85479188e-01 -5.69323897e-01 -4.51268464e-01 6.43618345e-01
-9.51855183e-01 -8.92656386e-01 -1.79215372e-01 3.31264853e-01
3.00598085e-01 2.98212022e-01 1.01109171e+00 2.29203001e-01
-3.08015853e-01 4.36882854e-01 -2.32097119e-01 -4.65830117e-01
7.70042002e-01 -1.44439530e+00 6.42715991e-01 8.27853560e-01
3.08437288e-01 1.10870314e+00 1.14659643e+00 -3.51983249e-01
-1.49012327e+00 -6.21132970e-01 1.78891349e+00 -8.99611831e-01
8.06306243e-01 -3.15831691e-01 -8.13862920e-01 5.83907306e-01
5.49299955e-01 -2.41479084e-01 5.22002995e-01 4.80135053e-01
-2.69509077e-01 1.78438097e-01 -1.02246547e+00 6.34210169e-01
1.23484910e+00 -8.02376568e-01 -7.65419424e-01 2.61808187e-01
6.29111111e-01 -4.45261031e-01 -7.17792928e-01 2.20903218e-01
6.15255654e-01 -9.90609288e-01 7.33098149e-01 -4.31030303e-01
-1.00161657e-01 -2.87062705e-01 1.67083651e-01 -1.02173245e+00
-1.63819656e-01 -9.52828467e-01 -2.25292698e-01 1.44549298e+00
5.47485471e-01 -5.75995386e-01 6.85702503e-01 8.96905243e-01
2.13651851e-01 -8.19762588e-01 -6.68698907e-01 -1.02006578e+00
-7.92638585e-02 -6.85506582e-01 5.03799319e-01 8.50731969e-01
6.35527492e-01 7.18886912e-01 -2.74200350e-01 -3.74981374e-01
6.61756843e-02 2.72443265e-01 1.00646794e+00 -1.46719718e+00
-1.16004512e-01 -7.45060921e-01 -4.45065320e-01 -1.02705443e+00
-1.86120138e-01 -8.97412658e-01 -1.08605810e-01 -1.68577683e+00
5.43677621e-02 -2.42116198e-01 -2.31114998e-01 4.32526588e-01
1.83983624e-01 5.11996374e-02 1.44158974e-01 3.73209089e-01
-1.02933848e+00 3.04525886e-02 7.30466127e-01 -8.98961052e-02
-5.84158659e-01 1.67181015e-01 -9.47138071e-01 7.40601659e-01
6.62677944e-01 -4.10154521e-01 -6.13697827e-01 -3.43626201e-01
8.39410663e-01 -2.32510746e-01 1.86504200e-02 -9.90374327e-01
5.86280525e-01 -1.30250648e-01 -4.42154199e-01 -3.62084448e-01
1.64701212e-02 -1.07343614e+00 2.49265209e-01 4.03523177e-01
-3.34376067e-01 4.86729294e-01 -2.87227314e-02 2.95883089e-01
-4.07429308e-01 -5.96914828e-01 4.20787960e-01 -2.12696880e-01
-7.34848320e-01 2.28209198e-01 -3.42618674e-01 1.96455836e-01
1.29868245e+00 -3.51818711e-01 7.78886303e-02 -1.91827372e-01
-7.29749501e-01 1.64425135e-01 5.60181737e-01 4.86632884e-01
9.48181301e-02 -1.05788970e+00 -5.20116687e-01 -2.28128016e-01
4.49718118e-01 -2.40781516e-01 5.02688810e-02 8.45143080e-01
-6.02088511e-01 7.34113455e-01 -2.70761270e-03 -5.17109632e-01
-1.48753226e+00 7.05765843e-01 1.62268266e-01 -6.49130464e-01
-5.75331688e-01 5.02456605e-01 -4.13646132e-01 -5.08753002e-01
4.10488993e-01 -3.79709333e-01 -4.38869387e-01 5.10822117e-01
3.32807779e-01 5.85982442e-01 2.33986616e-01 -6.01601601e-01
-4.85085607e-01 5.05173445e-01 -8.38331580e-02 -2.78212279e-01
1.18409002e+00 -3.35140973e-01 -3.31206471e-01 8.97645116e-01
1.01715040e+00 -7.94684514e-02 -5.40815234e-01 -4.23204005e-01
7.07050622e-01 -3.10271293e-01 -3.57726872e-01 -7.46417344e-01
-2.59733051e-01 7.44446516e-01 4.81110960e-01 6.95905924e-01
7.93834865e-01 -1.19271949e-01 5.55964768e-01 5.31527042e-01
6.57187164e-01 -1.33472753e+00 -2.24996760e-01 6.84451938e-01
9.38772619e-01 -9.02063906e-01 2.63605654e-01 -4.21461344e-01
-7.30571449e-01 9.43120062e-01 2.22193807e-01 1.00262307e-01
7.07493246e-01 3.38560224e-01 -4.59411554e-02 -2.55466074e-01
-7.63224900e-01 -2.28869721e-01 2.56528050e-01 5.78065097e-01
9.81873691e-01 -1.28494510e-02 -1.06673038e+00 4.66584861e-01
-3.35337907e-01 2.24189714e-01 2.08861634e-01 1.38779533e+00
-2.83650845e-01 -1.70580888e+00 -5.53184263e-02 1.95211917e-01
-2.83873558e-01 8.86460245e-02 -9.52662051e-01 8.86679769e-01
1.53545678e-01 9.91313517e-01 -3.19925070e-01 -3.02559823e-01
6.42559886e-01 3.54382128e-01 7.11964309e-01 -8.16758871e-01
-8.98953497e-01 -4.20094818e-01 3.29185307e-01 -6.49292588e-01
-5.93420446e-01 -9.45560098e-01 -8.47159445e-01 -3.84529769e-01
-3.35833043e-01 5.00309289e-01 5.72649062e-01 1.03458822e+00
3.01414520e-01 9.17858109e-02 3.27839553e-01 -7.47777879e-01
-5.44572651e-01 -9.98558223e-01 -3.78696293e-01 4.42811728e-01
9.13893506e-02 -4.20312166e-01 -3.03818911e-01 8.51688683e-02] | [11.211568832397461, 8.860475540161133] |
6ed035b3-4193-4175-b187-ac26dc7abe79 | flexible-job-classification-with-zero-shot | 2209.12678 | null | https://arxiv.org/abs/2209.12678v1 | https://arxiv.org/pdf/2209.12678v1.pdf | Flexible Job Classification with Zero-Shot Learning | Using a taxonomy to organize information requires classifying objects (documents, images, etc) with appropriate taxonomic classes. The flexible nature of zero-shot learning is appealing for this task because it allows classifiers to naturally adapt to taxonomy modifications. This work studies zero-shot multi-label document classification with fine-tuned language models under realistic taxonomy expansion scenarios in the human resource domain. Experiments show that zero-shot learning can be highly effective in this setting. When controlling for training data budget, zero-shot classifiers achieve a 12% relative increase in macro-AP when compared to a traditional multi-label classifier trained on all classes. Counterintuitively, these results suggest in some settings it would be preferable to adopt zero-shot techniques and spend resources annotating more documents with an incomplete set of classes, rather than spreading the labeling budget uniformly over all classes and using traditional classification techniques. Additional experiments demonstrate that adopting the well-known filter/re-rank decomposition from the recommender systems literature can significantly reduce the computational burden of high-performance zero-shot classifiers, empirically resulting in a 98% reduction in computational overhead for only a 2% relative decrease in performance. The evidence presented here demonstrates that zero-shot learning has the potential to significantly increase the flexibility of taxonomies and highlights directions for future research. | ['Thom Lake'] | 2022-08-30 | null | null | null | null | ['document-classification', 'job-classification', 'taxonomy-expansion'] | ['natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [ 2.63813853e-01 4.52869236e-02 -4.18364584e-01 -3.61538619e-01
-7.32883215e-01 -5.90099394e-01 5.95999956e-01 5.26206195e-01
-6.29853249e-01 4.11625922e-01 2.14780688e-01 -2.00245559e-01
-5.68190873e-01 -7.35595882e-01 7.16618747e-02 -4.50966030e-01
5.30774612e-03 6.29333913e-01 2.83652753e-01 -5.27845994e-02
5.57669222e-01 2.84036845e-01 -2.20863223e+00 4.96639520e-01
4.42025870e-01 6.92169845e-01 1.45445988e-01 6.49736226e-01
-4.25279856e-01 6.99265242e-01 -5.35855234e-01 -5.23192406e-01
3.91091794e-01 -9.03060567e-03 -9.12788391e-01 4.30804521e-01
6.86955392e-01 -2.21823290e-01 1.02932639e-01 9.31429327e-01
3.70397300e-01 8.33864987e-01 1.00793278e+00 -1.03589249e+00
-6.76873982e-01 5.82423031e-01 -5.09227276e-01 4.86743778e-01
3.88304561e-01 -2.05754519e-01 1.35497701e+00 -9.77006257e-01
6.27952456e-01 1.35996807e+00 8.14207077e-01 3.79889905e-01
-1.46937358e+00 -6.45914197e-01 2.82618523e-01 2.47708023e-01
-1.52342057e+00 -4.16262418e-01 1.39758289e-01 -7.81658292e-01
1.09274471e+00 4.92973685e-01 3.17838818e-01 8.28057408e-01
-2.26085633e-02 2.06068546e-01 1.03785706e+00 -8.33192170e-01
4.31652695e-01 5.63605130e-01 8.19031298e-01 5.02530038e-01
4.19279784e-01 -1.42276242e-01 -3.92885298e-01 -4.73022461e-01
1.77546870e-02 1.47520438e-01 -3.62725668e-02 -5.20957649e-01
-7.55484939e-01 1.09863150e+00 -5.18863648e-03 3.31881255e-01
-1.91110626e-01 -2.91814476e-01 5.94775379e-01 4.25909162e-01
7.36138523e-01 9.48457241e-01 -4.09793973e-01 -7.32898489e-02
-1.13245475e+00 1.30093172e-01 7.49766946e-01 1.10121703e+00
6.75234437e-01 -3.02640289e-01 -2.81986237e-01 1.25968206e+00
-7.28522167e-02 -2.12843552e-01 8.76689970e-01 -1.18371880e+00
-1.09493993e-01 4.24374104e-01 8.94663781e-02 -7.33516395e-01
-6.64362907e-01 -4.58761662e-01 -2.91942239e-01 1.12814076e-01
2.67819017e-01 -4.57902551e-02 -9.18127418e-01 1.36104786e+00
4.15465415e-01 4.26407941e-02 4.22280747e-03 4.13952470e-01
5.29853880e-01 3.73404205e-01 3.28021735e-01 -6.17365777e-01
1.61376643e+00 -7.63907731e-01 -5.32311141e-01 -1.01178102e-01
9.04742241e-01 -7.24215746e-01 1.24041593e+00 5.24052441e-01
-5.68556607e-01 -4.43521261e-01 -1.08902943e+00 5.83227798e-02
-7.26359606e-01 -4.83230114e-01 8.86289597e-01 1.12299168e+00
-6.34697318e-01 7.23959386e-01 -3.42282861e-01 -7.84954846e-01
1.81692824e-01 2.44642138e-01 -8.33935440e-02 -4.07905668e-01
-1.02101207e+00 9.82605100e-01 5.91514051e-01 -8.97018850e-01
-3.76032233e-01 -8.14115703e-01 -5.91873467e-01 4.78559524e-01
6.56072378e-01 -3.48845094e-01 1.41989553e+00 -6.98800325e-01
-1.19748473e+00 6.08810365e-01 2.63384134e-01 -2.46160164e-01
5.73401451e-02 1.71745062e-01 -3.39698881e-01 7.73135424e-02
2.83422172e-01 6.03350341e-01 5.21348596e-01 -9.85340536e-01
-1.12230968e+00 -1.32448465e-01 3.22320133e-01 3.61782759e-01
-9.92471695e-01 1.90755621e-01 -1.18988521e-01 -7.39364922e-01
1.46712852e-03 -9.89723921e-01 -2.08884552e-01 -1.56689256e-01
4.52583760e-01 -4.83686119e-01 4.47645366e-01 -2.64334660e-02
1.27740693e+00 -2.05622435e+00 2.73060799e-03 3.95173542e-02
1.47161469e-01 2.27064150e-03 -1.76462024e-01 2.55661547e-01
-2.15356778e-02 2.87083924e-01 2.21420258e-01 -1.35835961e-01
-5.19054942e-02 2.37895444e-01 -1.88611507e-01 3.78695399e-01
-4.09006983e-01 2.79413909e-01 -1.04750144e+00 -4.72891510e-01
1.78682864e-01 2.35168934e-01 -7.05443919e-01 -2.43815586e-01
2.53317668e-03 -4.30006683e-01 1.42111912e-01 6.73423529e-01
2.08261922e-01 -4.06228960e-01 4.31722134e-01 3.15601602e-02
-6.12067617e-02 -1.07405029e-01 -1.50346327e+00 1.43056548e+00
-6.28967285e-01 5.02347589e-01 -3.09228241e-01 -9.65418637e-01
5.09318948e-01 3.77234727e-01 6.72719717e-01 -2.48829961e-01
1.76249743e-01 -4.53935713e-02 1.07838057e-01 -1.52966499e-01
5.38138211e-01 -4.88785297e-01 -1.42236605e-01 7.01187313e-01
5.73897898e-01 -4.56735380e-02 4.80219364e-01 1.87338039e-01
1.02280807e+00 -4.94575411e-01 6.04925036e-01 -4.23429459e-01
6.43868819e-02 4.19774652e-02 3.85644883e-01 1.03615558e+00
-1.40638232e-01 2.86580592e-01 -8.90897512e-02 -3.80992770e-01
-9.67687070e-01 -6.94932759e-01 -5.14998734e-01 2.12534976e+00
-1.45819545e-01 -6.79230034e-01 -3.39254498e-01 -5.08423030e-01
2.28760749e-01 1.06865633e+00 -6.51331782e-01 -2.40302175e-01
1.42993629e-01 -9.33138669e-01 2.63479322e-01 3.40349555e-01
-1.22722447e-01 -3.96381050e-01 -7.19128251e-01 3.37565362e-01
6.96525797e-02 -6.38529539e-01 -3.26524973e-01 6.21098995e-01
-7.11151779e-01 -8.99681151e-01 -6.89021885e-01 -4.36364174e-01
3.19869637e-01 8.77656519e-01 8.38430285e-01 -1.22160144e-01
-4.13579166e-01 6.81752861e-01 -7.05381751e-01 -4.27055329e-01
-1.78003520e-01 2.48810813e-01 4.41760153e-01 -1.87054425e-01
7.65025973e-01 -4.73834366e-01 -5.57436198e-02 3.88057142e-01
-7.32230246e-01 -3.25574756e-01 1.88751295e-01 1.03150785e+00
2.68593431e-01 3.35221976e-01 7.72642732e-01 -1.19697630e+00
8.86509836e-01 -7.24812865e-01 -3.52605045e-01 5.12582898e-01
-1.35394061e+00 4.54128645e-02 3.03745121e-01 -9.51185405e-01
-1.03562355e+00 -3.43747474e-02 4.08658355e-01 -4.17597026e-01
-1.55548574e-02 5.22005737e-01 2.89699078e-01 -2.41944715e-01
1.21032929e+00 -3.92967165e-01 -2.80059874e-01 -5.82275033e-01
7.11480439e-01 1.10180640e+00 2.84214109e-01 -5.02614856e-01
4.15116042e-01 2.86491483e-01 -1.17458992e-01 -8.13309312e-01
-1.26020932e+00 -1.21696913e+00 -7.45168090e-01 -2.21434027e-01
5.39533734e-01 -9.42110837e-01 -4.57334608e-01 -4.28161263e-01
-5.40758967e-01 3.13668884e-02 -4.45138305e-01 4.91960913e-01
-4.67099488e-01 3.93041790e-01 -4.82381433e-01 -7.15781271e-01
-3.57475668e-01 -7.45310605e-01 6.48501575e-01 1.35192156e-01
-6.00352585e-01 -8.78309786e-01 -2.18209345e-03 3.63127261e-01
2.07136154e-01 -3.73691231e-01 1.09008658e+00 -1.02159226e+00
1.57032926e-02 -3.89950395e-01 -9.06030685e-02 -5.86826280e-02
4.33472753e-01 -2.79490620e-01 -1.08938408e+00 -5.77927113e-01
-1.66654103e-02 -4.47764814e-01 8.69767308e-01 1.05618685e-01
9.86404598e-01 -1.65874615e-01 -4.43120182e-01 3.12244266e-01
1.29545546e+00 3.03769559e-01 -1.32358387e-01 4.14167494e-01
4.52861637e-01 6.59871876e-01 9.42555010e-01 8.23157728e-01
1.24104433e-01 7.32837081e-01 -9.27445814e-02 5.62482417e-01
-1.13430761e-01 2.93245643e-01 -2.00278074e-01 5.65207720e-01
1.94319803e-02 -1.91396594e-01 -1.06163442e+00 4.22306299e-01
-1.67053926e+00 -1.22424233e+00 3.87965977e-01 2.31016994e+00
8.47402692e-01 2.18896404e-01 2.22156301e-01 2.43239686e-01
7.07090855e-01 -1.90064773e-01 -3.84514630e-01 -4.83870298e-01
2.41983265e-01 1.20124623e-01 5.63512385e-01 5.66017091e-01
-1.17591727e+00 7.46763587e-01 7.13538027e+00 8.89399588e-01
-7.26815939e-01 3.99039060e-01 2.35139027e-01 -6.30988538e-01
1.00021586e-02 -4.53102216e-03 -9.55876052e-01 3.72162253e-01
1.48317063e+00 -7.90059149e-01 3.76586348e-01 1.20538664e+00
-2.61058033e-01 -1.86835676e-01 -1.14736152e+00 9.95954096e-01
3.06170106e-01 -1.42008471e+00 2.45480657e-01 2.22923487e-01
6.39311075e-01 -1.62530228e-01 -1.42675593e-01 7.29346991e-01
6.18447959e-01 -6.13821626e-01 4.82576340e-01 2.63676673e-01
9.76999521e-01 -8.55495095e-01 5.08733094e-01 5.37021458e-01
-1.07582521e+00 -6.95131361e-01 -8.24040592e-01 -3.13561976e-01
-1.94422632e-01 2.97976077e-01 -9.61046159e-01 2.45361567e-01
6.16471767e-01 4.34673250e-01 -7.54805624e-01 1.17599118e+00
6.19208813e-01 3.91740471e-01 -1.85594931e-01 -6.54803887e-02
1.67264462e-01 1.89723402e-01 2.01027453e-01 1.55813420e+00
2.55028188e-01 4.80592817e-01 5.31730294e-01 -7.56772384e-02
1.68061808e-01 4.75129932e-01 -6.28273845e-01 1.46814408e-02
7.76142359e-01 1.27138138e+00 -1.07791579e+00 -7.94989347e-01
-6.50209665e-01 7.08100617e-01 2.45707139e-01 -7.18428346e-04
-3.27420115e-01 -5.64582586e-01 7.06986010e-01 -1.27323996e-02
2.07414746e-01 1.82218775e-01 -3.11290443e-01 -1.22474909e+00
-6.44864380e-01 -7.78910637e-01 8.97384703e-01 -5.33795536e-01
-1.40290511e+00 3.28274935e-01 2.99771070e-01 -1.19839752e+00
-3.57626379e-01 -5.24156272e-01 5.81236510e-03 5.91363013e-01
-8.09867024e-01 -8.78043711e-01 -2.30654091e-01 3.11093301e-01
1.01952493e+00 -3.52973968e-01 1.31974173e+00 3.76731753e-01
-3.70292068e-01 7.56751001e-01 4.71825540e-01 -3.32513303e-01
9.95571315e-01 -1.24887609e+00 -1.60898715e-01 5.42965114e-01
3.88866365e-01 7.73435533e-01 8.49507451e-01 -4.73715633e-01
-8.62833619e-01 -9.28829670e-01 6.79131091e-01 -5.03217578e-01
7.14096725e-01 -1.78310469e-01 -1.02296579e+00 6.42391622e-01
5.92831373e-02 -1.59236968e-01 1.51069033e+00 7.08795726e-01
-7.17690706e-01 2.98274141e-02 -1.23897099e+00 4.81437176e-01
9.22060132e-01 -5.35515428e-01 -7.91761994e-01 7.76113272e-01
6.64839566e-01 1.12301670e-01 -1.16592252e+00 2.06228569e-01
7.29617119e-01 -5.39168358e-01 9.69625175e-01 -7.17247069e-01
1.79126542e-02 1.24405995e-01 -5.56230247e-01 -1.40704727e+00
-1.10716939e+00 -1.49539128e-01 -1.25888526e-01 1.21546209e+00
1.42277539e-01 -3.27873558e-01 5.92838466e-01 8.18498194e-01
2.09923927e-02 -4.70675379e-01 -8.11227083e-01 -1.08581519e+00
1.95971221e-01 -2.80215710e-01 2.98177689e-01 1.31986856e+00
4.39245850e-01 6.47257090e-01 -2.25697249e-01 -3.22157517e-02
6.88632309e-01 5.46429977e-02 3.71518493e-01 -1.90263057e+00
-4.21484083e-01 -6.16566300e-01 -4.54463422e-01 -4.12459433e-01
2.93641418e-01 -1.23372483e+00 -1.73057854e-01 -1.29568875e+00
4.09988284e-01 -4.72157121e-01 -6.76300287e-01 5.37912846e-01
-9.37640518e-02 1.69213817e-01 2.74317622e-01 4.44090515e-01
-6.33460343e-01 3.78903635e-02 5.16048968e-01 -4.09290552e-01
-2.38862529e-01 4.09255177e-02 -1.03024042e+00 7.37919748e-01
4.57644075e-01 -5.37986338e-01 -7.37280309e-01 5.00356518e-02
-1.56423301e-01 -2.14571267e-01 -3.78978372e-01 -1.10974920e+00
3.47869933e-01 -4.65226114e-01 3.74831080e-01 -1.39023542e-01
5.08677542e-01 -6.43607140e-01 5.89279607e-02 3.01352173e-01
-6.08786881e-01 -2.11172596e-01 2.57287920e-01 8.26138854e-01
2.64161885e-01 -7.23854542e-01 9.49225843e-01 -4.38193917e-01
-9.46145773e-01 -4.69928235e-02 -8.58963549e-01 -1.73443720e-01
1.19666231e+00 -2.56845325e-01 -1.65305391e-01 -3.74406972e-03
-9.08104539e-01 7.10950000e-03 4.77632284e-01 5.90609372e-01
8.89118165e-02 -1.19151700e+00 -2.52925158e-01 -1.75070971e-01
6.54932201e-01 -7.14023888e-01 8.65830332e-02 2.45606586e-01
8.64511542e-03 4.85276043e-01 -3.18620563e-01 -3.15224290e-01
-1.55271184e+00 1.13883185e+00 2.23911360e-01 1.31242061e-02
-5.85659027e-01 9.61943746e-01 -5.34982495e-02 -2.70139456e-01
4.46845204e-01 6.66550621e-02 -4.75320518e-01 9.39875960e-01
7.20018446e-01 5.58825195e-01 1.66286543e-01 -2.39710018e-01
-2.91219234e-01 2.93723971e-01 -5.28263867e-01 -2.19662100e-01
1.17196882e+00 -2.01917350e-01 1.61561176e-01 8.94511282e-01
8.06388557e-01 -2.46294409e-01 -8.50068986e-01 -4.32566941e-01
5.70104182e-01 -6.33784294e-01 3.39675963e-01 -7.46129334e-01
-3.78024101e-01 6.07980013e-01 6.88468456e-01 3.60362202e-01
9.13492799e-01 -5.72483875e-02 1.68980919e-02 7.07938492e-01
7.58073509e-01 -1.34588480e+00 1.39462739e-01 3.86503011e-01
3.90764564e-01 -1.12648702e+00 4.47805345e-01 -2.80768096e-01
-4.87628132e-01 1.11843252e+00 5.34605205e-01 2.11631775e-01
8.16667736e-01 7.82318041e-02 -1.54716834e-01 -1.02400720e-01
-1.04258633e+00 -3.03910077e-01 2.26480842e-01 4.10579145e-01
4.46320415e-01 2.79247284e-01 -4.75135297e-01 4.94283974e-01
-6.46256879e-02 -2.23935217e-01 8.37801218e-01 1.01530457e+00
-1.05656815e+00 -1.04937458e+00 -3.61886919e-01 1.01354992e+00
-2.46978551e-01 -1.15178503e-01 -1.87658653e-01 4.22177374e-01
3.48898798e-01 1.33570540e+00 2.16219127e-01 -4.83422250e-01
1.36159524e-01 7.26552129e-01 3.95635158e-01 -1.51030123e+00
-4.41719323e-01 9.82994735e-02 1.09978072e-01 -2.29747102e-01
-3.23934674e-01 -9.02821898e-01 -8.61652136e-01 -2.32019901e-01
-6.41145825e-01 1.59245074e-01 3.57656956e-01 9.02844548e-01
3.21675062e-01 4.82689083e-01 3.75246912e-01 -7.87877917e-01
-8.99576068e-01 -1.18694711e+00 -6.08430982e-01 2.57941633e-01
-6.27209246e-02 -1.44116378e+00 -4.74491000e-01 -1.30898943e-02] | [10.04108715057373, 3.4288876056671143] |
2a483645-9adb-4381-8f60-e5acb153b965 | crformer-a-cross-region-transformer-for | 2207.01600 | null | https://arxiv.org/abs/2207.01600v1 | https://arxiv.org/pdf/2207.01600v1.pdf | CRFormer: A Cross-Region Transformer for Shadow Removal | Aiming to restore the original intensity of shadow regions in an image and make them compatible with the remaining non-shadow regions without a trace, shadow removal is a very challenging problem that benefits many downstream image/video-related tasks. Recently, transformers have shown their strong capability in various applications by capturing global pixel interactions and this capability is highly desirable in shadow removal. However, applying transformers to promote shadow removal is non-trivial for the following two reasons: 1) The patchify operation is not suitable for shadow removal due to irregular shadow shapes; 2) shadow removal only needs one-way interaction from the non-shadow region to the shadow region instead of the common two-way interactions among all pixels in the image. In this paper, we propose a novel cross-region transformer, namely CRFormer, for shadow removal which differs from existing transformers by only considering the pixel interactions from the non-shadow region to the shadow region without splitting images into patches. This is achieved by a carefully designed region-aware cross-attention operation that can aggregate the recovered shadow region features conditioned on the non-shadow region features. Extensive experiments on ISTD, AISTD, SRD, and Video Shadow Removal datasets demonstrate the superiority of our method compared to other state-of-the-art methods. | ['Song Wang', 'Zhihao Liu', 'Xinyi Wu', 'Zhenyao Wu', 'Hui Yin', 'Jin Wan'] | 2022-07-04 | null | null | null | null | ['shadow-removal'] | ['computer-vision'] | [ 7.04842746e-01 -1.66281268e-01 3.07117611e-01 -2.33165771e-01
-2.61427999e-01 -1.69918254e-01 4.04883415e-01 -3.94164324e-01
-3.64260077e-02 6.98138237e-01 2.67365575e-01 -3.91609579e-01
-5.21428809e-02 -6.42416060e-01 -6.44616067e-01 -1.32831824e+00
4.01371062e-01 9.24618170e-02 1.00418913e+00 -4.82917398e-01
1.46572351e-01 5.67918599e-01 -1.46871364e+00 2.28196055e-01
1.36400568e+00 9.72272575e-01 7.71643817e-01 1.64344370e-01
-2.16068387e-01 6.86356246e-01 -6.33842766e-01 -3.67816095e-03
3.63074601e-01 -5.30168474e-01 -4.09681424e-02 2.20231995e-01
3.91936183e-01 -6.40595078e-01 -3.87650907e-01 7.46639788e-01
4.26757038e-01 2.66861200e-01 5.42876959e-01 -1.18887949e+00
-5.91311157e-01 8.80115777e-02 -1.07454515e+00 1.96334437e-01
2.60653789e-03 -2.60164272e-02 5.77498019e-01 -9.23145771e-01
2.90345639e-01 1.19167268e+00 4.67929214e-01 9.96675193e-02
-7.91235387e-01 -8.49230528e-01 3.14107686e-01 4.25762922e-01
-1.16864061e+00 -1.93214893e-01 8.93109620e-01 -2.16814429e-02
4.07197475e-01 4.66015428e-01 5.16547203e-01 7.99293697e-01
4.21452820e-01 9.43588853e-01 1.49443686e+00 -2.07126439e-01
-8.56352225e-02 1.83454931e-01 3.81758437e-02 5.26806831e-01
4.72355857e-02 1.05098508e-01 -4.90521312e-01 2.26317090e-03
6.59366846e-01 2.65649498e-01 -9.56721783e-01 -4.73939657e-01
-1.00205481e+00 3.51258129e-01 5.54215252e-01 2.23729357e-01
-3.62869680e-01 -1.48910299e-01 7.90868625e-02 -3.36754210e-02
5.81117511e-01 -1.30833089e-01 -5.03467500e-01 2.83182889e-01
-8.94785345e-01 6.84510591e-03 4.35178220e-01 8.89166296e-01
8.71617019e-01 1.50311142e-01 -4.35800046e-01 8.01982284e-01
-1.25407457e-01 7.53966153e-01 1.81432962e-02 -5.38171828e-01
4.56422955e-01 5.74591935e-01 3.37704748e-01 -9.74390149e-01
-1.90888971e-01 -5.29604912e-01 -1.12873256e+00 3.78073752e-01
2.29344308e-01 1.30803779e-01 -1.08595526e+00 1.32753491e+00
5.98968089e-01 7.16135561e-01 7.70913158e-03 1.10969317e+00
9.45356846e-01 9.23434138e-01 -3.27810526e-01 -5.68154633e-01
1.24486756e+00 -1.08285761e+00 -8.51065099e-01 -2.23151445e-01
-4.16268371e-02 -8.89484227e-01 1.11690998e+00 2.83326328e-01
-6.52986586e-01 -4.88109797e-01 -1.05886602e+00 -5.07119149e-02
-2.22604409e-01 1.47963971e-01 5.47053933e-01 4.27731365e-01
-7.86790252e-01 1.72961831e-01 -2.09032878e-01 -1.59371227e-01
4.76704538e-01 1.21039756e-01 -8.36939588e-02 -3.13147992e-01
-1.03817964e+00 6.08813763e-01 1.17590234e-01 4.97098356e-01
-8.23599875e-01 -8.28554034e-01 -4.93001372e-01 2.74652660e-01
1.11118710e+00 -2.97612160e-01 7.40354538e-01 -1.20581794e+00
-1.31560051e+00 2.09285766e-01 -5.25492489e-01 -6.15013540e-02
5.80593050e-01 -2.85556048e-01 -2.76039720e-01 -7.11152628e-02
5.13574779e-02 -4.72415891e-03 1.15508366e+00 -1.82842112e+00
-7.35030949e-01 -2.60110795e-01 6.31313100e-02 6.46732330e-01
-3.14081341e-01 -1.77847564e-01 -8.99298012e-01 -8.23305309e-01
1.37464315e-01 -8.60394657e-01 2.14158706e-02 7.03516901e-02
-6.48415744e-01 6.17519058e-02 1.73095846e+00 -8.79422307e-01
1.08329141e+00 -2.18574595e+00 -5.27783893e-02 2.46396944e-01
2.66717523e-01 5.30932665e-01 -4.14482467e-02 2.69859791e-01
1.15599476e-01 -3.92785490e-01 -7.17216313e-01 -2.18834162e-01
-1.67157948e-01 2.00389326e-01 -6.69140756e-01 4.88815933e-01
-2.45680198e-01 6.36157572e-01 -8.50058436e-01 -5.69581568e-01
5.98056793e-01 7.24415362e-01 -6.55610207e-03 4.07815307e-01
-8.41512438e-03 5.75491190e-01 -5.60617983e-01 7.44669139e-01
1.30573618e+00 1.28057927e-01 -4.34618890e-02 -3.75494927e-01
-2.24340096e-01 -1.53855085e-01 -9.70122457e-01 1.05123973e+00
-5.21358073e-01 5.95408499e-01 3.08692008e-01 -6.14959121e-01
7.28701115e-01 -4.95779924e-02 5.86965978e-01 -8.80375326e-01
-9.87989735e-03 1.36241719e-01 -2.14211136e-01 -3.95453930e-01
4.82104570e-01 -5.04887737e-02 1.52554736e-01 4.03288841e-01
-6.95100129e-01 9.05175135e-03 -2.97594875e-01 2.42190331e-01
9.22275722e-01 2.40050286e-01 1.58407509e-01 -3.38289678e-01
6.81182623e-01 -3.49777222e-01 1.01735711e+00 6.83178604e-01
-1.26589552e-01 8.47674370e-01 2.51837999e-01 -3.09134852e-02
-4.86977607e-01 -1.15057969e+00 -4.32211347e-02 1.05674040e+00
8.64716887e-01 -1.73963699e-02 -7.03782380e-01 -8.34616363e-01
7.48683512e-02 7.56841958e-01 -6.87457860e-01 -1.74698651e-01
-5.33316553e-01 -7.78965294e-01 6.14206344e-02 3.98883313e-01
9.80184257e-01 -1.09523416e+00 -6.78189278e-01 1.88110322e-02
-4.63212699e-01 -1.11368930e+00 -8.57539594e-01 1.82476237e-01
-4.22970504e-01 -1.02681994e+00 -1.00551009e+00 -6.22374058e-01
7.34526634e-01 1.28159654e+00 7.36782432e-01 3.89660150e-01
-2.55507320e-01 3.08900476e-02 -5.34169972e-01 -4.83152241e-01
7.87617043e-02 -4.96440917e-01 -3.16199601e-01 5.71153879e-01
-1.73123077e-01 -6.32515967e-01 -8.76696587e-01 6.13352835e-01
-1.08343434e+00 4.31603670e-01 1.02800620e+00 8.92764568e-01
5.33051729e-01 5.70931613e-01 2.32941881e-01 -1.18824947e+00
2.14948043e-01 -4.32227135e-01 -4.85076427e-01 4.25391167e-01
-5.73863685e-01 -3.07557523e-01 6.03443861e-01 -2.04082727e-01
-1.92773342e+00 7.92757496e-02 3.70421588e-01 -5.61631322e-01
7.60077089e-02 -2.30366901e-01 -8.23270798e-01 -1.55799747e-01
8.93251300e-02 7.02012539e-01 -3.62966627e-01 -4.24561173e-01
2.28607312e-01 5.29984057e-01 5.22496223e-01 -2.42731884e-01
1.25960708e+00 9.47559416e-01 1.12749584e-01 -1.03237474e+00
-7.85452902e-01 -6.51854396e-01 -5.24975359e-01 -3.03474575e-01
7.01403260e-01 -6.28723204e-01 -5.41538656e-01 8.03242266e-01
-9.48193491e-01 -5.19913435e-01 -6.35562390e-02 -8.30144212e-02
-7.60641769e-02 8.90421927e-01 -1.39076993e-01 -8.77524853e-01
-4.31282520e-01 -1.27705300e+00 1.33679307e+00 4.47713047e-01
5.59877276e-01 -6.23434246e-01 -4.71541226e-01 5.02502084e-01
4.55290258e-01 9.07499716e-02 9.42993999e-01 2.69916002e-02
-1.11232853e+00 4.10646826e-01 -8.39992940e-01 4.68879849e-01
4.50837553e-01 -2.99835443e-01 -9.92828369e-01 -1.38248518e-01
3.50680798e-02 2.24341542e-01 1.22667444e+00 5.12649417e-01
1.26066101e+00 -1.12681426e-01 -5.44178009e-01 6.74653053e-01
1.47202933e+00 2.34412342e-01 9.73154008e-01 1.45634025e-01
1.02875471e+00 6.50014043e-01 1.15873325e+00 4.73442733e-01
2.68753409e-01 8.15607786e-01 4.77254778e-01 -7.62874603e-01
-4.62086707e-01 9.07540396e-02 4.56733435e-01 2.16482863e-01
-8.53441060e-02 -6.29891455e-01 -3.44088048e-01 4.43466783e-01
-1.88416791e+00 -7.29649663e-01 -6.02844477e-01 2.23948622e+00
5.14491737e-01 -9.38702449e-02 -4.17141467e-01 -1.79672427e-03
6.12510026e-01 5.32312036e-01 -6.20599329e-01 -5.20717390e-02
-6.06570125e-01 3.79252024e-02 7.78503299e-01 4.95786011e-01
-8.43072236e-01 1.03837931e+00 4.41798544e+00 1.28285372e+00
-8.25555503e-01 1.68118522e-01 3.92092615e-01 2.75347710e-01
-4.34314996e-01 3.68601769e-01 -5.78474164e-01 6.32007360e-01
-2.07496416e-02 3.16805810e-01 4.30145562e-01 4.20802146e-01
5.12144446e-01 -8.42878222e-01 -3.93510878e-01 8.23257327e-01
3.11835021e-01 -8.80872130e-01 1.86190680e-02 -2.95345532e-03
8.00941408e-01 -4.04632509e-01 -4.38829325e-02 9.09417048e-02
2.11732537e-02 -5.92284918e-01 6.48708344e-01 5.73513627e-01
6.69112444e-01 -6.71203256e-01 7.48667657e-01 2.29940712e-01
-1.46982026e+00 -1.16060987e-01 -2.39933312e-01 3.09421390e-01
2.07383275e-01 1.18914139e+00 -8.63357544e-01 8.78078580e-01
1.03491068e+00 6.27951801e-01 -5.28487563e-01 8.50971401e-01
-3.33288908e-01 4.52655554e-01 -1.40436426e-01 4.07791466e-01
-2.25532409e-02 -7.16752291e-01 7.36704648e-01 1.15336621e+00
2.93895364e-01 3.72203261e-01 1.64509639e-01 7.23092139e-01
1.51796624e-01 4.27909605e-02 -4.87083644e-01 6.47290230e-01
3.59094620e-01 1.28370869e+00 -9.34649885e-01 -3.68949652e-01
-4.70781267e-01 1.48806155e+00 -2.89462477e-01 6.89723194e-01
-1.23198223e+00 -5.02622724e-01 6.51379943e-01 2.04007700e-01
4.30815756e-01 4.20095064e-02 -2.30584517e-01 -8.42528760e-01
2.70579368e-01 -7.33389437e-01 2.72695180e-02 -1.11931598e+00
-9.91309106e-01 4.52094465e-01 9.93221253e-03 -1.25457931e+00
5.30887008e-01 -1.23461097e-01 -8.32203090e-01 8.34050238e-01
-2.04740691e+00 -1.45201242e+00 -9.99029696e-01 9.65770066e-01
6.19119465e-01 2.18665242e-01 1.87226087e-01 3.03959876e-01
-5.14494061e-01 3.21227252e-01 2.86643088e-01 -2.70621181e-01
9.41008925e-01 -1.09781611e+00 -5.47944382e-03 1.14348316e+00
-4.15218174e-01 2.13107705e-01 7.99297333e-01 -9.43431199e-01
-1.37018740e+00 -1.20376778e+00 5.05209029e-01 -6.74885064e-02
1.34788752e-01 -4.11077350e-01 -1.11031854e+00 3.80299240e-01
1.92083016e-01 -5.17671481e-02 1.16192810e-01 -2.82929301e-01
-1.51212826e-01 -6.29243612e-01 -8.91104221e-01 7.24010468e-01
1.17642963e+00 -2.20325038e-01 -3.03596497e-01 3.84377360e-01
5.59395313e-01 -4.56368446e-01 -1.69636607e-01 7.28209674e-01
4.17270660e-01 -1.56190932e+00 1.13782811e+00 4.65442479e-01
2.33568937e-01 -6.90807939e-01 -1.80949152e-01 -1.22214389e+00
-1.03822470e-01 -6.53793275e-01 -2.57211700e-02 1.58067191e+00
-2.46558823e-02 -9.21717465e-01 6.89963043e-01 1.72561258e-01
-6.15544260e-01 -7.16839015e-01 -8.27034891e-01 -6.05617821e-01
-5.65953672e-01 -1.27638534e-01 6.23209059e-01 5.16182125e-01
-7.27082610e-01 5.68674840e-02 -7.60150850e-01 5.26615202e-01
8.39598656e-01 7.66143262e-01 1.01936030e+00 -1.00736654e+00
-3.45659792e-01 -1.50694162e-01 1.37980774e-01 -1.30950797e+00
1.32014722e-01 -2.40034983e-01 6.26311660e-01 -1.74520910e+00
5.73516369e-01 -7.68692732e-01 -2.50639617e-01 5.49802065e-01
-5.78110874e-01 3.25475872e-01 1.86841801e-01 2.53415853e-01
-4.48246390e-01 8.91447663e-01 1.70456409e+00 -2.98450775e-02
-2.55878180e-01 2.05353439e-01 -6.61497116e-01 6.00396931e-01
4.02284861e-01 -2.89220005e-01 -7.04767883e-01 -2.51127243e-01
-5.04462302e-01 7.46167228e-02 4.97148603e-01 -8.96794856e-01
6.68798536e-02 -4.66824502e-01 3.74946892e-01 -1.00966978e+00
8.55809569e-01 -1.08042920e+00 7.59490728e-02 1.90708742e-01
3.55612487e-01 -5.54761291e-01 1.01353519e-03 1.02969277e+00
-9.96661279e-03 3.06149036e-01 7.85426676e-01 1.04924560e-01
-1.03816903e+00 2.48493105e-01 -3.18753093e-01 -2.71495849e-01
1.29066598e+00 -4.20499265e-01 -4.72262233e-01 -5.53846717e-01
-5.09444289e-02 2.11860597e-01 6.15627289e-01 3.54013681e-01
8.55791152e-01 -8.26401949e-01 -6.42433167e-01 2.73728251e-01
-7.33823627e-02 1.22156501e-01 7.60836720e-01 1.21828234e+00
-1.61724120e-01 2.36400723e-01 -5.62650710e-02 -5.45834601e-01
-1.69764185e+00 4.63748515e-01 2.50556886e-01 -1.61274984e-01
-1.10594547e+00 7.29604661e-01 1.38082933e+00 -9.09932777e-02
1.43227533e-01 -3.17350417e-01 -8.70046020e-02 -1.57088995e-01
3.43374729e-01 3.42916429e-01 7.85286948e-02 -7.89309263e-01
-2.78154790e-01 7.14424312e-01 1.05922990e-01 3.10046583e-01
1.14823925e+00 -5.23841560e-01 -2.05926403e-01 2.06780866e-01
9.40049231e-01 3.79950732e-01 -1.65718925e+00 -4.28070456e-01
-5.58413744e-01 -9.79159296e-01 3.09265345e-01 -7.33157218e-01
-1.53053904e+00 6.44184828e-01 5.79453468e-01 -1.56714186e-01
1.76105976e+00 -1.55172125e-01 1.30896592e+00 -4.07121703e-02
3.87942642e-01 -9.44272161e-01 1.62652627e-01 3.15497220e-01
9.12326217e-01 -1.11440587e+00 3.25241327e-01 -9.87301528e-01
-7.98558533e-01 8.02290916e-01 5.44600487e-01 5.14970161e-02
6.36480331e-01 3.92293781e-01 2.41405237e-02 -5.99119999e-02
-2.26733461e-01 -3.43511462e-01 2.87654489e-01 6.35868132e-01
-1.97398722e-01 1.30889434e-02 -1.45880565e-01 3.90086293e-01
3.67145032e-01 -2.73175538e-01 6.34564221e-01 9.15896833e-01
-4.87144828e-01 -9.40578341e-01 -7.96783149e-01 6.07061386e-01
-2.17015091e-02 -3.56407881e-01 -4.08082604e-01 7.75923967e-01
3.48581642e-01 1.16146207e+00 -3.58101249e-01 -3.58979821e-01
2.82882065e-01 -4.84807968e-01 2.71750212e-01 -3.55740935e-01
-2.42274299e-01 3.86712074e-01 -1.32263616e-01 -6.90748692e-01
-4.26269889e-01 -8.09826732e-01 -1.40123427e+00 -2.48642921e-01
-5.73416889e-01 -3.50996219e-02 3.72174531e-01 1.07640207e+00
2.89391935e-01 7.80925691e-01 8.03332210e-01 -1.14618683e+00
-7.16693327e-02 -7.02534676e-01 -8.89040291e-01 3.48788023e-01
5.26755154e-01 -1.01972067e+00 -4.88932103e-01 -7.15154186e-02] | [10.844086647033691, -4.086328506469727] |
9222b2d3-b914-4437-8e51-0437fb315765 | discriminative-region-based-multi-label-zero | 2108.09301 | null | https://arxiv.org/abs/2108.09301v1 | https://arxiv.org/pdf/2108.09301v1.pdf | Discriminative Region-based Multi-Label Zero-Shot Learning | Multi-label zero-shot learning (ZSL) is a more realistic counter-part of standard single-label ZSL since several objects can co-exist in a natural image. However, the occurrence of multiple objects complicates the reasoning and requires region-specific processing of visual features to preserve their contextual cues. We note that the best existing multi-label ZSL method takes a shared approach towards attending to region features with a common set of attention maps for all the classes. Such shared maps lead to diffused attention, which does not discriminatively focus on relevant locations when the number of classes are large. Moreover, mapping spatially-pooled visual features to the class semantics leads to inter-class feature entanglement, thus hampering the classification. Here, we propose an alternate approach towards region-based discriminability-preserving multi-label zero-shot classification. Our approach maintains the spatial resolution to preserve region-level characteristics and utilizes a bi-level attention module (BiAM) to enrich the features by incorporating both region and scene context information. The enriched region-level features are then mapped to the class semantics and only their class predictions are spatially pooled to obtain image-level predictions, thereby keeping the multi-class features disentangled. Our approach sets a new state of the art on two large-scale multi-label zero-shot benchmarks: NUS-WIDE and Open Images. On NUS-WIDE, our approach achieves an absolute gain of 6.9% mAP for ZSL, compared to the best published results. | ['Mubarak Shah', 'Ling Shao', 'Fahad Shahbaz Khan', 'Salman Khan', 'Akshita Gupta', 'Sanath Narayan'] | 2021-08-20 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Narayan_Discriminative_Region-Based_Multi-Label_Zero-Shot_Learning_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Narayan_Discriminative_Region-Based_Multi-Label_Zero-Shot_Learning_ICCV_2021_paper.pdf | iccv-2021-1 | ['multi-label-zero-shot-learning'] | ['computer-vision'] | [ 5.09980321e-01 -3.85787152e-02 -3.95296663e-01 -3.60447735e-01
-1.05925643e+00 -5.09552479e-01 8.19717824e-01 6.14463747e-01
-3.85386407e-01 4.74902838e-01 9.50969830e-02 2.83880174e-01
-1.46545783e-01 -7.48641491e-01 -6.38207734e-01 -8.56455922e-01
2.80588120e-01 6.90447092e-02 7.87583888e-01 -6.87642545e-02
1.97730005e-01 4.58476692e-01 -2.06149006e+00 8.34169567e-01
4.26682591e-01 1.16234291e+00 3.16882312e-01 4.25098687e-01
-1.56177387e-01 8.37103665e-01 -3.01625371e-01 -1.41257748e-01
3.11450452e-01 -3.22519690e-01 -7.82378137e-01 1.22721300e-01
1.04190528e+00 -2.72860751e-02 -1.82605252e-01 1.22968745e+00
2.55342901e-01 4.40327108e-01 9.00678575e-01 -1.26340258e+00
-7.43917048e-01 9.97741446e-02 -8.64393473e-01 1.63420677e-01
1.13602228e-01 1.54045835e-01 1.38171351e+00 -1.08406544e+00
8.62440765e-01 1.21486187e+00 4.39587593e-01 3.73693883e-01
-1.55325627e+00 -5.36049604e-01 2.99654275e-01 3.50705147e-01
-1.64305127e+00 -3.19617331e-01 7.74855554e-01 -6.48907781e-01
8.18930924e-01 4.50403810e-01 6.26845717e-01 9.58176076e-01
4.12868440e-01 5.75604320e-01 1.39837646e+00 -4.52394068e-01
3.83176506e-01 1.78575084e-01 2.20596686e-01 6.93817377e-01
6.34631887e-02 8.55724700e-03 -7.53876269e-01 -6.77694604e-02
5.35693407e-01 5.32623470e-01 -1.03730381e-01 -8.50789130e-01
-1.33691943e+00 8.73116553e-01 8.26045871e-01 4.67308789e-01
-1.85579404e-01 8.13534036e-02 3.98937434e-01 6.94994256e-02
4.79748875e-01 4.64130044e-01 -2.61321366e-01 4.48104858e-01
-1.12188232e+00 2.74775207e-01 1.09673776e-01 8.81137192e-01
1.21542513e+00 -4.15996820e-01 -7.39897132e-01 7.97918260e-01
1.42479822e-01 2.68055171e-01 3.08212072e-01 -8.68854761e-01
2.08191559e-01 8.72241139e-01 -2.28766337e-01 -1.12619495e+00
-4.02172714e-01 -4.88390714e-01 -8.64061713e-01 5.83003938e-01
3.23435575e-01 5.77070773e-01 -1.14297664e+00 1.83240175e+00
3.17150950e-01 3.01899999e-01 -1.89383596e-01 9.18312490e-01
8.97988617e-01 4.57345814e-01 3.01538676e-01 -5.82258217e-02
1.58037913e+00 -1.20499527e+00 -4.68195617e-01 -3.11259925e-01
6.28418386e-01 -4.64207411e-01 1.20342338e+00 -1.09271042e-01
-8.02788019e-01 -4.78371412e-01 -1.14105070e+00 -2.70757556e-01
-9.52915430e-01 -3.76853734e-01 3.86185378e-01 3.40110451e-01
-8.23489428e-01 5.24711192e-01 -2.57590324e-01 -3.13830346e-01
7.96234310e-01 2.20575988e-01 -6.57347381e-01 -4.30272847e-01
-1.17573690e+00 8.02226722e-01 4.33954537e-01 -5.17045915e-01
-8.58753800e-01 -8.99907947e-01 -9.66247737e-01 2.59612292e-01
5.14062226e-01 -4.27236497e-01 6.75043404e-01 -7.67035425e-01
-8.82963896e-01 1.14872372e+00 -7.37433806e-02 -4.09532934e-02
2.48419479e-01 2.76440024e-01 -3.31297755e-01 3.02487165e-01
5.24113119e-01 1.05141616e+00 7.44264841e-01 -1.43368661e+00
-8.26111019e-01 -3.77624214e-01 1.95260316e-01 2.67172426e-01
-2.28350773e-01 -1.25695169e-01 -2.95532733e-01 -7.56801307e-01
2.09689528e-01 -7.56353140e-01 -7.81313628e-02 3.65098178e-01
-2.39360273e-01 -1.19526265e-02 7.92267740e-01 -3.41704637e-02
1.10229063e+00 -2.24412012e+00 1.76567838e-01 -2.01229081e-02
5.56415737e-01 -2.56426190e-03 -2.61378318e-01 2.68078983e-01
-1.61388502e-01 1.18022621e-01 -3.53560805e-01 -3.27318162e-01
-1.03412434e-01 1.25846222e-01 -1.20260634e-01 7.25915372e-01
2.35402465e-01 1.15620410e+00 -1.17148876e+00 -7.44013548e-01
5.57946861e-01 3.34939808e-01 -5.89160502e-01 -1.56937942e-01
-1.03454497e-02 3.20593953e-01 -1.23904541e-01 7.96704650e-01
5.77807426e-01 -4.67147768e-01 -9.74781290e-02 -3.39018852e-01
-5.16433045e-02 -2.41609633e-01 -1.04823732e+00 1.89185905e+00
-3.23446989e-01 4.98133123e-01 -2.54320025e-01 -8.63951743e-01
6.74967110e-01 1.06863394e-01 6.60791874e-01 -8.00025880e-01
1.07800663e-01 6.60225227e-02 -2.91093290e-01 -1.65254623e-01
3.45805436e-01 -4.20842856e-01 -4.03299451e-01 3.67750138e-01
4.91629064e-01 2.86454540e-02 -2.62223482e-02 2.94280678e-01
1.00887132e+00 -1.07877992e-01 7.45432079e-01 -5.38985193e-01
3.08926433e-01 -2.93669283e-01 5.65091312e-01 9.39440548e-01
-4.55560595e-01 8.63360584e-01 4.89608616e-01 -4.05206740e-01
-9.78188574e-01 -1.26681042e+00 -2.64764994e-01 1.45916665e+00
6.12944067e-01 -4.05202299e-01 -4.09187526e-01 -9.65505123e-01
4.15010080e-02 6.13036752e-01 -1.11822176e+00 -4.98471946e-01
-6.00800198e-03 -4.89967704e-01 2.59828001e-01 3.29883218e-01
2.13530615e-01 -9.35839653e-01 -8.08369696e-01 6.23230226e-02
5.87092992e-03 -8.76699150e-01 -4.17900234e-01 5.27204931e-01
-2.97709227e-01 -1.06560266e+00 -7.45184958e-01 -6.49931192e-01
5.54909706e-01 6.61935329e-01 1.04005170e+00 -7.54630938e-02
-5.92489302e-01 1.08556516e-01 -3.74000877e-01 -1.42150000e-01
-3.49005461e-02 -5.71645796e-02 -2.34124810e-01 3.45048517e-01
2.25823402e-01 -5.78446984e-01 -6.41045272e-01 3.12074006e-01
-9.40388322e-01 2.68176079e-01 5.49769759e-01 1.10088646e+00
8.68412435e-01 -1.80946797e-01 7.31662154e-01 -9.20125067e-01
-1.07120890e-02 -5.89069784e-01 -2.80810654e-01 5.52053213e-01
-4.37853515e-01 -1.11930938e-02 3.42114002e-01 -4.78403211e-01
-8.79803836e-01 1.93042889e-01 3.74566048e-01 -6.93863690e-01
-2.96342760e-01 8.95654596e-03 -3.53957683e-01 -2.72360682e-01
7.05854356e-01 5.70194013e-02 -2.33722344e-01 -2.63791323e-01
7.14127779e-01 4.17395115e-01 3.94225866e-01 -4.23173219e-01
3.90269637e-01 7.85609066e-01 1.67206943e-01 -5.94244599e-01
-1.23985887e+00 -8.75819623e-01 -9.97339547e-01 -3.60123873e-01
1.16799104e+00 -8.47499549e-01 -3.37208122e-01 3.00425470e-01
-8.83081436e-01 -1.44195929e-01 -6.30052805e-01 1.49686530e-01
-7.03626990e-01 2.00377747e-01 -3.94172281e-01 -3.88831556e-01
8.91988128e-02 -1.27057314e+00 1.51144981e+00 1.13118328e-01
-1.38613865e-01 -8.12650144e-01 1.09387845e-01 1.40541866e-01
2.33061075e-01 3.50557238e-01 1.06970024e+00 -5.77118635e-01
-5.57348549e-01 -1.29772559e-01 -6.93937302e-01 -4.76170368e-02
5.88264093e-02 -4.21085089e-01 -1.33588767e+00 -3.10453832e-01
-3.08694959e-01 -5.17783344e-01 1.21139467e+00 1.98366508e-01
1.17809749e+00 4.79647070e-02 -4.34186816e-01 5.36886275e-01
1.66885126e+00 -1.92683950e-01 3.78126472e-01 1.05901659e-01
9.05815482e-01 7.03847647e-01 8.18637788e-01 3.87718946e-01
2.61989862e-01 9.08194661e-01 6.33109212e-01 -2.75254190e-01
-3.71033102e-01 -3.54543254e-02 -8.75671729e-02 3.18165600e-01
1.58254415e-01 -6.53186813e-02 -7.17838705e-01 5.84592044e-01
-1.98094523e+00 -1.11829853e+00 2.95935161e-02 2.21180892e+00
7.02101588e-01 -7.76222646e-02 -4.32031378e-02 -3.03240102e-02
8.43634844e-01 5.14120698e-01 -5.23685873e-01 -1.05959415e-01
-2.99844176e-01 1.06455080e-01 5.57499230e-01 3.56467545e-01
-1.42468929e+00 9.52309608e-01 5.80962372e+00 1.32585824e+00
-1.04043257e+00 5.91038644e-01 6.44725144e-01 -5.69542766e-01
-3.93091381e-01 3.29545862e-03 -7.32082367e-01 3.14451456e-01
5.38766801e-01 -2.04961114e-02 1.82977125e-01 7.05990076e-01
-3.64984334e-01 -2.90854782e-01 -1.09221351e+00 9.41035390e-01
2.52051800e-01 -1.50918376e+00 4.45321649e-02 2.08012640e-01
9.71689939e-01 -1.17360859e-03 4.76758890e-02 3.15852314e-01
1.79185939e-03 -9.99486804e-01 1.02562201e+00 6.33873701e-01
1.08641946e+00 -8.04941952e-01 5.20587265e-01 2.16042355e-01
-1.56826425e+00 -1.87960669e-01 -4.88573730e-01 9.78071988e-02
9.74944010e-02 4.19363558e-01 -2.78784633e-01 5.49776971e-01
5.73438704e-01 7.69859612e-01 -9.82410431e-01 8.79417539e-01
2.07934573e-01 2.11642180e-02 4.91430052e-02 2.85531282e-01
4.78025585e-01 2.27215573e-01 4.55565959e-01 1.14906597e+00
8.57904851e-02 -6.31490350e-02 4.41987753e-01 9.09471273e-01
5.45962565e-02 2.07683593e-01 -8.20601404e-01 3.22735310e-01
2.58608431e-01 1.37920177e+00 -1.09460938e+00 -5.23090899e-01
-7.21095145e-01 1.11924744e+00 6.16943955e-01 2.87021607e-01
-9.81118679e-01 -2.33310133e-01 7.34348238e-01 1.81714132e-01
3.27222168e-01 1.73939779e-01 -4.22201544e-01 -1.15950608e+00
-2.95250714e-01 -4.08783853e-01 5.71271956e-01 -6.55873775e-01
-1.31825280e+00 4.72172707e-01 7.40403458e-02 -1.29807413e+00
1.36749372e-01 -4.87851381e-01 -2.92134732e-01 7.79326379e-01
-1.58252370e+00 -1.67970860e+00 -3.04660082e-01 6.16564751e-01
6.28256440e-01 -3.98789831e-02 1.07691026e+00 4.17428344e-01
-2.90173769e-01 6.84415579e-01 7.19626471e-02 -3.11100870e-01
9.13583457e-01 -1.21811283e+00 -1.27483711e-01 6.51832044e-01
4.02198434e-01 4.10408497e-01 3.49666625e-01 -5.46501100e-01
-7.50372529e-01 -1.37558520e+00 7.66563535e-01 -4.99912679e-01
5.97722888e-01 -6.76794112e-01 -1.09029067e+00 4.21718776e-01
3.34637128e-02 8.47780883e-01 7.91084170e-01 -4.67021614e-02
-8.16601396e-01 5.67088835e-02 -1.10051370e+00 5.62741995e-01
1.18685222e+00 -8.42149317e-01 -5.28974652e-01 1.36975691e-01
8.69910479e-01 1.01591900e-01 -8.27781737e-01 4.42800760e-01
6.45751774e-01 -1.15212595e+00 1.15627694e+00 -5.00509322e-01
4.86763239e-01 -4.37550366e-01 -5.72144568e-01 -1.07036304e+00
-8.81468356e-01 8.12552646e-02 -8.86987522e-03 1.20028889e+00
1.73571289e-01 -4.74800199e-01 4.92080986e-01 3.71344477e-01
-1.29613772e-01 -9.71626341e-01 -1.09030557e+00 -7.48113632e-01
3.82755883e-02 -1.66779146e-01 3.85689020e-01 1.15458965e+00
-8.28212649e-02 3.69281650e-01 -3.82391065e-01 3.58435349e-03
6.84664369e-01 4.51935202e-01 2.06720024e-01 -1.36739361e+00
-1.30549893e-01 -6.45151377e-01 -7.39374101e-01 -4.27867085e-01
4.06002074e-01 -1.22350597e+00 5.35418727e-02 -1.29125607e+00
7.20386505e-01 -4.08943266e-01 -8.27018321e-01 7.14414597e-01
-2.73106694e-01 7.92839408e-01 4.21292990e-01 1.74432173e-01
-1.10331178e+00 5.39226234e-01 1.23857582e+00 -3.93668383e-01
8.19770768e-02 -5.59892297e-01 -6.31072879e-01 6.06351376e-01
4.05712634e-01 -6.14803612e-01 -4.09435987e-01 1.80162191e-01
-1.39224127e-01 -2.96108484e-01 7.33861685e-01 -1.14215958e+00
1.91866696e-01 -4.78421450e-01 5.01292348e-01 -5.41125536e-01
5.35547733e-01 -8.94412696e-01 1.47566702e-02 2.20418081e-01
-5.35180271e-01 -5.07304609e-01 1.75532084e-02 9.01315033e-01
-2.04452381e-01 -1.02701150e-01 1.23838675e+00 -2.09296644e-01
-1.16464114e+00 4.01266813e-01 -2.05858812e-01 4.61125225e-02
1.50381041e+00 -2.73949057e-01 -4.41335112e-01 -6.66818290e-04
-7.28718638e-01 3.02822571e-02 9.01192546e-01 5.37205160e-01
5.21951914e-01 -1.60101557e+00 -3.81146103e-01 4.29231703e-01
9.01212037e-01 -1.53181180e-01 7.88502395e-01 7.99012423e-01
5.77730238e-02 3.28592122e-01 -5.70345759e-01 -7.76923358e-01
-1.32234120e+00 1.07481873e+00 2.26019830e-01 -3.16388935e-01
-8.08936894e-01 8.76759589e-01 9.70488429e-01 -1.96023762e-01
-1.21761411e-01 -3.36081609e-02 -2.57382095e-01 4.55176562e-01
6.77650571e-01 1.85269415e-01 5.80771305e-02 -1.13284934e+00
-5.15230656e-01 7.57888734e-01 -3.81828845e-02 -6.22019991e-02
1.20218754e+00 -3.08067173e-01 -6.53693378e-02 7.97466338e-01
1.48947048e+00 -7.30659440e-02 -1.36899173e+00 -4.20502633e-01
-9.13825408e-02 -6.52479231e-01 2.74239838e-01 -6.34423018e-01
-1.01001239e+00 9.98567462e-01 6.40040219e-01 -1.09578617e-01
9.56827462e-01 4.48170125e-01 3.77235055e-01 -5.53417839e-02
5.06477475e-01 -8.99539769e-01 3.06961119e-01 3.21248293e-01
7.65007317e-01 -1.44850349e+00 -5.95611930e-02 -5.16629755e-01
-6.72152698e-01 7.58256137e-01 6.49445057e-01 -1.91896945e-01
6.26195133e-01 2.63508577e-02 -2.35640109e-01 -4.98237997e-01
-6.82971358e-01 -5.18609405e-01 6.86292291e-01 5.35213649e-01
9.22873467e-02 1.73004374e-01 3.28584798e-02 6.46356881e-01
5.01966715e-01 -4.68842149e-01 1.48304552e-01 8.66046846e-01
-6.94362223e-01 -7.81560302e-01 -2.60868818e-01 5.52632570e-01
-1.57241300e-01 -2.17513159e-01 -2.40014702e-01 6.84645057e-01
5.01744986e-01 6.56820655e-01 1.52830720e-01 -3.44298691e-01
1.52770400e-01 2.16036230e-01 4.49571878e-01 -9.74559188e-01
-4.38206583e-01 6.84407074e-03 -3.67269784e-01 -8.49056900e-01
-3.73123467e-01 -5.17824113e-01 -1.16685390e+00 1.16157062e-01
-2.36637726e-01 -3.94843757e-01 2.95253932e-01 8.32467437e-01
4.05442804e-01 7.27490485e-01 4.22158033e-01 -1.08421731e+00
-1.48192614e-01 -6.02703571e-01 -8.77052665e-01 8.00322831e-01
3.92598897e-01 -1.27640927e+00 -2.42449895e-01 -1.32727802e-01] | [9.846894264221191, 2.506770372390747] |
eea1a8d8-e558-4b25-9ba0-bc8c568e9f99 | imagination-is-all-you-need-curved | 2211.07591 | null | https://arxiv.org/abs/2211.07591v2 | https://arxiv.org/pdf/2211.07591v2.pdf | Imagination is All You Need! Curved Contrastive Learning for Abstract Sequence Modeling Utilized on Long Short-Term Dialogue Planning | Inspired by the curvature of space-time (Einstein, 1921), we introduce Curved Contrastive Learning (CCL), a novel representation learning technique for learning the relative turn distance between utterance pairs in multi-turn dialogues. The resulting bi-encoder models can guide transformers as a response ranking model towards a goal in a zero-shot fashion by projecting the goal utterance and the corresponding reply candidates into a latent space. Here the cosine similarity indicates the distance/reachability of a candidate utterance toward the corresponding goal. Furthermore, we explore how these forward-entailing language representations can be utilized for assessing the likelihood of sequences by the entailment strength i.e. through the cosine similarity of its individual members (encoded separately) as an emergent property in the curved space. These non-local properties allow us to imagine the likelihood of future patterns in dialogues, specifically by ordering/identifying future goal utterances that are multiple turns away, given a dialogue context. As part of our analysis, we investigate characteristics that make conversations (un)plannable and find strong evidence of planning capability over multiple turns (in 61.56% over 3 turns) in conversations from the DailyDialog (Li et al., 2017) dataset. Finally, we show how we achieve higher efficiency in sequence modeling tasks compared to previous work thanks to our relativistic approach, where only the last utterance needs to be encoded and computed during inference. | ['Gerasimos Spanakis', 'Stefan Schaffer', 'Justus-Jonas Erker'] | 2022-11-14 | null | null | null | null | ['sentence-embeddings', 'sentence-embeddings'] | ['methodology', 'natural-language-processing'] | [ 1.52118713e-01 7.32304752e-01 -8.35154504e-02 -5.86866260e-01
-8.90787125e-01 -9.18555021e-01 1.25723505e+00 1.60383329e-01
-2.30663568e-01 4.66911793e-01 1.21162581e+00 -4.74719524e-01
-2.78460622e-01 -6.09638870e-01 -4.76626694e-01 -6.71336412e-01
-4.41081226e-01 6.11499071e-01 -4.18854058e-01 -7.03234255e-01
3.93787891e-01 2.06945077e-01 -1.13741136e+00 6.22770429e-01
5.06651878e-01 4.35363084e-01 2.59516418e-01 9.43569779e-01
7.22408220e-02 9.57524896e-01 -3.52895260e-01 -3.38688254e-01
-8.88197199e-02 -6.55425608e-01 -1.45518064e+00 3.86327989e-02
-1.47710601e-02 -1.93424702e-01 -4.32021499e-01 4.92953837e-01
2.21380129e-01 5.91161430e-01 8.72421682e-01 -7.52006173e-01
-3.68796229e-01 9.19471323e-01 2.42507666e-01 3.54348689e-01
8.80773127e-01 1.86512604e-01 1.46996987e+00 -7.53167272e-01
7.92873025e-01 1.52415168e+00 3.53519022e-01 5.04853785e-01
-1.23829699e+00 6.99179918e-02 1.82697758e-01 4.03749824e-01
-7.60432363e-01 -6.14479542e-01 8.76229823e-01 -5.15880287e-01
1.15025759e+00 5.36220849e-01 4.96407688e-01 1.30378616e+00
7.60545060e-02 8.46357763e-01 9.10937428e-01 -4.00105327e-01
1.08284533e-01 -1.16287060e-01 1.83277801e-01 6.58433914e-01
-9.30601954e-01 3.63390408e-02 -6.56508029e-01 -1.85769126e-01
3.45677704e-01 -3.73791218e-01 -1.93548799e-01 3.41677889e-02
-1.37258971e+00 1.04545116e+00 5.33122718e-01 5.56605935e-01
-4.69568938e-01 -9.82978866e-02 3.94070715e-01 4.97994125e-01
4.82760906e-01 6.30018651e-01 -5.31301081e-01 -6.87658310e-01
-4.88676220e-01 3.88705254e-01 8.75454187e-01 6.16681337e-01
7.36621499e-01 -3.97460341e-01 -4.30681318e-01 6.80328012e-01
3.43696892e-01 -2.31231958e-01 5.18512964e-01 -1.16227078e+00
5.83297312e-01 4.83858675e-01 9.89070684e-02 -1.01523638e+00
-7.52240658e-01 -1.57738045e-01 -4.63048249e-01 -5.32579780e-01
6.52007997e-01 -2.48481959e-01 -1.53713878e-02 2.19771767e+00
2.02361897e-01 6.96493313e-02 3.05372477e-01 7.39557922e-01
5.30305862e-01 8.06163788e-01 -2.28099361e-01 -5.07977307e-01
1.35791183e+00 -6.64350927e-01 -5.74712813e-01 3.68142640e-03
1.19278753e+00 -5.60039461e-01 1.27334940e+00 1.98747829e-01
-1.08572531e+00 -3.99939746e-01 -6.37861431e-01 -1.51789784e-01
-9.00136083e-02 -2.26729438e-01 6.25934780e-01 2.41847262e-01
-9.63465214e-01 8.05949032e-01 -6.09956741e-01 -3.63607287e-01
-2.43471876e-01 -2.55926661e-02 -2.11430058e-01 3.62985164e-01
-1.40579677e+00 1.10445344e+00 4.23975848e-02 -9.82024241e-03
-7.10342765e-01 -3.70726973e-01 -8.87441814e-01 5.58132641e-02
7.14249089e-02 -2.75167435e-01 1.30717528e+00 -5.24771333e-01
-1.88533604e+00 6.51892543e-01 -4.41182017e-01 -5.70571840e-01
3.34504247e-01 -2.67657894e-03 -1.97785601e-01 1.43566221e-01
-1.68392345e-01 5.25622845e-01 4.67152506e-01 -8.50869596e-01
-1.51195049e-01 -2.83193856e-01 5.39784491e-01 5.33644259e-01
5.76931462e-02 2.57942639e-02 1.88890442e-01 -1.43752351e-01
3.15280229e-01 -1.07406056e+00 -4.09086905e-02 -6.06803894e-01
-4.67446059e-01 -9.02588964e-01 3.01941246e-01 -6.85422301e-01
1.23103821e+00 -1.88748169e+00 6.80944800e-01 -9.37090255e-03
1.04823217e-01 -2.64196843e-01 -1.76927060e-01 9.83879507e-01
-2.67944913e-02 1.05013311e-01 -3.09929013e-01 -3.86606455e-01
1.99703217e-01 1.18998490e-01 -5.84974170e-01 7.42552996e-01
8.42761695e-02 9.00763869e-01 -1.17127287e+00 -1.38555259e-01
3.12585801e-01 2.93623090e-01 -7.60703623e-01 4.97742891e-01
-3.67430836e-01 9.45949256e-01 -5.76499067e-02 -4.47207168e-02
3.49458940e-02 -1.99292958e-01 6.69416070e-01 2.96541192e-02
-3.33208501e-01 1.28584135e+00 -5.59729755e-01 1.71332526e+00
-8.51013303e-01 8.68299782e-01 -2.93101817e-01 -1.00423372e+00
1.00071895e+00 5.20431638e-01 2.09007621e-01 -6.02587163e-01
-4.83045727e-02 -6.97309971e-02 4.05674398e-01 -7.25088656e-01
6.52722776e-01 -2.72806019e-01 -4.48792726e-01 8.47141623e-01
5.52736819e-02 -1.31863415e-01 9.68106836e-02 4.22850341e-01
8.95769358e-01 1.39794707e-01 2.14890003e-01 -2.31991544e-01
6.25069857e-01 -5.32003164e-01 2.22296461e-01 5.05363762e-01
-6.12416454e-02 4.10955191e-01 1.01070774e+00 -4.74204272e-01
-9.05938387e-01 -8.43514800e-01 -1.01353399e-01 1.43878436e+00
-1.93712473e-01 -6.97514653e-01 -5.39859116e-01 -4.14147228e-01
-4.71707404e-01 1.17680943e+00 -6.99799299e-01 -2.45323390e-01
-1.01865256e+00 -3.46771836e-01 5.82185030e-01 1.55422091e-01
-3.97302657e-02 -1.27776861e+00 -5.94302714e-01 2.35958844e-01
-8.61341357e-01 -8.97354960e-01 -4.37166393e-01 7.15021193e-02
-6.60174727e-01 -8.12562943e-01 -2.56281734e-01 -3.93825322e-01
1.98582724e-01 -9.27839503e-02 1.01137531e+00 3.42142805e-02
2.12735385e-01 4.64999765e-01 -4.30870414e-01 2.57251620e-01
-6.97741866e-01 -1.74759477e-01 3.24069887e-01 1.26850322e-01
7.66165704e-02 -5.87747395e-01 -4.80775863e-01 1.45006523e-01
-3.23302656e-01 1.55221120e-01 -4.97457851e-03 8.79439592e-01
-7.73741603e-02 -6.69982374e-01 5.33045292e-01 -4.69064802e-01
1.02885568e+00 -9.57117438e-01 -2.05646321e-01 2.51693040e-01
-1.83833241e-01 3.88291270e-01 5.88453233e-01 -2.20545560e-01
-1.02746499e+00 -4.75369990e-01 -2.21242055e-01 3.27235729e-01
-7.76207894e-02 5.81379950e-01 2.37960070e-01 6.11571908e-01
6.27083778e-01 3.72469097e-01 1.37700319e-01 -3.94183218e-01
7.75624394e-01 6.69817746e-01 1.73396796e-01 -7.14915574e-01
4.36964706e-02 1.49863899e-01 -1.78661212e-01 -1.03864920e+00
-7.30997741e-01 -5.12964249e-01 -7.85252035e-01 -4.36553419e-01
7.41173625e-01 -5.48705757e-01 -1.40572584e+00 -8.43765587e-02
-1.34121585e+00 -4.97176379e-01 6.74404502e-02 6.21306777e-01
-1.13610017e+00 5.66282630e-01 -7.51503289e-01 -1.20738912e+00
-5.17420247e-02 -1.10573077e+00 8.36410642e-01 -2.33446825e-02
-9.97244835e-01 -1.23245263e+00 2.36623526e-01 3.63068074e-01
1.95682049e-01 -7.85846710e-02 1.14426124e+00 -8.18719745e-01
-3.84379774e-01 2.37894997e-01 2.92202085e-01 -2.24261031e-01
-2.37801950e-02 -1.55182794e-01 -8.57046723e-01 -1.47525817e-01
2.85266012e-01 -5.28798878e-01 6.54247940e-01 1.93406865e-01
8.67106318e-01 -8.77125084e-01 4.77214456e-02 2.99566656e-01
7.11204290e-01 -3.97855788e-03 5.33650756e-01 8.32448006e-02
2.89218366e-01 1.22817290e+00 5.89145720e-01 6.72670722e-01
6.91875815e-01 9.17845905e-01 3.15848589e-01 5.75158477e-01
2.31318250e-01 -4.58612472e-01 6.51636899e-01 1.22349930e+00
-6.88672438e-02 -3.36977601e-01 -8.29212129e-01 5.45485497e-01
-1.82628143e+00 -1.32873690e+00 -1.53034970e-01 2.09021306e+00
9.88720000e-01 -5.67290746e-02 2.44409487e-01 -1.80908054e-01
4.03791130e-01 4.76779342e-01 -1.62073091e-01 -8.18149567e-01
1.16195358e-01 -2.55575906e-02 -1.80607438e-01 1.06622505e+00
-6.24553621e-01 8.89534295e-01 6.24882126e+00 3.15331221e-01
-9.10128951e-01 -1.93380881e-02 6.11937940e-01 -7.55426055e-03
-7.44005144e-01 1.93947971e-01 -4.08014446e-01 4.55319643e-01
1.40290201e+00 -1.33624583e-01 7.31564999e-01 4.25886840e-01
4.44765180e-01 6.58478588e-02 -1.35707438e+00 6.58957124e-01
-1.21208392e-01 -1.52569532e+00 -3.58161986e-01 5.55907423e-03
2.96563059e-01 3.83944660e-02 -7.97571708e-03 5.28207362e-01
3.36464733e-01 -1.05449283e+00 6.52673185e-01 8.14079583e-01
4.39397186e-01 -5.59665740e-01 4.32932794e-01 8.05430532e-01
-8.86994064e-01 -1.83106005e-01 -1.73295096e-01 -5.32429218e-01
3.81773293e-01 7.70696476e-02 -1.45871353e+00 3.31212789e-01
1.22814380e-01 5.53214967e-01 -6.55497685e-02 2.40218982e-01
-3.03098768e-01 8.74559224e-01 -1.73890769e-01 -3.82611543e-01
4.47216362e-01 -5.54418147e-01 8.39663446e-01 1.23065376e+00
1.27084568e-01 2.28599712e-01 -1.10267233e-02 9.90858257e-01
4.54960279e-02 4.92449403e-02 -6.78601980e-01 -8.31862539e-02
5.65083385e-01 1.11150920e+00 -3.37583899e-01 -1.33695155e-01
-1.63718145e-02 7.37160504e-01 6.98023856e-01 1.82092056e-01
-5.42715430e-01 9.01425853e-02 8.05090725e-01 -3.14898908e-01
-7.92683512e-02 -4.30375636e-01 -8.90982226e-02 -9.38098729e-01
-1.07572757e-01 -7.57976174e-01 1.23241358e-01 -5.52660167e-01
-1.04680920e+00 6.23326063e-01 9.01161600e-03 -1.05465746e+00
-9.80157256e-01 -1.75721183e-01 -7.90432692e-01 8.77431929e-01
-1.06288421e+00 -8.25375378e-01 2.46891558e-01 3.22916508e-01
8.24588597e-01 -5.77444956e-02 1.12922943e+00 -4.68246043e-01
-2.42047697e-01 4.22231674e-01 -1.24543086e-01 6.13981448e-02
3.60880882e-01 -1.28704619e+00 5.95957637e-01 4.22508299e-01
4.24823105e-01 9.92680311e-01 8.73960435e-01 -3.52293521e-01
-1.25171590e+00 -4.04599428e-01 1.50017035e+00 -6.60887837e-01
7.70697892e-01 -5.35341799e-01 -9.46242094e-01 7.56732762e-01
3.18213403e-01 -4.02472705e-01 6.82462692e-01 6.77643120e-01
-3.67253661e-01 4.57921624e-01 -8.65801752e-01 8.51169169e-01
1.07625413e+00 -1.04219651e+00 -1.03707111e+00 5.81525266e-01
1.03585768e+00 -2.65297711e-01 -7.81457901e-01 2.14104608e-01
5.32725275e-01 -1.25197375e+00 7.82421827e-01 -1.01967347e+00
5.61649382e-01 2.53770322e-01 -3.84537220e-01 -1.29187202e+00
-3.94462764e-01 -1.09740937e+00 -1.67393968e-01 9.94407237e-01
4.06065464e-01 -3.54815930e-01 4.08304781e-01 4.72238183e-01
-2.97530472e-01 -9.89350319e-01 -1.13845563e+00 -3.06822032e-01
3.06971014e-01 -5.69824100e-01 5.60638666e-01 8.44155490e-01
8.59193087e-01 5.64548969e-01 -4.95466471e-01 -1.67530939e-01
-4.58634738e-03 2.18381777e-01 5.31337678e-01 -9.60142970e-01
-4.19253021e-01 -4.96641338e-01 -8.98839235e-02 -1.56771159e+00
4.16600943e-01 -1.24094641e+00 2.62713313e-01 -1.25026178e+00
-1.59264222e-01 -3.95152181e-01 1.52947530e-01 -2.38473210e-02
-1.68971717e-02 -4.96943027e-01 3.34704548e-01 2.87410796e-01
-5.37285686e-01 7.68626988e-01 1.04100311e+00 1.15382798e-01
-2.14788377e-01 1.29454464e-01 -4.99007106e-01 7.44692445e-01
8.27248454e-01 -2.01262668e-01 -2.79655457e-01 -1.13817111e-01
3.48311186e-01 8.47408593e-01 2.31094167e-01 -3.49379867e-01
8.50105137e-02 -1.90792918e-01 -3.00081104e-01 -5.36171377e-01
6.23836040e-01 -2.64909506e-01 -3.68659705e-01 3.65593553e-01
-1.17747307e+00 5.50767295e-02 -3.42356056e-01 5.79445958e-01
1.84389558e-02 -2.58610070e-01 4.08067346e-01 -2.10154504e-01
-4.15389419e-01 -1.60118014e-01 -7.68758118e-01 1.23848498e-01
6.35710895e-01 2.84604013e-01 -1.59432828e-01 -9.54961240e-01
-8.51753116e-01 4.84258495e-02 4.87390123e-02 4.65538085e-01
4.99496460e-01 -1.09017837e+00 -8.36539924e-01 2.99195610e-02
-6.77677104e-04 -4.29358184e-01 3.74687672e-01 1.09548891e+00
-2.12035757e-02 6.86951220e-01 1.35120591e-02 -5.83456039e-01
-1.09258282e+00 3.03676486e-01 3.36863786e-01 -3.81964982e-01
-6.56889737e-01 1.15580952e+00 1.15063548e-01 -9.35226202e-01
1.12080425e-01 -2.66384542e-01 -4.90759075e-01 3.53884399e-01
4.35529351e-01 4.25774753e-01 -1.41764164e-01 -8.61435711e-01
-3.07357013e-01 2.24202141e-01 1.31719317e-02 -5.27807236e-01
1.27742565e+00 -4.00438607e-01 -1.98447913e-01 1.01419199e+00
1.50344265e+00 1.15425691e-01 -1.25325501e+00 -3.37821543e-01
2.78083473e-01 -1.79963768e-01 -5.37025094e-01 -4.85373795e-01
-1.23846620e-01 9.02113020e-01 4.20393422e-02 6.26278639e-01
3.59432638e-01 5.47509789e-01 5.49373150e-01 5.29599071e-01
2.76934236e-01 -8.82056177e-01 3.84980142e-01 1.10119569e+00
1.24010789e+00 -1.11755002e+00 -3.95339757e-01 2.28310050e-03
-9.26752865e-01 1.39825046e+00 6.81779236e-02 -4.87545095e-02
3.37405860e-01 -2.32688978e-01 -1.65359765e-01 -2.83681631e-01
-1.32120788e+00 -1.34335428e-01 2.44364217e-01 3.39289993e-01
9.47987735e-01 4.44703221e-01 -4.04179543e-01 3.04999173e-01
-8.48408878e-01 -7.09800959e-01 7.11456180e-01 3.29572827e-01
-4.15911704e-01 -9.11206305e-01 1.08275628e-02 2.10896209e-01
1.36136357e-02 -5.11104353e-02 -3.98900211e-01 2.89925307e-01
-3.40495616e-01 1.37048101e+00 3.45899165e-01 -5.60372710e-01
-4.09042127e-02 4.84711558e-01 4.15597856e-01 -7.49286830e-01
-6.50403321e-01 -2.91497231e-01 5.63608170e-01 -6.13720596e-01
-3.24869722e-01 -1.00664961e+00 -1.42842150e+00 -2.24927172e-01
-2.16258794e-01 2.80295908e-01 5.40101767e-01 1.40439606e+00
2.98152089e-01 4.01541322e-01 9.85882401e-01 -8.02461088e-01
-8.49683940e-01 -1.38559175e+00 -4.81388085e-02 5.14268160e-01
5.19358873e-01 -5.47819972e-01 -5.79747617e-01 -2.09425911e-01] | [12.727585792541504, 7.959446907043457] |
4cc49707-e7a3-47aa-902a-1903da53555a | emergency-action-termination-for-immediate | 2211.06351 | null | https://arxiv.org/abs/2211.06351v1 | https://arxiv.org/pdf/2211.06351v1.pdf | Emergency action termination for immediate reaction in hierarchical reinforcement learning | Hierarchical decomposition of control is unavoidable in large dynamical systems. In reinforcement learning (RL), it is usually solved with subgoals defined at higher policy levels and achieved at lower policy levels. Reaching these goals can take a substantial amount of time, during which it is not verified whether they are still worth pursuing. However, due to the randomness of the environment, these goals may become obsolete. In this paper, we address this gap in the state-of-the-art approaches and propose a method in which the validity of higher-level actions (thus lower-level goals) is constantly verified at the higher level. If the actions, i.e. lower level goals, become inadequate, they are replaced by more appropriate ones. This way we combine the advantages of hierarchical RL, which is fast training, and flat RL, which is immediate reactivity. We study our approach experimentally on seven benchmark environments. | ['Tomasz Trzciński', 'Artur Grudkowski', 'Mateusz Ostaszewski', 'Paweł Wawrzyński', 'Jakub Łyskawa', 'Michał Bortkiewicz'] | 2022-11-11 | null | null | null | null | ['hierarchical-reinforcement-learning'] | ['methodology'] | [ 5.13405129e-02 6.29525632e-02 -1.34815797e-01 1.95575163e-01
-2.18174294e-01 -6.74390376e-01 5.42587459e-01 5.37126660e-01
-6.20709658e-01 1.50234330e+00 -1.16198100e-01 -2.72015721e-01
-3.44483078e-01 -9.51683283e-01 -6.58800781e-01 -1.02608216e+00
-2.34294057e-01 5.97887695e-01 6.97096467e-01 -4.54062343e-01
2.76864380e-01 4.62964207e-01 -1.78334749e+00 -1.16195511e-02
8.14065039e-01 7.82915831e-01 1.28682747e-01 5.47449589e-01
1.88514832e-02 9.23824012e-01 -6.56139553e-01 3.30447972e-01
2.36646429e-01 -6.33840978e-01 -8.65225673e-01 1.74726203e-01
-2.46190935e-01 -2.92396694e-01 7.46911243e-02 1.09996653e+00
1.52242213e-01 4.33691084e-01 5.69428355e-02 -1.16420937e+00
8.33136737e-02 6.72187984e-01 -2.51003116e-01 -5.66412136e-02
1.77983493e-01 3.30616176e-01 9.30605352e-01 -8.49138852e-03
5.42394638e-01 1.22264826e+00 1.07348122e-01 6.93736434e-01
-1.37783158e+00 -4.09904122e-01 7.51825333e-01 9.24920142e-02
-8.93324494e-01 -2.31501237e-01 5.92167199e-01 -2.40128547e-01
7.29899883e-01 1.70773745e-01 8.32548559e-01 6.67580605e-01
3.23017806e-01 3.71766657e-01 1.64294291e+00 -3.84908259e-01
7.38902211e-01 2.77478597e-03 -2.41182968e-01 5.66539645e-01
3.74970794e-01 5.53289115e-01 -2.45070472e-01 -1.19588561e-01
6.83839381e-01 -1.61840484e-01 -1.64765179e-01 -6.28903627e-01
-8.89901280e-01 6.82271183e-01 3.12304437e-01 6.26299083e-01
-8.99387300e-01 1.88823864e-01 2.25359887e-01 5.36695600e-01
-4.08362970e-02 7.70758092e-01 -3.89326364e-01 -3.63020837e-01
-7.13581264e-01 3.38370770e-01 6.67701542e-01 3.36824179e-01
7.89186299e-01 1.26422092e-01 -1.41107649e-01 3.15687954e-01
-2.09401160e-01 6.18757727e-03 3.34058076e-01 -1.01115346e+00
1.34799123e-01 7.64905214e-01 6.62007093e-01 -7.41216123e-01
-4.08858806e-01 -6.11676097e-01 -6.50431097e-01 8.62473905e-01
5.42472899e-01 -3.81965339e-01 -7.44314551e-01 1.91758680e+00
5.71017504e-01 2.55366772e-01 4.29672569e-01 8.10043812e-01
1.34876054e-02 7.19318509e-01 7.10593387e-02 -7.41746008e-01
8.47228646e-01 -7.95262516e-01 -6.57384634e-01 -2.54484192e-02
5.32713711e-01 -3.91526818e-01 8.39176059e-01 5.96333921e-01
-1.19465888e+00 -6.67061031e-01 -8.91810238e-01 7.92650640e-01
-2.03911379e-01 -1.85892746e-01 5.15487671e-01 3.07959765e-01
-1.01092231e+00 1.05907035e+00 -7.70124316e-01 -1.28839552e-01
-2.33868673e-01 4.99499857e-01 -1.70763895e-01 2.53000528e-01
-1.36266840e+00 9.18149531e-01 6.74042583e-01 9.01810005e-02
-1.38196540e+00 -1.38314858e-01 -3.90666872e-01 6.94812313e-02
1.09633482e+00 -1.59927592e-01 1.35761952e+00 -9.19718742e-01
-1.83438206e+00 1.83871120e-01 2.09181607e-01 -4.92929369e-01
6.63292050e-01 6.07127026e-02 6.55945316e-02 -1.51397467e-01
-1.37882292e-01 4.14951563e-01 1.03350401e+00 -1.54177332e+00
-1.12320566e+00 -2.15947449e-01 7.99378574e-01 4.46972251e-01
-3.54438573e-01 -4.09688801e-01 -2.42674351e-02 9.16068405e-02
1.84187844e-01 -1.17370379e+00 -6.65001631e-01 -6.02810860e-01
-1.54262662e-01 -4.08294588e-01 5.42690754e-01 -2.56913215e-01
1.32183230e+00 -1.94063234e+00 5.79713643e-01 2.85979837e-01
2.36626387e-01 2.99266636e-01 8.98330063e-02 5.05472898e-01
1.10709794e-01 3.51206549e-02 -3.65073793e-02 5.15538119e-02
-9.34782773e-02 4.91868794e-01 -9.44493413e-02 2.17258006e-01
-8.75923112e-02 4.36951995e-01 -1.20377421e+00 -2.92631060e-01
3.01855266e-01 -1.71636492e-01 -5.85554719e-01 2.94407308e-01
-5.19989610e-01 8.38155091e-01 -8.53403091e-01 2.56421655e-01
2.60053396e-01 3.35103758e-02 5.38917720e-01 1.23314828e-01
-6.73317552e-01 4.36115950e-01 -1.38010180e+00 1.06124234e+00
-5.38880110e-01 -1.00194462e-01 1.86448082e-01 -1.30459642e+00
9.12539482e-01 3.78789991e-01 6.70835018e-01 -5.00615716e-01
-2.48773284e-02 1.92625746e-01 3.81316960e-01 -2.96496600e-01
3.39105606e-01 -1.02865085e-01 -1.90625072e-01 3.18518639e-01
-3.81733894e-01 -2.26568341e-01 6.98405027e-01 -1.90144077e-01
1.14212418e+00 6.21220052e-01 5.40939271e-01 -2.18204871e-01
9.10415590e-01 1.23762995e-01 7.63327837e-01 1.01981843e+00
-2.57344931e-01 -3.05461198e-01 9.51740742e-01 -4.01995331e-01
-8.31091285e-01 -7.28443980e-01 2.64103025e-01 1.08209479e+00
1.88906088e-01 -4.39999491e-01 -7.13017762e-01 -7.39158273e-01
-1.52397044e-02 5.63866973e-01 -7.72858679e-01 -1.44798130e-01
-7.44910121e-01 -3.87529403e-01 6.77770004e-02 2.04141602e-01
6.86836481e-01 -1.33134687e+00 -1.49291337e+00 6.81152284e-01
-5.90895340e-02 -7.67733395e-01 1.54471144e-01 4.58876044e-01
-9.78069246e-01 -1.02768350e+00 -3.86674911e-01 -3.23840618e-01
6.73922479e-01 1.27610177e-01 9.38206196e-01 3.87818277e-01
2.54614621e-01 6.49577081e-02 -4.61627603e-01 -1.13259321e-02
-7.48245656e-01 7.75182098e-02 2.28455618e-01 -1.26872271e-01
-2.78077006e-01 -5.39925098e-01 -2.74540246e-01 2.52969027e-01
-8.40306282e-01 -6.07022271e-02 6.13617361e-01 8.63853693e-01
6.78476095e-01 7.31851578e-01 6.28383696e-01 -6.40692651e-01
8.62052500e-01 -8.40846151e-02 -1.14597428e+00 2.63109207e-01
-4.77295667e-01 4.50036168e-01 1.05879557e+00 -6.11641884e-01
-9.02762115e-01 -2.73286421e-02 2.57227500e-03 -2.64574647e-01
-2.10119635e-01 4.87512499e-01 -2.07642838e-02 4.18770202e-02
4.65811372e-01 2.85274625e-01 -4.05268669e-02 -2.58050531e-01
-3.24683711e-02 2.52407283e-01 9.47553366e-02 -8.44820797e-01
6.80454612e-01 5.04382253e-02 3.10214043e-01 -5.68569720e-01
-6.89169049e-01 -1.67103097e-01 -3.61368239e-01 -6.57571435e-01
5.38449705e-01 -3.16910177e-01 -9.75201845e-01 2.19397351e-01
-7.62958884e-01 -6.16149545e-01 -4.38841134e-01 2.86090821e-01
-6.09425306e-01 2.43318945e-01 -2.31280774e-01 -1.07349241e+00
1.45660713e-01 -1.46136212e+00 6.00520432e-01 3.39605570e-01
-1.07988112e-01 -7.77270913e-01 5.32042794e-02 -2.65599698e-01
2.86275566e-01 4.50170994e-01 8.92809331e-01 -4.04082388e-01
-4.80033576e-01 1.52495667e-01 5.33410251e-01 2.64112979e-01
3.92108619e-01 -8.49234834e-02 -4.29467976e-01 -6.15434706e-01
1.26219332e-01 -4.54246014e-01 5.49981177e-01 2.27989107e-01
9.55796659e-01 -4.24623460e-01 -2.89000005e-01 -3.74339223e-02
1.42946064e+00 6.58142924e-01 5.46974182e-01 6.98925734e-01
7.64127970e-02 6.91867650e-01 9.85657096e-01 6.00206375e-01
1.37761896e-02 7.79512465e-01 8.08707356e-01 1.01966038e-01
3.90714020e-01 -5.15784919e-02 4.83281374e-01 3.54816318e-01
-4.27296042e-01 -1.49358243e-01 -5.75175107e-01 2.89515495e-01
-2.07390404e+00 -1.04689586e+00 1.56135514e-01 2.54368806e+00
9.81158912e-01 6.81770802e-01 2.83954710e-01 3.70279133e-01
6.09506547e-01 2.48872921e-01 -8.36348891e-01 -4.63138133e-01
2.44155779e-01 -2.50326078e-02 2.26723790e-01 6.25383794e-01
-9.15214539e-01 1.13386190e+00 5.31678343e+00 5.31939626e-01
-1.18875110e+00 -5.58502600e-02 3.18779886e-01 9.05215740e-02
-5.20401523e-02 2.62752295e-01 -7.73288190e-01 4.17673409e-01
6.80221975e-01 -1.91999480e-01 1.00167572e+00 8.69751394e-01
3.62357318e-01 -4.78067487e-01 -1.10357106e+00 2.81073689e-01
-3.92844498e-01 -9.76561487e-01 -3.09249461e-01 6.86572120e-02
7.23454893e-01 -4.74596500e-01 -2.22434551e-01 6.65590167e-01
6.27908766e-01 -7.09030390e-01 6.25272453e-01 3.10910016e-01
2.00258628e-01 -9.88601625e-01 6.46732807e-01 8.98940206e-01
-1.08503902e+00 -4.25720394e-01 -4.48968858e-01 -2.37666652e-01
-9.81188565e-02 1.99156284e-01 -5.00616491e-01 7.70754993e-01
4.17446434e-01 2.30456859e-01 -1.26474068e-01 7.98571706e-01
-6.84851408e-01 4.25857753e-01 -3.50804657e-01 -4.06302512e-01
7.78851986e-01 -3.81556839e-01 5.37813246e-01 4.31678325e-01
1.58345997e-01 1.04773492e-01 6.30334020e-01 6.64697587e-01
3.57117683e-01 -1.14533760e-01 -5.36120534e-01 7.15145022e-02
3.07029635e-01 1.17575037e+00 -8.26348424e-01 -5.35527885e-01
-6.31685928e-02 5.83401322e-01 4.49212134e-01 2.43368581e-01
-9.18892801e-01 1.02813631e-01 6.20570660e-01 -1.54810950e-01
1.98165402e-01 -3.24724376e-01 2.17726335e-01 -9.24829781e-01
-1.62751243e-01 -1.20001268e+00 3.22107017e-01 -2.58818090e-01
-6.23482466e-01 6.57833219e-01 1.07578531e-01 -1.29476094e+00
-6.16168797e-01 -1.98409259e-01 -2.60904223e-01 5.10288179e-01
-1.49655557e+00 -3.72391731e-01 -7.97662437e-02 6.89038157e-01
5.41780353e-01 1.15104303e-01 7.91818202e-01 -1.85375154e-01
-3.17814052e-01 5.51126935e-02 1.17333822e-01 -4.74023968e-01
1.41108975e-01 -1.37445104e+00 -1.89881623e-01 8.07730913e-01
-1.74074724e-01 3.95084590e-01 1.07301760e+00 -7.90446877e-01
-1.11420107e+00 -7.57242858e-01 6.72425747e-01 1.42706633e-01
7.43426263e-01 -6.15790486e-02 -9.11617160e-01 2.50037432e-01
4.55524288e-02 -4.50999588e-01 1.17766686e-01 -7.81919714e-03
2.86434114e-01 -2.43762001e-01 -1.05427301e+00 8.54976356e-01
7.81142771e-01 6.40753955e-02 -5.30352235e-01 3.28211397e-01
4.11015570e-01 -3.34020495e-01 -7.58973300e-01 5.79639554e-01
4.10187125e-01 -1.10648286e+00 6.14919543e-01 -7.92075872e-01
2.47521192e-01 -4.72630233e-01 1.51219189e-01 -1.71722209e+00
-3.29896092e-01 -7.35678554e-01 -1.37651309e-01 7.40748763e-01
1.36540502e-01 -7.64782488e-01 6.30713761e-01 2.89630324e-01
1.08182907e-01 -8.52113843e-01 -1.09856951e+00 -1.10951924e+00
9.34055075e-02 9.83548760e-02 7.17604697e-01 7.28877187e-01
1.69302240e-01 3.16895872e-01 -5.82228243e-01 2.55974412e-01
5.62844574e-01 5.93634248e-01 6.40050232e-01 -1.17964542e+00
-5.62904835e-01 -4.59893435e-01 -2.88443454e-02 -8.50709319e-01
2.62982041e-01 -2.53110617e-01 2.80355871e-01 -1.54372263e+00
-2.89540499e-01 -6.07303679e-01 -6.06682956e-01 6.23522460e-01
-1.79772824e-01 -3.29132915e-01 3.33432466e-01 1.43899843e-01
-6.65919304e-01 5.42026222e-01 1.40987360e+00 1.22026250e-01
-7.43080080e-01 2.28840798e-01 -2.05461845e-01 7.52424657e-01
1.37538791e+00 -5.28396606e-01 -5.28637648e-01 1.16746739e-01
4.77861129e-02 5.32419324e-01 9.72627103e-02 -1.28709292e+00
1.24077983e-01 -7.00930834e-01 -1.64163895e-02 -3.40931475e-01
2.94090450e-01 -9.54220653e-01 1.44068405e-01 1.16609240e+00
-5.99683225e-01 1.83211297e-01 3.33520658e-02 3.42029274e-01
2.62807263e-03 -4.92164791e-01 8.74616206e-01 -4.26589847e-01
-5.59909582e-01 2.60367077e-02 -6.45473003e-01 -1.65068775e-01
1.42786896e+00 -1.29240751e-01 -2.93084830e-02 -4.44338471e-01
-9.53362763e-01 4.78852957e-01 3.59154165e-01 2.31194079e-01
4.44583863e-01 -8.68009031e-01 -4.86622393e-01 -2.34440994e-02
-2.62643903e-01 -2.57087737e-01 -1.36235744e-01 6.81174755e-01
1.99236255e-02 4.13285166e-01 -4.60834473e-01 -3.71981353e-01
-1.13131940e+00 9.12395298e-01 5.32237232e-01 -8.59367728e-01
-5.47531307e-01 1.94218650e-01 -4.41887602e-02 -1.49443150e-01
2.10203722e-01 -4.32015628e-01 -5.88996470e-01 1.19577786e-02
2.76642531e-01 2.47715369e-01 -8.59419927e-02 -2.00175315e-01
-1.54822469e-01 4.83689755e-01 8.39568228e-02 -2.50777692e-01
1.29582989e+00 -5.35183363e-02 -1.71929479e-01 4.14717227e-01
1.87773824e-01 -6.67831255e-03 -1.48156178e+00 6.24144152e-02
8.17776173e-02 -4.64119315e-01 1.19303510e-01 -6.88772857e-01
-7.84243166e-01 5.26962996e-01 3.52167219e-01 7.42543221e-01
1.27086449e+00 -3.34347606e-01 2.68180400e-01 6.43315196e-01
1.01682401e+00 -1.37817204e+00 1.52669013e-01 6.92530632e-01
7.20691860e-01 -7.24013388e-01 -3.98582630e-02 -8.29653591e-02
-5.17108381e-01 1.03493237e+00 8.67844939e-01 -1.50654256e-01
1.09839169e-02 2.75101155e-01 -4.06714231e-01 1.83193371e-01
-1.08296168e+00 -6.28238857e-01 -3.33275497e-01 3.46600592e-01
-9.32494625e-02 3.32197435e-02 -7.34952688e-01 -7.48343840e-02
8.45902041e-02 -7.59578543e-04 6.63880885e-01 1.16936612e+00
-8.44037056e-01 -1.39086843e+00 -5.36581039e-01 2.44170040e-01
-2.97005862e-01 1.89183310e-01 -1.65401176e-01 1.05794108e+00
2.94179887e-01 1.10777032e+00 -2.35846281e-01 -1.09046452e-01
4.93025333e-01 -7.79368281e-02 5.99705756e-01 -4.94523406e-01
-5.95781565e-01 1.73422784e-01 3.11448365e-01 -6.30598843e-01
-4.06542450e-01 -7.27417052e-01 -1.47339404e+00 -1.41909733e-01
-6.57205582e-02 6.24438763e-01 3.72692287e-01 9.14910257e-01
-7.67334551e-02 8.78612220e-01 8.10861886e-01 -6.18878722e-01
-1.11297202e+00 -5.50498009e-01 -3.49893630e-01 2.06984609e-01
4.42251563e-01 -1.09074962e+00 -3.59379679e-01 -2.91961104e-01] | [4.329390525817871, 2.0075623989105225] |
15d3b112-26be-4c6e-8500-9d05f2e74280 | improving-feature-generalizability-with | 2204.12915 | null | https://arxiv.org/abs/2204.12915v1 | https://arxiv.org/pdf/2204.12915v1.pdf | Improving Feature Generalizability with Multitask Learning in Class Incremental Learning | Many deep learning applications, like keyword spotting, require the incorporation of new concepts (classes) over time, referred to as Class Incremental Learning (CIL). The major challenge in CIL is catastrophic forgetting, i.e., preserving as much of the old knowledge as possible while learning new tasks. Various techniques, such as regularization, knowledge distillation, and the use of exemplars, have been proposed to resolve this issue. However, prior works primarily focus on the incremental learning step, while ignoring the optimization during the base model training. We hypothesize that a more transferable and generalizable feature representation from the base model would be beneficial to incremental learning. In this work, we adopt multitask learning during base model training to improve the feature generalizability. Specifically, instead of training a single model with all the base classes, we decompose the base classes into multiple subsets and regard each of them as a task. These tasks are trained concurrently and a shared feature extractor is obtained for incremental learning. We evaluate our approach on two datasets under various configurations. The results show that our approach enhances the average incremental learning accuracy by up to 5.5%, which enables more reliable and accurate keyword spotting over time. Moreover, the proposed approach can be combined with many existing techniques and provides additional performance gain. | ['Cecilia Mascolo', 'Chi Ian Tang', 'Dong Ma'] | 2022-04-26 | null | null | null | null | ['keyword-spotting'] | ['speech'] | [ 2.09508866e-01 -1.67664006e-01 -3.23441118e-01 -2.47416988e-01
-6.68274283e-01 -4.40666378e-01 5.33126593e-01 3.29633266e-01
-6.55172884e-01 9.63282824e-01 -5.05036004e-02 -1.89911142e-01
-4.79956977e-02 -7.10476279e-01 -8.54878843e-01 -6.38821661e-01
2.11173803e-01 2.82866329e-01 4.64043647e-01 7.57211447e-02
2.30940357e-01 3.97710562e-01 -1.62644243e+00 3.80215764e-01
1.22376871e+00 9.25778449e-01 4.96571571e-01 2.36287355e-01
-3.33024353e-01 6.75472498e-01 -6.13609135e-01 -3.03702325e-01
6.02675565e-02 -6.55988231e-02 -7.23086953e-01 7.51522109e-02
3.74458671e-01 -3.51692140e-01 -3.79509896e-01 8.56775284e-01
4.89736736e-01 3.59579176e-01 3.25184911e-01 -1.14274788e+00
-6.07512474e-01 7.60615885e-01 -4.55853999e-01 1.00894116e-01
1.07818328e-01 4.82564792e-02 7.86437929e-01 -1.41629422e+00
3.91713887e-01 9.58645165e-01 6.47358954e-01 4.78955865e-01
-1.19579434e+00 -1.05167639e+00 6.25613928e-01 5.69838583e-01
-1.54704130e+00 -6.03466213e-01 9.66410995e-01 -1.92095548e-01
8.43919694e-01 7.20442310e-02 6.15748703e-01 9.14321780e-01
-5.34006171e-02 1.10150361e+00 8.78877938e-01 -5.47467530e-01
2.35406354e-01 4.43987131e-01 3.10548365e-01 5.49274206e-01
4.40559804e-01 -2.06195846e-01 -7.16245711e-01 -1.44967809e-01
3.04530174e-01 5.15049994e-01 -2.67681301e-01 -3.55824322e-01
-1.09060872e+00 5.44319630e-01 3.18199724e-01 3.06920052e-01
-3.08044493e-01 5.41548361e-04 4.85492826e-01 3.91282916e-01
4.61580038e-01 3.63680571e-01 -7.37688482e-01 1.01627763e-02
-9.14268255e-01 2.02311859e-01 5.90563774e-01 9.56779122e-01
1.18324482e+00 8.00726935e-02 -2.55398273e-01 1.03444004e+00
-1.48705512e-01 3.04365724e-01 8.83017361e-01 -4.82113063e-01
5.23531914e-01 6.93825603e-01 -3.23666930e-02 -9.03122962e-01
-1.86074972e-01 -8.60977888e-01 -6.23117685e-01 -1.93043396e-01
-2.40693968e-02 -1.17778040e-01 -1.13051784e+00 1.85847437e+00
3.74918729e-01 5.03297508e-01 5.89686260e-02 3.50335389e-01
6.08073592e-01 5.89676023e-01 9.71262828e-02 -3.99845481e-01
9.62213337e-01 -1.09814656e+00 -6.28491580e-01 -3.28432828e-01
5.86546123e-01 -6.84874535e-01 1.27064192e+00 5.62650621e-01
-6.82793677e-01 -6.14316642e-01 -1.10158074e+00 1.15889549e-01
-4.59756404e-01 3.48220393e-02 5.65770805e-01 4.06389177e-01
-6.30497456e-01 4.49022919e-01 -8.01567078e-01 -8.75361040e-02
4.78543878e-01 4.05756325e-01 -2.94245929e-01 -2.79935598e-01
-1.19704115e+00 6.76330090e-01 9.59217370e-01 -5.98613620e-02
-1.08929408e+00 -8.87422144e-01 -5.18569410e-01 1.98318049e-01
7.59916663e-01 -5.22383511e-01 1.30946541e+00 -8.61565888e-01
-1.26308584e+00 3.37594360e-01 -3.28319311e-01 -5.06837547e-01
2.36896843e-01 -6.22538865e-01 -2.64723748e-01 -2.42427826e-01
-5.85750081e-02 4.17670965e-01 1.05686486e+00 -1.31972551e+00
-8.88063431e-01 -2.19123453e-01 2.34679773e-01 4.20374393e-01
-1.00579536e+00 -3.98077190e-01 -7.32103765e-01 -8.42413962e-01
7.93520063e-02 -9.81387854e-01 9.61671472e-02 -2.85921842e-01
-3.30962777e-01 -2.25526184e-01 9.45927024e-01 -6.53126359e-01
1.62128425e+00 -2.18026161e+00 1.88211352e-01 -3.49805579e-02
4.10365343e-01 6.04905546e-01 -6.47695735e-02 2.19116747e-01
7.27287307e-02 -7.60490522e-02 -2.52850026e-01 -4.60003257e-01
-2.96367973e-01 2.13362813e-01 -3.97516668e-01 7.62708560e-02
-8.13941509e-02 9.69109058e-01 -8.82905185e-01 -2.20979482e-01
-3.24695446e-02 2.75131762e-01 -5.95033586e-01 1.61369145e-01
-3.16773981e-01 2.32437372e-01 -2.83604145e-01 5.96581757e-01
7.46303678e-01 -1.40159816e-01 1.17596984e-01 -1.21802039e-01
5.22853471e-02 2.24774450e-01 -1.18144023e+00 1.67267203e+00
-7.27811158e-01 3.99860740e-01 -3.56822819e-01 -1.23411071e+00
9.08639729e-01 1.98765814e-01 3.28568995e-01 -5.99081099e-01
-2.96089709e-01 2.40388781e-01 9.10768658e-03 -3.43837172e-01
5.30792356e-01 -2.84338873e-02 -1.72121208e-02 5.39128482e-01
1.92521423e-01 2.99351186e-01 5.03605865e-02 1.81492105e-01
9.16437507e-01 -7.58590922e-02 3.88571531e-01 1.42451867e-01
4.15778190e-01 -2.36876681e-01 9.06655371e-01 8.91880929e-01
1.28011733e-01 1.81641847e-01 1.56758947e-03 -5.59119225e-01
-7.85832942e-01 -9.82326865e-01 1.49568647e-01 1.34743750e+00
6.71573132e-02 -5.87091863e-01 -3.42504144e-01 -9.71382856e-01
3.08146954e-01 8.63011897e-01 -4.42024529e-01 -8.71038258e-01
-6.36870027e-01 -8.68600845e-01 3.34337920e-01 4.97056782e-01
6.02410436e-01 -9.39834774e-01 -4.88058150e-01 3.79372597e-01
-1.10712618e-01 -8.52701724e-01 -4.58647162e-01 3.65966827e-01
-1.08618486e+00 -9.39270496e-01 -4.89541382e-01 -7.18409538e-01
6.86029375e-01 5.87783635e-01 7.63212800e-01 1.73252001e-01
-1.74847350e-01 3.20503622e-01 -4.65550840e-01 -3.96771550e-01
-8.17180276e-02 5.37599206e-01 3.29407245e-01 2.60296077e-01
2.85182595e-01 -6.80140197e-01 -4.88206297e-01 3.59705091e-02
-9.80267048e-01 1.64704755e-01 8.87047529e-01 1.13675070e+00
5.92389643e-01 2.50008047e-01 9.36052442e-01 -1.28830481e+00
6.45004988e-01 -5.98311424e-01 -2.18920857e-01 5.55492520e-01
-9.92665350e-01 1.87305406e-01 7.47883976e-01 -9.08761501e-01
-1.24431932e+00 9.05004591e-02 1.16135523e-01 -5.83440840e-01
1.59291610e-01 8.01400244e-01 -2.93001086e-01 -2.65368998e-01
5.46762645e-01 5.65221846e-01 -2.91547775e-01 -7.30590343e-01
4.51947004e-01 6.51401997e-01 3.76367092e-01 -7.43836582e-01
8.75577092e-01 1.62435979e-01 -4.84795958e-01 -6.51084244e-01
-1.01192153e+00 -3.31676602e-01 -6.32501185e-01 1.79771464e-02
-4.31124028e-03 -9.08448040e-01 -2.34182552e-01 6.68422818e-01
-9.40799117e-01 -2.19138727e-01 -3.93857598e-01 4.29164827e-01
-1.68793723e-01 2.65750468e-01 -1.93578839e-01 -6.72722816e-01
-4.13244575e-01 -9.64018047e-01 6.57731771e-01 3.20802808e-01
1.18948206e-01 -8.06986153e-01 -4.15810011e-02 -1.53691843e-02
6.85506403e-01 -1.40594646e-01 1.10046017e+00 -9.93748724e-01
-5.70715368e-01 -2.33077869e-01 -7.11345226e-02 4.05759126e-01
5.71451724e-01 -4.83436823e-01 -1.04271495e+00 -6.77901626e-01
7.10169459e-03 -4.56902921e-01 1.27568650e+00 -1.57041639e-01
1.44271433e+00 -4.35046583e-01 -5.85918427e-01 5.53818405e-01
1.19192636e+00 5.01762629e-01 3.31195742e-01 3.32847476e-01
7.84615397e-01 1.99668810e-01 6.27101660e-01 4.98056471e-01
3.69386822e-01 6.54015064e-01 1.29261062e-01 1.27462208e-01
-3.90972823e-01 -3.90965670e-01 1.70020550e-01 9.73407865e-01
2.29709744e-01 3.31462435e-02 -9.79717255e-01 6.08835697e-01
-1.92840457e+00 -6.06127858e-01 8.06808591e-01 2.42206264e+00
1.23041260e+00 3.21617961e-01 -2.40726560e-01 2.37798378e-01
6.53466940e-01 -4.77604344e-02 -1.02400458e+00 1.09259211e-01
4.93358634e-02 2.61468321e-01 1.50408983e-01 5.52940249e-01
-1.04914153e+00 1.29692769e+00 5.75613976e+00 9.76736367e-01
-1.43166459e+00 2.34088838e-01 5.80699503e-01 -2.67060906e-01
-3.23411465e-01 2.46736676e-01 -1.13785255e+00 4.76787984e-01
6.31966054e-01 -6.41948998e-01 5.12148559e-01 9.22198534e-01
-1.39570415e-01 7.73354322e-02 -1.00222754e+00 9.61006641e-01
2.86478281e-01 -1.08114278e+00 4.61729735e-01 -3.66473883e-01
7.87061989e-01 -3.23821664e-01 1.54314131e-01 8.25283766e-01
1.11928582e-02 -6.08933866e-01 5.40762067e-01 7.32769728e-01
8.62764478e-01 -9.13537443e-01 4.91496474e-01 5.52489758e-01
-1.22973061e+00 -3.73544961e-01 -3.12137127e-01 2.63080746e-02
-2.33474791e-01 8.61206770e-01 -1.00584054e+00 5.83186150e-01
5.92482448e-01 7.08376527e-01 -9.21163380e-01 1.29571950e+00
-1.41583666e-01 7.18510270e-01 -3.60400915e-01 1.37706473e-01
-2.10376501e-01 3.26071024e-01 4.33116555e-01 9.84511197e-01
2.66965240e-01 -2.08898813e-01 5.14502227e-01 3.96964908e-01
-4.48615134e-01 1.58723593e-01 -4.48825479e-01 -9.16286837e-04
8.74817252e-01 9.95032132e-01 -4.21624035e-01 -5.99362791e-01
-5.87734818e-01 1.11176050e+00 7.68256664e-01 4.96489197e-01
-6.20462477e-01 -6.70327961e-01 4.73259211e-01 -6.44098744e-02
3.79278779e-01 -2.32341781e-01 -1.67789221e-01 -1.40469265e+00
2.23633751e-01 -8.41606200e-01 4.26247478e-01 -3.41015965e-01
-1.01796710e+00 5.46383679e-01 1.47881940e-01 -9.89262104e-01
-1.60093859e-01 -2.67270714e-01 -4.41992611e-01 5.83724976e-01
-1.63329208e+00 -1.10827804e+00 -4.45985675e-01 6.40006900e-01
8.22511256e-01 -4.02146339e-01 6.89837039e-01 4.64820296e-01
-6.36607587e-01 1.08311498e+00 2.53827870e-01 -2.60918528e-01
8.79499257e-01 -1.00547063e+00 2.28138179e-01 8.02758992e-01
2.85471648e-01 9.07571554e-01 4.09154028e-01 -7.33613193e-01
-1.26558197e+00 -1.38633859e+00 8.19491327e-01 -9.63260904e-02
3.91363084e-01 -6.15965843e-01 -1.27484155e+00 7.88670599e-01
-1.00743398e-01 -5.16017005e-02 5.48112810e-01 2.84845978e-01
-4.48717207e-01 -5.64595282e-01 -8.26163411e-01 5.27069330e-01
1.02271497e+00 -5.78621387e-01 -7.74345160e-01 2.00477853e-01
1.16269088e+00 -3.42875659e-01 -4.31587726e-01 5.29041171e-01
5.71406841e-01 -4.74781901e-01 8.71417463e-01 -7.32442796e-01
-1.96153730e-01 -2.76076555e-01 -2.87663937e-03 -1.43218458e+00
-3.04386824e-01 -3.85133833e-01 -6.49456978e-01 1.26546991e+00
5.19402504e-01 -7.57940829e-01 7.60747135e-01 3.56661886e-01
-1.54525906e-01 -9.25989568e-01 -9.11703467e-01 -8.72348070e-01
-1.59306657e-02 -4.23931926e-01 6.87473238e-01 1.06000960e+00
-2.00442940e-01 4.82822388e-01 -6.16955996e-01 -6.23984970e-02
3.85344833e-01 2.40464702e-01 8.16301763e-01 -1.37742996e+00
-3.52624685e-01 -8.21108446e-02 -8.19925144e-02 -1.18397486e+00
1.90257818e-01 -1.09504485e+00 -4.64657061e-02 -1.31652212e+00
3.98473233e-01 -7.32164502e-01 -8.83370936e-01 9.33648944e-01
-6.05223358e-01 -6.93880916e-02 2.43184865e-01 3.95510823e-01
-7.67691016e-01 7.45003343e-01 9.06049550e-01 -3.26623231e-01
-4.90057170e-01 1.66947246e-01 -8.79714787e-01 5.78860462e-01
8.27570677e-01 -6.07558250e-01 -6.19019210e-01 -5.49597502e-01
3.32182050e-02 -3.90358835e-01 3.90284806e-02 -1.13019955e+00
6.23168826e-01 -1.20080583e-01 4.28271919e-01 -3.83509964e-01
3.32443863e-01 -7.69780934e-01 -1.15633383e-02 5.01547396e-01
-2.59319484e-01 -1.18651278e-01 4.95393813e-01 7.00423539e-01
-2.28208452e-01 -2.76689291e-01 5.48709333e-01 -1.82346493e-01
-9.35603738e-01 4.65911120e-01 2.00817119e-02 -8.91122818e-02
1.10175717e+00 2.38745119e-02 -2.17322126e-01 -1.04592994e-01
-9.02262926e-01 2.71484941e-01 2.70078450e-01 6.84581161e-01
8.62531841e-01 -1.40232754e+00 -5.87021589e-01 3.24476451e-01
3.37948084e-01 9.99188721e-02 1.51115596e-01 5.68975866e-01
3.27470601e-02 4.38782245e-01 3.66772451e-02 -3.71037513e-01
-1.10675097e+00 7.93249667e-01 1.73138797e-01 -3.25299591e-01
-5.80121636e-01 7.61234403e-01 1.84032127e-01 -3.70740443e-01
5.58499634e-01 -1.10040218e-01 -1.85984299e-01 1.76953301e-01
5.42555273e-01 2.43536800e-01 2.68507838e-01 -1.42067358e-01
-3.03640217e-01 4.21809912e-01 -8.76833498e-01 1.62809312e-01
1.16412687e+00 -1.17970534e-01 -1.35996090e-02 7.07663000e-01
1.04644918e+00 -3.09771225e-02 -1.08999395e+00 -9.01260316e-01
1.95422396e-01 -5.15510678e-01 4.60916497e-02 -8.91724348e-01
-9.82502341e-01 7.46809065e-01 6.16847634e-01 -2.42814302e-01
1.33772027e+00 -1.72490433e-01 8.83536279e-01 8.91535878e-01
5.21921813e-01 -1.12671292e+00 2.72239119e-01 6.26766622e-01
7.71901309e-01 -1.10091257e+00 5.95672056e-02 -1.59262136e-01
-3.95266682e-01 9.39824402e-01 8.20946455e-01 1.18684396e-01
6.46820962e-01 4.02835123e-02 -3.68581831e-01 1.88241020e-01
-1.02708411e+00 -8.65013152e-02 2.72897601e-01 2.44521201e-01
2.28848532e-01 -8.68958160e-02 -3.25608730e-01 8.38385403e-01
7.74064511e-02 1.73670068e-01 1.42107800e-01 1.21611404e+00
-6.17575586e-01 -1.21333838e+00 -5.83479851e-02 7.02053070e-01
-1.80839553e-01 -3.14104378e-01 -2.35712066e-01 5.44148982e-01
3.55564922e-01 5.74269712e-01 -1.85607508e-01 -6.80267632e-01
4.92751211e-01 5.32592177e-01 2.82398313e-01 -9.83975291e-01
-3.03133935e-01 -2.17643023e-01 -1.76972941e-01 -2.06736133e-01
-8.26281146e-04 -4.55206752e-01 -1.05171716e+00 -1.01778090e-01
-5.96776485e-01 3.19479793e-01 4.26733911e-01 1.00713098e+00
5.37928820e-01 5.83105505e-01 7.50930071e-01 -5.43779314e-01
-5.43623865e-01 -1.03548896e+00 -2.61292309e-01 2.19594970e-01
4.45535034e-01 -9.42434192e-01 -3.46692622e-01 6.95675388e-02] | [9.759343147277832, 3.5003929138183594] |
cdc340e9-1b7d-40bd-8827-ed46b488418f | 3d-consistent-robust-segmentation-of-cardiac | 1804.09400 | null | http://arxiv.org/abs/1804.09400v1 | http://arxiv.org/pdf/1804.09400v1.pdf | 3D Consistent & Robust Segmentation of Cardiac Images by Deep Learning with Spatial Propagation | We propose a method based on deep learning to perform cardiac segmentation on
short axis MRI image stacks iteratively from the top slice (around the base) to
the bottom slice (around the apex). At each iteration, a novel variant of U-net
is applied to propagate the segmentation of a slice to the adjacent slice below
it. In other words, the prediction of a segmentation of a slice is dependent
upon the already existing segmentation of an adjacent slice. 3D-consistency is
hence explicitly enforced. The method is trained on a large database of 3078
cases from UK Biobank. It is then tested on 756 different cases from UK Biobank
and three other state-of-the-art cohorts (ACDC with 100 cases, Sunnybrook with
30 cases, RVSC with 16 cases). Results comparable or even better than the
state-of-the-art in terms of distance measures are achieved. They also
emphasize the assets of our method, namely enhanced spatial consistency
(currently neither considered nor achieved by the state-of-the-art), and the
generalization ability to unseen cases even from other databases. | ['Nicholas Ayache', 'Hervé Delingette', 'Qiao Zheng', 'Nicolas Duchateau'] | 2018-04-25 | null | null | null | null | ['cardiac-segmentation'] | ['medical'] | [ 2.72847921e-01 4.02005434e-01 1.24134436e-01 -4.32788223e-01
-7.56231904e-01 -4.58347142e-01 4.25436109e-01 5.32894969e-01
-6.85894489e-01 8.99133384e-01 2.03760520e-01 -1.89144999e-01
-3.69024813e-01 -6.35594845e-01 -6.03300512e-01 -7.74051070e-01
-4.82023388e-01 8.93765271e-01 7.88789272e-01 1.11592129e-01
5.43069951e-02 6.85355961e-01 -7.15412438e-01 3.28567982e-01
4.45182353e-01 9.44757402e-01 1.15977719e-01 6.46852076e-01
2.92759955e-01 3.75671864e-01 -2.21738800e-01 -3.42015803e-01
1.41989753e-01 -3.08653027e-01 -1.09376562e+00 7.85654634e-02
4.16498035e-01 -2.13527560e-01 -1.40756577e-01 6.71623647e-01
5.72516561e-01 -9.88790393e-02 5.78796923e-01 -4.69790578e-01
-2.99478531e-01 4.65422422e-01 -4.88891602e-01 6.34624481e-01
-2.36652553e-01 7.88216665e-02 7.22285748e-01 -4.90226001e-01
1.20378995e+00 3.32365990e-01 9.48887885e-01 3.31312209e-01
-1.14177406e+00 -2.03641713e-01 -5.59672303e-02 -4.65159155e-02
-1.01778948e+00 -1.23248853e-01 4.95610923e-01 -6.52559102e-01
7.63475657e-01 8.50246176e-02 7.63236821e-01 5.83692491e-01
3.10484558e-01 4.04641181e-01 1.16590130e+00 -3.56969595e-01
3.11309338e-01 -2.25928053e-01 1.64479077e-01 7.43531406e-01
-6.47519827e-02 8.37035626e-02 -7.86021426e-02 -1.95680767e-01
7.24061966e-01 -2.23971888e-01 -3.19686860e-01 -4.98600215e-01
-1.36745989e+00 7.18761981e-01 7.56001234e-01 7.28777230e-01
-5.62909186e-01 -1.84953883e-01 5.74998081e-01 -5.29482886e-02
5.06359100e-01 2.01070458e-01 -4.79839623e-01 1.38771370e-01
-1.36577916e+00 1.62318349e-01 4.31242347e-01 3.68516237e-01
2.47173712e-01 -3.12030405e-01 -1.13248467e-01 5.71200073e-01
1.58628389e-01 -6.30078539e-02 8.09960425e-01 -7.34007895e-01
3.41920555e-01 4.37436253e-01 -2.44037122e-01 -6.24796331e-01
-1.07851768e+00 -8.44364345e-01 -9.34603095e-01 1.77850321e-01
7.25609720e-01 -3.45379442e-01 -9.69488084e-01 1.47200716e+00
4.96904880e-01 3.49337280e-01 -1.34973451e-01 9.58829045e-01
7.48629153e-01 1.87444001e-01 -2.90086508e-01 -2.95125544e-01
1.20619667e+00 -7.90421128e-01 -3.28299433e-01 -2.81882342e-02
8.05706859e-01 -4.41781521e-01 5.46974003e-01 4.24637556e-01
-1.11506259e+00 -4.48578328e-01 -9.67124641e-01 3.19478631e-01
-1.73305660e-01 3.30087319e-02 4.07884270e-01 6.63419962e-01
-1.37251532e+00 8.99002552e-01 -1.07717848e+00 -2.37559691e-01
6.93660319e-01 3.89257461e-01 -5.99826157e-01 4.32521813e-02
-1.12571704e+00 8.70687246e-01 3.93745720e-01 1.76986963e-01
-6.91824198e-01 -9.97622490e-01 -6.89938068e-01 -2.24295020e-01
2.57964998e-01 -5.87102473e-01 8.59560549e-01 -7.47260273e-01
-1.22572398e+00 1.22852612e+00 2.03780785e-01 -8.39805305e-01
1.07170630e+00 1.56308308e-01 -3.94101381e-01 5.72777927e-01
1.81660518e-01 6.72522843e-01 3.27630937e-01 -9.71605897e-01
-2.59255439e-01 -7.10427999e-01 -3.22948664e-01 -8.01853761e-02
1.95226341e-01 -1.53562743e-02 -4.39883322e-01 -7.15433419e-01
4.31993067e-01 -1.07556796e+00 -3.41190159e-01 9.55594778e-02
-5.32925487e-01 1.49532065e-01 5.94422817e-01 -1.15031767e+00
9.86809969e-01 -2.00373030e+00 1.30835818e-02 3.43579799e-01
4.55260426e-01 2.86356211e-01 1.56955540e-01 9.80160907e-02
-5.08383811e-01 -6.26359358e-02 -6.63740993e-01 -1.45111620e-01
-4.71531302e-01 -3.77094671e-02 2.77972728e-01 1.02008402e+00
-7.46677741e-02 8.45249474e-01 -8.49410415e-01 -5.59998214e-01
1.84104055e-01 4.32444841e-01 -5.65430164e-01 -1.31618693e-01
3.03279310e-01 7.79305995e-01 -2.06744015e-01 2.07937866e-01
6.46838188e-01 -3.55504245e-01 3.47776264e-01 -1.90287009e-01
-1.40824661e-01 3.11255101e-02 -9.52506721e-01 1.91647470e+00
-2.12239757e-01 3.84057194e-01 -6.09241799e-03 -1.16657019e+00
6.69228196e-01 5.39061487e-01 6.48067772e-01 -5.25042951e-01
-2.14338556e-01 2.50453055e-01 5.68159223e-01 -4.41181600e-01
4.54707630e-02 -4.67149973e-01 1.85982049e-01 5.10935187e-01
9.17500183e-02 2.69151200e-02 2.53828019e-01 1.19908825e-01
8.85924339e-01 1.75085530e-01 2.68045664e-01 -5.47014773e-01
5.73066950e-01 -2.43617043e-01 5.91356158e-01 7.34181464e-01
-5.28298557e-01 9.80932415e-01 6.98544085e-01 -6.76766157e-01
-1.10179460e+00 -1.18894625e+00 -6.09686494e-01 5.31134963e-01
-4.26669359e-01 -1.30207241e-01 -1.08399761e+00 -9.98995066e-01
-2.51340747e-01 3.02869618e-01 -1.09610760e+00 2.19620228e-01
-8.38699818e-01 -8.14491212e-01 6.00627065e-01 5.69764376e-01
4.51112181e-01 -1.14085615e+00 -9.61504579e-01 4.27338243e-01
-5.82879893e-02 -1.08699870e+00 -2.95873880e-01 9.12045836e-02
-1.04759872e+00 -1.12985110e+00 -1.20985067e+00 -5.44010997e-01
5.77090919e-01 -5.85087240e-01 1.12195802e+00 2.42270142e-01
-4.22061533e-01 -2.94039130e-01 -1.83829635e-01 -1.12227470e-01
-4.92308766e-01 1.56749606e-01 -5.12598753e-02 -4.92554605e-02
-2.82348573e-01 -4.57749754e-01 -8.09949219e-01 1.66605011e-01
-8.65670204e-01 -8.06855857e-02 4.32468355e-01 9.65101480e-01
7.72885442e-01 -1.78273305e-01 6.38601542e-01 -1.05789649e+00
4.21067476e-02 -2.52277732e-01 -5.92522264e-01 3.10360283e-01
-4.22660917e-01 -2.51473188e-01 4.75259304e-01 1.13143966e-01
-7.91965306e-01 7.96794295e-02 -3.66721869e-01 -1.93774313e-01
-4.56499398e-01 4.05457169e-01 2.89087504e-01 2.79964507e-02
4.90094513e-01 1.06342182e-01 1.35252476e-01 -4.69618887e-01
1.73996255e-01 2.62764424e-01 6.98631763e-01 -2.56308556e-01
1.56408772e-01 8.71738970e-01 3.91417414e-01 -8.00064027e-01
-6.20280862e-01 -3.23746473e-01 -1.48934424e+00 -2.38691553e-01
1.17569876e+00 -2.13511124e-01 -2.69273281e-01 5.02602041e-01
-9.18062747e-01 -4.94090140e-01 -3.28854322e-01 4.85516727e-01
-5.11898935e-01 3.92396092e-01 -8.46660435e-01 -1.41453877e-01
-4.91563559e-01 -1.50195193e+00 9.36691582e-01 5.95074296e-02
-2.34829232e-01 -1.24840021e+00 2.99861282e-01 3.23763013e-01
2.55779356e-01 5.72218597e-01 1.06134653e+00 -9.41459835e-01
-1.06220417e-01 -2.59730577e-01 -1.43626943e-01 2.86052227e-01
7.57246278e-03 -9.54208300e-02 -8.92883420e-01 -1.61500335e-01
-1.18253022e-01 5.61898053e-02 9.48017597e-01 8.52332711e-01
1.08225715e+00 2.27862701e-01 -3.19688410e-01 5.74597716e-01
1.32706189e+00 1.47720203e-01 6.04339302e-01 3.95212412e-01
4.28711265e-01 4.55543756e-01 1.50607809e-01 8.19963664e-02
2.10386127e-01 7.77685165e-01 3.40132505e-01 -4.45396990e-01
-1.76154017e-01 3.09558243e-01 -3.24336797e-01 4.62617487e-01
-2.23968297e-01 1.79487675e-01 -1.21113575e+00 7.21973836e-01
-1.46580160e+00 -6.42788470e-01 -2.98944056e-01 2.34927678e+00
6.48366928e-01 4.53647166e-01 3.79986554e-01 1.12232260e-01
6.39794588e-01 2.51673758e-01 -3.29167902e-01 -2.85453886e-01
4.60606925e-02 4.35767651e-01 3.14002901e-01 4.91138339e-01
-1.23361635e+00 5.72760224e-01 6.46231985e+00 4.85573798e-01
-1.38089705e+00 3.09536695e-01 1.27658284e+00 -1.90843388e-01
2.57226199e-01 -2.72651374e-01 -3.79235029e-01 4.65039432e-01
8.51900995e-01 4.29419875e-01 3.67207117e-02 3.51203084e-01
1.04011737e-01 -3.30826283e-01 -1.03013849e+00 4.78314668e-01
3.18872556e-02 -1.78602874e+00 -4.80525285e-01 1.33022711e-01
7.29599416e-01 4.10643578e-01 -1.82190597e-01 -9.35774073e-02
-2.57890224e-01 -1.06227207e+00 5.75378299e-01 6.51131392e-01
9.71165180e-01 -7.68798649e-01 1.04339159e+00 3.30742687e-01
-8.06454062e-01 3.30734581e-01 -1.33368999e-01 2.05151528e-01
5.29513299e-01 8.07863593e-01 -1.10405469e+00 8.34314346e-01
8.04057539e-01 4.28169489e-01 -6.63707852e-01 1.06354630e+00
-7.85750747e-02 6.82825565e-01 -2.79473096e-01 3.62064242e-01
5.72750092e-01 -1.64576948e-01 4.49382484e-01 1.37421227e+00
1.14732489e-01 4.52567078e-03 7.14824349e-02 6.53129518e-01
6.69129044e-02 2.08875552e-01 -4.00405586e-01 5.89528024e-01
9.79915727e-03 1.35958946e+00 -1.30835819e+00 -5.17381132e-01
-3.52240145e-01 7.81755745e-01 2.90563583e-01 1.22691065e-01
-5.44563711e-01 -1.51019424e-01 7.55248517e-02 3.85612488e-01
4.97780204e-01 2.38417640e-01 -3.63054127e-01 -6.31852627e-01
1.08712882e-01 -4.54418778e-01 7.35631168e-01 -6.64565682e-01
-1.24547958e+00 8.55457306e-01 -6.35249838e-02 -1.00347209e+00
-3.36796224e-01 -5.22775173e-01 -6.22254670e-01 1.05300665e+00
-1.19601107e+00 -9.09549952e-01 1.42887607e-01 1.90413848e-01
1.52816726e-02 -2.86663119e-02 7.93274045e-01 2.35575169e-01
-2.47423679e-01 4.08367485e-01 4.98804525e-02 4.89555418e-01
5.11290908e-01 -1.23160362e+00 3.52973789e-01 7.13065386e-01
-5.21676056e-03 4.89407480e-01 3.68663847e-01 -6.99419737e-01
-4.97656465e-01 -9.39009011e-01 9.70801711e-01 -2.99171031e-01
6.94108188e-01 -9.92587283e-02 -1.01887894e+00 7.29473770e-01
2.15874732e-01 5.98274827e-01 6.95014238e-01 -1.54597476e-01
1.10847272e-01 3.75521891e-02 -1.40075076e+00 2.08697349e-01
7.01163292e-01 -2.46387258e-01 -6.65391684e-01 2.47236475e-01
3.62510920e-01 -7.54241884e-01 -1.21054316e+00 4.94089603e-01
6.69516325e-01 -1.41973054e+00 1.02760780e+00 -6.80525720e-01
5.22327304e-01 -1.89624429e-01 4.68007475e-02 -1.41258836e+00
-2.44230017e-01 -3.73218685e-01 2.55250424e-01 7.33421922e-01
6.13295794e-01 -6.55933917e-01 8.35820377e-01 3.65153283e-01
-4.49368924e-01 -1.21698320e+00 -1.35609424e+00 -4.93896961e-01
6.00063145e-01 -4.16772425e-01 4.07606661e-01 1.12349010e+00
-4.48662229e-02 -4.76061366e-02 1.21185541e-01 8.01183283e-03
5.50678074e-01 8.82334933e-02 1.17017031e-01 -1.15536380e+00
-4.81745660e-01 -4.97681081e-01 -6.93473637e-01 -4.06528413e-01
5.07254824e-02 -1.11748672e+00 -2.73658395e-01 -1.62941170e+00
3.06452751e-01 -5.40954411e-01 -4.36693221e-01 3.06915760e-01
-1.71358824e-01 5.50261676e-01 2.31216088e-01 5.46729900e-02
-2.73780167e-01 -9.79676098e-02 1.53087831e+00 2.14101568e-01
-4.61923005e-03 1.05290428e-01 -3.73665094e-01 9.38195467e-01
7.14134634e-01 -3.93995285e-01 -7.03562871e-02 -8.29397440e-02
-1.98517486e-01 5.04456103e-01 5.35024583e-01 -1.17628670e+00
5.70848435e-02 5.42149186e-01 7.48787105e-01 -7.80752242e-01
1.07769594e-01 -5.16819179e-01 5.71418740e-02 7.45823205e-01
-4.78883743e-01 2.92260259e-01 1.66297168e-01 1.81038022e-01
-1.08266780e-02 -4.36868101e-01 1.01675546e+00 -4.50700998e-01
-2.73308307e-01 3.91109526e-01 -2.77815551e-01 2.27104366e-01
1.03170443e+00 -2.52242908e-02 -1.26266912e-01 -5.13424128e-02
-1.39758325e+00 -1.30109727e-01 3.45743388e-01 -8.69107023e-02
3.20005506e-01 -9.81230438e-01 -9.30706978e-01 1.79118872e-01
-2.14454338e-01 3.95598896e-02 5.60106456e-01 1.48881471e+00
-8.38170111e-01 5.60561717e-01 -1.62430033e-01 -8.51268888e-01
-1.07469094e+00 3.85801017e-01 7.99526691e-01 -8.01569879e-01
-1.23873401e+00 7.86953807e-01 3.62097532e-01 -6.77516460e-01
-1.69270754e-01 -3.34315091e-01 -1.59730747e-01 5.20527028e-02
5.32223940e-01 2.76661366e-01 6.13729715e-01 -8.63562882e-01
-5.20950258e-01 5.20226181e-01 -1.33865312e-01 -3.10140878e-01
1.48079824e+00 1.72954112e-01 -1.68903247e-01 3.72713804e-01
1.25042737e+00 -1.14808522e-01 -1.26472819e+00 -1.84275344e-01
1.91920972e-03 -1.33819148e-01 1.97376803e-01 -1.02301145e+00
-1.45973349e+00 9.24646258e-01 9.94354248e-01 1.00081312e-02
7.68859863e-01 4.15340550e-02 7.13900506e-01 -1.17390476e-01
1.74797058e-01 -9.74376678e-01 -2.43137047e-01 4.02804106e-01
6.92481756e-01 -9.65863287e-01 1.21303378e-02 -1.73371434e-01
-7.35979378e-01 1.10356820e+00 8.52552801e-02 -2.01441616e-01
9.26286042e-01 3.97168636e-01 2.68580675e-01 -3.88618827e-01
-3.74734074e-01 2.05425367e-01 4.57851648e-01 7.00197279e-01
7.36532986e-01 9.76874754e-02 -2.42366374e-01 4.16500717e-01
-1.05395377e-01 1.06951803e-01 4.93390322e-01 7.21637666e-01
-3.86779457e-02 -8.23544145e-01 -2.87630975e-01 7.33960330e-01
-8.65875363e-01 -2.18411773e-01 -1.10668205e-01 1.06388569e+00
3.37858349e-01 4.38392997e-01 1.86951384e-01 2.82733679e-01
2.09750876e-01 1.17390156e-01 4.61300761e-01 -4.35244113e-01
-8.52229714e-01 3.05543751e-01 9.14109126e-02 -4.30932343e-01
-3.57643425e-01 -1.19095290e+00 -1.47954798e+00 1.67627901e-01
1.09231388e-02 1.54677201e-02 4.72205549e-01 1.08427465e+00
7.91141689e-02 7.84458756e-01 2.94422001e-01 -8.41290534e-01
-1.91656202e-01 -9.78793025e-01 -8.11149538e-01 4.30415750e-01
3.08999538e-01 -5.22615016e-01 -6.15899041e-02 9.98012722e-03] | [14.25141429901123, -2.374495506286621] |
1fd3aba0-af60-448f-9f56-cf93d064d58a | wisenetmd-motion-detection-using-dynamic | 1805.09277 | null | http://arxiv.org/abs/1805.09277v1 | http://arxiv.org/pdf/1805.09277v1.pdf | WisenetMD: Motion Detection Using Dynamic Background Region Analysis | Motion detection algorithms that can be applied to surveillance cameras such
as CCTV (Closed Circuit Television) have been studied extensively. Motion
detection algorithm is mostly based on background subtraction. One main issue
in this technique is that false positives of dynamic backgrounds such as wind
shaking trees and flowing rivers might occur. In this paper, we proposed a
method to search for dynamic background region by analyzing the video and
removing false positives by re-checking false positives. The proposed method
was evaluated based on CDnet 2012/2014 dataset obtained at
"changedetection.net" site. We also compared its processing speed with other
algorithms. | ['Jeong-Eun Lim', 'Jin-Wook Shim', 'Soon-Chul Kwon', 'Sang-Ha Lee', 'Jisang Yoo'] | 2018-05-23 | null | null | null | null | ['motion-detection'] | ['computer-vision'] | [ 3.76143247e-01 -8.51183653e-01 1.85037673e-01 5.93621358e-02
7.08782896e-02 -7.18973279e-01 2.63813227e-01 -1.25215203e-01
-6.92829967e-01 8.46367002e-01 -8.03217366e-02 -6.30571187e-01
5.14055490e-01 -7.70297647e-01 -2.50921309e-01 -8.73205304e-01
4.41620201e-02 -2.02316448e-01 1.21334910e+00 5.99375851e-02
2.07323596e-01 5.91682255e-01 -1.38077796e+00 3.10383260e-01
4.37604040e-01 4.48356509e-01 4.89560574e-01 1.27525449e+00
-1.77461103e-01 1.11402738e+00 -9.97214794e-01 1.08719068e-02
4.07435119e-01 -4.95025933e-01 -4.60249126e-01 3.68261725e-01
4.98553187e-01 -7.04467118e-01 -4.65789199e-01 1.37515950e+00
4.15658563e-01 8.20574015e-02 2.24311769e-01 -1.23198795e+00
1.72262862e-01 -8.65082350e-03 -9.14378583e-01 1.39637601e+00
3.06882113e-01 2.78338164e-01 -7.67604336e-02 -5.22266448e-01
7.29894936e-01 1.16162014e+00 5.23289740e-01 7.34214365e-01
-7.18081832e-01 -7.54131198e-01 1.36556342e-01 5.83801031e-01
-1.39203548e+00 -3.31914127e-01 5.84607244e-01 -4.43349659e-01
9.10532355e-01 4.65649962e-01 7.66701460e-01 8.38083446e-01
4.90026087e-01 7.23716140e-01 9.38389838e-01 -5.34173310e-01
2.65421271e-01 -9.86432582e-02 5.44190884e-01 6.46436036e-01
6.27731025e-01 -1.92946076e-01 1.44022211e-01 -2.37648338e-01
7.21303284e-01 -1.07937396e-01 -6.20033503e-01 -2.63117440e-02
-1.01541936e+00 5.88923812e-01 -2.95468241e-01 2.88205296e-01
-2.76918113e-01 2.02866085e-02 6.13901675e-01 3.04669030e-02
1.56630486e-01 -6.24784052e-01 -5.38878918e-01 -1.71776474e-01
-8.47280681e-01 -2.07987409e-02 7.19385087e-01 9.73843873e-01
1.86916918e-01 2.30317265e-01 -6.08102903e-02 4.25367475e-01
3.21258903e-01 7.56830096e-01 3.97928506e-01 -6.43998504e-01
4.15146410e-01 3.19769949e-01 3.40272278e-01 -1.22035217e+00
-3.10260355e-01 9.10719410e-02 -8.93005490e-01 3.95761371e-01
4.00392085e-01 -6.24828577e-01 -1.17418516e+00 9.30967212e-01
5.53966165e-01 7.43718445e-01 9.85972583e-02 6.69214368e-01
1.00410450e+00 9.44021523e-01 2.49489229e-02 -7.63812542e-01
1.25613952e+00 -8.03329349e-01 -1.02136517e+00 -6.52754232e-02
3.52680176e-01 -1.02255297e+00 3.96138638e-01 6.35935307e-01
-5.96546113e-01 -4.79131788e-01 -8.08095276e-01 5.82208455e-01
-1.47966504e-01 1.11785010e-01 1.54296622e-01 1.12959242e+00
-9.81158912e-01 2.05949381e-01 -9.67690527e-01 -8.63080204e-01
2.09449112e-01 3.81289512e-01 3.96845490e-02 -2.75006946e-02
-8.23307395e-01 8.95956695e-01 4.39621300e-01 3.38412136e-01
-1.04067945e+00 3.48097593e-01 -5.09954154e-01 -3.89993161e-01
4.37031209e-01 -4.28654879e-01 1.22941101e+00 -1.32269323e+00
-1.24372840e+00 6.55276656e-01 -6.08825028e-01 -4.18697506e-01
7.54816055e-01 -1.94750756e-01 -9.57385361e-01 3.54824126e-01
-2.94236448e-02 -2.22326592e-02 8.04673076e-01 -8.63740504e-01
-8.46809149e-01 -1.93338200e-01 -2.55166411e-01 8.56566951e-02
3.22836220e-01 7.48827100e-01 -4.55956489e-01 -5.66225290e-01
1.65135130e-01 -8.82110059e-01 -2.94423491e-01 -8.96870419e-02
-1.06566362e-01 1.84144586e-01 1.73353410e+00 -8.54601979e-01
1.31271923e+00 -1.89696145e+00 -6.04455650e-01 2.33852379e-02
-7.69760692e-03 8.69727492e-01 2.42163002e-01 -1.16573177e-01
1.20878376e-01 9.30370241e-02 1.03711821e-01 4.97522920e-01
-1.08759010e+00 1.84955165e-01 3.08244657e-02 7.33411133e-01
9.37202126e-02 6.99051917e-02 -8.35729897e-01 -6.21112466e-01
6.00363970e-01 3.09447497e-01 1.41417014e-03 -9.76568535e-02
2.40741357e-01 4.58723247e-01 -3.29055727e-01 9.79054809e-01
1.22139835e+00 2.00208396e-01 5.92270903e-02 -1.22506231e-01
-4.85755771e-01 -2.09654003e-01 -1.37807000e+00 7.08484173e-01
3.04333240e-01 1.28705919e+00 3.87260675e-01 -7.17008889e-01
5.09244680e-01 4.87830400e-01 2.76844203e-01 -1.83836728e-01
4.31686699e-01 -2.73777187e-01 9.28108692e-02 -1.10662007e+00
5.33080995e-01 2.53332794e-01 6.65316820e-01 -2.48201758e-01
-3.86825800e-01 3.21610153e-01 3.77590477e-01 5.60239106e-02
1.48030734e+00 6.49671927e-02 3.91241997e-01 -2.60014951e-01
8.24659348e-01 5.19705355e-01 1.00241160e+00 8.54004145e-01
-9.72218931e-01 3.59381348e-01 4.36559506e-03 -5.35256743e-01
-7.74451077e-01 -9.59823608e-01 6.38347343e-02 4.03252333e-01
4.76757050e-01 -1.36890009e-01 -6.62131190e-01 -5.40793002e-01
-6.16435409e-01 4.38665479e-01 -2.96719283e-01 2.39813551e-01
-8.71188104e-01 -1.16314220e+00 3.64784062e-01 3.98050725e-01
9.74697828e-01 -8.64476740e-01 -1.16877675e+00 4.33683187e-01
-2.23278224e-01 -1.44101322e+00 -1.50463670e-01 1.09862173e-02
-1.20810175e+00 -1.32117069e+00 -6.20379567e-01 -8.95378172e-01
6.73570693e-01 1.15104342e+00 9.37817276e-01 2.72895396e-01
-5.86075544e-01 2.90415555e-01 -4.87999946e-01 -6.97100282e-01
-4.30326819e-01 -6.50253236e-01 -4.66991104e-02 -1.76700145e-01
6.79309785e-01 6.94447085e-02 -6.27701402e-01 5.59832215e-01
-8.06352496e-01 -5.71346432e-02 1.87620565e-01 2.90664136e-01
1.10843092e-01 5.95257401e-01 -3.89626101e-02 -6.51636541e-01
3.25797617e-01 -3.26321304e-01 -1.11152005e+00 2.78015077e-01
1.62632629e-01 -6.03227079e-01 2.32232958e-01 -6.26046956e-01
-1.07504475e+00 3.13937396e-01 2.44722024e-01 -4.32068288e-01
-5.52434564e-01 -3.20891708e-01 -2.17128456e-01 -1.91954792e-01
2.59453893e-01 2.19641507e-01 -3.30300033e-01 -2.03857854e-01
-2.65133649e-01 6.95785463e-01 5.82683206e-01 1.82052687e-01
5.98040819e-01 7.99806416e-01 -1.21203035e-01 -1.60129988e+00
-8.00058153e-03 -9.56946731e-01 -6.14685416e-01 -6.22593522e-01
1.27963293e+00 -9.73374665e-01 -4.77371633e-01 1.00290799e+00
-1.51356626e+00 7.69391060e-02 6.14795208e-01 7.97677577e-01
1.97139770e-01 8.84609222e-01 -5.41903317e-01 -1.36134386e+00
-3.12653959e-01 -9.22071874e-01 4.77276146e-01 6.03816152e-01
1.02868743e-01 -8.41456115e-01 2.16556713e-01 8.83658677e-02
4.07108039e-01 3.37474257e-01 1.95868164e-01 -2.72741258e-01
-6.94079101e-01 -9.66931507e-02 -1.28693730e-01 2.87065864e-01
3.67972046e-01 7.64629245e-01 -9.18540061e-01 -1.70201451e-01
3.78057748e-01 7.58084357e-01 8.32822204e-01 7.55286634e-01
7.33375847e-01 -1.65065825e-01 -6.80245459e-01 2.50796467e-01
1.62820685e+00 1.01717687e+00 9.71705079e-01 6.09155834e-01
6.29093647e-01 1.47194847e-01 6.73739254e-01 3.47482175e-01
-1.30436644e-01 4.70667124e-01 3.84596109e-01 -8.63468349e-02
1.46952748e-01 5.44877470e-01 5.64497530e-01 5.31244040e-01
-3.79878014e-01 -5.31414390e-01 -1.05406857e+00 5.23434281e-01
-1.83674133e+00 -1.53483117e+00 -1.09215021e+00 2.13539982e+00
1.10467888e-01 2.64287144e-01 1.44313991e-01 1.03523150e-01
1.43349361e+00 -1.40461981e-01 -1.50239304e-01 -2.91457206e-01
-3.88121635e-01 2.48606317e-02 9.36458409e-01 3.45535219e-01
-1.56178033e+00 8.25634241e-01 6.51097584e+00 2.48568997e-01
-1.23829269e+00 2.47028679e-01 4.67793614e-01 -1.46873798e-02
6.61249459e-01 3.25169112e-03 -9.22844529e-01 6.70342803e-01
6.76023901e-01 3.10114641e-02 -1.17894575e-01 6.47101283e-01
8.25774550e-01 -8.97363484e-01 -2.13628575e-01 1.13864660e+00
1.36583671e-01 -9.72981274e-01 -1.06570505e-01 -1.88656598e-01
5.26606321e-01 5.09562120e-02 -6.69332206e-01 6.90237135e-02
1.73483863e-01 -5.97897947e-01 4.65474635e-01 1.52702972e-01
-9.96577665e-02 -4.22753274e-01 1.04670453e+00 2.59680629e-01
-1.35168087e+00 8.12188759e-02 -5.52321017e-01 -2.72137612e-01
4.33812737e-01 4.67928767e-01 -8.08019876e-01 2.06199691e-01
9.75648344e-01 4.45308328e-01 -5.33730984e-01 1.55825365e+00
1.50678366e-01 8.82907748e-01 -4.44745839e-01 -1.10732928e-01
1.90593347e-01 -3.77180874e-01 8.70080411e-01 1.54220092e+00
4.16630179e-01 4.33118820e-01 1.22584008e-01 1.71274439e-01
4.12853867e-01 -2.28326432e-02 -8.85931790e-01 5.12752771e-01
2.02154100e-01 9.95481491e-01 -1.39392626e+00 -8.19115520e-01
-7.82256603e-01 1.09503222e+00 -6.00897253e-01 3.70351553e-01
-1.23101676e+00 -2.78131545e-01 3.04206043e-01 2.40976676e-01
5.46798170e-01 -4.79942262e-01 1.94967449e-01 -1.30239964e+00
-2.20168322e-01 -6.43841982e-01 4.54449326e-01 -8.21560383e-01
-7.11839974e-01 6.35402322e-01 2.94779241e-01 -1.38662350e+00
4.72793460e-01 -7.53846228e-01 -1.21934927e+00 5.45482874e-01
-9.05517817e-01 -6.35885179e-01 -8.55983377e-01 8.20661008e-01
1.03938353e+00 -1.00482076e-01 2.68099338e-01 4.60369676e-01
-1.01180971e+00 -1.54597253e-01 1.42030284e-01 3.55677783e-01
3.19422811e-01 -6.84382260e-01 2.33773381e-01 1.71660304e+00
-2.45719329e-01 3.18649799e-01 9.93925631e-01 -1.04918158e+00
-1.03506935e+00 -1.02658832e+00 3.21318030e-01 -2.88803458e-01
5.06906331e-01 -9.16243717e-02 -8.51739466e-01 4.92928535e-01
5.02193809e-01 1.14278391e-01 2.89774507e-01 -8.27364981e-01
4.96820539e-01 2.27125868e-01 -1.28636265e+00 5.14991581e-01
7.22067297e-01 2.33922482e-01 -3.84317666e-01 1.98581412e-01
1.43768609e-01 -3.45115483e-01 1.24080040e-01 4.64618623e-01
3.41736108e-01 -1.00295234e+00 6.50793254e-01 -2.80316204e-01
-3.63095909e-01 -9.09591913e-01 -1.88614979e-01 -5.93480766e-01
-2.12413788e-01 -6.92714095e-01 5.91665246e-02 1.20177257e+00
-3.43445726e-02 -3.44343841e-01 8.34573925e-01 4.67437088e-01
4.28312629e-01 1.69395015e-01 -8.53650570e-01 -8.57451856e-01
-6.76915884e-01 -2.90178299e-01 -1.78024262e-01 9.94171739e-01
-5.46418786e-01 2.01966092e-01 -5.92628241e-01 5.44917464e-01
5.86829007e-01 -5.97858310e-01 8.43724966e-01 -1.03949893e+00
-7.41936713e-02 -1.70419421e-02 -9.32336271e-01 -5.94426215e-01
-5.03952622e-01 7.66292214e-02 -1.34984227e-02 -1.45771098e+00
3.47718507e-01 2.61283189e-01 -1.04986437e-01 3.46847773e-02
-3.74763817e-01 1.75045654e-01 1.79980621e-01 6.75242618e-02
-5.08230388e-01 -9.48953331e-02 7.15557814e-01 -4.50772382e-02
-3.50927860e-01 1.10602282e-01 2.88244069e-01 1.19386017e+00
1.34277117e+00 -5.84626734e-01 -2.07854792e-01 -4.25220937e-01
-3.76315445e-01 1.84588861e-02 5.14292419e-01 -1.27501965e+00
2.53186584e-01 -3.91838461e-01 5.87025821e-01 -9.75600958e-01
-7.56720901e-02 -1.19225442e+00 3.43839705e-01 8.95063341e-01
4.93473798e-01 7.55659223e-01 7.16244340e-01 8.39361012e-01
-7.82171860e-02 -4.95487571e-01 8.78567755e-01 -4.45739180e-01
-1.17744935e+00 -3.02910209e-01 -1.34831834e+00 -2.72942126e-01
1.54657233e+00 -6.81521475e-01 -5.89906335e-01 -3.14700514e-01
-5.74219286e-01 2.06476934e-02 5.01760662e-01 1.69674322e-01
6.69557989e-01 -8.63161266e-01 -5.62977672e-01 5.63100986e-02
-3.00477266e-01 -3.19682121e-01 2.40787733e-02 9.64636505e-01
-1.46093309e+00 1.98022932e-01 -4.52179581e-01 -6.74872756e-01
-2.41502047e+00 6.33486569e-01 3.28640014e-01 1.12744302e-01
-8.17072690e-01 4.48862642e-01 3.04226503e-02 6.47276044e-01
1.92908779e-01 -6.53336465e-01 -3.95557761e-01 -3.50786358e-01
9.23324704e-01 8.70912373e-01 -1.24870285e-01 -7.77838826e-01
-8.74391854e-01 6.88495874e-01 5.59644913e-03 -1.16519220e-01
8.17207515e-01 -3.26038599e-01 -9.08320472e-02 1.95285454e-01
7.75734067e-01 2.14115113e-01 -9.62437153e-01 3.84902209e-01
1.49241641e-01 -8.15207839e-01 1.08952075e-01 -3.84257406e-01
-1.09947181e+00 6.36712730e-01 1.30598843e+00 1.48811281e-01
1.42837632e+00 -5.97348750e-01 5.00333130e-01 3.32693398e-01
4.05759186e-01 -1.12859094e+00 -2.93445915e-01 4.63143855e-01
2.35771433e-01 -1.41488087e+00 1.08459242e-01 -7.20968008e-01
-6.54778540e-01 1.24656260e+00 7.32542813e-01 -1.59339786e-01
5.85224390e-01 5.75367332e-01 3.27211589e-01 1.20195329e-01
-7.28736222e-01 -1.96766734e-01 -1.98300943e-01 9.15250301e-01
4.03870434e-01 -1.01953484e-01 -5.08479655e-01 -1.82674676e-01
6.70360088e-01 2.36580372e-01 1.17651868e+00 1.61985362e+00
-8.00757349e-01 -6.40521288e-01 -1.08320630e+00 3.25013697e-01
-9.88793790e-01 -4.31395210e-02 -5.24974644e-01 8.51615727e-01
1.86606809e-01 1.40051043e+00 8.21064636e-02 -2.46323928e-01
1.33265778e-01 -2.46423379e-01 4.08477366e-01 -2.89994568e-01
-4.97024536e-01 3.44170570e-01 3.24895740e-01 -3.68349105e-01
-9.36016798e-01 -6.22821689e-01 -1.05338156e+00 -3.69047344e-01
-6.02697551e-01 -2.35820971e-02 6.23599410e-01 4.37483102e-01
-5.38282320e-02 5.72351277e-01 1.48961961e-01 -5.16467214e-01
1.98103026e-01 -8.13078821e-01 -4.37455565e-01 2.24881098e-01
5.10193586e-01 -4.61359590e-01 -4.58065271e-01 5.30134320e-01] | [8.835376739501953, -0.945231020450592] |
b75ebc4e-683e-4b6b-9585-fa103d281def | shielded-decision-making-in-mdps | 1807.06096 | null | https://arxiv.org/abs/1807.06096v2 | https://arxiv.org/pdf/1807.06096v2.pdf | Safe Reinforcement Learning via Probabilistic Shields | This paper targets the efficient construction of a safety shield for decision making in scenarios that incorporate uncertainty. Markov decision processes (MDPs) are prominent models to capture such planning problems. Reinforcement learning (RL) is a machine learning technique to determine near-optimal policies in MDPs that may be unknown prior to exploring the model. However, during exploration, RL is prone to induce behavior that is undesirable or not allowed in safety- or mission-critical contexts. We introduce the concept of a probabilistic shield that enables decision-making to adhere to safety constraints with high probability. In a separation of concerns, we employ formal verification to efficiently compute the probabilities of critical decisions within a safety-relevant fragment of the MDP. We use these results to realize a shield that is applied to an RL algorithm which then optimizes the actual performance objective. We discuss tradeoffs between sufficient progress in exploration of the environment and ensuring safety. In our experiments, we demonstrate on the arcade game PAC-MAN and on a case study involving service robots that the learning efficiency increases as the learning needs orders of magnitude fewer episodes. | ['Bettina Könighofer', 'Roderick Bloem', 'Nils Jansen', 'Alexandru C. Serban', 'Sebastian Junges'] | 2018-07-16 | null | null | null | null | ['safe-exploration'] | ['robots'] | [ 1.90908119e-01 7.44327247e-01 -2.86495775e-01 -1.84304282e-01
-9.47147965e-01 -5.74350238e-01 4.19701874e-01 4.49107498e-01
-4.69968557e-01 1.06258798e+00 -4.05081436e-02 -8.51252556e-01
-5.98130584e-01 -9.42237377e-01 -8.55514824e-01 -5.60027957e-01
-8.87606978e-01 7.27459013e-01 4.28160727e-01 -2.12486267e-01
2.37618312e-01 5.38350701e-01 -1.47061574e+00 -1.24763139e-01
6.96751475e-01 7.80233324e-01 1.49013206e-01 7.83428788e-01
2.56097317e-01 9.55990851e-01 -3.14678997e-01 8.26365799e-02
5.28417706e-01 -6.75227959e-03 -1.10115337e+00 8.01160485e-02
-7.47135043e-01 -6.27291083e-01 -1.31549099e-02 1.05853522e+00
1.52620515e-02 2.19801173e-01 6.84216559e-01 -1.81623042e+00
3.97960424e-01 1.00143552e+00 -4.41831440e-01 -9.17433798e-02
3.14668000e-01 5.91141462e-01 7.07864523e-01 1.29266128e-01
2.98098624e-01 1.50811541e+00 1.46830425e-01 5.79234779e-01
-1.13399613e+00 -3.84024471e-01 5.28044105e-01 7.30385706e-02
-1.26712477e+00 -1.60791781e-02 4.38223481e-01 -3.56596500e-01
1.06754696e+00 6.56740293e-02 6.66139245e-01 8.73214364e-01
9.95123506e-01 8.74687254e-01 1.23384881e+00 -5.05624056e-01
1.00116742e+00 -2.35509038e-01 -3.63975376e-01 4.12244380e-01
4.61960822e-01 9.73400295e-01 -1.17510758e-01 -3.71553659e-01
4.77287382e-01 -1.87728003e-01 2.14200243e-01 -6.52924538e-01
-7.70097196e-01 9.28091705e-01 -1.36827171e-01 -3.14355046e-01
-6.37507439e-01 4.69304442e-01 2.65299559e-01 3.34588915e-01
-3.00858438e-01 7.25137353e-01 -4.14433718e-01 -4.50765669e-01
-3.72377306e-01 6.95354640e-01 1.11286688e+00 9.97532725e-01
5.42919099e-01 7.66177773e-02 -1.55177921e-01 -1.13547534e-01
4.96656030e-01 3.82421404e-01 -2.81550974e-01 -1.41221726e+00
5.54250956e-01 1.48788348e-01 9.00108755e-01 -4.44732904e-01
-4.43472922e-01 2.40289550e-02 -6.55461848e-02 9.85460520e-01
7.54808411e-02 -5.97525716e-01 -9.49363410e-01 1.77460551e+00
4.18172926e-01 -8.63976851e-02 4.70902056e-01 5.86611569e-01
-5.75493872e-01 7.36781120e-01 3.30134094e-01 -5.31283438e-01
9.24399614e-01 -3.99523139e-01 -7.28722155e-01 -2.95245767e-01
4.92543161e-01 -8.38471949e-02 8.25170338e-01 7.12813377e-01
-1.11216509e+00 1.07233189e-01 -1.31300342e+00 9.77192700e-01
2.23426327e-01 -8.15560520e-01 7.07682550e-01 5.92543721e-01
-8.79435301e-01 7.22437143e-01 -1.30509877e+00 -1.42811522e-01
1.76767558e-01 4.55170244e-01 1.39265910e-01 2.98945773e-02
-1.16955817e+00 1.30895638e+00 7.32250571e-01 -6.91648200e-02
-2.07731438e+00 -2.27624938e-01 -9.23406363e-01 2.33733617e-02
1.07947063e+00 -1.33699760e-01 1.88123012e+00 -4.51045126e-01
-1.57972407e+00 2.30670460e-02 3.34373385e-01 -8.58666241e-01
7.14996755e-01 -2.31948018e-01 -2.43582547e-01 1.66621715e-01
1.50047094e-01 5.72181225e-01 7.73830473e-01 -1.50514495e+00
-1.21354675e+00 -9.69052762e-02 6.78479493e-01 3.26993257e-01
3.06334436e-01 -1.53944746e-01 2.38502041e-01 6.47027837e-03
-1.92720234e-01 -1.29677486e+00 -1.21557081e+00 -5.68437397e-01
-4.54555243e-01 -1.22745246e-01 3.12839657e-01 -1.25109851e-01
1.02382457e+00 -1.70885706e+00 -5.11950962e-02 5.70120513e-01
-3.11727822e-01 -3.50419492e-01 -1.96716171e-02 8.51820529e-01
3.03954184e-01 5.90662919e-02 -4.31647211e-01 1.20908163e-01
2.43179202e-01 6.52106404e-01 -5.72463453e-01 4.62982178e-01
2.47972399e-01 4.00702029e-01 -9.53932047e-01 -3.68838429e-01
1.45962566e-01 -3.79162878e-01 -6.26144290e-01 5.67837358e-01
-6.88672543e-01 3.67280662e-01 -7.65703976e-01 6.58587277e-01
3.58494401e-01 3.87122691e-01 4.03503150e-01 8.38884652e-01
-3.10740858e-01 1.95187271e-01 -1.34655404e+00 1.24330842e+00
-3.95519465e-01 -1.34206101e-01 2.10436106e-01 -6.99840605e-01
5.51436424e-01 2.11369559e-01 4.73310828e-01 -2.72939414e-01
2.14906871e-01 3.83983254e-02 1.09539978e-01 -5.25963247e-01
4.98968363e-01 -4.14081633e-01 -6.11675322e-01 7.38522887e-01
-3.62374067e-01 -5.19134879e-01 6.05237558e-02 -8.33247006e-02
1.41303015e+00 3.44074249e-01 5.56103051e-01 -5.09548247e-01
1.80078998e-01 4.51276273e-01 9.09620523e-01 1.11474597e+00
-4.23380613e-01 -2.44348809e-01 9.53820825e-01 -3.60155702e-01
-6.82899177e-01 -9.29051340e-01 2.84579903e-01 7.23105013e-01
5.15523136e-01 -1.54733300e-01 -6.39201522e-01 -7.51884818e-01
1.41169624e-02 1.51523924e+00 -5.91070473e-01 -3.21600884e-01
-3.70098740e-01 -4.45923418e-01 2.91231722e-01 3.75619233e-01
1.09019801e-01 -1.00740814e+00 -1.55828357e+00 4.06972677e-01
1.74274966e-01 -8.27263713e-01 -5.95635660e-02 7.10817397e-01
-6.84455991e-01 -1.27327359e+00 9.61403251e-02 -3.10754806e-01
7.42286742e-01 -1.07563965e-01 6.55354857e-01 -3.27012151e-01
3.77798900e-02 6.65604055e-01 -2.43352816e-01 -8.73288095e-01
-7.98069477e-01 -2.51576096e-01 4.37394172e-01 -5.37070632e-01
1.15075260e-01 -4.84368503e-01 -2.91835219e-01 2.32744202e-01
-8.86198580e-01 -3.72793376e-02 5.11044741e-01 5.26974201e-01
5.93166709e-01 7.97493398e-01 3.92632574e-01 -4.77013677e-01
9.30836201e-01 -5.40795922e-01 -1.25717819e+00 4.50377375e-01
-7.85953045e-01 6.55867577e-01 3.79075497e-01 -4.55528617e-01
-1.14386404e+00 2.74871230e-01 3.16624671e-01 -2.70133376e-01
-2.78030485e-01 4.72667754e-01 -3.27505350e-01 2.47119278e-01
4.35703695e-01 -4.56089415e-02 4.73038713e-03 9.35110301e-02
1.18858740e-01 2.32806429e-01 1.41674787e-01 -1.32318521e+00
6.53887033e-01 1.55186102e-01 3.76926392e-01 -1.39427751e-01
-3.49215478e-01 7.43780136e-02 -1.00379445e-01 -4.07030284e-01
5.77385128e-01 -6.44585490e-01 -1.26149654e+00 -8.67143422e-02
-7.19547153e-01 -7.96390474e-01 -4.59208190e-01 4.16374385e-01
-1.37366974e+00 5.98181374e-02 -3.23320091e-01 -1.69406521e+00
2.22103462e-01 -1.43947268e+00 6.41425312e-01 2.37311184e-01
-3.08446646e-01 -4.34222311e-01 2.88026035e-01 -3.07046175e-01
-2.92661730e-02 4.45522726e-01 9.00982857e-01 -5.00878990e-01
-7.04565823e-01 1.07485995e-01 5.49740911e-01 -1.21295378e-01
-4.28533852e-02 -5.75420372e-02 -6.04908347e-01 -4.55013335e-01
1.27396226e-01 -3.40954244e-01 1.32081792e-01 3.76509011e-01
1.05065715e+00 -9.10954058e-01 -4.18864310e-01 -5.12089208e-02
1.45237148e+00 1.07669318e+00 5.11700630e-01 7.74032593e-01
-1.40915006e-01 1.06655288e+00 1.50238216e+00 8.51044953e-01
5.33929110e-01 4.69987005e-01 1.05533278e+00 5.79042435e-01
8.82178724e-01 -5.67052066e-01 6.29558980e-01 -2.77299970e-01
-1.02764368e-02 -5.47317863e-02 -1.17527938e+00 7.07293153e-01
-2.11742854e+00 -9.22782063e-01 5.44551551e-01 2.40629745e+00
8.63805175e-01 6.83236480e-01 2.53086060e-01 1.55981839e-01
5.33819139e-01 -2.80918151e-01 -7.53266990e-01 -8.94724667e-01
4.89919692e-01 -1.87221244e-01 1.00230587e+00 9.78909552e-01
-9.51382399e-01 8.59593570e-01 6.87168360e+00 5.89819074e-01
-5.25610268e-01 -1.91855103e-01 5.75780451e-01 -1.21189073e-01
-4.35676843e-01 3.90068740e-01 -7.21179366e-01 1.75565332e-01
1.12119567e+00 -3.49277079e-01 6.18294358e-01 1.16490591e+00
5.53615868e-01 -6.85558438e-01 -1.36731648e+00 2.16734439e-01
-7.46623278e-01 -9.86545384e-01 -2.30040774e-01 1.39319524e-01
6.16571903e-01 -3.80174071e-01 -1.72174215e-01 5.70285261e-01
1.31996179e+00 -1.05772316e+00 1.26933086e+00 1.93393186e-01
4.35989022e-01 -1.63823497e+00 6.11477733e-01 7.74619997e-01
-7.54347384e-01 -7.58074820e-01 -6.96189189e-03 -2.49489054e-01
5.08597612e-01 1.60276860e-01 -1.38852239e+00 3.69327188e-01
6.36079252e-01 -1.43013701e-01 1.62663445e-01 8.58109415e-01
-6.55638516e-01 3.93218517e-01 -3.48320067e-01 -2.07642809e-01
6.54700220e-01 -1.10766180e-01 6.45761430e-01 7.47576654e-01
1.74847081e-01 4.24983561e-01 6.14626110e-01 7.31630564e-01
8.63548696e-01 -6.24027371e-01 -8.49052727e-01 -5.71285002e-02
6.36919498e-01 7.11694777e-01 -7.36886859e-01 2.39757840e-02
1.74066544e-01 3.61500770e-01 7.77494311e-02 3.95771027e-01
-9.45341229e-01 -1.82225585e-01 8.92628193e-01 -9.73913372e-02
1.26041949e-01 -3.65223438e-01 -3.66571456e-01 -3.52034330e-01
-1.40585050e-01 -8.40229392e-01 3.83249521e-01 -4.17907774e-01
-7.23038971e-01 4.89782423e-01 6.12806678e-01 -1.25795221e+00
-8.16206217e-01 -2.77182341e-01 -5.30955017e-01 5.18964291e-01
-1.48933244e+00 -6.68609083e-01 4.22322720e-01 3.48107666e-01
5.37462115e-01 4.18021642e-02 6.47696495e-01 -6.07405663e-01
-4.41114575e-01 -6.57877401e-02 -3.20149183e-01 -5.11484325e-01
1.85904722e-03 -1.23350191e+00 2.03641191e-01 1.04639828e+00
-6.94191813e-01 2.85069287e-01 1.31517613e+00 -1.09313548e+00
-1.55500877e+00 -1.03103817e+00 1.32443950e-01 -2.66970813e-01
6.24064505e-01 -1.03481477e-02 -4.28657264e-01 8.77899289e-01
-2.41377447e-02 -5.85261822e-01 2.92204350e-01 -9.73745883e-02
1.90677002e-01 4.60540503e-02 -1.42550206e+00 9.57757890e-01
9.13437605e-01 -6.31023720e-02 -8.97390902e-01 5.50682563e-03
9.77501452e-01 -3.75899643e-01 -5.28853774e-01 5.55729091e-01
2.83843040e-01 -6.16467416e-01 4.40435201e-01 -9.29266632e-01
1.22195698e-01 -5.04366279e-01 -2.78364092e-01 -1.39642692e+00
-1.21030793e-01 -1.21284127e+00 -2.71104634e-01 7.75484979e-01
4.07604069e-01 -4.16326255e-01 6.31566882e-01 1.05649340e+00
-1.97473973e-01 -7.85019279e-01 -1.22326851e+00 -1.12394297e+00
4.41807657e-02 -6.79574311e-01 7.91114509e-01 3.51474196e-01
5.60935199e-01 -3.91814828e-01 -3.13829660e-01 7.03107953e-01
9.40843225e-01 1.06263414e-01 4.13219452e-01 -6.67810082e-01
-3.19955647e-01 -7.59928301e-02 8.16024840e-02 -5.37838638e-01
2.08303258e-01 -7.46627450e-02 9.97103214e-01 -1.43681145e+00
-1.19262531e-01 -9.13593113e-01 -2.98078775e-01 5.63521743e-01
2.35751241e-01 -1.16595185e+00 2.15119675e-01 -3.76807638e-02
-8.37584615e-01 6.91969037e-01 1.14469576e+00 -1.43494859e-01
-5.26648939e-01 4.63791251e-01 -7.11370170e-01 7.28204429e-01
1.01927698e+00 -5.86955428e-01 -7.54369378e-01 -8.95675123e-02
2.55294502e-01 8.66193831e-01 -1.78995028e-01 -9.18240786e-01
1.44963771e-01 -1.30916405e+00 -3.05618376e-01 -6.43634737e-01
2.16081902e-01 -1.16359162e+00 4.09149587e-01 1.10011947e+00
-7.80846477e-01 2.22421363e-01 2.98081964e-01 9.19055223e-01
1.26985103e-01 -4.55049962e-01 6.02664471e-01 -1.33094564e-01
-9.12253320e-01 2.18463883e-01 -1.10808384e+00 -1.95817769e-01
1.76001227e+00 1.62266061e-01 1.99275434e-01 -5.93000650e-01
-5.80694258e-01 9.07439232e-01 3.35740358e-01 3.17603558e-01
7.18316078e-01 -9.57938135e-01 -4.03238386e-01 -1.76622525e-01
-7.57715777e-02 2.81238049e-01 -3.00460476e-02 3.89001906e-01
-4.19440120e-01 3.34721595e-01 -4.95462179e-01 -2.74643719e-01
-7.51623929e-01 8.24334323e-01 4.93938714e-01 -5.56661546e-01
-3.63761842e-01 4.54799563e-01 4.12176028e-02 -2.00790823e-01
5.42594194e-01 -4.55177456e-01 -1.38985857e-01 -3.56045663e-01
6.24683321e-01 4.10920739e-01 -2.59611934e-01 1.14227504e-01
-4.76125300e-01 -7.51371533e-02 2.17244402e-01 -8.50837350e-01
1.20641649e+00 -2.75662988e-01 2.53544658e-01 2.98226207e-01
9.01649445e-02 -3.22610676e-01 -2.03243971e+00 2.14866146e-01
5.60831785e-01 -4.04581904e-01 -1.70368642e-01 -7.85893798e-01
-4.26263958e-01 4.98334378e-01 2.89084584e-01 2.92218179e-01
9.51435685e-01 -2.11882889e-01 2.20247313e-01 6.00241840e-01
1.23182952e+00 -1.45056331e+00 -9.89843905e-02 5.98966956e-01
8.20451558e-01 -9.54562962e-01 -1.60329372e-01 -1.52909726e-01
-1.15566909e+00 9.11094785e-01 8.61880898e-01 -1.68832541e-02
4.21104342e-01 7.00615466e-01 -2.43966669e-01 1.74156055e-02
-1.07470608e+00 -1.37795731e-01 -6.05447948e-01 8.10016930e-01
-8.02961528e-01 5.20282030e-01 -1.46132320e-01 5.83166301e-01
-5.24815768e-02 -7.07422420e-02 9.26348388e-01 1.69999516e+00
-9.78576958e-01 -9.99132454e-01 -6.87624276e-01 1.33852497e-01
-2.75982946e-01 4.36300188e-01 4.05066982e-02 8.42058063e-01
-1.00117728e-01 1.36344087e+00 -8.87910649e-02 -3.68318915e-01
2.97370732e-01 -1.63510442e-01 4.83269870e-01 -6.11812770e-01
-6.83229640e-02 -3.33180092e-02 5.81962407e-01 -1.00814879e+00
-1.68539435e-01 -8.51395071e-01 -1.64566565e+00 -1.24939919e-01
1.42186135e-01 5.09141684e-01 5.41626155e-01 1.15564764e+00
-1.22551642e-01 5.78022361e-01 9.21890676e-01 -5.65367520e-01
-1.33679199e+00 -3.52872819e-01 -6.47846043e-01 -4.83414263e-01
4.12429750e-01 -8.98567557e-01 -2.56406903e-01 -4.03317124e-01] | [4.541505336761475, 2.104837417602539] |
ae6fc0e5-9556-46b5-a0a4-97e0a65098c4 | polyphone-disambiguation-and-accent | 2201.09427 | null | https://arxiv.org/abs/2201.09427v1 | https://arxiv.org/pdf/2201.09427v1.pdf | Polyphone disambiguation and accent prediction using pre-trained language models in Japanese TTS front-end | Although end-to-end text-to-speech (TTS) models can generate natural speech, challenges still remain when it comes to estimating sentence-level phonetic and prosodic information from raw text in Japanese TTS systems. In this paper, we propose a method for polyphone disambiguation (PD) and accent prediction (AP). The proposed method incorporates explicit features extracted from morphological analysis and implicit features extracted from pre-trained language models (PLMs). We use BERT and Flair embeddings as implicit features and examine how to combine them with explicit features. Our objective evaluation results showed that the proposed method improved the accuracy by 5.7 points in PD and 6.0 points in AP. Moreover, the perceptual listening test results confirmed that a TTS system employing our proposed model as a front-end achieved a mean opinion score close to that of synthesized speech with ground-truth pronunciation and accent in terms of naturalness. | ['Toshiyuki Kumakura', 'Toshiyuki Sekiya', 'Emiru Tsunoo', 'Chie Kamada', 'Masaki Hamada', 'Rem Hida'] | 2022-01-24 | null | null | null | null | ['morphological-analysis', 'polyphone-disambiguation'] | ['natural-language-processing', 'natural-language-processing'] | [ 9.53101367e-03 4.09658343e-01 2.64133543e-01 -5.01036584e-01
-1.28762591e+00 -5.03336191e-01 3.88539910e-01 -1.34759858e-01
-3.36394519e-01 7.62615919e-01 6.01753891e-01 -2.88701981e-01
3.64683628e-01 -2.13442415e-01 -3.85458976e-01 -6.04697049e-01
3.20221752e-01 2.12101921e-01 2.04542354e-01 -2.98618287e-01
7.00921286e-03 2.64467001e-01 -1.27082729e+00 2.77281910e-01
9.05938983e-01 1.18605900e+00 6.31016374e-01 1.08410144e+00
-5.44059463e-02 6.02315605e-01 -1.02841330e+00 -4.35345352e-01
5.67634255e-02 -2.90471792e-01 -6.84424102e-01 -3.44254337e-02
1.95287436e-01 -1.65602654e-01 -5.56531809e-02 9.84481573e-01
9.12379980e-01 3.24457228e-01 5.46768844e-01 -7.21457422e-01
-6.62774503e-01 7.84867942e-01 2.39774212e-02 7.40095377e-02
4.12716120e-01 1.58698991e-01 1.05455005e+00 -1.11579633e+00
2.49313831e-01 1.10515428e+00 5.11444390e-01 5.44227183e-01
-1.14641881e+00 -4.69015032e-01 -2.96116825e-02 1.21939287e-01
-1.25613141e+00 -9.51802492e-01 7.50223279e-01 -3.22094023e-01
1.17032719e+00 4.40149844e-01 2.62806237e-01 1.01228762e+00
1.70423239e-01 7.75324821e-01 1.17200875e+00 -7.49077916e-01
1.98584437e-01 3.79746079e-01 3.36012356e-02 2.02698708e-01
-5.88096023e-01 1.49301633e-01 -6.03578627e-01 2.20936000e-01
3.88615519e-01 -8.99095714e-01 -3.07545960e-01 5.77830791e-01
-1.11788988e+00 6.54115975e-01 4.11959291e-02 5.73429942e-01
-5.67859650e-01 -3.39763373e-01 3.62079948e-01 1.05266511e-01
4.30143297e-01 4.98327166e-01 -7.05834031e-01 -4.31347758e-01
-1.06634462e+00 -1.16982803e-01 7.78980851e-01 9.12132680e-01
3.10386986e-01 6.74398065e-01 -3.27258170e-01 1.25132060e+00
1.72135681e-01 8.10689151e-01 7.71414697e-01 -8.54629517e-01
3.85164738e-01 -2.16636851e-01 2.44773552e-01 -7.57496595e-01
-7.50982985e-02 -4.50839341e-01 -5.00352621e-01 -4.59587239e-02
2.50076801e-01 -5.34304261e-01 -8.99423480e-01 1.64900470e+00
9.36214179e-02 2.66539138e-02 5.12799859e-01 9.25518155e-01
6.17108166e-01 1.19509518e+00 3.45717072e-02 -4.52974379e-01
1.24221492e+00 -1.21442902e+00 -1.25354517e+00 -3.25174898e-01
3.60314488e-01 -1.37681305e+00 1.37914979e+00 4.54965651e-01
-1.27918541e+00 -1.16715431e+00 -1.00807953e+00 1.49758369e-01
-2.19314769e-01 4.54398990e-01 -2.40893677e-01 8.38525891e-01
-1.27393091e+00 1.99588194e-01 -4.49781179e-01 -1.05828114e-01
-5.52678645e-01 4.77995545e-01 -2.10020021e-01 5.86053550e-01
-1.54329133e+00 8.42429101e-01 3.92660022e-01 2.31356040e-01
-5.24063349e-01 -3.79156768e-01 -8.29829574e-01 1.46545544e-01
1.01851560e-01 -3.11900795e-01 1.84490716e+00 -9.36145723e-01
-2.33892703e+00 3.86634231e-01 -4.21964943e-01 -5.73426604e-01
9.59734321e-02 -4.95272905e-01 -1.06285334e+00 2.57972926e-01
4.47809175e-02 7.22490132e-01 8.33690166e-01 -1.08969271e+00
-8.25596213e-01 9.79172736e-02 -4.33108985e-01 5.01173019e-01
-2.66691804e-01 1.93257391e-01 -1.14154488e-01 -8.64712834e-01
8.72170832e-03 -7.34043539e-01 -7.65211508e-02 -6.70289516e-01
-6.75029993e-01 -1.89165428e-01 6.94299340e-01 -1.21410525e+00
1.33275366e+00 -2.23572445e+00 -1.65578708e-01 4.61502858e-02
-3.51763606e-01 7.14642644e-01 -5.32798767e-02 2.82841533e-01
1.46171227e-01 6.12199008e-02 7.57767931e-02 -5.58875978e-01
2.78530538e-01 1.01250941e-02 -4.92525190e-01 -2.42315382e-01
3.47674310e-01 5.88200688e-01 -7.36375391e-01 -5.34127831e-01
6.29327357e-01 4.81657296e-01 -3.36752921e-01 6.45138085e-01
-9.78806168e-02 5.48649788e-01 -3.63741117e-03 3.38497788e-01
4.84009832e-01 6.80725098e-01 7.53664076e-02 -1.06392220e-01
-3.48807216e-01 8.71622384e-01 -1.06841302e+00 1.23942578e+00
-8.33978593e-01 6.86709046e-01 1.52085036e-01 -4.52248931e-01
1.24979675e+00 8.49560857e-01 -1.41398713e-01 -7.82048583e-01
5.10898791e-02 4.32595849e-01 2.24488661e-01 -4.57732201e-01
7.85372376e-01 -2.75558203e-01 -2.36379817e-01 -1.21227860e-01
2.04578221e-01 -4.54610527e-01 -3.00720245e-01 -2.21133143e-01
5.59221029e-01 -3.09885263e-01 2.74387449e-01 -2.84272224e-01
8.59158158e-01 -2.82879531e-01 5.23155272e-01 5.35804331e-01
-3.91481936e-01 7.75050938e-01 9.44484472e-02 -2.32471395e-02
-8.54550958e-01 -1.37598443e+00 3.48290242e-02 9.76808906e-01
-2.99188733e-01 -3.79669547e-01 -1.10109079e+00 -4.51425672e-01
-5.29545069e-01 1.25534034e+00 2.37188768e-02 1.75456777e-01
-5.51884294e-01 -3.23272422e-02 6.11636281e-01 5.70948541e-01
2.75229573e-01 -1.27435148e+00 1.44988492e-01 6.49495065e-01
-4.99347299e-01 -1.43402195e+00 -8.97399187e-01 1.92045778e-01
-3.69232148e-01 -1.91320807e-01 -7.36650169e-01 -1.19122612e+00
1.93314239e-01 -2.31607229e-01 5.90413570e-01 -7.03742623e-01
3.96797419e-01 -2.96266042e-02 -4.74639744e-01 -3.85816455e-01
-9.73073661e-01 1.57458186e-01 4.87028509e-01 1.30834416e-01
3.47239703e-01 -3.56411606e-01 -4.43887651e-01 5.02284288e-01
-5.43708861e-01 1.08875306e-02 6.19964957e-01 8.29201162e-01
6.30131960e-01 1.83801334e-02 1.06923127e+00 -4.06800121e-01
8.56173754e-01 6.08643740e-02 -4.66089249e-01 -8.73452425e-03
-4.56401020e-01 -1.37553796e-01 9.96177554e-01 -4.71808761e-01
-1.36428142e+00 7.25881085e-02 -9.20834661e-01 -1.37532309e-01
-4.81331080e-01 4.58736807e-01 -5.33643723e-01 4.19460922e-01
3.44815195e-01 3.37270081e-01 -7.29407147e-02 -4.43510354e-01
2.83538073e-01 1.56242740e+00 7.50642598e-01 -4.40576732e-01
5.45567691e-01 -4.23952162e-01 -7.28818417e-01 -1.36077416e+00
-7.84594178e-01 -3.94385934e-01 -6.39176786e-01 -2.15386271e-01
9.28308129e-01 -9.17142689e-01 -5.89829981e-01 5.27674139e-01
-1.32511675e+00 -2.65545577e-01 -2.45445013e-01 8.58621895e-01
-7.24330544e-01 3.32616270e-01 -7.84863532e-01 -1.13573265e+00
-4.36005026e-01 -1.28137076e+00 1.07019293e+00 1.21995538e-01
-5.52875996e-01 -8.64173949e-01 -7.59473592e-02 6.76761150e-01
6.19876146e-01 -2.85319418e-01 6.31648898e-01 -6.68724954e-01
-2.42746510e-02 -5.82481697e-02 2.72197157e-01 1.14779520e+00
3.85154635e-01 -1.22242935e-01 -1.53436947e+00 6.08627908e-02
3.05312097e-01 4.02507000e-03 2.19044447e-01 6.41324162e-01
7.01290488e-01 -5.69705606e-01 3.80130470e-01 1.13134831e-01
9.90263760e-01 7.09968746e-01 6.89420283e-01 -8.53049159e-02
4.43697423e-01 7.24096835e-01 8.25979173e-01 1.29315376e-01
9.00976360e-02 8.59859109e-01 -1.75927401e-01 -1.07242227e-01
-4.02116716e-01 -3.99401754e-01 8.93118739e-01 1.83310556e+00
4.09695894e-01 -4.36159700e-01 -7.30966151e-01 6.69642866e-01
-1.37831903e+00 -7.27029383e-01 -7.23545179e-02 2.13979578e+00
1.15969372e+00 5.64822435e-01 1.04845129e-01 3.83217365e-01
1.09488559e+00 1.63229331e-01 -6.02931567e-02 -1.15147460e+00
-1.22101232e-01 3.14603925e-01 1.71185344e-01 1.10283303e+00
-9.75092530e-01 1.29490936e+00 6.01475191e+00 1.10238492e+00
-1.29827356e+00 -2.58980040e-02 6.91866875e-01 2.55782694e-01
-8.45363885e-02 -3.27670157e-01 -9.22339499e-01 4.54891652e-01
1.61015296e+00 -6.69308230e-02 3.21346194e-01 7.44417310e-01
6.32449925e-01 1.53768659e-01 -8.32718611e-01 6.93555474e-01
-9.92547274e-02 -9.64541852e-01 -1.00576527e-01 -2.41682544e-01
7.33253539e-01 -2.37970859e-01 3.14305693e-01 4.63629991e-01
-1.61732659e-01 -9.15542424e-01 1.11141503e+00 3.41556162e-01
9.17783916e-01 -6.98382258e-01 1.03322530e+00 1.90788910e-01
-1.20275402e+00 2.92233884e-01 -6.67364597e-02 -1.09610051e-01
5.23328185e-01 4.04945970e-01 -1.78788364e+00 4.63818014e-01
1.60342813e-01 -1.10022962e-01 -1.86988145e-01 6.84597969e-01
-4.11368251e-01 1.15109098e+00 -2.28431225e-01 -2.43102849e-01
3.15080822e-01 1.41059309e-01 6.28904164e-01 1.61482644e+00
4.38526928e-01 -1.26255363e-01 -2.50634290e-02 5.23711145e-01
1.13570914e-01 3.41527969e-01 -2.64321387e-01 -2.99401402e-01
6.61511600e-01 9.50791299e-01 -3.87954086e-01 -2.71743268e-01
-2.51137942e-01 1.14813304e+00 -1.11140653e-01 4.35694993e-01
-7.40973830e-01 -7.33547509e-01 6.36385739e-01 -4.59454134e-02
4.64654535e-01 -4.19520825e-01 -2.32827887e-01 -7.01894701e-01
-1.01137189e-02 -8.17614257e-01 -2.58808643e-01 -9.25931454e-01
-1.32965636e+00 1.17495906e+00 -3.95635605e-01 -1.12925661e+00
-5.77818394e-01 -7.39835560e-01 -5.63447177e-01 1.26440549e+00
-1.39812076e+00 -8.80420327e-01 3.75897706e-01 1.63593829e-01
1.03455985e+00 -1.07808724e-01 1.10461390e+00 2.59400487e-01
-3.66635621e-01 8.59586596e-01 1.75916746e-01 1.82044983e-01
7.13348091e-01 -1.38244784e+00 5.14395177e-01 8.53450656e-01
2.51158237e-01 3.65770191e-01 8.89327586e-01 -3.64940941e-01
-8.51222754e-01 -1.09456646e+00 1.48207712e+00 -2.64383890e-02
6.21439934e-01 -4.14582819e-01 -7.45294809e-01 4.55081731e-01
5.72579026e-01 -3.35708380e-01 7.58771658e-01 -5.86993294e-03
2.02219322e-01 -2.43800700e-01 -1.05083060e+00 6.97875559e-01
4.48982030e-01 -7.65916288e-01 -9.52956975e-01 -2.09322318e-01
1.16422224e+00 -4.12569165e-01 -8.54824841e-01 2.95053482e-01
4.38814402e-01 -7.66785562e-01 5.67541063e-01 -1.31002337e-01
1.76013857e-02 -4.61930335e-01 -7.19159186e-01 -1.53883433e+00
-1.61435530e-01 -9.32674766e-01 1.99035108e-01 1.56799769e+00
7.01085567e-01 -4.91192251e-01 2.91853935e-01 2.83145636e-01
-7.19148636e-01 -4.75494623e-01 -8.86033535e-01 -9.54905570e-01
-5.88198639e-02 -6.76370084e-01 4.05195147e-01 6.05909586e-01
7.25030228e-02 6.36321366e-01 -3.65149200e-01 5.66667497e-01
3.83574665e-02 -2.43545786e-01 4.38226432e-01 -7.87463188e-01
-4.05715555e-01 -2.30185300e-01 -4.32692796e-01 -1.23068142e+00
2.63508826e-01 -5.47249019e-01 4.59455013e-01 -1.30181170e+00
-6.71525478e-01 -1.32332683e-01 -4.80247051e-01 1.12699695e-01
-1.99708834e-01 5.84082492e-02 3.87953579e-01 -3.87593925e-01
-9.98487249e-02 9.32885706e-01 1.09485424e+00 -8.41571426e-04
-3.37498814e-01 2.88424373e-01 -4.81857657e-01 6.85730875e-01
1.11573100e+00 -2.89191037e-01 -3.63864213e-01 -4.44293469e-01
-6.40319526e-01 4.38803315e-01 -1.01840504e-01 -1.20860827e+00
1.69446498e-01 -8.97717699e-02 1.36112526e-01 -6.53254986e-01
8.16447258e-01 -6.21231139e-01 -9.27480534e-02 1.50761113e-01
-4.42562073e-01 -1.57430023e-01 4.46415961e-01 1.35280207e-01
-6.86608374e-01 -2.40796193e-01 7.73784518e-01 3.23630780e-01
-6.02293909e-01 -3.35609674e-01 -7.85431564e-01 -1.07046589e-01
6.87042713e-01 -3.37413132e-01 -2.67109498e-02 -5.91194749e-01
-1.10367811e+00 -2.23237157e-01 -3.11562400e-02 4.91753191e-01
6.30641222e-01 -1.31327248e+00 -6.69647515e-01 4.01085496e-01
-1.81129143e-01 -3.68898571e-01 2.21666232e-01 6.08660817e-01
-2.63535291e-01 7.64278412e-01 8.90842080e-02 -3.83117348e-01
-1.21759284e+00 1.24965757e-01 1.89393967e-01 1.12114914e-01
-1.71516880e-01 8.59616756e-01 1.70045137e-01 -5.62975168e-01
4.38554853e-01 -5.68005085e-01 -3.39860306e-03 -2.32889563e-01
3.11158150e-01 1.55990347e-01 2.49585494e-01 -9.84128714e-01
-3.34228009e-01 2.89657205e-01 -2.49690432e-02 -7.55831718e-01
9.14803147e-01 -3.96711648e-01 4.56010431e-01 8.26090813e-01
1.23236191e+00 6.47026420e-01 -1.06345487e+00 -2.84357816e-02
1.27541825e-01 -2.10234672e-01 2.50411630e-01 -1.29856133e+00
-5.43094873e-01 1.07869768e+00 6.01423740e-01 3.59415054e-01
1.13783860e+00 -1.62681222e-01 1.28967750e+00 2.33828679e-01
1.17718853e-01 -1.50203848e+00 -6.61011413e-02 8.28305304e-01
1.00988853e+00 -1.00516343e+00 -1.02774620e+00 -4.00916040e-01
-1.06404781e+00 1.09671545e+00 5.77673495e-01 1.80646822e-01
6.14397764e-01 3.35222185e-01 5.70352674e-01 5.02272308e-01
-6.74101770e-01 -1.19323611e-01 4.00396377e-01 7.14676321e-01
6.77676678e-01 4.70428139e-01 -2.01540187e-01 9.19294775e-01
-8.80512655e-01 -4.34617937e-01 4.09412652e-01 4.66277212e-01
-6.66851163e-01 -1.13964283e+00 -5.28885901e-01 6.17239289e-02
-6.13102674e-01 -3.26418519e-01 -3.22582334e-01 3.76590282e-01
1.21147253e-01 1.47192430e+00 2.78628152e-02 -7.47126400e-01
4.95978922e-01 4.60450083e-01 5.28354011e-02 -6.73594713e-01
-5.17193198e-01 5.66182017e-01 4.69578981e-01 -1.08265923e-02
-2.22011004e-02 -5.09564519e-01 -1.32186186e+00 8.79380479e-02
-5.25588214e-01 5.88232160e-01 9.21274543e-01 8.30216348e-01
3.05498958e-01 6.92462385e-01 1.09990180e+00 -6.96867526e-01
-5.97883701e-01 -1.33456290e+00 -7.00637400e-01 9.53503326e-03
5.65802991e-01 -9.53021795e-02 -4.85908538e-01 2.89745778e-01] | [14.732780456542969, 6.678366661071777] |
fc10e311-d11a-4e1e-950e-b89cfc6a37a2 | 1st-solution-places-for-cvpr-2023-ug-2 | 2306.09379 | null | https://arxiv.org/abs/2306.09379v1 | https://arxiv.org/pdf/2306.09379v1.pdf | 1st Solution Places for CVPR 2023 UG$^2$+ Challenge Track 2.2-Coded Target Restoration through Atmospheric Turbulence | In this technical report, we briefly introduce the solution of our team VIELab-HUST for coded target restoration through atmospheric turbulence in CVPR 2023 UG$^2$+ Track 2.2. In this task, we propose an efficient multi-stage framework to restore a high quality image from distorted frames. Specifically, each distorted frame is initially aligned using image registration to suppress geometric distortion. We subsequently select the sharpest set of registered frames by employing a frame selection approach based on image sharpness, and average them to produce an image that is largely free of geometric distortion, albeit with blurriness. A learning-based deblurring method is then applied to remove the residual blur in the averaged image. Finally, post-processing techniques are utilized to further enhance the quality of the output image. Our framework is capable of handling different kinds of coded target dataset provided in the final testing phase, and ranked 1st on the final leaderboard. Our code will be available at https://github.com/xsqhust/Turbulence_Removal. | ['Luxin Yan', 'Yi Chang', 'Xueyao Xiao', 'Haoyue Liu', 'Shuning Cao', 'Shengqi Xu'] | 2023-06-15 | null | null | null | null | ['deblurring', 'image-registration'] | ['computer-vision', 'computer-vision'] | [ 5.35046935e-01 -5.96573710e-01 4.35787588e-01 -1.27045900e-01
-9.28240180e-01 -7.05275714e-01 3.30749065e-01 -4.28208441e-01
-1.19785979e-01 7.36941695e-01 2.37849027e-01 -1.63589641e-01
-8.19863975e-02 -4.15962636e-01 -5.21888137e-01 -9.31403875e-01
-2.76651774e-02 -3.41544330e-01 1.14950493e-01 -1.58660159e-01
4.98473555e-01 5.21020949e-01 -1.49064291e+00 3.14738691e-01
9.98852849e-01 9.29372251e-01 4.48866040e-01 1.01284134e+00
2.82152593e-01 9.73503232e-01 -4.74457324e-01 -1.85938299e-01
5.88309169e-01 -6.17093563e-01 -5.95087647e-01 5.21657825e-01
7.81438291e-01 -4.07408178e-01 -3.21174204e-01 1.23585105e+00
6.16824627e-01 3.46041620e-01 4.15047765e-01 -6.26348495e-01
-3.43630463e-01 1.31798923e-01 -6.56851947e-01 5.89742124e-01
4.68218178e-01 3.13608527e-01 5.03671706e-01 -1.27642596e+00
5.96008301e-01 1.11166322e+00 5.52657545e-01 3.26590955e-01
-1.15886676e+00 -5.22764325e-01 -1.52482405e-01 4.79840130e-01
-1.34589159e+00 -1.02899194e+00 6.95768893e-01 -3.86768699e-01
5.55002451e-01 5.66531181e-01 5.13357937e-01 6.96146131e-01
2.44415075e-01 3.84948254e-01 1.28775978e+00 -4.78047669e-01
2.60452479e-02 -3.04607928e-01 -9.00619403e-02 4.10287946e-01
9.56358537e-02 2.82679528e-01 -5.53627014e-01 -1.38252974e-02
6.50599957e-01 -1.33491322e-01 -9.89557505e-01 -6.12988062e-02
-1.25726902e+00 3.59149724e-01 3.74390900e-01 2.37335205e-01
-4.06252980e-01 -5.06626582e-03 1.60157070e-01 4.46832597e-01
8.86022389e-01 3.76279473e-01 -2.82254398e-01 -8.87788907e-02
-1.29657483e+00 2.40815908e-01 4.01959509e-01 6.35404646e-01
6.03796005e-01 1.51886865e-01 -2.89318919e-01 9.52456057e-01
1.62274078e-01 6.66267693e-01 2.25883558e-01 -1.49618804e+00
2.05357581e-01 -1.78933188e-01 4.08227205e-01 -9.51245070e-01
1.23316400e-01 -4.65873599e-01 -1.02837110e+00 4.50325072e-01
1.00607738e-01 -1.21936657e-01 -7.23164737e-01 1.12496412e+00
2.43002445e-01 6.20316744e-01 1.88473523e-01 1.21705997e+00
6.16673231e-01 7.63174474e-01 -4.56589311e-01 -6.69135332e-01
1.01385331e+00 -1.02954638e+00 -7.82282531e-01 -1.82030722e-02
1.00145955e-02 -1.40538764e+00 5.94579756e-01 6.62391186e-01
-1.37120938e+00 -7.98146248e-01 -9.46984887e-01 7.68175125e-02
2.74465680e-01 1.86452389e-01 4.60278131e-02 5.91954768e-01
-1.43515599e+00 7.76480377e-01 -7.39451289e-01 -1.89605594e-01
3.29894871e-01 -3.77731100e-02 -2.24688083e-01 -4.61404294e-01
-8.38270009e-01 9.70579743e-01 2.56729145e-02 2.55799025e-01
-9.34665084e-01 -6.88966572e-01 -8.28010678e-01 -3.05595636e-01
1.95894137e-01 -6.81103885e-01 1.06561995e+00 -9.41158354e-01
-1.55218434e+00 7.14846015e-01 -4.94872779e-01 -4.79515731e-01
7.10744023e-01 -5.13487637e-01 -3.39732856e-01 2.50509828e-01
3.90256345e-02 2.34936699e-01 1.40863132e+00 -1.41921067e+00
-6.36479020e-01 3.91572900e-03 -2.67974198e-01 5.62003851e-01
8.80914927e-02 2.93004364e-01 -6.63997769e-01 -9.47419584e-01
2.80617952e-01 -8.26518357e-01 -1.08784162e-01 -2.31752560e-01
-3.06838930e-01 5.60311317e-01 9.22275424e-01 -1.10323083e+00
1.03871489e+00 -2.37528229e+00 4.05420989e-01 -1.95548445e-01
3.24823022e-01 3.67254913e-01 -1.98321179e-01 1.57800138e-01
-2.56082952e-01 -2.50223935e-01 -5.04486680e-01 -5.76455653e-01
-5.96627891e-01 -3.16579401e-01 -5.34513474e-01 7.86920965e-01
6.39414415e-02 3.56992602e-01 -7.61701822e-01 -1.26361683e-01
6.42877996e-01 5.80646813e-01 -3.62431794e-01 3.49085748e-01
1.97395325e-01 8.62390101e-01 -2.78397780e-02 5.71952164e-01
1.19727314e+00 9.22672078e-02 -5.06055988e-02 -5.37259281e-01
-4.72354561e-01 -1.29131600e-01 -1.10061860e+00 1.51241481e+00
-4.67282265e-01 9.81188297e-01 5.02720177e-01 -7.47472644e-01
8.56227636e-01 2.58696437e-01 6.37121975e-01 -3.25611860e-01
2.27734502e-02 2.77662575e-01 -1.31539181e-01 -4.47659701e-01
6.59784138e-01 1.38802469e-01 4.68666285e-01 3.98745984e-02
-3.09379935e-01 -6.46400869e-01 2.20340028e-01 1.44669920e-01
1.14619255e+00 2.55480230e-01 -3.13444529e-03 -1.99841380e-01
8.55695963e-01 -8.92182440e-02 4.83599842e-01 7.07635283e-01
-4.91907388e-01 1.33182776e+00 -2.20732763e-01 -4.98351127e-01
-1.11610067e+00 -9.87297297e-01 -1.27833977e-01 5.72357595e-01
3.03709090e-01 -2.81479269e-01 -7.72109866e-01 -2.08778650e-01
-4.82777864e-01 5.58439136e-01 -2.64671981e-01 3.99907231e-02
-6.68929935e-01 -7.15130389e-01 2.22936437e-01 -2.34660640e-01
8.14376116e-01 -9.44465995e-01 -5.18643081e-01 8.36236924e-02
-6.40556455e-01 -1.15980661e+00 -6.80667281e-01 -1.68811560e-01
-7.66158104e-01 -1.14956522e+00 -9.62058365e-01 -5.76491475e-01
5.58266938e-01 8.59642565e-01 9.32575464e-01 3.23016942e-01
-2.51983523e-01 2.61397511e-01 -5.74375987e-01 1.48330167e-01
-4.10371900e-01 -6.26326680e-01 7.64355883e-02 3.19084167e-01
-3.68185788e-01 -4.20481056e-01 -7.73197114e-01 2.82432646e-01
-8.71157944e-01 2.02385664e-01 2.17201173e-01 7.82606483e-01
5.10445714e-01 4.95434672e-01 -1.23675935e-01 -3.85455012e-01
4.78342950e-01 -1.86199434e-02 -7.00999677e-01 -5.62368780e-02
-2.08431870e-01 -4.46231961e-01 5.22235513e-01 -2.23490044e-01
-1.27505720e+00 -3.78495292e-03 1.52898699e-01 -6.47032022e-01
-9.18535367e-02 2.50486016e-01 6.98536187e-02 -3.84320110e-01
6.27667546e-01 4.29821759e-01 -1.73411798e-02 -4.75443631e-01
3.29171300e-01 5.66790044e-01 8.47716987e-01 -2.49890670e-01
1.11620140e+00 5.84967077e-01 -1.67586207e-01 -9.44467485e-01
-7.58020222e-01 -4.86816943e-01 -5.77467740e-01 -5.37759840e-01
7.79470623e-01 -1.20331955e+00 -2.49577835e-01 1.00699151e+00
-1.25528204e+00 -4.39334244e-01 -1.50340959e-01 6.43198252e-01
-5.02569556e-01 6.47884846e-01 -6.86421573e-01 -6.47696197e-01
-4.68711585e-01 -1.33505082e+00 8.89861405e-01 3.84872973e-01
2.48509392e-01 -6.36745930e-01 1.27619579e-01 5.75990498e-01
7.48408735e-01 3.15026902e-02 1.55078575e-01 1.09423071e-01
-6.29403651e-01 1.03301920e-01 -2.70900965e-01 6.50596321e-01
4.13346529e-01 5.62725551e-02 -1.02729666e+00 -6.74116492e-01
4.02587354e-01 1.06021531e-01 1.08196461e+00 8.59579921e-01
9.03121531e-01 -8.59060362e-02 5.62134646e-02 8.75401616e-01
1.32100654e+00 1.21331938e-01 9.19380724e-01 5.43788910e-01
5.66978931e-01 2.93331116e-01 7.67601669e-01 3.69328201e-01
6.18647672e-02 8.25733244e-01 4.18059736e-01 -2.11219802e-01
-6.88531876e-01 5.14681816e-01 7.22715378e-01 5.74983656e-01
-3.19060147e-01 -3.51628095e-01 -6.52349651e-01 4.46786404e-01
-1.57915068e+00 -1.16563594e+00 -4.03878987e-01 2.21880198e+00
7.01743841e-01 -1.60787463e-01 -2.99569726e-01 1.05263121e-01
1.03306472e+00 3.56450468e-01 -2.78915048e-01 -5.68529926e-02
-4.51330662e-01 2.35416025e-01 4.35839117e-01 9.07330573e-01
-1.24799478e+00 1.07892835e+00 5.61454105e+00 7.82983482e-01
-1.17416561e+00 9.95309800e-02 7.29395032e-01 -9.95532200e-02
1.66620627e-01 7.85091966e-02 -2.76646286e-01 5.10076761e-01
7.97461689e-01 -1.23703867e-01 7.07666874e-01 2.03454539e-01
8.17926228e-01 -2.70955563e-01 -3.48881632e-01 1.00354362e+00
2.20348701e-01 -1.27133691e+00 -3.48105907e-01 -2.90879626e-02
8.34906638e-01 1.27125263e-01 1.55510023e-01 -1.45198613e-01
2.08046630e-01 -8.39722931e-01 7.70386696e-01 7.21495688e-01
8.96679640e-01 -4.99525011e-01 6.72650397e-01 6.23004399e-02
-1.19496953e+00 7.87906125e-02 -3.81072432e-01 6.03175201e-02
2.99341768e-01 8.64029408e-01 -3.43822718e-01 1.03964627e+00
9.13275003e-01 9.78235245e-01 -5.10021806e-01 1.41996074e+00
-1.16690524e-01 4.91447270e-01 -3.27288657e-02 8.99783254e-01
-2.86689520e-01 -3.82792443e-01 9.79663908e-01 1.21309173e+00
8.56857955e-01 4.27672684e-01 1.73037350e-01 2.91191876e-01
1.55272543e-01 -9.83570069e-02 -4.35713887e-01 5.29686630e-01
3.50218952e-01 1.36858618e+00 -5.31807601e-01 -4.72183764e-01
-1.97834581e-01 1.43177330e+00 -1.60194710e-01 4.47250575e-01
-7.82197535e-01 -8.66368264e-02 7.68453777e-01 -5.13321050e-02
3.45508635e-01 -2.79435337e-01 -1.23196766e-01 -1.50758457e+00
-2.16903035e-02 -1.13126767e+00 1.24909818e-01 -1.18936241e+00
-9.56612110e-01 8.73795807e-01 -2.36701339e-01 -1.65368271e+00
6.77386969e-02 -2.23140955e-01 -6.47949219e-01 1.06791985e+00
-1.66171372e+00 -7.61105537e-01 -6.18860066e-01 6.07809842e-01
9.42648113e-01 -1.11049101e-01 4.99400556e-01 4.10354614e-01
-6.55674458e-01 -4.12529297e-02 3.20414305e-01 -1.92027971e-01
1.13486755e+00 -1.03692579e+00 3.24983358e-01 1.56508195e+00
-1.31183222e-01 4.33715194e-01 9.87881184e-01 -7.18715727e-01
-1.26641035e+00 -1.20535553e+00 6.37237430e-01 -1.02738179e-01
4.11651939e-01 2.06526592e-01 -9.34508562e-01 3.79685998e-01
4.81943190e-01 2.73710042e-01 2.24988103e-01 -6.77228868e-01
-1.27287537e-01 -5.97446524e-02 -1.06288803e+00 4.12829787e-01
7.26243973e-01 -2.95238316e-01 -3.55389893e-01 3.43001962e-01
6.35410845e-01 -9.03486609e-01 -6.81327343e-01 4.50439602e-01
1.64735466e-01 -1.30106485e+00 1.10459948e+00 1.18036672e-01
4.45903927e-01 -9.31505620e-01 -3.38586777e-01 -1.47106981e+00
-5.27854741e-01 -9.96195257e-01 4.78786975e-02 1.13211250e+00
7.41165429e-02 -3.56731892e-01 3.23555827e-01 1.43750817e-01
-3.25999737e-01 -5.53488433e-02 -7.62798250e-01 -6.05162203e-01
-4.03458983e-01 -2.96325624e-01 2.42008809e-02 8.25134695e-01
-4.57533032e-01 -1.13265254e-01 -8.44027460e-01 4.60857213e-01
1.07032228e+00 1.95227563e-01 8.39961529e-01 -7.66294837e-01
-1.82447493e-01 -1.71367988e-01 -2.34982908e-01 -9.25724387e-01
-1.90707847e-01 -4.62057382e-01 3.43311340e-01 -1.33130610e+00
9.45146754e-02 -9.94470809e-03 -1.53663531e-01 2.87226140e-02
-3.65284503e-01 7.56738305e-01 5.32042265e-01 6.38269246e-01
-3.72705311e-01 3.59024793e-01 1.24958396e+00 1.59132443e-02
-2.13588148e-01 8.05246010e-02 -5.63670993e-01 5.70808589e-01
9.24006760e-01 -1.99987873e-01 -4.14028950e-02 -5.88987112e-01
-4.41960335e-01 1.99520707e-01 5.70140123e-01 -1.20982146e+00
1.32772222e-01 -1.69052958e-01 4.92405057e-01 -3.91795784e-01
4.95542407e-01 -6.85539603e-01 4.05602723e-01 4.53787386e-01
-7.71255344e-02 -3.19725438e-03 2.50674874e-01 3.27503800e-01
-3.17279607e-01 -5.47424816e-02 1.15387392e+00 -2.92119524e-03
-5.85549772e-01 2.80211538e-01 -4.03533876e-01 -3.46093625e-01
7.49929190e-01 -5.24089709e-02 -4.40723658e-01 -5.61044693e-01
-7.65948355e-01 -1.56293899e-01 7.69771338e-01 1.33179054e-01
6.85000479e-01 -9.75601315e-01 -1.14835453e+00 8.04646090e-02
-3.53784829e-01 -3.31758976e-01 5.54952264e-01 1.04894245e+00
-8.57266068e-01 -1.09521955e-01 -6.51413351e-02 -6.44136071e-01
-1.54811108e+00 2.53759861e-01 4.85310256e-01 -1.14113018e-01
-8.57576132e-01 8.84398222e-01 7.69637823e-02 1.65656343e-01
-7.47887492e-02 4.90299277e-02 -1.34154856e-01 -2.36828327e-01
9.75688338e-01 4.11269009e-01 2.51896471e-01 -1.02182734e+00
-3.45764518e-01 7.70985246e-01 3.05870380e-02 -3.33913386e-01
1.44337761e+00 -7.46275246e-01 -2.77133077e-01 -7.84705728e-02
1.01483822e+00 3.42291951e-01 -1.62719226e+00 -1.97909787e-01
-3.32067788e-01 -1.00093031e+00 3.82303715e-01 -6.55526280e-01
-1.32352734e+00 5.41993022e-01 9.20657516e-01 2.18139753e-01
1.66180325e+00 -4.98901099e-01 6.28773093e-01 -1.28123447e-01
8.94781947e-03 -4.56166446e-01 1.44449333e-02 6.24939084e-01
1.05764067e+00 -1.06063437e+00 3.16355526e-01 -4.97026324e-01
-4.46517974e-01 1.26547825e+00 2.83805192e-01 -1.36216775e-01
3.80156517e-01 2.89598882e-01 4.11289394e-01 3.15772891e-02
-5.79111457e-01 -3.25081080e-01 1.58636063e-01 6.79536045e-01
5.02504110e-01 -1.98947385e-01 -2.36603394e-01 -1.84505895e-01
-1.44616202e-01 8.57915357e-03 8.38078976e-01 7.46431828e-01
-6.53667212e-01 -9.45280671e-01 -9.35263813e-01 -1.16894133e-01
-5.63655317e-01 -4.87047762e-01 -2.31197733e-03 7.52635226e-02
2.74258517e-02 1.41905713e+00 -1.62317216e-01 -3.21879894e-01
3.01352471e-01 -3.39883298e-01 4.59432334e-01 -3.62478346e-01
-4.30847943e-01 5.47439158e-01 3.69240455e-02 -7.30527639e-01
-6.44887388e-01 -8.35024416e-01 -6.83912218e-01 -5.11254072e-01
-1.97670117e-01 1.24350451e-01 4.01584804e-01 6.05912149e-01
3.35443050e-01 4.57103550e-01 9.13493872e-01 -1.46872413e+00
4.76159640e-02 -8.60331237e-01 -3.78050089e-01 3.35842460e-01
8.28576207e-01 -3.50946963e-01 -6.97233438e-01 7.41422236e-01] | [11.231415748596191, -2.237999677658081] |
9e18ed22-e7cf-45d7-89b3-45365a0b86fa | learning-with-noisy-labels-over-imbalanced | 2211.08722 | null | https://arxiv.org/abs/2211.08722v1 | https://arxiv.org/pdf/2211.08722v1.pdf | Learning with Noisy Labels over Imbalanced Subpopulations | Learning with Noisy Labels (LNL) has attracted significant attention from the research community. Many recent LNL methods rely on the assumption that clean samples tend to have "small loss". However, this assumption always fails to generalize to some real-world cases with imbalanced subpopulations, i.e., training subpopulations varying in sample size or recognition difficulty. Therefore, recent LNL methods face the risk of misclassifying those "informative" samples (e.g., hard samples or samples in the tail subpopulations) into noisy samples, leading to poor generalization performance. To address the above issue, we propose a novel LNL method to simultaneously deal with noisy labels and imbalanced subpopulations. It first leverages sample correlation to estimate samples' clean probabilities for label correction and then utilizes corrected labels for Distributionally Robust Optimization (DRO) to further improve the robustness. Specifically, in contrast to previous works using classification loss as the selection criterion, we introduce a feature-based metric that takes the sample correlation into account for estimating samples' clean probabilities. Then, we refurbish the noisy labels using the estimated clean probabilities and the pseudo-labels from the model's predictions. With refurbished labels, we use DRO to train the model to be robust to subpopulation imbalance. Extensive experiments on a wide range of benchmarks demonstrate that our technique can consistently improve current state-of-the-art robust learning paradigms against noisy labels, especially when encountering imbalanced subpopulations. | ['Jianhua Yao', 'Bingzhe Wu', 'Zongbo Han', 'Bing He', 'Yu Zhao', 'Mingcai Chen'] | 2022-11-16 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [ 2.26152807e-01 -1.85293376e-01 -4.57328498e-01 -6.03053927e-01
-1.39535713e+00 -4.32151705e-01 3.08011323e-01 5.02933860e-01
-2.86562651e-01 9.54754412e-01 2.55451817e-02 -5.01498394e-02
-8.24754089e-02 -6.87974632e-01 -6.70379460e-01 -1.13052130e+00
3.86698961e-01 3.34862202e-01 -1.61379918e-01 3.01958740e-01
1.83756262e-01 9.75920409e-02 -1.55465150e+00 8.55637714e-02
1.32839143e+00 1.04939103e+00 -3.53851944e-01 4.50094827e-02
-1.33993670e-01 5.37759423e-01 -8.95777702e-01 -3.81342590e-01
2.60266304e-01 -2.64442891e-01 -5.50360441e-01 7.82879442e-02
4.90898222e-01 3.80476080e-02 1.99601188e-01 1.42683661e+00
7.36768067e-01 1.69012234e-01 8.12590122e-01 -1.25956857e+00
-6.07081413e-01 7.98021615e-01 -9.09745038e-01 -1.81014510e-03
-3.24495807e-02 1.97942927e-01 9.59030509e-01 -9.11651909e-01
2.31936619e-01 1.41814899e+00 9.62592304e-01 5.53627193e-01
-1.51909006e+00 -1.00147271e+00 4.77184564e-01 -1.56948566e-01
-1.61170626e+00 -5.22515178e-01 6.66120410e-01 -4.79532301e-01
3.25903565e-01 4.25749004e-01 9.60580185e-02 1.10303450e+00
1.11729935e-01 7.92042315e-01 1.25637364e+00 -2.34773323e-01
4.10612762e-01 2.14511573e-01 5.50900280e-01 2.05727041e-01
5.52436233e-01 5.04956245e-02 -4.31263328e-01 -2.98840821e-01
-8.31283554e-02 -5.90896085e-02 -3.02242637e-01 -1.40378162e-01
-1.09422135e+00 7.76754439e-01 2.35924989e-01 -8.11291188e-02
3.29869874e-02 -9.57521796e-02 6.18010283e-01 1.34957403e-01
1.00022042e+00 3.93504858e-01 -4.21401054e-01 1.17325194e-01
-9.71825302e-01 2.46810108e-01 5.41464925e-01 7.11892009e-01
7.17278719e-01 -1.78640768e-01 -5.23075640e-01 1.23020935e+00
2.36498624e-01 5.78883767e-01 6.19236469e-01 -6.31361783e-01
7.06025183e-01 8.17676127e-01 6.50665984e-02 -9.23693776e-01
-3.78484815e-01 -8.83853853e-01 -1.06270516e+00 -4.08378169e-02
6.13076091e-01 -3.21772359e-02 -7.77572811e-01 1.91371810e+00
6.14420652e-01 2.46034533e-01 9.90090985e-03 7.52657652e-01
6.47242427e-01 4.34230477e-01 1.48489639e-01 -2.57029295e-01
1.01055706e+00 -7.85751522e-01 -7.02796459e-01 -1.29215404e-01
8.56005907e-01 -5.54181218e-01 1.38803887e+00 3.98483366e-01
-6.40475750e-01 -4.54691023e-01 -9.66307759e-01 8.57500732e-02
-4.23923165e-01 2.29677528e-01 2.73391545e-01 9.58703458e-01
-4.97340173e-01 5.50536990e-01 -6.06657803e-01 -1.13330064e-02
7.09785521e-01 3.98861974e-01 -1.25924081e-01 -2.49851555e-01
-9.39412594e-01 2.68760383e-01 3.46446931e-01 1.64532244e-01
-7.35614896e-01 -9.88518715e-01 -9.40306365e-01 -3.91001254e-02
3.58279824e-01 -4.95290816e-01 1.02595389e+00 -9.09639955e-01
-1.22479868e+00 7.18212783e-01 -2.42848530e-01 -2.35064358e-01
7.17305660e-01 -1.56258970e-01 -2.98082590e-01 -4.23512280e-01
2.46895432e-01 4.88588095e-01 6.59597337e-01 -1.49048424e+00
-6.63377762e-01 -4.68831509e-01 -2.91472763e-01 5.43945888e-03
-5.39879262e-01 -3.26989412e-01 -9.99183431e-02 -8.97111714e-01
2.58067369e-01 -7.79384017e-01 -6.24795854e-02 -2.69883871e-01
-8.33504736e-01 -4.76408601e-01 6.18944585e-01 -3.08590919e-01
1.42493546e+00 -2.22647119e+00 -1.78295299e-01 3.21463376e-01
3.10004979e-01 2.00767547e-01 -2.39656150e-01 -8.25992003e-02
-6.57222569e-02 4.27485377e-01 -3.72404099e-01 -7.97903717e-01
1.95894226e-01 1.87453292e-02 -2.73113787e-01 7.87057936e-01
3.53246331e-01 5.96991122e-01 -1.07147098e+00 -5.33764541e-01
-1.33575112e-01 2.76761115e-01 -5.51488996e-01 1.53495431e-01
-4.22617719e-02 5.67237020e-01 -3.21184129e-01 9.06494677e-01
1.03950119e+00 -2.03775287e-01 -1.63549080e-01 -1.85025290e-01
1.01214230e-01 2.16329589e-01 -1.28190577e+00 1.02668965e+00
-4.49417830e-01 1.62260145e-01 -1.21822186e-01 -1.06637275e+00
1.02651274e+00 -6.64523095e-02 2.72089511e-01 -3.87843400e-01
2.38371193e-01 5.05540848e-01 -2.19844982e-01 -4.22733098e-01
3.20626467e-01 -3.56287867e-01 -2.45011598e-01 3.84221971e-01
-2.80894667e-01 9.07515287e-02 4.47991528e-02 -2.12243661e-01
8.07702303e-01 -6.42185435e-02 8.04284438e-02 -2.21967548e-01
5.22766888e-01 -3.99504691e-01 1.08087945e+00 8.92494202e-01
-3.83436978e-01 8.49773109e-01 5.57830155e-01 -8.15710127e-02
-7.32978582e-01 -1.06085861e+00 -5.24851382e-01 1.01602232e+00
2.19490498e-01 -1.85103387e-01 -7.52789080e-01 -1.05093420e+00
1.16739839e-01 6.97669208e-01 -6.85920656e-01 -5.36795139e-01
-2.30900660e-01 -1.40446472e+00 6.29745722e-01 3.02602589e-01
3.00767928e-01 -7.26714492e-01 9.91065651e-02 3.21872905e-03
-2.52922505e-01 -7.96954453e-01 -6.42141104e-01 3.09609145e-01
-7.62390018e-01 -1.00252831e+00 -6.05940759e-01 -6.64050102e-01
9.75787699e-01 1.17971636e-01 1.12685025e+00 2.11904105e-02
-5.27489297e-02 -1.60629615e-01 -4.00210917e-01 -3.48966539e-01
-4.10974473e-01 3.22658271e-01 2.10789174e-01 2.36531779e-01
4.82820123e-01 -2.65276670e-01 -3.16250056e-01 5.32102644e-01
-1.10634267e+00 -3.92759472e-01 3.52867305e-01 1.22011697e+00
1.00293291e+00 3.80029589e-01 1.22970915e+00 -1.23381937e+00
5.46009541e-01 -8.26405168e-01 -4.69491750e-01 2.87151098e-01
-6.21375084e-01 1.01640441e-01 9.01903391e-01 -7.66226351e-01
-9.09893990e-01 -2.02215225e-01 -1.60360664e-01 -2.59212881e-01
-2.12380856e-01 3.83268923e-01 -6.69823766e-01 1.12739123e-01
5.13800621e-01 -5.80093972e-02 -1.72983482e-01 -6.54212415e-01
-6.31432682e-02 1.02965724e+00 2.52444804e-01 -7.80983806e-01
6.40505135e-01 4.57314700e-01 -8.63444507e-02 -3.77176523e-01
-1.51675832e+00 -4.67658550e-01 -3.44950914e-01 6.47937432e-02
2.78892010e-01 -9.59223926e-01 -6.43515289e-01 8.37321103e-01
-6.53523505e-01 -3.42925102e-01 -5.02991319e-01 4.77022499e-01
-2.93360800e-01 4.63694900e-01 -3.96367669e-01 -9.84640300e-01
-2.78749108e-01 -1.30625594e+00 1.37192070e+00 3.80407959e-01
-1.00389369e-01 -9.46126282e-01 -9.51479282e-03 4.60834295e-01
1.67157844e-01 3.02278042e-01 1.03041768e+00 -9.71766770e-01
-9.50492248e-02 -2.99808949e-01 -3.25636536e-01 7.60394931e-01
3.37017447e-01 -1.00845501e-01 -1.12984598e+00 -4.59024489e-01
-1.63640007e-02 -5.76246440e-01 9.95961249e-01 3.19974661e-01
1.45428336e+00 -2.23572165e-01 -3.74415249e-01 6.66786432e-01
1.25303435e+00 -2.12822825e-01 2.77295619e-01 1.27028197e-01
7.34801590e-01 7.08935499e-01 8.38951170e-01 5.11838496e-01
3.74313205e-01 4.53233898e-01 2.95973569e-01 7.44912922e-02
-8.82048234e-02 -3.70736748e-01 4.70855415e-01 9.23685431e-01
6.84001744e-01 -3.78977239e-01 -7.82969713e-01 4.78114575e-01
-1.75516939e+00 -5.25412917e-01 -4.55406355e-03 2.62414670e+00
1.24070680e+00 1.64859086e-01 2.45814864e-02 2.62389123e-01
1.01398623e+00 1.01590320e-03 -1.00766063e+00 1.64799839e-02
-3.63798916e-01 -1.72122449e-01 5.06353974e-01 3.91239166e-01
-1.30612922e+00 6.49894178e-01 5.54827738e+00 1.35280812e+00
-1.04662764e+00 1.22691050e-01 1.24049294e+00 -1.72435895e-01
-4.82202888e-01 -2.15956137e-01 -1.20466256e+00 9.15810227e-01
6.91288054e-01 1.11919232e-01 1.32176071e-01 6.13238990e-01
2.69820899e-01 -1.39668643e-01 -1.04218352e+00 9.67857897e-01
1.28329739e-01 -6.12594128e-01 -3.26763056e-02 -3.04418597e-02
1.02355146e+00 -1.95516527e-01 3.89668912e-01 5.16541958e-01
2.55339056e-01 -1.08315253e+00 7.03835785e-01 5.42015076e-01
7.27977574e-01 -8.83738220e-01 1.20282352e+00 5.37605762e-01
-8.14279914e-01 -2.58929610e-01 -6.17026627e-01 1.74617320e-01
-1.46259323e-01 1.36630404e+00 -4.66777802e-01 3.38791162e-01
6.18328929e-01 7.91842222e-01 -8.16917300e-01 1.27049899e+00
8.37374851e-02 9.23652709e-01 -4.84587729e-01 1.60882995e-01
-3.68625671e-02 -1.66458726e-01 4.04224694e-01 1.03183210e+00
4.46579039e-01 -3.91380996e-01 3.45662624e-01 7.55990922e-01
-4.37034726e-01 3.78766149e-01 -3.88274014e-01 2.48975679e-01
7.47952998e-01 1.05323601e+00 -5.76309681e-01 -2.53502250e-01
-8.73864442e-02 5.11946797e-01 4.50750560e-01 3.18436623e-01
-7.60342002e-01 -4.71592456e-01 6.42696619e-01 -1.17475115e-01
1.92857385e-02 3.44738752e-01 -6.26700938e-01 -1.26565218e+00
1.98170274e-01 -9.52704430e-01 3.91565382e-01 -1.03157468e-01
-1.95791137e+00 3.48159611e-01 -2.09512815e-01 -1.33209789e+00
2.86543489e-01 -2.89166301e-01 -4.18642223e-01 7.36774564e-01
-1.75764835e+00 -9.31867242e-01 -2.73760021e-01 3.90727669e-02
4.81445611e-01 9.89889055e-02 3.66840273e-01 5.99507630e-01
-1.17188394e+00 1.25558138e+00 4.98920053e-01 -1.22994399e-02
1.12088227e+00 -1.28891039e+00 -2.03188453e-02 5.26864350e-01
-2.93805987e-01 5.15840828e-01 5.09234428e-01 -6.19912028e-01
-8.43280554e-01 -1.65657306e+00 9.21705544e-01 -4.20194447e-01
3.92110556e-01 -4.49980229e-01 -1.12968278e+00 4.22629714e-01
-4.36306179e-01 3.77785176e-01 9.95564640e-01 7.59675726e-02
-5.86942077e-01 -5.05166769e-01 -1.38540947e+00 4.38254714e-01
9.36800480e-01 -2.53173351e-01 -2.57037431e-01 5.05585968e-01
6.95430100e-01 -3.11291367e-01 -8.70187759e-01 5.96405625e-01
3.26906502e-01 -7.21674383e-01 9.06173050e-01 -4.79047924e-01
1.37058064e-01 -4.66617405e-01 -1.33923769e-01 -1.50804555e+00
4.58450435e-04 -2.72081584e-01 1.54125333e-01 1.89319599e+00
5.07200241e-01 -8.46286178e-01 6.42298877e-01 4.62711006e-01
-1.24867178e-01 -8.83121848e-01 -1.14194274e+00 -9.81068075e-01
3.44538063e-01 -3.51554275e-01 9.07862544e-01 8.74156356e-01
-2.06728712e-01 6.58367798e-02 -4.14656878e-01 1.51073784e-01
8.10116887e-01 1.42071351e-01 6.92308426e-01 -1.36620891e+00
-2.67528910e-02 -5.06046653e-01 -1.46200404e-01 -9.68919873e-01
6.67297184e-01 -1.02355683e+00 5.06771088e-01 -9.21810091e-01
3.71446729e-01 -1.04012179e+00 -5.61694801e-01 4.11959141e-01
-8.05858612e-01 6.02861524e-01 -6.07649237e-02 2.46848911e-01
-9.50846314e-01 8.21087778e-01 1.03131473e+00 -3.32875162e-01
-2.35060185e-01 2.63454854e-01 -1.01899981e+00 6.29603148e-01
8.87830496e-01 -8.83374333e-01 -1.64895087e-01 -7.96796530e-02
2.77081877e-01 -7.06274688e-01 1.36695415e-01 -1.01541972e+00
-7.11556748e-02 -1.98316902e-01 2.71418452e-01 -4.06405717e-01
-1.56036578e-02 -6.23601496e-01 -1.26610234e-01 2.13332251e-01
-6.71035707e-01 -2.82003760e-01 -6.63642660e-02 6.67180061e-01
-2.83841580e-01 -3.85715544e-01 9.26477015e-01 3.26211184e-01
1.27763316e-01 1.96071014e-01 -1.16466038e-01 5.07026672e-01
9.10901606e-01 1.56469010e-02 -5.17762780e-01 -7.73420334e-02
-3.66531581e-01 6.37699664e-01 5.84808230e-01 2.43504003e-01
2.60189325e-01 -1.31832898e+00 -8.38855326e-01 2.07954451e-01
4.04842466e-01 4.14908439e-01 2.64274001e-01 1.03781831e+00
-5.40773980e-02 6.44145384e-02 4.86973464e-01 -7.15631366e-01
-9.08208430e-01 5.62287927e-01 3.71091783e-01 -4.46257710e-01
-1.37553081e-01 7.76491284e-01 3.21424782e-01 -8.98180664e-01
4.54098493e-01 -5.19239008e-01 -1.31477296e-01 3.05582047e-01
4.72240448e-01 5.58200955e-01 1.96607471e-01 -6.75810277e-01
-2.66752273e-01 6.33593559e-01 -2.99724285e-02 4.37892199e-01
1.12212300e+00 -3.22499722e-01 -8.73296484e-02 8.12010586e-01
1.38458693e+00 2.43528426e-01 -1.30600262e+00 -3.67317110e-01
1.55295447e-01 -5.49493432e-01 -8.58037919e-02 -7.78606892e-01
-1.22698689e+00 9.12699521e-01 6.02060914e-01 1.23233393e-01
1.11292183e+00 -1.25578254e-01 8.62050772e-01 1.11813441e-01
1.05507664e-01 -1.09620225e+00 -1.21132538e-01 3.99872899e-01
3.62586230e-01 -1.50701368e+00 -4.54219505e-02 -4.98199046e-01
-3.83458614e-01 7.77797163e-01 6.83428347e-01 1.10243723e-01
5.92054427e-01 1.45919695e-01 2.43513927e-01 2.72939593e-01
-4.58881885e-01 6.12478666e-02 2.81468511e-01 5.95931709e-01
3.00160259e-01 1.64976984e-01 -3.94040048e-01 1.02716446e+00
-8.96750838e-02 -1.24082230e-01 3.22760880e-01 4.91517037e-01
-2.61461437e-01 -1.11465466e+00 -5.77096045e-01 8.96620929e-01
-6.00275159e-01 3.40907811e-03 -1.39497101e-01 3.70827138e-01
5.42210519e-01 1.09484470e+00 5.73149100e-02 -2.56023824e-01
3.38112831e-01 1.76247552e-01 -7.81440511e-02 -7.19701409e-01
-7.80530334e-01 -4.44084518e-02 -2.68528640e-01 -2.65974075e-01
-3.13766778e-01 -6.79790914e-01 -9.75514293e-01 -1.01917334e-01
-7.67097712e-01 2.37824678e-01 3.41798127e-01 8.95861745e-01
4.41210955e-01 4.42943484e-01 7.90426493e-01 -6.61095738e-01
-1.01630485e+00 -9.63951230e-01 -8.18682909e-01 6.85694814e-01
5.55138767e-01 -8.18025589e-01 -8.58440697e-01 -2.53919810e-01] | [9.439525604248047, 3.990262269973755] |
51a741e3-7d36-4436-8c59-ee30eb7b2c90 | learning-based-robust-speaker-counting-and | 2303.06867 | null | https://arxiv.org/abs/2303.06867v1 | https://arxiv.org/pdf/2303.06867v1.pdf | Learning-based Robust Speaker Counting and Separation with the Aid of Spatial Coherence | A two-stage approach is proposed for speaker counting and speech separation in noisy and reverberant environments. A spatial coherence matrix (SCM) is computed using whitened relative transfer functions (wRTFs) across time frames. The global activity functions of each speaker are estimated on the basis of a simplex constructed using the eigenvectors of the SCM, while the local coherence functions are computed from the coherence between the wRTFs of a time-frequency bin and the global activity function-weighted RTF of the target speaker. In speaker counting, we use the eigenvalues of the SCM and the maximum similarity of the interframe global activity distributions between two speakers as the input features to the speaker counting network (SCnet). In speaker separation, a global and local activity-driven network (GLADnet) is utilized to estimate a speaker mask, which is particularly useful for highly overlapping speech signals. Experimental results obtained from the real meeting recordings demonstrated the superior speaker counting and speaker separation performance achieved by the proposed learning-based system without prior knowledge of the array configurations. | ['Mingsian Bai', 'Yicheng Hsu'] | 2023-03-13 | null | null | null | null | ['speech-separation', 'speaker-separation'] | ['speech', 'speech'] | [ 2.53924757e-01 -3.61235648e-01 2.83708900e-01 -1.89283162e-01
-8.93860579e-01 -3.46640825e-01 3.67579788e-01 -1.25156865e-01
-3.26223254e-01 4.47171539e-01 5.78052282e-01 1.23316415e-01
-3.14573109e-01 -9.80915874e-02 -6.52892143e-02 -1.11858773e+00
-4.50993001e-01 5.06423227e-02 -6.05048798e-02 1.75218970e-01
2.40780741e-01 4.67483163e-01 -1.69181144e+00 1.11446083e-01
7.00515807e-01 1.02338696e+00 3.09375376e-01 8.43502402e-01
-6.90587936e-03 4.70716834e-01 -1.27497435e+00 6.36041701e-01
-9.05584693e-02 -6.85342312e-01 -2.20580444e-01 6.80121705e-02
1.98025420e-01 3.10726494e-01 -9.75900516e-02 1.01365876e+00
8.22226703e-01 6.49371803e-01 8.14062893e-01 -8.51244867e-01
2.50033975e-01 7.18831360e-01 -5.72080791e-01 5.75622141e-01
3.83725137e-01 -3.26128155e-01 7.13914454e-01 -1.13760042e+00
-2.36692876e-02 1.22140217e+00 5.14978409e-01 1.57476619e-01
-1.11121309e+00 -1.02886271e+00 1.02475755e-01 1.64122835e-01
-1.74553990e+00 -8.19355369e-01 7.95093954e-01 -3.97811949e-01
9.28935945e-01 6.77473426e-01 5.79590619e-01 5.86836934e-01
-1.34298012e-01 3.21842253e-01 7.34374702e-01 -7.20465064e-01
2.81444013e-01 1.00656144e-01 2.93754548e-01 4.70678777e-01
-2.14122310e-01 2.33196039e-02 -1.05683517e+00 -3.56616259e-01
6.95279300e-01 -5.68894625e-01 -6.54689670e-01 1.09311670e-01
-1.29304647e+00 5.31339228e-01 1.70209482e-01 8.23971093e-01
-2.63395429e-01 -1.48568749e-01 1.56445399e-01 1.36327639e-01
7.24379480e-01 1.48009613e-01 2.48158813e-01 1.27705604e-01
-1.30027974e+00 -4.13103521e-01 1.01554179e+00 5.36328971e-01
5.13339877e-01 5.86263597e-01 -1.46454617e-01 9.41471756e-01
4.23104256e-01 7.95026720e-01 5.00562906e-01 -8.26759875e-01
3.07947576e-01 8.99298564e-02 1.35734022e-01 -1.18424010e+00
-4.54396635e-01 -8.10028851e-01 -8.60888302e-01 -4.24855426e-02
4.15730298e-01 -2.43976131e-01 -5.21063864e-01 1.80019689e+00
3.36172521e-01 6.88042879e-01 4.30924930e-02 1.05052006e+00
6.10984802e-01 8.08930457e-01 -2.12386683e-01 -7.46051908e-01
1.16548085e+00 -6.44679606e-01 -9.26361501e-01 -2.87637562e-01
1.23323034e-02 -8.11908007e-01 5.22437572e-01 2.83532023e-01
-1.01093900e+00 -7.03569889e-01 -1.26442742e+00 7.37096131e-01
-2.36589313e-02 2.01935589e-01 -8.56860280e-02 8.15213919e-01
-8.94118369e-01 8.06948319e-02 -8.83099735e-01 -3.58986408e-02
-3.83022577e-01 3.82839471e-01 -6.23698644e-02 4.98322546e-01
-1.09535372e+00 7.62041330e-01 -2.29724050e-02 5.60849369e-01
-9.85879838e-01 -3.10354650e-01 -6.46814466e-01 3.65831435e-01
-2.48748995e-02 -2.26469889e-01 9.51481283e-01 -1.14640832e+00
-1.50588834e+00 3.47902507e-01 -6.70303762e-01 -2.69121140e-01
3.16014118e-03 1.25368372e-01 -7.96401203e-01 1.95283189e-01
8.20433646e-02 -3.72064329e-04 1.40385211e+00 -1.20852351e+00
-4.65488613e-01 -2.53520817e-01 -9.94913161e-01 4.97910321e-01
-4.41620290e-01 1.76517799e-01 4.99801114e-02 -5.96448839e-01
5.63676000e-01 -3.59747767e-01 1.10861391e-01 -6.01346314e-01
-4.66657370e-01 4.36686911e-02 4.79347020e-01 -1.04331172e+00
1.52174449e+00 -2.71194506e+00 3.00905377e-01 7.60261953e-01
8.77783000e-02 -1.11121222e-01 -1.72370851e-01 2.12991744e-01
-2.43279710e-01 -5.61581552e-01 -2.31672078e-01 -3.07510883e-01
-3.01959842e-01 -4.42721546e-01 -1.64317206e-01 7.90391028e-01
-2.02603370e-01 1.95805077e-02 -6.70026541e-01 -3.77558827e-01
1.84883088e-01 7.11672366e-01 1.50811374e-01 2.63616443e-01
3.63085717e-01 7.32768297e-01 -7.79374987e-02 1.42098248e-01
5.78305483e-01 1.29250705e-01 3.18547249e-01 -2.10739732e-01
-3.87201577e-01 3.10466588e-01 -1.65582931e+00 1.31625676e+00
-5.76734126e-01 7.63109267e-01 6.65784538e-01 -8.25063467e-01
1.13831365e+00 7.51842618e-01 3.62081200e-01 -3.82720888e-01
2.19053805e-01 2.82640487e-01 2.77689457e-01 -1.60698920e-01
1.61658656e-02 8.43169093e-02 -1.70365646e-02 6.10150039e-01
2.93054432e-01 1.12592116e-01 7.69990459e-02 -5.83231039e-02
7.19927907e-01 -6.41970813e-01 4.83459890e-01 -5.24255812e-01
1.03980744e+00 -6.93108618e-01 3.73443544e-01 5.78866124e-01
-7.82349557e-02 4.28148389e-01 1.33725852e-01 4.60780188e-02
-3.62717062e-01 -1.27727485e+00 -5.93076907e-02 1.34226072e+00
6.54509105e-03 8.63386765e-02 -1.04626560e+00 2.52366394e-01
-3.83198857e-01 8.03338528e-01 -1.36339918e-01 7.48115778e-03
-8.16147327e-01 -5.48139989e-01 4.73491848e-01 2.70830572e-01
3.33876580e-01 -7.58261263e-01 -3.61901283e-01 4.34351146e-01
-5.99448144e-01 -1.00441849e+00 -1.01667488e+00 3.17260802e-01
-6.01780057e-01 -5.16830802e-01 -5.58167517e-01 -1.10268033e+00
5.31946063e-01 3.94787252e-01 5.42710423e-01 -4.35042560e-01
-2.06365898e-01 3.97857815e-01 1.88286573e-01 -3.14745724e-01
-3.77511978e-01 -2.89508551e-01 4.00067300e-01 6.02432549e-01
2.91494280e-01 -8.97230029e-01 -4.54287499e-01 6.90322876e-01
-1.97805375e-01 -2.63475537e-01 2.68659592e-02 5.80634773e-01
1.23441204e-01 3.26646775e-01 8.42932224e-01 1.57858357e-01
9.57915068e-01 -2.51086295e-01 -7.70802259e-01 1.78866416e-01
4.71021570e-02 -3.37348670e-01 2.74579018e-01 -7.71110237e-01
-1.35982668e+00 -1.03377819e-01 3.22132349e-01 -8.99623558e-02
8.23941901e-02 3.39265436e-01 -1.50067970e-01 8.71172994e-02
7.51100838e-01 4.34325069e-01 8.46735109e-03 -1.59199327e-01
7.00413529e-03 1.06740773e+00 7.61958778e-01 -2.24321246e-01
5.29351890e-01 3.57600957e-01 -3.70423108e-01 -1.39223027e+00
-4.29044515e-01 -1.03465462e+00 -4.46946651e-01 -5.42655826e-01
9.42717969e-01 -1.18282294e+00 -8.91227722e-01 6.25840902e-01
-1.12130404e+00 8.14029574e-02 -1.28190577e-01 1.02912879e+00
-3.32516789e-01 2.24825963e-01 -4.87563014e-01 -1.61184216e+00
-5.24069548e-01 -7.84792602e-01 8.45715225e-01 1.21235803e-01
-5.73500574e-01 -8.88320446e-01 1.27171412e-01 1.77755386e-01
4.43199098e-01 -1.63398404e-02 7.83371329e-01 -6.91337049e-01
-3.83460015e-01 -2.49320954e-01 3.06530267e-01 4.62464780e-01
3.73425335e-01 -2.81401455e-01 -1.39626241e+00 -2.95463324e-01
6.82135820e-01 3.82603884e-01 5.83087802e-01 1.00919068e+00
4.66505051e-01 -1.69388264e-01 -4.08664942e-01 3.09684873e-01
9.91935253e-01 4.07187790e-01 1.17617644e-01 -3.78278911e-01
4.97281522e-01 7.13237047e-01 2.24200040e-01 4.99976128e-01
-1.84693590e-01 5.35846710e-01 -4.72551174e-02 1.20457634e-02
-1.21752493e-01 1.56992733e-01 6.42269850e-01 1.48147261e+00
-6.03640229e-02 -2.32157424e-01 -7.52761900e-01 5.38419962e-01
-1.50870287e+00 -9.92312133e-01 3.49727608e-02 2.61034608e+00
5.45905590e-01 1.89729948e-02 1.19627103e-01 4.15187031e-01
1.22257507e+00 2.16490999e-01 -4.23008680e-01 -1.70839280e-01
-8.95554498e-02 2.29549691e-01 1.80647269e-01 1.02412462e+00
-8.25014353e-01 3.08572531e-01 6.33752537e+00 8.45919132e-01
-9.92129624e-01 1.99995026e-01 1.98361292e-01 -3.37867320e-01
1.26113385e-01 -6.86972320e-01 -6.01811588e-01 3.38101417e-01
1.17105055e+00 -1.28610596e-01 7.48926222e-01 2.88585216e-01
5.60472012e-01 -4.30347830e-01 -1.08984065e+00 1.08535612e+00
2.19987735e-01 -6.59418225e-01 -5.93131244e-01 -8.90622288e-02
3.83950919e-01 -4.92646508e-02 1.89698085e-01 -1.05542757e-01
-9.92760137e-02 -1.05135727e+00 7.51017690e-01 5.21321416e-01
7.41040945e-01 -7.94582665e-01 5.92230380e-01 5.78494370e-01
-1.76409125e+00 -2.82027483e-01 -8.69365036e-02 1.54729709e-01
1.21149607e-01 7.36522973e-01 -1.18093669e+00 5.41980527e-02
4.49727088e-01 2.10951239e-01 1.76811308e-01 8.35669637e-01
2.22927313e-02 7.40424037e-01 -4.90421981e-01 -1.00315688e-02
-1.90606937e-01 -3.35809290e-01 1.02754784e+00 1.41807544e+00
4.69716668e-01 -5.55429384e-02 2.13621631e-02 7.57369339e-01
1.76238805e-01 -2.36838963e-02 -1.20699175e-01 1.94570515e-02
8.34868908e-01 1.40621853e+00 -7.29576528e-01 -1.26251057e-01
-8.16923752e-03 5.17988563e-01 -2.83375651e-01 7.64219761e-01
-5.60182631e-01 -4.03332710e-01 5.51244676e-01 2.43905075e-02
7.69948289e-02 -3.41598094e-01 -2.24069521e-01 -7.54793525e-01
-1.47769442e-02 -8.59727442e-01 1.58863246e-01 -4.89386290e-01
-1.07469583e+00 7.21484780e-01 -1.86384078e-02 -1.41563594e+00
-4.82150316e-01 -1.75204813e-01 -8.78838420e-01 1.52971244e+00
-9.46664453e-01 -5.01583815e-01 -1.21038899e-01 7.05451429e-01
6.53295398e-01 -4.49432015e-01 1.06434524e+00 2.17900008e-01
-4.39473599e-01 4.92047906e-01 1.18575223e-01 3.59812416e-02
4.12378818e-01 -9.49969113e-01 2.59863883e-01 7.05442667e-01
1.35765493e-01 6.34259105e-01 7.47451365e-01 -4.20718372e-01
-7.80801535e-01 -7.01845944e-01 8.27535212e-01 -3.02214380e-02
5.29398859e-01 -7.29374766e-01 -7.97312558e-01 2.28420675e-01
2.41345987e-01 -3.60940009e-01 8.53000700e-01 5.71101457e-02
-7.80201107e-02 -3.72510552e-01 -8.95201921e-01 2.97527730e-01
6.02266848e-01 -7.94918418e-01 -4.19926673e-01 2.85927057e-01
4.49664533e-01 -1.50497898e-01 -5.94982803e-01 2.36711558e-03
6.52533829e-01 -7.90349960e-01 1.00256431e+00 1.88668832e-01
-5.44777036e-01 -5.13680160e-01 -3.07095855e-01 -1.38123834e+00
-4.46357131e-01 -6.20619059e-01 7.65817463e-02 1.09352124e+00
4.08772498e-01 -8.08248818e-01 3.49537402e-01 -2.75448471e-01
-5.97585216e-02 -1.41463673e-03 -1.28057182e+00 -7.16268659e-01
-6.22940898e-01 -1.49052829e-01 3.28476578e-01 7.01558709e-01
2.90777892e-01 7.16245294e-01 -2.09020585e-01 6.29309177e-01
7.56446958e-01 -2.15345979e-01 1.16983347e-01 -1.21681666e+00
-3.59060496e-01 -2.93132901e-01 -2.00355276e-01 -7.94585168e-01
1.90463915e-01 -6.52853787e-01 6.26871943e-01 -1.22649431e+00
-7.91026875e-02 1.60757273e-01 -4.34136271e-01 -3.22068393e-01
6.42041937e-02 -2.22761378e-01 -7.56347030e-02 1.91281766e-01
2.30725743e-02 2.85054475e-01 8.80448163e-01 -1.59717694e-01
-5.87576449e-01 5.22594988e-01 -1.18236877e-01 8.35906625e-01
5.28842211e-01 -4.48309511e-01 -6.06042504e-01 -1.82778850e-01
-3.74521375e-01 5.55694878e-01 6.98803738e-03 -1.43860066e+00
5.15808344e-01 3.04681987e-01 3.97735000e-01 -7.94881761e-01
7.81982720e-01 -7.71177351e-01 3.20587933e-01 3.62689972e-01
-3.19795758e-01 -3.78607899e-01 5.56923002e-02 6.60299659e-01
-5.29360712e-01 -1.21185318e-01 8.29620242e-01 8.98588821e-02
-3.24604847e-02 -3.85268897e-01 -7.98224747e-01 -3.68193239e-01
6.08564615e-01 -4.27151144e-01 1.11012636e-02 -7.11775482e-01
-9.17942584e-01 -1.74570218e-01 -6.38396680e-01 2.82457024e-02
6.68790698e-01 -1.18516409e+00 -8.00975502e-01 5.60558200e-01
-3.22527766e-01 -1.93645135e-01 3.45374614e-01 9.46306288e-01
-4.65270430e-02 4.79972005e-01 2.11574193e-02 -8.48850965e-01
-1.51969230e+00 9.71807633e-03 6.83293641e-01 -5.19722290e-02
-1.25239581e-01 1.17169869e+00 5.29797971e-01 -2.06880778e-01
6.23085082e-01 -2.95773268e-01 -6.06292367e-01 4.57624495e-01
7.85419941e-01 8.74780595e-01 2.27127433e-01 -9.70352113e-01
-5.30363023e-01 6.00611806e-01 4.15378064e-01 -8.00432622e-01
1.05605102e+00 -4.34860229e-01 -2.78022796e-01 7.88869798e-01
1.25458252e+00 2.81901777e-01 -1.00669158e+00 -3.62211376e-01
-4.30311263e-02 -1.56699955e-01 9.96874496e-02 -5.33376217e-01
-7.28827536e-01 8.01992416e-01 9.63115633e-01 2.99172997e-01
1.45747602e+00 -1.98182508e-01 2.54313588e-01 7.67082572e-02
3.05400699e-01 -1.10623932e+00 2.38565132e-01 3.95918071e-01
8.67467880e-01 -5.42919934e-01 -3.39256138e-01 -4.88288343e-01
-3.76262814e-01 1.07090032e+00 1.54153377e-01 2.07019560e-02
8.07585120e-01 4.09713894e-01 2.33082071e-01 3.51772495e-02
-3.51624668e-01 9.18400288e-02 4.67219502e-01 5.69426775e-01
4.29050505e-01 2.39848629e-01 1.40237868e-01 3.61380756e-01
-3.17118973e-01 -7.03660071e-01 1.60681754e-01 4.81457263e-01
-9.02558982e-01 -4.11663681e-01 -1.08644986e+00 1.71170831e-01
-9.20246243e-02 -8.18924904e-02 -1.47819698e-01 2.10115090e-01
-8.82764235e-02 1.47889817e+00 3.93481135e-01 -2.91405857e-01
2.54190117e-01 3.67005467e-01 3.21383864e-01 -5.70046246e-01
-5.83053648e-01 9.38803494e-01 1.88751444e-01 -6.88135773e-02
-4.92585897e-01 -7.31380582e-01 -1.22184706e+00 4.47452962e-02
-7.84206688e-01 3.55382293e-01 9.55195010e-01 8.06726813e-01
-1.05906576e-01 7.93783665e-01 9.46794271e-01 -8.91755283e-01
-3.72116089e-01 -1.34420681e+00 -1.04882097e+00 -3.91838700e-01
6.64750636e-01 -2.56153882e-01 -8.41444075e-01 1.53505430e-01] | [14.912898063659668, 5.799304008483887] |
c09ec03a-6a05-47c5-9d91-59d3bd796c1f | on-systematic-style-differences-between | null | null | https://openreview.net/forum?id=mRUrRIL-jXh | https://openreview.net/pdf?id=mRUrRIL-jXh | On Systematic Style Differences between Unsupervised and Supervised MT and an Application for High-Resource Machine Translation | Modern unsupervised machine translation (MT) systems reach reasonable translation quality under clean and controlled data conditions. As the performance gap between supervised and unsupervised MT narrows, it is interesting to ask whether the different training methods result in systematically different output beyond what is visible via quality metrics like adequacy or BLEU. We compare translations from supervised and unsupervised MT systems of similar quality, finding that unsupervised output is more fluent and more structurally different in comparison to human translation than is supervised MT. We then demonstrate a way to combine the benefits of both methods into a single system which results in improved adequacy and fluency as rated by human evaluators. Our results open the door to interesting discussions about how supervised and unsupervised MT might be different yet mutually-beneficial. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['unsupervised-machine-translation'] | ['natural-language-processing'] | [ 4.15772736e-01 4.52893227e-01 -4.54611897e-01 -5.08262277e-01
-1.15438879e+00 -1.07643366e+00 9.79743958e-01 1.71038464e-01
-3.42774361e-01 1.08618736e+00 7.93681920e-01 -7.65185833e-01
1.73029765e-01 -4.91486013e-01 -4.38455433e-01 -2.94902593e-01
5.86203277e-01 8.83098900e-01 -2.25105137e-01 -5.43161094e-01
2.97873527e-01 1.01166815e-01 -9.88202453e-01 2.88341314e-01
1.49985385e+00 -1.29871294e-01 2.25856736e-01 5.41464865e-01
-1.58307925e-01 4.90290344e-01 -5.96580684e-01 -9.36696351e-01
2.00046539e-01 -8.28154564e-01 -1.20623648e+00 9.25453156e-02
4.87773180e-01 4.31872010e-02 3.82571816e-02 1.16005564e+00
4.64783698e-01 -1.54854372e-01 8.02353859e-01 -7.28621066e-01
-1.00628769e+00 1.03030229e+00 1.04486085e-02 1.87559843e-01
5.29131830e-01 3.62293154e-01 1.04960358e+00 -1.02323568e+00
8.89591813e-01 1.15025377e+00 5.32707274e-01 3.61198455e-01
-1.50036824e+00 -1.66885898e-01 -3.65771264e-01 -2.20042303e-01
-8.38815928e-01 -9.73682404e-01 9.05631557e-02 -4.75588709e-01
1.12802684e+00 2.19434783e-01 3.36925089e-01 1.05481005e+00
3.99675310e-01 4.60008472e-01 1.49484980e+00 -9.04549897e-01
-1.91572472e-01 4.87717241e-01 -1.83322299e-02 5.13488114e-01
4.97683227e-01 8.49210545e-02 -5.96959829e-01 8.34562108e-02
3.93230140e-01 -3.46720457e-01 -2.68900156e-01 1.12131968e-01
-1.87599146e+00 6.91240609e-01 -5.48048653e-02 6.40499830e-01
-3.08555573e-01 -2.60115534e-01 3.49878937e-01 9.00296569e-01
5.38217366e-01 1.01571941e+00 -5.63287318e-01 -5.07392168e-01
-1.09009326e+00 -9.56104025e-02 8.59024227e-01 1.13460934e+00
8.60372603e-01 1.64245050e-02 -2.77742743e-01 8.82859290e-01
3.43669541e-02 8.70608747e-01 6.34630024e-01 -9.66502428e-01
7.40679801e-01 4.39811558e-01 2.83687741e-01 -7.69176900e-01
-2.38033161e-02 -4.80722636e-01 -4.37539339e-01 -7.17973337e-02
4.21443433e-01 -2.78477430e-01 -8.91587257e-01 1.75942731e+00
-1.70406178e-01 -1.05702710e+00 2.19537586e-01 7.33459592e-01
6.13393188e-01 5.87576389e-01 -1.01312967e-02 -5.41247785e-01
1.07780492e+00 -1.23829257e+00 -9.57837462e-01 -3.06599379e-01
1.05367780e+00 -1.26076603e+00 1.45990074e+00 -6.83869608e-03
-1.44596910e+00 -5.17368197e-01 -9.55529392e-01 -1.77529439e-01
-2.73208261e-01 1.72175263e-04 5.09410918e-01 7.96079576e-01
-1.35377383e+00 8.57620418e-01 -5.94483376e-01 -8.69890392e-01
1.10118592e-03 2.80926257e-01 -5.28934479e-01 1.28732711e-01
-1.05238962e+00 1.43290198e+00 1.90504387e-01 -1.90615609e-01
-4.95193839e-01 -1.87269360e-01 -7.33212292e-01 -1.94819048e-01
5.31175584e-02 -7.16438472e-01 1.69649053e+00 -1.36655414e+00
-1.60918081e+00 9.58081305e-01 -5.57118893e-01 -1.70830339e-01
4.62756068e-01 -1.15749046e-01 -4.88617152e-01 -1.62144676e-02
4.15176094e-01 5.98599434e-01 4.76481766e-01 -9.31260109e-01
-5.71110129e-01 -1.26153737e-01 -1.58965588e-01 5.46658874e-01
-3.46138418e-01 4.78634745e-01 -1.16513073e-01 -6.07118130e-01
1.89679012e-01 -9.13750291e-01 1.01044821e-02 -6.63086891e-01
-3.60950053e-01 -1.23777248e-01 9.17126313e-02 -8.88807118e-01
1.14100587e+00 -1.49238050e+00 4.15669352e-01 -6.27670810e-02
2.14942247e-01 8.27534422e-02 -2.57887036e-01 9.51032877e-01
7.85305724e-02 6.15445197e-01 -2.13712126e-01 -1.16696849e-01
9.06193703e-02 2.11947128e-01 -1.64091036e-01 2.77133524e-01
2.05745488e-01 1.31363392e+00 -1.18334556e+00 -6.17380023e-01
-2.14062855e-01 -1.38697937e-01 -1.71907082e-01 -2.93366034e-02
4.98432014e-03 8.03500056e-01 -2.45135024e-01 7.75762737e-01
2.33259127e-01 -2.48712748e-01 3.35626334e-01 4.42699283e-01
-1.66556254e-01 1.04459262e+00 -2.42790535e-01 1.65739334e+00
-2.99885869e-01 9.66817617e-01 -3.40032011e-01 -5.68711162e-01
8.57719898e-01 7.11820662e-01 -1.97462514e-01 -8.94757628e-01
-6.33929595e-02 6.91643417e-01 3.65004510e-01 -4.17234451e-01
8.22622359e-01 -5.23598671e-01 -6.09533079e-02 9.92614865e-01
1.33374542e-01 -5.52518785e-01 3.19244146e-01 3.34192187e-01
9.05129194e-01 2.79606674e-02 1.48244515e-01 -5.96173048e-01
1.98307902e-01 6.15407348e-01 4.22495306e-01 8.39385629e-01
-2.59521008e-01 4.76201147e-01 1.66228011e-01 -8.27332810e-02
-1.57848668e+00 -1.05631459e+00 1.68322772e-02 1.02189410e+00
-1.39176711e-01 -4.43970859e-01 -9.44280446e-01 -7.62736559e-01
-4.28051203e-01 6.55347526e-01 -3.18435848e-01 2.94141229e-02
-6.36765540e-01 -4.23323721e-01 6.65528476e-01 2.49580801e-01
1.41392678e-01 -1.19140542e+00 -1.19984806e-01 2.00597435e-01
-8.35353136e-01 -9.70792115e-01 -5.37668467e-01 2.49864370e-01
-1.38531971e+00 -3.58361721e-01 -7.77208507e-01 -9.84016001e-01
6.66306138e-01 2.64708817e-01 1.56591570e+00 1.87743455e-01
5.65552056e-01 -3.29812467e-02 -6.27057016e-01 -3.18381071e-01
-1.23900855e+00 5.49686551e-01 4.33067381e-01 -7.17942774e-01
5.20386517e-01 -5.52743256e-01 -2.54924685e-01 3.60327631e-01
-7.22226143e-01 3.03157717e-01 7.37505615e-01 8.30471992e-01
1.52881488e-01 -4.51693982e-01 6.80789948e-01 -9.99601424e-01
1.09477091e+00 -2.45309338e-01 8.10894296e-02 4.33447182e-01
-1.09987378e+00 2.23725528e-01 6.84108078e-01 -2.46712044e-01
-9.50870812e-01 -3.42825979e-01 7.87310153e-02 4.25817490e-01
-5.69399372e-02 6.67179227e-01 -1.75878778e-02 3.07558268e-01
1.01268208e+00 2.06556499e-01 4.86838222e-02 -5.33782482e-01
4.51156050e-01 1.06809354e+00 4.32677299e-01 -6.01739347e-01
1.08714890e+00 -2.71459762e-02 -7.16459453e-01 -5.31407714e-01
-5.43933332e-01 -1.73722371e-01 -8.22362304e-01 -2.99374666e-02
6.24391258e-01 -8.27763557e-01 2.77954310e-01 6.35474622e-02
-1.09391952e+00 -5.11357367e-01 -4.05382633e-01 5.79058349e-01
-6.74292505e-01 3.89611721e-01 -9.50440586e-01 -4.50696468e-01
-4.91583228e-01 -1.33488357e+00 9.48945940e-01 1.61145359e-01
-8.67329359e-01 -1.05286324e+00 3.21271539e-01 6.33030117e-01
4.73539829e-01 -1.64963618e-01 1.11255920e+00 -7.27667153e-01
-2.54766732e-01 -1.17361069e-01 -6.14383817e-03 1.83996096e-01
5.59659123e-01 1.02314435e-01 -5.44944763e-01 -1.61628962e-01
1.86801367e-02 -4.45459306e-01 3.60396057e-01 1.49382222e-02
-1.03284061e-01 -5.06759822e-01 -8.15022215e-02 1.55324206e-01
1.14792788e+00 -6.44750819e-02 7.51924753e-01 3.96181583e-01
4.82973874e-01 9.02797401e-01 6.26584828e-01 -4.25309271e-01
1.15982972e-01 6.63557410e-01 -5.11238873e-01 -2.11876109e-01
-1.89766467e-01 -5.66076100e-01 7.89518118e-01 1.75123429e+00
-2.03818709e-01 -2.37688482e-01 -1.04126346e+00 6.81459069e-01
-1.67043340e+00 -1.01185966e+00 -3.34324181e-01 2.16369867e+00
1.23930168e+00 1.70903623e-01 2.21302912e-01 -1.52390689e-01
8.99460256e-01 -1.89439401e-01 -8.58554617e-02 -1.00700605e+00
-3.74836534e-01 3.01252693e-01 4.28053528e-01 6.30517364e-01
-4.84373808e-01 1.34367383e+00 8.05783176e+00 5.61171174e-01
-8.66601884e-01 3.17658573e-01 6.50414765e-01 9.32831988e-02
-7.46011078e-01 3.79638016e-01 -3.57059389e-01 3.63520026e-01
1.43073666e+00 -5.49569905e-01 4.33990121e-01 3.04272354e-01
6.34994328e-01 1.10299714e-01 -1.23767734e+00 4.21516061e-01
-2.20772147e-01 -1.16978717e+00 1.22740075e-01 2.99236923e-01
1.04956782e+00 3.62061590e-01 -1.24591943e-02 3.29398751e-01
6.97081566e-01 -1.25524402e+00 8.77176642e-01 3.03983480e-01
1.01412070e+00 -4.19306844e-01 6.05637133e-01 5.74808836e-01
-5.89404941e-01 3.26745987e-01 -4.64475006e-01 -4.19992507e-01
2.14397505e-01 4.96851742e-01 -8.48199487e-01 5.35006166e-01
2.53331631e-01 4.22235906e-01 -7.07244933e-01 6.28857195e-01
-5.13728201e-01 8.55030537e-01 1.43342063e-01 -1.49633646e-01
2.21540347e-01 -4.52818066e-01 6.72944844e-01 1.39267647e+00
2.55343884e-01 -2.35501781e-01 -5.77122830e-02 6.14046991e-01
-1.97132289e-01 5.52897573e-01 -9.76439595e-01 -6.75245047e-01
4.34377193e-01 9.89679277e-01 -7.64172018e-01 -7.71631896e-01
-3.26829493e-01 1.30938184e+00 4.20578420e-01 4.54859972e-01
-3.21981311e-01 -1.61860690e-01 2.73729533e-01 -4.31650406e-04
-1.27853528e-01 -4.02549446e-01 -6.85437024e-01 -1.42252946e+00
2.28149980e-01 -1.58040369e+00 -3.28292437e-02 -7.64551640e-01
-1.07552314e+00 7.76057422e-01 -3.75929385e-01 -1.25361979e+00
-4.71684754e-01 -2.31029376e-01 -3.64803165e-01 1.19635916e+00
-1.09915733e+00 -6.54016078e-01 1.76051870e-01 -1.21237766e-02
7.95470059e-01 -1.37374163e-01 9.00567949e-01 -8.36083200e-03
-2.66684026e-01 7.58864045e-01 3.28689456e-01 2.22989134e-02
1.06831074e+00 -1.28060937e+00 8.81123126e-01 1.06160295e+00
5.41432500e-01 1.03426337e+00 8.77205491e-01 -8.60272586e-01
-1.26058817e+00 -8.52252364e-01 1.81951451e+00 -1.09399188e+00
5.82232833e-01 -9.71280336e-02 -6.18148923e-01 6.54977441e-01
8.40048254e-01 -8.46032023e-01 6.50859833e-01 2.93629825e-01
-3.75837237e-01 4.66043741e-01 -1.02248216e+00 9.60325301e-01
1.18714142e+00 -7.98342645e-01 -9.29539680e-01 5.91894507e-01
1.07075346e+00 2.09442787e-02 -9.69297349e-01 2.79476643e-01
4.42681074e-01 -7.74871290e-01 3.50352705e-01 -8.78007948e-01
9.15965319e-01 -1.15819916e-01 -1.20878980e-01 -1.62427294e+00
-4.88326669e-01 -9.77425218e-01 3.15756947e-01 1.09508812e+00
1.26678360e+00 -6.65609539e-01 3.91894430e-01 5.35678983e-01
-2.85078555e-01 -4.25804198e-01 -3.71593207e-01 -9.52711046e-01
6.73666835e-01 -1.24748657e-02 5.98475456e-01 1.27410793e+00
6.89263165e-01 8.59737456e-01 -1.22196510e-01 -2.98887759e-01
1.26025423e-01 2.41605155e-02 8.35220337e-01 -1.02880239e+00
-3.20690483e-01 -5.86835444e-01 -1.81691602e-01 -1.11363029e+00
-1.17422931e-01 -1.28900695e+00 2.94565558e-01 -1.78060830e+00
5.30882120e-01 -2.30222940e-01 7.09619150e-02 3.90730351e-01
-4.06296164e-01 5.30467927e-01 1.65730342e-01 7.75777519e-01
-3.54088485e-01 1.76769748e-01 1.33653605e+00 -1.71716571e-01
-3.44532937e-01 -2.68606097e-01 -1.13866782e+00 3.31999481e-01
9.04255748e-01 -6.20973766e-01 -3.26742083e-01 -9.92349446e-01
3.28990370e-01 3.63872424e-02 -4.13705349e-01 -5.79258263e-01
7.56136794e-03 -3.58316600e-01 5.08862361e-02 -1.19900770e-01
-3.11978221e-01 -2.31896818e-01 1.23894423e-01 2.53945172e-01
-5.74116349e-01 6.43245995e-01 -1.10744551e-01 9.08369496e-02
-1.60590291e-01 -2.87594974e-01 5.24424732e-01 -3.40515763e-01
-1.21111996e-01 -8.22608843e-02 -7.94894397e-01 1.61905542e-01
2.32312560e-01 -5.55920839e-01 -4.44747210e-01 -7.36412466e-01
-5.78022718e-01 -9.37099308e-02 1.06308556e+00 4.30849105e-01
6.76837116e-02 -1.28449762e+00 -1.16715026e+00 -2.93648183e-01
3.50837499e-01 -5.99420667e-01 -6.02073669e-01 1.12044895e+00
-4.94574845e-01 6.50993526e-01 -5.72518483e-02 -5.64089060e-01
-1.01799965e+00 2.06349894e-01 1.93472147e-01 -3.28195870e-01
-5.02276897e-01 4.73009676e-01 -2.15412930e-01 -8.59219313e-01
-9.02011618e-02 -1.67988602e-03 1.92689151e-01 -1.75650731e-01
3.10045958e-01 2.17174485e-01 2.94489503e-01 -9.02619243e-01
-8.00720379e-02 2.02591613e-01 -2.21434101e-01 -8.36756408e-01
1.11226034e+00 -4.70843345e-01 -3.00726503e-01 5.71507454e-01
8.96219313e-01 5.35665989e-01 -6.04127944e-01 -2.55795717e-01
3.90087694e-01 -4.13343489e-01 -4.02994484e-01 -1.13783860e+00
-4.07561213e-01 7.21773744e-01 2.19343290e-01 2.19083637e-01
8.09515417e-01 1.97814759e-02 8.54406238e-01 6.44917846e-01
5.42347550e-01 -1.34517002e+00 -1.10324629e-01 7.83701837e-01
6.42897606e-01 -1.30274916e+00 -1.87388167e-01 -1.69831604e-01
-7.20663309e-01 8.94684851e-01 1.00452401e-01 3.77679199e-01
-1.63253754e-01 -8.92772004e-02 5.34688413e-01 -2.65790597e-02
-7.46810615e-01 -2.84554034e-01 1.65741205e-01 7.33614624e-01
1.03464019e+00 3.16926777e-01 -9.31538343e-01 6.28474122e-03
-8.23397756e-01 -1.92375287e-01 6.03458703e-01 9.05744612e-01
-6.79337502e-01 -1.63168180e+00 -2.42435142e-01 4.02183115e-01
-5.58758199e-01 -4.78072703e-01 -9.79722738e-01 5.49393177e-01
-2.11589202e-01 1.45624673e+00 -2.10077137e-01 -4.45705712e-01
8.77621025e-02 5.44453382e-01 6.98751450e-01 -1.16911793e+00
-7.85019040e-01 1.25191733e-01 5.71388423e-01 -1.63007304e-01
-3.00013423e-01 -5.92840433e-01 -9.42761421e-01 -5.43176293e-01
-5.69237769e-01 6.81339622e-01 6.38445675e-01 1.21038425e+00
2.71386445e-01 -1.02234066e-01 6.68869793e-01 -3.62816900e-01
-5.82227588e-01 -1.41460538e+00 -5.59397638e-02 4.42546338e-01
1.69318900e-01 4.24849391e-02 -3.06129456e-01 3.84696513e-01] | [11.581535339355469, 10.202765464782715] |
8f7281d5-0acd-4bc6-9073-e247d97b5f72 | towards-accurate-instance-segmentation-in | 2307.02877 | null | https://arxiv.org/abs/2307.02877v1 | https://arxiv.org/pdf/2307.02877v1.pdf | Towards accurate instance segmentation in large-scale LiDAR point clouds | Panoptic segmentation is the combination of semantic and instance segmentation: assign the points in a 3D point cloud to semantic categories and partition them into distinct object instances. It has many obvious applications for outdoor scene understanding, from city mapping to forest management. Existing methods struggle to segment nearby instances of the same semantic category, like adjacent pieces of street furniture or neighbouring trees, which limits their usability for inventory- or management-type applications that rely on object instances. This study explores the steps of the panoptic segmentation pipeline concerned with clustering points into object instances, with the goal to alleviate that bottleneck. We find that a carefully designed clustering strategy, which leverages multiple types of learned point embeddings, significantly improves instance segmentation. Experiments on the NPM3D urban mobile mapping dataset and the FOR-instance forest dataset demonstrate the effectiveness and versatility of the proposed strategy. | ['Konrad Schindler', 'Rasmus Astrup', 'Stefano Puliti', 'Frawa Vetterli', 'Theodora Kontogianni', 'Torben Peters', 'Binbin Xiang'] | 2023-07-06 | null | null | null | null | ['panoptic-segmentation', 'instance-segmentation', 'scene-understanding', 'clustering', 'management'] | ['computer-vision', 'computer-vision', 'computer-vision', 'methodology', 'miscellaneous'] | [ 3.85779113e-01 4.70101684e-02 -4.26047146e-01 -4.35350090e-01
-3.78431469e-01 -7.92656898e-01 6.34773195e-01 4.36588943e-01
-1.20991617e-01 4.11236286e-01 -3.97624448e-02 -5.15458107e-01
-4.96332794e-01 -1.26510012e+00 -5.06375253e-01 -6.24041915e-01
-2.21931368e-01 7.33298957e-01 4.03046131e-01 -2.01868154e-02
4.11466807e-01 7.85490513e-01 -1.73701262e+00 -7.32571408e-02
9.64856982e-01 1.11436057e+00 5.42212009e-01 2.94421673e-01
-7.21031129e-01 7.17014894e-02 -3.86450201e-01 1.10659248e-03
7.46082902e-01 1.46796003e-01 -8.91915798e-01 5.65215528e-01
4.83334988e-01 -2.22062096e-01 2.76610374e-01 1.06490505e+00
4.02048789e-02 1.93019226e-01 7.51304030e-01 -1.26818848e+00
-2.17172891e-01 9.65409800e-02 -9.09193039e-01 7.02631176e-02
-3.55733693e-01 -9.64605585e-02 1.26647878e+00 -5.33276796e-01
3.63112301e-01 1.07137382e+00 9.16770339e-01 -4.28227410e-02
-1.29883838e+00 -4.24854159e-01 5.37478328e-01 -2.96420101e-02
-1.55919075e+00 -1.12216860e-01 7.06375480e-01 -5.70415199e-01
8.25043499e-01 4.98805076e-01 8.61071944e-01 1.91042319e-01
-2.76448160e-01 7.17123032e-01 1.01543880e+00 -3.60619314e-02
4.03521925e-01 9.16188955e-02 4.26068842e-01 2.21273735e-01
4.28668052e-01 -2.86693335e-01 -1.19084239e-01 -2.68380474e-02
8.96800339e-01 4.75690722e-01 -8.56220797e-02 -6.62889361e-01
-9.46492493e-01 7.99492478e-01 8.09581637e-01 8.17151815e-02
-4.72622305e-01 3.59891467e-02 1.27782643e-01 -6.99363649e-02
7.81577528e-01 5.11651754e-01 -6.40385091e-01 1.58876032e-01
-1.35891461e+00 2.63197988e-01 4.81279045e-01 1.24746060e+00
1.15634310e+00 -4.74131078e-01 1.86248675e-01 1.02425265e+00
2.07005158e-01 4.19094115e-01 -5.69597334e-02 -1.06914532e+00
4.68247473e-01 1.02076876e+00 2.90235311e-01 -1.01061416e+00
-3.35812271e-01 -3.95417660e-01 -4.55185473e-01 1.15637772e-01
2.87210643e-01 1.96243137e-01 -1.30746055e+00 1.27668154e+00
7.10525692e-01 3.60364944e-01 -1.13004491e-01 9.80452776e-01
7.96219230e-01 6.36467040e-01 1.88184142e-01 1.91954166e-01
1.35546803e+00 -7.11557209e-01 -1.89606443e-01 -3.24800998e-01
1.69290170e-01 -5.35595715e-01 1.12151754e+00 -1.53981417e-01
-4.99271601e-01 -4.90470707e-01 -8.34766328e-01 -1.75422691e-02
-7.79332459e-01 -1.51404649e-01 9.34112906e-01 6.83364689e-01
-7.86711156e-01 4.58090067e-01 -7.36832738e-01 -6.25529468e-01
8.46464157e-01 2.64525592e-01 -1.88941061e-02 7.22032189e-02
-6.31293118e-01 3.96987885e-01 3.64589304e-01 1.31884053e-01
-5.67498624e-01 -9.52732980e-01 -6.60282850e-01 1.53598234e-01
4.03371841e-01 -4.93079484e-01 9.57375526e-01 -6.95998967e-01
-9.84083235e-01 1.14696825e+00 -1.86606571e-01 -4.06623632e-01
2.42834553e-01 -3.21105868e-01 -1.90243602e-01 1.62648201e-01
6.10569477e-01 1.01312304e+00 6.96076512e-01 -1.63481748e+00
-1.22310853e+00 -7.31230259e-01 3.33907545e-01 3.51207346e-01
4.35119271e-02 -4.59731430e-01 -6.00722194e-01 -4.76556331e-01
8.53810966e-01 -1.03658915e+00 -4.89454061e-01 -2.57234871e-01
-6.90430403e-01 -1.20364673e-01 9.60403800e-01 -3.85871619e-01
9.98435378e-01 -2.01807523e+00 -1.04672730e-01 4.55396712e-01
1.28788888e-01 -1.06789470e-01 2.69137800e-01 3.86332899e-01
6.17681481e-02 4.33911949e-01 -6.20053232e-01 2.17144676e-02
5.52760772e-02 4.36979890e-01 -4.22308296e-01 2.86375165e-01
2.44117051e-01 8.07812393e-01 -9.28072155e-01 -4.76218730e-01
5.50757110e-01 1.44270305e-02 -3.92832339e-01 3.66464257e-02
-5.84309042e-01 3.74927640e-01 -7.36158192e-01 9.32854950e-01
1.08470607e+00 -5.41654341e-02 1.07909068e-01 -1.12074852e-01
-4.01124269e-01 4.08491760e-01 -1.26290047e+00 1.51173556e+00
-3.19922119e-01 5.45677781e-01 8.76113214e-03 -1.09913290e+00
7.99471378e-01 -2.24525020e-01 7.62275279e-01 -3.73915255e-01
-2.83461481e-01 4.79658544e-02 -4.52876747e-01 -4.12608802e-01
9.11337078e-01 -1.53742760e-01 -2.37352178e-01 1.57835454e-01
-2.74781674e-01 -6.44329906e-01 -1.47922903e-01 -4.99058701e-02
7.23369956e-01 1.73649594e-01 1.92409828e-01 -4.61368442e-01
1.22870065e-01 7.59087026e-01 4.11519617e-01 8.28689039e-01
-2.82026589e-01 8.45876217e-01 3.37248653e-01 -4.28435147e-01
-9.64394033e-01 -1.45207763e+00 -6.01519883e-01 9.80396509e-01
6.75284147e-01 -2.08705992e-01 -6.13328516e-01 -6.50518239e-01
3.09666872e-01 6.73422039e-01 -4.74690944e-01 4.07643497e-01
-2.77377963e-01 -9.36998546e-01 1.14600398e-01 5.54163277e-01
5.88420987e-01 -7.41623282e-01 -8.06152165e-01 1.31105825e-01
-1.69640198e-01 -8.92889082e-01 4.46489453e-02 3.19475502e-01
-1.00592828e+00 -1.22553754e+00 -4.84452665e-01 -7.84627616e-01
7.18788743e-01 1.01346433e+00 1.13310730e+00 -8.22791159e-02
-9.66081396e-02 5.10689199e-01 -4.28854436e-01 -2.42086306e-01
2.43093535e-01 4.53102112e-01 -2.35278487e-01 3.27067599e-02
5.85314035e-01 -8.32037210e-01 -6.23551488e-01 4.87414569e-01
-6.71452284e-01 2.48935387e-01 5.17788231e-01 4.32504326e-01
1.04180491e+00 3.99096161e-01 1.99990630e-01 -1.05600870e+00
8.82168338e-02 -8.26851666e-01 -5.85389614e-01 1.85508460e-01
-2.76478022e-01 -3.35067779e-01 7.28866458e-02 -1.76577698e-02
-8.23604882e-01 2.27845639e-01 -2.34198868e-02 -2.63219446e-01
-6.61407292e-01 4.24060941e-01 -4.69988585e-01 1.79817155e-01
2.42827669e-01 1.07494265e-01 -3.30738276e-01 -5.48856378e-01
7.84673154e-01 7.85560369e-01 6.46199644e-01 -5.74714720e-01
1.01449931e+00 8.74605656e-01 -4.35142703e-02 -1.22453678e+00
-7.82340229e-01 -1.11085093e+00 -1.02534223e+00 -1.73574120e-01
1.03385210e+00 -8.47038984e-01 -2.79537171e-01 2.32733227e-02
-8.06039572e-01 -3.44500721e-01 -4.72619563e-01 2.88841128e-01
-6.70764327e-01 1.01495363e-01 -8.24651718e-02 -6.84598982e-01
7.21862018e-02 -8.85708690e-01 1.43827367e+00 3.42991769e-01
2.29331814e-02 -7.12151587e-01 -1.00023031e-01 5.43492794e-01
-2.30741259e-02 3.11829388e-01 9.69604492e-01 -3.92897785e-01
-1.08159924e+00 2.77325530e-02 -5.64225674e-01 8.70897025e-02
4.22349244e-01 -8.56212452e-02 -1.04111135e+00 1.23185247e-01
-2.09536389e-01 1.65436119e-01 1.04080713e+00 6.46125793e-01
1.42102730e+00 -8.59308764e-02 -5.66148162e-01 7.06147790e-01
1.57933521e+00 1.71877041e-01 5.09231389e-01 4.80612993e-01
9.20586050e-01 8.42390656e-01 8.46930742e-01 3.42268616e-01
4.17300403e-01 6.14805222e-01 7.44109035e-01 -1.99226156e-01
6.00944050e-02 -2.57022738e-01 -2.98391312e-01 3.06531191e-01
-8.21249261e-02 -3.02936584e-02 -1.11729312e+00 9.60415900e-01
-1.82378232e+00 -9.67927098e-01 -4.24085945e-01 2.13738990e+00
4.60473776e-01 1.74271137e-01 1.57684628e-02 1.95821255e-01
8.38844895e-01 3.45717520e-01 -5.74570596e-01 -1.59683579e-03
9.76429880e-02 2.47409597e-01 1.05424941e+00 4.00861025e-01
-1.48579144e+00 1.30010855e+00 5.80210400e+00 9.34457183e-01
-9.34304953e-01 6.47545606e-02 5.22372484e-01 1.39939710e-01
-2.59914100e-01 1.45593673e-01 -9.44276929e-01 3.82270098e-01
5.88817716e-01 3.09617877e-01 2.33231142e-01 9.76381838e-01
4.38883215e-01 -3.58663023e-01 -6.80015624e-01 6.24044061e-01
-4.85536337e-01 -1.44965541e+00 -1.27494326e-02 1.69707596e-01
8.06022286e-01 1.88627928e-01 6.66040033e-02 7.16995820e-02
3.96997899e-01 -8.87714803e-01 7.32865632e-01 1.76053897e-01
6.01965845e-01 -6.08114123e-01 2.82949418e-01 2.73283422e-01
-1.51875889e+00 -7.73345828e-02 -4.03272957e-01 -4.86947410e-02
2.27040261e-01 6.81887448e-01 -1.02221787e+00 6.76751435e-01
1.00374925e+00 7.60166049e-01 -5.70804298e-01 1.25243843e+00
-9.69335586e-02 7.92533100e-01 -6.08342946e-01 4.05172467e-01
6.78223014e-01 -6.71192884e-01 4.18956727e-01 8.90551329e-01
1.95051521e-01 2.51900330e-02 4.03628290e-01 9.02168810e-01
1.94670454e-01 -8.42488036e-02 -6.45581841e-01 1.04268029e-01
7.08891869e-01 1.25894856e+00 -1.27381563e+00 -2.67305821e-01
-2.70294487e-01 6.16408646e-01 1.19891547e-01 2.36961275e-01
-8.19878042e-01 -2.46427923e-01 1.08575392e+00 4.70908165e-01
6.67800605e-01 -6.18762851e-01 -9.57001865e-01 -5.90986073e-01
-1.05650285e-02 -2.41215855e-01 2.11243615e-01 -6.75061285e-01
-1.05004334e+00 -2.65528560e-02 3.22995454e-01 -1.33925962e+00
2.72913516e-01 -3.88271987e-01 -6.04470551e-01 5.35756290e-01
-1.70668340e+00 -1.51402724e+00 -5.88183463e-01 2.53051102e-01
7.43527830e-01 3.38737756e-01 5.12735426e-01 1.64339542e-01
-4.21934545e-01 -1.56632811e-01 4.35238510e-01 -1.07226700e-01
2.20795516e-02 -1.43506634e+00 3.32947493e-01 7.27446973e-01
2.52425373e-01 4.75607336e-01 5.13451576e-01 -7.34202266e-01
-7.74580956e-01 -1.66945255e+00 6.98206246e-01 -4.51764286e-01
6.01970434e-01 -4.21407640e-01 -7.77921736e-01 5.04205346e-01
-3.29462290e-01 -3.86667132e-01 6.15896285e-01 3.82643998e-01
-4.79303300e-02 -2.48999670e-01 -1.20658362e+00 5.43771923e-01
1.43195951e+00 -5.08895218e-01 -6.98038340e-01 6.17132604e-01
7.13246942e-01 -1.08161651e-01 -8.36424530e-01 3.90049249e-01
3.37364823e-01 -9.85996187e-01 1.33143091e+00 -3.98909122e-01
5.39554000e-01 -5.57273865e-01 -5.09053290e-01 -1.24530733e+00
-3.62445921e-01 -1.14717968e-01 3.86288464e-01 1.38804078e+00
1.98217943e-01 -5.35079539e-01 9.68731761e-01 2.82320052e-01
-4.60226506e-01 -4.68504727e-01 -8.68144155e-01 -8.29880655e-01
-5.41752996e-03 -6.45011485e-01 1.04124057e+00 9.64800596e-01
-6.92037582e-01 7.70277008e-02 2.24214196e-01 8.33598852e-01
4.93375540e-01 7.80180156e-01 1.11660099e+00 -1.60560679e+00
2.60431945e-01 -6.11139894e-01 -4.54126388e-01 -1.34078562e+00
1.27576590e-01 -8.07325959e-01 1.82116687e-01 -1.84869838e+00
-8.11761171e-02 -1.41182983e+00 -7.69818202e-02 3.81353825e-01
-2.11353540e-01 2.73817956e-01 -8.00697319e-03 6.13751650e-01
-3.79428208e-01 4.52553332e-01 7.79211998e-01 -4.14424688e-01
-5.70679963e-01 3.46170604e-01 -7.37161815e-01 6.11324966e-01
8.86490822e-01 -4.94873613e-01 -5.36724567e-01 -5.22931218e-01
-2.54471272e-01 -4.30619597e-01 5.01994550e-01 -1.20757711e+00
-6.09184913e-02 -5.76074183e-01 1.63714886e-01 -1.21943438e+00
4.32926178e-01 -1.26070118e+00 3.45964134e-01 1.29596725e-01
5.88460229e-02 -4.10471350e-01 1.03563502e-01 6.22821927e-01
-3.17679532e-02 -2.20792413e-01 6.91724956e-01 -1.96113482e-01
-1.16345310e+00 4.04308528e-01 -1.76419109e-01 2.38660444e-02
1.12580407e+00 -6.86848283e-01 -9.18912143e-02 6.48283586e-02
-6.32283270e-01 4.52542007e-01 7.42822289e-01 3.77528578e-01
2.54023015e-01 -9.77135658e-01 -2.80019045e-01 2.87226140e-01
2.89226323e-01 7.02723265e-01 9.26341340e-02 6.79831207e-01
-7.19541490e-01 3.96035612e-01 -1.52024701e-01 -1.06510961e+00
-9.25304830e-01 2.80833453e-01 2.99012363e-01 1.02776825e-01
-8.24075043e-01 6.69337809e-01 4.55088824e-01 -6.73609078e-01
-1.05409943e-01 -7.07585156e-01 -2.20223114e-01 3.96898210e-01
-1.34238347e-01 3.64238024e-01 -3.66993845e-02 -7.69883633e-01
-2.90277094e-01 6.92986250e-01 3.82596910e-01 5.87089211e-02
1.56098819e+00 -3.43773544e-01 -8.69114697e-02 5.92709601e-01
8.17983866e-01 -2.06793159e-01 -1.36308026e+00 -1.39488682e-01
3.18956614e-01 -7.35719085e-01 2.42580876e-01 -5.16848505e-01
-1.12429035e+00 8.19817901e-01 6.34683549e-01 5.26226103e-01
9.23365116e-01 3.20568204e-01 7.12395668e-01 2.01454118e-01
5.56182504e-01 -1.20286250e+00 -5.36139965e-01 3.12493742e-01
4.21587884e-01 -1.27827144e+00 8.47673640e-02 -7.76467323e-01
-3.77933860e-01 8.02910805e-01 3.61433178e-01 -2.38976404e-02
8.45053077e-01 -2.65588015e-01 -2.34679412e-02 -3.89346868e-01
1.24834433e-01 -8.14887226e-01 1.46612525e-01 7.60430455e-01
-1.77982673e-01 5.80972970e-01 -8.81925970e-02 4.78649676e-01
-3.48614693e-01 -3.17146868e-01 4.91982102e-02 1.01669312e+00
-9.20003891e-01 -7.13757575e-01 -5.46789825e-01 8.48349273e-01
8.93199965e-02 -8.12457278e-02 -2.94810534e-01 9.56491411e-01
5.52938819e-01 8.89267027e-01 6.17505848e-01 -2.19368771e-01
2.61420220e-01 -1.14322223e-01 1.34400308e-01 -7.70885348e-01
-3.31222862e-01 3.80095467e-02 -1.55634526e-02 -4.11617666e-01
-7.02541411e-01 -8.06351602e-01 -1.27461624e+00 -2.90485352e-01
-2.33681649e-01 8.93556029e-02 7.26763248e-01 8.03310096e-01
2.85007954e-01 1.31571248e-01 7.95472622e-01 -1.18214869e+00
3.60541157e-02 -7.63734400e-01 -8.26542497e-01 3.65063906e-01
1.99808106e-01 -9.56221402e-01 -2.13048175e-01 1.25069082e-01] | [8.34992790222168, -2.522951602935791] |
80bef646-700e-498d-a95f-bc8be226ea5c | 190508611 | 1905.08611 | null | https://arxiv.org/abs/1905.08611v1 | https://arxiv.org/pdf/1905.08611v1.pdf | Machine learning approach for segmenting glands in colon histology images using local intensity and texture features | Colon Cancer is one of the most common types of cancer. The treatment is planned to depend on the grade or stage of cancer. One of the preconditions for grading of colon cancer is to segment the glandular structures of tissues. Manual segmentation method is very time-consuming, and it leads to life risk for the patients. The principal objective of this project is to assist the pathologist to accurate detection of colon cancer. In this paper, the authors have proposed an algorithm for an automatic segmentation of glands in colon histology using local intensity and texture features. Here the dataset images are cropped into patches with different window sizes and taken the intensity of those patches, and also calculated texture-based features. Random forest classifier has been used to classify this patch into different labels. A multilevel random forest technique in a hierarchical way is proposed. This solution is fast, accurate and it is very much applicable in a clinical setup. | ['Soumick Chatterjee', 'Rupali Khatun'] | 2019-05-15 | null | null | null | null | ['colorectal-gland-segmentation'] | ['medical'] | [ 2.68125802e-01 -2.45439019e-02 -1.91755101e-01 -3.47571224e-01
-2.15030253e-01 -4.95210677e-01 2.44232133e-01 6.91186905e-01
-5.79414010e-01 6.03035748e-01 -2.05833316e-01 -3.45566124e-01
-1.01669870e-01 -1.13633633e+00 1.11438315e-02 -9.49120760e-01
-6.35041222e-02 8.10710013e-01 5.75895667e-01 1.00006349e-01
4.49532956e-01 7.56016254e-01 -1.33271790e+00 4.08864826e-01
9.09339368e-01 5.95777333e-01 4.38108295e-01 9.10730898e-01
-2.98829973e-01 3.84890407e-01 -1.90814018e-01 2.08034933e-01
1.14505991e-01 -6.44582689e-01 -9.22926605e-01 3.58583122e-01
-5.04994579e-02 1.49687648e-01 7.35586286e-01 1.23428071e+00
1.67694777e-01 -3.19210619e-01 1.29958475e+00 -5.61415493e-01
4.49272811e-01 3.66820931e-01 -7.39917636e-01 8.85481834e-02
1.12743460e-01 -3.48706067e-01 4.57763642e-01 -4.68630493e-01
6.28912628e-01 8.95740271e-01 5.75973690e-01 2.48160198e-01
-1.25764537e+00 -4.17198718e-01 -3.57507616e-01 2.71008521e-01
-1.25271940e+00 1.80333972e-01 2.00589195e-01 -8.02562177e-01
3.19519877e-01 5.52389324e-01 1.14940023e+00 1.37359142e-01
8.71966720e-01 2.95428485e-01 1.90366638e+00 -6.89418375e-01
4.20804977e-01 4.54971969e-01 4.53258663e-01 8.44291747e-01
6.18527651e-01 -2.29002178e-01 1.95404157e-01 -7.10253641e-02
6.03517473e-01 3.51102293e-01 -2.43888497e-02 -2.99119383e-01
-8.39920640e-01 8.86858463e-01 5.05482912e-01 1.00872195e+00
-3.45963627e-01 -9.25709382e-02 3.38332623e-01 9.44236889e-02
2.15656608e-01 3.06377858e-01 -4.06460971e-01 3.21042329e-01
-1.13205683e+00 -8.17546770e-02 7.45250821e-01 1.98902979e-01
6.00493729e-01 -6.86474919e-01 1.47043720e-01 4.66712505e-01
6.04567707e-01 4.46938187e-01 6.84682131e-01 -2.61074901e-01
-3.77528250e-01 1.23235345e+00 -2.20058590e-01 -1.02125311e+00
-6.10788226e-01 -2.79014826e-01 -9.78308439e-01 8.24615836e-01
7.49204993e-01 8.77121910e-02 -1.35725045e+00 7.41446257e-01
5.89568913e-01 -4.86953974e-01 -6.59744889e-02 6.65537953e-01
5.75528622e-01 6.51077271e-01 2.56141484e-01 -1.83566511e-01
1.68897378e+00 -6.24208331e-01 -6.93157434e-01 3.55895489e-01
6.45595074e-01 -1.14698505e+00 6.48243427e-01 7.28460133e-01
-7.36821711e-01 -2.44945377e-01 -8.93925846e-01 2.38262400e-01
-4.79568690e-01 3.81207854e-01 5.10762990e-01 8.96727085e-01
-8.52897167e-01 5.23132682e-01 -9.46766317e-01 -7.22944438e-01
-8.65435079e-02 5.71381092e-01 -3.69915456e-01 8.31613038e-03
-5.77231348e-01 8.90763879e-01 4.97176707e-01 1.03755578e-01
-4.41392392e-01 -9.13219228e-02 -5.03423631e-01 -2.00742483e-01
-2.41579950e-01 -6.47289515e-01 7.86167443e-01 -1.27317595e+00
-1.17081797e+00 1.24753034e+00 -2.35189036e-01 -3.11286211e-01
6.49181247e-01 6.38449013e-01 4.90645841e-02 3.56327325e-01
1.91761658e-01 4.71908569e-01 6.40064180e-01 -1.16122985e+00
-9.96219635e-01 -5.54364979e-01 -5.30618310e-01 2.68803567e-01
7.80157670e-02 -9.27344263e-02 -5.61190359e-02 -2.95192569e-01
5.78843653e-01 -1.02484596e+00 -7.36532271e-01 -1.37371421e-01
-3.02284062e-01 -5.85165843e-02 5.18325984e-01 -8.87583911e-01
9.86953497e-01 -1.99636066e+00 -2.25601792e-02 7.76017666e-01
-2.15868615e-02 -2.01547980e-01 7.94772863e-01 1.61054701e-01
4.19143997e-02 2.55349278e-01 -3.81973922e-01 2.59244174e-01
-4.22477514e-01 1.52111948e-01 4.71633494e-01 5.71374238e-01
-3.04837257e-01 2.00360008e-02 -4.55931365e-01 -1.19528401e+00
2.60177374e-01 3.85183871e-01 -3.28923911e-01 2.35726144e-02
-1.20345756e-01 5.64321935e-01 -4.72254246e-01 6.63051188e-01
8.02185476e-01 -1.54795619e-02 3.37454379e-01 -2.72295266e-01
-2.73268938e-01 -1.52493387e-01 -1.12173033e+00 1.12539816e+00
-2.81105787e-01 1.75330937e-01 1.21530958e-01 -8.09436023e-01
9.54322338e-01 4.58601743e-01 6.46525621e-01 -1.25924185e-01
3.47760946e-01 6.36473715e-01 8.71188492e-02 -6.45650148e-01
3.97989638e-02 -3.72044176e-01 1.18064158e-01 3.93042639e-02
-3.87786686e-01 -3.94198596e-01 5.43021977e-01 -2.37619117e-01
7.45293617e-01 -8.07805285e-02 9.76316214e-01 -7.33516932e-01
7.53684282e-01 7.02758133e-01 2.79109597e-01 1.26650631e-02
4.12863605e-02 2.83513457e-01 6.19079530e-01 -4.58519191e-01
-7.77345717e-01 -6.34003043e-01 -6.26675725e-01 5.23202956e-01
1.04658790e-01 3.55259418e-01 -9.77566361e-01 -4.77510601e-01
-1.73994914e-01 1.96918383e-01 -7.12012947e-01 5.31088352e-01
-4.81584579e-01 -1.16620195e+00 -1.07925273e-01 -2.33654454e-01
5.31088769e-01 -8.17560732e-01 -7.62192190e-01 2.54807085e-01
-4.75656837e-02 -2.43615225e-01 8.25089514e-02 4.08973724e-01
-1.42244935e+00 -1.27085841e+00 -9.04510856e-01 -1.18564868e+00
1.27746809e+00 1.31738752e-01 8.67512643e-01 4.68381852e-01
-7.84016550e-01 -2.34127432e-01 -4.67555195e-01 -3.11208457e-01
-6.97353065e-01 2.92543411e-01 -7.11656392e-01 -6.54485375e-02
2.70630896e-01 -1.08748175e-01 -7.52484441e-01 4.62008595e-01
-7.76498914e-01 -5.96208535e-02 9.42819238e-01 7.04963326e-01
9.41937745e-01 5.23314178e-01 -6.65328801e-02 -1.36310768e+00
3.65061611e-01 -2.91270405e-01 -6.62167013e-01 2.66329125e-02
-6.47940040e-01 -1.22321591e-01 4.27557796e-01 7.20518157e-02
-9.65143979e-01 4.77672666e-01 -1.98451355e-01 6.94090068e-01
-5.10972559e-01 5.06479144e-01 6.21529743e-02 -3.06986719e-01
6.76597118e-01 -1.15045577e-01 -4.39502671e-03 -3.32527071e-01
-4.32361782e-01 6.91821754e-01 -9.71950218e-02 -4.59239772e-03
3.70098591e-01 5.07789075e-01 5.23515999e-01 -8.67801726e-01
-2.45659202e-01 -9.45833445e-01 -6.42056644e-01 -3.68487149e-01
1.26872087e+00 -5.31418562e-01 -2.75369048e-01 4.50688750e-01
-7.15426207e-01 -3.35592806e-01 8.13328773e-02 5.82474113e-01
-2.25781575e-01 3.94570768e-01 -7.30192840e-01 -5.89692891e-01
-3.34544241e-01 -1.16309881e+00 5.50977468e-01 5.97053349e-01
-1.33629546e-01 -1.21654201e+00 1.70025662e-01 2.23607868e-01
1.65004194e-01 5.36303818e-01 9.99181807e-01 -3.58971179e-01
-3.67006987e-01 -4.93918210e-01 -1.39876738e-01 -4.47612032e-02
1.34737000e-01 4.33672190e-01 -6.46151900e-01 -9.75010842e-02
1.67518839e-01 1.44901991e-01 9.59019661e-01 7.48539388e-01
8.59769523e-01 1.16327316e-01 -8.05721164e-01 4.28342253e-01
1.97453988e+00 5.26736617e-01 5.27996421e-01 4.75027949e-01
1.88703299e-01 6.34971023e-01 1.05662215e+00 2.59200502e-02
-5.77438474e-02 2.79547334e-01 4.75894332e-01 -5.68563521e-01
-4.09568325e-02 1.64656088e-01 1.09417789e-01 5.19999087e-01
-3.22690487e-01 -1.96028911e-02 -9.06395912e-01 6.53427958e-01
-1.17988276e+00 -8.36303711e-01 -9.17570531e-01 2.16742492e+00
8.81739795e-01 8.99348110e-02 1.04984686e-01 7.06225574e-01
8.52825403e-01 -5.39189160e-01 1.55449495e-01 -7.16348827e-01
4.42366242e-01 2.24060714e-01 8.58757615e-01 8.39945257e-01
-1.07832706e+00 2.82009602e-01 5.98645878e+00 7.75303781e-01
-1.44311488e+00 -2.52718795e-02 8.51923108e-01 6.48984492e-01
-1.45051815e-02 1.07842520e-01 -8.12229633e-01 5.08923233e-01
5.09835541e-01 2.71663368e-01 -2.05833226e-01 6.25519395e-01
3.05756181e-01 -1.16009009e+00 -5.66262841e-01 2.70537317e-01
-2.05580920e-01 -7.24432886e-01 -1.85726270e-01 3.80212575e-01
7.26372242e-01 -4.22203362e-01 -2.17092901e-01 -2.88748056e-01
1.24112166e-01 -1.09478366e+00 5.36358133e-02 5.52644491e-01
6.90472960e-01 -6.63793147e-01 1.34798276e+00 3.97202641e-01
-1.01032901e+00 1.15508147e-01 -2.89205432e-01 -4.64801341e-02
-1.57870442e-01 8.98818910e-01 -1.44266951e+00 1.81462631e-01
6.31456852e-01 1.45295829e-01 -6.60777867e-01 1.49845266e+00
9.53120962e-02 6.82329237e-01 -4.27811325e-01 -5.87480366e-01
8.85956213e-02 -4.42870110e-01 1.23672403e-01 1.30350387e+00
3.62388104e-01 -1.97041690e-01 2.60611534e-01 3.09340775e-01
7.96486378e-01 9.44468498e-01 -3.43717754e-01 2.31719077e-01
1.03022121e-02 1.52510190e+00 -1.69545639e+00 -3.84543866e-01
4.88096029e-02 8.03174734e-01 -3.87298763e-01 -2.93162465e-01
-4.42652941e-01 -5.42524278e-01 -4.38534707e-01 6.42042518e-01
-1.49337456e-01 1.89274892e-01 -5.41535974e-01 -6.36969328e-01
-3.91624808e-01 -6.78942561e-01 5.79716504e-01 -1.20642319e-01
-9.36998963e-01 4.57593709e-01 -7.05815554e-02 -1.18844771e+00
4.01245877e-02 -8.50316405e-01 -5.62329113e-01 9.26247180e-01
-1.30034530e+00 -1.16609740e+00 -6.01053417e-01 4.23982203e-01
5.20859361e-01 1.48306832e-01 1.15056181e+00 7.59510249e-02
-2.27769494e-01 -7.93906674e-02 1.07904695e-01 4.03924324e-02
7.02389956e-01 -1.80951428e+00 -7.20123470e-01 4.47038800e-01
-4.47336257e-01 3.32109302e-01 8.80929828e-01 -7.84366429e-01
-5.41817129e-01 -6.59320414e-01 1.04423618e+00 3.44386160e-01
3.84572059e-01 1.43211380e-01 -4.39614862e-01 2.34188139e-01
3.88261080e-01 -2.69040525e-01 8.94173265e-01 -1.67545423e-01
4.96921927e-01 -1.16351359e-01 -1.51201737e+00 2.85006315e-01
-1.94007054e-01 2.90686935e-01 -2.87992924e-01 5.02451003e-01
-2.47370169e-01 -3.01549375e-01 -1.14221311e+00 1.90453500e-01
5.74507415e-01 -1.37736320e+00 5.21479368e-01 2.24329725e-01
1.99067891e-01 -4.91213858e-01 3.61435711e-01 -1.10959816e+00
-2.53678679e-01 3.24254595e-02 1.15465629e+00 6.99437976e-01
7.28254139e-01 -6.51916087e-01 1.12633646e+00 6.48137778e-02
2.80206710e-01 -6.33578897e-01 -6.86333239e-01 -8.72663129e-03
1.19883031e-01 5.59817731e-01 -3.11020277e-02 7.73465216e-01
1.40520796e-01 1.16043985e-01 3.43272120e-01 2.03464329e-02
7.52629936e-01 3.67177933e-01 5.35724759e-01 -1.60101283e+00
-2.70012081e-01 -4.02934462e-01 -6.65006936e-01 -3.34115118e-01
-6.50030911e-01 -7.30992138e-01 -6.08490519e-02 -1.86217260e+00
3.39987457e-01 -7.95189857e-01 9.67330709e-02 1.08394049e-01
-1.22364990e-01 2.50152767e-01 -3.16305429e-01 2.97380179e-01
2.13257462e-01 -6.46209657e-01 1.47563016e+00 -1.88485794e-02
-2.85662293e-01 6.76320672e-01 -9.30538476e-02 8.71815264e-01
1.10922635e+00 -6.52011812e-01 -1.75995573e-01 3.71711195e-01
1.65879250e-01 1.93618521e-01 1.28321320e-01 -1.11960876e+00
9.56290886e-02 -3.39915395e-01 6.48032963e-01 -8.01850379e-01
-4.35595587e-02 -1.39131784e+00 4.31736469e-01 1.36416972e+00
-2.23144159e-01 -2.90562119e-02 -3.32644284e-01 4.13815856e-01
-3.85622501e-01 -9.56573009e-01 1.21096909e+00 -6.84241295e-01
-3.34059656e-01 -2.80855864e-01 -8.69482100e-01 -6.51225150e-01
1.50071704e+00 -5.03344357e-01 2.71929443e-01 2.75740754e-02
-9.43831146e-01 -1.12201065e-01 7.53655612e-01 -8.03836465e-01
9.73953158e-02 -7.52774537e-01 -6.03341401e-01 -5.56315258e-02
-1.16993941e-01 2.57481426e-01 2.52132118e-01 1.26267123e+00
-1.62703776e+00 3.05962235e-01 -5.17027795e-01 -7.01334357e-01
-1.93982828e+00 4.05509949e-01 6.18520319e-01 -8.06592286e-01
-3.48213971e-01 6.86356962e-01 -1.64377868e-01 -5.82308397e-02
-5.19700535e-02 -7.64487624e-01 -9.55890775e-01 2.43196264e-01
1.47507027e-01 3.84332716e-01 1.14233401e-02 -4.72245842e-01
-8.84534195e-02 1.01318991e+00 -6.90770373e-02 -1.95488662e-01
8.69799256e-01 8.44706967e-02 -6.52099431e-01 5.24398625e-01
8.57149601e-01 5.72556734e-01 -5.13807952e-01 4.24430311e-01
1.18255824e-01 -3.16255659e-01 1.48056477e-01 -9.08804059e-01
-7.40157664e-01 8.36544037e-01 9.15373325e-01 5.37508190e-01
1.38646734e+00 -4.94863123e-01 3.43885303e-01 -6.15375349e-03
3.02087128e-01 -1.06844616e+00 -4.14209515e-01 -7.52091557e-02
3.91079724e-01 -1.15004504e+00 4.97966915e-01 -1.07153869e+00
-5.06370902e-01 1.50113547e+00 1.07637666e-01 -4.66621935e-01
7.84754336e-01 4.34508711e-01 4.15487230e-01 -2.16882095e-01
-3.72104406e-01 -2.30472252e-01 -2.22424441e-03 4.24644917e-01
8.59143734e-01 4.36815619e-01 -1.26979709e+00 7.88159370e-02
-4.32071052e-02 2.58163512e-01 5.24784803e-01 9.29669678e-01
-1.03659451e+00 -1.32015634e+00 -1.01580882e+00 5.72575331e-01
-9.39033628e-01 2.21377373e-01 -3.46676111e-01 8.13699245e-01
5.93110621e-01 7.48870373e-01 -1.59900278e-01 1.52156457e-01
-9.78225321e-02 -1.83305770e-01 4.22435790e-01 -3.89021009e-01
-8.62799227e-01 4.86243784e-01 1.79032162e-02 9.29569751e-02
-6.24148011e-01 -8.40897441e-01 -1.59310663e+00 1.11158378e-01
-4.11805898e-01 5.08559287e-01 1.25086987e+00 5.21558702e-01
-5.81563056e-01 3.25820357e-01 7.08943784e-01 -4.05735105e-01
-2.25138009e-01 -9.99552906e-01 -8.58281136e-01 2.68059283e-01
-1.14442185e-02 -4.92770165e-01 -5.09658158e-01 3.45098317e-01] | [15.170065879821777, -2.902688503265381] |
adc82b24-6660-42fe-99be-376870420f93 | porous-lattice-based-transformer-encoder-for | 1911.02733 | null | https://arxiv.org/abs/1911.02733v3 | https://arxiv.org/pdf/1911.02733v3.pdf | Porous Lattice-based Transformer Encoder for Chinese NER | Incorporating lattices into character-level Chinese named entity recognition is an effective method to exploit explicit word information. Recent works extend recurrent and convolutional neural networks to model lattice inputs. However, due to the DAG structure or the variable-sized potential word set for lattice inputs, these models prevent the convenient use of batched computation, resulting in serious inefficient. In this paper, we propose a porous lattice-based transformer encoder for Chinese named entity recognition, which is capable to better exploit the GPU parallelism and batch the computation owing to the mask mechanism in transformer. We first investigate the lattice-aware self-attention coupled with relative position representations to explore effective word information in the lattice structure. Besides, to strengthen the local dependencies among neighboring tokens, we propose a novel porous structure during self-attentional computation processing, in which every two non-neighboring tokens are connected through a shared pivot node. Experimental results on four datasets show that our model performs up to 9.47 times faster than state-of-the-art models, while is roughly on a par with its performance. The source code of this paper can be obtained from https://github.com/xxx/xxx. | ['Zhang Yue', 'Yu Bowen', 'Xue Mengge', 'Wang Bin', 'Meng Erli', 'Liu Tingwen'] | 2019-11-07 | null | null | null | null | ['chinese-named-entity-recognition'] | ['natural-language-processing'] | [-2.03601778e-01 -2.20955640e-01 -2.31949046e-01 -1.57450601e-01
-6.44737482e-01 -4.44332838e-01 2.34723136e-01 1.30760461e-01
-5.18907070e-01 6.22563601e-01 6.31734550e-01 -5.76406419e-01
3.87331367e-01 -9.84740078e-01 -6.80288136e-01 -7.85719812e-01
-8.12233835e-02 2.30867296e-01 3.21018577e-01 -1.19292818e-01
3.95403385e-01 4.83591855e-01 -1.11737442e+00 6.18612051e-01
9.25369263e-01 6.56242132e-01 4.93829191e-01 3.17208916e-01
-8.75905097e-01 6.27967298e-01 -2.94907182e-01 -3.60926658e-01
8.68098636e-04 -2.78400779e-01 -6.74388826e-01 -2.40978897e-01
7.60499313e-02 -2.22810254e-01 -2.77334452e-01 9.11502719e-01
6.13418102e-01 -1.21736653e-01 1.41867980e-01 -5.61447203e-01
-8.11672091e-01 1.03999746e+00 -4.13790494e-01 3.99710745e-01
-7.88015649e-02 -7.44034871e-02 1.29678786e+00 -1.20340836e+00
5.59444308e-01 9.92457867e-01 5.37832022e-01 3.75135452e-01
-7.99723327e-01 -8.66030097e-01 5.28872609e-01 2.30618522e-01
-1.64283586e+00 -3.35010558e-01 6.85249925e-01 7.61761367e-02
1.45467806e+00 1.58569574e-01 6.63439512e-01 7.67568946e-01
2.60917902e-01 1.02954412e+00 8.51998746e-01 -4.35221672e-01
3.59112211e-02 -2.71798223e-01 3.38096142e-01 9.21838820e-01
2.60175109e-01 -1.32549748e-01 -6.55396402e-01 -1.99015737e-01
1.05979645e+00 7.35577941e-02 3.13399062e-02 4.94349599e-02
-1.13458180e+00 6.10948801e-01 7.33709812e-01 5.80799580e-01
-3.04417253e-01 1.71270013e-01 5.32628179e-01 -9.48166475e-02
5.57894707e-01 1.76163748e-01 -5.05672514e-01 -1.76683456e-01
-6.92851901e-01 -1.99516162e-01 8.11874926e-01 9.50703681e-01
7.89378703e-01 9.96170938e-02 -2.93082893e-01 5.84816575e-01
2.75818646e-01 2.38653153e-01 4.82019037e-01 -2.28617474e-01
7.35598028e-01 8.21482658e-01 -2.50462592e-01 -6.92683339e-01
-3.81579131e-01 -6.40886486e-01 -1.17954040e+00 -5.58865070e-01
8.28132555e-02 -8.03264976e-02 -7.17656434e-01 1.21410036e+00
2.81668246e-01 4.95798051e-01 -9.16269496e-02 9.14002001e-01
7.11966813e-01 8.81206930e-01 1.41967461e-01 -1.92990974e-01
1.61454093e+00 -1.36759281e+00 -7.01208472e-01 -1.25926509e-01
1.06817842e+00 -7.21496940e-01 1.06536078e+00 1.07625857e-01
-8.96090746e-01 -4.70905155e-01 -7.80591309e-01 -3.57742190e-01
-3.20579678e-01 4.21659768e-01 8.18781197e-01 6.04678154e-01
-9.85591829e-01 3.95417035e-01 -1.17752922e+00 -3.25065255e-02
3.17620069e-01 3.09122145e-01 3.23344916e-02 -1.98984449e-03
-1.40469456e+00 3.94708961e-01 5.90178311e-01 4.67921734e-01
-4.01429772e-01 -5.73810935e-01 -7.31066465e-01 3.61413032e-01
3.01933855e-01 -1.93533957e-01 9.42437351e-01 -5.97031176e-01
-1.47133207e+00 4.33192134e-01 -6.43261433e-01 -3.36105049e-01
1.80338830e-01 -3.57480675e-01 -2.04579696e-01 -2.08519951e-01
1.92292966e-02 4.18318272e-01 2.99256593e-01 -5.54627240e-01
-3.65954906e-01 -3.40694278e-01 -2.27877334e-01 3.53109151e-01
-6.78924441e-01 3.01156551e-01 -9.31246996e-01 -8.40372980e-01
3.04049581e-01 -7.40332246e-01 -6.76797748e-01 -5.13662279e-01
-4.35558498e-01 -4.63946193e-01 5.03926992e-01 -5.14819264e-01
1.75675917e+00 -2.16011643e+00 -2.69817233e-01 -3.33646461e-02
9.59113911e-02 3.58282059e-01 -8.45246911e-02 6.54538572e-01
1.33452609e-01 4.72893685e-01 -9.96475890e-02 -3.12106669e-01
-1.38775334e-01 2.10962966e-01 -4.30157870e-01 2.78536111e-01
5.97952783e-01 1.19222105e+00 -8.36993694e-01 -4.61055249e-01
-1.23024195e-01 4.90229249e-01 -6.21541739e-01 9.41698924e-02
-1.36858091e-01 3.40773076e-01 -8.77859771e-01 5.26852429e-01
9.45675194e-01 -4.39012587e-01 5.37390530e-01 7.18771517e-02
-6.61692262e-01 9.84921634e-01 -1.13800824e+00 1.66753709e+00
-5.21813512e-01 2.19487727e-01 -1.35909945e-01 -6.86235309e-01
1.14589846e+00 2.30652601e-01 -3.10125854e-02 -9.77180183e-01
-2.45422691e-01 3.27752918e-01 -6.55161664e-02 -2.61626933e-02
6.14511371e-01 2.81467706e-01 -2.08227530e-01 6.24029338e-01
-3.62798929e-01 5.65196514e-01 1.08334400e-01 1.53997630e-01
1.13599741e+00 7.93357044e-02 2.02100039e-01 -5.33489704e-01
7.10423648e-01 -1.69446781e-01 1.03476512e+00 7.26484060e-01
4.93917353e-02 2.98406273e-01 5.81556976e-01 -9.14302111e-01
-1.14365208e+00 -8.65567088e-01 -4.52399105e-02 1.38460910e+00
2.26403717e-02 -7.91471183e-01 -5.99982679e-01 -4.63485241e-01
-3.95169616e-01 3.43530297e-01 -2.90235043e-01 2.01477364e-01
-1.30658031e+00 -8.09986413e-01 8.70877385e-01 9.59810793e-01
8.68356586e-01 -1.28544807e+00 -4.20401186e-01 6.49358749e-01
-1.85039505e-01 -8.82669628e-01 -7.35963821e-01 3.79528075e-01
-1.16775727e+00 -5.00139296e-01 -6.09591663e-01 -1.07003105e+00
7.82950878e-01 1.08952038e-01 1.11143291e+00 3.47332418e-01
-5.86280897e-02 -4.52377796e-01 -4.68652785e-01 -1.24862246e-01
1.20583527e-01 7.47650862e-01 -1.75956979e-01 -2.78405994e-01
5.02430975e-01 -3.72371554e-01 -5.40144920e-01 1.44000277e-01
-7.98682928e-01 3.84584248e-01 9.22436535e-01 1.20365500e+00
6.52656019e-01 -1.43528774e-01 2.27113456e-01 -9.73074973e-01
4.17419314e-01 -2.77013689e-01 -6.63516581e-01 2.75695831e-01
-4.19856131e-01 5.23204327e-01 7.74282396e-01 -3.15445870e-01
-1.30689478e+00 1.86659023e-01 -4.11894113e-01 -1.09836943e-01
2.59849690e-02 7.12210119e-01 -3.46431553e-01 2.90253997e-01
3.51018086e-02 5.28508484e-01 -6.19907141e-01 -7.01223314e-01
2.04220578e-01 5.05649507e-01 -1.46749495e-02 -7.37899303e-01
2.39444092e-01 4.39365923e-01 -3.12401742e-01 -7.71525800e-01
-5.78584671e-01 -4.42126930e-01 -8.37176621e-01 3.00452113e-01
7.42631257e-01 -9.83836472e-01 -6.57769918e-01 6.27370656e-01
-1.56570292e+00 -3.49209994e-01 2.27410346e-01 3.18442464e-01
9.26530436e-02 4.43999171e-01 -1.28008389e+00 -9.50385153e-01
-7.06747472e-01 -1.11134970e+00 1.09264076e+00 1.01406500e-01
3.49760443e-01 -8.39473426e-01 -5.37098832e-02 4.31557149e-02
3.87476325e-01 -4.51285094e-01 9.06130373e-01 -7.40443766e-01
-1.02390385e+00 2.25249872e-01 -4.21121150e-01 -7.50713572e-02
-1.56268865e-01 -3.05903435e-01 -7.03553200e-01 -2.40076020e-01
-1.79018438e-01 -6.37959018e-02 1.19787288e+00 -5.77347018e-02
1.26524591e+00 -3.86206359e-01 -4.09178376e-01 6.61532819e-01
1.33221102e+00 2.36904975e-02 8.01215708e-01 3.02036732e-01
9.84096944e-01 2.54927814e-01 3.25132042e-01 6.39029860e-01
4.97445256e-01 4.93140638e-01 1.38532773e-01 -1.28284812e-01
1.56210214e-01 -4.39253777e-01 5.18460512e-01 1.67291355e+00
-5.80535121e-02 -2.91029453e-01 -1.24642730e+00 5.50738513e-01
-1.81001687e+00 -5.74915469e-01 -2.83555299e-01 2.02034593e+00
8.56965780e-01 2.72012264e-01 -4.53624517e-01 -1.81050643e-01
8.45186710e-01 3.76314104e-01 -3.42124671e-01 -6.48020089e-01
-1.48167387e-01 4.82147723e-01 8.18333924e-01 3.02529246e-01
-1.02560365e+00 1.46138597e+00 5.11093569e+00 1.09828317e+00
-1.09121561e+00 2.72302747e-01 8.50017846e-01 1.95894420e-01
-3.99771571e-01 5.42480387e-02 -1.39162171e+00 2.76406646e-01
9.49370742e-01 9.86350328e-02 1.56272501e-01 5.59705317e-01
1.44191280e-01 2.44793430e-01 -7.26178765e-01 5.97518861e-01
-8.95288810e-02 -1.54535758e+00 1.76857740e-01 1.43441081e-01
6.88376427e-01 2.61776328e-01 -3.82868946e-02 3.34046096e-01
2.20254734e-01 -8.37449431e-01 5.43630600e-01 3.54439467e-01
7.88732767e-01 -9.10483003e-01 7.63993263e-01 4.91395146e-01
-1.87434185e+00 6.28880560e-02 -5.98835886e-01 -3.75175476e-01
1.18972942e-01 6.37989461e-01 -7.55649805e-01 5.69304824e-01
6.68736994e-01 6.53560221e-01 -3.33736002e-01 7.99269915e-01
-4.52582151e-01 8.47027302e-01 -4.10487562e-01 -5.67212999e-01
6.21499360e-01 -1.65046141e-01 1.50722414e-01 1.64028156e+00
2.82933295e-01 3.83324772e-01 2.23870263e-01 8.57120097e-01
-1.91066757e-01 4.64630425e-01 -4.32426691e-01 -4.26000357e-02
6.59661710e-01 1.06229627e+00 -9.85839844e-01 -3.77524763e-01
-5.09800971e-01 1.08715022e+00 7.88270473e-01 2.01350495e-01
-7.19555140e-01 -4.39524382e-01 4.57258552e-01 -1.15218706e-01
8.14167261e-01 -6.67095423e-01 -5.42273879e-01 -1.35714066e+00
1.74713001e-01 -6.25464678e-01 2.25814417e-01 -2.51560718e-01
-1.03972256e+00 7.47162223e-01 -5.89655340e-01 -9.78178084e-01
1.73440024e-01 -5.63606501e-01 -7.39688039e-01 1.01313031e+00
-1.50563884e+00 -1.08500040e+00 1.88796401e-01 2.82092333e-01
5.39431036e-01 8.45762268e-02 9.42068875e-01 3.52481544e-01
-7.39710033e-01 7.96203256e-01 3.22205454e-01 6.08545125e-01
4.01204139e-01 -8.64263952e-01 1.16231644e+00 9.46740627e-01
4.45152044e-01 9.93019938e-01 -1.28291592e-01 -8.88319910e-01
-1.67866921e+00 -1.14252627e+00 1.46511388e+00 -1.84557289e-01
6.96441889e-01 -8.67243528e-01 -1.18262911e+00 4.94873047e-01
2.55915463e-01 8.67651254e-02 6.69853091e-01 2.42077053e-01
-3.98800313e-01 2.41464358e-02 -3.93056184e-01 7.99493074e-01
1.27264452e+00 -5.81143260e-01 -4.47972715e-01 7.80131668e-02
1.01280761e+00 -5.67708611e-01 -6.08801365e-01 2.45792821e-01
4.50923055e-01 -8.52167070e-01 6.36673570e-01 -4.94531393e-01
2.90098608e-01 -2.75832474e-01 2.09602669e-01 -8.77129734e-01
-6.72355950e-01 -3.48829269e-01 4.91138659e-02 1.35941684e+00
6.00778401e-01 -7.01279521e-01 1.02705586e+00 2.17035174e-01
-2.06298053e-01 -9.79210556e-01 -1.05279732e+00 -5.11644304e-01
1.62341505e-01 -5.99856555e-01 8.16718578e-01 8.20850015e-01
4.24553528e-02 4.16108608e-01 -2.28484035e-01 3.28392118e-01
3.08355719e-01 2.82529712e-01 2.17429340e-01 -7.53158092e-01
-1.14474453e-01 -4.28783238e-01 -7.85751790e-02 -1.66784644e+00
2.69660085e-01 -1.07418275e+00 -8.31185747e-03 -1.33079898e+00
2.56009728e-01 -9.41369653e-01 -7.00873852e-01 6.43013239e-01
-2.50221372e-01 1.16905533e-01 2.73845166e-01 3.63900751e-01
-7.19774902e-01 7.18971610e-01 1.29825449e+00 -2.51537804e-02
-2.72877902e-01 -3.03617120e-01 -4.36341852e-01 4.22770709e-01
9.20287430e-01 -4.87644106e-01 5.67902140e-02 -1.01224875e+00
5.12442231e-01 1.58240478e-02 -3.41094546e-02 -7.03087986e-01
6.64912224e-01 1.49013087e-01 4.00828630e-01 -8.85849059e-01
-4.09220681e-02 -4.06378001e-01 -2.20061839e-01 6.92138672e-01
-3.29395473e-01 4.20081019e-01 3.88379663e-01 4.34931964e-01
-2.09980190e-01 -9.29509401e-02 2.33070597e-01 -3.53356212e-01
-7.94440150e-01 3.56610507e-01 -5.14210224e-01 -8.21360853e-03
4.90898281e-01 9.90581214e-02 -1.99298456e-01 1.11167066e-01
-3.23539853e-01 2.38316551e-01 1.24942787e-01 3.19263965e-01
6.25391603e-01 -1.38802195e+00 -8.05394173e-01 4.95626926e-01
-1.42020851e-01 1.50920182e-01 2.95370728e-01 6.30248785e-01
-7.17764914e-01 8.95463288e-01 6.93578972e-03 -3.79942179e-01
-9.47769105e-01 4.91926432e-01 -3.26582976e-02 -8.78718019e-01
-6.71560526e-01 9.82513487e-01 3.60698432e-01 -4.42958176e-01
5.44595718e-03 -5.99163592e-01 -1.13841452e-01 -2.14422956e-01
6.50494397e-01 2.12248191e-01 1.55190319e-01 -3.75844598e-01
-5.27226090e-01 3.12523931e-01 -3.87279004e-01 1.22581989e-01
1.11687613e+00 4.03716415e-02 -5.46975851e-01 2.75272727e-01
1.06998813e+00 1.38488770e-01 -9.34087873e-01 -4.18700904e-01
3.55311066e-01 -1.26226366e-01 -3.83617617e-02 -3.07801545e-01
-1.02910113e+00 1.00608754e+00 1.96589872e-01 -1.42472178e-01
1.08123016e+00 -2.01997817e-01 1.04596961e+00 6.57603085e-01
4.66834217e-01 -9.47744787e-01 -1.14769220e-01 1.31842387e+00
4.44851160e-01 -8.87026846e-01 -2.10383162e-01 -6.18120909e-01
-5.02330065e-01 1.17084098e+00 7.67045438e-01 -2.80116349e-01
6.41506135e-01 6.70801342e-01 4.27874066e-02 1.49417847e-01
-1.07817721e+00 -9.45526138e-02 8.69816095e-02 1.33406837e-02
9.57524836e-01 1.42165333e-01 -7.25075603e-01 8.45400393e-01
1.34892568e-01 -2.15804070e-01 1.32644370e-01 8.90406251e-01
-3.20427865e-01 -1.44492877e+00 -2.92458028e-01 3.62886608e-01
-5.42441130e-01 -9.14333403e-01 -4.50961068e-02 4.07720894e-01
8.95596147e-02 5.55926681e-01 4.42319095e-01 -1.59046009e-01
-3.64248380e-02 1.26917526e-01 -1.85941327e-02 -7.25634336e-01
-9.33376968e-01 4.55900788e-01 5.68201393e-02 -5.58283865e-01
-9.77825522e-02 -6.19731963e-01 -1.60066938e+00 -1.47784457e-01
-4.23317224e-01 2.62143284e-01 3.35458487e-01 7.20430553e-01
7.15364873e-01 5.23084641e-01 4.75161791e-01 -5.85581183e-01
-3.79189163e-01 -9.47329104e-01 -5.06078780e-01 -2.05955639e-01
-1.36167854e-01 -2.41711929e-01 7.18517005e-02 -2.73298472e-01] | [9.839982986450195, 9.791244506835938] |
714abf11-2c9d-4561-9476-3705bc7dc830 | semantic-hypergraphs | 1908.10784 | null | https://arxiv.org/abs/1908.10784v2 | https://arxiv.org/pdf/1908.10784v2.pdf | Semantic Hypergraphs | Approaches to Natural language processing (NLP) may be classified along a double dichotomy open/opaque - strict/adaptive. The former axis relates to the possibility of inspecting the underlying processing rules, the latter to the use of fixed or adaptive rules. We argue that many techniques fall into either the open-strict or opaque-adaptive categories. Our contribution takes steps in the open-adaptive direction, which we suggest is likely to provide key instruments for interdisciplinary research. The central idea of our approach is the Semantic Hypergraph (SH), a novel knowledge representation model that is intrinsically recursive and accommodates the natural hierarchical richness of natural language. The SH model is hybrid in two senses. First, it attempts to combine the strengths of ML and symbolic approaches. Second, it is a formal language representation that reduces but tolerates ambiguity and structural variability. We will see that SH enables simple yet powerful methods of pattern detection, and features a good compromise for intelligibility both for humans and machines. It also provides a semantically deep starting point (in terms of explicit meaning) for further algorithms to operate and collaborate on. We show how modern NLP ML-based building blocks can be used in combination with a random forest classifier and a simple search tree to parse NL to SH, and that this parser can achieve high precision in a diversity of text categories. We define a pattern language representable in SH itself, and a process to discover knowledge inference rules. We then illustrate the efficiency of the SH framework in a variety of tasks, including conjunction decomposition, open information extraction, concept taxonomy inference and co-reference resolution, and an applied example of claim and conflict analysis in a news corpus. | ['Camille Roth', 'Telmo Menezes'] | 2019-08-28 | null | null | null | null | ['open-information-extraction'] | ['natural-language-processing'] | [ 4.16800201e-01 6.47586107e-01 -4.20422643e-01 -1.96994632e-01
-4.33274090e-01 -1.00611782e+00 1.01379001e+00 5.38272083e-01
-1.09321982e-01 6.53627217e-01 4.95406687e-01 -9.67836380e-01
-6.97284639e-01 -1.03757191e+00 -1.04797386e-01 -3.84681582e-01
-4.46167216e-02 8.15019846e-01 4.58359450e-01 -6.13819540e-01
4.28617090e-01 6.77867651e-01 -1.72412503e+00 5.30044258e-01
5.68760693e-01 9.86896992e-01 -5.88257797e-02 3.25690418e-01
-8.08298171e-01 1.07600701e+00 -2.49408573e-01 -6.02117002e-01
1.12155475e-01 -1.34767294e-01 -1.50684559e+00 -1.07767314e-01
-2.04342261e-01 5.68114221e-01 3.95416170e-01 9.09319162e-01
-2.37412658e-02 -1.27506062e-01 7.05403745e-01 -1.06775320e+00
-3.10540438e-01 1.01841724e+00 -1.96034297e-01 6.90325052e-02
1.10652053e+00 -1.75540745e-01 1.25489330e+00 -6.46502137e-01
9.79991257e-01 1.50332892e+00 6.78383589e-01 3.28389972e-01
-1.32769108e+00 -1.83315918e-01 3.09465509e-02 2.42649883e-01
-1.31818330e+00 -4.15250778e-01 4.71681416e-01 -7.11169422e-01
1.09975350e+00 5.49398839e-01 6.39329433e-01 8.43920946e-01
9.05875340e-02 4.97406185e-01 1.43870115e+00 -1.13124681e+00
3.08480591e-01 2.66598642e-01 7.45391667e-01 7.73372471e-01
3.54875416e-01 8.37876871e-02 -5.28985262e-01 -5.12525558e-01
5.26010454e-01 -4.17601913e-01 -2.16359809e-01 -2.25911990e-01
-1.08950901e+00 1.03603745e+00 1.14693893e-02 7.62047648e-01
-1.70579419e-01 -4.67938602e-01 4.88694191e-01 4.01138812e-01
-5.52748516e-02 8.72245908e-01 -5.89278758e-01 -3.54935378e-02
-7.92166173e-01 4.58233833e-01 1.41637588e+00 6.35363281e-01
5.56979597e-01 -4.65672165e-01 9.53624547e-02 6.23744726e-01
5.12454808e-01 5.93420118e-02 5.12340426e-01 -9.54981327e-01
1.47638500e-01 9.44802403e-01 -6.72311410e-02 -1.14130974e+00
-5.28230250e-01 -1.50130495e-01 -3.61259162e-01 2.14458108e-01
5.81056416e-01 2.61237234e-01 -5.58794498e-01 1.68706882e+00
3.83909941e-01 -6.65493846e-01 2.42050573e-01 3.79521251e-01
6.54583573e-01 4.58824188e-01 2.63133138e-01 -6.67336762e-01
1.86580527e+00 -3.36182982e-01 -7.20954716e-01 -5.23665212e-02
6.81861520e-01 -5.30153334e-01 1.10068786e+00 5.48803151e-01
-1.20700347e+00 -7.19202608e-02 -1.00370574e+00 -2.08088890e-01
-8.99938047e-01 -5.37044644e-01 9.42377746e-01 6.20216370e-01
-1.01796317e+00 4.81328219e-01 -3.95174742e-01 -5.44489861e-01
2.69946396e-01 3.19379598e-01 -1.97223738e-01 1.37184441e-01
-1.31534660e+00 1.11363399e+00 8.35614920e-01 -2.40762636e-01
1.03880025e-01 -2.98158675e-01 -9.44992661e-01 1.72532618e-01
9.34564769e-01 -8.92574787e-01 1.14244473e+00 -9.16753650e-01
-1.28337502e+00 1.24084830e+00 -2.75854737e-01 -6.24365926e-01
2.59451747e-01 6.96799085e-02 -5.86971819e-01 2.51515746e-01
3.98097277e-01 3.62701237e-01 5.68785548e-01 -1.36452448e+00
-5.91003120e-01 -5.66948712e-01 1.79726675e-01 1.87785979e-02
1.06691435e-01 5.79956830e-01 1.56938106e-01 -5.65060318e-01
5.71410656e-01 -5.80320120e-01 -1.28691569e-01 -3.43062222e-01
-5.59841990e-01 -6.05103791e-01 2.54848182e-01 -3.41064215e-01
1.59480643e+00 -1.78220773e+00 1.72161832e-01 6.76309347e-01
4.21249151e-01 4.13455553e-02 5.33497095e-01 7.82742620e-01
-2.53056914e-01 5.71587145e-01 -4.54719216e-01 3.18103373e-01
4.00435269e-01 6.06136262e-01 -8.00301313e-01 -5.94330281e-02
8.72159749e-02 9.87430274e-01 -7.21306205e-01 -8.14149141e-01
-1.25408754e-01 -1.08513884e-01 -2.91669399e-01 -2.03546852e-01
-4.74679410e-01 -7.13096857e-02 -4.21416014e-01 6.75745666e-01
2.36230507e-01 -1.55213714e-01 7.04232872e-01 1.07291199e-01
-5.05189717e-01 7.61685848e-01 -1.38185561e+00 1.31506419e+00
-6.76586553e-02 3.28803807e-01 1.39914498e-01 -9.32654679e-01
8.40178370e-01 3.35310817e-01 9.37933475e-02 -3.65105510e-01
1.39649972e-01 3.36409867e-01 -1.19353980e-02 -5.76850235e-01
4.30809140e-01 -5.22188246e-01 -4.03851300e-01 4.83768553e-01
1.46530010e-03 -1.60839051e-01 3.59381169e-01 3.19747388e-01
1.07298028e+00 1.30055279e-01 1.22484732e+00 -6.37778342e-01
6.91927552e-01 4.47040856e-01 5.69212854e-01 8.02827656e-01
4.82201055e-02 -6.99437549e-03 8.42010379e-01 -7.47780621e-01
-6.84866130e-01 -1.12845111e+00 -4.02663589e-01 1.06837666e+00
-1.33646384e-01 -7.93301642e-01 -4.88520503e-01 -6.19121432e-01
-2.56869674e-01 8.46893430e-01 -4.59002584e-01 4.11966473e-01
-5.23310900e-01 -5.57353377e-01 7.06917167e-01 1.63365260e-01
2.19618797e-01 -1.22580874e+00 -8.02479565e-01 2.39059716e-01
-2.58537382e-01 -1.00854433e+00 4.93842870e-01 4.82995957e-01
-6.53166294e-01 -1.31683064e+00 2.88824290e-01 -7.66444802e-01
2.67654240e-01 -1.65972367e-01 1.26622021e+00 1.73810735e-01
-9.37468484e-02 2.80582458e-01 -6.50031388e-01 -6.22386038e-01
-7.39424706e-01 -8.28090161e-02 -2.37167738e-02 -1.83098152e-01
7.24749446e-01 -8.40611279e-01 1.34713322e-01 -1.93664152e-02
-1.03761446e+00 3.67619097e-02 2.92072684e-01 6.57457411e-01
3.90486538e-01 2.87034571e-01 3.66583288e-01 -1.13679969e+00
8.85371625e-01 -4.79180902e-01 -1.94815531e-01 5.02916992e-01
-8.18884611e-01 3.14907640e-01 3.37170959e-01 4.08935882e-02
-1.03273726e+00 -8.23242143e-02 -2.59112239e-01 2.75556028e-01
-5.48920393e-01 8.99243474e-01 -3.71278048e-01 1.06069319e-01
8.87479246e-01 8.67903307e-02 -7.13775009e-02 -4.27159071e-01
6.33512020e-01 5.55513084e-01 6.37735248e-01 -1.07250881e+00
8.41390848e-01 4.95726049e-01 1.53515175e-01 -8.40885520e-01
-8.40703130e-01 -4.20776010e-01 -8.42061758e-01 1.35455981e-01
8.29145908e-01 -2.86143631e-01 -7.41340637e-01 -2.02232286e-01
-1.29608226e+00 -4.32762578e-02 -7.42183626e-01 -2.29302142e-02
-6.34037673e-01 6.83474302e-01 -4.56030101e-01 -9.12377059e-01
-1.34494975e-01 -5.87002695e-01 7.91253805e-01 -6.48948178e-02
-9.01491046e-01 -1.02591753e+00 -5.83472885e-02 3.02019894e-01
-2.25086417e-02 4.89506841e-01 1.59987795e+00 -1.20869756e+00
-1.41673326e-01 7.36307427e-02 -4.02196590e-03 -2.19964430e-01
-1.56134799e-01 -1.00456148e-01 -7.01023102e-01 3.13264668e-01
3.91012549e-01 -4.01918143e-01 5.15719354e-01 -1.00496374e-01
7.15725005e-01 -7.89591908e-01 -4.18484330e-01 2.33386621e-01
1.58014333e+00 2.44065508e-01 8.11410785e-01 6.64920270e-01
2.41205782e-01 1.19699311e+00 3.20397288e-01 2.34079976e-02
5.33183098e-01 7.42823124e-01 -5.90828955e-02 2.44058087e-01
1.12199351e-01 -3.22210401e-01 8.60791504e-02 6.61504626e-01
-2.34404743e-01 2.26024389e-01 -1.43297637e+00 3.10504735e-01
-1.97225380e+00 -1.35483122e+00 -3.55938107e-01 1.93293381e+00
1.02459300e+00 4.76723433e-01 2.60747880e-01 7.15914011e-01
4.74831879e-01 -9.86421406e-02 2.71442652e-01 -1.00748503e+00
-3.34260643e-01 4.34626579e-01 -5.59591688e-02 6.38747036e-01
-8.60318065e-01 1.02091622e+00 6.73891592e+00 9.32891726e-01
-5.38041472e-01 2.25917567e-02 1.12019844e-01 4.87889051e-01
-4.38255161e-01 3.93855333e-01 -7.68517375e-01 5.39952181e-02
9.10502017e-01 -7.76991621e-02 3.44791174e-01 6.64578855e-01
-9.35996622e-02 -2.16809615e-01 -1.20134461e+00 6.80900991e-01
-1.52470870e-02 -1.65526509e+00 4.45620388e-01 1.07104652e-01
1.40976310e-01 -3.75553429e-01 -6.16092145e-01 1.52892068e-01
4.49211508e-01 -1.05760419e+00 9.69032645e-01 3.29173058e-01
4.72843349e-01 -3.70249957e-01 5.45786381e-01 5.77757001e-01
-1.19143915e+00 -4.48074937e-01 6.85934946e-02 -5.19072473e-01
2.11972296e-01 3.25840652e-01 -6.20172679e-01 9.25212443e-01
2.99849182e-01 2.03364164e-01 -4.10426110e-01 6.41639709e-01
-4.85800147e-01 3.60354692e-01 -3.12708288e-01 1.02921044e-02
1.61595121e-01 -1.53655395e-01 9.23261225e-01 1.55829060e+00
-2.29957268e-01 4.88937527e-01 3.82626086e-01 1.01616466e+00
5.63714027e-01 2.24880755e-01 -7.73724854e-01 7.14598671e-02
5.65068841e-01 9.10179615e-01 -1.05594122e+00 -5.08047581e-01
-2.51237899e-01 2.84237742e-01 1.81510106e-01 1.11962460e-01
-2.76484966e-01 -3.52705151e-01 1.86367109e-01 2.26606742e-01
1.08393528e-01 -2.76960373e-01 -6.71800554e-01 -1.22324717e+00
7.99574926e-02 -1.10189199e+00 8.60905230e-01 -6.56674147e-01
-1.23520005e+00 6.14679337e-01 4.22204703e-01 -7.08416700e-01
-5.44868171e-01 -8.96320283e-01 -5.04361749e-01 6.73906922e-01
-1.21865702e+00 -1.19750559e+00 2.33743325e-01 6.09839737e-01
4.01294142e-01 1.17972806e-01 1.30230987e+00 -3.94225895e-01
-1.87118277e-01 1.13868475e-01 -5.42679846e-01 1.28499344e-01
-4.33935691e-03 -1.30247581e+00 4.12071757e-02 8.30775619e-01
4.00536418e-01 1.03531897e+00 7.59088278e-01 -5.01584470e-01
-1.24364209e+00 -4.58225012e-01 1.57619143e+00 -6.55388713e-01
1.00466299e+00 -3.60519707e-01 -1.03293693e+00 5.91373980e-01
9.70503688e-02 -5.26111007e-01 9.15855885e-01 4.81983006e-01
-7.61508107e-01 2.68582404e-01 -1.15386987e+00 6.52469695e-01
1.11201167e+00 -6.56620383e-01 -1.71535265e+00 2.59750903e-01
7.09564507e-01 2.61179078e-02 -9.06886458e-01 3.58724982e-01
5.96646190e-01 -1.08464313e+00 8.27312589e-01 -8.61951351e-01
3.05239350e-01 -3.40147436e-01 -3.65445346e-01 -6.35228395e-01
-3.45315367e-01 -1.01688015e+00 -8.23030472e-02 1.34235692e+00
5.85657597e-01 -8.02134812e-01 2.90999591e-01 6.87052906e-01
3.80548611e-02 -8.55316699e-01 -9.18416262e-01 -6.09889388e-01
1.63211361e-01 -9.54776227e-01 5.92328370e-01 1.18665135e+00
8.74714553e-01 7.28180766e-01 4.67841774e-01 -1.93216652e-01
4.23185617e-01 5.10524511e-01 2.96735197e-01 -1.75695479e+00
-3.99861336e-01 -7.26438880e-01 -4.76259172e-01 -6.82745636e-01
2.24854261e-01 -1.19529498e+00 -3.49950105e-01 -1.54897380e+00
1.90528557e-02 -4.83143985e-01 8.80096182e-02 7.84553528e-01
4.40123707e-01 -1.52367696e-01 1.62566230e-01 5.42005956e-01
-4.60286140e-01 -1.06795393e-01 6.03882313e-01 7.72076473e-02
-4.22384202e-01 -2.65976191e-01 -1.14437270e+00 1.20212829e+00
7.50364423e-01 -4.65967149e-01 -2.57058740e-01 4.93221804e-02
8.21626842e-01 -1.31826043e-01 4.29173589e-01 -6.22392774e-01
3.80094439e-01 -3.54239404e-01 -1.54387886e-02 -7.79445469e-02
-7.46653154e-02 -7.06391037e-01 9.06618461e-02 4.17915195e-01
-3.95720720e-01 -9.10232365e-02 4.12079655e-02 3.39226097e-01
-1.61378458e-01 -4.52325284e-01 4.62392300e-01 -4.55653250e-01
-8.31011832e-01 -4.37762290e-01 -4.63214040e-01 1.89693883e-01
7.43388891e-01 -5.48178613e-01 -3.69049132e-01 1.31904095e-01
-9.41030443e-01 3.09432056e-02 4.96451318e-01 2.88057327e-01
4.43153441e-01 -9.66509879e-01 -3.74900490e-01 1.46126896e-01
1.42509148e-01 -2.28173226e-01 -4.06746894e-01 7.72031069e-01
-4.37425107e-01 5.94042778e-01 -1.53380752e-01 -2.64164060e-01
-1.08214748e+00 8.16708863e-01 1.54675409e-01 -5.48193336e-01
-6.57536387e-01 4.08737242e-01 -4.65776660e-02 -2.95763791e-01
1.39723137e-01 -3.11633319e-01 -6.94903970e-01 2.22219631e-01
7.13417649e-01 1.03013083e-01 -1.41149759e-01 -8.17021430e-01
-4.98817563e-01 5.87688148e-01 1.98771000e-01 -3.07547718e-01
1.17327380e+00 -2.37190038e-01 -7.47471273e-01 6.75300300e-01
5.15384734e-01 2.39930034e-01 -1.65842474e-01 -2.53573149e-01
7.32504845e-01 -8.98594782e-02 -3.16086054e-01 -1.07858598e+00
5.91323003e-02 4.32544321e-01 -9.47437286e-02 1.09119439e+00
1.02977693e+00 4.49606508e-01 3.40973914e-01 4.33294326e-01
5.82080901e-01 -8.26436937e-01 -5.80078244e-01 6.23735487e-01
9.16419804e-01 -6.94556832e-01 5.08040413e-02 -9.57433522e-01
-4.15915012e-01 1.45411170e+00 -6.40459880e-02 2.30417818e-01
5.75668335e-01 5.03884852e-01 -1.01169132e-01 -8.11197162e-01
-7.53248692e-01 -5.24668038e-01 1.76298738e-01 5.88440835e-01
3.06819528e-01 2.08908334e-01 -8.80597770e-01 9.04569805e-01
-6.22023702e-01 -1.06065519e-01 3.56354386e-01 1.18309665e+00
-7.53504455e-01 -1.45581317e+00 -5.69495797e-01 1.99414670e-01
-6.51987374e-01 -1.74552515e-01 -9.21043336e-01 1.06055927e+00
5.43832660e-01 1.18484795e+00 -2.88042843e-01 -3.00265491e-01
2.23397031e-01 7.10179925e-01 5.53378224e-01 -7.48456299e-01
-6.97881758e-01 -2.13670492e-01 6.39039874e-01 -4.30428177e-01
-7.61933029e-01 -6.48738623e-01 -1.38456953e+00 -2.20549896e-01
-2.14932293e-01 5.79172492e-01 5.31350851e-01 1.37287819e+00
9.31401104e-02 3.05739120e-02 1.92367479e-01 -4.42683280e-01
-5.79534113e-01 -5.69661558e-01 -5.43775618e-01 2.48035401e-01
-1.69810709e-02 -6.54610097e-01 -4.45972472e-01 1.46515921e-01] | [9.852252960205078, 8.755498886108398] |
fc7a2064-d887-4e3b-851d-4f1ea0da9d19 | modelling-customer-churn-for-the-retail | 2304.00575 | null | https://arxiv.org/abs/2304.00575v1 | https://arxiv.org/pdf/2304.00575v1.pdf | Modelling customer churn for the retail industry in a deep learning based sequential framework | As retailers around the world increase efforts in developing targeted marketing campaigns for different audiences, predicting accurately which customers are most likely to churn ahead of time is crucial for marketing teams in order to increase business profits. This work presents a deep survival framework to predict which customers are at risk of stopping to purchase with retail companies in non-contractual settings. By leveraging the survival model parameters to be learnt by recurrent neural networks, we are able to obtain individual level survival models for purchasing behaviour based only on individual customer behaviour and avoid time-consuming feature engineering processes usually done when training machine learning models. | ['Berthold Lausen', 'Maged Ali', 'Henrik Nordmark', 'Juan Pablo Equihua'] | 2023-04-02 | null | null | null | null | ['feature-engineering', 'marketing'] | ['methodology', 'miscellaneous'] | [-7.67445043e-02 -5.86528145e-02 -3.91532838e-01 -9.15353894e-01
-5.41054726e-01 -2.53779948e-01 2.00512558e-01 3.90623748e-01
-3.46374810e-01 1.35245726e-01 6.98599517e-02 -6.17498875e-01
-1.50892347e-01 -9.79761064e-01 -5.40721655e-01 -4.87673700e-01
-3.77905369e-01 8.81314814e-01 -4.94629711e-01 -4.51680154e-01
2.90477693e-01 4.24467057e-01 -1.51586592e+00 3.51561576e-01
3.86639982e-01 1.13326049e+00 -8.91423151e-02 5.43083191e-01
7.70442933e-02 8.43517244e-01 -1.62918404e-01 -5.48754692e-01
2.49753028e-01 -2.82341063e-01 -3.90774637e-01 3.60593642e-03
-3.68395954e-01 -5.75229287e-01 -6.14449829e-02 7.38879979e-01
1.68654501e-01 -3.17476578e-02 5.90556264e-01 -9.41390216e-01
-4.52547759e-01 1.37865353e+00 -1.72327772e-01 3.13987225e-01
-2.60156333e-01 -2.00351775e-02 1.48286617e+00 -3.78935337e-01
1.42222196e-01 1.02140033e+00 6.04593873e-01 2.30501994e-01
-1.57828784e+00 -5.94979823e-01 1.80524081e-01 2.12326273e-01
-9.27047968e-01 5.33610675e-03 8.60992610e-01 -3.53483915e-01
1.12563324e+00 8.80946871e-04 9.04815257e-01 1.48448908e+00
7.40565181e-01 9.43818510e-01 6.95871770e-01 -3.94877568e-02
1.89365819e-01 3.05484563e-01 4.19798255e-01 -7.17945099e-02
-3.45038297e-03 5.41372597e-01 -1.10085301e-01 1.17445029e-01
6.55267537e-01 3.85539562e-01 5.17195702e-01 -1.47674188e-01
-4.40501899e-01 1.69218397e+00 4.39443469e-01 5.59091806e-01
-7.55991936e-01 3.76630545e-01 3.39790136e-01 5.69628000e-01
4.12027776e-01 5.42316914e-01 -7.09080815e-01 -4.85510141e-01
-9.23976839e-01 4.67530698e-01 9.33088958e-01 7.43822813e-01
3.67431998e-01 3.70765254e-02 3.72515142e-01 4.94976819e-01
3.72392744e-01 1.23446420e-01 6.06940150e-01 -5.43281078e-01
-1.58179909e-01 6.03841841e-01 1.53750703e-01 -9.66484427e-01
-7.99527287e-01 -9.54131186e-01 -3.55831325e-01 -3.22910547e-01
2.22663239e-01 -5.46077900e-02 -7.65815973e-01 1.19434512e+00
-4.22738224e-01 -2.86051571e-01 1.65515497e-01 5.24343312e-01
2.67772049e-01 6.49293602e-01 2.45308101e-01 -2.90177256e-01
1.04403937e+00 -5.39413929e-01 -3.46194088e-01 -5.04355192e-01
8.94407332e-01 -5.47072351e-01 8.48525465e-01 8.35260212e-01
-9.33620870e-01 -3.54516447e-01 -8.47065449e-01 4.01977599e-01
-2.32850149e-01 -4.18666333e-01 1.02177680e+00 5.83649218e-01
-5.09889662e-01 9.38832283e-01 -9.30085719e-01 -2.99303800e-01
5.42820811e-01 6.74077392e-01 3.09874952e-01 9.48347673e-02
-1.29027247e+00 1.05403590e+00 2.99572676e-01 4.05284941e-01
-1.08495331e+00 -4.15056139e-01 -4.85825568e-01 2.86829591e-01
2.52313823e-01 -1.02444060e-01 1.66290653e+00 -1.11359489e+00
-9.88269627e-01 2.82724977e-01 2.86724001e-01 -1.03078663e+00
3.76445770e-01 -1.66737258e-01 -5.52600384e-01 -5.13661027e-01
-2.18335748e-01 3.10783684e-01 7.24675059e-01 -1.30687070e+00
-1.04162061e+00 -4.16359812e-01 -3.72053176e-01 -5.30086197e-02
-1.36746228e-01 4.92640026e-02 2.64145821e-01 -2.13031143e-01
1.25395402e-01 -9.98009861e-01 -6.93188846e-01 -1.27283204e+00
-3.05834413e-01 -3.56718332e-01 6.82308733e-01 -7.26920068e-01
1.05317354e+00 -2.07474756e+00 -1.10217400e-01 5.05998015e-01
8.76325369e-02 -1.69090591e-02 1.14158466e-01 6.41058266e-01
1.56472743e-01 2.36150920e-01 6.42081082e-01 -1.40911981e-01
3.21604967e-01 4.04205799e-01 -1.66628167e-01 2.17068493e-01
2.17147484e-01 1.00600290e+00 -6.02928281e-01 -3.31520550e-02
1.55303866e-01 4.56191510e-01 -4.76954252e-01 3.49744201e-01
-3.94193530e-01 4.37580973e-01 -5.45154631e-01 8.21285665e-01
2.34626919e-01 -2.80273050e-01 2.12060884e-01 2.49390498e-01
1.74877495e-01 3.94025654e-01 -6.43841624e-01 8.02522957e-01
-7.44888663e-01 2.00421676e-01 -9.17098299e-02 -1.28829873e+00
1.15209961e+00 5.84003888e-02 5.86138487e-01 -9.32278693e-01
8.87278795e-01 3.33991200e-01 3.48209381e-01 -2.42353559e-01
6.04612768e-01 -7.80104220e-01 -6.24767423e-01 3.91056687e-01
-4.61879857e-02 3.83328766e-01 -3.77352871e-02 -1.64910331e-01
7.79400945e-01 -3.25754255e-01 -2.45087549e-01 -9.73557904e-02
-1.50012434e-01 1.37798309e-01 4.75799263e-01 6.88654721e-01
-2.05506608e-01 1.73336156e-02 7.41911113e-01 -7.87503481e-01
-1.24941504e+00 -6.16429090e-01 -3.19368094e-01 1.23306179e+00
-2.35981330e-01 8.15767124e-02 -2.79215902e-01 -3.41408551e-01
2.90552348e-01 1.46887112e+00 -8.39418650e-01 -4.60302293e-01
-4.58065659e-01 -7.43944824e-01 -8.28064755e-02 8.77442300e-01
-1.55360714e-01 -1.12963092e+00 -6.15756929e-01 8.64803135e-01
2.68085718e-01 -7.26827383e-01 -1.01142839e-01 9.43187296e-01
-8.24959159e-01 -6.71577930e-01 -2.74220586e-01 -7.97108054e-01
1.63396299e-01 -3.06663662e-01 1.11769032e+00 -7.55258501e-02
-2.87822604e-01 -1.45210279e-02 -5.13568461e-01 -5.91996491e-01
-7.75425375e-01 5.10567069e-01 -1.59057185e-01 -7.15804026e-02
1.20357001e+00 -4.47480112e-01 -6.83953822e-01 4.06696767e-01
-6.27804637e-01 -3.14941466e-01 7.08499491e-01 9.22190011e-01
3.40143204e-01 7.00638175e-01 9.63708460e-01 -9.96138811e-01
7.17410028e-01 -7.00763285e-01 -5.14116526e-01 -2.45587453e-01
-1.40203035e+00 -3.93624902e-02 5.21033704e-01 -5.71376383e-01
-6.47581279e-01 2.51281895e-02 -3.00926566e-01 -2.16204092e-01
1.61340963e-02 7.56718934e-01 3.65924537e-01 4.58396912e-01
2.10429847e-01 1.78695023e-02 2.23291114e-01 -4.25811023e-01
2.15837151e-01 6.83382869e-01 -3.15469392e-02 8.33864436e-02
2.91590840e-01 -7.59237632e-02 -2.59893835e-01 -3.88103247e-01
-7.61556089e-01 -4.62507784e-01 -5.12547970e-01 -2.45971024e-01
6.60084665e-01 -7.21495688e-01 -1.25732791e+00 3.00681144e-01
-4.38970149e-01 -1.11489281e-01 -3.41560006e-01 5.21502852e-01
-5.98439813e-01 -5.44476807e-01 -7.89890587e-01 -1.00745654e+00
-2.67290145e-01 -1.11105275e+00 5.89544475e-01 2.04145566e-01
-4.03811544e-01 -1.17882872e+00 -2.48112187e-01 6.21746004e-01
5.40684581e-01 1.71112314e-01 1.14308333e+00 -1.38762176e+00
-2.98333019e-01 -8.79949749e-01 2.81592697e-01 6.76890492e-01
-3.06144327e-01 -3.99113335e-02 -3.92228901e-01 -4.95467633e-01
1.77455172e-01 -2.40612611e-01 6.17658019e-01 9.94369984e-01
6.60220802e-01 -2.95096129e-01 -4.49096441e-01 2.02782199e-01
1.47670233e+00 7.03756452e-01 4.55499858e-01 5.57382822e-01
4.04198289e-01 1.14881086e+00 7.87980437e-01 5.16525984e-01
5.36293805e-01 4.12549824e-01 7.29704201e-01 1.07139491e-01
9.29895282e-01 -4.89349544e-01 4.17993903e-01 5.22601902e-01
1.93874329e-01 -2.10184500e-01 -8.77037048e-01 6.56882405e-01
-1.88372862e+00 -8.61881018e-01 -1.56983882e-01 1.93810129e+00
4.07321155e-01 6.74761176e-01 6.03484511e-01 -4.34576087e-02
4.16048825e-01 -4.16194230e-01 -7.60568559e-01 -1.08494616e+00
8.79982337e-02 1.83963716e-01 9.77746487e-01 3.49952698e-01
-9.14361537e-01 1.04331827e+00 6.77239752e+00 3.85549635e-01
-1.05289829e+00 -2.52640516e-01 1.21950638e+00 -4.21359926e-01
-4.34665680e-01 2.95120060e-01 -1.26308274e+00 3.00971538e-01
1.65880895e+00 -6.58958927e-02 4.33274448e-01 9.72480178e-01
4.64884281e-01 1.61017403e-01 -1.25235331e+00 4.33624893e-01
-3.12809050e-01 -1.06330693e+00 -4.84333605e-01 4.90454704e-01
3.62803578e-01 -5.14717959e-02 4.05538946e-01 7.45748699e-01
7.69165158e-01 -1.57324040e+00 5.69002748e-01 4.52258348e-01
-1.71153963e-01 -1.40669847e+00 1.33944249e+00 4.35880482e-01
-5.15667021e-01 -8.25536132e-01 -1.94226712e-01 -1.62010044e-01
6.31563425e-01 4.79470611e-01 -1.27751803e+00 8.70278571e-03
7.83389628e-01 5.96864283e-01 -2.71605313e-01 3.67767900e-01
3.72777462e-01 8.68980050e-01 -2.39255190e-01 -4.17737991e-01
7.11319983e-01 -2.32894391e-01 7.14514181e-02 6.39179230e-01
5.09230018e-01 -2.68848419e-01 1.21589027e-01 9.37188566e-01
2.67509848e-01 1.40454561e-01 -3.77395511e-01 -6.56795323e-01
-5.06351776e-02 1.11666524e+00 -6.68206513e-01 2.95918822e-01
-2.93419033e-01 5.84997058e-01 -1.80092573e-01 -8.03786796e-03
-7.03757644e-01 -1.34957612e-01 4.16085184e-01 6.11124396e-01
7.78656006e-01 -2.37741560e-01 -3.71225417e-01 -5.72532117e-01
-4.95905012e-01 -8.62307072e-01 2.34800857e-02 -1.19138755e-01
-1.46213746e+00 5.28379023e-01 -3.90429258e-01 -8.06068897e-01
-6.49853051e-01 -5.06748319e-01 -4.60414082e-01 7.30275810e-01
-1.18891644e+00 -1.43055069e+00 3.43020827e-01 1.09680794e-01
3.13554883e-01 2.82631591e-02 6.65598631e-01 8.71488750e-02
-6.43802702e-01 4.87076670e-01 2.50902265e-01 -1.12195872e-03
9.03331563e-02 -1.12525344e+00 3.29903841e-01 1.57271817e-01
-3.62343877e-03 6.07454777e-01 1.23839676e+00 -6.78604722e-01
-1.22408736e+00 -7.48999238e-01 1.18778539e+00 -2.78077811e-01
8.73885870e-01 -6.00268245e-01 -6.87847853e-01 8.85831416e-01
2.60919780e-02 -7.86735535e-01 9.68697727e-01 6.50435507e-01
2.72382259e-01 -2.23132372e-01 -8.34350467e-01 2.19234362e-01
3.36942852e-01 2.19074748e-02 -2.66563863e-01 2.82366067e-01
6.59364820e-01 6.11819662e-02 -1.31555605e+00 1.83250234e-01
5.80452859e-01 -9.55305278e-01 6.55576885e-01 -7.31685340e-01
3.53899717e-01 6.75652325e-01 -1.16600238e-01 -1.04175436e+00
-4.26023453e-01 -5.24737597e-01 2.28732631e-01 1.05400288e+00
8.35669875e-01 -6.16682410e-01 1.07919657e+00 9.46271718e-01
1.95251122e-01 -1.09121120e+00 -8.00014436e-01 -6.58054888e-01
3.87321353e-01 -4.23716635e-01 6.80077493e-01 5.31006396e-01
4.54825466e-04 5.11815786e-01 -6.15261197e-01 -1.12684987e-01
3.65862399e-01 2.56641686e-01 5.20078242e-01 -1.50732279e+00
-4.14062202e-01 -6.31415725e-01 -4.02349561e-01 -8.18204522e-01
-8.69314149e-02 -6.55708134e-01 -4.18562889e-02 -1.29400563e+00
1.48393005e-01 -6.25075817e-01 -6.52792156e-01 2.77824163e-01
4.52263802e-01 -1.32883355e-01 -8.41287822e-02 1.70744047e-01
-1.39591873e-01 3.24910522e-01 1.12432802e+00 1.54047579e-01
-5.21606028e-01 6.80565655e-01 -1.16758120e+00 4.30038184e-01
7.92764544e-01 -4.94976729e-01 -3.48685026e-01 2.98667878e-01
6.92184269e-01 3.51679713e-01 9.29516330e-02 -4.54696149e-01
-2.22281948e-01 -1.21532969e-01 2.65994549e-01 -9.15050447e-01
1.94982246e-01 -9.38129842e-01 3.85608971e-01 6.41687870e-01
-6.84210837e-01 2.29124054e-01 5.45208715e-02 6.63544595e-01
-1.66279435e-01 -4.69281703e-01 5.15976131e-01 -8.68019015e-02
-4.51496392e-01 3.11766833e-01 -7.49376178e-01 -6.29224181e-01
1.30194271e+00 -1.75804704e-01 2.14827582e-01 -8.54425013e-01
-1.21967530e+00 4.71686810e-01 2.82401353e-01 5.94791055e-01
3.22945952e-01 -9.74929690e-01 -7.00047851e-01 2.42239356e-01
-1.11209854e-01 -2.39093468e-01 3.18850100e-01 6.95960045e-01
-3.94738048e-01 7.29028642e-01 -8.23187903e-02 -5.82740963e-01
-8.53240013e-01 6.91436410e-01 2.83905923e-01 -5.81670523e-01
-4.95599329e-01 1.18075633e+00 -4.53098029e-01 -7.09316805e-02
6.81640282e-02 -3.53412807e-01 -3.80580515e-01 4.30483848e-01
4.34602983e-03 1.19849049e-01 1.94296718e-01 -6.46730125e-01
-1.66698396e-01 -1.39703736e-01 -8.75140607e-01 4.83431704e-02
1.83702183e+00 -1.01377800e-01 2.81455606e-01 8.06541800e-01
1.43882298e+00 -6.98833704e-01 -1.45745873e+00 -3.31871994e-02
2.99837410e-01 -3.52599680e-01 5.82593799e-01 -9.61626470e-01
-1.15618932e+00 8.28635693e-01 7.19631672e-01 9.87824500e-01
1.03212559e+00 1.95250869e-01 1.26396346e+00 -1.69850081e-01
3.14497620e-01 -1.32559478e+00 -5.30847237e-02 1.63293198e-01
3.92733753e-01 -1.41905963e+00 -3.44533384e-01 3.27686876e-01
-1.04577410e+00 1.08055103e+00 2.97632767e-03 -4.44411188e-01
1.04973245e+00 1.47749513e-01 1.54487863e-01 -4.60625619e-01
-9.34866965e-01 -8.82055145e-03 -2.06535995e-01 3.47827524e-01
5.11840522e-01 4.73153055e-01 -3.31954777e-01 1.09573352e+00
-4.83784229e-01 -2.28353068e-01 5.34560561e-01 9.17936802e-01
-5.63506842e-01 -1.50064766e+00 2.86194295e-01 1.05817699e+00
-5.82539976e-01 -1.62183478e-01 -7.45480508e-02 7.56351948e-01
-1.74886197e-01 1.07597232e+00 2.40665421e-01 -8.06620598e-01
4.57842261e-01 7.98016712e-02 2.53728777e-01 -4.42005247e-01
-8.94366264e-01 5.47657192e-01 2.98932254e-01 -3.13703060e-01
9.76043642e-02 -1.24566650e+00 -1.28273761e+00 -8.24430764e-01
-7.66108632e-01 6.97815269e-02 9.34247255e-01 7.80673146e-01
2.17390090e-01 5.62562704e-01 9.96928334e-01 -6.29613578e-01
-1.23920333e+00 -1.00477147e+00 -1.08770323e+00 3.42650563e-01
1.65479377e-01 -5.68375826e-01 -5.64856470e-01 -2.78892279e-01] | [9.393022537231445, 5.881396770477295] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.